Adding upstream version 6.1.76.upstream/6.1.76

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

76 files changed, 63662 insertions, 0 deletions

diff --git a/fs/xfs/libxfs/xfs_ag.c b/fs/xfs/libxfs/xfs_ag.c
new file mode 100644
index 000000000..bb0c700af
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_ag.c
@@ -0,0 +1,1042 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * Copyright (c) 2018 Red Hat, Inc.
+ * All rights reserved.
+ */
+
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_bit.h"
+#include "xfs_sb.h"
+#include "xfs_mount.h"
+#include "xfs_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_rmap_btree.h"
+#include "xfs_alloc.h"
+#include "xfs_ialloc.h"
+#include "xfs_rmap.h"
+#include "xfs_ag.h"
+#include "xfs_ag_resv.h"
+#include "xfs_health.h"
+#include "xfs_error.h"
+#include "xfs_bmap.h"
+#include "xfs_defer.h"
+#include "xfs_log_format.h"
+#include "xfs_trans.h"
+#include "xfs_trace.h"
+#include "xfs_inode.h"
+#include "xfs_icache.h"
+
+
+/*
+ * Passive reference counting access wrappers to the perag structures.  If the
+ * per-ag structure is to be freed, the freeing code is responsible for cleaning
+ * up objects with passive references before freeing the structure. This is
+ * things like cached buffers.
+ */
+struct xfs_perag *
+xfs_perag_get(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		agno)
+{
+	struct xfs_perag	*pag;
+	int			ref = 0;
+
+	rcu_read_lock();
+	pag = radix_tree_lookup(&mp->m_perag_tree, agno);
+	if (pag) {
+		ASSERT(atomic_read(&pag->pag_ref) >= 0);
+		ref = atomic_inc_return(&pag->pag_ref);
+	}
+	rcu_read_unlock();
+	trace_xfs_perag_get(mp, agno, ref, _RET_IP_);
+	return pag;
+}
+
+/*
+ * search from @first to find the next perag with the given tag set.
+ */
+struct xfs_perag *
+xfs_perag_get_tag(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		first,
+	unsigned int		tag)
+{
+	struct xfs_perag	*pag;
+	int			found;
+	int			ref;
+
+	rcu_read_lock();
+	found = radix_tree_gang_lookup_tag(&mp->m_perag_tree,
+					(void **)&pag, first, 1, tag);
+	if (found <= 0) {
+		rcu_read_unlock();
+		return NULL;
+	}
+	ref = atomic_inc_return(&pag->pag_ref);
+	rcu_read_unlock();
+	trace_xfs_perag_get_tag(mp, pag->pag_agno, ref, _RET_IP_);
+	return pag;
+}
+
+void
+xfs_perag_put(
+	struct xfs_perag	*pag)
+{
+	int	ref;
+
+	ASSERT(atomic_read(&pag->pag_ref) > 0);
+	ref = atomic_dec_return(&pag->pag_ref);
+	trace_xfs_perag_put(pag->pag_mount, pag->pag_agno, ref, _RET_IP_);
+}
+
+/*
+ * xfs_initialize_perag_data
+ *
+ * Read in each per-ag structure so we can count up the number of
+ * allocated inodes, free inodes and used filesystem blocks as this
+ * information is no longer persistent in the superblock. Once we have
+ * this information, write it into the in-core superblock structure.
+ */
+int
+xfs_initialize_perag_data(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		agcount)
+{
+	xfs_agnumber_t		index;
+	struct xfs_perag	*pag;
+	struct xfs_sb		*sbp = &mp->m_sb;
+	uint64_t		ifree = 0;
+	uint64_t		ialloc = 0;
+	uint64_t		bfree = 0;
+	uint64_t		bfreelst = 0;
+	uint64_t		btree = 0;
+	uint64_t		fdblocks;
+	int			error = 0;
+
+	for (index = 0; index < agcount; index++) {
+		/*
+		 * Read the AGF and AGI buffers to populate the per-ag
+		 * structures for us.
+		 */
+		pag = xfs_perag_get(mp, index);
+		error = xfs_alloc_read_agf(pag, NULL, 0, NULL);
+		if (!error)
+			error = xfs_ialloc_read_agi(pag, NULL, NULL);
+		if (error) {
+			xfs_perag_put(pag);
+			return error;
+		}
+
+		ifree += pag->pagi_freecount;
+		ialloc += pag->pagi_count;
+		bfree += pag->pagf_freeblks;
+		bfreelst += pag->pagf_flcount;
+		btree += pag->pagf_btreeblks;
+		xfs_perag_put(pag);
+	}
+	fdblocks = bfree + bfreelst + btree;
+
+	/*
+	 * If the new summary counts are obviously incorrect, fail the
+	 * mount operation because that implies the AGFs are also corrupt.
+	 * Clear FS_COUNTERS so that we don't unmount with a dirty log, which
+	 * will prevent xfs_repair from fixing anything.
+	 */
+	if (fdblocks > sbp->sb_dblocks || ifree > ialloc) {
+		xfs_alert(mp, "AGF corruption. Please run xfs_repair.");
+		error = -EFSCORRUPTED;
+		goto out;
+	}
+
+	/* Overwrite incore superblock counters with just-read data */
+	spin_lock(&mp->m_sb_lock);
+	sbp->sb_ifree = ifree;
+	sbp->sb_icount = ialloc;
+	sbp->sb_fdblocks = fdblocks;
+	spin_unlock(&mp->m_sb_lock);
+
+	xfs_reinit_percpu_counters(mp);
+out:
+	xfs_fs_mark_healthy(mp, XFS_SICK_FS_COUNTERS);
+	return error;
+}
+
+STATIC void
+__xfs_free_perag(
+	struct rcu_head	*head)
+{
+	struct xfs_perag *pag = container_of(head, struct xfs_perag, rcu_head);
+
+	ASSERT(!delayed_work_pending(&pag->pag_blockgc_work));
+	kmem_free(pag);
+}
+
+/*
+ * Free up the per-ag resources associated with the mount structure.
+ */
+void
+xfs_free_perag(
+	struct xfs_mount	*mp)
+{
+	struct xfs_perag	*pag;
+	xfs_agnumber_t		agno;
+
+	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
+		spin_lock(&mp->m_perag_lock);
+		pag = radix_tree_delete(&mp->m_perag_tree, agno);
+		spin_unlock(&mp->m_perag_lock);
+		ASSERT(pag);
+		XFS_IS_CORRUPT(pag->pag_mount, atomic_read(&pag->pag_ref) != 0);
+
+		cancel_delayed_work_sync(&pag->pag_blockgc_work);
+		xfs_buf_hash_destroy(pag);
+
+		call_rcu(&pag->rcu_head, __xfs_free_perag);
+	}
+}
+
+/* Find the size of the AG, in blocks. */
+static xfs_agblock_t
+__xfs_ag_block_count(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		agno,
+	xfs_agnumber_t		agcount,
+	xfs_rfsblock_t		dblocks)
+{
+	ASSERT(agno < agcount);
+
+	if (agno < agcount - 1)
+		return mp->m_sb.sb_agblocks;
+	return dblocks - (agno * mp->m_sb.sb_agblocks);
+}
+
+xfs_agblock_t
+xfs_ag_block_count(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		agno)
+{
+	return __xfs_ag_block_count(mp, agno, mp->m_sb.sb_agcount,
+			mp->m_sb.sb_dblocks);
+}
+
+/* Calculate the first and last possible inode number in an AG. */
+static void
+__xfs_agino_range(
+	struct xfs_mount	*mp,
+	xfs_agblock_t		eoag,
+	xfs_agino_t		*first,
+	xfs_agino_t		*last)
+{
+	xfs_agblock_t		bno;
+
+	/*
+	 * Calculate the first inode, which will be in the first
+	 * cluster-aligned block after the AGFL.
+	 */
+	bno = round_up(XFS_AGFL_BLOCK(mp) + 1, M_IGEO(mp)->cluster_align);
+	*first = XFS_AGB_TO_AGINO(mp, bno);
+
+	/*
+	 * Calculate the last inode, which will be at the end of the
+	 * last (aligned) cluster that can be allocated in the AG.
+	 */
+	bno = round_down(eoag, M_IGEO(mp)->cluster_align);
+	*last = XFS_AGB_TO_AGINO(mp, bno) - 1;
+}
+
+void
+xfs_agino_range(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		agno,
+	xfs_agino_t		*first,
+	xfs_agino_t		*last)
+{
+	return __xfs_agino_range(mp, xfs_ag_block_count(mp, agno), first, last);
+}
+
+int
+xfs_initialize_perag(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		agcount,
+	xfs_rfsblock_t		dblocks,
+	xfs_agnumber_t		*maxagi)
+{
+	struct xfs_perag	*pag;
+	xfs_agnumber_t		index;
+	xfs_agnumber_t		first_initialised = NULLAGNUMBER;
+	int			error;
+
+	/*
+	 * Walk the current per-ag tree so we don't try to initialise AGs
+	 * that already exist (growfs case). Allocate and insert all the
+	 * AGs we don't find ready for initialisation.
+	 */
+	for (index = 0; index < agcount; index++) {
+		pag = xfs_perag_get(mp, index);
+		if (pag) {
+			xfs_perag_put(pag);
+			continue;
+		}
+
+		pag = kmem_zalloc(sizeof(*pag), KM_MAYFAIL);
+		if (!pag) {
+			error = -ENOMEM;
+			goto out_unwind_new_pags;
+		}
+		pag->pag_agno = index;
+		pag->pag_mount = mp;
+
+		error = radix_tree_preload(GFP_NOFS);
+		if (error)
+			goto out_free_pag;
+
+		spin_lock(&mp->m_perag_lock);
+		if (radix_tree_insert(&mp->m_perag_tree, index, pag)) {
+			WARN_ON_ONCE(1);
+			spin_unlock(&mp->m_perag_lock);
+			radix_tree_preload_end();
+			error = -EEXIST;
+			goto out_free_pag;
+		}
+		spin_unlock(&mp->m_perag_lock);
+		radix_tree_preload_end();
+
+#ifdef __KERNEL__
+		/* Place kernel structure only init below this point. */
+		spin_lock_init(&pag->pag_ici_lock);
+		spin_lock_init(&pag->pagb_lock);
+		spin_lock_init(&pag->pag_state_lock);
+		INIT_DELAYED_WORK(&pag->pag_blockgc_work, xfs_blockgc_worker);
+		INIT_RADIX_TREE(&pag->pag_ici_root, GFP_ATOMIC);
+		init_waitqueue_head(&pag->pagb_wait);
+		pag->pagb_count = 0;
+		pag->pagb_tree = RB_ROOT;
+#endif /* __KERNEL__ */
+
+		error = xfs_buf_hash_init(pag);
+		if (error)
+			goto out_remove_pag;
+
+		/* first new pag is fully initialized */
+		if (first_initialised == NULLAGNUMBER)
+			first_initialised = index;
+
+		/*
+		 * Pre-calculated geometry
+		 */
+		pag->block_count = __xfs_ag_block_count(mp, index, agcount,
+				dblocks);
+		pag->min_block = XFS_AGFL_BLOCK(mp);
+		__xfs_agino_range(mp, pag->block_count, &pag->agino_min,
+				&pag->agino_max);
+	}
+
+	index = xfs_set_inode_alloc(mp, agcount);
+
+	if (maxagi)
+		*maxagi = index;
+
+	mp->m_ag_prealloc_blocks = xfs_prealloc_blocks(mp);
+	return 0;
+
+out_remove_pag:
+	radix_tree_delete(&mp->m_perag_tree, index);
+out_free_pag:
+	kmem_free(pag);
+out_unwind_new_pags:
+	/* unwind any prior newly initialized pags */
+	for (index = first_initialised; index < agcount; index++) {
+		pag = radix_tree_delete(&mp->m_perag_tree, index);
+		if (!pag)
+			break;
+		xfs_buf_hash_destroy(pag);
+		kmem_free(pag);
+	}
+	return error;
+}
+
+static int
+xfs_get_aghdr_buf(
+	struct xfs_mount	*mp,
+	xfs_daddr_t		blkno,
+	size_t			numblks,
+	struct xfs_buf		**bpp,
+	const struct xfs_buf_ops *ops)
+{
+	struct xfs_buf		*bp;
+	int			error;
+
+	error = xfs_buf_get_uncached(mp->m_ddev_targp, numblks, 0, &bp);
+	if (error)
+		return error;
+
+	bp->b_maps[0].bm_bn = blkno;
+	bp->b_ops = ops;
+
+	*bpp = bp;
+	return 0;
+}
+
+/*
+ * Generic btree root block init function
+ */
+static void
+xfs_btroot_init(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp,
+	struct aghdr_init_data	*id)
+{
+	xfs_btree_init_block(mp, bp, id->type, 0, 0, id->agno);
+}
+
+/* Finish initializing a free space btree. */
+static void
+xfs_freesp_init_recs(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp,
+	struct aghdr_init_data	*id)
+{
+	struct xfs_alloc_rec	*arec;
+	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
+
+	arec = XFS_ALLOC_REC_ADDR(mp, XFS_BUF_TO_BLOCK(bp), 1);
+	arec->ar_startblock = cpu_to_be32(mp->m_ag_prealloc_blocks);
+
+	if (xfs_ag_contains_log(mp, id->agno)) {
+		struct xfs_alloc_rec	*nrec;
+		xfs_agblock_t		start = XFS_FSB_TO_AGBNO(mp,
+							mp->m_sb.sb_logstart);
+
+		ASSERT(start >= mp->m_ag_prealloc_blocks);
+		if (start != mp->m_ag_prealloc_blocks) {
+			/*
+			 * Modify first record to pad stripe align of log
+			 */
+			arec->ar_blockcount = cpu_to_be32(start -
+						mp->m_ag_prealloc_blocks);
+			nrec = arec + 1;
+
+			/*
+			 * Insert second record at start of internal log
+			 * which then gets trimmed.
+			 */
+			nrec->ar_startblock = cpu_to_be32(
+					be32_to_cpu(arec->ar_startblock) +
+					be32_to_cpu(arec->ar_blockcount));
+			arec = nrec;
+			be16_add_cpu(&block->bb_numrecs, 1);
+		}
+		/*
+		 * Change record start to after the internal log
+		 */
+		be32_add_cpu(&arec->ar_startblock, mp->m_sb.sb_logblocks);
+	}
+
+	/*
+	 * Calculate the record block count and check for the case where
+	 * the log might have consumed all available space in the AG. If
+	 * so, reset the record count to 0 to avoid exposure of an invalid
+	 * record start block.
+	 */
+	arec->ar_blockcount = cpu_to_be32(id->agsize -
+					  be32_to_cpu(arec->ar_startblock));
+	if (!arec->ar_blockcount)
+		block->bb_numrecs = 0;
+}
+
+/*
+ * Alloc btree root block init functions
+ */
+static void
+xfs_bnoroot_init(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp,
+	struct aghdr_init_data	*id)
+{
+	xfs_btree_init_block(mp, bp, XFS_BTNUM_BNO, 0, 1, id->agno);
+	xfs_freesp_init_recs(mp, bp, id);
+}
+
+static void
+xfs_cntroot_init(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp,
+	struct aghdr_init_data	*id)
+{
+	xfs_btree_init_block(mp, bp, XFS_BTNUM_CNT, 0, 1, id->agno);
+	xfs_freesp_init_recs(mp, bp, id);
+}
+
+/*
+ * Reverse map root block init
+ */
+static void
+xfs_rmaproot_init(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp,
+	struct aghdr_init_data	*id)
+{
+	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
+	struct xfs_rmap_rec	*rrec;
+
+	xfs_btree_init_block(mp, bp, XFS_BTNUM_RMAP, 0, 4, id->agno);
+
+	/*
+	 * mark the AG header regions as static metadata The BNO
+	 * btree block is the first block after the headers, so
+	 * it's location defines the size of region the static
+	 * metadata consumes.
+	 *
+	 * Note: unlike mkfs, we never have to account for log
+	 * space when growing the data regions
+	 */
+	rrec = XFS_RMAP_REC_ADDR(block, 1);
+	rrec->rm_startblock = 0;
+	rrec->rm_blockcount = cpu_to_be32(XFS_BNO_BLOCK(mp));
+	rrec->rm_owner = cpu_to_be64(XFS_RMAP_OWN_FS);
+	rrec->rm_offset = 0;
+
+	/* account freespace btree root blocks */
+	rrec = XFS_RMAP_REC_ADDR(block, 2);
+	rrec->rm_startblock = cpu_to_be32(XFS_BNO_BLOCK(mp));
+	rrec->rm_blockcount = cpu_to_be32(2);
+	rrec->rm_owner = cpu_to_be64(XFS_RMAP_OWN_AG);
+	rrec->rm_offset = 0;
+
+	/* account inode btree root blocks */
+	rrec = XFS_RMAP_REC_ADDR(block, 3);
+	rrec->rm_startblock = cpu_to_be32(XFS_IBT_BLOCK(mp));
+	rrec->rm_blockcount = cpu_to_be32(XFS_RMAP_BLOCK(mp) -
+					  XFS_IBT_BLOCK(mp));
+	rrec->rm_owner = cpu_to_be64(XFS_RMAP_OWN_INOBT);
+	rrec->rm_offset = 0;
+
+	/* account for rmap btree root */
+	rrec = XFS_RMAP_REC_ADDR(block, 4);
+	rrec->rm_startblock = cpu_to_be32(XFS_RMAP_BLOCK(mp));
+	rrec->rm_blockcount = cpu_to_be32(1);
+	rrec->rm_owner = cpu_to_be64(XFS_RMAP_OWN_AG);
+	rrec->rm_offset = 0;
+
+	/* account for refc btree root */
+	if (xfs_has_reflink(mp)) {
+		rrec = XFS_RMAP_REC_ADDR(block, 5);
+		rrec->rm_startblock = cpu_to_be32(xfs_refc_block(mp));
+		rrec->rm_blockcount = cpu_to_be32(1);
+		rrec->rm_owner = cpu_to_be64(XFS_RMAP_OWN_REFC);
+		rrec->rm_offset = 0;
+		be16_add_cpu(&block->bb_numrecs, 1);
+	}
+
+	/* account for the log space */
+	if (xfs_ag_contains_log(mp, id->agno)) {
+		rrec = XFS_RMAP_REC_ADDR(block,
+				be16_to_cpu(block->bb_numrecs) + 1);
+		rrec->rm_startblock = cpu_to_be32(
+				XFS_FSB_TO_AGBNO(mp, mp->m_sb.sb_logstart));
+		rrec->rm_blockcount = cpu_to_be32(mp->m_sb.sb_logblocks);
+		rrec->rm_owner = cpu_to_be64(XFS_RMAP_OWN_LOG);
+		rrec->rm_offset = 0;
+		be16_add_cpu(&block->bb_numrecs, 1);
+	}
+}
+
+/*
+ * Initialise new secondary superblocks with the pre-grow geometry, but mark
+ * them as "in progress" so we know they haven't yet been activated. This will
+ * get cleared when the update with the new geometry information is done after
+ * changes to the primary are committed. This isn't strictly necessary, but we
+ * get it for free with the delayed buffer write lists and it means we can tell
+ * if a grow operation didn't complete properly after the fact.
+ */
+static void
+xfs_sbblock_init(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp,
+	struct aghdr_init_data	*id)
+{
+	struct xfs_dsb		*dsb = bp->b_addr;
+
+	xfs_sb_to_disk(dsb, &mp->m_sb);
+	dsb->sb_inprogress = 1;
+}
+
+static void
+xfs_agfblock_init(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp,
+	struct aghdr_init_data	*id)
+{
+	struct xfs_agf		*agf = bp->b_addr;
+	xfs_extlen_t		tmpsize;
+
+	agf->agf_magicnum = cpu_to_be32(XFS_AGF_MAGIC);
+	agf->agf_versionnum = cpu_to_be32(XFS_AGF_VERSION);
+	agf->agf_seqno = cpu_to_be32(id->agno);
+	agf->agf_length = cpu_to_be32(id->agsize);
+	agf->agf_roots[XFS_BTNUM_BNOi] = cpu_to_be32(XFS_BNO_BLOCK(mp));
+	agf->agf_roots[XFS_BTNUM_CNTi] = cpu_to_be32(XFS_CNT_BLOCK(mp));
+	agf->agf_levels[XFS_BTNUM_BNOi] = cpu_to_be32(1);
+	agf->agf_levels[XFS_BTNUM_CNTi] = cpu_to_be32(1);
+	if (xfs_has_rmapbt(mp)) {
+		agf->agf_roots[XFS_BTNUM_RMAPi] =
+					cpu_to_be32(XFS_RMAP_BLOCK(mp));
+		agf->agf_levels[XFS_BTNUM_RMAPi] = cpu_to_be32(1);
+		agf->agf_rmap_blocks = cpu_to_be32(1);
+	}
+
+	agf->agf_flfirst = cpu_to_be32(1);
+	agf->agf_fllast = 0;
+	agf->agf_flcount = 0;
+	tmpsize = id->agsize - mp->m_ag_prealloc_blocks;
+	agf->agf_freeblks = cpu_to_be32(tmpsize);
+	agf->agf_longest = cpu_to_be32(tmpsize);
+	if (xfs_has_crc(mp))
+		uuid_copy(&agf->agf_uuid, &mp->m_sb.sb_meta_uuid);
+	if (xfs_has_reflink(mp)) {
+		agf->agf_refcount_root = cpu_to_be32(
+				xfs_refc_block(mp));
+		agf->agf_refcount_level = cpu_to_be32(1);
+		agf->agf_refcount_blocks = cpu_to_be32(1);
+	}
+
+	if (xfs_ag_contains_log(mp, id->agno)) {
+		int64_t	logblocks = mp->m_sb.sb_logblocks;
+
+		be32_add_cpu(&agf->agf_freeblks, -logblocks);
+		agf->agf_longest = cpu_to_be32(id->agsize -
+			XFS_FSB_TO_AGBNO(mp, mp->m_sb.sb_logstart) - logblocks);
+	}
+}
+
+static void
+xfs_agflblock_init(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp,
+	struct aghdr_init_data	*id)
+{
+	struct xfs_agfl		*agfl = XFS_BUF_TO_AGFL(bp);
+	__be32			*agfl_bno;
+	int			bucket;
+
+	if (xfs_has_crc(mp)) {
+		agfl->agfl_magicnum = cpu_to_be32(XFS_AGFL_MAGIC);
+		agfl->agfl_seqno = cpu_to_be32(id->agno);
+		uuid_copy(&agfl->agfl_uuid, &mp->m_sb.sb_meta_uuid);
+	}
+
+	agfl_bno = xfs_buf_to_agfl_bno(bp);
+	for (bucket = 0; bucket < xfs_agfl_size(mp); bucket++)
+		agfl_bno[bucket] = cpu_to_be32(NULLAGBLOCK);
+}
+
+static void
+xfs_agiblock_init(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp,
+	struct aghdr_init_data	*id)
+{
+	struct xfs_agi		*agi = bp->b_addr;
+	int			bucket;
+
+	agi->agi_magicnum = cpu_to_be32(XFS_AGI_MAGIC);
+	agi->agi_versionnum = cpu_to_be32(XFS_AGI_VERSION);
+	agi->agi_seqno = cpu_to_be32(id->agno);
+	agi->agi_length = cpu_to_be32(id->agsize);
+	agi->agi_count = 0;
+	agi->agi_root = cpu_to_be32(XFS_IBT_BLOCK(mp));
+	agi->agi_level = cpu_to_be32(1);
+	agi->agi_freecount = 0;
+	agi->agi_newino = cpu_to_be32(NULLAGINO);
+	agi->agi_dirino = cpu_to_be32(NULLAGINO);
+	if (xfs_has_crc(mp))
+		uuid_copy(&agi->agi_uuid, &mp->m_sb.sb_meta_uuid);
+	if (xfs_has_finobt(mp)) {
+		agi->agi_free_root = cpu_to_be32(XFS_FIBT_BLOCK(mp));
+		agi->agi_free_level = cpu_to_be32(1);
+	}
+	for (bucket = 0; bucket < XFS_AGI_UNLINKED_BUCKETS; bucket++)
+		agi->agi_unlinked[bucket] = cpu_to_be32(NULLAGINO);
+	if (xfs_has_inobtcounts(mp)) {
+		agi->agi_iblocks = cpu_to_be32(1);
+		if (xfs_has_finobt(mp))
+			agi->agi_fblocks = cpu_to_be32(1);
+	}
+}
+
+typedef void (*aghdr_init_work_f)(struct xfs_mount *mp, struct xfs_buf *bp,
+				  struct aghdr_init_data *id);
+static int
+xfs_ag_init_hdr(
+	struct xfs_mount	*mp,
+	struct aghdr_init_data	*id,
+	aghdr_init_work_f	work,
+	const struct xfs_buf_ops *ops)
+{
+	struct xfs_buf		*bp;
+	int			error;
+
+	error = xfs_get_aghdr_buf(mp, id->daddr, id->numblks, &bp, ops);
+	if (error)
+		return error;
+
+	(*work)(mp, bp, id);
+
+	xfs_buf_delwri_queue(bp, &id->buffer_list);
+	xfs_buf_relse(bp);
+	return 0;
+}
+
+struct xfs_aghdr_grow_data {
+	xfs_daddr_t		daddr;
+	size_t			numblks;
+	const struct xfs_buf_ops *ops;
+	aghdr_init_work_f	work;
+	xfs_btnum_t		type;
+	bool			need_init;
+};
+
+/*
+ * Prepare new AG headers to be written to disk. We use uncached buffers here,
+ * as it is assumed these new AG headers are currently beyond the currently
+ * valid filesystem address space. Using cached buffers would trip over EOFS
+ * corruption detection alogrithms in the buffer cache lookup routines.
+ *
+ * This is a non-transactional function, but the prepared buffers are added to a
+ * delayed write buffer list supplied by the caller so they can submit them to
+ * disk and wait on them as required.
+ */
+int
+xfs_ag_init_headers(
+	struct xfs_mount	*mp,
+	struct aghdr_init_data	*id)
+
+{
+	struct xfs_aghdr_grow_data aghdr_data[] = {
+	{ /* SB */
+		.daddr = XFS_AG_DADDR(mp, id->agno, XFS_SB_DADDR),
+		.numblks = XFS_FSS_TO_BB(mp, 1),
+		.ops = &xfs_sb_buf_ops,
+		.work = &xfs_sbblock_init,
+		.need_init = true
+	},
+	{ /* AGF */
+		.daddr = XFS_AG_DADDR(mp, id->agno, XFS_AGF_DADDR(mp)),
+		.numblks = XFS_FSS_TO_BB(mp, 1),
+		.ops = &xfs_agf_buf_ops,
+		.work = &xfs_agfblock_init,
+		.need_init = true
+	},
+	{ /* AGFL */
+		.daddr = XFS_AG_DADDR(mp, id->agno, XFS_AGFL_DADDR(mp)),
+		.numblks = XFS_FSS_TO_BB(mp, 1),
+		.ops = &xfs_agfl_buf_ops,
+		.work = &xfs_agflblock_init,
+		.need_init = true
+	},
+	{ /* AGI */
+		.daddr = XFS_AG_DADDR(mp, id->agno, XFS_AGI_DADDR(mp)),
+		.numblks = XFS_FSS_TO_BB(mp, 1),
+		.ops = &xfs_agi_buf_ops,
+		.work = &xfs_agiblock_init,
+		.need_init = true
+	},
+	{ /* BNO root block */
+		.daddr = XFS_AGB_TO_DADDR(mp, id->agno, XFS_BNO_BLOCK(mp)),
+		.numblks = BTOBB(mp->m_sb.sb_blocksize),
+		.ops = &xfs_bnobt_buf_ops,
+		.work = &xfs_bnoroot_init,
+		.need_init = true
+	},
+	{ /* CNT root block */
+		.daddr = XFS_AGB_TO_DADDR(mp, id->agno, XFS_CNT_BLOCK(mp)),
+		.numblks = BTOBB(mp->m_sb.sb_blocksize),
+		.ops = &xfs_cntbt_buf_ops,
+		.work = &xfs_cntroot_init,
+		.need_init = true
+	},
+	{ /* INO root block */
+		.daddr = XFS_AGB_TO_DADDR(mp, id->agno, XFS_IBT_BLOCK(mp)),
+		.numblks = BTOBB(mp->m_sb.sb_blocksize),
+		.ops = &xfs_inobt_buf_ops,
+		.work = &xfs_btroot_init,
+		.type = XFS_BTNUM_INO,
+		.need_init = true
+	},
+	{ /* FINO root block */
+		.daddr = XFS_AGB_TO_DADDR(mp, id->agno, XFS_FIBT_BLOCK(mp)),
+		.numblks = BTOBB(mp->m_sb.sb_blocksize),
+		.ops = &xfs_finobt_buf_ops,
+		.work = &xfs_btroot_init,
+		.type = XFS_BTNUM_FINO,
+		.need_init =  xfs_has_finobt(mp)
+	},
+	{ /* RMAP root block */
+		.daddr = XFS_AGB_TO_DADDR(mp, id->agno, XFS_RMAP_BLOCK(mp)),
+		.numblks = BTOBB(mp->m_sb.sb_blocksize),
+		.ops = &xfs_rmapbt_buf_ops,
+		.work = &xfs_rmaproot_init,
+		.need_init = xfs_has_rmapbt(mp)
+	},
+	{ /* REFC root block */
+		.daddr = XFS_AGB_TO_DADDR(mp, id->agno, xfs_refc_block(mp)),
+		.numblks = BTOBB(mp->m_sb.sb_blocksize),
+		.ops = &xfs_refcountbt_buf_ops,
+		.work = &xfs_btroot_init,
+		.type = XFS_BTNUM_REFC,
+		.need_init = xfs_has_reflink(mp)
+	},
+	{ /* NULL terminating block */
+		.daddr = XFS_BUF_DADDR_NULL,
+	}
+	};
+	struct  xfs_aghdr_grow_data *dp;
+	int			error = 0;
+
+	/* Account for AG free space in new AG */
+	id->nfree += id->agsize - mp->m_ag_prealloc_blocks;
+	for (dp = &aghdr_data[0]; dp->daddr != XFS_BUF_DADDR_NULL; dp++) {
+		if (!dp->need_init)
+			continue;
+
+		id->daddr = dp->daddr;
+		id->numblks = dp->numblks;
+		id->type = dp->type;
+		error = xfs_ag_init_hdr(mp, id, dp->work, dp->ops);
+		if (error)
+			break;
+	}
+	return error;
+}
+
+int
+xfs_ag_shrink_space(
+	struct xfs_perag	*pag,
+	struct xfs_trans	**tpp,
+	xfs_extlen_t		delta)
+{
+	struct xfs_mount	*mp = pag->pag_mount;
+	struct xfs_alloc_arg	args = {
+		.tp	= *tpp,
+		.mp	= mp,
+		.type	= XFS_ALLOCTYPE_THIS_BNO,
+		.minlen = delta,
+		.maxlen = delta,
+		.oinfo	= XFS_RMAP_OINFO_SKIP_UPDATE,
+		.resv	= XFS_AG_RESV_NONE,
+		.prod	= 1
+	};
+	struct xfs_buf		*agibp, *agfbp;
+	struct xfs_agi		*agi;
+	struct xfs_agf		*agf;
+	xfs_agblock_t		aglen;
+	int			error, err2;
+
+	ASSERT(pag->pag_agno == mp->m_sb.sb_agcount - 1);
+	error = xfs_ialloc_read_agi(pag, *tpp, &agibp);
+	if (error)
+		return error;
+
+	agi = agibp->b_addr;
+
+	error = xfs_alloc_read_agf(pag, *tpp, 0, &agfbp);
+	if (error)
+		return error;
+
+	agf = agfbp->b_addr;
+	aglen = be32_to_cpu(agi->agi_length);
+	/* some extra paranoid checks before we shrink the ag */
+	if (XFS_IS_CORRUPT(mp, agf->agf_length != agi->agi_length))
+		return -EFSCORRUPTED;
+	if (delta >= aglen)
+		return -EINVAL;
+
+	args.fsbno = XFS_AGB_TO_FSB(mp, pag->pag_agno, aglen - delta);
+
+	/*
+	 * Make sure that the last inode cluster cannot overlap with the new
+	 * end of the AG, even if it's sparse.
+	 */
+	error = xfs_ialloc_check_shrink(*tpp, pag->pag_agno, agibp,
+			aglen - delta);
+	if (error)
+		return error;
+
+	/*
+	 * Disable perag reservations so it doesn't cause the allocation request
+	 * to fail. We'll reestablish reservation before we return.
+	 */
+	error = xfs_ag_resv_free(pag);
+	if (error)
+		return error;
+
+	/* internal log shouldn't also show up in the free space btrees */
+	error = xfs_alloc_vextent(&args);
+	if (!error && args.agbno == NULLAGBLOCK)
+		error = -ENOSPC;
+
+	if (error) {
+		/*
+		 * if extent allocation fails, need to roll the transaction to
+		 * ensure that the AGFL fixup has been committed anyway.
+		 */
+		xfs_trans_bhold(*tpp, agfbp);
+		err2 = xfs_trans_roll(tpp);
+		if (err2)
+			return err2;
+		xfs_trans_bjoin(*tpp, agfbp);
+		goto resv_init_out;
+	}
+
+	/*
+	 * if successfully deleted from freespace btrees, need to confirm
+	 * per-AG reservation works as expected.
+	 */
+	be32_add_cpu(&agi->agi_length, -delta);
+	be32_add_cpu(&agf->agf_length, -delta);
+
+	err2 = xfs_ag_resv_init(pag, *tpp);
+	if (err2) {
+		be32_add_cpu(&agi->agi_length, delta);
+		be32_add_cpu(&agf->agf_length, delta);
+		if (err2 != -ENOSPC)
+			goto resv_err;
+
+		__xfs_free_extent_later(*tpp, args.fsbno, delta, NULL, true);
+
+		/*
+		 * Roll the transaction before trying to re-init the per-ag
+		 * reservation. The new transaction is clean so it will cancel
+		 * without any side effects.
+		 */
+		error = xfs_defer_finish(tpp);
+		if (error)
+			return error;
+
+		error = -ENOSPC;
+		goto resv_init_out;
+	}
+	xfs_ialloc_log_agi(*tpp, agibp, XFS_AGI_LENGTH);
+	xfs_alloc_log_agf(*tpp, agfbp, XFS_AGF_LENGTH);
+	return 0;
+
+resv_init_out:
+	err2 = xfs_ag_resv_init(pag, *tpp);
+	if (!err2)
+		return error;
+resv_err:
+	xfs_warn(mp, "Error %d reserving per-AG metadata reserve pool.", err2);
+	xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
+	return err2;
+}
+
+/*
+ * Extent the AG indicated by the @id by the length passed in
+ */
+int
+xfs_ag_extend_space(
+	struct xfs_perag	*pag,
+	struct xfs_trans	*tp,
+	xfs_extlen_t		len)
+{
+	struct xfs_buf		*bp;
+	struct xfs_agi		*agi;
+	struct xfs_agf		*agf;
+	int			error;
+
+	ASSERT(pag->pag_agno == pag->pag_mount->m_sb.sb_agcount - 1);
+
+	error = xfs_ialloc_read_agi(pag, tp, &bp);
+	if (error)
+		return error;
+
+	agi = bp->b_addr;
+	be32_add_cpu(&agi->agi_length, len);
+	xfs_ialloc_log_agi(tp, bp, XFS_AGI_LENGTH);
+
+	/*
+	 * Change agf length.
+	 */
+	error = xfs_alloc_read_agf(pag, tp, 0, &bp);
+	if (error)
+		return error;
+
+	agf = bp->b_addr;
+	be32_add_cpu(&agf->agf_length, len);
+	ASSERT(agf->agf_length == agi->agi_length);
+	xfs_alloc_log_agf(tp, bp, XFS_AGF_LENGTH);
+
+	/*
+	 * Free the new space.
+	 *
+	 * XFS_RMAP_OINFO_SKIP_UPDATE is used here to tell the rmap btree that
+	 * this doesn't actually exist in the rmap btree.
+	 */
+	error = xfs_rmap_free(tp, bp, pag, be32_to_cpu(agf->agf_length) - len,
+				len, &XFS_RMAP_OINFO_SKIP_UPDATE);
+	if (error)
+		return error;
+
+	error = xfs_free_extent(tp, XFS_AGB_TO_FSB(pag->pag_mount, pag->pag_agno,
+					be32_to_cpu(agf->agf_length) - len),
+				len, &XFS_RMAP_OINFO_SKIP_UPDATE,
+				XFS_AG_RESV_NONE);
+	if (error)
+		return error;
+
+	/* Update perag geometry */
+	pag->block_count = be32_to_cpu(agf->agf_length);
+	__xfs_agino_range(pag->pag_mount, pag->block_count, &pag->agino_min,
+				&pag->agino_max);
+	return 0;
+}
+
+/* Retrieve AG geometry. */
+int
+xfs_ag_get_geometry(
+	struct xfs_perag	*pag,
+	struct xfs_ag_geometry	*ageo)
+{
+	struct xfs_buf		*agi_bp;
+	struct xfs_buf		*agf_bp;
+	struct xfs_agi		*agi;
+	struct xfs_agf		*agf;
+	unsigned int		freeblks;
+	int			error;
+
+	/* Lock the AG headers. */
+	error = xfs_ialloc_read_agi(pag, NULL, &agi_bp);
+	if (error)
+		return error;
+	error = xfs_alloc_read_agf(pag, NULL, 0, &agf_bp);
+	if (error)
+		goto out_agi;
+
+	/* Fill out form. */
+	memset(ageo, 0, sizeof(*ageo));
+	ageo->ag_number = pag->pag_agno;
+
+	agi = agi_bp->b_addr;
+	ageo->ag_icount = be32_to_cpu(agi->agi_count);
+	ageo->ag_ifree = be32_to_cpu(agi->agi_freecount);
+
+	agf = agf_bp->b_addr;
+	ageo->ag_length = be32_to_cpu(agf->agf_length);
+	freeblks = pag->pagf_freeblks +
+		   pag->pagf_flcount +
+		   pag->pagf_btreeblks -
+		   xfs_ag_resv_needed(pag, XFS_AG_RESV_NONE);
+	ageo->ag_freeblks = freeblks;
+	xfs_ag_geom_health(pag, ageo);
+
+	/* Release resources. */
+	xfs_buf_relse(agf_bp);
+out_agi:
+	xfs_buf_relse(agi_bp);
+	return error;
+}
diff --git a/fs/xfs/libxfs/xfs_ag.h b/fs/xfs/libxfs/xfs_ag.h
new file mode 100644
index 000000000..191b22b9a
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_ag.h
@@ -0,0 +1,242 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2018 Red Hat, Inc.
+ * All rights reserved.
+ */
+
+#ifndef __LIBXFS_AG_H
+#define __LIBXFS_AG_H 1
+
+struct xfs_mount;
+struct xfs_trans;
+struct xfs_perag;
+
+/*
+ * Per-ag infrastructure
+ */
+
+/* per-AG block reservation data structures*/
+struct xfs_ag_resv {
+	/* number of blocks originally reserved here */
+	xfs_extlen_t			ar_orig_reserved;
+	/* number of blocks reserved here */
+	xfs_extlen_t			ar_reserved;
+	/* number of blocks originally asked for */
+	xfs_extlen_t			ar_asked;
+};
+
+/*
+ * Per-ag incore structure, copies of information in agf and agi, to improve the
+ * performance of allocation group selection.
+ */
+struct xfs_perag {
+	struct xfs_mount *pag_mount;	/* owner filesystem */
+	xfs_agnumber_t	pag_agno;	/* AG this structure belongs to */
+	atomic_t	pag_ref;	/* perag reference count */
+	char		pagf_init;	/* this agf's entry is initialized */
+	char		pagi_init;	/* this agi's entry is initialized */
+	char		pagf_metadata;	/* the agf is preferred to be metadata */
+	char		pagi_inodeok;	/* The agi is ok for inodes */
+	uint8_t		pagf_levels[XFS_BTNUM_AGF];
+					/* # of levels in bno & cnt btree */
+	bool		pagf_agflreset; /* agfl requires reset before use */
+	uint32_t	pagf_flcount;	/* count of blocks in freelist */
+	xfs_extlen_t	pagf_freeblks;	/* total free blocks */
+	xfs_extlen_t	pagf_longest;	/* longest free space */
+	uint32_t	pagf_btreeblks;	/* # of blocks held in AGF btrees */
+	xfs_agino_t	pagi_freecount;	/* number of free inodes */
+	xfs_agino_t	pagi_count;	/* number of allocated inodes */
+
+	/*
+	 * Inode allocation search lookup optimisation.
+	 * If the pagino matches, the search for new inodes
+	 * doesn't need to search the near ones again straight away
+	 */
+	xfs_agino_t	pagl_pagino;
+	xfs_agino_t	pagl_leftrec;
+	xfs_agino_t	pagl_rightrec;
+
+	int		pagb_count;	/* pagb slots in use */
+	uint8_t		pagf_refcount_level; /* recount btree height */
+
+	/* Blocks reserved for all kinds of metadata. */
+	struct xfs_ag_resv	pag_meta_resv;
+	/* Blocks reserved for the reverse mapping btree. */
+	struct xfs_ag_resv	pag_rmapbt_resv;
+
+	/* for rcu-safe freeing */
+	struct rcu_head	rcu_head;
+
+	/* Precalculated geometry info */
+	xfs_agblock_t		block_count;
+	xfs_agblock_t		min_block;
+	xfs_agino_t		agino_min;
+	xfs_agino_t		agino_max;
+
+#ifdef __KERNEL__
+	/* -- kernel only structures below this line -- */
+
+	/*
+	 * Bitsets of per-ag metadata that have been checked and/or are sick.
+	 * Callers should hold pag_state_lock before accessing this field.
+	 */
+	uint16_t	pag_checked;
+	uint16_t	pag_sick;
+	spinlock_t	pag_state_lock;
+
+	spinlock_t	pagb_lock;	/* lock for pagb_tree */
+	struct rb_root	pagb_tree;	/* ordered tree of busy extents */
+	unsigned int	pagb_gen;	/* generation count for pagb_tree */
+	wait_queue_head_t pagb_wait;	/* woken when pagb_gen changes */
+
+	atomic_t        pagf_fstrms;    /* # of filestreams active in this AG */
+
+	spinlock_t	pag_ici_lock;	/* incore inode cache lock */
+	struct radix_tree_root pag_ici_root;	/* incore inode cache root */
+	int		pag_ici_reclaimable;	/* reclaimable inodes */
+	unsigned long	pag_ici_reclaim_cursor;	/* reclaim restart point */
+
+	/* buffer cache index */
+	spinlock_t	pag_buf_lock;	/* lock for pag_buf_hash */
+	struct rhashtable pag_buf_hash;
+
+	/* background prealloc block trimming */
+	struct delayed_work	pag_blockgc_work;
+
+#endif /* __KERNEL__ */
+};
+
+int xfs_initialize_perag(struct xfs_mount *mp, xfs_agnumber_t agcount,
+			xfs_rfsblock_t dcount, xfs_agnumber_t *maxagi);
+int xfs_initialize_perag_data(struct xfs_mount *mp, xfs_agnumber_t agno);
+void xfs_free_perag(struct xfs_mount *mp);
+
+struct xfs_perag *xfs_perag_get(struct xfs_mount *mp, xfs_agnumber_t agno);
+struct xfs_perag *xfs_perag_get_tag(struct xfs_mount *mp, xfs_agnumber_t agno,
+		unsigned int tag);
+void xfs_perag_put(struct xfs_perag *pag);
+
+/*
+ * Per-ag geometry infomation and validation
+ */
+xfs_agblock_t xfs_ag_block_count(struct xfs_mount *mp, xfs_agnumber_t agno);
+void xfs_agino_range(struct xfs_mount *mp, xfs_agnumber_t agno,
+		xfs_agino_t *first, xfs_agino_t *last);
+
+static inline bool
+xfs_verify_agbno(struct xfs_perag *pag, xfs_agblock_t agbno)
+{
+	if (agbno >= pag->block_count)
+		return false;
+	if (agbno <= pag->min_block)
+		return false;
+	return true;
+}
+
+static inline bool
+xfs_verify_agbext(
+	struct xfs_perag	*pag,
+	xfs_agblock_t		agbno,
+	xfs_agblock_t		len)
+{
+	if (agbno + len <= agbno)
+		return false;
+
+	if (!xfs_verify_agbno(pag, agbno))
+		return false;
+
+	return xfs_verify_agbno(pag, agbno + len - 1);
+}
+
+/*
+ * Verify that an AG inode number pointer neither points outside the AG
+ * nor points at static metadata.
+ */
+static inline bool
+xfs_verify_agino(struct xfs_perag *pag, xfs_agino_t agino)
+{
+	if (agino < pag->agino_min)
+		return false;
+	if (agino > pag->agino_max)
+		return false;
+	return true;
+}
+
+/*
+ * Verify that an AG inode number pointer neither points outside the AG
+ * nor points at static metadata, or is NULLAGINO.
+ */
+static inline bool
+xfs_verify_agino_or_null(struct xfs_perag *pag, xfs_agino_t agino)
+{
+	if (agino == NULLAGINO)
+		return true;
+	return xfs_verify_agino(pag, agino);
+}
+
+static inline bool
+xfs_ag_contains_log(struct xfs_mount *mp, xfs_agnumber_t agno)
+{
+	return mp->m_sb.sb_logstart > 0 &&
+	       agno == XFS_FSB_TO_AGNO(mp, mp->m_sb.sb_logstart);
+}
+
+/*
+ * Perag iteration APIs
+ */
+static inline struct xfs_perag *
+xfs_perag_next(
+	struct xfs_perag	*pag,
+	xfs_agnumber_t		*agno,
+	xfs_agnumber_t		end_agno)
+{
+	struct xfs_mount	*mp = pag->pag_mount;
+
+	*agno = pag->pag_agno + 1;
+	xfs_perag_put(pag);
+	if (*agno > end_agno)
+		return NULL;
+	return xfs_perag_get(mp, *agno);
+}
+
+#define for_each_perag_range(mp, agno, end_agno, pag) \
+	for ((pag) = xfs_perag_get((mp), (agno)); \
+		(pag) != NULL; \
+		(pag) = xfs_perag_next((pag), &(agno), (end_agno)))
+
+#define for_each_perag_from(mp, agno, pag) \
+	for_each_perag_range((mp), (agno), (mp)->m_sb.sb_agcount - 1, (pag))
+
+
+#define for_each_perag(mp, agno, pag) \
+	(agno) = 0; \
+	for_each_perag_from((mp), (agno), (pag))
+
+#define for_each_perag_tag(mp, agno, pag, tag) \
+	for ((agno) = 0, (pag) = xfs_perag_get_tag((mp), 0, (tag)); \
+		(pag) != NULL; \
+		(agno) = (pag)->pag_agno + 1, \
+		xfs_perag_put(pag), \
+		(pag) = xfs_perag_get_tag((mp), (agno), (tag)))
+
+struct aghdr_init_data {
+	/* per ag data */
+	xfs_agblock_t		agno;		/* ag to init */
+	xfs_extlen_t		agsize;		/* new AG size */
+	struct list_head	buffer_list;	/* buffer writeback list */
+	xfs_rfsblock_t		nfree;		/* cumulative new free space */
+
+	/* per header data */
+	xfs_daddr_t		daddr;		/* header location */
+	size_t			numblks;	/* size of header */
+	xfs_btnum_t		type;		/* type of btree root block */
+};
+
+int xfs_ag_init_headers(struct xfs_mount *mp, struct aghdr_init_data *id);
+int xfs_ag_shrink_space(struct xfs_perag *pag, struct xfs_trans **tpp,
+			xfs_extlen_t delta);
+int xfs_ag_extend_space(struct xfs_perag *pag, struct xfs_trans *tp,
+			xfs_extlen_t len);
+int xfs_ag_get_geometry(struct xfs_perag *pag, struct xfs_ag_geometry *ageo);
+
+#endif /* __LIBXFS_AG_H */
diff --git a/fs/xfs/libxfs/xfs_ag_resv.c b/fs/xfs/libxfs/xfs_ag_resv.c
new file mode 100644
index 000000000..5af123d13
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_ag_resv.c
@@ -0,0 +1,426 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2016 Oracle.  All Rights Reserved.
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_alloc.h"
+#include "xfs_errortag.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+#include "xfs_trans.h"
+#include "xfs_rmap_btree.h"
+#include "xfs_btree.h"
+#include "xfs_refcount_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_ag.h"
+#include "xfs_ag_resv.h"
+
+/*
+ * Per-AG Block Reservations
+ *
+ * For some kinds of allocation group metadata structures, it is advantageous
+ * to reserve a small number of blocks in each AG so that future expansions of
+ * that data structure do not encounter ENOSPC because errors during a btree
+ * split cause the filesystem to go offline.
+ *
+ * Prior to the introduction of reflink, this wasn't an issue because the free
+ * space btrees maintain a reserve of space (the AGFL) to handle any expansion
+ * that may be necessary; and allocations of other metadata (inodes, BMBT,
+ * dir/attr) aren't restricted to a single AG.  However, with reflink it is
+ * possible to allocate all the space in an AG, have subsequent reflink/CoW
+ * activity expand the refcount btree, and discover that there's no space left
+ * to handle that expansion.  Since we can calculate the maximum size of the
+ * refcount btree, we can reserve space for it and avoid ENOSPC.
+ *
+ * Handling per-AG reservations consists of three changes to the allocator's
+ * behavior:  First, because these reservations are always needed, we decrease
+ * the ag_max_usable counter to reflect the size of the AG after the reserved
+ * blocks are taken.  Second, the reservations must be reflected in the
+ * fdblocks count to maintain proper accounting.  Third, each AG must maintain
+ * its own reserved block counter so that we can calculate the amount of space
+ * that must remain free to maintain the reservations.  Fourth, the "remaining
+ * reserved blocks" count must be used when calculating the length of the
+ * longest free extent in an AG and to clamp maxlen in the per-AG allocation
+ * functions.  In other words, we maintain a virtual allocation via in-core
+ * accounting tricks so that we don't have to clean up after a crash. :)
+ *
+ * Reserved blocks can be managed by passing one of the enum xfs_ag_resv_type
+ * values via struct xfs_alloc_arg or directly to the xfs_free_extent
+ * function.  It might seem a little funny to maintain a reservoir of blocks
+ * to feed another reservoir, but the AGFL only holds enough blocks to get
+ * through the next transaction.  The per-AG reservation is to ensure (we
+ * hope) that each AG never runs out of blocks.  Each data structure wanting
+ * to use the reservation system should update ask/used in xfs_ag_resv_init.
+ */
+
+/*
+ * Are we critically low on blocks?  For now we'll define that as the number
+ * of blocks we can get our hands on being less than 10% of what we reserved
+ * or less than some arbitrary number (maximum btree height).
+ */
+bool
+xfs_ag_resv_critical(
+	struct xfs_perag		*pag,
+	enum xfs_ag_resv_type		type)
+{
+	xfs_extlen_t			avail;
+	xfs_extlen_t			orig;
+
+	switch (type) {
+	case XFS_AG_RESV_METADATA:
+		avail = pag->pagf_freeblks - pag->pag_rmapbt_resv.ar_reserved;
+		orig = pag->pag_meta_resv.ar_asked;
+		break;
+	case XFS_AG_RESV_RMAPBT:
+		avail = pag->pagf_freeblks + pag->pagf_flcount -
+			pag->pag_meta_resv.ar_reserved;
+		orig = pag->pag_rmapbt_resv.ar_asked;
+		break;
+	default:
+		ASSERT(0);
+		return false;
+	}
+
+	trace_xfs_ag_resv_critical(pag, type, avail);
+
+	/* Critically low if less than 10% or max btree height remains. */
+	return XFS_TEST_ERROR(avail < orig / 10 ||
+			      avail < pag->pag_mount->m_agbtree_maxlevels,
+			pag->pag_mount, XFS_ERRTAG_AG_RESV_CRITICAL);
+}
+
+/*
+ * How many blocks are reserved but not used, and therefore must not be
+ * allocated away?
+ */
+xfs_extlen_t
+xfs_ag_resv_needed(
+	struct xfs_perag		*pag,
+	enum xfs_ag_resv_type		type)
+{
+	xfs_extlen_t			len;
+
+	len = pag->pag_meta_resv.ar_reserved + pag->pag_rmapbt_resv.ar_reserved;
+	switch (type) {
+	case XFS_AG_RESV_METADATA:
+	case XFS_AG_RESV_RMAPBT:
+		len -= xfs_perag_resv(pag, type)->ar_reserved;
+		break;
+	case XFS_AG_RESV_NONE:
+		/* empty */
+		break;
+	default:
+		ASSERT(0);
+	}
+
+	trace_xfs_ag_resv_needed(pag, type, len);
+
+	return len;
+}
+
+/* Clean out a reservation */
+static int
+__xfs_ag_resv_free(
+	struct xfs_perag		*pag,
+	enum xfs_ag_resv_type		type)
+{
+	struct xfs_ag_resv		*resv;
+	xfs_extlen_t			oldresv;
+	int				error;
+
+	trace_xfs_ag_resv_free(pag, type, 0);
+
+	resv = xfs_perag_resv(pag, type);
+	if (pag->pag_agno == 0)
+		pag->pag_mount->m_ag_max_usable += resv->ar_asked;
+	/*
+	 * RMAPBT blocks come from the AGFL and AGFL blocks are always
+	 * considered "free", so whatever was reserved at mount time must be
+	 * given back at umount.
+	 */
+	if (type == XFS_AG_RESV_RMAPBT)
+		oldresv = resv->ar_orig_reserved;
+	else
+		oldresv = resv->ar_reserved;
+	error = xfs_mod_fdblocks(pag->pag_mount, oldresv, true);
+	resv->ar_reserved = 0;
+	resv->ar_asked = 0;
+	resv->ar_orig_reserved = 0;
+
+	if (error)
+		trace_xfs_ag_resv_free_error(pag->pag_mount, pag->pag_agno,
+				error, _RET_IP_);
+	return error;
+}
+
+/* Free a per-AG reservation. */
+int
+xfs_ag_resv_free(
+	struct xfs_perag		*pag)
+{
+	int				error;
+	int				err2;
+
+	error = __xfs_ag_resv_free(pag, XFS_AG_RESV_RMAPBT);
+	err2 = __xfs_ag_resv_free(pag, XFS_AG_RESV_METADATA);
+	if (err2 && !error)
+		error = err2;
+	return error;
+}
+
+static int
+__xfs_ag_resv_init(
+	struct xfs_perag		*pag,
+	enum xfs_ag_resv_type		type,
+	xfs_extlen_t			ask,
+	xfs_extlen_t			used)
+{
+	struct xfs_mount		*mp = pag->pag_mount;
+	struct xfs_ag_resv		*resv;
+	int				error;
+	xfs_extlen_t			hidden_space;
+
+	if (used > ask)
+		ask = used;
+
+	switch (type) {
+	case XFS_AG_RESV_RMAPBT:
+		/*
+		 * Space taken by the rmapbt is not subtracted from fdblocks
+		 * because the rmapbt lives in the free space.  Here we must
+		 * subtract the entire reservation from fdblocks so that we
+		 * always have blocks available for rmapbt expansion.
+		 */
+		hidden_space = ask;
+		break;
+	case XFS_AG_RESV_METADATA:
+		/*
+		 * Space taken by all other metadata btrees are accounted
+		 * on-disk as used space.  We therefore only hide the space
+		 * that is reserved but not used by the trees.
+		 */
+		hidden_space = ask - used;
+		break;
+	default:
+		ASSERT(0);
+		return -EINVAL;
+	}
+
+	if (XFS_TEST_ERROR(false, mp, XFS_ERRTAG_AG_RESV_FAIL))
+		error = -ENOSPC;
+	else
+		error = xfs_mod_fdblocks(mp, -(int64_t)hidden_space, true);
+	if (error) {
+		trace_xfs_ag_resv_init_error(pag->pag_mount, pag->pag_agno,
+				error, _RET_IP_);
+		xfs_warn(mp,
+"Per-AG reservation for AG %u failed.  Filesystem may run out of space.",
+				pag->pag_agno);
+		return error;
+	}
+
+	/*
+	 * Reduce the maximum per-AG allocation length by however much we're
+	 * trying to reserve for an AG.  Since this is a filesystem-wide
+	 * counter, we only make the adjustment for AG 0.  This assumes that
+	 * there aren't any AGs hungrier for per-AG reservation than AG 0.
+	 */
+	if (pag->pag_agno == 0)
+		mp->m_ag_max_usable -= ask;
+
+	resv = xfs_perag_resv(pag, type);
+	resv->ar_asked = ask;
+	resv->ar_orig_reserved = hidden_space;
+	resv->ar_reserved = ask - used;
+
+	trace_xfs_ag_resv_init(pag, type, ask);
+	return 0;
+}
+
+/* Create a per-AG block reservation. */
+int
+xfs_ag_resv_init(
+	struct xfs_perag		*pag,
+	struct xfs_trans		*tp)
+{
+	struct xfs_mount		*mp = pag->pag_mount;
+	xfs_extlen_t			ask;
+	xfs_extlen_t			used;
+	int				error = 0, error2;
+	bool				has_resv = false;
+
+	/* Create the metadata reservation. */
+	if (pag->pag_meta_resv.ar_asked == 0) {
+		ask = used = 0;
+
+		error = xfs_refcountbt_calc_reserves(mp, tp, pag, &ask, &used);
+		if (error)
+			goto out;
+
+		error = xfs_finobt_calc_reserves(mp, tp, pag, &ask, &used);
+		if (error)
+			goto out;
+
+		error = __xfs_ag_resv_init(pag, XFS_AG_RESV_METADATA,
+				ask, used);
+		if (error) {
+			/*
+			 * Because we didn't have per-AG reservations when the
+			 * finobt feature was added we might not be able to
+			 * reserve all needed blocks.  Warn and fall back to the
+			 * old and potentially buggy code in that case, but
+			 * ensure we do have the reservation for the refcountbt.
+			 */
+			ask = used = 0;
+
+			mp->m_finobt_nores = true;
+
+			error = xfs_refcountbt_calc_reserves(mp, tp, pag, &ask,
+					&used);
+			if (error)
+				goto out;
+
+			error = __xfs_ag_resv_init(pag, XFS_AG_RESV_METADATA,
+					ask, used);
+			if (error)
+				goto out;
+		}
+		if (ask)
+			has_resv = true;
+	}
+
+	/* Create the RMAPBT metadata reservation */
+	if (pag->pag_rmapbt_resv.ar_asked == 0) {
+		ask = used = 0;
+
+		error = xfs_rmapbt_calc_reserves(mp, tp, pag, &ask, &used);
+		if (error)
+			goto out;
+
+		error = __xfs_ag_resv_init(pag, XFS_AG_RESV_RMAPBT, ask, used);
+		if (error)
+			goto out;
+		if (ask)
+			has_resv = true;
+	}
+
+out:
+	/*
+	 * Initialize the pagf if we have at least one active reservation on the
+	 * AG. This may have occurred already via reservation calculation, but
+	 * fall back to an explicit init to ensure the in-core allocbt usage
+	 * counters are initialized as soon as possible. This is important
+	 * because filesystems with large perag reservations are susceptible to
+	 * free space reservation problems that the allocbt counter is used to
+	 * address.
+	 */
+	if (has_resv) {
+		error2 = xfs_alloc_read_agf(pag, tp, 0, NULL);
+		if (error2)
+			return error2;
+
+		/*
+		 * If there isn't enough space in the AG to satisfy the
+		 * reservation, let the caller know that there wasn't enough
+		 * space.  Callers are responsible for deciding what to do
+		 * next, since (in theory) we can stumble along with
+		 * insufficient reservation if data blocks are being freed to
+		 * replenish the AG's free space.
+		 */
+		if (!error &&
+		    xfs_perag_resv(pag, XFS_AG_RESV_METADATA)->ar_reserved +
+		    xfs_perag_resv(pag, XFS_AG_RESV_RMAPBT)->ar_reserved >
+		    pag->pagf_freeblks + pag->pagf_flcount)
+			error = -ENOSPC;
+	}
+
+	return error;
+}
+
+/* Allocate a block from the reservation. */
+void
+xfs_ag_resv_alloc_extent(
+	struct xfs_perag		*pag,
+	enum xfs_ag_resv_type		type,
+	struct xfs_alloc_arg		*args)
+{
+	struct xfs_ag_resv		*resv;
+	xfs_extlen_t			len;
+	uint				field;
+
+	trace_xfs_ag_resv_alloc_extent(pag, type, args->len);
+
+	switch (type) {
+	case XFS_AG_RESV_AGFL:
+		return;
+	case XFS_AG_RESV_METADATA:
+	case XFS_AG_RESV_RMAPBT:
+		resv = xfs_perag_resv(pag, type);
+		break;
+	default:
+		ASSERT(0);
+		fallthrough;
+	case XFS_AG_RESV_NONE:
+		field = args->wasdel ? XFS_TRANS_SB_RES_FDBLOCKS :
+				       XFS_TRANS_SB_FDBLOCKS;
+		xfs_trans_mod_sb(args->tp, field, -(int64_t)args->len);
+		return;
+	}
+
+	len = min_t(xfs_extlen_t, args->len, resv->ar_reserved);
+	resv->ar_reserved -= len;
+	if (type == XFS_AG_RESV_RMAPBT)
+		return;
+	/* Allocations of reserved blocks only need on-disk sb updates... */
+	xfs_trans_mod_sb(args->tp, XFS_TRANS_SB_RES_FDBLOCKS, -(int64_t)len);
+	/* ...but non-reserved blocks need in-core and on-disk updates. */
+	if (args->len > len)
+		xfs_trans_mod_sb(args->tp, XFS_TRANS_SB_FDBLOCKS,
+				-((int64_t)args->len - len));
+}
+
+/* Free a block to the reservation. */
+void
+xfs_ag_resv_free_extent(
+	struct xfs_perag		*pag,
+	enum xfs_ag_resv_type		type,
+	struct xfs_trans		*tp,
+	xfs_extlen_t			len)
+{
+	xfs_extlen_t			leftover;
+	struct xfs_ag_resv		*resv;
+
+	trace_xfs_ag_resv_free_extent(pag, type, len);
+
+	switch (type) {
+	case XFS_AG_RESV_AGFL:
+		return;
+	case XFS_AG_RESV_METADATA:
+	case XFS_AG_RESV_RMAPBT:
+		resv = xfs_perag_resv(pag, type);
+		break;
+	default:
+		ASSERT(0);
+		fallthrough;
+	case XFS_AG_RESV_NONE:
+		xfs_trans_mod_sb(tp, XFS_TRANS_SB_FDBLOCKS, (int64_t)len);
+		return;
+	}
+
+	leftover = min_t(xfs_extlen_t, len, resv->ar_asked - resv->ar_reserved);
+	resv->ar_reserved += leftover;
+	if (type == XFS_AG_RESV_RMAPBT)
+		return;
+	/* Freeing into the reserved pool only requires on-disk update... */
+	xfs_trans_mod_sb(tp, XFS_TRANS_SB_RES_FDBLOCKS, len);
+	/* ...but freeing beyond that requires in-core and on-disk update. */
+	if (len > leftover)
+		xfs_trans_mod_sb(tp, XFS_TRANS_SB_FDBLOCKS, len - leftover);
+}
diff --git a/fs/xfs/libxfs/xfs_ag_resv.h b/fs/xfs/libxfs/xfs_ag_resv.h
new file mode 100644
index 000000000..b74b21000
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_ag_resv.h
@@ -0,0 +1,55 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Copyright (C) 2016 Oracle.  All Rights Reserved.
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ */
+#ifndef __XFS_AG_RESV_H__
+#define	__XFS_AG_RESV_H__
+
+int xfs_ag_resv_free(struct xfs_perag *pag);
+int xfs_ag_resv_init(struct xfs_perag *pag, struct xfs_trans *tp);
+
+bool xfs_ag_resv_critical(struct xfs_perag *pag, enum xfs_ag_resv_type type);
+xfs_extlen_t xfs_ag_resv_needed(struct xfs_perag *pag,
+		enum xfs_ag_resv_type type);
+
+void xfs_ag_resv_alloc_extent(struct xfs_perag *pag, enum xfs_ag_resv_type type,
+		struct xfs_alloc_arg *args);
+void xfs_ag_resv_free_extent(struct xfs_perag *pag, enum xfs_ag_resv_type type,
+		struct xfs_trans *tp, xfs_extlen_t len);
+
+static inline struct xfs_ag_resv *
+xfs_perag_resv(
+	struct xfs_perag	*pag,
+	enum xfs_ag_resv_type	type)
+{
+	switch (type) {
+	case XFS_AG_RESV_METADATA:
+		return &pag->pag_meta_resv;
+	case XFS_AG_RESV_RMAPBT:
+		return &pag->pag_rmapbt_resv;
+	default:
+		return NULL;
+	}
+}
+
+/*
+ * RMAPBT reservation accounting wrappers. Since rmapbt blocks are sourced from
+ * the AGFL, they are allocated one at a time and the reservation updates don't
+ * require a transaction.
+ */
+static inline void
+xfs_ag_resv_rmapbt_alloc(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		agno)
+{
+	struct xfs_alloc_arg	args = { NULL };
+	struct xfs_perag	*pag;
+
+	args.len = 1;
+	pag = xfs_perag_get(mp, agno);
+	xfs_ag_resv_alloc_extent(pag, XFS_AG_RESV_RMAPBT, &args);
+	xfs_perag_put(pag);
+}
+
+#endif	/* __XFS_AG_RESV_H__ */
diff --git a/fs/xfs/libxfs/xfs_alloc.c b/fs/xfs/libxfs/xfs_alloc.c
new file mode 100644
index 000000000..de79f5d07
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_alloc.c
@@ -0,0 +1,3568 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_shared.h"
+#include "xfs_trans_resv.h"
+#include "xfs_bit.h"
+#include "xfs_mount.h"
+#include "xfs_defer.h"
+#include "xfs_btree.h"
+#include "xfs_rmap.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_alloc.h"
+#include "xfs_extent_busy.h"
+#include "xfs_errortag.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+#include "xfs_trans.h"
+#include "xfs_buf_item.h"
+#include "xfs_log.h"
+#include "xfs_ag.h"
+#include "xfs_ag_resv.h"
+#include "xfs_bmap.h"
+
+struct kmem_cache	*xfs_extfree_item_cache;
+
+struct workqueue_struct *xfs_alloc_wq;
+
+#define XFS_ABSDIFF(a,b)	(((a) <= (b)) ? ((b) - (a)) : ((a) - (b)))
+
+#define	XFSA_FIXUP_BNO_OK	1
+#define	XFSA_FIXUP_CNT_OK	2
+
+STATIC int xfs_alloc_ag_vextent_exact(xfs_alloc_arg_t *);
+STATIC int xfs_alloc_ag_vextent_near(xfs_alloc_arg_t *);
+STATIC int xfs_alloc_ag_vextent_size(xfs_alloc_arg_t *);
+
+/*
+ * Size of the AGFL.  For CRC-enabled filesystes we steal a couple of slots in
+ * the beginning of the block for a proper header with the location information
+ * and CRC.
+ */
+unsigned int
+xfs_agfl_size(
+	struct xfs_mount	*mp)
+{
+	unsigned int		size = mp->m_sb.sb_sectsize;
+
+	if (xfs_has_crc(mp))
+		size -= sizeof(struct xfs_agfl);
+
+	return size / sizeof(xfs_agblock_t);
+}
+
+unsigned int
+xfs_refc_block(
+	struct xfs_mount	*mp)
+{
+	if (xfs_has_rmapbt(mp))
+		return XFS_RMAP_BLOCK(mp) + 1;
+	if (xfs_has_finobt(mp))
+		return XFS_FIBT_BLOCK(mp) + 1;
+	return XFS_IBT_BLOCK(mp) + 1;
+}
+
+xfs_extlen_t
+xfs_prealloc_blocks(
+	struct xfs_mount	*mp)
+{
+	if (xfs_has_reflink(mp))
+		return xfs_refc_block(mp) + 1;
+	if (xfs_has_rmapbt(mp))
+		return XFS_RMAP_BLOCK(mp) + 1;
+	if (xfs_has_finobt(mp))
+		return XFS_FIBT_BLOCK(mp) + 1;
+	return XFS_IBT_BLOCK(mp) + 1;
+}
+
+/*
+ * The number of blocks per AG that we withhold from xfs_mod_fdblocks to
+ * guarantee that we can refill the AGFL prior to allocating space in a nearly
+ * full AG.  Although the space described by the free space btrees, the
+ * blocks used by the freesp btrees themselves, and the blocks owned by the
+ * AGFL are counted in the ondisk fdblocks, it's a mistake to let the ondisk
+ * free space in the AG drop so low that the free space btrees cannot refill an
+ * empty AGFL up to the minimum level.  Rather than grind through empty AGs
+ * until the fs goes down, we subtract this many AG blocks from the incore
+ * fdblocks to ensure user allocation does not overcommit the space the
+ * filesystem needs for the AGFLs.  The rmap btree uses a per-AG reservation to
+ * withhold space from xfs_mod_fdblocks, so we do not account for that here.
+ */
+#define XFS_ALLOCBT_AGFL_RESERVE	4
+
+/*
+ * Compute the number of blocks that we set aside to guarantee the ability to
+ * refill the AGFL and handle a full bmap btree split.
+ *
+ * In order to avoid ENOSPC-related deadlock caused by out-of-order locking of
+ * AGF buffer (PV 947395), we place constraints on the relationship among
+ * actual allocations for data blocks, freelist blocks, and potential file data
+ * bmap btree blocks. However, these restrictions may result in no actual space
+ * allocated for a delayed extent, for example, a data block in a certain AG is
+ * allocated but there is no additional block for the additional bmap btree
+ * block due to a split of the bmap btree of the file. The result of this may
+ * lead to an infinite loop when the file gets flushed to disk and all delayed
+ * extents need to be actually allocated. To get around this, we explicitly set
+ * aside a few blocks which will not be reserved in delayed allocation.
+ *
+ * For each AG, we need to reserve enough blocks to replenish a totally empty
+ * AGFL and 4 more to handle a potential split of the file's bmap btree.
+ */
+unsigned int
+xfs_alloc_set_aside(
+	struct xfs_mount	*mp)
+{
+	return mp->m_sb.sb_agcount * (XFS_ALLOCBT_AGFL_RESERVE + 4);
+}
+
+/*
+ * When deciding how much space to allocate out of an AG, we limit the
+ * allocation maximum size to the size the AG. However, we cannot use all the
+ * blocks in the AG - some are permanently used by metadata. These
+ * blocks are generally:
+ *	- the AG superblock, AGF, AGI and AGFL
+ *	- the AGF (bno and cnt) and AGI btree root blocks, and optionally
+ *	  the AGI free inode and rmap btree root blocks.
+ *	- blocks on the AGFL according to xfs_alloc_set_aside() limits
+ *	- the rmapbt root block
+ *
+ * The AG headers are sector sized, so the amount of space they take up is
+ * dependent on filesystem geometry. The others are all single blocks.
+ */
+unsigned int
+xfs_alloc_ag_max_usable(
+	struct xfs_mount	*mp)
+{
+	unsigned int		blocks;
+
+	blocks = XFS_BB_TO_FSB(mp, XFS_FSS_TO_BB(mp, 4)); /* ag headers */
+	blocks += XFS_ALLOCBT_AGFL_RESERVE;
+	blocks += 3;			/* AGF, AGI btree root blocks */
+	if (xfs_has_finobt(mp))
+		blocks++;		/* finobt root block */
+	if (xfs_has_rmapbt(mp))
+		blocks++;		/* rmap root block */
+	if (xfs_has_reflink(mp))
+		blocks++;		/* refcount root block */
+
+	return mp->m_sb.sb_agblocks - blocks;
+}
+
+/*
+ * Lookup the record equal to [bno, len] in the btree given by cur.
+ */
+STATIC int				/* error */
+xfs_alloc_lookup_eq(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* starting block of extent */
+	xfs_extlen_t		len,	/* length of extent */
+	int			*stat)	/* success/failure */
+{
+	int			error;
+
+	cur->bc_rec.a.ar_startblock = bno;
+	cur->bc_rec.a.ar_blockcount = len;
+	error = xfs_btree_lookup(cur, XFS_LOOKUP_EQ, stat);
+	cur->bc_ag.abt.active = (*stat == 1);
+	return error;
+}
+
+/*
+ * Lookup the first record greater than or equal to [bno, len]
+ * in the btree given by cur.
+ */
+int				/* error */
+xfs_alloc_lookup_ge(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* starting block of extent */
+	xfs_extlen_t		len,	/* length of extent */
+	int			*stat)	/* success/failure */
+{
+	int			error;
+
+	cur->bc_rec.a.ar_startblock = bno;
+	cur->bc_rec.a.ar_blockcount = len;
+	error = xfs_btree_lookup(cur, XFS_LOOKUP_GE, stat);
+	cur->bc_ag.abt.active = (*stat == 1);
+	return error;
+}
+
+/*
+ * Lookup the first record less than or equal to [bno, len]
+ * in the btree given by cur.
+ */
+int					/* error */
+xfs_alloc_lookup_le(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* starting block of extent */
+	xfs_extlen_t		len,	/* length of extent */
+	int			*stat)	/* success/failure */
+{
+	int			error;
+	cur->bc_rec.a.ar_startblock = bno;
+	cur->bc_rec.a.ar_blockcount = len;
+	error = xfs_btree_lookup(cur, XFS_LOOKUP_LE, stat);
+	cur->bc_ag.abt.active = (*stat == 1);
+	return error;
+}
+
+static inline bool
+xfs_alloc_cur_active(
+	struct xfs_btree_cur	*cur)
+{
+	return cur && cur->bc_ag.abt.active;
+}
+
+/*
+ * Update the record referred to by cur to the value given
+ * by [bno, len].
+ * This either works (return 0) or gets an EFSCORRUPTED error.
+ */
+STATIC int				/* error */
+xfs_alloc_update(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* starting block of extent */
+	xfs_extlen_t		len)	/* length of extent */
+{
+	union xfs_btree_rec	rec;
+
+	rec.alloc.ar_startblock = cpu_to_be32(bno);
+	rec.alloc.ar_blockcount = cpu_to_be32(len);
+	return xfs_btree_update(cur, &rec);
+}
+
+/*
+ * Get the data from the pointed-to record.
+ */
+int					/* error */
+xfs_alloc_get_rec(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		*bno,	/* output: starting block of extent */
+	xfs_extlen_t		*len,	/* output: length of extent */
+	int			*stat)	/* output: success/failure */
+{
+	struct xfs_mount	*mp = cur->bc_mp;
+	struct xfs_perag	*pag = cur->bc_ag.pag;
+	union xfs_btree_rec	*rec;
+	int			error;
+
+	error = xfs_btree_get_rec(cur, &rec, stat);
+	if (error || !(*stat))
+		return error;
+
+	*bno = be32_to_cpu(rec->alloc.ar_startblock);
+	*len = be32_to_cpu(rec->alloc.ar_blockcount);
+
+	if (*len == 0)
+		goto out_bad_rec;
+
+	/* check for valid extent range, including overflow */
+	if (!xfs_verify_agbext(pag, *bno, *len))
+		goto out_bad_rec;
+
+	return 0;
+
+out_bad_rec:
+	xfs_warn(mp,
+		"%s Freespace BTree record corruption in AG %d detected!",
+		cur->bc_btnum == XFS_BTNUM_BNO ? "Block" : "Size",
+		pag->pag_agno);
+	xfs_warn(mp,
+		"start block 0x%x block count 0x%x", *bno, *len);
+	return -EFSCORRUPTED;
+}
+
+/*
+ * Compute aligned version of the found extent.
+ * Takes alignment and min length into account.
+ */
+STATIC bool
+xfs_alloc_compute_aligned(
+	xfs_alloc_arg_t	*args,		/* allocation argument structure */
+	xfs_agblock_t	foundbno,	/* starting block in found extent */
+	xfs_extlen_t	foundlen,	/* length in found extent */
+	xfs_agblock_t	*resbno,	/* result block number */
+	xfs_extlen_t	*reslen,	/* result length */
+	unsigned	*busy_gen)
+{
+	xfs_agblock_t	bno = foundbno;
+	xfs_extlen_t	len = foundlen;
+	xfs_extlen_t	diff;
+	bool		busy;
+
+	/* Trim busy sections out of found extent */
+	busy = xfs_extent_busy_trim(args, &bno, &len, busy_gen);
+
+	/*
+	 * If we have a largish extent that happens to start before min_agbno,
+	 * see if we can shift it into range...
+	 */
+	if (bno < args->min_agbno && bno + len > args->min_agbno) {
+		diff = args->min_agbno - bno;
+		if (len > diff) {
+			bno += diff;
+			len -= diff;
+		}
+	}
+
+	if (args->alignment > 1 && len >= args->minlen) {
+		xfs_agblock_t	aligned_bno = roundup(bno, args->alignment);
+
+		diff = aligned_bno - bno;
+
+		*resbno = aligned_bno;
+		*reslen = diff >= len ? 0 : len - diff;
+	} else {
+		*resbno = bno;
+		*reslen = len;
+	}
+
+	return busy;
+}
+
+/*
+ * Compute best start block and diff for "near" allocations.
+ * freelen >= wantlen already checked by caller.
+ */
+STATIC xfs_extlen_t			/* difference value (absolute) */
+xfs_alloc_compute_diff(
+	xfs_agblock_t	wantbno,	/* target starting block */
+	xfs_extlen_t	wantlen,	/* target length */
+	xfs_extlen_t	alignment,	/* target alignment */
+	int		datatype,	/* are we allocating data? */
+	xfs_agblock_t	freebno,	/* freespace's starting block */
+	xfs_extlen_t	freelen,	/* freespace's length */
+	xfs_agblock_t	*newbnop)	/* result: best start block from free */
+{
+	xfs_agblock_t	freeend;	/* end of freespace extent */
+	xfs_agblock_t	newbno1;	/* return block number */
+	xfs_agblock_t	newbno2;	/* other new block number */
+	xfs_extlen_t	newlen1=0;	/* length with newbno1 */
+	xfs_extlen_t	newlen2=0;	/* length with newbno2 */
+	xfs_agblock_t	wantend;	/* end of target extent */
+	bool		userdata = datatype & XFS_ALLOC_USERDATA;
+
+	ASSERT(freelen >= wantlen);
+	freeend = freebno + freelen;
+	wantend = wantbno + wantlen;
+	/*
+	 * We want to allocate from the start of a free extent if it is past
+	 * the desired block or if we are allocating user data and the free
+	 * extent is before desired block. The second case is there to allow
+	 * for contiguous allocation from the remaining free space if the file
+	 * grows in the short term.
+	 */
+	if (freebno >= wantbno || (userdata && freeend < wantend)) {
+		if ((newbno1 = roundup(freebno, alignment)) >= freeend)
+			newbno1 = NULLAGBLOCK;
+	} else if (freeend >= wantend && alignment > 1) {
+		newbno1 = roundup(wantbno, alignment);
+		newbno2 = newbno1 - alignment;
+		if (newbno1 >= freeend)
+			newbno1 = NULLAGBLOCK;
+		else
+			newlen1 = XFS_EXTLEN_MIN(wantlen, freeend - newbno1);
+		if (newbno2 < freebno)
+			newbno2 = NULLAGBLOCK;
+		else
+			newlen2 = XFS_EXTLEN_MIN(wantlen, freeend - newbno2);
+		if (newbno1 != NULLAGBLOCK && newbno2 != NULLAGBLOCK) {
+			if (newlen1 < newlen2 ||
+			    (newlen1 == newlen2 &&
+			     XFS_ABSDIFF(newbno1, wantbno) >
+			     XFS_ABSDIFF(newbno2, wantbno)))
+				newbno1 = newbno2;
+		} else if (newbno2 != NULLAGBLOCK)
+			newbno1 = newbno2;
+	} else if (freeend >= wantend) {
+		newbno1 = wantbno;
+	} else if (alignment > 1) {
+		newbno1 = roundup(freeend - wantlen, alignment);
+		if (newbno1 > freeend - wantlen &&
+		    newbno1 - alignment >= freebno)
+			newbno1 -= alignment;
+		else if (newbno1 >= freeend)
+			newbno1 = NULLAGBLOCK;
+	} else
+		newbno1 = freeend - wantlen;
+	*newbnop = newbno1;
+	return newbno1 == NULLAGBLOCK ? 0 : XFS_ABSDIFF(newbno1, wantbno);
+}
+
+/*
+ * Fix up the length, based on mod and prod.
+ * len should be k * prod + mod for some k.
+ * If len is too small it is returned unchanged.
+ * If len hits maxlen it is left alone.
+ */
+STATIC void
+xfs_alloc_fix_len(
+	xfs_alloc_arg_t	*args)		/* allocation argument structure */
+{
+	xfs_extlen_t	k;
+	xfs_extlen_t	rlen;
+
+	ASSERT(args->mod < args->prod);
+	rlen = args->len;
+	ASSERT(rlen >= args->minlen);
+	ASSERT(rlen <= args->maxlen);
+	if (args->prod <= 1 || rlen < args->mod || rlen == args->maxlen ||
+	    (args->mod == 0 && rlen < args->prod))
+		return;
+	k = rlen % args->prod;
+	if (k == args->mod)
+		return;
+	if (k > args->mod)
+		rlen = rlen - (k - args->mod);
+	else
+		rlen = rlen - args->prod + (args->mod - k);
+	/* casts to (int) catch length underflows */
+	if ((int)rlen < (int)args->minlen)
+		return;
+	ASSERT(rlen >= args->minlen && rlen <= args->maxlen);
+	ASSERT(rlen % args->prod == args->mod);
+	ASSERT(args->pag->pagf_freeblks + args->pag->pagf_flcount >=
+		rlen + args->minleft);
+	args->len = rlen;
+}
+
+/*
+ * Update the two btrees, logically removing from freespace the extent
+ * starting at rbno, rlen blocks.  The extent is contained within the
+ * actual (current) free extent fbno for flen blocks.
+ * Flags are passed in indicating whether the cursors are set to the
+ * relevant records.
+ */
+STATIC int				/* error code */
+xfs_alloc_fixup_trees(
+	struct xfs_btree_cur *cnt_cur,	/* cursor for by-size btree */
+	struct xfs_btree_cur *bno_cur,	/* cursor for by-block btree */
+	xfs_agblock_t	fbno,		/* starting block of free extent */
+	xfs_extlen_t	flen,		/* length of free extent */
+	xfs_agblock_t	rbno,		/* starting block of returned extent */
+	xfs_extlen_t	rlen,		/* length of returned extent */
+	int		flags)		/* flags, XFSA_FIXUP_... */
+{
+	int		error;		/* error code */
+	int		i;		/* operation results */
+	xfs_agblock_t	nfbno1;		/* first new free startblock */
+	xfs_agblock_t	nfbno2;		/* second new free startblock */
+	xfs_extlen_t	nflen1=0;	/* first new free length */
+	xfs_extlen_t	nflen2=0;	/* second new free length */
+	struct xfs_mount *mp;
+
+	mp = cnt_cur->bc_mp;
+
+	/*
+	 * Look up the record in the by-size tree if necessary.
+	 */
+	if (flags & XFSA_FIXUP_CNT_OK) {
+#ifdef DEBUG
+		if ((error = xfs_alloc_get_rec(cnt_cur, &nfbno1, &nflen1, &i)))
+			return error;
+		if (XFS_IS_CORRUPT(mp,
+				   i != 1 ||
+				   nfbno1 != fbno ||
+				   nflen1 != flen))
+			return -EFSCORRUPTED;
+#endif
+	} else {
+		if ((error = xfs_alloc_lookup_eq(cnt_cur, fbno, flen, &i)))
+			return error;
+		if (XFS_IS_CORRUPT(mp, i != 1))
+			return -EFSCORRUPTED;
+	}
+	/*
+	 * Look up the record in the by-block tree if necessary.
+	 */
+	if (flags & XFSA_FIXUP_BNO_OK) {
+#ifdef DEBUG
+		if ((error = xfs_alloc_get_rec(bno_cur, &nfbno1, &nflen1, &i)))
+			return error;
+		if (XFS_IS_CORRUPT(mp,
+				   i != 1 ||
+				   nfbno1 != fbno ||
+				   nflen1 != flen))
+			return -EFSCORRUPTED;
+#endif
+	} else {
+		if ((error = xfs_alloc_lookup_eq(bno_cur, fbno, flen, &i)))
+			return error;
+		if (XFS_IS_CORRUPT(mp, i != 1))
+			return -EFSCORRUPTED;
+	}
+
+#ifdef DEBUG
+	if (bno_cur->bc_nlevels == 1 && cnt_cur->bc_nlevels == 1) {
+		struct xfs_btree_block	*bnoblock;
+		struct xfs_btree_block	*cntblock;
+
+		bnoblock = XFS_BUF_TO_BLOCK(bno_cur->bc_levels[0].bp);
+		cntblock = XFS_BUF_TO_BLOCK(cnt_cur->bc_levels[0].bp);
+
+		if (XFS_IS_CORRUPT(mp,
+				   bnoblock->bb_numrecs !=
+				   cntblock->bb_numrecs))
+			return -EFSCORRUPTED;
+	}
+#endif
+
+	/*
+	 * Deal with all four cases: the allocated record is contained
+	 * within the freespace record, so we can have new freespace
+	 * at either (or both) end, or no freespace remaining.
+	 */
+	if (rbno == fbno && rlen == flen)
+		nfbno1 = nfbno2 = NULLAGBLOCK;
+	else if (rbno == fbno) {
+		nfbno1 = rbno + rlen;
+		nflen1 = flen - rlen;
+		nfbno2 = NULLAGBLOCK;
+	} else if (rbno + rlen == fbno + flen) {
+		nfbno1 = fbno;
+		nflen1 = flen - rlen;
+		nfbno2 = NULLAGBLOCK;
+	} else {
+		nfbno1 = fbno;
+		nflen1 = rbno - fbno;
+		nfbno2 = rbno + rlen;
+		nflen2 = (fbno + flen) - nfbno2;
+	}
+	/*
+	 * Delete the entry from the by-size btree.
+	 */
+	if ((error = xfs_btree_delete(cnt_cur, &i)))
+		return error;
+	if (XFS_IS_CORRUPT(mp, i != 1))
+		return -EFSCORRUPTED;
+	/*
+	 * Add new by-size btree entry(s).
+	 */
+	if (nfbno1 != NULLAGBLOCK) {
+		if ((error = xfs_alloc_lookup_eq(cnt_cur, nfbno1, nflen1, &i)))
+			return error;
+		if (XFS_IS_CORRUPT(mp, i != 0))
+			return -EFSCORRUPTED;
+		if ((error = xfs_btree_insert(cnt_cur, &i)))
+			return error;
+		if (XFS_IS_CORRUPT(mp, i != 1))
+			return -EFSCORRUPTED;
+	}
+	if (nfbno2 != NULLAGBLOCK) {
+		if ((error = xfs_alloc_lookup_eq(cnt_cur, nfbno2, nflen2, &i)))
+			return error;
+		if (XFS_IS_CORRUPT(mp, i != 0))
+			return -EFSCORRUPTED;
+		if ((error = xfs_btree_insert(cnt_cur, &i)))
+			return error;
+		if (XFS_IS_CORRUPT(mp, i != 1))
+			return -EFSCORRUPTED;
+	}
+	/*
+	 * Fix up the by-block btree entry(s).
+	 */
+	if (nfbno1 == NULLAGBLOCK) {
+		/*
+		 * No remaining freespace, just delete the by-block tree entry.
+		 */
+		if ((error = xfs_btree_delete(bno_cur, &i)))
+			return error;
+		if (XFS_IS_CORRUPT(mp, i != 1))
+			return -EFSCORRUPTED;
+	} else {
+		/*
+		 * Update the by-block entry to start later|be shorter.
+		 */
+		if ((error = xfs_alloc_update(bno_cur, nfbno1, nflen1)))
+			return error;
+	}
+	if (nfbno2 != NULLAGBLOCK) {
+		/*
+		 * 2 resulting free entries, need to add one.
+		 */
+		if ((error = xfs_alloc_lookup_eq(bno_cur, nfbno2, nflen2, &i)))
+			return error;
+		if (XFS_IS_CORRUPT(mp, i != 0))
+			return -EFSCORRUPTED;
+		if ((error = xfs_btree_insert(bno_cur, &i)))
+			return error;
+		if (XFS_IS_CORRUPT(mp, i != 1))
+			return -EFSCORRUPTED;
+	}
+	return 0;
+}
+
+static xfs_failaddr_t
+xfs_agfl_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount *mp = bp->b_mount;
+	struct xfs_agfl	*agfl = XFS_BUF_TO_AGFL(bp);
+	__be32		*agfl_bno = xfs_buf_to_agfl_bno(bp);
+	int		i;
+
+	/*
+	 * There is no verification of non-crc AGFLs because mkfs does not
+	 * initialise the AGFL to zero or NULL. Hence the only valid part of the
+	 * AGFL is what the AGF says is active. We can't get to the AGF, so we
+	 * can't verify just those entries are valid.
+	 */
+	if (!xfs_has_crc(mp))
+		return NULL;
+
+	if (!xfs_verify_magic(bp, agfl->agfl_magicnum))
+		return __this_address;
+	if (!uuid_equal(&agfl->agfl_uuid, &mp->m_sb.sb_meta_uuid))
+		return __this_address;
+	/*
+	 * during growfs operations, the perag is not fully initialised,
+	 * so we can't use it for any useful checking. growfs ensures we can't
+	 * use it by using uncached buffers that don't have the perag attached
+	 * so we can detect and avoid this problem.
+	 */
+	if (bp->b_pag && be32_to_cpu(agfl->agfl_seqno) != bp->b_pag->pag_agno)
+		return __this_address;
+
+	for (i = 0; i < xfs_agfl_size(mp); i++) {
+		if (be32_to_cpu(agfl_bno[i]) != NULLAGBLOCK &&
+		    be32_to_cpu(agfl_bno[i]) >= mp->m_sb.sb_agblocks)
+			return __this_address;
+	}
+
+	if (!xfs_log_check_lsn(mp, be64_to_cpu(XFS_BUF_TO_AGFL(bp)->agfl_lsn)))
+		return __this_address;
+	return NULL;
+}
+
+static void
+xfs_agfl_read_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount *mp = bp->b_mount;
+	xfs_failaddr_t	fa;
+
+	/*
+	 * There is no verification of non-crc AGFLs because mkfs does not
+	 * initialise the AGFL to zero or NULL. Hence the only valid part of the
+	 * AGFL is what the AGF says is active. We can't get to the AGF, so we
+	 * can't verify just those entries are valid.
+	 */
+	if (!xfs_has_crc(mp))
+		return;
+
+	if (!xfs_buf_verify_cksum(bp, XFS_AGFL_CRC_OFF))
+		xfs_verifier_error(bp, -EFSBADCRC, __this_address);
+	else {
+		fa = xfs_agfl_verify(bp);
+		if (fa)
+			xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+	}
+}
+
+static void
+xfs_agfl_write_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_buf_log_item	*bip = bp->b_log_item;
+	xfs_failaddr_t		fa;
+
+	/* no verification of non-crc AGFLs */
+	if (!xfs_has_crc(mp))
+		return;
+
+	fa = xfs_agfl_verify(bp);
+	if (fa) {
+		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+		return;
+	}
+
+	if (bip)
+		XFS_BUF_TO_AGFL(bp)->agfl_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+	xfs_buf_update_cksum(bp, XFS_AGFL_CRC_OFF);
+}
+
+const struct xfs_buf_ops xfs_agfl_buf_ops = {
+	.name = "xfs_agfl",
+	.magic = { cpu_to_be32(XFS_AGFL_MAGIC), cpu_to_be32(XFS_AGFL_MAGIC) },
+	.verify_read = xfs_agfl_read_verify,
+	.verify_write = xfs_agfl_write_verify,
+	.verify_struct = xfs_agfl_verify,
+};
+
+/*
+ * Read in the allocation group free block array.
+ */
+int
+xfs_alloc_read_agfl(
+	struct xfs_perag	*pag,
+	struct xfs_trans	*tp,
+	struct xfs_buf		**bpp)
+{
+	struct xfs_mount	*mp = pag->pag_mount;
+	struct xfs_buf		*bp;
+	int			error;
+
+	error = xfs_trans_read_buf(
+			mp, tp, mp->m_ddev_targp,
+			XFS_AG_DADDR(mp, pag->pag_agno, XFS_AGFL_DADDR(mp)),
+			XFS_FSS_TO_BB(mp, 1), 0, &bp, &xfs_agfl_buf_ops);
+	if (error)
+		return error;
+	xfs_buf_set_ref(bp, XFS_AGFL_REF);
+	*bpp = bp;
+	return 0;
+}
+
+STATIC int
+xfs_alloc_update_counters(
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp,
+	long			len)
+{
+	struct xfs_agf		*agf = agbp->b_addr;
+
+	agbp->b_pag->pagf_freeblks += len;
+	be32_add_cpu(&agf->agf_freeblks, len);
+
+	if (unlikely(be32_to_cpu(agf->agf_freeblks) >
+		     be32_to_cpu(agf->agf_length))) {
+		xfs_buf_mark_corrupt(agbp);
+		return -EFSCORRUPTED;
+	}
+
+	xfs_alloc_log_agf(tp, agbp, XFS_AGF_FREEBLKS);
+	return 0;
+}
+
+/*
+ * Block allocation algorithm and data structures.
+ */
+struct xfs_alloc_cur {
+	struct xfs_btree_cur		*cnt;	/* btree cursors */
+	struct xfs_btree_cur		*bnolt;
+	struct xfs_btree_cur		*bnogt;
+	xfs_extlen_t			cur_len;/* current search length */
+	xfs_agblock_t			rec_bno;/* extent startblock */
+	xfs_extlen_t			rec_len;/* extent length */
+	xfs_agblock_t			bno;	/* alloc bno */
+	xfs_extlen_t			len;	/* alloc len */
+	xfs_extlen_t			diff;	/* diff from search bno */
+	unsigned int			busy_gen;/* busy state */
+	bool				busy;
+};
+
+/*
+ * Set up cursors, etc. in the extent allocation cursor. This function can be
+ * called multiple times to reset an initialized structure without having to
+ * reallocate cursors.
+ */
+static int
+xfs_alloc_cur_setup(
+	struct xfs_alloc_arg	*args,
+	struct xfs_alloc_cur	*acur)
+{
+	int			error;
+	int			i;
+
+	ASSERT(args->alignment == 1 || args->type != XFS_ALLOCTYPE_THIS_BNO);
+
+	acur->cur_len = args->maxlen;
+	acur->rec_bno = 0;
+	acur->rec_len = 0;
+	acur->bno = 0;
+	acur->len = 0;
+	acur->diff = -1;
+	acur->busy = false;
+	acur->busy_gen = 0;
+
+	/*
+	 * Perform an initial cntbt lookup to check for availability of maxlen
+	 * extents. If this fails, we'll return -ENOSPC to signal the caller to
+	 * attempt a small allocation.
+	 */
+	if (!acur->cnt)
+		acur->cnt = xfs_allocbt_init_cursor(args->mp, args->tp,
+					args->agbp, args->pag, XFS_BTNUM_CNT);
+	error = xfs_alloc_lookup_ge(acur->cnt, 0, args->maxlen, &i);
+	if (error)
+		return error;
+
+	/*
+	 * Allocate the bnobt left and right search cursors.
+	 */
+	if (!acur->bnolt)
+		acur->bnolt = xfs_allocbt_init_cursor(args->mp, args->tp,
+					args->agbp, args->pag, XFS_BTNUM_BNO);
+	if (!acur->bnogt)
+		acur->bnogt = xfs_allocbt_init_cursor(args->mp, args->tp,
+					args->agbp, args->pag, XFS_BTNUM_BNO);
+	return i == 1 ? 0 : -ENOSPC;
+}
+
+static void
+xfs_alloc_cur_close(
+	struct xfs_alloc_cur	*acur,
+	bool			error)
+{
+	int			cur_error = XFS_BTREE_NOERROR;
+
+	if (error)
+		cur_error = XFS_BTREE_ERROR;
+
+	if (acur->cnt)
+		xfs_btree_del_cursor(acur->cnt, cur_error);
+	if (acur->bnolt)
+		xfs_btree_del_cursor(acur->bnolt, cur_error);
+	if (acur->bnogt)
+		xfs_btree_del_cursor(acur->bnogt, cur_error);
+	acur->cnt = acur->bnolt = acur->bnogt = NULL;
+}
+
+/*
+ * Check an extent for allocation and track the best available candidate in the
+ * allocation structure. The cursor is deactivated if it has entered an out of
+ * range state based on allocation arguments. Optionally return the extent
+ * extent geometry and allocation status if requested by the caller.
+ */
+static int
+xfs_alloc_cur_check(
+	struct xfs_alloc_arg	*args,
+	struct xfs_alloc_cur	*acur,
+	struct xfs_btree_cur	*cur,
+	int			*new)
+{
+	int			error, i;
+	xfs_agblock_t		bno, bnoa, bnew;
+	xfs_extlen_t		len, lena, diff = -1;
+	bool			busy;
+	unsigned		busy_gen = 0;
+	bool			deactivate = false;
+	bool			isbnobt = cur->bc_btnum == XFS_BTNUM_BNO;
+
+	*new = 0;
+
+	error = xfs_alloc_get_rec(cur, &bno, &len, &i);
+	if (error)
+		return error;
+	if (XFS_IS_CORRUPT(args->mp, i != 1))
+		return -EFSCORRUPTED;
+
+	/*
+	 * Check minlen and deactivate a cntbt cursor if out of acceptable size
+	 * range (i.e., walking backwards looking for a minlen extent).
+	 */
+	if (len < args->minlen) {
+		deactivate = !isbnobt;
+		goto out;
+	}
+
+	busy = xfs_alloc_compute_aligned(args, bno, len, &bnoa, &lena,
+					 &busy_gen);
+	acur->busy |= busy;
+	if (busy)
+		acur->busy_gen = busy_gen;
+	/* deactivate a bnobt cursor outside of locality range */
+	if (bnoa < args->min_agbno || bnoa > args->max_agbno) {
+		deactivate = isbnobt;
+		goto out;
+	}
+	if (lena < args->minlen)
+		goto out;
+
+	args->len = XFS_EXTLEN_MIN(lena, args->maxlen);
+	xfs_alloc_fix_len(args);
+	ASSERT(args->len >= args->minlen);
+	if (args->len < acur->len)
+		goto out;
+
+	/*
+	 * We have an aligned record that satisfies minlen and beats or matches
+	 * the candidate extent size. Compare locality for near allocation mode.
+	 */
+	ASSERT(args->type == XFS_ALLOCTYPE_NEAR_BNO);
+	diff = xfs_alloc_compute_diff(args->agbno, args->len,
+				      args->alignment, args->datatype,
+				      bnoa, lena, &bnew);
+	if (bnew == NULLAGBLOCK)
+		goto out;
+
+	/*
+	 * Deactivate a bnobt cursor with worse locality than the current best.
+	 */
+	if (diff > acur->diff) {
+		deactivate = isbnobt;
+		goto out;
+	}
+
+	ASSERT(args->len > acur->len ||
+	       (args->len == acur->len && diff <= acur->diff));
+	acur->rec_bno = bno;
+	acur->rec_len = len;
+	acur->bno = bnew;
+	acur->len = args->len;
+	acur->diff = diff;
+	*new = 1;
+
+	/*
+	 * We're done if we found a perfect allocation. This only deactivates
+	 * the current cursor, but this is just an optimization to terminate a
+	 * cntbt search that otherwise runs to the edge of the tree.
+	 */
+	if (acur->diff == 0 && acur->len == args->maxlen)
+		deactivate = true;
+out:
+	if (deactivate)
+		cur->bc_ag.abt.active = false;
+	trace_xfs_alloc_cur_check(args->mp, cur->bc_btnum, bno, len, diff,
+				  *new);
+	return 0;
+}
+
+/*
+ * Complete an allocation of a candidate extent. Remove the extent from both
+ * trees and update the args structure.
+ */
+STATIC int
+xfs_alloc_cur_finish(
+	struct xfs_alloc_arg	*args,
+	struct xfs_alloc_cur	*acur)
+{
+	struct xfs_agf __maybe_unused *agf = args->agbp->b_addr;
+	int			error;
+
+	ASSERT(acur->cnt && acur->bnolt);
+	ASSERT(acur->bno >= acur->rec_bno);
+	ASSERT(acur->bno + acur->len <= acur->rec_bno + acur->rec_len);
+	ASSERT(acur->rec_bno + acur->rec_len <= be32_to_cpu(agf->agf_length));
+
+	error = xfs_alloc_fixup_trees(acur->cnt, acur->bnolt, acur->rec_bno,
+				      acur->rec_len, acur->bno, acur->len, 0);
+	if (error)
+		return error;
+
+	args->agbno = acur->bno;
+	args->len = acur->len;
+	args->wasfromfl = 0;
+
+	trace_xfs_alloc_cur(args);
+	return 0;
+}
+
+/*
+ * Locality allocation lookup algorithm. This expects a cntbt cursor and uses
+ * bno optimized lookup to search for extents with ideal size and locality.
+ */
+STATIC int
+xfs_alloc_cntbt_iter(
+	struct xfs_alloc_arg		*args,
+	struct xfs_alloc_cur		*acur)
+{
+	struct xfs_btree_cur	*cur = acur->cnt;
+	xfs_agblock_t		bno;
+	xfs_extlen_t		len, cur_len;
+	int			error;
+	int			i;
+
+	if (!xfs_alloc_cur_active(cur))
+		return 0;
+
+	/* locality optimized lookup */
+	cur_len = acur->cur_len;
+	error = xfs_alloc_lookup_ge(cur, args->agbno, cur_len, &i);
+	if (error)
+		return error;
+	if (i == 0)
+		return 0;
+	error = xfs_alloc_get_rec(cur, &bno, &len, &i);
+	if (error)
+		return error;
+
+	/* check the current record and update search length from it */
+	error = xfs_alloc_cur_check(args, acur, cur, &i);
+	if (error)
+		return error;
+	ASSERT(len >= acur->cur_len);
+	acur->cur_len = len;
+
+	/*
+	 * We looked up the first record >= [agbno, len] above. The agbno is a
+	 * secondary key and so the current record may lie just before or after
+	 * agbno. If it is past agbno, check the previous record too so long as
+	 * the length matches as it may be closer. Don't check a smaller record
+	 * because that could deactivate our cursor.
+	 */
+	if (bno > args->agbno) {
+		error = xfs_btree_decrement(cur, 0, &i);
+		if (!error && i) {
+			error = xfs_alloc_get_rec(cur, &bno, &len, &i);
+			if (!error && i && len == acur->cur_len)
+				error = xfs_alloc_cur_check(args, acur, cur,
+							    &i);
+		}
+		if (error)
+			return error;
+	}
+
+	/*
+	 * Increment the search key until we find at least one allocation
+	 * candidate or if the extent we found was larger. Otherwise, double the
+	 * search key to optimize the search. Efficiency is more important here
+	 * than absolute best locality.
+	 */
+	cur_len <<= 1;
+	if (!acur->len || acur->cur_len >= cur_len)
+		acur->cur_len++;
+	else
+		acur->cur_len = cur_len;
+
+	return error;
+}
+
+/*
+ * Deal with the case where only small freespaces remain. Either return the
+ * contents of the last freespace record, or allocate space from the freelist if
+ * there is nothing in the tree.
+ */
+STATIC int			/* error */
+xfs_alloc_ag_vextent_small(
+	struct xfs_alloc_arg	*args,	/* allocation argument structure */
+	struct xfs_btree_cur	*ccur,	/* optional by-size cursor */
+	xfs_agblock_t		*fbnop,	/* result block number */
+	xfs_extlen_t		*flenp,	/* result length */
+	int			*stat)	/* status: 0-freelist, 1-normal/none */
+{
+	struct xfs_agf		*agf = args->agbp->b_addr;
+	int			error = 0;
+	xfs_agblock_t		fbno = NULLAGBLOCK;
+	xfs_extlen_t		flen = 0;
+	int			i = 0;
+
+	/*
+	 * If a cntbt cursor is provided, try to allocate the largest record in
+	 * the tree. Try the AGFL if the cntbt is empty, otherwise fail the
+	 * allocation. Make sure to respect minleft even when pulling from the
+	 * freelist.
+	 */
+	if (ccur)
+		error = xfs_btree_decrement(ccur, 0, &i);
+	if (error)
+		goto error;
+	if (i) {
+		error = xfs_alloc_get_rec(ccur, &fbno, &flen, &i);
+		if (error)
+			goto error;
+		if (XFS_IS_CORRUPT(args->mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error;
+		}
+		goto out;
+	}
+
+	if (args->minlen != 1 || args->alignment != 1 ||
+	    args->resv == XFS_AG_RESV_AGFL ||
+	    be32_to_cpu(agf->agf_flcount) <= args->minleft)
+		goto out;
+
+	error = xfs_alloc_get_freelist(args->pag, args->tp, args->agbp,
+			&fbno, 0);
+	if (error)
+		goto error;
+	if (fbno == NULLAGBLOCK)
+		goto out;
+
+	xfs_extent_busy_reuse(args->mp, args->pag, fbno, 1,
+			      (args->datatype & XFS_ALLOC_NOBUSY));
+
+	if (args->datatype & XFS_ALLOC_USERDATA) {
+		struct xfs_buf	*bp;
+
+		error = xfs_trans_get_buf(args->tp, args->mp->m_ddev_targp,
+				XFS_AGB_TO_DADDR(args->mp, args->agno, fbno),
+				args->mp->m_bsize, 0, &bp);
+		if (error)
+			goto error;
+		xfs_trans_binval(args->tp, bp);
+	}
+	*fbnop = args->agbno = fbno;
+	*flenp = args->len = 1;
+	if (XFS_IS_CORRUPT(args->mp, fbno >= be32_to_cpu(agf->agf_length))) {
+		error = -EFSCORRUPTED;
+		goto error;
+	}
+	args->wasfromfl = 1;
+	trace_xfs_alloc_small_freelist(args);
+
+	/*
+	 * If we're feeding an AGFL block to something that doesn't live in the
+	 * free space, we need to clear out the OWN_AG rmap.
+	 */
+	error = xfs_rmap_free(args->tp, args->agbp, args->pag, fbno, 1,
+			      &XFS_RMAP_OINFO_AG);
+	if (error)
+		goto error;
+
+	*stat = 0;
+	return 0;
+
+out:
+	/*
+	 * Can't do the allocation, give up.
+	 */
+	if (flen < args->minlen) {
+		args->agbno = NULLAGBLOCK;
+		trace_xfs_alloc_small_notenough(args);
+		flen = 0;
+	}
+	*fbnop = fbno;
+	*flenp = flen;
+	*stat = 1;
+	trace_xfs_alloc_small_done(args);
+	return 0;
+
+error:
+	trace_xfs_alloc_small_error(args);
+	return error;
+}
+
+/*
+ * Allocate a variable extent in the allocation group agno.
+ * Type and bno are used to determine where in the allocation group the
+ * extent will start.
+ * Extent's length (returned in *len) will be between minlen and maxlen,
+ * and of the form k * prod + mod unless there's nothing that large.
+ * Return the starting a.g. block, or NULLAGBLOCK if we can't do it.
+ */
+STATIC int			/* error */
+xfs_alloc_ag_vextent(
+	xfs_alloc_arg_t	*args)	/* argument structure for allocation */
+{
+	int		error=0;
+
+	ASSERT(args->minlen > 0);
+	ASSERT(args->maxlen > 0);
+	ASSERT(args->minlen <= args->maxlen);
+	ASSERT(args->mod < args->prod);
+	ASSERT(args->alignment > 0);
+
+	/*
+	 * Branch to correct routine based on the type.
+	 */
+	args->wasfromfl = 0;
+	switch (args->type) {
+	case XFS_ALLOCTYPE_THIS_AG:
+		error = xfs_alloc_ag_vextent_size(args);
+		break;
+	case XFS_ALLOCTYPE_NEAR_BNO:
+		error = xfs_alloc_ag_vextent_near(args);
+		break;
+	case XFS_ALLOCTYPE_THIS_BNO:
+		error = xfs_alloc_ag_vextent_exact(args);
+		break;
+	default:
+		ASSERT(0);
+		/* NOTREACHED */
+	}
+
+	if (error || args->agbno == NULLAGBLOCK)
+		return error;
+
+	ASSERT(args->len >= args->minlen);
+	ASSERT(args->len <= args->maxlen);
+	ASSERT(!args->wasfromfl || args->resv != XFS_AG_RESV_AGFL);
+	ASSERT(args->agbno % args->alignment == 0);
+
+	/* if not file data, insert new block into the reverse map btree */
+	if (!xfs_rmap_should_skip_owner_update(&args->oinfo)) {
+		error = xfs_rmap_alloc(args->tp, args->agbp, args->pag,
+				       args->agbno, args->len, &args->oinfo);
+		if (error)
+			return error;
+	}
+
+	if (!args->wasfromfl) {
+		error = xfs_alloc_update_counters(args->tp, args->agbp,
+						  -((long)(args->len)));
+		if (error)
+			return error;
+
+		ASSERT(!xfs_extent_busy_search(args->mp, args->pag,
+					      args->agbno, args->len));
+	}
+
+	xfs_ag_resv_alloc_extent(args->pag, args->resv, args);
+
+	XFS_STATS_INC(args->mp, xs_allocx);
+	XFS_STATS_ADD(args->mp, xs_allocb, args->len);
+	return error;
+}
+
+/*
+ * Allocate a variable extent at exactly agno/bno.
+ * Extent's length (returned in *len) will be between minlen and maxlen,
+ * and of the form k * prod + mod unless there's nothing that large.
+ * Return the starting a.g. block (bno), or NULLAGBLOCK if we can't do it.
+ */
+STATIC int			/* error */
+xfs_alloc_ag_vextent_exact(
+	xfs_alloc_arg_t	*args)	/* allocation argument structure */
+{
+	struct xfs_agf __maybe_unused *agf = args->agbp->b_addr;
+	struct xfs_btree_cur *bno_cur;/* by block-number btree cursor */
+	struct xfs_btree_cur *cnt_cur;/* by count btree cursor */
+	int		error;
+	xfs_agblock_t	fbno;	/* start block of found extent */
+	xfs_extlen_t	flen;	/* length of found extent */
+	xfs_agblock_t	tbno;	/* start block of busy extent */
+	xfs_extlen_t	tlen;	/* length of busy extent */
+	xfs_agblock_t	tend;	/* end block of busy extent */
+	int		i;	/* success/failure of operation */
+	unsigned	busy_gen;
+
+	ASSERT(args->alignment == 1);
+
+	/*
+	 * Allocate/initialize a cursor for the by-number freespace btree.
+	 */
+	bno_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
+					  args->pag, XFS_BTNUM_BNO);
+
+	/*
+	 * Lookup bno and minlen in the btree (minlen is irrelevant, really).
+	 * Look for the closest free block <= bno, it must contain bno
+	 * if any free block does.
+	 */
+	error = xfs_alloc_lookup_le(bno_cur, args->agbno, args->minlen, &i);
+	if (error)
+		goto error0;
+	if (!i)
+		goto not_found;
+
+	/*
+	 * Grab the freespace record.
+	 */
+	error = xfs_alloc_get_rec(bno_cur, &fbno, &flen, &i);
+	if (error)
+		goto error0;
+	if (XFS_IS_CORRUPT(args->mp, i != 1)) {
+		error = -EFSCORRUPTED;
+		goto error0;
+	}
+	ASSERT(fbno <= args->agbno);
+
+	/*
+	 * Check for overlapping busy extents.
+	 */
+	tbno = fbno;
+	tlen = flen;
+	xfs_extent_busy_trim(args, &tbno, &tlen, &busy_gen);
+
+	/*
+	 * Give up if the start of the extent is busy, or the freespace isn't
+	 * long enough for the minimum request.
+	 */
+	if (tbno > args->agbno)
+		goto not_found;
+	if (tlen < args->minlen)
+		goto not_found;
+	tend = tbno + tlen;
+	if (tend < args->agbno + args->minlen)
+		goto not_found;
+
+	/*
+	 * End of extent will be smaller of the freespace end and the
+	 * maximal requested end.
+	 *
+	 * Fix the length according to mod and prod if given.
+	 */
+	args->len = XFS_AGBLOCK_MIN(tend, args->agbno + args->maxlen)
+						- args->agbno;
+	xfs_alloc_fix_len(args);
+	ASSERT(args->agbno + args->len <= tend);
+
+	/*
+	 * We are allocating agbno for args->len
+	 * Allocate/initialize a cursor for the by-size btree.
+	 */
+	cnt_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
+					args->pag, XFS_BTNUM_CNT);
+	ASSERT(args->agbno + args->len <= be32_to_cpu(agf->agf_length));
+	error = xfs_alloc_fixup_trees(cnt_cur, bno_cur, fbno, flen, args->agbno,
+				      args->len, XFSA_FIXUP_BNO_OK);
+	if (error) {
+		xfs_btree_del_cursor(cnt_cur, XFS_BTREE_ERROR);
+		goto error0;
+	}
+
+	xfs_btree_del_cursor(bno_cur, XFS_BTREE_NOERROR);
+	xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+
+	args->wasfromfl = 0;
+	trace_xfs_alloc_exact_done(args);
+	return 0;
+
+not_found:
+	/* Didn't find it, return null. */
+	xfs_btree_del_cursor(bno_cur, XFS_BTREE_NOERROR);
+	args->agbno = NULLAGBLOCK;
+	trace_xfs_alloc_exact_notfound(args);
+	return 0;
+
+error0:
+	xfs_btree_del_cursor(bno_cur, XFS_BTREE_ERROR);
+	trace_xfs_alloc_exact_error(args);
+	return error;
+}
+
+/*
+ * Search a given number of btree records in a given direction. Check each
+ * record against the good extent we've already found.
+ */
+STATIC int
+xfs_alloc_walk_iter(
+	struct xfs_alloc_arg	*args,
+	struct xfs_alloc_cur	*acur,
+	struct xfs_btree_cur	*cur,
+	bool			increment,
+	bool			find_one, /* quit on first candidate */
+	int			count,    /* rec count (-1 for infinite) */
+	int			*stat)
+{
+	int			error;
+	int			i;
+
+	*stat = 0;
+
+	/*
+	 * Search so long as the cursor is active or we find a better extent.
+	 * The cursor is deactivated if it extends beyond the range of the
+	 * current allocation candidate.
+	 */
+	while (xfs_alloc_cur_active(cur) && count) {
+		error = xfs_alloc_cur_check(args, acur, cur, &i);
+		if (error)
+			return error;
+		if (i == 1) {
+			*stat = 1;
+			if (find_one)
+				break;
+		}
+		if (!xfs_alloc_cur_active(cur))
+			break;
+
+		if (increment)
+			error = xfs_btree_increment(cur, 0, &i);
+		else
+			error = xfs_btree_decrement(cur, 0, &i);
+		if (error)
+			return error;
+		if (i == 0)
+			cur->bc_ag.abt.active = false;
+
+		if (count > 0)
+			count--;
+	}
+
+	return 0;
+}
+
+/*
+ * Search the by-bno and by-size btrees in parallel in search of an extent with
+ * ideal locality based on the NEAR mode ->agbno locality hint.
+ */
+STATIC int
+xfs_alloc_ag_vextent_locality(
+	struct xfs_alloc_arg	*args,
+	struct xfs_alloc_cur	*acur,
+	int			*stat)
+{
+	struct xfs_btree_cur	*fbcur = NULL;
+	int			error;
+	int			i;
+	bool			fbinc;
+
+	ASSERT(acur->len == 0);
+	ASSERT(args->type == XFS_ALLOCTYPE_NEAR_BNO);
+
+	*stat = 0;
+
+	error = xfs_alloc_lookup_ge(acur->cnt, args->agbno, acur->cur_len, &i);
+	if (error)
+		return error;
+	error = xfs_alloc_lookup_le(acur->bnolt, args->agbno, 0, &i);
+	if (error)
+		return error;
+	error = xfs_alloc_lookup_ge(acur->bnogt, args->agbno, 0, &i);
+	if (error)
+		return error;
+
+	/*
+	 * Search the bnobt and cntbt in parallel. Search the bnobt left and
+	 * right and lookup the closest extent to the locality hint for each
+	 * extent size key in the cntbt. The entire search terminates
+	 * immediately on a bnobt hit because that means we've found best case
+	 * locality. Otherwise the search continues until the cntbt cursor runs
+	 * off the end of the tree. If no allocation candidate is found at this
+	 * point, give up on locality, walk backwards from the end of the cntbt
+	 * and take the first available extent.
+	 *
+	 * The parallel tree searches balance each other out to provide fairly
+	 * consistent performance for various situations. The bnobt search can
+	 * have pathological behavior in the worst case scenario of larger
+	 * allocation requests and fragmented free space. On the other hand, the
+	 * bnobt is able to satisfy most smaller allocation requests much more
+	 * quickly than the cntbt. The cntbt search can sift through fragmented
+	 * free space and sets of free extents for larger allocation requests
+	 * more quickly than the bnobt. Since the locality hint is just a hint
+	 * and we don't want to scan the entire bnobt for perfect locality, the
+	 * cntbt search essentially bounds the bnobt search such that we can
+	 * find good enough locality at reasonable performance in most cases.
+	 */
+	while (xfs_alloc_cur_active(acur->bnolt) ||
+	       xfs_alloc_cur_active(acur->bnogt) ||
+	       xfs_alloc_cur_active(acur->cnt)) {
+
+		trace_xfs_alloc_cur_lookup(args);
+
+		/*
+		 * Search the bnobt left and right. In the case of a hit, finish
+		 * the search in the opposite direction and we're done.
+		 */
+		error = xfs_alloc_walk_iter(args, acur, acur->bnolt, false,
+					    true, 1, &i);
+		if (error)
+			return error;
+		if (i == 1) {
+			trace_xfs_alloc_cur_left(args);
+			fbcur = acur->bnogt;
+			fbinc = true;
+			break;
+		}
+		error = xfs_alloc_walk_iter(args, acur, acur->bnogt, true, true,
+					    1, &i);
+		if (error)
+			return error;
+		if (i == 1) {
+			trace_xfs_alloc_cur_right(args);
+			fbcur = acur->bnolt;
+			fbinc = false;
+			break;
+		}
+
+		/*
+		 * Check the extent with best locality based on the current
+		 * extent size search key and keep track of the best candidate.
+		 */
+		error = xfs_alloc_cntbt_iter(args, acur);
+		if (error)
+			return error;
+		if (!xfs_alloc_cur_active(acur->cnt)) {
+			trace_xfs_alloc_cur_lookup_done(args);
+			break;
+		}
+	}
+
+	/*
+	 * If we failed to find anything due to busy extents, return empty
+	 * handed so the caller can flush and retry. If no busy extents were
+	 * found, walk backwards from the end of the cntbt as a last resort.
+	 */
+	if (!xfs_alloc_cur_active(acur->cnt) && !acur->len && !acur->busy) {
+		error = xfs_btree_decrement(acur->cnt, 0, &i);
+		if (error)
+			return error;
+		if (i) {
+			acur->cnt->bc_ag.abt.active = true;
+			fbcur = acur->cnt;
+			fbinc = false;
+		}
+	}
+
+	/*
+	 * Search in the opposite direction for a better entry in the case of
+	 * a bnobt hit or walk backwards from the end of the cntbt.
+	 */
+	if (fbcur) {
+		error = xfs_alloc_walk_iter(args, acur, fbcur, fbinc, true, -1,
+					    &i);
+		if (error)
+			return error;
+	}
+
+	if (acur->len)
+		*stat = 1;
+
+	return 0;
+}
+
+/* Check the last block of the cnt btree for allocations. */
+static int
+xfs_alloc_ag_vextent_lastblock(
+	struct xfs_alloc_arg	*args,
+	struct xfs_alloc_cur	*acur,
+	xfs_agblock_t		*bno,
+	xfs_extlen_t		*len,
+	bool			*allocated)
+{
+	int			error;
+	int			i;
+
+#ifdef DEBUG
+	/* Randomly don't execute the first algorithm. */
+	if (prandom_u32_max(2))
+		return 0;
+#endif
+
+	/*
+	 * Start from the entry that lookup found, sequence through all larger
+	 * free blocks.  If we're actually pointing at a record smaller than
+	 * maxlen, go to the start of this block, and skip all those smaller
+	 * than minlen.
+	 */
+	if (*len || args->alignment > 1) {
+		acur->cnt->bc_levels[0].ptr = 1;
+		do {
+			error = xfs_alloc_get_rec(acur->cnt, bno, len, &i);
+			if (error)
+				return error;
+			if (XFS_IS_CORRUPT(args->mp, i != 1))
+				return -EFSCORRUPTED;
+			if (*len >= args->minlen)
+				break;
+			error = xfs_btree_increment(acur->cnt, 0, &i);
+			if (error)
+				return error;
+		} while (i);
+		ASSERT(*len >= args->minlen);
+		if (!i)
+			return 0;
+	}
+
+	error = xfs_alloc_walk_iter(args, acur, acur->cnt, true, false, -1, &i);
+	if (error)
+		return error;
+
+	/*
+	 * It didn't work.  We COULD be in a case where there's a good record
+	 * somewhere, so try again.
+	 */
+	if (acur->len == 0)
+		return 0;
+
+	trace_xfs_alloc_near_first(args);
+	*allocated = true;
+	return 0;
+}
+
+/*
+ * Allocate a variable extent near bno in the allocation group agno.
+ * Extent's length (returned in len) will be between minlen and maxlen,
+ * and of the form k * prod + mod unless there's nothing that large.
+ * Return the starting a.g. block, or NULLAGBLOCK if we can't do it.
+ */
+STATIC int
+xfs_alloc_ag_vextent_near(
+	struct xfs_alloc_arg	*args)
+{
+	struct xfs_alloc_cur	acur = {};
+	int			error;		/* error code */
+	int			i;		/* result code, temporary */
+	xfs_agblock_t		bno;
+	xfs_extlen_t		len;
+
+	/* handle uninitialized agbno range so caller doesn't have to */
+	if (!args->min_agbno && !args->max_agbno)
+		args->max_agbno = args->mp->m_sb.sb_agblocks - 1;
+	ASSERT(args->min_agbno <= args->max_agbno);
+
+	/* clamp agbno to the range if it's outside */
+	if (args->agbno < args->min_agbno)
+		args->agbno = args->min_agbno;
+	if (args->agbno > args->max_agbno)
+		args->agbno = args->max_agbno;
+
+restart:
+	len = 0;
+
+	/*
+	 * Set up cursors and see if there are any free extents as big as
+	 * maxlen. If not, pick the last entry in the tree unless the tree is
+	 * empty.
+	 */
+	error = xfs_alloc_cur_setup(args, &acur);
+	if (error == -ENOSPC) {
+		error = xfs_alloc_ag_vextent_small(args, acur.cnt, &bno,
+				&len, &i);
+		if (error)
+			goto out;
+		if (i == 0 || len == 0) {
+			trace_xfs_alloc_near_noentry(args);
+			goto out;
+		}
+		ASSERT(i == 1);
+	} else if (error) {
+		goto out;
+	}
+
+	/*
+	 * First algorithm.
+	 * If the requested extent is large wrt the freespaces available
+	 * in this a.g., then the cursor will be pointing to a btree entry
+	 * near the right edge of the tree.  If it's in the last btree leaf
+	 * block, then we just examine all the entries in that block
+	 * that are big enough, and pick the best one.
+	 */
+	if (xfs_btree_islastblock(acur.cnt, 0)) {
+		bool		allocated = false;
+
+		error = xfs_alloc_ag_vextent_lastblock(args, &acur, &bno, &len,
+				&allocated);
+		if (error)
+			goto out;
+		if (allocated)
+			goto alloc_finish;
+	}
+
+	/*
+	 * Second algorithm. Combined cntbt and bnobt search to find ideal
+	 * locality.
+	 */
+	error = xfs_alloc_ag_vextent_locality(args, &acur, &i);
+	if (error)
+		goto out;
+
+	/*
+	 * If we couldn't get anything, give up.
+	 */
+	if (!acur.len) {
+		if (acur.busy) {
+			trace_xfs_alloc_near_busy(args);
+			xfs_extent_busy_flush(args->mp, args->pag,
+					      acur.busy_gen);
+			goto restart;
+		}
+		trace_xfs_alloc_size_neither(args);
+		args->agbno = NULLAGBLOCK;
+		goto out;
+	}
+
+alloc_finish:
+	/* fix up btrees on a successful allocation */
+	error = xfs_alloc_cur_finish(args, &acur);
+
+out:
+	xfs_alloc_cur_close(&acur, error);
+	return error;
+}
+
+/*
+ * Allocate a variable extent anywhere in the allocation group agno.
+ * Extent's length (returned in len) will be between minlen and maxlen,
+ * and of the form k * prod + mod unless there's nothing that large.
+ * Return the starting a.g. block, or NULLAGBLOCK if we can't do it.
+ */
+STATIC int				/* error */
+xfs_alloc_ag_vextent_size(
+	xfs_alloc_arg_t	*args)		/* allocation argument structure */
+{
+	struct xfs_agf	*agf = args->agbp->b_addr;
+	struct xfs_btree_cur *bno_cur;	/* cursor for bno btree */
+	struct xfs_btree_cur *cnt_cur;	/* cursor for cnt btree */
+	int		error;		/* error result */
+	xfs_agblock_t	fbno;		/* start of found freespace */
+	xfs_extlen_t	flen;		/* length of found freespace */
+	int		i;		/* temp status variable */
+	xfs_agblock_t	rbno;		/* returned block number */
+	xfs_extlen_t	rlen;		/* length of returned extent */
+	bool		busy;
+	unsigned	busy_gen;
+
+restart:
+	/*
+	 * Allocate and initialize a cursor for the by-size btree.
+	 */
+	cnt_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
+					args->pag, XFS_BTNUM_CNT);
+	bno_cur = NULL;
+
+	/*
+	 * Look for an entry >= maxlen+alignment-1 blocks.
+	 */
+	if ((error = xfs_alloc_lookup_ge(cnt_cur, 0,
+			args->maxlen + args->alignment - 1, &i)))
+		goto error0;
+
+	/*
+	 * If none then we have to settle for a smaller extent. In the case that
+	 * there are no large extents, this will return the last entry in the
+	 * tree unless the tree is empty. In the case that there are only busy
+	 * large extents, this will return the largest small extent unless there
+	 * are no smaller extents available.
+	 */
+	if (!i) {
+		error = xfs_alloc_ag_vextent_small(args, cnt_cur,
+						   &fbno, &flen, &i);
+		if (error)
+			goto error0;
+		if (i == 0 || flen == 0) {
+			xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+			trace_xfs_alloc_size_noentry(args);
+			return 0;
+		}
+		ASSERT(i == 1);
+		busy = xfs_alloc_compute_aligned(args, fbno, flen, &rbno,
+				&rlen, &busy_gen);
+	} else {
+		/*
+		 * Search for a non-busy extent that is large enough.
+		 */
+		for (;;) {
+			error = xfs_alloc_get_rec(cnt_cur, &fbno, &flen, &i);
+			if (error)
+				goto error0;
+			if (XFS_IS_CORRUPT(args->mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto error0;
+			}
+
+			busy = xfs_alloc_compute_aligned(args, fbno, flen,
+					&rbno, &rlen, &busy_gen);
+
+			if (rlen >= args->maxlen)
+				break;
+
+			error = xfs_btree_increment(cnt_cur, 0, &i);
+			if (error)
+				goto error0;
+			if (i == 0) {
+				/*
+				 * Our only valid extents must have been busy.
+				 * Make it unbusy by forcing the log out and
+				 * retrying.
+				 */
+				xfs_btree_del_cursor(cnt_cur,
+						     XFS_BTREE_NOERROR);
+				trace_xfs_alloc_size_busy(args);
+				xfs_extent_busy_flush(args->mp,
+							args->pag, busy_gen);
+				goto restart;
+			}
+		}
+	}
+
+	/*
+	 * In the first case above, we got the last entry in the
+	 * by-size btree.  Now we check to see if the space hits maxlen
+	 * once aligned; if not, we search left for something better.
+	 * This can't happen in the second case above.
+	 */
+	rlen = XFS_EXTLEN_MIN(args->maxlen, rlen);
+	if (XFS_IS_CORRUPT(args->mp,
+			   rlen != 0 &&
+			   (rlen > flen ||
+			    rbno + rlen > fbno + flen))) {
+		error = -EFSCORRUPTED;
+		goto error0;
+	}
+	if (rlen < args->maxlen) {
+		xfs_agblock_t	bestfbno;
+		xfs_extlen_t	bestflen;
+		xfs_agblock_t	bestrbno;
+		xfs_extlen_t	bestrlen;
+
+		bestrlen = rlen;
+		bestrbno = rbno;
+		bestflen = flen;
+		bestfbno = fbno;
+		for (;;) {
+			if ((error = xfs_btree_decrement(cnt_cur, 0, &i)))
+				goto error0;
+			if (i == 0)
+				break;
+			if ((error = xfs_alloc_get_rec(cnt_cur, &fbno, &flen,
+					&i)))
+				goto error0;
+			if (XFS_IS_CORRUPT(args->mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto error0;
+			}
+			if (flen < bestrlen)
+				break;
+			busy = xfs_alloc_compute_aligned(args, fbno, flen,
+					&rbno, &rlen, &busy_gen);
+			rlen = XFS_EXTLEN_MIN(args->maxlen, rlen);
+			if (XFS_IS_CORRUPT(args->mp,
+					   rlen != 0 &&
+					   (rlen > flen ||
+					    rbno + rlen > fbno + flen))) {
+				error = -EFSCORRUPTED;
+				goto error0;
+			}
+			if (rlen > bestrlen) {
+				bestrlen = rlen;
+				bestrbno = rbno;
+				bestflen = flen;
+				bestfbno = fbno;
+				if (rlen == args->maxlen)
+					break;
+			}
+		}
+		if ((error = xfs_alloc_lookup_eq(cnt_cur, bestfbno, bestflen,
+				&i)))
+			goto error0;
+		if (XFS_IS_CORRUPT(args->mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+		rlen = bestrlen;
+		rbno = bestrbno;
+		flen = bestflen;
+		fbno = bestfbno;
+	}
+	args->wasfromfl = 0;
+	/*
+	 * Fix up the length.
+	 */
+	args->len = rlen;
+	if (rlen < args->minlen) {
+		if (busy) {
+			xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+			trace_xfs_alloc_size_busy(args);
+			xfs_extent_busy_flush(args->mp, args->pag, busy_gen);
+			goto restart;
+		}
+		goto out_nominleft;
+	}
+	xfs_alloc_fix_len(args);
+
+	rlen = args->len;
+	if (XFS_IS_CORRUPT(args->mp, rlen > flen)) {
+		error = -EFSCORRUPTED;
+		goto error0;
+	}
+	/*
+	 * Allocate and initialize a cursor for the by-block tree.
+	 */
+	bno_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
+					args->pag, XFS_BTNUM_BNO);
+	if ((error = xfs_alloc_fixup_trees(cnt_cur, bno_cur, fbno, flen,
+			rbno, rlen, XFSA_FIXUP_CNT_OK)))
+		goto error0;
+	xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+	xfs_btree_del_cursor(bno_cur, XFS_BTREE_NOERROR);
+	cnt_cur = bno_cur = NULL;
+	args->len = rlen;
+	args->agbno = rbno;
+	if (XFS_IS_CORRUPT(args->mp,
+			   args->agbno + args->len >
+			   be32_to_cpu(agf->agf_length))) {
+		error = -EFSCORRUPTED;
+		goto error0;
+	}
+	trace_xfs_alloc_size_done(args);
+	return 0;
+
+error0:
+	trace_xfs_alloc_size_error(args);
+	if (cnt_cur)
+		xfs_btree_del_cursor(cnt_cur, XFS_BTREE_ERROR);
+	if (bno_cur)
+		xfs_btree_del_cursor(bno_cur, XFS_BTREE_ERROR);
+	return error;
+
+out_nominleft:
+	xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+	trace_xfs_alloc_size_nominleft(args);
+	args->agbno = NULLAGBLOCK;
+	return 0;
+}
+
+/*
+ * Free the extent starting at agno/bno for length.
+ */
+STATIC int
+xfs_free_ag_extent(
+	struct xfs_trans		*tp,
+	struct xfs_buf			*agbp,
+	xfs_agnumber_t			agno,
+	xfs_agblock_t			bno,
+	xfs_extlen_t			len,
+	const struct xfs_owner_info	*oinfo,
+	enum xfs_ag_resv_type		type)
+{
+	struct xfs_mount		*mp;
+	struct xfs_btree_cur		*bno_cur;
+	struct xfs_btree_cur		*cnt_cur;
+	xfs_agblock_t			gtbno; /* start of right neighbor */
+	xfs_extlen_t			gtlen; /* length of right neighbor */
+	xfs_agblock_t			ltbno; /* start of left neighbor */
+	xfs_extlen_t			ltlen; /* length of left neighbor */
+	xfs_agblock_t			nbno; /* new starting block of freesp */
+	xfs_extlen_t			nlen; /* new length of freespace */
+	int				haveleft; /* have a left neighbor */
+	int				haveright; /* have a right neighbor */
+	int				i;
+	int				error;
+	struct xfs_perag		*pag = agbp->b_pag;
+
+	bno_cur = cnt_cur = NULL;
+	mp = tp->t_mountp;
+
+	if (!xfs_rmap_should_skip_owner_update(oinfo)) {
+		error = xfs_rmap_free(tp, agbp, pag, bno, len, oinfo);
+		if (error)
+			goto error0;
+	}
+
+	/*
+	 * Allocate and initialize a cursor for the by-block btree.
+	 */
+	bno_cur = xfs_allocbt_init_cursor(mp, tp, agbp, pag, XFS_BTNUM_BNO);
+	/*
+	 * Look for a neighboring block on the left (lower block numbers)
+	 * that is contiguous with this space.
+	 */
+	if ((error = xfs_alloc_lookup_le(bno_cur, bno, len, &haveleft)))
+		goto error0;
+	if (haveleft) {
+		/*
+		 * There is a block to our left.
+		 */
+		if ((error = xfs_alloc_get_rec(bno_cur, &ltbno, &ltlen, &i)))
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+		/*
+		 * It's not contiguous, though.
+		 */
+		if (ltbno + ltlen < bno)
+			haveleft = 0;
+		else {
+			/*
+			 * If this failure happens the request to free this
+			 * space was invalid, it's (partly) already free.
+			 * Very bad.
+			 */
+			if (XFS_IS_CORRUPT(mp, ltbno + ltlen > bno)) {
+				error = -EFSCORRUPTED;
+				goto error0;
+			}
+		}
+	}
+	/*
+	 * Look for a neighboring block on the right (higher block numbers)
+	 * that is contiguous with this space.
+	 */
+	if ((error = xfs_btree_increment(bno_cur, 0, &haveright)))
+		goto error0;
+	if (haveright) {
+		/*
+		 * There is a block to our right.
+		 */
+		if ((error = xfs_alloc_get_rec(bno_cur, &gtbno, &gtlen, &i)))
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+		/*
+		 * It's not contiguous, though.
+		 */
+		if (bno + len < gtbno)
+			haveright = 0;
+		else {
+			/*
+			 * If this failure happens the request to free this
+			 * space was invalid, it's (partly) already free.
+			 * Very bad.
+			 */
+			if (XFS_IS_CORRUPT(mp, bno + len > gtbno)) {
+				error = -EFSCORRUPTED;
+				goto error0;
+			}
+		}
+	}
+	/*
+	 * Now allocate and initialize a cursor for the by-size tree.
+	 */
+	cnt_cur = xfs_allocbt_init_cursor(mp, tp, agbp, pag, XFS_BTNUM_CNT);
+	/*
+	 * Have both left and right contiguous neighbors.
+	 * Merge all three into a single free block.
+	 */
+	if (haveleft && haveright) {
+		/*
+		 * Delete the old by-size entry on the left.
+		 */
+		if ((error = xfs_alloc_lookup_eq(cnt_cur, ltbno, ltlen, &i)))
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+		if ((error = xfs_btree_delete(cnt_cur, &i)))
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+		/*
+		 * Delete the old by-size entry on the right.
+		 */
+		if ((error = xfs_alloc_lookup_eq(cnt_cur, gtbno, gtlen, &i)))
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+		if ((error = xfs_btree_delete(cnt_cur, &i)))
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+		/*
+		 * Delete the old by-block entry for the right block.
+		 */
+		if ((error = xfs_btree_delete(bno_cur, &i)))
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+		/*
+		 * Move the by-block cursor back to the left neighbor.
+		 */
+		if ((error = xfs_btree_decrement(bno_cur, 0, &i)))
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+#ifdef DEBUG
+		/*
+		 * Check that this is the right record: delete didn't
+		 * mangle the cursor.
+		 */
+		{
+			xfs_agblock_t	xxbno;
+			xfs_extlen_t	xxlen;
+
+			if ((error = xfs_alloc_get_rec(bno_cur, &xxbno, &xxlen,
+					&i)))
+				goto error0;
+			if (XFS_IS_CORRUPT(mp,
+					   i != 1 ||
+					   xxbno != ltbno ||
+					   xxlen != ltlen)) {
+				error = -EFSCORRUPTED;
+				goto error0;
+			}
+		}
+#endif
+		/*
+		 * Update remaining by-block entry to the new, joined block.
+		 */
+		nbno = ltbno;
+		nlen = len + ltlen + gtlen;
+		if ((error = xfs_alloc_update(bno_cur, nbno, nlen)))
+			goto error0;
+	}
+	/*
+	 * Have only a left contiguous neighbor.
+	 * Merge it together with the new freespace.
+	 */
+	else if (haveleft) {
+		/*
+		 * Delete the old by-size entry on the left.
+		 */
+		if ((error = xfs_alloc_lookup_eq(cnt_cur, ltbno, ltlen, &i)))
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+		if ((error = xfs_btree_delete(cnt_cur, &i)))
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+		/*
+		 * Back up the by-block cursor to the left neighbor, and
+		 * update its length.
+		 */
+		if ((error = xfs_btree_decrement(bno_cur, 0, &i)))
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+		nbno = ltbno;
+		nlen = len + ltlen;
+		if ((error = xfs_alloc_update(bno_cur, nbno, nlen)))
+			goto error0;
+	}
+	/*
+	 * Have only a right contiguous neighbor.
+	 * Merge it together with the new freespace.
+	 */
+	else if (haveright) {
+		/*
+		 * Delete the old by-size entry on the right.
+		 */
+		if ((error = xfs_alloc_lookup_eq(cnt_cur, gtbno, gtlen, &i)))
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+		if ((error = xfs_btree_delete(cnt_cur, &i)))
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+		/*
+		 * Update the starting block and length of the right
+		 * neighbor in the by-block tree.
+		 */
+		nbno = bno;
+		nlen = len + gtlen;
+		if ((error = xfs_alloc_update(bno_cur, nbno, nlen)))
+			goto error0;
+	}
+	/*
+	 * No contiguous neighbors.
+	 * Insert the new freespace into the by-block tree.
+	 */
+	else {
+		nbno = bno;
+		nlen = len;
+		if ((error = xfs_btree_insert(bno_cur, &i)))
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+	}
+	xfs_btree_del_cursor(bno_cur, XFS_BTREE_NOERROR);
+	bno_cur = NULL;
+	/*
+	 * In all cases we need to insert the new freespace in the by-size tree.
+	 */
+	if ((error = xfs_alloc_lookup_eq(cnt_cur, nbno, nlen, &i)))
+		goto error0;
+	if (XFS_IS_CORRUPT(mp, i != 0)) {
+		error = -EFSCORRUPTED;
+		goto error0;
+	}
+	if ((error = xfs_btree_insert(cnt_cur, &i)))
+		goto error0;
+	if (XFS_IS_CORRUPT(mp, i != 1)) {
+		error = -EFSCORRUPTED;
+		goto error0;
+	}
+	xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+	cnt_cur = NULL;
+
+	/*
+	 * Update the freespace totals in the ag and superblock.
+	 */
+	error = xfs_alloc_update_counters(tp, agbp, len);
+	xfs_ag_resv_free_extent(agbp->b_pag, type, tp, len);
+	if (error)
+		goto error0;
+
+	XFS_STATS_INC(mp, xs_freex);
+	XFS_STATS_ADD(mp, xs_freeb, len);
+
+	trace_xfs_free_extent(mp, agno, bno, len, type, haveleft, haveright);
+
+	return 0;
+
+ error0:
+	trace_xfs_free_extent(mp, agno, bno, len, type, -1, -1);
+	if (bno_cur)
+		xfs_btree_del_cursor(bno_cur, XFS_BTREE_ERROR);
+	if (cnt_cur)
+		xfs_btree_del_cursor(cnt_cur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/*
+ * Visible (exported) allocation/free functions.
+ * Some of these are used just by xfs_alloc_btree.c and this file.
+ */
+
+/*
+ * Compute and fill in value of m_alloc_maxlevels.
+ */
+void
+xfs_alloc_compute_maxlevels(
+	xfs_mount_t	*mp)	/* file system mount structure */
+{
+	mp->m_alloc_maxlevels = xfs_btree_compute_maxlevels(mp->m_alloc_mnr,
+			(mp->m_sb.sb_agblocks + 1) / 2);
+	ASSERT(mp->m_alloc_maxlevels <= xfs_allocbt_maxlevels_ondisk());
+}
+
+/*
+ * Find the length of the longest extent in an AG.  The 'need' parameter
+ * specifies how much space we're going to need for the AGFL and the
+ * 'reserved' parameter tells us how many blocks in this AG are reserved for
+ * other callers.
+ */
+xfs_extlen_t
+xfs_alloc_longest_free_extent(
+	struct xfs_perag	*pag,
+	xfs_extlen_t		need,
+	xfs_extlen_t		reserved)
+{
+	xfs_extlen_t		delta = 0;
+
+	/*
+	 * If the AGFL needs a recharge, we'll have to subtract that from the
+	 * longest extent.
+	 */
+	if (need > pag->pagf_flcount)
+		delta = need - pag->pagf_flcount;
+
+	/*
+	 * If we cannot maintain others' reservations with space from the
+	 * not-longest freesp extents, we'll have to subtract /that/ from
+	 * the longest extent too.
+	 */
+	if (pag->pagf_freeblks - pag->pagf_longest < reserved)
+		delta += reserved - (pag->pagf_freeblks - pag->pagf_longest);
+
+	/*
+	 * If the longest extent is long enough to satisfy all the
+	 * reservations and AGFL rules in place, we can return this extent.
+	 */
+	if (pag->pagf_longest > delta)
+		return min_t(xfs_extlen_t, pag->pag_mount->m_ag_max_usable,
+				pag->pagf_longest - delta);
+
+	/* Otherwise, let the caller try for 1 block if there's space. */
+	return pag->pagf_flcount > 0 || pag->pagf_longest > 0;
+}
+
+/*
+ * Compute the minimum length of the AGFL in the given AG.  If @pag is NULL,
+ * return the largest possible minimum length.
+ */
+unsigned int
+xfs_alloc_min_freelist(
+	struct xfs_mount	*mp,
+	struct xfs_perag	*pag)
+{
+	/* AG btrees have at least 1 level. */
+	static const uint8_t	fake_levels[XFS_BTNUM_AGF] = {1, 1, 1};
+	const uint8_t		*levels = pag ? pag->pagf_levels : fake_levels;
+	unsigned int		min_free;
+
+	ASSERT(mp->m_alloc_maxlevels > 0);
+
+	/* space needed by-bno freespace btree */
+	min_free = min_t(unsigned int, levels[XFS_BTNUM_BNOi] + 1,
+				       mp->m_alloc_maxlevels);
+	/* space needed by-size freespace btree */
+	min_free += min_t(unsigned int, levels[XFS_BTNUM_CNTi] + 1,
+				       mp->m_alloc_maxlevels);
+	/* space needed reverse mapping used space btree */
+	if (xfs_has_rmapbt(mp))
+		min_free += min_t(unsigned int, levels[XFS_BTNUM_RMAPi] + 1,
+						mp->m_rmap_maxlevels);
+
+	return min_free;
+}
+
+/*
+ * Check if the operation we are fixing up the freelist for should go ahead or
+ * not. If we are freeing blocks, we always allow it, otherwise the allocation
+ * is dependent on whether the size and shape of free space available will
+ * permit the requested allocation to take place.
+ */
+static bool
+xfs_alloc_space_available(
+	struct xfs_alloc_arg	*args,
+	xfs_extlen_t		min_free,
+	int			flags)
+{
+	struct xfs_perag	*pag = args->pag;
+	xfs_extlen_t		alloc_len, longest;
+	xfs_extlen_t		reservation; /* blocks that are still reserved */
+	int			available;
+	xfs_extlen_t		agflcount;
+
+	if (flags & XFS_ALLOC_FLAG_FREEING)
+		return true;
+
+	reservation = xfs_ag_resv_needed(pag, args->resv);
+
+	/* do we have enough contiguous free space for the allocation? */
+	alloc_len = args->minlen + (args->alignment - 1) + args->minalignslop;
+	longest = xfs_alloc_longest_free_extent(pag, min_free, reservation);
+	if (longest < alloc_len)
+		return false;
+
+	/*
+	 * Do we have enough free space remaining for the allocation? Don't
+	 * account extra agfl blocks because we are about to defer free them,
+	 * making them unavailable until the current transaction commits.
+	 */
+	agflcount = min_t(xfs_extlen_t, pag->pagf_flcount, min_free);
+	available = (int)(pag->pagf_freeblks + agflcount -
+			  reservation - min_free - args->minleft);
+	if (available < (int)max(args->total, alloc_len))
+		return false;
+
+	/*
+	 * Clamp maxlen to the amount of free space available for the actual
+	 * extent allocation.
+	 */
+	if (available < (int)args->maxlen && !(flags & XFS_ALLOC_FLAG_CHECK)) {
+		args->maxlen = available;
+		ASSERT(args->maxlen > 0);
+		ASSERT(args->maxlen >= args->minlen);
+	}
+
+	return true;
+}
+
+int
+xfs_free_agfl_block(
+	struct xfs_trans	*tp,
+	xfs_agnumber_t		agno,
+	xfs_agblock_t		agbno,
+	struct xfs_buf		*agbp,
+	struct xfs_owner_info	*oinfo)
+{
+	int			error;
+	struct xfs_buf		*bp;
+
+	error = xfs_free_ag_extent(tp, agbp, agno, agbno, 1, oinfo,
+				   XFS_AG_RESV_AGFL);
+	if (error)
+		return error;
+
+	error = xfs_trans_get_buf(tp, tp->t_mountp->m_ddev_targp,
+			XFS_AGB_TO_DADDR(tp->t_mountp, agno, agbno),
+			tp->t_mountp->m_bsize, 0, &bp);
+	if (error)
+		return error;
+	xfs_trans_binval(tp, bp);
+
+	return 0;
+}
+
+/*
+ * Check the agfl fields of the agf for inconsistency or corruption. The purpose
+ * is to detect an agfl header padding mismatch between current and early v5
+ * kernels. This problem manifests as a 1-slot size difference between the
+ * on-disk flcount and the active [first, last] range of a wrapped agfl. This
+ * may also catch variants of agfl count corruption unrelated to padding. Either
+ * way, we'll reset the agfl and warn the user.
+ *
+ * Return true if a reset is required before the agfl can be used, false
+ * otherwise.
+ */
+static bool
+xfs_agfl_needs_reset(
+	struct xfs_mount	*mp,
+	struct xfs_agf		*agf)
+{
+	uint32_t		f = be32_to_cpu(agf->agf_flfirst);
+	uint32_t		l = be32_to_cpu(agf->agf_fllast);
+	uint32_t		c = be32_to_cpu(agf->agf_flcount);
+	int			agfl_size = xfs_agfl_size(mp);
+	int			active;
+
+	/* no agfl header on v4 supers */
+	if (!xfs_has_crc(mp))
+		return false;
+
+	/*
+	 * The agf read verifier catches severe corruption of these fields.
+	 * Repeat some sanity checks to cover a packed -> unpacked mismatch if
+	 * the verifier allows it.
+	 */
+	if (f >= agfl_size || l >= agfl_size)
+		return true;
+	if (c > agfl_size)
+		return true;
+
+	/*
+	 * Check consistency between the on-disk count and the active range. An
+	 * agfl padding mismatch manifests as an inconsistent flcount.
+	 */
+	if (c && l >= f)
+		active = l - f + 1;
+	else if (c)
+		active = agfl_size - f + l + 1;
+	else
+		active = 0;
+
+	return active != c;
+}
+
+/*
+ * Reset the agfl to an empty state. Ignore/drop any existing blocks since the
+ * agfl content cannot be trusted. Warn the user that a repair is required to
+ * recover leaked blocks.
+ *
+ * The purpose of this mechanism is to handle filesystems affected by the agfl
+ * header padding mismatch problem. A reset keeps the filesystem online with a
+ * relatively minor free space accounting inconsistency rather than suffer the
+ * inevitable crash from use of an invalid agfl block.
+ */
+static void
+xfs_agfl_reset(
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp,
+	struct xfs_perag	*pag)
+{
+	struct xfs_mount	*mp = tp->t_mountp;
+	struct xfs_agf		*agf = agbp->b_addr;
+
+	ASSERT(pag->pagf_agflreset);
+	trace_xfs_agfl_reset(mp, agf, 0, _RET_IP_);
+
+	xfs_warn(mp,
+	       "WARNING: Reset corrupted AGFL on AG %u. %d blocks leaked. "
+	       "Please unmount and run xfs_repair.",
+	         pag->pag_agno, pag->pagf_flcount);
+
+	agf->agf_flfirst = 0;
+	agf->agf_fllast = cpu_to_be32(xfs_agfl_size(mp) - 1);
+	agf->agf_flcount = 0;
+	xfs_alloc_log_agf(tp, agbp, XFS_AGF_FLFIRST | XFS_AGF_FLLAST |
+				    XFS_AGF_FLCOUNT);
+
+	pag->pagf_flcount = 0;
+	pag->pagf_agflreset = false;
+}
+
+/*
+ * Defer an AGFL block free. This is effectively equivalent to
+ * xfs_free_extent_later() with some special handling particular to AGFL blocks.
+ *
+ * Deferring AGFL frees helps prevent log reservation overruns due to too many
+ * allocation operations in a transaction. AGFL frees are prone to this problem
+ * because for one they are always freed one at a time. Further, an immediate
+ * AGFL block free can cause a btree join and require another block free before
+ * the real allocation can proceed. Deferring the free disconnects freeing up
+ * the AGFL slot from freeing the block.
+ */
+STATIC void
+xfs_defer_agfl_block(
+	struct xfs_trans		*tp,
+	xfs_agnumber_t			agno,
+	xfs_fsblock_t			agbno,
+	struct xfs_owner_info		*oinfo)
+{
+	struct xfs_mount		*mp = tp->t_mountp;
+	struct xfs_extent_free_item	*new;		/* new element */
+
+	ASSERT(xfs_extfree_item_cache != NULL);
+	ASSERT(oinfo != NULL);
+
+	new = kmem_cache_zalloc(xfs_extfree_item_cache,
+			       GFP_KERNEL | __GFP_NOFAIL);
+	new->xefi_startblock = XFS_AGB_TO_FSB(mp, agno, agbno);
+	new->xefi_blockcount = 1;
+	new->xefi_owner = oinfo->oi_owner;
+
+	trace_xfs_agfl_free_defer(mp, agno, 0, agbno, 1);
+
+	xfs_defer_add(tp, XFS_DEFER_OPS_TYPE_AGFL_FREE, &new->xefi_list);
+}
+
+/*
+ * Add the extent to the list of extents to be free at transaction end.
+ * The list is maintained sorted (by block number).
+ */
+void
+__xfs_free_extent_later(
+	struct xfs_trans		*tp,
+	xfs_fsblock_t			bno,
+	xfs_filblks_t			len,
+	const struct xfs_owner_info	*oinfo,
+	bool				skip_discard)
+{
+	struct xfs_extent_free_item	*new;		/* new element */
+#ifdef DEBUG
+	struct xfs_mount		*mp = tp->t_mountp;
+	xfs_agnumber_t			agno;
+	xfs_agblock_t			agbno;
+
+	ASSERT(bno != NULLFSBLOCK);
+	ASSERT(len > 0);
+	ASSERT(len <= XFS_MAX_BMBT_EXTLEN);
+	ASSERT(!isnullstartblock(bno));
+	agno = XFS_FSB_TO_AGNO(mp, bno);
+	agbno = XFS_FSB_TO_AGBNO(mp, bno);
+	ASSERT(agno < mp->m_sb.sb_agcount);
+	ASSERT(agbno < mp->m_sb.sb_agblocks);
+	ASSERT(len < mp->m_sb.sb_agblocks);
+	ASSERT(agbno + len <= mp->m_sb.sb_agblocks);
+#endif
+	ASSERT(xfs_extfree_item_cache != NULL);
+
+	new = kmem_cache_zalloc(xfs_extfree_item_cache,
+			       GFP_KERNEL | __GFP_NOFAIL);
+	new->xefi_startblock = bno;
+	new->xefi_blockcount = (xfs_extlen_t)len;
+	if (skip_discard)
+		new->xefi_flags |= XFS_EFI_SKIP_DISCARD;
+	if (oinfo) {
+		ASSERT(oinfo->oi_offset == 0);
+
+		if (oinfo->oi_flags & XFS_OWNER_INFO_ATTR_FORK)
+			new->xefi_flags |= XFS_EFI_ATTR_FORK;
+		if (oinfo->oi_flags & XFS_OWNER_INFO_BMBT_BLOCK)
+			new->xefi_flags |= XFS_EFI_BMBT_BLOCK;
+		new->xefi_owner = oinfo->oi_owner;
+	} else {
+		new->xefi_owner = XFS_RMAP_OWN_NULL;
+	}
+	trace_xfs_bmap_free_defer(tp->t_mountp,
+			XFS_FSB_TO_AGNO(tp->t_mountp, bno), 0,
+			XFS_FSB_TO_AGBNO(tp->t_mountp, bno), len);
+	xfs_defer_add(tp, XFS_DEFER_OPS_TYPE_FREE, &new->xefi_list);
+}
+
+#ifdef DEBUG
+/*
+ * Check if an AGF has a free extent record whose length is equal to
+ * args->minlen.
+ */
+STATIC int
+xfs_exact_minlen_extent_available(
+	struct xfs_alloc_arg	*args,
+	struct xfs_buf		*agbp,
+	int			*stat)
+{
+	struct xfs_btree_cur	*cnt_cur;
+	xfs_agblock_t		fbno;
+	xfs_extlen_t		flen;
+	int			error = 0;
+
+	cnt_cur = xfs_allocbt_init_cursor(args->mp, args->tp, agbp,
+					args->pag, XFS_BTNUM_CNT);
+	error = xfs_alloc_lookup_ge(cnt_cur, 0, args->minlen, stat);
+	if (error)
+		goto out;
+
+	if (*stat == 0) {
+		error = -EFSCORRUPTED;
+		goto out;
+	}
+
+	error = xfs_alloc_get_rec(cnt_cur, &fbno, &flen, stat);
+	if (error)
+		goto out;
+
+	if (*stat == 1 && flen != args->minlen)
+		*stat = 0;
+
+out:
+	xfs_btree_del_cursor(cnt_cur, error);
+
+	return error;
+}
+#endif
+
+/*
+ * Decide whether to use this allocation group for this allocation.
+ * If so, fix up the btree freelist's size.
+ */
+int			/* error */
+xfs_alloc_fix_freelist(
+	struct xfs_alloc_arg	*args,	/* allocation argument structure */
+	int			flags)	/* XFS_ALLOC_FLAG_... */
+{
+	struct xfs_mount	*mp = args->mp;
+	struct xfs_perag	*pag = args->pag;
+	struct xfs_trans	*tp = args->tp;
+	struct xfs_buf		*agbp = NULL;
+	struct xfs_buf		*agflbp = NULL;
+	struct xfs_alloc_arg	targs;	/* local allocation arguments */
+	xfs_agblock_t		bno;	/* freelist block */
+	xfs_extlen_t		need;	/* total blocks needed in freelist */
+	int			error = 0;
+
+	/* deferred ops (AGFL block frees) require permanent transactions */
+	ASSERT(tp->t_flags & XFS_TRANS_PERM_LOG_RES);
+
+	if (!pag->pagf_init) {
+		error = xfs_alloc_read_agf(pag, tp, flags, &agbp);
+		if (error) {
+			/* Couldn't lock the AGF so skip this AG. */
+			if (error == -EAGAIN)
+				error = 0;
+			goto out_no_agbp;
+		}
+	}
+
+	/*
+	 * If this is a metadata preferred pag and we are user data then try
+	 * somewhere else if we are not being asked to try harder at this
+	 * point
+	 */
+	if (pag->pagf_metadata && (args->datatype & XFS_ALLOC_USERDATA) &&
+	    (flags & XFS_ALLOC_FLAG_TRYLOCK)) {
+		ASSERT(!(flags & XFS_ALLOC_FLAG_FREEING));
+		goto out_agbp_relse;
+	}
+
+	need = xfs_alloc_min_freelist(mp, pag);
+	if (!xfs_alloc_space_available(args, need, flags |
+			XFS_ALLOC_FLAG_CHECK))
+		goto out_agbp_relse;
+
+	/*
+	 * Get the a.g. freespace buffer.
+	 * Can fail if we're not blocking on locks, and it's held.
+	 */
+	if (!agbp) {
+		error = xfs_alloc_read_agf(pag, tp, flags, &agbp);
+		if (error) {
+			/* Couldn't lock the AGF so skip this AG. */
+			if (error == -EAGAIN)
+				error = 0;
+			goto out_no_agbp;
+		}
+	}
+
+	/* reset a padding mismatched agfl before final free space check */
+	if (pag->pagf_agflreset)
+		xfs_agfl_reset(tp, agbp, pag);
+
+	/* If there isn't enough total space or single-extent, reject it. */
+	need = xfs_alloc_min_freelist(mp, pag);
+	if (!xfs_alloc_space_available(args, need, flags))
+		goto out_agbp_relse;
+
+#ifdef DEBUG
+	if (args->alloc_minlen_only) {
+		int stat;
+
+		error = xfs_exact_minlen_extent_available(args, agbp, &stat);
+		if (error || !stat)
+			goto out_agbp_relse;
+	}
+#endif
+	/*
+	 * Make the freelist shorter if it's too long.
+	 *
+	 * Note that from this point onwards, we will always release the agf and
+	 * agfl buffers on error. This handles the case where we error out and
+	 * the buffers are clean or may not have been joined to the transaction
+	 * and hence need to be released manually. If they have been joined to
+	 * the transaction, then xfs_trans_brelse() will handle them
+	 * appropriately based on the recursion count and dirty state of the
+	 * buffer.
+	 *
+	 * XXX (dgc): When we have lots of free space, does this buy us
+	 * anything other than extra overhead when we need to put more blocks
+	 * back on the free list? Maybe we should only do this when space is
+	 * getting low or the AGFL is more than half full?
+	 *
+	 * The NOSHRINK flag prevents the AGFL from being shrunk if it's too
+	 * big; the NORMAP flag prevents AGFL expand/shrink operations from
+	 * updating the rmapbt.  Both flags are used in xfs_repair while we're
+	 * rebuilding the rmapbt, and neither are used by the kernel.  They're
+	 * both required to ensure that rmaps are correctly recorded for the
+	 * regenerated AGFL, bnobt, and cntbt.  See repair/phase5.c and
+	 * repair/rmap.c in xfsprogs for details.
+	 */
+	memset(&targs, 0, sizeof(targs));
+	/* struct copy below */
+	if (flags & XFS_ALLOC_FLAG_NORMAP)
+		targs.oinfo = XFS_RMAP_OINFO_SKIP_UPDATE;
+	else
+		targs.oinfo = XFS_RMAP_OINFO_AG;
+	while (!(flags & XFS_ALLOC_FLAG_NOSHRINK) && pag->pagf_flcount > need) {
+		error = xfs_alloc_get_freelist(pag, tp, agbp, &bno, 0);
+		if (error)
+			goto out_agbp_relse;
+
+		/* defer agfl frees */
+		xfs_defer_agfl_block(tp, args->agno, bno, &targs.oinfo);
+	}
+
+	targs.tp = tp;
+	targs.mp = mp;
+	targs.agbp = agbp;
+	targs.agno = args->agno;
+	targs.alignment = targs.minlen = targs.prod = 1;
+	targs.type = XFS_ALLOCTYPE_THIS_AG;
+	targs.pag = pag;
+	error = xfs_alloc_read_agfl(pag, tp, &agflbp);
+	if (error)
+		goto out_agbp_relse;
+
+	/* Make the freelist longer if it's too short. */
+	while (pag->pagf_flcount < need) {
+		targs.agbno = 0;
+		targs.maxlen = need - pag->pagf_flcount;
+		targs.resv = XFS_AG_RESV_AGFL;
+
+		/* Allocate as many blocks as possible at once. */
+		error = xfs_alloc_ag_vextent(&targs);
+		if (error)
+			goto out_agflbp_relse;
+
+		/*
+		 * Stop if we run out.  Won't happen if callers are obeying
+		 * the restrictions correctly.  Can happen for free calls
+		 * on a completely full ag.
+		 */
+		if (targs.agbno == NULLAGBLOCK) {
+			if (flags & XFS_ALLOC_FLAG_FREEING)
+				break;
+			goto out_agflbp_relse;
+		}
+		/*
+		 * Put each allocated block on the list.
+		 */
+		for (bno = targs.agbno; bno < targs.agbno + targs.len; bno++) {
+			error = xfs_alloc_put_freelist(pag, tp, agbp,
+							agflbp, bno, 0);
+			if (error)
+				goto out_agflbp_relse;
+		}
+	}
+	xfs_trans_brelse(tp, agflbp);
+	args->agbp = agbp;
+	return 0;
+
+out_agflbp_relse:
+	xfs_trans_brelse(tp, agflbp);
+out_agbp_relse:
+	if (agbp)
+		xfs_trans_brelse(tp, agbp);
+out_no_agbp:
+	args->agbp = NULL;
+	return error;
+}
+
+/*
+ * Get a block from the freelist.
+ * Returns with the buffer for the block gotten.
+ */
+int
+xfs_alloc_get_freelist(
+	struct xfs_perag	*pag,
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp,
+	xfs_agblock_t		*bnop,
+	int			btreeblk)
+{
+	struct xfs_agf		*agf = agbp->b_addr;
+	struct xfs_buf		*agflbp;
+	xfs_agblock_t		bno;
+	__be32			*agfl_bno;
+	int			error;
+	uint32_t		logflags;
+	struct xfs_mount	*mp = tp->t_mountp;
+
+	/*
+	 * Freelist is empty, give up.
+	 */
+	if (!agf->agf_flcount) {
+		*bnop = NULLAGBLOCK;
+		return 0;
+	}
+	/*
+	 * Read the array of free blocks.
+	 */
+	error = xfs_alloc_read_agfl(pag, tp, &agflbp);
+	if (error)
+		return error;
+
+
+	/*
+	 * Get the block number and update the data structures.
+	 */
+	agfl_bno = xfs_buf_to_agfl_bno(agflbp);
+	bno = be32_to_cpu(agfl_bno[be32_to_cpu(agf->agf_flfirst)]);
+	be32_add_cpu(&agf->agf_flfirst, 1);
+	xfs_trans_brelse(tp, agflbp);
+	if (be32_to_cpu(agf->agf_flfirst) == xfs_agfl_size(mp))
+		agf->agf_flfirst = 0;
+
+	ASSERT(!pag->pagf_agflreset);
+	be32_add_cpu(&agf->agf_flcount, -1);
+	pag->pagf_flcount--;
+
+	logflags = XFS_AGF_FLFIRST | XFS_AGF_FLCOUNT;
+	if (btreeblk) {
+		be32_add_cpu(&agf->agf_btreeblks, 1);
+		pag->pagf_btreeblks++;
+		logflags |= XFS_AGF_BTREEBLKS;
+	}
+
+	xfs_alloc_log_agf(tp, agbp, logflags);
+	*bnop = bno;
+
+	return 0;
+}
+
+/*
+ * Log the given fields from the agf structure.
+ */
+void
+xfs_alloc_log_agf(
+	struct xfs_trans	*tp,
+	struct xfs_buf		*bp,
+	uint32_t		fields)
+{
+	int	first;		/* first byte offset */
+	int	last;		/* last byte offset */
+	static const short	offsets[] = {
+		offsetof(xfs_agf_t, agf_magicnum),
+		offsetof(xfs_agf_t, agf_versionnum),
+		offsetof(xfs_agf_t, agf_seqno),
+		offsetof(xfs_agf_t, agf_length),
+		offsetof(xfs_agf_t, agf_roots[0]),
+		offsetof(xfs_agf_t, agf_levels[0]),
+		offsetof(xfs_agf_t, agf_flfirst),
+		offsetof(xfs_agf_t, agf_fllast),
+		offsetof(xfs_agf_t, agf_flcount),
+		offsetof(xfs_agf_t, agf_freeblks),
+		offsetof(xfs_agf_t, agf_longest),
+		offsetof(xfs_agf_t, agf_btreeblks),
+		offsetof(xfs_agf_t, agf_uuid),
+		offsetof(xfs_agf_t, agf_rmap_blocks),
+		offsetof(xfs_agf_t, agf_refcount_blocks),
+		offsetof(xfs_agf_t, agf_refcount_root),
+		offsetof(xfs_agf_t, agf_refcount_level),
+		/* needed so that we don't log the whole rest of the structure: */
+		offsetof(xfs_agf_t, agf_spare64),
+		sizeof(xfs_agf_t)
+	};
+
+	trace_xfs_agf(tp->t_mountp, bp->b_addr, fields, _RET_IP_);
+
+	xfs_trans_buf_set_type(tp, bp, XFS_BLFT_AGF_BUF);
+
+	xfs_btree_offsets(fields, offsets, XFS_AGF_NUM_BITS, &first, &last);
+	xfs_trans_log_buf(tp, bp, (uint)first, (uint)last);
+}
+
+/*
+ * Put the block on the freelist for the allocation group.
+ */
+int
+xfs_alloc_put_freelist(
+	struct xfs_perag	*pag,
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp,
+	struct xfs_buf		*agflbp,
+	xfs_agblock_t		bno,
+	int			btreeblk)
+{
+	struct xfs_mount	*mp = tp->t_mountp;
+	struct xfs_agf		*agf = agbp->b_addr;
+	__be32			*blockp;
+	int			error;
+	uint32_t		logflags;
+	__be32			*agfl_bno;
+	int			startoff;
+
+	if (!agflbp) {
+		error = xfs_alloc_read_agfl(pag, tp, &agflbp);
+		if (error)
+			return error;
+	}
+
+	be32_add_cpu(&agf->agf_fllast, 1);
+	if (be32_to_cpu(agf->agf_fllast) == xfs_agfl_size(mp))
+		agf->agf_fllast = 0;
+
+	ASSERT(!pag->pagf_agflreset);
+	be32_add_cpu(&agf->agf_flcount, 1);
+	pag->pagf_flcount++;
+
+	logflags = XFS_AGF_FLLAST | XFS_AGF_FLCOUNT;
+	if (btreeblk) {
+		be32_add_cpu(&agf->agf_btreeblks, -1);
+		pag->pagf_btreeblks--;
+		logflags |= XFS_AGF_BTREEBLKS;
+	}
+
+	xfs_alloc_log_agf(tp, agbp, logflags);
+
+	ASSERT(be32_to_cpu(agf->agf_flcount) <= xfs_agfl_size(mp));
+
+	agfl_bno = xfs_buf_to_agfl_bno(agflbp);
+	blockp = &agfl_bno[be32_to_cpu(agf->agf_fllast)];
+	*blockp = cpu_to_be32(bno);
+	startoff = (char *)blockp - (char *)agflbp->b_addr;
+
+	xfs_alloc_log_agf(tp, agbp, logflags);
+
+	xfs_trans_buf_set_type(tp, agflbp, XFS_BLFT_AGFL_BUF);
+	xfs_trans_log_buf(tp, agflbp, startoff,
+			  startoff + sizeof(xfs_agblock_t) - 1);
+	return 0;
+}
+
+static xfs_failaddr_t
+xfs_agf_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_agf		*agf = bp->b_addr;
+
+	if (xfs_has_crc(mp)) {
+		if (!uuid_equal(&agf->agf_uuid, &mp->m_sb.sb_meta_uuid))
+			return __this_address;
+		if (!xfs_log_check_lsn(mp, be64_to_cpu(agf->agf_lsn)))
+			return __this_address;
+	}
+
+	if (!xfs_verify_magic(bp, agf->agf_magicnum))
+		return __this_address;
+
+	if (!(XFS_AGF_GOOD_VERSION(be32_to_cpu(agf->agf_versionnum)) &&
+	      be32_to_cpu(agf->agf_freeblks) <= be32_to_cpu(agf->agf_length) &&
+	      be32_to_cpu(agf->agf_flfirst) < xfs_agfl_size(mp) &&
+	      be32_to_cpu(agf->agf_fllast) < xfs_agfl_size(mp) &&
+	      be32_to_cpu(agf->agf_flcount) <= xfs_agfl_size(mp)))
+		return __this_address;
+
+	if (be32_to_cpu(agf->agf_length) > mp->m_sb.sb_dblocks)
+		return __this_address;
+
+	if (be32_to_cpu(agf->agf_freeblks) < be32_to_cpu(agf->agf_longest) ||
+	    be32_to_cpu(agf->agf_freeblks) > be32_to_cpu(agf->agf_length))
+		return __this_address;
+
+	if (be32_to_cpu(agf->agf_levels[XFS_BTNUM_BNO]) < 1 ||
+	    be32_to_cpu(agf->agf_levels[XFS_BTNUM_CNT]) < 1 ||
+	    be32_to_cpu(agf->agf_levels[XFS_BTNUM_BNO]) >
+						mp->m_alloc_maxlevels ||
+	    be32_to_cpu(agf->agf_levels[XFS_BTNUM_CNT]) >
+						mp->m_alloc_maxlevels)
+		return __this_address;
+
+	if (xfs_has_rmapbt(mp) &&
+	    (be32_to_cpu(agf->agf_levels[XFS_BTNUM_RMAP]) < 1 ||
+	     be32_to_cpu(agf->agf_levels[XFS_BTNUM_RMAP]) >
+						mp->m_rmap_maxlevels))
+		return __this_address;
+
+	if (xfs_has_rmapbt(mp) &&
+	    be32_to_cpu(agf->agf_rmap_blocks) > be32_to_cpu(agf->agf_length))
+		return __this_address;
+
+	/*
+	 * during growfs operations, the perag is not fully initialised,
+	 * so we can't use it for any useful checking. growfs ensures we can't
+	 * use it by using uncached buffers that don't have the perag attached
+	 * so we can detect and avoid this problem.
+	 */
+	if (bp->b_pag && be32_to_cpu(agf->agf_seqno) != bp->b_pag->pag_agno)
+		return __this_address;
+
+	if (xfs_has_lazysbcount(mp) &&
+	    be32_to_cpu(agf->agf_btreeblks) > be32_to_cpu(agf->agf_length))
+		return __this_address;
+
+	if (xfs_has_reflink(mp) &&
+	    be32_to_cpu(agf->agf_refcount_blocks) >
+	    be32_to_cpu(agf->agf_length))
+		return __this_address;
+
+	if (xfs_has_reflink(mp) &&
+	    (be32_to_cpu(agf->agf_refcount_level) < 1 ||
+	     be32_to_cpu(agf->agf_refcount_level) > mp->m_refc_maxlevels))
+		return __this_address;
+
+	return NULL;
+
+}
+
+static void
+xfs_agf_read_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount *mp = bp->b_mount;
+	xfs_failaddr_t	fa;
+
+	if (xfs_has_crc(mp) &&
+	    !xfs_buf_verify_cksum(bp, XFS_AGF_CRC_OFF))
+		xfs_verifier_error(bp, -EFSBADCRC, __this_address);
+	else {
+		fa = xfs_agf_verify(bp);
+		if (XFS_TEST_ERROR(fa, mp, XFS_ERRTAG_ALLOC_READ_AGF))
+			xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+	}
+}
+
+static void
+xfs_agf_write_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_buf_log_item	*bip = bp->b_log_item;
+	struct xfs_agf		*agf = bp->b_addr;
+	xfs_failaddr_t		fa;
+
+	fa = xfs_agf_verify(bp);
+	if (fa) {
+		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+		return;
+	}
+
+	if (!xfs_has_crc(mp))
+		return;
+
+	if (bip)
+		agf->agf_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+	xfs_buf_update_cksum(bp, XFS_AGF_CRC_OFF);
+}
+
+const struct xfs_buf_ops xfs_agf_buf_ops = {
+	.name = "xfs_agf",
+	.magic = { cpu_to_be32(XFS_AGF_MAGIC), cpu_to_be32(XFS_AGF_MAGIC) },
+	.verify_read = xfs_agf_read_verify,
+	.verify_write = xfs_agf_write_verify,
+	.verify_struct = xfs_agf_verify,
+};
+
+/*
+ * Read in the allocation group header (free/alloc section).
+ */
+int
+xfs_read_agf(
+	struct xfs_perag	*pag,
+	struct xfs_trans	*tp,
+	int			flags,
+	struct xfs_buf		**agfbpp)
+{
+	struct xfs_mount	*mp = pag->pag_mount;
+	int			error;
+
+	trace_xfs_read_agf(pag->pag_mount, pag->pag_agno);
+
+	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
+			XFS_AG_DADDR(mp, pag->pag_agno, XFS_AGF_DADDR(mp)),
+			XFS_FSS_TO_BB(mp, 1), flags, agfbpp, &xfs_agf_buf_ops);
+	if (error)
+		return error;
+
+	xfs_buf_set_ref(*agfbpp, XFS_AGF_REF);
+	return 0;
+}
+
+/*
+ * Read in the allocation group header (free/alloc section) and initialise the
+ * perag structure if necessary. If the caller provides @agfbpp, then return the
+ * locked buffer to the caller, otherwise free it.
+ */
+int
+xfs_alloc_read_agf(
+	struct xfs_perag	*pag,
+	struct xfs_trans	*tp,
+	int			flags,
+	struct xfs_buf		**agfbpp)
+{
+	struct xfs_buf		*agfbp;
+	struct xfs_agf		*agf;
+	int			error;
+	int			allocbt_blks;
+
+	trace_xfs_alloc_read_agf(pag->pag_mount, pag->pag_agno);
+
+	/* We don't support trylock when freeing. */
+	ASSERT((flags & (XFS_ALLOC_FLAG_FREEING | XFS_ALLOC_FLAG_TRYLOCK)) !=
+			(XFS_ALLOC_FLAG_FREEING | XFS_ALLOC_FLAG_TRYLOCK));
+	error = xfs_read_agf(pag, tp,
+			(flags & XFS_ALLOC_FLAG_TRYLOCK) ? XBF_TRYLOCK : 0,
+			&agfbp);
+	if (error)
+		return error;
+
+	agf = agfbp->b_addr;
+	if (!pag->pagf_init) {
+		pag->pagf_freeblks = be32_to_cpu(agf->agf_freeblks);
+		pag->pagf_btreeblks = be32_to_cpu(agf->agf_btreeblks);
+		pag->pagf_flcount = be32_to_cpu(agf->agf_flcount);
+		pag->pagf_longest = be32_to_cpu(agf->agf_longest);
+		pag->pagf_levels[XFS_BTNUM_BNOi] =
+			be32_to_cpu(agf->agf_levels[XFS_BTNUM_BNOi]);
+		pag->pagf_levels[XFS_BTNUM_CNTi] =
+			be32_to_cpu(agf->agf_levels[XFS_BTNUM_CNTi]);
+		pag->pagf_levels[XFS_BTNUM_RMAPi] =
+			be32_to_cpu(agf->agf_levels[XFS_BTNUM_RMAPi]);
+		pag->pagf_refcount_level = be32_to_cpu(agf->agf_refcount_level);
+		pag->pagf_init = 1;
+		pag->pagf_agflreset = xfs_agfl_needs_reset(pag->pag_mount, agf);
+
+		/*
+		 * Update the in-core allocbt counter. Filter out the rmapbt
+		 * subset of the btreeblks counter because the rmapbt is managed
+		 * by perag reservation. Subtract one for the rmapbt root block
+		 * because the rmap counter includes it while the btreeblks
+		 * counter only tracks non-root blocks.
+		 */
+		allocbt_blks = pag->pagf_btreeblks;
+		if (xfs_has_rmapbt(pag->pag_mount))
+			allocbt_blks -= be32_to_cpu(agf->agf_rmap_blocks) - 1;
+		if (allocbt_blks > 0)
+			atomic64_add(allocbt_blks,
+					&pag->pag_mount->m_allocbt_blks);
+	}
+#ifdef DEBUG
+	else if (!xfs_is_shutdown(pag->pag_mount)) {
+		ASSERT(pag->pagf_freeblks == be32_to_cpu(agf->agf_freeblks));
+		ASSERT(pag->pagf_btreeblks == be32_to_cpu(agf->agf_btreeblks));
+		ASSERT(pag->pagf_flcount == be32_to_cpu(agf->agf_flcount));
+		ASSERT(pag->pagf_longest == be32_to_cpu(agf->agf_longest));
+		ASSERT(pag->pagf_levels[XFS_BTNUM_BNOi] ==
+		       be32_to_cpu(agf->agf_levels[XFS_BTNUM_BNOi]));
+		ASSERT(pag->pagf_levels[XFS_BTNUM_CNTi] ==
+		       be32_to_cpu(agf->agf_levels[XFS_BTNUM_CNTi]));
+	}
+#endif
+	if (agfbpp)
+		*agfbpp = agfbp;
+	else
+		xfs_trans_brelse(tp, agfbp);
+	return 0;
+}
+
+/*
+ * Allocate an extent (variable-size).
+ * Depending on the allocation type, we either look in a single allocation
+ * group or loop over the allocation groups to find the result.
+ */
+int				/* error */
+xfs_alloc_vextent(
+	struct xfs_alloc_arg	*args)	/* allocation argument structure */
+{
+	xfs_agblock_t		agsize;	/* allocation group size */
+	int			error;
+	int			flags;	/* XFS_ALLOC_FLAG_... locking flags */
+	struct xfs_mount	*mp;	/* mount structure pointer */
+	xfs_agnumber_t		sagno;	/* starting allocation group number */
+	xfs_alloctype_t		type;	/* input allocation type */
+	int			bump_rotor = 0;
+	xfs_agnumber_t		rotorstep = xfs_rotorstep; /* inode32 agf stepper */
+
+	mp = args->mp;
+	type = args->otype = args->type;
+	args->agbno = NULLAGBLOCK;
+	/*
+	 * Just fix this up, for the case where the last a.g. is shorter
+	 * (or there's only one a.g.) and the caller couldn't easily figure
+	 * that out (xfs_bmap_alloc).
+	 */
+	agsize = mp->m_sb.sb_agblocks;
+	if (args->maxlen > agsize)
+		args->maxlen = agsize;
+	if (args->alignment == 0)
+		args->alignment = 1;
+	ASSERT(XFS_FSB_TO_AGNO(mp, args->fsbno) < mp->m_sb.sb_agcount);
+	ASSERT(XFS_FSB_TO_AGBNO(mp, args->fsbno) < agsize);
+	ASSERT(args->minlen <= args->maxlen);
+	ASSERT(args->minlen <= agsize);
+	ASSERT(args->mod < args->prod);
+	if (XFS_FSB_TO_AGNO(mp, args->fsbno) >= mp->m_sb.sb_agcount ||
+	    XFS_FSB_TO_AGBNO(mp, args->fsbno) >= agsize ||
+	    args->minlen > args->maxlen || args->minlen > agsize ||
+	    args->mod >= args->prod) {
+		args->fsbno = NULLFSBLOCK;
+		trace_xfs_alloc_vextent_badargs(args);
+		return 0;
+	}
+
+	switch (type) {
+	case XFS_ALLOCTYPE_THIS_AG:
+	case XFS_ALLOCTYPE_NEAR_BNO:
+	case XFS_ALLOCTYPE_THIS_BNO:
+		/*
+		 * These three force us into a single a.g.
+		 */
+		args->agno = XFS_FSB_TO_AGNO(mp, args->fsbno);
+		args->pag = xfs_perag_get(mp, args->agno);
+		error = xfs_alloc_fix_freelist(args, 0);
+		if (error) {
+			trace_xfs_alloc_vextent_nofix(args);
+			goto error0;
+		}
+		if (!args->agbp) {
+			trace_xfs_alloc_vextent_noagbp(args);
+			break;
+		}
+		args->agbno = XFS_FSB_TO_AGBNO(mp, args->fsbno);
+		if ((error = xfs_alloc_ag_vextent(args)))
+			goto error0;
+		break;
+	case XFS_ALLOCTYPE_START_BNO:
+		/*
+		 * Try near allocation first, then anywhere-in-ag after
+		 * the first a.g. fails.
+		 */
+		if ((args->datatype & XFS_ALLOC_INITIAL_USER_DATA) &&
+		    xfs_is_inode32(mp)) {
+			args->fsbno = XFS_AGB_TO_FSB(mp,
+					((mp->m_agfrotor / rotorstep) %
+					mp->m_sb.sb_agcount), 0);
+			bump_rotor = 1;
+		}
+		args->agbno = XFS_FSB_TO_AGBNO(mp, args->fsbno);
+		args->type = XFS_ALLOCTYPE_NEAR_BNO;
+		fallthrough;
+	case XFS_ALLOCTYPE_FIRST_AG:
+		/*
+		 * Rotate through the allocation groups looking for a winner.
+		 */
+		if (type == XFS_ALLOCTYPE_FIRST_AG) {
+			/*
+			 * Start with allocation group given by bno.
+			 */
+			args->agno = XFS_FSB_TO_AGNO(mp, args->fsbno);
+			args->type = XFS_ALLOCTYPE_THIS_AG;
+			sagno = 0;
+			flags = 0;
+		} else {
+			/*
+			 * Start with the given allocation group.
+			 */
+			args->agno = sagno = XFS_FSB_TO_AGNO(mp, args->fsbno);
+			flags = XFS_ALLOC_FLAG_TRYLOCK;
+		}
+		/*
+		 * Loop over allocation groups twice; first time with
+		 * trylock set, second time without.
+		 */
+		for (;;) {
+			args->pag = xfs_perag_get(mp, args->agno);
+			error = xfs_alloc_fix_freelist(args, flags);
+			if (error) {
+				trace_xfs_alloc_vextent_nofix(args);
+				goto error0;
+			}
+			/*
+			 * If we get a buffer back then the allocation will fly.
+			 */
+			if (args->agbp) {
+				if ((error = xfs_alloc_ag_vextent(args)))
+					goto error0;
+				break;
+			}
+
+			trace_xfs_alloc_vextent_loopfailed(args);
+
+			/*
+			 * Didn't work, figure out the next iteration.
+			 */
+			if (args->agno == sagno &&
+			    type == XFS_ALLOCTYPE_START_BNO)
+				args->type = XFS_ALLOCTYPE_THIS_AG;
+			/*
+			* For the first allocation, we can try any AG to get
+			* space.  However, if we already have allocated a
+			* block, we don't want to try AGs whose number is below
+			* sagno. Otherwise, we may end up with out-of-order
+			* locking of AGF, which might cause deadlock.
+			*/
+			if (++(args->agno) == mp->m_sb.sb_agcount) {
+				if (args->tp->t_firstblock != NULLFSBLOCK)
+					args->agno = sagno;
+				else
+					args->agno = 0;
+			}
+			/*
+			 * Reached the starting a.g., must either be done
+			 * or switch to non-trylock mode.
+			 */
+			if (args->agno == sagno) {
+				if (flags == 0) {
+					args->agbno = NULLAGBLOCK;
+					trace_xfs_alloc_vextent_allfailed(args);
+					break;
+				}
+
+				flags = 0;
+				if (type == XFS_ALLOCTYPE_START_BNO) {
+					args->agbno = XFS_FSB_TO_AGBNO(mp,
+						args->fsbno);
+					args->type = XFS_ALLOCTYPE_NEAR_BNO;
+				}
+			}
+			xfs_perag_put(args->pag);
+		}
+		if (bump_rotor) {
+			if (args->agno == sagno)
+				mp->m_agfrotor = (mp->m_agfrotor + 1) %
+					(mp->m_sb.sb_agcount * rotorstep);
+			else
+				mp->m_agfrotor = (args->agno * rotorstep + 1) %
+					(mp->m_sb.sb_agcount * rotorstep);
+		}
+		break;
+	default:
+		ASSERT(0);
+		/* NOTREACHED */
+	}
+	if (args->agbno == NULLAGBLOCK)
+		args->fsbno = NULLFSBLOCK;
+	else {
+		args->fsbno = XFS_AGB_TO_FSB(mp, args->agno, args->agbno);
+#ifdef DEBUG
+		ASSERT(args->len >= args->minlen);
+		ASSERT(args->len <= args->maxlen);
+		ASSERT(args->agbno % args->alignment == 0);
+		XFS_AG_CHECK_DADDR(mp, XFS_FSB_TO_DADDR(mp, args->fsbno),
+			args->len);
+#endif
+
+	}
+	xfs_perag_put(args->pag);
+	return 0;
+error0:
+	xfs_perag_put(args->pag);
+	return error;
+}
+
+/* Ensure that the freelist is at full capacity. */
+int
+xfs_free_extent_fix_freelist(
+	struct xfs_trans	*tp,
+	struct xfs_perag	*pag,
+	struct xfs_buf		**agbp)
+{
+	struct xfs_alloc_arg	args;
+	int			error;
+
+	memset(&args, 0, sizeof(struct xfs_alloc_arg));
+	args.tp = tp;
+	args.mp = tp->t_mountp;
+	args.agno = pag->pag_agno;
+	args.pag = pag;
+
+	/*
+	 * validate that the block number is legal - the enables us to detect
+	 * and handle a silent filesystem corruption rather than crashing.
+	 */
+	if (args.agno >= args.mp->m_sb.sb_agcount)
+		return -EFSCORRUPTED;
+
+	error = xfs_alloc_fix_freelist(&args, XFS_ALLOC_FLAG_FREEING);
+	if (error)
+		return error;
+
+	*agbp = args.agbp;
+	return 0;
+}
+
+/*
+ * Free an extent.
+ * Just break up the extent address and hand off to xfs_free_ag_extent
+ * after fixing up the freelist.
+ */
+int
+__xfs_free_extent(
+	struct xfs_trans		*tp,
+	xfs_fsblock_t			bno,
+	xfs_extlen_t			len,
+	const struct xfs_owner_info	*oinfo,
+	enum xfs_ag_resv_type		type,
+	bool				skip_discard)
+{
+	struct xfs_mount		*mp = tp->t_mountp;
+	struct xfs_buf			*agbp;
+	xfs_agnumber_t			agno = XFS_FSB_TO_AGNO(mp, bno);
+	xfs_agblock_t			agbno = XFS_FSB_TO_AGBNO(mp, bno);
+	struct xfs_agf			*agf;
+	int				error;
+	unsigned int			busy_flags = 0;
+	struct xfs_perag		*pag;
+
+	ASSERT(len != 0);
+	ASSERT(type != XFS_AG_RESV_AGFL);
+
+	if (XFS_TEST_ERROR(false, mp,
+			XFS_ERRTAG_FREE_EXTENT))
+		return -EIO;
+
+	pag = xfs_perag_get(mp, agno);
+	error = xfs_free_extent_fix_freelist(tp, pag, &agbp);
+	if (error)
+		goto err;
+	agf = agbp->b_addr;
+
+	if (XFS_IS_CORRUPT(mp, agbno >= mp->m_sb.sb_agblocks)) {
+		error = -EFSCORRUPTED;
+		goto err_release;
+	}
+
+	/* validate the extent size is legal now we have the agf locked */
+	if (XFS_IS_CORRUPT(mp, agbno + len > be32_to_cpu(agf->agf_length))) {
+		error = -EFSCORRUPTED;
+		goto err_release;
+	}
+
+	error = xfs_free_ag_extent(tp, agbp, agno, agbno, len, oinfo, type);
+	if (error)
+		goto err_release;
+
+	if (skip_discard)
+		busy_flags |= XFS_EXTENT_BUSY_SKIP_DISCARD;
+	xfs_extent_busy_insert(tp, pag, agbno, len, busy_flags);
+	xfs_perag_put(pag);
+	return 0;
+
+err_release:
+	xfs_trans_brelse(tp, agbp);
+err:
+	xfs_perag_put(pag);
+	return error;
+}
+
+struct xfs_alloc_query_range_info {
+	xfs_alloc_query_range_fn	fn;
+	void				*priv;
+};
+
+/* Format btree record and pass to our callback. */
+STATIC int
+xfs_alloc_query_range_helper(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_rec	*rec,
+	void				*priv)
+{
+	struct xfs_alloc_query_range_info	*query = priv;
+	struct xfs_alloc_rec_incore		irec;
+
+	irec.ar_startblock = be32_to_cpu(rec->alloc.ar_startblock);
+	irec.ar_blockcount = be32_to_cpu(rec->alloc.ar_blockcount);
+	return query->fn(cur, &irec, query->priv);
+}
+
+/* Find all free space within a given range of blocks. */
+int
+xfs_alloc_query_range(
+	struct xfs_btree_cur			*cur,
+	const struct xfs_alloc_rec_incore	*low_rec,
+	const struct xfs_alloc_rec_incore	*high_rec,
+	xfs_alloc_query_range_fn		fn,
+	void					*priv)
+{
+	union xfs_btree_irec			low_brec;
+	union xfs_btree_irec			high_brec;
+	struct xfs_alloc_query_range_info	query;
+
+	ASSERT(cur->bc_btnum == XFS_BTNUM_BNO);
+	low_brec.a = *low_rec;
+	high_brec.a = *high_rec;
+	query.priv = priv;
+	query.fn = fn;
+	return xfs_btree_query_range(cur, &low_brec, &high_brec,
+			xfs_alloc_query_range_helper, &query);
+}
+
+/* Find all free space records. */
+int
+xfs_alloc_query_all(
+	struct xfs_btree_cur			*cur,
+	xfs_alloc_query_range_fn		fn,
+	void					*priv)
+{
+	struct xfs_alloc_query_range_info	query;
+
+	ASSERT(cur->bc_btnum == XFS_BTNUM_BNO);
+	query.priv = priv;
+	query.fn = fn;
+	return xfs_btree_query_all(cur, xfs_alloc_query_range_helper, &query);
+}
+
+/* Is there a record covering a given extent? */
+int
+xfs_alloc_has_record(
+	struct xfs_btree_cur	*cur,
+	xfs_agblock_t		bno,
+	xfs_extlen_t		len,
+	bool			*exists)
+{
+	union xfs_btree_irec	low;
+	union xfs_btree_irec	high;
+
+	memset(&low, 0, sizeof(low));
+	low.a.ar_startblock = bno;
+	memset(&high, 0xFF, sizeof(high));
+	high.a.ar_startblock = bno + len - 1;
+
+	return xfs_btree_has_record(cur, &low, &high, exists);
+}
+
+/*
+ * Walk all the blocks in the AGFL.  The @walk_fn can return any negative
+ * error code or XFS_ITER_*.
+ */
+int
+xfs_agfl_walk(
+	struct xfs_mount	*mp,
+	struct xfs_agf		*agf,
+	struct xfs_buf		*agflbp,
+	xfs_agfl_walk_fn	walk_fn,
+	void			*priv)
+{
+	__be32			*agfl_bno;
+	unsigned int		i;
+	int			error;
+
+	agfl_bno = xfs_buf_to_agfl_bno(agflbp);
+	i = be32_to_cpu(agf->agf_flfirst);
+
+	/* Nothing to walk in an empty AGFL. */
+	if (agf->agf_flcount == cpu_to_be32(0))
+		return 0;
+
+	/* Otherwise, walk from first to last, wrapping as needed. */
+	for (;;) {
+		error = walk_fn(mp, be32_to_cpu(agfl_bno[i]), priv);
+		if (error)
+			return error;
+		if (i == be32_to_cpu(agf->agf_fllast))
+			break;
+		if (++i == xfs_agfl_size(mp))
+			i = 0;
+	}
+
+	return 0;
+}
+
+int __init
+xfs_extfree_intent_init_cache(void)
+{
+	xfs_extfree_item_cache = kmem_cache_create("xfs_extfree_intent",
+			sizeof(struct xfs_extent_free_item),
+			0, 0, NULL);
+
+	return xfs_extfree_item_cache != NULL ? 0 : -ENOMEM;
+}
+
+void
+xfs_extfree_intent_destroy_cache(void)
+{
+	kmem_cache_destroy(xfs_extfree_item_cache);
+	xfs_extfree_item_cache = NULL;
+}
diff --git a/fs/xfs/libxfs/xfs_alloc.h b/fs/xfs/libxfs/xfs_alloc.h
new file mode 100644
index 000000000..2c3f762df
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_alloc.h
@@ -0,0 +1,250 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_ALLOC_H__
+#define	__XFS_ALLOC_H__
+
+struct xfs_buf;
+struct xfs_btree_cur;
+struct xfs_mount;
+struct xfs_perag;
+struct xfs_trans;
+
+extern struct workqueue_struct *xfs_alloc_wq;
+
+unsigned int xfs_agfl_size(struct xfs_mount *mp);
+
+/*
+ * Freespace allocation types.  Argument to xfs_alloc_[v]extent.
+ */
+#define XFS_ALLOCTYPE_FIRST_AG	0x02	/* ... start at ag 0 */
+#define XFS_ALLOCTYPE_THIS_AG	0x08	/* anywhere in this a.g. */
+#define XFS_ALLOCTYPE_START_BNO	0x10	/* near this block else anywhere */
+#define XFS_ALLOCTYPE_NEAR_BNO	0x20	/* in this a.g. and near this block */
+#define XFS_ALLOCTYPE_THIS_BNO	0x40	/* at exactly this block */
+
+/* this should become an enum again when the tracing code is fixed */
+typedef unsigned int xfs_alloctype_t;
+
+#define XFS_ALLOC_TYPES \
+	{ XFS_ALLOCTYPE_FIRST_AG,	"FIRST_AG" }, \
+	{ XFS_ALLOCTYPE_THIS_AG,	"THIS_AG" }, \
+	{ XFS_ALLOCTYPE_START_BNO,	"START_BNO" }, \
+	{ XFS_ALLOCTYPE_NEAR_BNO,	"NEAR_BNO" }, \
+	{ XFS_ALLOCTYPE_THIS_BNO,	"THIS_BNO" }
+
+/*
+ * Flags for xfs_alloc_fix_freelist.
+ */
+#define	XFS_ALLOC_FLAG_TRYLOCK	0x00000001  /* use trylock for buffer locking */
+#define	XFS_ALLOC_FLAG_FREEING	0x00000002  /* indicate caller is freeing extents*/
+#define	XFS_ALLOC_FLAG_NORMAP	0x00000004  /* don't modify the rmapbt */
+#define	XFS_ALLOC_FLAG_NOSHRINK	0x00000008  /* don't shrink the freelist */
+#define	XFS_ALLOC_FLAG_CHECK	0x00000010  /* test only, don't modify args */
+
+/*
+ * Argument structure for xfs_alloc routines.
+ * This is turned into a structure to avoid having 20 arguments passed
+ * down several levels of the stack.
+ */
+typedef struct xfs_alloc_arg {
+	struct xfs_trans *tp;		/* transaction pointer */
+	struct xfs_mount *mp;		/* file system mount point */
+	struct xfs_buf	*agbp;		/* buffer for a.g. freelist header */
+	struct xfs_perag *pag;		/* per-ag struct for this agno */
+	xfs_fsblock_t	fsbno;		/* file system block number */
+	xfs_agnumber_t	agno;		/* allocation group number */
+	xfs_agblock_t	agbno;		/* allocation group-relative block # */
+	xfs_extlen_t	minlen;		/* minimum size of extent */
+	xfs_extlen_t	maxlen;		/* maximum size of extent */
+	xfs_extlen_t	mod;		/* mod value for extent size */
+	xfs_extlen_t	prod;		/* prod value for extent size */
+	xfs_extlen_t	minleft;	/* min blocks must be left after us */
+	xfs_extlen_t	total;		/* total blocks needed in xaction */
+	xfs_extlen_t	alignment;	/* align answer to multiple of this */
+	xfs_extlen_t	minalignslop;	/* slop for minlen+alignment calcs */
+	xfs_agblock_t	min_agbno;	/* set an agbno range for NEAR allocs */
+	xfs_agblock_t	max_agbno;	/* ... */
+	xfs_extlen_t	len;		/* output: actual size of extent */
+	xfs_alloctype_t	type;		/* allocation type XFS_ALLOCTYPE_... */
+	xfs_alloctype_t	otype;		/* original allocation type */
+	int		datatype;	/* mask defining data type treatment */
+	char		wasdel;		/* set if allocation was prev delayed */
+	char		wasfromfl;	/* set if allocation is from freelist */
+	struct xfs_owner_info	oinfo;	/* owner of blocks being allocated */
+	enum xfs_ag_resv_type	resv;	/* block reservation to use */
+#ifdef DEBUG
+	bool		alloc_minlen_only; /* allocate exact minlen extent */
+#endif
+} xfs_alloc_arg_t;
+
+/*
+ * Defines for datatype
+ */
+#define XFS_ALLOC_USERDATA		(1 << 0)/* allocation is for user data*/
+#define XFS_ALLOC_INITIAL_USER_DATA	(1 << 1)/* special case start of file */
+#define XFS_ALLOC_NOBUSY		(1 << 2)/* Busy extents not allowed */
+
+/* freespace limit calculations */
+unsigned int xfs_alloc_set_aside(struct xfs_mount *mp);
+unsigned int xfs_alloc_ag_max_usable(struct xfs_mount *mp);
+
+xfs_extlen_t xfs_alloc_longest_free_extent(struct xfs_perag *pag,
+		xfs_extlen_t need, xfs_extlen_t reserved);
+unsigned int xfs_alloc_min_freelist(struct xfs_mount *mp,
+		struct xfs_perag *pag);
+int xfs_alloc_get_freelist(struct xfs_perag *pag, struct xfs_trans *tp,
+		struct xfs_buf *agfbp, xfs_agblock_t *bnop, int	 btreeblk);
+int xfs_alloc_put_freelist(struct xfs_perag *pag, struct xfs_trans *tp,
+		struct xfs_buf *agfbp, struct xfs_buf *agflbp,
+		xfs_agblock_t bno, int btreeblk);
+
+/*
+ * Compute and fill in value of m_alloc_maxlevels.
+ */
+void
+xfs_alloc_compute_maxlevels(
+	struct xfs_mount	*mp);	/* file system mount structure */
+
+/*
+ * Log the given fields from the agf structure.
+ */
+void
+xfs_alloc_log_agf(
+	struct xfs_trans *tp,	/* transaction pointer */
+	struct xfs_buf	*bp,	/* buffer for a.g. freelist header */
+	uint32_t	fields);/* mask of fields to be logged (XFS_AGF_...) */
+
+/*
+ * Allocate an extent (variable-size).
+ */
+int				/* error */
+xfs_alloc_vextent(
+	xfs_alloc_arg_t	*args);	/* allocation argument structure */
+
+/*
+ * Free an extent.
+ */
+int				/* error */
+__xfs_free_extent(
+	struct xfs_trans	*tp,	/* transaction pointer */
+	xfs_fsblock_t		bno,	/* starting block number of extent */
+	xfs_extlen_t		len,	/* length of extent */
+	const struct xfs_owner_info	*oinfo,	/* extent owner */
+	enum xfs_ag_resv_type	type,	/* block reservation type */
+	bool			skip_discard);
+
+static inline int
+xfs_free_extent(
+	struct xfs_trans	*tp,
+	xfs_fsblock_t		bno,
+	xfs_extlen_t		len,
+	const struct xfs_owner_info	*oinfo,
+	enum xfs_ag_resv_type	type)
+{
+	return __xfs_free_extent(tp, bno, len, oinfo, type, false);
+}
+
+int				/* error */
+xfs_alloc_lookup_le(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* starting block of extent */
+	xfs_extlen_t		len,	/* length of extent */
+	int			*stat);	/* success/failure */
+
+int				/* error */
+xfs_alloc_lookup_ge(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* starting block of extent */
+	xfs_extlen_t		len,	/* length of extent */
+	int			*stat);	/* success/failure */
+
+int					/* error */
+xfs_alloc_get_rec(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		*bno,	/* output: starting block of extent */
+	xfs_extlen_t		*len,	/* output: length of extent */
+	int			*stat);	/* output: success/failure */
+
+int xfs_read_agf(struct xfs_perag *pag, struct xfs_trans *tp, int flags,
+		struct xfs_buf **agfbpp);
+int xfs_alloc_read_agf(struct xfs_perag *pag, struct xfs_trans *tp, int flags,
+		struct xfs_buf **agfbpp);
+int xfs_alloc_read_agfl(struct xfs_perag *pag, struct xfs_trans *tp,
+		struct xfs_buf **bpp);
+int xfs_free_agfl_block(struct xfs_trans *, xfs_agnumber_t, xfs_agblock_t,
+			struct xfs_buf *, struct xfs_owner_info *);
+int xfs_alloc_fix_freelist(struct xfs_alloc_arg *args, int flags);
+int xfs_free_extent_fix_freelist(struct xfs_trans *tp, struct xfs_perag *pag,
+		struct xfs_buf **agbp);
+
+xfs_extlen_t xfs_prealloc_blocks(struct xfs_mount *mp);
+
+typedef int (*xfs_alloc_query_range_fn)(
+	struct xfs_btree_cur			*cur,
+	const struct xfs_alloc_rec_incore	*rec,
+	void					*priv);
+
+int xfs_alloc_query_range(struct xfs_btree_cur *cur,
+		const struct xfs_alloc_rec_incore *low_rec,
+		const struct xfs_alloc_rec_incore *high_rec,
+		xfs_alloc_query_range_fn fn, void *priv);
+int xfs_alloc_query_all(struct xfs_btree_cur *cur, xfs_alloc_query_range_fn fn,
+		void *priv);
+
+int xfs_alloc_has_record(struct xfs_btree_cur *cur, xfs_agblock_t bno,
+		xfs_extlen_t len, bool *exist);
+
+typedef int (*xfs_agfl_walk_fn)(struct xfs_mount *mp, xfs_agblock_t bno,
+		void *priv);
+int xfs_agfl_walk(struct xfs_mount *mp, struct xfs_agf *agf,
+		struct xfs_buf *agflbp, xfs_agfl_walk_fn walk_fn, void *priv);
+
+static inline __be32 *
+xfs_buf_to_agfl_bno(
+	struct xfs_buf		*bp)
+{
+	if (xfs_has_crc(bp->b_mount))
+		return bp->b_addr + sizeof(struct xfs_agfl);
+	return bp->b_addr;
+}
+
+void __xfs_free_extent_later(struct xfs_trans *tp, xfs_fsblock_t bno,
+		xfs_filblks_t len, const struct xfs_owner_info *oinfo,
+		bool skip_discard);
+
+/*
+ * List of extents to be free "later".
+ * The list is kept sorted on xbf_startblock.
+ */
+struct xfs_extent_free_item {
+	struct list_head	xefi_list;
+	uint64_t		xefi_owner;
+	xfs_fsblock_t		xefi_startblock;/* starting fs block number */
+	xfs_extlen_t		xefi_blockcount;/* number of blocks in extent */
+	unsigned int		xefi_flags;
+};
+
+#define XFS_EFI_SKIP_DISCARD	(1U << 0) /* don't issue discard */
+#define XFS_EFI_ATTR_FORK	(1U << 1) /* freeing attr fork block */
+#define XFS_EFI_BMBT_BLOCK	(1U << 2) /* freeing bmap btree block */
+
+static inline void
+xfs_free_extent_later(
+	struct xfs_trans		*tp,
+	xfs_fsblock_t			bno,
+	xfs_filblks_t			len,
+	const struct xfs_owner_info	*oinfo)
+{
+	__xfs_free_extent_later(tp, bno, len, oinfo, false);
+}
+
+
+extern struct kmem_cache	*xfs_extfree_item_cache;
+
+int __init xfs_extfree_intent_init_cache(void);
+void xfs_extfree_intent_destroy_cache(void);
+
+#endif	/* __XFS_ALLOC_H__ */
diff --git a/fs/xfs/libxfs/xfs_alloc_btree.c b/fs/xfs/libxfs/xfs_alloc_btree.c
new file mode 100644
index 000000000..549a3cba0
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_alloc_btree.c
@@ -0,0 +1,641 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_btree.h"
+#include "xfs_btree_staging.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_alloc.h"
+#include "xfs_extent_busy.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+#include "xfs_trans.h"
+#include "xfs_ag.h"
+
+static struct kmem_cache	*xfs_allocbt_cur_cache;
+
+STATIC struct xfs_btree_cur *
+xfs_allocbt_dup_cursor(
+	struct xfs_btree_cur	*cur)
+{
+	return xfs_allocbt_init_cursor(cur->bc_mp, cur->bc_tp,
+			cur->bc_ag.agbp, cur->bc_ag.pag, cur->bc_btnum);
+}
+
+STATIC void
+xfs_allocbt_set_root(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*ptr,
+	int				inc)
+{
+	struct xfs_buf		*agbp = cur->bc_ag.agbp;
+	struct xfs_agf		*agf = agbp->b_addr;
+	int			btnum = cur->bc_btnum;
+
+	ASSERT(ptr->s != 0);
+
+	agf->agf_roots[btnum] = ptr->s;
+	be32_add_cpu(&agf->agf_levels[btnum], inc);
+	cur->bc_ag.pag->pagf_levels[btnum] += inc;
+
+	xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_ROOTS | XFS_AGF_LEVELS);
+}
+
+STATIC int
+xfs_allocbt_alloc_block(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*start,
+	union xfs_btree_ptr		*new,
+	int				*stat)
+{
+	int			error;
+	xfs_agblock_t		bno;
+
+	/* Allocate the new block from the freelist. If we can't, give up.  */
+	error = xfs_alloc_get_freelist(cur->bc_ag.pag, cur->bc_tp,
+			cur->bc_ag.agbp, &bno, 1);
+	if (error)
+		return error;
+
+	if (bno == NULLAGBLOCK) {
+		*stat = 0;
+		return 0;
+	}
+
+	atomic64_inc(&cur->bc_mp->m_allocbt_blks);
+	xfs_extent_busy_reuse(cur->bc_mp, cur->bc_ag.pag, bno, 1, false);
+
+	new->s = cpu_to_be32(bno);
+
+	*stat = 1;
+	return 0;
+}
+
+STATIC int
+xfs_allocbt_free_block(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp)
+{
+	struct xfs_buf		*agbp = cur->bc_ag.agbp;
+	xfs_agblock_t		bno;
+	int			error;
+
+	bno = xfs_daddr_to_agbno(cur->bc_mp, xfs_buf_daddr(bp));
+	error = xfs_alloc_put_freelist(cur->bc_ag.pag, cur->bc_tp, agbp, NULL,
+			bno, 1);
+	if (error)
+		return error;
+
+	atomic64_dec(&cur->bc_mp->m_allocbt_blks);
+	xfs_extent_busy_insert(cur->bc_tp, agbp->b_pag, bno, 1,
+			      XFS_EXTENT_BUSY_SKIP_DISCARD);
+	return 0;
+}
+
+/*
+ * Update the longest extent in the AGF
+ */
+STATIC void
+xfs_allocbt_update_lastrec(
+	struct xfs_btree_cur		*cur,
+	const struct xfs_btree_block	*block,
+	const union xfs_btree_rec	*rec,
+	int				ptr,
+	int				reason)
+{
+	struct xfs_agf		*agf = cur->bc_ag.agbp->b_addr;
+	struct xfs_perag	*pag;
+	__be32			len;
+	int			numrecs;
+
+	ASSERT(cur->bc_btnum == XFS_BTNUM_CNT);
+
+	switch (reason) {
+	case LASTREC_UPDATE:
+		/*
+		 * If this is the last leaf block and it's the last record,
+		 * then update the size of the longest extent in the AG.
+		 */
+		if (ptr != xfs_btree_get_numrecs(block))
+			return;
+		len = rec->alloc.ar_blockcount;
+		break;
+	case LASTREC_INSREC:
+		if (be32_to_cpu(rec->alloc.ar_blockcount) <=
+		    be32_to_cpu(agf->agf_longest))
+			return;
+		len = rec->alloc.ar_blockcount;
+		break;
+	case LASTREC_DELREC:
+		numrecs = xfs_btree_get_numrecs(block);
+		if (ptr <= numrecs)
+			return;
+		ASSERT(ptr == numrecs + 1);
+
+		if (numrecs) {
+			xfs_alloc_rec_t *rrp;
+
+			rrp = XFS_ALLOC_REC_ADDR(cur->bc_mp, block, numrecs);
+			len = rrp->ar_blockcount;
+		} else {
+			len = 0;
+		}
+
+		break;
+	default:
+		ASSERT(0);
+		return;
+	}
+
+	agf->agf_longest = len;
+	pag = cur->bc_ag.agbp->b_pag;
+	pag->pagf_longest = be32_to_cpu(len);
+	xfs_alloc_log_agf(cur->bc_tp, cur->bc_ag.agbp, XFS_AGF_LONGEST);
+}
+
+STATIC int
+xfs_allocbt_get_minrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	return cur->bc_mp->m_alloc_mnr[level != 0];
+}
+
+STATIC int
+xfs_allocbt_get_maxrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	return cur->bc_mp->m_alloc_mxr[level != 0];
+}
+
+STATIC void
+xfs_allocbt_init_key_from_rec(
+	union xfs_btree_key		*key,
+	const union xfs_btree_rec	*rec)
+{
+	key->alloc.ar_startblock = rec->alloc.ar_startblock;
+	key->alloc.ar_blockcount = rec->alloc.ar_blockcount;
+}
+
+STATIC void
+xfs_bnobt_init_high_key_from_rec(
+	union xfs_btree_key		*key,
+	const union xfs_btree_rec	*rec)
+{
+	__u32				x;
+
+	x = be32_to_cpu(rec->alloc.ar_startblock);
+	x += be32_to_cpu(rec->alloc.ar_blockcount) - 1;
+	key->alloc.ar_startblock = cpu_to_be32(x);
+	key->alloc.ar_blockcount = 0;
+}
+
+STATIC void
+xfs_cntbt_init_high_key_from_rec(
+	union xfs_btree_key		*key,
+	const union xfs_btree_rec	*rec)
+{
+	key->alloc.ar_blockcount = rec->alloc.ar_blockcount;
+	key->alloc.ar_startblock = 0;
+}
+
+STATIC void
+xfs_allocbt_init_rec_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec)
+{
+	rec->alloc.ar_startblock = cpu_to_be32(cur->bc_rec.a.ar_startblock);
+	rec->alloc.ar_blockcount = cpu_to_be32(cur->bc_rec.a.ar_blockcount);
+}
+
+STATIC void
+xfs_allocbt_init_ptr_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	struct xfs_agf		*agf = cur->bc_ag.agbp->b_addr;
+
+	ASSERT(cur->bc_ag.pag->pag_agno == be32_to_cpu(agf->agf_seqno));
+
+	ptr->s = agf->agf_roots[cur->bc_btnum];
+}
+
+STATIC int64_t
+xfs_bnobt_key_diff(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*key)
+{
+	struct xfs_alloc_rec_incore	*rec = &cur->bc_rec.a;
+	const struct xfs_alloc_rec	*kp = &key->alloc;
+
+	return (int64_t)be32_to_cpu(kp->ar_startblock) - rec->ar_startblock;
+}
+
+STATIC int64_t
+xfs_cntbt_key_diff(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*key)
+{
+	struct xfs_alloc_rec_incore	*rec = &cur->bc_rec.a;
+	const struct xfs_alloc_rec	*kp = &key->alloc;
+	int64_t				diff;
+
+	diff = (int64_t)be32_to_cpu(kp->ar_blockcount) - rec->ar_blockcount;
+	if (diff)
+		return diff;
+
+	return (int64_t)be32_to_cpu(kp->ar_startblock) - rec->ar_startblock;
+}
+
+STATIC int64_t
+xfs_bnobt_diff_two_keys(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*k1,
+	const union xfs_btree_key	*k2)
+{
+	return (int64_t)be32_to_cpu(k1->alloc.ar_startblock) -
+			  be32_to_cpu(k2->alloc.ar_startblock);
+}
+
+STATIC int64_t
+xfs_cntbt_diff_two_keys(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*k1,
+	const union xfs_btree_key	*k2)
+{
+	int64_t				diff;
+
+	diff =  be32_to_cpu(k1->alloc.ar_blockcount) -
+		be32_to_cpu(k2->alloc.ar_blockcount);
+	if (diff)
+		return diff;
+
+	return  be32_to_cpu(k1->alloc.ar_startblock) -
+		be32_to_cpu(k2->alloc.ar_startblock);
+}
+
+static xfs_failaddr_t
+xfs_allocbt_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
+	struct xfs_perag	*pag = bp->b_pag;
+	xfs_failaddr_t		fa;
+	unsigned int		level;
+	xfs_btnum_t		btnum = XFS_BTNUM_BNOi;
+
+	if (!xfs_verify_magic(bp, block->bb_magic))
+		return __this_address;
+
+	if (xfs_has_crc(mp)) {
+		fa = xfs_btree_sblock_v5hdr_verify(bp);
+		if (fa)
+			return fa;
+	}
+
+	/*
+	 * The perag may not be attached during grow operations or fully
+	 * initialized from the AGF during log recovery. Therefore we can only
+	 * check against maximum tree depth from those contexts.
+	 *
+	 * Otherwise check against the per-tree limit. Peek at one of the
+	 * verifier magic values to determine the type of tree we're verifying
+	 * against.
+	 */
+	level = be16_to_cpu(block->bb_level);
+	if (bp->b_ops->magic[0] == cpu_to_be32(XFS_ABTC_MAGIC))
+		btnum = XFS_BTNUM_CNTi;
+	if (pag && pag->pagf_init) {
+		if (level >= pag->pagf_levels[btnum])
+			return __this_address;
+	} else if (level >= mp->m_alloc_maxlevels)
+		return __this_address;
+
+	return xfs_btree_sblock_verify(bp, mp->m_alloc_mxr[level != 0]);
+}
+
+static void
+xfs_allocbt_read_verify(
+	struct xfs_buf	*bp)
+{
+	xfs_failaddr_t	fa;
+
+	if (!xfs_btree_sblock_verify_crc(bp))
+		xfs_verifier_error(bp, -EFSBADCRC, __this_address);
+	else {
+		fa = xfs_allocbt_verify(bp);
+		if (fa)
+			xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+	}
+
+	if (bp->b_error)
+		trace_xfs_btree_corrupt(bp, _RET_IP_);
+}
+
+static void
+xfs_allocbt_write_verify(
+	struct xfs_buf	*bp)
+{
+	xfs_failaddr_t	fa;
+
+	fa = xfs_allocbt_verify(bp);
+	if (fa) {
+		trace_xfs_btree_corrupt(bp, _RET_IP_);
+		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+		return;
+	}
+	xfs_btree_sblock_calc_crc(bp);
+
+}
+
+const struct xfs_buf_ops xfs_bnobt_buf_ops = {
+	.name = "xfs_bnobt",
+	.magic = { cpu_to_be32(XFS_ABTB_MAGIC),
+		   cpu_to_be32(XFS_ABTB_CRC_MAGIC) },
+	.verify_read = xfs_allocbt_read_verify,
+	.verify_write = xfs_allocbt_write_verify,
+	.verify_struct = xfs_allocbt_verify,
+};
+
+const struct xfs_buf_ops xfs_cntbt_buf_ops = {
+	.name = "xfs_cntbt",
+	.magic = { cpu_to_be32(XFS_ABTC_MAGIC),
+		   cpu_to_be32(XFS_ABTC_CRC_MAGIC) },
+	.verify_read = xfs_allocbt_read_verify,
+	.verify_write = xfs_allocbt_write_verify,
+	.verify_struct = xfs_allocbt_verify,
+};
+
+STATIC int
+xfs_bnobt_keys_inorder(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*k1,
+	const union xfs_btree_key	*k2)
+{
+	return be32_to_cpu(k1->alloc.ar_startblock) <
+	       be32_to_cpu(k2->alloc.ar_startblock);
+}
+
+STATIC int
+xfs_bnobt_recs_inorder(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_rec	*r1,
+	const union xfs_btree_rec	*r2)
+{
+	return be32_to_cpu(r1->alloc.ar_startblock) +
+		be32_to_cpu(r1->alloc.ar_blockcount) <=
+		be32_to_cpu(r2->alloc.ar_startblock);
+}
+
+STATIC int
+xfs_cntbt_keys_inorder(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*k1,
+	const union xfs_btree_key	*k2)
+{
+	return be32_to_cpu(k1->alloc.ar_blockcount) <
+		be32_to_cpu(k2->alloc.ar_blockcount) ||
+		(k1->alloc.ar_blockcount == k2->alloc.ar_blockcount &&
+		 be32_to_cpu(k1->alloc.ar_startblock) <
+		 be32_to_cpu(k2->alloc.ar_startblock));
+}
+
+STATIC int
+xfs_cntbt_recs_inorder(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_rec	*r1,
+	const union xfs_btree_rec	*r2)
+{
+	return be32_to_cpu(r1->alloc.ar_blockcount) <
+		be32_to_cpu(r2->alloc.ar_blockcount) ||
+		(r1->alloc.ar_blockcount == r2->alloc.ar_blockcount &&
+		 be32_to_cpu(r1->alloc.ar_startblock) <
+		 be32_to_cpu(r2->alloc.ar_startblock));
+}
+
+static const struct xfs_btree_ops xfs_bnobt_ops = {
+	.rec_len		= sizeof(xfs_alloc_rec_t),
+	.key_len		= sizeof(xfs_alloc_key_t),
+
+	.dup_cursor		= xfs_allocbt_dup_cursor,
+	.set_root		= xfs_allocbt_set_root,
+	.alloc_block		= xfs_allocbt_alloc_block,
+	.free_block		= xfs_allocbt_free_block,
+	.update_lastrec		= xfs_allocbt_update_lastrec,
+	.get_minrecs		= xfs_allocbt_get_minrecs,
+	.get_maxrecs		= xfs_allocbt_get_maxrecs,
+	.init_key_from_rec	= xfs_allocbt_init_key_from_rec,
+	.init_high_key_from_rec	= xfs_bnobt_init_high_key_from_rec,
+	.init_rec_from_cur	= xfs_allocbt_init_rec_from_cur,
+	.init_ptr_from_cur	= xfs_allocbt_init_ptr_from_cur,
+	.key_diff		= xfs_bnobt_key_diff,
+	.buf_ops		= &xfs_bnobt_buf_ops,
+	.diff_two_keys		= xfs_bnobt_diff_two_keys,
+	.keys_inorder		= xfs_bnobt_keys_inorder,
+	.recs_inorder		= xfs_bnobt_recs_inorder,
+};
+
+static const struct xfs_btree_ops xfs_cntbt_ops = {
+	.rec_len		= sizeof(xfs_alloc_rec_t),
+	.key_len		= sizeof(xfs_alloc_key_t),
+
+	.dup_cursor		= xfs_allocbt_dup_cursor,
+	.set_root		= xfs_allocbt_set_root,
+	.alloc_block		= xfs_allocbt_alloc_block,
+	.free_block		= xfs_allocbt_free_block,
+	.update_lastrec		= xfs_allocbt_update_lastrec,
+	.get_minrecs		= xfs_allocbt_get_minrecs,
+	.get_maxrecs		= xfs_allocbt_get_maxrecs,
+	.init_key_from_rec	= xfs_allocbt_init_key_from_rec,
+	.init_high_key_from_rec	= xfs_cntbt_init_high_key_from_rec,
+	.init_rec_from_cur	= xfs_allocbt_init_rec_from_cur,
+	.init_ptr_from_cur	= xfs_allocbt_init_ptr_from_cur,
+	.key_diff		= xfs_cntbt_key_diff,
+	.buf_ops		= &xfs_cntbt_buf_ops,
+	.diff_two_keys		= xfs_cntbt_diff_two_keys,
+	.keys_inorder		= xfs_cntbt_keys_inorder,
+	.recs_inorder		= xfs_cntbt_recs_inorder,
+};
+
+/* Allocate most of a new allocation btree cursor. */
+STATIC struct xfs_btree_cur *
+xfs_allocbt_init_common(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_perag	*pag,
+	xfs_btnum_t		btnum)
+{
+	struct xfs_btree_cur	*cur;
+
+	ASSERT(btnum == XFS_BTNUM_BNO || btnum == XFS_BTNUM_CNT);
+
+	cur = xfs_btree_alloc_cursor(mp, tp, btnum, mp->m_alloc_maxlevels,
+			xfs_allocbt_cur_cache);
+	cur->bc_ag.abt.active = false;
+
+	if (btnum == XFS_BTNUM_CNT) {
+		cur->bc_ops = &xfs_cntbt_ops;
+		cur->bc_statoff = XFS_STATS_CALC_INDEX(xs_abtc_2);
+		cur->bc_flags = XFS_BTREE_LASTREC_UPDATE;
+	} else {
+		cur->bc_ops = &xfs_bnobt_ops;
+		cur->bc_statoff = XFS_STATS_CALC_INDEX(xs_abtb_2);
+	}
+
+	/* take a reference for the cursor */
+	atomic_inc(&pag->pag_ref);
+	cur->bc_ag.pag = pag;
+
+	if (xfs_has_crc(mp))
+		cur->bc_flags |= XFS_BTREE_CRC_BLOCKS;
+
+	return cur;
+}
+
+/*
+ * Allocate a new allocation btree cursor.
+ */
+struct xfs_btree_cur *			/* new alloc btree cursor */
+xfs_allocbt_init_cursor(
+	struct xfs_mount	*mp,		/* file system mount point */
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_buf		*agbp,		/* buffer for agf structure */
+	struct xfs_perag	*pag,
+	xfs_btnum_t		btnum)		/* btree identifier */
+{
+	struct xfs_agf		*agf = agbp->b_addr;
+	struct xfs_btree_cur	*cur;
+
+	cur = xfs_allocbt_init_common(mp, tp, pag, btnum);
+	if (btnum == XFS_BTNUM_CNT)
+		cur->bc_nlevels = be32_to_cpu(agf->agf_levels[XFS_BTNUM_CNT]);
+	else
+		cur->bc_nlevels = be32_to_cpu(agf->agf_levels[XFS_BTNUM_BNO]);
+
+	cur->bc_ag.agbp = agbp;
+
+	return cur;
+}
+
+/* Create a free space btree cursor with a fake root for staging. */
+struct xfs_btree_cur *
+xfs_allocbt_stage_cursor(
+	struct xfs_mount	*mp,
+	struct xbtree_afakeroot	*afake,
+	struct xfs_perag	*pag,
+	xfs_btnum_t		btnum)
+{
+	struct xfs_btree_cur	*cur;
+
+	cur = xfs_allocbt_init_common(mp, NULL, pag, btnum);
+	xfs_btree_stage_afakeroot(cur, afake);
+	return cur;
+}
+
+/*
+ * Install a new free space btree root.  Caller is responsible for invalidating
+ * and freeing the old btree blocks.
+ */
+void
+xfs_allocbt_commit_staged_btree(
+	struct xfs_btree_cur	*cur,
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp)
+{
+	struct xfs_agf		*agf = agbp->b_addr;
+	struct xbtree_afakeroot	*afake = cur->bc_ag.afake;
+
+	ASSERT(cur->bc_flags & XFS_BTREE_STAGING);
+
+	agf->agf_roots[cur->bc_btnum] = cpu_to_be32(afake->af_root);
+	agf->agf_levels[cur->bc_btnum] = cpu_to_be32(afake->af_levels);
+	xfs_alloc_log_agf(tp, agbp, XFS_AGF_ROOTS | XFS_AGF_LEVELS);
+
+	if (cur->bc_btnum == XFS_BTNUM_BNO) {
+		xfs_btree_commit_afakeroot(cur, tp, agbp, &xfs_bnobt_ops);
+	} else {
+		cur->bc_flags |= XFS_BTREE_LASTREC_UPDATE;
+		xfs_btree_commit_afakeroot(cur, tp, agbp, &xfs_cntbt_ops);
+	}
+}
+
+/* Calculate number of records in an alloc btree block. */
+static inline unsigned int
+xfs_allocbt_block_maxrecs(
+	unsigned int		blocklen,
+	bool			leaf)
+{
+	if (leaf)
+		return blocklen / sizeof(xfs_alloc_rec_t);
+	return blocklen / (sizeof(xfs_alloc_key_t) + sizeof(xfs_alloc_ptr_t));
+}
+
+/*
+ * Calculate number of records in an alloc btree block.
+ */
+int
+xfs_allocbt_maxrecs(
+	struct xfs_mount	*mp,
+	int			blocklen,
+	int			leaf)
+{
+	blocklen -= XFS_ALLOC_BLOCK_LEN(mp);
+	return xfs_allocbt_block_maxrecs(blocklen, leaf);
+}
+
+/* Free space btrees are at their largest when every other block is free. */
+#define XFS_MAX_FREESP_RECORDS	((XFS_MAX_AG_BLOCKS + 1) / 2)
+
+/* Compute the max possible height for free space btrees. */
+unsigned int
+xfs_allocbt_maxlevels_ondisk(void)
+{
+	unsigned int		minrecs[2];
+	unsigned int		blocklen;
+
+	blocklen = min(XFS_MIN_BLOCKSIZE - XFS_BTREE_SBLOCK_LEN,
+		       XFS_MIN_CRC_BLOCKSIZE - XFS_BTREE_SBLOCK_CRC_LEN);
+
+	minrecs[0] = xfs_allocbt_block_maxrecs(blocklen, true) / 2;
+	minrecs[1] = xfs_allocbt_block_maxrecs(blocklen, false) / 2;
+
+	return xfs_btree_compute_maxlevels(minrecs, XFS_MAX_FREESP_RECORDS);
+}
+
+/* Calculate the freespace btree size for some records. */
+xfs_extlen_t
+xfs_allocbt_calc_size(
+	struct xfs_mount	*mp,
+	unsigned long long	len)
+{
+	return xfs_btree_calc_size(mp->m_alloc_mnr, len);
+}
+
+int __init
+xfs_allocbt_init_cur_cache(void)
+{
+	xfs_allocbt_cur_cache = kmem_cache_create("xfs_bnobt_cur",
+			xfs_btree_cur_sizeof(xfs_allocbt_maxlevels_ondisk()),
+			0, 0, NULL);
+
+	if (!xfs_allocbt_cur_cache)
+		return -ENOMEM;
+	return 0;
+}
+
+void
+xfs_allocbt_destroy_cur_cache(void)
+{
+	kmem_cache_destroy(xfs_allocbt_cur_cache);
+	xfs_allocbt_cur_cache = NULL;
+}
diff --git a/fs/xfs/libxfs/xfs_alloc_btree.h b/fs/xfs/libxfs/xfs_alloc_btree.h
new file mode 100644
index 000000000..45df893ef
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_alloc_btree.h
@@ -0,0 +1,68 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_ALLOC_BTREE_H__
+#define	__XFS_ALLOC_BTREE_H__
+
+/*
+ * Freespace on-disk structures
+ */
+
+struct xfs_buf;
+struct xfs_btree_cur;
+struct xfs_mount;
+struct xfs_perag;
+struct xbtree_afakeroot;
+
+/*
+ * Btree block header size depends on a superblock flag.
+ */
+#define XFS_ALLOC_BLOCK_LEN(mp) \
+	(xfs_has_crc(((mp))) ? \
+		XFS_BTREE_SBLOCK_CRC_LEN : XFS_BTREE_SBLOCK_LEN)
+
+/*
+ * Record, key, and pointer address macros for btree blocks.
+ *
+ * (note that some of these may appear unused, but they are used in userspace)
+ */
+#define XFS_ALLOC_REC_ADDR(mp, block, index) \
+	((xfs_alloc_rec_t *) \
+		((char *)(block) + \
+		 XFS_ALLOC_BLOCK_LEN(mp) + \
+		 (((index) - 1) * sizeof(xfs_alloc_rec_t))))
+
+#define XFS_ALLOC_KEY_ADDR(mp, block, index) \
+	((xfs_alloc_key_t *) \
+		((char *)(block) + \
+		 XFS_ALLOC_BLOCK_LEN(mp) + \
+		 ((index) - 1) * sizeof(xfs_alloc_key_t)))
+
+#define XFS_ALLOC_PTR_ADDR(mp, block, index, maxrecs) \
+	((xfs_alloc_ptr_t *) \
+		((char *)(block) + \
+		 XFS_ALLOC_BLOCK_LEN(mp) + \
+		 (maxrecs) * sizeof(xfs_alloc_key_t) + \
+		 ((index) - 1) * sizeof(xfs_alloc_ptr_t)))
+
+extern struct xfs_btree_cur *xfs_allocbt_init_cursor(struct xfs_mount *mp,
+		struct xfs_trans *tp, struct xfs_buf *bp,
+		struct xfs_perag *pag, xfs_btnum_t btnum);
+struct xfs_btree_cur *xfs_allocbt_stage_cursor(struct xfs_mount *mp,
+		struct xbtree_afakeroot *afake, struct xfs_perag *pag,
+		xfs_btnum_t btnum);
+extern int xfs_allocbt_maxrecs(struct xfs_mount *, int, int);
+extern xfs_extlen_t xfs_allocbt_calc_size(struct xfs_mount *mp,
+		unsigned long long len);
+
+void xfs_allocbt_commit_staged_btree(struct xfs_btree_cur *cur,
+		struct xfs_trans *tp, struct xfs_buf *agbp);
+
+unsigned int xfs_allocbt_maxlevels_ondisk(void);
+
+int __init xfs_allocbt_init_cur_cache(void);
+void xfs_allocbt_destroy_cur_cache(void);
+
+#endif	/* __XFS_ALLOC_BTREE_H__ */
diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c
new file mode 100644
index 000000000..e28d93d23
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_attr.c
@@ -0,0 +1,1600 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_defer.h"
+#include "xfs_da_format.h"
+#include "xfs_da_btree.h"
+#include "xfs_attr_sf.h"
+#include "xfs_inode.h"
+#include "xfs_trans.h"
+#include "xfs_bmap.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_attr.h"
+#include "xfs_attr_leaf.h"
+#include "xfs_attr_remote.h"
+#include "xfs_quota.h"
+#include "xfs_trans_space.h"
+#include "xfs_trace.h"
+#include "xfs_attr_item.h"
+#include "xfs_xattr.h"
+
+struct kmem_cache		*xfs_attr_intent_cache;
+
+/*
+ * xfs_attr.c
+ *
+ * Provide the external interfaces to manage attribute lists.
+ */
+
+/*========================================================================
+ * Function prototypes for the kernel.
+ *========================================================================*/
+
+/*
+ * Internal routines when attribute list fits inside the inode.
+ */
+STATIC int xfs_attr_shortform_addname(xfs_da_args_t *args);
+
+/*
+ * Internal routines when attribute list is one block.
+ */
+STATIC int xfs_attr_leaf_get(xfs_da_args_t *args);
+STATIC int xfs_attr_leaf_removename(xfs_da_args_t *args);
+STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp);
+STATIC int xfs_attr_leaf_try_add(struct xfs_da_args *args);
+
+/*
+ * Internal routines when attribute list is more than one block.
+ */
+STATIC int xfs_attr_node_get(xfs_da_args_t *args);
+STATIC void xfs_attr_restore_rmt_blk(struct xfs_da_args *args);
+static int xfs_attr_node_try_addname(struct xfs_attr_intent *attr);
+STATIC int xfs_attr_node_addname_find_attr(struct xfs_attr_intent *attr);
+STATIC int xfs_attr_node_remove_attr(struct xfs_attr_intent *attr);
+STATIC int xfs_attr_node_lookup(struct xfs_da_args *args,
+		struct xfs_da_state *state);
+
+int
+xfs_inode_hasattr(
+	struct xfs_inode	*ip)
+{
+	if (!xfs_inode_has_attr_fork(ip))
+		return 0;
+	if (ip->i_af.if_format == XFS_DINODE_FMT_EXTENTS &&
+	    ip->i_af.if_nextents == 0)
+		return 0;
+	return 1;
+}
+
+/*
+ * Returns true if the there is exactly only block in the attr fork, in which
+ * case the attribute fork consists of a single leaf block entry.
+ */
+bool
+xfs_attr_is_leaf(
+	struct xfs_inode	*ip)
+{
+	struct xfs_ifork	*ifp = &ip->i_af;
+	struct xfs_iext_cursor	icur;
+	struct xfs_bmbt_irec	imap;
+
+	if (ifp->if_nextents != 1 || ifp->if_format != XFS_DINODE_FMT_EXTENTS)
+		return false;
+
+	xfs_iext_first(ifp, &icur);
+	xfs_iext_get_extent(ifp, &icur, &imap);
+	return imap.br_startoff == 0 && imap.br_blockcount == 1;
+}
+
+/*
+ * XXX (dchinner): name path state saving and refilling is an optimisation to
+ * avoid needing to look up name entries after rolling transactions removing
+ * remote xattr blocks between the name entry lookup and name entry removal.
+ * This optimisation got sidelined when combining the set and remove state
+ * machines, but the code has been left in place because it is worthwhile to
+ * restore the optimisation once the combined state machine paths have settled.
+ *
+ * This comment is a public service announcement to remind Future Dave that he
+ * still needs to restore this code to working order.
+ */
+#if 0
+/*
+ * Fill in the disk block numbers in the state structure for the buffers
+ * that are attached to the state structure.
+ * This is done so that we can quickly reattach ourselves to those buffers
+ * after some set of transaction commits have released these buffers.
+ */
+static int
+xfs_attr_fillstate(xfs_da_state_t *state)
+{
+	xfs_da_state_path_t *path;
+	xfs_da_state_blk_t *blk;
+	int level;
+
+	trace_xfs_attr_fillstate(state->args);
+
+	/*
+	 * Roll down the "path" in the state structure, storing the on-disk
+	 * block number for those buffers in the "path".
+	 */
+	path = &state->path;
+	ASSERT((path->active >= 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
+	for (blk = path->blk, level = 0; level < path->active; blk++, level++) {
+		if (blk->bp) {
+			blk->disk_blkno = xfs_buf_daddr(blk->bp);
+			blk->bp = NULL;
+		} else {
+			blk->disk_blkno = 0;
+		}
+	}
+
+	/*
+	 * Roll down the "altpath" in the state structure, storing the on-disk
+	 * block number for those buffers in the "altpath".
+	 */
+	path = &state->altpath;
+	ASSERT((path->active >= 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
+	for (blk = path->blk, level = 0; level < path->active; blk++, level++) {
+		if (blk->bp) {
+			blk->disk_blkno = xfs_buf_daddr(blk->bp);
+			blk->bp = NULL;
+		} else {
+			blk->disk_blkno = 0;
+		}
+	}
+
+	return 0;
+}
+
+/*
+ * Reattach the buffers to the state structure based on the disk block
+ * numbers stored in the state structure.
+ * This is done after some set of transaction commits have released those
+ * buffers from our grip.
+ */
+static int
+xfs_attr_refillstate(xfs_da_state_t *state)
+{
+	xfs_da_state_path_t *path;
+	xfs_da_state_blk_t *blk;
+	int level, error;
+
+	trace_xfs_attr_refillstate(state->args);
+
+	/*
+	 * Roll down the "path" in the state structure, storing the on-disk
+	 * block number for those buffers in the "path".
+	 */
+	path = &state->path;
+	ASSERT((path->active >= 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
+	for (blk = path->blk, level = 0; level < path->active; blk++, level++) {
+		if (blk->disk_blkno) {
+			error = xfs_da3_node_read_mapped(state->args->trans,
+					state->args->dp, blk->disk_blkno,
+					&blk->bp, XFS_ATTR_FORK);
+			if (error)
+				return error;
+		} else {
+			blk->bp = NULL;
+		}
+	}
+
+	/*
+	 * Roll down the "altpath" in the state structure, storing the on-disk
+	 * block number for those buffers in the "altpath".
+	 */
+	path = &state->altpath;
+	ASSERT((path->active >= 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
+	for (blk = path->blk, level = 0; level < path->active; blk++, level++) {
+		if (blk->disk_blkno) {
+			error = xfs_da3_node_read_mapped(state->args->trans,
+					state->args->dp, blk->disk_blkno,
+					&blk->bp, XFS_ATTR_FORK);
+			if (error)
+				return error;
+		} else {
+			blk->bp = NULL;
+		}
+	}
+
+	return 0;
+}
+#else
+static int xfs_attr_fillstate(xfs_da_state_t *state) { return 0; }
+#endif
+
+/*========================================================================
+ * Overall external interface routines.
+ *========================================================================*/
+
+/*
+ * Retrieve an extended attribute and its value.  Must have ilock.
+ * Returns 0 on successful retrieval, otherwise an error.
+ */
+int
+xfs_attr_get_ilocked(
+	struct xfs_da_args	*args)
+{
+	ASSERT(xfs_isilocked(args->dp, XFS_ILOCK_SHARED | XFS_ILOCK_EXCL));
+
+	if (!xfs_inode_hasattr(args->dp))
+		return -ENOATTR;
+
+	if (args->dp->i_af.if_format == XFS_DINODE_FMT_LOCAL)
+		return xfs_attr_shortform_getvalue(args);
+	if (xfs_attr_is_leaf(args->dp))
+		return xfs_attr_leaf_get(args);
+	return xfs_attr_node_get(args);
+}
+
+/*
+ * Retrieve an extended attribute by name, and its value if requested.
+ *
+ * If args->valuelen is zero, then the caller does not want the value, just an
+ * indication whether the attribute exists and the size of the value if it
+ * exists. The size is returned in args.valuelen.
+ *
+ * If args->value is NULL but args->valuelen is non-zero, allocate the buffer
+ * for the value after existence of the attribute has been determined. The
+ * caller always has to free args->value if it is set, no matter if this
+ * function was successful or not.
+ *
+ * If the attribute is found, but exceeds the size limit set by the caller in
+ * args->valuelen, return -ERANGE with the size of the attribute that was found
+ * in args->valuelen.
+ */
+int
+xfs_attr_get(
+	struct xfs_da_args	*args)
+{
+	uint			lock_mode;
+	int			error;
+
+	XFS_STATS_INC(args->dp->i_mount, xs_attr_get);
+
+	if (xfs_is_shutdown(args->dp->i_mount))
+		return -EIO;
+
+	args->geo = args->dp->i_mount->m_attr_geo;
+	args->whichfork = XFS_ATTR_FORK;
+	args->hashval = xfs_da_hashname(args->name, args->namelen);
+
+	/* Entirely possible to look up a name which doesn't exist */
+	args->op_flags = XFS_DA_OP_OKNOENT;
+
+	lock_mode = xfs_ilock_attr_map_shared(args->dp);
+	error = xfs_attr_get_ilocked(args);
+	xfs_iunlock(args->dp, lock_mode);
+
+	return error;
+}
+
+/*
+ * Calculate how many blocks we need for the new attribute,
+ */
+int
+xfs_attr_calc_size(
+	struct xfs_da_args	*args,
+	int			*local)
+{
+	struct xfs_mount	*mp = args->dp->i_mount;
+	int			size;
+	int			nblks;
+
+	/*
+	 * Determine space new attribute will use, and if it would be
+	 * "local" or "remote" (note: local != inline).
+	 */
+	size = xfs_attr_leaf_newentsize(args, local);
+	nblks = XFS_DAENTER_SPACE_RES(mp, XFS_ATTR_FORK);
+	if (*local) {
+		if (size > (args->geo->blksize / 2)) {
+			/* Double split possible */
+			nblks *= 2;
+		}
+	} else {
+		/*
+		 * Out of line attribute, cannot double split, but
+		 * make room for the attribute value itself.
+		 */
+		uint	dblocks = xfs_attr3_rmt_blocks(mp, args->valuelen);
+		nblks += dblocks;
+		nblks += XFS_NEXTENTADD_SPACE_RES(mp, dblocks, XFS_ATTR_FORK);
+	}
+
+	return nblks;
+}
+
+/* Initialize transaction reservation for attr operations */
+void
+xfs_init_attr_trans(
+	struct xfs_da_args	*args,
+	struct xfs_trans_res	*tres,
+	unsigned int		*total)
+{
+	struct xfs_mount	*mp = args->dp->i_mount;
+
+	if (args->value) {
+		tres->tr_logres = M_RES(mp)->tr_attrsetm.tr_logres +
+				 M_RES(mp)->tr_attrsetrt.tr_logres *
+				 args->total;
+		tres->tr_logcount = XFS_ATTRSET_LOG_COUNT;
+		tres->tr_logflags = XFS_TRANS_PERM_LOG_RES;
+		*total = args->total;
+	} else {
+		*tres = M_RES(mp)->tr_attrrm;
+		*total = XFS_ATTRRM_SPACE_RES(mp);
+	}
+}
+
+/*
+ * Add an attr to a shortform fork. If there is no space,
+ * xfs_attr_shortform_addname() will convert to leaf format and return -ENOSPC.
+ * to use.
+ */
+STATIC int
+xfs_attr_try_sf_addname(
+	struct xfs_inode	*dp,
+	struct xfs_da_args	*args)
+{
+
+	int			error;
+
+	/*
+	 * Build initial attribute list (if required).
+	 */
+	if (dp->i_af.if_format == XFS_DINODE_FMT_EXTENTS)
+		xfs_attr_shortform_create(args);
+
+	error = xfs_attr_shortform_addname(args);
+	if (error == -ENOSPC)
+		return error;
+
+	/*
+	 * Commit the shortform mods, and we're done.
+	 * NOTE: this is also the error path (EEXIST, etc).
+	 */
+	if (!error && !(args->op_flags & XFS_DA_OP_NOTIME))
+		xfs_trans_ichgtime(args->trans, dp, XFS_ICHGTIME_CHG);
+
+	if (xfs_has_wsync(dp->i_mount))
+		xfs_trans_set_sync(args->trans);
+
+	return error;
+}
+
+static int
+xfs_attr_sf_addname(
+	struct xfs_attr_intent		*attr)
+{
+	struct xfs_da_args		*args = attr->xattri_da_args;
+	struct xfs_inode		*dp = args->dp;
+	int				error = 0;
+
+	error = xfs_attr_try_sf_addname(dp, args);
+	if (error != -ENOSPC) {
+		ASSERT(!error || error == -EEXIST);
+		attr->xattri_dela_state = XFS_DAS_DONE;
+		goto out;
+	}
+
+	/*
+	 * It won't fit in the shortform, transform to a leaf block.  GROT:
+	 * another possible req'mt for a double-split btree op.
+	 */
+	error = xfs_attr_shortform_to_leaf(args);
+	if (error)
+		return error;
+
+	attr->xattri_dela_state = XFS_DAS_LEAF_ADD;
+out:
+	trace_xfs_attr_sf_addname_return(attr->xattri_dela_state, args->dp);
+	return error;
+}
+
+/*
+ * Handle the state change on completion of a multi-state attr operation.
+ *
+ * If the XFS_DA_OP_REPLACE flag is set, this means the operation was the first
+ * modification in a attr replace operation and we still have to do the second
+ * state, indicated by @replace_state.
+ *
+ * We consume the XFS_DA_OP_REPLACE flag so that when we are called again on
+ * completion of the second half of the attr replace operation we correctly
+ * signal that it is done.
+ */
+static enum xfs_delattr_state
+xfs_attr_complete_op(
+	struct xfs_attr_intent	*attr,
+	enum xfs_delattr_state	replace_state)
+{
+	struct xfs_da_args	*args = attr->xattri_da_args;
+	bool			do_replace = args->op_flags & XFS_DA_OP_REPLACE;
+
+	args->op_flags &= ~XFS_DA_OP_REPLACE;
+	if (do_replace) {
+		args->attr_filter &= ~XFS_ATTR_INCOMPLETE;
+		return replace_state;
+	}
+	return XFS_DAS_DONE;
+}
+
+static int
+xfs_attr_leaf_addname(
+	struct xfs_attr_intent	*attr)
+{
+	struct xfs_da_args	*args = attr->xattri_da_args;
+	int			error;
+
+	ASSERT(xfs_attr_is_leaf(args->dp));
+
+	/*
+	 * Use the leaf buffer we may already hold locked as a result of
+	 * a sf-to-leaf conversion.
+	 */
+	error = xfs_attr_leaf_try_add(args);
+
+	if (error == -ENOSPC) {
+		error = xfs_attr3_leaf_to_node(args);
+		if (error)
+			return error;
+
+		/*
+		 * We're not in leaf format anymore, so roll the transaction and
+		 * retry the add to the newly allocated node block.
+		 */
+		attr->xattri_dela_state = XFS_DAS_NODE_ADD;
+		goto out;
+	}
+	if (error)
+		return error;
+
+	/*
+	 * We need to commit and roll if we need to allocate remote xattr blocks
+	 * or perform more xattr manipulations. Otherwise there is nothing more
+	 * to do and we can return success.
+	 */
+	if (args->rmtblkno)
+		attr->xattri_dela_state = XFS_DAS_LEAF_SET_RMT;
+	else
+		attr->xattri_dela_state = xfs_attr_complete_op(attr,
+							XFS_DAS_LEAF_REPLACE);
+out:
+	trace_xfs_attr_leaf_addname_return(attr->xattri_dela_state, args->dp);
+	return error;
+}
+
+/*
+ * Add an entry to a node format attr tree.
+ *
+ * Note that we might still have a leaf here - xfs_attr_is_leaf() cannot tell
+ * the difference between leaf + remote attr blocks and a node format tree,
+ * so we may still end up having to convert from leaf to node format here.
+ */
+static int
+xfs_attr_node_addname(
+	struct xfs_attr_intent	*attr)
+{
+	struct xfs_da_args	*args = attr->xattri_da_args;
+	int			error;
+
+	error = xfs_attr_node_addname_find_attr(attr);
+	if (error)
+		return error;
+
+	error = xfs_attr_node_try_addname(attr);
+	if (error == -ENOSPC) {
+		error = xfs_attr3_leaf_to_node(args);
+		if (error)
+			return error;
+		/*
+		 * No state change, we really are in node form now
+		 * but we need the transaction rolled to continue.
+		 */
+		goto out;
+	}
+	if (error)
+		return error;
+
+	if (args->rmtblkno)
+		attr->xattri_dela_state = XFS_DAS_NODE_SET_RMT;
+	else
+		attr->xattri_dela_state = xfs_attr_complete_op(attr,
+							XFS_DAS_NODE_REPLACE);
+out:
+	trace_xfs_attr_node_addname_return(attr->xattri_dela_state, args->dp);
+	return error;
+}
+
+static int
+xfs_attr_rmtval_alloc(
+	struct xfs_attr_intent		*attr)
+{
+	struct xfs_da_args              *args = attr->xattri_da_args;
+	int				error = 0;
+
+	/*
+	 * If there was an out-of-line value, allocate the blocks we
+	 * identified for its storage and copy the value.  This is done
+	 * after we create the attribute so that we don't overflow the
+	 * maximum size of a transaction and/or hit a deadlock.
+	 */
+	if (attr->xattri_blkcnt > 0) {
+		error = xfs_attr_rmtval_set_blk(attr);
+		if (error)
+			return error;
+		/* Roll the transaction only if there is more to allocate. */
+		if (attr->xattri_blkcnt > 0)
+			goto out;
+	}
+
+	error = xfs_attr_rmtval_set_value(args);
+	if (error)
+		return error;
+
+	attr->xattri_dela_state = xfs_attr_complete_op(attr,
+						++attr->xattri_dela_state);
+	/*
+	 * If we are not doing a rename, we've finished the operation but still
+	 * have to clear the incomplete flag protecting the new attr from
+	 * exposing partially initialised state if we crash during creation.
+	 */
+	if (attr->xattri_dela_state == XFS_DAS_DONE)
+		error = xfs_attr3_leaf_clearflag(args);
+out:
+	trace_xfs_attr_rmtval_alloc(attr->xattri_dela_state, args->dp);
+	return error;
+}
+
+/*
+ * Mark an attribute entry INCOMPLETE and save pointers to the relevant buffers
+ * for later deletion of the entry.
+ */
+static int
+xfs_attr_leaf_mark_incomplete(
+	struct xfs_da_args	*args,
+	struct xfs_da_state	*state)
+{
+	int			error;
+
+	/*
+	 * Fill in disk block numbers in the state structure
+	 * so that we can get the buffers back after we commit
+	 * several transactions in the following calls.
+	 */
+	error = xfs_attr_fillstate(state);
+	if (error)
+		return error;
+
+	/*
+	 * Mark the attribute as INCOMPLETE
+	 */
+	return xfs_attr3_leaf_setflag(args);
+}
+
+/* Ensure the da state of an xattr deferred work item is ready to go. */
+static inline void
+xfs_attr_item_init_da_state(
+	struct xfs_attr_intent	*attr)
+{
+	struct xfs_da_args	*args = attr->xattri_da_args;
+
+	if (!attr->xattri_da_state)
+		attr->xattri_da_state = xfs_da_state_alloc(args);
+	else
+		xfs_da_state_reset(attr->xattri_da_state, args);
+}
+
+/*
+ * Initial setup for xfs_attr_node_removename.  Make sure the attr is there and
+ * the blocks are valid.  Attr keys with remote blocks will be marked
+ * incomplete.
+ */
+static
+int xfs_attr_node_removename_setup(
+	struct xfs_attr_intent		*attr)
+{
+	struct xfs_da_args		*args = attr->xattri_da_args;
+	struct xfs_da_state		*state;
+	int				error;
+
+	xfs_attr_item_init_da_state(attr);
+	error = xfs_attr_node_lookup(args, attr->xattri_da_state);
+	if (error != -EEXIST)
+		goto out;
+	error = 0;
+
+	state = attr->xattri_da_state;
+	ASSERT(state->path.blk[state->path.active - 1].bp != NULL);
+	ASSERT(state->path.blk[state->path.active - 1].magic ==
+		XFS_ATTR_LEAF_MAGIC);
+
+	error = xfs_attr_leaf_mark_incomplete(args, state);
+	if (error)
+		goto out;
+	if (args->rmtblkno > 0)
+		error = xfs_attr_rmtval_invalidate(args);
+out:
+	if (error) {
+		xfs_da_state_free(attr->xattri_da_state);
+		attr->xattri_da_state = NULL;
+	}
+
+	return error;
+}
+
+/*
+ * Remove the original attr we have just replaced. This is dependent on the
+ * original lookup and insert placing the old attr in args->blkno/args->index
+ * and the new attr in args->blkno2/args->index2.
+ */
+static int
+xfs_attr_leaf_remove_attr(
+	struct xfs_attr_intent		*attr)
+{
+	struct xfs_da_args              *args = attr->xattri_da_args;
+	struct xfs_inode		*dp = args->dp;
+	struct xfs_buf			*bp = NULL;
+	int				forkoff;
+	int				error;
+
+	error = xfs_attr3_leaf_read(args->trans, args->dp, args->blkno,
+				   &bp);
+	if (error)
+		return error;
+
+	xfs_attr3_leaf_remove(bp, args);
+
+	forkoff = xfs_attr_shortform_allfit(bp, dp);
+	if (forkoff)
+		error = xfs_attr3_leaf_to_shortform(bp, args, forkoff);
+		/* bp is gone due to xfs_da_shrink_inode */
+
+	return error;
+}
+
+/*
+ * Shrink an attribute from leaf to shortform. Used by the node format remove
+ * path when the node format collapses to a single block and so we have to check
+ * if it can be collapsed further.
+ */
+static int
+xfs_attr_leaf_shrink(
+	struct xfs_da_args	*args)
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_buf		*bp;
+	int			forkoff;
+	int			error;
+
+	if (!xfs_attr_is_leaf(dp))
+		return 0;
+
+	error = xfs_attr3_leaf_read(args->trans, args->dp, 0, &bp);
+	if (error)
+		return error;
+
+	forkoff = xfs_attr_shortform_allfit(bp, dp);
+	if (forkoff) {
+		error = xfs_attr3_leaf_to_shortform(bp, args, forkoff);
+		/* bp is gone due to xfs_da_shrink_inode */
+	} else {
+		xfs_trans_brelse(args->trans, bp);
+	}
+
+	return error;
+}
+
+/*
+ * Run the attribute operation specified in @attr.
+ *
+ * This routine is meant to function as a delayed operation and will set the
+ * state to XFS_DAS_DONE when the operation is complete.  Calling functions will
+ * need to handle this, and recall the function until either an error or
+ * XFS_DAS_DONE is detected.
+ */
+int
+xfs_attr_set_iter(
+	struct xfs_attr_intent		*attr)
+{
+	struct xfs_da_args              *args = attr->xattri_da_args;
+	int				error = 0;
+
+	/* State machine switch */
+next_state:
+	switch (attr->xattri_dela_state) {
+	case XFS_DAS_UNINIT:
+		ASSERT(0);
+		return -EFSCORRUPTED;
+	case XFS_DAS_SF_ADD:
+		return xfs_attr_sf_addname(attr);
+	case XFS_DAS_LEAF_ADD:
+		return xfs_attr_leaf_addname(attr);
+	case XFS_DAS_NODE_ADD:
+		return xfs_attr_node_addname(attr);
+
+	case XFS_DAS_SF_REMOVE:
+		error = xfs_attr_sf_removename(args);
+		attr->xattri_dela_state = xfs_attr_complete_op(attr,
+						xfs_attr_init_add_state(args));
+		break;
+	case XFS_DAS_LEAF_REMOVE:
+		error = xfs_attr_leaf_removename(args);
+		attr->xattri_dela_state = xfs_attr_complete_op(attr,
+						xfs_attr_init_add_state(args));
+		break;
+	case XFS_DAS_NODE_REMOVE:
+		error = xfs_attr_node_removename_setup(attr);
+		if (error == -ENOATTR &&
+		    (args->op_flags & XFS_DA_OP_RECOVERY)) {
+			attr->xattri_dela_state = xfs_attr_complete_op(attr,
+						xfs_attr_init_add_state(args));
+			error = 0;
+			break;
+		}
+		if (error)
+			return error;
+		attr->xattri_dela_state = XFS_DAS_NODE_REMOVE_RMT;
+		if (args->rmtblkno == 0)
+			attr->xattri_dela_state++;
+		break;
+
+	case XFS_DAS_LEAF_SET_RMT:
+	case XFS_DAS_NODE_SET_RMT:
+		error = xfs_attr_rmtval_find_space(attr);
+		if (error)
+			return error;
+		attr->xattri_dela_state++;
+		fallthrough;
+
+	case XFS_DAS_LEAF_ALLOC_RMT:
+	case XFS_DAS_NODE_ALLOC_RMT:
+		error = xfs_attr_rmtval_alloc(attr);
+		if (error)
+			return error;
+		if (attr->xattri_dela_state == XFS_DAS_DONE)
+			break;
+		goto next_state;
+
+	case XFS_DAS_LEAF_REPLACE:
+	case XFS_DAS_NODE_REPLACE:
+		/*
+		 * We must "flip" the incomplete flags on the "new" and "old"
+		 * attribute/value pairs so that one disappears and one appears
+		 * atomically.
+		 */
+		error = xfs_attr3_leaf_flipflags(args);
+		if (error)
+			return error;
+		/*
+		 * We must commit the flag value change now to make it atomic
+		 * and then we can start the next trans in series at REMOVE_OLD.
+		 */
+		attr->xattri_dela_state++;
+		break;
+
+	case XFS_DAS_LEAF_REMOVE_OLD:
+	case XFS_DAS_NODE_REMOVE_OLD:
+		/*
+		 * If we have a remote attr, start the process of removing it
+		 * by invalidating any cached buffers.
+		 *
+		 * If we don't have a remote attr, we skip the remote block
+		 * removal state altogether with a second state increment.
+		 */
+		xfs_attr_restore_rmt_blk(args);
+		if (args->rmtblkno) {
+			error = xfs_attr_rmtval_invalidate(args);
+			if (error)
+				return error;
+		} else {
+			attr->xattri_dela_state++;
+		}
+
+		attr->xattri_dela_state++;
+		goto next_state;
+
+	case XFS_DAS_LEAF_REMOVE_RMT:
+	case XFS_DAS_NODE_REMOVE_RMT:
+		error = xfs_attr_rmtval_remove(attr);
+		if (error == -EAGAIN) {
+			error = 0;
+			break;
+		}
+		if (error)
+			return error;
+
+		/*
+		 * We've finished removing the remote attr blocks, so commit the
+		 * transaction and move on to removing the attr name from the
+		 * leaf/node block. Removing the attr might require a full
+		 * transaction reservation for btree block freeing, so we
+		 * can't do that in the same transaction where we removed the
+		 * remote attr blocks.
+		 */
+		attr->xattri_dela_state++;
+		break;
+
+	case XFS_DAS_LEAF_REMOVE_ATTR:
+		error = xfs_attr_leaf_remove_attr(attr);
+		attr->xattri_dela_state = xfs_attr_complete_op(attr,
+						xfs_attr_init_add_state(args));
+		break;
+
+	case XFS_DAS_NODE_REMOVE_ATTR:
+		error = xfs_attr_node_remove_attr(attr);
+		if (!error)
+			error = xfs_attr_leaf_shrink(args);
+		attr->xattri_dela_state = xfs_attr_complete_op(attr,
+						xfs_attr_init_add_state(args));
+		break;
+	default:
+		ASSERT(0);
+		break;
+	}
+
+	trace_xfs_attr_set_iter_return(attr->xattri_dela_state, args->dp);
+	return error;
+}
+
+
+/*
+ * Return EEXIST if attr is found, or ENOATTR if not
+ */
+static int
+xfs_attr_lookup(
+	struct xfs_da_args	*args)
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_buf		*bp = NULL;
+	struct xfs_da_state	*state;
+	int			error;
+
+	if (!xfs_inode_hasattr(dp))
+		return -ENOATTR;
+
+	if (dp->i_af.if_format == XFS_DINODE_FMT_LOCAL)
+		return xfs_attr_sf_findname(args, NULL, NULL);
+
+	if (xfs_attr_is_leaf(dp)) {
+		error = xfs_attr_leaf_hasname(args, &bp);
+
+		if (bp)
+			xfs_trans_brelse(args->trans, bp);
+
+		return error;
+	}
+
+	state = xfs_da_state_alloc(args);
+	error = xfs_attr_node_lookup(args, state);
+	xfs_da_state_free(state);
+	return error;
+}
+
+static int
+xfs_attr_intent_init(
+	struct xfs_da_args	*args,
+	unsigned int		op_flags,	/* op flag (set or remove) */
+	struct xfs_attr_intent	**attr)		/* new xfs_attr_intent */
+{
+
+	struct xfs_attr_intent	*new;
+
+	new = kmem_cache_zalloc(xfs_attr_intent_cache, GFP_NOFS | __GFP_NOFAIL);
+	new->xattri_op_flags = op_flags;
+	new->xattri_da_args = args;
+
+	*attr = new;
+	return 0;
+}
+
+/* Sets an attribute for an inode as a deferred operation */
+static int
+xfs_attr_defer_add(
+	struct xfs_da_args	*args)
+{
+	struct xfs_attr_intent	*new;
+	int			error = 0;
+
+	error = xfs_attr_intent_init(args, XFS_ATTRI_OP_FLAGS_SET, &new);
+	if (error)
+		return error;
+
+	new->xattri_dela_state = xfs_attr_init_add_state(args);
+	xfs_defer_add(args->trans, XFS_DEFER_OPS_TYPE_ATTR, &new->xattri_list);
+	trace_xfs_attr_defer_add(new->xattri_dela_state, args->dp);
+
+	return 0;
+}
+
+/* Sets an attribute for an inode as a deferred operation */
+static int
+xfs_attr_defer_replace(
+	struct xfs_da_args	*args)
+{
+	struct xfs_attr_intent	*new;
+	int			error = 0;
+
+	error = xfs_attr_intent_init(args, XFS_ATTRI_OP_FLAGS_REPLACE, &new);
+	if (error)
+		return error;
+
+	new->xattri_dela_state = xfs_attr_init_replace_state(args);
+	xfs_defer_add(args->trans, XFS_DEFER_OPS_TYPE_ATTR, &new->xattri_list);
+	trace_xfs_attr_defer_replace(new->xattri_dela_state, args->dp);
+
+	return 0;
+}
+
+/* Removes an attribute for an inode as a deferred operation */
+static int
+xfs_attr_defer_remove(
+	struct xfs_da_args	*args)
+{
+
+	struct xfs_attr_intent	*new;
+	int			error;
+
+	error  = xfs_attr_intent_init(args, XFS_ATTRI_OP_FLAGS_REMOVE, &new);
+	if (error)
+		return error;
+
+	new->xattri_dela_state = xfs_attr_init_remove_state(args);
+	xfs_defer_add(args->trans, XFS_DEFER_OPS_TYPE_ATTR, &new->xattri_list);
+	trace_xfs_attr_defer_remove(new->xattri_dela_state, args->dp);
+
+	return 0;
+}
+
+/*
+ * Note: If args->value is NULL the attribute will be removed, just like the
+ * Linux ->setattr API.
+ */
+int
+xfs_attr_set(
+	struct xfs_da_args	*args)
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	struct xfs_trans_res	tres;
+	bool			rsvd = (args->attr_filter & XFS_ATTR_ROOT);
+	int			error, local;
+	int			rmt_blks = 0;
+	unsigned int		total;
+
+	if (xfs_is_shutdown(dp->i_mount))
+		return -EIO;
+
+	error = xfs_qm_dqattach(dp);
+	if (error)
+		return error;
+
+	args->geo = mp->m_attr_geo;
+	args->whichfork = XFS_ATTR_FORK;
+	args->hashval = xfs_da_hashname(args->name, args->namelen);
+
+	/*
+	 * We have no control over the attribute names that userspace passes us
+	 * to remove, so we have to allow the name lookup prior to attribute
+	 * removal to fail as well.  Preserve the logged flag, since we need
+	 * to pass that through to the logging code.
+	 */
+	args->op_flags = XFS_DA_OP_OKNOENT |
+					(args->op_flags & XFS_DA_OP_LOGGED);
+
+	if (args->value) {
+		XFS_STATS_INC(mp, xs_attr_set);
+		args->total = xfs_attr_calc_size(args, &local);
+
+		/*
+		 * If the inode doesn't have an attribute fork, add one.
+		 * (inode must not be locked when we call this routine)
+		 */
+		if (xfs_inode_has_attr_fork(dp) == 0) {
+			int sf_size = sizeof(struct xfs_attr_sf_hdr) +
+				xfs_attr_sf_entsize_byname(args->namelen,
+						args->valuelen);
+
+			error = xfs_bmap_add_attrfork(dp, sf_size, rsvd);
+			if (error)
+				return error;
+		}
+
+		if (!local)
+			rmt_blks = xfs_attr3_rmt_blocks(mp, args->valuelen);
+	} else {
+		XFS_STATS_INC(mp, xs_attr_remove);
+		rmt_blks = xfs_attr3_rmt_blocks(mp, XFS_XATTR_SIZE_MAX);
+	}
+
+	/*
+	 * Root fork attributes can use reserved data blocks for this
+	 * operation if necessary
+	 */
+	xfs_init_attr_trans(args, &tres, &total);
+	error = xfs_trans_alloc_inode(dp, &tres, total, 0, rsvd, &args->trans);
+	if (error)
+		return error;
+
+	if (args->value || xfs_inode_hasattr(dp)) {
+		error = xfs_iext_count_may_overflow(dp, XFS_ATTR_FORK,
+				XFS_IEXT_ATTR_MANIP_CNT(rmt_blks));
+		if (error == -EFBIG)
+			error = xfs_iext_count_upgrade(args->trans, dp,
+					XFS_IEXT_ATTR_MANIP_CNT(rmt_blks));
+		if (error)
+			goto out_trans_cancel;
+	}
+
+	error = xfs_attr_lookup(args);
+	switch (error) {
+	case -EEXIST:
+		/* if no value, we are performing a remove operation */
+		if (!args->value) {
+			error = xfs_attr_defer_remove(args);
+			break;
+		}
+		/* Pure create fails if the attr already exists */
+		if (args->attr_flags & XATTR_CREATE)
+			goto out_trans_cancel;
+
+		error = xfs_attr_defer_replace(args);
+		break;
+	case -ENOATTR:
+		/* Can't remove what isn't there. */
+		if (!args->value)
+			goto out_trans_cancel;
+
+		/* Pure replace fails if no existing attr to replace. */
+		if (args->attr_flags & XATTR_REPLACE)
+			goto out_trans_cancel;
+
+		error = xfs_attr_defer_add(args);
+		break;
+	default:
+		goto out_trans_cancel;
+	}
+	if (error)
+		goto out_trans_cancel;
+
+	/*
+	 * If this is a synchronous mount, make sure that the
+	 * transaction goes to disk before returning to the user.
+	 */
+	if (xfs_has_wsync(mp))
+		xfs_trans_set_sync(args->trans);
+
+	if (!(args->op_flags & XFS_DA_OP_NOTIME))
+		xfs_trans_ichgtime(args->trans, dp, XFS_ICHGTIME_CHG);
+
+	/*
+	 * Commit the last in the sequence of transactions.
+	 */
+	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE);
+	error = xfs_trans_commit(args->trans);
+out_unlock:
+	xfs_iunlock(dp, XFS_ILOCK_EXCL);
+	return error;
+
+out_trans_cancel:
+	if (args->trans)
+		xfs_trans_cancel(args->trans);
+	goto out_unlock;
+}
+
+/*========================================================================
+ * External routines when attribute list is inside the inode
+ *========================================================================*/
+
+static inline int xfs_attr_sf_totsize(struct xfs_inode *dp)
+{
+	struct xfs_attr_shortform *sf;
+
+	sf = (struct xfs_attr_shortform *)dp->i_af.if_u1.if_data;
+	return be16_to_cpu(sf->hdr.totsize);
+}
+
+/*
+ * Add a name to the shortform attribute list structure
+ * This is the external routine.
+ */
+static int
+xfs_attr_shortform_addname(
+	struct xfs_da_args	*args)
+{
+	int			newsize, forkoff;
+	int			error;
+
+	trace_xfs_attr_sf_addname(args);
+
+	error = xfs_attr_shortform_lookup(args);
+	switch (error) {
+	case -ENOATTR:
+		if (args->op_flags & XFS_DA_OP_REPLACE)
+			return error;
+		break;
+	case -EEXIST:
+		if (!(args->op_flags & XFS_DA_OP_REPLACE))
+			return error;
+
+		error = xfs_attr_sf_removename(args);
+		if (error)
+			return error;
+
+		/*
+		 * Since we have removed the old attr, clear XFS_DA_OP_REPLACE
+		 * so that the new attr doesn't fit in shortform format, the
+		 * leaf format add routine won't trip over the attr not being
+		 * around.
+		 */
+		args->op_flags &= ~XFS_DA_OP_REPLACE;
+		break;
+	case 0:
+		break;
+	default:
+		return error;
+	}
+
+	if (args->namelen >= XFS_ATTR_SF_ENTSIZE_MAX ||
+	    args->valuelen >= XFS_ATTR_SF_ENTSIZE_MAX)
+		return -ENOSPC;
+
+	newsize = xfs_attr_sf_totsize(args->dp);
+	newsize += xfs_attr_sf_entsize_byname(args->namelen, args->valuelen);
+
+	forkoff = xfs_attr_shortform_bytesfit(args->dp, newsize);
+	if (!forkoff)
+		return -ENOSPC;
+
+	xfs_attr_shortform_add(args, forkoff);
+	return 0;
+}
+
+
+/*========================================================================
+ * External routines when attribute list is one block
+ *========================================================================*/
+
+/* Save the current remote block info and clear the current pointers. */
+static void
+xfs_attr_save_rmt_blk(
+	struct xfs_da_args	*args)
+{
+	args->blkno2 = args->blkno;
+	args->index2 = args->index;
+	args->rmtblkno2 = args->rmtblkno;
+	args->rmtblkcnt2 = args->rmtblkcnt;
+	args->rmtvaluelen2 = args->rmtvaluelen;
+	args->rmtblkno = 0;
+	args->rmtblkcnt = 0;
+	args->rmtvaluelen = 0;
+}
+
+/* Set stored info about a remote block */
+static void
+xfs_attr_restore_rmt_blk(
+	struct xfs_da_args	*args)
+{
+	args->blkno = args->blkno2;
+	args->index = args->index2;
+	args->rmtblkno = args->rmtblkno2;
+	args->rmtblkcnt = args->rmtblkcnt2;
+	args->rmtvaluelen = args->rmtvaluelen2;
+}
+
+/*
+ * Tries to add an attribute to an inode in leaf form
+ *
+ * This function is meant to execute as part of a delayed operation and leaves
+ * the transaction handling to the caller.  On success the attribute is added
+ * and the inode and transaction are left dirty.  If there is not enough space,
+ * the attr data is converted to node format and -ENOSPC is returned. Caller is
+ * responsible for handling the dirty inode and transaction or adding the attr
+ * in node format.
+ */
+STATIC int
+xfs_attr_leaf_try_add(
+	struct xfs_da_args	*args)
+{
+	struct xfs_buf		*bp;
+	int			error;
+
+	error = xfs_attr3_leaf_read(args->trans, args->dp, 0, &bp);
+	if (error)
+		return error;
+
+	/*
+	 * Look up the xattr name to set the insertion point for the new xattr.
+	 */
+	error = xfs_attr3_leaf_lookup_int(bp, args);
+	switch (error) {
+	case -ENOATTR:
+		if (args->op_flags & XFS_DA_OP_REPLACE)
+			goto out_brelse;
+		break;
+	case -EEXIST:
+		if (!(args->op_flags & XFS_DA_OP_REPLACE))
+			goto out_brelse;
+
+		trace_xfs_attr_leaf_replace(args);
+		/*
+		 * Save the existing remote attr state so that the current
+		 * values reflect the state of the new attribute we are about to
+		 * add, not the attribute we just found and will remove later.
+		 */
+		xfs_attr_save_rmt_blk(args);
+		break;
+	case 0:
+		break;
+	default:
+		goto out_brelse;
+	}
+
+	return xfs_attr3_leaf_add(bp, args);
+
+out_brelse:
+	xfs_trans_brelse(args->trans, bp);
+	return error;
+}
+
+/*
+ * Return EEXIST if attr is found, or ENOATTR if not
+ */
+STATIC int
+xfs_attr_leaf_hasname(
+	struct xfs_da_args	*args,
+	struct xfs_buf		**bp)
+{
+	int                     error = 0;
+
+	error = xfs_attr3_leaf_read(args->trans, args->dp, 0, bp);
+	if (error)
+		return error;
+
+	error = xfs_attr3_leaf_lookup_int(*bp, args);
+	if (error != -ENOATTR && error != -EEXIST)
+		xfs_trans_brelse(args->trans, *bp);
+
+	return error;
+}
+
+/*
+ * Remove a name from the leaf attribute list structure
+ *
+ * This leaf block cannot have a "remote" value, we only call this routine
+ * if bmap_one_block() says there is only one block (ie: no remote blks).
+ */
+STATIC int
+xfs_attr_leaf_removename(
+	struct xfs_da_args	*args)
+{
+	struct xfs_inode	*dp;
+	struct xfs_buf		*bp;
+	int			error, forkoff;
+
+	trace_xfs_attr_leaf_removename(args);
+
+	/*
+	 * Remove the attribute.
+	 */
+	dp = args->dp;
+
+	error = xfs_attr_leaf_hasname(args, &bp);
+	if (error == -ENOATTR) {
+		xfs_trans_brelse(args->trans, bp);
+		if (args->op_flags & XFS_DA_OP_RECOVERY)
+			return 0;
+		return error;
+	} else if (error != -EEXIST)
+		return error;
+
+	xfs_attr3_leaf_remove(bp, args);
+
+	/*
+	 * If the result is small enough, shrink it all into the inode.
+	 */
+	forkoff = xfs_attr_shortform_allfit(bp, dp);
+	if (forkoff)
+		return xfs_attr3_leaf_to_shortform(bp, args, forkoff);
+		/* bp is gone due to xfs_da_shrink_inode */
+
+	return 0;
+}
+
+/*
+ * Look up a name in a leaf attribute list structure.
+ *
+ * This leaf block cannot have a "remote" value, we only call this routine
+ * if bmap_one_block() says there is only one block (ie: no remote blks).
+ *
+ * Returns 0 on successful retrieval, otherwise an error.
+ */
+STATIC int
+xfs_attr_leaf_get(xfs_da_args_t *args)
+{
+	struct xfs_buf *bp;
+	int error;
+
+	trace_xfs_attr_leaf_get(args);
+
+	error = xfs_attr_leaf_hasname(args, &bp);
+
+	if (error == -ENOATTR)  {
+		xfs_trans_brelse(args->trans, bp);
+		return error;
+	} else if (error != -EEXIST)
+		return error;
+
+
+	error = xfs_attr3_leaf_getvalue(bp, args);
+	xfs_trans_brelse(args->trans, bp);
+	return error;
+}
+
+/* Return EEXIST if attr is found, or ENOATTR if not. */
+STATIC int
+xfs_attr_node_lookup(
+	struct xfs_da_args	*args,
+	struct xfs_da_state	*state)
+{
+	int			retval, error;
+
+	/*
+	 * Search to see if name exists, and get back a pointer to it.
+	 */
+	error = xfs_da3_node_lookup_int(state, &retval);
+	if (error)
+		return error;
+
+	return retval;
+}
+
+/*========================================================================
+ * External routines when attribute list size > geo->blksize
+ *========================================================================*/
+
+STATIC int
+xfs_attr_node_addname_find_attr(
+	 struct xfs_attr_intent	*attr)
+{
+	struct xfs_da_args	*args = attr->xattri_da_args;
+	int			error;
+
+	/*
+	 * Search to see if name already exists, and get back a pointer
+	 * to where it should go.
+	 */
+	xfs_attr_item_init_da_state(attr);
+	error = xfs_attr_node_lookup(args, attr->xattri_da_state);
+	switch (error) {
+	case -ENOATTR:
+		if (args->op_flags & XFS_DA_OP_REPLACE)
+			goto error;
+		break;
+	case -EEXIST:
+		if (!(args->op_flags & XFS_DA_OP_REPLACE))
+			goto error;
+
+
+		trace_xfs_attr_node_replace(args);
+		/*
+		 * Save the existing remote attr state so that the current
+		 * values reflect the state of the new attribute we are about to
+		 * add, not the attribute we just found and will remove later.
+		 */
+		xfs_attr_save_rmt_blk(args);
+		break;
+	case 0:
+		break;
+	default:
+		goto error;
+	}
+
+	return 0;
+error:
+	if (attr->xattri_da_state) {
+		xfs_da_state_free(attr->xattri_da_state);
+		attr->xattri_da_state = NULL;
+	}
+	return error;
+}
+
+/*
+ * Add a name to a Btree-format attribute list.
+ *
+ * This will involve walking down the Btree, and may involve splitting
+ * leaf nodes and even splitting intermediate nodes up to and including
+ * the root node (a special case of an intermediate node).
+ */
+static int
+xfs_attr_node_try_addname(
+	struct xfs_attr_intent		*attr)
+{
+	struct xfs_da_state		*state = attr->xattri_da_state;
+	struct xfs_da_state_blk		*blk;
+	int				error;
+
+	trace_xfs_attr_node_addname(state->args);
+
+	blk = &state->path.blk[state->path.active-1];
+	ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
+
+	error = xfs_attr3_leaf_add(blk->bp, state->args);
+	if (error == -ENOSPC) {
+		if (state->path.active == 1) {
+			/*
+			 * Its really a single leaf node, but it had
+			 * out-of-line values so it looked like it *might*
+			 * have been a b-tree. Let the caller deal with this.
+			 */
+			goto out;
+		}
+
+		/*
+		 * Split as many Btree elements as required.
+		 * This code tracks the new and old attr's location
+		 * in the index/blkno/rmtblkno/rmtblkcnt fields and
+		 * in the index2/blkno2/rmtblkno2/rmtblkcnt2 fields.
+		 */
+		error = xfs_da3_split(state);
+		if (error)
+			goto out;
+	} else {
+		/*
+		 * Addition succeeded, update Btree hashvals.
+		 */
+		xfs_da3_fixhashpath(state, &state->path);
+	}
+
+out:
+	xfs_da_state_free(state);
+	attr->xattri_da_state = NULL;
+	return error;
+}
+
+static int
+xfs_attr_node_removename(
+	struct xfs_da_args	*args,
+	struct xfs_da_state	*state)
+{
+	struct xfs_da_state_blk	*blk;
+	int			retval;
+
+	/*
+	 * Remove the name and update the hashvals in the tree.
+	 */
+	blk = &state->path.blk[state->path.active-1];
+	ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
+	retval = xfs_attr3_leaf_remove(blk->bp, args);
+	xfs_da3_fixhashpath(state, &state->path);
+
+	return retval;
+}
+
+static int
+xfs_attr_node_remove_attr(
+	struct xfs_attr_intent		*attr)
+{
+	struct xfs_da_args		*args = attr->xattri_da_args;
+	struct xfs_da_state		*state = xfs_da_state_alloc(args);
+	int				retval = 0;
+	int				error = 0;
+
+	/*
+	 * The attr we are removing has already been marked incomplete, so
+	 * we need to set the filter appropriately to re-find the "old"
+	 * attribute entry after any split ops.
+	 */
+	args->attr_filter |= XFS_ATTR_INCOMPLETE;
+	error = xfs_da3_node_lookup_int(state, &retval);
+	if (error)
+		goto out;
+
+	error = xfs_attr_node_removename(args, state);
+
+	/*
+	 * Check to see if the tree needs to be collapsed.
+	 */
+	if (retval && (state->path.active > 1)) {
+		error = xfs_da3_join(state);
+		if (error)
+			goto out;
+	}
+	retval = error = 0;
+
+out:
+	xfs_da_state_free(state);
+	if (error)
+		return error;
+	return retval;
+}
+
+/*
+ * Retrieve the attribute data from a node attribute list.
+ *
+ * This routine gets called for any attribute fork that has more than one
+ * block, ie: both true Btree attr lists and for single-leaf-blocks with
+ * "remote" values taking up more blocks.
+ *
+ * Returns 0 on successful retrieval, otherwise an error.
+ */
+STATIC int
+xfs_attr_node_get(
+	struct xfs_da_args	*args)
+{
+	struct xfs_da_state	*state;
+	struct xfs_da_state_blk	*blk;
+	int			i;
+	int			error;
+
+	trace_xfs_attr_node_get(args);
+
+	/*
+	 * Search to see if name exists, and get back a pointer to it.
+	 */
+	state = xfs_da_state_alloc(args);
+	error = xfs_attr_node_lookup(args, state);
+	if (error != -EEXIST)
+		goto out_release;
+
+	/*
+	 * Get the value, local or "remote"
+	 */
+	blk = &state->path.blk[state->path.active - 1];
+	error = xfs_attr3_leaf_getvalue(blk->bp, args);
+
+	/*
+	 * If not in a transaction, we have to release all the buffers.
+	 */
+out_release:
+	for (i = 0; i < state->path.active; i++) {
+		xfs_trans_brelse(args->trans, state->path.blk[i].bp);
+		state->path.blk[i].bp = NULL;
+	}
+
+	xfs_da_state_free(state);
+	return error;
+}
+
+/* Returns true if the attribute entry name is valid. */
+bool
+xfs_attr_namecheck(
+	const void	*name,
+	size_t		length)
+{
+	/*
+	 * MAXNAMELEN includes the trailing null, but (name/length) leave it
+	 * out, so use >= for the length check.
+	 */
+	if (length >= MAXNAMELEN)
+		return false;
+
+	/* There shouldn't be any nulls here */
+	return !memchr(name, 0, length);
+}
+
+int __init
+xfs_attr_intent_init_cache(void)
+{
+	xfs_attr_intent_cache = kmem_cache_create("xfs_attr_intent",
+			sizeof(struct xfs_attr_intent),
+			0, 0, NULL);
+
+	return xfs_attr_intent_cache != NULL ? 0 : -ENOMEM;
+}
+
+void
+xfs_attr_intent_destroy_cache(void)
+{
+	kmem_cache_destroy(xfs_attr_intent_cache);
+	xfs_attr_intent_cache = NULL;
+}
diff --git a/fs/xfs/libxfs/xfs_attr.h b/fs/xfs/libxfs/xfs_attr.h
new file mode 100644
index 000000000..81be9b3e4
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_attr.h
@@ -0,0 +1,621 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2000,2002-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_ATTR_H__
+#define	__XFS_ATTR_H__
+
+struct xfs_inode;
+struct xfs_da_args;
+struct xfs_attr_list_context;
+
+/*
+ * Large attribute lists are structured around Btrees where all the data
+ * elements are in the leaf nodes.  Attribute names are hashed into an int,
+ * then that int is used as the index into the Btree.  Since the hashval
+ * of an attribute name may not be unique, we may have duplicate keys.
+ * The internal links in the Btree are logical block offsets into the file.
+ *
+ * Small attribute lists use a different format and are packed as tightly
+ * as possible so as to fit into the literal area of the inode.
+ */
+
+/*
+ * The maximum size (into the kernel or returned from the kernel) of an
+ * attribute value or the buffer used for an attr_list() call.  Larger
+ * sizes will result in an ERANGE return code.
+ */
+#define	ATTR_MAX_VALUELEN	(64*1024)	/* max length of a value */
+
+/*
+ * Kernel-internal version of the attrlist cursor.
+ */
+struct xfs_attrlist_cursor_kern {
+	__u32	hashval;	/* hash value of next entry to add */
+	__u32	blkno;		/* block containing entry (suggestion) */
+	__u32	offset;		/* offset in list of equal-hashvals */
+	__u16	pad1;		/* padding to match user-level */
+	__u8	pad2;		/* padding to match user-level */
+	__u8	initted;	/* T/F: cursor has been initialized */
+};
+
+
+/*========================================================================
+ * Structure used to pass context around among the routines.
+ *========================================================================*/
+
+
+/* void; state communicated via *context */
+typedef void (*put_listent_func_t)(struct xfs_attr_list_context *, int,
+			      unsigned char *, int, int);
+
+struct xfs_attr_list_context {
+	struct xfs_trans	*tp;
+	struct xfs_inode	*dp;		/* inode */
+	struct xfs_attrlist_cursor_kern cursor;	/* position in list */
+	void			*buffer;	/* output buffer */
+
+	/*
+	 * Abort attribute list iteration if non-zero.  Can be used to pass
+	 * error values to the xfs_attr_list caller.
+	 */
+	int			seen_enough;
+	bool			allow_incomplete;
+
+	ssize_t			count;		/* num used entries */
+	int			dupcnt;		/* count dup hashvals seen */
+	int			bufsize;	/* total buffer size */
+	int			firstu;		/* first used byte in buffer */
+	unsigned int		attr_filter;	/* XFS_ATTR_{ROOT,SECURE} */
+	int			resynch;	/* T/F: resynch with cursor */
+	put_listent_func_t	put_listent;	/* list output fmt function */
+	int			index;		/* index into output buffer */
+};
+
+
+/*
+ * ========================================================================
+ * Structure used to pass context around among the delayed routines.
+ * ========================================================================
+ */
+
+/*
+ * Below is a state machine diagram for attr remove operations. The  XFS_DAS_*
+ * states indicate places where the function would return -EAGAIN, and then
+ * immediately resume from after being called by the calling function. States
+ * marked as a "subroutine state" indicate that they belong to a subroutine, and
+ * so the calling function needs to pass them back to that subroutine to allow
+ * it to finish where it left off. But they otherwise do not have a role in the
+ * calling function other than just passing through.
+ *
+ * xfs_attr_remove_iter()
+ *              │
+ *              v
+ *        have attr to remove? ──n──> done
+ *              │
+ *              y
+ *              │
+ *              v
+ *        are we short form? ──y──> xfs_attr_shortform_remove ──> done
+ *              │
+ *              n
+ *              │
+ *              V
+ *        are we leaf form? ──y──> xfs_attr_leaf_removename ──> done
+ *              │
+ *              n
+ *              │
+ *              V
+ *   ┌── need to setup state?
+ *   │          │
+ *   n          y
+ *   │          │
+ *   │          v
+ *   │ find attr and get state
+ *   │ attr has remote blks? ──n─┐
+ *   │          │                v
+ *   │          │         find and invalidate
+ *   │          y         the remote blocks.
+ *   │          │         mark attr incomplete
+ *   │          ├────────────────┘
+ *   └──────────┤
+ *              │
+ *              v
+ *   Have remote blks to remove? ───y─────┐
+ *              │        ^          remove the blks
+ *              │        │                │
+ *              │        │                v
+ *              │  XFS_DAS_RMTBLK <─n── done?
+ *              │  re-enter with          │
+ *              │  one less blk to        y
+ *              │      remove             │
+ *              │                         V
+ *              │                  refill the state
+ *              n                         │
+ *              │                         v
+ *              │                   XFS_DAS_RM_NAME
+ *              │                         │
+ *              ├─────────────────────────┘
+ *              │
+ *              v
+ *       remove leaf and
+ *       update hash with
+ *   xfs_attr_node_remove_cleanup
+ *              │
+ *              v
+ *           need to
+ *        shrink tree? ─n─┐
+ *              │         │
+ *              y         │
+ *              │         │
+ *              v         │
+ *          join leaf     │
+ *              │         │
+ *              v         │
+ *      XFS_DAS_RM_SHRINK │
+ *              │         │
+ *              v         │
+ *       do the shrink    │
+ *              │         │
+ *              v         │
+ *          free state <──┘
+ *              │
+ *              v
+ *            done
+ *
+ *
+ * Below is a state machine diagram for attr set operations.
+ *
+ * It seems the challenge with understanding this system comes from trying to
+ * absorb the state machine all at once, when really one should only be looking
+ * at it with in the context of a single function. Once a state sensitive
+ * function is called, the idea is that it "takes ownership" of the
+ * state machine. It isn't concerned with the states that may have belonged to
+ * it's calling parent. Only the states relevant to itself or any other
+ * subroutines there in. Once a calling function hands off the state machine to
+ * a subroutine, it needs to respect the simple rule that it doesn't "own" the
+ * state machine anymore, and it's the responsibility of that calling function
+ * to propagate the -EAGAIN back up the call stack. Upon reentry, it is
+ * committed to re-calling that subroutine until it returns something other than
+ * -EAGAIN. Once that subroutine signals completion (by returning anything other
+ * than -EAGAIN), the calling function can resume using the state machine.
+ *
+ *  xfs_attr_set_iter()
+ *              │
+ *              v
+ *   ┌─y─ has an attr fork?
+ *   │          |
+ *   │          n
+ *   │          |
+ *   │          V
+ *   │       add a fork
+ *   │          │
+ *   └──────────┤
+ *              │
+ *              V
+ *   ┌─── is shortform?
+ *   │          │
+ *   │          y
+ *   │          │
+ *   │          V
+ *   │   xfs_attr_set_fmt
+ *   │          |
+ *   │          V
+ *   │ xfs_attr_try_sf_addname
+ *   │          │
+ *   │          V
+ *   │      had enough ──y──> done
+ *   │        space?
+ *   n          │
+ *   │          n
+ *   │          │
+ *   │          V
+ *   │   transform to leaf
+ *   │          │
+ *   │          V
+ *   │   hold the leaf buffer
+ *   │          │
+ *   │          V
+ *   │     return -EAGAIN
+ *   │      Re-enter in
+ *   │       leaf form
+ *   │
+ *   └─> release leaf buffer
+ *          if needed
+ *              │
+ *              V
+ *   ┌───n── fork has
+ *   │      only 1 blk?
+ *   │          │
+ *   │          y
+ *   │          │
+ *   │          v
+ *   │ xfs_attr_leaf_try_add()
+ *   │          │
+ *   │          v
+ *   │      had enough ──────────────y─────────────┐
+ *   │        space?                               │
+ *   │          │                                  │
+ *   │          n                                  │
+ *   │          │                                  │
+ *   │          v                                  │
+ *   │    return -EAGAIN                           │
+ *   │      re-enter in                            │
+ *   │        node form                            │
+ *   │          │                                  │
+ *   └──────────┤                                  │
+ *              │                                  │
+ *              V                                  │
+ * xfs_attr_node_addname_find_attr                 │
+ *        determines if this                       │
+ *       is create or rename                       │
+ *     find space to store attr                    │
+ *              │                                  │
+ *              v                                  │
+ *     xfs_attr_node_addname                       │
+ *              │                                  │
+ *              v                                  │
+ *   fits in a node leaf? ────n─────┐              │
+ *              │     ^             v              │
+ *              │     │       single leaf node?    │
+ *              │     │         │            │     │
+ *              y     │         y            n     │
+ *              │     │         │            │     │
+ *              v     │         v            v     │
+ *            update  │    grow the leaf  split if │
+ *           hashvals └── return -EAGAIN   needed  │
+ *              │         retry leaf add     │     │
+ *              │           on reentry       │     │
+ *              ├────────────────────────────┘     │
+ *              │                                  │
+ *              v                                  │
+ *         need to alloc                           │
+ *   ┌─y── or flip flag?                           │
+ *   │          │                                  │
+ *   │          n                                  │
+ *   │          │                                  │
+ *   │          v                                  │
+ *   │         done                                │
+ *   │                                             │
+ *   │                                             │
+ *   │         XFS_DAS_FOUND_LBLK <────────────────┘
+ *   │                  │
+ *   │                  V
+ *   │        xfs_attr_leaf_addname()
+ *   │                  │
+ *   │                  v
+ *   │      ┌──first time through?
+ *   │      │          │
+ *   │      │          y
+ *   │      │          │
+ *   │      n          v
+ *   │      │    if we have rmt blks
+ *   │      │    find space for them
+ *   │      │          │
+ *   │      └──────────┤
+ *   │                 │
+ *   │                 v
+ *   │            still have
+ *   │      ┌─n─ blks to alloc? <──┐
+ *   │      │          │           │
+ *   │      │          y           │
+ *   │      │          │           │
+ *   │      │          v           │
+ *   │      │     alloc one blk    │
+ *   │      │     return -EAGAIN ──┘
+ *   │      │    re-enter with one
+ *   │      │    less blk to alloc
+ *   │      │
+ *   │      │
+ *   │      └───> set the rmt
+ *   │               value
+ *   │                 │
+ *   │                 v
+ *   │               was this
+ *   │              a rename? ──n─┐
+ *   │                 │          │
+ *   │                 y          │
+ *   │                 │          │
+ *   │                 v          │
+ *   │           flip incomplete  │
+ *   │               flag         │
+ *   │                 │          │
+ *   │                 v          │
+ *   │         XFS_DAS_FLIP_LFLAG │
+ *   │                 │          │
+ *   │                 v          │
+ *   │          need to remove    │
+ *   │              old bks? ──n──┤
+ *   │                 │          │
+ *   │                 y          │
+ *   │                 │          │
+ *   │                 V          │
+ *   │               remove       │
+ *   │        ┌───> old blks      │
+ *   │        │        │          │
+ *   │ XFS_DAS_RM_LBLK │          │
+ *   │        ^        │          │
+ *   │        │        v          │
+ *   │        └──y── more to      │
+ *   │              remove?       │
+ *   │                 │          │
+ *   │                 n          │
+ *   │                 │          │
+ *   │                 v          │
+ *   │          XFS_DAS_RD_LEAF   │
+ *   │                 │          │
+ *   │                 v          │
+ *   │            remove leaf     │
+ *   │                 │          │
+ *   │                 v          │
+ *   │            shrink to sf    │
+ *   │             if needed      │
+ *   │                 │          │
+ *   │                 v          │
+ *   │                done <──────┘
+ *   │
+ *   └──────> XFS_DAS_FOUND_NBLK
+ *                     │
+ *                     v
+ *       ┌─────n──  need to
+ *       │        alloc blks?
+ *       │             │
+ *       │             y
+ *       │             │
+ *       │             v
+ *       │        find space
+ *       │             │
+ *       │             v
+ *       │  ┌─>XFS_DAS_ALLOC_NODE
+ *       │  │          │
+ *       │  │          v
+ *       │  │      alloc blk
+ *       │  │          │
+ *       │  │          v
+ *       │  └──y── need to alloc
+ *       │         more blocks?
+ *       │             │
+ *       │             n
+ *       │             │
+ *       │             v
+ *       │      set the rmt value
+ *       │             │
+ *       │             v
+ *       │          was this
+ *       └────────> a rename? ──n─┐
+ *                     │          │
+ *                     y          │
+ *                     │          │
+ *                     v          │
+ *               flip incomplete  │
+ *                   flag         │
+ *                     │          │
+ *                     v          │
+ *             XFS_DAS_FLIP_NFLAG │
+ *                     │          │
+ *                     v          │
+ *                 need to        │
+ *               remove blks? ─n──┤
+ *                     │          │
+ *                     y          │
+ *                     │          │
+ *                     v          │
+ *                   remove       │
+ *        ┌────────> old blks     │
+ *        │            │          │
+ *  XFS_DAS_RM_NBLK    │          │
+ *        ^            │          │
+ *        │            v          │
+ *        └──────y── more to      │
+ *                   remove       │
+ *                     │          │
+ *                     n          │
+ *                     │          │
+ *                     v          │
+ *              XFS_DAS_CLR_FLAG  │
+ *                     │          │
+ *                     v          │
+ *                clear flags     │
+ *                     │          │
+ *                     ├──────────┘
+ *                     │
+ *                     v
+ *                   done
+ */
+
+/*
+ * Enum values for xfs_attr_intent.xattri_da_state
+ *
+ * These values are used by delayed attribute operations to keep track  of where
+ * they were before they returned -EAGAIN.  A return code of -EAGAIN signals the
+ * calling function to roll the transaction, and then call the subroutine to
+ * finish the operation.  The enum is then used by the subroutine to jump back
+ * to where it was and resume executing where it left off.
+ */
+enum xfs_delattr_state {
+	XFS_DAS_UNINIT		= 0,	/* No state has been set yet */
+
+	/*
+	 * Initial sequence states. The replace setup code relies on the
+	 * ADD and REMOVE states for a specific format to be sequential so
+	 * that we can transform the initial operation to be performed
+	 * according to the xfs_has_larp() state easily.
+	 */
+	XFS_DAS_SF_ADD,			/* Initial sf add state */
+	XFS_DAS_SF_REMOVE,		/* Initial sf replace/remove state */
+
+	XFS_DAS_LEAF_ADD,		/* Initial leaf add state */
+	XFS_DAS_LEAF_REMOVE,		/* Initial leaf replace/remove state */
+
+	XFS_DAS_NODE_ADD,		/* Initial node add state */
+	XFS_DAS_NODE_REMOVE,		/* Initial node replace/remove state */
+
+	/* Leaf state set/replace/remove sequence */
+	XFS_DAS_LEAF_SET_RMT,		/* set a remote xattr from a leaf */
+	XFS_DAS_LEAF_ALLOC_RMT,		/* We are allocating remote blocks */
+	XFS_DAS_LEAF_REPLACE,		/* Perform replace ops on a leaf */
+	XFS_DAS_LEAF_REMOVE_OLD,	/* Start removing old attr from leaf */
+	XFS_DAS_LEAF_REMOVE_RMT,	/* A rename is removing remote blocks */
+	XFS_DAS_LEAF_REMOVE_ATTR,	/* Remove the old attr from a leaf */
+
+	/* Node state sequence, must match leaf state above */
+	XFS_DAS_NODE_SET_RMT,		/* set a remote xattr from a node */
+	XFS_DAS_NODE_ALLOC_RMT,		/* We are allocating remote blocks */
+	XFS_DAS_NODE_REPLACE,		/* Perform replace ops on a node */
+	XFS_DAS_NODE_REMOVE_OLD,	/* Start removing old attr from node */
+	XFS_DAS_NODE_REMOVE_RMT,	/* A rename is removing remote blocks */
+	XFS_DAS_NODE_REMOVE_ATTR,	/* Remove the old attr from a node */
+
+	XFS_DAS_DONE,			/* finished operation */
+};
+
+#define XFS_DAS_STRINGS	\
+	{ XFS_DAS_UNINIT,		"XFS_DAS_UNINIT" }, \
+	{ XFS_DAS_SF_ADD,		"XFS_DAS_SF_ADD" }, \
+	{ XFS_DAS_SF_REMOVE,		"XFS_DAS_SF_REMOVE" }, \
+	{ XFS_DAS_LEAF_ADD,		"XFS_DAS_LEAF_ADD" }, \
+	{ XFS_DAS_LEAF_REMOVE,		"XFS_DAS_LEAF_REMOVE" }, \
+	{ XFS_DAS_NODE_ADD,		"XFS_DAS_NODE_ADD" }, \
+	{ XFS_DAS_NODE_REMOVE,		"XFS_DAS_NODE_REMOVE" }, \
+	{ XFS_DAS_LEAF_SET_RMT,		"XFS_DAS_LEAF_SET_RMT" }, \
+	{ XFS_DAS_LEAF_ALLOC_RMT,	"XFS_DAS_LEAF_ALLOC_RMT" }, \
+	{ XFS_DAS_LEAF_REPLACE,		"XFS_DAS_LEAF_REPLACE" }, \
+	{ XFS_DAS_LEAF_REMOVE_OLD,	"XFS_DAS_LEAF_REMOVE_OLD" }, \
+	{ XFS_DAS_LEAF_REMOVE_RMT,	"XFS_DAS_LEAF_REMOVE_RMT" }, \
+	{ XFS_DAS_LEAF_REMOVE_ATTR,	"XFS_DAS_LEAF_REMOVE_ATTR" }, \
+	{ XFS_DAS_NODE_SET_RMT,		"XFS_DAS_NODE_SET_RMT" }, \
+	{ XFS_DAS_NODE_ALLOC_RMT,	"XFS_DAS_NODE_ALLOC_RMT" },  \
+	{ XFS_DAS_NODE_REPLACE,		"XFS_DAS_NODE_REPLACE" },  \
+	{ XFS_DAS_NODE_REMOVE_OLD,	"XFS_DAS_NODE_REMOVE_OLD" }, \
+	{ XFS_DAS_NODE_REMOVE_RMT,	"XFS_DAS_NODE_REMOVE_RMT" }, \
+	{ XFS_DAS_NODE_REMOVE_ATTR,	"XFS_DAS_NODE_REMOVE_ATTR" }, \
+	{ XFS_DAS_DONE,			"XFS_DAS_DONE" }
+
+struct xfs_attri_log_nameval;
+
+/*
+ * Context used for keeping track of delayed attribute operations
+ */
+struct xfs_attr_intent {
+	/*
+	 * used to log this item to an intent containing a list of attrs to
+	 * commit later
+	 */
+	struct list_head		xattri_list;
+
+	/* Used in xfs_attr_node_removename to roll through removing blocks */
+	struct xfs_da_state		*xattri_da_state;
+
+	struct xfs_da_args		*xattri_da_args;
+
+	/*
+	 * Shared buffer containing the attr name and value so that the logging
+	 * code can share large memory buffers between log items.
+	 */
+	struct xfs_attri_log_nameval	*xattri_nameval;
+
+	/* Used to keep track of current state of delayed operation */
+	enum xfs_delattr_state		xattri_dela_state;
+
+	/*
+	 * Attr operation being performed - XFS_ATTRI_OP_FLAGS_*
+	 */
+	unsigned int			xattri_op_flags;
+
+	/* Used in xfs_attr_rmtval_set_blk to roll through allocating blocks */
+	xfs_dablk_t			xattri_lblkno;
+	int				xattri_blkcnt;
+	struct xfs_bmbt_irec		xattri_map;
+};
+
+
+/*========================================================================
+ * Function prototypes for the kernel.
+ *========================================================================*/
+
+/*
+ * Overall external interface routines.
+ */
+int xfs_attr_inactive(struct xfs_inode *dp);
+int xfs_attr_list_ilocked(struct xfs_attr_list_context *);
+int xfs_attr_list(struct xfs_attr_list_context *);
+int xfs_inode_hasattr(struct xfs_inode *ip);
+bool xfs_attr_is_leaf(struct xfs_inode *ip);
+int xfs_attr_get_ilocked(struct xfs_da_args *args);
+int xfs_attr_get(struct xfs_da_args *args);
+int xfs_attr_set(struct xfs_da_args *args);
+int xfs_attr_set_iter(struct xfs_attr_intent *attr);
+int xfs_attr_remove_iter(struct xfs_attr_intent *attr);
+bool xfs_attr_namecheck(const void *name, size_t length);
+int xfs_attr_calc_size(struct xfs_da_args *args, int *local);
+void xfs_init_attr_trans(struct xfs_da_args *args, struct xfs_trans_res *tres,
+			 unsigned int *total);
+
+/*
+ * Check to see if the attr should be upgraded from non-existent or shortform to
+ * single-leaf-block attribute list.
+ */
+static inline bool
+xfs_attr_is_shortform(
+	struct xfs_inode    *ip)
+{
+	return ip->i_af.if_format == XFS_DINODE_FMT_LOCAL ||
+	       (ip->i_af.if_format == XFS_DINODE_FMT_EXTENTS &&
+		ip->i_af.if_nextents == 0);
+}
+
+static inline enum xfs_delattr_state
+xfs_attr_init_add_state(struct xfs_da_args *args)
+{
+	/*
+	 * When called from the completion of a attr remove to determine the
+	 * next state, the attribute fork may be null. This can occur only occur
+	 * on a pure remove, but we grab the next state before we check if a
+	 * replace operation is being performed. If we are called from any other
+	 * context, i_af is guaranteed to exist. Hence if the attr fork is
+	 * null, we were called from a pure remove operation and so we are done.
+	 */
+	if (!xfs_inode_has_attr_fork(args->dp))
+		return XFS_DAS_DONE;
+
+	args->op_flags |= XFS_DA_OP_ADDNAME;
+	if (xfs_attr_is_shortform(args->dp))
+		return XFS_DAS_SF_ADD;
+	if (xfs_attr_is_leaf(args->dp))
+		return XFS_DAS_LEAF_ADD;
+	return XFS_DAS_NODE_ADD;
+}
+
+static inline enum xfs_delattr_state
+xfs_attr_init_remove_state(struct xfs_da_args *args)
+{
+	args->op_flags |= XFS_DA_OP_REMOVE;
+	if (xfs_attr_is_shortform(args->dp))
+		return XFS_DAS_SF_REMOVE;
+	if (xfs_attr_is_leaf(args->dp))
+		return XFS_DAS_LEAF_REMOVE;
+	return XFS_DAS_NODE_REMOVE;
+}
+
+/*
+ * If we are logging the attributes, then we have to start with removal of the
+ * old attribute so that there is always consistent state that we can recover
+ * from if the system goes down part way through. We always log the new attr
+ * value, so even when we remove the attr first we still have the information in
+ * the log to finish the replace operation atomically.
+ */
+static inline enum xfs_delattr_state
+xfs_attr_init_replace_state(struct xfs_da_args *args)
+{
+	args->op_flags |= XFS_DA_OP_ADDNAME | XFS_DA_OP_REPLACE;
+	if (args->op_flags & XFS_DA_OP_LOGGED)
+		return xfs_attr_init_remove_state(args);
+	return xfs_attr_init_add_state(args);
+}
+
+extern struct kmem_cache *xfs_attr_intent_cache;
+int __init xfs_attr_intent_init_cache(void);
+void xfs_attr_intent_destroy_cache(void);
+
+#endif	/* __XFS_ATTR_H__ */
diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c
new file mode 100644
index 000000000..beee51ad7
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_attr_leaf.c
@@ -0,0 +1,3002 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_sb.h"
+#include "xfs_mount.h"
+#include "xfs_da_format.h"
+#include "xfs_da_btree.h"
+#include "xfs_inode.h"
+#include "xfs_trans.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_bmap.h"
+#include "xfs_attr_sf.h"
+#include "xfs_attr.h"
+#include "xfs_attr_remote.h"
+#include "xfs_attr_leaf.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+#include "xfs_buf_item.h"
+#include "xfs_dir2.h"
+#include "xfs_log.h"
+#include "xfs_ag.h"
+#include "xfs_errortag.h"
+
+
+/*
+ * xfs_attr_leaf.c
+ *
+ * Routines to implement leaf blocks of attributes as Btrees of hashed names.
+ */
+
+/*========================================================================
+ * Function prototypes for the kernel.
+ *========================================================================*/
+
+/*
+ * Routines used for growing the Btree.
+ */
+STATIC int xfs_attr3_leaf_create(struct xfs_da_args *args,
+				 xfs_dablk_t which_block, struct xfs_buf **bpp);
+STATIC int xfs_attr3_leaf_add_work(struct xfs_buf *leaf_buffer,
+				   struct xfs_attr3_icleaf_hdr *ichdr,
+				   struct xfs_da_args *args, int freemap_index);
+STATIC void xfs_attr3_leaf_compact(struct xfs_da_args *args,
+				   struct xfs_attr3_icleaf_hdr *ichdr,
+				   struct xfs_buf *leaf_buffer);
+STATIC void xfs_attr3_leaf_rebalance(xfs_da_state_t *state,
+						   xfs_da_state_blk_t *blk1,
+						   xfs_da_state_blk_t *blk2);
+STATIC int xfs_attr3_leaf_figure_balance(xfs_da_state_t *state,
+			xfs_da_state_blk_t *leaf_blk_1,
+			struct xfs_attr3_icleaf_hdr *ichdr1,
+			xfs_da_state_blk_t *leaf_blk_2,
+			struct xfs_attr3_icleaf_hdr *ichdr2,
+			int *number_entries_in_blk1,
+			int *number_usedbytes_in_blk1);
+
+/*
+ * Utility routines.
+ */
+STATIC void xfs_attr3_leaf_moveents(struct xfs_da_args *args,
+			struct xfs_attr_leafblock *src_leaf,
+			struct xfs_attr3_icleaf_hdr *src_ichdr, int src_start,
+			struct xfs_attr_leafblock *dst_leaf,
+			struct xfs_attr3_icleaf_hdr *dst_ichdr, int dst_start,
+			int move_count);
+STATIC int xfs_attr_leaf_entsize(xfs_attr_leafblock_t *leaf, int index);
+
+/*
+ * attr3 block 'firstused' conversion helpers.
+ *
+ * firstused refers to the offset of the first used byte of the nameval region
+ * of an attr leaf block. The region starts at the tail of the block and expands
+ * backwards towards the middle. As such, firstused is initialized to the block
+ * size for an empty leaf block and is reduced from there.
+ *
+ * The attr3 block size is pegged to the fsb size and the maximum fsb is 64k.
+ * The in-core firstused field is 32-bit and thus supports the maximum fsb size.
+ * The on-disk field is only 16-bit, however, and overflows at 64k. Since this
+ * only occurs at exactly 64k, we use zero as a magic on-disk value to represent
+ * the attr block size. The following helpers manage the conversion between the
+ * in-core and on-disk formats.
+ */
+
+static void
+xfs_attr3_leaf_firstused_from_disk(
+	struct xfs_da_geometry		*geo,
+	struct xfs_attr3_icleaf_hdr	*to,
+	struct xfs_attr_leafblock	*from)
+{
+	struct xfs_attr3_leaf_hdr	*hdr3;
+
+	if (from->hdr.info.magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC)) {
+		hdr3 = (struct xfs_attr3_leaf_hdr *) from;
+		to->firstused = be16_to_cpu(hdr3->firstused);
+	} else {
+		to->firstused = be16_to_cpu(from->hdr.firstused);
+	}
+
+	/*
+	 * Convert from the magic fsb size value to actual blocksize. This
+	 * should only occur for empty blocks when the block size overflows
+	 * 16-bits.
+	 */
+	if (to->firstused == XFS_ATTR3_LEAF_NULLOFF) {
+		ASSERT(!to->count && !to->usedbytes);
+		ASSERT(geo->blksize > USHRT_MAX);
+		to->firstused = geo->blksize;
+	}
+}
+
+static void
+xfs_attr3_leaf_firstused_to_disk(
+	struct xfs_da_geometry		*geo,
+	struct xfs_attr_leafblock	*to,
+	struct xfs_attr3_icleaf_hdr	*from)
+{
+	struct xfs_attr3_leaf_hdr	*hdr3;
+	uint32_t			firstused;
+
+	/* magic value should only be seen on disk */
+	ASSERT(from->firstused != XFS_ATTR3_LEAF_NULLOFF);
+
+	/*
+	 * Scale down the 32-bit in-core firstused value to the 16-bit on-disk
+	 * value. This only overflows at the max supported value of 64k. Use the
+	 * magic on-disk value to represent block size in this case.
+	 */
+	firstused = from->firstused;
+	if (firstused > USHRT_MAX) {
+		ASSERT(from->firstused == geo->blksize);
+		firstused = XFS_ATTR3_LEAF_NULLOFF;
+	}
+
+	if (from->magic == XFS_ATTR3_LEAF_MAGIC) {
+		hdr3 = (struct xfs_attr3_leaf_hdr *) to;
+		hdr3->firstused = cpu_to_be16(firstused);
+	} else {
+		to->hdr.firstused = cpu_to_be16(firstused);
+	}
+}
+
+void
+xfs_attr3_leaf_hdr_from_disk(
+	struct xfs_da_geometry		*geo,
+	struct xfs_attr3_icleaf_hdr	*to,
+	struct xfs_attr_leafblock	*from)
+{
+	int	i;
+
+	ASSERT(from->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC) ||
+	       from->hdr.info.magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC));
+
+	if (from->hdr.info.magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC)) {
+		struct xfs_attr3_leaf_hdr *hdr3 = (struct xfs_attr3_leaf_hdr *)from;
+
+		to->forw = be32_to_cpu(hdr3->info.hdr.forw);
+		to->back = be32_to_cpu(hdr3->info.hdr.back);
+		to->magic = be16_to_cpu(hdr3->info.hdr.magic);
+		to->count = be16_to_cpu(hdr3->count);
+		to->usedbytes = be16_to_cpu(hdr3->usedbytes);
+		xfs_attr3_leaf_firstused_from_disk(geo, to, from);
+		to->holes = hdr3->holes;
+
+		for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; i++) {
+			to->freemap[i].base = be16_to_cpu(hdr3->freemap[i].base);
+			to->freemap[i].size = be16_to_cpu(hdr3->freemap[i].size);
+		}
+		return;
+	}
+	to->forw = be32_to_cpu(from->hdr.info.forw);
+	to->back = be32_to_cpu(from->hdr.info.back);
+	to->magic = be16_to_cpu(from->hdr.info.magic);
+	to->count = be16_to_cpu(from->hdr.count);
+	to->usedbytes = be16_to_cpu(from->hdr.usedbytes);
+	xfs_attr3_leaf_firstused_from_disk(geo, to, from);
+	to->holes = from->hdr.holes;
+
+	for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; i++) {
+		to->freemap[i].base = be16_to_cpu(from->hdr.freemap[i].base);
+		to->freemap[i].size = be16_to_cpu(from->hdr.freemap[i].size);
+	}
+}
+
+void
+xfs_attr3_leaf_hdr_to_disk(
+	struct xfs_da_geometry		*geo,
+	struct xfs_attr_leafblock	*to,
+	struct xfs_attr3_icleaf_hdr	*from)
+{
+	int				i;
+
+	ASSERT(from->magic == XFS_ATTR_LEAF_MAGIC ||
+	       from->magic == XFS_ATTR3_LEAF_MAGIC);
+
+	if (from->magic == XFS_ATTR3_LEAF_MAGIC) {
+		struct xfs_attr3_leaf_hdr *hdr3 = (struct xfs_attr3_leaf_hdr *)to;
+
+		hdr3->info.hdr.forw = cpu_to_be32(from->forw);
+		hdr3->info.hdr.back = cpu_to_be32(from->back);
+		hdr3->info.hdr.magic = cpu_to_be16(from->magic);
+		hdr3->count = cpu_to_be16(from->count);
+		hdr3->usedbytes = cpu_to_be16(from->usedbytes);
+		xfs_attr3_leaf_firstused_to_disk(geo, to, from);
+		hdr3->holes = from->holes;
+		hdr3->pad1 = 0;
+
+		for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; i++) {
+			hdr3->freemap[i].base = cpu_to_be16(from->freemap[i].base);
+			hdr3->freemap[i].size = cpu_to_be16(from->freemap[i].size);
+		}
+		return;
+	}
+	to->hdr.info.forw = cpu_to_be32(from->forw);
+	to->hdr.info.back = cpu_to_be32(from->back);
+	to->hdr.info.magic = cpu_to_be16(from->magic);
+	to->hdr.count = cpu_to_be16(from->count);
+	to->hdr.usedbytes = cpu_to_be16(from->usedbytes);
+	xfs_attr3_leaf_firstused_to_disk(geo, to, from);
+	to->hdr.holes = from->holes;
+	to->hdr.pad1 = 0;
+
+	for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; i++) {
+		to->hdr.freemap[i].base = cpu_to_be16(from->freemap[i].base);
+		to->hdr.freemap[i].size = cpu_to_be16(from->freemap[i].size);
+	}
+}
+
+static xfs_failaddr_t
+xfs_attr3_leaf_verify_entry(
+	struct xfs_mount			*mp,
+	char					*buf_end,
+	struct xfs_attr_leafblock		*leaf,
+	struct xfs_attr3_icleaf_hdr		*leafhdr,
+	struct xfs_attr_leaf_entry		*ent,
+	int					idx,
+	__u32					*last_hashval)
+{
+	struct xfs_attr_leaf_name_local		*lentry;
+	struct xfs_attr_leaf_name_remote	*rentry;
+	char					*name_end;
+	unsigned int				nameidx;
+	unsigned int				namesize;
+	__u32					hashval;
+
+	/* hash order check */
+	hashval = be32_to_cpu(ent->hashval);
+	if (hashval < *last_hashval)
+		return __this_address;
+	*last_hashval = hashval;
+
+	nameidx = be16_to_cpu(ent->nameidx);
+	if (nameidx < leafhdr->firstused || nameidx >= mp->m_attr_geo->blksize)
+		return __this_address;
+
+	/*
+	 * Check the name information.  The namelen fields are u8 so we can't
+	 * possibly exceed the maximum name length of 255 bytes.
+	 */
+	if (ent->flags & XFS_ATTR_LOCAL) {
+		lentry = xfs_attr3_leaf_name_local(leaf, idx);
+		namesize = xfs_attr_leaf_entsize_local(lentry->namelen,
+				be16_to_cpu(lentry->valuelen));
+		name_end = (char *)lentry + namesize;
+		if (lentry->namelen == 0)
+			return __this_address;
+	} else {
+		rentry = xfs_attr3_leaf_name_remote(leaf, idx);
+		namesize = xfs_attr_leaf_entsize_remote(rentry->namelen);
+		name_end = (char *)rentry + namesize;
+		if (rentry->namelen == 0)
+			return __this_address;
+		if (!(ent->flags & XFS_ATTR_INCOMPLETE) &&
+		    rentry->valueblk == 0)
+			return __this_address;
+	}
+
+	if (name_end > buf_end)
+		return __this_address;
+
+	return NULL;
+}
+
+/*
+ * Validate an attribute leaf block.
+ *
+ * Empty leaf blocks can occur under the following circumstances:
+ *
+ * 1. setxattr adds a new extended attribute to a file;
+ * 2. The file has zero existing attributes;
+ * 3. The attribute is too large to fit in the attribute fork;
+ * 4. The attribute is small enough to fit in a leaf block;
+ * 5. A log flush occurs after committing the transaction that creates
+ *    the (empty) leaf block; and
+ * 6. The filesystem goes down after the log flush but before the new
+ *    attribute can be committed to the leaf block.
+ *
+ * Hence we need to ensure that we don't fail the validation purely
+ * because the leaf is empty.
+ */
+static xfs_failaddr_t
+xfs_attr3_leaf_verify(
+	struct xfs_buf			*bp)
+{
+	struct xfs_attr3_icleaf_hdr	ichdr;
+	struct xfs_mount		*mp = bp->b_mount;
+	struct xfs_attr_leafblock	*leaf = bp->b_addr;
+	struct xfs_attr_leaf_entry	*entries;
+	struct xfs_attr_leaf_entry	*ent;
+	char				*buf_end;
+	uint32_t			end;	/* must be 32bit - see below */
+	__u32				last_hashval = 0;
+	int				i;
+	xfs_failaddr_t			fa;
+
+	xfs_attr3_leaf_hdr_from_disk(mp->m_attr_geo, &ichdr, leaf);
+
+	fa = xfs_da3_blkinfo_verify(bp, bp->b_addr);
+	if (fa)
+		return fa;
+
+	/*
+	 * firstused is the block offset of the first name info structure.
+	 * Make sure it doesn't go off the block or crash into the header.
+	 */
+	if (ichdr.firstused > mp->m_attr_geo->blksize)
+		return __this_address;
+	if (ichdr.firstused < xfs_attr3_leaf_hdr_size(leaf))
+		return __this_address;
+
+	/* Make sure the entries array doesn't crash into the name info. */
+	entries = xfs_attr3_leaf_entryp(bp->b_addr);
+	if ((char *)&entries[ichdr.count] >
+	    (char *)bp->b_addr + ichdr.firstused)
+		return __this_address;
+
+	/*
+	 * NOTE: This verifier historically failed empty leaf buffers because
+	 * we expect the fork to be in another format. Empty attr fork format
+	 * conversions are possible during xattr set, however, and format
+	 * conversion is not atomic with the xattr set that triggers it. We
+	 * cannot assume leaf blocks are non-empty until that is addressed.
+	*/
+	buf_end = (char *)bp->b_addr + mp->m_attr_geo->blksize;
+	for (i = 0, ent = entries; i < ichdr.count; ent++, i++) {
+		fa = xfs_attr3_leaf_verify_entry(mp, buf_end, leaf, &ichdr,
+				ent, i, &last_hashval);
+		if (fa)
+			return fa;
+	}
+
+	/*
+	 * Quickly check the freemap information.  Attribute data has to be
+	 * aligned to 4-byte boundaries, and likewise for the free space.
+	 *
+	 * Note that for 64k block size filesystems, the freemap entries cannot
+	 * overflow as they are only be16 fields. However, when checking end
+	 * pointer of the freemap, we have to be careful to detect overflows and
+	 * so use uint32_t for those checks.
+	 */
+	for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; i++) {
+		if (ichdr.freemap[i].base > mp->m_attr_geo->blksize)
+			return __this_address;
+		if (ichdr.freemap[i].base & 0x3)
+			return __this_address;
+		if (ichdr.freemap[i].size > mp->m_attr_geo->blksize)
+			return __this_address;
+		if (ichdr.freemap[i].size & 0x3)
+			return __this_address;
+
+		/* be care of 16 bit overflows here */
+		end = (uint32_t)ichdr.freemap[i].base + ichdr.freemap[i].size;
+		if (end < ichdr.freemap[i].base)
+			return __this_address;
+		if (end > mp->m_attr_geo->blksize)
+			return __this_address;
+	}
+
+	return NULL;
+}
+
+static void
+xfs_attr3_leaf_write_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_buf_log_item	*bip = bp->b_log_item;
+	struct xfs_attr3_leaf_hdr *hdr3 = bp->b_addr;
+	xfs_failaddr_t		fa;
+
+	fa = xfs_attr3_leaf_verify(bp);
+	if (fa) {
+		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+		return;
+	}
+
+	if (!xfs_has_crc(mp))
+		return;
+
+	if (bip)
+		hdr3->info.lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+	xfs_buf_update_cksum(bp, XFS_ATTR3_LEAF_CRC_OFF);
+}
+
+/*
+ * leaf/node format detection on trees is sketchy, so a node read can be done on
+ * leaf level blocks when detection identifies the tree as a node format tree
+ * incorrectly. In this case, we need to swap the verifier to match the correct
+ * format of the block being read.
+ */
+static void
+xfs_attr3_leaf_read_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	xfs_failaddr_t		fa;
+
+	if (xfs_has_crc(mp) &&
+	     !xfs_buf_verify_cksum(bp, XFS_ATTR3_LEAF_CRC_OFF))
+		xfs_verifier_error(bp, -EFSBADCRC, __this_address);
+	else {
+		fa = xfs_attr3_leaf_verify(bp);
+		if (fa)
+			xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+	}
+}
+
+const struct xfs_buf_ops xfs_attr3_leaf_buf_ops = {
+	.name = "xfs_attr3_leaf",
+	.magic16 = { cpu_to_be16(XFS_ATTR_LEAF_MAGIC),
+		     cpu_to_be16(XFS_ATTR3_LEAF_MAGIC) },
+	.verify_read = xfs_attr3_leaf_read_verify,
+	.verify_write = xfs_attr3_leaf_write_verify,
+	.verify_struct = xfs_attr3_leaf_verify,
+};
+
+int
+xfs_attr3_leaf_read(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*dp,
+	xfs_dablk_t		bno,
+	struct xfs_buf		**bpp)
+{
+	int			err;
+
+	err = xfs_da_read_buf(tp, dp, bno, 0, bpp, XFS_ATTR_FORK,
+			&xfs_attr3_leaf_buf_ops);
+	if (!err && tp && *bpp)
+		xfs_trans_buf_set_type(tp, *bpp, XFS_BLFT_ATTR_LEAF_BUF);
+	return err;
+}
+
+/*========================================================================
+ * Namespace helper routines
+ *========================================================================*/
+
+/*
+ * If we are in log recovery, then we want the lookup to ignore the INCOMPLETE
+ * flag on disk - if there's an incomplete attr then recovery needs to tear it
+ * down. If there's no incomplete attr, then recovery needs to tear that attr
+ * down to replace it with the attr that has been logged. In this case, the
+ * INCOMPLETE flag will not be set in attr->attr_filter, but rather
+ * XFS_DA_OP_RECOVERY will be set in args->op_flags.
+ */
+static bool
+xfs_attr_match(
+	struct xfs_da_args	*args,
+	uint8_t			namelen,
+	unsigned char		*name,
+	int			flags)
+{
+
+	if (args->namelen != namelen)
+		return false;
+	if (memcmp(args->name, name, namelen) != 0)
+		return false;
+
+	/* Recovery ignores the INCOMPLETE flag. */
+	if ((args->op_flags & XFS_DA_OP_RECOVERY) &&
+	    args->attr_filter == (flags & XFS_ATTR_NSP_ONDISK_MASK))
+		return true;
+
+	/* All remaining matches need to be filtered by INCOMPLETE state. */
+	if (args->attr_filter !=
+	    (flags & (XFS_ATTR_NSP_ONDISK_MASK | XFS_ATTR_INCOMPLETE)))
+		return false;
+	return true;
+}
+
+static int
+xfs_attr_copy_value(
+	struct xfs_da_args	*args,
+	unsigned char		*value,
+	int			valuelen)
+{
+	/*
+	 * No copy if all we have to do is get the length
+	 */
+	if (!args->valuelen) {
+		args->valuelen = valuelen;
+		return 0;
+	}
+
+	/*
+	 * No copy if the length of the existing buffer is too small
+	 */
+	if (args->valuelen < valuelen) {
+		args->valuelen = valuelen;
+		return -ERANGE;
+	}
+
+	if (!args->value) {
+		args->value = kvmalloc(valuelen, GFP_KERNEL | __GFP_NOLOCKDEP);
+		if (!args->value)
+			return -ENOMEM;
+	}
+	args->valuelen = valuelen;
+
+	/* remote block xattr requires IO for copy-in */
+	if (args->rmtblkno)
+		return xfs_attr_rmtval_get(args);
+
+	/*
+	 * This is to prevent a GCC warning because the remote xattr case
+	 * doesn't have a value to pass in. In that case, we never reach here,
+	 * but GCC can't work that out and so throws a "passing NULL to
+	 * memcpy" warning.
+	 */
+	if (!value)
+		return -EINVAL;
+	memcpy(args->value, value, valuelen);
+	return 0;
+}
+
+/*========================================================================
+ * External routines when attribute fork size < XFS_LITINO(mp).
+ *========================================================================*/
+
+/*
+ * Query whether the total requested number of attr fork bytes of extended
+ * attribute space will be able to fit inline.
+ *
+ * Returns zero if not, else the i_forkoff fork offset to be used in the
+ * literal area for attribute data once the new bytes have been added.
+ *
+ * i_forkoff must be 8 byte aligned, hence is stored as a >>3 value;
+ * special case for dev/uuid inodes, they have fixed size data forks.
+ */
+int
+xfs_attr_shortform_bytesfit(
+	struct xfs_inode	*dp,
+	int			bytes)
+{
+	struct xfs_mount	*mp = dp->i_mount;
+	int64_t			dsize;
+	int			minforkoff;
+	int			maxforkoff;
+	int			offset;
+
+	/*
+	 * Check if the new size could fit at all first:
+	 */
+	if (bytes > XFS_LITINO(mp))
+		return 0;
+
+	/* rounded down */
+	offset = (XFS_LITINO(mp) - bytes) >> 3;
+
+	if (dp->i_df.if_format == XFS_DINODE_FMT_DEV) {
+		minforkoff = roundup(sizeof(xfs_dev_t), 8) >> 3;
+		return (offset >= minforkoff) ? minforkoff : 0;
+	}
+
+	/*
+	 * If the requested numbers of bytes is smaller or equal to the
+	 * current attribute fork size we can always proceed.
+	 *
+	 * Note that if_bytes in the data fork might actually be larger than
+	 * the current data fork size is due to delalloc extents. In that
+	 * case either the extent count will go down when they are converted
+	 * to real extents, or the delalloc conversion will take care of the
+	 * literal area rebalancing.
+	 */
+	if (bytes <= xfs_inode_attr_fork_size(dp))
+		return dp->i_forkoff;
+
+	/*
+	 * For attr2 we can try to move the forkoff if there is space in the
+	 * literal area, but for the old format we are done if there is no
+	 * space in the fixed attribute fork.
+	 */
+	if (!xfs_has_attr2(mp))
+		return 0;
+
+	dsize = dp->i_df.if_bytes;
+
+	switch (dp->i_df.if_format) {
+	case XFS_DINODE_FMT_EXTENTS:
+		/*
+		 * If there is no attr fork and the data fork is extents,
+		 * determine if creating the default attr fork will result
+		 * in the extents form migrating to btree. If so, the
+		 * minimum offset only needs to be the space required for
+		 * the btree root.
+		 */
+		if (!dp->i_forkoff && dp->i_df.if_bytes >
+		    xfs_default_attroffset(dp))
+			dsize = XFS_BMDR_SPACE_CALC(MINDBTPTRS);
+		break;
+	case XFS_DINODE_FMT_BTREE:
+		/*
+		 * If we have a data btree then keep forkoff if we have one,
+		 * otherwise we are adding a new attr, so then we set
+		 * minforkoff to where the btree root can finish so we have
+		 * plenty of room for attrs
+		 */
+		if (dp->i_forkoff) {
+			if (offset < dp->i_forkoff)
+				return 0;
+			return dp->i_forkoff;
+		}
+		dsize = XFS_BMAP_BROOT_SPACE(mp, dp->i_df.if_broot);
+		break;
+	}
+
+	/*
+	 * A data fork btree root must have space for at least
+	 * MINDBTPTRS key/ptr pairs if the data fork is small or empty.
+	 */
+	minforkoff = max_t(int64_t, dsize, XFS_BMDR_SPACE_CALC(MINDBTPTRS));
+	minforkoff = roundup(minforkoff, 8) >> 3;
+
+	/* attr fork btree root can have at least this many key/ptr pairs */
+	maxforkoff = XFS_LITINO(mp) - XFS_BMDR_SPACE_CALC(MINABTPTRS);
+	maxforkoff = maxforkoff >> 3;	/* rounded down */
+
+	if (offset >= maxforkoff)
+		return maxforkoff;
+	if (offset >= minforkoff)
+		return offset;
+	return 0;
+}
+
+/*
+ * Switch on the ATTR2 superblock bit (implies also FEATURES2) unless:
+ * - noattr2 mount option is set,
+ * - on-disk version bit says it is already set, or
+ * - the attr2 mount option is not set to enable automatic upgrade from attr1.
+ */
+STATIC void
+xfs_sbversion_add_attr2(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp)
+{
+	if (xfs_has_noattr2(mp))
+		return;
+	if (mp->m_sb.sb_features2 & XFS_SB_VERSION2_ATTR2BIT)
+		return;
+	if (!xfs_has_attr2(mp))
+		return;
+
+	spin_lock(&mp->m_sb_lock);
+	xfs_add_attr2(mp);
+	spin_unlock(&mp->m_sb_lock);
+	xfs_log_sb(tp);
+}
+
+/*
+ * Create the initial contents of a shortform attribute list.
+ */
+void
+xfs_attr_shortform_create(
+	struct xfs_da_args	*args)
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_ifork	*ifp = &dp->i_af;
+	struct xfs_attr_sf_hdr	*hdr;
+
+	trace_xfs_attr_sf_create(args);
+
+	ASSERT(ifp->if_bytes == 0);
+	if (ifp->if_format == XFS_DINODE_FMT_EXTENTS)
+		ifp->if_format = XFS_DINODE_FMT_LOCAL;
+	xfs_idata_realloc(dp, sizeof(*hdr), XFS_ATTR_FORK);
+	hdr = (struct xfs_attr_sf_hdr *)ifp->if_u1.if_data;
+	memset(hdr, 0, sizeof(*hdr));
+	hdr->totsize = cpu_to_be16(sizeof(*hdr));
+	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_ADATA);
+}
+
+/*
+ * Return -EEXIST if attr is found, or -ENOATTR if not
+ * args:  args containing attribute name and namelen
+ * sfep:  If not null, pointer will be set to the last attr entry found on
+	  -EEXIST.  On -ENOATTR pointer is left at the last entry in the list
+ * basep: If not null, pointer is set to the byte offset of the entry in the
+ *	  list on -EEXIST.  On -ENOATTR, pointer is left at the byte offset of
+ *	  the last entry in the list
+ */
+int
+xfs_attr_sf_findname(
+	struct xfs_da_args	 *args,
+	struct xfs_attr_sf_entry **sfep,
+	unsigned int		 *basep)
+{
+	struct xfs_attr_shortform *sf;
+	struct xfs_attr_sf_entry *sfe;
+	unsigned int		base = sizeof(struct xfs_attr_sf_hdr);
+	int			size = 0;
+	int			end;
+	int			i;
+
+	sf = (struct xfs_attr_shortform *)args->dp->i_af.if_u1.if_data;
+	sfe = &sf->list[0];
+	end = sf->hdr.count;
+	for (i = 0; i < end; sfe = xfs_attr_sf_nextentry(sfe),
+			     base += size, i++) {
+		size = xfs_attr_sf_entsize(sfe);
+		if (!xfs_attr_match(args, sfe->namelen, sfe->nameval,
+				    sfe->flags))
+			continue;
+		break;
+	}
+
+	if (sfep != NULL)
+		*sfep = sfe;
+
+	if (basep != NULL)
+		*basep = base;
+
+	if (i == end)
+		return -ENOATTR;
+	return -EEXIST;
+}
+
+/*
+ * Add a name/value pair to the shortform attribute list.
+ * Overflow from the inode has already been checked for.
+ */
+void
+xfs_attr_shortform_add(
+	struct xfs_da_args		*args,
+	int				forkoff)
+{
+	struct xfs_attr_shortform	*sf;
+	struct xfs_attr_sf_entry	*sfe;
+	int				offset, size;
+	struct xfs_mount		*mp;
+	struct xfs_inode		*dp;
+	struct xfs_ifork		*ifp;
+
+	trace_xfs_attr_sf_add(args);
+
+	dp = args->dp;
+	mp = dp->i_mount;
+	dp->i_forkoff = forkoff;
+
+	ifp = &dp->i_af;
+	ASSERT(ifp->if_format == XFS_DINODE_FMT_LOCAL);
+	sf = (struct xfs_attr_shortform *)ifp->if_u1.if_data;
+	if (xfs_attr_sf_findname(args, &sfe, NULL) == -EEXIST)
+		ASSERT(0);
+
+	offset = (char *)sfe - (char *)sf;
+	size = xfs_attr_sf_entsize_byname(args->namelen, args->valuelen);
+	xfs_idata_realloc(dp, size, XFS_ATTR_FORK);
+	sf = (struct xfs_attr_shortform *)ifp->if_u1.if_data;
+	sfe = (struct xfs_attr_sf_entry *)((char *)sf + offset);
+
+	sfe->namelen = args->namelen;
+	sfe->valuelen = args->valuelen;
+	sfe->flags = args->attr_filter;
+	memcpy(sfe->nameval, args->name, args->namelen);
+	memcpy(&sfe->nameval[args->namelen], args->value, args->valuelen);
+	sf->hdr.count++;
+	be16_add_cpu(&sf->hdr.totsize, size);
+	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_ADATA);
+
+	xfs_sbversion_add_attr2(mp, args->trans);
+}
+
+/*
+ * After the last attribute is removed revert to original inode format,
+ * making all literal area available to the data fork once more.
+ */
+void
+xfs_attr_fork_remove(
+	struct xfs_inode	*ip,
+	struct xfs_trans	*tp)
+{
+	ASSERT(ip->i_af.if_nextents == 0);
+
+	xfs_ifork_zap_attr(ip);
+	ip->i_forkoff = 0;
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+}
+
+/*
+ * Remove an attribute from the shortform attribute list structure.
+ */
+int
+xfs_attr_sf_removename(
+	struct xfs_da_args		*args)
+{
+	struct xfs_attr_shortform	*sf;
+	struct xfs_attr_sf_entry	*sfe;
+	int				size = 0, end, totsize;
+	unsigned int			base;
+	struct xfs_mount		*mp;
+	struct xfs_inode		*dp;
+	int				error;
+
+	trace_xfs_attr_sf_remove(args);
+
+	dp = args->dp;
+	mp = dp->i_mount;
+	sf = (struct xfs_attr_shortform *)dp->i_af.if_u1.if_data;
+
+	error = xfs_attr_sf_findname(args, &sfe, &base);
+
+	/*
+	 * If we are recovering an operation, finding nothing to
+	 * remove is not an error - it just means there was nothing
+	 * to clean up.
+	 */
+	if (error == -ENOATTR && (args->op_flags & XFS_DA_OP_RECOVERY))
+		return 0;
+	if (error != -EEXIST)
+		return error;
+	size = xfs_attr_sf_entsize(sfe);
+
+	/*
+	 * Fix up the attribute fork data, covering the hole
+	 */
+	end = base + size;
+	totsize = be16_to_cpu(sf->hdr.totsize);
+	if (end != totsize)
+		memmove(&((char *)sf)[base], &((char *)sf)[end], totsize - end);
+	sf->hdr.count--;
+	be16_add_cpu(&sf->hdr.totsize, -size);
+
+	/*
+	 * Fix up the start offset of the attribute fork
+	 */
+	totsize -= size;
+	if (totsize == sizeof(xfs_attr_sf_hdr_t) && xfs_has_attr2(mp) &&
+	    (dp->i_df.if_format != XFS_DINODE_FMT_BTREE) &&
+	    !(args->op_flags & (XFS_DA_OP_ADDNAME | XFS_DA_OP_REPLACE))) {
+		xfs_attr_fork_remove(dp, args->trans);
+	} else {
+		xfs_idata_realloc(dp, -size, XFS_ATTR_FORK);
+		dp->i_forkoff = xfs_attr_shortform_bytesfit(dp, totsize);
+		ASSERT(dp->i_forkoff);
+		ASSERT(totsize > sizeof(xfs_attr_sf_hdr_t) ||
+				(args->op_flags & XFS_DA_OP_ADDNAME) ||
+				!xfs_has_attr2(mp) ||
+				dp->i_df.if_format == XFS_DINODE_FMT_BTREE);
+		xfs_trans_log_inode(args->trans, dp,
+					XFS_ILOG_CORE | XFS_ILOG_ADATA);
+	}
+
+	xfs_sbversion_add_attr2(mp, args->trans);
+
+	return 0;
+}
+
+/*
+ * Look up a name in a shortform attribute list structure.
+ */
+/*ARGSUSED*/
+int
+xfs_attr_shortform_lookup(xfs_da_args_t *args)
+{
+	struct xfs_attr_shortform *sf;
+	struct xfs_attr_sf_entry *sfe;
+	int i;
+	struct xfs_ifork *ifp;
+
+	trace_xfs_attr_sf_lookup(args);
+
+	ifp = &args->dp->i_af;
+	ASSERT(ifp->if_format == XFS_DINODE_FMT_LOCAL);
+	sf = (struct xfs_attr_shortform *)ifp->if_u1.if_data;
+	sfe = &sf->list[0];
+	for (i = 0; i < sf->hdr.count;
+				sfe = xfs_attr_sf_nextentry(sfe), i++) {
+		if (xfs_attr_match(args, sfe->namelen, sfe->nameval,
+				sfe->flags))
+			return -EEXIST;
+	}
+	return -ENOATTR;
+}
+
+/*
+ * Retrieve the attribute value and length.
+ *
+ * If args->valuelen is zero, only the length needs to be returned.  Unlike a
+ * lookup, we only return an error if the attribute does not exist or we can't
+ * retrieve the value.
+ */
+int
+xfs_attr_shortform_getvalue(
+	struct xfs_da_args	*args)
+{
+	struct xfs_attr_shortform *sf;
+	struct xfs_attr_sf_entry *sfe;
+	int			i;
+
+	ASSERT(args->dp->i_af.if_format == XFS_DINODE_FMT_LOCAL);
+	sf = (struct xfs_attr_shortform *)args->dp->i_af.if_u1.if_data;
+	sfe = &sf->list[0];
+	for (i = 0; i < sf->hdr.count;
+				sfe = xfs_attr_sf_nextentry(sfe), i++) {
+		if (xfs_attr_match(args, sfe->namelen, sfe->nameval,
+				sfe->flags))
+			return xfs_attr_copy_value(args,
+				&sfe->nameval[args->namelen], sfe->valuelen);
+	}
+	return -ENOATTR;
+}
+
+/* Convert from using the shortform to the leaf format. */
+int
+xfs_attr_shortform_to_leaf(
+	struct xfs_da_args		*args)
+{
+	struct xfs_inode		*dp;
+	struct xfs_attr_shortform	*sf;
+	struct xfs_attr_sf_entry	*sfe;
+	struct xfs_da_args		nargs;
+	char				*tmpbuffer;
+	int				error, i, size;
+	xfs_dablk_t			blkno;
+	struct xfs_buf			*bp;
+	struct xfs_ifork		*ifp;
+
+	trace_xfs_attr_sf_to_leaf(args);
+
+	dp = args->dp;
+	ifp = &dp->i_af;
+	sf = (struct xfs_attr_shortform *)ifp->if_u1.if_data;
+	size = be16_to_cpu(sf->hdr.totsize);
+	tmpbuffer = kmem_alloc(size, 0);
+	ASSERT(tmpbuffer != NULL);
+	memcpy(tmpbuffer, ifp->if_u1.if_data, size);
+	sf = (struct xfs_attr_shortform *)tmpbuffer;
+
+	xfs_idata_realloc(dp, -size, XFS_ATTR_FORK);
+	xfs_bmap_local_to_extents_empty(args->trans, dp, XFS_ATTR_FORK);
+
+	bp = NULL;
+	error = xfs_da_grow_inode(args, &blkno);
+	if (error)
+		goto out;
+
+	ASSERT(blkno == 0);
+	error = xfs_attr3_leaf_create(args, blkno, &bp);
+	if (error)
+		goto out;
+
+	memset((char *)&nargs, 0, sizeof(nargs));
+	nargs.dp = dp;
+	nargs.geo = args->geo;
+	nargs.total = args->total;
+	nargs.whichfork = XFS_ATTR_FORK;
+	nargs.trans = args->trans;
+	nargs.op_flags = XFS_DA_OP_OKNOENT;
+
+	sfe = &sf->list[0];
+	for (i = 0; i < sf->hdr.count; i++) {
+		nargs.name = sfe->nameval;
+		nargs.namelen = sfe->namelen;
+		nargs.value = &sfe->nameval[nargs.namelen];
+		nargs.valuelen = sfe->valuelen;
+		nargs.hashval = xfs_da_hashname(sfe->nameval,
+						sfe->namelen);
+		nargs.attr_filter = sfe->flags & XFS_ATTR_NSP_ONDISK_MASK;
+		error = xfs_attr3_leaf_lookup_int(bp, &nargs); /* set a->index */
+		ASSERT(error == -ENOATTR);
+		error = xfs_attr3_leaf_add(bp, &nargs);
+		ASSERT(error != -ENOSPC);
+		if (error)
+			goto out;
+		sfe = xfs_attr_sf_nextentry(sfe);
+	}
+	error = 0;
+out:
+	kmem_free(tmpbuffer);
+	return error;
+}
+
+/*
+ * Check a leaf attribute block to see if all the entries would fit into
+ * a shortform attribute list.
+ */
+int
+xfs_attr_shortform_allfit(
+	struct xfs_buf		*bp,
+	struct xfs_inode	*dp)
+{
+	struct xfs_attr_leafblock *leaf;
+	struct xfs_attr_leaf_entry *entry;
+	xfs_attr_leaf_name_local_t *name_loc;
+	struct xfs_attr3_icleaf_hdr leafhdr;
+	int			bytes;
+	int			i;
+	struct xfs_mount	*mp = bp->b_mount;
+
+	leaf = bp->b_addr;
+	xfs_attr3_leaf_hdr_from_disk(mp->m_attr_geo, &leafhdr, leaf);
+	entry = xfs_attr3_leaf_entryp(leaf);
+
+	bytes = sizeof(struct xfs_attr_sf_hdr);
+	for (i = 0; i < leafhdr.count; entry++, i++) {
+		if (entry->flags & XFS_ATTR_INCOMPLETE)
+			continue;		/* don't copy partial entries */
+		if (!(entry->flags & XFS_ATTR_LOCAL))
+			return 0;
+		name_loc = xfs_attr3_leaf_name_local(leaf, i);
+		if (name_loc->namelen >= XFS_ATTR_SF_ENTSIZE_MAX)
+			return 0;
+		if (be16_to_cpu(name_loc->valuelen) >= XFS_ATTR_SF_ENTSIZE_MAX)
+			return 0;
+		bytes += xfs_attr_sf_entsize_byname(name_loc->namelen,
+					be16_to_cpu(name_loc->valuelen));
+	}
+	if (xfs_has_attr2(dp->i_mount) &&
+	    (dp->i_df.if_format != XFS_DINODE_FMT_BTREE) &&
+	    (bytes == sizeof(struct xfs_attr_sf_hdr)))
+		return -1;
+	return xfs_attr_shortform_bytesfit(dp, bytes);
+}
+
+/* Verify the consistency of an inline attribute fork. */
+xfs_failaddr_t
+xfs_attr_shortform_verify(
+	struct xfs_inode		*ip)
+{
+	struct xfs_attr_shortform	*sfp;
+	struct xfs_attr_sf_entry	*sfep;
+	struct xfs_attr_sf_entry	*next_sfep;
+	char				*endp;
+	struct xfs_ifork		*ifp;
+	int				i;
+	int64_t				size;
+
+	ASSERT(ip->i_af.if_format == XFS_DINODE_FMT_LOCAL);
+	ifp = xfs_ifork_ptr(ip, XFS_ATTR_FORK);
+	sfp = (struct xfs_attr_shortform *)ifp->if_u1.if_data;
+	size = ifp->if_bytes;
+
+	/*
+	 * Give up if the attribute is way too short.
+	 */
+	if (size < sizeof(struct xfs_attr_sf_hdr))
+		return __this_address;
+
+	endp = (char *)sfp + size;
+
+	/* Check all reported entries */
+	sfep = &sfp->list[0];
+	for (i = 0; i < sfp->hdr.count; i++) {
+		/*
+		 * struct xfs_attr_sf_entry has a variable length.
+		 * Check the fixed-offset parts of the structure are
+		 * within the data buffer.
+		 * xfs_attr_sf_entry is defined with a 1-byte variable
+		 * array at the end, so we must subtract that off.
+		 */
+		if (((char *)sfep + sizeof(*sfep)) >= endp)
+			return __this_address;
+
+		/* Don't allow names with known bad length. */
+		if (sfep->namelen == 0)
+			return __this_address;
+
+		/*
+		 * Check that the variable-length part of the structure is
+		 * within the data buffer.  The next entry starts after the
+		 * name component, so nextentry is an acceptable test.
+		 */
+		next_sfep = xfs_attr_sf_nextentry(sfep);
+		if ((char *)next_sfep > endp)
+			return __this_address;
+
+		/*
+		 * Check for unknown flags.  Short form doesn't support
+		 * the incomplete or local bits, so we can use the namespace
+		 * mask here.
+		 */
+		if (sfep->flags & ~XFS_ATTR_NSP_ONDISK_MASK)
+			return __this_address;
+
+		/*
+		 * Check for invalid namespace combinations.  We only allow
+		 * one namespace flag per xattr, so we can just count the
+		 * bits (i.e. hweight) here.
+		 */
+		if (hweight8(sfep->flags & XFS_ATTR_NSP_ONDISK_MASK) > 1)
+			return __this_address;
+
+		sfep = next_sfep;
+	}
+	if ((void *)sfep != (void *)endp)
+		return __this_address;
+
+	return NULL;
+}
+
+/*
+ * Convert a leaf attribute list to shortform attribute list
+ */
+int
+xfs_attr3_leaf_to_shortform(
+	struct xfs_buf		*bp,
+	struct xfs_da_args	*args,
+	int			forkoff)
+{
+	struct xfs_attr_leafblock *leaf;
+	struct xfs_attr3_icleaf_hdr ichdr;
+	struct xfs_attr_leaf_entry *entry;
+	struct xfs_attr_leaf_name_local *name_loc;
+	struct xfs_da_args	nargs;
+	struct xfs_inode	*dp = args->dp;
+	char			*tmpbuffer;
+	int			error;
+	int			i;
+
+	trace_xfs_attr_leaf_to_sf(args);
+
+	tmpbuffer = kmem_alloc(args->geo->blksize, 0);
+	if (!tmpbuffer)
+		return -ENOMEM;
+
+	memcpy(tmpbuffer, bp->b_addr, args->geo->blksize);
+
+	leaf = (xfs_attr_leafblock_t *)tmpbuffer;
+	xfs_attr3_leaf_hdr_from_disk(args->geo, &ichdr, leaf);
+	entry = xfs_attr3_leaf_entryp(leaf);
+
+	/* XXX (dgc): buffer is about to be marked stale - why zero it? */
+	memset(bp->b_addr, 0, args->geo->blksize);
+
+	/*
+	 * Clean out the prior contents of the attribute list.
+	 */
+	error = xfs_da_shrink_inode(args, 0, bp);
+	if (error)
+		goto out;
+
+	if (forkoff == -1) {
+		/*
+		 * Don't remove the attr fork if this operation is the first
+		 * part of a attr replace operations. We're going to add a new
+		 * attr immediately, so we need to keep the attr fork around in
+		 * this case.
+		 */
+		if (!(args->op_flags & XFS_DA_OP_REPLACE)) {
+			ASSERT(xfs_has_attr2(dp->i_mount));
+			ASSERT(dp->i_df.if_format != XFS_DINODE_FMT_BTREE);
+			xfs_attr_fork_remove(dp, args->trans);
+		}
+		goto out;
+	}
+
+	xfs_attr_shortform_create(args);
+
+	/*
+	 * Copy the attributes
+	 */
+	memset((char *)&nargs, 0, sizeof(nargs));
+	nargs.geo = args->geo;
+	nargs.dp = dp;
+	nargs.total = args->total;
+	nargs.whichfork = XFS_ATTR_FORK;
+	nargs.trans = args->trans;
+	nargs.op_flags = XFS_DA_OP_OKNOENT;
+
+	for (i = 0; i < ichdr.count; entry++, i++) {
+		if (entry->flags & XFS_ATTR_INCOMPLETE)
+			continue;	/* don't copy partial entries */
+		if (!entry->nameidx)
+			continue;
+		ASSERT(entry->flags & XFS_ATTR_LOCAL);
+		name_loc = xfs_attr3_leaf_name_local(leaf, i);
+		nargs.name = name_loc->nameval;
+		nargs.namelen = name_loc->namelen;
+		nargs.value = &name_loc->nameval[nargs.namelen];
+		nargs.valuelen = be16_to_cpu(name_loc->valuelen);
+		nargs.hashval = be32_to_cpu(entry->hashval);
+		nargs.attr_filter = entry->flags & XFS_ATTR_NSP_ONDISK_MASK;
+		xfs_attr_shortform_add(&nargs, forkoff);
+	}
+	error = 0;
+
+out:
+	kmem_free(tmpbuffer);
+	return error;
+}
+
+/*
+ * Convert from using a single leaf to a root node and a leaf.
+ */
+int
+xfs_attr3_leaf_to_node(
+	struct xfs_da_args	*args)
+{
+	struct xfs_attr_leafblock *leaf;
+	struct xfs_attr3_icleaf_hdr icleafhdr;
+	struct xfs_attr_leaf_entry *entries;
+	struct xfs_da3_icnode_hdr icnodehdr;
+	struct xfs_da_intnode	*node;
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	struct xfs_buf		*bp1 = NULL;
+	struct xfs_buf		*bp2 = NULL;
+	xfs_dablk_t		blkno;
+	int			error;
+
+	trace_xfs_attr_leaf_to_node(args);
+
+	if (XFS_TEST_ERROR(false, mp, XFS_ERRTAG_ATTR_LEAF_TO_NODE)) {
+		error = -EIO;
+		goto out;
+	}
+
+	error = xfs_da_grow_inode(args, &blkno);
+	if (error)
+		goto out;
+	error = xfs_attr3_leaf_read(args->trans, dp, 0, &bp1);
+	if (error)
+		goto out;
+
+	error = xfs_da_get_buf(args->trans, dp, blkno, &bp2, XFS_ATTR_FORK);
+	if (error)
+		goto out;
+
+	/* copy leaf to new buffer, update identifiers */
+	xfs_trans_buf_set_type(args->trans, bp2, XFS_BLFT_ATTR_LEAF_BUF);
+	bp2->b_ops = bp1->b_ops;
+	memcpy(bp2->b_addr, bp1->b_addr, args->geo->blksize);
+	if (xfs_has_crc(mp)) {
+		struct xfs_da3_blkinfo *hdr3 = bp2->b_addr;
+		hdr3->blkno = cpu_to_be64(xfs_buf_daddr(bp2));
+	}
+	xfs_trans_log_buf(args->trans, bp2, 0, args->geo->blksize - 1);
+
+	/*
+	 * Set up the new root node.
+	 */
+	error = xfs_da3_node_create(args, 0, 1, &bp1, XFS_ATTR_FORK);
+	if (error)
+		goto out;
+	node = bp1->b_addr;
+	xfs_da3_node_hdr_from_disk(mp, &icnodehdr, node);
+
+	leaf = bp2->b_addr;
+	xfs_attr3_leaf_hdr_from_disk(args->geo, &icleafhdr, leaf);
+	entries = xfs_attr3_leaf_entryp(leaf);
+
+	/* both on-disk, don't endian-flip twice */
+	icnodehdr.btree[0].hashval = entries[icleafhdr.count - 1].hashval;
+	icnodehdr.btree[0].before = cpu_to_be32(blkno);
+	icnodehdr.count = 1;
+	xfs_da3_node_hdr_to_disk(dp->i_mount, node, &icnodehdr);
+	xfs_trans_log_buf(args->trans, bp1, 0, args->geo->blksize - 1);
+	error = 0;
+out:
+	return error;
+}
+
+/*========================================================================
+ * Routines used for growing the Btree.
+ *========================================================================*/
+
+/*
+ * Create the initial contents of a leaf attribute list
+ * or a leaf in a node attribute list.
+ */
+STATIC int
+xfs_attr3_leaf_create(
+	struct xfs_da_args	*args,
+	xfs_dablk_t		blkno,
+	struct xfs_buf		**bpp)
+{
+	struct xfs_attr_leafblock *leaf;
+	struct xfs_attr3_icleaf_hdr ichdr;
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	struct xfs_buf		*bp;
+	int			error;
+
+	trace_xfs_attr_leaf_create(args);
+
+	error = xfs_da_get_buf(args->trans, args->dp, blkno, &bp,
+					    XFS_ATTR_FORK);
+	if (error)
+		return error;
+	bp->b_ops = &xfs_attr3_leaf_buf_ops;
+	xfs_trans_buf_set_type(args->trans, bp, XFS_BLFT_ATTR_LEAF_BUF);
+	leaf = bp->b_addr;
+	memset(leaf, 0, args->geo->blksize);
+
+	memset(&ichdr, 0, sizeof(ichdr));
+	ichdr.firstused = args->geo->blksize;
+
+	if (xfs_has_crc(mp)) {
+		struct xfs_da3_blkinfo *hdr3 = bp->b_addr;
+
+		ichdr.magic = XFS_ATTR3_LEAF_MAGIC;
+
+		hdr3->blkno = cpu_to_be64(xfs_buf_daddr(bp));
+		hdr3->owner = cpu_to_be64(dp->i_ino);
+		uuid_copy(&hdr3->uuid, &mp->m_sb.sb_meta_uuid);
+
+		ichdr.freemap[0].base = sizeof(struct xfs_attr3_leaf_hdr);
+	} else {
+		ichdr.magic = XFS_ATTR_LEAF_MAGIC;
+		ichdr.freemap[0].base = sizeof(struct xfs_attr_leaf_hdr);
+	}
+	ichdr.freemap[0].size = ichdr.firstused - ichdr.freemap[0].base;
+
+	xfs_attr3_leaf_hdr_to_disk(args->geo, leaf, &ichdr);
+	xfs_trans_log_buf(args->trans, bp, 0, args->geo->blksize - 1);
+
+	*bpp = bp;
+	return 0;
+}
+
+/*
+ * Split the leaf node, rebalance, then add the new entry.
+ */
+int
+xfs_attr3_leaf_split(
+	struct xfs_da_state	*state,
+	struct xfs_da_state_blk	*oldblk,
+	struct xfs_da_state_blk	*newblk)
+{
+	xfs_dablk_t blkno;
+	int error;
+
+	trace_xfs_attr_leaf_split(state->args);
+
+	/*
+	 * Allocate space for a new leaf node.
+	 */
+	ASSERT(oldblk->magic == XFS_ATTR_LEAF_MAGIC);
+	error = xfs_da_grow_inode(state->args, &blkno);
+	if (error)
+		return error;
+	error = xfs_attr3_leaf_create(state->args, blkno, &newblk->bp);
+	if (error)
+		return error;
+	newblk->blkno = blkno;
+	newblk->magic = XFS_ATTR_LEAF_MAGIC;
+
+	/*
+	 * Rebalance the entries across the two leaves.
+	 * NOTE: rebalance() currently depends on the 2nd block being empty.
+	 */
+	xfs_attr3_leaf_rebalance(state, oldblk, newblk);
+	error = xfs_da3_blk_link(state, oldblk, newblk);
+	if (error)
+		return error;
+
+	/*
+	 * Save info on "old" attribute for "atomic rename" ops, leaf_add()
+	 * modifies the index/blkno/rmtblk/rmtblkcnt fields to show the
+	 * "new" attrs info.  Will need the "old" info to remove it later.
+	 *
+	 * Insert the "new" entry in the correct block.
+	 */
+	if (state->inleaf) {
+		trace_xfs_attr_leaf_add_old(state->args);
+		error = xfs_attr3_leaf_add(oldblk->bp, state->args);
+	} else {
+		trace_xfs_attr_leaf_add_new(state->args);
+		error = xfs_attr3_leaf_add(newblk->bp, state->args);
+	}
+
+	/*
+	 * Update last hashval in each block since we added the name.
+	 */
+	oldblk->hashval = xfs_attr_leaf_lasthash(oldblk->bp, NULL);
+	newblk->hashval = xfs_attr_leaf_lasthash(newblk->bp, NULL);
+	return error;
+}
+
+/*
+ * Add a name to the leaf attribute list structure.
+ */
+int
+xfs_attr3_leaf_add(
+	struct xfs_buf		*bp,
+	struct xfs_da_args	*args)
+{
+	struct xfs_attr_leafblock *leaf;
+	struct xfs_attr3_icleaf_hdr ichdr;
+	int			tablesize;
+	int			entsize;
+	int			sum;
+	int			tmp;
+	int			i;
+
+	trace_xfs_attr_leaf_add(args);
+
+	leaf = bp->b_addr;
+	xfs_attr3_leaf_hdr_from_disk(args->geo, &ichdr, leaf);
+	ASSERT(args->index >= 0 && args->index <= ichdr.count);
+	entsize = xfs_attr_leaf_newentsize(args, NULL);
+
+	/*
+	 * Search through freemap for first-fit on new name length.
+	 * (may need to figure in size of entry struct too)
+	 */
+	tablesize = (ichdr.count + 1) * sizeof(xfs_attr_leaf_entry_t)
+					+ xfs_attr3_leaf_hdr_size(leaf);
+	for (sum = 0, i = XFS_ATTR_LEAF_MAPSIZE - 1; i >= 0; i--) {
+		if (tablesize > ichdr.firstused) {
+			sum += ichdr.freemap[i].size;
+			continue;
+		}
+		if (!ichdr.freemap[i].size)
+			continue;	/* no space in this map */
+		tmp = entsize;
+		if (ichdr.freemap[i].base < ichdr.firstused)
+			tmp += sizeof(xfs_attr_leaf_entry_t);
+		if (ichdr.freemap[i].size >= tmp) {
+			tmp = xfs_attr3_leaf_add_work(bp, &ichdr, args, i);
+			goto out_log_hdr;
+		}
+		sum += ichdr.freemap[i].size;
+	}
+
+	/*
+	 * If there are no holes in the address space of the block,
+	 * and we don't have enough freespace, then compaction will do us
+	 * no good and we should just give up.
+	 */
+	if (!ichdr.holes && sum < entsize)
+		return -ENOSPC;
+
+	/*
+	 * Compact the entries to coalesce free space.
+	 * This may change the hdr->count via dropping INCOMPLETE entries.
+	 */
+	xfs_attr3_leaf_compact(args, &ichdr, bp);
+
+	/*
+	 * After compaction, the block is guaranteed to have only one
+	 * free region, in freemap[0].  If it is not big enough, give up.
+	 */
+	if (ichdr.freemap[0].size < (entsize + sizeof(xfs_attr_leaf_entry_t))) {
+		tmp = -ENOSPC;
+		goto out_log_hdr;
+	}
+
+	tmp = xfs_attr3_leaf_add_work(bp, &ichdr, args, 0);
+
+out_log_hdr:
+	xfs_attr3_leaf_hdr_to_disk(args->geo, leaf, &ichdr);
+	xfs_trans_log_buf(args->trans, bp,
+		XFS_DA_LOGRANGE(leaf, &leaf->hdr,
+				xfs_attr3_leaf_hdr_size(leaf)));
+	return tmp;
+}
+
+/*
+ * Add a name to a leaf attribute list structure.
+ */
+STATIC int
+xfs_attr3_leaf_add_work(
+	struct xfs_buf		*bp,
+	struct xfs_attr3_icleaf_hdr *ichdr,
+	struct xfs_da_args	*args,
+	int			mapindex)
+{
+	struct xfs_attr_leafblock *leaf;
+	struct xfs_attr_leaf_entry *entry;
+	struct xfs_attr_leaf_name_local *name_loc;
+	struct xfs_attr_leaf_name_remote *name_rmt;
+	struct xfs_mount	*mp;
+	int			tmp;
+	int			i;
+
+	trace_xfs_attr_leaf_add_work(args);
+
+	leaf = bp->b_addr;
+	ASSERT(mapindex >= 0 && mapindex < XFS_ATTR_LEAF_MAPSIZE);
+	ASSERT(args->index >= 0 && args->index <= ichdr->count);
+
+	/*
+	 * Force open some space in the entry array and fill it in.
+	 */
+	entry = &xfs_attr3_leaf_entryp(leaf)[args->index];
+	if (args->index < ichdr->count) {
+		tmp  = ichdr->count - args->index;
+		tmp *= sizeof(xfs_attr_leaf_entry_t);
+		memmove(entry + 1, entry, tmp);
+		xfs_trans_log_buf(args->trans, bp,
+		    XFS_DA_LOGRANGE(leaf, entry, tmp + sizeof(*entry)));
+	}
+	ichdr->count++;
+
+	/*
+	 * Allocate space for the new string (at the end of the run).
+	 */
+	mp = args->trans->t_mountp;
+	ASSERT(ichdr->freemap[mapindex].base < args->geo->blksize);
+	ASSERT((ichdr->freemap[mapindex].base & 0x3) == 0);
+	ASSERT(ichdr->freemap[mapindex].size >=
+		xfs_attr_leaf_newentsize(args, NULL));
+	ASSERT(ichdr->freemap[mapindex].size < args->geo->blksize);
+	ASSERT((ichdr->freemap[mapindex].size & 0x3) == 0);
+
+	ichdr->freemap[mapindex].size -= xfs_attr_leaf_newentsize(args, &tmp);
+
+	entry->nameidx = cpu_to_be16(ichdr->freemap[mapindex].base +
+				     ichdr->freemap[mapindex].size);
+	entry->hashval = cpu_to_be32(args->hashval);
+	entry->flags = args->attr_filter;
+	if (tmp)
+		entry->flags |= XFS_ATTR_LOCAL;
+	if (args->op_flags & XFS_DA_OP_REPLACE) {
+		if (!(args->op_flags & XFS_DA_OP_LOGGED))
+			entry->flags |= XFS_ATTR_INCOMPLETE;
+		if ((args->blkno2 == args->blkno) &&
+		    (args->index2 <= args->index)) {
+			args->index2++;
+		}
+	}
+	xfs_trans_log_buf(args->trans, bp,
+			  XFS_DA_LOGRANGE(leaf, entry, sizeof(*entry)));
+	ASSERT((args->index == 0) ||
+	       (be32_to_cpu(entry->hashval) >= be32_to_cpu((entry-1)->hashval)));
+	ASSERT((args->index == ichdr->count - 1) ||
+	       (be32_to_cpu(entry->hashval) <= be32_to_cpu((entry+1)->hashval)));
+
+	/*
+	 * For "remote" attribute values, simply note that we need to
+	 * allocate space for the "remote" value.  We can't actually
+	 * allocate the extents in this transaction, and we can't decide
+	 * which blocks they should be as we might allocate more blocks
+	 * as part of this transaction (a split operation for example).
+	 */
+	if (entry->flags & XFS_ATTR_LOCAL) {
+		name_loc = xfs_attr3_leaf_name_local(leaf, args->index);
+		name_loc->namelen = args->namelen;
+		name_loc->valuelen = cpu_to_be16(args->valuelen);
+		memcpy((char *)name_loc->nameval, args->name, args->namelen);
+		memcpy((char *)&name_loc->nameval[args->namelen], args->value,
+				   be16_to_cpu(name_loc->valuelen));
+	} else {
+		name_rmt = xfs_attr3_leaf_name_remote(leaf, args->index);
+		name_rmt->namelen = args->namelen;
+		memcpy((char *)name_rmt->name, args->name, args->namelen);
+		entry->flags |= XFS_ATTR_INCOMPLETE;
+		/* just in case */
+		name_rmt->valuelen = 0;
+		name_rmt->valueblk = 0;
+		args->rmtblkno = 1;
+		args->rmtblkcnt = xfs_attr3_rmt_blocks(mp, args->valuelen);
+		args->rmtvaluelen = args->valuelen;
+	}
+	xfs_trans_log_buf(args->trans, bp,
+	     XFS_DA_LOGRANGE(leaf, xfs_attr3_leaf_name(leaf, args->index),
+				   xfs_attr_leaf_entsize(leaf, args->index)));
+
+	/*
+	 * Update the control info for this leaf node
+	 */
+	if (be16_to_cpu(entry->nameidx) < ichdr->firstused)
+		ichdr->firstused = be16_to_cpu(entry->nameidx);
+
+	ASSERT(ichdr->firstused >= ichdr->count * sizeof(xfs_attr_leaf_entry_t)
+					+ xfs_attr3_leaf_hdr_size(leaf));
+	tmp = (ichdr->count - 1) * sizeof(xfs_attr_leaf_entry_t)
+					+ xfs_attr3_leaf_hdr_size(leaf);
+
+	for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; i++) {
+		if (ichdr->freemap[i].base == tmp) {
+			ichdr->freemap[i].base += sizeof(xfs_attr_leaf_entry_t);
+			ichdr->freemap[i].size -=
+				min_t(uint16_t, ichdr->freemap[i].size,
+						sizeof(xfs_attr_leaf_entry_t));
+		}
+	}
+	ichdr->usedbytes += xfs_attr_leaf_entsize(leaf, args->index);
+	return 0;
+}
+
+/*
+ * Garbage collect a leaf attribute list block by copying it to a new buffer.
+ */
+STATIC void
+xfs_attr3_leaf_compact(
+	struct xfs_da_args	*args,
+	struct xfs_attr3_icleaf_hdr *ichdr_dst,
+	struct xfs_buf		*bp)
+{
+	struct xfs_attr_leafblock *leaf_src;
+	struct xfs_attr_leafblock *leaf_dst;
+	struct xfs_attr3_icleaf_hdr ichdr_src;
+	struct xfs_trans	*trans = args->trans;
+	char			*tmpbuffer;
+
+	trace_xfs_attr_leaf_compact(args);
+
+	tmpbuffer = kmem_alloc(args->geo->blksize, 0);
+	memcpy(tmpbuffer, bp->b_addr, args->geo->blksize);
+	memset(bp->b_addr, 0, args->geo->blksize);
+	leaf_src = (xfs_attr_leafblock_t *)tmpbuffer;
+	leaf_dst = bp->b_addr;
+
+	/*
+	 * Copy the on-disk header back into the destination buffer to ensure
+	 * all the information in the header that is not part of the incore
+	 * header structure is preserved.
+	 */
+	memcpy(bp->b_addr, tmpbuffer, xfs_attr3_leaf_hdr_size(leaf_src));
+
+	/* Initialise the incore headers */
+	ichdr_src = *ichdr_dst;	/* struct copy */
+	ichdr_dst->firstused = args->geo->blksize;
+	ichdr_dst->usedbytes = 0;
+	ichdr_dst->count = 0;
+	ichdr_dst->holes = 0;
+	ichdr_dst->freemap[0].base = xfs_attr3_leaf_hdr_size(leaf_src);
+	ichdr_dst->freemap[0].size = ichdr_dst->firstused -
+						ichdr_dst->freemap[0].base;
+
+	/* write the header back to initialise the underlying buffer */
+	xfs_attr3_leaf_hdr_to_disk(args->geo, leaf_dst, ichdr_dst);
+
+	/*
+	 * Copy all entry's in the same (sorted) order,
+	 * but allocate name/value pairs packed and in sequence.
+	 */
+	xfs_attr3_leaf_moveents(args, leaf_src, &ichdr_src, 0,
+				leaf_dst, ichdr_dst, 0, ichdr_src.count);
+	/*
+	 * this logs the entire buffer, but the caller must write the header
+	 * back to the buffer when it is finished modifying it.
+	 */
+	xfs_trans_log_buf(trans, bp, 0, args->geo->blksize - 1);
+
+	kmem_free(tmpbuffer);
+}
+
+/*
+ * Compare two leaf blocks "order".
+ * Return 0 unless leaf2 should go before leaf1.
+ */
+static int
+xfs_attr3_leaf_order(
+	struct xfs_buf	*leaf1_bp,
+	struct xfs_attr3_icleaf_hdr *leaf1hdr,
+	struct xfs_buf	*leaf2_bp,
+	struct xfs_attr3_icleaf_hdr *leaf2hdr)
+{
+	struct xfs_attr_leaf_entry *entries1;
+	struct xfs_attr_leaf_entry *entries2;
+
+	entries1 = xfs_attr3_leaf_entryp(leaf1_bp->b_addr);
+	entries2 = xfs_attr3_leaf_entryp(leaf2_bp->b_addr);
+	if (leaf1hdr->count > 0 && leaf2hdr->count > 0 &&
+	    ((be32_to_cpu(entries2[0].hashval) <
+	      be32_to_cpu(entries1[0].hashval)) ||
+	     (be32_to_cpu(entries2[leaf2hdr->count - 1].hashval) <
+	      be32_to_cpu(entries1[leaf1hdr->count - 1].hashval)))) {
+		return 1;
+	}
+	return 0;
+}
+
+int
+xfs_attr_leaf_order(
+	struct xfs_buf	*leaf1_bp,
+	struct xfs_buf	*leaf2_bp)
+{
+	struct xfs_attr3_icleaf_hdr ichdr1;
+	struct xfs_attr3_icleaf_hdr ichdr2;
+	struct xfs_mount *mp = leaf1_bp->b_mount;
+
+	xfs_attr3_leaf_hdr_from_disk(mp->m_attr_geo, &ichdr1, leaf1_bp->b_addr);
+	xfs_attr3_leaf_hdr_from_disk(mp->m_attr_geo, &ichdr2, leaf2_bp->b_addr);
+	return xfs_attr3_leaf_order(leaf1_bp, &ichdr1, leaf2_bp, &ichdr2);
+}
+
+/*
+ * Redistribute the attribute list entries between two leaf nodes,
+ * taking into account the size of the new entry.
+ *
+ * NOTE: if new block is empty, then it will get the upper half of the
+ * old block.  At present, all (one) callers pass in an empty second block.
+ *
+ * This code adjusts the args->index/blkno and args->index2/blkno2 fields
+ * to match what it is doing in splitting the attribute leaf block.  Those
+ * values are used in "atomic rename" operations on attributes.  Note that
+ * the "new" and "old" values can end up in different blocks.
+ */
+STATIC void
+xfs_attr3_leaf_rebalance(
+	struct xfs_da_state	*state,
+	struct xfs_da_state_blk	*blk1,
+	struct xfs_da_state_blk	*blk2)
+{
+	struct xfs_da_args	*args;
+	struct xfs_attr_leafblock *leaf1;
+	struct xfs_attr_leafblock *leaf2;
+	struct xfs_attr3_icleaf_hdr ichdr1;
+	struct xfs_attr3_icleaf_hdr ichdr2;
+	struct xfs_attr_leaf_entry *entries1;
+	struct xfs_attr_leaf_entry *entries2;
+	int			count;
+	int			totallen;
+	int			max;
+	int			space;
+	int			swap;
+
+	/*
+	 * Set up environment.
+	 */
+	ASSERT(blk1->magic == XFS_ATTR_LEAF_MAGIC);
+	ASSERT(blk2->magic == XFS_ATTR_LEAF_MAGIC);
+	leaf1 = blk1->bp->b_addr;
+	leaf2 = blk2->bp->b_addr;
+	xfs_attr3_leaf_hdr_from_disk(state->args->geo, &ichdr1, leaf1);
+	xfs_attr3_leaf_hdr_from_disk(state->args->geo, &ichdr2, leaf2);
+	ASSERT(ichdr2.count == 0);
+	args = state->args;
+
+	trace_xfs_attr_leaf_rebalance(args);
+
+	/*
+	 * Check ordering of blocks, reverse if it makes things simpler.
+	 *
+	 * NOTE: Given that all (current) callers pass in an empty
+	 * second block, this code should never set "swap".
+	 */
+	swap = 0;
+	if (xfs_attr3_leaf_order(blk1->bp, &ichdr1, blk2->bp, &ichdr2)) {
+		swap(blk1, blk2);
+
+		/* swap structures rather than reconverting them */
+		swap(ichdr1, ichdr2);
+
+		leaf1 = blk1->bp->b_addr;
+		leaf2 = blk2->bp->b_addr;
+		swap = 1;
+	}
+
+	/*
+	 * Examine entries until we reduce the absolute difference in
+	 * byte usage between the two blocks to a minimum.  Then get
+	 * the direction to copy and the number of elements to move.
+	 *
+	 * "inleaf" is true if the new entry should be inserted into blk1.
+	 * If "swap" is also true, then reverse the sense of "inleaf".
+	 */
+	state->inleaf = xfs_attr3_leaf_figure_balance(state, blk1, &ichdr1,
+						      blk2, &ichdr2,
+						      &count, &totallen);
+	if (swap)
+		state->inleaf = !state->inleaf;
+
+	/*
+	 * Move any entries required from leaf to leaf:
+	 */
+	if (count < ichdr1.count) {
+		/*
+		 * Figure the total bytes to be added to the destination leaf.
+		 */
+		/* number entries being moved */
+		count = ichdr1.count - count;
+		space  = ichdr1.usedbytes - totallen;
+		space += count * sizeof(xfs_attr_leaf_entry_t);
+
+		/*
+		 * leaf2 is the destination, compact it if it looks tight.
+		 */
+		max  = ichdr2.firstused - xfs_attr3_leaf_hdr_size(leaf1);
+		max -= ichdr2.count * sizeof(xfs_attr_leaf_entry_t);
+		if (space > max)
+			xfs_attr3_leaf_compact(args, &ichdr2, blk2->bp);
+
+		/*
+		 * Move high entries from leaf1 to low end of leaf2.
+		 */
+		xfs_attr3_leaf_moveents(args, leaf1, &ichdr1,
+				ichdr1.count - count, leaf2, &ichdr2, 0, count);
+
+	} else if (count > ichdr1.count) {
+		/*
+		 * I assert that since all callers pass in an empty
+		 * second buffer, this code should never execute.
+		 */
+		ASSERT(0);
+
+		/*
+		 * Figure the total bytes to be added to the destination leaf.
+		 */
+		/* number entries being moved */
+		count -= ichdr1.count;
+		space  = totallen - ichdr1.usedbytes;
+		space += count * sizeof(xfs_attr_leaf_entry_t);
+
+		/*
+		 * leaf1 is the destination, compact it if it looks tight.
+		 */
+		max  = ichdr1.firstused - xfs_attr3_leaf_hdr_size(leaf1);
+		max -= ichdr1.count * sizeof(xfs_attr_leaf_entry_t);
+		if (space > max)
+			xfs_attr3_leaf_compact(args, &ichdr1, blk1->bp);
+
+		/*
+		 * Move low entries from leaf2 to high end of leaf1.
+		 */
+		xfs_attr3_leaf_moveents(args, leaf2, &ichdr2, 0, leaf1, &ichdr1,
+					ichdr1.count, count);
+	}
+
+	xfs_attr3_leaf_hdr_to_disk(state->args->geo, leaf1, &ichdr1);
+	xfs_attr3_leaf_hdr_to_disk(state->args->geo, leaf2, &ichdr2);
+	xfs_trans_log_buf(args->trans, blk1->bp, 0, args->geo->blksize - 1);
+	xfs_trans_log_buf(args->trans, blk2->bp, 0, args->geo->blksize - 1);
+
+	/*
+	 * Copy out last hashval in each block for B-tree code.
+	 */
+	entries1 = xfs_attr3_leaf_entryp(leaf1);
+	entries2 = xfs_attr3_leaf_entryp(leaf2);
+	blk1->hashval = be32_to_cpu(entries1[ichdr1.count - 1].hashval);
+	blk2->hashval = be32_to_cpu(entries2[ichdr2.count - 1].hashval);
+
+	/*
+	 * Adjust the expected index for insertion.
+	 * NOTE: this code depends on the (current) situation that the
+	 * second block was originally empty.
+	 *
+	 * If the insertion point moved to the 2nd block, we must adjust
+	 * the index.  We must also track the entry just following the
+	 * new entry for use in an "atomic rename" operation, that entry
+	 * is always the "old" entry and the "new" entry is what we are
+	 * inserting.  The index/blkno fields refer to the "old" entry,
+	 * while the index2/blkno2 fields refer to the "new" entry.
+	 */
+	if (blk1->index > ichdr1.count) {
+		ASSERT(state->inleaf == 0);
+		blk2->index = blk1->index - ichdr1.count;
+		args->index = args->index2 = blk2->index;
+		args->blkno = args->blkno2 = blk2->blkno;
+	} else if (blk1->index == ichdr1.count) {
+		if (state->inleaf) {
+			args->index = blk1->index;
+			args->blkno = blk1->blkno;
+			args->index2 = 0;
+			args->blkno2 = blk2->blkno;
+		} else {
+			/*
+			 * On a double leaf split, the original attr location
+			 * is already stored in blkno2/index2, so don't
+			 * overwrite it overwise we corrupt the tree.
+			 */
+			blk2->index = blk1->index - ichdr1.count;
+			args->index = blk2->index;
+			args->blkno = blk2->blkno;
+			if (!state->extravalid) {
+				/*
+				 * set the new attr location to match the old
+				 * one and let the higher level split code
+				 * decide where in the leaf to place it.
+				 */
+				args->index2 = blk2->index;
+				args->blkno2 = blk2->blkno;
+			}
+		}
+	} else {
+		ASSERT(state->inleaf == 1);
+		args->index = args->index2 = blk1->index;
+		args->blkno = args->blkno2 = blk1->blkno;
+	}
+}
+
+/*
+ * Examine entries until we reduce the absolute difference in
+ * byte usage between the two blocks to a minimum.
+ * GROT: Is this really necessary?  With other than a 512 byte blocksize,
+ * GROT: there will always be enough room in either block for a new entry.
+ * GROT: Do a double-split for this case?
+ */
+STATIC int
+xfs_attr3_leaf_figure_balance(
+	struct xfs_da_state		*state,
+	struct xfs_da_state_blk		*blk1,
+	struct xfs_attr3_icleaf_hdr	*ichdr1,
+	struct xfs_da_state_blk		*blk2,
+	struct xfs_attr3_icleaf_hdr	*ichdr2,
+	int				*countarg,
+	int				*usedbytesarg)
+{
+	struct xfs_attr_leafblock	*leaf1 = blk1->bp->b_addr;
+	struct xfs_attr_leafblock	*leaf2 = blk2->bp->b_addr;
+	struct xfs_attr_leaf_entry	*entry;
+	int				count;
+	int				max;
+	int				index;
+	int				totallen = 0;
+	int				half;
+	int				lastdelta;
+	int				foundit = 0;
+	int				tmp;
+
+	/*
+	 * Examine entries until we reduce the absolute difference in
+	 * byte usage between the two blocks to a minimum.
+	 */
+	max = ichdr1->count + ichdr2->count;
+	half = (max + 1) * sizeof(*entry);
+	half += ichdr1->usedbytes + ichdr2->usedbytes +
+			xfs_attr_leaf_newentsize(state->args, NULL);
+	half /= 2;
+	lastdelta = state->args->geo->blksize;
+	entry = xfs_attr3_leaf_entryp(leaf1);
+	for (count = index = 0; count < max; entry++, index++, count++) {
+
+#define XFS_ATTR_ABS(A)	(((A) < 0) ? -(A) : (A))
+		/*
+		 * The new entry is in the first block, account for it.
+		 */
+		if (count == blk1->index) {
+			tmp = totallen + sizeof(*entry) +
+				xfs_attr_leaf_newentsize(state->args, NULL);
+			if (XFS_ATTR_ABS(half - tmp) > lastdelta)
+				break;
+			lastdelta = XFS_ATTR_ABS(half - tmp);
+			totallen = tmp;
+			foundit = 1;
+		}
+
+		/*
+		 * Wrap around into the second block if necessary.
+		 */
+		if (count == ichdr1->count) {
+			leaf1 = leaf2;
+			entry = xfs_attr3_leaf_entryp(leaf1);
+			index = 0;
+		}
+
+		/*
+		 * Figure out if next leaf entry would be too much.
+		 */
+		tmp = totallen + sizeof(*entry) + xfs_attr_leaf_entsize(leaf1,
+									index);
+		if (XFS_ATTR_ABS(half - tmp) > lastdelta)
+			break;
+		lastdelta = XFS_ATTR_ABS(half - tmp);
+		totallen = tmp;
+#undef XFS_ATTR_ABS
+	}
+
+	/*
+	 * Calculate the number of usedbytes that will end up in lower block.
+	 * If new entry not in lower block, fix up the count.
+	 */
+	totallen -= count * sizeof(*entry);
+	if (foundit) {
+		totallen -= sizeof(*entry) +
+				xfs_attr_leaf_newentsize(state->args, NULL);
+	}
+
+	*countarg = count;
+	*usedbytesarg = totallen;
+	return foundit;
+}
+
+/*========================================================================
+ * Routines used for shrinking the Btree.
+ *========================================================================*/
+
+/*
+ * Check a leaf block and its neighbors to see if the block should be
+ * collapsed into one or the other neighbor.  Always keep the block
+ * with the smaller block number.
+ * If the current block is over 50% full, don't try to join it, return 0.
+ * If the block is empty, fill in the state structure and return 2.
+ * If it can be collapsed, fill in the state structure and return 1.
+ * If nothing can be done, return 0.
+ *
+ * GROT: allow for INCOMPLETE entries in calculation.
+ */
+int
+xfs_attr3_leaf_toosmall(
+	struct xfs_da_state	*state,
+	int			*action)
+{
+	struct xfs_attr_leafblock *leaf;
+	struct xfs_da_state_blk	*blk;
+	struct xfs_attr3_icleaf_hdr ichdr;
+	struct xfs_buf		*bp;
+	xfs_dablk_t		blkno;
+	int			bytes;
+	int			forward;
+	int			error;
+	int			retval;
+	int			i;
+
+	trace_xfs_attr_leaf_toosmall(state->args);
+
+	/*
+	 * Check for the degenerate case of the block being over 50% full.
+	 * If so, it's not worth even looking to see if we might be able
+	 * to coalesce with a sibling.
+	 */
+	blk = &state->path.blk[ state->path.active-1 ];
+	leaf = blk->bp->b_addr;
+	xfs_attr3_leaf_hdr_from_disk(state->args->geo, &ichdr, leaf);
+	bytes = xfs_attr3_leaf_hdr_size(leaf) +
+		ichdr.count * sizeof(xfs_attr_leaf_entry_t) +
+		ichdr.usedbytes;
+	if (bytes > (state->args->geo->blksize >> 1)) {
+		*action = 0;	/* blk over 50%, don't try to join */
+		return 0;
+	}
+
+	/*
+	 * Check for the degenerate case of the block being empty.
+	 * If the block is empty, we'll simply delete it, no need to
+	 * coalesce it with a sibling block.  We choose (arbitrarily)
+	 * to merge with the forward block unless it is NULL.
+	 */
+	if (ichdr.count == 0) {
+		/*
+		 * Make altpath point to the block we want to keep and
+		 * path point to the block we want to drop (this one).
+		 */
+		forward = (ichdr.forw != 0);
+		memcpy(&state->altpath, &state->path, sizeof(state->path));
+		error = xfs_da3_path_shift(state, &state->altpath, forward,
+						 0, &retval);
+		if (error)
+			return error;
+		if (retval) {
+			*action = 0;
+		} else {
+			*action = 2;
+		}
+		return 0;
+	}
+
+	/*
+	 * Examine each sibling block to see if we can coalesce with
+	 * at least 25% free space to spare.  We need to figure out
+	 * whether to merge with the forward or the backward block.
+	 * We prefer coalescing with the lower numbered sibling so as
+	 * to shrink an attribute list over time.
+	 */
+	/* start with smaller blk num */
+	forward = ichdr.forw < ichdr.back;
+	for (i = 0; i < 2; forward = !forward, i++) {
+		struct xfs_attr3_icleaf_hdr ichdr2;
+		if (forward)
+			blkno = ichdr.forw;
+		else
+			blkno = ichdr.back;
+		if (blkno == 0)
+			continue;
+		error = xfs_attr3_leaf_read(state->args->trans, state->args->dp,
+					blkno, &bp);
+		if (error)
+			return error;
+
+		xfs_attr3_leaf_hdr_from_disk(state->args->geo, &ichdr2, bp->b_addr);
+
+		bytes = state->args->geo->blksize -
+			(state->args->geo->blksize >> 2) -
+			ichdr.usedbytes - ichdr2.usedbytes -
+			((ichdr.count + ichdr2.count) *
+					sizeof(xfs_attr_leaf_entry_t)) -
+			xfs_attr3_leaf_hdr_size(leaf);
+
+		xfs_trans_brelse(state->args->trans, bp);
+		if (bytes >= 0)
+			break;	/* fits with at least 25% to spare */
+	}
+	if (i >= 2) {
+		*action = 0;
+		return 0;
+	}
+
+	/*
+	 * Make altpath point to the block we want to keep (the lower
+	 * numbered block) and path point to the block we want to drop.
+	 */
+	memcpy(&state->altpath, &state->path, sizeof(state->path));
+	if (blkno < blk->blkno) {
+		error = xfs_da3_path_shift(state, &state->altpath, forward,
+						 0, &retval);
+	} else {
+		error = xfs_da3_path_shift(state, &state->path, forward,
+						 0, &retval);
+	}
+	if (error)
+		return error;
+	if (retval) {
+		*action = 0;
+	} else {
+		*action = 1;
+	}
+	return 0;
+}
+
+/*
+ * Remove a name from the leaf attribute list structure.
+ *
+ * Return 1 if leaf is less than 37% full, 0 if >= 37% full.
+ * If two leaves are 37% full, when combined they will leave 25% free.
+ */
+int
+xfs_attr3_leaf_remove(
+	struct xfs_buf		*bp,
+	struct xfs_da_args	*args)
+{
+	struct xfs_attr_leafblock *leaf;
+	struct xfs_attr3_icleaf_hdr ichdr;
+	struct xfs_attr_leaf_entry *entry;
+	int			before;
+	int			after;
+	int			smallest;
+	int			entsize;
+	int			tablesize;
+	int			tmp;
+	int			i;
+
+	trace_xfs_attr_leaf_remove(args);
+
+	leaf = bp->b_addr;
+	xfs_attr3_leaf_hdr_from_disk(args->geo, &ichdr, leaf);
+
+	ASSERT(ichdr.count > 0 && ichdr.count < args->geo->blksize / 8);
+	ASSERT(args->index >= 0 && args->index < ichdr.count);
+	ASSERT(ichdr.firstused >= ichdr.count * sizeof(*entry) +
+					xfs_attr3_leaf_hdr_size(leaf));
+
+	entry = &xfs_attr3_leaf_entryp(leaf)[args->index];
+
+	ASSERT(be16_to_cpu(entry->nameidx) >= ichdr.firstused);
+	ASSERT(be16_to_cpu(entry->nameidx) < args->geo->blksize);
+
+	/*
+	 * Scan through free region table:
+	 *    check for adjacency of free'd entry with an existing one,
+	 *    find smallest free region in case we need to replace it,
+	 *    adjust any map that borders the entry table,
+	 */
+	tablesize = ichdr.count * sizeof(xfs_attr_leaf_entry_t)
+					+ xfs_attr3_leaf_hdr_size(leaf);
+	tmp = ichdr.freemap[0].size;
+	before = after = -1;
+	smallest = XFS_ATTR_LEAF_MAPSIZE - 1;
+	entsize = xfs_attr_leaf_entsize(leaf, args->index);
+	for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; i++) {
+		ASSERT(ichdr.freemap[i].base < args->geo->blksize);
+		ASSERT(ichdr.freemap[i].size < args->geo->blksize);
+		if (ichdr.freemap[i].base == tablesize) {
+			ichdr.freemap[i].base -= sizeof(xfs_attr_leaf_entry_t);
+			ichdr.freemap[i].size += sizeof(xfs_attr_leaf_entry_t);
+		}
+
+		if (ichdr.freemap[i].base + ichdr.freemap[i].size ==
+				be16_to_cpu(entry->nameidx)) {
+			before = i;
+		} else if (ichdr.freemap[i].base ==
+				(be16_to_cpu(entry->nameidx) + entsize)) {
+			after = i;
+		} else if (ichdr.freemap[i].size < tmp) {
+			tmp = ichdr.freemap[i].size;
+			smallest = i;
+		}
+	}
+
+	/*
+	 * Coalesce adjacent freemap regions,
+	 * or replace the smallest region.
+	 */
+	if ((before >= 0) || (after >= 0)) {
+		if ((before >= 0) && (after >= 0)) {
+			ichdr.freemap[before].size += entsize;
+			ichdr.freemap[before].size += ichdr.freemap[after].size;
+			ichdr.freemap[after].base = 0;
+			ichdr.freemap[after].size = 0;
+		} else if (before >= 0) {
+			ichdr.freemap[before].size += entsize;
+		} else {
+			ichdr.freemap[after].base = be16_to_cpu(entry->nameidx);
+			ichdr.freemap[after].size += entsize;
+		}
+	} else {
+		/*
+		 * Replace smallest region (if it is smaller than free'd entry)
+		 */
+		if (ichdr.freemap[smallest].size < entsize) {
+			ichdr.freemap[smallest].base = be16_to_cpu(entry->nameidx);
+			ichdr.freemap[smallest].size = entsize;
+		}
+	}
+
+	/*
+	 * Did we remove the first entry?
+	 */
+	if (be16_to_cpu(entry->nameidx) == ichdr.firstused)
+		smallest = 1;
+	else
+		smallest = 0;
+
+	/*
+	 * Compress the remaining entries and zero out the removed stuff.
+	 */
+	memset(xfs_attr3_leaf_name(leaf, args->index), 0, entsize);
+	ichdr.usedbytes -= entsize;
+	xfs_trans_log_buf(args->trans, bp,
+	     XFS_DA_LOGRANGE(leaf, xfs_attr3_leaf_name(leaf, args->index),
+				   entsize));
+
+	tmp = (ichdr.count - args->index) * sizeof(xfs_attr_leaf_entry_t);
+	memmove(entry, entry + 1, tmp);
+	ichdr.count--;
+	xfs_trans_log_buf(args->trans, bp,
+	    XFS_DA_LOGRANGE(leaf, entry, tmp + sizeof(xfs_attr_leaf_entry_t)));
+
+	entry = &xfs_attr3_leaf_entryp(leaf)[ichdr.count];
+	memset(entry, 0, sizeof(xfs_attr_leaf_entry_t));
+
+	/*
+	 * If we removed the first entry, re-find the first used byte
+	 * in the name area.  Note that if the entry was the "firstused",
+	 * then we don't have a "hole" in our block resulting from
+	 * removing the name.
+	 */
+	if (smallest) {
+		tmp = args->geo->blksize;
+		entry = xfs_attr3_leaf_entryp(leaf);
+		for (i = ichdr.count - 1; i >= 0; entry++, i--) {
+			ASSERT(be16_to_cpu(entry->nameidx) >= ichdr.firstused);
+			ASSERT(be16_to_cpu(entry->nameidx) < args->geo->blksize);
+
+			if (be16_to_cpu(entry->nameidx) < tmp)
+				tmp = be16_to_cpu(entry->nameidx);
+		}
+		ichdr.firstused = tmp;
+		ASSERT(ichdr.firstused != 0);
+	} else {
+		ichdr.holes = 1;	/* mark as needing compaction */
+	}
+	xfs_attr3_leaf_hdr_to_disk(args->geo, leaf, &ichdr);
+	xfs_trans_log_buf(args->trans, bp,
+			  XFS_DA_LOGRANGE(leaf, &leaf->hdr,
+					  xfs_attr3_leaf_hdr_size(leaf)));
+
+	/*
+	 * Check if leaf is less than 50% full, caller may want to
+	 * "join" the leaf with a sibling if so.
+	 */
+	tmp = ichdr.usedbytes + xfs_attr3_leaf_hdr_size(leaf) +
+	      ichdr.count * sizeof(xfs_attr_leaf_entry_t);
+
+	return tmp < args->geo->magicpct; /* leaf is < 37% full */
+}
+
+/*
+ * Move all the attribute list entries from drop_leaf into save_leaf.
+ */
+void
+xfs_attr3_leaf_unbalance(
+	struct xfs_da_state	*state,
+	struct xfs_da_state_blk	*drop_blk,
+	struct xfs_da_state_blk	*save_blk)
+{
+	struct xfs_attr_leafblock *drop_leaf = drop_blk->bp->b_addr;
+	struct xfs_attr_leafblock *save_leaf = save_blk->bp->b_addr;
+	struct xfs_attr3_icleaf_hdr drophdr;
+	struct xfs_attr3_icleaf_hdr savehdr;
+	struct xfs_attr_leaf_entry *entry;
+
+	trace_xfs_attr_leaf_unbalance(state->args);
+
+	drop_leaf = drop_blk->bp->b_addr;
+	save_leaf = save_blk->bp->b_addr;
+	xfs_attr3_leaf_hdr_from_disk(state->args->geo, &drophdr, drop_leaf);
+	xfs_attr3_leaf_hdr_from_disk(state->args->geo, &savehdr, save_leaf);
+	entry = xfs_attr3_leaf_entryp(drop_leaf);
+
+	/*
+	 * Save last hashval from dying block for later Btree fixup.
+	 */
+	drop_blk->hashval = be32_to_cpu(entry[drophdr.count - 1].hashval);
+
+	/*
+	 * Check if we need a temp buffer, or can we do it in place.
+	 * Note that we don't check "leaf" for holes because we will
+	 * always be dropping it, toosmall() decided that for us already.
+	 */
+	if (savehdr.holes == 0) {
+		/*
+		 * dest leaf has no holes, so we add there.  May need
+		 * to make some room in the entry array.
+		 */
+		if (xfs_attr3_leaf_order(save_blk->bp, &savehdr,
+					 drop_blk->bp, &drophdr)) {
+			xfs_attr3_leaf_moveents(state->args,
+						drop_leaf, &drophdr, 0,
+						save_leaf, &savehdr, 0,
+						drophdr.count);
+		} else {
+			xfs_attr3_leaf_moveents(state->args,
+						drop_leaf, &drophdr, 0,
+						save_leaf, &savehdr,
+						savehdr.count, drophdr.count);
+		}
+	} else {
+		/*
+		 * Destination has holes, so we make a temporary copy
+		 * of the leaf and add them both to that.
+		 */
+		struct xfs_attr_leafblock *tmp_leaf;
+		struct xfs_attr3_icleaf_hdr tmphdr;
+
+		tmp_leaf = kmem_zalloc(state->args->geo->blksize, 0);
+
+		/*
+		 * Copy the header into the temp leaf so that all the stuff
+		 * not in the incore header is present and gets copied back in
+		 * once we've moved all the entries.
+		 */
+		memcpy(tmp_leaf, save_leaf, xfs_attr3_leaf_hdr_size(save_leaf));
+
+		memset(&tmphdr, 0, sizeof(tmphdr));
+		tmphdr.magic = savehdr.magic;
+		tmphdr.forw = savehdr.forw;
+		tmphdr.back = savehdr.back;
+		tmphdr.firstused = state->args->geo->blksize;
+
+		/* write the header to the temp buffer to initialise it */
+		xfs_attr3_leaf_hdr_to_disk(state->args->geo, tmp_leaf, &tmphdr);
+
+		if (xfs_attr3_leaf_order(save_blk->bp, &savehdr,
+					 drop_blk->bp, &drophdr)) {
+			xfs_attr3_leaf_moveents(state->args,
+						drop_leaf, &drophdr, 0,
+						tmp_leaf, &tmphdr, 0,
+						drophdr.count);
+			xfs_attr3_leaf_moveents(state->args,
+						save_leaf, &savehdr, 0,
+						tmp_leaf, &tmphdr, tmphdr.count,
+						savehdr.count);
+		} else {
+			xfs_attr3_leaf_moveents(state->args,
+						save_leaf, &savehdr, 0,
+						tmp_leaf, &tmphdr, 0,
+						savehdr.count);
+			xfs_attr3_leaf_moveents(state->args,
+						drop_leaf, &drophdr, 0,
+						tmp_leaf, &tmphdr, tmphdr.count,
+						drophdr.count);
+		}
+		memcpy(save_leaf, tmp_leaf, state->args->geo->blksize);
+		savehdr = tmphdr; /* struct copy */
+		kmem_free(tmp_leaf);
+	}
+
+	xfs_attr3_leaf_hdr_to_disk(state->args->geo, save_leaf, &savehdr);
+	xfs_trans_log_buf(state->args->trans, save_blk->bp, 0,
+					   state->args->geo->blksize - 1);
+
+	/*
+	 * Copy out last hashval in each block for B-tree code.
+	 */
+	entry = xfs_attr3_leaf_entryp(save_leaf);
+	save_blk->hashval = be32_to_cpu(entry[savehdr.count - 1].hashval);
+}
+
+/*========================================================================
+ * Routines used for finding things in the Btree.
+ *========================================================================*/
+
+/*
+ * Look up a name in a leaf attribute list structure.
+ * This is the internal routine, it uses the caller's buffer.
+ *
+ * Note that duplicate keys are allowed, but only check within the
+ * current leaf node.  The Btree code must check in adjacent leaf nodes.
+ *
+ * Return in args->index the index into the entry[] array of either
+ * the found entry, or where the entry should have been (insert before
+ * that entry).
+ *
+ * Don't change the args->value unless we find the attribute.
+ */
+int
+xfs_attr3_leaf_lookup_int(
+	struct xfs_buf		*bp,
+	struct xfs_da_args	*args)
+{
+	struct xfs_attr_leafblock *leaf;
+	struct xfs_attr3_icleaf_hdr ichdr;
+	struct xfs_attr_leaf_entry *entry;
+	struct xfs_attr_leaf_entry *entries;
+	struct xfs_attr_leaf_name_local *name_loc;
+	struct xfs_attr_leaf_name_remote *name_rmt;
+	xfs_dahash_t		hashval;
+	int			probe;
+	int			span;
+
+	trace_xfs_attr_leaf_lookup(args);
+
+	leaf = bp->b_addr;
+	xfs_attr3_leaf_hdr_from_disk(args->geo, &ichdr, leaf);
+	entries = xfs_attr3_leaf_entryp(leaf);
+	if (ichdr.count >= args->geo->blksize / 8) {
+		xfs_buf_mark_corrupt(bp);
+		return -EFSCORRUPTED;
+	}
+
+	/*
+	 * Binary search.  (note: small blocks will skip this loop)
+	 */
+	hashval = args->hashval;
+	probe = span = ichdr.count / 2;
+	for (entry = &entries[probe]; span > 4; entry = &entries[probe]) {
+		span /= 2;
+		if (be32_to_cpu(entry->hashval) < hashval)
+			probe += span;
+		else if (be32_to_cpu(entry->hashval) > hashval)
+			probe -= span;
+		else
+			break;
+	}
+	if (!(probe >= 0 && (!ichdr.count || probe < ichdr.count))) {
+		xfs_buf_mark_corrupt(bp);
+		return -EFSCORRUPTED;
+	}
+	if (!(span <= 4 || be32_to_cpu(entry->hashval) == hashval)) {
+		xfs_buf_mark_corrupt(bp);
+		return -EFSCORRUPTED;
+	}
+
+	/*
+	 * Since we may have duplicate hashval's, find the first matching
+	 * hashval in the leaf.
+	 */
+	while (probe > 0 && be32_to_cpu(entry->hashval) >= hashval) {
+		entry--;
+		probe--;
+	}
+	while (probe < ichdr.count &&
+	       be32_to_cpu(entry->hashval) < hashval) {
+		entry++;
+		probe++;
+	}
+	if (probe == ichdr.count || be32_to_cpu(entry->hashval) != hashval) {
+		args->index = probe;
+		return -ENOATTR;
+	}
+
+	/*
+	 * Duplicate keys may be present, so search all of them for a match.
+	 */
+	for (; probe < ichdr.count && (be32_to_cpu(entry->hashval) == hashval);
+			entry++, probe++) {
+/*
+ * GROT: Add code to remove incomplete entries.
+ */
+		if (entry->flags & XFS_ATTR_LOCAL) {
+			name_loc = xfs_attr3_leaf_name_local(leaf, probe);
+			if (!xfs_attr_match(args, name_loc->namelen,
+					name_loc->nameval, entry->flags))
+				continue;
+			args->index = probe;
+			return -EEXIST;
+		} else {
+			name_rmt = xfs_attr3_leaf_name_remote(leaf, probe);
+			if (!xfs_attr_match(args, name_rmt->namelen,
+					name_rmt->name, entry->flags))
+				continue;
+			args->index = probe;
+			args->rmtvaluelen = be32_to_cpu(name_rmt->valuelen);
+			args->rmtblkno = be32_to_cpu(name_rmt->valueblk);
+			args->rmtblkcnt = xfs_attr3_rmt_blocks(
+							args->dp->i_mount,
+							args->rmtvaluelen);
+			return -EEXIST;
+		}
+	}
+	args->index = probe;
+	return -ENOATTR;
+}
+
+/*
+ * Get the value associated with an attribute name from a leaf attribute
+ * list structure.
+ *
+ * If args->valuelen is zero, only the length needs to be returned.  Unlike a
+ * lookup, we only return an error if the attribute does not exist or we can't
+ * retrieve the value.
+ */
+int
+xfs_attr3_leaf_getvalue(
+	struct xfs_buf		*bp,
+	struct xfs_da_args	*args)
+{
+	struct xfs_attr_leafblock *leaf;
+	struct xfs_attr3_icleaf_hdr ichdr;
+	struct xfs_attr_leaf_entry *entry;
+	struct xfs_attr_leaf_name_local *name_loc;
+	struct xfs_attr_leaf_name_remote *name_rmt;
+
+	leaf = bp->b_addr;
+	xfs_attr3_leaf_hdr_from_disk(args->geo, &ichdr, leaf);
+	ASSERT(ichdr.count < args->geo->blksize / 8);
+	ASSERT(args->index < ichdr.count);
+
+	entry = &xfs_attr3_leaf_entryp(leaf)[args->index];
+	if (entry->flags & XFS_ATTR_LOCAL) {
+		name_loc = xfs_attr3_leaf_name_local(leaf, args->index);
+		ASSERT(name_loc->namelen == args->namelen);
+		ASSERT(memcmp(args->name, name_loc->nameval, args->namelen) == 0);
+		return xfs_attr_copy_value(args,
+					&name_loc->nameval[args->namelen],
+					be16_to_cpu(name_loc->valuelen));
+	}
+
+	name_rmt = xfs_attr3_leaf_name_remote(leaf, args->index);
+	ASSERT(name_rmt->namelen == args->namelen);
+	ASSERT(memcmp(args->name, name_rmt->name, args->namelen) == 0);
+	args->rmtvaluelen = be32_to_cpu(name_rmt->valuelen);
+	args->rmtblkno = be32_to_cpu(name_rmt->valueblk);
+	args->rmtblkcnt = xfs_attr3_rmt_blocks(args->dp->i_mount,
+					       args->rmtvaluelen);
+	return xfs_attr_copy_value(args, NULL, args->rmtvaluelen);
+}
+
+/*========================================================================
+ * Utility routines.
+ *========================================================================*/
+
+/*
+ * Move the indicated entries from one leaf to another.
+ * NOTE: this routine modifies both source and destination leaves.
+ */
+/*ARGSUSED*/
+STATIC void
+xfs_attr3_leaf_moveents(
+	struct xfs_da_args		*args,
+	struct xfs_attr_leafblock	*leaf_s,
+	struct xfs_attr3_icleaf_hdr	*ichdr_s,
+	int				start_s,
+	struct xfs_attr_leafblock	*leaf_d,
+	struct xfs_attr3_icleaf_hdr	*ichdr_d,
+	int				start_d,
+	int				count)
+{
+	struct xfs_attr_leaf_entry	*entry_s;
+	struct xfs_attr_leaf_entry	*entry_d;
+	int				desti;
+	int				tmp;
+	int				i;
+
+	/*
+	 * Check for nothing to do.
+	 */
+	if (count == 0)
+		return;
+
+	/*
+	 * Set up environment.
+	 */
+	ASSERT(ichdr_s->magic == XFS_ATTR_LEAF_MAGIC ||
+	       ichdr_s->magic == XFS_ATTR3_LEAF_MAGIC);
+	ASSERT(ichdr_s->magic == ichdr_d->magic);
+	ASSERT(ichdr_s->count > 0 && ichdr_s->count < args->geo->blksize / 8);
+	ASSERT(ichdr_s->firstused >= (ichdr_s->count * sizeof(*entry_s))
+					+ xfs_attr3_leaf_hdr_size(leaf_s));
+	ASSERT(ichdr_d->count < args->geo->blksize / 8);
+	ASSERT(ichdr_d->firstused >= (ichdr_d->count * sizeof(*entry_d))
+					+ xfs_attr3_leaf_hdr_size(leaf_d));
+
+	ASSERT(start_s < ichdr_s->count);
+	ASSERT(start_d <= ichdr_d->count);
+	ASSERT(count <= ichdr_s->count);
+
+
+	/*
+	 * Move the entries in the destination leaf up to make a hole?
+	 */
+	if (start_d < ichdr_d->count) {
+		tmp  = ichdr_d->count - start_d;
+		tmp *= sizeof(xfs_attr_leaf_entry_t);
+		entry_s = &xfs_attr3_leaf_entryp(leaf_d)[start_d];
+		entry_d = &xfs_attr3_leaf_entryp(leaf_d)[start_d + count];
+		memmove(entry_d, entry_s, tmp);
+	}
+
+	/*
+	 * Copy all entry's in the same (sorted) order,
+	 * but allocate attribute info packed and in sequence.
+	 */
+	entry_s = &xfs_attr3_leaf_entryp(leaf_s)[start_s];
+	entry_d = &xfs_attr3_leaf_entryp(leaf_d)[start_d];
+	desti = start_d;
+	for (i = 0; i < count; entry_s++, entry_d++, desti++, i++) {
+		ASSERT(be16_to_cpu(entry_s->nameidx) >= ichdr_s->firstused);
+		tmp = xfs_attr_leaf_entsize(leaf_s, start_s + i);
+#ifdef GROT
+		/*
+		 * Code to drop INCOMPLETE entries.  Difficult to use as we
+		 * may also need to change the insertion index.  Code turned
+		 * off for 6.2, should be revisited later.
+		 */
+		if (entry_s->flags & XFS_ATTR_INCOMPLETE) { /* skip partials? */
+			memset(xfs_attr3_leaf_name(leaf_s, start_s + i), 0, tmp);
+			ichdr_s->usedbytes -= tmp;
+			ichdr_s->count -= 1;
+			entry_d--;	/* to compensate for ++ in loop hdr */
+			desti--;
+			if ((start_s + i) < offset)
+				result++;	/* insertion index adjustment */
+		} else {
+#endif /* GROT */
+			ichdr_d->firstused -= tmp;
+			/* both on-disk, don't endian flip twice */
+			entry_d->hashval = entry_s->hashval;
+			entry_d->nameidx = cpu_to_be16(ichdr_d->firstused);
+			entry_d->flags = entry_s->flags;
+			ASSERT(be16_to_cpu(entry_d->nameidx) + tmp
+							<= args->geo->blksize);
+			memmove(xfs_attr3_leaf_name(leaf_d, desti),
+				xfs_attr3_leaf_name(leaf_s, start_s + i), tmp);
+			ASSERT(be16_to_cpu(entry_s->nameidx) + tmp
+							<= args->geo->blksize);
+			memset(xfs_attr3_leaf_name(leaf_s, start_s + i), 0, tmp);
+			ichdr_s->usedbytes -= tmp;
+			ichdr_d->usedbytes += tmp;
+			ichdr_s->count -= 1;
+			ichdr_d->count += 1;
+			tmp = ichdr_d->count * sizeof(xfs_attr_leaf_entry_t)
+					+ xfs_attr3_leaf_hdr_size(leaf_d);
+			ASSERT(ichdr_d->firstused >= tmp);
+#ifdef GROT
+		}
+#endif /* GROT */
+	}
+
+	/*
+	 * Zero out the entries we just copied.
+	 */
+	if (start_s == ichdr_s->count) {
+		tmp = count * sizeof(xfs_attr_leaf_entry_t);
+		entry_s = &xfs_attr3_leaf_entryp(leaf_s)[start_s];
+		ASSERT(((char *)entry_s + tmp) <=
+		       ((char *)leaf_s + args->geo->blksize));
+		memset(entry_s, 0, tmp);
+	} else {
+		/*
+		 * Move the remaining entries down to fill the hole,
+		 * then zero the entries at the top.
+		 */
+		tmp  = (ichdr_s->count - count) * sizeof(xfs_attr_leaf_entry_t);
+		entry_s = &xfs_attr3_leaf_entryp(leaf_s)[start_s + count];
+		entry_d = &xfs_attr3_leaf_entryp(leaf_s)[start_s];
+		memmove(entry_d, entry_s, tmp);
+
+		tmp = count * sizeof(xfs_attr_leaf_entry_t);
+		entry_s = &xfs_attr3_leaf_entryp(leaf_s)[ichdr_s->count];
+		ASSERT(((char *)entry_s + tmp) <=
+		       ((char *)leaf_s + args->geo->blksize));
+		memset(entry_s, 0, tmp);
+	}
+
+	/*
+	 * Fill in the freemap information
+	 */
+	ichdr_d->freemap[0].base = xfs_attr3_leaf_hdr_size(leaf_d);
+	ichdr_d->freemap[0].base += ichdr_d->count * sizeof(xfs_attr_leaf_entry_t);
+	ichdr_d->freemap[0].size = ichdr_d->firstused - ichdr_d->freemap[0].base;
+	ichdr_d->freemap[1].base = 0;
+	ichdr_d->freemap[2].base = 0;
+	ichdr_d->freemap[1].size = 0;
+	ichdr_d->freemap[2].size = 0;
+	ichdr_s->holes = 1;	/* leaf may not be compact */
+}
+
+/*
+ * Pick up the last hashvalue from a leaf block.
+ */
+xfs_dahash_t
+xfs_attr_leaf_lasthash(
+	struct xfs_buf	*bp,
+	int		*count)
+{
+	struct xfs_attr3_icleaf_hdr ichdr;
+	struct xfs_attr_leaf_entry *entries;
+	struct xfs_mount *mp = bp->b_mount;
+
+	xfs_attr3_leaf_hdr_from_disk(mp->m_attr_geo, &ichdr, bp->b_addr);
+	entries = xfs_attr3_leaf_entryp(bp->b_addr);
+	if (count)
+		*count = ichdr.count;
+	if (!ichdr.count)
+		return 0;
+	return be32_to_cpu(entries[ichdr.count - 1].hashval);
+}
+
+/*
+ * Calculate the number of bytes used to store the indicated attribute
+ * (whether local or remote only calculate bytes in this block).
+ */
+STATIC int
+xfs_attr_leaf_entsize(xfs_attr_leafblock_t *leaf, int index)
+{
+	struct xfs_attr_leaf_entry *entries;
+	xfs_attr_leaf_name_local_t *name_loc;
+	xfs_attr_leaf_name_remote_t *name_rmt;
+	int size;
+
+	entries = xfs_attr3_leaf_entryp(leaf);
+	if (entries[index].flags & XFS_ATTR_LOCAL) {
+		name_loc = xfs_attr3_leaf_name_local(leaf, index);
+		size = xfs_attr_leaf_entsize_local(name_loc->namelen,
+						   be16_to_cpu(name_loc->valuelen));
+	} else {
+		name_rmt = xfs_attr3_leaf_name_remote(leaf, index);
+		size = xfs_attr_leaf_entsize_remote(name_rmt->namelen);
+	}
+	return size;
+}
+
+/*
+ * Calculate the number of bytes that would be required to store the new
+ * attribute (whether local or remote only calculate bytes in this block).
+ * This routine decides as a side effect whether the attribute will be
+ * a "local" or a "remote" attribute.
+ */
+int
+xfs_attr_leaf_newentsize(
+	struct xfs_da_args	*args,
+	int			*local)
+{
+	int			size;
+
+	size = xfs_attr_leaf_entsize_local(args->namelen, args->valuelen);
+	if (size < xfs_attr_leaf_entsize_local_max(args->geo->blksize)) {
+		if (local)
+			*local = 1;
+		return size;
+	}
+	if (local)
+		*local = 0;
+	return xfs_attr_leaf_entsize_remote(args->namelen);
+}
+
+
+/*========================================================================
+ * Manage the INCOMPLETE flag in a leaf entry
+ *========================================================================*/
+
+/*
+ * Clear the INCOMPLETE flag on an entry in a leaf block.
+ */
+int
+xfs_attr3_leaf_clearflag(
+	struct xfs_da_args	*args)
+{
+	struct xfs_attr_leafblock *leaf;
+	struct xfs_attr_leaf_entry *entry;
+	struct xfs_attr_leaf_name_remote *name_rmt;
+	struct xfs_buf		*bp;
+	int			error;
+#ifdef DEBUG
+	struct xfs_attr3_icleaf_hdr ichdr;
+	xfs_attr_leaf_name_local_t *name_loc;
+	int namelen;
+	char *name;
+#endif /* DEBUG */
+
+	trace_xfs_attr_leaf_clearflag(args);
+	/*
+	 * Set up the operation.
+	 */
+	error = xfs_attr3_leaf_read(args->trans, args->dp, args->blkno, &bp);
+	if (error)
+		return error;
+
+	leaf = bp->b_addr;
+	entry = &xfs_attr3_leaf_entryp(leaf)[args->index];
+	ASSERT(entry->flags & XFS_ATTR_INCOMPLETE);
+
+#ifdef DEBUG
+	xfs_attr3_leaf_hdr_from_disk(args->geo, &ichdr, leaf);
+	ASSERT(args->index < ichdr.count);
+	ASSERT(args->index >= 0);
+
+	if (entry->flags & XFS_ATTR_LOCAL) {
+		name_loc = xfs_attr3_leaf_name_local(leaf, args->index);
+		namelen = name_loc->namelen;
+		name = (char *)name_loc->nameval;
+	} else {
+		name_rmt = xfs_attr3_leaf_name_remote(leaf, args->index);
+		namelen = name_rmt->namelen;
+		name = (char *)name_rmt->name;
+	}
+	ASSERT(be32_to_cpu(entry->hashval) == args->hashval);
+	ASSERT(namelen == args->namelen);
+	ASSERT(memcmp(name, args->name, namelen) == 0);
+#endif /* DEBUG */
+
+	entry->flags &= ~XFS_ATTR_INCOMPLETE;
+	xfs_trans_log_buf(args->trans, bp,
+			 XFS_DA_LOGRANGE(leaf, entry, sizeof(*entry)));
+
+	if (args->rmtblkno) {
+		ASSERT((entry->flags & XFS_ATTR_LOCAL) == 0);
+		name_rmt = xfs_attr3_leaf_name_remote(leaf, args->index);
+		name_rmt->valueblk = cpu_to_be32(args->rmtblkno);
+		name_rmt->valuelen = cpu_to_be32(args->rmtvaluelen);
+		xfs_trans_log_buf(args->trans, bp,
+			 XFS_DA_LOGRANGE(leaf, name_rmt, sizeof(*name_rmt)));
+	}
+
+	return 0;
+}
+
+/*
+ * Set the INCOMPLETE flag on an entry in a leaf block.
+ */
+int
+xfs_attr3_leaf_setflag(
+	struct xfs_da_args	*args)
+{
+	struct xfs_attr_leafblock *leaf;
+	struct xfs_attr_leaf_entry *entry;
+	struct xfs_attr_leaf_name_remote *name_rmt;
+	struct xfs_buf		*bp;
+	int error;
+#ifdef DEBUG
+	struct xfs_attr3_icleaf_hdr ichdr;
+#endif
+
+	trace_xfs_attr_leaf_setflag(args);
+
+	/*
+	 * Set up the operation.
+	 */
+	error = xfs_attr3_leaf_read(args->trans, args->dp, args->blkno, &bp);
+	if (error)
+		return error;
+
+	leaf = bp->b_addr;
+#ifdef DEBUG
+	xfs_attr3_leaf_hdr_from_disk(args->geo, &ichdr, leaf);
+	ASSERT(args->index < ichdr.count);
+	ASSERT(args->index >= 0);
+#endif
+	entry = &xfs_attr3_leaf_entryp(leaf)[args->index];
+
+	ASSERT((entry->flags & XFS_ATTR_INCOMPLETE) == 0);
+	entry->flags |= XFS_ATTR_INCOMPLETE;
+	xfs_trans_log_buf(args->trans, bp,
+			XFS_DA_LOGRANGE(leaf, entry, sizeof(*entry)));
+	if ((entry->flags & XFS_ATTR_LOCAL) == 0) {
+		name_rmt = xfs_attr3_leaf_name_remote(leaf, args->index);
+		name_rmt->valueblk = 0;
+		name_rmt->valuelen = 0;
+		xfs_trans_log_buf(args->trans, bp,
+			 XFS_DA_LOGRANGE(leaf, name_rmt, sizeof(*name_rmt)));
+	}
+
+	return 0;
+}
+
+/*
+ * In a single transaction, clear the INCOMPLETE flag on the leaf entry
+ * given by args->blkno/index and set the INCOMPLETE flag on the leaf
+ * entry given by args->blkno2/index2.
+ *
+ * Note that they could be in different blocks, or in the same block.
+ */
+int
+xfs_attr3_leaf_flipflags(
+	struct xfs_da_args	*args)
+{
+	struct xfs_attr_leafblock *leaf1;
+	struct xfs_attr_leafblock *leaf2;
+	struct xfs_attr_leaf_entry *entry1;
+	struct xfs_attr_leaf_entry *entry2;
+	struct xfs_attr_leaf_name_remote *name_rmt;
+	struct xfs_buf		*bp1;
+	struct xfs_buf		*bp2;
+	int error;
+#ifdef DEBUG
+	struct xfs_attr3_icleaf_hdr ichdr1;
+	struct xfs_attr3_icleaf_hdr ichdr2;
+	xfs_attr_leaf_name_local_t *name_loc;
+	int namelen1, namelen2;
+	char *name1, *name2;
+#endif /* DEBUG */
+
+	trace_xfs_attr_leaf_flipflags(args);
+
+	/*
+	 * Read the block containing the "old" attr
+	 */
+	error = xfs_attr3_leaf_read(args->trans, args->dp, args->blkno, &bp1);
+	if (error)
+		return error;
+
+	/*
+	 * Read the block containing the "new" attr, if it is different
+	 */
+	if (args->blkno2 != args->blkno) {
+		error = xfs_attr3_leaf_read(args->trans, args->dp, args->blkno2,
+					   &bp2);
+		if (error)
+			return error;
+	} else {
+		bp2 = bp1;
+	}
+
+	leaf1 = bp1->b_addr;
+	entry1 = &xfs_attr3_leaf_entryp(leaf1)[args->index];
+
+	leaf2 = bp2->b_addr;
+	entry2 = &xfs_attr3_leaf_entryp(leaf2)[args->index2];
+
+#ifdef DEBUG
+	xfs_attr3_leaf_hdr_from_disk(args->geo, &ichdr1, leaf1);
+	ASSERT(args->index < ichdr1.count);
+	ASSERT(args->index >= 0);
+
+	xfs_attr3_leaf_hdr_from_disk(args->geo, &ichdr2, leaf2);
+	ASSERT(args->index2 < ichdr2.count);
+	ASSERT(args->index2 >= 0);
+
+	if (entry1->flags & XFS_ATTR_LOCAL) {
+		name_loc = xfs_attr3_leaf_name_local(leaf1, args->index);
+		namelen1 = name_loc->namelen;
+		name1 = (char *)name_loc->nameval;
+	} else {
+		name_rmt = xfs_attr3_leaf_name_remote(leaf1, args->index);
+		namelen1 = name_rmt->namelen;
+		name1 = (char *)name_rmt->name;
+	}
+	if (entry2->flags & XFS_ATTR_LOCAL) {
+		name_loc = xfs_attr3_leaf_name_local(leaf2, args->index2);
+		namelen2 = name_loc->namelen;
+		name2 = (char *)name_loc->nameval;
+	} else {
+		name_rmt = xfs_attr3_leaf_name_remote(leaf2, args->index2);
+		namelen2 = name_rmt->namelen;
+		name2 = (char *)name_rmt->name;
+	}
+	ASSERT(be32_to_cpu(entry1->hashval) == be32_to_cpu(entry2->hashval));
+	ASSERT(namelen1 == namelen2);
+	ASSERT(memcmp(name1, name2, namelen1) == 0);
+#endif /* DEBUG */
+
+	ASSERT(entry1->flags & XFS_ATTR_INCOMPLETE);
+	ASSERT((entry2->flags & XFS_ATTR_INCOMPLETE) == 0);
+
+	entry1->flags &= ~XFS_ATTR_INCOMPLETE;
+	xfs_trans_log_buf(args->trans, bp1,
+			  XFS_DA_LOGRANGE(leaf1, entry1, sizeof(*entry1)));
+	if (args->rmtblkno) {
+		ASSERT((entry1->flags & XFS_ATTR_LOCAL) == 0);
+		name_rmt = xfs_attr3_leaf_name_remote(leaf1, args->index);
+		name_rmt->valueblk = cpu_to_be32(args->rmtblkno);
+		name_rmt->valuelen = cpu_to_be32(args->rmtvaluelen);
+		xfs_trans_log_buf(args->trans, bp1,
+			 XFS_DA_LOGRANGE(leaf1, name_rmt, sizeof(*name_rmt)));
+	}
+
+	entry2->flags |= XFS_ATTR_INCOMPLETE;
+	xfs_trans_log_buf(args->trans, bp2,
+			  XFS_DA_LOGRANGE(leaf2, entry2, sizeof(*entry2)));
+	if ((entry2->flags & XFS_ATTR_LOCAL) == 0) {
+		name_rmt = xfs_attr3_leaf_name_remote(leaf2, args->index2);
+		name_rmt->valueblk = 0;
+		name_rmt->valuelen = 0;
+		xfs_trans_log_buf(args->trans, bp2,
+			 XFS_DA_LOGRANGE(leaf2, name_rmt, sizeof(*name_rmt)));
+	}
+
+	return 0;
+}
diff --git a/fs/xfs/libxfs/xfs_attr_leaf.h b/fs/xfs/libxfs/xfs_attr_leaf.h
new file mode 100644
index 000000000..368f4d9fa
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_attr_leaf.h
@@ -0,0 +1,111 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2000,2002-2003,2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_ATTR_LEAF_H__
+#define	__XFS_ATTR_LEAF_H__
+
+struct attrlist;
+struct xfs_attr_list_context;
+struct xfs_da_args;
+struct xfs_da_state;
+struct xfs_da_state_blk;
+struct xfs_inode;
+struct xfs_trans;
+
+/*
+ * Incore version of the attribute leaf header.
+ */
+struct xfs_attr3_icleaf_hdr {
+	uint32_t	forw;
+	uint32_t	back;
+	uint16_t	magic;
+	uint16_t	count;
+	uint16_t	usedbytes;
+	/*
+	 * Firstused is 32-bit here instead of 16-bit like the on-disk variant
+	 * to support maximum fsb size of 64k without overflow issues throughout
+	 * the attr code. Instead, the overflow condition is handled on
+	 * conversion to/from disk.
+	 */
+	uint32_t	firstused;
+	__u8		holes;
+	struct {
+		uint16_t	base;
+		uint16_t	size;
+	} freemap[XFS_ATTR_LEAF_MAPSIZE];
+};
+
+/*========================================================================
+ * Function prototypes for the kernel.
+ *========================================================================*/
+
+/*
+ * Internal routines when attribute fork size < XFS_LITINO(mp).
+ */
+void	xfs_attr_shortform_create(struct xfs_da_args *args);
+void	xfs_attr_shortform_add(struct xfs_da_args *args, int forkoff);
+int	xfs_attr_shortform_lookup(struct xfs_da_args *args);
+int	xfs_attr_shortform_getvalue(struct xfs_da_args *args);
+int	xfs_attr_shortform_to_leaf(struct xfs_da_args *args);
+int	xfs_attr_sf_removename(struct xfs_da_args *args);
+int	xfs_attr_sf_findname(struct xfs_da_args *args,
+			     struct xfs_attr_sf_entry **sfep,
+			     unsigned int *basep);
+int	xfs_attr_shortform_allfit(struct xfs_buf *bp, struct xfs_inode *dp);
+int	xfs_attr_shortform_bytesfit(struct xfs_inode *dp, int bytes);
+xfs_failaddr_t xfs_attr_shortform_verify(struct xfs_inode *ip);
+void	xfs_attr_fork_remove(struct xfs_inode *ip, struct xfs_trans *tp);
+
+/*
+ * Internal routines when attribute fork size == XFS_LBSIZE(mp).
+ */
+int	xfs_attr3_leaf_to_node(struct xfs_da_args *args);
+int	xfs_attr3_leaf_to_shortform(struct xfs_buf *bp,
+				   struct xfs_da_args *args, int forkoff);
+int	xfs_attr3_leaf_clearflag(struct xfs_da_args *args);
+int	xfs_attr3_leaf_setflag(struct xfs_da_args *args);
+int	xfs_attr3_leaf_flipflags(struct xfs_da_args *args);
+
+/*
+ * Routines used for growing the Btree.
+ */
+int	xfs_attr3_leaf_split(struct xfs_da_state *state,
+				   struct xfs_da_state_blk *oldblk,
+				   struct xfs_da_state_blk *newblk);
+int	xfs_attr3_leaf_lookup_int(struct xfs_buf *leaf,
+					struct xfs_da_args *args);
+int	xfs_attr3_leaf_getvalue(struct xfs_buf *bp, struct xfs_da_args *args);
+int	xfs_attr3_leaf_add(struct xfs_buf *leaf_buffer,
+				 struct xfs_da_args *args);
+int	xfs_attr3_leaf_remove(struct xfs_buf *leaf_buffer,
+				    struct xfs_da_args *args);
+int	xfs_attr3_leaf_list_int(struct xfs_buf *bp,
+				struct xfs_attr_list_context *context);
+
+/*
+ * Routines used for shrinking the Btree.
+ */
+int	xfs_attr3_leaf_toosmall(struct xfs_da_state *state, int *retval);
+void	xfs_attr3_leaf_unbalance(struct xfs_da_state *state,
+				       struct xfs_da_state_blk *drop_blk,
+				       struct xfs_da_state_blk *save_blk);
+/*
+ * Utility routines.
+ */
+xfs_dahash_t	xfs_attr_leaf_lasthash(struct xfs_buf *bp, int *count);
+int	xfs_attr_leaf_order(struct xfs_buf *leaf1_bp,
+				   struct xfs_buf *leaf2_bp);
+int	xfs_attr_leaf_newentsize(struct xfs_da_args *args, int *local);
+int	xfs_attr3_leaf_read(struct xfs_trans *tp, struct xfs_inode *dp,
+			xfs_dablk_t bno, struct xfs_buf **bpp);
+void	xfs_attr3_leaf_hdr_from_disk(struct xfs_da_geometry *geo,
+				     struct xfs_attr3_icleaf_hdr *to,
+				     struct xfs_attr_leafblock *from);
+void	xfs_attr3_leaf_hdr_to_disk(struct xfs_da_geometry *geo,
+				   struct xfs_attr_leafblock *to,
+				   struct xfs_attr3_icleaf_hdr *from);
+
+#endif	/* __XFS_ATTR_LEAF_H__ */
diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c
new file mode 100644
index 000000000..d440393b4
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_attr_remote.c
@@ -0,0 +1,712 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_bit.h"
+#include "xfs_mount.h"
+#include "xfs_defer.h"
+#include "xfs_da_format.h"
+#include "xfs_da_btree.h"
+#include "xfs_inode.h"
+#include "xfs_trans.h"
+#include "xfs_bmap.h"
+#include "xfs_attr.h"
+#include "xfs_attr_remote.h"
+#include "xfs_trace.h"
+#include "xfs_error.h"
+
+#define ATTR_RMTVALUE_MAPSIZE	1	/* # of map entries at once */
+
+/*
+ * Remote Attribute Values
+ * =======================
+ *
+ * Remote extended attribute values are conceptually simple -- they're written
+ * to data blocks mapped by an inode's attribute fork, and they have an upper
+ * size limit of 64k.  Setting a value does not involve the XFS log.
+ *
+ * However, on a v5 filesystem, maximally sized remote attr values require one
+ * block more than 64k worth of space to hold both the remote attribute value
+ * header (64 bytes).  On a 4k block filesystem this results in a 68k buffer;
+ * on a 64k block filesystem, this would be a 128k buffer.  Note that the log
+ * format can only handle a dirty buffer of XFS_MAX_BLOCKSIZE length (64k).
+ * Therefore, we /must/ ensure that remote attribute value buffers never touch
+ * the logging system and therefore never have a log item.
+ */
+
+/*
+ * Each contiguous block has a header, so it is not just a simple attribute
+ * length to FSB conversion.
+ */
+int
+xfs_attr3_rmt_blocks(
+	struct xfs_mount *mp,
+	int		attrlen)
+{
+	if (xfs_has_crc(mp)) {
+		int buflen = XFS_ATTR3_RMT_BUF_SPACE(mp, mp->m_sb.sb_blocksize);
+		return (attrlen + buflen - 1) / buflen;
+	}
+	return XFS_B_TO_FSB(mp, attrlen);
+}
+
+/*
+ * Checking of the remote attribute header is split into two parts. The verifier
+ * does CRC, location and bounds checking, the unpacking function checks the
+ * attribute parameters and owner.
+ */
+static xfs_failaddr_t
+xfs_attr3_rmt_hdr_ok(
+	void			*ptr,
+	xfs_ino_t		ino,
+	uint32_t		offset,
+	uint32_t		size,
+	xfs_daddr_t		bno)
+{
+	struct xfs_attr3_rmt_hdr *rmt = ptr;
+
+	if (bno != be64_to_cpu(rmt->rm_blkno))
+		return __this_address;
+	if (offset != be32_to_cpu(rmt->rm_offset))
+		return __this_address;
+	if (size != be32_to_cpu(rmt->rm_bytes))
+		return __this_address;
+	if (ino != be64_to_cpu(rmt->rm_owner))
+		return __this_address;
+
+	/* ok */
+	return NULL;
+}
+
+static xfs_failaddr_t
+xfs_attr3_rmt_verify(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp,
+	void			*ptr,
+	int			fsbsize,
+	xfs_daddr_t		bno)
+{
+	struct xfs_attr3_rmt_hdr *rmt = ptr;
+
+	if (!xfs_verify_magic(bp, rmt->rm_magic))
+		return __this_address;
+	if (!uuid_equal(&rmt->rm_uuid, &mp->m_sb.sb_meta_uuid))
+		return __this_address;
+	if (be64_to_cpu(rmt->rm_blkno) != bno)
+		return __this_address;
+	if (be32_to_cpu(rmt->rm_bytes) > fsbsize - sizeof(*rmt))
+		return __this_address;
+	if (be32_to_cpu(rmt->rm_offset) +
+				be32_to_cpu(rmt->rm_bytes) > XFS_XATTR_SIZE_MAX)
+		return __this_address;
+	if (rmt->rm_owner == 0)
+		return __this_address;
+
+	return NULL;
+}
+
+static int
+__xfs_attr3_rmt_read_verify(
+	struct xfs_buf	*bp,
+	bool		check_crc,
+	xfs_failaddr_t	*failaddr)
+{
+	struct xfs_mount *mp = bp->b_mount;
+	char		*ptr;
+	int		len;
+	xfs_daddr_t	bno;
+	int		blksize = mp->m_attr_geo->blksize;
+
+	/* no verification of non-crc buffers */
+	if (!xfs_has_crc(mp))
+		return 0;
+
+	ptr = bp->b_addr;
+	bno = xfs_buf_daddr(bp);
+	len = BBTOB(bp->b_length);
+	ASSERT(len >= blksize);
+
+	while (len > 0) {
+		if (check_crc &&
+		    !xfs_verify_cksum(ptr, blksize, XFS_ATTR3_RMT_CRC_OFF)) {
+			*failaddr = __this_address;
+			return -EFSBADCRC;
+		}
+		*failaddr = xfs_attr3_rmt_verify(mp, bp, ptr, blksize, bno);
+		if (*failaddr)
+			return -EFSCORRUPTED;
+		len -= blksize;
+		ptr += blksize;
+		bno += BTOBB(blksize);
+	}
+
+	if (len != 0) {
+		*failaddr = __this_address;
+		return -EFSCORRUPTED;
+	}
+
+	return 0;
+}
+
+static void
+xfs_attr3_rmt_read_verify(
+	struct xfs_buf	*bp)
+{
+	xfs_failaddr_t	fa;
+	int		error;
+
+	error = __xfs_attr3_rmt_read_verify(bp, true, &fa);
+	if (error)
+		xfs_verifier_error(bp, error, fa);
+}
+
+static xfs_failaddr_t
+xfs_attr3_rmt_verify_struct(
+	struct xfs_buf	*bp)
+{
+	xfs_failaddr_t	fa;
+	int		error;
+
+	error = __xfs_attr3_rmt_read_verify(bp, false, &fa);
+	return error ? fa : NULL;
+}
+
+static void
+xfs_attr3_rmt_write_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount *mp = bp->b_mount;
+	xfs_failaddr_t	fa;
+	int		blksize = mp->m_attr_geo->blksize;
+	char		*ptr;
+	int		len;
+	xfs_daddr_t	bno;
+
+	/* no verification of non-crc buffers */
+	if (!xfs_has_crc(mp))
+		return;
+
+	ptr = bp->b_addr;
+	bno = xfs_buf_daddr(bp);
+	len = BBTOB(bp->b_length);
+	ASSERT(len >= blksize);
+
+	while (len > 0) {
+		struct xfs_attr3_rmt_hdr *rmt = (struct xfs_attr3_rmt_hdr *)ptr;
+
+		fa = xfs_attr3_rmt_verify(mp, bp, ptr, blksize, bno);
+		if (fa) {
+			xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+			return;
+		}
+
+		/*
+		 * Ensure we aren't writing bogus LSNs to disk. See
+		 * xfs_attr3_rmt_hdr_set() for the explanation.
+		 */
+		if (rmt->rm_lsn != cpu_to_be64(NULLCOMMITLSN)) {
+			xfs_verifier_error(bp, -EFSCORRUPTED, __this_address);
+			return;
+		}
+		xfs_update_cksum(ptr, blksize, XFS_ATTR3_RMT_CRC_OFF);
+
+		len -= blksize;
+		ptr += blksize;
+		bno += BTOBB(blksize);
+	}
+
+	if (len != 0)
+		xfs_verifier_error(bp, -EFSCORRUPTED, __this_address);
+}
+
+const struct xfs_buf_ops xfs_attr3_rmt_buf_ops = {
+	.name = "xfs_attr3_rmt",
+	.magic = { 0, cpu_to_be32(XFS_ATTR3_RMT_MAGIC) },
+	.verify_read = xfs_attr3_rmt_read_verify,
+	.verify_write = xfs_attr3_rmt_write_verify,
+	.verify_struct = xfs_attr3_rmt_verify_struct,
+};
+
+STATIC int
+xfs_attr3_rmt_hdr_set(
+	struct xfs_mount	*mp,
+	void			*ptr,
+	xfs_ino_t		ino,
+	uint32_t		offset,
+	uint32_t		size,
+	xfs_daddr_t		bno)
+{
+	struct xfs_attr3_rmt_hdr *rmt = ptr;
+
+	if (!xfs_has_crc(mp))
+		return 0;
+
+	rmt->rm_magic = cpu_to_be32(XFS_ATTR3_RMT_MAGIC);
+	rmt->rm_offset = cpu_to_be32(offset);
+	rmt->rm_bytes = cpu_to_be32(size);
+	uuid_copy(&rmt->rm_uuid, &mp->m_sb.sb_meta_uuid);
+	rmt->rm_owner = cpu_to_be64(ino);
+	rmt->rm_blkno = cpu_to_be64(bno);
+
+	/*
+	 * Remote attribute blocks are written synchronously, so we don't
+	 * have an LSN that we can stamp in them that makes any sense to log
+	 * recovery. To ensure that log recovery handles overwrites of these
+	 * blocks sanely (i.e. once they've been freed and reallocated as some
+	 * other type of metadata) we need to ensure that the LSN has a value
+	 * that tells log recovery to ignore the LSN and overwrite the buffer
+	 * with whatever is in it's log. To do this, we use the magic
+	 * NULLCOMMITLSN to indicate that the LSN is invalid.
+	 */
+	rmt->rm_lsn = cpu_to_be64(NULLCOMMITLSN);
+
+	return sizeof(struct xfs_attr3_rmt_hdr);
+}
+
+/*
+ * Helper functions to copy attribute data in and out of the one disk extents
+ */
+STATIC int
+xfs_attr_rmtval_copyout(
+	struct xfs_mount *mp,
+	struct xfs_buf	*bp,
+	xfs_ino_t	ino,
+	int		*offset,
+	int		*valuelen,
+	uint8_t		**dst)
+{
+	char		*src = bp->b_addr;
+	xfs_daddr_t	bno = xfs_buf_daddr(bp);
+	int		len = BBTOB(bp->b_length);
+	int		blksize = mp->m_attr_geo->blksize;
+
+	ASSERT(len >= blksize);
+
+	while (len > 0 && *valuelen > 0) {
+		int hdr_size = 0;
+		int byte_cnt = XFS_ATTR3_RMT_BUF_SPACE(mp, blksize);
+
+		byte_cnt = min(*valuelen, byte_cnt);
+
+		if (xfs_has_crc(mp)) {
+			if (xfs_attr3_rmt_hdr_ok(src, ino, *offset,
+						  byte_cnt, bno)) {
+				xfs_alert(mp,
+"remote attribute header mismatch bno/off/len/owner (0x%llx/0x%x/Ox%x/0x%llx)",
+					bno, *offset, byte_cnt, ino);
+				return -EFSCORRUPTED;
+			}
+			hdr_size = sizeof(struct xfs_attr3_rmt_hdr);
+		}
+
+		memcpy(*dst, src + hdr_size, byte_cnt);
+
+		/* roll buffer forwards */
+		len -= blksize;
+		src += blksize;
+		bno += BTOBB(blksize);
+
+		/* roll attribute data forwards */
+		*valuelen -= byte_cnt;
+		*dst += byte_cnt;
+		*offset += byte_cnt;
+	}
+	return 0;
+}
+
+STATIC void
+xfs_attr_rmtval_copyin(
+	struct xfs_mount *mp,
+	struct xfs_buf	*bp,
+	xfs_ino_t	ino,
+	int		*offset,
+	int		*valuelen,
+	uint8_t		**src)
+{
+	char		*dst = bp->b_addr;
+	xfs_daddr_t	bno = xfs_buf_daddr(bp);
+	int		len = BBTOB(bp->b_length);
+	int		blksize = mp->m_attr_geo->blksize;
+
+	ASSERT(len >= blksize);
+
+	while (len > 0 && *valuelen > 0) {
+		int hdr_size;
+		int byte_cnt = XFS_ATTR3_RMT_BUF_SPACE(mp, blksize);
+
+		byte_cnt = min(*valuelen, byte_cnt);
+		hdr_size = xfs_attr3_rmt_hdr_set(mp, dst, ino, *offset,
+						 byte_cnt, bno);
+
+		memcpy(dst + hdr_size, *src, byte_cnt);
+
+		/*
+		 * If this is the last block, zero the remainder of it.
+		 * Check that we are actually the last block, too.
+		 */
+		if (byte_cnt + hdr_size < blksize) {
+			ASSERT(*valuelen - byte_cnt == 0);
+			ASSERT(len == blksize);
+			memset(dst + hdr_size + byte_cnt, 0,
+					blksize - hdr_size - byte_cnt);
+		}
+
+		/* roll buffer forwards */
+		len -= blksize;
+		dst += blksize;
+		bno += BTOBB(blksize);
+
+		/* roll attribute data forwards */
+		*valuelen -= byte_cnt;
+		*src += byte_cnt;
+		*offset += byte_cnt;
+	}
+}
+
+/*
+ * Read the value associated with an attribute from the out-of-line buffer
+ * that we stored it in.
+ *
+ * Returns 0 on successful retrieval, otherwise an error.
+ */
+int
+xfs_attr_rmtval_get(
+	struct xfs_da_args	*args)
+{
+	struct xfs_bmbt_irec	map[ATTR_RMTVALUE_MAPSIZE];
+	struct xfs_mount	*mp = args->dp->i_mount;
+	struct xfs_buf		*bp;
+	xfs_dablk_t		lblkno = args->rmtblkno;
+	uint8_t			*dst = args->value;
+	int			valuelen;
+	int			nmap;
+	int			error;
+	int			blkcnt = args->rmtblkcnt;
+	int			i;
+	int			offset = 0;
+
+	trace_xfs_attr_rmtval_get(args);
+
+	ASSERT(args->valuelen != 0);
+	ASSERT(args->rmtvaluelen == args->valuelen);
+
+	valuelen = args->rmtvaluelen;
+	while (valuelen > 0) {
+		nmap = ATTR_RMTVALUE_MAPSIZE;
+		error = xfs_bmapi_read(args->dp, (xfs_fileoff_t)lblkno,
+				       blkcnt, map, &nmap,
+				       XFS_BMAPI_ATTRFORK);
+		if (error)
+			return error;
+		ASSERT(nmap >= 1);
+
+		for (i = 0; (i < nmap) && (valuelen > 0); i++) {
+			xfs_daddr_t	dblkno;
+			int		dblkcnt;
+
+			ASSERT((map[i].br_startblock != DELAYSTARTBLOCK) &&
+			       (map[i].br_startblock != HOLESTARTBLOCK));
+			dblkno = XFS_FSB_TO_DADDR(mp, map[i].br_startblock);
+			dblkcnt = XFS_FSB_TO_BB(mp, map[i].br_blockcount);
+			error = xfs_buf_read(mp->m_ddev_targp, dblkno, dblkcnt,
+					0, &bp, &xfs_attr3_rmt_buf_ops);
+			if (error)
+				return error;
+
+			error = xfs_attr_rmtval_copyout(mp, bp, args->dp->i_ino,
+							&offset, &valuelen,
+							&dst);
+			xfs_buf_relse(bp);
+			if (error)
+				return error;
+
+			/* roll attribute extent map forwards */
+			lblkno += map[i].br_blockcount;
+			blkcnt -= map[i].br_blockcount;
+		}
+	}
+	ASSERT(valuelen == 0);
+	return 0;
+}
+
+/*
+ * Find a "hole" in the attribute address space large enough for us to drop the
+ * new attributes value into
+ */
+int
+xfs_attr_rmt_find_hole(
+	struct xfs_da_args	*args)
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	int			error;
+	int			blkcnt;
+	xfs_fileoff_t		lfileoff = 0;
+
+	/*
+	 * Because CRC enable attributes have headers, we can't just do a
+	 * straight byte to FSB conversion and have to take the header space
+	 * into account.
+	 */
+	blkcnt = xfs_attr3_rmt_blocks(mp, args->rmtvaluelen);
+	error = xfs_bmap_first_unused(args->trans, args->dp, blkcnt, &lfileoff,
+						   XFS_ATTR_FORK);
+	if (error)
+		return error;
+
+	args->rmtblkno = (xfs_dablk_t)lfileoff;
+	args->rmtblkcnt = blkcnt;
+
+	return 0;
+}
+
+int
+xfs_attr_rmtval_set_value(
+	struct xfs_da_args	*args)
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	struct xfs_bmbt_irec	map;
+	xfs_dablk_t		lblkno;
+	uint8_t			*src = args->value;
+	int			blkcnt;
+	int			valuelen;
+	int			nmap;
+	int			error;
+	int			offset = 0;
+
+	/*
+	 * Roll through the "value", copying the attribute value to the
+	 * already-allocated blocks.  Blocks are written synchronously
+	 * so that we can know they are all on disk before we turn off
+	 * the INCOMPLETE flag.
+	 */
+	lblkno = args->rmtblkno;
+	blkcnt = args->rmtblkcnt;
+	valuelen = args->rmtvaluelen;
+	while (valuelen > 0) {
+		struct xfs_buf	*bp;
+		xfs_daddr_t	dblkno;
+		int		dblkcnt;
+
+		ASSERT(blkcnt > 0);
+
+		nmap = 1;
+		error = xfs_bmapi_read(dp, (xfs_fileoff_t)lblkno,
+				       blkcnt, &map, &nmap,
+				       XFS_BMAPI_ATTRFORK);
+		if (error)
+			return error;
+		ASSERT(nmap == 1);
+		ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
+		       (map.br_startblock != HOLESTARTBLOCK));
+
+		dblkno = XFS_FSB_TO_DADDR(mp, map.br_startblock),
+		dblkcnt = XFS_FSB_TO_BB(mp, map.br_blockcount);
+
+		error = xfs_buf_get(mp->m_ddev_targp, dblkno, dblkcnt, &bp);
+		if (error)
+			return error;
+		bp->b_ops = &xfs_attr3_rmt_buf_ops;
+
+		xfs_attr_rmtval_copyin(mp, bp, args->dp->i_ino, &offset,
+				       &valuelen, &src);
+
+		error = xfs_bwrite(bp);	/* GROT: NOTE: synchronous write */
+		xfs_buf_relse(bp);
+		if (error)
+			return error;
+
+
+		/* roll attribute extent map forwards */
+		lblkno += map.br_blockcount;
+		blkcnt -= map.br_blockcount;
+	}
+	ASSERT(valuelen == 0);
+	return 0;
+}
+
+/* Mark stale any incore buffers for the remote value. */
+int
+xfs_attr_rmtval_stale(
+	struct xfs_inode	*ip,
+	struct xfs_bmbt_irec	*map,
+	xfs_buf_flags_t		incore_flags)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_buf		*bp;
+	int			error;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+
+	if (XFS_IS_CORRUPT(mp, map->br_startblock == DELAYSTARTBLOCK) ||
+	    XFS_IS_CORRUPT(mp, map->br_startblock == HOLESTARTBLOCK))
+		return -EFSCORRUPTED;
+
+	error = xfs_buf_incore(mp->m_ddev_targp,
+			XFS_FSB_TO_DADDR(mp, map->br_startblock),
+			XFS_FSB_TO_BB(mp, map->br_blockcount),
+			incore_flags, &bp);
+	if (error) {
+		if (error == -ENOENT)
+			return 0;
+		return error;
+	}
+
+	xfs_buf_stale(bp);
+	xfs_buf_relse(bp);
+	return 0;
+}
+
+/*
+ * Find a hole for the attr and store it in the delayed attr context.  This
+ * initializes the context to roll through allocating an attr extent for a
+ * delayed attr operation
+ */
+int
+xfs_attr_rmtval_find_space(
+	struct xfs_attr_intent		*attr)
+{
+	struct xfs_da_args		*args = attr->xattri_da_args;
+	struct xfs_bmbt_irec		*map = &attr->xattri_map;
+	int				error;
+
+	attr->xattri_lblkno = 0;
+	attr->xattri_blkcnt = 0;
+	args->rmtblkcnt = 0;
+	args->rmtblkno = 0;
+	memset(map, 0, sizeof(struct xfs_bmbt_irec));
+
+	error = xfs_attr_rmt_find_hole(args);
+	if (error)
+		return error;
+
+	attr->xattri_blkcnt = args->rmtblkcnt;
+	attr->xattri_lblkno = args->rmtblkno;
+
+	return 0;
+}
+
+/*
+ * Write one block of the value associated with an attribute into the
+ * out-of-line buffer that we have defined for it. This is similar to a subset
+ * of xfs_attr_rmtval_set, but records the current block to the delayed attr
+ * context, and leaves transaction handling to the caller.
+ */
+int
+xfs_attr_rmtval_set_blk(
+	struct xfs_attr_intent		*attr)
+{
+	struct xfs_da_args		*args = attr->xattri_da_args;
+	struct xfs_inode		*dp = args->dp;
+	struct xfs_bmbt_irec		*map = &attr->xattri_map;
+	int nmap;
+	int error;
+
+	nmap = 1;
+	error = xfs_bmapi_write(args->trans, dp,
+			(xfs_fileoff_t)attr->xattri_lblkno,
+			attr->xattri_blkcnt, XFS_BMAPI_ATTRFORK, args->total,
+			map, &nmap);
+	if (error)
+		return error;
+
+	ASSERT(nmap == 1);
+	ASSERT((map->br_startblock != DELAYSTARTBLOCK) &&
+	       (map->br_startblock != HOLESTARTBLOCK));
+
+	/* roll attribute extent map forwards */
+	attr->xattri_lblkno += map->br_blockcount;
+	attr->xattri_blkcnt -= map->br_blockcount;
+
+	return 0;
+}
+
+/*
+ * Remove the value associated with an attribute by deleting the
+ * out-of-line buffer that it is stored on.
+ */
+int
+xfs_attr_rmtval_invalidate(
+	struct xfs_da_args	*args)
+{
+	xfs_dablk_t		lblkno;
+	int			blkcnt;
+	int			error;
+
+	/*
+	 * Roll through the "value", invalidating the attribute value's blocks.
+	 */
+	lblkno = args->rmtblkno;
+	blkcnt = args->rmtblkcnt;
+	while (blkcnt > 0) {
+		struct xfs_bmbt_irec	map;
+		int			nmap;
+
+		/*
+		 * Try to remember where we decided to put the value.
+		 */
+		nmap = 1;
+		error = xfs_bmapi_read(args->dp, (xfs_fileoff_t)lblkno,
+				       blkcnt, &map, &nmap, XFS_BMAPI_ATTRFORK);
+		if (error)
+			return error;
+		if (XFS_IS_CORRUPT(args->dp->i_mount, nmap != 1))
+			return -EFSCORRUPTED;
+		error = xfs_attr_rmtval_stale(args->dp, &map, XBF_TRYLOCK);
+		if (error)
+			return error;
+
+		lblkno += map.br_blockcount;
+		blkcnt -= map.br_blockcount;
+	}
+	return 0;
+}
+
+/*
+ * Remove the value associated with an attribute by deleting the out-of-line
+ * buffer that it is stored on. Returns -EAGAIN for the caller to refresh the
+ * transaction and re-call the function.  Callers should keep calling this
+ * routine until it returns something other than -EAGAIN.
+ */
+int
+xfs_attr_rmtval_remove(
+	struct xfs_attr_intent		*attr)
+{
+	struct xfs_da_args		*args = attr->xattri_da_args;
+	int				error, done;
+
+	/*
+	 * Unmap value blocks for this attr.
+	 */
+	error = xfs_bunmapi(args->trans, args->dp, args->rmtblkno,
+			    args->rmtblkcnt, XFS_BMAPI_ATTRFORK, 1, &done);
+	if (error)
+		return error;
+
+	/*
+	 * We don't need an explicit state here to pick up where we left off. We
+	 * can figure it out using the !done return code. The actual value of
+	 * attr->xattri_dela_state may be some value reminiscent of the calling
+	 * function, but it's value is irrelevant with in the context of this
+	 * function. Once we are done here, the next state is set as needed by
+	 * the parent
+	 */
+	if (!done) {
+		trace_xfs_attr_rmtval_remove_return(attr->xattri_dela_state,
+						    args->dp);
+		return -EAGAIN;
+	}
+
+	args->rmtblkno = 0;
+	args->rmtblkcnt = 0;
+	return 0;
+}
diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h
new file mode 100644
index 000000000..d097ec6c4
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_attr_remote.h
@@ -0,0 +1,20 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2013 Red Hat, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_ATTR_REMOTE_H__
+#define	__XFS_ATTR_REMOTE_H__
+
+int xfs_attr3_rmt_blocks(struct xfs_mount *mp, int attrlen);
+
+int xfs_attr_rmtval_get(struct xfs_da_args *args);
+int xfs_attr_rmtval_stale(struct xfs_inode *ip, struct xfs_bmbt_irec *map,
+		xfs_buf_flags_t incore_flags);
+int xfs_attr_rmtval_invalidate(struct xfs_da_args *args);
+int xfs_attr_rmtval_remove(struct xfs_attr_intent *attr);
+int xfs_attr_rmt_find_hole(struct xfs_da_args *args);
+int xfs_attr_rmtval_set_value(struct xfs_da_args *args);
+int xfs_attr_rmtval_set_blk(struct xfs_attr_intent *attr);
+int xfs_attr_rmtval_find_space(struct xfs_attr_intent *attr);
+#endif /* __XFS_ATTR_REMOTE_H__ */
diff --git a/fs/xfs/libxfs/xfs_attr_sf.h b/fs/xfs/libxfs/xfs_attr_sf.h
new file mode 100644
index 000000000..37578b369
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_attr_sf.h
@@ -0,0 +1,51 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2000,2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_ATTR_SF_H__
+#define	__XFS_ATTR_SF_H__
+
+/*
+ * Attribute storage when stored inside the inode.
+ *
+ * Small attribute lists are packed as tightly as possible so as
+ * to fit into the literal area of the inode.
+ */
+typedef struct xfs_attr_sf_hdr xfs_attr_sf_hdr_t;
+
+/*
+ * We generate this then sort it, attr_list() must return things in hash-order.
+ */
+typedef struct xfs_attr_sf_sort {
+	uint8_t		entno;		/* entry number in original list */
+	uint8_t		namelen;	/* length of name value (no null) */
+	uint8_t		valuelen;	/* length of value */
+	uint8_t		flags;		/* flags bits (see xfs_attr_leaf.h) */
+	xfs_dahash_t	hash;		/* this entry's hash value */
+	unsigned char	*name;		/* name value, pointer into buffer */
+} xfs_attr_sf_sort_t;
+
+#define XFS_ATTR_SF_ENTSIZE_MAX			/* max space for name&value */ \
+	((1 << (NBBY*(int)sizeof(uint8_t))) - 1)
+
+/* space name/value uses */
+static inline int xfs_attr_sf_entsize_byname(uint8_t nlen, uint8_t vlen)
+{
+	return sizeof(struct xfs_attr_sf_entry) + nlen + vlen;
+}
+
+/* space an entry uses */
+static inline int xfs_attr_sf_entsize(struct xfs_attr_sf_entry *sfep)
+{
+	return struct_size(sfep, nameval, sfep->namelen + sfep->valuelen);
+}
+
+/* next entry in struct */
+static inline struct xfs_attr_sf_entry *
+xfs_attr_sf_nextentry(struct xfs_attr_sf_entry *sfep)
+{
+	return (void *)sfep + xfs_attr_sf_entsize(sfep);
+}
+
+#endif	/* __XFS_ATTR_SF_H__ */
diff --git a/fs/xfs/libxfs/xfs_bit.c b/fs/xfs/libxfs/xfs_bit.c
new file mode 100644
index 000000000..40ce5f309
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_bit.c
@@ -0,0 +1,106 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_log_format.h"
+#include "xfs_bit.h"
+
+/*
+ * XFS bit manipulation routines, used in non-realtime code.
+ */
+
+/*
+ * Return whether bitmap is empty.
+ * Size is number of words in the bitmap, which is padded to word boundary
+ * Returns 1 for empty, 0 for non-empty.
+ */
+int
+xfs_bitmap_empty(uint *map, uint size)
+{
+	uint i;
+
+	for (i = 0; i < size; i++) {
+		if (map[i] != 0)
+			return 0;
+	}
+
+	return 1;
+}
+
+/*
+ * Count the number of contiguous bits set in the bitmap starting with bit
+ * start_bit.  Size is the size of the bitmap in words.
+ */
+int
+xfs_contig_bits(uint *map, uint	size, uint start_bit)
+{
+	uint * p = ((unsigned int *) map) + (start_bit >> BIT_TO_WORD_SHIFT);
+	uint result = 0;
+	uint tmp;
+
+	size <<= BIT_TO_WORD_SHIFT;
+
+	ASSERT(start_bit < size);
+	size -= start_bit & ~(NBWORD - 1);
+	start_bit &= (NBWORD - 1);
+	if (start_bit) {
+		tmp = *p++;
+		/* set to one first offset bits prior to start */
+		tmp |= (~0U >> (NBWORD-start_bit));
+		if (tmp != ~0U)
+			goto found;
+		result += NBWORD;
+		size -= NBWORD;
+	}
+	while (size) {
+		if ((tmp = *p++) != ~0U)
+			goto found;
+		result += NBWORD;
+		size -= NBWORD;
+	}
+	return result - start_bit;
+found:
+	return result + ffz(tmp) - start_bit;
+}
+
+/*
+ * This takes the bit number to start looking from and
+ * returns the next set bit from there.  It returns -1
+ * if there are no more bits set or the start bit is
+ * beyond the end of the bitmap.
+ *
+ * Size is the number of words, not bytes, in the bitmap.
+ */
+int xfs_next_bit(uint *map, uint size, uint start_bit)
+{
+	uint * p = ((unsigned int *) map) + (start_bit >> BIT_TO_WORD_SHIFT);
+	uint result = start_bit & ~(NBWORD - 1);
+	uint tmp;
+
+	size <<= BIT_TO_WORD_SHIFT;
+
+	if (start_bit >= size)
+		return -1;
+	size -= result;
+	start_bit &= (NBWORD - 1);
+	if (start_bit) {
+		tmp = *p++;
+		/* set to zero first offset bits prior to start */
+		tmp &= (~0U << start_bit);
+		if (tmp != 0U)
+			goto found;
+		result += NBWORD;
+		size -= NBWORD;
+	}
+	while (size) {
+		if ((tmp = *p++) != 0U)
+			goto found;
+		result += NBWORD;
+		size -= NBWORD;
+	}
+	return -1;
+found:
+	return result + ffs(tmp) - 1;
+}
diff --git a/fs/xfs/libxfs/xfs_bit.h b/fs/xfs/libxfs/xfs_bit.h
new file mode 100644
index 000000000..a04f266ae
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_bit.h
@@ -0,0 +1,75 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2000,2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_BIT_H__
+#define	__XFS_BIT_H__
+
+/*
+ * XFS bit manipulation routines.
+ */
+
+/*
+ * masks with n high/low bits set, 64-bit values
+ */
+static inline uint64_t xfs_mask64hi(int n)
+{
+	return (uint64_t)-1 << (64 - (n));
+}
+static inline uint32_t xfs_mask32lo(int n)
+{
+	return ((uint32_t)1 << (n)) - 1;
+}
+static inline uint64_t xfs_mask64lo(int n)
+{
+	return ((uint64_t)1 << (n)) - 1;
+}
+
+/* Get high bit set out of 32-bit argument, -1 if none set */
+static inline int xfs_highbit32(uint32_t v)
+{
+	return fls(v) - 1;
+}
+
+/* Get high bit set out of 64-bit argument, -1 if none set */
+static inline int xfs_highbit64(uint64_t v)
+{
+	return fls64(v) - 1;
+}
+
+/* Get low bit set out of 32-bit argument, -1 if none set */
+static inline int xfs_lowbit32(uint32_t v)
+{
+	return ffs(v) - 1;
+}
+
+/* Get low bit set out of 64-bit argument, -1 if none set */
+static inline int xfs_lowbit64(uint64_t v)
+{
+	uint32_t	w = (uint32_t)v;
+	int		n = 0;
+
+	if (w) {	/* lower bits */
+		n = ffs(w);
+	} else {	/* upper bits */
+		w = (uint32_t)(v >> 32);
+		if (w) {
+			n = ffs(w);
+			if (n)
+				n += 32;
+		}
+	}
+	return n - 1;
+}
+
+/* Return whether bitmap is empty (1 == empty) */
+extern int xfs_bitmap_empty(uint *map, uint size);
+
+/* Count continuous one bits in map starting with start_bit */
+extern int xfs_contig_bits(uint *map, uint size, uint start_bit);
+
+/* Find next set bit in map */
+extern int xfs_next_bit(uint *map, uint size, uint start_bit);
+
+#endif	/* __XFS_BIT_H__ */
diff --git a/fs/xfs/libxfs/xfs_bmap.c b/fs/xfs/libxfs/xfs_bmap.c
new file mode 100644
index 000000000..49d0d4ea6
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_bmap.c
@@ -0,0 +1,6230 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_bit.h"
+#include "xfs_sb.h"
+#include "xfs_mount.h"
+#include "xfs_defer.h"
+#include "xfs_dir2.h"
+#include "xfs_inode.h"
+#include "xfs_btree.h"
+#include "xfs_trans.h"
+#include "xfs_alloc.h"
+#include "xfs_bmap.h"
+#include "xfs_bmap_util.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_rtalloc.h"
+#include "xfs_errortag.h"
+#include "xfs_error.h"
+#include "xfs_quota.h"
+#include "xfs_trans_space.h"
+#include "xfs_buf_item.h"
+#include "xfs_trace.h"
+#include "xfs_attr_leaf.h"
+#include "xfs_filestream.h"
+#include "xfs_rmap.h"
+#include "xfs_ag.h"
+#include "xfs_ag_resv.h"
+#include "xfs_refcount.h"
+#include "xfs_icache.h"
+#include "xfs_iomap.h"
+
+struct kmem_cache		*xfs_bmap_intent_cache;
+
+/*
+ * Miscellaneous helper functions
+ */
+
+/*
+ * Compute and fill in the value of the maximum depth of a bmap btree
+ * in this filesystem.  Done once, during mount.
+ */
+void
+xfs_bmap_compute_maxlevels(
+	xfs_mount_t	*mp,		/* file system mount structure */
+	int		whichfork)	/* data or attr fork */
+{
+	uint64_t	maxblocks;	/* max blocks at this level */
+	xfs_extnum_t	maxleafents;	/* max leaf entries possible */
+	int		level;		/* btree level */
+	int		maxrootrecs;	/* max records in root block */
+	int		minleafrecs;	/* min records in leaf block */
+	int		minnoderecs;	/* min records in node block */
+	int		sz;		/* root block size */
+
+	/*
+	 * The maximum number of extents in a fork, hence the maximum number of
+	 * leaf entries, is controlled by the size of the on-disk extent count.
+	 *
+	 * Note that we can no longer assume that if we are in ATTR1 that the
+	 * fork offset of all the inodes will be
+	 * (xfs_default_attroffset(ip) >> 3) because we could have mounted with
+	 * ATTR2 and then mounted back with ATTR1, keeping the i_forkoff's fixed
+	 * but probably at various positions. Therefore, for both ATTR1 and
+	 * ATTR2 we have to assume the worst case scenario of a minimum size
+	 * available.
+	 */
+	maxleafents = xfs_iext_max_nextents(xfs_has_large_extent_counts(mp),
+				whichfork);
+	if (whichfork == XFS_DATA_FORK)
+		sz = XFS_BMDR_SPACE_CALC(MINDBTPTRS);
+	else
+		sz = XFS_BMDR_SPACE_CALC(MINABTPTRS);
+
+	maxrootrecs = xfs_bmdr_maxrecs(sz, 0);
+	minleafrecs = mp->m_bmap_dmnr[0];
+	minnoderecs = mp->m_bmap_dmnr[1];
+	maxblocks = howmany_64(maxleafents, minleafrecs);
+	for (level = 1; maxblocks > 1; level++) {
+		if (maxblocks <= maxrootrecs)
+			maxblocks = 1;
+		else
+			maxblocks = howmany_64(maxblocks, minnoderecs);
+	}
+	mp->m_bm_maxlevels[whichfork] = level;
+	ASSERT(mp->m_bm_maxlevels[whichfork] <= xfs_bmbt_maxlevels_ondisk());
+}
+
+unsigned int
+xfs_bmap_compute_attr_offset(
+	struct xfs_mount	*mp)
+{
+	if (mp->m_sb.sb_inodesize == 256)
+		return XFS_LITINO(mp) - XFS_BMDR_SPACE_CALC(MINABTPTRS);
+	return XFS_BMDR_SPACE_CALC(6 * MINABTPTRS);
+}
+
+STATIC int				/* error */
+xfs_bmbt_lookup_eq(
+	struct xfs_btree_cur	*cur,
+	struct xfs_bmbt_irec	*irec,
+	int			*stat)	/* success/failure */
+{
+	cur->bc_rec.b = *irec;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_EQ, stat);
+}
+
+STATIC int				/* error */
+xfs_bmbt_lookup_first(
+	struct xfs_btree_cur	*cur,
+	int			*stat)	/* success/failure */
+{
+	cur->bc_rec.b.br_startoff = 0;
+	cur->bc_rec.b.br_startblock = 0;
+	cur->bc_rec.b.br_blockcount = 0;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_GE, stat);
+}
+
+/*
+ * Check if the inode needs to be converted to btree format.
+ */
+static inline bool xfs_bmap_needs_btree(struct xfs_inode *ip, int whichfork)
+{
+	struct xfs_ifork *ifp = xfs_ifork_ptr(ip, whichfork);
+
+	return whichfork != XFS_COW_FORK &&
+		ifp->if_format == XFS_DINODE_FMT_EXTENTS &&
+		ifp->if_nextents > XFS_IFORK_MAXEXT(ip, whichfork);
+}
+
+/*
+ * Check if the inode should be converted to extent format.
+ */
+static inline bool xfs_bmap_wants_extents(struct xfs_inode *ip, int whichfork)
+{
+	struct xfs_ifork *ifp = xfs_ifork_ptr(ip, whichfork);
+
+	return whichfork != XFS_COW_FORK &&
+		ifp->if_format == XFS_DINODE_FMT_BTREE &&
+		ifp->if_nextents <= XFS_IFORK_MAXEXT(ip, whichfork);
+}
+
+/*
+ * Update the record referred to by cur to the value given by irec
+ * This either works (return 0) or gets an EFSCORRUPTED error.
+ */
+STATIC int
+xfs_bmbt_update(
+	struct xfs_btree_cur	*cur,
+	struct xfs_bmbt_irec	*irec)
+{
+	union xfs_btree_rec	rec;
+
+	xfs_bmbt_disk_set_all(&rec.bmbt, irec);
+	return xfs_btree_update(cur, &rec);
+}
+
+/*
+ * Compute the worst-case number of indirect blocks that will be used
+ * for ip's delayed extent of length "len".
+ */
+STATIC xfs_filblks_t
+xfs_bmap_worst_indlen(
+	xfs_inode_t	*ip,		/* incore inode pointer */
+	xfs_filblks_t	len)		/* delayed extent length */
+{
+	int		level;		/* btree level number */
+	int		maxrecs;	/* maximum record count at this level */
+	xfs_mount_t	*mp;		/* mount structure */
+	xfs_filblks_t	rval;		/* return value */
+
+	mp = ip->i_mount;
+	maxrecs = mp->m_bmap_dmxr[0];
+	for (level = 0, rval = 0;
+	     level < XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK);
+	     level++) {
+		len += maxrecs - 1;
+		do_div(len, maxrecs);
+		rval += len;
+		if (len == 1)
+			return rval + XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK) -
+				level - 1;
+		if (level == 0)
+			maxrecs = mp->m_bmap_dmxr[1];
+	}
+	return rval;
+}
+
+/*
+ * Calculate the default attribute fork offset for newly created inodes.
+ */
+uint
+xfs_default_attroffset(
+	struct xfs_inode	*ip)
+{
+	if (ip->i_df.if_format == XFS_DINODE_FMT_DEV)
+		return roundup(sizeof(xfs_dev_t), 8);
+	return M_IGEO(ip->i_mount)->attr_fork_offset;
+}
+
+/*
+ * Helper routine to reset inode i_forkoff field when switching attribute fork
+ * from local to extent format - we reset it where possible to make space
+ * available for inline data fork extents.
+ */
+STATIC void
+xfs_bmap_forkoff_reset(
+	xfs_inode_t	*ip,
+	int		whichfork)
+{
+	if (whichfork == XFS_ATTR_FORK &&
+	    ip->i_df.if_format != XFS_DINODE_FMT_DEV &&
+	    ip->i_df.if_format != XFS_DINODE_FMT_BTREE) {
+		uint	dfl_forkoff = xfs_default_attroffset(ip) >> 3;
+
+		if (dfl_forkoff > ip->i_forkoff)
+			ip->i_forkoff = dfl_forkoff;
+	}
+}
+
+#ifdef DEBUG
+STATIC struct xfs_buf *
+xfs_bmap_get_bp(
+	struct xfs_btree_cur	*cur,
+	xfs_fsblock_t		bno)
+{
+	struct xfs_log_item	*lip;
+	int			i;
+
+	if (!cur)
+		return NULL;
+
+	for (i = 0; i < cur->bc_maxlevels; i++) {
+		if (!cur->bc_levels[i].bp)
+			break;
+		if (xfs_buf_daddr(cur->bc_levels[i].bp) == bno)
+			return cur->bc_levels[i].bp;
+	}
+
+	/* Chase down all the log items to see if the bp is there */
+	list_for_each_entry(lip, &cur->bc_tp->t_items, li_trans) {
+		struct xfs_buf_log_item	*bip = (struct xfs_buf_log_item *)lip;
+
+		if (bip->bli_item.li_type == XFS_LI_BUF &&
+		    xfs_buf_daddr(bip->bli_buf) == bno)
+			return bip->bli_buf;
+	}
+
+	return NULL;
+}
+
+STATIC void
+xfs_check_block(
+	struct xfs_btree_block	*block,
+	xfs_mount_t		*mp,
+	int			root,
+	short			sz)
+{
+	int			i, j, dmxr;
+	__be64			*pp, *thispa;	/* pointer to block address */
+	xfs_bmbt_key_t		*prevp, *keyp;
+
+	ASSERT(be16_to_cpu(block->bb_level) > 0);
+
+	prevp = NULL;
+	for( i = 1; i <= xfs_btree_get_numrecs(block); i++) {
+		dmxr = mp->m_bmap_dmxr[0];
+		keyp = XFS_BMBT_KEY_ADDR(mp, block, i);
+
+		if (prevp) {
+			ASSERT(be64_to_cpu(prevp->br_startoff) <
+			       be64_to_cpu(keyp->br_startoff));
+		}
+		prevp = keyp;
+
+		/*
+		 * Compare the block numbers to see if there are dups.
+		 */
+		if (root)
+			pp = XFS_BMAP_BROOT_PTR_ADDR(mp, block, i, sz);
+		else
+			pp = XFS_BMBT_PTR_ADDR(mp, block, i, dmxr);
+
+		for (j = i+1; j <= be16_to_cpu(block->bb_numrecs); j++) {
+			if (root)
+				thispa = XFS_BMAP_BROOT_PTR_ADDR(mp, block, j, sz);
+			else
+				thispa = XFS_BMBT_PTR_ADDR(mp, block, j, dmxr);
+			if (*thispa == *pp) {
+				xfs_warn(mp, "%s: thispa(%d) == pp(%d) %lld",
+					__func__, j, i,
+					(unsigned long long)be64_to_cpu(*thispa));
+				xfs_err(mp, "%s: ptrs are equal in node\n",
+					__func__);
+				xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
+			}
+		}
+	}
+}
+
+/*
+ * Check that the extents for the inode ip are in the right order in all
+ * btree leaves. THis becomes prohibitively expensive for large extent count
+ * files, so don't bother with inodes that have more than 10,000 extents in
+ * them. The btree record ordering checks will still be done, so for such large
+ * bmapbt constructs that is going to catch most corruptions.
+ */
+STATIC void
+xfs_bmap_check_leaf_extents(
+	struct xfs_btree_cur	*cur,	/* btree cursor or null */
+	xfs_inode_t		*ip,		/* incore inode pointer */
+	int			whichfork)	/* data or attr fork */
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_btree_block	*block;	/* current btree block */
+	xfs_fsblock_t		bno;	/* block # of "block" */
+	struct xfs_buf		*bp;	/* buffer for "block" */
+	int			error;	/* error return value */
+	xfs_extnum_t		i=0, j;	/* index into the extents list */
+	int			level;	/* btree level, for checking */
+	__be64			*pp;	/* pointer to block address */
+	xfs_bmbt_rec_t		*ep;	/* pointer to current extent */
+	xfs_bmbt_rec_t		last = {0, 0}; /* last extent in prev block */
+	xfs_bmbt_rec_t		*nextp;	/* pointer to next extent */
+	int			bp_release = 0;
+
+	if (ifp->if_format != XFS_DINODE_FMT_BTREE)
+		return;
+
+	/* skip large extent count inodes */
+	if (ip->i_df.if_nextents > 10000)
+		return;
+
+	bno = NULLFSBLOCK;
+	block = ifp->if_broot;
+	/*
+	 * Root level must use BMAP_BROOT_PTR_ADDR macro to get ptr out.
+	 */
+	level = be16_to_cpu(block->bb_level);
+	ASSERT(level > 0);
+	xfs_check_block(block, mp, 1, ifp->if_broot_bytes);
+	pp = XFS_BMAP_BROOT_PTR_ADDR(mp, block, 1, ifp->if_broot_bytes);
+	bno = be64_to_cpu(*pp);
+
+	ASSERT(bno != NULLFSBLOCK);
+	ASSERT(XFS_FSB_TO_AGNO(mp, bno) < mp->m_sb.sb_agcount);
+	ASSERT(XFS_FSB_TO_AGBNO(mp, bno) < mp->m_sb.sb_agblocks);
+
+	/*
+	 * Go down the tree until leaf level is reached, following the first
+	 * pointer (leftmost) at each level.
+	 */
+	while (level-- > 0) {
+		/* See if buf is in cur first */
+		bp_release = 0;
+		bp = xfs_bmap_get_bp(cur, XFS_FSB_TO_DADDR(mp, bno));
+		if (!bp) {
+			bp_release = 1;
+			error = xfs_btree_read_bufl(mp, NULL, bno, &bp,
+						XFS_BMAP_BTREE_REF,
+						&xfs_bmbt_buf_ops);
+			if (error)
+				goto error_norelse;
+		}
+		block = XFS_BUF_TO_BLOCK(bp);
+		if (level == 0)
+			break;
+
+		/*
+		 * Check this block for basic sanity (increasing keys and
+		 * no duplicate blocks).
+		 */
+
+		xfs_check_block(block, mp, 0, 0);
+		pp = XFS_BMBT_PTR_ADDR(mp, block, 1, mp->m_bmap_dmxr[1]);
+		bno = be64_to_cpu(*pp);
+		if (XFS_IS_CORRUPT(mp, !xfs_verify_fsbno(mp, bno))) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+		if (bp_release) {
+			bp_release = 0;
+			xfs_trans_brelse(NULL, bp);
+		}
+	}
+
+	/*
+	 * Here with bp and block set to the leftmost leaf node in the tree.
+	 */
+	i = 0;
+
+	/*
+	 * Loop over all leaf nodes checking that all extents are in the right order.
+	 */
+	for (;;) {
+		xfs_fsblock_t	nextbno;
+		xfs_extnum_t	num_recs;
+
+
+		num_recs = xfs_btree_get_numrecs(block);
+
+		/*
+		 * Read-ahead the next leaf block, if any.
+		 */
+
+		nextbno = be64_to_cpu(block->bb_u.l.bb_rightsib);
+
+		/*
+		 * Check all the extents to make sure they are OK.
+		 * If we had a previous block, the last entry should
+		 * conform with the first entry in this one.
+		 */
+
+		ep = XFS_BMBT_REC_ADDR(mp, block, 1);
+		if (i) {
+			ASSERT(xfs_bmbt_disk_get_startoff(&last) +
+			       xfs_bmbt_disk_get_blockcount(&last) <=
+			       xfs_bmbt_disk_get_startoff(ep));
+		}
+		for (j = 1; j < num_recs; j++) {
+			nextp = XFS_BMBT_REC_ADDR(mp, block, j + 1);
+			ASSERT(xfs_bmbt_disk_get_startoff(ep) +
+			       xfs_bmbt_disk_get_blockcount(ep) <=
+			       xfs_bmbt_disk_get_startoff(nextp));
+			ep = nextp;
+		}
+
+		last = *ep;
+		i += num_recs;
+		if (bp_release) {
+			bp_release = 0;
+			xfs_trans_brelse(NULL, bp);
+		}
+		bno = nextbno;
+		/*
+		 * If we've reached the end, stop.
+		 */
+		if (bno == NULLFSBLOCK)
+			break;
+
+		bp_release = 0;
+		bp = xfs_bmap_get_bp(cur, XFS_FSB_TO_DADDR(mp, bno));
+		if (!bp) {
+			bp_release = 1;
+			error = xfs_btree_read_bufl(mp, NULL, bno, &bp,
+						XFS_BMAP_BTREE_REF,
+						&xfs_bmbt_buf_ops);
+			if (error)
+				goto error_norelse;
+		}
+		block = XFS_BUF_TO_BLOCK(bp);
+	}
+
+	return;
+
+error0:
+	xfs_warn(mp, "%s: at error0", __func__);
+	if (bp_release)
+		xfs_trans_brelse(NULL, bp);
+error_norelse:
+	xfs_warn(mp, "%s: BAD after btree leaves for %llu extents",
+		__func__, i);
+	xfs_err(mp, "%s: CORRUPTED BTREE OR SOMETHING", __func__);
+	xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
+	return;
+}
+
+/*
+ * Validate that the bmbt_irecs being returned from bmapi are valid
+ * given the caller's original parameters.  Specifically check the
+ * ranges of the returned irecs to ensure that they only extend beyond
+ * the given parameters if the XFS_BMAPI_ENTIRE flag was set.
+ */
+STATIC void
+xfs_bmap_validate_ret(
+	xfs_fileoff_t		bno,
+	xfs_filblks_t		len,
+	uint32_t		flags,
+	xfs_bmbt_irec_t		*mval,
+	int			nmap,
+	int			ret_nmap)
+{
+	int			i;		/* index to map values */
+
+	ASSERT(ret_nmap <= nmap);
+
+	for (i = 0; i < ret_nmap; i++) {
+		ASSERT(mval[i].br_blockcount > 0);
+		if (!(flags & XFS_BMAPI_ENTIRE)) {
+			ASSERT(mval[i].br_startoff >= bno);
+			ASSERT(mval[i].br_blockcount <= len);
+			ASSERT(mval[i].br_startoff + mval[i].br_blockcount <=
+			       bno + len);
+		} else {
+			ASSERT(mval[i].br_startoff < bno + len);
+			ASSERT(mval[i].br_startoff + mval[i].br_blockcount >
+			       bno);
+		}
+		ASSERT(i == 0 ||
+		       mval[i - 1].br_startoff + mval[i - 1].br_blockcount ==
+		       mval[i].br_startoff);
+		ASSERT(mval[i].br_startblock != DELAYSTARTBLOCK &&
+		       mval[i].br_startblock != HOLESTARTBLOCK);
+		ASSERT(mval[i].br_state == XFS_EXT_NORM ||
+		       mval[i].br_state == XFS_EXT_UNWRITTEN);
+	}
+}
+
+#else
+#define xfs_bmap_check_leaf_extents(cur, ip, whichfork)		do { } while (0)
+#define	xfs_bmap_validate_ret(bno,len,flags,mval,onmap,nmap)	do { } while (0)
+#endif /* DEBUG */
+
+/*
+ * Inode fork format manipulation functions
+ */
+
+/*
+ * Convert the inode format to extent format if it currently is in btree format,
+ * but the extent list is small enough that it fits into the extent format.
+ *
+ * Since the extents are already in-core, all we have to do is give up the space
+ * for the btree root and pitch the leaf block.
+ */
+STATIC int				/* error */
+xfs_bmap_btree_to_extents(
+	struct xfs_trans	*tp,	/* transaction pointer */
+	struct xfs_inode	*ip,	/* incore inode pointer */
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			*logflagsp, /* inode logging flags */
+	int			whichfork)  /* data or attr fork */
+{
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_btree_block	*rblock = ifp->if_broot;
+	struct xfs_btree_block	*cblock;/* child btree block */
+	xfs_fsblock_t		cbno;	/* child block number */
+	struct xfs_buf		*cbp;	/* child block's buffer */
+	int			error;	/* error return value */
+	__be64			*pp;	/* ptr to block address */
+	struct xfs_owner_info	oinfo;
+
+	/* check if we actually need the extent format first: */
+	if (!xfs_bmap_wants_extents(ip, whichfork))
+		return 0;
+
+	ASSERT(cur);
+	ASSERT(whichfork != XFS_COW_FORK);
+	ASSERT(ifp->if_format == XFS_DINODE_FMT_BTREE);
+	ASSERT(be16_to_cpu(rblock->bb_level) == 1);
+	ASSERT(be16_to_cpu(rblock->bb_numrecs) == 1);
+	ASSERT(xfs_bmbt_maxrecs(mp, ifp->if_broot_bytes, 0) == 1);
+
+	pp = XFS_BMAP_BROOT_PTR_ADDR(mp, rblock, 1, ifp->if_broot_bytes);
+	cbno = be64_to_cpu(*pp);
+#ifdef DEBUG
+	if (XFS_IS_CORRUPT(cur->bc_mp, !xfs_btree_check_lptr(cur, cbno, 1)))
+		return -EFSCORRUPTED;
+#endif
+	error = xfs_btree_read_bufl(mp, tp, cbno, &cbp, XFS_BMAP_BTREE_REF,
+				&xfs_bmbt_buf_ops);
+	if (error)
+		return error;
+	cblock = XFS_BUF_TO_BLOCK(cbp);
+	if ((error = xfs_btree_check_block(cur, cblock, 0, cbp)))
+		return error;
+	xfs_rmap_ino_bmbt_owner(&oinfo, ip->i_ino, whichfork);
+	xfs_free_extent_later(cur->bc_tp, cbno, 1, &oinfo);
+	ip->i_nblocks--;
+	xfs_trans_mod_dquot_byino(tp, ip, XFS_TRANS_DQ_BCOUNT, -1L);
+	xfs_trans_binval(tp, cbp);
+	if (cur->bc_levels[0].bp == cbp)
+		cur->bc_levels[0].bp = NULL;
+	xfs_iroot_realloc(ip, -1, whichfork);
+	ASSERT(ifp->if_broot == NULL);
+	ifp->if_format = XFS_DINODE_FMT_EXTENTS;
+	*logflagsp |= XFS_ILOG_CORE | xfs_ilog_fext(whichfork);
+	return 0;
+}
+
+/*
+ * Convert an extents-format file into a btree-format file.
+ * The new file will have a root block (in the inode) and a single child block.
+ */
+STATIC int					/* error */
+xfs_bmap_extents_to_btree(
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* incore inode pointer */
+	struct xfs_btree_cur	**curp,		/* cursor returned to caller */
+	int			wasdel,		/* converting a delayed alloc */
+	int			*logflagsp,	/* inode logging flags */
+	int			whichfork)	/* data or attr fork */
+{
+	struct xfs_btree_block	*ablock;	/* allocated (child) bt block */
+	struct xfs_buf		*abp;		/* buffer for ablock */
+	struct xfs_alloc_arg	args;		/* allocation arguments */
+	struct xfs_bmbt_rec	*arp;		/* child record pointer */
+	struct xfs_btree_block	*block;		/* btree root block */
+	struct xfs_btree_cur	*cur;		/* bmap btree cursor */
+	int			error;		/* error return value */
+	struct xfs_ifork	*ifp;		/* inode fork pointer */
+	struct xfs_bmbt_key	*kp;		/* root block key pointer */
+	struct xfs_mount	*mp;		/* mount structure */
+	xfs_bmbt_ptr_t		*pp;		/* root block address pointer */
+	struct xfs_iext_cursor	icur;
+	struct xfs_bmbt_irec	rec;
+	xfs_extnum_t		cnt = 0;
+
+	mp = ip->i_mount;
+	ASSERT(whichfork != XFS_COW_FORK);
+	ifp = xfs_ifork_ptr(ip, whichfork);
+	ASSERT(ifp->if_format == XFS_DINODE_FMT_EXTENTS);
+
+	/*
+	 * Make space in the inode incore. This needs to be undone if we fail
+	 * to expand the root.
+	 */
+	xfs_iroot_realloc(ip, 1, whichfork);
+
+	/*
+	 * Fill in the root.
+	 */
+	block = ifp->if_broot;
+	xfs_btree_init_block_int(mp, block, XFS_BUF_DADDR_NULL,
+				 XFS_BTNUM_BMAP, 1, 1, ip->i_ino,
+				 XFS_BTREE_LONG_PTRS);
+	/*
+	 * Need a cursor.  Can't allocate until bb_level is filled in.
+	 */
+	cur = xfs_bmbt_init_cursor(mp, tp, ip, whichfork);
+	cur->bc_ino.flags = wasdel ? XFS_BTCUR_BMBT_WASDEL : 0;
+	/*
+	 * Convert to a btree with two levels, one record in root.
+	 */
+	ifp->if_format = XFS_DINODE_FMT_BTREE;
+	memset(&args, 0, sizeof(args));
+	args.tp = tp;
+	args.mp = mp;
+	xfs_rmap_ino_bmbt_owner(&args.oinfo, ip->i_ino, whichfork);
+	if (tp->t_firstblock == NULLFSBLOCK) {
+		args.type = XFS_ALLOCTYPE_START_BNO;
+		args.fsbno = XFS_INO_TO_FSB(mp, ip->i_ino);
+	} else if (tp->t_flags & XFS_TRANS_LOWMODE) {
+		args.type = XFS_ALLOCTYPE_START_BNO;
+		args.fsbno = tp->t_firstblock;
+	} else {
+		args.type = XFS_ALLOCTYPE_NEAR_BNO;
+		args.fsbno = tp->t_firstblock;
+	}
+	args.minlen = args.maxlen = args.prod = 1;
+	args.wasdel = wasdel;
+	*logflagsp = 0;
+	error = xfs_alloc_vextent(&args);
+	if (error)
+		goto out_root_realloc;
+
+	if (WARN_ON_ONCE(args.fsbno == NULLFSBLOCK)) {
+		error = -ENOSPC;
+		goto out_root_realloc;
+	}
+
+	/*
+	 * Allocation can't fail, the space was reserved.
+	 */
+	ASSERT(tp->t_firstblock == NULLFSBLOCK ||
+	       args.agno >= XFS_FSB_TO_AGNO(mp, tp->t_firstblock));
+	tp->t_firstblock = args.fsbno;
+	cur->bc_ino.allocated++;
+	ip->i_nblocks++;
+	xfs_trans_mod_dquot_byino(tp, ip, XFS_TRANS_DQ_BCOUNT, 1L);
+	error = xfs_trans_get_buf(tp, mp->m_ddev_targp,
+			XFS_FSB_TO_DADDR(mp, args.fsbno),
+			mp->m_bsize, 0, &abp);
+	if (error)
+		goto out_unreserve_dquot;
+
+	/*
+	 * Fill in the child block.
+	 */
+	abp->b_ops = &xfs_bmbt_buf_ops;
+	ablock = XFS_BUF_TO_BLOCK(abp);
+	xfs_btree_init_block_int(mp, ablock, xfs_buf_daddr(abp),
+				XFS_BTNUM_BMAP, 0, 0, ip->i_ino,
+				XFS_BTREE_LONG_PTRS);
+
+	for_each_xfs_iext(ifp, &icur, &rec) {
+		if (isnullstartblock(rec.br_startblock))
+			continue;
+		arp = XFS_BMBT_REC_ADDR(mp, ablock, 1 + cnt);
+		xfs_bmbt_disk_set_all(arp, &rec);
+		cnt++;
+	}
+	ASSERT(cnt == ifp->if_nextents);
+	xfs_btree_set_numrecs(ablock, cnt);
+
+	/*
+	 * Fill in the root key and pointer.
+	 */
+	kp = XFS_BMBT_KEY_ADDR(mp, block, 1);
+	arp = XFS_BMBT_REC_ADDR(mp, ablock, 1);
+	kp->br_startoff = cpu_to_be64(xfs_bmbt_disk_get_startoff(arp));
+	pp = XFS_BMBT_PTR_ADDR(mp, block, 1, xfs_bmbt_get_maxrecs(cur,
+						be16_to_cpu(block->bb_level)));
+	*pp = cpu_to_be64(args.fsbno);
+
+	/*
+	 * Do all this logging at the end so that
+	 * the root is at the right level.
+	 */
+	xfs_btree_log_block(cur, abp, XFS_BB_ALL_BITS);
+	xfs_btree_log_recs(cur, abp, 1, be16_to_cpu(ablock->bb_numrecs));
+	ASSERT(*curp == NULL);
+	*curp = cur;
+	*logflagsp = XFS_ILOG_CORE | xfs_ilog_fbroot(whichfork);
+	return 0;
+
+out_unreserve_dquot:
+	xfs_trans_mod_dquot_byino(tp, ip, XFS_TRANS_DQ_BCOUNT, -1L);
+out_root_realloc:
+	xfs_iroot_realloc(ip, -1, whichfork);
+	ifp->if_format = XFS_DINODE_FMT_EXTENTS;
+	ASSERT(ifp->if_broot == NULL);
+	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+
+	return error;
+}
+
+/*
+ * Convert a local file to an extents file.
+ * This code is out of bounds for data forks of regular files,
+ * since the file data needs to get logged so things will stay consistent.
+ * (The bmap-level manipulations are ok, though).
+ */
+void
+xfs_bmap_local_to_extents_empty(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	int			whichfork)
+{
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+
+	ASSERT(whichfork != XFS_COW_FORK);
+	ASSERT(ifp->if_format == XFS_DINODE_FMT_LOCAL);
+	ASSERT(ifp->if_bytes == 0);
+	ASSERT(ifp->if_nextents == 0);
+
+	xfs_bmap_forkoff_reset(ip, whichfork);
+	ifp->if_u1.if_root = NULL;
+	ifp->if_height = 0;
+	ifp->if_format = XFS_DINODE_FMT_EXTENTS;
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+}
+
+
+STATIC int				/* error */
+xfs_bmap_local_to_extents(
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_inode_t	*ip,		/* incore inode pointer */
+	xfs_extlen_t	total,		/* total blocks needed by transaction */
+	int		*logflagsp,	/* inode logging flags */
+	int		whichfork,
+	void		(*init_fn)(struct xfs_trans *tp,
+				   struct xfs_buf *bp,
+				   struct xfs_inode *ip,
+				   struct xfs_ifork *ifp))
+{
+	int		error = 0;
+	int		flags;		/* logging flags returned */
+	struct xfs_ifork *ifp;		/* inode fork pointer */
+	xfs_alloc_arg_t	args;		/* allocation arguments */
+	struct xfs_buf	*bp;		/* buffer for extent block */
+	struct xfs_bmbt_irec rec;
+	struct xfs_iext_cursor icur;
+
+	/*
+	 * We don't want to deal with the case of keeping inode data inline yet.
+	 * So sending the data fork of a regular inode is invalid.
+	 */
+	ASSERT(!(S_ISREG(VFS_I(ip)->i_mode) && whichfork == XFS_DATA_FORK));
+	ifp = xfs_ifork_ptr(ip, whichfork);
+	ASSERT(ifp->if_format == XFS_DINODE_FMT_LOCAL);
+
+	if (!ifp->if_bytes) {
+		xfs_bmap_local_to_extents_empty(tp, ip, whichfork);
+		flags = XFS_ILOG_CORE;
+		goto done;
+	}
+
+	flags = 0;
+	error = 0;
+	memset(&args, 0, sizeof(args));
+	args.tp = tp;
+	args.mp = ip->i_mount;
+	xfs_rmap_ino_owner(&args.oinfo, ip->i_ino, whichfork, 0);
+	/*
+	 * Allocate a block.  We know we need only one, since the
+	 * file currently fits in an inode.
+	 */
+	if (tp->t_firstblock == NULLFSBLOCK) {
+		args.fsbno = XFS_INO_TO_FSB(args.mp, ip->i_ino);
+		args.type = XFS_ALLOCTYPE_START_BNO;
+	} else {
+		args.fsbno = tp->t_firstblock;
+		args.type = XFS_ALLOCTYPE_NEAR_BNO;
+	}
+	args.total = total;
+	args.minlen = args.maxlen = args.prod = 1;
+	error = xfs_alloc_vextent(&args);
+	if (error)
+		goto done;
+
+	/* Can't fail, the space was reserved. */
+	ASSERT(args.fsbno != NULLFSBLOCK);
+	ASSERT(args.len == 1);
+	tp->t_firstblock = args.fsbno;
+	error = xfs_trans_get_buf(tp, args.mp->m_ddev_targp,
+			XFS_FSB_TO_DADDR(args.mp, args.fsbno),
+			args.mp->m_bsize, 0, &bp);
+	if (error)
+		goto done;
+
+	/*
+	 * Initialize the block, copy the data and log the remote buffer.
+	 *
+	 * The callout is responsible for logging because the remote format
+	 * might differ from the local format and thus we don't know how much to
+	 * log here. Note that init_fn must also set the buffer log item type
+	 * correctly.
+	 */
+	init_fn(tp, bp, ip, ifp);
+
+	/* account for the change in fork size */
+	xfs_idata_realloc(ip, -ifp->if_bytes, whichfork);
+	xfs_bmap_local_to_extents_empty(tp, ip, whichfork);
+	flags |= XFS_ILOG_CORE;
+
+	ifp->if_u1.if_root = NULL;
+	ifp->if_height = 0;
+
+	rec.br_startoff = 0;
+	rec.br_startblock = args.fsbno;
+	rec.br_blockcount = 1;
+	rec.br_state = XFS_EXT_NORM;
+	xfs_iext_first(ifp, &icur);
+	xfs_iext_insert(ip, &icur, &rec, 0);
+
+	ifp->if_nextents = 1;
+	ip->i_nblocks = 1;
+	xfs_trans_mod_dquot_byino(tp, ip,
+		XFS_TRANS_DQ_BCOUNT, 1L);
+	flags |= xfs_ilog_fext(whichfork);
+
+done:
+	*logflagsp = flags;
+	return error;
+}
+
+/*
+ * Called from xfs_bmap_add_attrfork to handle btree format files.
+ */
+STATIC int					/* error */
+xfs_bmap_add_attrfork_btree(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_inode_t		*ip,		/* incore inode pointer */
+	int			*flags)		/* inode logging flags */
+{
+	struct xfs_btree_block	*block = ip->i_df.if_broot;
+	struct xfs_btree_cur	*cur;		/* btree cursor */
+	int			error;		/* error return value */
+	xfs_mount_t		*mp;		/* file system mount struct */
+	int			stat;		/* newroot status */
+
+	mp = ip->i_mount;
+
+	if (XFS_BMAP_BMDR_SPACE(block) <= xfs_inode_data_fork_size(ip))
+		*flags |= XFS_ILOG_DBROOT;
+	else {
+		cur = xfs_bmbt_init_cursor(mp, tp, ip, XFS_DATA_FORK);
+		error = xfs_bmbt_lookup_first(cur, &stat);
+		if (error)
+			goto error0;
+		/* must be at least one entry */
+		if (XFS_IS_CORRUPT(mp, stat != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+		if ((error = xfs_btree_new_iroot(cur, flags, &stat)))
+			goto error0;
+		if (stat == 0) {
+			xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+			return -ENOSPC;
+		}
+		cur->bc_ino.allocated = 0;
+		xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	}
+	return 0;
+error0:
+	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/*
+ * Called from xfs_bmap_add_attrfork to handle extents format files.
+ */
+STATIC int					/* error */
+xfs_bmap_add_attrfork_extents(
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* incore inode pointer */
+	int			*flags)		/* inode logging flags */
+{
+	struct xfs_btree_cur	*cur;		/* bmap btree cursor */
+	int			error;		/* error return value */
+
+	if (ip->i_df.if_nextents * sizeof(struct xfs_bmbt_rec) <=
+	    xfs_inode_data_fork_size(ip))
+		return 0;
+	cur = NULL;
+	error = xfs_bmap_extents_to_btree(tp, ip, &cur, 0, flags,
+					  XFS_DATA_FORK);
+	if (cur) {
+		cur->bc_ino.allocated = 0;
+		xfs_btree_del_cursor(cur, error);
+	}
+	return error;
+}
+
+/*
+ * Called from xfs_bmap_add_attrfork to handle local format files. Each
+ * different data fork content type needs a different callout to do the
+ * conversion. Some are basic and only require special block initialisation
+ * callouts for the data formating, others (directories) are so specialised they
+ * handle everything themselves.
+ *
+ * XXX (dgc): investigate whether directory conversion can use the generic
+ * formatting callout. It should be possible - it's just a very complex
+ * formatter.
+ */
+STATIC int					/* error */
+xfs_bmap_add_attrfork_local(
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* incore inode pointer */
+	int			*flags)		/* inode logging flags */
+{
+	struct xfs_da_args	dargs;		/* args for dir/attr code */
+
+	if (ip->i_df.if_bytes <= xfs_inode_data_fork_size(ip))
+		return 0;
+
+	if (S_ISDIR(VFS_I(ip)->i_mode)) {
+		memset(&dargs, 0, sizeof(dargs));
+		dargs.geo = ip->i_mount->m_dir_geo;
+		dargs.dp = ip;
+		dargs.total = dargs.geo->fsbcount;
+		dargs.whichfork = XFS_DATA_FORK;
+		dargs.trans = tp;
+		return xfs_dir2_sf_to_block(&dargs);
+	}
+
+	if (S_ISLNK(VFS_I(ip)->i_mode))
+		return xfs_bmap_local_to_extents(tp, ip, 1, flags,
+						 XFS_DATA_FORK,
+						 xfs_symlink_local_to_remote);
+
+	/* should only be called for types that support local format data */
+	ASSERT(0);
+	return -EFSCORRUPTED;
+}
+
+/*
+ * Set an inode attr fork offset based on the format of the data fork.
+ */
+static int
+xfs_bmap_set_attrforkoff(
+	struct xfs_inode	*ip,
+	int			size,
+	int			*version)
+{
+	int			default_size = xfs_default_attroffset(ip) >> 3;
+
+	switch (ip->i_df.if_format) {
+	case XFS_DINODE_FMT_DEV:
+		ip->i_forkoff = default_size;
+		break;
+	case XFS_DINODE_FMT_LOCAL:
+	case XFS_DINODE_FMT_EXTENTS:
+	case XFS_DINODE_FMT_BTREE:
+		ip->i_forkoff = xfs_attr_shortform_bytesfit(ip, size);
+		if (!ip->i_forkoff)
+			ip->i_forkoff = default_size;
+		else if (xfs_has_attr2(ip->i_mount) && version)
+			*version = 2;
+		break;
+	default:
+		ASSERT(0);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+/*
+ * Convert inode from non-attributed to attributed.
+ * Must not be in a transaction, ip must not be locked.
+ */
+int						/* error code */
+xfs_bmap_add_attrfork(
+	xfs_inode_t		*ip,		/* incore inode pointer */
+	int			size,		/* space new attribute needs */
+	int			rsvd)		/* xact may use reserved blks */
+{
+	xfs_mount_t		*mp;		/* mount structure */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	int			blks;		/* space reservation */
+	int			version = 1;	/* superblock attr version */
+	int			logflags;	/* logging flags */
+	int			error;		/* error return value */
+
+	ASSERT(xfs_inode_has_attr_fork(ip) == 0);
+
+	mp = ip->i_mount;
+	ASSERT(!XFS_NOT_DQATTACHED(mp, ip));
+
+	blks = XFS_ADDAFORK_SPACE_RES(mp);
+
+	error = xfs_trans_alloc_inode(ip, &M_RES(mp)->tr_addafork, blks, 0,
+			rsvd, &tp);
+	if (error)
+		return error;
+	if (xfs_inode_has_attr_fork(ip))
+		goto trans_cancel;
+
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+	error = xfs_bmap_set_attrforkoff(ip, size, &version);
+	if (error)
+		goto trans_cancel;
+
+	xfs_ifork_init_attr(ip, XFS_DINODE_FMT_EXTENTS, 0);
+	logflags = 0;
+	switch (ip->i_df.if_format) {
+	case XFS_DINODE_FMT_LOCAL:
+		error = xfs_bmap_add_attrfork_local(tp, ip, &logflags);
+		break;
+	case XFS_DINODE_FMT_EXTENTS:
+		error = xfs_bmap_add_attrfork_extents(tp, ip, &logflags);
+		break;
+	case XFS_DINODE_FMT_BTREE:
+		error = xfs_bmap_add_attrfork_btree(tp, ip, &logflags);
+		break;
+	default:
+		error = 0;
+		break;
+	}
+	if (logflags)
+		xfs_trans_log_inode(tp, ip, logflags);
+	if (error)
+		goto trans_cancel;
+	if (!xfs_has_attr(mp) ||
+	   (!xfs_has_attr2(mp) && version == 2)) {
+		bool log_sb = false;
+
+		spin_lock(&mp->m_sb_lock);
+		if (!xfs_has_attr(mp)) {
+			xfs_add_attr(mp);
+			log_sb = true;
+		}
+		if (!xfs_has_attr2(mp) && version == 2) {
+			xfs_add_attr2(mp);
+			log_sb = true;
+		}
+		spin_unlock(&mp->m_sb_lock);
+		if (log_sb)
+			xfs_log_sb(tp);
+	}
+
+	error = xfs_trans_commit(tp);
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	return error;
+
+trans_cancel:
+	xfs_trans_cancel(tp);
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	return error;
+}
+
+/*
+ * Internal and external extent tree search functions.
+ */
+
+struct xfs_iread_state {
+	struct xfs_iext_cursor	icur;
+	xfs_extnum_t		loaded;
+};
+
+/* Stuff every bmbt record from this block into the incore extent map. */
+static int
+xfs_iread_bmbt_block(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	void			*priv)
+{
+	struct xfs_iread_state	*ir = priv;
+	struct xfs_mount	*mp = cur->bc_mp;
+	struct xfs_inode	*ip = cur->bc_ino.ip;
+	struct xfs_btree_block	*block;
+	struct xfs_buf		*bp;
+	struct xfs_bmbt_rec	*frp;
+	xfs_extnum_t		num_recs;
+	xfs_extnum_t		j;
+	int			whichfork = cur->bc_ino.whichfork;
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+
+	block = xfs_btree_get_block(cur, level, &bp);
+
+	/* Abort if we find more records than nextents. */
+	num_recs = xfs_btree_get_numrecs(block);
+	if (unlikely(ir->loaded + num_recs > ifp->if_nextents)) {
+		xfs_warn(ip->i_mount, "corrupt dinode %llu, (btree extents).",
+				(unsigned long long)ip->i_ino);
+		xfs_inode_verifier_error(ip, -EFSCORRUPTED, __func__, block,
+				sizeof(*block), __this_address);
+		return -EFSCORRUPTED;
+	}
+
+	/* Copy records into the incore cache. */
+	frp = XFS_BMBT_REC_ADDR(mp, block, 1);
+	for (j = 0; j < num_recs; j++, frp++, ir->loaded++) {
+		struct xfs_bmbt_irec	new;
+		xfs_failaddr_t		fa;
+
+		xfs_bmbt_disk_get_all(frp, &new);
+		fa = xfs_bmap_validate_extent(ip, whichfork, &new);
+		if (fa) {
+			xfs_inode_verifier_error(ip, -EFSCORRUPTED,
+					"xfs_iread_extents(2)", frp,
+					sizeof(*frp), fa);
+			return -EFSCORRUPTED;
+		}
+		xfs_iext_insert(ip, &ir->icur, &new,
+				xfs_bmap_fork_to_state(whichfork));
+		trace_xfs_read_extent(ip, &ir->icur,
+				xfs_bmap_fork_to_state(whichfork), _THIS_IP_);
+		xfs_iext_next(ifp, &ir->icur);
+	}
+
+	return 0;
+}
+
+/*
+ * Read in extents from a btree-format inode.
+ */
+int
+xfs_iread_extents(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	int			whichfork)
+{
+	struct xfs_iread_state	ir;
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_btree_cur	*cur;
+	int			error;
+
+	if (!xfs_need_iread_extents(ifp))
+		return 0;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+
+	ir.loaded = 0;
+	xfs_iext_first(ifp, &ir.icur);
+	cur = xfs_bmbt_init_cursor(mp, tp, ip, whichfork);
+	error = xfs_btree_visit_blocks(cur, xfs_iread_bmbt_block,
+			XFS_BTREE_VISIT_RECORDS, &ir);
+	xfs_btree_del_cursor(cur, error);
+	if (error)
+		goto out;
+
+	if (XFS_IS_CORRUPT(mp, ir.loaded != ifp->if_nextents)) {
+		error = -EFSCORRUPTED;
+		goto out;
+	}
+	ASSERT(ir.loaded == xfs_iext_count(ifp));
+	return 0;
+out:
+	xfs_iext_destroy(ifp);
+	return error;
+}
+
+/*
+ * Returns the relative block number of the first unused block(s) in the given
+ * fork with at least "len" logically contiguous blocks free.  This is the
+ * lowest-address hole if the fork has holes, else the first block past the end
+ * of fork.  Return 0 if the fork is currently local (in-inode).
+ */
+int						/* error */
+xfs_bmap_first_unused(
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* incore inode */
+	xfs_extlen_t		len,		/* size of hole to find */
+	xfs_fileoff_t		*first_unused,	/* unused block */
+	int			whichfork)	/* data or attr fork */
+{
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_bmbt_irec	got;
+	struct xfs_iext_cursor	icur;
+	xfs_fileoff_t		lastaddr = 0;
+	xfs_fileoff_t		lowest, max;
+	int			error;
+
+	if (ifp->if_format == XFS_DINODE_FMT_LOCAL) {
+		*first_unused = 0;
+		return 0;
+	}
+
+	ASSERT(xfs_ifork_has_extents(ifp));
+
+	error = xfs_iread_extents(tp, ip, whichfork);
+	if (error)
+		return error;
+
+	lowest = max = *first_unused;
+	for_each_xfs_iext(ifp, &icur, &got) {
+		/*
+		 * See if the hole before this extent will work.
+		 */
+		if (got.br_startoff >= lowest + len &&
+		    got.br_startoff - max >= len)
+			break;
+		lastaddr = got.br_startoff + got.br_blockcount;
+		max = XFS_FILEOFF_MAX(lastaddr, lowest);
+	}
+
+	*first_unused = max;
+	return 0;
+}
+
+/*
+ * Returns the file-relative block number of the last block - 1 before
+ * last_block (input value) in the file.
+ * This is not based on i_size, it is based on the extent records.
+ * Returns 0 for local files, as they do not have extent records.
+ */
+int						/* error */
+xfs_bmap_last_before(
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* incore inode */
+	xfs_fileoff_t		*last_block,	/* last block */
+	int			whichfork)	/* data or attr fork */
+{
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_bmbt_irec	got;
+	struct xfs_iext_cursor	icur;
+	int			error;
+
+	switch (ifp->if_format) {
+	case XFS_DINODE_FMT_LOCAL:
+		*last_block = 0;
+		return 0;
+	case XFS_DINODE_FMT_BTREE:
+	case XFS_DINODE_FMT_EXTENTS:
+		break;
+	default:
+		ASSERT(0);
+		return -EFSCORRUPTED;
+	}
+
+	error = xfs_iread_extents(tp, ip, whichfork);
+	if (error)
+		return error;
+
+	if (!xfs_iext_lookup_extent_before(ip, ifp, last_block, &icur, &got))
+		*last_block = 0;
+	return 0;
+}
+
+int
+xfs_bmap_last_extent(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	int			whichfork,
+	struct xfs_bmbt_irec	*rec,
+	int			*is_empty)
+{
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_iext_cursor	icur;
+	int			error;
+
+	error = xfs_iread_extents(tp, ip, whichfork);
+	if (error)
+		return error;
+
+	xfs_iext_last(ifp, &icur);
+	if (!xfs_iext_get_extent(ifp, &icur, rec))
+		*is_empty = 1;
+	else
+		*is_empty = 0;
+	return 0;
+}
+
+/*
+ * Check the last inode extent to determine whether this allocation will result
+ * in blocks being allocated at the end of the file. When we allocate new data
+ * blocks at the end of the file which do not start at the previous data block,
+ * we will try to align the new blocks at stripe unit boundaries.
+ *
+ * Returns 1 in bma->aeof if the file (fork) is empty as any new write will be
+ * at, or past the EOF.
+ */
+STATIC int
+xfs_bmap_isaeof(
+	struct xfs_bmalloca	*bma,
+	int			whichfork)
+{
+	struct xfs_bmbt_irec	rec;
+	int			is_empty;
+	int			error;
+
+	bma->aeof = false;
+	error = xfs_bmap_last_extent(NULL, bma->ip, whichfork, &rec,
+				     &is_empty);
+	if (error)
+		return error;
+
+	if (is_empty) {
+		bma->aeof = true;
+		return 0;
+	}
+
+	/*
+	 * Check if we are allocation or past the last extent, or at least into
+	 * the last delayed allocated extent.
+	 */
+	bma->aeof = bma->offset >= rec.br_startoff + rec.br_blockcount ||
+		(bma->offset >= rec.br_startoff &&
+		 isnullstartblock(rec.br_startblock));
+	return 0;
+}
+
+/*
+ * Returns the file-relative block number of the first block past eof in
+ * the file.  This is not based on i_size, it is based on the extent records.
+ * Returns 0 for local files, as they do not have extent records.
+ */
+int
+xfs_bmap_last_offset(
+	struct xfs_inode	*ip,
+	xfs_fileoff_t		*last_block,
+	int			whichfork)
+{
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_bmbt_irec	rec;
+	int			is_empty;
+	int			error;
+
+	*last_block = 0;
+
+	if (ifp->if_format == XFS_DINODE_FMT_LOCAL)
+		return 0;
+
+	if (XFS_IS_CORRUPT(ip->i_mount, !xfs_ifork_has_extents(ifp)))
+		return -EFSCORRUPTED;
+
+	error = xfs_bmap_last_extent(NULL, ip, whichfork, &rec, &is_empty);
+	if (error || is_empty)
+		return error;
+
+	*last_block = rec.br_startoff + rec.br_blockcount;
+	return 0;
+}
+
+/*
+ * Extent tree manipulation functions used during allocation.
+ */
+
+/*
+ * Convert a delayed allocation to a real allocation.
+ */
+STATIC int				/* error */
+xfs_bmap_add_extent_delay_real(
+	struct xfs_bmalloca	*bma,
+	int			whichfork)
+{
+	struct xfs_mount	*mp = bma->ip->i_mount;
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(bma->ip, whichfork);
+	struct xfs_bmbt_irec	*new = &bma->got;
+	int			error;	/* error return value */
+	int			i;	/* temp state */
+	xfs_fileoff_t		new_endoff;	/* end offset of new entry */
+	xfs_bmbt_irec_t		r[3];	/* neighbor extent entries */
+					/* left is 0, right is 1, prev is 2 */
+	int			rval=0;	/* return value (logging flags) */
+	uint32_t		state = xfs_bmap_fork_to_state(whichfork);
+	xfs_filblks_t		da_new; /* new count del alloc blocks used */
+	xfs_filblks_t		da_old; /* old count del alloc blocks used */
+	xfs_filblks_t		temp=0;	/* value for da_new calculations */
+	int			tmp_rval;	/* partial logging flags */
+	struct xfs_bmbt_irec	old;
+
+	ASSERT(whichfork != XFS_ATTR_FORK);
+	ASSERT(!isnullstartblock(new->br_startblock));
+	ASSERT(!bma->cur ||
+	       (bma->cur->bc_ino.flags & XFS_BTCUR_BMBT_WASDEL));
+
+	XFS_STATS_INC(mp, xs_add_exlist);
+
+#define	LEFT		r[0]
+#define	RIGHT		r[1]
+#define	PREV		r[2]
+
+	/*
+	 * Set up a bunch of variables to make the tests simpler.
+	 */
+	xfs_iext_get_extent(ifp, &bma->icur, &PREV);
+	new_endoff = new->br_startoff + new->br_blockcount;
+	ASSERT(isnullstartblock(PREV.br_startblock));
+	ASSERT(PREV.br_startoff <= new->br_startoff);
+	ASSERT(PREV.br_startoff + PREV.br_blockcount >= new_endoff);
+
+	da_old = startblockval(PREV.br_startblock);
+	da_new = 0;
+
+	/*
+	 * Set flags determining what part of the previous delayed allocation
+	 * extent is being replaced by a real allocation.
+	 */
+	if (PREV.br_startoff == new->br_startoff)
+		state |= BMAP_LEFT_FILLING;
+	if (PREV.br_startoff + PREV.br_blockcount == new_endoff)
+		state |= BMAP_RIGHT_FILLING;
+
+	/*
+	 * Check and set flags if this segment has a left neighbor.
+	 * Don't set contiguous if the combined extent would be too large.
+	 */
+	if (xfs_iext_peek_prev_extent(ifp, &bma->icur, &LEFT)) {
+		state |= BMAP_LEFT_VALID;
+		if (isnullstartblock(LEFT.br_startblock))
+			state |= BMAP_LEFT_DELAY;
+	}
+
+	if ((state & BMAP_LEFT_VALID) && !(state & BMAP_LEFT_DELAY) &&
+	    LEFT.br_startoff + LEFT.br_blockcount == new->br_startoff &&
+	    LEFT.br_startblock + LEFT.br_blockcount == new->br_startblock &&
+	    LEFT.br_state == new->br_state &&
+	    LEFT.br_blockcount + new->br_blockcount <= XFS_MAX_BMBT_EXTLEN)
+		state |= BMAP_LEFT_CONTIG;
+
+	/*
+	 * Check and set flags if this segment has a right neighbor.
+	 * Don't set contiguous if the combined extent would be too large.
+	 * Also check for all-three-contiguous being too large.
+	 */
+	if (xfs_iext_peek_next_extent(ifp, &bma->icur, &RIGHT)) {
+		state |= BMAP_RIGHT_VALID;
+		if (isnullstartblock(RIGHT.br_startblock))
+			state |= BMAP_RIGHT_DELAY;
+	}
+
+	if ((state & BMAP_RIGHT_VALID) && !(state & BMAP_RIGHT_DELAY) &&
+	    new_endoff == RIGHT.br_startoff &&
+	    new->br_startblock + new->br_blockcount == RIGHT.br_startblock &&
+	    new->br_state == RIGHT.br_state &&
+	    new->br_blockcount + RIGHT.br_blockcount <= XFS_MAX_BMBT_EXTLEN &&
+	    ((state & (BMAP_LEFT_CONTIG | BMAP_LEFT_FILLING |
+		       BMAP_RIGHT_FILLING)) !=
+		      (BMAP_LEFT_CONTIG | BMAP_LEFT_FILLING |
+		       BMAP_RIGHT_FILLING) ||
+	     LEFT.br_blockcount + new->br_blockcount + RIGHT.br_blockcount
+			<= XFS_MAX_BMBT_EXTLEN))
+		state |= BMAP_RIGHT_CONTIG;
+
+	error = 0;
+	/*
+	 * Switch out based on the FILLING and CONTIG state bits.
+	 */
+	switch (state & (BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG |
+			 BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG)) {
+	case BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG |
+	     BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG:
+		/*
+		 * Filling in all of a previously delayed allocation extent.
+		 * The left and right neighbors are both contiguous with new.
+		 */
+		LEFT.br_blockcount += PREV.br_blockcount + RIGHT.br_blockcount;
+
+		xfs_iext_remove(bma->ip, &bma->icur, state);
+		xfs_iext_remove(bma->ip, &bma->icur, state);
+		xfs_iext_prev(ifp, &bma->icur);
+		xfs_iext_update_extent(bma->ip, state, &bma->icur, &LEFT);
+		ifp->if_nextents--;
+
+		if (bma->cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			error = xfs_bmbt_lookup_eq(bma->cur, &RIGHT, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_btree_delete(bma->cur, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_btree_decrement(bma->cur, 0, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_bmbt_update(bma->cur, &LEFT);
+			if (error)
+				goto done;
+		}
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING | BMAP_LEFT_CONTIG:
+		/*
+		 * Filling in all of a previously delayed allocation extent.
+		 * The left neighbor is contiguous, the right is not.
+		 */
+		old = LEFT;
+		LEFT.br_blockcount += PREV.br_blockcount;
+
+		xfs_iext_remove(bma->ip, &bma->icur, state);
+		xfs_iext_prev(ifp, &bma->icur);
+		xfs_iext_update_extent(bma->ip, state, &bma->icur, &LEFT);
+
+		if (bma->cur == NULL)
+			rval = XFS_ILOG_DEXT;
+		else {
+			rval = 0;
+			error = xfs_bmbt_lookup_eq(bma->cur, &old, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_bmbt_update(bma->cur, &LEFT);
+			if (error)
+				goto done;
+		}
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG:
+		/*
+		 * Filling in all of a previously delayed allocation extent.
+		 * The right neighbor is contiguous, the left is not. Take care
+		 * with delay -> unwritten extent allocation here because the
+		 * delalloc record we are overwriting is always written.
+		 */
+		PREV.br_startblock = new->br_startblock;
+		PREV.br_blockcount += RIGHT.br_blockcount;
+		PREV.br_state = new->br_state;
+
+		xfs_iext_next(ifp, &bma->icur);
+		xfs_iext_remove(bma->ip, &bma->icur, state);
+		xfs_iext_prev(ifp, &bma->icur);
+		xfs_iext_update_extent(bma->ip, state, &bma->icur, &PREV);
+
+		if (bma->cur == NULL)
+			rval = XFS_ILOG_DEXT;
+		else {
+			rval = 0;
+			error = xfs_bmbt_lookup_eq(bma->cur, &RIGHT, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_bmbt_update(bma->cur, &PREV);
+			if (error)
+				goto done;
+		}
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING:
+		/*
+		 * Filling in all of a previously delayed allocation extent.
+		 * Neither the left nor right neighbors are contiguous with
+		 * the new one.
+		 */
+		PREV.br_startblock = new->br_startblock;
+		PREV.br_state = new->br_state;
+		xfs_iext_update_extent(bma->ip, state, &bma->icur, &PREV);
+		ifp->if_nextents++;
+
+		if (bma->cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			error = xfs_bmbt_lookup_eq(bma->cur, new, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 0)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_btree_insert(bma->cur, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+		}
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG:
+		/*
+		 * Filling in the first part of a previous delayed allocation.
+		 * The left neighbor is contiguous.
+		 */
+		old = LEFT;
+		temp = PREV.br_blockcount - new->br_blockcount;
+		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(bma->ip, temp),
+				startblockval(PREV.br_startblock));
+
+		LEFT.br_blockcount += new->br_blockcount;
+
+		PREV.br_blockcount = temp;
+		PREV.br_startoff += new->br_blockcount;
+		PREV.br_startblock = nullstartblock(da_new);
+
+		xfs_iext_update_extent(bma->ip, state, &bma->icur, &PREV);
+		xfs_iext_prev(ifp, &bma->icur);
+		xfs_iext_update_extent(bma->ip, state, &bma->icur, &LEFT);
+
+		if (bma->cur == NULL)
+			rval = XFS_ILOG_DEXT;
+		else {
+			rval = 0;
+			error = xfs_bmbt_lookup_eq(bma->cur, &old, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_bmbt_update(bma->cur, &LEFT);
+			if (error)
+				goto done;
+		}
+		break;
+
+	case BMAP_LEFT_FILLING:
+		/*
+		 * Filling in the first part of a previous delayed allocation.
+		 * The left neighbor is not contiguous.
+		 */
+		xfs_iext_update_extent(bma->ip, state, &bma->icur, new);
+		ifp->if_nextents++;
+
+		if (bma->cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			error = xfs_bmbt_lookup_eq(bma->cur, new, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 0)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_btree_insert(bma->cur, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+		}
+
+		if (xfs_bmap_needs_btree(bma->ip, whichfork)) {
+			error = xfs_bmap_extents_to_btree(bma->tp, bma->ip,
+					&bma->cur, 1, &tmp_rval, whichfork);
+			rval |= tmp_rval;
+			if (error)
+				goto done;
+		}
+
+		temp = PREV.br_blockcount - new->br_blockcount;
+		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(bma->ip, temp),
+			startblockval(PREV.br_startblock) -
+			(bma->cur ? bma->cur->bc_ino.allocated : 0));
+
+		PREV.br_startoff = new_endoff;
+		PREV.br_blockcount = temp;
+		PREV.br_startblock = nullstartblock(da_new);
+		xfs_iext_next(ifp, &bma->icur);
+		xfs_iext_insert(bma->ip, &bma->icur, &PREV, state);
+		xfs_iext_prev(ifp, &bma->icur);
+		break;
+
+	case BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG:
+		/*
+		 * Filling in the last part of a previous delayed allocation.
+		 * The right neighbor is contiguous with the new allocation.
+		 */
+		old = RIGHT;
+		RIGHT.br_startoff = new->br_startoff;
+		RIGHT.br_startblock = new->br_startblock;
+		RIGHT.br_blockcount += new->br_blockcount;
+
+		if (bma->cur == NULL)
+			rval = XFS_ILOG_DEXT;
+		else {
+			rval = 0;
+			error = xfs_bmbt_lookup_eq(bma->cur, &old, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_bmbt_update(bma->cur, &RIGHT);
+			if (error)
+				goto done;
+		}
+
+		temp = PREV.br_blockcount - new->br_blockcount;
+		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(bma->ip, temp),
+			startblockval(PREV.br_startblock));
+
+		PREV.br_blockcount = temp;
+		PREV.br_startblock = nullstartblock(da_new);
+
+		xfs_iext_update_extent(bma->ip, state, &bma->icur, &PREV);
+		xfs_iext_next(ifp, &bma->icur);
+		xfs_iext_update_extent(bma->ip, state, &bma->icur, &RIGHT);
+		break;
+
+	case BMAP_RIGHT_FILLING:
+		/*
+		 * Filling in the last part of a previous delayed allocation.
+		 * The right neighbor is not contiguous.
+		 */
+		xfs_iext_update_extent(bma->ip, state, &bma->icur, new);
+		ifp->if_nextents++;
+
+		if (bma->cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			error = xfs_bmbt_lookup_eq(bma->cur, new, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 0)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_btree_insert(bma->cur, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+		}
+
+		if (xfs_bmap_needs_btree(bma->ip, whichfork)) {
+			error = xfs_bmap_extents_to_btree(bma->tp, bma->ip,
+				&bma->cur, 1, &tmp_rval, whichfork);
+			rval |= tmp_rval;
+			if (error)
+				goto done;
+		}
+
+		temp = PREV.br_blockcount - new->br_blockcount;
+		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(bma->ip, temp),
+			startblockval(PREV.br_startblock) -
+			(bma->cur ? bma->cur->bc_ino.allocated : 0));
+
+		PREV.br_startblock = nullstartblock(da_new);
+		PREV.br_blockcount = temp;
+		xfs_iext_insert(bma->ip, &bma->icur, &PREV, state);
+		xfs_iext_next(ifp, &bma->icur);
+		break;
+
+	case 0:
+		/*
+		 * Filling in the middle part of a previous delayed allocation.
+		 * Contiguity is impossible here.
+		 * This case is avoided almost all the time.
+		 *
+		 * We start with a delayed allocation:
+		 *
+		 * +ddddddddddddddddddddddddddddddddddddddddddddddddddddddd+
+		 *  PREV @ idx
+		 *
+	         * and we are allocating:
+		 *                     +rrrrrrrrrrrrrrrrr+
+		 *			      new
+		 *
+		 * and we set it up for insertion as:
+		 * +ddddddddddddddddddd+rrrrrrrrrrrrrrrrr+ddddddddddddddddd+
+		 *                            new
+		 *  PREV @ idx          LEFT              RIGHT
+		 *                      inserted at idx + 1
+		 */
+		old = PREV;
+
+		/* LEFT is the new middle */
+		LEFT = *new;
+
+		/* RIGHT is the new right */
+		RIGHT.br_state = PREV.br_state;
+		RIGHT.br_startoff = new_endoff;
+		RIGHT.br_blockcount =
+			PREV.br_startoff + PREV.br_blockcount - new_endoff;
+		RIGHT.br_startblock =
+			nullstartblock(xfs_bmap_worst_indlen(bma->ip,
+					RIGHT.br_blockcount));
+
+		/* truncate PREV */
+		PREV.br_blockcount = new->br_startoff - PREV.br_startoff;
+		PREV.br_startblock =
+			nullstartblock(xfs_bmap_worst_indlen(bma->ip,
+					PREV.br_blockcount));
+		xfs_iext_update_extent(bma->ip, state, &bma->icur, &PREV);
+
+		xfs_iext_next(ifp, &bma->icur);
+		xfs_iext_insert(bma->ip, &bma->icur, &RIGHT, state);
+		xfs_iext_insert(bma->ip, &bma->icur, &LEFT, state);
+		ifp->if_nextents++;
+
+		if (bma->cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			error = xfs_bmbt_lookup_eq(bma->cur, new, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 0)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_btree_insert(bma->cur, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+		}
+
+		if (xfs_bmap_needs_btree(bma->ip, whichfork)) {
+			error = xfs_bmap_extents_to_btree(bma->tp, bma->ip,
+					&bma->cur, 1, &tmp_rval, whichfork);
+			rval |= tmp_rval;
+			if (error)
+				goto done;
+		}
+
+		da_new = startblockval(PREV.br_startblock) +
+			 startblockval(RIGHT.br_startblock);
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG:
+	case BMAP_RIGHT_FILLING | BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG:
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_CONTIG:
+	case BMAP_RIGHT_FILLING | BMAP_LEFT_CONTIG:
+	case BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG:
+	case BMAP_LEFT_CONTIG:
+	case BMAP_RIGHT_CONTIG:
+		/*
+		 * These cases are all impossible.
+		 */
+		ASSERT(0);
+	}
+
+	/* add reverse mapping unless caller opted out */
+	if (!(bma->flags & XFS_BMAPI_NORMAP))
+		xfs_rmap_map_extent(bma->tp, bma->ip, whichfork, new);
+
+	/* convert to a btree if necessary */
+	if (xfs_bmap_needs_btree(bma->ip, whichfork)) {
+		int	tmp_logflags;	/* partial log flag return val */
+
+		ASSERT(bma->cur == NULL);
+		error = xfs_bmap_extents_to_btree(bma->tp, bma->ip,
+				&bma->cur, da_old > 0, &tmp_logflags,
+				whichfork);
+		bma->logflags |= tmp_logflags;
+		if (error)
+			goto done;
+	}
+
+	if (da_new != da_old)
+		xfs_mod_delalloc(mp, (int64_t)da_new - da_old);
+
+	if (bma->cur) {
+		da_new += bma->cur->bc_ino.allocated;
+		bma->cur->bc_ino.allocated = 0;
+	}
+
+	/* adjust for changes in reserved delayed indirect blocks */
+	if (da_new != da_old) {
+		ASSERT(state == 0 || da_new < da_old);
+		error = xfs_mod_fdblocks(mp, (int64_t)(da_old - da_new),
+				false);
+	}
+
+	xfs_bmap_check_leaf_extents(bma->cur, bma->ip, whichfork);
+done:
+	if (whichfork != XFS_COW_FORK)
+		bma->logflags |= rval;
+	return error;
+#undef	LEFT
+#undef	RIGHT
+#undef	PREV
+}
+
+/*
+ * Convert an unwritten allocation to a real allocation or vice versa.
+ */
+int					/* error */
+xfs_bmap_add_extent_unwritten_real(
+	struct xfs_trans	*tp,
+	xfs_inode_t		*ip,	/* incore inode pointer */
+	int			whichfork,
+	struct xfs_iext_cursor	*icur,
+	struct xfs_btree_cur	**curp,	/* if *curp is null, not a btree */
+	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
+	int			*logflagsp) /* inode logging flags */
+{
+	struct xfs_btree_cur	*cur;	/* btree cursor */
+	int			error;	/* error return value */
+	int			i;	/* temp state */
+	struct xfs_ifork	*ifp;	/* inode fork pointer */
+	xfs_fileoff_t		new_endoff;	/* end offset of new entry */
+	xfs_bmbt_irec_t		r[3];	/* neighbor extent entries */
+					/* left is 0, right is 1, prev is 2 */
+	int			rval=0;	/* return value (logging flags) */
+	uint32_t		state = xfs_bmap_fork_to_state(whichfork);
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_bmbt_irec	old;
+
+	*logflagsp = 0;
+
+	cur = *curp;
+	ifp = xfs_ifork_ptr(ip, whichfork);
+
+	ASSERT(!isnullstartblock(new->br_startblock));
+
+	XFS_STATS_INC(mp, xs_add_exlist);
+
+#define	LEFT		r[0]
+#define	RIGHT		r[1]
+#define	PREV		r[2]
+
+	/*
+	 * Set up a bunch of variables to make the tests simpler.
+	 */
+	error = 0;
+	xfs_iext_get_extent(ifp, icur, &PREV);
+	ASSERT(new->br_state != PREV.br_state);
+	new_endoff = new->br_startoff + new->br_blockcount;
+	ASSERT(PREV.br_startoff <= new->br_startoff);
+	ASSERT(PREV.br_startoff + PREV.br_blockcount >= new_endoff);
+
+	/*
+	 * Set flags determining what part of the previous oldext allocation
+	 * extent is being replaced by a newext allocation.
+	 */
+	if (PREV.br_startoff == new->br_startoff)
+		state |= BMAP_LEFT_FILLING;
+	if (PREV.br_startoff + PREV.br_blockcount == new_endoff)
+		state |= BMAP_RIGHT_FILLING;
+
+	/*
+	 * Check and set flags if this segment has a left neighbor.
+	 * Don't set contiguous if the combined extent would be too large.
+	 */
+	if (xfs_iext_peek_prev_extent(ifp, icur, &LEFT)) {
+		state |= BMAP_LEFT_VALID;
+		if (isnullstartblock(LEFT.br_startblock))
+			state |= BMAP_LEFT_DELAY;
+	}
+
+	if ((state & BMAP_LEFT_VALID) && !(state & BMAP_LEFT_DELAY) &&
+	    LEFT.br_startoff + LEFT.br_blockcount == new->br_startoff &&
+	    LEFT.br_startblock + LEFT.br_blockcount == new->br_startblock &&
+	    LEFT.br_state == new->br_state &&
+	    LEFT.br_blockcount + new->br_blockcount <= XFS_MAX_BMBT_EXTLEN)
+		state |= BMAP_LEFT_CONTIG;
+
+	/*
+	 * Check and set flags if this segment has a right neighbor.
+	 * Don't set contiguous if the combined extent would be too large.
+	 * Also check for all-three-contiguous being too large.
+	 */
+	if (xfs_iext_peek_next_extent(ifp, icur, &RIGHT)) {
+		state |= BMAP_RIGHT_VALID;
+		if (isnullstartblock(RIGHT.br_startblock))
+			state |= BMAP_RIGHT_DELAY;
+	}
+
+	if ((state & BMAP_RIGHT_VALID) && !(state & BMAP_RIGHT_DELAY) &&
+	    new_endoff == RIGHT.br_startoff &&
+	    new->br_startblock + new->br_blockcount == RIGHT.br_startblock &&
+	    new->br_state == RIGHT.br_state &&
+	    new->br_blockcount + RIGHT.br_blockcount <= XFS_MAX_BMBT_EXTLEN &&
+	    ((state & (BMAP_LEFT_CONTIG | BMAP_LEFT_FILLING |
+		       BMAP_RIGHT_FILLING)) !=
+		      (BMAP_LEFT_CONTIG | BMAP_LEFT_FILLING |
+		       BMAP_RIGHT_FILLING) ||
+	     LEFT.br_blockcount + new->br_blockcount + RIGHT.br_blockcount
+			<= XFS_MAX_BMBT_EXTLEN))
+		state |= BMAP_RIGHT_CONTIG;
+
+	/*
+	 * Switch out based on the FILLING and CONTIG state bits.
+	 */
+	switch (state & (BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG |
+			 BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG)) {
+	case BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG |
+	     BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG:
+		/*
+		 * Setting all of a previous oldext extent to newext.
+		 * The left and right neighbors are both contiguous with new.
+		 */
+		LEFT.br_blockcount += PREV.br_blockcount + RIGHT.br_blockcount;
+
+		xfs_iext_remove(ip, icur, state);
+		xfs_iext_remove(ip, icur, state);
+		xfs_iext_prev(ifp, icur);
+		xfs_iext_update_extent(ip, state, icur, &LEFT);
+		ifp->if_nextents -= 2;
+		if (cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			error = xfs_bmbt_lookup_eq(cur, &RIGHT, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			if ((error = xfs_btree_delete(cur, &i)))
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			if ((error = xfs_btree_delete(cur, &i)))
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_bmbt_update(cur, &LEFT);
+			if (error)
+				goto done;
+		}
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING | BMAP_LEFT_CONTIG:
+		/*
+		 * Setting all of a previous oldext extent to newext.
+		 * The left neighbor is contiguous, the right is not.
+		 */
+		LEFT.br_blockcount += PREV.br_blockcount;
+
+		xfs_iext_remove(ip, icur, state);
+		xfs_iext_prev(ifp, icur);
+		xfs_iext_update_extent(ip, state, icur, &LEFT);
+		ifp->if_nextents--;
+		if (cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			error = xfs_bmbt_lookup_eq(cur, &PREV, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			if ((error = xfs_btree_delete(cur, &i)))
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_bmbt_update(cur, &LEFT);
+			if (error)
+				goto done;
+		}
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG:
+		/*
+		 * Setting all of a previous oldext extent to newext.
+		 * The right neighbor is contiguous, the left is not.
+		 */
+		PREV.br_blockcount += RIGHT.br_blockcount;
+		PREV.br_state = new->br_state;
+
+		xfs_iext_next(ifp, icur);
+		xfs_iext_remove(ip, icur, state);
+		xfs_iext_prev(ifp, icur);
+		xfs_iext_update_extent(ip, state, icur, &PREV);
+		ifp->if_nextents--;
+
+		if (cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			error = xfs_bmbt_lookup_eq(cur, &RIGHT, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			if ((error = xfs_btree_delete(cur, &i)))
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_bmbt_update(cur, &PREV);
+			if (error)
+				goto done;
+		}
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING:
+		/*
+		 * Setting all of a previous oldext extent to newext.
+		 * Neither the left nor right neighbors are contiguous with
+		 * the new one.
+		 */
+		PREV.br_state = new->br_state;
+		xfs_iext_update_extent(ip, state, icur, &PREV);
+
+		if (cur == NULL)
+			rval = XFS_ILOG_DEXT;
+		else {
+			rval = 0;
+			error = xfs_bmbt_lookup_eq(cur, new, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_bmbt_update(cur, &PREV);
+			if (error)
+				goto done;
+		}
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG:
+		/*
+		 * Setting the first part of a previous oldext extent to newext.
+		 * The left neighbor is contiguous.
+		 */
+		LEFT.br_blockcount += new->br_blockcount;
+
+		old = PREV;
+		PREV.br_startoff += new->br_blockcount;
+		PREV.br_startblock += new->br_blockcount;
+		PREV.br_blockcount -= new->br_blockcount;
+
+		xfs_iext_update_extent(ip, state, icur, &PREV);
+		xfs_iext_prev(ifp, icur);
+		xfs_iext_update_extent(ip, state, icur, &LEFT);
+
+		if (cur == NULL)
+			rval = XFS_ILOG_DEXT;
+		else {
+			rval = 0;
+			error = xfs_bmbt_lookup_eq(cur, &old, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_bmbt_update(cur, &PREV);
+			if (error)
+				goto done;
+			error = xfs_btree_decrement(cur, 0, &i);
+			if (error)
+				goto done;
+			error = xfs_bmbt_update(cur, &LEFT);
+			if (error)
+				goto done;
+		}
+		break;
+
+	case BMAP_LEFT_FILLING:
+		/*
+		 * Setting the first part of a previous oldext extent to newext.
+		 * The left neighbor is not contiguous.
+		 */
+		old = PREV;
+		PREV.br_startoff += new->br_blockcount;
+		PREV.br_startblock += new->br_blockcount;
+		PREV.br_blockcount -= new->br_blockcount;
+
+		xfs_iext_update_extent(ip, state, icur, &PREV);
+		xfs_iext_insert(ip, icur, new, state);
+		ifp->if_nextents++;
+
+		if (cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			error = xfs_bmbt_lookup_eq(cur, &old, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_bmbt_update(cur, &PREV);
+			if (error)
+				goto done;
+			cur->bc_rec.b = *new;
+			if ((error = xfs_btree_insert(cur, &i)))
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+		}
+		break;
+
+	case BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG:
+		/*
+		 * Setting the last part of a previous oldext extent to newext.
+		 * The right neighbor is contiguous with the new allocation.
+		 */
+		old = PREV;
+		PREV.br_blockcount -= new->br_blockcount;
+
+		RIGHT.br_startoff = new->br_startoff;
+		RIGHT.br_startblock = new->br_startblock;
+		RIGHT.br_blockcount += new->br_blockcount;
+
+		xfs_iext_update_extent(ip, state, icur, &PREV);
+		xfs_iext_next(ifp, icur);
+		xfs_iext_update_extent(ip, state, icur, &RIGHT);
+
+		if (cur == NULL)
+			rval = XFS_ILOG_DEXT;
+		else {
+			rval = 0;
+			error = xfs_bmbt_lookup_eq(cur, &old, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_bmbt_update(cur, &PREV);
+			if (error)
+				goto done;
+			error = xfs_btree_increment(cur, 0, &i);
+			if (error)
+				goto done;
+			error = xfs_bmbt_update(cur, &RIGHT);
+			if (error)
+				goto done;
+		}
+		break;
+
+	case BMAP_RIGHT_FILLING:
+		/*
+		 * Setting the last part of a previous oldext extent to newext.
+		 * The right neighbor is not contiguous.
+		 */
+		old = PREV;
+		PREV.br_blockcount -= new->br_blockcount;
+
+		xfs_iext_update_extent(ip, state, icur, &PREV);
+		xfs_iext_next(ifp, icur);
+		xfs_iext_insert(ip, icur, new, state);
+		ifp->if_nextents++;
+
+		if (cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			error = xfs_bmbt_lookup_eq(cur, &old, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_bmbt_update(cur, &PREV);
+			if (error)
+				goto done;
+			error = xfs_bmbt_lookup_eq(cur, new, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 0)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			if ((error = xfs_btree_insert(cur, &i)))
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+		}
+		break;
+
+	case 0:
+		/*
+		 * Setting the middle part of a previous oldext extent to
+		 * newext.  Contiguity is impossible here.
+		 * One extent becomes three extents.
+		 */
+		old = PREV;
+		PREV.br_blockcount = new->br_startoff - PREV.br_startoff;
+
+		r[0] = *new;
+		r[1].br_startoff = new_endoff;
+		r[1].br_blockcount =
+			old.br_startoff + old.br_blockcount - new_endoff;
+		r[1].br_startblock = new->br_startblock + new->br_blockcount;
+		r[1].br_state = PREV.br_state;
+
+		xfs_iext_update_extent(ip, state, icur, &PREV);
+		xfs_iext_next(ifp, icur);
+		xfs_iext_insert(ip, icur, &r[1], state);
+		xfs_iext_insert(ip, icur, &r[0], state);
+		ifp->if_nextents += 2;
+
+		if (cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			error = xfs_bmbt_lookup_eq(cur, &old, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			/* new right extent - oldext */
+			error = xfs_bmbt_update(cur, &r[1]);
+			if (error)
+				goto done;
+			/* new left extent - oldext */
+			cur->bc_rec.b = PREV;
+			if ((error = xfs_btree_insert(cur, &i)))
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			/*
+			 * Reset the cursor to the position of the new extent
+			 * we are about to insert as we can't trust it after
+			 * the previous insert.
+			 */
+			error = xfs_bmbt_lookup_eq(cur, new, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 0)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			/* new middle extent - newext */
+			if ((error = xfs_btree_insert(cur, &i)))
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+		}
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG:
+	case BMAP_RIGHT_FILLING | BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG:
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_CONTIG:
+	case BMAP_RIGHT_FILLING | BMAP_LEFT_CONTIG:
+	case BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG:
+	case BMAP_LEFT_CONTIG:
+	case BMAP_RIGHT_CONTIG:
+		/*
+		 * These cases are all impossible.
+		 */
+		ASSERT(0);
+	}
+
+	/* update reverse mappings */
+	xfs_rmap_convert_extent(mp, tp, ip, whichfork, new);
+
+	/* convert to a btree if necessary */
+	if (xfs_bmap_needs_btree(ip, whichfork)) {
+		int	tmp_logflags;	/* partial log flag return val */
+
+		ASSERT(cur == NULL);
+		error = xfs_bmap_extents_to_btree(tp, ip, &cur, 0,
+				&tmp_logflags, whichfork);
+		*logflagsp |= tmp_logflags;
+		if (error)
+			goto done;
+	}
+
+	/* clear out the allocated field, done with it now in any case. */
+	if (cur) {
+		cur->bc_ino.allocated = 0;
+		*curp = cur;
+	}
+
+	xfs_bmap_check_leaf_extents(*curp, ip, whichfork);
+done:
+	*logflagsp |= rval;
+	return error;
+#undef	LEFT
+#undef	RIGHT
+#undef	PREV
+}
+
+/*
+ * Convert a hole to a delayed allocation.
+ */
+STATIC void
+xfs_bmap_add_extent_hole_delay(
+	xfs_inode_t		*ip,	/* incore inode pointer */
+	int			whichfork,
+	struct xfs_iext_cursor	*icur,
+	xfs_bmbt_irec_t		*new)	/* new data to add to file extents */
+{
+	struct xfs_ifork	*ifp;	/* inode fork pointer */
+	xfs_bmbt_irec_t		left;	/* left neighbor extent entry */
+	xfs_filblks_t		newlen=0;	/* new indirect size */
+	xfs_filblks_t		oldlen=0;	/* old indirect size */
+	xfs_bmbt_irec_t		right;	/* right neighbor extent entry */
+	uint32_t		state = xfs_bmap_fork_to_state(whichfork);
+	xfs_filblks_t		temp;	 /* temp for indirect calculations */
+
+	ifp = xfs_ifork_ptr(ip, whichfork);
+	ASSERT(isnullstartblock(new->br_startblock));
+
+	/*
+	 * Check and set flags if this segment has a left neighbor
+	 */
+	if (xfs_iext_peek_prev_extent(ifp, icur, &left)) {
+		state |= BMAP_LEFT_VALID;
+		if (isnullstartblock(left.br_startblock))
+			state |= BMAP_LEFT_DELAY;
+	}
+
+	/*
+	 * Check and set flags if the current (right) segment exists.
+	 * If it doesn't exist, we're converting the hole at end-of-file.
+	 */
+	if (xfs_iext_get_extent(ifp, icur, &right)) {
+		state |= BMAP_RIGHT_VALID;
+		if (isnullstartblock(right.br_startblock))
+			state |= BMAP_RIGHT_DELAY;
+	}
+
+	/*
+	 * Set contiguity flags on the left and right neighbors.
+	 * Don't let extents get too large, even if the pieces are contiguous.
+	 */
+	if ((state & BMAP_LEFT_VALID) && (state & BMAP_LEFT_DELAY) &&
+	    left.br_startoff + left.br_blockcount == new->br_startoff &&
+	    left.br_blockcount + new->br_blockcount <= XFS_MAX_BMBT_EXTLEN)
+		state |= BMAP_LEFT_CONTIG;
+
+	if ((state & BMAP_RIGHT_VALID) && (state & BMAP_RIGHT_DELAY) &&
+	    new->br_startoff + new->br_blockcount == right.br_startoff &&
+	    new->br_blockcount + right.br_blockcount <= XFS_MAX_BMBT_EXTLEN &&
+	    (!(state & BMAP_LEFT_CONTIG) ||
+	     (left.br_blockcount + new->br_blockcount +
+	      right.br_blockcount <= XFS_MAX_BMBT_EXTLEN)))
+		state |= BMAP_RIGHT_CONTIG;
+
+	/*
+	 * Switch out based on the contiguity flags.
+	 */
+	switch (state & (BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG)) {
+	case BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG:
+		/*
+		 * New allocation is contiguous with delayed allocations
+		 * on the left and on the right.
+		 * Merge all three into a single extent record.
+		 */
+		temp = left.br_blockcount + new->br_blockcount +
+			right.br_blockcount;
+
+		oldlen = startblockval(left.br_startblock) +
+			startblockval(new->br_startblock) +
+			startblockval(right.br_startblock);
+		newlen = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
+					 oldlen);
+		left.br_startblock = nullstartblock(newlen);
+		left.br_blockcount = temp;
+
+		xfs_iext_remove(ip, icur, state);
+		xfs_iext_prev(ifp, icur);
+		xfs_iext_update_extent(ip, state, icur, &left);
+		break;
+
+	case BMAP_LEFT_CONTIG:
+		/*
+		 * New allocation is contiguous with a delayed allocation
+		 * on the left.
+		 * Merge the new allocation with the left neighbor.
+		 */
+		temp = left.br_blockcount + new->br_blockcount;
+
+		oldlen = startblockval(left.br_startblock) +
+			startblockval(new->br_startblock);
+		newlen = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
+					 oldlen);
+		left.br_blockcount = temp;
+		left.br_startblock = nullstartblock(newlen);
+
+		xfs_iext_prev(ifp, icur);
+		xfs_iext_update_extent(ip, state, icur, &left);
+		break;
+
+	case BMAP_RIGHT_CONTIG:
+		/*
+		 * New allocation is contiguous with a delayed allocation
+		 * on the right.
+		 * Merge the new allocation with the right neighbor.
+		 */
+		temp = new->br_blockcount + right.br_blockcount;
+		oldlen = startblockval(new->br_startblock) +
+			startblockval(right.br_startblock);
+		newlen = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
+					 oldlen);
+		right.br_startoff = new->br_startoff;
+		right.br_startblock = nullstartblock(newlen);
+		right.br_blockcount = temp;
+		xfs_iext_update_extent(ip, state, icur, &right);
+		break;
+
+	case 0:
+		/*
+		 * New allocation is not contiguous with another
+		 * delayed allocation.
+		 * Insert a new entry.
+		 */
+		oldlen = newlen = 0;
+		xfs_iext_insert(ip, icur, new, state);
+		break;
+	}
+	if (oldlen != newlen) {
+		ASSERT(oldlen > newlen);
+		xfs_mod_fdblocks(ip->i_mount, (int64_t)(oldlen - newlen),
+				 false);
+		/*
+		 * Nothing to do for disk quota accounting here.
+		 */
+		xfs_mod_delalloc(ip->i_mount, (int64_t)newlen - oldlen);
+	}
+}
+
+/*
+ * Convert a hole to a real allocation.
+ */
+STATIC int				/* error */
+xfs_bmap_add_extent_hole_real(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	int			whichfork,
+	struct xfs_iext_cursor	*icur,
+	struct xfs_btree_cur	**curp,
+	struct xfs_bmbt_irec	*new,
+	int			*logflagsp,
+	uint32_t		flags)
+{
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_btree_cur	*cur = *curp;
+	int			error;	/* error return value */
+	int			i;	/* temp state */
+	xfs_bmbt_irec_t		left;	/* left neighbor extent entry */
+	xfs_bmbt_irec_t		right;	/* right neighbor extent entry */
+	int			rval=0;	/* return value (logging flags) */
+	uint32_t		state = xfs_bmap_fork_to_state(whichfork);
+	struct xfs_bmbt_irec	old;
+
+	ASSERT(!isnullstartblock(new->br_startblock));
+	ASSERT(!cur || !(cur->bc_ino.flags & XFS_BTCUR_BMBT_WASDEL));
+
+	XFS_STATS_INC(mp, xs_add_exlist);
+
+	/*
+	 * Check and set flags if this segment has a left neighbor.
+	 */
+	if (xfs_iext_peek_prev_extent(ifp, icur, &left)) {
+		state |= BMAP_LEFT_VALID;
+		if (isnullstartblock(left.br_startblock))
+			state |= BMAP_LEFT_DELAY;
+	}
+
+	/*
+	 * Check and set flags if this segment has a current value.
+	 * Not true if we're inserting into the "hole" at eof.
+	 */
+	if (xfs_iext_get_extent(ifp, icur, &right)) {
+		state |= BMAP_RIGHT_VALID;
+		if (isnullstartblock(right.br_startblock))
+			state |= BMAP_RIGHT_DELAY;
+	}
+
+	/*
+	 * We're inserting a real allocation between "left" and "right".
+	 * Set the contiguity flags.  Don't let extents get too large.
+	 */
+	if ((state & BMAP_LEFT_VALID) && !(state & BMAP_LEFT_DELAY) &&
+	    left.br_startoff + left.br_blockcount == new->br_startoff &&
+	    left.br_startblock + left.br_blockcount == new->br_startblock &&
+	    left.br_state == new->br_state &&
+	    left.br_blockcount + new->br_blockcount <= XFS_MAX_BMBT_EXTLEN)
+		state |= BMAP_LEFT_CONTIG;
+
+	if ((state & BMAP_RIGHT_VALID) && !(state & BMAP_RIGHT_DELAY) &&
+	    new->br_startoff + new->br_blockcount == right.br_startoff &&
+	    new->br_startblock + new->br_blockcount == right.br_startblock &&
+	    new->br_state == right.br_state &&
+	    new->br_blockcount + right.br_blockcount <= XFS_MAX_BMBT_EXTLEN &&
+	    (!(state & BMAP_LEFT_CONTIG) ||
+	     left.br_blockcount + new->br_blockcount +
+	     right.br_blockcount <= XFS_MAX_BMBT_EXTLEN))
+		state |= BMAP_RIGHT_CONTIG;
+
+	error = 0;
+	/*
+	 * Select which case we're in here, and implement it.
+	 */
+	switch (state & (BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG)) {
+	case BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG:
+		/*
+		 * New allocation is contiguous with real allocations on the
+		 * left and on the right.
+		 * Merge all three into a single extent record.
+		 */
+		left.br_blockcount += new->br_blockcount + right.br_blockcount;
+
+		xfs_iext_remove(ip, icur, state);
+		xfs_iext_prev(ifp, icur);
+		xfs_iext_update_extent(ip, state, icur, &left);
+		ifp->if_nextents--;
+
+		if (cur == NULL) {
+			rval = XFS_ILOG_CORE | xfs_ilog_fext(whichfork);
+		} else {
+			rval = XFS_ILOG_CORE;
+			error = xfs_bmbt_lookup_eq(cur, &right, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_btree_delete(cur, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_btree_decrement(cur, 0, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_bmbt_update(cur, &left);
+			if (error)
+				goto done;
+		}
+		break;
+
+	case BMAP_LEFT_CONTIG:
+		/*
+		 * New allocation is contiguous with a real allocation
+		 * on the left.
+		 * Merge the new allocation with the left neighbor.
+		 */
+		old = left;
+		left.br_blockcount += new->br_blockcount;
+
+		xfs_iext_prev(ifp, icur);
+		xfs_iext_update_extent(ip, state, icur, &left);
+
+		if (cur == NULL) {
+			rval = xfs_ilog_fext(whichfork);
+		} else {
+			rval = 0;
+			error = xfs_bmbt_lookup_eq(cur, &old, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_bmbt_update(cur, &left);
+			if (error)
+				goto done;
+		}
+		break;
+
+	case BMAP_RIGHT_CONTIG:
+		/*
+		 * New allocation is contiguous with a real allocation
+		 * on the right.
+		 * Merge the new allocation with the right neighbor.
+		 */
+		old = right;
+
+		right.br_startoff = new->br_startoff;
+		right.br_startblock = new->br_startblock;
+		right.br_blockcount += new->br_blockcount;
+		xfs_iext_update_extent(ip, state, icur, &right);
+
+		if (cur == NULL) {
+			rval = xfs_ilog_fext(whichfork);
+		} else {
+			rval = 0;
+			error = xfs_bmbt_lookup_eq(cur, &old, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_bmbt_update(cur, &right);
+			if (error)
+				goto done;
+		}
+		break;
+
+	case 0:
+		/*
+		 * New allocation is not contiguous with another
+		 * real allocation.
+		 * Insert a new entry.
+		 */
+		xfs_iext_insert(ip, icur, new, state);
+		ifp->if_nextents++;
+
+		if (cur == NULL) {
+			rval = XFS_ILOG_CORE | xfs_ilog_fext(whichfork);
+		} else {
+			rval = XFS_ILOG_CORE;
+			error = xfs_bmbt_lookup_eq(cur, new, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 0)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+			error = xfs_btree_insert(cur, &i);
+			if (error)
+				goto done;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+		}
+		break;
+	}
+
+	/* add reverse mapping unless caller opted out */
+	if (!(flags & XFS_BMAPI_NORMAP))
+		xfs_rmap_map_extent(tp, ip, whichfork, new);
+
+	/* convert to a btree if necessary */
+	if (xfs_bmap_needs_btree(ip, whichfork)) {
+		int	tmp_logflags;	/* partial log flag return val */
+
+		ASSERT(cur == NULL);
+		error = xfs_bmap_extents_to_btree(tp, ip, curp, 0,
+				&tmp_logflags, whichfork);
+		*logflagsp |= tmp_logflags;
+		cur = *curp;
+		if (error)
+			goto done;
+	}
+
+	/* clear out the allocated field, done with it now in any case. */
+	if (cur)
+		cur->bc_ino.allocated = 0;
+
+	xfs_bmap_check_leaf_extents(cur, ip, whichfork);
+done:
+	*logflagsp |= rval;
+	return error;
+}
+
+/*
+ * Functions used in the extent read, allocate and remove paths
+ */
+
+/*
+ * Adjust the size of the new extent based on i_extsize and rt extsize.
+ */
+int
+xfs_bmap_extsize_align(
+	xfs_mount_t	*mp,
+	xfs_bmbt_irec_t	*gotp,		/* next extent pointer */
+	xfs_bmbt_irec_t	*prevp,		/* previous extent pointer */
+	xfs_extlen_t	extsz,		/* align to this extent size */
+	int		rt,		/* is this a realtime inode? */
+	int		eof,		/* is extent at end-of-file? */
+	int		delay,		/* creating delalloc extent? */
+	int		convert,	/* overwriting unwritten extent? */
+	xfs_fileoff_t	*offp,		/* in/out: aligned offset */
+	xfs_extlen_t	*lenp)		/* in/out: aligned length */
+{
+	xfs_fileoff_t	orig_off;	/* original offset */
+	xfs_extlen_t	orig_alen;	/* original length */
+	xfs_fileoff_t	orig_end;	/* original off+len */
+	xfs_fileoff_t	nexto;		/* next file offset */
+	xfs_fileoff_t	prevo;		/* previous file offset */
+	xfs_fileoff_t	align_off;	/* temp for offset */
+	xfs_extlen_t	align_alen;	/* temp for length */
+	xfs_extlen_t	temp;		/* temp for calculations */
+
+	if (convert)
+		return 0;
+
+	orig_off = align_off = *offp;
+	orig_alen = align_alen = *lenp;
+	orig_end = orig_off + orig_alen;
+
+	/*
+	 * If this request overlaps an existing extent, then don't
+	 * attempt to perform any additional alignment.
+	 */
+	if (!delay && !eof &&
+	    (orig_off >= gotp->br_startoff) &&
+	    (orig_end <= gotp->br_startoff + gotp->br_blockcount)) {
+		return 0;
+	}
+
+	/*
+	 * If the file offset is unaligned vs. the extent size
+	 * we need to align it.  This will be possible unless
+	 * the file was previously written with a kernel that didn't
+	 * perform this alignment, or if a truncate shot us in the
+	 * foot.
+	 */
+	div_u64_rem(orig_off, extsz, &temp);
+	if (temp) {
+		align_alen += temp;
+		align_off -= temp;
+	}
+
+	/* Same adjustment for the end of the requested area. */
+	temp = (align_alen % extsz);
+	if (temp)
+		align_alen += extsz - temp;
+
+	/*
+	 * For large extent hint sizes, the aligned extent might be larger than
+	 * XFS_BMBT_MAX_EXTLEN. In that case, reduce the size by an extsz so
+	 * that it pulls the length back under XFS_BMBT_MAX_EXTLEN. The outer
+	 * allocation loops handle short allocation just fine, so it is safe to
+	 * do this. We only want to do it when we are forced to, though, because
+	 * it means more allocation operations are required.
+	 */
+	while (align_alen > XFS_MAX_BMBT_EXTLEN)
+		align_alen -= extsz;
+	ASSERT(align_alen <= XFS_MAX_BMBT_EXTLEN);
+
+	/*
+	 * If the previous block overlaps with this proposed allocation
+	 * then move the start forward without adjusting the length.
+	 */
+	if (prevp->br_startoff != NULLFILEOFF) {
+		if (prevp->br_startblock == HOLESTARTBLOCK)
+			prevo = prevp->br_startoff;
+		else
+			prevo = prevp->br_startoff + prevp->br_blockcount;
+	} else
+		prevo = 0;
+	if (align_off != orig_off && align_off < prevo)
+		align_off = prevo;
+	/*
+	 * If the next block overlaps with this proposed allocation
+	 * then move the start back without adjusting the length,
+	 * but not before offset 0.
+	 * This may of course make the start overlap previous block,
+	 * and if we hit the offset 0 limit then the next block
+	 * can still overlap too.
+	 */
+	if (!eof && gotp->br_startoff != NULLFILEOFF) {
+		if ((delay && gotp->br_startblock == HOLESTARTBLOCK) ||
+		    (!delay && gotp->br_startblock == DELAYSTARTBLOCK))
+			nexto = gotp->br_startoff + gotp->br_blockcount;
+		else
+			nexto = gotp->br_startoff;
+	} else
+		nexto = NULLFILEOFF;
+	if (!eof &&
+	    align_off + align_alen != orig_end &&
+	    align_off + align_alen > nexto)
+		align_off = nexto > align_alen ? nexto - align_alen : 0;
+	/*
+	 * If we're now overlapping the next or previous extent that
+	 * means we can't fit an extsz piece in this hole.  Just move
+	 * the start forward to the first valid spot and set
+	 * the length so we hit the end.
+	 */
+	if (align_off != orig_off && align_off < prevo)
+		align_off = prevo;
+	if (align_off + align_alen != orig_end &&
+	    align_off + align_alen > nexto &&
+	    nexto != NULLFILEOFF) {
+		ASSERT(nexto > prevo);
+		align_alen = nexto - align_off;
+	}
+
+	/*
+	 * If realtime, and the result isn't a multiple of the realtime
+	 * extent size we need to remove blocks until it is.
+	 */
+	if (rt && (temp = (align_alen % mp->m_sb.sb_rextsize))) {
+		/*
+		 * We're not covering the original request, or
+		 * we won't be able to once we fix the length.
+		 */
+		if (orig_off < align_off ||
+		    orig_end > align_off + align_alen ||
+		    align_alen - temp < orig_alen)
+			return -EINVAL;
+		/*
+		 * Try to fix it by moving the start up.
+		 */
+		if (align_off + temp <= orig_off) {
+			align_alen -= temp;
+			align_off += temp;
+		}
+		/*
+		 * Try to fix it by moving the end in.
+		 */
+		else if (align_off + align_alen - temp >= orig_end)
+			align_alen -= temp;
+		/*
+		 * Set the start to the minimum then trim the length.
+		 */
+		else {
+			align_alen -= orig_off - align_off;
+			align_off = orig_off;
+			align_alen -= align_alen % mp->m_sb.sb_rextsize;
+		}
+		/*
+		 * Result doesn't cover the request, fail it.
+		 */
+		if (orig_off < align_off || orig_end > align_off + align_alen)
+			return -EINVAL;
+	} else {
+		ASSERT(orig_off >= align_off);
+		/* see XFS_BMBT_MAX_EXTLEN handling above */
+		ASSERT(orig_end <= align_off + align_alen ||
+		       align_alen + extsz > XFS_MAX_BMBT_EXTLEN);
+	}
+
+#ifdef DEBUG
+	if (!eof && gotp->br_startoff != NULLFILEOFF)
+		ASSERT(align_off + align_alen <= gotp->br_startoff);
+	if (prevp->br_startoff != NULLFILEOFF)
+		ASSERT(align_off >= prevp->br_startoff + prevp->br_blockcount);
+#endif
+
+	*lenp = align_alen;
+	*offp = align_off;
+	return 0;
+}
+
+#define XFS_ALLOC_GAP_UNITS	4
+
+void
+xfs_bmap_adjacent(
+	struct xfs_bmalloca	*ap)	/* bmap alloc argument struct */
+{
+	xfs_fsblock_t	adjust;		/* adjustment to block numbers */
+	xfs_agnumber_t	fb_agno;	/* ag number of ap->firstblock */
+	xfs_mount_t	*mp;		/* mount point structure */
+	int		nullfb;		/* true if ap->firstblock isn't set */
+	int		rt;		/* true if inode is realtime */
+
+#define	ISVALID(x,y)	\
+	(rt ? \
+		(x) < mp->m_sb.sb_rblocks : \
+		XFS_FSB_TO_AGNO(mp, x) == XFS_FSB_TO_AGNO(mp, y) && \
+		XFS_FSB_TO_AGNO(mp, x) < mp->m_sb.sb_agcount && \
+		XFS_FSB_TO_AGBNO(mp, x) < mp->m_sb.sb_agblocks)
+
+	mp = ap->ip->i_mount;
+	nullfb = ap->tp->t_firstblock == NULLFSBLOCK;
+	rt = XFS_IS_REALTIME_INODE(ap->ip) &&
+		(ap->datatype & XFS_ALLOC_USERDATA);
+	fb_agno = nullfb ? NULLAGNUMBER : XFS_FSB_TO_AGNO(mp,
+							ap->tp->t_firstblock);
+	/*
+	 * If allocating at eof, and there's a previous real block,
+	 * try to use its last block as our starting point.
+	 */
+	if (ap->eof && ap->prev.br_startoff != NULLFILEOFF &&
+	    !isnullstartblock(ap->prev.br_startblock) &&
+	    ISVALID(ap->prev.br_startblock + ap->prev.br_blockcount,
+		    ap->prev.br_startblock)) {
+		ap->blkno = ap->prev.br_startblock + ap->prev.br_blockcount;
+		/*
+		 * Adjust for the gap between prevp and us.
+		 */
+		adjust = ap->offset -
+			(ap->prev.br_startoff + ap->prev.br_blockcount);
+		if (adjust &&
+		    ISVALID(ap->blkno + adjust, ap->prev.br_startblock))
+			ap->blkno += adjust;
+	}
+	/*
+	 * If not at eof, then compare the two neighbor blocks.
+	 * Figure out whether either one gives us a good starting point,
+	 * and pick the better one.
+	 */
+	else if (!ap->eof) {
+		xfs_fsblock_t	gotbno;		/* right side block number */
+		xfs_fsblock_t	gotdiff=0;	/* right side difference */
+		xfs_fsblock_t	prevbno;	/* left side block number */
+		xfs_fsblock_t	prevdiff=0;	/* left side difference */
+
+		/*
+		 * If there's a previous (left) block, select a requested
+		 * start block based on it.
+		 */
+		if (ap->prev.br_startoff != NULLFILEOFF &&
+		    !isnullstartblock(ap->prev.br_startblock) &&
+		    (prevbno = ap->prev.br_startblock +
+			       ap->prev.br_blockcount) &&
+		    ISVALID(prevbno, ap->prev.br_startblock)) {
+			/*
+			 * Calculate gap to end of previous block.
+			 */
+			adjust = prevdiff = ap->offset -
+				(ap->prev.br_startoff +
+				 ap->prev.br_blockcount);
+			/*
+			 * Figure the startblock based on the previous block's
+			 * end and the gap size.
+			 * Heuristic!
+			 * If the gap is large relative to the piece we're
+			 * allocating, or using it gives us an invalid block
+			 * number, then just use the end of the previous block.
+			 */
+			if (prevdiff <= XFS_ALLOC_GAP_UNITS * ap->length &&
+			    ISVALID(prevbno + prevdiff,
+				    ap->prev.br_startblock))
+				prevbno += adjust;
+			else
+				prevdiff += adjust;
+			/*
+			 * If the firstblock forbids it, can't use it,
+			 * must use default.
+			 */
+			if (!rt && !nullfb &&
+			    XFS_FSB_TO_AGNO(mp, prevbno) != fb_agno)
+				prevbno = NULLFSBLOCK;
+		}
+		/*
+		 * No previous block or can't follow it, just default.
+		 */
+		else
+			prevbno = NULLFSBLOCK;
+		/*
+		 * If there's a following (right) block, select a requested
+		 * start block based on it.
+		 */
+		if (!isnullstartblock(ap->got.br_startblock)) {
+			/*
+			 * Calculate gap to start of next block.
+			 */
+			adjust = gotdiff = ap->got.br_startoff - ap->offset;
+			/*
+			 * Figure the startblock based on the next block's
+			 * start and the gap size.
+			 */
+			gotbno = ap->got.br_startblock;
+			/*
+			 * Heuristic!
+			 * If the gap is large relative to the piece we're
+			 * allocating, or using it gives us an invalid block
+			 * number, then just use the start of the next block
+			 * offset by our length.
+			 */
+			if (gotdiff <= XFS_ALLOC_GAP_UNITS * ap->length &&
+			    ISVALID(gotbno - gotdiff, gotbno))
+				gotbno -= adjust;
+			else if (ISVALID(gotbno - ap->length, gotbno)) {
+				gotbno -= ap->length;
+				gotdiff += adjust - ap->length;
+			} else
+				gotdiff += adjust;
+			/*
+			 * If the firstblock forbids it, can't use it,
+			 * must use default.
+			 */
+			if (!rt && !nullfb &&
+			    XFS_FSB_TO_AGNO(mp, gotbno) != fb_agno)
+				gotbno = NULLFSBLOCK;
+		}
+		/*
+		 * No next block, just default.
+		 */
+		else
+			gotbno = NULLFSBLOCK;
+		/*
+		 * If both valid, pick the better one, else the only good
+		 * one, else ap->blkno is already set (to 0 or the inode block).
+		 */
+		if (prevbno != NULLFSBLOCK && gotbno != NULLFSBLOCK)
+			ap->blkno = prevdiff <= gotdiff ? prevbno : gotbno;
+		else if (prevbno != NULLFSBLOCK)
+			ap->blkno = prevbno;
+		else if (gotbno != NULLFSBLOCK)
+			ap->blkno = gotbno;
+	}
+#undef ISVALID
+}
+
+static int
+xfs_bmap_longest_free_extent(
+	struct xfs_trans	*tp,
+	xfs_agnumber_t		ag,
+	xfs_extlen_t		*blen,
+	int			*notinit)
+{
+	struct xfs_mount	*mp = tp->t_mountp;
+	struct xfs_perag	*pag;
+	xfs_extlen_t		longest;
+	int			error = 0;
+
+	pag = xfs_perag_get(mp, ag);
+	if (!pag->pagf_init) {
+		error = xfs_alloc_read_agf(pag, tp, XFS_ALLOC_FLAG_TRYLOCK,
+				NULL);
+		if (error) {
+			/* Couldn't lock the AGF, so skip this AG. */
+			if (error == -EAGAIN) {
+				*notinit = 1;
+				error = 0;
+			}
+			goto out;
+		}
+	}
+
+	longest = xfs_alloc_longest_free_extent(pag,
+				xfs_alloc_min_freelist(mp, pag),
+				xfs_ag_resv_needed(pag, XFS_AG_RESV_NONE));
+	if (*blen < longest)
+		*blen = longest;
+
+out:
+	xfs_perag_put(pag);
+	return error;
+}
+
+static void
+xfs_bmap_select_minlen(
+	struct xfs_bmalloca	*ap,
+	struct xfs_alloc_arg	*args,
+	xfs_extlen_t		*blen,
+	int			notinit)
+{
+	if (notinit || *blen < ap->minlen) {
+		/*
+		 * Since we did a BUF_TRYLOCK above, it is possible that
+		 * there is space for this request.
+		 */
+		args->minlen = ap->minlen;
+	} else if (*blen < args->maxlen) {
+		/*
+		 * If the best seen length is less than the request length,
+		 * use the best as the minimum.
+		 */
+		args->minlen = *blen;
+	} else {
+		/*
+		 * Otherwise we've seen an extent as big as maxlen, use that
+		 * as the minimum.
+		 */
+		args->minlen = args->maxlen;
+	}
+}
+
+STATIC int
+xfs_bmap_btalloc_nullfb(
+	struct xfs_bmalloca	*ap,
+	struct xfs_alloc_arg	*args,
+	xfs_extlen_t		*blen)
+{
+	struct xfs_mount	*mp = ap->ip->i_mount;
+	xfs_agnumber_t		ag, startag;
+	int			notinit = 0;
+	int			error;
+
+	args->type = XFS_ALLOCTYPE_START_BNO;
+	args->total = ap->total;
+
+	startag = ag = XFS_FSB_TO_AGNO(mp, args->fsbno);
+	if (startag == NULLAGNUMBER)
+		startag = ag = 0;
+
+	while (*blen < args->maxlen) {
+		error = xfs_bmap_longest_free_extent(args->tp, ag, blen,
+						     &notinit);
+		if (error)
+			return error;
+
+		if (++ag == mp->m_sb.sb_agcount)
+			ag = 0;
+		if (ag == startag)
+			break;
+	}
+
+	xfs_bmap_select_minlen(ap, args, blen, notinit);
+	return 0;
+}
+
+STATIC int
+xfs_bmap_btalloc_filestreams(
+	struct xfs_bmalloca	*ap,
+	struct xfs_alloc_arg	*args,
+	xfs_extlen_t		*blen)
+{
+	struct xfs_mount	*mp = ap->ip->i_mount;
+	xfs_agnumber_t		ag;
+	int			notinit = 0;
+	int			error;
+
+	args->type = XFS_ALLOCTYPE_NEAR_BNO;
+	args->total = ap->total;
+
+	ag = XFS_FSB_TO_AGNO(mp, args->fsbno);
+	if (ag == NULLAGNUMBER)
+		ag = 0;
+
+	error = xfs_bmap_longest_free_extent(args->tp, ag, blen, &notinit);
+	if (error)
+		return error;
+
+	if (*blen < args->maxlen) {
+		error = xfs_filestream_new_ag(ap, &ag);
+		if (error)
+			return error;
+
+		error = xfs_bmap_longest_free_extent(args->tp, ag, blen,
+						     &notinit);
+		if (error)
+			return error;
+
+	}
+
+	xfs_bmap_select_minlen(ap, args, blen, notinit);
+
+	/*
+	 * Set the failure fallback case to look in the selected AG as stream
+	 * may have moved.
+	 */
+	ap->blkno = args->fsbno = XFS_AGB_TO_FSB(mp, ag, 0);
+	return 0;
+}
+
+/* Update all inode and quota accounting for the allocation we just did. */
+static void
+xfs_bmap_btalloc_accounting(
+	struct xfs_bmalloca	*ap,
+	struct xfs_alloc_arg	*args)
+{
+	if (ap->flags & XFS_BMAPI_COWFORK) {
+		/*
+		 * COW fork blocks are in-core only and thus are treated as
+		 * in-core quota reservation (like delalloc blocks) even when
+		 * converted to real blocks. The quota reservation is not
+		 * accounted to disk until blocks are remapped to the data
+		 * fork. So if these blocks were previously delalloc, we
+		 * already have quota reservation and there's nothing to do
+		 * yet.
+		 */
+		if (ap->wasdel) {
+			xfs_mod_delalloc(ap->ip->i_mount, -(int64_t)args->len);
+			return;
+		}
+
+		/*
+		 * Otherwise, we've allocated blocks in a hole. The transaction
+		 * has acquired in-core quota reservation for this extent.
+		 * Rather than account these as real blocks, however, we reduce
+		 * the transaction quota reservation based on the allocation.
+		 * This essentially transfers the transaction quota reservation
+		 * to that of a delalloc extent.
+		 */
+		ap->ip->i_delayed_blks += args->len;
+		xfs_trans_mod_dquot_byino(ap->tp, ap->ip, XFS_TRANS_DQ_RES_BLKS,
+				-(long)args->len);
+		return;
+	}
+
+	/* data/attr fork only */
+	ap->ip->i_nblocks += args->len;
+	xfs_trans_log_inode(ap->tp, ap->ip, XFS_ILOG_CORE);
+	if (ap->wasdel) {
+		ap->ip->i_delayed_blks -= args->len;
+		xfs_mod_delalloc(ap->ip->i_mount, -(int64_t)args->len);
+	}
+	xfs_trans_mod_dquot_byino(ap->tp, ap->ip,
+		ap->wasdel ? XFS_TRANS_DQ_DELBCOUNT : XFS_TRANS_DQ_BCOUNT,
+		args->len);
+}
+
+static int
+xfs_bmap_compute_alignments(
+	struct xfs_bmalloca	*ap,
+	struct xfs_alloc_arg	*args)
+{
+	struct xfs_mount	*mp = args->mp;
+	xfs_extlen_t		align = 0; /* minimum allocation alignment */
+	int			stripe_align = 0;
+
+	/* stripe alignment for allocation is determined by mount parameters */
+	if (mp->m_swidth && xfs_has_swalloc(mp))
+		stripe_align = mp->m_swidth;
+	else if (mp->m_dalign)
+		stripe_align = mp->m_dalign;
+
+	if (ap->flags & XFS_BMAPI_COWFORK)
+		align = xfs_get_cowextsz_hint(ap->ip);
+	else if (ap->datatype & XFS_ALLOC_USERDATA)
+		align = xfs_get_extsz_hint(ap->ip);
+	if (align) {
+		if (xfs_bmap_extsize_align(mp, &ap->got, &ap->prev, align, 0,
+					ap->eof, 0, ap->conv, &ap->offset,
+					&ap->length))
+			ASSERT(0);
+		ASSERT(ap->length);
+	}
+
+	/* apply extent size hints if obtained earlier */
+	if (align) {
+		args->prod = align;
+		div_u64_rem(ap->offset, args->prod, &args->mod);
+		if (args->mod)
+			args->mod = args->prod - args->mod;
+	} else if (mp->m_sb.sb_blocksize >= PAGE_SIZE) {
+		args->prod = 1;
+		args->mod = 0;
+	} else {
+		args->prod = PAGE_SIZE >> mp->m_sb.sb_blocklog;
+		div_u64_rem(ap->offset, args->prod, &args->mod);
+		if (args->mod)
+			args->mod = args->prod - args->mod;
+	}
+
+	return stripe_align;
+}
+
+static void
+xfs_bmap_process_allocated_extent(
+	struct xfs_bmalloca	*ap,
+	struct xfs_alloc_arg	*args,
+	xfs_fileoff_t		orig_offset,
+	xfs_extlen_t		orig_length)
+{
+	int			nullfb;
+
+	nullfb = ap->tp->t_firstblock == NULLFSBLOCK;
+
+	/*
+	 * check the allocation happened at the same or higher AG than
+	 * the first block that was allocated.
+	 */
+	ASSERT(nullfb ||
+		XFS_FSB_TO_AGNO(args->mp, ap->tp->t_firstblock) <=
+		XFS_FSB_TO_AGNO(args->mp, args->fsbno));
+
+	ap->blkno = args->fsbno;
+	if (nullfb)
+		ap->tp->t_firstblock = args->fsbno;
+	ap->length = args->len;
+	/*
+	 * If the extent size hint is active, we tried to round the
+	 * caller's allocation request offset down to extsz and the
+	 * length up to another extsz boundary.  If we found a free
+	 * extent we mapped it in starting at this new offset.  If the
+	 * newly mapped space isn't long enough to cover any of the
+	 * range of offsets that was originally requested, move the
+	 * mapping up so that we can fill as much of the caller's
+	 * original request as possible.  Free space is apparently
+	 * very fragmented so we're unlikely to be able to satisfy the
+	 * hints anyway.
+	 */
+	if (ap->length <= orig_length)
+		ap->offset = orig_offset;
+	else if (ap->offset + ap->length < orig_offset + orig_length)
+		ap->offset = orig_offset + orig_length - ap->length;
+	xfs_bmap_btalloc_accounting(ap, args);
+}
+
+#ifdef DEBUG
+static int
+xfs_bmap_exact_minlen_extent_alloc(
+	struct xfs_bmalloca	*ap)
+{
+	struct xfs_mount	*mp = ap->ip->i_mount;
+	struct xfs_alloc_arg	args = { .tp = ap->tp, .mp = mp };
+	xfs_fileoff_t		orig_offset;
+	xfs_extlen_t		orig_length;
+	int			error;
+
+	ASSERT(ap->length);
+
+	if (ap->minlen != 1) {
+		ap->blkno = NULLFSBLOCK;
+		ap->length = 0;
+		return 0;
+	}
+
+	orig_offset = ap->offset;
+	orig_length = ap->length;
+
+	args.alloc_minlen_only = 1;
+
+	xfs_bmap_compute_alignments(ap, &args);
+
+	if (ap->tp->t_firstblock == NULLFSBLOCK) {
+		/*
+		 * Unlike the longest extent available in an AG, we don't track
+		 * the length of an AG's shortest extent.
+		 * XFS_ERRTAG_BMAP_ALLOC_MINLEN_EXTENT is a debug only knob and
+		 * hence we can afford to start traversing from the 0th AG since
+		 * we need not be concerned about a drop in performance in
+		 * "debug only" code paths.
+		 */
+		ap->blkno = XFS_AGB_TO_FSB(mp, 0, 0);
+	} else {
+		ap->blkno = ap->tp->t_firstblock;
+	}
+
+	args.fsbno = ap->blkno;
+	args.oinfo = XFS_RMAP_OINFO_SKIP_UPDATE;
+	args.type = XFS_ALLOCTYPE_FIRST_AG;
+	args.minlen = args.maxlen = ap->minlen;
+	args.total = ap->total;
+
+	args.alignment = 1;
+	args.minalignslop = 0;
+
+	args.minleft = ap->minleft;
+	args.wasdel = ap->wasdel;
+	args.resv = XFS_AG_RESV_NONE;
+	args.datatype = ap->datatype;
+
+	error = xfs_alloc_vextent(&args);
+	if (error)
+		return error;
+
+	if (args.fsbno != NULLFSBLOCK) {
+		xfs_bmap_process_allocated_extent(ap, &args, orig_offset,
+			orig_length);
+	} else {
+		ap->blkno = NULLFSBLOCK;
+		ap->length = 0;
+	}
+
+	return 0;
+}
+#else
+
+#define xfs_bmap_exact_minlen_extent_alloc(bma) (-EFSCORRUPTED)
+
+#endif
+
+STATIC int
+xfs_bmap_btalloc(
+	struct xfs_bmalloca	*ap)
+{
+	struct xfs_mount	*mp = ap->ip->i_mount;
+	struct xfs_alloc_arg	args = { .tp = ap->tp, .mp = mp };
+	xfs_alloctype_t		atype = 0;
+	xfs_agnumber_t		fb_agno;	/* ag number of ap->firstblock */
+	xfs_agnumber_t		ag;
+	xfs_fileoff_t		orig_offset;
+	xfs_extlen_t		orig_length;
+	xfs_extlen_t		blen;
+	xfs_extlen_t		nextminlen = 0;
+	int			nullfb; /* true if ap->firstblock isn't set */
+	int			isaligned;
+	int			tryagain;
+	int			error;
+	int			stripe_align;
+
+	ASSERT(ap->length);
+	orig_offset = ap->offset;
+	orig_length = ap->length;
+
+	stripe_align = xfs_bmap_compute_alignments(ap, &args);
+
+	nullfb = ap->tp->t_firstblock == NULLFSBLOCK;
+	fb_agno = nullfb ? NULLAGNUMBER : XFS_FSB_TO_AGNO(mp,
+							ap->tp->t_firstblock);
+	if (nullfb) {
+		if ((ap->datatype & XFS_ALLOC_USERDATA) &&
+		    xfs_inode_is_filestream(ap->ip)) {
+			ag = xfs_filestream_lookup_ag(ap->ip);
+			ag = (ag != NULLAGNUMBER) ? ag : 0;
+			ap->blkno = XFS_AGB_TO_FSB(mp, ag, 0);
+		} else {
+			ap->blkno = XFS_INO_TO_FSB(mp, ap->ip->i_ino);
+		}
+	} else
+		ap->blkno = ap->tp->t_firstblock;
+
+	xfs_bmap_adjacent(ap);
+
+	/*
+	 * If allowed, use ap->blkno; otherwise must use firstblock since
+	 * it's in the right allocation group.
+	 */
+	if (nullfb || XFS_FSB_TO_AGNO(mp, ap->blkno) == fb_agno)
+		;
+	else
+		ap->blkno = ap->tp->t_firstblock;
+	/*
+	 * Normal allocation, done through xfs_alloc_vextent.
+	 */
+	tryagain = isaligned = 0;
+	args.fsbno = ap->blkno;
+	args.oinfo = XFS_RMAP_OINFO_SKIP_UPDATE;
+
+	/* Trim the allocation back to the maximum an AG can fit. */
+	args.maxlen = min(ap->length, mp->m_ag_max_usable);
+	blen = 0;
+	if (nullfb) {
+		/*
+		 * Search for an allocation group with a single extent large
+		 * enough for the request.  If one isn't found, then adjust
+		 * the minimum allocation size to the largest space found.
+		 */
+		if ((ap->datatype & XFS_ALLOC_USERDATA) &&
+		    xfs_inode_is_filestream(ap->ip))
+			error = xfs_bmap_btalloc_filestreams(ap, &args, &blen);
+		else
+			error = xfs_bmap_btalloc_nullfb(ap, &args, &blen);
+		if (error)
+			return error;
+	} else if (ap->tp->t_flags & XFS_TRANS_LOWMODE) {
+		if (xfs_inode_is_filestream(ap->ip))
+			args.type = XFS_ALLOCTYPE_FIRST_AG;
+		else
+			args.type = XFS_ALLOCTYPE_START_BNO;
+		args.total = args.minlen = ap->minlen;
+	} else {
+		args.type = XFS_ALLOCTYPE_NEAR_BNO;
+		args.total = ap->total;
+		args.minlen = ap->minlen;
+	}
+
+	/*
+	 * If we are not low on available data blocks, and the underlying
+	 * logical volume manager is a stripe, and the file offset is zero then
+	 * try to allocate data blocks on stripe unit boundary. NOTE: ap->aeof
+	 * is only set if the allocation length is >= the stripe unit and the
+	 * allocation offset is at the end of file.
+	 */
+	if (!(ap->tp->t_flags & XFS_TRANS_LOWMODE) && ap->aeof) {
+		if (!ap->offset) {
+			args.alignment = stripe_align;
+			atype = args.type;
+			isaligned = 1;
+			/*
+			 * Adjust minlen to try and preserve alignment if we
+			 * can't guarantee an aligned maxlen extent.
+			 */
+			if (blen > args.alignment &&
+			    blen <= args.maxlen + args.alignment)
+				args.minlen = blen - args.alignment;
+			args.minalignslop = 0;
+		} else {
+			/*
+			 * First try an exact bno allocation.
+			 * If it fails then do a near or start bno
+			 * allocation with alignment turned on.
+			 */
+			atype = args.type;
+			tryagain = 1;
+			args.type = XFS_ALLOCTYPE_THIS_BNO;
+			args.alignment = 1;
+			/*
+			 * Compute the minlen+alignment for the
+			 * next case.  Set slop so that the value
+			 * of minlen+alignment+slop doesn't go up
+			 * between the calls.
+			 */
+			if (blen > stripe_align && blen <= args.maxlen)
+				nextminlen = blen - stripe_align;
+			else
+				nextminlen = args.minlen;
+			if (nextminlen + stripe_align > args.minlen + 1)
+				args.minalignslop =
+					nextminlen + stripe_align -
+					args.minlen - 1;
+			else
+				args.minalignslop = 0;
+		}
+	} else {
+		args.alignment = 1;
+		args.minalignslop = 0;
+	}
+	args.minleft = ap->minleft;
+	args.wasdel = ap->wasdel;
+	args.resv = XFS_AG_RESV_NONE;
+	args.datatype = ap->datatype;
+
+	error = xfs_alloc_vextent(&args);
+	if (error)
+		return error;
+
+	if (tryagain && args.fsbno == NULLFSBLOCK) {
+		/*
+		 * Exact allocation failed. Now try with alignment
+		 * turned on.
+		 */
+		args.type = atype;
+		args.fsbno = ap->blkno;
+		args.alignment = stripe_align;
+		args.minlen = nextminlen;
+		args.minalignslop = 0;
+		isaligned = 1;
+		if ((error = xfs_alloc_vextent(&args)))
+			return error;
+	}
+	if (isaligned && args.fsbno == NULLFSBLOCK) {
+		/*
+		 * allocation failed, so turn off alignment and
+		 * try again.
+		 */
+		args.type = atype;
+		args.fsbno = ap->blkno;
+		args.alignment = 0;
+		if ((error = xfs_alloc_vextent(&args)))
+			return error;
+	}
+	if (args.fsbno == NULLFSBLOCK && nullfb &&
+	    args.minlen > ap->minlen) {
+		args.minlen = ap->minlen;
+		args.type = XFS_ALLOCTYPE_START_BNO;
+		args.fsbno = ap->blkno;
+		if ((error = xfs_alloc_vextent(&args)))
+			return error;
+	}
+	if (args.fsbno == NULLFSBLOCK && nullfb) {
+		args.fsbno = 0;
+		args.type = XFS_ALLOCTYPE_FIRST_AG;
+		args.total = ap->minlen;
+		if ((error = xfs_alloc_vextent(&args)))
+			return error;
+		ap->tp->t_flags |= XFS_TRANS_LOWMODE;
+	}
+
+	if (args.fsbno != NULLFSBLOCK) {
+		xfs_bmap_process_allocated_extent(ap, &args, orig_offset,
+			orig_length);
+	} else {
+		ap->blkno = NULLFSBLOCK;
+		ap->length = 0;
+	}
+	return 0;
+}
+
+/* Trim extent to fit a logical block range. */
+void
+xfs_trim_extent(
+	struct xfs_bmbt_irec	*irec,
+	xfs_fileoff_t		bno,
+	xfs_filblks_t		len)
+{
+	xfs_fileoff_t		distance;
+	xfs_fileoff_t		end = bno + len;
+
+	if (irec->br_startoff + irec->br_blockcount <= bno ||
+	    irec->br_startoff >= end) {
+		irec->br_blockcount = 0;
+		return;
+	}
+
+	if (irec->br_startoff < bno) {
+		distance = bno - irec->br_startoff;
+		if (isnullstartblock(irec->br_startblock))
+			irec->br_startblock = DELAYSTARTBLOCK;
+		if (irec->br_startblock != DELAYSTARTBLOCK &&
+		    irec->br_startblock != HOLESTARTBLOCK)
+			irec->br_startblock += distance;
+		irec->br_startoff += distance;
+		irec->br_blockcount -= distance;
+	}
+
+	if (end < irec->br_startoff + irec->br_blockcount) {
+		distance = irec->br_startoff + irec->br_blockcount - end;
+		irec->br_blockcount -= distance;
+	}
+}
+
+/*
+ * Trim the returned map to the required bounds
+ */
+STATIC void
+xfs_bmapi_trim_map(
+	struct xfs_bmbt_irec	*mval,
+	struct xfs_bmbt_irec	*got,
+	xfs_fileoff_t		*bno,
+	xfs_filblks_t		len,
+	xfs_fileoff_t		obno,
+	xfs_fileoff_t		end,
+	int			n,
+	uint32_t		flags)
+{
+	if ((flags & XFS_BMAPI_ENTIRE) ||
+	    got->br_startoff + got->br_blockcount <= obno) {
+		*mval = *got;
+		if (isnullstartblock(got->br_startblock))
+			mval->br_startblock = DELAYSTARTBLOCK;
+		return;
+	}
+
+	if (obno > *bno)
+		*bno = obno;
+	ASSERT((*bno >= obno) || (n == 0));
+	ASSERT(*bno < end);
+	mval->br_startoff = *bno;
+	if (isnullstartblock(got->br_startblock))
+		mval->br_startblock = DELAYSTARTBLOCK;
+	else
+		mval->br_startblock = got->br_startblock +
+					(*bno - got->br_startoff);
+	/*
+	 * Return the minimum of what we got and what we asked for for
+	 * the length.  We can use the len variable here because it is
+	 * modified below and we could have been there before coming
+	 * here if the first part of the allocation didn't overlap what
+	 * was asked for.
+	 */
+	mval->br_blockcount = XFS_FILBLKS_MIN(end - *bno,
+			got->br_blockcount - (*bno - got->br_startoff));
+	mval->br_state = got->br_state;
+	ASSERT(mval->br_blockcount <= len);
+	return;
+}
+
+/*
+ * Update and validate the extent map to return
+ */
+STATIC void
+xfs_bmapi_update_map(
+	struct xfs_bmbt_irec	**map,
+	xfs_fileoff_t		*bno,
+	xfs_filblks_t		*len,
+	xfs_fileoff_t		obno,
+	xfs_fileoff_t		end,
+	int			*n,
+	uint32_t		flags)
+{
+	xfs_bmbt_irec_t	*mval = *map;
+
+	ASSERT((flags & XFS_BMAPI_ENTIRE) ||
+	       ((mval->br_startoff + mval->br_blockcount) <= end));
+	ASSERT((flags & XFS_BMAPI_ENTIRE) || (mval->br_blockcount <= *len) ||
+	       (mval->br_startoff < obno));
+
+	*bno = mval->br_startoff + mval->br_blockcount;
+	*len = end - *bno;
+	if (*n > 0 && mval->br_startoff == mval[-1].br_startoff) {
+		/* update previous map with new information */
+		ASSERT(mval->br_startblock == mval[-1].br_startblock);
+		ASSERT(mval->br_blockcount > mval[-1].br_blockcount);
+		ASSERT(mval->br_state == mval[-1].br_state);
+		mval[-1].br_blockcount = mval->br_blockcount;
+		mval[-1].br_state = mval->br_state;
+	} else if (*n > 0 && mval->br_startblock != DELAYSTARTBLOCK &&
+		   mval[-1].br_startblock != DELAYSTARTBLOCK &&
+		   mval[-1].br_startblock != HOLESTARTBLOCK &&
+		   mval->br_startblock == mval[-1].br_startblock +
+					  mval[-1].br_blockcount &&
+		   mval[-1].br_state == mval->br_state) {
+		ASSERT(mval->br_startoff ==
+		       mval[-1].br_startoff + mval[-1].br_blockcount);
+		mval[-1].br_blockcount += mval->br_blockcount;
+	} else if (*n > 0 &&
+		   mval->br_startblock == DELAYSTARTBLOCK &&
+		   mval[-1].br_startblock == DELAYSTARTBLOCK &&
+		   mval->br_startoff ==
+		   mval[-1].br_startoff + mval[-1].br_blockcount) {
+		mval[-1].br_blockcount += mval->br_blockcount;
+		mval[-1].br_state = mval->br_state;
+	} else if (!((*n == 0) &&
+		     ((mval->br_startoff + mval->br_blockcount) <=
+		      obno))) {
+		mval++;
+		(*n)++;
+	}
+	*map = mval;
+}
+
+/*
+ * Map file blocks to filesystem blocks without allocation.
+ */
+int
+xfs_bmapi_read(
+	struct xfs_inode	*ip,
+	xfs_fileoff_t		bno,
+	xfs_filblks_t		len,
+	struct xfs_bmbt_irec	*mval,
+	int			*nmap,
+	uint32_t		flags)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	int			whichfork = xfs_bmapi_whichfork(flags);
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_bmbt_irec	got;
+	xfs_fileoff_t		obno;
+	xfs_fileoff_t		end;
+	struct xfs_iext_cursor	icur;
+	int			error;
+	bool			eof = false;
+	int			n = 0;
+
+	ASSERT(*nmap >= 1);
+	ASSERT(!(flags & ~(XFS_BMAPI_ATTRFORK | XFS_BMAPI_ENTIRE)));
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_SHARED|XFS_ILOCK_EXCL));
+
+	if (WARN_ON_ONCE(!ifp))
+		return -EFSCORRUPTED;
+
+	if (XFS_IS_CORRUPT(mp, !xfs_ifork_has_extents(ifp)) ||
+	    XFS_TEST_ERROR(false, mp, XFS_ERRTAG_BMAPIFORMAT))
+		return -EFSCORRUPTED;
+
+	if (xfs_is_shutdown(mp))
+		return -EIO;
+
+	XFS_STATS_INC(mp, xs_blk_mapr);
+
+	error = xfs_iread_extents(NULL, ip, whichfork);
+	if (error)
+		return error;
+
+	if (!xfs_iext_lookup_extent(ip, ifp, bno, &icur, &got))
+		eof = true;
+	end = bno + len;
+	obno = bno;
+
+	while (bno < end && n < *nmap) {
+		/* Reading past eof, act as though there's a hole up to end. */
+		if (eof)
+			got.br_startoff = end;
+		if (got.br_startoff > bno) {
+			/* Reading in a hole.  */
+			mval->br_startoff = bno;
+			mval->br_startblock = HOLESTARTBLOCK;
+			mval->br_blockcount =
+				XFS_FILBLKS_MIN(len, got.br_startoff - bno);
+			mval->br_state = XFS_EXT_NORM;
+			bno += mval->br_blockcount;
+			len -= mval->br_blockcount;
+			mval++;
+			n++;
+			continue;
+		}
+
+		/* set up the extent map to return. */
+		xfs_bmapi_trim_map(mval, &got, &bno, len, obno, end, n, flags);
+		xfs_bmapi_update_map(&mval, &bno, &len, obno, end, &n, flags);
+
+		/* If we're done, stop now. */
+		if (bno >= end || n >= *nmap)
+			break;
+
+		/* Else go on to the next record. */
+		if (!xfs_iext_next_extent(ifp, &icur, &got))
+			eof = true;
+	}
+	*nmap = n;
+	return 0;
+}
+
+/*
+ * Add a delayed allocation extent to an inode. Blocks are reserved from the
+ * global pool and the extent inserted into the inode in-core extent tree.
+ *
+ * On entry, got refers to the first extent beyond the offset of the extent to
+ * allocate or eof is specified if no such extent exists. On return, got refers
+ * to the extent record that was inserted to the inode fork.
+ *
+ * Note that the allocated extent may have been merged with contiguous extents
+ * during insertion into the inode fork. Thus, got does not reflect the current
+ * state of the inode fork on return. If necessary, the caller can use lastx to
+ * look up the updated record in the inode fork.
+ */
+int
+xfs_bmapi_reserve_delalloc(
+	struct xfs_inode	*ip,
+	int			whichfork,
+	xfs_fileoff_t		off,
+	xfs_filblks_t		len,
+	xfs_filblks_t		prealloc,
+	struct xfs_bmbt_irec	*got,
+	struct xfs_iext_cursor	*icur,
+	int			eof)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	xfs_extlen_t		alen;
+	xfs_extlen_t		indlen;
+	int			error;
+	xfs_fileoff_t		aoff = off;
+
+	/*
+	 * Cap the alloc length. Keep track of prealloc so we know whether to
+	 * tag the inode before we return.
+	 */
+	alen = XFS_FILBLKS_MIN(len + prealloc, XFS_MAX_BMBT_EXTLEN);
+	if (!eof)
+		alen = XFS_FILBLKS_MIN(alen, got->br_startoff - aoff);
+	if (prealloc && alen >= len)
+		prealloc = alen - len;
+
+	/* Figure out the extent size, adjust alen */
+	if (whichfork == XFS_COW_FORK) {
+		struct xfs_bmbt_irec	prev;
+		xfs_extlen_t		extsz = xfs_get_cowextsz_hint(ip);
+
+		if (!xfs_iext_peek_prev_extent(ifp, icur, &prev))
+			prev.br_startoff = NULLFILEOFF;
+
+		error = xfs_bmap_extsize_align(mp, got, &prev, extsz, 0, eof,
+					       1, 0, &aoff, &alen);
+		ASSERT(!error);
+	}
+
+	/*
+	 * Make a transaction-less quota reservation for delayed allocation
+	 * blocks.  This number gets adjusted later.  We return if we haven't
+	 * allocated blocks already inside this loop.
+	 */
+	error = xfs_quota_reserve_blkres(ip, alen);
+	if (error)
+		return error;
+
+	/*
+	 * Split changing sb for alen and indlen since they could be coming
+	 * from different places.
+	 */
+	indlen = (xfs_extlen_t)xfs_bmap_worst_indlen(ip, alen);
+	ASSERT(indlen > 0);
+
+	error = xfs_mod_fdblocks(mp, -((int64_t)alen), false);
+	if (error)
+		goto out_unreserve_quota;
+
+	error = xfs_mod_fdblocks(mp, -((int64_t)indlen), false);
+	if (error)
+		goto out_unreserve_blocks;
+
+
+	ip->i_delayed_blks += alen;
+	xfs_mod_delalloc(ip->i_mount, alen + indlen);
+
+	got->br_startoff = aoff;
+	got->br_startblock = nullstartblock(indlen);
+	got->br_blockcount = alen;
+	got->br_state = XFS_EXT_NORM;
+
+	xfs_bmap_add_extent_hole_delay(ip, whichfork, icur, got);
+
+	/*
+	 * Tag the inode if blocks were preallocated. Note that COW fork
+	 * preallocation can occur at the start or end of the extent, even when
+	 * prealloc == 0, so we must also check the aligned offset and length.
+	 */
+	if (whichfork == XFS_DATA_FORK && prealloc)
+		xfs_inode_set_eofblocks_tag(ip);
+	if (whichfork == XFS_COW_FORK && (prealloc || aoff < off || alen > len))
+		xfs_inode_set_cowblocks_tag(ip);
+
+	return 0;
+
+out_unreserve_blocks:
+	xfs_mod_fdblocks(mp, alen, false);
+out_unreserve_quota:
+	if (XFS_IS_QUOTA_ON(mp))
+		xfs_quota_unreserve_blkres(ip, alen);
+	return error;
+}
+
+static int
+xfs_bmap_alloc_userdata(
+	struct xfs_bmalloca	*bma)
+{
+	struct xfs_mount	*mp = bma->ip->i_mount;
+	int			whichfork = xfs_bmapi_whichfork(bma->flags);
+	int			error;
+
+	/*
+	 * Set the data type being allocated. For the data fork, the first data
+	 * in the file is treated differently to all other allocations. For the
+	 * attribute fork, we only need to ensure the allocated range is not on
+	 * the busy list.
+	 */
+	bma->datatype = XFS_ALLOC_NOBUSY;
+	if (whichfork == XFS_DATA_FORK) {
+		bma->datatype |= XFS_ALLOC_USERDATA;
+		if (bma->offset == 0)
+			bma->datatype |= XFS_ALLOC_INITIAL_USER_DATA;
+
+		if (mp->m_dalign && bma->length >= mp->m_dalign) {
+			error = xfs_bmap_isaeof(bma, whichfork);
+			if (error)
+				return error;
+		}
+
+		if (XFS_IS_REALTIME_INODE(bma->ip))
+			return xfs_bmap_rtalloc(bma);
+	}
+
+	if (unlikely(XFS_TEST_ERROR(false, mp,
+			XFS_ERRTAG_BMAP_ALLOC_MINLEN_EXTENT)))
+		return xfs_bmap_exact_minlen_extent_alloc(bma);
+
+	return xfs_bmap_btalloc(bma);
+}
+
+static int
+xfs_bmapi_allocate(
+	struct xfs_bmalloca	*bma)
+{
+	struct xfs_mount	*mp = bma->ip->i_mount;
+	int			whichfork = xfs_bmapi_whichfork(bma->flags);
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(bma->ip, whichfork);
+	int			tmp_logflags = 0;
+	int			error;
+
+	ASSERT(bma->length > 0);
+
+	/*
+	 * For the wasdelay case, we could also just allocate the stuff asked
+	 * for in this bmap call but that wouldn't be as good.
+	 */
+	if (bma->wasdel) {
+		bma->length = (xfs_extlen_t)bma->got.br_blockcount;
+		bma->offset = bma->got.br_startoff;
+		if (!xfs_iext_peek_prev_extent(ifp, &bma->icur, &bma->prev))
+			bma->prev.br_startoff = NULLFILEOFF;
+	} else {
+		bma->length = XFS_FILBLKS_MIN(bma->length, XFS_MAX_BMBT_EXTLEN);
+		if (!bma->eof)
+			bma->length = XFS_FILBLKS_MIN(bma->length,
+					bma->got.br_startoff - bma->offset);
+	}
+
+	if (bma->flags & XFS_BMAPI_CONTIG)
+		bma->minlen = bma->length;
+	else
+		bma->minlen = 1;
+
+	if (bma->flags & XFS_BMAPI_METADATA) {
+		if (unlikely(XFS_TEST_ERROR(false, mp,
+				XFS_ERRTAG_BMAP_ALLOC_MINLEN_EXTENT)))
+			error = xfs_bmap_exact_minlen_extent_alloc(bma);
+		else
+			error = xfs_bmap_btalloc(bma);
+	} else {
+		error = xfs_bmap_alloc_userdata(bma);
+	}
+	if (error || bma->blkno == NULLFSBLOCK)
+		return error;
+
+	if (bma->flags & XFS_BMAPI_ZERO) {
+		error = xfs_zero_extent(bma->ip, bma->blkno, bma->length);
+		if (error)
+			return error;
+	}
+
+	if (ifp->if_format == XFS_DINODE_FMT_BTREE && !bma->cur)
+		bma->cur = xfs_bmbt_init_cursor(mp, bma->tp, bma->ip, whichfork);
+	/*
+	 * Bump the number of extents we've allocated
+	 * in this call.
+	 */
+	bma->nallocs++;
+
+	if (bma->cur)
+		bma->cur->bc_ino.flags =
+			bma->wasdel ? XFS_BTCUR_BMBT_WASDEL : 0;
+
+	bma->got.br_startoff = bma->offset;
+	bma->got.br_startblock = bma->blkno;
+	bma->got.br_blockcount = bma->length;
+	bma->got.br_state = XFS_EXT_NORM;
+
+	if (bma->flags & XFS_BMAPI_PREALLOC)
+		bma->got.br_state = XFS_EXT_UNWRITTEN;
+
+	if (bma->wasdel)
+		error = xfs_bmap_add_extent_delay_real(bma, whichfork);
+	else
+		error = xfs_bmap_add_extent_hole_real(bma->tp, bma->ip,
+				whichfork, &bma->icur, &bma->cur, &bma->got,
+				&bma->logflags, bma->flags);
+
+	bma->logflags |= tmp_logflags;
+	if (error)
+		return error;
+
+	/*
+	 * Update our extent pointer, given that xfs_bmap_add_extent_delay_real
+	 * or xfs_bmap_add_extent_hole_real might have merged it into one of
+	 * the neighbouring ones.
+	 */
+	xfs_iext_get_extent(ifp, &bma->icur, &bma->got);
+
+	ASSERT(bma->got.br_startoff <= bma->offset);
+	ASSERT(bma->got.br_startoff + bma->got.br_blockcount >=
+	       bma->offset + bma->length);
+	ASSERT(bma->got.br_state == XFS_EXT_NORM ||
+	       bma->got.br_state == XFS_EXT_UNWRITTEN);
+	return 0;
+}
+
+STATIC int
+xfs_bmapi_convert_unwritten(
+	struct xfs_bmalloca	*bma,
+	struct xfs_bmbt_irec	*mval,
+	xfs_filblks_t		len,
+	uint32_t		flags)
+{
+	int			whichfork = xfs_bmapi_whichfork(flags);
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(bma->ip, whichfork);
+	int			tmp_logflags = 0;
+	int			error;
+
+	/* check if we need to do unwritten->real conversion */
+	if (mval->br_state == XFS_EXT_UNWRITTEN &&
+	    (flags & XFS_BMAPI_PREALLOC))
+		return 0;
+
+	/* check if we need to do real->unwritten conversion */
+	if (mval->br_state == XFS_EXT_NORM &&
+	    (flags & (XFS_BMAPI_PREALLOC | XFS_BMAPI_CONVERT)) !=
+			(XFS_BMAPI_PREALLOC | XFS_BMAPI_CONVERT))
+		return 0;
+
+	/*
+	 * Modify (by adding) the state flag, if writing.
+	 */
+	ASSERT(mval->br_blockcount <= len);
+	if (ifp->if_format == XFS_DINODE_FMT_BTREE && !bma->cur) {
+		bma->cur = xfs_bmbt_init_cursor(bma->ip->i_mount, bma->tp,
+					bma->ip, whichfork);
+	}
+	mval->br_state = (mval->br_state == XFS_EXT_UNWRITTEN)
+				? XFS_EXT_NORM : XFS_EXT_UNWRITTEN;
+
+	/*
+	 * Before insertion into the bmbt, zero the range being converted
+	 * if required.
+	 */
+	if (flags & XFS_BMAPI_ZERO) {
+		error = xfs_zero_extent(bma->ip, mval->br_startblock,
+					mval->br_blockcount);
+		if (error)
+			return error;
+	}
+
+	error = xfs_bmap_add_extent_unwritten_real(bma->tp, bma->ip, whichfork,
+			&bma->icur, &bma->cur, mval, &tmp_logflags);
+	/*
+	 * Log the inode core unconditionally in the unwritten extent conversion
+	 * path because the conversion might not have done so (e.g., if the
+	 * extent count hasn't changed). We need to make sure the inode is dirty
+	 * in the transaction for the sake of fsync(), even if nothing has
+	 * changed, because fsync() will not force the log for this transaction
+	 * unless it sees the inode pinned.
+	 *
+	 * Note: If we're only converting cow fork extents, there aren't
+	 * any on-disk updates to make, so we don't need to log anything.
+	 */
+	if (whichfork != XFS_COW_FORK)
+		bma->logflags |= tmp_logflags | XFS_ILOG_CORE;
+	if (error)
+		return error;
+
+	/*
+	 * Update our extent pointer, given that
+	 * xfs_bmap_add_extent_unwritten_real might have merged it into one
+	 * of the neighbouring ones.
+	 */
+	xfs_iext_get_extent(ifp, &bma->icur, &bma->got);
+
+	/*
+	 * We may have combined previously unwritten space with written space,
+	 * so generate another request.
+	 */
+	if (mval->br_blockcount < len)
+		return -EAGAIN;
+	return 0;
+}
+
+static inline xfs_extlen_t
+xfs_bmapi_minleft(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	int			fork)
+{
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, fork);
+
+	if (tp && tp->t_firstblock != NULLFSBLOCK)
+		return 0;
+	if (ifp->if_format != XFS_DINODE_FMT_BTREE)
+		return 1;
+	return be16_to_cpu(ifp->if_broot->bb_level) + 1;
+}
+
+/*
+ * Log whatever the flags say, even if error.  Otherwise we might miss detecting
+ * a case where the data is changed, there's an error, and it's not logged so we
+ * don't shutdown when we should.  Don't bother logging extents/btree changes if
+ * we converted to the other format.
+ */
+static void
+xfs_bmapi_finish(
+	struct xfs_bmalloca	*bma,
+	int			whichfork,
+	int			error)
+{
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(bma->ip, whichfork);
+
+	if ((bma->logflags & xfs_ilog_fext(whichfork)) &&
+	    ifp->if_format != XFS_DINODE_FMT_EXTENTS)
+		bma->logflags &= ~xfs_ilog_fext(whichfork);
+	else if ((bma->logflags & xfs_ilog_fbroot(whichfork)) &&
+		 ifp->if_format != XFS_DINODE_FMT_BTREE)
+		bma->logflags &= ~xfs_ilog_fbroot(whichfork);
+
+	if (bma->logflags)
+		xfs_trans_log_inode(bma->tp, bma->ip, bma->logflags);
+	if (bma->cur)
+		xfs_btree_del_cursor(bma->cur, error);
+}
+
+/*
+ * Map file blocks to filesystem blocks, and allocate blocks or convert the
+ * extent state if necessary.  Details behaviour is controlled by the flags
+ * parameter.  Only allocates blocks from a single allocation group, to avoid
+ * locking problems.
+ */
+int
+xfs_bmapi_write(
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* incore inode */
+	xfs_fileoff_t		bno,		/* starting file offs. mapped */
+	xfs_filblks_t		len,		/* length to map in file */
+	uint32_t		flags,		/* XFS_BMAPI_... */
+	xfs_extlen_t		total,		/* total blocks needed */
+	struct xfs_bmbt_irec	*mval,		/* output: map values */
+	int			*nmap)		/* i/o: mval size/count */
+{
+	struct xfs_bmalloca	bma = {
+		.tp		= tp,
+		.ip		= ip,
+		.total		= total,
+	};
+	struct xfs_mount	*mp = ip->i_mount;
+	int			whichfork = xfs_bmapi_whichfork(flags);
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	xfs_fileoff_t		end;		/* end of mapped file region */
+	bool			eof = false;	/* after the end of extents */
+	int			error;		/* error return */
+	int			n;		/* current extent index */
+	xfs_fileoff_t		obno;		/* old block number (offset) */
+
+#ifdef DEBUG
+	xfs_fileoff_t		orig_bno;	/* original block number value */
+	int			orig_flags;	/* original flags arg value */
+	xfs_filblks_t		orig_len;	/* original value of len arg */
+	struct xfs_bmbt_irec	*orig_mval;	/* original value of mval */
+	int			orig_nmap;	/* original value of *nmap */
+
+	orig_bno = bno;
+	orig_len = len;
+	orig_flags = flags;
+	orig_mval = mval;
+	orig_nmap = *nmap;
+#endif
+
+	ASSERT(*nmap >= 1);
+	ASSERT(*nmap <= XFS_BMAP_MAX_NMAP);
+	ASSERT(tp != NULL);
+	ASSERT(len > 0);
+	ASSERT(ifp->if_format != XFS_DINODE_FMT_LOCAL);
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+	ASSERT(!(flags & XFS_BMAPI_REMAP));
+
+	/* zeroing is for currently only for data extents, not metadata */
+	ASSERT((flags & (XFS_BMAPI_METADATA | XFS_BMAPI_ZERO)) !=
+			(XFS_BMAPI_METADATA | XFS_BMAPI_ZERO));
+	/*
+	 * we can allocate unwritten extents or pre-zero allocated blocks,
+	 * but it makes no sense to do both at once. This would result in
+	 * zeroing the unwritten extent twice, but it still being an
+	 * unwritten extent....
+	 */
+	ASSERT((flags & (XFS_BMAPI_PREALLOC | XFS_BMAPI_ZERO)) !=
+			(XFS_BMAPI_PREALLOC | XFS_BMAPI_ZERO));
+
+	if (XFS_IS_CORRUPT(mp, !xfs_ifork_has_extents(ifp)) ||
+	    XFS_TEST_ERROR(false, mp, XFS_ERRTAG_BMAPIFORMAT)) {
+		return -EFSCORRUPTED;
+	}
+
+	if (xfs_is_shutdown(mp))
+		return -EIO;
+
+	XFS_STATS_INC(mp, xs_blk_mapw);
+
+	error = xfs_iread_extents(tp, ip, whichfork);
+	if (error)
+		goto error0;
+
+	if (!xfs_iext_lookup_extent(ip, ifp, bno, &bma.icur, &bma.got))
+		eof = true;
+	if (!xfs_iext_peek_prev_extent(ifp, &bma.icur, &bma.prev))
+		bma.prev.br_startoff = NULLFILEOFF;
+	bma.minleft = xfs_bmapi_minleft(tp, ip, whichfork);
+
+	n = 0;
+	end = bno + len;
+	obno = bno;
+	while (bno < end && n < *nmap) {
+		bool			need_alloc = false, wasdelay = false;
+
+		/* in hole or beyond EOF? */
+		if (eof || bma.got.br_startoff > bno) {
+			/*
+			 * CoW fork conversions should /never/ hit EOF or
+			 * holes.  There should always be something for us
+			 * to work on.
+			 */
+			ASSERT(!((flags & XFS_BMAPI_CONVERT) &&
+			         (flags & XFS_BMAPI_COWFORK)));
+
+			need_alloc = true;
+		} else if (isnullstartblock(bma.got.br_startblock)) {
+			wasdelay = true;
+		}
+
+		/*
+		 * First, deal with the hole before the allocated space
+		 * that we found, if any.
+		 */
+		if (need_alloc || wasdelay) {
+			bma.eof = eof;
+			bma.conv = !!(flags & XFS_BMAPI_CONVERT);
+			bma.wasdel = wasdelay;
+			bma.offset = bno;
+			bma.flags = flags;
+
+			/*
+			 * There's a 32/64 bit type mismatch between the
+			 * allocation length request (which can be 64 bits in
+			 * length) and the bma length request, which is
+			 * xfs_extlen_t and therefore 32 bits. Hence we have to
+			 * check for 32-bit overflows and handle them here.
+			 */
+			if (len > (xfs_filblks_t)XFS_MAX_BMBT_EXTLEN)
+				bma.length = XFS_MAX_BMBT_EXTLEN;
+			else
+				bma.length = len;
+
+			ASSERT(len > 0);
+			ASSERT(bma.length > 0);
+			error = xfs_bmapi_allocate(&bma);
+			if (error)
+				goto error0;
+			if (bma.blkno == NULLFSBLOCK)
+				break;
+
+			/*
+			 * If this is a CoW allocation, record the data in
+			 * the refcount btree for orphan recovery.
+			 */
+			if (whichfork == XFS_COW_FORK)
+				xfs_refcount_alloc_cow_extent(tp, bma.blkno,
+						bma.length);
+		}
+
+		/* Deal with the allocated space we found.  */
+		xfs_bmapi_trim_map(mval, &bma.got, &bno, len, obno,
+							end, n, flags);
+
+		/* Execute unwritten extent conversion if necessary */
+		error = xfs_bmapi_convert_unwritten(&bma, mval, len, flags);
+		if (error == -EAGAIN)
+			continue;
+		if (error)
+			goto error0;
+
+		/* update the extent map to return */
+		xfs_bmapi_update_map(&mval, &bno, &len, obno, end, &n, flags);
+
+		/*
+		 * If we're done, stop now.  Stop when we've allocated
+		 * XFS_BMAP_MAX_NMAP extents no matter what.  Otherwise
+		 * the transaction may get too big.
+		 */
+		if (bno >= end || n >= *nmap || bma.nallocs >= *nmap)
+			break;
+
+		/* Else go on to the next record. */
+		bma.prev = bma.got;
+		if (!xfs_iext_next_extent(ifp, &bma.icur, &bma.got))
+			eof = true;
+	}
+	*nmap = n;
+
+	error = xfs_bmap_btree_to_extents(tp, ip, bma.cur, &bma.logflags,
+			whichfork);
+	if (error)
+		goto error0;
+
+	ASSERT(ifp->if_format != XFS_DINODE_FMT_BTREE ||
+	       ifp->if_nextents > XFS_IFORK_MAXEXT(ip, whichfork));
+	xfs_bmapi_finish(&bma, whichfork, 0);
+	xfs_bmap_validate_ret(orig_bno, orig_len, orig_flags, orig_mval,
+		orig_nmap, *nmap);
+	return 0;
+error0:
+	xfs_bmapi_finish(&bma, whichfork, error);
+	return error;
+}
+
+/*
+ * Convert an existing delalloc extent to real blocks based on file offset. This
+ * attempts to allocate the entire delalloc extent and may require multiple
+ * invocations to allocate the target offset if a large enough physical extent
+ * is not available.
+ */
+int
+xfs_bmapi_convert_delalloc(
+	struct xfs_inode	*ip,
+	int			whichfork,
+	xfs_off_t		offset,
+	struct iomap		*iomap,
+	unsigned int		*seq)
+{
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_mount	*mp = ip->i_mount;
+	xfs_fileoff_t		offset_fsb = XFS_B_TO_FSBT(mp, offset);
+	struct xfs_bmalloca	bma = { NULL };
+	uint16_t		flags = 0;
+	struct xfs_trans	*tp;
+	int			error;
+
+	if (whichfork == XFS_COW_FORK)
+		flags |= IOMAP_F_SHARED;
+
+	/*
+	 * Space for the extent and indirect blocks was reserved when the
+	 * delalloc extent was created so there's no need to do so here.
+	 */
+	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_write, 0, 0,
+				XFS_TRANS_RESERVE, &tp);
+	if (error)
+		return error;
+
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	xfs_trans_ijoin(tp, ip, 0);
+
+	error = xfs_iext_count_may_overflow(ip, whichfork,
+			XFS_IEXT_ADD_NOSPLIT_CNT);
+	if (error == -EFBIG)
+		error = xfs_iext_count_upgrade(tp, ip,
+				XFS_IEXT_ADD_NOSPLIT_CNT);
+	if (error)
+		goto out_trans_cancel;
+
+	if (!xfs_iext_lookup_extent(ip, ifp, offset_fsb, &bma.icur, &bma.got) ||
+	    bma.got.br_startoff > offset_fsb) {
+		/*
+		 * No extent found in the range we are trying to convert.  This
+		 * should only happen for the COW fork, where another thread
+		 * might have moved the extent to the data fork in the meantime.
+		 */
+		WARN_ON_ONCE(whichfork != XFS_COW_FORK);
+		error = -EAGAIN;
+		goto out_trans_cancel;
+	}
+
+	/*
+	 * If we find a real extent here we raced with another thread converting
+	 * the extent.  Just return the real extent at this offset.
+	 */
+	if (!isnullstartblock(bma.got.br_startblock)) {
+		xfs_bmbt_to_iomap(ip, iomap, &bma.got, 0, flags);
+		*seq = READ_ONCE(ifp->if_seq);
+		goto out_trans_cancel;
+	}
+
+	bma.tp = tp;
+	bma.ip = ip;
+	bma.wasdel = true;
+	bma.offset = bma.got.br_startoff;
+	bma.length = max_t(xfs_filblks_t, bma.got.br_blockcount,
+			XFS_MAX_BMBT_EXTLEN);
+	bma.minleft = xfs_bmapi_minleft(tp, ip, whichfork);
+
+	/*
+	 * When we're converting the delalloc reservations backing dirty pages
+	 * in the page cache, we must be careful about how we create the new
+	 * extents:
+	 *
+	 * New CoW fork extents are created unwritten, turned into real extents
+	 * when we're about to write the data to disk, and mapped into the data
+	 * fork after the write finishes.  End of story.
+	 *
+	 * New data fork extents must be mapped in as unwritten and converted
+	 * to real extents after the write succeeds to avoid exposing stale
+	 * disk contents if we crash.
+	 */
+	bma.flags = XFS_BMAPI_PREALLOC;
+	if (whichfork == XFS_COW_FORK)
+		bma.flags |= XFS_BMAPI_COWFORK;
+
+	if (!xfs_iext_peek_prev_extent(ifp, &bma.icur, &bma.prev))
+		bma.prev.br_startoff = NULLFILEOFF;
+
+	error = xfs_bmapi_allocate(&bma);
+	if (error)
+		goto out_finish;
+
+	error = -ENOSPC;
+	if (WARN_ON_ONCE(bma.blkno == NULLFSBLOCK))
+		goto out_finish;
+	error = -EFSCORRUPTED;
+	if (WARN_ON_ONCE(!xfs_valid_startblock(ip, bma.got.br_startblock)))
+		goto out_finish;
+
+	XFS_STATS_ADD(mp, xs_xstrat_bytes, XFS_FSB_TO_B(mp, bma.length));
+	XFS_STATS_INC(mp, xs_xstrat_quick);
+
+	ASSERT(!isnullstartblock(bma.got.br_startblock));
+	xfs_bmbt_to_iomap(ip, iomap, &bma.got, 0, flags);
+	*seq = READ_ONCE(ifp->if_seq);
+
+	if (whichfork == XFS_COW_FORK)
+		xfs_refcount_alloc_cow_extent(tp, bma.blkno, bma.length);
+
+	error = xfs_bmap_btree_to_extents(tp, ip, bma.cur, &bma.logflags,
+			whichfork);
+	if (error)
+		goto out_finish;
+
+	xfs_bmapi_finish(&bma, whichfork, 0);
+	error = xfs_trans_commit(tp);
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	return error;
+
+out_finish:
+	xfs_bmapi_finish(&bma, whichfork, error);
+out_trans_cancel:
+	xfs_trans_cancel(tp);
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	return error;
+}
+
+int
+xfs_bmapi_remap(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	xfs_fileoff_t		bno,
+	xfs_filblks_t		len,
+	xfs_fsblock_t		startblock,
+	uint32_t		flags)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_ifork	*ifp;
+	struct xfs_btree_cur	*cur = NULL;
+	struct xfs_bmbt_irec	got;
+	struct xfs_iext_cursor	icur;
+	int			whichfork = xfs_bmapi_whichfork(flags);
+	int			logflags = 0, error;
+
+	ifp = xfs_ifork_ptr(ip, whichfork);
+	ASSERT(len > 0);
+	ASSERT(len <= (xfs_filblks_t)XFS_MAX_BMBT_EXTLEN);
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+	ASSERT(!(flags & ~(XFS_BMAPI_ATTRFORK | XFS_BMAPI_PREALLOC |
+			   XFS_BMAPI_NORMAP)));
+	ASSERT((flags & (XFS_BMAPI_ATTRFORK | XFS_BMAPI_PREALLOC)) !=
+			(XFS_BMAPI_ATTRFORK | XFS_BMAPI_PREALLOC));
+
+	if (XFS_IS_CORRUPT(mp, !xfs_ifork_has_extents(ifp)) ||
+	    XFS_TEST_ERROR(false, mp, XFS_ERRTAG_BMAPIFORMAT)) {
+		return -EFSCORRUPTED;
+	}
+
+	if (xfs_is_shutdown(mp))
+		return -EIO;
+
+	error = xfs_iread_extents(tp, ip, whichfork);
+	if (error)
+		return error;
+
+	if (xfs_iext_lookup_extent(ip, ifp, bno, &icur, &got)) {
+		/* make sure we only reflink into a hole. */
+		ASSERT(got.br_startoff > bno);
+		ASSERT(got.br_startoff - bno >= len);
+	}
+
+	ip->i_nblocks += len;
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+
+	if (ifp->if_format == XFS_DINODE_FMT_BTREE) {
+		cur = xfs_bmbt_init_cursor(mp, tp, ip, whichfork);
+		cur->bc_ino.flags = 0;
+	}
+
+	got.br_startoff = bno;
+	got.br_startblock = startblock;
+	got.br_blockcount = len;
+	if (flags & XFS_BMAPI_PREALLOC)
+		got.br_state = XFS_EXT_UNWRITTEN;
+	else
+		got.br_state = XFS_EXT_NORM;
+
+	error = xfs_bmap_add_extent_hole_real(tp, ip, whichfork, &icur,
+			&cur, &got, &logflags, flags);
+	if (error)
+		goto error0;
+
+	error = xfs_bmap_btree_to_extents(tp, ip, cur, &logflags, whichfork);
+
+error0:
+	if (ip->i_df.if_format != XFS_DINODE_FMT_EXTENTS)
+		logflags &= ~XFS_ILOG_DEXT;
+	else if (ip->i_df.if_format != XFS_DINODE_FMT_BTREE)
+		logflags &= ~XFS_ILOG_DBROOT;
+
+	if (logflags)
+		xfs_trans_log_inode(tp, ip, logflags);
+	if (cur)
+		xfs_btree_del_cursor(cur, error);
+	return error;
+}
+
+/*
+ * When a delalloc extent is split (e.g., due to a hole punch), the original
+ * indlen reservation must be shared across the two new extents that are left
+ * behind.
+ *
+ * Given the original reservation and the worst case indlen for the two new
+ * extents (as calculated by xfs_bmap_worst_indlen()), split the original
+ * reservation fairly across the two new extents. If necessary, steal available
+ * blocks from a deleted extent to make up a reservation deficiency (e.g., if
+ * ores == 1). The number of stolen blocks is returned. The availability and
+ * subsequent accounting of stolen blocks is the responsibility of the caller.
+ */
+static xfs_filblks_t
+xfs_bmap_split_indlen(
+	xfs_filblks_t			ores,		/* original res. */
+	xfs_filblks_t			*indlen1,	/* ext1 worst indlen */
+	xfs_filblks_t			*indlen2,	/* ext2 worst indlen */
+	xfs_filblks_t			avail)		/* stealable blocks */
+{
+	xfs_filblks_t			len1 = *indlen1;
+	xfs_filblks_t			len2 = *indlen2;
+	xfs_filblks_t			nres = len1 + len2; /* new total res. */
+	xfs_filblks_t			stolen = 0;
+	xfs_filblks_t			resfactor;
+
+	/*
+	 * Steal as many blocks as we can to try and satisfy the worst case
+	 * indlen for both new extents.
+	 */
+	if (ores < nres && avail)
+		stolen = XFS_FILBLKS_MIN(nres - ores, avail);
+	ores += stolen;
+
+	 /* nothing else to do if we've satisfied the new reservation */
+	if (ores >= nres)
+		return stolen;
+
+	/*
+	 * We can't meet the total required reservation for the two extents.
+	 * Calculate the percent of the overall shortage between both extents
+	 * and apply this percentage to each of the requested indlen values.
+	 * This distributes the shortage fairly and reduces the chances that one
+	 * of the two extents is left with nothing when extents are repeatedly
+	 * split.
+	 */
+	resfactor = (ores * 100);
+	do_div(resfactor, nres);
+	len1 *= resfactor;
+	do_div(len1, 100);
+	len2 *= resfactor;
+	do_div(len2, 100);
+	ASSERT(len1 + len2 <= ores);
+	ASSERT(len1 < *indlen1 && len2 < *indlen2);
+
+	/*
+	 * Hand out the remainder to each extent. If one of the two reservations
+	 * is zero, we want to make sure that one gets a block first. The loop
+	 * below starts with len1, so hand len2 a block right off the bat if it
+	 * is zero.
+	 */
+	ores -= (len1 + len2);
+	ASSERT((*indlen1 - len1) + (*indlen2 - len2) >= ores);
+	if (ores && !len2 && *indlen2) {
+		len2++;
+		ores--;
+	}
+	while (ores) {
+		if (len1 < *indlen1) {
+			len1++;
+			ores--;
+		}
+		if (!ores)
+			break;
+		if (len2 < *indlen2) {
+			len2++;
+			ores--;
+		}
+	}
+
+	*indlen1 = len1;
+	*indlen2 = len2;
+
+	return stolen;
+}
+
+int
+xfs_bmap_del_extent_delay(
+	struct xfs_inode	*ip,
+	int			whichfork,
+	struct xfs_iext_cursor	*icur,
+	struct xfs_bmbt_irec	*got,
+	struct xfs_bmbt_irec	*del)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_bmbt_irec	new;
+	int64_t			da_old, da_new, da_diff = 0;
+	xfs_fileoff_t		del_endoff, got_endoff;
+	xfs_filblks_t		got_indlen, new_indlen, stolen;
+	uint32_t		state = xfs_bmap_fork_to_state(whichfork);
+	int			error = 0;
+	bool			isrt;
+
+	XFS_STATS_INC(mp, xs_del_exlist);
+
+	isrt = (whichfork == XFS_DATA_FORK) && XFS_IS_REALTIME_INODE(ip);
+	del_endoff = del->br_startoff + del->br_blockcount;
+	got_endoff = got->br_startoff + got->br_blockcount;
+	da_old = startblockval(got->br_startblock);
+	da_new = 0;
+
+	ASSERT(del->br_blockcount > 0);
+	ASSERT(got->br_startoff <= del->br_startoff);
+	ASSERT(got_endoff >= del_endoff);
+
+	if (isrt) {
+		uint64_t rtexts = XFS_FSB_TO_B(mp, del->br_blockcount);
+
+		do_div(rtexts, mp->m_sb.sb_rextsize);
+		xfs_mod_frextents(mp, rtexts);
+	}
+
+	/*
+	 * Update the inode delalloc counter now and wait to update the
+	 * sb counters as we might have to borrow some blocks for the
+	 * indirect block accounting.
+	 */
+	ASSERT(!isrt);
+	error = xfs_quota_unreserve_blkres(ip, del->br_blockcount);
+	if (error)
+		return error;
+	ip->i_delayed_blks -= del->br_blockcount;
+
+	if (got->br_startoff == del->br_startoff)
+		state |= BMAP_LEFT_FILLING;
+	if (got_endoff == del_endoff)
+		state |= BMAP_RIGHT_FILLING;
+
+	switch (state & (BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING)) {
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING:
+		/*
+		 * Matches the whole extent.  Delete the entry.
+		 */
+		xfs_iext_remove(ip, icur, state);
+		xfs_iext_prev(ifp, icur);
+		break;
+	case BMAP_LEFT_FILLING:
+		/*
+		 * Deleting the first part of the extent.
+		 */
+		got->br_startoff = del_endoff;
+		got->br_blockcount -= del->br_blockcount;
+		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip,
+				got->br_blockcount), da_old);
+		got->br_startblock = nullstartblock((int)da_new);
+		xfs_iext_update_extent(ip, state, icur, got);
+		break;
+	case BMAP_RIGHT_FILLING:
+		/*
+		 * Deleting the last part of the extent.
+		 */
+		got->br_blockcount = got->br_blockcount - del->br_blockcount;
+		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip,
+				got->br_blockcount), da_old);
+		got->br_startblock = nullstartblock((int)da_new);
+		xfs_iext_update_extent(ip, state, icur, got);
+		break;
+	case 0:
+		/*
+		 * Deleting the middle of the extent.
+		 *
+		 * Distribute the original indlen reservation across the two new
+		 * extents.  Steal blocks from the deleted extent if necessary.
+		 * Stealing blocks simply fudges the fdblocks accounting below.
+		 * Warn if either of the new indlen reservations is zero as this
+		 * can lead to delalloc problems.
+		 */
+		got->br_blockcount = del->br_startoff - got->br_startoff;
+		got_indlen = xfs_bmap_worst_indlen(ip, got->br_blockcount);
+
+		new.br_blockcount = got_endoff - del_endoff;
+		new_indlen = xfs_bmap_worst_indlen(ip, new.br_blockcount);
+
+		WARN_ON_ONCE(!got_indlen || !new_indlen);
+		stolen = xfs_bmap_split_indlen(da_old, &got_indlen, &new_indlen,
+						       del->br_blockcount);
+
+		got->br_startblock = nullstartblock((int)got_indlen);
+
+		new.br_startoff = del_endoff;
+		new.br_state = got->br_state;
+		new.br_startblock = nullstartblock((int)new_indlen);
+
+		xfs_iext_update_extent(ip, state, icur, got);
+		xfs_iext_next(ifp, icur);
+		xfs_iext_insert(ip, icur, &new, state);
+
+		da_new = got_indlen + new_indlen - stolen;
+		del->br_blockcount -= stolen;
+		break;
+	}
+
+	ASSERT(da_old >= da_new);
+	da_diff = da_old - da_new;
+	if (!isrt)
+		da_diff += del->br_blockcount;
+	if (da_diff) {
+		xfs_mod_fdblocks(mp, da_diff, false);
+		xfs_mod_delalloc(mp, -da_diff);
+	}
+	return error;
+}
+
+void
+xfs_bmap_del_extent_cow(
+	struct xfs_inode	*ip,
+	struct xfs_iext_cursor	*icur,
+	struct xfs_bmbt_irec	*got,
+	struct xfs_bmbt_irec	*del)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, XFS_COW_FORK);
+	struct xfs_bmbt_irec	new;
+	xfs_fileoff_t		del_endoff, got_endoff;
+	uint32_t		state = BMAP_COWFORK;
+
+	XFS_STATS_INC(mp, xs_del_exlist);
+
+	del_endoff = del->br_startoff + del->br_blockcount;
+	got_endoff = got->br_startoff + got->br_blockcount;
+
+	ASSERT(del->br_blockcount > 0);
+	ASSERT(got->br_startoff <= del->br_startoff);
+	ASSERT(got_endoff >= del_endoff);
+	ASSERT(!isnullstartblock(got->br_startblock));
+
+	if (got->br_startoff == del->br_startoff)
+		state |= BMAP_LEFT_FILLING;
+	if (got_endoff == del_endoff)
+		state |= BMAP_RIGHT_FILLING;
+
+	switch (state & (BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING)) {
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING:
+		/*
+		 * Matches the whole extent.  Delete the entry.
+		 */
+		xfs_iext_remove(ip, icur, state);
+		xfs_iext_prev(ifp, icur);
+		break;
+	case BMAP_LEFT_FILLING:
+		/*
+		 * Deleting the first part of the extent.
+		 */
+		got->br_startoff = del_endoff;
+		got->br_blockcount -= del->br_blockcount;
+		got->br_startblock = del->br_startblock + del->br_blockcount;
+		xfs_iext_update_extent(ip, state, icur, got);
+		break;
+	case BMAP_RIGHT_FILLING:
+		/*
+		 * Deleting the last part of the extent.
+		 */
+		got->br_blockcount -= del->br_blockcount;
+		xfs_iext_update_extent(ip, state, icur, got);
+		break;
+	case 0:
+		/*
+		 * Deleting the middle of the extent.
+		 */
+		got->br_blockcount = del->br_startoff - got->br_startoff;
+
+		new.br_startoff = del_endoff;
+		new.br_blockcount = got_endoff - del_endoff;
+		new.br_state = got->br_state;
+		new.br_startblock = del->br_startblock + del->br_blockcount;
+
+		xfs_iext_update_extent(ip, state, icur, got);
+		xfs_iext_next(ifp, icur);
+		xfs_iext_insert(ip, icur, &new, state);
+		break;
+	}
+	ip->i_delayed_blks -= del->br_blockcount;
+}
+
+/*
+ * Called by xfs_bmapi to update file extent records and the btree
+ * after removing space.
+ */
+STATIC int				/* error */
+xfs_bmap_del_extent_real(
+	xfs_inode_t		*ip,	/* incore inode pointer */
+	xfs_trans_t		*tp,	/* current transaction pointer */
+	struct xfs_iext_cursor	*icur,
+	struct xfs_btree_cur	*cur,	/* if null, not a btree */
+	xfs_bmbt_irec_t		*del,	/* data to remove from extents */
+	int			*logflagsp, /* inode logging flags */
+	int			whichfork, /* data or attr fork */
+	uint32_t		bflags)	/* bmapi flags */
+{
+	xfs_fsblock_t		del_endblock=0;	/* first block past del */
+	xfs_fileoff_t		del_endoff;	/* first offset past del */
+	int			do_fx;	/* free extent at end of routine */
+	int			error;	/* error return value */
+	int			flags = 0;/* inode logging flags */
+	struct xfs_bmbt_irec	got;	/* current extent entry */
+	xfs_fileoff_t		got_endoff;	/* first offset past got */
+	int			i;	/* temp state */
+	struct xfs_ifork	*ifp;	/* inode fork pointer */
+	xfs_mount_t		*mp;	/* mount structure */
+	xfs_filblks_t		nblks;	/* quota/sb block count */
+	xfs_bmbt_irec_t		new;	/* new record to be inserted */
+	/* REFERENCED */
+	uint			qfield;	/* quota field to update */
+	uint32_t		state = xfs_bmap_fork_to_state(whichfork);
+	struct xfs_bmbt_irec	old;
+
+	mp = ip->i_mount;
+	XFS_STATS_INC(mp, xs_del_exlist);
+
+	ifp = xfs_ifork_ptr(ip, whichfork);
+	ASSERT(del->br_blockcount > 0);
+	xfs_iext_get_extent(ifp, icur, &got);
+	ASSERT(got.br_startoff <= del->br_startoff);
+	del_endoff = del->br_startoff + del->br_blockcount;
+	got_endoff = got.br_startoff + got.br_blockcount;
+	ASSERT(got_endoff >= del_endoff);
+	ASSERT(!isnullstartblock(got.br_startblock));
+	qfield = 0;
+	error = 0;
+
+	/*
+	 * If it's the case where the directory code is running with no block
+	 * reservation, and the deleted block is in the middle of its extent,
+	 * and the resulting insert of an extent would cause transformation to
+	 * btree format, then reject it.  The calling code will then swap blocks
+	 * around instead.  We have to do this now, rather than waiting for the
+	 * conversion to btree format, since the transaction will be dirty then.
+	 */
+	if (tp->t_blk_res == 0 &&
+	    ifp->if_format == XFS_DINODE_FMT_EXTENTS &&
+	    ifp->if_nextents >= XFS_IFORK_MAXEXT(ip, whichfork) &&
+	    del->br_startoff > got.br_startoff && del_endoff < got_endoff)
+		return -ENOSPC;
+
+	flags = XFS_ILOG_CORE;
+	if (whichfork == XFS_DATA_FORK && XFS_IS_REALTIME_INODE(ip)) {
+		xfs_filblks_t	len;
+		xfs_extlen_t	mod;
+
+		len = div_u64_rem(del->br_blockcount, mp->m_sb.sb_rextsize,
+				  &mod);
+		ASSERT(mod == 0);
+
+		if (!(bflags & XFS_BMAPI_REMAP)) {
+			xfs_fsblock_t	bno;
+
+			bno = div_u64_rem(del->br_startblock,
+					mp->m_sb.sb_rextsize, &mod);
+			ASSERT(mod == 0);
+
+			error = xfs_rtfree_extent(tp, bno, (xfs_extlen_t)len);
+			if (error)
+				goto done;
+		}
+
+		do_fx = 0;
+		nblks = len * mp->m_sb.sb_rextsize;
+		qfield = XFS_TRANS_DQ_RTBCOUNT;
+	} else {
+		do_fx = 1;
+		nblks = del->br_blockcount;
+		qfield = XFS_TRANS_DQ_BCOUNT;
+	}
+
+	del_endblock = del->br_startblock + del->br_blockcount;
+	if (cur) {
+		error = xfs_bmbt_lookup_eq(cur, &got, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+	}
+
+	if (got.br_startoff == del->br_startoff)
+		state |= BMAP_LEFT_FILLING;
+	if (got_endoff == del_endoff)
+		state |= BMAP_RIGHT_FILLING;
+
+	switch (state & (BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING)) {
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING:
+		/*
+		 * Matches the whole extent.  Delete the entry.
+		 */
+		xfs_iext_remove(ip, icur, state);
+		xfs_iext_prev(ifp, icur);
+		ifp->if_nextents--;
+
+		flags |= XFS_ILOG_CORE;
+		if (!cur) {
+			flags |= xfs_ilog_fext(whichfork);
+			break;
+		}
+		if ((error = xfs_btree_delete(cur, &i)))
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		break;
+	case BMAP_LEFT_FILLING:
+		/*
+		 * Deleting the first part of the extent.
+		 */
+		got.br_startoff = del_endoff;
+		got.br_startblock = del_endblock;
+		got.br_blockcount -= del->br_blockcount;
+		xfs_iext_update_extent(ip, state, icur, &got);
+		if (!cur) {
+			flags |= xfs_ilog_fext(whichfork);
+			break;
+		}
+		error = xfs_bmbt_update(cur, &got);
+		if (error)
+			goto done;
+		break;
+	case BMAP_RIGHT_FILLING:
+		/*
+		 * Deleting the last part of the extent.
+		 */
+		got.br_blockcount -= del->br_blockcount;
+		xfs_iext_update_extent(ip, state, icur, &got);
+		if (!cur) {
+			flags |= xfs_ilog_fext(whichfork);
+			break;
+		}
+		error = xfs_bmbt_update(cur, &got);
+		if (error)
+			goto done;
+		break;
+	case 0:
+		/*
+		 * Deleting the middle of the extent.
+		 */
+
+		old = got;
+
+		got.br_blockcount = del->br_startoff - got.br_startoff;
+		xfs_iext_update_extent(ip, state, icur, &got);
+
+		new.br_startoff = del_endoff;
+		new.br_blockcount = got_endoff - del_endoff;
+		new.br_state = got.br_state;
+		new.br_startblock = del_endblock;
+
+		flags |= XFS_ILOG_CORE;
+		if (cur) {
+			error = xfs_bmbt_update(cur, &got);
+			if (error)
+				goto done;
+			error = xfs_btree_increment(cur, 0, &i);
+			if (error)
+				goto done;
+			cur->bc_rec.b = new;
+			error = xfs_btree_insert(cur, &i);
+			if (error && error != -ENOSPC)
+				goto done;
+			/*
+			 * If get no-space back from btree insert, it tried a
+			 * split, and we have a zero block reservation.  Fix up
+			 * our state and return the error.
+			 */
+			if (error == -ENOSPC) {
+				/*
+				 * Reset the cursor, don't trust it after any
+				 * insert operation.
+				 */
+				error = xfs_bmbt_lookup_eq(cur, &got, &i);
+				if (error)
+					goto done;
+				if (XFS_IS_CORRUPT(mp, i != 1)) {
+					error = -EFSCORRUPTED;
+					goto done;
+				}
+				/*
+				 * Update the btree record back
+				 * to the original value.
+				 */
+				error = xfs_bmbt_update(cur, &old);
+				if (error)
+					goto done;
+				/*
+				 * Reset the extent record back
+				 * to the original value.
+				 */
+				xfs_iext_update_extent(ip, state, icur, &old);
+				flags = 0;
+				error = -ENOSPC;
+				goto done;
+			}
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto done;
+			}
+		} else
+			flags |= xfs_ilog_fext(whichfork);
+
+		ifp->if_nextents++;
+		xfs_iext_next(ifp, icur);
+		xfs_iext_insert(ip, icur, &new, state);
+		break;
+	}
+
+	/* remove reverse mapping */
+	xfs_rmap_unmap_extent(tp, ip, whichfork, del);
+
+	/*
+	 * If we need to, add to list of extents to delete.
+	 */
+	if (do_fx && !(bflags & XFS_BMAPI_REMAP)) {
+		if (xfs_is_reflink_inode(ip) && whichfork == XFS_DATA_FORK) {
+			xfs_refcount_decrease_extent(tp, del);
+		} else {
+			__xfs_free_extent_later(tp, del->br_startblock,
+					del->br_blockcount, NULL,
+					(bflags & XFS_BMAPI_NODISCARD) ||
+					del->br_state == XFS_EXT_UNWRITTEN);
+		}
+	}
+
+	/*
+	 * Adjust inode # blocks in the file.
+	 */
+	if (nblks)
+		ip->i_nblocks -= nblks;
+	/*
+	 * Adjust quota data.
+	 */
+	if (qfield && !(bflags & XFS_BMAPI_REMAP))
+		xfs_trans_mod_dquot_byino(tp, ip, qfield, (long)-nblks);
+
+done:
+	*logflagsp = flags;
+	return error;
+}
+
+/*
+ * Unmap (remove) blocks from a file.
+ * If nexts is nonzero then the number of extents to remove is limited to
+ * that value.  If not all extents in the block range can be removed then
+ * *done is set.
+ */
+int						/* error */
+__xfs_bunmapi(
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* incore inode */
+	xfs_fileoff_t		start,		/* first file offset deleted */
+	xfs_filblks_t		*rlen,		/* i/o: amount remaining */
+	uint32_t		flags,		/* misc flags */
+	xfs_extnum_t		nexts)		/* number of extents max */
+{
+	struct xfs_btree_cur	*cur;		/* bmap btree cursor */
+	struct xfs_bmbt_irec	del;		/* extent being deleted */
+	int			error;		/* error return value */
+	xfs_extnum_t		extno;		/* extent number in list */
+	struct xfs_bmbt_irec	got;		/* current extent record */
+	struct xfs_ifork	*ifp;		/* inode fork pointer */
+	int			isrt;		/* freeing in rt area */
+	int			logflags;	/* transaction logging flags */
+	xfs_extlen_t		mod;		/* rt extent offset */
+	struct xfs_mount	*mp = ip->i_mount;
+	int			tmp_logflags;	/* partial logging flags */
+	int			wasdel;		/* was a delayed alloc extent */
+	int			whichfork;	/* data or attribute fork */
+	xfs_fsblock_t		sum;
+	xfs_filblks_t		len = *rlen;	/* length to unmap in file */
+	xfs_fileoff_t		end;
+	struct xfs_iext_cursor	icur;
+	bool			done = false;
+
+	trace_xfs_bunmap(ip, start, len, flags, _RET_IP_);
+
+	whichfork = xfs_bmapi_whichfork(flags);
+	ASSERT(whichfork != XFS_COW_FORK);
+	ifp = xfs_ifork_ptr(ip, whichfork);
+	if (XFS_IS_CORRUPT(mp, !xfs_ifork_has_extents(ifp)))
+		return -EFSCORRUPTED;
+	if (xfs_is_shutdown(mp))
+		return -EIO;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+	ASSERT(len > 0);
+	ASSERT(nexts >= 0);
+
+	error = xfs_iread_extents(tp, ip, whichfork);
+	if (error)
+		return error;
+
+	if (xfs_iext_count(ifp) == 0) {
+		*rlen = 0;
+		return 0;
+	}
+	XFS_STATS_INC(mp, xs_blk_unmap);
+	isrt = (whichfork == XFS_DATA_FORK) && XFS_IS_REALTIME_INODE(ip);
+	end = start + len;
+
+	if (!xfs_iext_lookup_extent_before(ip, ifp, &end, &icur, &got)) {
+		*rlen = 0;
+		return 0;
+	}
+	end--;
+
+	logflags = 0;
+	if (ifp->if_format == XFS_DINODE_FMT_BTREE) {
+		ASSERT(ifp->if_format == XFS_DINODE_FMT_BTREE);
+		cur = xfs_bmbt_init_cursor(mp, tp, ip, whichfork);
+		cur->bc_ino.flags = 0;
+	} else
+		cur = NULL;
+
+	if (isrt) {
+		/*
+		 * Synchronize by locking the bitmap inode.
+		 */
+		xfs_ilock(mp->m_rbmip, XFS_ILOCK_EXCL|XFS_ILOCK_RTBITMAP);
+		xfs_trans_ijoin(tp, mp->m_rbmip, XFS_ILOCK_EXCL);
+		xfs_ilock(mp->m_rsumip, XFS_ILOCK_EXCL|XFS_ILOCK_RTSUM);
+		xfs_trans_ijoin(tp, mp->m_rsumip, XFS_ILOCK_EXCL);
+	}
+
+	extno = 0;
+	while (end != (xfs_fileoff_t)-1 && end >= start &&
+	       (nexts == 0 || extno < nexts)) {
+		/*
+		 * Is the found extent after a hole in which end lives?
+		 * Just back up to the previous extent, if so.
+		 */
+		if (got.br_startoff > end &&
+		    !xfs_iext_prev_extent(ifp, &icur, &got)) {
+			done = true;
+			break;
+		}
+		/*
+		 * Is the last block of this extent before the range
+		 * we're supposed to delete?  If so, we're done.
+		 */
+		end = XFS_FILEOFF_MIN(end,
+			got.br_startoff + got.br_blockcount - 1);
+		if (end < start)
+			break;
+		/*
+		 * Then deal with the (possibly delayed) allocated space
+		 * we found.
+		 */
+		del = got;
+		wasdel = isnullstartblock(del.br_startblock);
+
+		if (got.br_startoff < start) {
+			del.br_startoff = start;
+			del.br_blockcount -= start - got.br_startoff;
+			if (!wasdel)
+				del.br_startblock += start - got.br_startoff;
+		}
+		if (del.br_startoff + del.br_blockcount > end + 1)
+			del.br_blockcount = end + 1 - del.br_startoff;
+
+		if (!isrt)
+			goto delete;
+
+		sum = del.br_startblock + del.br_blockcount;
+		div_u64_rem(sum, mp->m_sb.sb_rextsize, &mod);
+		if (mod) {
+			/*
+			 * Realtime extent not lined up at the end.
+			 * The extent could have been split into written
+			 * and unwritten pieces, or we could just be
+			 * unmapping part of it.  But we can't really
+			 * get rid of part of a realtime extent.
+			 */
+			if (del.br_state == XFS_EXT_UNWRITTEN) {
+				/*
+				 * This piece is unwritten, or we're not
+				 * using unwritten extents.  Skip over it.
+				 */
+				ASSERT(end >= mod);
+				end -= mod > del.br_blockcount ?
+					del.br_blockcount : mod;
+				if (end < got.br_startoff &&
+				    !xfs_iext_prev_extent(ifp, &icur, &got)) {
+					done = true;
+					break;
+				}
+				continue;
+			}
+			/*
+			 * It's written, turn it unwritten.
+			 * This is better than zeroing it.
+			 */
+			ASSERT(del.br_state == XFS_EXT_NORM);
+			ASSERT(tp->t_blk_res > 0);
+			/*
+			 * If this spans a realtime extent boundary,
+			 * chop it back to the start of the one we end at.
+			 */
+			if (del.br_blockcount > mod) {
+				del.br_startoff += del.br_blockcount - mod;
+				del.br_startblock += del.br_blockcount - mod;
+				del.br_blockcount = mod;
+			}
+			del.br_state = XFS_EXT_UNWRITTEN;
+			error = xfs_bmap_add_extent_unwritten_real(tp, ip,
+					whichfork, &icur, &cur, &del,
+					&logflags);
+			if (error)
+				goto error0;
+			goto nodelete;
+		}
+		div_u64_rem(del.br_startblock, mp->m_sb.sb_rextsize, &mod);
+		if (mod) {
+			xfs_extlen_t off = mp->m_sb.sb_rextsize - mod;
+
+			/*
+			 * Realtime extent is lined up at the end but not
+			 * at the front.  We'll get rid of full extents if
+			 * we can.
+			 */
+			if (del.br_blockcount > off) {
+				del.br_blockcount -= off;
+				del.br_startoff += off;
+				del.br_startblock += off;
+			} else if (del.br_startoff == start &&
+				   (del.br_state == XFS_EXT_UNWRITTEN ||
+				    tp->t_blk_res == 0)) {
+				/*
+				 * Can't make it unwritten.  There isn't
+				 * a full extent here so just skip it.
+				 */
+				ASSERT(end >= del.br_blockcount);
+				end -= del.br_blockcount;
+				if (got.br_startoff > end &&
+				    !xfs_iext_prev_extent(ifp, &icur, &got)) {
+					done = true;
+					break;
+				}
+				continue;
+			} else if (del.br_state == XFS_EXT_UNWRITTEN) {
+				struct xfs_bmbt_irec	prev;
+				xfs_fileoff_t		unwrite_start;
+
+				/*
+				 * This one is already unwritten.
+				 * It must have a written left neighbor.
+				 * Unwrite the killed part of that one and
+				 * try again.
+				 */
+				if (!xfs_iext_prev_extent(ifp, &icur, &prev))
+					ASSERT(0);
+				ASSERT(prev.br_state == XFS_EXT_NORM);
+				ASSERT(!isnullstartblock(prev.br_startblock));
+				ASSERT(del.br_startblock ==
+				       prev.br_startblock + prev.br_blockcount);
+				unwrite_start = max3(start,
+						     del.br_startoff - mod,
+						     prev.br_startoff);
+				mod = unwrite_start - prev.br_startoff;
+				prev.br_startoff = unwrite_start;
+				prev.br_startblock += mod;
+				prev.br_blockcount -= mod;
+				prev.br_state = XFS_EXT_UNWRITTEN;
+				error = xfs_bmap_add_extent_unwritten_real(tp,
+						ip, whichfork, &icur, &cur,
+						&prev, &logflags);
+				if (error)
+					goto error0;
+				goto nodelete;
+			} else {
+				ASSERT(del.br_state == XFS_EXT_NORM);
+				del.br_state = XFS_EXT_UNWRITTEN;
+				error = xfs_bmap_add_extent_unwritten_real(tp,
+						ip, whichfork, &icur, &cur,
+						&del, &logflags);
+				if (error)
+					goto error0;
+				goto nodelete;
+			}
+		}
+
+delete:
+		if (wasdel) {
+			error = xfs_bmap_del_extent_delay(ip, whichfork, &icur,
+					&got, &del);
+		} else {
+			error = xfs_bmap_del_extent_real(ip, tp, &icur, cur,
+					&del, &tmp_logflags, whichfork,
+					flags);
+			logflags |= tmp_logflags;
+		}
+
+		if (error)
+			goto error0;
+
+		end = del.br_startoff - 1;
+nodelete:
+		/*
+		 * If not done go on to the next (previous) record.
+		 */
+		if (end != (xfs_fileoff_t)-1 && end >= start) {
+			if (!xfs_iext_get_extent(ifp, &icur, &got) ||
+			    (got.br_startoff > end &&
+			     !xfs_iext_prev_extent(ifp, &icur, &got))) {
+				done = true;
+				break;
+			}
+			extno++;
+		}
+	}
+	if (done || end == (xfs_fileoff_t)-1 || end < start)
+		*rlen = 0;
+	else
+		*rlen = end - start + 1;
+
+	/*
+	 * Convert to a btree if necessary.
+	 */
+	if (xfs_bmap_needs_btree(ip, whichfork)) {
+		ASSERT(cur == NULL);
+		error = xfs_bmap_extents_to_btree(tp, ip, &cur, 0,
+				&tmp_logflags, whichfork);
+		logflags |= tmp_logflags;
+	} else {
+		error = xfs_bmap_btree_to_extents(tp, ip, cur, &logflags,
+			whichfork);
+	}
+
+error0:
+	/*
+	 * Log everything.  Do this after conversion, there's no point in
+	 * logging the extent records if we've converted to btree format.
+	 */
+	if ((logflags & xfs_ilog_fext(whichfork)) &&
+	    ifp->if_format != XFS_DINODE_FMT_EXTENTS)
+		logflags &= ~xfs_ilog_fext(whichfork);
+	else if ((logflags & xfs_ilog_fbroot(whichfork)) &&
+		 ifp->if_format != XFS_DINODE_FMT_BTREE)
+		logflags &= ~xfs_ilog_fbroot(whichfork);
+	/*
+	 * Log inode even in the error case, if the transaction
+	 * is dirty we'll need to shut down the filesystem.
+	 */
+	if (logflags)
+		xfs_trans_log_inode(tp, ip, logflags);
+	if (cur) {
+		if (!error)
+			cur->bc_ino.allocated = 0;
+		xfs_btree_del_cursor(cur, error);
+	}
+	return error;
+}
+
+/* Unmap a range of a file. */
+int
+xfs_bunmapi(
+	xfs_trans_t		*tp,
+	struct xfs_inode	*ip,
+	xfs_fileoff_t		bno,
+	xfs_filblks_t		len,
+	uint32_t		flags,
+	xfs_extnum_t		nexts,
+	int			*done)
+{
+	int			error;
+
+	error = __xfs_bunmapi(tp, ip, bno, &len, flags, nexts);
+	*done = (len == 0);
+	return error;
+}
+
+/*
+ * Determine whether an extent shift can be accomplished by a merge with the
+ * extent that precedes the target hole of the shift.
+ */
+STATIC bool
+xfs_bmse_can_merge(
+	struct xfs_bmbt_irec	*left,	/* preceding extent */
+	struct xfs_bmbt_irec	*got,	/* current extent to shift */
+	xfs_fileoff_t		shift)	/* shift fsb */
+{
+	xfs_fileoff_t		startoff;
+
+	startoff = got->br_startoff - shift;
+
+	/*
+	 * The extent, once shifted, must be adjacent in-file and on-disk with
+	 * the preceding extent.
+	 */
+	if ((left->br_startoff + left->br_blockcount != startoff) ||
+	    (left->br_startblock + left->br_blockcount != got->br_startblock) ||
+	    (left->br_state != got->br_state) ||
+	    (left->br_blockcount + got->br_blockcount > XFS_MAX_BMBT_EXTLEN))
+		return false;
+
+	return true;
+}
+
+/*
+ * A bmap extent shift adjusts the file offset of an extent to fill a preceding
+ * hole in the file. If an extent shift would result in the extent being fully
+ * adjacent to the extent that currently precedes the hole, we can merge with
+ * the preceding extent rather than do the shift.
+ *
+ * This function assumes the caller has verified a shift-by-merge is possible
+ * with the provided extents via xfs_bmse_can_merge().
+ */
+STATIC int
+xfs_bmse_merge(
+	struct xfs_trans		*tp,
+	struct xfs_inode		*ip,
+	int				whichfork,
+	xfs_fileoff_t			shift,		/* shift fsb */
+	struct xfs_iext_cursor		*icur,
+	struct xfs_bmbt_irec		*got,		/* extent to shift */
+	struct xfs_bmbt_irec		*left,		/* preceding extent */
+	struct xfs_btree_cur		*cur,
+	int				*logflags)	/* output */
+{
+	struct xfs_ifork		*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_bmbt_irec		new;
+	xfs_filblks_t			blockcount;
+	int				error, i;
+	struct xfs_mount		*mp = ip->i_mount;
+
+	blockcount = left->br_blockcount + got->br_blockcount;
+
+	ASSERT(xfs_isilocked(ip, XFS_IOLOCK_EXCL));
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+	ASSERT(xfs_bmse_can_merge(left, got, shift));
+
+	new = *left;
+	new.br_blockcount = blockcount;
+
+	/*
+	 * Update the on-disk extent count, the btree if necessary and log the
+	 * inode.
+	 */
+	ifp->if_nextents--;
+	*logflags |= XFS_ILOG_CORE;
+	if (!cur) {
+		*logflags |= XFS_ILOG_DEXT;
+		goto done;
+	}
+
+	/* lookup and remove the extent to merge */
+	error = xfs_bmbt_lookup_eq(cur, got, &i);
+	if (error)
+		return error;
+	if (XFS_IS_CORRUPT(mp, i != 1))
+		return -EFSCORRUPTED;
+
+	error = xfs_btree_delete(cur, &i);
+	if (error)
+		return error;
+	if (XFS_IS_CORRUPT(mp, i != 1))
+		return -EFSCORRUPTED;
+
+	/* lookup and update size of the previous extent */
+	error = xfs_bmbt_lookup_eq(cur, left, &i);
+	if (error)
+		return error;
+	if (XFS_IS_CORRUPT(mp, i != 1))
+		return -EFSCORRUPTED;
+
+	error = xfs_bmbt_update(cur, &new);
+	if (error)
+		return error;
+
+	/* change to extent format if required after extent removal */
+	error = xfs_bmap_btree_to_extents(tp, ip, cur, logflags, whichfork);
+	if (error)
+		return error;
+
+done:
+	xfs_iext_remove(ip, icur, 0);
+	xfs_iext_prev(ifp, icur);
+	xfs_iext_update_extent(ip, xfs_bmap_fork_to_state(whichfork), icur,
+			&new);
+
+	/* update reverse mapping. rmap functions merge the rmaps for us */
+	xfs_rmap_unmap_extent(tp, ip, whichfork, got);
+	memcpy(&new, got, sizeof(new));
+	new.br_startoff = left->br_startoff + left->br_blockcount;
+	xfs_rmap_map_extent(tp, ip, whichfork, &new);
+	return 0;
+}
+
+static int
+xfs_bmap_shift_update_extent(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	int			whichfork,
+	struct xfs_iext_cursor	*icur,
+	struct xfs_bmbt_irec	*got,
+	struct xfs_btree_cur	*cur,
+	int			*logflags,
+	xfs_fileoff_t		startoff)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_bmbt_irec	prev = *got;
+	int			error, i;
+
+	*logflags |= XFS_ILOG_CORE;
+
+	got->br_startoff = startoff;
+
+	if (cur) {
+		error = xfs_bmbt_lookup_eq(cur, &prev, &i);
+		if (error)
+			return error;
+		if (XFS_IS_CORRUPT(mp, i != 1))
+			return -EFSCORRUPTED;
+
+		error = xfs_bmbt_update(cur, got);
+		if (error)
+			return error;
+	} else {
+		*logflags |= XFS_ILOG_DEXT;
+	}
+
+	xfs_iext_update_extent(ip, xfs_bmap_fork_to_state(whichfork), icur,
+			got);
+
+	/* update reverse mapping */
+	xfs_rmap_unmap_extent(tp, ip, whichfork, &prev);
+	xfs_rmap_map_extent(tp, ip, whichfork, got);
+	return 0;
+}
+
+int
+xfs_bmap_collapse_extents(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	xfs_fileoff_t		*next_fsb,
+	xfs_fileoff_t		offset_shift_fsb,
+	bool			*done)
+{
+	int			whichfork = XFS_DATA_FORK;
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_btree_cur	*cur = NULL;
+	struct xfs_bmbt_irec	got, prev;
+	struct xfs_iext_cursor	icur;
+	xfs_fileoff_t		new_startoff;
+	int			error = 0;
+	int			logflags = 0;
+
+	if (XFS_IS_CORRUPT(mp, !xfs_ifork_has_extents(ifp)) ||
+	    XFS_TEST_ERROR(false, mp, XFS_ERRTAG_BMAPIFORMAT)) {
+		return -EFSCORRUPTED;
+	}
+
+	if (xfs_is_shutdown(mp))
+		return -EIO;
+
+	ASSERT(xfs_isilocked(ip, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL));
+
+	error = xfs_iread_extents(tp, ip, whichfork);
+	if (error)
+		return error;
+
+	if (ifp->if_format == XFS_DINODE_FMT_BTREE) {
+		cur = xfs_bmbt_init_cursor(mp, tp, ip, whichfork);
+		cur->bc_ino.flags = 0;
+	}
+
+	if (!xfs_iext_lookup_extent(ip, ifp, *next_fsb, &icur, &got)) {
+		*done = true;
+		goto del_cursor;
+	}
+	if (XFS_IS_CORRUPT(mp, isnullstartblock(got.br_startblock))) {
+		error = -EFSCORRUPTED;
+		goto del_cursor;
+	}
+
+	new_startoff = got.br_startoff - offset_shift_fsb;
+	if (xfs_iext_peek_prev_extent(ifp, &icur, &prev)) {
+		if (new_startoff < prev.br_startoff + prev.br_blockcount) {
+			error = -EINVAL;
+			goto del_cursor;
+		}
+
+		if (xfs_bmse_can_merge(&prev, &got, offset_shift_fsb)) {
+			error = xfs_bmse_merge(tp, ip, whichfork,
+					offset_shift_fsb, &icur, &got, &prev,
+					cur, &logflags);
+			if (error)
+				goto del_cursor;
+			goto done;
+		}
+	} else {
+		if (got.br_startoff < offset_shift_fsb) {
+			error = -EINVAL;
+			goto del_cursor;
+		}
+	}
+
+	error = xfs_bmap_shift_update_extent(tp, ip, whichfork, &icur, &got,
+			cur, &logflags, new_startoff);
+	if (error)
+		goto del_cursor;
+
+done:
+	if (!xfs_iext_next_extent(ifp, &icur, &got)) {
+		*done = true;
+		goto del_cursor;
+	}
+
+	*next_fsb = got.br_startoff;
+del_cursor:
+	if (cur)
+		xfs_btree_del_cursor(cur, error);
+	if (logflags)
+		xfs_trans_log_inode(tp, ip, logflags);
+	return error;
+}
+
+/* Make sure we won't be right-shifting an extent past the maximum bound. */
+int
+xfs_bmap_can_insert_extents(
+	struct xfs_inode	*ip,
+	xfs_fileoff_t		off,
+	xfs_fileoff_t		shift)
+{
+	struct xfs_bmbt_irec	got;
+	int			is_empty;
+	int			error = 0;
+
+	ASSERT(xfs_isilocked(ip, XFS_IOLOCK_EXCL));
+
+	if (xfs_is_shutdown(ip->i_mount))
+		return -EIO;
+
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	error = xfs_bmap_last_extent(NULL, ip, XFS_DATA_FORK, &got, &is_empty);
+	if (!error && !is_empty && got.br_startoff >= off &&
+	    ((got.br_startoff + shift) & BMBT_STARTOFF_MASK) < got.br_startoff)
+		error = -EINVAL;
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+
+	return error;
+}
+
+int
+xfs_bmap_insert_extents(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	xfs_fileoff_t		*next_fsb,
+	xfs_fileoff_t		offset_shift_fsb,
+	bool			*done,
+	xfs_fileoff_t		stop_fsb)
+{
+	int			whichfork = XFS_DATA_FORK;
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_btree_cur	*cur = NULL;
+	struct xfs_bmbt_irec	got, next;
+	struct xfs_iext_cursor	icur;
+	xfs_fileoff_t		new_startoff;
+	int			error = 0;
+	int			logflags = 0;
+
+	if (XFS_IS_CORRUPT(mp, !xfs_ifork_has_extents(ifp)) ||
+	    XFS_TEST_ERROR(false, mp, XFS_ERRTAG_BMAPIFORMAT)) {
+		return -EFSCORRUPTED;
+	}
+
+	if (xfs_is_shutdown(mp))
+		return -EIO;
+
+	ASSERT(xfs_isilocked(ip, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL));
+
+	error = xfs_iread_extents(tp, ip, whichfork);
+	if (error)
+		return error;
+
+	if (ifp->if_format == XFS_DINODE_FMT_BTREE) {
+		cur = xfs_bmbt_init_cursor(mp, tp, ip, whichfork);
+		cur->bc_ino.flags = 0;
+	}
+
+	if (*next_fsb == NULLFSBLOCK) {
+		xfs_iext_last(ifp, &icur);
+		if (!xfs_iext_get_extent(ifp, &icur, &got) ||
+		    stop_fsb > got.br_startoff) {
+			*done = true;
+			goto del_cursor;
+		}
+	} else {
+		if (!xfs_iext_lookup_extent(ip, ifp, *next_fsb, &icur, &got)) {
+			*done = true;
+			goto del_cursor;
+		}
+	}
+	if (XFS_IS_CORRUPT(mp, isnullstartblock(got.br_startblock))) {
+		error = -EFSCORRUPTED;
+		goto del_cursor;
+	}
+
+	if (XFS_IS_CORRUPT(mp, stop_fsb > got.br_startoff)) {
+		error = -EFSCORRUPTED;
+		goto del_cursor;
+	}
+
+	new_startoff = got.br_startoff + offset_shift_fsb;
+	if (xfs_iext_peek_next_extent(ifp, &icur, &next)) {
+		if (new_startoff + got.br_blockcount > next.br_startoff) {
+			error = -EINVAL;
+			goto del_cursor;
+		}
+
+		/*
+		 * Unlike a left shift (which involves a hole punch), a right
+		 * shift does not modify extent neighbors in any way.  We should
+		 * never find mergeable extents in this scenario.  Check anyways
+		 * and warn if we encounter two extents that could be one.
+		 */
+		if (xfs_bmse_can_merge(&got, &next, offset_shift_fsb))
+			WARN_ON_ONCE(1);
+	}
+
+	error = xfs_bmap_shift_update_extent(tp, ip, whichfork, &icur, &got,
+			cur, &logflags, new_startoff);
+	if (error)
+		goto del_cursor;
+
+	if (!xfs_iext_prev_extent(ifp, &icur, &got) ||
+	    stop_fsb >= got.br_startoff + got.br_blockcount) {
+		*done = true;
+		goto del_cursor;
+	}
+
+	*next_fsb = got.br_startoff;
+del_cursor:
+	if (cur)
+		xfs_btree_del_cursor(cur, error);
+	if (logflags)
+		xfs_trans_log_inode(tp, ip, logflags);
+	return error;
+}
+
+/*
+ * Splits an extent into two extents at split_fsb block such that it is the
+ * first block of the current_ext. @ext is a target extent to be split.
+ * @split_fsb is a block where the extents is split.  If split_fsb lies in a
+ * hole or the first block of extents, just return 0.
+ */
+int
+xfs_bmap_split_extent(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	xfs_fileoff_t		split_fsb)
+{
+	int				whichfork = XFS_DATA_FORK;
+	struct xfs_ifork		*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_btree_cur		*cur = NULL;
+	struct xfs_bmbt_irec		got;
+	struct xfs_bmbt_irec		new; /* split extent */
+	struct xfs_mount		*mp = ip->i_mount;
+	xfs_fsblock_t			gotblkcnt; /* new block count for got */
+	struct xfs_iext_cursor		icur;
+	int				error = 0;
+	int				logflags = 0;
+	int				i = 0;
+
+	if (XFS_IS_CORRUPT(mp, !xfs_ifork_has_extents(ifp)) ||
+	    XFS_TEST_ERROR(false, mp, XFS_ERRTAG_BMAPIFORMAT)) {
+		return -EFSCORRUPTED;
+	}
+
+	if (xfs_is_shutdown(mp))
+		return -EIO;
+
+	/* Read in all the extents */
+	error = xfs_iread_extents(tp, ip, whichfork);
+	if (error)
+		return error;
+
+	/*
+	 * If there are not extents, or split_fsb lies in a hole we are done.
+	 */
+	if (!xfs_iext_lookup_extent(ip, ifp, split_fsb, &icur, &got) ||
+	    got.br_startoff >= split_fsb)
+		return 0;
+
+	gotblkcnt = split_fsb - got.br_startoff;
+	new.br_startoff = split_fsb;
+	new.br_startblock = got.br_startblock + gotblkcnt;
+	new.br_blockcount = got.br_blockcount - gotblkcnt;
+	new.br_state = got.br_state;
+
+	if (ifp->if_format == XFS_DINODE_FMT_BTREE) {
+		cur = xfs_bmbt_init_cursor(mp, tp, ip, whichfork);
+		cur->bc_ino.flags = 0;
+		error = xfs_bmbt_lookup_eq(cur, &got, &i);
+		if (error)
+			goto del_cursor;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto del_cursor;
+		}
+	}
+
+	got.br_blockcount = gotblkcnt;
+	xfs_iext_update_extent(ip, xfs_bmap_fork_to_state(whichfork), &icur,
+			&got);
+
+	logflags = XFS_ILOG_CORE;
+	if (cur) {
+		error = xfs_bmbt_update(cur, &got);
+		if (error)
+			goto del_cursor;
+	} else
+		logflags |= XFS_ILOG_DEXT;
+
+	/* Add new extent */
+	xfs_iext_next(ifp, &icur);
+	xfs_iext_insert(ip, &icur, &new, 0);
+	ifp->if_nextents++;
+
+	if (cur) {
+		error = xfs_bmbt_lookup_eq(cur, &new, &i);
+		if (error)
+			goto del_cursor;
+		if (XFS_IS_CORRUPT(mp, i != 0)) {
+			error = -EFSCORRUPTED;
+			goto del_cursor;
+		}
+		error = xfs_btree_insert(cur, &i);
+		if (error)
+			goto del_cursor;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto del_cursor;
+		}
+	}
+
+	/*
+	 * Convert to a btree if necessary.
+	 */
+	if (xfs_bmap_needs_btree(ip, whichfork)) {
+		int tmp_logflags; /* partial log flag return val */
+
+		ASSERT(cur == NULL);
+		error = xfs_bmap_extents_to_btree(tp, ip, &cur, 0,
+				&tmp_logflags, whichfork);
+		logflags |= tmp_logflags;
+	}
+
+del_cursor:
+	if (cur) {
+		cur->bc_ino.allocated = 0;
+		xfs_btree_del_cursor(cur, error);
+	}
+
+	if (logflags)
+		xfs_trans_log_inode(tp, ip, logflags);
+	return error;
+}
+
+/* Deferred mapping is only for real extents in the data fork. */
+static bool
+xfs_bmap_is_update_needed(
+	struct xfs_bmbt_irec	*bmap)
+{
+	return  bmap->br_startblock != HOLESTARTBLOCK &&
+		bmap->br_startblock != DELAYSTARTBLOCK;
+}
+
+/* Record a bmap intent. */
+static int
+__xfs_bmap_add(
+	struct xfs_trans		*tp,
+	enum xfs_bmap_intent_type	type,
+	struct xfs_inode		*ip,
+	int				whichfork,
+	struct xfs_bmbt_irec		*bmap)
+{
+	struct xfs_bmap_intent		*bi;
+
+	trace_xfs_bmap_defer(tp->t_mountp,
+			XFS_FSB_TO_AGNO(tp->t_mountp, bmap->br_startblock),
+			type,
+			XFS_FSB_TO_AGBNO(tp->t_mountp, bmap->br_startblock),
+			ip->i_ino, whichfork,
+			bmap->br_startoff,
+			bmap->br_blockcount,
+			bmap->br_state);
+
+	bi = kmem_cache_alloc(xfs_bmap_intent_cache, GFP_NOFS | __GFP_NOFAIL);
+	INIT_LIST_HEAD(&bi->bi_list);
+	bi->bi_type = type;
+	bi->bi_owner = ip;
+	bi->bi_whichfork = whichfork;
+	bi->bi_bmap = *bmap;
+
+	xfs_defer_add(tp, XFS_DEFER_OPS_TYPE_BMAP, &bi->bi_list);
+	return 0;
+}
+
+/* Map an extent into a file. */
+void
+xfs_bmap_map_extent(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	struct xfs_bmbt_irec	*PREV)
+{
+	if (!xfs_bmap_is_update_needed(PREV))
+		return;
+
+	__xfs_bmap_add(tp, XFS_BMAP_MAP, ip, XFS_DATA_FORK, PREV);
+}
+
+/* Unmap an extent out of a file. */
+void
+xfs_bmap_unmap_extent(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	struct xfs_bmbt_irec	*PREV)
+{
+	if (!xfs_bmap_is_update_needed(PREV))
+		return;
+
+	__xfs_bmap_add(tp, XFS_BMAP_UNMAP, ip, XFS_DATA_FORK, PREV);
+}
+
+/*
+ * Process one of the deferred bmap operations.  We pass back the
+ * btree cursor to maintain our lock on the bmapbt between calls.
+ */
+int
+xfs_bmap_finish_one(
+	struct xfs_trans		*tp,
+	struct xfs_inode		*ip,
+	enum xfs_bmap_intent_type	type,
+	int				whichfork,
+	xfs_fileoff_t			startoff,
+	xfs_fsblock_t			startblock,
+	xfs_filblks_t			*blockcount,
+	xfs_exntst_t			state)
+{
+	int				error = 0;
+
+	ASSERT(tp->t_firstblock == NULLFSBLOCK);
+
+	trace_xfs_bmap_deferred(tp->t_mountp,
+			XFS_FSB_TO_AGNO(tp->t_mountp, startblock), type,
+			XFS_FSB_TO_AGBNO(tp->t_mountp, startblock),
+			ip->i_ino, whichfork, startoff, *blockcount, state);
+
+	if (WARN_ON_ONCE(whichfork != XFS_DATA_FORK))
+		return -EFSCORRUPTED;
+
+	if (XFS_TEST_ERROR(false, tp->t_mountp,
+			XFS_ERRTAG_BMAP_FINISH_ONE))
+		return -EIO;
+
+	switch (type) {
+	case XFS_BMAP_MAP:
+		error = xfs_bmapi_remap(tp, ip, startoff, *blockcount,
+				startblock, 0);
+		*blockcount = 0;
+		break;
+	case XFS_BMAP_UNMAP:
+		error = __xfs_bunmapi(tp, ip, startoff, blockcount,
+				XFS_BMAPI_REMAP, 1);
+		break;
+	default:
+		ASSERT(0);
+		error = -EFSCORRUPTED;
+	}
+
+	return error;
+}
+
+/* Check that an inode's extent does not have invalid flags or bad ranges. */
+xfs_failaddr_t
+xfs_bmap_validate_extent(
+	struct xfs_inode	*ip,
+	int			whichfork,
+	struct xfs_bmbt_irec	*irec)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+
+	if (!xfs_verify_fileext(mp, irec->br_startoff, irec->br_blockcount))
+		return __this_address;
+
+	if (XFS_IS_REALTIME_INODE(ip) && whichfork == XFS_DATA_FORK) {
+		if (!xfs_verify_rtext(mp, irec->br_startblock,
+					  irec->br_blockcount))
+			return __this_address;
+	} else {
+		if (!xfs_verify_fsbext(mp, irec->br_startblock,
+					   irec->br_blockcount))
+			return __this_address;
+	}
+	if (irec->br_state != XFS_EXT_NORM && whichfork != XFS_DATA_FORK)
+		return __this_address;
+	return NULL;
+}
+
+int __init
+xfs_bmap_intent_init_cache(void)
+{
+	xfs_bmap_intent_cache = kmem_cache_create("xfs_bmap_intent",
+			sizeof(struct xfs_bmap_intent),
+			0, 0, NULL);
+
+	return xfs_bmap_intent_cache != NULL ? 0 : -ENOMEM;
+}
+
+void
+xfs_bmap_intent_destroy_cache(void)
+{
+	kmem_cache_destroy(xfs_bmap_intent_cache);
+	xfs_bmap_intent_cache = NULL;
+}
diff --git a/fs/xfs/libxfs/xfs_bmap.h b/fs/xfs/libxfs/xfs_bmap.h
new file mode 100644
index 000000000..16db95b11
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_bmap.h
@@ -0,0 +1,270 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_BMAP_H__
+#define	__XFS_BMAP_H__
+
+struct getbmap;
+struct xfs_bmbt_irec;
+struct xfs_ifork;
+struct xfs_inode;
+struct xfs_mount;
+struct xfs_trans;
+
+/*
+ * Argument structure for xfs_bmap_alloc.
+ */
+struct xfs_bmalloca {
+	struct xfs_trans	*tp;	/* transaction pointer */
+	struct xfs_inode	*ip;	/* incore inode pointer */
+	struct xfs_bmbt_irec	prev;	/* extent before the new one */
+	struct xfs_bmbt_irec	got;	/* extent after, or delayed */
+
+	xfs_fileoff_t		offset;	/* offset in file filling in */
+	xfs_extlen_t		length;	/* i/o length asked/allocated */
+	xfs_fsblock_t		blkno;	/* starting block of new extent */
+
+	struct xfs_btree_cur	*cur;	/* btree cursor */
+	struct xfs_iext_cursor	icur;	/* incore extent cursor */
+	int			nallocs;/* number of extents alloc'd */
+	int			logflags;/* flags for transaction logging */
+
+	xfs_extlen_t		total;	/* total blocks needed for xaction */
+	xfs_extlen_t		minlen;	/* minimum allocation size (blocks) */
+	xfs_extlen_t		minleft; /* amount must be left after alloc */
+	bool			eof;	/* set if allocating past last extent */
+	bool			wasdel;	/* replacing a delayed allocation */
+	bool			aeof;	/* allocated space at eof */
+	bool			conv;	/* overwriting unwritten extents */
+	int			datatype;/* data type being allocated */
+	uint32_t		flags;
+};
+
+#define	XFS_BMAP_MAX_NMAP	4
+
+/*
+ * Flags for xfs_bmapi_*
+ */
+#define XFS_BMAPI_ENTIRE	(1u << 0) /* return entire extent untrimmed */
+#define XFS_BMAPI_METADATA	(1u << 1) /* mapping metadata not user data */
+#define XFS_BMAPI_ATTRFORK	(1u << 2) /* use attribute fork not data */
+#define XFS_BMAPI_PREALLOC	(1u << 3) /* preallocating unwritten space */
+#define XFS_BMAPI_CONTIG	(1u << 4) /* must allocate only one extent */
+/*
+ * unwritten extent conversion - this needs write cache flushing and no additional
+ * allocation alignments. When specified with XFS_BMAPI_PREALLOC it converts
+ * from written to unwritten, otherwise convert from unwritten to written.
+ */
+#define XFS_BMAPI_CONVERT	(1u << 5)
+
+/*
+ * allocate zeroed extents - this requires all newly allocated user data extents
+ * to be initialised to zero. It will be ignored if XFS_BMAPI_METADATA is set.
+ * Use in conjunction with XFS_BMAPI_CONVERT to convert unwritten extents found
+ * during the allocation range to zeroed written extents.
+ */
+#define XFS_BMAPI_ZERO		(1u << 6)
+
+/*
+ * Map the inode offset to the block given in ap->firstblock.  Primarily
+ * used for reflink.  The range must be in a hole, and this flag cannot be
+ * turned on with PREALLOC or CONVERT, and cannot be used on the attr fork.
+ *
+ * For bunmapi, this flag unmaps the range without adjusting quota, reducing
+ * refcount, or freeing the blocks.
+ */
+#define XFS_BMAPI_REMAP		(1u << 7)
+
+/* Map something in the CoW fork. */
+#define XFS_BMAPI_COWFORK	(1u << 8)
+
+/* Skip online discard of freed extents */
+#define XFS_BMAPI_NODISCARD	(1u << 9)
+
+/* Do not update the rmap btree.  Used for reconstructing bmbt from rmapbt. */
+#define XFS_BMAPI_NORMAP	(1u << 10)
+
+#define XFS_BMAPI_FLAGS \
+	{ XFS_BMAPI_ENTIRE,	"ENTIRE" }, \
+	{ XFS_BMAPI_METADATA,	"METADATA" }, \
+	{ XFS_BMAPI_ATTRFORK,	"ATTRFORK" }, \
+	{ XFS_BMAPI_PREALLOC,	"PREALLOC" }, \
+	{ XFS_BMAPI_CONTIG,	"CONTIG" }, \
+	{ XFS_BMAPI_CONVERT,	"CONVERT" }, \
+	{ XFS_BMAPI_ZERO,	"ZERO" }, \
+	{ XFS_BMAPI_REMAP,	"REMAP" }, \
+	{ XFS_BMAPI_COWFORK,	"COWFORK" }, \
+	{ XFS_BMAPI_NODISCARD,	"NODISCARD" }, \
+	{ XFS_BMAPI_NORMAP,	"NORMAP" }
+
+
+static inline int xfs_bmapi_aflag(int w)
+{
+	return (w == XFS_ATTR_FORK ? XFS_BMAPI_ATTRFORK :
+	       (w == XFS_COW_FORK ? XFS_BMAPI_COWFORK : 0));
+}
+
+static inline int xfs_bmapi_whichfork(uint32_t bmapi_flags)
+{
+	if (bmapi_flags & XFS_BMAPI_COWFORK)
+		return XFS_COW_FORK;
+	else if (bmapi_flags & XFS_BMAPI_ATTRFORK)
+		return XFS_ATTR_FORK;
+	return XFS_DATA_FORK;
+}
+
+/*
+ * Special values for xfs_bmbt_irec_t br_startblock field.
+ */
+#define	DELAYSTARTBLOCK		((xfs_fsblock_t)-1LL)
+#define	HOLESTARTBLOCK		((xfs_fsblock_t)-2LL)
+
+/*
+ * Flags for xfs_bmap_add_extent*.
+ */
+#define BMAP_LEFT_CONTIG	(1u << 0)
+#define BMAP_RIGHT_CONTIG	(1u << 1)
+#define BMAP_LEFT_FILLING	(1u << 2)
+#define BMAP_RIGHT_FILLING	(1u << 3)
+#define BMAP_LEFT_DELAY		(1u << 4)
+#define BMAP_RIGHT_DELAY	(1u << 5)
+#define BMAP_LEFT_VALID		(1u << 6)
+#define BMAP_RIGHT_VALID	(1u << 7)
+#define BMAP_ATTRFORK		(1u << 8)
+#define BMAP_COWFORK		(1u << 9)
+
+#define XFS_BMAP_EXT_FLAGS \
+	{ BMAP_LEFT_CONTIG,	"LC" }, \
+	{ BMAP_RIGHT_CONTIG,	"RC" }, \
+	{ BMAP_LEFT_FILLING,	"LF" }, \
+	{ BMAP_RIGHT_FILLING,	"RF" }, \
+	{ BMAP_ATTRFORK,	"ATTR" }, \
+	{ BMAP_COWFORK,		"COW" }
+
+/* Return true if the extent is an allocated extent, written or not. */
+static inline bool xfs_bmap_is_real_extent(struct xfs_bmbt_irec *irec)
+{
+	return irec->br_startblock != HOLESTARTBLOCK &&
+		irec->br_startblock != DELAYSTARTBLOCK &&
+		!isnullstartblock(irec->br_startblock);
+}
+
+/*
+ * Return true if the extent is a real, allocated extent, or false if it is  a
+ * delayed allocation, and unwritten extent or a hole.
+ */
+static inline bool xfs_bmap_is_written_extent(struct xfs_bmbt_irec *irec)
+{
+	return xfs_bmap_is_real_extent(irec) &&
+	       irec->br_state != XFS_EXT_UNWRITTEN;
+}
+
+/*
+ * Check the mapping for obviously garbage allocations that could trash the
+ * filesystem immediately.
+ */
+#define xfs_valid_startblock(ip, startblock) \
+	((startblock) != 0 || XFS_IS_REALTIME_INODE(ip))
+
+void	xfs_trim_extent(struct xfs_bmbt_irec *irec, xfs_fileoff_t bno,
+		xfs_filblks_t len);
+unsigned int xfs_bmap_compute_attr_offset(struct xfs_mount *mp);
+int	xfs_bmap_add_attrfork(struct xfs_inode *ip, int size, int rsvd);
+void	xfs_bmap_local_to_extents_empty(struct xfs_trans *tp,
+		struct xfs_inode *ip, int whichfork);
+void	xfs_bmap_compute_maxlevels(struct xfs_mount *mp, int whichfork);
+int	xfs_bmap_first_unused(struct xfs_trans *tp, struct xfs_inode *ip,
+		xfs_extlen_t len, xfs_fileoff_t *unused, int whichfork);
+int	xfs_bmap_last_before(struct xfs_trans *tp, struct xfs_inode *ip,
+		xfs_fileoff_t *last_block, int whichfork);
+int	xfs_bmap_last_offset(struct xfs_inode *ip, xfs_fileoff_t *unused,
+		int whichfork);
+int	xfs_bmapi_read(struct xfs_inode *ip, xfs_fileoff_t bno,
+		xfs_filblks_t len, struct xfs_bmbt_irec *mval,
+		int *nmap, uint32_t flags);
+int	xfs_bmapi_write(struct xfs_trans *tp, struct xfs_inode *ip,
+		xfs_fileoff_t bno, xfs_filblks_t len, uint32_t flags,
+		xfs_extlen_t total, struct xfs_bmbt_irec *mval, int *nmap);
+int	__xfs_bunmapi(struct xfs_trans *tp, struct xfs_inode *ip,
+		xfs_fileoff_t bno, xfs_filblks_t *rlen, uint32_t flags,
+		xfs_extnum_t nexts);
+int	xfs_bunmapi(struct xfs_trans *tp, struct xfs_inode *ip,
+		xfs_fileoff_t bno, xfs_filblks_t len, uint32_t flags,
+		xfs_extnum_t nexts, int *done);
+int	xfs_bmap_del_extent_delay(struct xfs_inode *ip, int whichfork,
+		struct xfs_iext_cursor *cur, struct xfs_bmbt_irec *got,
+		struct xfs_bmbt_irec *del);
+void	xfs_bmap_del_extent_cow(struct xfs_inode *ip,
+		struct xfs_iext_cursor *cur, struct xfs_bmbt_irec *got,
+		struct xfs_bmbt_irec *del);
+uint	xfs_default_attroffset(struct xfs_inode *ip);
+int	xfs_bmap_collapse_extents(struct xfs_trans *tp, struct xfs_inode *ip,
+		xfs_fileoff_t *next_fsb, xfs_fileoff_t offset_shift_fsb,
+		bool *done);
+int	xfs_bmap_can_insert_extents(struct xfs_inode *ip, xfs_fileoff_t off,
+		xfs_fileoff_t shift);
+int	xfs_bmap_insert_extents(struct xfs_trans *tp, struct xfs_inode *ip,
+		xfs_fileoff_t *next_fsb, xfs_fileoff_t offset_shift_fsb,
+		bool *done, xfs_fileoff_t stop_fsb);
+int	xfs_bmap_split_extent(struct xfs_trans *tp, struct xfs_inode *ip,
+		xfs_fileoff_t split_offset);
+int	xfs_bmapi_reserve_delalloc(struct xfs_inode *ip, int whichfork,
+		xfs_fileoff_t off, xfs_filblks_t len, xfs_filblks_t prealloc,
+		struct xfs_bmbt_irec *got, struct xfs_iext_cursor *cur,
+		int eof);
+int	xfs_bmapi_convert_delalloc(struct xfs_inode *ip, int whichfork,
+		xfs_off_t offset, struct iomap *iomap, unsigned int *seq);
+int	xfs_bmap_add_extent_unwritten_real(struct xfs_trans *tp,
+		struct xfs_inode *ip, int whichfork,
+		struct xfs_iext_cursor *icur, struct xfs_btree_cur **curp,
+		struct xfs_bmbt_irec *new, int *logflagsp);
+
+enum xfs_bmap_intent_type {
+	XFS_BMAP_MAP = 1,
+	XFS_BMAP_UNMAP,
+};
+
+struct xfs_bmap_intent {
+	struct list_head			bi_list;
+	enum xfs_bmap_intent_type		bi_type;
+	int					bi_whichfork;
+	struct xfs_inode			*bi_owner;
+	struct xfs_bmbt_irec			bi_bmap;
+};
+
+int	xfs_bmap_finish_one(struct xfs_trans *tp, struct xfs_inode *ip,
+		enum xfs_bmap_intent_type type, int whichfork,
+		xfs_fileoff_t startoff, xfs_fsblock_t startblock,
+		xfs_filblks_t *blockcount, xfs_exntst_t state);
+void	xfs_bmap_map_extent(struct xfs_trans *tp, struct xfs_inode *ip,
+		struct xfs_bmbt_irec *imap);
+void	xfs_bmap_unmap_extent(struct xfs_trans *tp, struct xfs_inode *ip,
+		struct xfs_bmbt_irec *imap);
+
+static inline uint32_t xfs_bmap_fork_to_state(int whichfork)
+{
+	switch (whichfork) {
+	case XFS_ATTR_FORK:
+		return BMAP_ATTRFORK;
+	case XFS_COW_FORK:
+		return BMAP_COWFORK;
+	default:
+		return 0;
+	}
+}
+
+xfs_failaddr_t xfs_bmap_validate_extent(struct xfs_inode *ip, int whichfork,
+		struct xfs_bmbt_irec *irec);
+
+int	xfs_bmapi_remap(struct xfs_trans *tp, struct xfs_inode *ip,
+		xfs_fileoff_t bno, xfs_filblks_t len, xfs_fsblock_t startblock,
+		uint32_t flags);
+
+extern struct kmem_cache	*xfs_bmap_intent_cache;
+
+int __init xfs_bmap_intent_init_cache(void);
+void xfs_bmap_intent_destroy_cache(void);
+
+#endif	/* __XFS_BMAP_H__ */
diff --git a/fs/xfs/libxfs/xfs_bmap_btree.c b/fs/xfs/libxfs/xfs_bmap_btree.c
new file mode 100644
index 000000000..cfa052d40
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_bmap_btree.c
@@ -0,0 +1,703 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_bit.h"
+#include "xfs_mount.h"
+#include "xfs_inode.h"
+#include "xfs_trans.h"
+#include "xfs_alloc.h"
+#include "xfs_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_bmap.h"
+#include "xfs_error.h"
+#include "xfs_quota.h"
+#include "xfs_trace.h"
+#include "xfs_rmap.h"
+
+static struct kmem_cache	*xfs_bmbt_cur_cache;
+
+/*
+ * Convert on-disk form of btree root to in-memory form.
+ */
+void
+xfs_bmdr_to_bmbt(
+	struct xfs_inode	*ip,
+	xfs_bmdr_block_t	*dblock,
+	int			dblocklen,
+	struct xfs_btree_block	*rblock,
+	int			rblocklen)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	int			dmxr;
+	xfs_bmbt_key_t		*fkp;
+	__be64			*fpp;
+	xfs_bmbt_key_t		*tkp;
+	__be64			*tpp;
+
+	xfs_btree_init_block_int(mp, rblock, XFS_BUF_DADDR_NULL,
+				 XFS_BTNUM_BMAP, 0, 0, ip->i_ino,
+				 XFS_BTREE_LONG_PTRS);
+	rblock->bb_level = dblock->bb_level;
+	ASSERT(be16_to_cpu(rblock->bb_level) > 0);
+	rblock->bb_numrecs = dblock->bb_numrecs;
+	dmxr = xfs_bmdr_maxrecs(dblocklen, 0);
+	fkp = XFS_BMDR_KEY_ADDR(dblock, 1);
+	tkp = XFS_BMBT_KEY_ADDR(mp, rblock, 1);
+	fpp = XFS_BMDR_PTR_ADDR(dblock, 1, dmxr);
+	tpp = XFS_BMAP_BROOT_PTR_ADDR(mp, rblock, 1, rblocklen);
+	dmxr = be16_to_cpu(dblock->bb_numrecs);
+	memcpy(tkp, fkp, sizeof(*fkp) * dmxr);
+	memcpy(tpp, fpp, sizeof(*fpp) * dmxr);
+}
+
+void
+xfs_bmbt_disk_get_all(
+	const struct xfs_bmbt_rec *rec,
+	struct xfs_bmbt_irec	*irec)
+{
+	uint64_t		l0 = get_unaligned_be64(&rec->l0);
+	uint64_t		l1 = get_unaligned_be64(&rec->l1);
+
+	irec->br_startoff = (l0 & xfs_mask64lo(64 - BMBT_EXNTFLAG_BITLEN)) >> 9;
+	irec->br_startblock = ((l0 & xfs_mask64lo(9)) << 43) | (l1 >> 21);
+	irec->br_blockcount = l1 & xfs_mask64lo(21);
+	if (l0 >> (64 - BMBT_EXNTFLAG_BITLEN))
+		irec->br_state = XFS_EXT_UNWRITTEN;
+	else
+		irec->br_state = XFS_EXT_NORM;
+}
+
+/*
+ * Extract the blockcount field from an on disk bmap extent record.
+ */
+xfs_filblks_t
+xfs_bmbt_disk_get_blockcount(
+	const struct xfs_bmbt_rec	*r)
+{
+	return (xfs_filblks_t)(be64_to_cpu(r->l1) & xfs_mask64lo(21));
+}
+
+/*
+ * Extract the startoff field from a disk format bmap extent record.
+ */
+xfs_fileoff_t
+xfs_bmbt_disk_get_startoff(
+	const struct xfs_bmbt_rec	*r)
+{
+	return ((xfs_fileoff_t)be64_to_cpu(r->l0) &
+		 xfs_mask64lo(64 - BMBT_EXNTFLAG_BITLEN)) >> 9;
+}
+
+/*
+ * Set all the fields in a bmap extent record from the uncompressed form.
+ */
+void
+xfs_bmbt_disk_set_all(
+	struct xfs_bmbt_rec	*r,
+	struct xfs_bmbt_irec	*s)
+{
+	int			extent_flag = (s->br_state != XFS_EXT_NORM);
+
+	ASSERT(s->br_state == XFS_EXT_NORM || s->br_state == XFS_EXT_UNWRITTEN);
+	ASSERT(!(s->br_startoff & xfs_mask64hi(64-BMBT_STARTOFF_BITLEN)));
+	ASSERT(!(s->br_blockcount & xfs_mask64hi(64-BMBT_BLOCKCOUNT_BITLEN)));
+	ASSERT(!(s->br_startblock & xfs_mask64hi(64-BMBT_STARTBLOCK_BITLEN)));
+
+	put_unaligned_be64(
+		((xfs_bmbt_rec_base_t)extent_flag << 63) |
+		 ((xfs_bmbt_rec_base_t)s->br_startoff << 9) |
+		 ((xfs_bmbt_rec_base_t)s->br_startblock >> 43), &r->l0);
+	put_unaligned_be64(
+		((xfs_bmbt_rec_base_t)s->br_startblock << 21) |
+		 ((xfs_bmbt_rec_base_t)s->br_blockcount &
+		  (xfs_bmbt_rec_base_t)xfs_mask64lo(21)), &r->l1);
+}
+
+/*
+ * Convert in-memory form of btree root to on-disk form.
+ */
+void
+xfs_bmbt_to_bmdr(
+	struct xfs_mount	*mp,
+	struct xfs_btree_block	*rblock,
+	int			rblocklen,
+	xfs_bmdr_block_t	*dblock,
+	int			dblocklen)
+{
+	int			dmxr;
+	xfs_bmbt_key_t		*fkp;
+	__be64			*fpp;
+	xfs_bmbt_key_t		*tkp;
+	__be64			*tpp;
+
+	if (xfs_has_crc(mp)) {
+		ASSERT(rblock->bb_magic == cpu_to_be32(XFS_BMAP_CRC_MAGIC));
+		ASSERT(uuid_equal(&rblock->bb_u.l.bb_uuid,
+		       &mp->m_sb.sb_meta_uuid));
+		ASSERT(rblock->bb_u.l.bb_blkno ==
+		       cpu_to_be64(XFS_BUF_DADDR_NULL));
+	} else
+		ASSERT(rblock->bb_magic == cpu_to_be32(XFS_BMAP_MAGIC));
+	ASSERT(rblock->bb_u.l.bb_leftsib == cpu_to_be64(NULLFSBLOCK));
+	ASSERT(rblock->bb_u.l.bb_rightsib == cpu_to_be64(NULLFSBLOCK));
+	ASSERT(rblock->bb_level != 0);
+	dblock->bb_level = rblock->bb_level;
+	dblock->bb_numrecs = rblock->bb_numrecs;
+	dmxr = xfs_bmdr_maxrecs(dblocklen, 0);
+	fkp = XFS_BMBT_KEY_ADDR(mp, rblock, 1);
+	tkp = XFS_BMDR_KEY_ADDR(dblock, 1);
+	fpp = XFS_BMAP_BROOT_PTR_ADDR(mp, rblock, 1, rblocklen);
+	tpp = XFS_BMDR_PTR_ADDR(dblock, 1, dmxr);
+	dmxr = be16_to_cpu(dblock->bb_numrecs);
+	memcpy(tkp, fkp, sizeof(*fkp) * dmxr);
+	memcpy(tpp, fpp, sizeof(*fpp) * dmxr);
+}
+
+STATIC struct xfs_btree_cur *
+xfs_bmbt_dup_cursor(
+	struct xfs_btree_cur	*cur)
+{
+	struct xfs_btree_cur	*new;
+
+	new = xfs_bmbt_init_cursor(cur->bc_mp, cur->bc_tp,
+			cur->bc_ino.ip, cur->bc_ino.whichfork);
+
+	/*
+	 * Copy the firstblock, dfops, and flags values,
+	 * since init cursor doesn't get them.
+	 */
+	new->bc_ino.flags = cur->bc_ino.flags;
+
+	return new;
+}
+
+STATIC void
+xfs_bmbt_update_cursor(
+	struct xfs_btree_cur	*src,
+	struct xfs_btree_cur	*dst)
+{
+	ASSERT((dst->bc_tp->t_firstblock != NULLFSBLOCK) ||
+	       (dst->bc_ino.ip->i_diflags & XFS_DIFLAG_REALTIME));
+
+	dst->bc_ino.allocated += src->bc_ino.allocated;
+	dst->bc_tp->t_firstblock = src->bc_tp->t_firstblock;
+
+	src->bc_ino.allocated = 0;
+}
+
+STATIC int
+xfs_bmbt_alloc_block(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*start,
+	union xfs_btree_ptr		*new,
+	int				*stat)
+{
+	xfs_alloc_arg_t		args;		/* block allocation args */
+	int			error;		/* error return value */
+
+	memset(&args, 0, sizeof(args));
+	args.tp = cur->bc_tp;
+	args.mp = cur->bc_mp;
+	args.fsbno = cur->bc_tp->t_firstblock;
+	xfs_rmap_ino_bmbt_owner(&args.oinfo, cur->bc_ino.ip->i_ino,
+			cur->bc_ino.whichfork);
+
+	if (args.fsbno == NULLFSBLOCK) {
+		args.fsbno = be64_to_cpu(start->l);
+		args.type = XFS_ALLOCTYPE_START_BNO;
+		/*
+		 * Make sure there is sufficient room left in the AG to
+		 * complete a full tree split for an extent insert.  If
+		 * we are converting the middle part of an extent then
+		 * we may need space for two tree splits.
+		 *
+		 * We are relying on the caller to make the correct block
+		 * reservation for this operation to succeed.  If the
+		 * reservation amount is insufficient then we may fail a
+		 * block allocation here and corrupt the filesystem.
+		 */
+		args.minleft = args.tp->t_blk_res;
+	} else if (cur->bc_tp->t_flags & XFS_TRANS_LOWMODE) {
+		args.type = XFS_ALLOCTYPE_START_BNO;
+	} else {
+		args.type = XFS_ALLOCTYPE_NEAR_BNO;
+	}
+
+	args.minlen = args.maxlen = args.prod = 1;
+	args.wasdel = cur->bc_ino.flags & XFS_BTCUR_BMBT_WASDEL;
+	if (!args.wasdel && args.tp->t_blk_res == 0) {
+		error = -ENOSPC;
+		goto error0;
+	}
+	error = xfs_alloc_vextent(&args);
+	if (error)
+		goto error0;
+
+	if (args.fsbno == NULLFSBLOCK && args.minleft) {
+		/*
+		 * Could not find an AG with enough free space to satisfy
+		 * a full btree split.  Try again and if
+		 * successful activate the lowspace algorithm.
+		 */
+		args.fsbno = 0;
+		args.type = XFS_ALLOCTYPE_FIRST_AG;
+		error = xfs_alloc_vextent(&args);
+		if (error)
+			goto error0;
+		cur->bc_tp->t_flags |= XFS_TRANS_LOWMODE;
+	}
+	if (WARN_ON_ONCE(args.fsbno == NULLFSBLOCK)) {
+		*stat = 0;
+		return 0;
+	}
+
+	ASSERT(args.len == 1);
+	cur->bc_tp->t_firstblock = args.fsbno;
+	cur->bc_ino.allocated++;
+	cur->bc_ino.ip->i_nblocks++;
+	xfs_trans_log_inode(args.tp, cur->bc_ino.ip, XFS_ILOG_CORE);
+	xfs_trans_mod_dquot_byino(args.tp, cur->bc_ino.ip,
+			XFS_TRANS_DQ_BCOUNT, 1L);
+
+	new->l = cpu_to_be64(args.fsbno);
+
+	*stat = 1;
+	return 0;
+
+ error0:
+	return error;
+}
+
+STATIC int
+xfs_bmbt_free_block(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = cur->bc_mp;
+	struct xfs_inode	*ip = cur->bc_ino.ip;
+	struct xfs_trans	*tp = cur->bc_tp;
+	xfs_fsblock_t		fsbno = XFS_DADDR_TO_FSB(mp, xfs_buf_daddr(bp));
+	struct xfs_owner_info	oinfo;
+
+	xfs_rmap_ino_bmbt_owner(&oinfo, ip->i_ino, cur->bc_ino.whichfork);
+	xfs_free_extent_later(cur->bc_tp, fsbno, 1, &oinfo);
+	ip->i_nblocks--;
+
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+	xfs_trans_mod_dquot_byino(tp, ip, XFS_TRANS_DQ_BCOUNT, -1L);
+	return 0;
+}
+
+STATIC int
+xfs_bmbt_get_minrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	if (level == cur->bc_nlevels - 1) {
+		struct xfs_ifork	*ifp;
+
+		ifp = xfs_ifork_ptr(cur->bc_ino.ip,
+				    cur->bc_ino.whichfork);
+
+		return xfs_bmbt_maxrecs(cur->bc_mp,
+					ifp->if_broot_bytes, level == 0) / 2;
+	}
+
+	return cur->bc_mp->m_bmap_dmnr[level != 0];
+}
+
+int
+xfs_bmbt_get_maxrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	if (level == cur->bc_nlevels - 1) {
+		struct xfs_ifork	*ifp;
+
+		ifp = xfs_ifork_ptr(cur->bc_ino.ip,
+				    cur->bc_ino.whichfork);
+
+		return xfs_bmbt_maxrecs(cur->bc_mp,
+					ifp->if_broot_bytes, level == 0);
+	}
+
+	return cur->bc_mp->m_bmap_dmxr[level != 0];
+
+}
+
+/*
+ * Get the maximum records we could store in the on-disk format.
+ *
+ * For non-root nodes this is equivalent to xfs_bmbt_get_maxrecs, but
+ * for the root node this checks the available space in the dinode fork
+ * so that we can resize the in-memory buffer to match it.  After a
+ * resize to the maximum size this function returns the same value
+ * as xfs_bmbt_get_maxrecs for the root node, too.
+ */
+STATIC int
+xfs_bmbt_get_dmaxrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	if (level != cur->bc_nlevels - 1)
+		return cur->bc_mp->m_bmap_dmxr[level != 0];
+	return xfs_bmdr_maxrecs(cur->bc_ino.forksize, level == 0);
+}
+
+STATIC void
+xfs_bmbt_init_key_from_rec(
+	union xfs_btree_key		*key,
+	const union xfs_btree_rec	*rec)
+{
+	key->bmbt.br_startoff =
+		cpu_to_be64(xfs_bmbt_disk_get_startoff(&rec->bmbt));
+}
+
+STATIC void
+xfs_bmbt_init_high_key_from_rec(
+	union xfs_btree_key		*key,
+	const union xfs_btree_rec	*rec)
+{
+	key->bmbt.br_startoff = cpu_to_be64(
+			xfs_bmbt_disk_get_startoff(&rec->bmbt) +
+			xfs_bmbt_disk_get_blockcount(&rec->bmbt) - 1);
+}
+
+STATIC void
+xfs_bmbt_init_rec_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec)
+{
+	xfs_bmbt_disk_set_all(&rec->bmbt, &cur->bc_rec.b);
+}
+
+STATIC void
+xfs_bmbt_init_ptr_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	ptr->l = 0;
+}
+
+STATIC int64_t
+xfs_bmbt_key_diff(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*key)
+{
+	return (int64_t)be64_to_cpu(key->bmbt.br_startoff) -
+				      cur->bc_rec.b.br_startoff;
+}
+
+STATIC int64_t
+xfs_bmbt_diff_two_keys(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*k1,
+	const union xfs_btree_key	*k2)
+{
+	uint64_t			a = be64_to_cpu(k1->bmbt.br_startoff);
+	uint64_t			b = be64_to_cpu(k2->bmbt.br_startoff);
+
+	/*
+	 * Note: This routine previously casted a and b to int64 and subtracted
+	 * them to generate a result.  This lead to problems if b was the
+	 * "maximum" key value (all ones) being signed incorrectly, hence this
+	 * somewhat less efficient version.
+	 */
+	if (a > b)
+		return 1;
+	if (b > a)
+		return -1;
+	return 0;
+}
+
+static xfs_failaddr_t
+xfs_bmbt_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
+	xfs_failaddr_t		fa;
+	unsigned int		level;
+
+	if (!xfs_verify_magic(bp, block->bb_magic))
+		return __this_address;
+
+	if (xfs_has_crc(mp)) {
+		/*
+		 * XXX: need a better way of verifying the owner here. Right now
+		 * just make sure there has been one set.
+		 */
+		fa = xfs_btree_lblock_v5hdr_verify(bp, XFS_RMAP_OWN_UNKNOWN);
+		if (fa)
+			return fa;
+	}
+
+	/*
+	 * numrecs and level verification.
+	 *
+	 * We don't know what fork we belong to, so just verify that the level
+	 * is less than the maximum of the two. Later checks will be more
+	 * precise.
+	 */
+	level = be16_to_cpu(block->bb_level);
+	if (level > max(mp->m_bm_maxlevels[0], mp->m_bm_maxlevels[1]))
+		return __this_address;
+
+	return xfs_btree_lblock_verify(bp, mp->m_bmap_dmxr[level != 0]);
+}
+
+static void
+xfs_bmbt_read_verify(
+	struct xfs_buf	*bp)
+{
+	xfs_failaddr_t	fa;
+
+	if (!xfs_btree_lblock_verify_crc(bp))
+		xfs_verifier_error(bp, -EFSBADCRC, __this_address);
+	else {
+		fa = xfs_bmbt_verify(bp);
+		if (fa)
+			xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+	}
+
+	if (bp->b_error)
+		trace_xfs_btree_corrupt(bp, _RET_IP_);
+}
+
+static void
+xfs_bmbt_write_verify(
+	struct xfs_buf	*bp)
+{
+	xfs_failaddr_t	fa;
+
+	fa = xfs_bmbt_verify(bp);
+	if (fa) {
+		trace_xfs_btree_corrupt(bp, _RET_IP_);
+		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+		return;
+	}
+	xfs_btree_lblock_calc_crc(bp);
+}
+
+const struct xfs_buf_ops xfs_bmbt_buf_ops = {
+	.name = "xfs_bmbt",
+	.magic = { cpu_to_be32(XFS_BMAP_MAGIC),
+		   cpu_to_be32(XFS_BMAP_CRC_MAGIC) },
+	.verify_read = xfs_bmbt_read_verify,
+	.verify_write = xfs_bmbt_write_verify,
+	.verify_struct = xfs_bmbt_verify,
+};
+
+
+STATIC int
+xfs_bmbt_keys_inorder(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*k1,
+	const union xfs_btree_key	*k2)
+{
+	return be64_to_cpu(k1->bmbt.br_startoff) <
+		be64_to_cpu(k2->bmbt.br_startoff);
+}
+
+STATIC int
+xfs_bmbt_recs_inorder(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_rec	*r1,
+	const union xfs_btree_rec	*r2)
+{
+	return xfs_bmbt_disk_get_startoff(&r1->bmbt) +
+		xfs_bmbt_disk_get_blockcount(&r1->bmbt) <=
+		xfs_bmbt_disk_get_startoff(&r2->bmbt);
+}
+
+static const struct xfs_btree_ops xfs_bmbt_ops = {
+	.rec_len		= sizeof(xfs_bmbt_rec_t),
+	.key_len		= sizeof(xfs_bmbt_key_t),
+
+	.dup_cursor		= xfs_bmbt_dup_cursor,
+	.update_cursor		= xfs_bmbt_update_cursor,
+	.alloc_block		= xfs_bmbt_alloc_block,
+	.free_block		= xfs_bmbt_free_block,
+	.get_maxrecs		= xfs_bmbt_get_maxrecs,
+	.get_minrecs		= xfs_bmbt_get_minrecs,
+	.get_dmaxrecs		= xfs_bmbt_get_dmaxrecs,
+	.init_key_from_rec	= xfs_bmbt_init_key_from_rec,
+	.init_high_key_from_rec	= xfs_bmbt_init_high_key_from_rec,
+	.init_rec_from_cur	= xfs_bmbt_init_rec_from_cur,
+	.init_ptr_from_cur	= xfs_bmbt_init_ptr_from_cur,
+	.key_diff		= xfs_bmbt_key_diff,
+	.diff_two_keys		= xfs_bmbt_diff_two_keys,
+	.buf_ops		= &xfs_bmbt_buf_ops,
+	.keys_inorder		= xfs_bmbt_keys_inorder,
+	.recs_inorder		= xfs_bmbt_recs_inorder,
+};
+
+/*
+ * Allocate a new bmap btree cursor.
+ */
+struct xfs_btree_cur *				/* new bmap btree cursor */
+xfs_bmbt_init_cursor(
+	struct xfs_mount	*mp,		/* file system mount point */
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* inode owning the btree */
+	int			whichfork)	/* data or attr fork */
+{
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_btree_cur	*cur;
+	ASSERT(whichfork != XFS_COW_FORK);
+
+	cur = xfs_btree_alloc_cursor(mp, tp, XFS_BTNUM_BMAP,
+			mp->m_bm_maxlevels[whichfork], xfs_bmbt_cur_cache);
+	cur->bc_nlevels = be16_to_cpu(ifp->if_broot->bb_level) + 1;
+	cur->bc_statoff = XFS_STATS_CALC_INDEX(xs_bmbt_2);
+
+	cur->bc_ops = &xfs_bmbt_ops;
+	cur->bc_flags = XFS_BTREE_LONG_PTRS | XFS_BTREE_ROOT_IN_INODE;
+	if (xfs_has_crc(mp))
+		cur->bc_flags |= XFS_BTREE_CRC_BLOCKS;
+
+	cur->bc_ino.forksize = xfs_inode_fork_size(ip, whichfork);
+	cur->bc_ino.ip = ip;
+	cur->bc_ino.allocated = 0;
+	cur->bc_ino.flags = 0;
+	cur->bc_ino.whichfork = whichfork;
+
+	return cur;
+}
+
+/* Calculate number of records in a block mapping btree block. */
+static inline unsigned int
+xfs_bmbt_block_maxrecs(
+	unsigned int		blocklen,
+	bool			leaf)
+{
+	if (leaf)
+		return blocklen / sizeof(xfs_bmbt_rec_t);
+	return blocklen / (sizeof(xfs_bmbt_key_t) + sizeof(xfs_bmbt_ptr_t));
+}
+
+/*
+ * Calculate number of records in a bmap btree block.
+ */
+int
+xfs_bmbt_maxrecs(
+	struct xfs_mount	*mp,
+	int			blocklen,
+	int			leaf)
+{
+	blocklen -= XFS_BMBT_BLOCK_LEN(mp);
+	return xfs_bmbt_block_maxrecs(blocklen, leaf);
+}
+
+/*
+ * Calculate the maximum possible height of the btree that the on-disk format
+ * supports. This is used for sizing structures large enough to support every
+ * possible configuration of a filesystem that might get mounted.
+ */
+unsigned int
+xfs_bmbt_maxlevels_ondisk(void)
+{
+	unsigned int		minrecs[2];
+	unsigned int		blocklen;
+
+	blocklen = min(XFS_MIN_BLOCKSIZE - XFS_BTREE_SBLOCK_LEN,
+		       XFS_MIN_CRC_BLOCKSIZE - XFS_BTREE_SBLOCK_CRC_LEN);
+
+	minrecs[0] = xfs_bmbt_block_maxrecs(blocklen, true) / 2;
+	minrecs[1] = xfs_bmbt_block_maxrecs(blocklen, false) / 2;
+
+	/* One extra level for the inode root. */
+	return xfs_btree_compute_maxlevels(minrecs,
+			XFS_MAX_EXTCNT_DATA_FORK_LARGE) + 1;
+}
+
+/*
+ * Calculate number of records in a bmap btree inode root.
+ */
+int
+xfs_bmdr_maxrecs(
+	int			blocklen,
+	int			leaf)
+{
+	blocklen -= sizeof(xfs_bmdr_block_t);
+
+	if (leaf)
+		return blocklen / sizeof(xfs_bmdr_rec_t);
+	return blocklen / (sizeof(xfs_bmdr_key_t) + sizeof(xfs_bmdr_ptr_t));
+}
+
+/*
+ * Change the owner of a btree format fork fo the inode passed in. Change it to
+ * the owner of that is passed in so that we can change owners before or after
+ * we switch forks between inodes. The operation that the caller is doing will
+ * determine whether is needs to change owner before or after the switch.
+ *
+ * For demand paged transactional modification, the fork switch should be done
+ * after reading in all the blocks, modifying them and pinning them in the
+ * transaction. For modification when the buffers are already pinned in memory,
+ * the fork switch can be done before changing the owner as we won't need to
+ * validate the owner until the btree buffers are unpinned and writes can occur
+ * again.
+ *
+ * For recovery based ownership change, there is no transactional context and
+ * so a buffer list must be supplied so that we can record the buffers that we
+ * modified for the caller to issue IO on.
+ */
+int
+xfs_bmbt_change_owner(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	int			whichfork,
+	xfs_ino_t		new_owner,
+	struct list_head	*buffer_list)
+{
+	struct xfs_btree_cur	*cur;
+	int			error;
+
+	ASSERT(tp || buffer_list);
+	ASSERT(!(tp && buffer_list));
+	ASSERT(xfs_ifork_ptr(ip, whichfork)->if_format == XFS_DINODE_FMT_BTREE);
+
+	cur = xfs_bmbt_init_cursor(ip->i_mount, tp, ip, whichfork);
+	cur->bc_ino.flags |= XFS_BTCUR_BMBT_INVALID_OWNER;
+
+	error = xfs_btree_change_owner(cur, new_owner, buffer_list);
+	xfs_btree_del_cursor(cur, error);
+	return error;
+}
+
+/* Calculate the bmap btree size for some records. */
+unsigned long long
+xfs_bmbt_calc_size(
+	struct xfs_mount	*mp,
+	unsigned long long	len)
+{
+	return xfs_btree_calc_size(mp->m_bmap_dmnr, len);
+}
+
+int __init
+xfs_bmbt_init_cur_cache(void)
+{
+	xfs_bmbt_cur_cache = kmem_cache_create("xfs_bmbt_cur",
+			xfs_btree_cur_sizeof(xfs_bmbt_maxlevels_ondisk()),
+			0, 0, NULL);
+
+	if (!xfs_bmbt_cur_cache)
+		return -ENOMEM;
+	return 0;
+}
+
+void
+xfs_bmbt_destroy_cur_cache(void)
+{
+	kmem_cache_destroy(xfs_bmbt_cur_cache);
+	xfs_bmbt_cur_cache = NULL;
+}
diff --git a/fs/xfs/libxfs/xfs_bmap_btree.h b/fs/xfs/libxfs/xfs_bmap_btree.h
new file mode 100644
index 000000000..3e7a40a83
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_bmap_btree.h
@@ -0,0 +1,118 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2000,2002-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_BMAP_BTREE_H__
+#define __XFS_BMAP_BTREE_H__
+
+struct xfs_btree_cur;
+struct xfs_btree_block;
+struct xfs_mount;
+struct xfs_inode;
+struct xfs_trans;
+
+/*
+ * Btree block header size depends on a superblock flag.
+ */
+#define XFS_BMBT_BLOCK_LEN(mp) \
+	(xfs_has_crc(((mp))) ? \
+		XFS_BTREE_LBLOCK_CRC_LEN : XFS_BTREE_LBLOCK_LEN)
+
+#define XFS_BMBT_REC_ADDR(mp, block, index) \
+	((xfs_bmbt_rec_t *) \
+		((char *)(block) + \
+		 XFS_BMBT_BLOCK_LEN(mp) + \
+		 ((index) - 1) * sizeof(xfs_bmbt_rec_t)))
+
+#define XFS_BMBT_KEY_ADDR(mp, block, index) \
+	((xfs_bmbt_key_t *) \
+		((char *)(block) + \
+		 XFS_BMBT_BLOCK_LEN(mp) + \
+		 ((index) - 1) * sizeof(xfs_bmbt_key_t)))
+
+#define XFS_BMBT_PTR_ADDR(mp, block, index, maxrecs) \
+	((xfs_bmbt_ptr_t *) \
+		((char *)(block) + \
+		 XFS_BMBT_BLOCK_LEN(mp) + \
+		 (maxrecs) * sizeof(xfs_bmbt_key_t) + \
+		 ((index) - 1) * sizeof(xfs_bmbt_ptr_t)))
+
+#define XFS_BMDR_REC_ADDR(block, index) \
+	((xfs_bmdr_rec_t *) \
+		((char *)(block) + \
+		 sizeof(struct xfs_bmdr_block) + \
+	         ((index) - 1) * sizeof(xfs_bmdr_rec_t)))
+
+#define XFS_BMDR_KEY_ADDR(block, index) \
+	((xfs_bmdr_key_t *) \
+		((char *)(block) + \
+		 sizeof(struct xfs_bmdr_block) + \
+		 ((index) - 1) * sizeof(xfs_bmdr_key_t)))
+
+#define XFS_BMDR_PTR_ADDR(block, index, maxrecs) \
+	((xfs_bmdr_ptr_t *) \
+		((char *)(block) + \
+		 sizeof(struct xfs_bmdr_block) + \
+		 (maxrecs) * sizeof(xfs_bmdr_key_t) + \
+		 ((index) - 1) * sizeof(xfs_bmdr_ptr_t)))
+
+/*
+ * These are to be used when we know the size of the block and
+ * we don't have a cursor.
+ */
+#define XFS_BMAP_BROOT_PTR_ADDR(mp, bb, i, sz) \
+	XFS_BMBT_PTR_ADDR(mp, bb, i, xfs_bmbt_maxrecs(mp, sz, 0))
+
+#define XFS_BMAP_BROOT_SPACE_CALC(mp, nrecs) \
+	(int)(XFS_BMBT_BLOCK_LEN(mp) + \
+	       ((nrecs) * (sizeof(xfs_bmbt_key_t) + sizeof(xfs_bmbt_ptr_t))))
+
+#define XFS_BMAP_BROOT_SPACE(mp, bb) \
+	(XFS_BMAP_BROOT_SPACE_CALC(mp, be16_to_cpu((bb)->bb_numrecs)))
+#define XFS_BMDR_SPACE_CALC(nrecs) \
+	(int)(sizeof(xfs_bmdr_block_t) + \
+	       ((nrecs) * (sizeof(xfs_bmbt_key_t) + sizeof(xfs_bmbt_ptr_t))))
+#define XFS_BMAP_BMDR_SPACE(bb) \
+	(XFS_BMDR_SPACE_CALC(be16_to_cpu((bb)->bb_numrecs)))
+
+/*
+ * Maximum number of bmap btree levels.
+ */
+#define XFS_BM_MAXLEVELS(mp,w)		((mp)->m_bm_maxlevels[(w)])
+
+/*
+ * Prototypes for xfs_bmap.c to call.
+ */
+extern void xfs_bmdr_to_bmbt(struct xfs_inode *, xfs_bmdr_block_t *, int,
+			struct xfs_btree_block *, int);
+
+void xfs_bmbt_disk_set_all(struct xfs_bmbt_rec *r, struct xfs_bmbt_irec *s);
+extern xfs_filblks_t xfs_bmbt_disk_get_blockcount(const struct xfs_bmbt_rec *r);
+extern xfs_fileoff_t xfs_bmbt_disk_get_startoff(const struct xfs_bmbt_rec *r);
+void xfs_bmbt_disk_get_all(const struct xfs_bmbt_rec *r,
+		struct xfs_bmbt_irec *s);
+
+extern void xfs_bmbt_to_bmdr(struct xfs_mount *, struct xfs_btree_block *, int,
+			xfs_bmdr_block_t *, int);
+
+extern int xfs_bmbt_get_maxrecs(struct xfs_btree_cur *, int level);
+extern int xfs_bmdr_maxrecs(int blocklen, int leaf);
+extern int xfs_bmbt_maxrecs(struct xfs_mount *, int blocklen, int leaf);
+
+extern int xfs_bmbt_change_owner(struct xfs_trans *tp, struct xfs_inode *ip,
+				 int whichfork, xfs_ino_t new_owner,
+				 struct list_head *buffer_list);
+
+extern struct xfs_btree_cur *xfs_bmbt_init_cursor(struct xfs_mount *,
+		struct xfs_trans *, struct xfs_inode *, int);
+
+extern unsigned long long xfs_bmbt_calc_size(struct xfs_mount *mp,
+		unsigned long long len);
+
+unsigned int xfs_bmbt_maxlevels_ondisk(void);
+
+int __init xfs_bmbt_init_cur_cache(void);
+void xfs_bmbt_destroy_cur_cache(void);
+
+#endif	/* __XFS_BMAP_BTREE_H__ */
diff --git a/fs/xfs/libxfs/xfs_btree.c b/fs/xfs/libxfs/xfs_btree.c
new file mode 100644
index 000000000..4c16c8c31
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_btree.c
@@ -0,0 +1,5099 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_bit.h"
+#include "xfs_mount.h"
+#include "xfs_inode.h"
+#include "xfs_trans.h"
+#include "xfs_buf_item.h"
+#include "xfs_btree.h"
+#include "xfs_errortag.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+#include "xfs_alloc.h"
+#include "xfs_log.h"
+#include "xfs_btree_staging.h"
+#include "xfs_ag.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_rmap_btree.h"
+#include "xfs_refcount_btree.h"
+
+/*
+ * Btree magic numbers.
+ */
+static const uint32_t xfs_magics[2][XFS_BTNUM_MAX] = {
+	{ XFS_ABTB_MAGIC, XFS_ABTC_MAGIC, 0, XFS_BMAP_MAGIC, XFS_IBT_MAGIC,
+	  XFS_FIBT_MAGIC, 0 },
+	{ XFS_ABTB_CRC_MAGIC, XFS_ABTC_CRC_MAGIC, XFS_RMAP_CRC_MAGIC,
+	  XFS_BMAP_CRC_MAGIC, XFS_IBT_CRC_MAGIC, XFS_FIBT_CRC_MAGIC,
+	  XFS_REFC_CRC_MAGIC }
+};
+
+uint32_t
+xfs_btree_magic(
+	int			crc,
+	xfs_btnum_t		btnum)
+{
+	uint32_t		magic = xfs_magics[crc][btnum];
+
+	/* Ensure we asked for crc for crc-only magics. */
+	ASSERT(magic != 0);
+	return magic;
+}
+
+/*
+ * These sibling pointer checks are optimised for null sibling pointers. This
+ * happens a lot, and we don't need to byte swap at runtime if the sibling
+ * pointer is NULL.
+ *
+ * These are explicitly marked at inline because the cost of calling them as
+ * functions instead of inlining them is about 36 bytes extra code per call site
+ * on x86-64. Yes, gcc-11 fails to inline them, and explicit inlining of these
+ * two sibling check functions reduces the compiled code size by over 300
+ * bytes.
+ */
+static inline xfs_failaddr_t
+xfs_btree_check_lblock_siblings(
+	struct xfs_mount	*mp,
+	struct xfs_btree_cur	*cur,
+	int			level,
+	xfs_fsblock_t		fsb,
+	__be64			dsibling)
+{
+	xfs_fsblock_t		sibling;
+
+	if (dsibling == cpu_to_be64(NULLFSBLOCK))
+		return NULL;
+
+	sibling = be64_to_cpu(dsibling);
+	if (sibling == fsb)
+		return __this_address;
+	if (level >= 0) {
+		if (!xfs_btree_check_lptr(cur, sibling, level + 1))
+			return __this_address;
+	} else {
+		if (!xfs_verify_fsbno(mp, sibling))
+			return __this_address;
+	}
+
+	return NULL;
+}
+
+static inline xfs_failaddr_t
+xfs_btree_check_sblock_siblings(
+	struct xfs_perag	*pag,
+	struct xfs_btree_cur	*cur,
+	int			level,
+	xfs_agblock_t		agbno,
+	__be32			dsibling)
+{
+	xfs_agblock_t		sibling;
+
+	if (dsibling == cpu_to_be32(NULLAGBLOCK))
+		return NULL;
+
+	sibling = be32_to_cpu(dsibling);
+	if (sibling == agbno)
+		return __this_address;
+	if (level >= 0) {
+		if (!xfs_btree_check_sptr(cur, sibling, level + 1))
+			return __this_address;
+	} else {
+		if (!xfs_verify_agbno(pag, sibling))
+			return __this_address;
+	}
+	return NULL;
+}
+
+/*
+ * Check a long btree block header.  Return the address of the failing check,
+ * or NULL if everything is ok.
+ */
+xfs_failaddr_t
+__xfs_btree_check_lblock(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_block	*block,
+	int			level,
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = cur->bc_mp;
+	xfs_btnum_t		btnum = cur->bc_btnum;
+	int			crc = xfs_has_crc(mp);
+	xfs_failaddr_t		fa;
+	xfs_fsblock_t		fsb = NULLFSBLOCK;
+
+	if (crc) {
+		if (!uuid_equal(&block->bb_u.l.bb_uuid, &mp->m_sb.sb_meta_uuid))
+			return __this_address;
+		if (block->bb_u.l.bb_blkno !=
+		    cpu_to_be64(bp ? xfs_buf_daddr(bp) : XFS_BUF_DADDR_NULL))
+			return __this_address;
+		if (block->bb_u.l.bb_pad != cpu_to_be32(0))
+			return __this_address;
+	}
+
+	if (be32_to_cpu(block->bb_magic) != xfs_btree_magic(crc, btnum))
+		return __this_address;
+	if (be16_to_cpu(block->bb_level) != level)
+		return __this_address;
+	if (be16_to_cpu(block->bb_numrecs) >
+	    cur->bc_ops->get_maxrecs(cur, level))
+		return __this_address;
+
+	if (bp)
+		fsb = XFS_DADDR_TO_FSB(mp, xfs_buf_daddr(bp));
+
+	fa = xfs_btree_check_lblock_siblings(mp, cur, level, fsb,
+			block->bb_u.l.bb_leftsib);
+	if (!fa)
+		fa = xfs_btree_check_lblock_siblings(mp, cur, level, fsb,
+				block->bb_u.l.bb_rightsib);
+	return fa;
+}
+
+/* Check a long btree block header. */
+static int
+xfs_btree_check_lblock(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_block	*block,
+	int			level,
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = cur->bc_mp;
+	xfs_failaddr_t		fa;
+
+	fa = __xfs_btree_check_lblock(cur, block, level, bp);
+	if (XFS_IS_CORRUPT(mp, fa != NULL) ||
+	    XFS_TEST_ERROR(false, mp, XFS_ERRTAG_BTREE_CHECK_LBLOCK)) {
+		if (bp)
+			trace_xfs_btree_corrupt(bp, _RET_IP_);
+		return -EFSCORRUPTED;
+	}
+	return 0;
+}
+
+/*
+ * Check a short btree block header.  Return the address of the failing check,
+ * or NULL if everything is ok.
+ */
+xfs_failaddr_t
+__xfs_btree_check_sblock(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_block	*block,
+	int			level,
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = cur->bc_mp;
+	struct xfs_perag	*pag = cur->bc_ag.pag;
+	xfs_btnum_t		btnum = cur->bc_btnum;
+	int			crc = xfs_has_crc(mp);
+	xfs_failaddr_t		fa;
+	xfs_agblock_t		agbno = NULLAGBLOCK;
+
+	if (crc) {
+		if (!uuid_equal(&block->bb_u.s.bb_uuid, &mp->m_sb.sb_meta_uuid))
+			return __this_address;
+		if (block->bb_u.s.bb_blkno !=
+		    cpu_to_be64(bp ? xfs_buf_daddr(bp) : XFS_BUF_DADDR_NULL))
+			return __this_address;
+	}
+
+	if (be32_to_cpu(block->bb_magic) != xfs_btree_magic(crc, btnum))
+		return __this_address;
+	if (be16_to_cpu(block->bb_level) != level)
+		return __this_address;
+	if (be16_to_cpu(block->bb_numrecs) >
+	    cur->bc_ops->get_maxrecs(cur, level))
+		return __this_address;
+
+	if (bp)
+		agbno = xfs_daddr_to_agbno(mp, xfs_buf_daddr(bp));
+
+	fa = xfs_btree_check_sblock_siblings(pag, cur, level, agbno,
+			block->bb_u.s.bb_leftsib);
+	if (!fa)
+		fa = xfs_btree_check_sblock_siblings(pag, cur, level, agbno,
+				block->bb_u.s.bb_rightsib);
+	return fa;
+}
+
+/* Check a short btree block header. */
+STATIC int
+xfs_btree_check_sblock(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_block	*block,
+	int			level,
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = cur->bc_mp;
+	xfs_failaddr_t		fa;
+
+	fa = __xfs_btree_check_sblock(cur, block, level, bp);
+	if (XFS_IS_CORRUPT(mp, fa != NULL) ||
+	    XFS_TEST_ERROR(false, mp, XFS_ERRTAG_BTREE_CHECK_SBLOCK)) {
+		if (bp)
+			trace_xfs_btree_corrupt(bp, _RET_IP_);
+		return -EFSCORRUPTED;
+	}
+	return 0;
+}
+
+/*
+ * Debug routine: check that block header is ok.
+ */
+int
+xfs_btree_check_block(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	struct xfs_btree_block	*block,	/* generic btree block pointer */
+	int			level,	/* level of the btree block */
+	struct xfs_buf		*bp)	/* buffer containing block, if any */
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		return xfs_btree_check_lblock(cur, block, level, bp);
+	else
+		return xfs_btree_check_sblock(cur, block, level, bp);
+}
+
+/* Check that this long pointer is valid and points within the fs. */
+bool
+xfs_btree_check_lptr(
+	struct xfs_btree_cur	*cur,
+	xfs_fsblock_t		fsbno,
+	int			level)
+{
+	if (level <= 0)
+		return false;
+	return xfs_verify_fsbno(cur->bc_mp, fsbno);
+}
+
+/* Check that this short pointer is valid and points within the AG. */
+bool
+xfs_btree_check_sptr(
+	struct xfs_btree_cur	*cur,
+	xfs_agblock_t		agbno,
+	int			level)
+{
+	if (level <= 0)
+		return false;
+	return xfs_verify_agbno(cur->bc_ag.pag, agbno);
+}
+
+/*
+ * Check that a given (indexed) btree pointer at a certain level of a
+ * btree is valid and doesn't point past where it should.
+ */
+static int
+xfs_btree_check_ptr(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*ptr,
+	int				index,
+	int				level)
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		if (xfs_btree_check_lptr(cur, be64_to_cpu((&ptr->l)[index]),
+				level))
+			return 0;
+		xfs_err(cur->bc_mp,
+"Inode %llu fork %d: Corrupt btree %d pointer at level %d index %d.",
+				cur->bc_ino.ip->i_ino,
+				cur->bc_ino.whichfork, cur->bc_btnum,
+				level, index);
+	} else {
+		if (xfs_btree_check_sptr(cur, be32_to_cpu((&ptr->s)[index]),
+				level))
+			return 0;
+		xfs_err(cur->bc_mp,
+"AG %u: Corrupt btree %d pointer at level %d index %d.",
+				cur->bc_ag.pag->pag_agno, cur->bc_btnum,
+				level, index);
+	}
+
+	return -EFSCORRUPTED;
+}
+
+#ifdef DEBUG
+# define xfs_btree_debug_check_ptr	xfs_btree_check_ptr
+#else
+# define xfs_btree_debug_check_ptr(...)	(0)
+#endif
+
+/*
+ * Calculate CRC on the whole btree block and stuff it into the
+ * long-form btree header.
+ *
+ * Prior to calculting the CRC, pull the LSN out of the buffer log item and put
+ * it into the buffer so recovery knows what the last modification was that made
+ * it to disk.
+ */
+void
+xfs_btree_lblock_calc_crc(
+	struct xfs_buf		*bp)
+{
+	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
+	struct xfs_buf_log_item	*bip = bp->b_log_item;
+
+	if (!xfs_has_crc(bp->b_mount))
+		return;
+	if (bip)
+		block->bb_u.l.bb_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+	xfs_buf_update_cksum(bp, XFS_BTREE_LBLOCK_CRC_OFF);
+}
+
+bool
+xfs_btree_lblock_verify_crc(
+	struct xfs_buf		*bp)
+{
+	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
+	struct xfs_mount	*mp = bp->b_mount;
+
+	if (xfs_has_crc(mp)) {
+		if (!xfs_log_check_lsn(mp, be64_to_cpu(block->bb_u.l.bb_lsn)))
+			return false;
+		return xfs_buf_verify_cksum(bp, XFS_BTREE_LBLOCK_CRC_OFF);
+	}
+
+	return true;
+}
+
+/*
+ * Calculate CRC on the whole btree block and stuff it into the
+ * short-form btree header.
+ *
+ * Prior to calculting the CRC, pull the LSN out of the buffer log item and put
+ * it into the buffer so recovery knows what the last modification was that made
+ * it to disk.
+ */
+void
+xfs_btree_sblock_calc_crc(
+	struct xfs_buf		*bp)
+{
+	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
+	struct xfs_buf_log_item	*bip = bp->b_log_item;
+
+	if (!xfs_has_crc(bp->b_mount))
+		return;
+	if (bip)
+		block->bb_u.s.bb_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+	xfs_buf_update_cksum(bp, XFS_BTREE_SBLOCK_CRC_OFF);
+}
+
+bool
+xfs_btree_sblock_verify_crc(
+	struct xfs_buf		*bp)
+{
+	struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
+	struct xfs_mount	*mp = bp->b_mount;
+
+	if (xfs_has_crc(mp)) {
+		if (!xfs_log_check_lsn(mp, be64_to_cpu(block->bb_u.s.bb_lsn)))
+			return false;
+		return xfs_buf_verify_cksum(bp, XFS_BTREE_SBLOCK_CRC_OFF);
+	}
+
+	return true;
+}
+
+static int
+xfs_btree_free_block(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp)
+{
+	int			error;
+
+	error = cur->bc_ops->free_block(cur, bp);
+	if (!error) {
+		xfs_trans_binval(cur->bc_tp, bp);
+		XFS_BTREE_STATS_INC(cur, free);
+	}
+	return error;
+}
+
+/*
+ * Delete the btree cursor.
+ */
+void
+xfs_btree_del_cursor(
+	struct xfs_btree_cur	*cur,		/* btree cursor */
+	int			error)		/* del because of error */
+{
+	int			i;		/* btree level */
+
+	/*
+	 * Clear the buffer pointers and release the buffers. If we're doing
+	 * this because of an error, inspect all of the entries in the bc_bufs
+	 * array for buffers to be unlocked. This is because some of the btree
+	 * code works from level n down to 0, and if we get an error along the
+	 * way we won't have initialized all the entries down to 0.
+	 */
+	for (i = 0; i < cur->bc_nlevels; i++) {
+		if (cur->bc_levels[i].bp)
+			xfs_trans_brelse(cur->bc_tp, cur->bc_levels[i].bp);
+		else if (!error)
+			break;
+	}
+
+	/*
+	 * If we are doing a BMBT update, the number of unaccounted blocks
+	 * allocated during this cursor life time should be zero. If it's not
+	 * zero, then we should be shut down or on our way to shutdown due to
+	 * cancelling a dirty transaction on error.
+	 */
+	ASSERT(cur->bc_btnum != XFS_BTNUM_BMAP || cur->bc_ino.allocated == 0 ||
+	       xfs_is_shutdown(cur->bc_mp) || error != 0);
+	if (unlikely(cur->bc_flags & XFS_BTREE_STAGING))
+		kmem_free(cur->bc_ops);
+	if (!(cur->bc_flags & XFS_BTREE_LONG_PTRS) && cur->bc_ag.pag)
+		xfs_perag_put(cur->bc_ag.pag);
+	kmem_cache_free(cur->bc_cache, cur);
+}
+
+/*
+ * Duplicate the btree cursor.
+ * Allocate a new one, copy the record, re-get the buffers.
+ */
+int					/* error */
+xfs_btree_dup_cursor(
+	struct xfs_btree_cur *cur,		/* input cursor */
+	struct xfs_btree_cur **ncur)		/* output cursor */
+{
+	struct xfs_buf	*bp;		/* btree block's buffer pointer */
+	int		error;		/* error return value */
+	int		i;		/* level number of btree block */
+	xfs_mount_t	*mp;		/* mount structure for filesystem */
+	struct xfs_btree_cur *new;		/* new cursor value */
+	xfs_trans_t	*tp;		/* transaction pointer, can be NULL */
+
+	tp = cur->bc_tp;
+	mp = cur->bc_mp;
+
+	/*
+	 * Allocate a new cursor like the old one.
+	 */
+	new = cur->bc_ops->dup_cursor(cur);
+
+	/*
+	 * Copy the record currently in the cursor.
+	 */
+	new->bc_rec = cur->bc_rec;
+
+	/*
+	 * For each level current, re-get the buffer and copy the ptr value.
+	 */
+	for (i = 0; i < new->bc_nlevels; i++) {
+		new->bc_levels[i].ptr = cur->bc_levels[i].ptr;
+		new->bc_levels[i].ra = cur->bc_levels[i].ra;
+		bp = cur->bc_levels[i].bp;
+		if (bp) {
+			error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
+						   xfs_buf_daddr(bp), mp->m_bsize,
+						   0, &bp,
+						   cur->bc_ops->buf_ops);
+			if (error) {
+				xfs_btree_del_cursor(new, error);
+				*ncur = NULL;
+				return error;
+			}
+		}
+		new->bc_levels[i].bp = bp;
+	}
+	*ncur = new;
+	return 0;
+}
+
+/*
+ * XFS btree block layout and addressing:
+ *
+ * There are two types of blocks in the btree: leaf and non-leaf blocks.
+ *
+ * The leaf record start with a header then followed by records containing
+ * the values.  A non-leaf block also starts with the same header, and
+ * then first contains lookup keys followed by an equal number of pointers
+ * to the btree blocks at the previous level.
+ *
+ *		+--------+-------+-------+-------+-------+-------+-------+
+ * Leaf:	| header | rec 1 | rec 2 | rec 3 | rec 4 | rec 5 | rec N |
+ *		+--------+-------+-------+-------+-------+-------+-------+
+ *
+ *		+--------+-------+-------+-------+-------+-------+-------+
+ * Non-Leaf:	| header | key 1 | key 2 | key N | ptr 1 | ptr 2 | ptr N |
+ *		+--------+-------+-------+-------+-------+-------+-------+
+ *
+ * The header is called struct xfs_btree_block for reasons better left unknown
+ * and comes in different versions for short (32bit) and long (64bit) block
+ * pointers.  The record and key structures are defined by the btree instances
+ * and opaque to the btree core.  The block pointers are simple disk endian
+ * integers, available in a short (32bit) and long (64bit) variant.
+ *
+ * The helpers below calculate the offset of a given record, key or pointer
+ * into a btree block (xfs_btree_*_offset) or return a pointer to the given
+ * record, key or pointer (xfs_btree_*_addr).  Note that all addressing
+ * inside the btree block is done using indices starting at one, not zero!
+ *
+ * If XFS_BTREE_OVERLAPPING is set, then this btree supports keys containing
+ * overlapping intervals.  In such a tree, records are still sorted lowest to
+ * highest and indexed by the smallest key value that refers to the record.
+ * However, nodes are different: each pointer has two associated keys -- one
+ * indexing the lowest key available in the block(s) below (the same behavior
+ * as the key in a regular btree) and another indexing the highest key
+ * available in the block(s) below.  Because records are /not/ sorted by the
+ * highest key, all leaf block updates require us to compute the highest key
+ * that matches any record in the leaf and to recursively update the high keys
+ * in the nodes going further up in the tree, if necessary.  Nodes look like
+ * this:
+ *
+ *		+--------+-----+-----+-----+-----+-----+-------+-------+-----+
+ * Non-Leaf:	| header | lo1 | hi1 | lo2 | hi2 | ... | ptr 1 | ptr 2 | ... |
+ *		+--------+-----+-----+-----+-----+-----+-------+-------+-----+
+ *
+ * To perform an interval query on an overlapped tree, perform the usual
+ * depth-first search and use the low and high keys to decide if we can skip
+ * that particular node.  If a leaf node is reached, return the records that
+ * intersect the interval.  Note that an interval query may return numerous
+ * entries.  For a non-overlapped tree, simply search for the record associated
+ * with the lowest key and iterate forward until a non-matching record is
+ * found.  Section 14.3 ("Interval Trees") of _Introduction to Algorithms_ by
+ * Cormen, Leiserson, Rivest, and Stein (2nd or 3rd ed. only) discuss this in
+ * more detail.
+ *
+ * Why do we care about overlapping intervals?  Let's say you have a bunch of
+ * reverse mapping records on a reflink filesystem:
+ *
+ * 1: +- file A startblock B offset C length D -----------+
+ * 2:      +- file E startblock F offset G length H --------------+
+ * 3:      +- file I startblock F offset J length K --+
+ * 4:                                                        +- file L... --+
+ *
+ * Now say we want to map block (B+D) into file A at offset (C+D).  Ideally,
+ * we'd simply increment the length of record 1.  But how do we find the record
+ * that ends at (B+D-1) (i.e. record 1)?  A LE lookup of (B+D-1) would return
+ * record 3 because the keys are ordered first by startblock.  An interval
+ * query would return records 1 and 2 because they both overlap (B+D-1), and
+ * from that we can pick out record 1 as the appropriate left neighbor.
+ *
+ * In the non-overlapped case you can do a LE lookup and decrement the cursor
+ * because a record's interval must end before the next record.
+ */
+
+/*
+ * Return size of the btree block header for this btree instance.
+ */
+static inline size_t xfs_btree_block_len(struct xfs_btree_cur *cur)
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS)
+			return XFS_BTREE_LBLOCK_CRC_LEN;
+		return XFS_BTREE_LBLOCK_LEN;
+	}
+	if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS)
+		return XFS_BTREE_SBLOCK_CRC_LEN;
+	return XFS_BTREE_SBLOCK_LEN;
+}
+
+/*
+ * Return size of btree block pointers for this btree instance.
+ */
+static inline size_t xfs_btree_ptr_len(struct xfs_btree_cur *cur)
+{
+	return (cur->bc_flags & XFS_BTREE_LONG_PTRS) ?
+		sizeof(__be64) : sizeof(__be32);
+}
+
+/*
+ * Calculate offset of the n-th record in a btree block.
+ */
+STATIC size_t
+xfs_btree_rec_offset(
+	struct xfs_btree_cur	*cur,
+	int			n)
+{
+	return xfs_btree_block_len(cur) +
+		(n - 1) * cur->bc_ops->rec_len;
+}
+
+/*
+ * Calculate offset of the n-th key in a btree block.
+ */
+STATIC size_t
+xfs_btree_key_offset(
+	struct xfs_btree_cur	*cur,
+	int			n)
+{
+	return xfs_btree_block_len(cur) +
+		(n - 1) * cur->bc_ops->key_len;
+}
+
+/*
+ * Calculate offset of the n-th high key in a btree block.
+ */
+STATIC size_t
+xfs_btree_high_key_offset(
+	struct xfs_btree_cur	*cur,
+	int			n)
+{
+	return xfs_btree_block_len(cur) +
+		(n - 1) * cur->bc_ops->key_len + (cur->bc_ops->key_len / 2);
+}
+
+/*
+ * Calculate offset of the n-th block pointer in a btree block.
+ */
+STATIC size_t
+xfs_btree_ptr_offset(
+	struct xfs_btree_cur	*cur,
+	int			n,
+	int			level)
+{
+	return xfs_btree_block_len(cur) +
+		cur->bc_ops->get_maxrecs(cur, level) * cur->bc_ops->key_len +
+		(n - 1) * xfs_btree_ptr_len(cur);
+}
+
+/*
+ * Return a pointer to the n-th record in the btree block.
+ */
+union xfs_btree_rec *
+xfs_btree_rec_addr(
+	struct xfs_btree_cur	*cur,
+	int			n,
+	struct xfs_btree_block	*block)
+{
+	return (union xfs_btree_rec *)
+		((char *)block + xfs_btree_rec_offset(cur, n));
+}
+
+/*
+ * Return a pointer to the n-th key in the btree block.
+ */
+union xfs_btree_key *
+xfs_btree_key_addr(
+	struct xfs_btree_cur	*cur,
+	int			n,
+	struct xfs_btree_block	*block)
+{
+	return (union xfs_btree_key *)
+		((char *)block + xfs_btree_key_offset(cur, n));
+}
+
+/*
+ * Return a pointer to the n-th high key in the btree block.
+ */
+union xfs_btree_key *
+xfs_btree_high_key_addr(
+	struct xfs_btree_cur	*cur,
+	int			n,
+	struct xfs_btree_block	*block)
+{
+	return (union xfs_btree_key *)
+		((char *)block + xfs_btree_high_key_offset(cur, n));
+}
+
+/*
+ * Return a pointer to the n-th block pointer in the btree block.
+ */
+union xfs_btree_ptr *
+xfs_btree_ptr_addr(
+	struct xfs_btree_cur	*cur,
+	int			n,
+	struct xfs_btree_block	*block)
+{
+	int			level = xfs_btree_get_level(block);
+
+	ASSERT(block->bb_level != 0);
+
+	return (union xfs_btree_ptr *)
+		((char *)block + xfs_btree_ptr_offset(cur, n, level));
+}
+
+struct xfs_ifork *
+xfs_btree_ifork_ptr(
+	struct xfs_btree_cur	*cur)
+{
+	ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
+
+	if (cur->bc_flags & XFS_BTREE_STAGING)
+		return cur->bc_ino.ifake->if_fork;
+	return xfs_ifork_ptr(cur->bc_ino.ip, cur->bc_ino.whichfork);
+}
+
+/*
+ * Get the root block which is stored in the inode.
+ *
+ * For now this btree implementation assumes the btree root is always
+ * stored in the if_broot field of an inode fork.
+ */
+STATIC struct xfs_btree_block *
+xfs_btree_get_iroot(
+	struct xfs_btree_cur	*cur)
+{
+	struct xfs_ifork	*ifp = xfs_btree_ifork_ptr(cur);
+
+	return (struct xfs_btree_block *)ifp->if_broot;
+}
+
+/*
+ * Retrieve the block pointer from the cursor at the given level.
+ * This may be an inode btree root or from a buffer.
+ */
+struct xfs_btree_block *		/* generic btree block pointer */
+xfs_btree_get_block(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			level,	/* level in btree */
+	struct xfs_buf		**bpp)	/* buffer containing the block */
+{
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    (level == cur->bc_nlevels - 1)) {
+		*bpp = NULL;
+		return xfs_btree_get_iroot(cur);
+	}
+
+	*bpp = cur->bc_levels[level].bp;
+	return XFS_BUF_TO_BLOCK(*bpp);
+}
+
+/*
+ * Change the cursor to point to the first record at the given level.
+ * Other levels are unaffected.
+ */
+STATIC int				/* success=1, failure=0 */
+xfs_btree_firstrec(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			level)	/* level to change */
+{
+	struct xfs_btree_block	*block;	/* generic btree block pointer */
+	struct xfs_buf		*bp;	/* buffer containing block */
+
+	/*
+	 * Get the block pointer for this level.
+	 */
+	block = xfs_btree_get_block(cur, level, &bp);
+	if (xfs_btree_check_block(cur, block, level, bp))
+		return 0;
+	/*
+	 * It's empty, there is no such record.
+	 */
+	if (!block->bb_numrecs)
+		return 0;
+	/*
+	 * Set the ptr value to 1, that's the first record/key.
+	 */
+	cur->bc_levels[level].ptr = 1;
+	return 1;
+}
+
+/*
+ * Change the cursor to point to the last record in the current block
+ * at the given level.  Other levels are unaffected.
+ */
+STATIC int				/* success=1, failure=0 */
+xfs_btree_lastrec(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			level)	/* level to change */
+{
+	struct xfs_btree_block	*block;	/* generic btree block pointer */
+	struct xfs_buf		*bp;	/* buffer containing block */
+
+	/*
+	 * Get the block pointer for this level.
+	 */
+	block = xfs_btree_get_block(cur, level, &bp);
+	if (xfs_btree_check_block(cur, block, level, bp))
+		return 0;
+	/*
+	 * It's empty, there is no such record.
+	 */
+	if (!block->bb_numrecs)
+		return 0;
+	/*
+	 * Set the ptr value to numrecs, that's the last record/key.
+	 */
+	cur->bc_levels[level].ptr = be16_to_cpu(block->bb_numrecs);
+	return 1;
+}
+
+/*
+ * Compute first and last byte offsets for the fields given.
+ * Interprets the offsets table, which contains struct field offsets.
+ */
+void
+xfs_btree_offsets(
+	uint32_t	fields,		/* bitmask of fields */
+	const short	*offsets,	/* table of field offsets */
+	int		nbits,		/* number of bits to inspect */
+	int		*first,		/* output: first byte offset */
+	int		*last)		/* output: last byte offset */
+{
+	int		i;		/* current bit number */
+	uint32_t	imask;		/* mask for current bit number */
+
+	ASSERT(fields != 0);
+	/*
+	 * Find the lowest bit, so the first byte offset.
+	 */
+	for (i = 0, imask = 1u; ; i++, imask <<= 1) {
+		if (imask & fields) {
+			*first = offsets[i];
+			break;
+		}
+	}
+	/*
+	 * Find the highest bit, so the last byte offset.
+	 */
+	for (i = nbits - 1, imask = 1u << i; ; i--, imask >>= 1) {
+		if (imask & fields) {
+			*last = offsets[i + 1] - 1;
+			break;
+		}
+	}
+}
+
+/*
+ * Get a buffer for the block, return it read in.
+ * Long-form addressing.
+ */
+int
+xfs_btree_read_bufl(
+	struct xfs_mount	*mp,		/* file system mount point */
+	struct xfs_trans	*tp,		/* transaction pointer */
+	xfs_fsblock_t		fsbno,		/* file system block number */
+	struct xfs_buf		**bpp,		/* buffer for fsbno */
+	int			refval,		/* ref count value for buffer */
+	const struct xfs_buf_ops *ops)
+{
+	struct xfs_buf		*bp;		/* return value */
+	xfs_daddr_t		d;		/* real disk block address */
+	int			error;
+
+	if (!xfs_verify_fsbno(mp, fsbno))
+		return -EFSCORRUPTED;
+	d = XFS_FSB_TO_DADDR(mp, fsbno);
+	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, d,
+				   mp->m_bsize, 0, &bp, ops);
+	if (error)
+		return error;
+	if (bp)
+		xfs_buf_set_ref(bp, refval);
+	*bpp = bp;
+	return 0;
+}
+
+/*
+ * Read-ahead the block, don't wait for it, don't return a buffer.
+ * Long-form addressing.
+ */
+/* ARGSUSED */
+void
+xfs_btree_reada_bufl(
+	struct xfs_mount	*mp,		/* file system mount point */
+	xfs_fsblock_t		fsbno,		/* file system block number */
+	xfs_extlen_t		count,		/* count of filesystem blocks */
+	const struct xfs_buf_ops *ops)
+{
+	xfs_daddr_t		d;
+
+	ASSERT(fsbno != NULLFSBLOCK);
+	d = XFS_FSB_TO_DADDR(mp, fsbno);
+	xfs_buf_readahead(mp->m_ddev_targp, d, mp->m_bsize * count, ops);
+}
+
+/*
+ * Read-ahead the block, don't wait for it, don't return a buffer.
+ * Short-form addressing.
+ */
+/* ARGSUSED */
+void
+xfs_btree_reada_bufs(
+	struct xfs_mount	*mp,		/* file system mount point */
+	xfs_agnumber_t		agno,		/* allocation group number */
+	xfs_agblock_t		agbno,		/* allocation group block number */
+	xfs_extlen_t		count,		/* count of filesystem blocks */
+	const struct xfs_buf_ops *ops)
+{
+	xfs_daddr_t		d;
+
+	ASSERT(agno != NULLAGNUMBER);
+	ASSERT(agbno != NULLAGBLOCK);
+	d = XFS_AGB_TO_DADDR(mp, agno, agbno);
+	xfs_buf_readahead(mp->m_ddev_targp, d, mp->m_bsize * count, ops);
+}
+
+STATIC int
+xfs_btree_readahead_lblock(
+	struct xfs_btree_cur	*cur,
+	int			lr,
+	struct xfs_btree_block	*block)
+{
+	int			rval = 0;
+	xfs_fsblock_t		left = be64_to_cpu(block->bb_u.l.bb_leftsib);
+	xfs_fsblock_t		right = be64_to_cpu(block->bb_u.l.bb_rightsib);
+
+	if ((lr & XFS_BTCUR_LEFTRA) && left != NULLFSBLOCK) {
+		xfs_btree_reada_bufl(cur->bc_mp, left, 1,
+				     cur->bc_ops->buf_ops);
+		rval++;
+	}
+
+	if ((lr & XFS_BTCUR_RIGHTRA) && right != NULLFSBLOCK) {
+		xfs_btree_reada_bufl(cur->bc_mp, right, 1,
+				     cur->bc_ops->buf_ops);
+		rval++;
+	}
+
+	return rval;
+}
+
+STATIC int
+xfs_btree_readahead_sblock(
+	struct xfs_btree_cur	*cur,
+	int			lr,
+	struct xfs_btree_block *block)
+{
+	int			rval = 0;
+	xfs_agblock_t		left = be32_to_cpu(block->bb_u.s.bb_leftsib);
+	xfs_agblock_t		right = be32_to_cpu(block->bb_u.s.bb_rightsib);
+
+
+	if ((lr & XFS_BTCUR_LEFTRA) && left != NULLAGBLOCK) {
+		xfs_btree_reada_bufs(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+				     left, 1, cur->bc_ops->buf_ops);
+		rval++;
+	}
+
+	if ((lr & XFS_BTCUR_RIGHTRA) && right != NULLAGBLOCK) {
+		xfs_btree_reada_bufs(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+				     right, 1, cur->bc_ops->buf_ops);
+		rval++;
+	}
+
+	return rval;
+}
+
+/*
+ * Read-ahead btree blocks, at the given level.
+ * Bits in lr are set from XFS_BTCUR_{LEFT,RIGHT}RA.
+ */
+STATIC int
+xfs_btree_readahead(
+	struct xfs_btree_cur	*cur,		/* btree cursor */
+	int			lev,		/* level in btree */
+	int			lr)		/* left/right bits */
+{
+	struct xfs_btree_block	*block;
+
+	/*
+	 * No readahead needed if we are at the root level and the
+	 * btree root is stored in the inode.
+	 */
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    (lev == cur->bc_nlevels - 1))
+		return 0;
+
+	if ((cur->bc_levels[lev].ra | lr) == cur->bc_levels[lev].ra)
+		return 0;
+
+	cur->bc_levels[lev].ra |= lr;
+	block = XFS_BUF_TO_BLOCK(cur->bc_levels[lev].bp);
+
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		return xfs_btree_readahead_lblock(cur, lr, block);
+	return xfs_btree_readahead_sblock(cur, lr, block);
+}
+
+STATIC int
+xfs_btree_ptr_to_daddr(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*ptr,
+	xfs_daddr_t			*daddr)
+{
+	xfs_fsblock_t		fsbno;
+	xfs_agblock_t		agbno;
+	int			error;
+
+	error = xfs_btree_check_ptr(cur, ptr, 0, 1);
+	if (error)
+		return error;
+
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		fsbno = be64_to_cpu(ptr->l);
+		*daddr = XFS_FSB_TO_DADDR(cur->bc_mp, fsbno);
+	} else {
+		agbno = be32_to_cpu(ptr->s);
+		*daddr = XFS_AGB_TO_DADDR(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+				agbno);
+	}
+
+	return 0;
+}
+
+/*
+ * Readahead @count btree blocks at the given @ptr location.
+ *
+ * We don't need to care about long or short form btrees here as we have a
+ * method of converting the ptr directly to a daddr available to us.
+ */
+STATIC void
+xfs_btree_readahead_ptr(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr,
+	xfs_extlen_t		count)
+{
+	xfs_daddr_t		daddr;
+
+	if (xfs_btree_ptr_to_daddr(cur, ptr, &daddr))
+		return;
+	xfs_buf_readahead(cur->bc_mp->m_ddev_targp, daddr,
+			  cur->bc_mp->m_bsize * count, cur->bc_ops->buf_ops);
+}
+
+/*
+ * Set the buffer for level "lev" in the cursor to bp, releasing
+ * any previous buffer.
+ */
+STATIC void
+xfs_btree_setbuf(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			lev,	/* level in btree */
+	struct xfs_buf		*bp)	/* new buffer to set */
+{
+	struct xfs_btree_block	*b;	/* btree block */
+
+	if (cur->bc_levels[lev].bp)
+		xfs_trans_brelse(cur->bc_tp, cur->bc_levels[lev].bp);
+	cur->bc_levels[lev].bp = bp;
+	cur->bc_levels[lev].ra = 0;
+
+	b = XFS_BUF_TO_BLOCK(bp);
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		if (b->bb_u.l.bb_leftsib == cpu_to_be64(NULLFSBLOCK))
+			cur->bc_levels[lev].ra |= XFS_BTCUR_LEFTRA;
+		if (b->bb_u.l.bb_rightsib == cpu_to_be64(NULLFSBLOCK))
+			cur->bc_levels[lev].ra |= XFS_BTCUR_RIGHTRA;
+	} else {
+		if (b->bb_u.s.bb_leftsib == cpu_to_be32(NULLAGBLOCK))
+			cur->bc_levels[lev].ra |= XFS_BTCUR_LEFTRA;
+		if (b->bb_u.s.bb_rightsib == cpu_to_be32(NULLAGBLOCK))
+			cur->bc_levels[lev].ra |= XFS_BTCUR_RIGHTRA;
+	}
+}
+
+bool
+xfs_btree_ptr_is_null(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*ptr)
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		return ptr->l == cpu_to_be64(NULLFSBLOCK);
+	else
+		return ptr->s == cpu_to_be32(NULLAGBLOCK);
+}
+
+void
+xfs_btree_set_ptr_null(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		ptr->l = cpu_to_be64(NULLFSBLOCK);
+	else
+		ptr->s = cpu_to_be32(NULLAGBLOCK);
+}
+
+/*
+ * Get/set/init sibling pointers
+ */
+void
+xfs_btree_get_sibling(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_block	*block,
+	union xfs_btree_ptr	*ptr,
+	int			lr)
+{
+	ASSERT(lr == XFS_BB_LEFTSIB || lr == XFS_BB_RIGHTSIB);
+
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		if (lr == XFS_BB_RIGHTSIB)
+			ptr->l = block->bb_u.l.bb_rightsib;
+		else
+			ptr->l = block->bb_u.l.bb_leftsib;
+	} else {
+		if (lr == XFS_BB_RIGHTSIB)
+			ptr->s = block->bb_u.s.bb_rightsib;
+		else
+			ptr->s = block->bb_u.s.bb_leftsib;
+	}
+}
+
+void
+xfs_btree_set_sibling(
+	struct xfs_btree_cur		*cur,
+	struct xfs_btree_block		*block,
+	const union xfs_btree_ptr	*ptr,
+	int				lr)
+{
+	ASSERT(lr == XFS_BB_LEFTSIB || lr == XFS_BB_RIGHTSIB);
+
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		if (lr == XFS_BB_RIGHTSIB)
+			block->bb_u.l.bb_rightsib = ptr->l;
+		else
+			block->bb_u.l.bb_leftsib = ptr->l;
+	} else {
+		if (lr == XFS_BB_RIGHTSIB)
+			block->bb_u.s.bb_rightsib = ptr->s;
+		else
+			block->bb_u.s.bb_leftsib = ptr->s;
+	}
+}
+
+void
+xfs_btree_init_block_int(
+	struct xfs_mount	*mp,
+	struct xfs_btree_block	*buf,
+	xfs_daddr_t		blkno,
+	xfs_btnum_t		btnum,
+	__u16			level,
+	__u16			numrecs,
+	__u64			owner,
+	unsigned int		flags)
+{
+	int			crc = xfs_has_crc(mp);
+	__u32			magic = xfs_btree_magic(crc, btnum);
+
+	buf->bb_magic = cpu_to_be32(magic);
+	buf->bb_level = cpu_to_be16(level);
+	buf->bb_numrecs = cpu_to_be16(numrecs);
+
+	if (flags & XFS_BTREE_LONG_PTRS) {
+		buf->bb_u.l.bb_leftsib = cpu_to_be64(NULLFSBLOCK);
+		buf->bb_u.l.bb_rightsib = cpu_to_be64(NULLFSBLOCK);
+		if (crc) {
+			buf->bb_u.l.bb_blkno = cpu_to_be64(blkno);
+			buf->bb_u.l.bb_owner = cpu_to_be64(owner);
+			uuid_copy(&buf->bb_u.l.bb_uuid, &mp->m_sb.sb_meta_uuid);
+			buf->bb_u.l.bb_pad = 0;
+			buf->bb_u.l.bb_lsn = 0;
+		}
+	} else {
+		/* owner is a 32 bit value on short blocks */
+		__u32 __owner = (__u32)owner;
+
+		buf->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
+		buf->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
+		if (crc) {
+			buf->bb_u.s.bb_blkno = cpu_to_be64(blkno);
+			buf->bb_u.s.bb_owner = cpu_to_be32(__owner);
+			uuid_copy(&buf->bb_u.s.bb_uuid, &mp->m_sb.sb_meta_uuid);
+			buf->bb_u.s.bb_lsn = 0;
+		}
+	}
+}
+
+void
+xfs_btree_init_block(
+	struct xfs_mount *mp,
+	struct xfs_buf	*bp,
+	xfs_btnum_t	btnum,
+	__u16		level,
+	__u16		numrecs,
+	__u64		owner)
+{
+	xfs_btree_init_block_int(mp, XFS_BUF_TO_BLOCK(bp), xfs_buf_daddr(bp),
+				 btnum, level, numrecs, owner, 0);
+}
+
+void
+xfs_btree_init_block_cur(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp,
+	int			level,
+	int			numrecs)
+{
+	__u64			owner;
+
+	/*
+	 * we can pull the owner from the cursor right now as the different
+	 * owners align directly with the pointer size of the btree. This may
+	 * change in future, but is safe for current users of the generic btree
+	 * code.
+	 */
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		owner = cur->bc_ino.ip->i_ino;
+	else
+		owner = cur->bc_ag.pag->pag_agno;
+
+	xfs_btree_init_block_int(cur->bc_mp, XFS_BUF_TO_BLOCK(bp),
+				xfs_buf_daddr(bp), cur->bc_btnum, level,
+				numrecs, owner, cur->bc_flags);
+}
+
+/*
+ * Return true if ptr is the last record in the btree and
+ * we need to track updates to this record.  The decision
+ * will be further refined in the update_lastrec method.
+ */
+STATIC int
+xfs_btree_is_lastrec(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_block	*block,
+	int			level)
+{
+	union xfs_btree_ptr	ptr;
+
+	if (level > 0)
+		return 0;
+	if (!(cur->bc_flags & XFS_BTREE_LASTREC_UPDATE))
+		return 0;
+
+	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
+	if (!xfs_btree_ptr_is_null(cur, &ptr))
+		return 0;
+	return 1;
+}
+
+STATIC void
+xfs_btree_buf_to_ptr(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp,
+	union xfs_btree_ptr	*ptr)
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		ptr->l = cpu_to_be64(XFS_DADDR_TO_FSB(cur->bc_mp,
+					xfs_buf_daddr(bp)));
+	else {
+		ptr->s = cpu_to_be32(xfs_daddr_to_agbno(cur->bc_mp,
+					xfs_buf_daddr(bp)));
+	}
+}
+
+STATIC void
+xfs_btree_set_refs(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp)
+{
+	switch (cur->bc_btnum) {
+	case XFS_BTNUM_BNO:
+	case XFS_BTNUM_CNT:
+		xfs_buf_set_ref(bp, XFS_ALLOC_BTREE_REF);
+		break;
+	case XFS_BTNUM_INO:
+	case XFS_BTNUM_FINO:
+		xfs_buf_set_ref(bp, XFS_INO_BTREE_REF);
+		break;
+	case XFS_BTNUM_BMAP:
+		xfs_buf_set_ref(bp, XFS_BMAP_BTREE_REF);
+		break;
+	case XFS_BTNUM_RMAP:
+		xfs_buf_set_ref(bp, XFS_RMAP_BTREE_REF);
+		break;
+	case XFS_BTNUM_REFC:
+		xfs_buf_set_ref(bp, XFS_REFC_BTREE_REF);
+		break;
+	default:
+		ASSERT(0);
+	}
+}
+
+int
+xfs_btree_get_buf_block(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*ptr,
+	struct xfs_btree_block		**block,
+	struct xfs_buf			**bpp)
+{
+	struct xfs_mount	*mp = cur->bc_mp;
+	xfs_daddr_t		d;
+	int			error;
+
+	error = xfs_btree_ptr_to_daddr(cur, ptr, &d);
+	if (error)
+		return error;
+	error = xfs_trans_get_buf(cur->bc_tp, mp->m_ddev_targp, d, mp->m_bsize,
+			0, bpp);
+	if (error)
+		return error;
+
+	(*bpp)->b_ops = cur->bc_ops->buf_ops;
+	*block = XFS_BUF_TO_BLOCK(*bpp);
+	return 0;
+}
+
+/*
+ * Read in the buffer at the given ptr and return the buffer and
+ * the block pointer within the buffer.
+ */
+STATIC int
+xfs_btree_read_buf_block(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*ptr,
+	int				flags,
+	struct xfs_btree_block		**block,
+	struct xfs_buf			**bpp)
+{
+	struct xfs_mount	*mp = cur->bc_mp;
+	xfs_daddr_t		d;
+	int			error;
+
+	/* need to sort out how callers deal with failures first */
+	ASSERT(!(flags & XBF_TRYLOCK));
+
+	error = xfs_btree_ptr_to_daddr(cur, ptr, &d);
+	if (error)
+		return error;
+	error = xfs_trans_read_buf(mp, cur->bc_tp, mp->m_ddev_targp, d,
+				   mp->m_bsize, flags, bpp,
+				   cur->bc_ops->buf_ops);
+	if (error)
+		return error;
+
+	xfs_btree_set_refs(cur, *bpp);
+	*block = XFS_BUF_TO_BLOCK(*bpp);
+	return 0;
+}
+
+/*
+ * Copy keys from one btree block to another.
+ */
+void
+xfs_btree_copy_keys(
+	struct xfs_btree_cur		*cur,
+	union xfs_btree_key		*dst_key,
+	const union xfs_btree_key	*src_key,
+	int				numkeys)
+{
+	ASSERT(numkeys >= 0);
+	memcpy(dst_key, src_key, numkeys * cur->bc_ops->key_len);
+}
+
+/*
+ * Copy records from one btree block to another.
+ */
+STATIC void
+xfs_btree_copy_recs(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*dst_rec,
+	union xfs_btree_rec	*src_rec,
+	int			numrecs)
+{
+	ASSERT(numrecs >= 0);
+	memcpy(dst_rec, src_rec, numrecs * cur->bc_ops->rec_len);
+}
+
+/*
+ * Copy block pointers from one btree block to another.
+ */
+void
+xfs_btree_copy_ptrs(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*dst_ptr,
+	const union xfs_btree_ptr *src_ptr,
+	int			numptrs)
+{
+	ASSERT(numptrs >= 0);
+	memcpy(dst_ptr, src_ptr, numptrs * xfs_btree_ptr_len(cur));
+}
+
+/*
+ * Shift keys one index left/right inside a single btree block.
+ */
+STATIC void
+xfs_btree_shift_keys(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*key,
+	int			dir,
+	int			numkeys)
+{
+	char			*dst_key;
+
+	ASSERT(numkeys >= 0);
+	ASSERT(dir == 1 || dir == -1);
+
+	dst_key = (char *)key + (dir * cur->bc_ops->key_len);
+	memmove(dst_key, key, numkeys * cur->bc_ops->key_len);
+}
+
+/*
+ * Shift records one index left/right inside a single btree block.
+ */
+STATIC void
+xfs_btree_shift_recs(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec,
+	int			dir,
+	int			numrecs)
+{
+	char			*dst_rec;
+
+	ASSERT(numrecs >= 0);
+	ASSERT(dir == 1 || dir == -1);
+
+	dst_rec = (char *)rec + (dir * cur->bc_ops->rec_len);
+	memmove(dst_rec, rec, numrecs * cur->bc_ops->rec_len);
+}
+
+/*
+ * Shift block pointers one index left/right inside a single btree block.
+ */
+STATIC void
+xfs_btree_shift_ptrs(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr,
+	int			dir,
+	int			numptrs)
+{
+	char			*dst_ptr;
+
+	ASSERT(numptrs >= 0);
+	ASSERT(dir == 1 || dir == -1);
+
+	dst_ptr = (char *)ptr + (dir * xfs_btree_ptr_len(cur));
+	memmove(dst_ptr, ptr, numptrs * xfs_btree_ptr_len(cur));
+}
+
+/*
+ * Log key values from the btree block.
+ */
+STATIC void
+xfs_btree_log_keys(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp,
+	int			first,
+	int			last)
+{
+
+	if (bp) {
+		xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
+		xfs_trans_log_buf(cur->bc_tp, bp,
+				  xfs_btree_key_offset(cur, first),
+				  xfs_btree_key_offset(cur, last + 1) - 1);
+	} else {
+		xfs_trans_log_inode(cur->bc_tp, cur->bc_ino.ip,
+				xfs_ilog_fbroot(cur->bc_ino.whichfork));
+	}
+}
+
+/*
+ * Log record values from the btree block.
+ */
+void
+xfs_btree_log_recs(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp,
+	int			first,
+	int			last)
+{
+
+	xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
+	xfs_trans_log_buf(cur->bc_tp, bp,
+			  xfs_btree_rec_offset(cur, first),
+			  xfs_btree_rec_offset(cur, last + 1) - 1);
+
+}
+
+/*
+ * Log block pointer fields from a btree block (nonleaf).
+ */
+STATIC void
+xfs_btree_log_ptrs(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	struct xfs_buf		*bp,	/* buffer containing btree block */
+	int			first,	/* index of first pointer to log */
+	int			last)	/* index of last pointer to log */
+{
+
+	if (bp) {
+		struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
+		int			level = xfs_btree_get_level(block);
+
+		xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
+		xfs_trans_log_buf(cur->bc_tp, bp,
+				xfs_btree_ptr_offset(cur, first, level),
+				xfs_btree_ptr_offset(cur, last + 1, level) - 1);
+	} else {
+		xfs_trans_log_inode(cur->bc_tp, cur->bc_ino.ip,
+			xfs_ilog_fbroot(cur->bc_ino.whichfork));
+	}
+
+}
+
+/*
+ * Log fields from a btree block header.
+ */
+void
+xfs_btree_log_block(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	struct xfs_buf		*bp,	/* buffer containing btree block */
+	uint32_t		fields)	/* mask of fields: XFS_BB_... */
+{
+	int			first;	/* first byte offset logged */
+	int			last;	/* last byte offset logged */
+	static const short	soffsets[] = {	/* table of offsets (short) */
+		offsetof(struct xfs_btree_block, bb_magic),
+		offsetof(struct xfs_btree_block, bb_level),
+		offsetof(struct xfs_btree_block, bb_numrecs),
+		offsetof(struct xfs_btree_block, bb_u.s.bb_leftsib),
+		offsetof(struct xfs_btree_block, bb_u.s.bb_rightsib),
+		offsetof(struct xfs_btree_block, bb_u.s.bb_blkno),
+		offsetof(struct xfs_btree_block, bb_u.s.bb_lsn),
+		offsetof(struct xfs_btree_block, bb_u.s.bb_uuid),
+		offsetof(struct xfs_btree_block, bb_u.s.bb_owner),
+		offsetof(struct xfs_btree_block, bb_u.s.bb_crc),
+		XFS_BTREE_SBLOCK_CRC_LEN
+	};
+	static const short	loffsets[] = {	/* table of offsets (long) */
+		offsetof(struct xfs_btree_block, bb_magic),
+		offsetof(struct xfs_btree_block, bb_level),
+		offsetof(struct xfs_btree_block, bb_numrecs),
+		offsetof(struct xfs_btree_block, bb_u.l.bb_leftsib),
+		offsetof(struct xfs_btree_block, bb_u.l.bb_rightsib),
+		offsetof(struct xfs_btree_block, bb_u.l.bb_blkno),
+		offsetof(struct xfs_btree_block, bb_u.l.bb_lsn),
+		offsetof(struct xfs_btree_block, bb_u.l.bb_uuid),
+		offsetof(struct xfs_btree_block, bb_u.l.bb_owner),
+		offsetof(struct xfs_btree_block, bb_u.l.bb_crc),
+		offsetof(struct xfs_btree_block, bb_u.l.bb_pad),
+		XFS_BTREE_LBLOCK_CRC_LEN
+	};
+
+	if (bp) {
+		int nbits;
+
+		if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS) {
+			/*
+			 * We don't log the CRC when updating a btree
+			 * block but instead recreate it during log
+			 * recovery.  As the log buffers have checksums
+			 * of their own this is safe and avoids logging a crc
+			 * update in a lot of places.
+			 */
+			if (fields == XFS_BB_ALL_BITS)
+				fields = XFS_BB_ALL_BITS_CRC;
+			nbits = XFS_BB_NUM_BITS_CRC;
+		} else {
+			nbits = XFS_BB_NUM_BITS;
+		}
+		xfs_btree_offsets(fields,
+				  (cur->bc_flags & XFS_BTREE_LONG_PTRS) ?
+					loffsets : soffsets,
+				  nbits, &first, &last);
+		xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
+		xfs_trans_log_buf(cur->bc_tp, bp, first, last);
+	} else {
+		xfs_trans_log_inode(cur->bc_tp, cur->bc_ino.ip,
+			xfs_ilog_fbroot(cur->bc_ino.whichfork));
+	}
+}
+
+/*
+ * Increment cursor by one record at the level.
+ * For nonzero levels the leaf-ward information is untouched.
+ */
+int						/* error */
+xfs_btree_increment(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			*stat)		/* success/failure */
+{
+	struct xfs_btree_block	*block;
+	union xfs_btree_ptr	ptr;
+	struct xfs_buf		*bp;
+	int			error;		/* error return value */
+	int			lev;
+
+	ASSERT(level < cur->bc_nlevels);
+
+	/* Read-ahead to the right at this level. */
+	xfs_btree_readahead(cur, level, XFS_BTCUR_RIGHTRA);
+
+	/* Get a pointer to the btree block. */
+	block = xfs_btree_get_block(cur, level, &bp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, level, bp);
+	if (error)
+		goto error0;
+#endif
+
+	/* We're done if we remain in the block after the increment. */
+	if (++cur->bc_levels[level].ptr <= xfs_btree_get_numrecs(block))
+		goto out1;
+
+	/* Fail if we just went off the right edge of the tree. */
+	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
+	if (xfs_btree_ptr_is_null(cur, &ptr))
+		goto out0;
+
+	XFS_BTREE_STATS_INC(cur, increment);
+
+	/*
+	 * March up the tree incrementing pointers.
+	 * Stop when we don't go off the right edge of a block.
+	 */
+	for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
+		block = xfs_btree_get_block(cur, lev, &bp);
+
+#ifdef DEBUG
+		error = xfs_btree_check_block(cur, block, lev, bp);
+		if (error)
+			goto error0;
+#endif
+
+		if (++cur->bc_levels[lev].ptr <= xfs_btree_get_numrecs(block))
+			break;
+
+		/* Read-ahead the right block for the next loop. */
+		xfs_btree_readahead(cur, lev, XFS_BTCUR_RIGHTRA);
+	}
+
+	/*
+	 * If we went off the root then we are either seriously
+	 * confused or have the tree root in an inode.
+	 */
+	if (lev == cur->bc_nlevels) {
+		if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE)
+			goto out0;
+		ASSERT(0);
+		error = -EFSCORRUPTED;
+		goto error0;
+	}
+	ASSERT(lev < cur->bc_nlevels);
+
+	/*
+	 * Now walk back down the tree, fixing up the cursor's buffer
+	 * pointers and key numbers.
+	 */
+	for (block = xfs_btree_get_block(cur, lev, &bp); lev > level; ) {
+		union xfs_btree_ptr	*ptrp;
+
+		ptrp = xfs_btree_ptr_addr(cur, cur->bc_levels[lev].ptr, block);
+		--lev;
+		error = xfs_btree_read_buf_block(cur, ptrp, 0, &block, &bp);
+		if (error)
+			goto error0;
+
+		xfs_btree_setbuf(cur, lev, bp);
+		cur->bc_levels[lev].ptr = 1;
+	}
+out1:
+	*stat = 1;
+	return 0;
+
+out0:
+	*stat = 0;
+	return 0;
+
+error0:
+	return error;
+}
+
+/*
+ * Decrement cursor by one record at the level.
+ * For nonzero levels the leaf-ward information is untouched.
+ */
+int						/* error */
+xfs_btree_decrement(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			*stat)		/* success/failure */
+{
+	struct xfs_btree_block	*block;
+	struct xfs_buf		*bp;
+	int			error;		/* error return value */
+	int			lev;
+	union xfs_btree_ptr	ptr;
+
+	ASSERT(level < cur->bc_nlevels);
+
+	/* Read-ahead to the left at this level. */
+	xfs_btree_readahead(cur, level, XFS_BTCUR_LEFTRA);
+
+	/* We're done if we remain in the block after the decrement. */
+	if (--cur->bc_levels[level].ptr > 0)
+		goto out1;
+
+	/* Get a pointer to the btree block. */
+	block = xfs_btree_get_block(cur, level, &bp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, level, bp);
+	if (error)
+		goto error0;
+#endif
+
+	/* Fail if we just went off the left edge of the tree. */
+	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_LEFTSIB);
+	if (xfs_btree_ptr_is_null(cur, &ptr))
+		goto out0;
+
+	XFS_BTREE_STATS_INC(cur, decrement);
+
+	/*
+	 * March up the tree decrementing pointers.
+	 * Stop when we don't go off the left edge of a block.
+	 */
+	for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
+		if (--cur->bc_levels[lev].ptr > 0)
+			break;
+		/* Read-ahead the left block for the next loop. */
+		xfs_btree_readahead(cur, lev, XFS_BTCUR_LEFTRA);
+	}
+
+	/*
+	 * If we went off the root then we are seriously confused.
+	 * or the root of the tree is in an inode.
+	 */
+	if (lev == cur->bc_nlevels) {
+		if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE)
+			goto out0;
+		ASSERT(0);
+		error = -EFSCORRUPTED;
+		goto error0;
+	}
+	ASSERT(lev < cur->bc_nlevels);
+
+	/*
+	 * Now walk back down the tree, fixing up the cursor's buffer
+	 * pointers and key numbers.
+	 */
+	for (block = xfs_btree_get_block(cur, lev, &bp); lev > level; ) {
+		union xfs_btree_ptr	*ptrp;
+
+		ptrp = xfs_btree_ptr_addr(cur, cur->bc_levels[lev].ptr, block);
+		--lev;
+		error = xfs_btree_read_buf_block(cur, ptrp, 0, &block, &bp);
+		if (error)
+			goto error0;
+		xfs_btree_setbuf(cur, lev, bp);
+		cur->bc_levels[lev].ptr = xfs_btree_get_numrecs(block);
+	}
+out1:
+	*stat = 1;
+	return 0;
+
+out0:
+	*stat = 0;
+	return 0;
+
+error0:
+	return error;
+}
+
+int
+xfs_btree_lookup_get_block(
+	struct xfs_btree_cur		*cur,	/* btree cursor */
+	int				level,	/* level in the btree */
+	const union xfs_btree_ptr	*pp,	/* ptr to btree block */
+	struct xfs_btree_block		**blkp) /* return btree block */
+{
+	struct xfs_buf		*bp;	/* buffer pointer for btree block */
+	xfs_daddr_t		daddr;
+	int			error = 0;
+
+	/* special case the root block if in an inode */
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    (level == cur->bc_nlevels - 1)) {
+		*blkp = xfs_btree_get_iroot(cur);
+		return 0;
+	}
+
+	/*
+	 * If the old buffer at this level for the disk address we are
+	 * looking for re-use it.
+	 *
+	 * Otherwise throw it away and get a new one.
+	 */
+	bp = cur->bc_levels[level].bp;
+	error = xfs_btree_ptr_to_daddr(cur, pp, &daddr);
+	if (error)
+		return error;
+	if (bp && xfs_buf_daddr(bp) == daddr) {
+		*blkp = XFS_BUF_TO_BLOCK(bp);
+		return 0;
+	}
+
+	error = xfs_btree_read_buf_block(cur, pp, 0, blkp, &bp);
+	if (error)
+		return error;
+
+	/* Check the inode owner since the verifiers don't. */
+	if (xfs_has_crc(cur->bc_mp) &&
+	    !(cur->bc_ino.flags & XFS_BTCUR_BMBT_INVALID_OWNER) &&
+	    (cur->bc_flags & XFS_BTREE_LONG_PTRS) &&
+	    be64_to_cpu((*blkp)->bb_u.l.bb_owner) !=
+			cur->bc_ino.ip->i_ino)
+		goto out_bad;
+
+	/* Did we get the level we were looking for? */
+	if (be16_to_cpu((*blkp)->bb_level) != level)
+		goto out_bad;
+
+	/* Check that internal nodes have at least one record. */
+	if (level != 0 && be16_to_cpu((*blkp)->bb_numrecs) == 0)
+		goto out_bad;
+
+	xfs_btree_setbuf(cur, level, bp);
+	return 0;
+
+out_bad:
+	*blkp = NULL;
+	xfs_buf_mark_corrupt(bp);
+	xfs_trans_brelse(cur->bc_tp, bp);
+	return -EFSCORRUPTED;
+}
+
+/*
+ * Get current search key.  For level 0 we don't actually have a key
+ * structure so we make one up from the record.  For all other levels
+ * we just return the right key.
+ */
+STATIC union xfs_btree_key *
+xfs_lookup_get_search_key(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			keyno,
+	struct xfs_btree_block	*block,
+	union xfs_btree_key	*kp)
+{
+	if (level == 0) {
+		cur->bc_ops->init_key_from_rec(kp,
+				xfs_btree_rec_addr(cur, keyno, block));
+		return kp;
+	}
+
+	return xfs_btree_key_addr(cur, keyno, block);
+}
+
+/*
+ * Lookup the record.  The cursor is made to point to it, based on dir.
+ * stat is set to 0 if can't find any such record, 1 for success.
+ */
+int					/* error */
+xfs_btree_lookup(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_lookup_t		dir,	/* <=, ==, or >= */
+	int			*stat)	/* success/failure */
+{
+	struct xfs_btree_block	*block;	/* current btree block */
+	int64_t			diff;	/* difference for the current key */
+	int			error;	/* error return value */
+	int			keyno;	/* current key number */
+	int			level;	/* level in the btree */
+	union xfs_btree_ptr	*pp;	/* ptr to btree block */
+	union xfs_btree_ptr	ptr;	/* ptr to btree block */
+
+	XFS_BTREE_STATS_INC(cur, lookup);
+
+	/* No such thing as a zero-level tree. */
+	if (XFS_IS_CORRUPT(cur->bc_mp, cur->bc_nlevels == 0))
+		return -EFSCORRUPTED;
+
+	block = NULL;
+	keyno = 0;
+
+	/* initialise start pointer from cursor */
+	cur->bc_ops->init_ptr_from_cur(cur, &ptr);
+	pp = &ptr;
+
+	/*
+	 * Iterate over each level in the btree, starting at the root.
+	 * For each level above the leaves, find the key we need, based
+	 * on the lookup record, then follow the corresponding block
+	 * pointer down to the next level.
+	 */
+	for (level = cur->bc_nlevels - 1, diff = 1; level >= 0; level--) {
+		/* Get the block we need to do the lookup on. */
+		error = xfs_btree_lookup_get_block(cur, level, pp, &block);
+		if (error)
+			goto error0;
+
+		if (diff == 0) {
+			/*
+			 * If we already had a key match at a higher level, we
+			 * know we need to use the first entry in this block.
+			 */
+			keyno = 1;
+		} else {
+			/* Otherwise search this block. Do a binary search. */
+
+			int	high;	/* high entry number */
+			int	low;	/* low entry number */
+
+			/* Set low and high entry numbers, 1-based. */
+			low = 1;
+			high = xfs_btree_get_numrecs(block);
+			if (!high) {
+				/* Block is empty, must be an empty leaf. */
+				if (level != 0 || cur->bc_nlevels != 1) {
+					XFS_CORRUPTION_ERROR(__func__,
+							XFS_ERRLEVEL_LOW,
+							cur->bc_mp, block,
+							sizeof(*block));
+					return -EFSCORRUPTED;
+				}
+
+				cur->bc_levels[0].ptr = dir != XFS_LOOKUP_LE;
+				*stat = 0;
+				return 0;
+			}
+
+			/* Binary search the block. */
+			while (low <= high) {
+				union xfs_btree_key	key;
+				union xfs_btree_key	*kp;
+
+				XFS_BTREE_STATS_INC(cur, compare);
+
+				/* keyno is average of low and high. */
+				keyno = (low + high) >> 1;
+
+				/* Get current search key */
+				kp = xfs_lookup_get_search_key(cur, level,
+						keyno, block, &key);
+
+				/*
+				 * Compute difference to get next direction:
+				 *  - less than, move right
+				 *  - greater than, move left
+				 *  - equal, we're done
+				 */
+				diff = cur->bc_ops->key_diff(cur, kp);
+				if (diff < 0)
+					low = keyno + 1;
+				else if (diff > 0)
+					high = keyno - 1;
+				else
+					break;
+			}
+		}
+
+		/*
+		 * If there are more levels, set up for the next level
+		 * by getting the block number and filling in the cursor.
+		 */
+		if (level > 0) {
+			/*
+			 * If we moved left, need the previous key number,
+			 * unless there isn't one.
+			 */
+			if (diff > 0 && --keyno < 1)
+				keyno = 1;
+			pp = xfs_btree_ptr_addr(cur, keyno, block);
+
+			error = xfs_btree_debug_check_ptr(cur, pp, 0, level);
+			if (error)
+				goto error0;
+
+			cur->bc_levels[level].ptr = keyno;
+		}
+	}
+
+	/* Done with the search. See if we need to adjust the results. */
+	if (dir != XFS_LOOKUP_LE && diff < 0) {
+		keyno++;
+		/*
+		 * If ge search and we went off the end of the block, but it's
+		 * not the last block, we're in the wrong block.
+		 */
+		xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
+		if (dir == XFS_LOOKUP_GE &&
+		    keyno > xfs_btree_get_numrecs(block) &&
+		    !xfs_btree_ptr_is_null(cur, &ptr)) {
+			int	i;
+
+			cur->bc_levels[0].ptr = keyno;
+			error = xfs_btree_increment(cur, 0, &i);
+			if (error)
+				goto error0;
+			if (XFS_IS_CORRUPT(cur->bc_mp, i != 1))
+				return -EFSCORRUPTED;
+			*stat = 1;
+			return 0;
+		}
+	} else if (dir == XFS_LOOKUP_LE && diff > 0)
+		keyno--;
+	cur->bc_levels[0].ptr = keyno;
+
+	/* Return if we succeeded or not. */
+	if (keyno == 0 || keyno > xfs_btree_get_numrecs(block))
+		*stat = 0;
+	else if (dir != XFS_LOOKUP_EQ || diff == 0)
+		*stat = 1;
+	else
+		*stat = 0;
+	return 0;
+
+error0:
+	return error;
+}
+
+/* Find the high key storage area from a regular key. */
+union xfs_btree_key *
+xfs_btree_high_key_from_key(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*key)
+{
+	ASSERT(cur->bc_flags & XFS_BTREE_OVERLAPPING);
+	return (union xfs_btree_key *)((char *)key +
+			(cur->bc_ops->key_len / 2));
+}
+
+/* Determine the low (and high if overlapped) keys of a leaf block */
+STATIC void
+xfs_btree_get_leaf_keys(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_block	*block,
+	union xfs_btree_key	*key)
+{
+	union xfs_btree_key	max_hkey;
+	union xfs_btree_key	hkey;
+	union xfs_btree_rec	*rec;
+	union xfs_btree_key	*high;
+	int			n;
+
+	rec = xfs_btree_rec_addr(cur, 1, block);
+	cur->bc_ops->init_key_from_rec(key, rec);
+
+	if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
+
+		cur->bc_ops->init_high_key_from_rec(&max_hkey, rec);
+		for (n = 2; n <= xfs_btree_get_numrecs(block); n++) {
+			rec = xfs_btree_rec_addr(cur, n, block);
+			cur->bc_ops->init_high_key_from_rec(&hkey, rec);
+			if (cur->bc_ops->diff_two_keys(cur, &hkey, &max_hkey)
+					> 0)
+				max_hkey = hkey;
+		}
+
+		high = xfs_btree_high_key_from_key(cur, key);
+		memcpy(high, &max_hkey, cur->bc_ops->key_len / 2);
+	}
+}
+
+/* Determine the low (and high if overlapped) keys of a node block */
+STATIC void
+xfs_btree_get_node_keys(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_block	*block,
+	union xfs_btree_key	*key)
+{
+	union xfs_btree_key	*hkey;
+	union xfs_btree_key	*max_hkey;
+	union xfs_btree_key	*high;
+	int			n;
+
+	if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
+		memcpy(key, xfs_btree_key_addr(cur, 1, block),
+				cur->bc_ops->key_len / 2);
+
+		max_hkey = xfs_btree_high_key_addr(cur, 1, block);
+		for (n = 2; n <= xfs_btree_get_numrecs(block); n++) {
+			hkey = xfs_btree_high_key_addr(cur, n, block);
+			if (cur->bc_ops->diff_two_keys(cur, hkey, max_hkey) > 0)
+				max_hkey = hkey;
+		}
+
+		high = xfs_btree_high_key_from_key(cur, key);
+		memcpy(high, max_hkey, cur->bc_ops->key_len / 2);
+	} else {
+		memcpy(key, xfs_btree_key_addr(cur, 1, block),
+				cur->bc_ops->key_len);
+	}
+}
+
+/* Derive the keys for any btree block. */
+void
+xfs_btree_get_keys(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_block	*block,
+	union xfs_btree_key	*key)
+{
+	if (be16_to_cpu(block->bb_level) == 0)
+		xfs_btree_get_leaf_keys(cur, block, key);
+	else
+		xfs_btree_get_node_keys(cur, block, key);
+}
+
+/*
+ * Decide if we need to update the parent keys of a btree block.  For
+ * a standard btree this is only necessary if we're updating the first
+ * record/key.  For an overlapping btree, we must always update the
+ * keys because the highest key can be in any of the records or keys
+ * in the block.
+ */
+static inline bool
+xfs_btree_needs_key_update(
+	struct xfs_btree_cur	*cur,
+	int			ptr)
+{
+	return (cur->bc_flags & XFS_BTREE_OVERLAPPING) || ptr == 1;
+}
+
+/*
+ * Update the low and high parent keys of the given level, progressing
+ * towards the root.  If force_all is false, stop if the keys for a given
+ * level do not need updating.
+ */
+STATIC int
+__xfs_btree_updkeys(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	struct xfs_btree_block	*block,
+	struct xfs_buf		*bp0,
+	bool			force_all)
+{
+	union xfs_btree_key	key;	/* keys from current level */
+	union xfs_btree_key	*lkey;	/* keys from the next level up */
+	union xfs_btree_key	*hkey;
+	union xfs_btree_key	*nlkey;	/* keys from the next level up */
+	union xfs_btree_key	*nhkey;
+	struct xfs_buf		*bp;
+	int			ptr;
+
+	ASSERT(cur->bc_flags & XFS_BTREE_OVERLAPPING);
+
+	/* Exit if there aren't any parent levels to update. */
+	if (level + 1 >= cur->bc_nlevels)
+		return 0;
+
+	trace_xfs_btree_updkeys(cur, level, bp0);
+
+	lkey = &key;
+	hkey = xfs_btree_high_key_from_key(cur, lkey);
+	xfs_btree_get_keys(cur, block, lkey);
+	for (level++; level < cur->bc_nlevels; level++) {
+#ifdef DEBUG
+		int		error;
+#endif
+		block = xfs_btree_get_block(cur, level, &bp);
+		trace_xfs_btree_updkeys(cur, level, bp);
+#ifdef DEBUG
+		error = xfs_btree_check_block(cur, block, level, bp);
+		if (error)
+			return error;
+#endif
+		ptr = cur->bc_levels[level].ptr;
+		nlkey = xfs_btree_key_addr(cur, ptr, block);
+		nhkey = xfs_btree_high_key_addr(cur, ptr, block);
+		if (!force_all &&
+		    !(cur->bc_ops->diff_two_keys(cur, nlkey, lkey) != 0 ||
+		      cur->bc_ops->diff_two_keys(cur, nhkey, hkey) != 0))
+			break;
+		xfs_btree_copy_keys(cur, nlkey, lkey, 1);
+		xfs_btree_log_keys(cur, bp, ptr, ptr);
+		if (level + 1 >= cur->bc_nlevels)
+			break;
+		xfs_btree_get_node_keys(cur, block, lkey);
+	}
+
+	return 0;
+}
+
+/* Update all the keys from some level in cursor back to the root. */
+STATIC int
+xfs_btree_updkeys_force(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	struct xfs_buf		*bp;
+	struct xfs_btree_block	*block;
+
+	block = xfs_btree_get_block(cur, level, &bp);
+	return __xfs_btree_updkeys(cur, level, block, bp, true);
+}
+
+/*
+ * Update the parent keys of the given level, progressing towards the root.
+ */
+STATIC int
+xfs_btree_update_keys(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	struct xfs_btree_block	*block;
+	struct xfs_buf		*bp;
+	union xfs_btree_key	*kp;
+	union xfs_btree_key	key;
+	int			ptr;
+
+	ASSERT(level >= 0);
+
+	block = xfs_btree_get_block(cur, level, &bp);
+	if (cur->bc_flags & XFS_BTREE_OVERLAPPING)
+		return __xfs_btree_updkeys(cur, level, block, bp, false);
+
+	/*
+	 * Go up the tree from this level toward the root.
+	 * At each level, update the key value to the value input.
+	 * Stop when we reach a level where the cursor isn't pointing
+	 * at the first entry in the block.
+	 */
+	xfs_btree_get_keys(cur, block, &key);
+	for (level++, ptr = 1; ptr == 1 && level < cur->bc_nlevels; level++) {
+#ifdef DEBUG
+		int		error;
+#endif
+		block = xfs_btree_get_block(cur, level, &bp);
+#ifdef DEBUG
+		error = xfs_btree_check_block(cur, block, level, bp);
+		if (error)
+			return error;
+#endif
+		ptr = cur->bc_levels[level].ptr;
+		kp = xfs_btree_key_addr(cur, ptr, block);
+		xfs_btree_copy_keys(cur, kp, &key, 1);
+		xfs_btree_log_keys(cur, bp, ptr, ptr);
+	}
+
+	return 0;
+}
+
+/*
+ * Update the record referred to by cur to the value in the
+ * given record. This either works (return 0) or gets an
+ * EFSCORRUPTED error.
+ */
+int
+xfs_btree_update(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec)
+{
+	struct xfs_btree_block	*block;
+	struct xfs_buf		*bp;
+	int			error;
+	int			ptr;
+	union xfs_btree_rec	*rp;
+
+	/* Pick up the current block. */
+	block = xfs_btree_get_block(cur, 0, &bp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, 0, bp);
+	if (error)
+		goto error0;
+#endif
+	/* Get the address of the rec to be updated. */
+	ptr = cur->bc_levels[0].ptr;
+	rp = xfs_btree_rec_addr(cur, ptr, block);
+
+	/* Fill in the new contents and log them. */
+	xfs_btree_copy_recs(cur, rp, rec, 1);
+	xfs_btree_log_recs(cur, bp, ptr, ptr);
+
+	/*
+	 * If we are tracking the last record in the tree and
+	 * we are at the far right edge of the tree, update it.
+	 */
+	if (xfs_btree_is_lastrec(cur, block, 0)) {
+		cur->bc_ops->update_lastrec(cur, block, rec,
+					    ptr, LASTREC_UPDATE);
+	}
+
+	/* Pass new key value up to our parent. */
+	if (xfs_btree_needs_key_update(cur, ptr)) {
+		error = xfs_btree_update_keys(cur, 0);
+		if (error)
+			goto error0;
+	}
+
+	return 0;
+
+error0:
+	return error;
+}
+
+/*
+ * Move 1 record left from cur/level if possible.
+ * Update cur to reflect the new path.
+ */
+STATIC int					/* error */
+xfs_btree_lshift(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			*stat)		/* success/failure */
+{
+	struct xfs_buf		*lbp;		/* left buffer pointer */
+	struct xfs_btree_block	*left;		/* left btree block */
+	int			lrecs;		/* left record count */
+	struct xfs_buf		*rbp;		/* right buffer pointer */
+	struct xfs_btree_block	*right;		/* right btree block */
+	struct xfs_btree_cur	*tcur;		/* temporary btree cursor */
+	int			rrecs;		/* right record count */
+	union xfs_btree_ptr	lptr;		/* left btree pointer */
+	union xfs_btree_key	*rkp = NULL;	/* right btree key */
+	union xfs_btree_ptr	*rpp = NULL;	/* right address pointer */
+	union xfs_btree_rec	*rrp = NULL;	/* right record pointer */
+	int			error;		/* error return value */
+	int			i;
+
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    level == cur->bc_nlevels - 1)
+		goto out0;
+
+	/* Set up variables for this block as "right". */
+	right = xfs_btree_get_block(cur, level, &rbp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, right, level, rbp);
+	if (error)
+		goto error0;
+#endif
+
+	/* If we've got no left sibling then we can't shift an entry left. */
+	xfs_btree_get_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
+	if (xfs_btree_ptr_is_null(cur, &lptr))
+		goto out0;
+
+	/*
+	 * If the cursor entry is the one that would be moved, don't
+	 * do it... it's too complicated.
+	 */
+	if (cur->bc_levels[level].ptr <= 1)
+		goto out0;
+
+	/* Set up the left neighbor as "left". */
+	error = xfs_btree_read_buf_block(cur, &lptr, 0, &left, &lbp);
+	if (error)
+		goto error0;
+
+	/* If it's full, it can't take another entry. */
+	lrecs = xfs_btree_get_numrecs(left);
+	if (lrecs == cur->bc_ops->get_maxrecs(cur, level))
+		goto out0;
+
+	rrecs = xfs_btree_get_numrecs(right);
+
+	/*
+	 * We add one entry to the left side and remove one for the right side.
+	 * Account for it here, the changes will be updated on disk and logged
+	 * later.
+	 */
+	lrecs++;
+	rrecs--;
+
+	XFS_BTREE_STATS_INC(cur, lshift);
+	XFS_BTREE_STATS_ADD(cur, moves, 1);
+
+	/*
+	 * If non-leaf, copy a key and a ptr to the left block.
+	 * Log the changes to the left block.
+	 */
+	if (level > 0) {
+		/* It's a non-leaf.  Move keys and pointers. */
+		union xfs_btree_key	*lkp;	/* left btree key */
+		union xfs_btree_ptr	*lpp;	/* left address pointer */
+
+		lkp = xfs_btree_key_addr(cur, lrecs, left);
+		rkp = xfs_btree_key_addr(cur, 1, right);
+
+		lpp = xfs_btree_ptr_addr(cur, lrecs, left);
+		rpp = xfs_btree_ptr_addr(cur, 1, right);
+
+		error = xfs_btree_debug_check_ptr(cur, rpp, 0, level);
+		if (error)
+			goto error0;
+
+		xfs_btree_copy_keys(cur, lkp, rkp, 1);
+		xfs_btree_copy_ptrs(cur, lpp, rpp, 1);
+
+		xfs_btree_log_keys(cur, lbp, lrecs, lrecs);
+		xfs_btree_log_ptrs(cur, lbp, lrecs, lrecs);
+
+		ASSERT(cur->bc_ops->keys_inorder(cur,
+			xfs_btree_key_addr(cur, lrecs - 1, left), lkp));
+	} else {
+		/* It's a leaf.  Move records.  */
+		union xfs_btree_rec	*lrp;	/* left record pointer */
+
+		lrp = xfs_btree_rec_addr(cur, lrecs, left);
+		rrp = xfs_btree_rec_addr(cur, 1, right);
+
+		xfs_btree_copy_recs(cur, lrp, rrp, 1);
+		xfs_btree_log_recs(cur, lbp, lrecs, lrecs);
+
+		ASSERT(cur->bc_ops->recs_inorder(cur,
+			xfs_btree_rec_addr(cur, lrecs - 1, left), lrp));
+	}
+
+	xfs_btree_set_numrecs(left, lrecs);
+	xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS);
+
+	xfs_btree_set_numrecs(right, rrecs);
+	xfs_btree_log_block(cur, rbp, XFS_BB_NUMRECS);
+
+	/*
+	 * Slide the contents of right down one entry.
+	 */
+	XFS_BTREE_STATS_ADD(cur, moves, rrecs - 1);
+	if (level > 0) {
+		/* It's a nonleaf. operate on keys and ptrs */
+		for (i = 0; i < rrecs; i++) {
+			error = xfs_btree_debug_check_ptr(cur, rpp, i + 1, level);
+			if (error)
+				goto error0;
+		}
+
+		xfs_btree_shift_keys(cur,
+				xfs_btree_key_addr(cur, 2, right),
+				-1, rrecs);
+		xfs_btree_shift_ptrs(cur,
+				xfs_btree_ptr_addr(cur, 2, right),
+				-1, rrecs);
+
+		xfs_btree_log_keys(cur, rbp, 1, rrecs);
+		xfs_btree_log_ptrs(cur, rbp, 1, rrecs);
+	} else {
+		/* It's a leaf. operate on records */
+		xfs_btree_shift_recs(cur,
+			xfs_btree_rec_addr(cur, 2, right),
+			-1, rrecs);
+		xfs_btree_log_recs(cur, rbp, 1, rrecs);
+	}
+
+	/*
+	 * Using a temporary cursor, update the parent key values of the
+	 * block on the left.
+	 */
+	if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
+		error = xfs_btree_dup_cursor(cur, &tcur);
+		if (error)
+			goto error0;
+		i = xfs_btree_firstrec(tcur, level);
+		if (XFS_IS_CORRUPT(tcur->bc_mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+
+		error = xfs_btree_decrement(tcur, level, &i);
+		if (error)
+			goto error1;
+
+		/* Update the parent high keys of the left block, if needed. */
+		error = xfs_btree_update_keys(tcur, level);
+		if (error)
+			goto error1;
+
+		xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+	}
+
+	/* Update the parent keys of the right block. */
+	error = xfs_btree_update_keys(cur, level);
+	if (error)
+		goto error0;
+
+	/* Slide the cursor value left one. */
+	cur->bc_levels[level].ptr--;
+
+	*stat = 1;
+	return 0;
+
+out0:
+	*stat = 0;
+	return 0;
+
+error0:
+	return error;
+
+error1:
+	xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/*
+ * Move 1 record right from cur/level if possible.
+ * Update cur to reflect the new path.
+ */
+STATIC int					/* error */
+xfs_btree_rshift(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			*stat)		/* success/failure */
+{
+	struct xfs_buf		*lbp;		/* left buffer pointer */
+	struct xfs_btree_block	*left;		/* left btree block */
+	struct xfs_buf		*rbp;		/* right buffer pointer */
+	struct xfs_btree_block	*right;		/* right btree block */
+	struct xfs_btree_cur	*tcur;		/* temporary btree cursor */
+	union xfs_btree_ptr	rptr;		/* right block pointer */
+	union xfs_btree_key	*rkp;		/* right btree key */
+	int			rrecs;		/* right record count */
+	int			lrecs;		/* left record count */
+	int			error;		/* error return value */
+	int			i;		/* loop counter */
+
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    (level == cur->bc_nlevels - 1))
+		goto out0;
+
+	/* Set up variables for this block as "left". */
+	left = xfs_btree_get_block(cur, level, &lbp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, left, level, lbp);
+	if (error)
+		goto error0;
+#endif
+
+	/* If we've got no right sibling then we can't shift an entry right. */
+	xfs_btree_get_sibling(cur, left, &rptr, XFS_BB_RIGHTSIB);
+	if (xfs_btree_ptr_is_null(cur, &rptr))
+		goto out0;
+
+	/*
+	 * If the cursor entry is the one that would be moved, don't
+	 * do it... it's too complicated.
+	 */
+	lrecs = xfs_btree_get_numrecs(left);
+	if (cur->bc_levels[level].ptr >= lrecs)
+		goto out0;
+
+	/* Set up the right neighbor as "right". */
+	error = xfs_btree_read_buf_block(cur, &rptr, 0, &right, &rbp);
+	if (error)
+		goto error0;
+
+	/* If it's full, it can't take another entry. */
+	rrecs = xfs_btree_get_numrecs(right);
+	if (rrecs == cur->bc_ops->get_maxrecs(cur, level))
+		goto out0;
+
+	XFS_BTREE_STATS_INC(cur, rshift);
+	XFS_BTREE_STATS_ADD(cur, moves, rrecs);
+
+	/*
+	 * Make a hole at the start of the right neighbor block, then
+	 * copy the last left block entry to the hole.
+	 */
+	if (level > 0) {
+		/* It's a nonleaf. make a hole in the keys and ptrs */
+		union xfs_btree_key	*lkp;
+		union xfs_btree_ptr	*lpp;
+		union xfs_btree_ptr	*rpp;
+
+		lkp = xfs_btree_key_addr(cur, lrecs, left);
+		lpp = xfs_btree_ptr_addr(cur, lrecs, left);
+		rkp = xfs_btree_key_addr(cur, 1, right);
+		rpp = xfs_btree_ptr_addr(cur, 1, right);
+
+		for (i = rrecs - 1; i >= 0; i--) {
+			error = xfs_btree_debug_check_ptr(cur, rpp, i, level);
+			if (error)
+				goto error0;
+		}
+
+		xfs_btree_shift_keys(cur, rkp, 1, rrecs);
+		xfs_btree_shift_ptrs(cur, rpp, 1, rrecs);
+
+		error = xfs_btree_debug_check_ptr(cur, lpp, 0, level);
+		if (error)
+			goto error0;
+
+		/* Now put the new data in, and log it. */
+		xfs_btree_copy_keys(cur, rkp, lkp, 1);
+		xfs_btree_copy_ptrs(cur, rpp, lpp, 1);
+
+		xfs_btree_log_keys(cur, rbp, 1, rrecs + 1);
+		xfs_btree_log_ptrs(cur, rbp, 1, rrecs + 1);
+
+		ASSERT(cur->bc_ops->keys_inorder(cur, rkp,
+			xfs_btree_key_addr(cur, 2, right)));
+	} else {
+		/* It's a leaf. make a hole in the records */
+		union xfs_btree_rec	*lrp;
+		union xfs_btree_rec	*rrp;
+
+		lrp = xfs_btree_rec_addr(cur, lrecs, left);
+		rrp = xfs_btree_rec_addr(cur, 1, right);
+
+		xfs_btree_shift_recs(cur, rrp, 1, rrecs);
+
+		/* Now put the new data in, and log it. */
+		xfs_btree_copy_recs(cur, rrp, lrp, 1);
+		xfs_btree_log_recs(cur, rbp, 1, rrecs + 1);
+	}
+
+	/*
+	 * Decrement and log left's numrecs, bump and log right's numrecs.
+	 */
+	xfs_btree_set_numrecs(left, --lrecs);
+	xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS);
+
+	xfs_btree_set_numrecs(right, ++rrecs);
+	xfs_btree_log_block(cur, rbp, XFS_BB_NUMRECS);
+
+	/*
+	 * Using a temporary cursor, update the parent key values of the
+	 * block on the right.
+	 */
+	error = xfs_btree_dup_cursor(cur, &tcur);
+	if (error)
+		goto error0;
+	i = xfs_btree_lastrec(tcur, level);
+	if (XFS_IS_CORRUPT(tcur->bc_mp, i != 1)) {
+		error = -EFSCORRUPTED;
+		goto error0;
+	}
+
+	error = xfs_btree_increment(tcur, level, &i);
+	if (error)
+		goto error1;
+
+	/* Update the parent high keys of the left block, if needed. */
+	if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
+		error = xfs_btree_update_keys(cur, level);
+		if (error)
+			goto error1;
+	}
+
+	/* Update the parent keys of the right block. */
+	error = xfs_btree_update_keys(tcur, level);
+	if (error)
+		goto error1;
+
+	xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+
+	*stat = 1;
+	return 0;
+
+out0:
+	*stat = 0;
+	return 0;
+
+error0:
+	return error;
+
+error1:
+	xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/*
+ * Split cur/level block in half.
+ * Return new block number and the key to its first
+ * record (to be inserted into parent).
+ */
+STATIC int					/* error */
+__xfs_btree_split(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	union xfs_btree_ptr	*ptrp,
+	union xfs_btree_key	*key,
+	struct xfs_btree_cur	**curp,
+	int			*stat)		/* success/failure */
+{
+	union xfs_btree_ptr	lptr;		/* left sibling block ptr */
+	struct xfs_buf		*lbp;		/* left buffer pointer */
+	struct xfs_btree_block	*left;		/* left btree block */
+	union xfs_btree_ptr	rptr;		/* right sibling block ptr */
+	struct xfs_buf		*rbp;		/* right buffer pointer */
+	struct xfs_btree_block	*right;		/* right btree block */
+	union xfs_btree_ptr	rrptr;		/* right-right sibling ptr */
+	struct xfs_buf		*rrbp;		/* right-right buffer pointer */
+	struct xfs_btree_block	*rrblock;	/* right-right btree block */
+	int			lrecs;
+	int			rrecs;
+	int			src_index;
+	int			error;		/* error return value */
+	int			i;
+
+	XFS_BTREE_STATS_INC(cur, split);
+
+	/* Set up left block (current one). */
+	left = xfs_btree_get_block(cur, level, &lbp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, left, level, lbp);
+	if (error)
+		goto error0;
+#endif
+
+	xfs_btree_buf_to_ptr(cur, lbp, &lptr);
+
+	/* Allocate the new block. If we can't do it, we're toast. Give up. */
+	error = cur->bc_ops->alloc_block(cur, &lptr, &rptr, stat);
+	if (error)
+		goto error0;
+	if (*stat == 0)
+		goto out0;
+	XFS_BTREE_STATS_INC(cur, alloc);
+
+	/* Set up the new block as "right". */
+	error = xfs_btree_get_buf_block(cur, &rptr, &right, &rbp);
+	if (error)
+		goto error0;
+
+	/* Fill in the btree header for the new right block. */
+	xfs_btree_init_block_cur(cur, rbp, xfs_btree_get_level(left), 0);
+
+	/*
+	 * Split the entries between the old and the new block evenly.
+	 * Make sure that if there's an odd number of entries now, that
+	 * each new block will have the same number of entries.
+	 */
+	lrecs = xfs_btree_get_numrecs(left);
+	rrecs = lrecs / 2;
+	if ((lrecs & 1) && cur->bc_levels[level].ptr <= rrecs + 1)
+		rrecs++;
+	src_index = (lrecs - rrecs + 1);
+
+	XFS_BTREE_STATS_ADD(cur, moves, rrecs);
+
+	/* Adjust numrecs for the later get_*_keys() calls. */
+	lrecs -= rrecs;
+	xfs_btree_set_numrecs(left, lrecs);
+	xfs_btree_set_numrecs(right, xfs_btree_get_numrecs(right) + rrecs);
+
+	/*
+	 * Copy btree block entries from the left block over to the
+	 * new block, the right. Update the right block and log the
+	 * changes.
+	 */
+	if (level > 0) {
+		/* It's a non-leaf.  Move keys and pointers. */
+		union xfs_btree_key	*lkp;	/* left btree key */
+		union xfs_btree_ptr	*lpp;	/* left address pointer */
+		union xfs_btree_key	*rkp;	/* right btree key */
+		union xfs_btree_ptr	*rpp;	/* right address pointer */
+
+		lkp = xfs_btree_key_addr(cur, src_index, left);
+		lpp = xfs_btree_ptr_addr(cur, src_index, left);
+		rkp = xfs_btree_key_addr(cur, 1, right);
+		rpp = xfs_btree_ptr_addr(cur, 1, right);
+
+		for (i = src_index; i < rrecs; i++) {
+			error = xfs_btree_debug_check_ptr(cur, lpp, i, level);
+			if (error)
+				goto error0;
+		}
+
+		/* Copy the keys & pointers to the new block. */
+		xfs_btree_copy_keys(cur, rkp, lkp, rrecs);
+		xfs_btree_copy_ptrs(cur, rpp, lpp, rrecs);
+
+		xfs_btree_log_keys(cur, rbp, 1, rrecs);
+		xfs_btree_log_ptrs(cur, rbp, 1, rrecs);
+
+		/* Stash the keys of the new block for later insertion. */
+		xfs_btree_get_node_keys(cur, right, key);
+	} else {
+		/* It's a leaf.  Move records.  */
+		union xfs_btree_rec	*lrp;	/* left record pointer */
+		union xfs_btree_rec	*rrp;	/* right record pointer */
+
+		lrp = xfs_btree_rec_addr(cur, src_index, left);
+		rrp = xfs_btree_rec_addr(cur, 1, right);
+
+		/* Copy records to the new block. */
+		xfs_btree_copy_recs(cur, rrp, lrp, rrecs);
+		xfs_btree_log_recs(cur, rbp, 1, rrecs);
+
+		/* Stash the keys of the new block for later insertion. */
+		xfs_btree_get_leaf_keys(cur, right, key);
+	}
+
+	/*
+	 * Find the left block number by looking in the buffer.
+	 * Adjust sibling pointers.
+	 */
+	xfs_btree_get_sibling(cur, left, &rrptr, XFS_BB_RIGHTSIB);
+	xfs_btree_set_sibling(cur, right, &rrptr, XFS_BB_RIGHTSIB);
+	xfs_btree_set_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
+	xfs_btree_set_sibling(cur, left, &rptr, XFS_BB_RIGHTSIB);
+
+	xfs_btree_log_block(cur, rbp, XFS_BB_ALL_BITS);
+	xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
+
+	/*
+	 * If there's a block to the new block's right, make that block
+	 * point back to right instead of to left.
+	 */
+	if (!xfs_btree_ptr_is_null(cur, &rrptr)) {
+		error = xfs_btree_read_buf_block(cur, &rrptr,
+							0, &rrblock, &rrbp);
+		if (error)
+			goto error0;
+		xfs_btree_set_sibling(cur, rrblock, &rptr, XFS_BB_LEFTSIB);
+		xfs_btree_log_block(cur, rrbp, XFS_BB_LEFTSIB);
+	}
+
+	/* Update the parent high keys of the left block, if needed. */
+	if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
+		error = xfs_btree_update_keys(cur, level);
+		if (error)
+			goto error0;
+	}
+
+	/*
+	 * If the cursor is really in the right block, move it there.
+	 * If it's just pointing past the last entry in left, then we'll
+	 * insert there, so don't change anything in that case.
+	 */
+	if (cur->bc_levels[level].ptr > lrecs + 1) {
+		xfs_btree_setbuf(cur, level, rbp);
+		cur->bc_levels[level].ptr -= lrecs;
+	}
+	/*
+	 * If there are more levels, we'll need another cursor which refers
+	 * the right block, no matter where this cursor was.
+	 */
+	if (level + 1 < cur->bc_nlevels) {
+		error = xfs_btree_dup_cursor(cur, curp);
+		if (error)
+			goto error0;
+		(*curp)->bc_levels[level + 1].ptr++;
+	}
+	*ptrp = rptr;
+	*stat = 1;
+	return 0;
+out0:
+	*stat = 0;
+	return 0;
+
+error0:
+	return error;
+}
+
+#ifdef __KERNEL__
+struct xfs_btree_split_args {
+	struct xfs_btree_cur	*cur;
+	int			level;
+	union xfs_btree_ptr	*ptrp;
+	union xfs_btree_key	*key;
+	struct xfs_btree_cur	**curp;
+	int			*stat;		/* success/failure */
+	int			result;
+	bool			kswapd;	/* allocation in kswapd context */
+	struct completion	*done;
+	struct work_struct	work;
+};
+
+/*
+ * Stack switching interfaces for allocation
+ */
+static void
+xfs_btree_split_worker(
+	struct work_struct	*work)
+{
+	struct xfs_btree_split_args	*args = container_of(work,
+						struct xfs_btree_split_args, work);
+	unsigned long		pflags;
+	unsigned long		new_pflags = 0;
+
+	/*
+	 * we are in a transaction context here, but may also be doing work
+	 * in kswapd context, and hence we may need to inherit that state
+	 * temporarily to ensure that we don't block waiting for memory reclaim
+	 * in any way.
+	 */
+	if (args->kswapd)
+		new_pflags |= PF_MEMALLOC | PF_KSWAPD;
+
+	current_set_flags_nested(&pflags, new_pflags);
+	xfs_trans_set_context(args->cur->bc_tp);
+
+	args->result = __xfs_btree_split(args->cur, args->level, args->ptrp,
+					 args->key, args->curp, args->stat);
+
+	xfs_trans_clear_context(args->cur->bc_tp);
+	current_restore_flags_nested(&pflags, new_pflags);
+
+	/*
+	 * Do not access args after complete() has run here. We don't own args
+	 * and the owner may run and free args before we return here.
+	 */
+	complete(args->done);
+
+}
+
+/*
+ * BMBT split requests often come in with little stack to work on. Push
+ * them off to a worker thread so there is lots of stack to use. For the other
+ * btree types, just call directly to avoid the context switch overhead here.
+ */
+STATIC int					/* error */
+xfs_btree_split(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	union xfs_btree_ptr	*ptrp,
+	union xfs_btree_key	*key,
+	struct xfs_btree_cur	**curp,
+	int			*stat)		/* success/failure */
+{
+	struct xfs_btree_split_args	args;
+	DECLARE_COMPLETION_ONSTACK(done);
+
+	if (cur->bc_btnum != XFS_BTNUM_BMAP)
+		return __xfs_btree_split(cur, level, ptrp, key, curp, stat);
+
+	args.cur = cur;
+	args.level = level;
+	args.ptrp = ptrp;
+	args.key = key;
+	args.curp = curp;
+	args.stat = stat;
+	args.done = &done;
+	args.kswapd = current_is_kswapd();
+	INIT_WORK_ONSTACK(&args.work, xfs_btree_split_worker);
+	queue_work(xfs_alloc_wq, &args.work);
+	wait_for_completion(&done);
+	destroy_work_on_stack(&args.work);
+	return args.result;
+}
+#else
+#define xfs_btree_split	__xfs_btree_split
+#endif /* __KERNEL__ */
+
+
+/*
+ * Copy the old inode root contents into a real block and make the
+ * broot point to it.
+ */
+int						/* error */
+xfs_btree_new_iroot(
+	struct xfs_btree_cur	*cur,		/* btree cursor */
+	int			*logflags,	/* logging flags for inode */
+	int			*stat)		/* return status - 0 fail */
+{
+	struct xfs_buf		*cbp;		/* buffer for cblock */
+	struct xfs_btree_block	*block;		/* btree block */
+	struct xfs_btree_block	*cblock;	/* child btree block */
+	union xfs_btree_key	*ckp;		/* child key pointer */
+	union xfs_btree_ptr	*cpp;		/* child ptr pointer */
+	union xfs_btree_key	*kp;		/* pointer to btree key */
+	union xfs_btree_ptr	*pp;		/* pointer to block addr */
+	union xfs_btree_ptr	nptr;		/* new block addr */
+	int			level;		/* btree level */
+	int			error;		/* error return code */
+	int			i;		/* loop counter */
+
+	XFS_BTREE_STATS_INC(cur, newroot);
+
+	ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
+
+	level = cur->bc_nlevels - 1;
+
+	block = xfs_btree_get_iroot(cur);
+	pp = xfs_btree_ptr_addr(cur, 1, block);
+
+	/* Allocate the new block. If we can't do it, we're toast. Give up. */
+	error = cur->bc_ops->alloc_block(cur, pp, &nptr, stat);
+	if (error)
+		goto error0;
+	if (*stat == 0)
+		return 0;
+
+	XFS_BTREE_STATS_INC(cur, alloc);
+
+	/* Copy the root into a real block. */
+	error = xfs_btree_get_buf_block(cur, &nptr, &cblock, &cbp);
+	if (error)
+		goto error0;
+
+	/*
+	 * we can't just memcpy() the root in for CRC enabled btree blocks.
+	 * In that case have to also ensure the blkno remains correct
+	 */
+	memcpy(cblock, block, xfs_btree_block_len(cur));
+	if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS) {
+		__be64 bno = cpu_to_be64(xfs_buf_daddr(cbp));
+		if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+			cblock->bb_u.l.bb_blkno = bno;
+		else
+			cblock->bb_u.s.bb_blkno = bno;
+	}
+
+	be16_add_cpu(&block->bb_level, 1);
+	xfs_btree_set_numrecs(block, 1);
+	cur->bc_nlevels++;
+	ASSERT(cur->bc_nlevels <= cur->bc_maxlevels);
+	cur->bc_levels[level + 1].ptr = 1;
+
+	kp = xfs_btree_key_addr(cur, 1, block);
+	ckp = xfs_btree_key_addr(cur, 1, cblock);
+	xfs_btree_copy_keys(cur, ckp, kp, xfs_btree_get_numrecs(cblock));
+
+	cpp = xfs_btree_ptr_addr(cur, 1, cblock);
+	for (i = 0; i < be16_to_cpu(cblock->bb_numrecs); i++) {
+		error = xfs_btree_debug_check_ptr(cur, pp, i, level);
+		if (error)
+			goto error0;
+	}
+
+	xfs_btree_copy_ptrs(cur, cpp, pp, xfs_btree_get_numrecs(cblock));
+
+	error = xfs_btree_debug_check_ptr(cur, &nptr, 0, level);
+	if (error)
+		goto error0;
+
+	xfs_btree_copy_ptrs(cur, pp, &nptr, 1);
+
+	xfs_iroot_realloc(cur->bc_ino.ip,
+			  1 - xfs_btree_get_numrecs(cblock),
+			  cur->bc_ino.whichfork);
+
+	xfs_btree_setbuf(cur, level, cbp);
+
+	/*
+	 * Do all this logging at the end so that
+	 * the root is at the right level.
+	 */
+	xfs_btree_log_block(cur, cbp, XFS_BB_ALL_BITS);
+	xfs_btree_log_keys(cur, cbp, 1, be16_to_cpu(cblock->bb_numrecs));
+	xfs_btree_log_ptrs(cur, cbp, 1, be16_to_cpu(cblock->bb_numrecs));
+
+	*logflags |=
+		XFS_ILOG_CORE | xfs_ilog_fbroot(cur->bc_ino.whichfork);
+	*stat = 1;
+	return 0;
+error0:
+	return error;
+}
+
+/*
+ * Allocate a new root block, fill it in.
+ */
+STATIC int				/* error */
+xfs_btree_new_root(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			*stat)	/* success/failure */
+{
+	struct xfs_btree_block	*block;	/* one half of the old root block */
+	struct xfs_buf		*bp;	/* buffer containing block */
+	int			error;	/* error return value */
+	struct xfs_buf		*lbp;	/* left buffer pointer */
+	struct xfs_btree_block	*left;	/* left btree block */
+	struct xfs_buf		*nbp;	/* new (root) buffer */
+	struct xfs_btree_block	*new;	/* new (root) btree block */
+	int			nptr;	/* new value for key index, 1 or 2 */
+	struct xfs_buf		*rbp;	/* right buffer pointer */
+	struct xfs_btree_block	*right;	/* right btree block */
+	union xfs_btree_ptr	rptr;
+	union xfs_btree_ptr	lptr;
+
+	XFS_BTREE_STATS_INC(cur, newroot);
+
+	/* initialise our start point from the cursor */
+	cur->bc_ops->init_ptr_from_cur(cur, &rptr);
+
+	/* Allocate the new block. If we can't do it, we're toast. Give up. */
+	error = cur->bc_ops->alloc_block(cur, &rptr, &lptr, stat);
+	if (error)
+		goto error0;
+	if (*stat == 0)
+		goto out0;
+	XFS_BTREE_STATS_INC(cur, alloc);
+
+	/* Set up the new block. */
+	error = xfs_btree_get_buf_block(cur, &lptr, &new, &nbp);
+	if (error)
+		goto error0;
+
+	/* Set the root in the holding structure  increasing the level by 1. */
+	cur->bc_ops->set_root(cur, &lptr, 1);
+
+	/*
+	 * At the previous root level there are now two blocks: the old root,
+	 * and the new block generated when it was split.  We don't know which
+	 * one the cursor is pointing at, so we set up variables "left" and
+	 * "right" for each case.
+	 */
+	block = xfs_btree_get_block(cur, cur->bc_nlevels - 1, &bp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, cur->bc_nlevels - 1, bp);
+	if (error)
+		goto error0;
+#endif
+
+	xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
+	if (!xfs_btree_ptr_is_null(cur, &rptr)) {
+		/* Our block is left, pick up the right block. */
+		lbp = bp;
+		xfs_btree_buf_to_ptr(cur, lbp, &lptr);
+		left = block;
+		error = xfs_btree_read_buf_block(cur, &rptr, 0, &right, &rbp);
+		if (error)
+			goto error0;
+		bp = rbp;
+		nptr = 1;
+	} else {
+		/* Our block is right, pick up the left block. */
+		rbp = bp;
+		xfs_btree_buf_to_ptr(cur, rbp, &rptr);
+		right = block;
+		xfs_btree_get_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
+		error = xfs_btree_read_buf_block(cur, &lptr, 0, &left, &lbp);
+		if (error)
+			goto error0;
+		bp = lbp;
+		nptr = 2;
+	}
+
+	/* Fill in the new block's btree header and log it. */
+	xfs_btree_init_block_cur(cur, nbp, cur->bc_nlevels, 2);
+	xfs_btree_log_block(cur, nbp, XFS_BB_ALL_BITS);
+	ASSERT(!xfs_btree_ptr_is_null(cur, &lptr) &&
+			!xfs_btree_ptr_is_null(cur, &rptr));
+
+	/* Fill in the key data in the new root. */
+	if (xfs_btree_get_level(left) > 0) {
+		/*
+		 * Get the keys for the left block's keys and put them directly
+		 * in the parent block.  Do the same for the right block.
+		 */
+		xfs_btree_get_node_keys(cur, left,
+				xfs_btree_key_addr(cur, 1, new));
+		xfs_btree_get_node_keys(cur, right,
+				xfs_btree_key_addr(cur, 2, new));
+	} else {
+		/*
+		 * Get the keys for the left block's records and put them
+		 * directly in the parent block.  Do the same for the right
+		 * block.
+		 */
+		xfs_btree_get_leaf_keys(cur, left,
+			xfs_btree_key_addr(cur, 1, new));
+		xfs_btree_get_leaf_keys(cur, right,
+			xfs_btree_key_addr(cur, 2, new));
+	}
+	xfs_btree_log_keys(cur, nbp, 1, 2);
+
+	/* Fill in the pointer data in the new root. */
+	xfs_btree_copy_ptrs(cur,
+		xfs_btree_ptr_addr(cur, 1, new), &lptr, 1);
+	xfs_btree_copy_ptrs(cur,
+		xfs_btree_ptr_addr(cur, 2, new), &rptr, 1);
+	xfs_btree_log_ptrs(cur, nbp, 1, 2);
+
+	/* Fix up the cursor. */
+	xfs_btree_setbuf(cur, cur->bc_nlevels, nbp);
+	cur->bc_levels[cur->bc_nlevels].ptr = nptr;
+	cur->bc_nlevels++;
+	ASSERT(cur->bc_nlevels <= cur->bc_maxlevels);
+	*stat = 1;
+	return 0;
+error0:
+	return error;
+out0:
+	*stat = 0;
+	return 0;
+}
+
+STATIC int
+xfs_btree_make_block_unfull(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			level,	/* btree level */
+	int			numrecs,/* # of recs in block */
+	int			*oindex,/* old tree index */
+	int			*index,	/* new tree index */
+	union xfs_btree_ptr	*nptr,	/* new btree ptr */
+	struct xfs_btree_cur	**ncur,	/* new btree cursor */
+	union xfs_btree_key	*key,	/* key of new block */
+	int			*stat)
+{
+	int			error = 0;
+
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    level == cur->bc_nlevels - 1) {
+		struct xfs_inode *ip = cur->bc_ino.ip;
+
+		if (numrecs < cur->bc_ops->get_dmaxrecs(cur, level)) {
+			/* A root block that can be made bigger. */
+			xfs_iroot_realloc(ip, 1, cur->bc_ino.whichfork);
+			*stat = 1;
+		} else {
+			/* A root block that needs replacing */
+			int	logflags = 0;
+
+			error = xfs_btree_new_iroot(cur, &logflags, stat);
+			if (error || *stat == 0)
+				return error;
+
+			xfs_trans_log_inode(cur->bc_tp, ip, logflags);
+		}
+
+		return 0;
+	}
+
+	/* First, try shifting an entry to the right neighbor. */
+	error = xfs_btree_rshift(cur, level, stat);
+	if (error || *stat)
+		return error;
+
+	/* Next, try shifting an entry to the left neighbor. */
+	error = xfs_btree_lshift(cur, level, stat);
+	if (error)
+		return error;
+
+	if (*stat) {
+		*oindex = *index = cur->bc_levels[level].ptr;
+		return 0;
+	}
+
+	/*
+	 * Next, try splitting the current block in half.
+	 *
+	 * If this works we have to re-set our variables because we
+	 * could be in a different block now.
+	 */
+	error = xfs_btree_split(cur, level, nptr, key, ncur, stat);
+	if (error || *stat == 0)
+		return error;
+
+
+	*index = cur->bc_levels[level].ptr;
+	return 0;
+}
+
+/*
+ * Insert one record/level.  Return information to the caller
+ * allowing the next level up to proceed if necessary.
+ */
+STATIC int
+xfs_btree_insrec(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			level,	/* level to insert record at */
+	union xfs_btree_ptr	*ptrp,	/* i/o: block number inserted */
+	union xfs_btree_rec	*rec,	/* record to insert */
+	union xfs_btree_key	*key,	/* i/o: block key for ptrp */
+	struct xfs_btree_cur	**curp,	/* output: new cursor replacing cur */
+	int			*stat)	/* success/failure */
+{
+	struct xfs_btree_block	*block;	/* btree block */
+	struct xfs_buf		*bp;	/* buffer for block */
+	union xfs_btree_ptr	nptr;	/* new block ptr */
+	struct xfs_btree_cur	*ncur = NULL;	/* new btree cursor */
+	union xfs_btree_key	nkey;	/* new block key */
+	union xfs_btree_key	*lkey;
+	int			optr;	/* old key/record index */
+	int			ptr;	/* key/record index */
+	int			numrecs;/* number of records */
+	int			error;	/* error return value */
+	int			i;
+	xfs_daddr_t		old_bn;
+
+	ncur = NULL;
+	lkey = &nkey;
+
+	/*
+	 * If we have an external root pointer, and we've made it to the
+	 * root level, allocate a new root block and we're done.
+	 */
+	if (!(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    (level >= cur->bc_nlevels)) {
+		error = xfs_btree_new_root(cur, stat);
+		xfs_btree_set_ptr_null(cur, ptrp);
+
+		return error;
+	}
+
+	/* If we're off the left edge, return failure. */
+	ptr = cur->bc_levels[level].ptr;
+	if (ptr == 0) {
+		*stat = 0;
+		return 0;
+	}
+
+	optr = ptr;
+
+	XFS_BTREE_STATS_INC(cur, insrec);
+
+	/* Get pointers to the btree buffer and block. */
+	block = xfs_btree_get_block(cur, level, &bp);
+	old_bn = bp ? xfs_buf_daddr(bp) : XFS_BUF_DADDR_NULL;
+	numrecs = xfs_btree_get_numrecs(block);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, level, bp);
+	if (error)
+		goto error0;
+
+	/* Check that the new entry is being inserted in the right place. */
+	if (ptr <= numrecs) {
+		if (level == 0) {
+			ASSERT(cur->bc_ops->recs_inorder(cur, rec,
+				xfs_btree_rec_addr(cur, ptr, block)));
+		} else {
+			ASSERT(cur->bc_ops->keys_inorder(cur, key,
+				xfs_btree_key_addr(cur, ptr, block)));
+		}
+	}
+#endif
+
+	/*
+	 * If the block is full, we can't insert the new entry until we
+	 * make the block un-full.
+	 */
+	xfs_btree_set_ptr_null(cur, &nptr);
+	if (numrecs == cur->bc_ops->get_maxrecs(cur, level)) {
+		error = xfs_btree_make_block_unfull(cur, level, numrecs,
+					&optr, &ptr, &nptr, &ncur, lkey, stat);
+		if (error || *stat == 0)
+			goto error0;
+	}
+
+	/*
+	 * The current block may have changed if the block was
+	 * previously full and we have just made space in it.
+	 */
+	block = xfs_btree_get_block(cur, level, &bp);
+	numrecs = xfs_btree_get_numrecs(block);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, level, bp);
+	if (error)
+		goto error0;
+#endif
+
+	/*
+	 * At this point we know there's room for our new entry in the block
+	 * we're pointing at.
+	 */
+	XFS_BTREE_STATS_ADD(cur, moves, numrecs - ptr + 1);
+
+	if (level > 0) {
+		/* It's a nonleaf. make a hole in the keys and ptrs */
+		union xfs_btree_key	*kp;
+		union xfs_btree_ptr	*pp;
+
+		kp = xfs_btree_key_addr(cur, ptr, block);
+		pp = xfs_btree_ptr_addr(cur, ptr, block);
+
+		for (i = numrecs - ptr; i >= 0; i--) {
+			error = xfs_btree_debug_check_ptr(cur, pp, i, level);
+			if (error)
+				goto error0;
+		}
+
+		xfs_btree_shift_keys(cur, kp, 1, numrecs - ptr + 1);
+		xfs_btree_shift_ptrs(cur, pp, 1, numrecs - ptr + 1);
+
+		error = xfs_btree_debug_check_ptr(cur, ptrp, 0, level);
+		if (error)
+			goto error0;
+
+		/* Now put the new data in, bump numrecs and log it. */
+		xfs_btree_copy_keys(cur, kp, key, 1);
+		xfs_btree_copy_ptrs(cur, pp, ptrp, 1);
+		numrecs++;
+		xfs_btree_set_numrecs(block, numrecs);
+		xfs_btree_log_ptrs(cur, bp, ptr, numrecs);
+		xfs_btree_log_keys(cur, bp, ptr, numrecs);
+#ifdef DEBUG
+		if (ptr < numrecs) {
+			ASSERT(cur->bc_ops->keys_inorder(cur, kp,
+				xfs_btree_key_addr(cur, ptr + 1, block)));
+		}
+#endif
+	} else {
+		/* It's a leaf. make a hole in the records */
+		union xfs_btree_rec             *rp;
+
+		rp = xfs_btree_rec_addr(cur, ptr, block);
+
+		xfs_btree_shift_recs(cur, rp, 1, numrecs - ptr + 1);
+
+		/* Now put the new data in, bump numrecs and log it. */
+		xfs_btree_copy_recs(cur, rp, rec, 1);
+		xfs_btree_set_numrecs(block, ++numrecs);
+		xfs_btree_log_recs(cur, bp, ptr, numrecs);
+#ifdef DEBUG
+		if (ptr < numrecs) {
+			ASSERT(cur->bc_ops->recs_inorder(cur, rp,
+				xfs_btree_rec_addr(cur, ptr + 1, block)));
+		}
+#endif
+	}
+
+	/* Log the new number of records in the btree header. */
+	xfs_btree_log_block(cur, bp, XFS_BB_NUMRECS);
+
+	/*
+	 * If we just inserted into a new tree block, we have to
+	 * recalculate nkey here because nkey is out of date.
+	 *
+	 * Otherwise we're just updating an existing block (having shoved
+	 * some records into the new tree block), so use the regular key
+	 * update mechanism.
+	 */
+	if (bp && xfs_buf_daddr(bp) != old_bn) {
+		xfs_btree_get_keys(cur, block, lkey);
+	} else if (xfs_btree_needs_key_update(cur, optr)) {
+		error = xfs_btree_update_keys(cur, level);
+		if (error)
+			goto error0;
+	}
+
+	/*
+	 * If we are tracking the last record in the tree and
+	 * we are at the far right edge of the tree, update it.
+	 */
+	if (xfs_btree_is_lastrec(cur, block, level)) {
+		cur->bc_ops->update_lastrec(cur, block, rec,
+					    ptr, LASTREC_INSREC);
+	}
+
+	/*
+	 * Return the new block number, if any.
+	 * If there is one, give back a record value and a cursor too.
+	 */
+	*ptrp = nptr;
+	if (!xfs_btree_ptr_is_null(cur, &nptr)) {
+		xfs_btree_copy_keys(cur, key, lkey, 1);
+		*curp = ncur;
+	}
+
+	*stat = 1;
+	return 0;
+
+error0:
+	if (ncur)
+		xfs_btree_del_cursor(ncur, error);
+	return error;
+}
+
+/*
+ * Insert the record at the point referenced by cur.
+ *
+ * A multi-level split of the tree on insert will invalidate the original
+ * cursor.  All callers of this function should assume that the cursor is
+ * no longer valid and revalidate it.
+ */
+int
+xfs_btree_insert(
+	struct xfs_btree_cur	*cur,
+	int			*stat)
+{
+	int			error;	/* error return value */
+	int			i;	/* result value, 0 for failure */
+	int			level;	/* current level number in btree */
+	union xfs_btree_ptr	nptr;	/* new block number (split result) */
+	struct xfs_btree_cur	*ncur;	/* new cursor (split result) */
+	struct xfs_btree_cur	*pcur;	/* previous level's cursor */
+	union xfs_btree_key	bkey;	/* key of block to insert */
+	union xfs_btree_key	*key;
+	union xfs_btree_rec	rec;	/* record to insert */
+
+	level = 0;
+	ncur = NULL;
+	pcur = cur;
+	key = &bkey;
+
+	xfs_btree_set_ptr_null(cur, &nptr);
+
+	/* Make a key out of the record data to be inserted, and save it. */
+	cur->bc_ops->init_rec_from_cur(cur, &rec);
+	cur->bc_ops->init_key_from_rec(key, &rec);
+
+	/*
+	 * Loop going up the tree, starting at the leaf level.
+	 * Stop when we don't get a split block, that must mean that
+	 * the insert is finished with this level.
+	 */
+	do {
+		/*
+		 * Insert nrec/nptr into this level of the tree.
+		 * Note if we fail, nptr will be null.
+		 */
+		error = xfs_btree_insrec(pcur, level, &nptr, &rec, key,
+				&ncur, &i);
+		if (error) {
+			if (pcur != cur)
+				xfs_btree_del_cursor(pcur, XFS_BTREE_ERROR);
+			goto error0;
+		}
+
+		if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+		level++;
+
+		/*
+		 * See if the cursor we just used is trash.
+		 * Can't trash the caller's cursor, but otherwise we should
+		 * if ncur is a new cursor or we're about to be done.
+		 */
+		if (pcur != cur &&
+		    (ncur || xfs_btree_ptr_is_null(cur, &nptr))) {
+			/* Save the state from the cursor before we trash it */
+			if (cur->bc_ops->update_cursor)
+				cur->bc_ops->update_cursor(pcur, cur);
+			cur->bc_nlevels = pcur->bc_nlevels;
+			xfs_btree_del_cursor(pcur, XFS_BTREE_NOERROR);
+		}
+		/* If we got a new cursor, switch to it. */
+		if (ncur) {
+			pcur = ncur;
+			ncur = NULL;
+		}
+	} while (!xfs_btree_ptr_is_null(cur, &nptr));
+
+	*stat = i;
+	return 0;
+error0:
+	return error;
+}
+
+/*
+ * Try to merge a non-leaf block back into the inode root.
+ *
+ * Note: the killroot names comes from the fact that we're effectively
+ * killing the old root block.  But because we can't just delete the
+ * inode we have to copy the single block it was pointing to into the
+ * inode.
+ */
+STATIC int
+xfs_btree_kill_iroot(
+	struct xfs_btree_cur	*cur)
+{
+	int			whichfork = cur->bc_ino.whichfork;
+	struct xfs_inode	*ip = cur->bc_ino.ip;
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_btree_block	*block;
+	struct xfs_btree_block	*cblock;
+	union xfs_btree_key	*kp;
+	union xfs_btree_key	*ckp;
+	union xfs_btree_ptr	*pp;
+	union xfs_btree_ptr	*cpp;
+	struct xfs_buf		*cbp;
+	int			level;
+	int			index;
+	int			numrecs;
+	int			error;
+#ifdef DEBUG
+	union xfs_btree_ptr	ptr;
+#endif
+	int			i;
+
+	ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
+	ASSERT(cur->bc_nlevels > 1);
+
+	/*
+	 * Don't deal with the root block needs to be a leaf case.
+	 * We're just going to turn the thing back into extents anyway.
+	 */
+	level = cur->bc_nlevels - 1;
+	if (level == 1)
+		goto out0;
+
+	/*
+	 * Give up if the root has multiple children.
+	 */
+	block = xfs_btree_get_iroot(cur);
+	if (xfs_btree_get_numrecs(block) != 1)
+		goto out0;
+
+	cblock = xfs_btree_get_block(cur, level - 1, &cbp);
+	numrecs = xfs_btree_get_numrecs(cblock);
+
+	/*
+	 * Only do this if the next level will fit.
+	 * Then the data must be copied up to the inode,
+	 * instead of freeing the root you free the next level.
+	 */
+	if (numrecs > cur->bc_ops->get_dmaxrecs(cur, level))
+		goto out0;
+
+	XFS_BTREE_STATS_INC(cur, killroot);
+
+#ifdef DEBUG
+	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_LEFTSIB);
+	ASSERT(xfs_btree_ptr_is_null(cur, &ptr));
+	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
+	ASSERT(xfs_btree_ptr_is_null(cur, &ptr));
+#endif
+
+	index = numrecs - cur->bc_ops->get_maxrecs(cur, level);
+	if (index) {
+		xfs_iroot_realloc(cur->bc_ino.ip, index,
+				  cur->bc_ino.whichfork);
+		block = ifp->if_broot;
+	}
+
+	be16_add_cpu(&block->bb_numrecs, index);
+	ASSERT(block->bb_numrecs == cblock->bb_numrecs);
+
+	kp = xfs_btree_key_addr(cur, 1, block);
+	ckp = xfs_btree_key_addr(cur, 1, cblock);
+	xfs_btree_copy_keys(cur, kp, ckp, numrecs);
+
+	pp = xfs_btree_ptr_addr(cur, 1, block);
+	cpp = xfs_btree_ptr_addr(cur, 1, cblock);
+
+	for (i = 0; i < numrecs; i++) {
+		error = xfs_btree_debug_check_ptr(cur, cpp, i, level - 1);
+		if (error)
+			return error;
+	}
+
+	xfs_btree_copy_ptrs(cur, pp, cpp, numrecs);
+
+	error = xfs_btree_free_block(cur, cbp);
+	if (error)
+		return error;
+
+	cur->bc_levels[level - 1].bp = NULL;
+	be16_add_cpu(&block->bb_level, -1);
+	xfs_trans_log_inode(cur->bc_tp, ip,
+		XFS_ILOG_CORE | xfs_ilog_fbroot(cur->bc_ino.whichfork));
+	cur->bc_nlevels--;
+out0:
+	return 0;
+}
+
+/*
+ * Kill the current root node, and replace it with it's only child node.
+ */
+STATIC int
+xfs_btree_kill_root(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp,
+	int			level,
+	union xfs_btree_ptr	*newroot)
+{
+	int			error;
+
+	XFS_BTREE_STATS_INC(cur, killroot);
+
+	/*
+	 * Update the root pointer, decreasing the level by 1 and then
+	 * free the old root.
+	 */
+	cur->bc_ops->set_root(cur, newroot, -1);
+
+	error = xfs_btree_free_block(cur, bp);
+	if (error)
+		return error;
+
+	cur->bc_levels[level].bp = NULL;
+	cur->bc_levels[level].ra = 0;
+	cur->bc_nlevels--;
+
+	return 0;
+}
+
+STATIC int
+xfs_btree_dec_cursor(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			*stat)
+{
+	int			error;
+	int			i;
+
+	if (level > 0) {
+		error = xfs_btree_decrement(cur, level, &i);
+		if (error)
+			return error;
+	}
+
+	*stat = 1;
+	return 0;
+}
+
+/*
+ * Single level of the btree record deletion routine.
+ * Delete record pointed to by cur/level.
+ * Remove the record from its block then rebalance the tree.
+ * Return 0 for error, 1 for done, 2 to go on to the next level.
+ */
+STATIC int					/* error */
+xfs_btree_delrec(
+	struct xfs_btree_cur	*cur,		/* btree cursor */
+	int			level,		/* level removing record from */
+	int			*stat)		/* fail/done/go-on */
+{
+	struct xfs_btree_block	*block;		/* btree block */
+	union xfs_btree_ptr	cptr;		/* current block ptr */
+	struct xfs_buf		*bp;		/* buffer for block */
+	int			error;		/* error return value */
+	int			i;		/* loop counter */
+	union xfs_btree_ptr	lptr;		/* left sibling block ptr */
+	struct xfs_buf		*lbp;		/* left buffer pointer */
+	struct xfs_btree_block	*left;		/* left btree block */
+	int			lrecs = 0;	/* left record count */
+	int			ptr;		/* key/record index */
+	union xfs_btree_ptr	rptr;		/* right sibling block ptr */
+	struct xfs_buf		*rbp;		/* right buffer pointer */
+	struct xfs_btree_block	*right;		/* right btree block */
+	struct xfs_btree_block	*rrblock;	/* right-right btree block */
+	struct xfs_buf		*rrbp;		/* right-right buffer pointer */
+	int			rrecs = 0;	/* right record count */
+	struct xfs_btree_cur	*tcur;		/* temporary btree cursor */
+	int			numrecs;	/* temporary numrec count */
+
+	tcur = NULL;
+
+	/* Get the index of the entry being deleted, check for nothing there. */
+	ptr = cur->bc_levels[level].ptr;
+	if (ptr == 0) {
+		*stat = 0;
+		return 0;
+	}
+
+	/* Get the buffer & block containing the record or key/ptr. */
+	block = xfs_btree_get_block(cur, level, &bp);
+	numrecs = xfs_btree_get_numrecs(block);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, level, bp);
+	if (error)
+		goto error0;
+#endif
+
+	/* Fail if we're off the end of the block. */
+	if (ptr > numrecs) {
+		*stat = 0;
+		return 0;
+	}
+
+	XFS_BTREE_STATS_INC(cur, delrec);
+	XFS_BTREE_STATS_ADD(cur, moves, numrecs - ptr);
+
+	/* Excise the entries being deleted. */
+	if (level > 0) {
+		/* It's a nonleaf. operate on keys and ptrs */
+		union xfs_btree_key	*lkp;
+		union xfs_btree_ptr	*lpp;
+
+		lkp = xfs_btree_key_addr(cur, ptr + 1, block);
+		lpp = xfs_btree_ptr_addr(cur, ptr + 1, block);
+
+		for (i = 0; i < numrecs - ptr; i++) {
+			error = xfs_btree_debug_check_ptr(cur, lpp, i, level);
+			if (error)
+				goto error0;
+		}
+
+		if (ptr < numrecs) {
+			xfs_btree_shift_keys(cur, lkp, -1, numrecs - ptr);
+			xfs_btree_shift_ptrs(cur, lpp, -1, numrecs - ptr);
+			xfs_btree_log_keys(cur, bp, ptr, numrecs - 1);
+			xfs_btree_log_ptrs(cur, bp, ptr, numrecs - 1);
+		}
+	} else {
+		/* It's a leaf. operate on records */
+		if (ptr < numrecs) {
+			xfs_btree_shift_recs(cur,
+				xfs_btree_rec_addr(cur, ptr + 1, block),
+				-1, numrecs - ptr);
+			xfs_btree_log_recs(cur, bp, ptr, numrecs - 1);
+		}
+	}
+
+	/*
+	 * Decrement and log the number of entries in the block.
+	 */
+	xfs_btree_set_numrecs(block, --numrecs);
+	xfs_btree_log_block(cur, bp, XFS_BB_NUMRECS);
+
+	/*
+	 * If we are tracking the last record in the tree and
+	 * we are at the far right edge of the tree, update it.
+	 */
+	if (xfs_btree_is_lastrec(cur, block, level)) {
+		cur->bc_ops->update_lastrec(cur, block, NULL,
+					    ptr, LASTREC_DELREC);
+	}
+
+	/*
+	 * We're at the root level.  First, shrink the root block in-memory.
+	 * Try to get rid of the next level down.  If we can't then there's
+	 * nothing left to do.
+	 */
+	if (level == cur->bc_nlevels - 1) {
+		if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) {
+			xfs_iroot_realloc(cur->bc_ino.ip, -1,
+					  cur->bc_ino.whichfork);
+
+			error = xfs_btree_kill_iroot(cur);
+			if (error)
+				goto error0;
+
+			error = xfs_btree_dec_cursor(cur, level, stat);
+			if (error)
+				goto error0;
+			*stat = 1;
+			return 0;
+		}
+
+		/*
+		 * If this is the root level, and there's only one entry left,
+		 * and it's NOT the leaf level, then we can get rid of this
+		 * level.
+		 */
+		if (numrecs == 1 && level > 0) {
+			union xfs_btree_ptr	*pp;
+			/*
+			 * pp is still set to the first pointer in the block.
+			 * Make it the new root of the btree.
+			 */
+			pp = xfs_btree_ptr_addr(cur, 1, block);
+			error = xfs_btree_kill_root(cur, bp, level, pp);
+			if (error)
+				goto error0;
+		} else if (level > 0) {
+			error = xfs_btree_dec_cursor(cur, level, stat);
+			if (error)
+				goto error0;
+		}
+		*stat = 1;
+		return 0;
+	}
+
+	/*
+	 * If we deleted the leftmost entry in the block, update the
+	 * key values above us in the tree.
+	 */
+	if (xfs_btree_needs_key_update(cur, ptr)) {
+		error = xfs_btree_update_keys(cur, level);
+		if (error)
+			goto error0;
+	}
+
+	/*
+	 * If the number of records remaining in the block is at least
+	 * the minimum, we're done.
+	 */
+	if (numrecs >= cur->bc_ops->get_minrecs(cur, level)) {
+		error = xfs_btree_dec_cursor(cur, level, stat);
+		if (error)
+			goto error0;
+		return 0;
+	}
+
+	/*
+	 * Otherwise, we have to move some records around to keep the
+	 * tree balanced.  Look at the left and right sibling blocks to
+	 * see if we can re-balance by moving only one record.
+	 */
+	xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
+	xfs_btree_get_sibling(cur, block, &lptr, XFS_BB_LEFTSIB);
+
+	if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) {
+		/*
+		 * One child of root, need to get a chance to copy its contents
+		 * into the root and delete it. Can't go up to next level,
+		 * there's nothing to delete there.
+		 */
+		if (xfs_btree_ptr_is_null(cur, &rptr) &&
+		    xfs_btree_ptr_is_null(cur, &lptr) &&
+		    level == cur->bc_nlevels - 2) {
+			error = xfs_btree_kill_iroot(cur);
+			if (!error)
+				error = xfs_btree_dec_cursor(cur, level, stat);
+			if (error)
+				goto error0;
+			return 0;
+		}
+	}
+
+	ASSERT(!xfs_btree_ptr_is_null(cur, &rptr) ||
+	       !xfs_btree_ptr_is_null(cur, &lptr));
+
+	/*
+	 * Duplicate the cursor so our btree manipulations here won't
+	 * disrupt the next level up.
+	 */
+	error = xfs_btree_dup_cursor(cur, &tcur);
+	if (error)
+		goto error0;
+
+	/*
+	 * If there's a right sibling, see if it's ok to shift an entry
+	 * out of it.
+	 */
+	if (!xfs_btree_ptr_is_null(cur, &rptr)) {
+		/*
+		 * Move the temp cursor to the last entry in the next block.
+		 * Actually any entry but the first would suffice.
+		 */
+		i = xfs_btree_lastrec(tcur, level);
+		if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+
+		error = xfs_btree_increment(tcur, level, &i);
+		if (error)
+			goto error0;
+		if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+
+		i = xfs_btree_lastrec(tcur, level);
+		if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+
+		/* Grab a pointer to the block. */
+		right = xfs_btree_get_block(tcur, level, &rbp);
+#ifdef DEBUG
+		error = xfs_btree_check_block(tcur, right, level, rbp);
+		if (error)
+			goto error0;
+#endif
+		/* Grab the current block number, for future use. */
+		xfs_btree_get_sibling(tcur, right, &cptr, XFS_BB_LEFTSIB);
+
+		/*
+		 * If right block is full enough so that removing one entry
+		 * won't make it too empty, and left-shifting an entry out
+		 * of right to us works, we're done.
+		 */
+		if (xfs_btree_get_numrecs(right) - 1 >=
+		    cur->bc_ops->get_minrecs(tcur, level)) {
+			error = xfs_btree_lshift(tcur, level, &i);
+			if (error)
+				goto error0;
+			if (i) {
+				ASSERT(xfs_btree_get_numrecs(block) >=
+				       cur->bc_ops->get_minrecs(tcur, level));
+
+				xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+				tcur = NULL;
+
+				error = xfs_btree_dec_cursor(cur, level, stat);
+				if (error)
+					goto error0;
+				return 0;
+			}
+		}
+
+		/*
+		 * Otherwise, grab the number of records in right for
+		 * future reference, and fix up the temp cursor to point
+		 * to our block again (last record).
+		 */
+		rrecs = xfs_btree_get_numrecs(right);
+		if (!xfs_btree_ptr_is_null(cur, &lptr)) {
+			i = xfs_btree_firstrec(tcur, level);
+			if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto error0;
+			}
+
+			error = xfs_btree_decrement(tcur, level, &i);
+			if (error)
+				goto error0;
+			if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto error0;
+			}
+		}
+	}
+
+	/*
+	 * If there's a left sibling, see if it's ok to shift an entry
+	 * out of it.
+	 */
+	if (!xfs_btree_ptr_is_null(cur, &lptr)) {
+		/*
+		 * Move the temp cursor to the first entry in the
+		 * previous block.
+		 */
+		i = xfs_btree_firstrec(tcur, level);
+		if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+
+		error = xfs_btree_decrement(tcur, level, &i);
+		if (error)
+			goto error0;
+		i = xfs_btree_firstrec(tcur, level);
+		if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+
+		/* Grab a pointer to the block. */
+		left = xfs_btree_get_block(tcur, level, &lbp);
+#ifdef DEBUG
+		error = xfs_btree_check_block(cur, left, level, lbp);
+		if (error)
+			goto error0;
+#endif
+		/* Grab the current block number, for future use. */
+		xfs_btree_get_sibling(tcur, left, &cptr, XFS_BB_RIGHTSIB);
+
+		/*
+		 * If left block is full enough so that removing one entry
+		 * won't make it too empty, and right-shifting an entry out
+		 * of left to us works, we're done.
+		 */
+		if (xfs_btree_get_numrecs(left) - 1 >=
+		    cur->bc_ops->get_minrecs(tcur, level)) {
+			error = xfs_btree_rshift(tcur, level, &i);
+			if (error)
+				goto error0;
+			if (i) {
+				ASSERT(xfs_btree_get_numrecs(block) >=
+				       cur->bc_ops->get_minrecs(tcur, level));
+				xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+				tcur = NULL;
+				if (level == 0)
+					cur->bc_levels[0].ptr++;
+
+				*stat = 1;
+				return 0;
+			}
+		}
+
+		/*
+		 * Otherwise, grab the number of records in right for
+		 * future reference.
+		 */
+		lrecs = xfs_btree_get_numrecs(left);
+	}
+
+	/* Delete the temp cursor, we're done with it. */
+	xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+	tcur = NULL;
+
+	/* If here, we need to do a join to keep the tree balanced. */
+	ASSERT(!xfs_btree_ptr_is_null(cur, &cptr));
+
+	if (!xfs_btree_ptr_is_null(cur, &lptr) &&
+	    lrecs + xfs_btree_get_numrecs(block) <=
+			cur->bc_ops->get_maxrecs(cur, level)) {
+		/*
+		 * Set "right" to be the starting block,
+		 * "left" to be the left neighbor.
+		 */
+		rptr = cptr;
+		right = block;
+		rbp = bp;
+		error = xfs_btree_read_buf_block(cur, &lptr, 0, &left, &lbp);
+		if (error)
+			goto error0;
+
+	/*
+	 * If that won't work, see if we can join with the right neighbor block.
+	 */
+	} else if (!xfs_btree_ptr_is_null(cur, &rptr) &&
+		   rrecs + xfs_btree_get_numrecs(block) <=
+			cur->bc_ops->get_maxrecs(cur, level)) {
+		/*
+		 * Set "left" to be the starting block,
+		 * "right" to be the right neighbor.
+		 */
+		lptr = cptr;
+		left = block;
+		lbp = bp;
+		error = xfs_btree_read_buf_block(cur, &rptr, 0, &right, &rbp);
+		if (error)
+			goto error0;
+
+	/*
+	 * Otherwise, we can't fix the imbalance.
+	 * Just return.  This is probably a logic error, but it's not fatal.
+	 */
+	} else {
+		error = xfs_btree_dec_cursor(cur, level, stat);
+		if (error)
+			goto error0;
+		return 0;
+	}
+
+	rrecs = xfs_btree_get_numrecs(right);
+	lrecs = xfs_btree_get_numrecs(left);
+
+	/*
+	 * We're now going to join "left" and "right" by moving all the stuff
+	 * in "right" to "left" and deleting "right".
+	 */
+	XFS_BTREE_STATS_ADD(cur, moves, rrecs);
+	if (level > 0) {
+		/* It's a non-leaf.  Move keys and pointers. */
+		union xfs_btree_key	*lkp;	/* left btree key */
+		union xfs_btree_ptr	*lpp;	/* left address pointer */
+		union xfs_btree_key	*rkp;	/* right btree key */
+		union xfs_btree_ptr	*rpp;	/* right address pointer */
+
+		lkp = xfs_btree_key_addr(cur, lrecs + 1, left);
+		lpp = xfs_btree_ptr_addr(cur, lrecs + 1, left);
+		rkp = xfs_btree_key_addr(cur, 1, right);
+		rpp = xfs_btree_ptr_addr(cur, 1, right);
+
+		for (i = 1; i < rrecs; i++) {
+			error = xfs_btree_debug_check_ptr(cur, rpp, i, level);
+			if (error)
+				goto error0;
+		}
+
+		xfs_btree_copy_keys(cur, lkp, rkp, rrecs);
+		xfs_btree_copy_ptrs(cur, lpp, rpp, rrecs);
+
+		xfs_btree_log_keys(cur, lbp, lrecs + 1, lrecs + rrecs);
+		xfs_btree_log_ptrs(cur, lbp, lrecs + 1, lrecs + rrecs);
+	} else {
+		/* It's a leaf.  Move records.  */
+		union xfs_btree_rec	*lrp;	/* left record pointer */
+		union xfs_btree_rec	*rrp;	/* right record pointer */
+
+		lrp = xfs_btree_rec_addr(cur, lrecs + 1, left);
+		rrp = xfs_btree_rec_addr(cur, 1, right);
+
+		xfs_btree_copy_recs(cur, lrp, rrp, rrecs);
+		xfs_btree_log_recs(cur, lbp, lrecs + 1, lrecs + rrecs);
+	}
+
+	XFS_BTREE_STATS_INC(cur, join);
+
+	/*
+	 * Fix up the number of records and right block pointer in the
+	 * surviving block, and log it.
+	 */
+	xfs_btree_set_numrecs(left, lrecs + rrecs);
+	xfs_btree_get_sibling(cur, right, &cptr, XFS_BB_RIGHTSIB);
+	xfs_btree_set_sibling(cur, left, &cptr, XFS_BB_RIGHTSIB);
+	xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
+
+	/* If there is a right sibling, point it to the remaining block. */
+	xfs_btree_get_sibling(cur, left, &cptr, XFS_BB_RIGHTSIB);
+	if (!xfs_btree_ptr_is_null(cur, &cptr)) {
+		error = xfs_btree_read_buf_block(cur, &cptr, 0, &rrblock, &rrbp);
+		if (error)
+			goto error0;
+		xfs_btree_set_sibling(cur, rrblock, &lptr, XFS_BB_LEFTSIB);
+		xfs_btree_log_block(cur, rrbp, XFS_BB_LEFTSIB);
+	}
+
+	/* Free the deleted block. */
+	error = xfs_btree_free_block(cur, rbp);
+	if (error)
+		goto error0;
+
+	/*
+	 * If we joined with the left neighbor, set the buffer in the
+	 * cursor to the left block, and fix up the index.
+	 */
+	if (bp != lbp) {
+		cur->bc_levels[level].bp = lbp;
+		cur->bc_levels[level].ptr += lrecs;
+		cur->bc_levels[level].ra = 0;
+	}
+	/*
+	 * If we joined with the right neighbor and there's a level above
+	 * us, increment the cursor at that level.
+	 */
+	else if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) ||
+		   (level + 1 < cur->bc_nlevels)) {
+		error = xfs_btree_increment(cur, level + 1, &i);
+		if (error)
+			goto error0;
+	}
+
+	/*
+	 * Readjust the ptr at this level if it's not a leaf, since it's
+	 * still pointing at the deletion point, which makes the cursor
+	 * inconsistent.  If this makes the ptr 0, the caller fixes it up.
+	 * We can't use decrement because it would change the next level up.
+	 */
+	if (level > 0)
+		cur->bc_levels[level].ptr--;
+
+	/*
+	 * We combined blocks, so we have to update the parent keys if the
+	 * btree supports overlapped intervals.  However,
+	 * bc_levels[level + 1].ptr points to the old block so that the caller
+	 * knows which record to delete.  Therefore, the caller must be savvy
+	 * enough to call updkeys for us if we return stat == 2.  The other
+	 * exit points from this function don't require deletions further up
+	 * the tree, so they can call updkeys directly.
+	 */
+
+	/* Return value means the next level up has something to do. */
+	*stat = 2;
+	return 0;
+
+error0:
+	if (tcur)
+		xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/*
+ * Delete the record pointed to by cur.
+ * The cursor refers to the place where the record was (could be inserted)
+ * when the operation returns.
+ */
+int					/* error */
+xfs_btree_delete(
+	struct xfs_btree_cur	*cur,
+	int			*stat)	/* success/failure */
+{
+	int			error;	/* error return value */
+	int			level;
+	int			i;
+	bool			joined = false;
+
+	/*
+	 * Go up the tree, starting at leaf level.
+	 *
+	 * If 2 is returned then a join was done; go to the next level.
+	 * Otherwise we are done.
+	 */
+	for (level = 0, i = 2; i == 2; level++) {
+		error = xfs_btree_delrec(cur, level, &i);
+		if (error)
+			goto error0;
+		if (i == 2)
+			joined = true;
+	}
+
+	/*
+	 * If we combined blocks as part of deleting the record, delrec won't
+	 * have updated the parent high keys so we have to do that here.
+	 */
+	if (joined && (cur->bc_flags & XFS_BTREE_OVERLAPPING)) {
+		error = xfs_btree_updkeys_force(cur, 0);
+		if (error)
+			goto error0;
+	}
+
+	if (i == 0) {
+		for (level = 1; level < cur->bc_nlevels; level++) {
+			if (cur->bc_levels[level].ptr == 0) {
+				error = xfs_btree_decrement(cur, level, &i);
+				if (error)
+					goto error0;
+				break;
+			}
+		}
+	}
+
+	*stat = i;
+	return 0;
+error0:
+	return error;
+}
+
+/*
+ * Get the data from the pointed-to record.
+ */
+int					/* error */
+xfs_btree_get_rec(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	union xfs_btree_rec	**recp,	/* output: btree record */
+	int			*stat)	/* output: success/failure */
+{
+	struct xfs_btree_block	*block;	/* btree block */
+	struct xfs_buf		*bp;	/* buffer pointer */
+	int			ptr;	/* record number */
+#ifdef DEBUG
+	int			error;	/* error return value */
+#endif
+
+	ptr = cur->bc_levels[0].ptr;
+	block = xfs_btree_get_block(cur, 0, &bp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, 0, bp);
+	if (error)
+		return error;
+#endif
+
+	/*
+	 * Off the right end or left end, return failure.
+	 */
+	if (ptr > xfs_btree_get_numrecs(block) || ptr <= 0) {
+		*stat = 0;
+		return 0;
+	}
+
+	/*
+	 * Point to the record and extract its data.
+	 */
+	*recp = xfs_btree_rec_addr(cur, ptr, block);
+	*stat = 1;
+	return 0;
+}
+
+/* Visit a block in a btree. */
+STATIC int
+xfs_btree_visit_block(
+	struct xfs_btree_cur		*cur,
+	int				level,
+	xfs_btree_visit_blocks_fn	fn,
+	void				*data)
+{
+	struct xfs_btree_block		*block;
+	struct xfs_buf			*bp;
+	union xfs_btree_ptr		rptr;
+	int				error;
+
+	/* do right sibling readahead */
+	xfs_btree_readahead(cur, level, XFS_BTCUR_RIGHTRA);
+	block = xfs_btree_get_block(cur, level, &bp);
+
+	/* process the block */
+	error = fn(cur, level, data);
+	if (error)
+		return error;
+
+	/* now read rh sibling block for next iteration */
+	xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
+	if (xfs_btree_ptr_is_null(cur, &rptr))
+		return -ENOENT;
+
+	/*
+	 * We only visit blocks once in this walk, so we have to avoid the
+	 * internal xfs_btree_lookup_get_block() optimisation where it will
+	 * return the same block without checking if the right sibling points
+	 * back to us and creates a cyclic reference in the btree.
+	 */
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		if (be64_to_cpu(rptr.l) == XFS_DADDR_TO_FSB(cur->bc_mp,
+							xfs_buf_daddr(bp)))
+			return -EFSCORRUPTED;
+	} else {
+		if (be32_to_cpu(rptr.s) == xfs_daddr_to_agbno(cur->bc_mp,
+							xfs_buf_daddr(bp)))
+			return -EFSCORRUPTED;
+	}
+	return xfs_btree_lookup_get_block(cur, level, &rptr, &block);
+}
+
+
+/* Visit every block in a btree. */
+int
+xfs_btree_visit_blocks(
+	struct xfs_btree_cur		*cur,
+	xfs_btree_visit_blocks_fn	fn,
+	unsigned int			flags,
+	void				*data)
+{
+	union xfs_btree_ptr		lptr;
+	int				level;
+	struct xfs_btree_block		*block = NULL;
+	int				error = 0;
+
+	cur->bc_ops->init_ptr_from_cur(cur, &lptr);
+
+	/* for each level */
+	for (level = cur->bc_nlevels - 1; level >= 0; level--) {
+		/* grab the left hand block */
+		error = xfs_btree_lookup_get_block(cur, level, &lptr, &block);
+		if (error)
+			return error;
+
+		/* readahead the left most block for the next level down */
+		if (level > 0) {
+			union xfs_btree_ptr     *ptr;
+
+			ptr = xfs_btree_ptr_addr(cur, 1, block);
+			xfs_btree_readahead_ptr(cur, ptr, 1);
+
+			/* save for the next iteration of the loop */
+			xfs_btree_copy_ptrs(cur, &lptr, ptr, 1);
+
+			if (!(flags & XFS_BTREE_VISIT_LEAVES))
+				continue;
+		} else if (!(flags & XFS_BTREE_VISIT_RECORDS)) {
+			continue;
+		}
+
+		/* for each buffer in the level */
+		do {
+			error = xfs_btree_visit_block(cur, level, fn, data);
+		} while (!error);
+
+		if (error != -ENOENT)
+			return error;
+	}
+
+	return 0;
+}
+
+/*
+ * Change the owner of a btree.
+ *
+ * The mechanism we use here is ordered buffer logging. Because we don't know
+ * how many buffers were are going to need to modify, we don't really want to
+ * have to make transaction reservations for the worst case of every buffer in a
+ * full size btree as that may be more space that we can fit in the log....
+ *
+ * We do the btree walk in the most optimal manner possible - we have sibling
+ * pointers so we can just walk all the blocks on each level from left to right
+ * in a single pass, and then move to the next level and do the same. We can
+ * also do readahead on the sibling pointers to get IO moving more quickly,
+ * though for slow disks this is unlikely to make much difference to performance
+ * as the amount of CPU work we have to do before moving to the next block is
+ * relatively small.
+ *
+ * For each btree block that we load, modify the owner appropriately, set the
+ * buffer as an ordered buffer and log it appropriately. We need to ensure that
+ * we mark the region we change dirty so that if the buffer is relogged in
+ * a subsequent transaction the changes we make here as an ordered buffer are
+ * correctly relogged in that transaction.  If we are in recovery context, then
+ * just queue the modified buffer as delayed write buffer so the transaction
+ * recovery completion writes the changes to disk.
+ */
+struct xfs_btree_block_change_owner_info {
+	uint64_t		new_owner;
+	struct list_head	*buffer_list;
+};
+
+static int
+xfs_btree_block_change_owner(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	void			*data)
+{
+	struct xfs_btree_block_change_owner_info	*bbcoi = data;
+	struct xfs_btree_block	*block;
+	struct xfs_buf		*bp;
+
+	/* modify the owner */
+	block = xfs_btree_get_block(cur, level, &bp);
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		if (block->bb_u.l.bb_owner == cpu_to_be64(bbcoi->new_owner))
+			return 0;
+		block->bb_u.l.bb_owner = cpu_to_be64(bbcoi->new_owner);
+	} else {
+		if (block->bb_u.s.bb_owner == cpu_to_be32(bbcoi->new_owner))
+			return 0;
+		block->bb_u.s.bb_owner = cpu_to_be32(bbcoi->new_owner);
+	}
+
+	/*
+	 * If the block is a root block hosted in an inode, we might not have a
+	 * buffer pointer here and we shouldn't attempt to log the change as the
+	 * information is already held in the inode and discarded when the root
+	 * block is formatted into the on-disk inode fork. We still change it,
+	 * though, so everything is consistent in memory.
+	 */
+	if (!bp) {
+		ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
+		ASSERT(level == cur->bc_nlevels - 1);
+		return 0;
+	}
+
+	if (cur->bc_tp) {
+		if (!xfs_trans_ordered_buf(cur->bc_tp, bp)) {
+			xfs_btree_log_block(cur, bp, XFS_BB_OWNER);
+			return -EAGAIN;
+		}
+	} else {
+		xfs_buf_delwri_queue(bp, bbcoi->buffer_list);
+	}
+
+	return 0;
+}
+
+int
+xfs_btree_change_owner(
+	struct xfs_btree_cur	*cur,
+	uint64_t		new_owner,
+	struct list_head	*buffer_list)
+{
+	struct xfs_btree_block_change_owner_info	bbcoi;
+
+	bbcoi.new_owner = new_owner;
+	bbcoi.buffer_list = buffer_list;
+
+	return xfs_btree_visit_blocks(cur, xfs_btree_block_change_owner,
+			XFS_BTREE_VISIT_ALL, &bbcoi);
+}
+
+/* Verify the v5 fields of a long-format btree block. */
+xfs_failaddr_t
+xfs_btree_lblock_v5hdr_verify(
+	struct xfs_buf		*bp,
+	uint64_t		owner)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
+
+	if (!xfs_has_crc(mp))
+		return __this_address;
+	if (!uuid_equal(&block->bb_u.l.bb_uuid, &mp->m_sb.sb_meta_uuid))
+		return __this_address;
+	if (block->bb_u.l.bb_blkno != cpu_to_be64(xfs_buf_daddr(bp)))
+		return __this_address;
+	if (owner != XFS_RMAP_OWN_UNKNOWN &&
+	    be64_to_cpu(block->bb_u.l.bb_owner) != owner)
+		return __this_address;
+	return NULL;
+}
+
+/* Verify a long-format btree block. */
+xfs_failaddr_t
+xfs_btree_lblock_verify(
+	struct xfs_buf		*bp,
+	unsigned int		max_recs)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
+	xfs_fsblock_t		fsb;
+	xfs_failaddr_t		fa;
+
+	/* numrecs verification */
+	if (be16_to_cpu(block->bb_numrecs) > max_recs)
+		return __this_address;
+
+	/* sibling pointer verification */
+	fsb = XFS_DADDR_TO_FSB(mp, xfs_buf_daddr(bp));
+	fa = xfs_btree_check_lblock_siblings(mp, NULL, -1, fsb,
+			block->bb_u.l.bb_leftsib);
+	if (!fa)
+		fa = xfs_btree_check_lblock_siblings(mp, NULL, -1, fsb,
+				block->bb_u.l.bb_rightsib);
+	return fa;
+}
+
+/**
+ * xfs_btree_sblock_v5hdr_verify() -- verify the v5 fields of a short-format
+ *				      btree block
+ *
+ * @bp: buffer containing the btree block
+ */
+xfs_failaddr_t
+xfs_btree_sblock_v5hdr_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
+	struct xfs_perag	*pag = bp->b_pag;
+
+	if (!xfs_has_crc(mp))
+		return __this_address;
+	if (!uuid_equal(&block->bb_u.s.bb_uuid, &mp->m_sb.sb_meta_uuid))
+		return __this_address;
+	if (block->bb_u.s.bb_blkno != cpu_to_be64(xfs_buf_daddr(bp)))
+		return __this_address;
+	if (pag && be32_to_cpu(block->bb_u.s.bb_owner) != pag->pag_agno)
+		return __this_address;
+	return NULL;
+}
+
+/**
+ * xfs_btree_sblock_verify() -- verify a short-format btree block
+ *
+ * @bp: buffer containing the btree block
+ * @max_recs: maximum records allowed in this btree node
+ */
+xfs_failaddr_t
+xfs_btree_sblock_verify(
+	struct xfs_buf		*bp,
+	unsigned int		max_recs)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
+	xfs_agblock_t		agbno;
+	xfs_failaddr_t		fa;
+
+	/* numrecs verification */
+	if (be16_to_cpu(block->bb_numrecs) > max_recs)
+		return __this_address;
+
+	/* sibling pointer verification */
+	agbno = xfs_daddr_to_agbno(mp, xfs_buf_daddr(bp));
+	fa = xfs_btree_check_sblock_siblings(bp->b_pag, NULL, -1, agbno,
+			block->bb_u.s.bb_leftsib);
+	if (!fa)
+		fa = xfs_btree_check_sblock_siblings(bp->b_pag, NULL, -1, agbno,
+				block->bb_u.s.bb_rightsib);
+	return fa;
+}
+
+/*
+ * For the given limits on leaf and keyptr records per block, calculate the
+ * height of the tree needed to index the number of leaf records.
+ */
+unsigned int
+xfs_btree_compute_maxlevels(
+	const unsigned int	*limits,
+	unsigned long long	records)
+{
+	unsigned long long	level_blocks = howmany_64(records, limits[0]);
+	unsigned int		height = 1;
+
+	while (level_blocks > 1) {
+		level_blocks = howmany_64(level_blocks, limits[1]);
+		height++;
+	}
+
+	return height;
+}
+
+/*
+ * For the given limits on leaf and keyptr records per block, calculate the
+ * number of blocks needed to index the given number of leaf records.
+ */
+unsigned long long
+xfs_btree_calc_size(
+	const unsigned int	*limits,
+	unsigned long long	records)
+{
+	unsigned long long	level_blocks = howmany_64(records, limits[0]);
+	unsigned long long	blocks = level_blocks;
+
+	while (level_blocks > 1) {
+		level_blocks = howmany_64(level_blocks, limits[1]);
+		blocks += level_blocks;
+	}
+
+	return blocks;
+}
+
+/*
+ * Given a number of available blocks for the btree to consume with records and
+ * pointers, calculate the height of the tree needed to index all the records
+ * that space can hold based on the number of pointers each interior node
+ * holds.
+ *
+ * We start by assuming a single level tree consumes a single block, then track
+ * the number of blocks each node level consumes until we no longer have space
+ * to store the next node level. At this point, we are indexing all the leaf
+ * blocks in the space, and there's no more free space to split the tree any
+ * further. That's our maximum btree height.
+ */
+unsigned int
+xfs_btree_space_to_height(
+	const unsigned int	*limits,
+	unsigned long long	leaf_blocks)
+{
+	unsigned long long	node_blocks = limits[1];
+	unsigned long long	blocks_left = leaf_blocks - 1;
+	unsigned int		height = 1;
+
+	if (leaf_blocks < 1)
+		return 0;
+
+	while (node_blocks < blocks_left) {
+		blocks_left -= node_blocks;
+		node_blocks *= limits[1];
+		height++;
+	}
+
+	return height;
+}
+
+/*
+ * Query a regular btree for all records overlapping a given interval.
+ * Start with a LE lookup of the key of low_rec and return all records
+ * until we find a record with a key greater than the key of high_rec.
+ */
+STATIC int
+xfs_btree_simple_query_range(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*low_key,
+	const union xfs_btree_key	*high_key,
+	xfs_btree_query_range_fn	fn,
+	void				*priv)
+{
+	union xfs_btree_rec		*recp;
+	union xfs_btree_key		rec_key;
+	int64_t				diff;
+	int				stat;
+	bool				firstrec = true;
+	int				error;
+
+	ASSERT(cur->bc_ops->init_high_key_from_rec);
+	ASSERT(cur->bc_ops->diff_two_keys);
+
+	/*
+	 * Find the leftmost record.  The btree cursor must be set
+	 * to the low record used to generate low_key.
+	 */
+	stat = 0;
+	error = xfs_btree_lookup(cur, XFS_LOOKUP_LE, &stat);
+	if (error)
+		goto out;
+
+	/* Nothing?  See if there's anything to the right. */
+	if (!stat) {
+		error = xfs_btree_increment(cur, 0, &stat);
+		if (error)
+			goto out;
+	}
+
+	while (stat) {
+		/* Find the record. */
+		error = xfs_btree_get_rec(cur, &recp, &stat);
+		if (error || !stat)
+			break;
+
+		/* Skip if high_key(rec) < low_key. */
+		if (firstrec) {
+			cur->bc_ops->init_high_key_from_rec(&rec_key, recp);
+			firstrec = false;
+			diff = cur->bc_ops->diff_two_keys(cur, low_key,
+					&rec_key);
+			if (diff > 0)
+				goto advloop;
+		}
+
+		/* Stop if high_key < low_key(rec). */
+		cur->bc_ops->init_key_from_rec(&rec_key, recp);
+		diff = cur->bc_ops->diff_two_keys(cur, &rec_key, high_key);
+		if (diff > 0)
+			break;
+
+		/* Callback */
+		error = fn(cur, recp, priv);
+		if (error)
+			break;
+
+advloop:
+		/* Move on to the next record. */
+		error = xfs_btree_increment(cur, 0, &stat);
+		if (error)
+			break;
+	}
+
+out:
+	return error;
+}
+
+/*
+ * Query an overlapped interval btree for all records overlapping a given
+ * interval.  This function roughly follows the algorithm given in
+ * "Interval Trees" of _Introduction to Algorithms_, which is section
+ * 14.3 in the 2nd and 3rd editions.
+ *
+ * First, generate keys for the low and high records passed in.
+ *
+ * For any leaf node, generate the high and low keys for the record.
+ * If the record keys overlap with the query low/high keys, pass the
+ * record to the function iterator.
+ *
+ * For any internal node, compare the low and high keys of each
+ * pointer against the query low/high keys.  If there's an overlap,
+ * follow the pointer.
+ *
+ * As an optimization, we stop scanning a block when we find a low key
+ * that is greater than the query's high key.
+ */
+STATIC int
+xfs_btree_overlapped_query_range(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*low_key,
+	const union xfs_btree_key	*high_key,
+	xfs_btree_query_range_fn	fn,
+	void				*priv)
+{
+	union xfs_btree_ptr		ptr;
+	union xfs_btree_ptr		*pp;
+	union xfs_btree_key		rec_key;
+	union xfs_btree_key		rec_hkey;
+	union xfs_btree_key		*lkp;
+	union xfs_btree_key		*hkp;
+	union xfs_btree_rec		*recp;
+	struct xfs_btree_block		*block;
+	int64_t				ldiff;
+	int64_t				hdiff;
+	int				level;
+	struct xfs_buf			*bp;
+	int				i;
+	int				error;
+
+	/* Load the root of the btree. */
+	level = cur->bc_nlevels - 1;
+	cur->bc_ops->init_ptr_from_cur(cur, &ptr);
+	error = xfs_btree_lookup_get_block(cur, level, &ptr, &block);
+	if (error)
+		return error;
+	xfs_btree_get_block(cur, level, &bp);
+	trace_xfs_btree_overlapped_query_range(cur, level, bp);
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, level, bp);
+	if (error)
+		goto out;
+#endif
+	cur->bc_levels[level].ptr = 1;
+
+	while (level < cur->bc_nlevels) {
+		block = xfs_btree_get_block(cur, level, &bp);
+
+		/* End of node, pop back towards the root. */
+		if (cur->bc_levels[level].ptr >
+					be16_to_cpu(block->bb_numrecs)) {
+pop_up:
+			if (level < cur->bc_nlevels - 1)
+				cur->bc_levels[level + 1].ptr++;
+			level++;
+			continue;
+		}
+
+		if (level == 0) {
+			/* Handle a leaf node. */
+			recp = xfs_btree_rec_addr(cur, cur->bc_levels[0].ptr,
+					block);
+
+			cur->bc_ops->init_high_key_from_rec(&rec_hkey, recp);
+			ldiff = cur->bc_ops->diff_two_keys(cur, &rec_hkey,
+					low_key);
+
+			cur->bc_ops->init_key_from_rec(&rec_key, recp);
+			hdiff = cur->bc_ops->diff_two_keys(cur, high_key,
+					&rec_key);
+
+			/*
+			 * If (record's high key >= query's low key) and
+			 *    (query's high key >= record's low key), then
+			 * this record overlaps the query range; callback.
+			 */
+			if (ldiff >= 0 && hdiff >= 0) {
+				error = fn(cur, recp, priv);
+				if (error)
+					break;
+			} else if (hdiff < 0) {
+				/* Record is larger than high key; pop. */
+				goto pop_up;
+			}
+			cur->bc_levels[level].ptr++;
+			continue;
+		}
+
+		/* Handle an internal node. */
+		lkp = xfs_btree_key_addr(cur, cur->bc_levels[level].ptr, block);
+		hkp = xfs_btree_high_key_addr(cur, cur->bc_levels[level].ptr,
+				block);
+		pp = xfs_btree_ptr_addr(cur, cur->bc_levels[level].ptr, block);
+
+		ldiff = cur->bc_ops->diff_two_keys(cur, hkp, low_key);
+		hdiff = cur->bc_ops->diff_two_keys(cur, high_key, lkp);
+
+		/*
+		 * If (pointer's high key >= query's low key) and
+		 *    (query's high key >= pointer's low key), then
+		 * this record overlaps the query range; follow pointer.
+		 */
+		if (ldiff >= 0 && hdiff >= 0) {
+			level--;
+			error = xfs_btree_lookup_get_block(cur, level, pp,
+					&block);
+			if (error)
+				goto out;
+			xfs_btree_get_block(cur, level, &bp);
+			trace_xfs_btree_overlapped_query_range(cur, level, bp);
+#ifdef DEBUG
+			error = xfs_btree_check_block(cur, block, level, bp);
+			if (error)
+				goto out;
+#endif
+			cur->bc_levels[level].ptr = 1;
+			continue;
+		} else if (hdiff < 0) {
+			/* The low key is larger than the upper range; pop. */
+			goto pop_up;
+		}
+		cur->bc_levels[level].ptr++;
+	}
+
+out:
+	/*
+	 * If we don't end this function with the cursor pointing at a record
+	 * block, a subsequent non-error cursor deletion will not release
+	 * node-level buffers, causing a buffer leak.  This is quite possible
+	 * with a zero-results range query, so release the buffers if we
+	 * failed to return any results.
+	 */
+	if (cur->bc_levels[0].bp == NULL) {
+		for (i = 0; i < cur->bc_nlevels; i++) {
+			if (cur->bc_levels[i].bp) {
+				xfs_trans_brelse(cur->bc_tp,
+						cur->bc_levels[i].bp);
+				cur->bc_levels[i].bp = NULL;
+				cur->bc_levels[i].ptr = 0;
+				cur->bc_levels[i].ra = 0;
+			}
+		}
+	}
+
+	return error;
+}
+
+/*
+ * Query a btree for all records overlapping a given interval of keys.  The
+ * supplied function will be called with each record found; return one of the
+ * XFS_BTREE_QUERY_RANGE_{CONTINUE,ABORT} values or the usual negative error
+ * code.  This function returns -ECANCELED, zero, or a negative error code.
+ */
+int
+xfs_btree_query_range(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_irec	*low_rec,
+	const union xfs_btree_irec	*high_rec,
+	xfs_btree_query_range_fn	fn,
+	void				*priv)
+{
+	union xfs_btree_rec		rec;
+	union xfs_btree_key		low_key;
+	union xfs_btree_key		high_key;
+
+	/* Find the keys of both ends of the interval. */
+	cur->bc_rec = *high_rec;
+	cur->bc_ops->init_rec_from_cur(cur, &rec);
+	cur->bc_ops->init_key_from_rec(&high_key, &rec);
+
+	cur->bc_rec = *low_rec;
+	cur->bc_ops->init_rec_from_cur(cur, &rec);
+	cur->bc_ops->init_key_from_rec(&low_key, &rec);
+
+	/* Enforce low key < high key. */
+	if (cur->bc_ops->diff_two_keys(cur, &low_key, &high_key) > 0)
+		return -EINVAL;
+
+	if (!(cur->bc_flags & XFS_BTREE_OVERLAPPING))
+		return xfs_btree_simple_query_range(cur, &low_key,
+				&high_key, fn, priv);
+	return xfs_btree_overlapped_query_range(cur, &low_key, &high_key,
+			fn, priv);
+}
+
+/* Query a btree for all records. */
+int
+xfs_btree_query_all(
+	struct xfs_btree_cur		*cur,
+	xfs_btree_query_range_fn	fn,
+	void				*priv)
+{
+	union xfs_btree_key		low_key;
+	union xfs_btree_key		high_key;
+
+	memset(&cur->bc_rec, 0, sizeof(cur->bc_rec));
+	memset(&low_key, 0, sizeof(low_key));
+	memset(&high_key, 0xFF, sizeof(high_key));
+
+	return xfs_btree_simple_query_range(cur, &low_key, &high_key, fn, priv);
+}
+
+static int
+xfs_btree_count_blocks_helper(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	void			*data)
+{
+	xfs_extlen_t		*blocks = data;
+	(*blocks)++;
+
+	return 0;
+}
+
+/* Count the blocks in a btree and return the result in *blocks. */
+int
+xfs_btree_count_blocks(
+	struct xfs_btree_cur	*cur,
+	xfs_extlen_t		*blocks)
+{
+	*blocks = 0;
+	return xfs_btree_visit_blocks(cur, xfs_btree_count_blocks_helper,
+			XFS_BTREE_VISIT_ALL, blocks);
+}
+
+/* Compare two btree pointers. */
+int64_t
+xfs_btree_diff_two_ptrs(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*a,
+	const union xfs_btree_ptr	*b)
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		return (int64_t)be64_to_cpu(a->l) - be64_to_cpu(b->l);
+	return (int64_t)be32_to_cpu(a->s) - be32_to_cpu(b->s);
+}
+
+/* If there's an extent, we're done. */
+STATIC int
+xfs_btree_has_record_helper(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_rec	*rec,
+	void				*priv)
+{
+	return -ECANCELED;
+}
+
+/* Is there a record covering a given range of keys? */
+int
+xfs_btree_has_record(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_irec	*low,
+	const union xfs_btree_irec	*high,
+	bool				*exists)
+{
+	int				error;
+
+	error = xfs_btree_query_range(cur, low, high,
+			&xfs_btree_has_record_helper, NULL);
+	if (error == -ECANCELED) {
+		*exists = true;
+		return 0;
+	}
+	*exists = false;
+	return error;
+}
+
+/* Are there more records in this btree? */
+bool
+xfs_btree_has_more_records(
+	struct xfs_btree_cur	*cur)
+{
+	struct xfs_btree_block	*block;
+	struct xfs_buf		*bp;
+
+	block = xfs_btree_get_block(cur, 0, &bp);
+
+	/* There are still records in this block. */
+	if (cur->bc_levels[0].ptr < xfs_btree_get_numrecs(block))
+		return true;
+
+	/* There are more record blocks. */
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		return block->bb_u.l.bb_rightsib != cpu_to_be64(NULLFSBLOCK);
+	else
+		return block->bb_u.s.bb_rightsib != cpu_to_be32(NULLAGBLOCK);
+}
+
+/* Set up all the btree cursor caches. */
+int __init
+xfs_btree_init_cur_caches(void)
+{
+	int		error;
+
+	error = xfs_allocbt_init_cur_cache();
+	if (error)
+		return error;
+	error = xfs_inobt_init_cur_cache();
+	if (error)
+		goto err;
+	error = xfs_bmbt_init_cur_cache();
+	if (error)
+		goto err;
+	error = xfs_rmapbt_init_cur_cache();
+	if (error)
+		goto err;
+	error = xfs_refcountbt_init_cur_cache();
+	if (error)
+		goto err;
+
+	return 0;
+err:
+	xfs_btree_destroy_cur_caches();
+	return error;
+}
+
+/* Destroy all the btree cursor caches, if they've been allocated. */
+void
+xfs_btree_destroy_cur_caches(void)
+{
+	xfs_allocbt_destroy_cur_cache();
+	xfs_inobt_destroy_cur_cache();
+	xfs_bmbt_destroy_cur_cache();
+	xfs_rmapbt_destroy_cur_cache();
+	xfs_refcountbt_destroy_cur_cache();
+}
diff --git a/fs/xfs/libxfs/xfs_btree.h b/fs/xfs/libxfs/xfs_btree.h
new file mode 100644
index 000000000..eef27858a
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_btree.h
@@ -0,0 +1,606 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_BTREE_H__
+#define	__XFS_BTREE_H__
+
+struct xfs_buf;
+struct xfs_inode;
+struct xfs_mount;
+struct xfs_trans;
+struct xfs_ifork;
+struct xfs_perag;
+
+/*
+ * Generic key, ptr and record wrapper structures.
+ *
+ * These are disk format structures, and are converted where necessary
+ * by the btree specific code that needs to interpret them.
+ */
+union xfs_btree_ptr {
+	__be32			s;	/* short form ptr */
+	__be64			l;	/* long form ptr */
+};
+
+/*
+ * The in-core btree key.  Overlapping btrees actually store two keys
+ * per pointer, so we reserve enough memory to hold both.  The __*bigkey
+ * items should never be accessed directly.
+ */
+union xfs_btree_key {
+	struct xfs_bmbt_key		bmbt;
+	xfs_bmdr_key_t			bmbr;	/* bmbt root block */
+	xfs_alloc_key_t			alloc;
+	struct xfs_inobt_key		inobt;
+	struct xfs_rmap_key		rmap;
+	struct xfs_rmap_key		__rmap_bigkey[2];
+	struct xfs_refcount_key		refc;
+};
+
+union xfs_btree_rec {
+	struct xfs_bmbt_rec		bmbt;
+	xfs_bmdr_rec_t			bmbr;	/* bmbt root block */
+	struct xfs_alloc_rec		alloc;
+	struct xfs_inobt_rec		inobt;
+	struct xfs_rmap_rec		rmap;
+	struct xfs_refcount_rec		refc;
+};
+
+/*
+ * This nonsense is to make -wlint happy.
+ */
+#define	XFS_LOOKUP_EQ	((xfs_lookup_t)XFS_LOOKUP_EQi)
+#define	XFS_LOOKUP_LE	((xfs_lookup_t)XFS_LOOKUP_LEi)
+#define	XFS_LOOKUP_GE	((xfs_lookup_t)XFS_LOOKUP_GEi)
+
+#define	XFS_BTNUM_BNO	((xfs_btnum_t)XFS_BTNUM_BNOi)
+#define	XFS_BTNUM_CNT	((xfs_btnum_t)XFS_BTNUM_CNTi)
+#define	XFS_BTNUM_BMAP	((xfs_btnum_t)XFS_BTNUM_BMAPi)
+#define	XFS_BTNUM_INO	((xfs_btnum_t)XFS_BTNUM_INOi)
+#define	XFS_BTNUM_FINO	((xfs_btnum_t)XFS_BTNUM_FINOi)
+#define	XFS_BTNUM_RMAP	((xfs_btnum_t)XFS_BTNUM_RMAPi)
+#define	XFS_BTNUM_REFC	((xfs_btnum_t)XFS_BTNUM_REFCi)
+
+uint32_t xfs_btree_magic(int crc, xfs_btnum_t btnum);
+
+/*
+ * For logging record fields.
+ */
+#define	XFS_BB_MAGIC		(1u << 0)
+#define	XFS_BB_LEVEL		(1u << 1)
+#define	XFS_BB_NUMRECS		(1u << 2)
+#define	XFS_BB_LEFTSIB		(1u << 3)
+#define	XFS_BB_RIGHTSIB		(1u << 4)
+#define	XFS_BB_BLKNO		(1u << 5)
+#define	XFS_BB_LSN		(1u << 6)
+#define	XFS_BB_UUID		(1u << 7)
+#define	XFS_BB_OWNER		(1u << 8)
+#define	XFS_BB_NUM_BITS		5
+#define	XFS_BB_ALL_BITS		((1u << XFS_BB_NUM_BITS) - 1)
+#define	XFS_BB_NUM_BITS_CRC	9
+#define	XFS_BB_ALL_BITS_CRC	((1u << XFS_BB_NUM_BITS_CRC) - 1)
+
+/*
+ * Generic stats interface
+ */
+#define XFS_BTREE_STATS_INC(cur, stat)	\
+	XFS_STATS_INC_OFF((cur)->bc_mp, (cur)->bc_statoff + __XBTS_ ## stat)
+#define XFS_BTREE_STATS_ADD(cur, stat, val)	\
+	XFS_STATS_ADD_OFF((cur)->bc_mp, (cur)->bc_statoff + __XBTS_ ## stat, val)
+
+struct xfs_btree_ops {
+	/* size of the key and record structures */
+	size_t	key_len;
+	size_t	rec_len;
+
+	/* cursor operations */
+	struct xfs_btree_cur *(*dup_cursor)(struct xfs_btree_cur *);
+	void	(*update_cursor)(struct xfs_btree_cur *src,
+				 struct xfs_btree_cur *dst);
+
+	/* update btree root pointer */
+	void	(*set_root)(struct xfs_btree_cur *cur,
+			    const union xfs_btree_ptr *nptr, int level_change);
+
+	/* block allocation / freeing */
+	int	(*alloc_block)(struct xfs_btree_cur *cur,
+			       const union xfs_btree_ptr *start_bno,
+			       union xfs_btree_ptr *new_bno,
+			       int *stat);
+	int	(*free_block)(struct xfs_btree_cur *cur, struct xfs_buf *bp);
+
+	/* update last record information */
+	void	(*update_lastrec)(struct xfs_btree_cur *cur,
+				  const struct xfs_btree_block *block,
+				  const union xfs_btree_rec *rec,
+				  int ptr, int reason);
+
+	/* records in block/level */
+	int	(*get_minrecs)(struct xfs_btree_cur *cur, int level);
+	int	(*get_maxrecs)(struct xfs_btree_cur *cur, int level);
+
+	/* records on disk.  Matter for the root in inode case. */
+	int	(*get_dmaxrecs)(struct xfs_btree_cur *cur, int level);
+
+	/* init values of btree structures */
+	void	(*init_key_from_rec)(union xfs_btree_key *key,
+				     const union xfs_btree_rec *rec);
+	void	(*init_rec_from_cur)(struct xfs_btree_cur *cur,
+				     union xfs_btree_rec *rec);
+	void	(*init_ptr_from_cur)(struct xfs_btree_cur *cur,
+				     union xfs_btree_ptr *ptr);
+	void	(*init_high_key_from_rec)(union xfs_btree_key *key,
+					  const union xfs_btree_rec *rec);
+
+	/* difference between key value and cursor value */
+	int64_t (*key_diff)(struct xfs_btree_cur *cur,
+			    const union xfs_btree_key *key);
+
+	/*
+	 * Difference between key2 and key1 -- positive if key1 > key2,
+	 * negative if key1 < key2, and zero if equal.
+	 */
+	int64_t (*diff_two_keys)(struct xfs_btree_cur *cur,
+				 const union xfs_btree_key *key1,
+				 const union xfs_btree_key *key2);
+
+	const struct xfs_buf_ops	*buf_ops;
+
+	/* check that k1 is lower than k2 */
+	int	(*keys_inorder)(struct xfs_btree_cur *cur,
+				const union xfs_btree_key *k1,
+				const union xfs_btree_key *k2);
+
+	/* check that r1 is lower than r2 */
+	int	(*recs_inorder)(struct xfs_btree_cur *cur,
+				const union xfs_btree_rec *r1,
+				const union xfs_btree_rec *r2);
+};
+
+/*
+ * Reasons for the update_lastrec method to be called.
+ */
+#define LASTREC_UPDATE	0
+#define LASTREC_INSREC	1
+#define LASTREC_DELREC	2
+
+
+union xfs_btree_irec {
+	struct xfs_alloc_rec_incore	a;
+	struct xfs_bmbt_irec		b;
+	struct xfs_inobt_rec_incore	i;
+	struct xfs_rmap_irec		r;
+	struct xfs_refcount_irec	rc;
+};
+
+/* Per-AG btree information. */
+struct xfs_btree_cur_ag {
+	struct xfs_perag		*pag;
+	union {
+		struct xfs_buf		*agbp;
+		struct xbtree_afakeroot	*afake;	/* for staging cursor */
+	};
+	union {
+		struct {
+			unsigned int	nr_ops;	/* # record updates */
+			unsigned int	shape_changes;	/* # of extent splits */
+		} refc;
+		struct {
+			bool		active;	/* allocation cursor state */
+		} abt;
+	};
+};
+
+/* Btree-in-inode cursor information */
+struct xfs_btree_cur_ino {
+	struct xfs_inode		*ip;
+	struct xbtree_ifakeroot		*ifake;	/* for staging cursor */
+	int				allocated;
+	short				forksize;
+	char				whichfork;
+	char				flags;
+/* We are converting a delalloc reservation */
+#define	XFS_BTCUR_BMBT_WASDEL		(1 << 0)
+
+/* For extent swap, ignore owner check in verifier */
+#define	XFS_BTCUR_BMBT_INVALID_OWNER	(1 << 1)
+};
+
+struct xfs_btree_level {
+	/* buffer pointer */
+	struct xfs_buf		*bp;
+
+	/* key/record number */
+	uint16_t		ptr;
+
+	/* readahead info */
+#define XFS_BTCUR_LEFTRA	(1 << 0) /* left sibling has been read-ahead */
+#define XFS_BTCUR_RIGHTRA	(1 << 1) /* right sibling has been read-ahead */
+	uint16_t		ra;
+};
+
+/*
+ * Btree cursor structure.
+ * This collects all information needed by the btree code in one place.
+ */
+struct xfs_btree_cur
+{
+	struct xfs_trans	*bc_tp;	/* transaction we're in, if any */
+	struct xfs_mount	*bc_mp;	/* file system mount struct */
+	const struct xfs_btree_ops *bc_ops;
+	struct kmem_cache	*bc_cache; /* cursor cache */
+	unsigned int		bc_flags; /* btree features - below */
+	xfs_btnum_t		bc_btnum; /* identifies which btree type */
+	union xfs_btree_irec	bc_rec;	/* current insert/search record value */
+	uint8_t			bc_nlevels; /* number of levels in the tree */
+	uint8_t			bc_maxlevels; /* maximum levels for this btree type */
+	int			bc_statoff; /* offset of btree stats array */
+
+	/*
+	 * Short btree pointers need an agno to be able to turn the pointers
+	 * into physical addresses for IO, so the btree cursor switches between
+	 * bc_ino and bc_ag based on whether XFS_BTREE_LONG_PTRS is set for the
+	 * cursor.
+	 */
+	union {
+		struct xfs_btree_cur_ag	bc_ag;
+		struct xfs_btree_cur_ino bc_ino;
+	};
+
+	/* Must be at the end of the struct! */
+	struct xfs_btree_level	bc_levels[];
+};
+
+/*
+ * Compute the size of a btree cursor that can handle a btree of a given
+ * height.  The bc_levels array handles node and leaf blocks, so its size
+ * is exactly nlevels.
+ */
+static inline size_t
+xfs_btree_cur_sizeof(unsigned int nlevels)
+{
+	return struct_size((struct xfs_btree_cur *)NULL, bc_levels, nlevels);
+}
+
+/* cursor flags */
+#define XFS_BTREE_LONG_PTRS		(1<<0)	/* pointers are 64bits long */
+#define XFS_BTREE_ROOT_IN_INODE		(1<<1)	/* root may be variable size */
+#define XFS_BTREE_LASTREC_UPDATE	(1<<2)	/* track last rec externally */
+#define XFS_BTREE_CRC_BLOCKS		(1<<3)	/* uses extended btree blocks */
+#define XFS_BTREE_OVERLAPPING		(1<<4)	/* overlapping intervals */
+/*
+ * The root of this btree is a fakeroot structure so that we can stage a btree
+ * rebuild without leaving it accessible via primary metadata.  The ops struct
+ * is dynamically allocated and must be freed when the cursor is deleted.
+ */
+#define XFS_BTREE_STAGING		(1<<5)
+
+#define	XFS_BTREE_NOERROR	0
+#define	XFS_BTREE_ERROR		1
+
+/*
+ * Convert from buffer to btree block header.
+ */
+#define	XFS_BUF_TO_BLOCK(bp)	((struct xfs_btree_block *)((bp)->b_addr))
+
+/*
+ * Internal long and short btree block checks.  They return NULL if the
+ * block is ok or the address of the failed check otherwise.
+ */
+xfs_failaddr_t __xfs_btree_check_lblock(struct xfs_btree_cur *cur,
+		struct xfs_btree_block *block, int level, struct xfs_buf *bp);
+xfs_failaddr_t __xfs_btree_check_sblock(struct xfs_btree_cur *cur,
+		struct xfs_btree_block *block, int level, struct xfs_buf *bp);
+
+/*
+ * Check that block header is ok.
+ */
+int
+xfs_btree_check_block(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	struct xfs_btree_block	*block,	/* generic btree block pointer */
+	int			level,	/* level of the btree block */
+	struct xfs_buf		*bp);	/* buffer containing block, if any */
+
+/*
+ * Check that (long) pointer is ok.
+ */
+bool					/* error (0 or EFSCORRUPTED) */
+xfs_btree_check_lptr(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_fsblock_t		fsbno,	/* btree block disk address */
+	int			level);	/* btree block level */
+
+/*
+ * Check that (short) pointer is ok.
+ */
+bool					/* error (0 or EFSCORRUPTED) */
+xfs_btree_check_sptr(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		agbno,	/* btree block disk address */
+	int			level);	/* btree block level */
+
+/*
+ * Delete the btree cursor.
+ */
+void
+xfs_btree_del_cursor(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			error);	/* del because of error */
+
+/*
+ * Duplicate the btree cursor.
+ * Allocate a new one, copy the record, re-get the buffers.
+ */
+int					/* error */
+xfs_btree_dup_cursor(
+	struct xfs_btree_cur		*cur,	/* input cursor */
+	struct xfs_btree_cur		**ncur);/* output cursor */
+
+/*
+ * Compute first and last byte offsets for the fields given.
+ * Interprets the offsets table, which contains struct field offsets.
+ */
+void
+xfs_btree_offsets(
+	uint32_t		fields,	/* bitmask of fields */
+	const short		*offsets,/* table of field offsets */
+	int			nbits,	/* number of bits to inspect */
+	int			*first,	/* output: first byte offset */
+	int			*last);	/* output: last byte offset */
+
+/*
+ * Get a buffer for the block, return it read in.
+ * Long-form addressing.
+ */
+int					/* error */
+xfs_btree_read_bufl(
+	struct xfs_mount	*mp,	/* file system mount point */
+	struct xfs_trans	*tp,	/* transaction pointer */
+	xfs_fsblock_t		fsbno,	/* file system block number */
+	struct xfs_buf		**bpp,	/* buffer for fsbno */
+	int			refval,	/* ref count value for buffer */
+	const struct xfs_buf_ops *ops);
+
+/*
+ * Read-ahead the block, don't wait for it, don't return a buffer.
+ * Long-form addressing.
+ */
+void					/* error */
+xfs_btree_reada_bufl(
+	struct xfs_mount	*mp,	/* file system mount point */
+	xfs_fsblock_t		fsbno,	/* file system block number */
+	xfs_extlen_t		count,	/* count of filesystem blocks */
+	const struct xfs_buf_ops *ops);
+
+/*
+ * Read-ahead the block, don't wait for it, don't return a buffer.
+ * Short-form addressing.
+ */
+void					/* error */
+xfs_btree_reada_bufs(
+	struct xfs_mount	*mp,	/* file system mount point */
+	xfs_agnumber_t		agno,	/* allocation group number */
+	xfs_agblock_t		agbno,	/* allocation group block number */
+	xfs_extlen_t		count,	/* count of filesystem blocks */
+	const struct xfs_buf_ops *ops);
+
+/*
+ * Initialise a new btree block header
+ */
+void
+xfs_btree_init_block(
+	struct xfs_mount *mp,
+	struct xfs_buf	*bp,
+	xfs_btnum_t	btnum,
+	__u16		level,
+	__u16		numrecs,
+	__u64		owner);
+
+void
+xfs_btree_init_block_int(
+	struct xfs_mount	*mp,
+	struct xfs_btree_block	*buf,
+	xfs_daddr_t		blkno,
+	xfs_btnum_t		btnum,
+	__u16			level,
+	__u16			numrecs,
+	__u64			owner,
+	unsigned int		flags);
+
+/*
+ * Common btree core entry points.
+ */
+int xfs_btree_increment(struct xfs_btree_cur *, int, int *);
+int xfs_btree_decrement(struct xfs_btree_cur *, int, int *);
+int xfs_btree_lookup(struct xfs_btree_cur *, xfs_lookup_t, int *);
+int xfs_btree_update(struct xfs_btree_cur *, union xfs_btree_rec *);
+int xfs_btree_new_iroot(struct xfs_btree_cur *, int *, int *);
+int xfs_btree_insert(struct xfs_btree_cur *, int *);
+int xfs_btree_delete(struct xfs_btree_cur *, int *);
+int xfs_btree_get_rec(struct xfs_btree_cur *, union xfs_btree_rec **, int *);
+int xfs_btree_change_owner(struct xfs_btree_cur *cur, uint64_t new_owner,
+			   struct list_head *buffer_list);
+
+/*
+ * btree block CRC helpers
+ */
+void xfs_btree_lblock_calc_crc(struct xfs_buf *);
+bool xfs_btree_lblock_verify_crc(struct xfs_buf *);
+void xfs_btree_sblock_calc_crc(struct xfs_buf *);
+bool xfs_btree_sblock_verify_crc(struct xfs_buf *);
+
+/*
+ * Internal btree helpers also used by xfs_bmap.c.
+ */
+void xfs_btree_log_block(struct xfs_btree_cur *, struct xfs_buf *, uint32_t);
+void xfs_btree_log_recs(struct xfs_btree_cur *, struct xfs_buf *, int, int);
+
+/*
+ * Helpers.
+ */
+static inline int xfs_btree_get_numrecs(const struct xfs_btree_block *block)
+{
+	return be16_to_cpu(block->bb_numrecs);
+}
+
+static inline void xfs_btree_set_numrecs(struct xfs_btree_block *block,
+		uint16_t numrecs)
+{
+	block->bb_numrecs = cpu_to_be16(numrecs);
+}
+
+static inline int xfs_btree_get_level(const struct xfs_btree_block *block)
+{
+	return be16_to_cpu(block->bb_level);
+}
+
+
+/*
+ * Min and max functions for extlen, agblock, fileoff, and filblks types.
+ */
+#define	XFS_EXTLEN_MIN(a,b)	min_t(xfs_extlen_t, (a), (b))
+#define	XFS_EXTLEN_MAX(a,b)	max_t(xfs_extlen_t, (a), (b))
+#define	XFS_AGBLOCK_MIN(a,b)	min_t(xfs_agblock_t, (a), (b))
+#define	XFS_AGBLOCK_MAX(a,b)	max_t(xfs_agblock_t, (a), (b))
+#define	XFS_FILEOFF_MIN(a,b)	min_t(xfs_fileoff_t, (a), (b))
+#define	XFS_FILEOFF_MAX(a,b)	max_t(xfs_fileoff_t, (a), (b))
+#define	XFS_FILBLKS_MIN(a,b)	min_t(xfs_filblks_t, (a), (b))
+#define	XFS_FILBLKS_MAX(a,b)	max_t(xfs_filblks_t, (a), (b))
+
+xfs_failaddr_t xfs_btree_sblock_v5hdr_verify(struct xfs_buf *bp);
+xfs_failaddr_t xfs_btree_sblock_verify(struct xfs_buf *bp,
+		unsigned int max_recs);
+xfs_failaddr_t xfs_btree_lblock_v5hdr_verify(struct xfs_buf *bp,
+		uint64_t owner);
+xfs_failaddr_t xfs_btree_lblock_verify(struct xfs_buf *bp,
+		unsigned int max_recs);
+
+unsigned int xfs_btree_compute_maxlevels(const unsigned int *limits,
+		unsigned long long records);
+unsigned long long xfs_btree_calc_size(const unsigned int *limits,
+		unsigned long long records);
+unsigned int xfs_btree_space_to_height(const unsigned int *limits,
+		unsigned long long blocks);
+
+/*
+ * Return codes for the query range iterator function are 0 to continue
+ * iterating, and non-zero to stop iterating.  Any non-zero value will be
+ * passed up to the _query_range caller.  The special value -ECANCELED can be
+ * used to stop iteration, because _query_range never generates that error
+ * code on its own.
+ */
+typedef int (*xfs_btree_query_range_fn)(struct xfs_btree_cur *cur,
+		const union xfs_btree_rec *rec, void *priv);
+
+int xfs_btree_query_range(struct xfs_btree_cur *cur,
+		const union xfs_btree_irec *low_rec,
+		const union xfs_btree_irec *high_rec,
+		xfs_btree_query_range_fn fn, void *priv);
+int xfs_btree_query_all(struct xfs_btree_cur *cur, xfs_btree_query_range_fn fn,
+		void *priv);
+
+typedef int (*xfs_btree_visit_blocks_fn)(struct xfs_btree_cur *cur, int level,
+		void *data);
+/* Visit record blocks. */
+#define XFS_BTREE_VISIT_RECORDS		(1 << 0)
+/* Visit leaf blocks. */
+#define XFS_BTREE_VISIT_LEAVES		(1 << 1)
+/* Visit all blocks. */
+#define XFS_BTREE_VISIT_ALL		(XFS_BTREE_VISIT_RECORDS | \
+					 XFS_BTREE_VISIT_LEAVES)
+int xfs_btree_visit_blocks(struct xfs_btree_cur *cur,
+		xfs_btree_visit_blocks_fn fn, unsigned int flags, void *data);
+
+int xfs_btree_count_blocks(struct xfs_btree_cur *cur, xfs_extlen_t *blocks);
+
+union xfs_btree_rec *xfs_btree_rec_addr(struct xfs_btree_cur *cur, int n,
+		struct xfs_btree_block *block);
+union xfs_btree_key *xfs_btree_key_addr(struct xfs_btree_cur *cur, int n,
+		struct xfs_btree_block *block);
+union xfs_btree_key *xfs_btree_high_key_addr(struct xfs_btree_cur *cur, int n,
+		struct xfs_btree_block *block);
+union xfs_btree_ptr *xfs_btree_ptr_addr(struct xfs_btree_cur *cur, int n,
+		struct xfs_btree_block *block);
+int xfs_btree_lookup_get_block(struct xfs_btree_cur *cur, int level,
+		const union xfs_btree_ptr *pp, struct xfs_btree_block **blkp);
+struct xfs_btree_block *xfs_btree_get_block(struct xfs_btree_cur *cur,
+		int level, struct xfs_buf **bpp);
+bool xfs_btree_ptr_is_null(struct xfs_btree_cur *cur,
+		const union xfs_btree_ptr *ptr);
+int64_t xfs_btree_diff_two_ptrs(struct xfs_btree_cur *cur,
+				const union xfs_btree_ptr *a,
+				const union xfs_btree_ptr *b);
+void xfs_btree_get_sibling(struct xfs_btree_cur *cur,
+			   struct xfs_btree_block *block,
+			   union xfs_btree_ptr *ptr, int lr);
+void xfs_btree_get_keys(struct xfs_btree_cur *cur,
+		struct xfs_btree_block *block, union xfs_btree_key *key);
+union xfs_btree_key *xfs_btree_high_key_from_key(struct xfs_btree_cur *cur,
+		union xfs_btree_key *key);
+int xfs_btree_has_record(struct xfs_btree_cur *cur,
+		const union xfs_btree_irec *low,
+		const union xfs_btree_irec *high, bool *exists);
+bool xfs_btree_has_more_records(struct xfs_btree_cur *cur);
+struct xfs_ifork *xfs_btree_ifork_ptr(struct xfs_btree_cur *cur);
+
+/* Does this cursor point to the last block in the given level? */
+static inline bool
+xfs_btree_islastblock(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	struct xfs_btree_block	*block;
+	struct xfs_buf		*bp;
+
+	block = xfs_btree_get_block(cur, level, &bp);
+	ASSERT(block && xfs_btree_check_block(cur, block, level, bp) == 0);
+
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		return block->bb_u.l.bb_rightsib == cpu_to_be64(NULLFSBLOCK);
+	return block->bb_u.s.bb_rightsib == cpu_to_be32(NULLAGBLOCK);
+}
+
+void xfs_btree_set_ptr_null(struct xfs_btree_cur *cur,
+		union xfs_btree_ptr *ptr);
+int xfs_btree_get_buf_block(struct xfs_btree_cur *cur,
+		const union xfs_btree_ptr *ptr, struct xfs_btree_block **block,
+		struct xfs_buf **bpp);
+void xfs_btree_set_sibling(struct xfs_btree_cur *cur,
+		struct xfs_btree_block *block, const union xfs_btree_ptr *ptr,
+		int lr);
+void xfs_btree_init_block_cur(struct xfs_btree_cur *cur,
+		struct xfs_buf *bp, int level, int numrecs);
+void xfs_btree_copy_ptrs(struct xfs_btree_cur *cur,
+		union xfs_btree_ptr *dst_ptr,
+		const union xfs_btree_ptr *src_ptr, int numptrs);
+void xfs_btree_copy_keys(struct xfs_btree_cur *cur,
+		union xfs_btree_key *dst_key,
+		const union xfs_btree_key *src_key, int numkeys);
+
+static inline struct xfs_btree_cur *
+xfs_btree_alloc_cursor(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	xfs_btnum_t		btnum,
+	uint8_t			maxlevels,
+	struct kmem_cache	*cache)
+{
+	struct xfs_btree_cur	*cur;
+
+	cur = kmem_cache_zalloc(cache, GFP_NOFS | __GFP_NOFAIL);
+	cur->bc_tp = tp;
+	cur->bc_mp = mp;
+	cur->bc_btnum = btnum;
+	cur->bc_maxlevels = maxlevels;
+	cur->bc_cache = cache;
+
+	return cur;
+}
+
+int __init xfs_btree_init_cur_caches(void);
+void xfs_btree_destroy_cur_caches(void);
+
+#endif	/* __XFS_BTREE_H__ */
diff --git a/fs/xfs/libxfs/xfs_btree_staging.c b/fs/xfs/libxfs/xfs_btree_staging.c
new file mode 100644
index 000000000..dd75e208b
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_btree_staging.c
@@ -0,0 +1,880 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) 2020 Oracle.  All Rights Reserved.
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_bit.h"
+#include "xfs_mount.h"
+#include "xfs_inode.h"
+#include "xfs_trans.h"
+#include "xfs_btree.h"
+#include "xfs_trace.h"
+#include "xfs_btree_staging.h"
+
+/*
+ * Staging Cursors and Fake Roots for Btrees
+ * =========================================
+ *
+ * A staging btree cursor is a special type of btree cursor that callers must
+ * use to construct a new btree index using the btree bulk loader code.  The
+ * bulk loading code uses the staging btree cursor to abstract the details of
+ * initializing new btree blocks and filling them with records or key/ptr
+ * pairs.  Regular btree operations (e.g. queries and modifications) are not
+ * supported with staging cursors, and callers must not invoke them.
+ *
+ * Fake root structures contain all the information about a btree that is under
+ * construction by the bulk loading code.  Staging btree cursors point to fake
+ * root structures instead of the usual AG header or inode structure.
+ *
+ * Callers are expected to initialize a fake root structure and pass it into
+ * the _stage_cursor function for a specific btree type.  When bulk loading is
+ * complete, callers should call the _commit_staged_btree function for that
+ * specific btree type to commit the new btree into the filesystem.
+ */
+
+/*
+ * Don't allow staging cursors to be duplicated because they're supposed to be
+ * kept private to a single thread.
+ */
+STATIC struct xfs_btree_cur *
+xfs_btree_fakeroot_dup_cursor(
+	struct xfs_btree_cur	*cur)
+{
+	ASSERT(0);
+	return NULL;
+}
+
+/*
+ * Don't allow block allocation for a staging cursor, because staging cursors
+ * do not support regular btree modifications.
+ *
+ * Bulk loading uses a separate callback to obtain new blocks from a
+ * preallocated list, which prevents ENOSPC failures during loading.
+ */
+STATIC int
+xfs_btree_fakeroot_alloc_block(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*start_bno,
+	union xfs_btree_ptr		*new_bno,
+	int				*stat)
+{
+	ASSERT(0);
+	return -EFSCORRUPTED;
+}
+
+/*
+ * Don't allow block freeing for a staging cursor, because staging cursors
+ * do not support regular btree modifications.
+ */
+STATIC int
+xfs_btree_fakeroot_free_block(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp)
+{
+	ASSERT(0);
+	return -EFSCORRUPTED;
+}
+
+/* Initialize a pointer to the root block from the fakeroot. */
+STATIC void
+xfs_btree_fakeroot_init_ptr_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	struct xbtree_afakeroot	*afake;
+
+	ASSERT(cur->bc_flags & XFS_BTREE_STAGING);
+
+	afake = cur->bc_ag.afake;
+	ptr->s = cpu_to_be32(afake->af_root);
+}
+
+/*
+ * Bulk Loading for AG Btrees
+ * ==========================
+ *
+ * For a btree rooted in an AG header, pass a xbtree_afakeroot structure to the
+ * staging cursor.  Callers should initialize this to zero.
+ *
+ * The _stage_cursor() function for a specific btree type should call
+ * xfs_btree_stage_afakeroot to set up the in-memory cursor as a staging
+ * cursor.  The corresponding _commit_staged_btree() function should log the
+ * new root and call xfs_btree_commit_afakeroot() to transform the staging
+ * cursor into a regular btree cursor.
+ */
+
+/* Update the btree root information for a per-AG fake root. */
+STATIC void
+xfs_btree_afakeroot_set_root(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*ptr,
+	int				inc)
+{
+	struct xbtree_afakeroot	*afake = cur->bc_ag.afake;
+
+	ASSERT(cur->bc_flags & XFS_BTREE_STAGING);
+	afake->af_root = be32_to_cpu(ptr->s);
+	afake->af_levels += inc;
+}
+
+/*
+ * Initialize a AG-rooted btree cursor with the given AG btree fake root.
+ * The btree cursor's bc_ops will be overridden as needed to make the staging
+ * functionality work.
+ */
+void
+xfs_btree_stage_afakeroot(
+	struct xfs_btree_cur		*cur,
+	struct xbtree_afakeroot		*afake)
+{
+	struct xfs_btree_ops		*nops;
+
+	ASSERT(!(cur->bc_flags & XFS_BTREE_STAGING));
+	ASSERT(!(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE));
+	ASSERT(cur->bc_tp == NULL);
+
+	nops = kmem_alloc(sizeof(struct xfs_btree_ops), KM_NOFS);
+	memcpy(nops, cur->bc_ops, sizeof(struct xfs_btree_ops));
+	nops->alloc_block = xfs_btree_fakeroot_alloc_block;
+	nops->free_block = xfs_btree_fakeroot_free_block;
+	nops->init_ptr_from_cur = xfs_btree_fakeroot_init_ptr_from_cur;
+	nops->set_root = xfs_btree_afakeroot_set_root;
+	nops->dup_cursor = xfs_btree_fakeroot_dup_cursor;
+
+	cur->bc_ag.afake = afake;
+	cur->bc_nlevels = afake->af_levels;
+	cur->bc_ops = nops;
+	cur->bc_flags |= XFS_BTREE_STAGING;
+}
+
+/*
+ * Transform an AG-rooted staging btree cursor back into a regular cursor by
+ * substituting a real btree root for the fake one and restoring normal btree
+ * cursor ops.  The caller must log the btree root change prior to calling
+ * this.
+ */
+void
+xfs_btree_commit_afakeroot(
+	struct xfs_btree_cur		*cur,
+	struct xfs_trans		*tp,
+	struct xfs_buf			*agbp,
+	const struct xfs_btree_ops	*ops)
+{
+	ASSERT(cur->bc_flags & XFS_BTREE_STAGING);
+	ASSERT(cur->bc_tp == NULL);
+
+	trace_xfs_btree_commit_afakeroot(cur);
+
+	kmem_free((void *)cur->bc_ops);
+	cur->bc_ag.agbp = agbp;
+	cur->bc_ops = ops;
+	cur->bc_flags &= ~XFS_BTREE_STAGING;
+	cur->bc_tp = tp;
+}
+
+/*
+ * Bulk Loading for Inode-Rooted Btrees
+ * ====================================
+ *
+ * For a btree rooted in an inode fork, pass a xbtree_ifakeroot structure to
+ * the staging cursor.  This structure should be initialized as follows:
+ *
+ * - if_fork_size field should be set to the number of bytes available to the
+ *   fork in the inode.
+ *
+ * - if_fork should point to a freshly allocated struct xfs_ifork.
+ *
+ * - if_format should be set to the appropriate fork type (e.g.
+ *   XFS_DINODE_FMT_BTREE).
+ *
+ * All other fields must be zero.
+ *
+ * The _stage_cursor() function for a specific btree type should call
+ * xfs_btree_stage_ifakeroot to set up the in-memory cursor as a staging
+ * cursor.  The corresponding _commit_staged_btree() function should log the
+ * new root and call xfs_btree_commit_ifakeroot() to transform the staging
+ * cursor into a regular btree cursor.
+ */
+
+/*
+ * Initialize an inode-rooted btree cursor with the given inode btree fake
+ * root.  The btree cursor's bc_ops will be overridden as needed to make the
+ * staging functionality work.  If new_ops is not NULL, these new ops will be
+ * passed out to the caller for further overriding.
+ */
+void
+xfs_btree_stage_ifakeroot(
+	struct xfs_btree_cur		*cur,
+	struct xbtree_ifakeroot		*ifake,
+	struct xfs_btree_ops		**new_ops)
+{
+	struct xfs_btree_ops		*nops;
+
+	ASSERT(!(cur->bc_flags & XFS_BTREE_STAGING));
+	ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
+	ASSERT(cur->bc_tp == NULL);
+
+	nops = kmem_alloc(sizeof(struct xfs_btree_ops), KM_NOFS);
+	memcpy(nops, cur->bc_ops, sizeof(struct xfs_btree_ops));
+	nops->alloc_block = xfs_btree_fakeroot_alloc_block;
+	nops->free_block = xfs_btree_fakeroot_free_block;
+	nops->init_ptr_from_cur = xfs_btree_fakeroot_init_ptr_from_cur;
+	nops->dup_cursor = xfs_btree_fakeroot_dup_cursor;
+
+	cur->bc_ino.ifake = ifake;
+	cur->bc_nlevels = ifake->if_levels;
+	cur->bc_ops = nops;
+	cur->bc_flags |= XFS_BTREE_STAGING;
+
+	if (new_ops)
+		*new_ops = nops;
+}
+
+/*
+ * Transform an inode-rooted staging btree cursor back into a regular cursor by
+ * substituting a real btree root for the fake one and restoring normal btree
+ * cursor ops.  The caller must log the btree root change prior to calling
+ * this.
+ */
+void
+xfs_btree_commit_ifakeroot(
+	struct xfs_btree_cur		*cur,
+	struct xfs_trans		*tp,
+	int				whichfork,
+	const struct xfs_btree_ops	*ops)
+{
+	ASSERT(cur->bc_flags & XFS_BTREE_STAGING);
+	ASSERT(cur->bc_tp == NULL);
+
+	trace_xfs_btree_commit_ifakeroot(cur);
+
+	kmem_free((void *)cur->bc_ops);
+	cur->bc_ino.ifake = NULL;
+	cur->bc_ino.whichfork = whichfork;
+	cur->bc_ops = ops;
+	cur->bc_flags &= ~XFS_BTREE_STAGING;
+	cur->bc_tp = tp;
+}
+
+/*
+ * Bulk Loading of Staged Btrees
+ * =============================
+ *
+ * This interface is used with a staged btree cursor to create a totally new
+ * btree with a large number of records (i.e. more than what would fit in a
+ * single root block).  When the creation is complete, the new root can be
+ * linked atomically into the filesystem by committing the staged cursor.
+ *
+ * Creation of a new btree proceeds roughly as follows:
+ *
+ * The first step is to initialize an appropriate fake btree root structure and
+ * then construct a staged btree cursor.  Refer to the block comments about
+ * "Bulk Loading for AG Btrees" and "Bulk Loading for Inode-Rooted Btrees" for
+ * more information about how to do this.
+ *
+ * The second step is to initialize a struct xfs_btree_bload context as
+ * documented in the structure definition.
+ *
+ * The third step is to call xfs_btree_bload_compute_geometry to compute the
+ * height of and the number of blocks needed to construct the btree.  See the
+ * section "Computing the Geometry of the New Btree" for details about this
+ * computation.
+ *
+ * In step four, the caller must allocate xfs_btree_bload.nr_blocks blocks and
+ * save them for later use by ->claim_block().  Bulk loading requires all
+ * blocks to be allocated beforehand to avoid ENOSPC failures midway through a
+ * rebuild, and to minimize seek distances of the new btree.
+ *
+ * Step five is to call xfs_btree_bload() to start constructing the btree.
+ *
+ * The final step is to commit the staging btree cursor, which logs the new
+ * btree root and turns the staging cursor into a regular cursor.  The caller
+ * is responsible for cleaning up the previous btree blocks, if any.
+ *
+ * Computing the Geometry of the New Btree
+ * =======================================
+ *
+ * The number of items placed in each btree block is computed via the following
+ * algorithm: For leaf levels, the number of items for the level is nr_records
+ * in the bload structure.  For node levels, the number of items for the level
+ * is the number of blocks in the next lower level of the tree.  For each
+ * level, the desired number of items per block is defined as:
+ *
+ * desired = max(minrecs, maxrecs - slack factor)
+ *
+ * The number of blocks for the level is defined to be:
+ *
+ * blocks = floor(nr_items / desired)
+ *
+ * Note this is rounded down so that the npb calculation below will never fall
+ * below minrecs.  The number of items that will actually be loaded into each
+ * btree block is defined as:
+ *
+ * npb =  nr_items / blocks
+ *
+ * Some of the leftmost blocks in the level will contain one extra record as
+ * needed to handle uneven division.  If the number of records in any block
+ * would exceed maxrecs for that level, blocks is incremented and npb is
+ * recalculated.
+ *
+ * In other words, we compute the number of blocks needed to satisfy a given
+ * loading level, then spread the items as evenly as possible.
+ *
+ * The height and number of fs blocks required to create the btree are computed
+ * and returned via btree_height and nr_blocks.
+ */
+
+/*
+ * Put a btree block that we're loading onto the ordered list and release it.
+ * The btree blocks will be written to disk when bulk loading is finished.
+ */
+static void
+xfs_btree_bload_drop_buf(
+	struct list_head	*buffers_list,
+	struct xfs_buf		**bpp)
+{
+	if (*bpp == NULL)
+		return;
+
+	if (!xfs_buf_delwri_queue(*bpp, buffers_list))
+		ASSERT(0);
+
+	xfs_buf_relse(*bpp);
+	*bpp = NULL;
+}
+
+/*
+ * Allocate and initialize one btree block for bulk loading.
+ *
+ * The new btree block will have its level and numrecs fields set to the values
+ * of the level and nr_this_block parameters, respectively.
+ *
+ * The caller should ensure that ptrp, bpp, and blockp refer to the left
+ * sibling of the new block, if there is any.  On exit, ptrp, bpp, and blockp
+ * will all point to the new block.
+ */
+STATIC int
+xfs_btree_bload_prep_block(
+	struct xfs_btree_cur		*cur,
+	struct xfs_btree_bload		*bbl,
+	struct list_head		*buffers_list,
+	unsigned int			level,
+	unsigned int			nr_this_block,
+	union xfs_btree_ptr		*ptrp, /* in/out */
+	struct xfs_buf			**bpp, /* in/out */
+	struct xfs_btree_block		**blockp, /* in/out */
+	void				*priv)
+{
+	union xfs_btree_ptr		new_ptr;
+	struct xfs_buf			*new_bp;
+	struct xfs_btree_block		*new_block;
+	int				ret;
+
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    level == cur->bc_nlevels - 1) {
+		struct xfs_ifork	*ifp = xfs_btree_ifork_ptr(cur);
+		size_t			new_size;
+
+		ASSERT(*bpp == NULL);
+
+		/* Allocate a new incore btree root block. */
+		new_size = bbl->iroot_size(cur, nr_this_block, priv);
+		ifp->if_broot = kmem_zalloc(new_size, 0);
+		ifp->if_broot_bytes = (int)new_size;
+
+		/* Initialize it and send it out. */
+		xfs_btree_init_block_int(cur->bc_mp, ifp->if_broot,
+				XFS_BUF_DADDR_NULL, cur->bc_btnum, level,
+				nr_this_block, cur->bc_ino.ip->i_ino,
+				cur->bc_flags);
+
+		*bpp = NULL;
+		*blockp = ifp->if_broot;
+		xfs_btree_set_ptr_null(cur, ptrp);
+		return 0;
+	}
+
+	/* Claim one of the caller's preallocated blocks. */
+	xfs_btree_set_ptr_null(cur, &new_ptr);
+	ret = bbl->claim_block(cur, &new_ptr, priv);
+	if (ret)
+		return ret;
+
+	ASSERT(!xfs_btree_ptr_is_null(cur, &new_ptr));
+
+	ret = xfs_btree_get_buf_block(cur, &new_ptr, &new_block, &new_bp);
+	if (ret)
+		return ret;
+
+	/*
+	 * The previous block (if any) is the left sibling of the new block,
+	 * so set its right sibling pointer to the new block and drop it.
+	 */
+	if (*blockp)
+		xfs_btree_set_sibling(cur, *blockp, &new_ptr, XFS_BB_RIGHTSIB);
+	xfs_btree_bload_drop_buf(buffers_list, bpp);
+
+	/* Initialize the new btree block. */
+	xfs_btree_init_block_cur(cur, new_bp, level, nr_this_block);
+	xfs_btree_set_sibling(cur, new_block, ptrp, XFS_BB_LEFTSIB);
+
+	/* Set the out parameters. */
+	*bpp = new_bp;
+	*blockp = new_block;
+	xfs_btree_copy_ptrs(cur, ptrp, &new_ptr, 1);
+	return 0;
+}
+
+/* Load one leaf block. */
+STATIC int
+xfs_btree_bload_leaf(
+	struct xfs_btree_cur		*cur,
+	unsigned int			recs_this_block,
+	xfs_btree_bload_get_record_fn	get_record,
+	struct xfs_btree_block		*block,
+	void				*priv)
+{
+	unsigned int			j;
+	int				ret;
+
+	/* Fill the leaf block with records. */
+	for (j = 1; j <= recs_this_block; j++) {
+		union xfs_btree_rec	*block_rec;
+
+		ret = get_record(cur, priv);
+		if (ret)
+			return ret;
+		block_rec = xfs_btree_rec_addr(cur, j, block);
+		cur->bc_ops->init_rec_from_cur(cur, block_rec);
+	}
+
+	return 0;
+}
+
+/*
+ * Load one node block with key/ptr pairs.
+ *
+ * child_ptr must point to a block within the next level down in the tree.  A
+ * key/ptr entry will be created in the new node block to the block pointed to
+ * by child_ptr.  On exit, child_ptr points to the next block on the child
+ * level that needs processing.
+ */
+STATIC int
+xfs_btree_bload_node(
+	struct xfs_btree_cur	*cur,
+	unsigned int		recs_this_block,
+	union xfs_btree_ptr	*child_ptr,
+	struct xfs_btree_block	*block)
+{
+	unsigned int		j;
+	int			ret;
+
+	/* Fill the node block with keys and pointers. */
+	for (j = 1; j <= recs_this_block; j++) {
+		union xfs_btree_key	child_key;
+		union xfs_btree_ptr	*block_ptr;
+		union xfs_btree_key	*block_key;
+		struct xfs_btree_block	*child_block;
+		struct xfs_buf		*child_bp;
+
+		ASSERT(!xfs_btree_ptr_is_null(cur, child_ptr));
+
+		ret = xfs_btree_get_buf_block(cur, child_ptr, &child_block,
+				&child_bp);
+		if (ret)
+			return ret;
+
+		block_ptr = xfs_btree_ptr_addr(cur, j, block);
+		xfs_btree_copy_ptrs(cur, block_ptr, child_ptr, 1);
+
+		block_key = xfs_btree_key_addr(cur, j, block);
+		xfs_btree_get_keys(cur, child_block, &child_key);
+		xfs_btree_copy_keys(cur, block_key, &child_key, 1);
+
+		xfs_btree_get_sibling(cur, child_block, child_ptr,
+				XFS_BB_RIGHTSIB);
+		xfs_buf_relse(child_bp);
+	}
+
+	return 0;
+}
+
+/*
+ * Compute the maximum number of records (or keyptrs) per block that we want to
+ * install at this level in the btree.  Caller is responsible for having set
+ * @cur->bc_ino.forksize to the desired fork size, if appropriate.
+ */
+STATIC unsigned int
+xfs_btree_bload_max_npb(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_bload	*bbl,
+	unsigned int		level)
+{
+	unsigned int		ret;
+
+	if (level == cur->bc_nlevels - 1 && cur->bc_ops->get_dmaxrecs)
+		return cur->bc_ops->get_dmaxrecs(cur, level);
+
+	ret = cur->bc_ops->get_maxrecs(cur, level);
+	if (level == 0)
+		ret -= bbl->leaf_slack;
+	else
+		ret -= bbl->node_slack;
+	return ret;
+}
+
+/*
+ * Compute the desired number of records (or keyptrs) per block that we want to
+ * install at this level in the btree, which must be somewhere between minrecs
+ * and max_npb.  The caller is free to install fewer records per block.
+ */
+STATIC unsigned int
+xfs_btree_bload_desired_npb(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_bload	*bbl,
+	unsigned int		level)
+{
+	unsigned int		npb = xfs_btree_bload_max_npb(cur, bbl, level);
+
+	/* Root blocks are not subject to minrecs rules. */
+	if (level == cur->bc_nlevels - 1)
+		return max(1U, npb);
+
+	return max_t(unsigned int, cur->bc_ops->get_minrecs(cur, level), npb);
+}
+
+/*
+ * Compute the number of records to be stored in each block at this level and
+ * the number of blocks for this level.  For leaf levels, we must populate an
+ * empty root block even if there are no records, so we have to have at least
+ * one block.
+ */
+STATIC void
+xfs_btree_bload_level_geometry(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_bload	*bbl,
+	unsigned int		level,
+	uint64_t		nr_this_level,
+	unsigned int		*avg_per_block,
+	uint64_t		*blocks,
+	uint64_t		*blocks_with_extra)
+{
+	uint64_t		npb;
+	uint64_t		dontcare;
+	unsigned int		desired_npb;
+	unsigned int		maxnr;
+
+	maxnr = cur->bc_ops->get_maxrecs(cur, level);
+
+	/*
+	 * Compute the number of blocks we need to fill each block with the
+	 * desired number of records/keyptrs per block.  Because desired_npb
+	 * could be minrecs, we use regular integer division (which rounds
+	 * the block count down) so that in the next step the effective # of
+	 * items per block will never be less than desired_npb.
+	 */
+	desired_npb = xfs_btree_bload_desired_npb(cur, bbl, level);
+	*blocks = div64_u64_rem(nr_this_level, desired_npb, &dontcare);
+	*blocks = max(1ULL, *blocks);
+
+	/*
+	 * Compute the number of records that we will actually put in each
+	 * block, assuming that we want to spread the records evenly between
+	 * the blocks.  Take care that the effective # of items per block (npb)
+	 * won't exceed maxrecs even for the blocks that get an extra record,
+	 * since desired_npb could be maxrecs, and in the previous step we
+	 * rounded the block count down.
+	 */
+	npb = div64_u64_rem(nr_this_level, *blocks, blocks_with_extra);
+	if (npb > maxnr || (npb == maxnr && *blocks_with_extra > 0)) {
+		(*blocks)++;
+		npb = div64_u64_rem(nr_this_level, *blocks, blocks_with_extra);
+	}
+
+	*avg_per_block = min_t(uint64_t, npb, nr_this_level);
+
+	trace_xfs_btree_bload_level_geometry(cur, level, nr_this_level,
+			*avg_per_block, desired_npb, *blocks,
+			*blocks_with_extra);
+}
+
+/*
+ * Ensure a slack value is appropriate for the btree.
+ *
+ * If the slack value is negative, set slack so that we fill the block to
+ * halfway between minrecs and maxrecs.  Make sure the slack is never so large
+ * that we can underflow minrecs.
+ */
+static void
+xfs_btree_bload_ensure_slack(
+	struct xfs_btree_cur	*cur,
+	int			*slack,
+	int			level)
+{
+	int			maxr;
+	int			minr;
+
+	maxr = cur->bc_ops->get_maxrecs(cur, level);
+	minr = cur->bc_ops->get_minrecs(cur, level);
+
+	/*
+	 * If slack is negative, automatically set slack so that we load the
+	 * btree block approximately halfway between minrecs and maxrecs.
+	 * Generally, this will net us 75% loading.
+	 */
+	if (*slack < 0)
+		*slack = maxr - ((maxr + minr) >> 1);
+
+	*slack = min(*slack, maxr - minr);
+}
+
+/*
+ * Prepare a btree cursor for a bulk load operation by computing the geometry
+ * fields in bbl.  Caller must ensure that the btree cursor is a staging
+ * cursor.  This function can be called multiple times.
+ */
+int
+xfs_btree_bload_compute_geometry(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_bload	*bbl,
+	uint64_t		nr_records)
+{
+	uint64_t		nr_blocks = 0;
+	uint64_t		nr_this_level;
+
+	ASSERT(cur->bc_flags & XFS_BTREE_STAGING);
+
+	/*
+	 * Make sure that the slack values make sense for traditional leaf and
+	 * node blocks.  Inode-rooted btrees will return different minrecs and
+	 * maxrecs values for the root block (bc_nlevels == level - 1).  We're
+	 * checking levels 0 and 1 here, so set bc_nlevels such that the btree
+	 * code doesn't interpret either as the root level.
+	 */
+	cur->bc_nlevels = cur->bc_maxlevels - 1;
+	xfs_btree_bload_ensure_slack(cur, &bbl->leaf_slack, 0);
+	xfs_btree_bload_ensure_slack(cur, &bbl->node_slack, 1);
+
+	bbl->nr_records = nr_this_level = nr_records;
+	for (cur->bc_nlevels = 1; cur->bc_nlevels <= cur->bc_maxlevels;) {
+		uint64_t	level_blocks;
+		uint64_t	dontcare64;
+		unsigned int	level = cur->bc_nlevels - 1;
+		unsigned int	avg_per_block;
+
+		xfs_btree_bload_level_geometry(cur, bbl, level, nr_this_level,
+				&avg_per_block, &level_blocks, &dontcare64);
+
+		if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) {
+			/*
+			 * If all the items we want to store at this level
+			 * would fit in the inode root block, then we have our
+			 * btree root and are done.
+			 *
+			 * Note that bmap btrees forbid records in the root.
+			 */
+			if (level != 0 && nr_this_level <= avg_per_block) {
+				nr_blocks++;
+				break;
+			}
+
+			/*
+			 * Otherwise, we have to store all the items for this
+			 * level in traditional btree blocks and therefore need
+			 * another level of btree to point to those blocks.
+			 *
+			 * We have to re-compute the geometry for each level of
+			 * an inode-rooted btree because the geometry differs
+			 * between a btree root in an inode fork and a
+			 * traditional btree block.
+			 *
+			 * This distinction is made in the btree code based on
+			 * whether level == bc_nlevels - 1.  Based on the
+			 * previous root block size check against the root
+			 * block geometry, we know that we aren't yet ready to
+			 * populate the root.  Increment bc_nevels and
+			 * recalculate the geometry for a traditional
+			 * block-based btree level.
+			 */
+			cur->bc_nlevels++;
+			ASSERT(cur->bc_nlevels <= cur->bc_maxlevels);
+			xfs_btree_bload_level_geometry(cur, bbl, level,
+					nr_this_level, &avg_per_block,
+					&level_blocks, &dontcare64);
+		} else {
+			/*
+			 * If all the items we want to store at this level
+			 * would fit in a single root block, we're done.
+			 */
+			if (nr_this_level <= avg_per_block) {
+				nr_blocks++;
+				break;
+			}
+
+			/* Otherwise, we need another level of btree. */
+			cur->bc_nlevels++;
+			ASSERT(cur->bc_nlevels <= cur->bc_maxlevels);
+		}
+
+		nr_blocks += level_blocks;
+		nr_this_level = level_blocks;
+	}
+
+	if (cur->bc_nlevels > cur->bc_maxlevels)
+		return -EOVERFLOW;
+
+	bbl->btree_height = cur->bc_nlevels;
+	if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE)
+		bbl->nr_blocks = nr_blocks - 1;
+	else
+		bbl->nr_blocks = nr_blocks;
+	return 0;
+}
+
+/* Bulk load a btree given the parameters and geometry established in bbl. */
+int
+xfs_btree_bload(
+	struct xfs_btree_cur		*cur,
+	struct xfs_btree_bload		*bbl,
+	void				*priv)
+{
+	struct list_head		buffers_list;
+	union xfs_btree_ptr		child_ptr;
+	union xfs_btree_ptr		ptr;
+	struct xfs_buf			*bp = NULL;
+	struct xfs_btree_block		*block = NULL;
+	uint64_t			nr_this_level = bbl->nr_records;
+	uint64_t			blocks;
+	uint64_t			i;
+	uint64_t			blocks_with_extra;
+	uint64_t			total_blocks = 0;
+	unsigned int			avg_per_block;
+	unsigned int			level = 0;
+	int				ret;
+
+	ASSERT(cur->bc_flags & XFS_BTREE_STAGING);
+
+	INIT_LIST_HEAD(&buffers_list);
+	cur->bc_nlevels = bbl->btree_height;
+	xfs_btree_set_ptr_null(cur, &child_ptr);
+	xfs_btree_set_ptr_null(cur, &ptr);
+
+	xfs_btree_bload_level_geometry(cur, bbl, level, nr_this_level,
+			&avg_per_block, &blocks, &blocks_with_extra);
+
+	/* Load each leaf block. */
+	for (i = 0; i < blocks; i++) {
+		unsigned int		nr_this_block = avg_per_block;
+
+		/*
+		 * Due to rounding, btree blocks will not be evenly populated
+		 * in most cases.  blocks_with_extra tells us how many blocks
+		 * will receive an extra record to distribute the excess across
+		 * the current level as evenly as possible.
+		 */
+		if (i < blocks_with_extra)
+			nr_this_block++;
+
+		ret = xfs_btree_bload_prep_block(cur, bbl, &buffers_list, level,
+				nr_this_block, &ptr, &bp, &block, priv);
+		if (ret)
+			goto out;
+
+		trace_xfs_btree_bload_block(cur, level, i, blocks, &ptr,
+				nr_this_block);
+
+		ret = xfs_btree_bload_leaf(cur, nr_this_block, bbl->get_record,
+				block, priv);
+		if (ret)
+			goto out;
+
+		/*
+		 * Record the leftmost leaf pointer so we know where to start
+		 * with the first node level.
+		 */
+		if (i == 0)
+			xfs_btree_copy_ptrs(cur, &child_ptr, &ptr, 1);
+	}
+	total_blocks += blocks;
+	xfs_btree_bload_drop_buf(&buffers_list, &bp);
+
+	/* Populate the internal btree nodes. */
+	for (level = 1; level < cur->bc_nlevels; level++) {
+		union xfs_btree_ptr	first_ptr;
+
+		nr_this_level = blocks;
+		block = NULL;
+		xfs_btree_set_ptr_null(cur, &ptr);
+
+		xfs_btree_bload_level_geometry(cur, bbl, level, nr_this_level,
+				&avg_per_block, &blocks, &blocks_with_extra);
+
+		/* Load each node block. */
+		for (i = 0; i < blocks; i++) {
+			unsigned int	nr_this_block = avg_per_block;
+
+			if (i < blocks_with_extra)
+				nr_this_block++;
+
+			ret = xfs_btree_bload_prep_block(cur, bbl,
+					&buffers_list, level, nr_this_block,
+					&ptr, &bp, &block, priv);
+			if (ret)
+				goto out;
+
+			trace_xfs_btree_bload_block(cur, level, i, blocks,
+					&ptr, nr_this_block);
+
+			ret = xfs_btree_bload_node(cur, nr_this_block,
+					&child_ptr, block);
+			if (ret)
+				goto out;
+
+			/*
+			 * Record the leftmost node pointer so that we know
+			 * where to start the next node level above this one.
+			 */
+			if (i == 0)
+				xfs_btree_copy_ptrs(cur, &first_ptr, &ptr, 1);
+		}
+		total_blocks += blocks;
+		xfs_btree_bload_drop_buf(&buffers_list, &bp);
+		xfs_btree_copy_ptrs(cur, &child_ptr, &first_ptr, 1);
+	}
+
+	/* Initialize the new root. */
+	if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) {
+		ASSERT(xfs_btree_ptr_is_null(cur, &ptr));
+		cur->bc_ino.ifake->if_levels = cur->bc_nlevels;
+		cur->bc_ino.ifake->if_blocks = total_blocks - 1;
+	} else {
+		cur->bc_ag.afake->af_root = be32_to_cpu(ptr.s);
+		cur->bc_ag.afake->af_levels = cur->bc_nlevels;
+		cur->bc_ag.afake->af_blocks = total_blocks;
+	}
+
+	/*
+	 * Write the new blocks to disk.  If the ordered list isn't empty after
+	 * that, then something went wrong and we have to fail.  This should
+	 * never happen, but we'll check anyway.
+	 */
+	ret = xfs_buf_delwri_submit(&buffers_list);
+	if (ret)
+		goto out;
+	if (!list_empty(&buffers_list)) {
+		ASSERT(list_empty(&buffers_list));
+		ret = -EIO;
+	}
+
+out:
+	xfs_buf_delwri_cancel(&buffers_list);
+	if (bp)
+		xfs_buf_relse(bp);
+	return ret;
+}
diff --git a/fs/xfs/libxfs/xfs_btree_staging.h b/fs/xfs/libxfs/xfs_btree_staging.h
new file mode 100644
index 000000000..f0d297605
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_btree_staging.h
@@ -0,0 +1,123 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (C) 2020 Oracle.  All Rights Reserved.
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ */
+#ifndef __XFS_BTREE_STAGING_H__
+#define __XFS_BTREE_STAGING_H__
+
+/* Fake root for an AG-rooted btree. */
+struct xbtree_afakeroot {
+	/* AG block number of the new btree root. */
+	xfs_agblock_t		af_root;
+
+	/* Height of the new btree. */
+	unsigned int		af_levels;
+
+	/* Number of blocks used by the btree. */
+	unsigned int		af_blocks;
+};
+
+/* Cursor interactions with fake roots for AG-rooted btrees. */
+void xfs_btree_stage_afakeroot(struct xfs_btree_cur *cur,
+		struct xbtree_afakeroot *afake);
+void xfs_btree_commit_afakeroot(struct xfs_btree_cur *cur, struct xfs_trans *tp,
+		struct xfs_buf *agbp, const struct xfs_btree_ops *ops);
+
+/* Fake root for an inode-rooted btree. */
+struct xbtree_ifakeroot {
+	/* Fake inode fork. */
+	struct xfs_ifork	*if_fork;
+
+	/* Number of blocks used by the btree. */
+	int64_t			if_blocks;
+
+	/* Height of the new btree. */
+	unsigned int		if_levels;
+
+	/* Number of bytes available for this fork in the inode. */
+	unsigned int		if_fork_size;
+
+	/* Fork format. */
+	unsigned int		if_format;
+
+	/* Number of records. */
+	unsigned int		if_extents;
+};
+
+/* Cursor interactions with fake roots for inode-rooted btrees. */
+void xfs_btree_stage_ifakeroot(struct xfs_btree_cur *cur,
+		struct xbtree_ifakeroot *ifake,
+		struct xfs_btree_ops **new_ops);
+void xfs_btree_commit_ifakeroot(struct xfs_btree_cur *cur, struct xfs_trans *tp,
+		int whichfork, const struct xfs_btree_ops *ops);
+
+/* Bulk loading of staged btrees. */
+typedef int (*xfs_btree_bload_get_record_fn)(struct xfs_btree_cur *cur, void *priv);
+typedef int (*xfs_btree_bload_claim_block_fn)(struct xfs_btree_cur *cur,
+		union xfs_btree_ptr *ptr, void *priv);
+typedef size_t (*xfs_btree_bload_iroot_size_fn)(struct xfs_btree_cur *cur,
+		unsigned int nr_this_level, void *priv);
+
+struct xfs_btree_bload {
+	/*
+	 * This function will be called nr_records times to load records into
+	 * the btree.  The function does this by setting the cursor's bc_rec
+	 * field in in-core format.  Records must be returned in sort order.
+	 */
+	xfs_btree_bload_get_record_fn	get_record;
+
+	/*
+	 * This function will be called nr_blocks times to obtain a pointer
+	 * to a new btree block on disk.  Callers must preallocate all space
+	 * for the new btree before calling xfs_btree_bload, and this function
+	 * is what claims that reservation.
+	 */
+	xfs_btree_bload_claim_block_fn	claim_block;
+
+	/*
+	 * This function should return the size of the in-core btree root
+	 * block.  It is only necessary for XFS_BTREE_ROOT_IN_INODE btree
+	 * types.
+	 */
+	xfs_btree_bload_iroot_size_fn	iroot_size;
+
+	/*
+	 * The caller should set this to the number of records that will be
+	 * stored in the new btree.
+	 */
+	uint64_t			nr_records;
+
+	/*
+	 * Number of free records to leave in each leaf block.  If the caller
+	 * sets this to -1, the slack value will be calculated to be halfway
+	 * between maxrecs and minrecs.  This typically leaves the block 75%
+	 * full.  Note that slack values are not enforced on inode root blocks.
+	 */
+	int				leaf_slack;
+
+	/*
+	 * Number of free key/ptrs pairs to leave in each node block.  This
+	 * field has the same semantics as leaf_slack.
+	 */
+	int				node_slack;
+
+	/*
+	 * The xfs_btree_bload_compute_geometry function will set this to the
+	 * number of btree blocks needed to store nr_records records.
+	 */
+	uint64_t			nr_blocks;
+
+	/*
+	 * The xfs_btree_bload_compute_geometry function will set this to the
+	 * height of the new btree.
+	 */
+	unsigned int			btree_height;
+};
+
+int xfs_btree_bload_compute_geometry(struct xfs_btree_cur *cur,
+		struct xfs_btree_bload *bbl, uint64_t nr_records);
+int xfs_btree_bload(struct xfs_btree_cur *cur, struct xfs_btree_bload *bbl,
+		void *priv);
+
+#endif	/* __XFS_BTREE_STAGING_H__ */
diff --git a/fs/xfs/libxfs/xfs_cksum.h b/fs/xfs/libxfs/xfs_cksum.h
new file mode 100644
index 000000000..999a290cf
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_cksum.h
@@ -0,0 +1,82 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _XFS_CKSUM_H
+#define _XFS_CKSUM_H 1
+
+#define XFS_CRC_SEED	(~(uint32_t)0)
+
+/*
+ * Calculate the intermediate checksum for a buffer that has the CRC field
+ * inside it.  The offset of the 32bit crc fields is passed as the
+ * cksum_offset parameter. We do not modify the buffer during verification,
+ * hence we have to split the CRC calculation across the cksum_offset.
+ */
+static inline uint32_t
+xfs_start_cksum_safe(char *buffer, size_t length, unsigned long cksum_offset)
+{
+	uint32_t zero = 0;
+	uint32_t crc;
+
+	/* Calculate CRC up to the checksum. */
+	crc = crc32c(XFS_CRC_SEED, buffer, cksum_offset);
+
+	/* Skip checksum field */
+	crc = crc32c(crc, &zero, sizeof(__u32));
+
+	/* Calculate the rest of the CRC. */
+	return crc32c(crc, &buffer[cksum_offset + sizeof(__be32)],
+		      length - (cksum_offset + sizeof(__be32)));
+}
+
+/*
+ * Fast CRC method where the buffer is modified. Callers must have exclusive
+ * access to the buffer while the calculation takes place.
+ */
+static inline uint32_t
+xfs_start_cksum_update(char *buffer, size_t length, unsigned long cksum_offset)
+{
+	/* zero the CRC field */
+	*(__le32 *)(buffer + cksum_offset) = 0;
+
+	/* single pass CRC calculation for the entire buffer */
+	return crc32c(XFS_CRC_SEED, buffer, length);
+}
+
+/*
+ * Convert the intermediate checksum to the final ondisk format.
+ *
+ * The CRC32c calculation uses LE format even on BE machines, but returns the
+ * result in host endian format. Hence we need to byte swap it back to LE format
+ * so that it is consistent on disk.
+ */
+static inline __le32
+xfs_end_cksum(uint32_t crc)
+{
+	return ~cpu_to_le32(crc);
+}
+
+/*
+ * Helper to generate the checksum for a buffer.
+ *
+ * This modifies the buffer temporarily - callers must have exclusive
+ * access to the buffer while the calculation takes place.
+ */
+static inline void
+xfs_update_cksum(char *buffer, size_t length, unsigned long cksum_offset)
+{
+	uint32_t crc = xfs_start_cksum_update(buffer, length, cksum_offset);
+
+	*(__le32 *)(buffer + cksum_offset) = xfs_end_cksum(crc);
+}
+
+/*
+ * Helper to verify the checksum for a buffer.
+ */
+static inline int
+xfs_verify_cksum(char *buffer, size_t length, unsigned long cksum_offset)
+{
+	uint32_t crc = xfs_start_cksum_safe(buffer, length, cksum_offset);
+
+	return *(__le32 *)(buffer + cksum_offset) == xfs_end_cksum(crc);
+}
+
+#endif /* _XFS_CKSUM_H */
diff --git a/fs/xfs/libxfs/xfs_da_btree.c b/fs/xfs/libxfs/xfs_da_btree.c
new file mode 100644
index 000000000..e576560b4
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_da_btree.c
@@ -0,0 +1,2698 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_bit.h"
+#include "xfs_mount.h"
+#include "xfs_inode.h"
+#include "xfs_dir2.h"
+#include "xfs_dir2_priv.h"
+#include "xfs_trans.h"
+#include "xfs_bmap.h"
+#include "xfs_attr_leaf.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+#include "xfs_buf_item.h"
+#include "xfs_log.h"
+#include "xfs_errortag.h"
+
+/*
+ * xfs_da_btree.c
+ *
+ * Routines to implement directories as Btrees of hashed names.
+ */
+
+/*========================================================================
+ * Function prototypes for the kernel.
+ *========================================================================*/
+
+/*
+ * Routines used for growing the Btree.
+ */
+STATIC int xfs_da3_root_split(xfs_da_state_t *state,
+					    xfs_da_state_blk_t *existing_root,
+					    xfs_da_state_blk_t *new_child);
+STATIC int xfs_da3_node_split(xfs_da_state_t *state,
+					    xfs_da_state_blk_t *existing_blk,
+					    xfs_da_state_blk_t *split_blk,
+					    xfs_da_state_blk_t *blk_to_add,
+					    int treelevel,
+					    int *result);
+STATIC void xfs_da3_node_rebalance(xfs_da_state_t *state,
+					 xfs_da_state_blk_t *node_blk_1,
+					 xfs_da_state_blk_t *node_blk_2);
+STATIC void xfs_da3_node_add(xfs_da_state_t *state,
+				   xfs_da_state_blk_t *old_node_blk,
+				   xfs_da_state_blk_t *new_node_blk);
+
+/*
+ * Routines used for shrinking the Btree.
+ */
+STATIC int xfs_da3_root_join(xfs_da_state_t *state,
+					   xfs_da_state_blk_t *root_blk);
+STATIC int xfs_da3_node_toosmall(xfs_da_state_t *state, int *retval);
+STATIC void xfs_da3_node_remove(xfs_da_state_t *state,
+					      xfs_da_state_blk_t *drop_blk);
+STATIC void xfs_da3_node_unbalance(xfs_da_state_t *state,
+					 xfs_da_state_blk_t *src_node_blk,
+					 xfs_da_state_blk_t *dst_node_blk);
+
+/*
+ * Utility routines.
+ */
+STATIC int	xfs_da3_blk_unlink(xfs_da_state_t *state,
+				  xfs_da_state_blk_t *drop_blk,
+				  xfs_da_state_blk_t *save_blk);
+
+
+struct kmem_cache	*xfs_da_state_cache;	/* anchor for dir/attr state */
+
+/*
+ * Allocate a dir-state structure.
+ * We don't put them on the stack since they're large.
+ */
+struct xfs_da_state *
+xfs_da_state_alloc(
+	struct xfs_da_args	*args)
+{
+	struct xfs_da_state	*state;
+
+	state = kmem_cache_zalloc(xfs_da_state_cache, GFP_NOFS | __GFP_NOFAIL);
+	state->args = args;
+	state->mp = args->dp->i_mount;
+	return state;
+}
+
+/*
+ * Kill the altpath contents of a da-state structure.
+ */
+STATIC void
+xfs_da_state_kill_altpath(xfs_da_state_t *state)
+{
+	int	i;
+
+	for (i = 0; i < state->altpath.active; i++)
+		state->altpath.blk[i].bp = NULL;
+	state->altpath.active = 0;
+}
+
+/*
+ * Free a da-state structure.
+ */
+void
+xfs_da_state_free(xfs_da_state_t *state)
+{
+	xfs_da_state_kill_altpath(state);
+#ifdef DEBUG
+	memset((char *)state, 0, sizeof(*state));
+#endif /* DEBUG */
+	kmem_cache_free(xfs_da_state_cache, state);
+}
+
+void
+xfs_da_state_reset(
+	struct xfs_da_state	*state,
+	struct xfs_da_args	*args)
+{
+	xfs_da_state_kill_altpath(state);
+	memset(state, 0, sizeof(struct xfs_da_state));
+	state->args = args;
+	state->mp = state->args->dp->i_mount;
+}
+
+static inline int xfs_dabuf_nfsb(struct xfs_mount *mp, int whichfork)
+{
+	if (whichfork == XFS_DATA_FORK)
+		return mp->m_dir_geo->fsbcount;
+	return mp->m_attr_geo->fsbcount;
+}
+
+void
+xfs_da3_node_hdr_from_disk(
+	struct xfs_mount		*mp,
+	struct xfs_da3_icnode_hdr	*to,
+	struct xfs_da_intnode		*from)
+{
+	if (xfs_has_crc(mp)) {
+		struct xfs_da3_intnode	*from3 = (struct xfs_da3_intnode *)from;
+
+		to->forw = be32_to_cpu(from3->hdr.info.hdr.forw);
+		to->back = be32_to_cpu(from3->hdr.info.hdr.back);
+		to->magic = be16_to_cpu(from3->hdr.info.hdr.magic);
+		to->count = be16_to_cpu(from3->hdr.__count);
+		to->level = be16_to_cpu(from3->hdr.__level);
+		to->btree = from3->__btree;
+		ASSERT(to->magic == XFS_DA3_NODE_MAGIC);
+	} else {
+		to->forw = be32_to_cpu(from->hdr.info.forw);
+		to->back = be32_to_cpu(from->hdr.info.back);
+		to->magic = be16_to_cpu(from->hdr.info.magic);
+		to->count = be16_to_cpu(from->hdr.__count);
+		to->level = be16_to_cpu(from->hdr.__level);
+		to->btree = from->__btree;
+		ASSERT(to->magic == XFS_DA_NODE_MAGIC);
+	}
+}
+
+void
+xfs_da3_node_hdr_to_disk(
+	struct xfs_mount		*mp,
+	struct xfs_da_intnode		*to,
+	struct xfs_da3_icnode_hdr	*from)
+{
+	if (xfs_has_crc(mp)) {
+		struct xfs_da3_intnode	*to3 = (struct xfs_da3_intnode *)to;
+
+		ASSERT(from->magic == XFS_DA3_NODE_MAGIC);
+		to3->hdr.info.hdr.forw = cpu_to_be32(from->forw);
+		to3->hdr.info.hdr.back = cpu_to_be32(from->back);
+		to3->hdr.info.hdr.magic = cpu_to_be16(from->magic);
+		to3->hdr.__count = cpu_to_be16(from->count);
+		to3->hdr.__level = cpu_to_be16(from->level);
+	} else {
+		ASSERT(from->magic == XFS_DA_NODE_MAGIC);
+		to->hdr.info.forw = cpu_to_be32(from->forw);
+		to->hdr.info.back = cpu_to_be32(from->back);
+		to->hdr.info.magic = cpu_to_be16(from->magic);
+		to->hdr.__count = cpu_to_be16(from->count);
+		to->hdr.__level = cpu_to_be16(from->level);
+	}
+}
+
+/*
+ * Verify an xfs_da3_blkinfo structure. Note that the da3 fields are only
+ * accessible on v5 filesystems. This header format is common across da node,
+ * attr leaf and dir leaf blocks.
+ */
+xfs_failaddr_t
+xfs_da3_blkinfo_verify(
+	struct xfs_buf		*bp,
+	struct xfs_da3_blkinfo	*hdr3)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_da_blkinfo	*hdr = &hdr3->hdr;
+
+	if (!xfs_verify_magic16(bp, hdr->magic))
+		return __this_address;
+
+	if (xfs_has_crc(mp)) {
+		if (!uuid_equal(&hdr3->uuid, &mp->m_sb.sb_meta_uuid))
+			return __this_address;
+		if (be64_to_cpu(hdr3->blkno) != xfs_buf_daddr(bp))
+			return __this_address;
+		if (!xfs_log_check_lsn(mp, be64_to_cpu(hdr3->lsn)))
+			return __this_address;
+	}
+
+	return NULL;
+}
+
+static xfs_failaddr_t
+xfs_da3_node_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_da_intnode	*hdr = bp->b_addr;
+	struct xfs_da3_icnode_hdr ichdr;
+	xfs_failaddr_t		fa;
+
+	xfs_da3_node_hdr_from_disk(mp, &ichdr, hdr);
+
+	fa = xfs_da3_blkinfo_verify(bp, bp->b_addr);
+	if (fa)
+		return fa;
+
+	if (ichdr.level == 0)
+		return __this_address;
+	if (ichdr.level > XFS_DA_NODE_MAXDEPTH)
+		return __this_address;
+	if (ichdr.count == 0)
+		return __this_address;
+
+	/*
+	 * we don't know if the node is for and attribute or directory tree,
+	 * so only fail if the count is outside both bounds
+	 */
+	if (ichdr.count > mp->m_dir_geo->node_ents &&
+	    ichdr.count > mp->m_attr_geo->node_ents)
+		return __this_address;
+
+	/* XXX: hash order check? */
+
+	return NULL;
+}
+
+static void
+xfs_da3_node_write_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_buf_log_item	*bip = bp->b_log_item;
+	struct xfs_da3_node_hdr *hdr3 = bp->b_addr;
+	xfs_failaddr_t		fa;
+
+	fa = xfs_da3_node_verify(bp);
+	if (fa) {
+		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+		return;
+	}
+
+	if (!xfs_has_crc(mp))
+		return;
+
+	if (bip)
+		hdr3->info.lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+	xfs_buf_update_cksum(bp, XFS_DA3_NODE_CRC_OFF);
+}
+
+/*
+ * leaf/node format detection on trees is sketchy, so a node read can be done on
+ * leaf level blocks when detection identifies the tree as a node format tree
+ * incorrectly. In this case, we need to swap the verifier to match the correct
+ * format of the block being read.
+ */
+static void
+xfs_da3_node_read_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_da_blkinfo	*info = bp->b_addr;
+	xfs_failaddr_t		fa;
+
+	switch (be16_to_cpu(info->magic)) {
+		case XFS_DA3_NODE_MAGIC:
+			if (!xfs_buf_verify_cksum(bp, XFS_DA3_NODE_CRC_OFF)) {
+				xfs_verifier_error(bp, -EFSBADCRC,
+						__this_address);
+				break;
+			}
+			fallthrough;
+		case XFS_DA_NODE_MAGIC:
+			fa = xfs_da3_node_verify(bp);
+			if (fa)
+				xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+			return;
+		case XFS_ATTR_LEAF_MAGIC:
+		case XFS_ATTR3_LEAF_MAGIC:
+			bp->b_ops = &xfs_attr3_leaf_buf_ops;
+			bp->b_ops->verify_read(bp);
+			return;
+		case XFS_DIR2_LEAFN_MAGIC:
+		case XFS_DIR3_LEAFN_MAGIC:
+			bp->b_ops = &xfs_dir3_leafn_buf_ops;
+			bp->b_ops->verify_read(bp);
+			return;
+		default:
+			xfs_verifier_error(bp, -EFSCORRUPTED, __this_address);
+			break;
+	}
+}
+
+/* Verify the structure of a da3 block. */
+static xfs_failaddr_t
+xfs_da3_node_verify_struct(
+	struct xfs_buf		*bp)
+{
+	struct xfs_da_blkinfo	*info = bp->b_addr;
+
+	switch (be16_to_cpu(info->magic)) {
+	case XFS_DA3_NODE_MAGIC:
+	case XFS_DA_NODE_MAGIC:
+		return xfs_da3_node_verify(bp);
+	case XFS_ATTR_LEAF_MAGIC:
+	case XFS_ATTR3_LEAF_MAGIC:
+		bp->b_ops = &xfs_attr3_leaf_buf_ops;
+		return bp->b_ops->verify_struct(bp);
+	case XFS_DIR2_LEAFN_MAGIC:
+	case XFS_DIR3_LEAFN_MAGIC:
+		bp->b_ops = &xfs_dir3_leafn_buf_ops;
+		return bp->b_ops->verify_struct(bp);
+	default:
+		return __this_address;
+	}
+}
+
+const struct xfs_buf_ops xfs_da3_node_buf_ops = {
+	.name = "xfs_da3_node",
+	.magic16 = { cpu_to_be16(XFS_DA_NODE_MAGIC),
+		     cpu_to_be16(XFS_DA3_NODE_MAGIC) },
+	.verify_read = xfs_da3_node_read_verify,
+	.verify_write = xfs_da3_node_write_verify,
+	.verify_struct = xfs_da3_node_verify_struct,
+};
+
+static int
+xfs_da3_node_set_type(
+	struct xfs_trans	*tp,
+	struct xfs_buf		*bp)
+{
+	struct xfs_da_blkinfo	*info = bp->b_addr;
+
+	switch (be16_to_cpu(info->magic)) {
+	case XFS_DA_NODE_MAGIC:
+	case XFS_DA3_NODE_MAGIC:
+		xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DA_NODE_BUF);
+		return 0;
+	case XFS_ATTR_LEAF_MAGIC:
+	case XFS_ATTR3_LEAF_MAGIC:
+		xfs_trans_buf_set_type(tp, bp, XFS_BLFT_ATTR_LEAF_BUF);
+		return 0;
+	case XFS_DIR2_LEAFN_MAGIC:
+	case XFS_DIR3_LEAFN_MAGIC:
+		xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DIR_LEAFN_BUF);
+		return 0;
+	default:
+		XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, tp->t_mountp,
+				info, sizeof(*info));
+		xfs_trans_brelse(tp, bp);
+		return -EFSCORRUPTED;
+	}
+}
+
+int
+xfs_da3_node_read(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*dp,
+	xfs_dablk_t		bno,
+	struct xfs_buf		**bpp,
+	int			whichfork)
+{
+	int			error;
+
+	error = xfs_da_read_buf(tp, dp, bno, 0, bpp, whichfork,
+			&xfs_da3_node_buf_ops);
+	if (error || !*bpp || !tp)
+		return error;
+	return xfs_da3_node_set_type(tp, *bpp);
+}
+
+int
+xfs_da3_node_read_mapped(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*dp,
+	xfs_daddr_t		mappedbno,
+	struct xfs_buf		**bpp,
+	int			whichfork)
+{
+	struct xfs_mount	*mp = dp->i_mount;
+	int			error;
+
+	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, mappedbno,
+			XFS_FSB_TO_BB(mp, xfs_dabuf_nfsb(mp, whichfork)), 0,
+			bpp, &xfs_da3_node_buf_ops);
+	if (error || !*bpp)
+		return error;
+
+	if (whichfork == XFS_ATTR_FORK)
+		xfs_buf_set_ref(*bpp, XFS_ATTR_BTREE_REF);
+	else
+		xfs_buf_set_ref(*bpp, XFS_DIR_BTREE_REF);
+
+	if (!tp)
+		return 0;
+	return xfs_da3_node_set_type(tp, *bpp);
+}
+
+/*========================================================================
+ * Routines used for growing the Btree.
+ *========================================================================*/
+
+/*
+ * Create the initial contents of an intermediate node.
+ */
+int
+xfs_da3_node_create(
+	struct xfs_da_args	*args,
+	xfs_dablk_t		blkno,
+	int			level,
+	struct xfs_buf		**bpp,
+	int			whichfork)
+{
+	struct xfs_da_intnode	*node;
+	struct xfs_trans	*tp = args->trans;
+	struct xfs_mount	*mp = tp->t_mountp;
+	struct xfs_da3_icnode_hdr ichdr = {0};
+	struct xfs_buf		*bp;
+	int			error;
+	struct xfs_inode	*dp = args->dp;
+
+	trace_xfs_da_node_create(args);
+	ASSERT(level <= XFS_DA_NODE_MAXDEPTH);
+
+	error = xfs_da_get_buf(tp, dp, blkno, &bp, whichfork);
+	if (error)
+		return error;
+	bp->b_ops = &xfs_da3_node_buf_ops;
+	xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DA_NODE_BUF);
+	node = bp->b_addr;
+
+	if (xfs_has_crc(mp)) {
+		struct xfs_da3_node_hdr *hdr3 = bp->b_addr;
+
+		memset(hdr3, 0, sizeof(struct xfs_da3_node_hdr));
+		ichdr.magic = XFS_DA3_NODE_MAGIC;
+		hdr3->info.blkno = cpu_to_be64(xfs_buf_daddr(bp));
+		hdr3->info.owner = cpu_to_be64(args->dp->i_ino);
+		uuid_copy(&hdr3->info.uuid, &mp->m_sb.sb_meta_uuid);
+	} else {
+		ichdr.magic = XFS_DA_NODE_MAGIC;
+	}
+	ichdr.level = level;
+
+	xfs_da3_node_hdr_to_disk(dp->i_mount, node, &ichdr);
+	xfs_trans_log_buf(tp, bp,
+		XFS_DA_LOGRANGE(node, &node->hdr, args->geo->node_hdr_size));
+
+	*bpp = bp;
+	return 0;
+}
+
+/*
+ * Split a leaf node, rebalance, then possibly split
+ * intermediate nodes, rebalance, etc.
+ */
+int							/* error */
+xfs_da3_split(
+	struct xfs_da_state	*state)
+{
+	struct xfs_da_state_blk	*oldblk;
+	struct xfs_da_state_blk	*newblk;
+	struct xfs_da_state_blk	*addblk;
+	struct xfs_da_intnode	*node;
+	int			max;
+	int			action = 0;
+	int			error;
+	int			i;
+
+	trace_xfs_da_split(state->args);
+
+	if (XFS_TEST_ERROR(false, state->mp, XFS_ERRTAG_DA_LEAF_SPLIT))
+		return -EIO;
+
+	/*
+	 * Walk back up the tree splitting/inserting/adjusting as necessary.
+	 * If we need to insert and there isn't room, split the node, then
+	 * decide which fragment to insert the new block from below into.
+	 * Note that we may split the root this way, but we need more fixup.
+	 */
+	max = state->path.active - 1;
+	ASSERT((max >= 0) && (max < XFS_DA_NODE_MAXDEPTH));
+	ASSERT(state->path.blk[max].magic == XFS_ATTR_LEAF_MAGIC ||
+	       state->path.blk[max].magic == XFS_DIR2_LEAFN_MAGIC);
+
+	addblk = &state->path.blk[max];		/* initial dummy value */
+	for (i = max; (i >= 0) && addblk; state->path.active--, i--) {
+		oldblk = &state->path.blk[i];
+		newblk = &state->altpath.blk[i];
+
+		/*
+		 * If a leaf node then
+		 *     Allocate a new leaf node, then rebalance across them.
+		 * else if an intermediate node then
+		 *     We split on the last layer, must we split the node?
+		 */
+		switch (oldblk->magic) {
+		case XFS_ATTR_LEAF_MAGIC:
+			error = xfs_attr3_leaf_split(state, oldblk, newblk);
+			if ((error != 0) && (error != -ENOSPC)) {
+				return error;	/* GROT: attr is inconsistent */
+			}
+			if (!error) {
+				addblk = newblk;
+				break;
+			}
+			/*
+			 * Entry wouldn't fit, split the leaf again. The new
+			 * extrablk will be consumed by xfs_da3_node_split if
+			 * the node is split.
+			 */
+			state->extravalid = 1;
+			if (state->inleaf) {
+				state->extraafter = 0;	/* before newblk */
+				trace_xfs_attr_leaf_split_before(state->args);
+				error = xfs_attr3_leaf_split(state, oldblk,
+							    &state->extrablk);
+			} else {
+				state->extraafter = 1;	/* after newblk */
+				trace_xfs_attr_leaf_split_after(state->args);
+				error = xfs_attr3_leaf_split(state, newblk,
+							    &state->extrablk);
+			}
+			if (error)
+				return error;	/* GROT: attr inconsistent */
+			addblk = newblk;
+			break;
+		case XFS_DIR2_LEAFN_MAGIC:
+			error = xfs_dir2_leafn_split(state, oldblk, newblk);
+			if (error)
+				return error;
+			addblk = newblk;
+			break;
+		case XFS_DA_NODE_MAGIC:
+			error = xfs_da3_node_split(state, oldblk, newblk, addblk,
+							 max - i, &action);
+			addblk->bp = NULL;
+			if (error)
+				return error;	/* GROT: dir is inconsistent */
+			/*
+			 * Record the newly split block for the next time thru?
+			 */
+			if (action)
+				addblk = newblk;
+			else
+				addblk = NULL;
+			break;
+		}
+
+		/*
+		 * Update the btree to show the new hashval for this child.
+		 */
+		xfs_da3_fixhashpath(state, &state->path);
+	}
+	if (!addblk)
+		return 0;
+
+	/*
+	 * xfs_da3_node_split() should have consumed any extra blocks we added
+	 * during a double leaf split in the attr fork. This is guaranteed as
+	 * we can't be here if the attr fork only has a single leaf block.
+	 */
+	ASSERT(state->extravalid == 0 ||
+	       state->path.blk[max].magic == XFS_DIR2_LEAFN_MAGIC);
+
+	/*
+	 * Split the root node.
+	 */
+	ASSERT(state->path.active == 0);
+	oldblk = &state->path.blk[0];
+	error = xfs_da3_root_split(state, oldblk, addblk);
+	if (error)
+		goto out;
+
+	/*
+	 * Update pointers to the node which used to be block 0 and just got
+	 * bumped because of the addition of a new root node.  Note that the
+	 * original block 0 could be at any position in the list of blocks in
+	 * the tree.
+	 *
+	 * Note: the magic numbers and sibling pointers are in the same physical
+	 * place for both v2 and v3 headers (by design). Hence it doesn't matter
+	 * which version of the xfs_da_intnode structure we use here as the
+	 * result will be the same using either structure.
+	 */
+	node = oldblk->bp->b_addr;
+	if (node->hdr.info.forw) {
+		if (be32_to_cpu(node->hdr.info.forw) != addblk->blkno) {
+			xfs_buf_mark_corrupt(oldblk->bp);
+			error = -EFSCORRUPTED;
+			goto out;
+		}
+		node = addblk->bp->b_addr;
+		node->hdr.info.back = cpu_to_be32(oldblk->blkno);
+		xfs_trans_log_buf(state->args->trans, addblk->bp,
+				  XFS_DA_LOGRANGE(node, &node->hdr.info,
+				  sizeof(node->hdr.info)));
+	}
+	node = oldblk->bp->b_addr;
+	if (node->hdr.info.back) {
+		if (be32_to_cpu(node->hdr.info.back) != addblk->blkno) {
+			xfs_buf_mark_corrupt(oldblk->bp);
+			error = -EFSCORRUPTED;
+			goto out;
+		}
+		node = addblk->bp->b_addr;
+		node->hdr.info.forw = cpu_to_be32(oldblk->blkno);
+		xfs_trans_log_buf(state->args->trans, addblk->bp,
+				  XFS_DA_LOGRANGE(node, &node->hdr.info,
+				  sizeof(node->hdr.info)));
+	}
+out:
+	addblk->bp = NULL;
+	return error;
+}
+
+/*
+ * Split the root.  We have to create a new root and point to the two
+ * parts (the split old root) that we just created.  Copy block zero to
+ * the EOF, extending the inode in process.
+ */
+STATIC int						/* error */
+xfs_da3_root_split(
+	struct xfs_da_state	*state,
+	struct xfs_da_state_blk	*blk1,
+	struct xfs_da_state_blk	*blk2)
+{
+	struct xfs_da_intnode	*node;
+	struct xfs_da_intnode	*oldroot;
+	struct xfs_da_node_entry *btree;
+	struct xfs_da3_icnode_hdr nodehdr;
+	struct xfs_da_args	*args;
+	struct xfs_buf		*bp;
+	struct xfs_inode	*dp;
+	struct xfs_trans	*tp;
+	struct xfs_dir2_leaf	*leaf;
+	xfs_dablk_t		blkno;
+	int			level;
+	int			error;
+	int			size;
+
+	trace_xfs_da_root_split(state->args);
+
+	/*
+	 * Copy the existing (incorrect) block from the root node position
+	 * to a free space somewhere.
+	 */
+	args = state->args;
+	error = xfs_da_grow_inode(args, &blkno);
+	if (error)
+		return error;
+
+	dp = args->dp;
+	tp = args->trans;
+	error = xfs_da_get_buf(tp, dp, blkno, &bp, args->whichfork);
+	if (error)
+		return error;
+	node = bp->b_addr;
+	oldroot = blk1->bp->b_addr;
+	if (oldroot->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC) ||
+	    oldroot->hdr.info.magic == cpu_to_be16(XFS_DA3_NODE_MAGIC)) {
+		struct xfs_da3_icnode_hdr icnodehdr;
+
+		xfs_da3_node_hdr_from_disk(dp->i_mount, &icnodehdr, oldroot);
+		btree = icnodehdr.btree;
+		size = (int)((char *)&btree[icnodehdr.count] - (char *)oldroot);
+		level = icnodehdr.level;
+
+		/*
+		 * we are about to copy oldroot to bp, so set up the type
+		 * of bp while we know exactly what it will be.
+		 */
+		xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DA_NODE_BUF);
+	} else {
+		struct xfs_dir3_icleaf_hdr leafhdr;
+
+		leaf = (xfs_dir2_leaf_t *)oldroot;
+		xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &leafhdr, leaf);
+
+		ASSERT(leafhdr.magic == XFS_DIR2_LEAFN_MAGIC ||
+		       leafhdr.magic == XFS_DIR3_LEAFN_MAGIC);
+		size = (int)((char *)&leafhdr.ents[leafhdr.count] -
+			(char *)leaf);
+		level = 0;
+
+		/*
+		 * we are about to copy oldroot to bp, so set up the type
+		 * of bp while we know exactly what it will be.
+		 */
+		xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DIR_LEAFN_BUF);
+	}
+
+	/*
+	 * we can copy most of the information in the node from one block to
+	 * another, but for CRC enabled headers we have to make sure that the
+	 * block specific identifiers are kept intact. We update the buffer
+	 * directly for this.
+	 */
+	memcpy(node, oldroot, size);
+	if (oldroot->hdr.info.magic == cpu_to_be16(XFS_DA3_NODE_MAGIC) ||
+	    oldroot->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC)) {
+		struct xfs_da3_intnode *node3 = (struct xfs_da3_intnode *)node;
+
+		node3->hdr.info.blkno = cpu_to_be64(xfs_buf_daddr(bp));
+	}
+	xfs_trans_log_buf(tp, bp, 0, size - 1);
+
+	bp->b_ops = blk1->bp->b_ops;
+	xfs_trans_buf_copy_type(bp, blk1->bp);
+	blk1->bp = bp;
+	blk1->blkno = blkno;
+
+	/*
+	 * Set up the new root node.
+	 */
+	error = xfs_da3_node_create(args,
+		(args->whichfork == XFS_DATA_FORK) ? args->geo->leafblk : 0,
+		level + 1, &bp, args->whichfork);
+	if (error)
+		return error;
+
+	node = bp->b_addr;
+	xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, node);
+	btree = nodehdr.btree;
+	btree[0].hashval = cpu_to_be32(blk1->hashval);
+	btree[0].before = cpu_to_be32(blk1->blkno);
+	btree[1].hashval = cpu_to_be32(blk2->hashval);
+	btree[1].before = cpu_to_be32(blk2->blkno);
+	nodehdr.count = 2;
+	xfs_da3_node_hdr_to_disk(dp->i_mount, node, &nodehdr);
+
+#ifdef DEBUG
+	if (oldroot->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
+	    oldroot->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC)) {
+		ASSERT(blk1->blkno >= args->geo->leafblk &&
+		       blk1->blkno < args->geo->freeblk);
+		ASSERT(blk2->blkno >= args->geo->leafblk &&
+		       blk2->blkno < args->geo->freeblk);
+	}
+#endif
+
+	/* Header is already logged by xfs_da_node_create */
+	xfs_trans_log_buf(tp, bp,
+		XFS_DA_LOGRANGE(node, btree, sizeof(xfs_da_node_entry_t) * 2));
+
+	return 0;
+}
+
+/*
+ * Split the node, rebalance, then add the new entry.
+ */
+STATIC int						/* error */
+xfs_da3_node_split(
+	struct xfs_da_state	*state,
+	struct xfs_da_state_blk	*oldblk,
+	struct xfs_da_state_blk	*newblk,
+	struct xfs_da_state_blk	*addblk,
+	int			treelevel,
+	int			*result)
+{
+	struct xfs_da_intnode	*node;
+	struct xfs_da3_icnode_hdr nodehdr;
+	xfs_dablk_t		blkno;
+	int			newcount;
+	int			error;
+	int			useextra;
+	struct xfs_inode	*dp = state->args->dp;
+
+	trace_xfs_da_node_split(state->args);
+
+	node = oldblk->bp->b_addr;
+	xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, node);
+
+	/*
+	 * With V2 dirs the extra block is data or freespace.
+	 */
+	useextra = state->extravalid && state->args->whichfork == XFS_ATTR_FORK;
+	newcount = 1 + useextra;
+	/*
+	 * Do we have to split the node?
+	 */
+	if (nodehdr.count + newcount > state->args->geo->node_ents) {
+		/*
+		 * Allocate a new node, add to the doubly linked chain of
+		 * nodes, then move some of our excess entries into it.
+		 */
+		error = xfs_da_grow_inode(state->args, &blkno);
+		if (error)
+			return error;	/* GROT: dir is inconsistent */
+
+		error = xfs_da3_node_create(state->args, blkno, treelevel,
+					   &newblk->bp, state->args->whichfork);
+		if (error)
+			return error;	/* GROT: dir is inconsistent */
+		newblk->blkno = blkno;
+		newblk->magic = XFS_DA_NODE_MAGIC;
+		xfs_da3_node_rebalance(state, oldblk, newblk);
+		error = xfs_da3_blk_link(state, oldblk, newblk);
+		if (error)
+			return error;
+		*result = 1;
+	} else {
+		*result = 0;
+	}
+
+	/*
+	 * Insert the new entry(s) into the correct block
+	 * (updating last hashval in the process).
+	 *
+	 * xfs_da3_node_add() inserts BEFORE the given index,
+	 * and as a result of using node_lookup_int() we always
+	 * point to a valid entry (not after one), but a split
+	 * operation always results in a new block whose hashvals
+	 * FOLLOW the current block.
+	 *
+	 * If we had double-split op below us, then add the extra block too.
+	 */
+	node = oldblk->bp->b_addr;
+	xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, node);
+	if (oldblk->index <= nodehdr.count) {
+		oldblk->index++;
+		xfs_da3_node_add(state, oldblk, addblk);
+		if (useextra) {
+			if (state->extraafter)
+				oldblk->index++;
+			xfs_da3_node_add(state, oldblk, &state->extrablk);
+			state->extravalid = 0;
+		}
+	} else {
+		newblk->index++;
+		xfs_da3_node_add(state, newblk, addblk);
+		if (useextra) {
+			if (state->extraafter)
+				newblk->index++;
+			xfs_da3_node_add(state, newblk, &state->extrablk);
+			state->extravalid = 0;
+		}
+	}
+
+	return 0;
+}
+
+/*
+ * Balance the btree elements between two intermediate nodes,
+ * usually one full and one empty.
+ *
+ * NOTE: if blk2 is empty, then it will get the upper half of blk1.
+ */
+STATIC void
+xfs_da3_node_rebalance(
+	struct xfs_da_state	*state,
+	struct xfs_da_state_blk	*blk1,
+	struct xfs_da_state_blk	*blk2)
+{
+	struct xfs_da_intnode	*node1;
+	struct xfs_da_intnode	*node2;
+	struct xfs_da_node_entry *btree1;
+	struct xfs_da_node_entry *btree2;
+	struct xfs_da_node_entry *btree_s;
+	struct xfs_da_node_entry *btree_d;
+	struct xfs_da3_icnode_hdr nodehdr1;
+	struct xfs_da3_icnode_hdr nodehdr2;
+	struct xfs_trans	*tp;
+	int			count;
+	int			tmp;
+	int			swap = 0;
+	struct xfs_inode	*dp = state->args->dp;
+
+	trace_xfs_da_node_rebalance(state->args);
+
+	node1 = blk1->bp->b_addr;
+	node2 = blk2->bp->b_addr;
+	xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr1, node1);
+	xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr2, node2);
+	btree1 = nodehdr1.btree;
+	btree2 = nodehdr2.btree;
+
+	/*
+	 * Figure out how many entries need to move, and in which direction.
+	 * Swap the nodes around if that makes it simpler.
+	 */
+	if (nodehdr1.count > 0 && nodehdr2.count > 0 &&
+	    ((be32_to_cpu(btree2[0].hashval) < be32_to_cpu(btree1[0].hashval)) ||
+	     (be32_to_cpu(btree2[nodehdr2.count - 1].hashval) <
+			be32_to_cpu(btree1[nodehdr1.count - 1].hashval)))) {
+		swap(node1, node2);
+		xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr1, node1);
+		xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr2, node2);
+		btree1 = nodehdr1.btree;
+		btree2 = nodehdr2.btree;
+		swap = 1;
+	}
+
+	count = (nodehdr1.count - nodehdr2.count) / 2;
+	if (count == 0)
+		return;
+	tp = state->args->trans;
+	/*
+	 * Two cases: high-to-low and low-to-high.
+	 */
+	if (count > 0) {
+		/*
+		 * Move elements in node2 up to make a hole.
+		 */
+		tmp = nodehdr2.count;
+		if (tmp > 0) {
+			tmp *= (uint)sizeof(xfs_da_node_entry_t);
+			btree_s = &btree2[0];
+			btree_d = &btree2[count];
+			memmove(btree_d, btree_s, tmp);
+		}
+
+		/*
+		 * Move the req'd B-tree elements from high in node1 to
+		 * low in node2.
+		 */
+		nodehdr2.count += count;
+		tmp = count * (uint)sizeof(xfs_da_node_entry_t);
+		btree_s = &btree1[nodehdr1.count - count];
+		btree_d = &btree2[0];
+		memcpy(btree_d, btree_s, tmp);
+		nodehdr1.count -= count;
+	} else {
+		/*
+		 * Move the req'd B-tree elements from low in node2 to
+		 * high in node1.
+		 */
+		count = -count;
+		tmp = count * (uint)sizeof(xfs_da_node_entry_t);
+		btree_s = &btree2[0];
+		btree_d = &btree1[nodehdr1.count];
+		memcpy(btree_d, btree_s, tmp);
+		nodehdr1.count += count;
+
+		xfs_trans_log_buf(tp, blk1->bp,
+			XFS_DA_LOGRANGE(node1, btree_d, tmp));
+
+		/*
+		 * Move elements in node2 down to fill the hole.
+		 */
+		tmp  = nodehdr2.count - count;
+		tmp *= (uint)sizeof(xfs_da_node_entry_t);
+		btree_s = &btree2[count];
+		btree_d = &btree2[0];
+		memmove(btree_d, btree_s, tmp);
+		nodehdr2.count -= count;
+	}
+
+	/*
+	 * Log header of node 1 and all current bits of node 2.
+	 */
+	xfs_da3_node_hdr_to_disk(dp->i_mount, node1, &nodehdr1);
+	xfs_trans_log_buf(tp, blk1->bp,
+		XFS_DA_LOGRANGE(node1, &node1->hdr,
+				state->args->geo->node_hdr_size));
+
+	xfs_da3_node_hdr_to_disk(dp->i_mount, node2, &nodehdr2);
+	xfs_trans_log_buf(tp, blk2->bp,
+		XFS_DA_LOGRANGE(node2, &node2->hdr,
+				state->args->geo->node_hdr_size +
+				(sizeof(btree2[0]) * nodehdr2.count)));
+
+	/*
+	 * Record the last hashval from each block for upward propagation.
+	 * (note: don't use the swapped node pointers)
+	 */
+	if (swap) {
+		node1 = blk1->bp->b_addr;
+		node2 = blk2->bp->b_addr;
+		xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr1, node1);
+		xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr2, node2);
+		btree1 = nodehdr1.btree;
+		btree2 = nodehdr2.btree;
+	}
+	blk1->hashval = be32_to_cpu(btree1[nodehdr1.count - 1].hashval);
+	blk2->hashval = be32_to_cpu(btree2[nodehdr2.count - 1].hashval);
+
+	/*
+	 * Adjust the expected index for insertion.
+	 */
+	if (blk1->index >= nodehdr1.count) {
+		blk2->index = blk1->index - nodehdr1.count;
+		blk1->index = nodehdr1.count + 1;	/* make it invalid */
+	}
+}
+
+/*
+ * Add a new entry to an intermediate node.
+ */
+STATIC void
+xfs_da3_node_add(
+	struct xfs_da_state	*state,
+	struct xfs_da_state_blk	*oldblk,
+	struct xfs_da_state_blk	*newblk)
+{
+	struct xfs_da_intnode	*node;
+	struct xfs_da3_icnode_hdr nodehdr;
+	struct xfs_da_node_entry *btree;
+	int			tmp;
+	struct xfs_inode	*dp = state->args->dp;
+
+	trace_xfs_da_node_add(state->args);
+
+	node = oldblk->bp->b_addr;
+	xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, node);
+	btree = nodehdr.btree;
+
+	ASSERT(oldblk->index >= 0 && oldblk->index <= nodehdr.count);
+	ASSERT(newblk->blkno != 0);
+	if (state->args->whichfork == XFS_DATA_FORK)
+		ASSERT(newblk->blkno >= state->args->geo->leafblk &&
+		       newblk->blkno < state->args->geo->freeblk);
+
+	/*
+	 * We may need to make some room before we insert the new node.
+	 */
+	tmp = 0;
+	if (oldblk->index < nodehdr.count) {
+		tmp = (nodehdr.count - oldblk->index) * (uint)sizeof(*btree);
+		memmove(&btree[oldblk->index + 1], &btree[oldblk->index], tmp);
+	}
+	btree[oldblk->index].hashval = cpu_to_be32(newblk->hashval);
+	btree[oldblk->index].before = cpu_to_be32(newblk->blkno);
+	xfs_trans_log_buf(state->args->trans, oldblk->bp,
+		XFS_DA_LOGRANGE(node, &btree[oldblk->index],
+				tmp + sizeof(*btree)));
+
+	nodehdr.count += 1;
+	xfs_da3_node_hdr_to_disk(dp->i_mount, node, &nodehdr);
+	xfs_trans_log_buf(state->args->trans, oldblk->bp,
+		XFS_DA_LOGRANGE(node, &node->hdr,
+				state->args->geo->node_hdr_size));
+
+	/*
+	 * Copy the last hash value from the oldblk to propagate upwards.
+	 */
+	oldblk->hashval = be32_to_cpu(btree[nodehdr.count - 1].hashval);
+}
+
+/*========================================================================
+ * Routines used for shrinking the Btree.
+ *========================================================================*/
+
+/*
+ * Deallocate an empty leaf node, remove it from its parent,
+ * possibly deallocating that block, etc...
+ */
+int
+xfs_da3_join(
+	struct xfs_da_state	*state)
+{
+	struct xfs_da_state_blk	*drop_blk;
+	struct xfs_da_state_blk	*save_blk;
+	int			action = 0;
+	int			error;
+
+	trace_xfs_da_join(state->args);
+
+	drop_blk = &state->path.blk[ state->path.active-1 ];
+	save_blk = &state->altpath.blk[ state->path.active-1 ];
+	ASSERT(state->path.blk[0].magic == XFS_DA_NODE_MAGIC);
+	ASSERT(drop_blk->magic == XFS_ATTR_LEAF_MAGIC ||
+	       drop_blk->magic == XFS_DIR2_LEAFN_MAGIC);
+
+	/*
+	 * Walk back up the tree joining/deallocating as necessary.
+	 * When we stop dropping blocks, break out.
+	 */
+	for (  ; state->path.active >= 2; drop_blk--, save_blk--,
+		 state->path.active--) {
+		/*
+		 * See if we can combine the block with a neighbor.
+		 *   (action == 0) => no options, just leave
+		 *   (action == 1) => coalesce, then unlink
+		 *   (action == 2) => block empty, unlink it
+		 */
+		switch (drop_blk->magic) {
+		case XFS_ATTR_LEAF_MAGIC:
+			error = xfs_attr3_leaf_toosmall(state, &action);
+			if (error)
+				return error;
+			if (action == 0)
+				return 0;
+			xfs_attr3_leaf_unbalance(state, drop_blk, save_blk);
+			break;
+		case XFS_DIR2_LEAFN_MAGIC:
+			error = xfs_dir2_leafn_toosmall(state, &action);
+			if (error)
+				return error;
+			if (action == 0)
+				return 0;
+			xfs_dir2_leafn_unbalance(state, drop_blk, save_blk);
+			break;
+		case XFS_DA_NODE_MAGIC:
+			/*
+			 * Remove the offending node, fixup hashvals,
+			 * check for a toosmall neighbor.
+			 */
+			xfs_da3_node_remove(state, drop_blk);
+			xfs_da3_fixhashpath(state, &state->path);
+			error = xfs_da3_node_toosmall(state, &action);
+			if (error)
+				return error;
+			if (action == 0)
+				return 0;
+			xfs_da3_node_unbalance(state, drop_blk, save_blk);
+			break;
+		}
+		xfs_da3_fixhashpath(state, &state->altpath);
+		error = xfs_da3_blk_unlink(state, drop_blk, save_blk);
+		xfs_da_state_kill_altpath(state);
+		if (error)
+			return error;
+		error = xfs_da_shrink_inode(state->args, drop_blk->blkno,
+							 drop_blk->bp);
+		drop_blk->bp = NULL;
+		if (error)
+			return error;
+	}
+	/*
+	 * We joined all the way to the top.  If it turns out that
+	 * we only have one entry in the root, make the child block
+	 * the new root.
+	 */
+	xfs_da3_node_remove(state, drop_blk);
+	xfs_da3_fixhashpath(state, &state->path);
+	error = xfs_da3_root_join(state, &state->path.blk[0]);
+	return error;
+}
+
+#ifdef	DEBUG
+static void
+xfs_da_blkinfo_onlychild_validate(struct xfs_da_blkinfo *blkinfo, __u16 level)
+{
+	__be16	magic = blkinfo->magic;
+
+	if (level == 1) {
+		ASSERT(magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
+		       magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC) ||
+		       magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC) ||
+		       magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC));
+	} else {
+		ASSERT(magic == cpu_to_be16(XFS_DA_NODE_MAGIC) ||
+		       magic == cpu_to_be16(XFS_DA3_NODE_MAGIC));
+	}
+	ASSERT(!blkinfo->forw);
+	ASSERT(!blkinfo->back);
+}
+#else	/* !DEBUG */
+#define	xfs_da_blkinfo_onlychild_validate(blkinfo, level)
+#endif	/* !DEBUG */
+
+/*
+ * We have only one entry in the root.  Copy the only remaining child of
+ * the old root to block 0 as the new root node.
+ */
+STATIC int
+xfs_da3_root_join(
+	struct xfs_da_state	*state,
+	struct xfs_da_state_blk	*root_blk)
+{
+	struct xfs_da_intnode	*oldroot;
+	struct xfs_da_args	*args;
+	xfs_dablk_t		child;
+	struct xfs_buf		*bp;
+	struct xfs_da3_icnode_hdr oldroothdr;
+	int			error;
+	struct xfs_inode	*dp = state->args->dp;
+
+	trace_xfs_da_root_join(state->args);
+
+	ASSERT(root_blk->magic == XFS_DA_NODE_MAGIC);
+
+	args = state->args;
+	oldroot = root_blk->bp->b_addr;
+	xfs_da3_node_hdr_from_disk(dp->i_mount, &oldroothdr, oldroot);
+	ASSERT(oldroothdr.forw == 0);
+	ASSERT(oldroothdr.back == 0);
+
+	/*
+	 * If the root has more than one child, then don't do anything.
+	 */
+	if (oldroothdr.count > 1)
+		return 0;
+
+	/*
+	 * Read in the (only) child block, then copy those bytes into
+	 * the root block's buffer and free the original child block.
+	 */
+	child = be32_to_cpu(oldroothdr.btree[0].before);
+	ASSERT(child != 0);
+	error = xfs_da3_node_read(args->trans, dp, child, &bp, args->whichfork);
+	if (error)
+		return error;
+	xfs_da_blkinfo_onlychild_validate(bp->b_addr, oldroothdr.level);
+
+	/*
+	 * This could be copying a leaf back into the root block in the case of
+	 * there only being a single leaf block left in the tree. Hence we have
+	 * to update the b_ops pointer as well to match the buffer type change
+	 * that could occur. For dir3 blocks we also need to update the block
+	 * number in the buffer header.
+	 */
+	memcpy(root_blk->bp->b_addr, bp->b_addr, args->geo->blksize);
+	root_blk->bp->b_ops = bp->b_ops;
+	xfs_trans_buf_copy_type(root_blk->bp, bp);
+	if (oldroothdr.magic == XFS_DA3_NODE_MAGIC) {
+		struct xfs_da3_blkinfo *da3 = root_blk->bp->b_addr;
+		da3->blkno = cpu_to_be64(xfs_buf_daddr(root_blk->bp));
+	}
+	xfs_trans_log_buf(args->trans, root_blk->bp, 0,
+			  args->geo->blksize - 1);
+	error = xfs_da_shrink_inode(args, child, bp);
+	return error;
+}
+
+/*
+ * Check a node block and its neighbors to see if the block should be
+ * collapsed into one or the other neighbor.  Always keep the block
+ * with the smaller block number.
+ * If the current block is over 50% full, don't try to join it, return 0.
+ * If the block is empty, fill in the state structure and return 2.
+ * If it can be collapsed, fill in the state structure and return 1.
+ * If nothing can be done, return 0.
+ */
+STATIC int
+xfs_da3_node_toosmall(
+	struct xfs_da_state	*state,
+	int			*action)
+{
+	struct xfs_da_intnode	*node;
+	struct xfs_da_state_blk	*blk;
+	struct xfs_da_blkinfo	*info;
+	xfs_dablk_t		blkno;
+	struct xfs_buf		*bp;
+	struct xfs_da3_icnode_hdr nodehdr;
+	int			count;
+	int			forward;
+	int			error;
+	int			retval;
+	int			i;
+	struct xfs_inode	*dp = state->args->dp;
+
+	trace_xfs_da_node_toosmall(state->args);
+
+	/*
+	 * Check for the degenerate case of the block being over 50% full.
+	 * If so, it's not worth even looking to see if we might be able
+	 * to coalesce with a sibling.
+	 */
+	blk = &state->path.blk[ state->path.active-1 ];
+	info = blk->bp->b_addr;
+	node = (xfs_da_intnode_t *)info;
+	xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, node);
+	if (nodehdr.count > (state->args->geo->node_ents >> 1)) {
+		*action = 0;	/* blk over 50%, don't try to join */
+		return 0;	/* blk over 50%, don't try to join */
+	}
+
+	/*
+	 * Check for the degenerate case of the block being empty.
+	 * If the block is empty, we'll simply delete it, no need to
+	 * coalesce it with a sibling block.  We choose (arbitrarily)
+	 * to merge with the forward block unless it is NULL.
+	 */
+	if (nodehdr.count == 0) {
+		/*
+		 * Make altpath point to the block we want to keep and
+		 * path point to the block we want to drop (this one).
+		 */
+		forward = (info->forw != 0);
+		memcpy(&state->altpath, &state->path, sizeof(state->path));
+		error = xfs_da3_path_shift(state, &state->altpath, forward,
+						 0, &retval);
+		if (error)
+			return error;
+		if (retval) {
+			*action = 0;
+		} else {
+			*action = 2;
+		}
+		return 0;
+	}
+
+	/*
+	 * Examine each sibling block to see if we can coalesce with
+	 * at least 25% free space to spare.  We need to figure out
+	 * whether to merge with the forward or the backward block.
+	 * We prefer coalescing with the lower numbered sibling so as
+	 * to shrink a directory over time.
+	 */
+	count  = state->args->geo->node_ents;
+	count -= state->args->geo->node_ents >> 2;
+	count -= nodehdr.count;
+
+	/* start with smaller blk num */
+	forward = nodehdr.forw < nodehdr.back;
+	for (i = 0; i < 2; forward = !forward, i++) {
+		struct xfs_da3_icnode_hdr thdr;
+		if (forward)
+			blkno = nodehdr.forw;
+		else
+			blkno = nodehdr.back;
+		if (blkno == 0)
+			continue;
+		error = xfs_da3_node_read(state->args->trans, dp, blkno, &bp,
+				state->args->whichfork);
+		if (error)
+			return error;
+
+		node = bp->b_addr;
+		xfs_da3_node_hdr_from_disk(dp->i_mount, &thdr, node);
+		xfs_trans_brelse(state->args->trans, bp);
+
+		if (count - thdr.count >= 0)
+			break;	/* fits with at least 25% to spare */
+	}
+	if (i >= 2) {
+		*action = 0;
+		return 0;
+	}
+
+	/*
+	 * Make altpath point to the block we want to keep (the lower
+	 * numbered block) and path point to the block we want to drop.
+	 */
+	memcpy(&state->altpath, &state->path, sizeof(state->path));
+	if (blkno < blk->blkno) {
+		error = xfs_da3_path_shift(state, &state->altpath, forward,
+						 0, &retval);
+	} else {
+		error = xfs_da3_path_shift(state, &state->path, forward,
+						 0, &retval);
+	}
+	if (error)
+		return error;
+	if (retval) {
+		*action = 0;
+		return 0;
+	}
+	*action = 1;
+	return 0;
+}
+
+/*
+ * Pick up the last hashvalue from an intermediate node.
+ */
+STATIC uint
+xfs_da3_node_lasthash(
+	struct xfs_inode	*dp,
+	struct xfs_buf		*bp,
+	int			*count)
+{
+	struct xfs_da3_icnode_hdr nodehdr;
+
+	xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, bp->b_addr);
+	if (count)
+		*count = nodehdr.count;
+	if (!nodehdr.count)
+		return 0;
+	return be32_to_cpu(nodehdr.btree[nodehdr.count - 1].hashval);
+}
+
+/*
+ * Walk back up the tree adjusting hash values as necessary,
+ * when we stop making changes, return.
+ */
+void
+xfs_da3_fixhashpath(
+	struct xfs_da_state	*state,
+	struct xfs_da_state_path *path)
+{
+	struct xfs_da_state_blk	*blk;
+	struct xfs_da_intnode	*node;
+	struct xfs_da_node_entry *btree;
+	xfs_dahash_t		lasthash=0;
+	int			level;
+	int			count;
+	struct xfs_inode	*dp = state->args->dp;
+
+	trace_xfs_da_fixhashpath(state->args);
+
+	level = path->active-1;
+	blk = &path->blk[ level ];
+	switch (blk->magic) {
+	case XFS_ATTR_LEAF_MAGIC:
+		lasthash = xfs_attr_leaf_lasthash(blk->bp, &count);
+		if (count == 0)
+			return;
+		break;
+	case XFS_DIR2_LEAFN_MAGIC:
+		lasthash = xfs_dir2_leaf_lasthash(dp, blk->bp, &count);
+		if (count == 0)
+			return;
+		break;
+	case XFS_DA_NODE_MAGIC:
+		lasthash = xfs_da3_node_lasthash(dp, blk->bp, &count);
+		if (count == 0)
+			return;
+		break;
+	}
+	for (blk--, level--; level >= 0; blk--, level--) {
+		struct xfs_da3_icnode_hdr nodehdr;
+
+		node = blk->bp->b_addr;
+		xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, node);
+		btree = nodehdr.btree;
+		if (be32_to_cpu(btree[blk->index].hashval) == lasthash)
+			break;
+		blk->hashval = lasthash;
+		btree[blk->index].hashval = cpu_to_be32(lasthash);
+		xfs_trans_log_buf(state->args->trans, blk->bp,
+				  XFS_DA_LOGRANGE(node, &btree[blk->index],
+						  sizeof(*btree)));
+
+		lasthash = be32_to_cpu(btree[nodehdr.count - 1].hashval);
+	}
+}
+
+/*
+ * Remove an entry from an intermediate node.
+ */
+STATIC void
+xfs_da3_node_remove(
+	struct xfs_da_state	*state,
+	struct xfs_da_state_blk	*drop_blk)
+{
+	struct xfs_da_intnode	*node;
+	struct xfs_da3_icnode_hdr nodehdr;
+	struct xfs_da_node_entry *btree;
+	int			index;
+	int			tmp;
+	struct xfs_inode	*dp = state->args->dp;
+
+	trace_xfs_da_node_remove(state->args);
+
+	node = drop_blk->bp->b_addr;
+	xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, node);
+	ASSERT(drop_blk->index < nodehdr.count);
+	ASSERT(drop_blk->index >= 0);
+
+	/*
+	 * Copy over the offending entry, or just zero it out.
+	 */
+	index = drop_blk->index;
+	btree = nodehdr.btree;
+	if (index < nodehdr.count - 1) {
+		tmp  = nodehdr.count - index - 1;
+		tmp *= (uint)sizeof(xfs_da_node_entry_t);
+		memmove(&btree[index], &btree[index + 1], tmp);
+		xfs_trans_log_buf(state->args->trans, drop_blk->bp,
+		    XFS_DA_LOGRANGE(node, &btree[index], tmp));
+		index = nodehdr.count - 1;
+	}
+	memset(&btree[index], 0, sizeof(xfs_da_node_entry_t));
+	xfs_trans_log_buf(state->args->trans, drop_blk->bp,
+	    XFS_DA_LOGRANGE(node, &btree[index], sizeof(btree[index])));
+	nodehdr.count -= 1;
+	xfs_da3_node_hdr_to_disk(dp->i_mount, node, &nodehdr);
+	xfs_trans_log_buf(state->args->trans, drop_blk->bp,
+	    XFS_DA_LOGRANGE(node, &node->hdr, state->args->geo->node_hdr_size));
+
+	/*
+	 * Copy the last hash value from the block to propagate upwards.
+	 */
+	drop_blk->hashval = be32_to_cpu(btree[index - 1].hashval);
+}
+
+/*
+ * Unbalance the elements between two intermediate nodes,
+ * move all Btree elements from one node into another.
+ */
+STATIC void
+xfs_da3_node_unbalance(
+	struct xfs_da_state	*state,
+	struct xfs_da_state_blk	*drop_blk,
+	struct xfs_da_state_blk	*save_blk)
+{
+	struct xfs_da_intnode	*drop_node;
+	struct xfs_da_intnode	*save_node;
+	struct xfs_da_node_entry *drop_btree;
+	struct xfs_da_node_entry *save_btree;
+	struct xfs_da3_icnode_hdr drop_hdr;
+	struct xfs_da3_icnode_hdr save_hdr;
+	struct xfs_trans	*tp;
+	int			sindex;
+	int			tmp;
+	struct xfs_inode	*dp = state->args->dp;
+
+	trace_xfs_da_node_unbalance(state->args);
+
+	drop_node = drop_blk->bp->b_addr;
+	save_node = save_blk->bp->b_addr;
+	xfs_da3_node_hdr_from_disk(dp->i_mount, &drop_hdr, drop_node);
+	xfs_da3_node_hdr_from_disk(dp->i_mount, &save_hdr, save_node);
+	drop_btree = drop_hdr.btree;
+	save_btree = save_hdr.btree;
+	tp = state->args->trans;
+
+	/*
+	 * If the dying block has lower hashvals, then move all the
+	 * elements in the remaining block up to make a hole.
+	 */
+	if ((be32_to_cpu(drop_btree[0].hashval) <
+			be32_to_cpu(save_btree[0].hashval)) ||
+	    (be32_to_cpu(drop_btree[drop_hdr.count - 1].hashval) <
+			be32_to_cpu(save_btree[save_hdr.count - 1].hashval))) {
+		/* XXX: check this - is memmove dst correct? */
+		tmp = save_hdr.count * sizeof(xfs_da_node_entry_t);
+		memmove(&save_btree[drop_hdr.count], &save_btree[0], tmp);
+
+		sindex = 0;
+		xfs_trans_log_buf(tp, save_blk->bp,
+			XFS_DA_LOGRANGE(save_node, &save_btree[0],
+				(save_hdr.count + drop_hdr.count) *
+						sizeof(xfs_da_node_entry_t)));
+	} else {
+		sindex = save_hdr.count;
+		xfs_trans_log_buf(tp, save_blk->bp,
+			XFS_DA_LOGRANGE(save_node, &save_btree[sindex],
+				drop_hdr.count * sizeof(xfs_da_node_entry_t)));
+	}
+
+	/*
+	 * Move all the B-tree elements from drop_blk to save_blk.
+	 */
+	tmp = drop_hdr.count * (uint)sizeof(xfs_da_node_entry_t);
+	memcpy(&save_btree[sindex], &drop_btree[0], tmp);
+	save_hdr.count += drop_hdr.count;
+
+	xfs_da3_node_hdr_to_disk(dp->i_mount, save_node, &save_hdr);
+	xfs_trans_log_buf(tp, save_blk->bp,
+		XFS_DA_LOGRANGE(save_node, &save_node->hdr,
+				state->args->geo->node_hdr_size));
+
+	/*
+	 * Save the last hashval in the remaining block for upward propagation.
+	 */
+	save_blk->hashval = be32_to_cpu(save_btree[save_hdr.count - 1].hashval);
+}
+
+/*========================================================================
+ * Routines used for finding things in the Btree.
+ *========================================================================*/
+
+/*
+ * Walk down the Btree looking for a particular filename, filling
+ * in the state structure as we go.
+ *
+ * We will set the state structure to point to each of the elements
+ * in each of the nodes where either the hashval is or should be.
+ *
+ * We support duplicate hashval's so for each entry in the current
+ * node that could contain the desired hashval, descend.  This is a
+ * pruned depth-first tree search.
+ */
+int							/* error */
+xfs_da3_node_lookup_int(
+	struct xfs_da_state	*state,
+	int			*result)
+{
+	struct xfs_da_state_blk	*blk;
+	struct xfs_da_blkinfo	*curr;
+	struct xfs_da_intnode	*node;
+	struct xfs_da_node_entry *btree;
+	struct xfs_da3_icnode_hdr nodehdr;
+	struct xfs_da_args	*args;
+	xfs_dablk_t		blkno;
+	xfs_dahash_t		hashval;
+	xfs_dahash_t		btreehashval;
+	int			probe;
+	int			span;
+	int			max;
+	int			error;
+	int			retval;
+	unsigned int		expected_level = 0;
+	uint16_t		magic;
+	struct xfs_inode	*dp = state->args->dp;
+
+	args = state->args;
+
+	/*
+	 * Descend thru the B-tree searching each level for the right
+	 * node to use, until the right hashval is found.
+	 */
+	blkno = args->geo->leafblk;
+	for (blk = &state->path.blk[0], state->path.active = 1;
+			 state->path.active <= XFS_DA_NODE_MAXDEPTH;
+			 blk++, state->path.active++) {
+		/*
+		 * Read the next node down in the tree.
+		 */
+		blk->blkno = blkno;
+		error = xfs_da3_node_read(args->trans, args->dp, blkno,
+					&blk->bp, args->whichfork);
+		if (error) {
+			blk->blkno = 0;
+			state->path.active--;
+			return error;
+		}
+		curr = blk->bp->b_addr;
+		magic = be16_to_cpu(curr->magic);
+
+		if (magic == XFS_ATTR_LEAF_MAGIC ||
+		    magic == XFS_ATTR3_LEAF_MAGIC) {
+			blk->magic = XFS_ATTR_LEAF_MAGIC;
+			blk->hashval = xfs_attr_leaf_lasthash(blk->bp, NULL);
+			break;
+		}
+
+		if (magic == XFS_DIR2_LEAFN_MAGIC ||
+		    magic == XFS_DIR3_LEAFN_MAGIC) {
+			blk->magic = XFS_DIR2_LEAFN_MAGIC;
+			blk->hashval = xfs_dir2_leaf_lasthash(args->dp,
+							      blk->bp, NULL);
+			break;
+		}
+
+		if (magic != XFS_DA_NODE_MAGIC && magic != XFS_DA3_NODE_MAGIC) {
+			xfs_buf_mark_corrupt(blk->bp);
+			return -EFSCORRUPTED;
+		}
+
+		blk->magic = XFS_DA_NODE_MAGIC;
+
+		/*
+		 * Search an intermediate node for a match.
+		 */
+		node = blk->bp->b_addr;
+		xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, node);
+		btree = nodehdr.btree;
+
+		/* Tree taller than we can handle; bail out! */
+		if (nodehdr.level >= XFS_DA_NODE_MAXDEPTH) {
+			xfs_buf_mark_corrupt(blk->bp);
+			return -EFSCORRUPTED;
+		}
+
+		/* Check the level from the root. */
+		if (blkno == args->geo->leafblk)
+			expected_level = nodehdr.level - 1;
+		else if (expected_level != nodehdr.level) {
+			xfs_buf_mark_corrupt(blk->bp);
+			return -EFSCORRUPTED;
+		} else
+			expected_level--;
+
+		max = nodehdr.count;
+		blk->hashval = be32_to_cpu(btree[max - 1].hashval);
+
+		/*
+		 * Binary search.  (note: small blocks will skip loop)
+		 */
+		probe = span = max / 2;
+		hashval = args->hashval;
+		while (span > 4) {
+			span /= 2;
+			btreehashval = be32_to_cpu(btree[probe].hashval);
+			if (btreehashval < hashval)
+				probe += span;
+			else if (btreehashval > hashval)
+				probe -= span;
+			else
+				break;
+		}
+		ASSERT((probe >= 0) && (probe < max));
+		ASSERT((span <= 4) ||
+			(be32_to_cpu(btree[probe].hashval) == hashval));
+
+		/*
+		 * Since we may have duplicate hashval's, find the first
+		 * matching hashval in the node.
+		 */
+		while (probe > 0 &&
+		       be32_to_cpu(btree[probe].hashval) >= hashval) {
+			probe--;
+		}
+		while (probe < max &&
+		       be32_to_cpu(btree[probe].hashval) < hashval) {
+			probe++;
+		}
+
+		/*
+		 * Pick the right block to descend on.
+		 */
+		if (probe == max) {
+			blk->index = max - 1;
+			blkno = be32_to_cpu(btree[max - 1].before);
+		} else {
+			blk->index = probe;
+			blkno = be32_to_cpu(btree[probe].before);
+		}
+
+		/* We can't point back to the root. */
+		if (XFS_IS_CORRUPT(dp->i_mount, blkno == args->geo->leafblk))
+			return -EFSCORRUPTED;
+	}
+
+	if (XFS_IS_CORRUPT(dp->i_mount, expected_level != 0))
+		return -EFSCORRUPTED;
+
+	/*
+	 * A leaf block that ends in the hashval that we are interested in
+	 * (final hashval == search hashval) means that the next block may
+	 * contain more entries with the same hashval, shift upward to the
+	 * next leaf and keep searching.
+	 */
+	for (;;) {
+		if (blk->magic == XFS_DIR2_LEAFN_MAGIC) {
+			retval = xfs_dir2_leafn_lookup_int(blk->bp, args,
+							&blk->index, state);
+		} else if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
+			retval = xfs_attr3_leaf_lookup_int(blk->bp, args);
+			blk->index = args->index;
+			args->blkno = blk->blkno;
+		} else {
+			ASSERT(0);
+			return -EFSCORRUPTED;
+		}
+		if (((retval == -ENOENT) || (retval == -ENOATTR)) &&
+		    (blk->hashval == args->hashval)) {
+			error = xfs_da3_path_shift(state, &state->path, 1, 1,
+							 &retval);
+			if (error)
+				return error;
+			if (retval == 0) {
+				continue;
+			} else if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
+				/* path_shift() gives ENOENT */
+				retval = -ENOATTR;
+			}
+		}
+		break;
+	}
+	*result = retval;
+	return 0;
+}
+
+/*========================================================================
+ * Utility routines.
+ *========================================================================*/
+
+/*
+ * Compare two intermediate nodes for "order".
+ */
+STATIC int
+xfs_da3_node_order(
+	struct xfs_inode *dp,
+	struct xfs_buf	*node1_bp,
+	struct xfs_buf	*node2_bp)
+{
+	struct xfs_da_intnode	*node1;
+	struct xfs_da_intnode	*node2;
+	struct xfs_da_node_entry *btree1;
+	struct xfs_da_node_entry *btree2;
+	struct xfs_da3_icnode_hdr node1hdr;
+	struct xfs_da3_icnode_hdr node2hdr;
+
+	node1 = node1_bp->b_addr;
+	node2 = node2_bp->b_addr;
+	xfs_da3_node_hdr_from_disk(dp->i_mount, &node1hdr, node1);
+	xfs_da3_node_hdr_from_disk(dp->i_mount, &node2hdr, node2);
+	btree1 = node1hdr.btree;
+	btree2 = node2hdr.btree;
+
+	if (node1hdr.count > 0 && node2hdr.count > 0 &&
+	    ((be32_to_cpu(btree2[0].hashval) < be32_to_cpu(btree1[0].hashval)) ||
+	     (be32_to_cpu(btree2[node2hdr.count - 1].hashval) <
+	      be32_to_cpu(btree1[node1hdr.count - 1].hashval)))) {
+		return 1;
+	}
+	return 0;
+}
+
+/*
+ * Link a new block into a doubly linked list of blocks (of whatever type).
+ */
+int							/* error */
+xfs_da3_blk_link(
+	struct xfs_da_state	*state,
+	struct xfs_da_state_blk	*old_blk,
+	struct xfs_da_state_blk	*new_blk)
+{
+	struct xfs_da_blkinfo	*old_info;
+	struct xfs_da_blkinfo	*new_info;
+	struct xfs_da_blkinfo	*tmp_info;
+	struct xfs_da_args	*args;
+	struct xfs_buf		*bp;
+	int			before = 0;
+	int			error;
+	struct xfs_inode	*dp = state->args->dp;
+
+	/*
+	 * Set up environment.
+	 */
+	args = state->args;
+	ASSERT(args != NULL);
+	old_info = old_blk->bp->b_addr;
+	new_info = new_blk->bp->b_addr;
+	ASSERT(old_blk->magic == XFS_DA_NODE_MAGIC ||
+	       old_blk->magic == XFS_DIR2_LEAFN_MAGIC ||
+	       old_blk->magic == XFS_ATTR_LEAF_MAGIC);
+
+	switch (old_blk->magic) {
+	case XFS_ATTR_LEAF_MAGIC:
+		before = xfs_attr_leaf_order(old_blk->bp, new_blk->bp);
+		break;
+	case XFS_DIR2_LEAFN_MAGIC:
+		before = xfs_dir2_leafn_order(dp, old_blk->bp, new_blk->bp);
+		break;
+	case XFS_DA_NODE_MAGIC:
+		before = xfs_da3_node_order(dp, old_blk->bp, new_blk->bp);
+		break;
+	}
+
+	/*
+	 * Link blocks in appropriate order.
+	 */
+	if (before) {
+		/*
+		 * Link new block in before existing block.
+		 */
+		trace_xfs_da_link_before(args);
+		new_info->forw = cpu_to_be32(old_blk->blkno);
+		new_info->back = old_info->back;
+		if (old_info->back) {
+			error = xfs_da3_node_read(args->trans, dp,
+						be32_to_cpu(old_info->back),
+						&bp, args->whichfork);
+			if (error)
+				return error;
+			ASSERT(bp != NULL);
+			tmp_info = bp->b_addr;
+			ASSERT(tmp_info->magic == old_info->magic);
+			ASSERT(be32_to_cpu(tmp_info->forw) == old_blk->blkno);
+			tmp_info->forw = cpu_to_be32(new_blk->blkno);
+			xfs_trans_log_buf(args->trans, bp, 0, sizeof(*tmp_info)-1);
+		}
+		old_info->back = cpu_to_be32(new_blk->blkno);
+	} else {
+		/*
+		 * Link new block in after existing block.
+		 */
+		trace_xfs_da_link_after(args);
+		new_info->forw = old_info->forw;
+		new_info->back = cpu_to_be32(old_blk->blkno);
+		if (old_info->forw) {
+			error = xfs_da3_node_read(args->trans, dp,
+						be32_to_cpu(old_info->forw),
+						&bp, args->whichfork);
+			if (error)
+				return error;
+			ASSERT(bp != NULL);
+			tmp_info = bp->b_addr;
+			ASSERT(tmp_info->magic == old_info->magic);
+			ASSERT(be32_to_cpu(tmp_info->back) == old_blk->blkno);
+			tmp_info->back = cpu_to_be32(new_blk->blkno);
+			xfs_trans_log_buf(args->trans, bp, 0, sizeof(*tmp_info)-1);
+		}
+		old_info->forw = cpu_to_be32(new_blk->blkno);
+	}
+
+	xfs_trans_log_buf(args->trans, old_blk->bp, 0, sizeof(*tmp_info) - 1);
+	xfs_trans_log_buf(args->trans, new_blk->bp, 0, sizeof(*tmp_info) - 1);
+	return 0;
+}
+
+/*
+ * Unlink a block from a doubly linked list of blocks.
+ */
+STATIC int						/* error */
+xfs_da3_blk_unlink(
+	struct xfs_da_state	*state,
+	struct xfs_da_state_blk	*drop_blk,
+	struct xfs_da_state_blk	*save_blk)
+{
+	struct xfs_da_blkinfo	*drop_info;
+	struct xfs_da_blkinfo	*save_info;
+	struct xfs_da_blkinfo	*tmp_info;
+	struct xfs_da_args	*args;
+	struct xfs_buf		*bp;
+	int			error;
+
+	/*
+	 * Set up environment.
+	 */
+	args = state->args;
+	ASSERT(args != NULL);
+	save_info = save_blk->bp->b_addr;
+	drop_info = drop_blk->bp->b_addr;
+	ASSERT(save_blk->magic == XFS_DA_NODE_MAGIC ||
+	       save_blk->magic == XFS_DIR2_LEAFN_MAGIC ||
+	       save_blk->magic == XFS_ATTR_LEAF_MAGIC);
+	ASSERT(save_blk->magic == drop_blk->magic);
+	ASSERT((be32_to_cpu(save_info->forw) == drop_blk->blkno) ||
+	       (be32_to_cpu(save_info->back) == drop_blk->blkno));
+	ASSERT((be32_to_cpu(drop_info->forw) == save_blk->blkno) ||
+	       (be32_to_cpu(drop_info->back) == save_blk->blkno));
+
+	/*
+	 * Unlink the leaf block from the doubly linked chain of leaves.
+	 */
+	if (be32_to_cpu(save_info->back) == drop_blk->blkno) {
+		trace_xfs_da_unlink_back(args);
+		save_info->back = drop_info->back;
+		if (drop_info->back) {
+			error = xfs_da3_node_read(args->trans, args->dp,
+						be32_to_cpu(drop_info->back),
+						&bp, args->whichfork);
+			if (error)
+				return error;
+			ASSERT(bp != NULL);
+			tmp_info = bp->b_addr;
+			ASSERT(tmp_info->magic == save_info->magic);
+			ASSERT(be32_to_cpu(tmp_info->forw) == drop_blk->blkno);
+			tmp_info->forw = cpu_to_be32(save_blk->blkno);
+			xfs_trans_log_buf(args->trans, bp, 0,
+						    sizeof(*tmp_info) - 1);
+		}
+	} else {
+		trace_xfs_da_unlink_forward(args);
+		save_info->forw = drop_info->forw;
+		if (drop_info->forw) {
+			error = xfs_da3_node_read(args->trans, args->dp,
+						be32_to_cpu(drop_info->forw),
+						&bp, args->whichfork);
+			if (error)
+				return error;
+			ASSERT(bp != NULL);
+			tmp_info = bp->b_addr;
+			ASSERT(tmp_info->magic == save_info->magic);
+			ASSERT(be32_to_cpu(tmp_info->back) == drop_blk->blkno);
+			tmp_info->back = cpu_to_be32(save_blk->blkno);
+			xfs_trans_log_buf(args->trans, bp, 0,
+						    sizeof(*tmp_info) - 1);
+		}
+	}
+
+	xfs_trans_log_buf(args->trans, save_blk->bp, 0, sizeof(*save_info) - 1);
+	return 0;
+}
+
+/*
+ * Move a path "forward" or "!forward" one block at the current level.
+ *
+ * This routine will adjust a "path" to point to the next block
+ * "forward" (higher hashvalues) or "!forward" (lower hashvals) in the
+ * Btree, including updating pointers to the intermediate nodes between
+ * the new bottom and the root.
+ */
+int							/* error */
+xfs_da3_path_shift(
+	struct xfs_da_state	*state,
+	struct xfs_da_state_path *path,
+	int			forward,
+	int			release,
+	int			*result)
+{
+	struct xfs_da_state_blk	*blk;
+	struct xfs_da_blkinfo	*info;
+	struct xfs_da_args	*args;
+	struct xfs_da_node_entry *btree;
+	struct xfs_da3_icnode_hdr nodehdr;
+	struct xfs_buf		*bp;
+	xfs_dablk_t		blkno = 0;
+	int			level;
+	int			error;
+	struct xfs_inode	*dp = state->args->dp;
+
+	trace_xfs_da_path_shift(state->args);
+
+	/*
+	 * Roll up the Btree looking for the first block where our
+	 * current index is not at the edge of the block.  Note that
+	 * we skip the bottom layer because we want the sibling block.
+	 */
+	args = state->args;
+	ASSERT(args != NULL);
+	ASSERT(path != NULL);
+	ASSERT((path->active > 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
+	level = (path->active-1) - 1;	/* skip bottom layer in path */
+	for (; level >= 0; level--) {
+		blk = &path->blk[level];
+		xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr,
+					   blk->bp->b_addr);
+
+		if (forward && (blk->index < nodehdr.count - 1)) {
+			blk->index++;
+			blkno = be32_to_cpu(nodehdr.btree[blk->index].before);
+			break;
+		} else if (!forward && (blk->index > 0)) {
+			blk->index--;
+			blkno = be32_to_cpu(nodehdr.btree[blk->index].before);
+			break;
+		}
+	}
+	if (level < 0) {
+		*result = -ENOENT;	/* we're out of our tree */
+		ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
+		return 0;
+	}
+
+	/*
+	 * Roll down the edge of the subtree until we reach the
+	 * same depth we were at originally.
+	 */
+	for (blk++, level++; level < path->active; blk++, level++) {
+		/*
+		 * Read the next child block into a local buffer.
+		 */
+		error = xfs_da3_node_read(args->trans, dp, blkno, &bp,
+					  args->whichfork);
+		if (error)
+			return error;
+
+		/*
+		 * Release the old block (if it's dirty, the trans doesn't
+		 * actually let go) and swap the local buffer into the path
+		 * structure. This ensures failure of the above read doesn't set
+		 * a NULL buffer in an active slot in the path.
+		 */
+		if (release)
+			xfs_trans_brelse(args->trans, blk->bp);
+		blk->blkno = blkno;
+		blk->bp = bp;
+
+		info = blk->bp->b_addr;
+		ASSERT(info->magic == cpu_to_be16(XFS_DA_NODE_MAGIC) ||
+		       info->magic == cpu_to_be16(XFS_DA3_NODE_MAGIC) ||
+		       info->magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
+		       info->magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC) ||
+		       info->magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC) ||
+		       info->magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC));
+
+
+		/*
+		 * Note: we flatten the magic number to a single type so we
+		 * don't have to compare against crc/non-crc types elsewhere.
+		 */
+		switch (be16_to_cpu(info->magic)) {
+		case XFS_DA_NODE_MAGIC:
+		case XFS_DA3_NODE_MAGIC:
+			blk->magic = XFS_DA_NODE_MAGIC;
+			xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr,
+						   bp->b_addr);
+			btree = nodehdr.btree;
+			blk->hashval = be32_to_cpu(btree[nodehdr.count - 1].hashval);
+			if (forward)
+				blk->index = 0;
+			else
+				blk->index = nodehdr.count - 1;
+			blkno = be32_to_cpu(btree[blk->index].before);
+			break;
+		case XFS_ATTR_LEAF_MAGIC:
+		case XFS_ATTR3_LEAF_MAGIC:
+			blk->magic = XFS_ATTR_LEAF_MAGIC;
+			ASSERT(level == path->active-1);
+			blk->index = 0;
+			blk->hashval = xfs_attr_leaf_lasthash(blk->bp, NULL);
+			break;
+		case XFS_DIR2_LEAFN_MAGIC:
+		case XFS_DIR3_LEAFN_MAGIC:
+			blk->magic = XFS_DIR2_LEAFN_MAGIC;
+			ASSERT(level == path->active-1);
+			blk->index = 0;
+			blk->hashval = xfs_dir2_leaf_lasthash(args->dp,
+							      blk->bp, NULL);
+			break;
+		default:
+			ASSERT(0);
+			break;
+		}
+	}
+	*result = 0;
+	return 0;
+}
+
+
+/*========================================================================
+ * Utility routines.
+ *========================================================================*/
+
+/*
+ * Implement a simple hash on a character string.
+ * Rotate the hash value by 7 bits, then XOR each character in.
+ * This is implemented with some source-level loop unrolling.
+ */
+xfs_dahash_t
+xfs_da_hashname(const uint8_t *name, int namelen)
+{
+	xfs_dahash_t hash;
+
+	/*
+	 * Do four characters at a time as long as we can.
+	 */
+	for (hash = 0; namelen >= 4; namelen -= 4, name += 4)
+		hash = (name[0] << 21) ^ (name[1] << 14) ^ (name[2] << 7) ^
+		       (name[3] << 0) ^ rol32(hash, 7 * 4);
+
+	/*
+	 * Now do the rest of the characters.
+	 */
+	switch (namelen) {
+	case 3:
+		return (name[0] << 14) ^ (name[1] << 7) ^ (name[2] << 0) ^
+		       rol32(hash, 7 * 3);
+	case 2:
+		return (name[0] << 7) ^ (name[1] << 0) ^ rol32(hash, 7 * 2);
+	case 1:
+		return (name[0] << 0) ^ rol32(hash, 7 * 1);
+	default: /* case 0: */
+		return hash;
+	}
+}
+
+enum xfs_dacmp
+xfs_da_compname(
+	struct xfs_da_args *args,
+	const unsigned char *name,
+	int		len)
+{
+	return (args->namelen == len && memcmp(args->name, name, len) == 0) ?
+					XFS_CMP_EXACT : XFS_CMP_DIFFERENT;
+}
+
+int
+xfs_da_grow_inode_int(
+	struct xfs_da_args	*args,
+	xfs_fileoff_t		*bno,
+	int			count)
+{
+	struct xfs_trans	*tp = args->trans;
+	struct xfs_inode	*dp = args->dp;
+	int			w = args->whichfork;
+	xfs_rfsblock_t		nblks = dp->i_nblocks;
+	struct xfs_bmbt_irec	map, *mapp;
+	int			nmap, error, got, i, mapi;
+
+	/*
+	 * Find a spot in the file space to put the new block.
+	 */
+	error = xfs_bmap_first_unused(tp, dp, count, bno, w);
+	if (error)
+		return error;
+
+	/*
+	 * Try mapping it in one filesystem block.
+	 */
+	nmap = 1;
+	error = xfs_bmapi_write(tp, dp, *bno, count,
+			xfs_bmapi_aflag(w)|XFS_BMAPI_METADATA|XFS_BMAPI_CONTIG,
+			args->total, &map, &nmap);
+	if (error)
+		return error;
+
+	ASSERT(nmap <= 1);
+	if (nmap == 1) {
+		mapp = &map;
+		mapi = 1;
+	} else if (nmap == 0 && count > 1) {
+		xfs_fileoff_t		b;
+		int			c;
+
+		/*
+		 * If we didn't get it and the block might work if fragmented,
+		 * try without the CONTIG flag.  Loop until we get it all.
+		 */
+		mapp = kmem_alloc(sizeof(*mapp) * count, 0);
+		for (b = *bno, mapi = 0; b < *bno + count; ) {
+			c = (int)(*bno + count - b);
+			nmap = min(XFS_BMAP_MAX_NMAP, c);
+			error = xfs_bmapi_write(tp, dp, b, c,
+					xfs_bmapi_aflag(w)|XFS_BMAPI_METADATA,
+					args->total, &mapp[mapi], &nmap);
+			if (error)
+				goto out_free_map;
+			if (nmap < 1)
+				break;
+			mapi += nmap;
+			b = mapp[mapi - 1].br_startoff +
+			    mapp[mapi - 1].br_blockcount;
+		}
+	} else {
+		mapi = 0;
+		mapp = NULL;
+	}
+
+	/*
+	 * Count the blocks we got, make sure it matches the total.
+	 */
+	for (i = 0, got = 0; i < mapi; i++)
+		got += mapp[i].br_blockcount;
+	if (got != count || mapp[0].br_startoff != *bno ||
+	    mapp[mapi - 1].br_startoff + mapp[mapi - 1].br_blockcount !=
+	    *bno + count) {
+		error = -ENOSPC;
+		goto out_free_map;
+	}
+
+	/* account for newly allocated blocks in reserved blocks total */
+	args->total -= dp->i_nblocks - nblks;
+
+out_free_map:
+	if (mapp != &map)
+		kmem_free(mapp);
+	return error;
+}
+
+/*
+ * Add a block to the btree ahead of the file.
+ * Return the new block number to the caller.
+ */
+int
+xfs_da_grow_inode(
+	struct xfs_da_args	*args,
+	xfs_dablk_t		*new_blkno)
+{
+	xfs_fileoff_t		bno;
+	int			error;
+
+	trace_xfs_da_grow_inode(args);
+
+	bno = args->geo->leafblk;
+	error = xfs_da_grow_inode_int(args, &bno, args->geo->fsbcount);
+	if (!error)
+		*new_blkno = (xfs_dablk_t)bno;
+	return error;
+}
+
+/*
+ * Ick.  We need to always be able to remove a btree block, even
+ * if there's no space reservation because the filesystem is full.
+ * This is called if xfs_bunmapi on a btree block fails due to ENOSPC.
+ * It swaps the target block with the last block in the file.  The
+ * last block in the file can always be removed since it can't cause
+ * a bmap btree split to do that.
+ */
+STATIC int
+xfs_da3_swap_lastblock(
+	struct xfs_da_args	*args,
+	xfs_dablk_t		*dead_blknop,
+	struct xfs_buf		**dead_bufp)
+{
+	struct xfs_da_blkinfo	*dead_info;
+	struct xfs_da_blkinfo	*sib_info;
+	struct xfs_da_intnode	*par_node;
+	struct xfs_da_intnode	*dead_node;
+	struct xfs_dir2_leaf	*dead_leaf2;
+	struct xfs_da_node_entry *btree;
+	struct xfs_da3_icnode_hdr par_hdr;
+	struct xfs_inode	*dp;
+	struct xfs_trans	*tp;
+	struct xfs_mount	*mp;
+	struct xfs_buf		*dead_buf;
+	struct xfs_buf		*last_buf;
+	struct xfs_buf		*sib_buf;
+	struct xfs_buf		*par_buf;
+	xfs_dahash_t		dead_hash;
+	xfs_fileoff_t		lastoff;
+	xfs_dablk_t		dead_blkno;
+	xfs_dablk_t		last_blkno;
+	xfs_dablk_t		sib_blkno;
+	xfs_dablk_t		par_blkno;
+	int			error;
+	int			w;
+	int			entno;
+	int			level;
+	int			dead_level;
+
+	trace_xfs_da_swap_lastblock(args);
+
+	dead_buf = *dead_bufp;
+	dead_blkno = *dead_blknop;
+	tp = args->trans;
+	dp = args->dp;
+	w = args->whichfork;
+	ASSERT(w == XFS_DATA_FORK);
+	mp = dp->i_mount;
+	lastoff = args->geo->freeblk;
+	error = xfs_bmap_last_before(tp, dp, &lastoff, w);
+	if (error)
+		return error;
+	if (XFS_IS_CORRUPT(mp, lastoff == 0))
+		return -EFSCORRUPTED;
+	/*
+	 * Read the last block in the btree space.
+	 */
+	last_blkno = (xfs_dablk_t)lastoff - args->geo->fsbcount;
+	error = xfs_da3_node_read(tp, dp, last_blkno, &last_buf, w);
+	if (error)
+		return error;
+	/*
+	 * Copy the last block into the dead buffer and log it.
+	 */
+	memcpy(dead_buf->b_addr, last_buf->b_addr, args->geo->blksize);
+	xfs_trans_log_buf(tp, dead_buf, 0, args->geo->blksize - 1);
+	dead_info = dead_buf->b_addr;
+	/*
+	 * Get values from the moved block.
+	 */
+	if (dead_info->magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
+	    dead_info->magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC)) {
+		struct xfs_dir3_icleaf_hdr leafhdr;
+		struct xfs_dir2_leaf_entry *ents;
+
+		dead_leaf2 = (xfs_dir2_leaf_t *)dead_info;
+		xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &leafhdr,
+					    dead_leaf2);
+		ents = leafhdr.ents;
+		dead_level = 0;
+		dead_hash = be32_to_cpu(ents[leafhdr.count - 1].hashval);
+	} else {
+		struct xfs_da3_icnode_hdr deadhdr;
+
+		dead_node = (xfs_da_intnode_t *)dead_info;
+		xfs_da3_node_hdr_from_disk(dp->i_mount, &deadhdr, dead_node);
+		btree = deadhdr.btree;
+		dead_level = deadhdr.level;
+		dead_hash = be32_to_cpu(btree[deadhdr.count - 1].hashval);
+	}
+	sib_buf = par_buf = NULL;
+	/*
+	 * If the moved block has a left sibling, fix up the pointers.
+	 */
+	if ((sib_blkno = be32_to_cpu(dead_info->back))) {
+		error = xfs_da3_node_read(tp, dp, sib_blkno, &sib_buf, w);
+		if (error)
+			goto done;
+		sib_info = sib_buf->b_addr;
+		if (XFS_IS_CORRUPT(mp,
+				   be32_to_cpu(sib_info->forw) != last_blkno ||
+				   sib_info->magic != dead_info->magic)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		sib_info->forw = cpu_to_be32(dead_blkno);
+		xfs_trans_log_buf(tp, sib_buf,
+			XFS_DA_LOGRANGE(sib_info, &sib_info->forw,
+					sizeof(sib_info->forw)));
+		sib_buf = NULL;
+	}
+	/*
+	 * If the moved block has a right sibling, fix up the pointers.
+	 */
+	if ((sib_blkno = be32_to_cpu(dead_info->forw))) {
+		error = xfs_da3_node_read(tp, dp, sib_blkno, &sib_buf, w);
+		if (error)
+			goto done;
+		sib_info = sib_buf->b_addr;
+		if (XFS_IS_CORRUPT(mp,
+				   be32_to_cpu(sib_info->back) != last_blkno ||
+				   sib_info->magic != dead_info->magic)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		sib_info->back = cpu_to_be32(dead_blkno);
+		xfs_trans_log_buf(tp, sib_buf,
+			XFS_DA_LOGRANGE(sib_info, &sib_info->back,
+					sizeof(sib_info->back)));
+		sib_buf = NULL;
+	}
+	par_blkno = args->geo->leafblk;
+	level = -1;
+	/*
+	 * Walk down the tree looking for the parent of the moved block.
+	 */
+	for (;;) {
+		error = xfs_da3_node_read(tp, dp, par_blkno, &par_buf, w);
+		if (error)
+			goto done;
+		par_node = par_buf->b_addr;
+		xfs_da3_node_hdr_from_disk(dp->i_mount, &par_hdr, par_node);
+		if (XFS_IS_CORRUPT(mp,
+				   level >= 0 && level != par_hdr.level + 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		level = par_hdr.level;
+		btree = par_hdr.btree;
+		for (entno = 0;
+		     entno < par_hdr.count &&
+		     be32_to_cpu(btree[entno].hashval) < dead_hash;
+		     entno++)
+			continue;
+		if (XFS_IS_CORRUPT(mp, entno == par_hdr.count)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		par_blkno = be32_to_cpu(btree[entno].before);
+		if (level == dead_level + 1)
+			break;
+		xfs_trans_brelse(tp, par_buf);
+		par_buf = NULL;
+	}
+	/*
+	 * We're in the right parent block.
+	 * Look for the right entry.
+	 */
+	for (;;) {
+		for (;
+		     entno < par_hdr.count &&
+		     be32_to_cpu(btree[entno].before) != last_blkno;
+		     entno++)
+			continue;
+		if (entno < par_hdr.count)
+			break;
+		par_blkno = par_hdr.forw;
+		xfs_trans_brelse(tp, par_buf);
+		par_buf = NULL;
+		if (XFS_IS_CORRUPT(mp, par_blkno == 0)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		error = xfs_da3_node_read(tp, dp, par_blkno, &par_buf, w);
+		if (error)
+			goto done;
+		par_node = par_buf->b_addr;
+		xfs_da3_node_hdr_from_disk(dp->i_mount, &par_hdr, par_node);
+		if (XFS_IS_CORRUPT(mp, par_hdr.level != level)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		btree = par_hdr.btree;
+		entno = 0;
+	}
+	/*
+	 * Update the parent entry pointing to the moved block.
+	 */
+	btree[entno].before = cpu_to_be32(dead_blkno);
+	xfs_trans_log_buf(tp, par_buf,
+		XFS_DA_LOGRANGE(par_node, &btree[entno].before,
+				sizeof(btree[entno].before)));
+	*dead_blknop = last_blkno;
+	*dead_bufp = last_buf;
+	return 0;
+done:
+	if (par_buf)
+		xfs_trans_brelse(tp, par_buf);
+	if (sib_buf)
+		xfs_trans_brelse(tp, sib_buf);
+	xfs_trans_brelse(tp, last_buf);
+	return error;
+}
+
+/*
+ * Remove a btree block from a directory or attribute.
+ */
+int
+xfs_da_shrink_inode(
+	struct xfs_da_args	*args,
+	xfs_dablk_t		dead_blkno,
+	struct xfs_buf		*dead_buf)
+{
+	struct xfs_inode	*dp;
+	int			done, error, w, count;
+	struct xfs_trans	*tp;
+
+	trace_xfs_da_shrink_inode(args);
+
+	dp = args->dp;
+	w = args->whichfork;
+	tp = args->trans;
+	count = args->geo->fsbcount;
+	for (;;) {
+		/*
+		 * Remove extents.  If we get ENOSPC for a dir we have to move
+		 * the last block to the place we want to kill.
+		 */
+		error = xfs_bunmapi(tp, dp, dead_blkno, count,
+				    xfs_bmapi_aflag(w), 0, &done);
+		if (error == -ENOSPC) {
+			if (w != XFS_DATA_FORK)
+				break;
+			error = xfs_da3_swap_lastblock(args, &dead_blkno,
+						      &dead_buf);
+			if (error)
+				break;
+		} else {
+			break;
+		}
+	}
+	xfs_trans_binval(tp, dead_buf);
+	return error;
+}
+
+static int
+xfs_dabuf_map(
+	struct xfs_inode	*dp,
+	xfs_dablk_t		bno,
+	unsigned int		flags,
+	int			whichfork,
+	struct xfs_buf_map	**mapp,
+	int			*nmaps)
+{
+	struct xfs_mount	*mp = dp->i_mount;
+	int			nfsb = xfs_dabuf_nfsb(mp, whichfork);
+	struct xfs_bmbt_irec	irec, *irecs = &irec;
+	struct xfs_buf_map	*map = *mapp;
+	xfs_fileoff_t		off = bno;
+	int			error = 0, nirecs, i;
+
+	if (nfsb > 1)
+		irecs = kmem_zalloc(sizeof(irec) * nfsb, KM_NOFS);
+
+	nirecs = nfsb;
+	error = xfs_bmapi_read(dp, bno, nfsb, irecs, &nirecs,
+			xfs_bmapi_aflag(whichfork));
+	if (error)
+		goto out_free_irecs;
+
+	/*
+	 * Use the caller provided map for the single map case, else allocate a
+	 * larger one that needs to be free by the caller.
+	 */
+	if (nirecs > 1) {
+		map = kmem_zalloc(nirecs * sizeof(struct xfs_buf_map), KM_NOFS);
+		if (!map) {
+			error = -ENOMEM;
+			goto out_free_irecs;
+		}
+		*mapp = map;
+	}
+
+	for (i = 0; i < nirecs; i++) {
+		if (irecs[i].br_startblock == HOLESTARTBLOCK ||
+		    irecs[i].br_startblock == DELAYSTARTBLOCK)
+			goto invalid_mapping;
+		if (off != irecs[i].br_startoff)
+			goto invalid_mapping;
+
+		map[i].bm_bn = XFS_FSB_TO_DADDR(mp, irecs[i].br_startblock);
+		map[i].bm_len = XFS_FSB_TO_BB(mp, irecs[i].br_blockcount);
+		off += irecs[i].br_blockcount;
+	}
+
+	if (off != bno + nfsb)
+		goto invalid_mapping;
+
+	*nmaps = nirecs;
+out_free_irecs:
+	if (irecs != &irec)
+		kmem_free(irecs);
+	return error;
+
+invalid_mapping:
+	/* Caller ok with no mapping. */
+	if (XFS_IS_CORRUPT(mp, !(flags & XFS_DABUF_MAP_HOLE_OK))) {
+		error = -EFSCORRUPTED;
+		if (xfs_error_level >= XFS_ERRLEVEL_LOW) {
+			xfs_alert(mp, "%s: bno %u inode %llu",
+					__func__, bno, dp->i_ino);
+
+			for (i = 0; i < nirecs; i++) {
+				xfs_alert(mp,
+"[%02d] br_startoff %lld br_startblock %lld br_blockcount %lld br_state %d",
+					i, irecs[i].br_startoff,
+					irecs[i].br_startblock,
+					irecs[i].br_blockcount,
+					irecs[i].br_state);
+			}
+		}
+	} else {
+		*nmaps = 0;
+	}
+	goto out_free_irecs;
+}
+
+/*
+ * Get a buffer for the dir/attr block.
+ */
+int
+xfs_da_get_buf(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*dp,
+	xfs_dablk_t		bno,
+	struct xfs_buf		**bpp,
+	int			whichfork)
+{
+	struct xfs_mount	*mp = dp->i_mount;
+	struct xfs_buf		*bp;
+	struct xfs_buf_map	map, *mapp = &map;
+	int			nmap = 1;
+	int			error;
+
+	*bpp = NULL;
+	error = xfs_dabuf_map(dp, bno, 0, whichfork, &mapp, &nmap);
+	if (error || nmap == 0)
+		goto out_free;
+
+	error = xfs_trans_get_buf_map(tp, mp->m_ddev_targp, mapp, nmap, 0, &bp);
+	if (error)
+		goto out_free;
+
+	*bpp = bp;
+
+out_free:
+	if (mapp != &map)
+		kmem_free(mapp);
+
+	return error;
+}
+
+/*
+ * Get a buffer for the dir/attr block, fill in the contents.
+ */
+int
+xfs_da_read_buf(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*dp,
+	xfs_dablk_t		bno,
+	unsigned int		flags,
+	struct xfs_buf		**bpp,
+	int			whichfork,
+	const struct xfs_buf_ops *ops)
+{
+	struct xfs_mount	*mp = dp->i_mount;
+	struct xfs_buf		*bp;
+	struct xfs_buf_map	map, *mapp = &map;
+	int			nmap = 1;
+	int			error;
+
+	*bpp = NULL;
+	error = xfs_dabuf_map(dp, bno, flags, whichfork, &mapp, &nmap);
+	if (error || !nmap)
+		goto out_free;
+
+	error = xfs_trans_read_buf_map(mp, tp, mp->m_ddev_targp, mapp, nmap, 0,
+			&bp, ops);
+	if (error)
+		goto out_free;
+
+	if (whichfork == XFS_ATTR_FORK)
+		xfs_buf_set_ref(bp, XFS_ATTR_BTREE_REF);
+	else
+		xfs_buf_set_ref(bp, XFS_DIR_BTREE_REF);
+	*bpp = bp;
+out_free:
+	if (mapp != &map)
+		kmem_free(mapp);
+
+	return error;
+}
+
+/*
+ * Readahead the dir/attr block.
+ */
+int
+xfs_da_reada_buf(
+	struct xfs_inode	*dp,
+	xfs_dablk_t		bno,
+	unsigned int		flags,
+	int			whichfork,
+	const struct xfs_buf_ops *ops)
+{
+	struct xfs_buf_map	map;
+	struct xfs_buf_map	*mapp;
+	int			nmap;
+	int			error;
+
+	mapp = &map;
+	nmap = 1;
+	error = xfs_dabuf_map(dp, bno, flags, whichfork, &mapp, &nmap);
+	if (error || !nmap)
+		goto out_free;
+
+	xfs_buf_readahead_map(dp->i_mount->m_ddev_targp, mapp, nmap, ops);
+
+out_free:
+	if (mapp != &map)
+		kmem_free(mapp);
+
+	return error;
+}
diff --git a/fs/xfs/libxfs/xfs_da_btree.h b/fs/xfs/libxfs/xfs_da_btree.h
new file mode 100644
index 000000000..ffa3df5b2
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_da_btree.h
@@ -0,0 +1,239 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2000,2002,2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_DA_BTREE_H__
+#define	__XFS_DA_BTREE_H__
+
+struct xfs_inode;
+struct xfs_trans;
+
+/*
+ * Directory/attribute geometry information. There will be one of these for each
+ * data fork type, and it will be passed around via the xfs_da_args. Global
+ * structures will be attached to the xfs_mount.
+ */
+struct xfs_da_geometry {
+	unsigned int	blksize;	/* da block size in bytes */
+	unsigned int	fsbcount;	/* da block size in filesystem blocks */
+	uint8_t		fsblog;		/* log2 of _filesystem_ block size */
+	uint8_t		blklog;		/* log2 of da block size */
+	unsigned int	node_hdr_size;	/* danode header size in bytes */
+	unsigned int	node_ents;	/* # of entries in a danode */
+	unsigned int	magicpct;	/* 37% of block size in bytes */
+	xfs_dablk_t	datablk;	/* blockno of dir data v2 */
+	unsigned int	leaf_hdr_size;	/* dir2 leaf header size */
+	unsigned int	leaf_max_ents;	/* # of entries in dir2 leaf */
+	xfs_dablk_t	leafblk;	/* blockno of leaf data v2 */
+	unsigned int	free_hdr_size;	/* dir2 free header size */
+	unsigned int	free_max_bests;	/* # of bests entries in dir2 free */
+	xfs_dablk_t	freeblk;	/* blockno of free data v2 */
+	xfs_extnum_t	max_extents;	/* Max. extents in corresponding fork */
+
+	xfs_dir2_data_aoff_t data_first_offset;
+	size_t		data_entry_offset;
+};
+
+/*========================================================================
+ * Btree searching and modification structure definitions.
+ *========================================================================*/
+
+/*
+ * Search comparison results
+ */
+enum xfs_dacmp {
+	XFS_CMP_DIFFERENT,	/* names are completely different */
+	XFS_CMP_EXACT,		/* names are exactly the same */
+	XFS_CMP_CASE		/* names are same but differ in case */
+};
+
+/*
+ * Structure to ease passing around component names.
+ */
+typedef struct xfs_da_args {
+	struct xfs_da_geometry *geo;	/* da block geometry */
+	const uint8_t		*name;		/* string (maybe not NULL terminated) */
+	int		namelen;	/* length of string (maybe no NULL) */
+	uint8_t		filetype;	/* filetype of inode for directories */
+	void		*value;		/* set of bytes (maybe contain NULLs) */
+	int		valuelen;	/* length of value */
+	unsigned int	attr_filter;	/* XFS_ATTR_{ROOT,SECURE,INCOMPLETE} */
+	unsigned int	attr_flags;	/* XATTR_{CREATE,REPLACE} */
+	xfs_dahash_t	hashval;	/* hash value of name */
+	xfs_ino_t	inumber;	/* input/output inode number */
+	struct xfs_inode *dp;		/* directory inode to manipulate */
+	struct xfs_trans *trans;	/* current trans (changes over time) */
+	xfs_extlen_t	total;		/* total blocks needed, for 1st bmap */
+	int		whichfork;	/* data or attribute fork */
+	xfs_dablk_t	blkno;		/* blkno of attr leaf of interest */
+	int		index;		/* index of attr of interest in blk */
+	xfs_dablk_t	rmtblkno;	/* remote attr value starting blkno */
+	int		rmtblkcnt;	/* remote attr value block count */
+	int		rmtvaluelen;	/* remote attr value length in bytes */
+	xfs_dablk_t	blkno2;		/* blkno of 2nd attr leaf of interest */
+	int		index2;		/* index of 2nd attr in blk */
+	xfs_dablk_t	rmtblkno2;	/* remote attr value starting blkno */
+	int		rmtblkcnt2;	/* remote attr value block count */
+	int		rmtvaluelen2;	/* remote attr value length in bytes */
+	uint32_t	op_flags;	/* operation flags */
+	enum xfs_dacmp	cmpresult;	/* name compare result for lookups */
+} xfs_da_args_t;
+
+/*
+ * Operation flags:
+ */
+#define XFS_DA_OP_JUSTCHECK	(1u << 0) /* check for ok with no space */
+#define XFS_DA_OP_REPLACE	(1u << 1) /* this is an atomic replace op */
+#define XFS_DA_OP_ADDNAME	(1u << 2) /* this is an add operation */
+#define XFS_DA_OP_OKNOENT	(1u << 3) /* lookup op, ENOENT ok, else die */
+#define XFS_DA_OP_CILOOKUP	(1u << 4) /* lookup returns CI name if found */
+#define XFS_DA_OP_NOTIME	(1u << 5) /* don't update inode timestamps */
+#define XFS_DA_OP_REMOVE	(1u << 6) /* this is a remove operation */
+#define XFS_DA_OP_RECOVERY	(1u << 7) /* Log recovery operation */
+#define XFS_DA_OP_LOGGED	(1u << 8) /* Use intent items to track op */
+
+#define XFS_DA_OP_FLAGS \
+	{ XFS_DA_OP_JUSTCHECK,	"JUSTCHECK" }, \
+	{ XFS_DA_OP_REPLACE,	"REPLACE" }, \
+	{ XFS_DA_OP_ADDNAME,	"ADDNAME" }, \
+	{ XFS_DA_OP_OKNOENT,	"OKNOENT" }, \
+	{ XFS_DA_OP_CILOOKUP,	"CILOOKUP" }, \
+	{ XFS_DA_OP_NOTIME,	"NOTIME" }, \
+	{ XFS_DA_OP_REMOVE,	"REMOVE" }, \
+	{ XFS_DA_OP_RECOVERY,	"RECOVERY" }, \
+	{ XFS_DA_OP_LOGGED,	"LOGGED" }
+
+/*
+ * Storage for holding state during Btree searches and split/join ops.
+ *
+ * Only need space for 5 intermediate nodes.  With a minimum of 62-way
+ * fanout to the Btree, we can support over 900 million directory blocks,
+ * which is slightly more than enough.
+ */
+typedef struct xfs_da_state_blk {
+	struct xfs_buf	*bp;		/* buffer containing block */
+	xfs_dablk_t	blkno;		/* filesystem blkno of buffer */
+	xfs_daddr_t	disk_blkno;	/* on-disk blkno (in BBs) of buffer */
+	int		index;		/* relevant index into block */
+	xfs_dahash_t	hashval;	/* last hash value in block */
+	int		magic;		/* blk's magic number, ie: blk type */
+} xfs_da_state_blk_t;
+
+typedef struct xfs_da_state_path {
+	int			active;		/* number of active levels */
+	xfs_da_state_blk_t	blk[XFS_DA_NODE_MAXDEPTH];
+} xfs_da_state_path_t;
+
+typedef struct xfs_da_state {
+	xfs_da_args_t		*args;		/* filename arguments */
+	struct xfs_mount	*mp;		/* filesystem mount point */
+	xfs_da_state_path_t	path;		/* search/split paths */
+	xfs_da_state_path_t	altpath;	/* alternate path for join */
+	unsigned char		inleaf;		/* insert into 1->lf, 0->splf */
+	unsigned char		extravalid;	/* T/F: extrablk is in use */
+	unsigned char		extraafter;	/* T/F: extrablk is after new */
+	xfs_da_state_blk_t	extrablk;	/* for double-splits on leaves */
+						/* for dirv2 extrablk is data */
+} xfs_da_state_t;
+
+/*
+ * In-core version of the node header to abstract the differences in the v2 and
+ * v3 disk format of the headers. Callers need to convert to/from disk format as
+ * appropriate.
+ */
+struct xfs_da3_icnode_hdr {
+	uint32_t		forw;
+	uint32_t		back;
+	uint16_t		magic;
+	uint16_t		count;
+	uint16_t		level;
+
+	/*
+	 * Pointer to the on-disk format entries, which are behind the
+	 * variable size (v4 vs v5) header in the on-disk block.
+	 */
+	struct xfs_da_node_entry *btree;
+};
+
+/*
+ * Utility macros to aid in logging changed structure fields.
+ */
+#define XFS_DA_LOGOFF(BASE, ADDR)	((char *)(ADDR) - (char *)(BASE))
+#define XFS_DA_LOGRANGE(BASE, ADDR, SIZE)	\
+		(uint)(XFS_DA_LOGOFF(BASE, ADDR)), \
+		(uint)(XFS_DA_LOGOFF(BASE, ADDR)+(SIZE)-1)
+
+/*========================================================================
+ * Function prototypes.
+ *========================================================================*/
+
+/*
+ * Routines used for growing the Btree.
+ */
+int	xfs_da3_node_create(struct xfs_da_args *args, xfs_dablk_t blkno,
+			    int level, struct xfs_buf **bpp, int whichfork);
+int	xfs_da3_split(xfs_da_state_t *state);
+
+/*
+ * Routines used for shrinking the Btree.
+ */
+int	xfs_da3_join(xfs_da_state_t *state);
+void	xfs_da3_fixhashpath(struct xfs_da_state *state,
+			    struct xfs_da_state_path *path_to_to_fix);
+
+/*
+ * Routines used for finding things in the Btree.
+ */
+int	xfs_da3_node_lookup_int(xfs_da_state_t *state, int *result);
+int	xfs_da3_path_shift(xfs_da_state_t *state, xfs_da_state_path_t *path,
+					 int forward, int release, int *result);
+/*
+ * Utility routines.
+ */
+int	xfs_da3_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk,
+				       xfs_da_state_blk_t *new_blk);
+int	xfs_da3_node_read(struct xfs_trans *tp, struct xfs_inode *dp,
+			xfs_dablk_t bno, struct xfs_buf **bpp, int whichfork);
+int	xfs_da3_node_read_mapped(struct xfs_trans *tp, struct xfs_inode *dp,
+			xfs_daddr_t mappedbno, struct xfs_buf **bpp,
+			int whichfork);
+
+/*
+ * Utility routines.
+ */
+
+#define XFS_DABUF_MAP_HOLE_OK	(1u << 0)
+
+int	xfs_da_grow_inode(xfs_da_args_t *args, xfs_dablk_t *new_blkno);
+int	xfs_da_grow_inode_int(struct xfs_da_args *args, xfs_fileoff_t *bno,
+			      int count);
+int	xfs_da_get_buf(struct xfs_trans *trans, struct xfs_inode *dp,
+		xfs_dablk_t bno, struct xfs_buf **bp, int whichfork);
+int	xfs_da_read_buf(struct xfs_trans *trans, struct xfs_inode *dp,
+		xfs_dablk_t bno, unsigned int flags, struct xfs_buf **bpp,
+		int whichfork, const struct xfs_buf_ops *ops);
+int	xfs_da_reada_buf(struct xfs_inode *dp, xfs_dablk_t bno,
+		unsigned int flags, int whichfork,
+		const struct xfs_buf_ops *ops);
+int	xfs_da_shrink_inode(xfs_da_args_t *args, xfs_dablk_t dead_blkno,
+					  struct xfs_buf *dead_buf);
+
+uint xfs_da_hashname(const uint8_t *name_string, int name_length);
+enum xfs_dacmp xfs_da_compname(struct xfs_da_args *args,
+				const unsigned char *name, int len);
+
+
+struct xfs_da_state *xfs_da_state_alloc(struct xfs_da_args *args);
+void xfs_da_state_free(xfs_da_state_t *state);
+void xfs_da_state_reset(struct xfs_da_state *state, struct xfs_da_args *args);
+
+void	xfs_da3_node_hdr_from_disk(struct xfs_mount *mp,
+		struct xfs_da3_icnode_hdr *to, struct xfs_da_intnode *from);
+void	xfs_da3_node_hdr_to_disk(struct xfs_mount *mp,
+		struct xfs_da_intnode *to, struct xfs_da3_icnode_hdr *from);
+
+extern struct kmem_cache	*xfs_da_state_cache;
+
+#endif	/* __XFS_DA_BTREE_H__ */
diff --git a/fs/xfs/libxfs/xfs_da_format.h b/fs/xfs/libxfs/xfs_da_format.h
new file mode 100644
index 000000000..25e284108
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_da_format.h
@@ -0,0 +1,805 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_DA_FORMAT_H__
+#define __XFS_DA_FORMAT_H__
+
+/*
+ * This structure is common to both leaf nodes and non-leaf nodes in the Btree.
+ *
+ * It is used to manage a doubly linked list of all blocks at the same
+ * level in the Btree, and to identify which type of block this is.
+ */
+#define XFS_DA_NODE_MAGIC	0xfebe	/* magic number: non-leaf blocks */
+#define XFS_ATTR_LEAF_MAGIC	0xfbee	/* magic number: attribute leaf blks */
+#define XFS_DIR2_LEAF1_MAGIC	0xd2f1	/* magic number: v2 dirlf single blks */
+#define XFS_DIR2_LEAFN_MAGIC	0xd2ff	/* magic number: v2 dirlf multi blks */
+
+typedef struct xfs_da_blkinfo {
+	__be32		forw;			/* previous block in list */
+	__be32		back;			/* following block in list */
+	__be16		magic;			/* validity check on block */
+	__be16		pad;			/* unused */
+} xfs_da_blkinfo_t;
+
+/*
+ * CRC enabled directory structure types
+ *
+ * The headers change size for the additional verification information, but
+ * otherwise the tree layouts and contents are unchanged. Hence the da btree
+ * code can use the struct xfs_da_blkinfo for manipulating the tree links and
+ * magic numbers without modification for both v2 and v3 nodes.
+ */
+#define XFS_DA3_NODE_MAGIC	0x3ebe	/* magic number: non-leaf blocks */
+#define XFS_ATTR3_LEAF_MAGIC	0x3bee	/* magic number: attribute leaf blks */
+#define XFS_DIR3_LEAF1_MAGIC	0x3df1	/* magic number: v3 dirlf single blks */
+#define XFS_DIR3_LEAFN_MAGIC	0x3dff	/* magic number: v3 dirlf multi blks */
+
+struct xfs_da3_blkinfo {
+	/*
+	 * the node link manipulation code relies on the fact that the first
+	 * element of this structure is the struct xfs_da_blkinfo so it can
+	 * ignore the differences in the rest of the structures.
+	 */
+	struct xfs_da_blkinfo	hdr;
+	__be32			crc;	/* CRC of block */
+	__be64			blkno;	/* first block of the buffer */
+	__be64			lsn;	/* sequence number of last write */
+	uuid_t			uuid;	/* filesystem we belong to */
+	__be64			owner;	/* inode that owns the block */
+};
+
+/*
+ * This is the structure of the root and intermediate nodes in the Btree.
+ * The leaf nodes are defined above.
+ *
+ * Entries are not packed.
+ *
+ * Since we have duplicate keys, use a binary search but always follow
+ * all match in the block, not just the first match found.
+ */
+#define XFS_DA_NODE_MAXDEPTH	5	/* max depth of Btree */
+
+typedef struct xfs_da_node_hdr {
+	struct xfs_da_blkinfo	info;	/* block type, links, etc. */
+	__be16			__count; /* count of active entries */
+	__be16			__level; /* level above leaves (leaf == 0) */
+} xfs_da_node_hdr_t;
+
+struct xfs_da3_node_hdr {
+	struct xfs_da3_blkinfo	info;	/* block type, links, etc. */
+	__be16			__count; /* count of active entries */
+	__be16			__level; /* level above leaves (leaf == 0) */
+	__be32			__pad32;
+};
+
+#define XFS_DA3_NODE_CRC_OFF	(offsetof(struct xfs_da3_node_hdr, info.crc))
+
+typedef struct xfs_da_node_entry {
+	__be32	hashval;	/* hash value for this descendant */
+	__be32	before;		/* Btree block before this key */
+} xfs_da_node_entry_t;
+
+typedef struct xfs_da_intnode {
+	struct xfs_da_node_hdr	hdr;
+	struct xfs_da_node_entry __btree[];
+} xfs_da_intnode_t;
+
+struct xfs_da3_intnode {
+	struct xfs_da3_node_hdr	hdr;
+	struct xfs_da_node_entry __btree[];
+};
+
+/*
+ * Directory version 2.
+ *
+ * There are 4 possible formats:
+ *  - shortform - embedded into the inode
+ *  - single block - data with embedded leaf at the end
+ *  - multiple data blocks, single leaf+freeindex block
+ *  - data blocks, node and leaf blocks (btree), freeindex blocks
+ *
+ * Note: many node blocks structures and constants are shared with the attr
+ * code and defined in xfs_da_btree.h.
+ */
+
+#define	XFS_DIR2_BLOCK_MAGIC	0x58443242	/* XD2B: single block dirs */
+#define	XFS_DIR2_DATA_MAGIC	0x58443244	/* XD2D: multiblock dirs */
+#define	XFS_DIR2_FREE_MAGIC	0x58443246	/* XD2F: free index blocks */
+
+/*
+ * Directory Version 3 With CRCs.
+ *
+ * The tree formats are the same as for version 2 directories.  The difference
+ * is in the block header and dirent formats. In many cases the v3 structures
+ * use v2 definitions as they are no different and this makes code sharing much
+ * easier.
+ *
+ * Also, the xfs_dir3_*() functions handle both v2 and v3 formats - if the
+ * format is v2 then they switch to the existing v2 code, or the format is v3
+ * they implement the v3 functionality. This means the existing dir2 is a mix of
+ * xfs_dir2/xfs_dir3 calls and functions. The xfs_dir3 functions are called
+ * where there is a difference in the formats, otherwise the code is unchanged.
+ *
+ * Where it is possible, the code decides what to do based on the magic numbers
+ * in the blocks rather than feature bits in the superblock. This means the code
+ * is as independent of the external XFS code as possible as doesn't require
+ * passing struct xfs_mount pointers into places where it isn't really
+ * necessary.
+ *
+ * Version 3 includes:
+ *
+ *	- a larger block header for CRC and identification purposes and so the
+ *	offsets of all the structures inside the blocks are different.
+ *
+ *	- new magic numbers to be able to detect the v2/v3 types on the fly.
+ */
+
+#define	XFS_DIR3_BLOCK_MAGIC	0x58444233	/* XDB3: single block dirs */
+#define	XFS_DIR3_DATA_MAGIC	0x58444433	/* XDD3: multiblock dirs */
+#define	XFS_DIR3_FREE_MAGIC	0x58444633	/* XDF3: free index blocks */
+
+/*
+ * Dirents in version 3 directories have a file type field. Additions to this
+ * list are an on-disk format change, requiring feature bits. Valid values
+ * are as follows:
+ */
+#define XFS_DIR3_FT_UNKNOWN		0
+#define XFS_DIR3_FT_REG_FILE		1
+#define XFS_DIR3_FT_DIR			2
+#define XFS_DIR3_FT_CHRDEV		3
+#define XFS_DIR3_FT_BLKDEV		4
+#define XFS_DIR3_FT_FIFO		5
+#define XFS_DIR3_FT_SOCK		6
+#define XFS_DIR3_FT_SYMLINK		7
+#define XFS_DIR3_FT_WHT			8
+
+#define XFS_DIR3_FT_MAX			9
+
+/*
+ * Byte offset in data block and shortform entry.
+ */
+typedef uint16_t	xfs_dir2_data_off_t;
+#define	NULLDATAOFF	0xffffU
+typedef uint		xfs_dir2_data_aoff_t;	/* argument form */
+
+/*
+ * Offset in data space of a data entry.
+ */
+typedef uint32_t	xfs_dir2_dataptr_t;
+#define	XFS_DIR2_MAX_DATAPTR	((xfs_dir2_dataptr_t)0xffffffff)
+#define	XFS_DIR2_NULL_DATAPTR	((xfs_dir2_dataptr_t)0)
+
+/*
+ * Byte offset in a directory.
+ */
+typedef	xfs_off_t	xfs_dir2_off_t;
+
+/*
+ * Directory block number (logical dirblk in file)
+ */
+typedef uint32_t	xfs_dir2_db_t;
+
+#define XFS_INO32_SIZE	4
+#define XFS_INO64_SIZE	8
+#define XFS_INO64_DIFF	(XFS_INO64_SIZE - XFS_INO32_SIZE)
+
+#define	XFS_DIR2_MAX_SHORT_INUM	((xfs_ino_t)0xffffffffULL)
+
+/*
+ * Directory layout when stored internal to an inode.
+ *
+ * Small directories are packed as tightly as possible so as to fit into the
+ * literal area of the inode.  These "shortform" directories consist of a
+ * single xfs_dir2_sf_hdr header followed by zero or more xfs_dir2_sf_entry
+ * structures.  Due the different inode number storage size and the variable
+ * length name field in the xfs_dir2_sf_entry all these structure are
+ * variable length, and the accessors in this file should be used to iterate
+ * over them.
+ */
+typedef struct xfs_dir2_sf_hdr {
+	uint8_t			count;		/* count of entries */
+	uint8_t			i8count;	/* count of 8-byte inode #s */
+	uint8_t			parent[8];	/* parent dir inode number */
+} __packed xfs_dir2_sf_hdr_t;
+
+typedef struct xfs_dir2_sf_entry {
+	__u8			namelen;	/* actual name length */
+	__u8			offset[2];	/* saved offset */
+	__u8			name[];		/* name, variable size */
+	/*
+	 * A single byte containing the file type field follows the inode
+	 * number for version 3 directory entries.
+	 *
+	 * A 64-bit or 32-bit inode number follows here, at a variable offset
+	 * after the name.
+	 */
+} __packed xfs_dir2_sf_entry_t;
+
+static inline int xfs_dir2_sf_hdr_size(int i8count)
+{
+	return sizeof(struct xfs_dir2_sf_hdr) -
+		(i8count == 0) * XFS_INO64_DIFF;
+}
+
+static inline xfs_dir2_data_aoff_t
+xfs_dir2_sf_get_offset(xfs_dir2_sf_entry_t *sfep)
+{
+	return get_unaligned_be16(sfep->offset);
+}
+
+static inline void
+xfs_dir2_sf_put_offset(xfs_dir2_sf_entry_t *sfep, xfs_dir2_data_aoff_t off)
+{
+	put_unaligned_be16(off, sfep->offset);
+}
+
+static inline struct xfs_dir2_sf_entry *
+xfs_dir2_sf_firstentry(struct xfs_dir2_sf_hdr *hdr)
+{
+	return (struct xfs_dir2_sf_entry *)
+		((char *)hdr + xfs_dir2_sf_hdr_size(hdr->i8count));
+}
+
+/*
+ * Data block structures.
+ *
+ * A pure data block looks like the following drawing on disk:
+ *
+ *    +-------------------------------------------------+
+ *    | xfs_dir2_data_hdr_t                             |
+ *    +-------------------------------------------------+
+ *    | xfs_dir2_data_entry_t OR xfs_dir2_data_unused_t |
+ *    | xfs_dir2_data_entry_t OR xfs_dir2_data_unused_t |
+ *    | xfs_dir2_data_entry_t OR xfs_dir2_data_unused_t |
+ *    | ...                                             |
+ *    +-------------------------------------------------+
+ *    | unused space                                    |
+ *    +-------------------------------------------------+
+ *
+ * As all the entries are variable size structures the accessors below should
+ * be used to iterate over them.
+ *
+ * In addition to the pure data blocks for the data and node formats,
+ * most structures are also used for the combined data/freespace "block"
+ * format below.
+ */
+
+#define	XFS_DIR2_DATA_ALIGN_LOG	3		/* i.e., 8 bytes */
+#define	XFS_DIR2_DATA_ALIGN	(1 << XFS_DIR2_DATA_ALIGN_LOG)
+#define	XFS_DIR2_DATA_FREE_TAG	0xffff
+#define	XFS_DIR2_DATA_FD_COUNT	3
+
+/*
+ * Directory address space divided into sections,
+ * spaces separated by 32GB.
+ */
+#define	XFS_DIR2_MAX_SPACES	3
+#define	XFS_DIR2_SPACE_SIZE	(1ULL << (32 + XFS_DIR2_DATA_ALIGN_LOG))
+#define	XFS_DIR2_DATA_SPACE	0
+#define	XFS_DIR2_DATA_OFFSET	(XFS_DIR2_DATA_SPACE * XFS_DIR2_SPACE_SIZE)
+
+/*
+ * Describe a free area in the data block.
+ *
+ * The freespace will be formatted as a xfs_dir2_data_unused_t.
+ */
+typedef struct xfs_dir2_data_free {
+	__be16			offset;		/* start of freespace */
+	__be16			length;		/* length of freespace */
+} xfs_dir2_data_free_t;
+
+/*
+ * Header for the data blocks.
+ *
+ * The code knows that XFS_DIR2_DATA_FD_COUNT is 3.
+ */
+typedef struct xfs_dir2_data_hdr {
+	__be32			magic;		/* XFS_DIR2_DATA_MAGIC or */
+						/* XFS_DIR2_BLOCK_MAGIC */
+	xfs_dir2_data_free_t	bestfree[XFS_DIR2_DATA_FD_COUNT];
+} xfs_dir2_data_hdr_t;
+
+/*
+ * define a structure for all the verification fields we are adding to the
+ * directory block structures. This will be used in several structures.
+ * The magic number must be the first entry to align with all the dir2
+ * structures so we determine how to decode them just by the magic number.
+ */
+struct xfs_dir3_blk_hdr {
+	__be32			magic;	/* magic number */
+	__be32			crc;	/* CRC of block */
+	__be64			blkno;	/* first block of the buffer */
+	__be64			lsn;	/* sequence number of last write */
+	uuid_t			uuid;	/* filesystem we belong to */
+	__be64			owner;	/* inode that owns the block */
+};
+
+struct xfs_dir3_data_hdr {
+	struct xfs_dir3_blk_hdr	hdr;
+	xfs_dir2_data_free_t	best_free[XFS_DIR2_DATA_FD_COUNT];
+	__be32			pad;	/* 64 bit alignment */
+};
+
+#define XFS_DIR3_DATA_CRC_OFF  offsetof(struct xfs_dir3_data_hdr, hdr.crc)
+
+/*
+ * Active entry in a data block.
+ *
+ * Aligned to 8 bytes.  After the variable length name field there is a
+ * 2 byte tag field, which can be accessed using xfs_dir3_data_entry_tag_p.
+ *
+ * For dir3 structures, there is file type field between the name and the tag.
+ * This can only be manipulated by helper functions. It is packed hard against
+ * the end of the name so any padding for rounding is between the file type and
+ * the tag.
+ */
+typedef struct xfs_dir2_data_entry {
+	__be64			inumber;	/* inode number */
+	__u8			namelen;	/* name length */
+	__u8			name[];		/* name bytes, no null */
+     /* __u8			filetype; */	/* type of inode we point to */
+     /*	__be16                  tag; */		/* starting offset of us */
+} xfs_dir2_data_entry_t;
+
+/*
+ * Unused entry in a data block.
+ *
+ * Aligned to 8 bytes.  Tag appears as the last 2 bytes and must be accessed
+ * using xfs_dir2_data_unused_tag_p.
+ */
+typedef struct xfs_dir2_data_unused {
+	__be16			freetag;	/* XFS_DIR2_DATA_FREE_TAG */
+	__be16			length;		/* total free length */
+						/* variable offset */
+	__be16			tag;		/* starting offset of us */
+} xfs_dir2_data_unused_t;
+
+/*
+ * Pointer to a freespace's tag word.
+ */
+static inline __be16 *
+xfs_dir2_data_unused_tag_p(struct xfs_dir2_data_unused *dup)
+{
+	return (__be16 *)((char *)dup +
+			be16_to_cpu(dup->length) - sizeof(__be16));
+}
+
+/*
+ * Leaf block structures.
+ *
+ * A pure leaf block looks like the following drawing on disk:
+ *
+ *    +---------------------------+
+ *    | xfs_dir2_leaf_hdr_t       |
+ *    +---------------------------+
+ *    | xfs_dir2_leaf_entry_t     |
+ *    | xfs_dir2_leaf_entry_t     |
+ *    | xfs_dir2_leaf_entry_t     |
+ *    | xfs_dir2_leaf_entry_t     |
+ *    | ...                       |
+ *    +---------------------------+
+ *    | xfs_dir2_data_off_t       |
+ *    | xfs_dir2_data_off_t       |
+ *    | xfs_dir2_data_off_t       |
+ *    | ...                       |
+ *    +---------------------------+
+ *    | xfs_dir2_leaf_tail_t      |
+ *    +---------------------------+
+ *
+ * The xfs_dir2_data_off_t members (bests) and tail are at the end of the block
+ * for single-leaf (magic = XFS_DIR2_LEAF1_MAGIC) blocks only, but not present
+ * for directories with separate leaf nodes and free space blocks
+ * (magic = XFS_DIR2_LEAFN_MAGIC).
+ *
+ * As all the entries are variable size structures the accessors below should
+ * be used to iterate over them.
+ */
+
+/*
+ * Offset of the leaf/node space.  First block in this space
+ * is the btree root.
+ */
+#define	XFS_DIR2_LEAF_SPACE	1
+#define	XFS_DIR2_LEAF_OFFSET	(XFS_DIR2_LEAF_SPACE * XFS_DIR2_SPACE_SIZE)
+
+/*
+ * Leaf block header.
+ */
+typedef struct xfs_dir2_leaf_hdr {
+	xfs_da_blkinfo_t	info;		/* header for da routines */
+	__be16			count;		/* count of entries */
+	__be16			stale;		/* count of stale entries */
+} xfs_dir2_leaf_hdr_t;
+
+struct xfs_dir3_leaf_hdr {
+	struct xfs_da3_blkinfo	info;		/* header for da routines */
+	__be16			count;		/* count of entries */
+	__be16			stale;		/* count of stale entries */
+	__be32			pad;		/* 64 bit alignment */
+};
+
+/*
+ * Leaf block entry.
+ */
+typedef struct xfs_dir2_leaf_entry {
+	__be32			hashval;	/* hash value of name */
+	__be32			address;	/* address of data entry */
+} xfs_dir2_leaf_entry_t;
+
+/*
+ * Leaf block tail.
+ */
+typedef struct xfs_dir2_leaf_tail {
+	__be32			bestcount;
+} xfs_dir2_leaf_tail_t;
+
+/*
+ * Leaf block.
+ */
+typedef struct xfs_dir2_leaf {
+	xfs_dir2_leaf_hdr_t	hdr;			/* leaf header */
+	xfs_dir2_leaf_entry_t	__ents[];		/* entries */
+} xfs_dir2_leaf_t;
+
+struct xfs_dir3_leaf {
+	struct xfs_dir3_leaf_hdr	hdr;		/* leaf header */
+	struct xfs_dir2_leaf_entry	__ents[];	/* entries */
+};
+
+#define XFS_DIR3_LEAF_CRC_OFF  offsetof(struct xfs_dir3_leaf_hdr, info.crc)
+
+/*
+ * Get address of the bests array in the single-leaf block.
+ */
+static inline __be16 *
+xfs_dir2_leaf_bests_p(struct xfs_dir2_leaf_tail *ltp)
+{
+	return (__be16 *)ltp - be32_to_cpu(ltp->bestcount);
+}
+
+/*
+ * Free space block definitions for the node format.
+ */
+
+/*
+ * Offset of the freespace index.
+ */
+#define	XFS_DIR2_FREE_SPACE	2
+#define	XFS_DIR2_FREE_OFFSET	(XFS_DIR2_FREE_SPACE * XFS_DIR2_SPACE_SIZE)
+
+typedef	struct xfs_dir2_free_hdr {
+	__be32			magic;		/* XFS_DIR2_FREE_MAGIC */
+	__be32			firstdb;	/* db of first entry */
+	__be32			nvalid;		/* count of valid entries */
+	__be32			nused;		/* count of used entries */
+} xfs_dir2_free_hdr_t;
+
+typedef struct xfs_dir2_free {
+	xfs_dir2_free_hdr_t	hdr;		/* block header */
+	__be16			bests[];	/* best free counts */
+						/* unused entries are -1 */
+} xfs_dir2_free_t;
+
+struct xfs_dir3_free_hdr {
+	struct xfs_dir3_blk_hdr	hdr;
+	__be32			firstdb;	/* db of first entry */
+	__be32			nvalid;		/* count of valid entries */
+	__be32			nused;		/* count of used entries */
+	__be32			pad;		/* 64 bit alignment */
+};
+
+struct xfs_dir3_free {
+	struct xfs_dir3_free_hdr hdr;
+	__be16			bests[];	/* best free counts */
+						/* unused entries are -1 */
+};
+
+#define XFS_DIR3_FREE_CRC_OFF  offsetof(struct xfs_dir3_free, hdr.hdr.crc)
+
+/*
+ * Single block format.
+ *
+ * The single block format looks like the following drawing on disk:
+ *
+ *    +-------------------------------------------------+
+ *    | xfs_dir2_data_hdr_t                             |
+ *    +-------------------------------------------------+
+ *    | xfs_dir2_data_entry_t OR xfs_dir2_data_unused_t |
+ *    | xfs_dir2_data_entry_t OR xfs_dir2_data_unused_t |
+ *    | xfs_dir2_data_entry_t OR xfs_dir2_data_unused_t :
+ *    | ...                                             |
+ *    +-------------------------------------------------+
+ *    | unused space                                    |
+ *    +-------------------------------------------------+
+ *    | ...                                             |
+ *    | xfs_dir2_leaf_entry_t                           |
+ *    | xfs_dir2_leaf_entry_t                           |
+ *    +-------------------------------------------------+
+ *    | xfs_dir2_block_tail_t                           |
+ *    +-------------------------------------------------+
+ *
+ * As all the entries are variable size structures the accessors below should
+ * be used to iterate over them.
+ */
+
+typedef struct xfs_dir2_block_tail {
+	__be32		count;			/* count of leaf entries */
+	__be32		stale;			/* count of stale lf entries */
+} xfs_dir2_block_tail_t;
+
+/*
+ * Pointer to the leaf entries embedded in a data block (1-block format)
+ */
+static inline struct xfs_dir2_leaf_entry *
+xfs_dir2_block_leaf_p(struct xfs_dir2_block_tail *btp)
+{
+	return ((struct xfs_dir2_leaf_entry *)btp) - be32_to_cpu(btp->count);
+}
+
+
+/*
+ * Attribute storage layout
+ *
+ * Attribute lists are structured around Btrees where all the data
+ * elements are in the leaf nodes.  Attribute names are hashed into an int,
+ * then that int is used as the index into the Btree.  Since the hashval
+ * of an attribute name may not be unique, we may have duplicate keys.  The
+ * internal links in the Btree are logical block offsets into the file.
+ *
+ * Struct leaf_entry's are packed from the top.  Name/values grow from the
+ * bottom but are not packed.  The freemap contains run-length-encoded entries
+ * for the free bytes after the leaf_entry's, but only the N largest such,
+ * smaller runs are dropped.  When the freemap doesn't show enough space
+ * for an allocation, we compact the name/value area and try again.  If we
+ * still don't have enough space, then we have to split the block.  The
+ * name/value structs (both local and remote versions) must be 32bit aligned.
+ *
+ * Since we have duplicate hash keys, for each key that matches, compare
+ * the actual name string.  The root and intermediate node search always
+ * takes the first-in-the-block key match found, so we should only have
+ * to work "forw"ard.  If none matches, continue with the "forw"ard leaf
+ * nodes until the hash key changes or the attribute name is found.
+ *
+ * We store the fact that an attribute is a ROOT/USER/SECURE attribute in
+ * the leaf_entry.  The namespaces are independent only because we also look
+ * at the namespace bit when we are looking for a matching attribute name.
+ *
+ * We also store an "incomplete" bit in the leaf_entry.  It shows that an
+ * attribute is in the middle of being created and should not be shown to
+ * the user if we crash during the time that the bit is set.  We clear the
+ * bit when we have finished setting up the attribute.  We do this because
+ * we cannot create some large attributes inside a single transaction, and we
+ * need some indication that we weren't finished if we crash in the middle.
+ */
+#define XFS_ATTR_LEAF_MAPSIZE	3	/* how many freespace slots */
+
+/*
+ * Entries are packed toward the top as tight as possible.
+ */
+struct xfs_attr_shortform {
+	struct xfs_attr_sf_hdr {	/* constant-structure header block */
+		__be16	totsize;	/* total bytes in shortform list */
+		__u8	count;	/* count of active entries */
+		__u8	padding;
+	} hdr;
+	struct xfs_attr_sf_entry {
+		uint8_t namelen;	/* actual length of name (no NULL) */
+		uint8_t valuelen;	/* actual length of value (no NULL) */
+		uint8_t flags;	/* flags bits (see xfs_attr_leaf.h) */
+		uint8_t nameval[];	/* name & value bytes concatenated */
+	} list[1];			/* variable sized array */
+};
+
+typedef struct xfs_attr_leaf_map {	/* RLE map of free bytes */
+	__be16	base;			  /* base of free region */
+	__be16	size;			  /* length of free region */
+} xfs_attr_leaf_map_t;
+
+typedef struct xfs_attr_leaf_hdr {	/* constant-structure header block */
+	xfs_da_blkinfo_t info;		/* block type, links, etc. */
+	__be16	count;			/* count of active leaf_entry's */
+	__be16	usedbytes;		/* num bytes of names/values stored */
+	__be16	firstused;		/* first used byte in name area */
+	__u8	holes;			/* != 0 if blk needs compaction */
+	__u8	pad1;
+	xfs_attr_leaf_map_t freemap[XFS_ATTR_LEAF_MAPSIZE];
+					/* N largest free regions */
+} xfs_attr_leaf_hdr_t;
+
+typedef struct xfs_attr_leaf_entry {	/* sorted on key, not name */
+	__be32	hashval;		/* hash value of name */
+	__be16	nameidx;		/* index into buffer of name/value */
+	__u8	flags;			/* LOCAL/ROOT/SECURE/INCOMPLETE flag */
+	__u8	pad2;			/* unused pad byte */
+} xfs_attr_leaf_entry_t;
+
+typedef struct xfs_attr_leaf_name_local {
+	__be16	valuelen;		/* number of bytes in value */
+	__u8	namelen;		/* length of name bytes */
+	__u8	nameval[1];		/* name/value bytes */
+} xfs_attr_leaf_name_local_t;
+
+typedef struct xfs_attr_leaf_name_remote {
+	__be32	valueblk;		/* block number of value bytes */
+	__be32	valuelen;		/* number of bytes in value */
+	__u8	namelen;		/* length of name bytes */
+	__u8	name[1];		/* name bytes */
+} xfs_attr_leaf_name_remote_t;
+
+typedef struct xfs_attr_leafblock {
+	xfs_attr_leaf_hdr_t	hdr;	/* constant-structure header block */
+	xfs_attr_leaf_entry_t	entries[1];	/* sorted on key, not name */
+	/*
+	 * The rest of the block contains the following structures after the
+	 * leaf entries, growing from the bottom up. The variables are never
+	 * referenced and definining them can actually make gcc optimize away
+	 * accesses to the 'entries' array above index 0 so don't do that.
+	 *
+	 * xfs_attr_leaf_name_local_t namelist;
+	 * xfs_attr_leaf_name_remote_t valuelist;
+	 */
+} xfs_attr_leafblock_t;
+
+/*
+ * CRC enabled leaf structures. Called "version 3" structures to match the
+ * version number of the directory and dablk structures for this feature, and
+ * attr2 is already taken by the variable inode attribute fork size feature.
+ */
+struct xfs_attr3_leaf_hdr {
+	struct xfs_da3_blkinfo	info;
+	__be16			count;
+	__be16			usedbytes;
+	__be16			firstused;
+	__u8			holes;
+	__u8			pad1;
+	struct xfs_attr_leaf_map freemap[XFS_ATTR_LEAF_MAPSIZE];
+	__be32			pad2;		/* 64 bit alignment */
+};
+
+#define XFS_ATTR3_LEAF_CRC_OFF	(offsetof(struct xfs_attr3_leaf_hdr, info.crc))
+
+struct xfs_attr3_leafblock {
+	struct xfs_attr3_leaf_hdr	hdr;
+	struct xfs_attr_leaf_entry	entries[1];
+
+	/*
+	 * The rest of the block contains the following structures after the
+	 * leaf entries, growing from the bottom up. The variables are never
+	 * referenced, the locations accessed purely from helper functions.
+	 *
+	 * struct xfs_attr_leaf_name_local
+	 * struct xfs_attr_leaf_name_remote
+	 */
+};
+
+/*
+ * Special value to represent fs block size in the leaf header firstused field.
+ * Only used when block size overflows the 2-bytes available on disk.
+ */
+#define XFS_ATTR3_LEAF_NULLOFF	0
+
+/*
+ * Flags used in the leaf_entry[i].flags field.
+ */
+#define	XFS_ATTR_LOCAL_BIT	0	/* attr is stored locally */
+#define	XFS_ATTR_ROOT_BIT	1	/* limit access to trusted attrs */
+#define	XFS_ATTR_SECURE_BIT	2	/* limit access to secure attrs */
+#define	XFS_ATTR_INCOMPLETE_BIT	7	/* attr in middle of create/delete */
+#define XFS_ATTR_LOCAL		(1u << XFS_ATTR_LOCAL_BIT)
+#define XFS_ATTR_ROOT		(1u << XFS_ATTR_ROOT_BIT)
+#define XFS_ATTR_SECURE		(1u << XFS_ATTR_SECURE_BIT)
+#define XFS_ATTR_INCOMPLETE	(1u << XFS_ATTR_INCOMPLETE_BIT)
+#define XFS_ATTR_NSP_ONDISK_MASK	(XFS_ATTR_ROOT | XFS_ATTR_SECURE)
+
+/*
+ * Alignment for namelist and valuelist entries (since they are mixed
+ * there can be only one alignment value)
+ */
+#define	XFS_ATTR_LEAF_NAME_ALIGN	((uint)sizeof(xfs_dablk_t))
+
+static inline int
+xfs_attr3_leaf_hdr_size(struct xfs_attr_leafblock *leafp)
+{
+	if (leafp->hdr.info.magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC))
+		return sizeof(struct xfs_attr3_leaf_hdr);
+	return sizeof(struct xfs_attr_leaf_hdr);
+}
+
+static inline struct xfs_attr_leaf_entry *
+xfs_attr3_leaf_entryp(xfs_attr_leafblock_t *leafp)
+{
+	if (leafp->hdr.info.magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC))
+		return &((struct xfs_attr3_leafblock *)leafp)->entries[0];
+	return &leafp->entries[0];
+}
+
+/*
+ * Cast typed pointers for "local" and "remote" name/value structs.
+ */
+static inline char *
+xfs_attr3_leaf_name(xfs_attr_leafblock_t *leafp, int idx)
+{
+	struct xfs_attr_leaf_entry *entries = xfs_attr3_leaf_entryp(leafp);
+
+	return &((char *)leafp)[be16_to_cpu(entries[idx].nameidx)];
+}
+
+static inline xfs_attr_leaf_name_remote_t *
+xfs_attr3_leaf_name_remote(xfs_attr_leafblock_t *leafp, int idx)
+{
+	return (xfs_attr_leaf_name_remote_t *)xfs_attr3_leaf_name(leafp, idx);
+}
+
+static inline xfs_attr_leaf_name_local_t *
+xfs_attr3_leaf_name_local(xfs_attr_leafblock_t *leafp, int idx)
+{
+	return (xfs_attr_leaf_name_local_t *)xfs_attr3_leaf_name(leafp, idx);
+}
+
+/*
+ * Calculate total bytes used (including trailing pad for alignment) for
+ * a "local" name/value structure, a "remote" name/value structure, and
+ * a pointer which might be either.
+ */
+static inline int xfs_attr_leaf_entsize_remote(int nlen)
+{
+	return round_up(sizeof(struct xfs_attr_leaf_name_remote) - 1 +
+			nlen, XFS_ATTR_LEAF_NAME_ALIGN);
+}
+
+static inline int xfs_attr_leaf_entsize_local(int nlen, int vlen)
+{
+	return round_up(sizeof(struct xfs_attr_leaf_name_local) - 1 +
+			nlen + vlen, XFS_ATTR_LEAF_NAME_ALIGN);
+}
+
+static inline int xfs_attr_leaf_entsize_local_max(int bsize)
+{
+	return (((bsize) >> 1) + ((bsize) >> 2));
+}
+
+
+
+/*
+ * Remote attribute block format definition
+ *
+ * There is one of these headers per filesystem block in a remote attribute.
+ * This is done to ensure there is a 1:1 mapping between the attribute value
+ * length and the number of blocks needed to store the attribute. This makes the
+ * verification of a buffer a little more complex, but greatly simplifies the
+ * allocation, reading and writing of these attributes as we don't have to guess
+ * the number of blocks needed to store the attribute data.
+ */
+#define XFS_ATTR3_RMT_MAGIC	0x5841524d	/* XARM */
+
+struct xfs_attr3_rmt_hdr {
+	__be32	rm_magic;
+	__be32	rm_offset;
+	__be32	rm_bytes;
+	__be32	rm_crc;
+	uuid_t	rm_uuid;
+	__be64	rm_owner;
+	__be64	rm_blkno;
+	__be64	rm_lsn;
+};
+
+#define XFS_ATTR3_RMT_CRC_OFF	offsetof(struct xfs_attr3_rmt_hdr, rm_crc)
+
+#define XFS_ATTR3_RMT_BUF_SPACE(mp, bufsize)	\
+	((bufsize) - (xfs_has_crc((mp)) ? \
+			sizeof(struct xfs_attr3_rmt_hdr) : 0))
+
+/* Number of bytes in a directory block. */
+static inline unsigned int xfs_dir2_dirblock_bytes(struct xfs_sb *sbp)
+{
+	return 1 << (sbp->sb_blocklog + sbp->sb_dirblklog);
+}
+
+xfs_failaddr_t xfs_da3_blkinfo_verify(struct xfs_buf *bp,
+				      struct xfs_da3_blkinfo *hdr3);
+
+#endif /* __XFS_DA_FORMAT_H__ */
diff --git a/fs/xfs/libxfs/xfs_defer.c b/fs/xfs/libxfs/xfs_defer.c
new file mode 100644
index 000000000..5a321b783
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_defer.c
@@ -0,0 +1,930 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2016 Oracle.  All Rights Reserved.
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_defer.h"
+#include "xfs_trans.h"
+#include "xfs_buf_item.h"
+#include "xfs_inode.h"
+#include "xfs_inode_item.h"
+#include "xfs_trace.h"
+#include "xfs_icache.h"
+#include "xfs_log.h"
+#include "xfs_rmap.h"
+#include "xfs_refcount.h"
+#include "xfs_bmap.h"
+#include "xfs_alloc.h"
+#include "xfs_buf.h"
+#include "xfs_da_format.h"
+#include "xfs_da_btree.h"
+#include "xfs_attr.h"
+
+static struct kmem_cache	*xfs_defer_pending_cache;
+
+/*
+ * Deferred Operations in XFS
+ *
+ * Due to the way locking rules work in XFS, certain transactions (block
+ * mapping and unmapping, typically) have permanent reservations so that
+ * we can roll the transaction to adhere to AG locking order rules and
+ * to unlock buffers between metadata updates.  Prior to rmap/reflink,
+ * the mapping code had a mechanism to perform these deferrals for
+ * extents that were going to be freed; this code makes that facility
+ * more generic.
+ *
+ * When adding the reverse mapping and reflink features, it became
+ * necessary to perform complex remapping multi-transactions to comply
+ * with AG locking order rules, and to be able to spread a single
+ * refcount update operation (an operation on an n-block extent can
+ * update as many as n records!) among multiple transactions.  XFS can
+ * roll a transaction to facilitate this, but using this facility
+ * requires us to log "intent" items in case log recovery needs to
+ * redo the operation, and to log "done" items to indicate that redo
+ * is not necessary.
+ *
+ * Deferred work is tracked in xfs_defer_pending items.  Each pending
+ * item tracks one type of deferred work.  Incoming work items (which
+ * have not yet had an intent logged) are attached to a pending item
+ * on the dop_intake list, where they wait for the caller to finish
+ * the deferred operations.
+ *
+ * Finishing a set of deferred operations is an involved process.  To
+ * start, we define "rolling a deferred-op transaction" as follows:
+ *
+ * > For each xfs_defer_pending item on the dop_intake list,
+ *   - Sort the work items in AG order.  XFS locking
+ *     order rules require us to lock buffers in AG order.
+ *   - Create a log intent item for that type.
+ *   - Attach it to the pending item.
+ *   - Move the pending item from the dop_intake list to the
+ *     dop_pending list.
+ * > Roll the transaction.
+ *
+ * NOTE: To avoid exceeding the transaction reservation, we limit the
+ * number of items that we attach to a given xfs_defer_pending.
+ *
+ * The actual finishing process looks like this:
+ *
+ * > For each xfs_defer_pending in the dop_pending list,
+ *   - Roll the deferred-op transaction as above.
+ *   - Create a log done item for that type, and attach it to the
+ *     log intent item.
+ *   - For each work item attached to the log intent item,
+ *     * Perform the described action.
+ *     * Attach the work item to the log done item.
+ *     * If the result of doing the work was -EAGAIN, ->finish work
+ *       wants a new transaction.  See the "Requesting a Fresh
+ *       Transaction while Finishing Deferred Work" section below for
+ *       details.
+ *
+ * The key here is that we must log an intent item for all pending
+ * work items every time we roll the transaction, and that we must log
+ * a done item as soon as the work is completed.  With this mechanism
+ * we can perform complex remapping operations, chaining intent items
+ * as needed.
+ *
+ * Requesting a Fresh Transaction while Finishing Deferred Work
+ *
+ * If ->finish_item decides that it needs a fresh transaction to
+ * finish the work, it must ask its caller (xfs_defer_finish) for a
+ * continuation.  The most likely cause of this circumstance are the
+ * refcount adjust functions deciding that they've logged enough items
+ * to be at risk of exceeding the transaction reservation.
+ *
+ * To get a fresh transaction, we want to log the existing log done
+ * item to prevent the log intent item from replaying, immediately log
+ * a new log intent item with the unfinished work items, roll the
+ * transaction, and re-call ->finish_item wherever it left off.  The
+ * log done item and the new log intent item must be in the same
+ * transaction or atomicity cannot be guaranteed; defer_finish ensures
+ * that this happens.
+ *
+ * This requires some coordination between ->finish_item and
+ * defer_finish.  Upon deciding to request a new transaction,
+ * ->finish_item should update the current work item to reflect the
+ * unfinished work.  Next, it should reset the log done item's list
+ * count to the number of items finished, and return -EAGAIN.
+ * defer_finish sees the -EAGAIN, logs the new log intent item
+ * with the remaining work items, and leaves the xfs_defer_pending
+ * item at the head of the dop_work queue.  Then it rolls the
+ * transaction and picks up processing where it left off.  It is
+ * required that ->finish_item must be careful to leave enough
+ * transaction reservation to fit the new log intent item.
+ *
+ * This is an example of remapping the extent (E, E+B) into file X at
+ * offset A and dealing with the extent (C, C+B) already being mapped
+ * there:
+ * +-------------------------------------------------+
+ * | Unmap file X startblock C offset A length B     | t0
+ * | Intent to reduce refcount for extent (C, B)     |
+ * | Intent to remove rmap (X, C, A, B)              |
+ * | Intent to free extent (D, 1) (bmbt block)       |
+ * | Intent to map (X, A, B) at startblock E         |
+ * +-------------------------------------------------+
+ * | Map file X startblock E offset A length B       | t1
+ * | Done mapping (X, E, A, B)                       |
+ * | Intent to increase refcount for extent (E, B)   |
+ * | Intent to add rmap (X, E, A, B)                 |
+ * +-------------------------------------------------+
+ * | Reduce refcount for extent (C, B)               | t2
+ * | Done reducing refcount for extent (C, 9)        |
+ * | Intent to reduce refcount for extent (C+9, B-9) |
+ * | (ran out of space after 9 refcount updates)     |
+ * +-------------------------------------------------+
+ * | Reduce refcount for extent (C+9, B+9)           | t3
+ * | Done reducing refcount for extent (C+9, B-9)    |
+ * | Increase refcount for extent (E, B)             |
+ * | Done increasing refcount for extent (E, B)      |
+ * | Intent to free extent (C, B)                    |
+ * | Intent to free extent (F, 1) (refcountbt block) |
+ * | Intent to remove rmap (F, 1, REFC)              |
+ * +-------------------------------------------------+
+ * | Remove rmap (X, C, A, B)                        | t4
+ * | Done removing rmap (X, C, A, B)                 |
+ * | Add rmap (X, E, A, B)                           |
+ * | Done adding rmap (X, E, A, B)                   |
+ * | Remove rmap (F, 1, REFC)                        |
+ * | Done removing rmap (F, 1, REFC)                 |
+ * +-------------------------------------------------+
+ * | Free extent (C, B)                              | t5
+ * | Done freeing extent (C, B)                      |
+ * | Free extent (D, 1)                              |
+ * | Done freeing extent (D, 1)                      |
+ * | Free extent (F, 1)                              |
+ * | Done freeing extent (F, 1)                      |
+ * +-------------------------------------------------+
+ *
+ * If we should crash before t2 commits, log recovery replays
+ * the following intent items:
+ *
+ * - Intent to reduce refcount for extent (C, B)
+ * - Intent to remove rmap (X, C, A, B)
+ * - Intent to free extent (D, 1) (bmbt block)
+ * - Intent to increase refcount for extent (E, B)
+ * - Intent to add rmap (X, E, A, B)
+ *
+ * In the process of recovering, it should also generate and take care
+ * of these intent items:
+ *
+ * - Intent to free extent (C, B)
+ * - Intent to free extent (F, 1) (refcountbt block)
+ * - Intent to remove rmap (F, 1, REFC)
+ *
+ * Note that the continuation requested between t2 and t3 is likely to
+ * reoccur.
+ */
+
+static const struct xfs_defer_op_type *defer_op_types[] = {
+	[XFS_DEFER_OPS_TYPE_BMAP]	= &xfs_bmap_update_defer_type,
+	[XFS_DEFER_OPS_TYPE_REFCOUNT]	= &xfs_refcount_update_defer_type,
+	[XFS_DEFER_OPS_TYPE_RMAP]	= &xfs_rmap_update_defer_type,
+	[XFS_DEFER_OPS_TYPE_FREE]	= &xfs_extent_free_defer_type,
+	[XFS_DEFER_OPS_TYPE_AGFL_FREE]	= &xfs_agfl_free_defer_type,
+	[XFS_DEFER_OPS_TYPE_ATTR]	= &xfs_attr_defer_type,
+};
+
+/*
+ * Ensure there's a log intent item associated with this deferred work item if
+ * the operation must be restarted on crash.  Returns 1 if there's a log item;
+ * 0 if there isn't; or a negative errno.
+ */
+static int
+xfs_defer_create_intent(
+	struct xfs_trans		*tp,
+	struct xfs_defer_pending	*dfp,
+	bool				sort)
+{
+	const struct xfs_defer_op_type	*ops = defer_op_types[dfp->dfp_type];
+	struct xfs_log_item		*lip;
+
+	if (dfp->dfp_intent)
+		return 1;
+
+	lip = ops->create_intent(tp, &dfp->dfp_work, dfp->dfp_count, sort);
+	if (!lip)
+		return 0;
+	if (IS_ERR(lip))
+		return PTR_ERR(lip);
+
+	dfp->dfp_intent = lip;
+	return 1;
+}
+
+/*
+ * For each pending item in the intake list, log its intent item and the
+ * associated extents, then add the entire intake list to the end of
+ * the pending list.
+ *
+ * Returns 1 if at least one log item was associated with the deferred work;
+ * 0 if there are no log items; or a negative errno.
+ */
+static int
+xfs_defer_create_intents(
+	struct xfs_trans		*tp)
+{
+	struct xfs_defer_pending	*dfp;
+	int				ret = 0;
+
+	list_for_each_entry(dfp, &tp->t_dfops, dfp_list) {
+		int			ret2;
+
+		trace_xfs_defer_create_intent(tp->t_mountp, dfp);
+		ret2 = xfs_defer_create_intent(tp, dfp, true);
+		if (ret2 < 0)
+			return ret2;
+		ret |= ret2;
+	}
+	return ret;
+}
+
+/* Abort all the intents that were committed. */
+STATIC void
+xfs_defer_trans_abort(
+	struct xfs_trans		*tp,
+	struct list_head		*dop_pending)
+{
+	struct xfs_defer_pending	*dfp;
+	const struct xfs_defer_op_type	*ops;
+
+	trace_xfs_defer_trans_abort(tp, _RET_IP_);
+
+	/* Abort intent items that don't have a done item. */
+	list_for_each_entry(dfp, dop_pending, dfp_list) {
+		ops = defer_op_types[dfp->dfp_type];
+		trace_xfs_defer_pending_abort(tp->t_mountp, dfp);
+		if (dfp->dfp_intent && !dfp->dfp_done) {
+			ops->abort_intent(dfp->dfp_intent);
+			dfp->dfp_intent = NULL;
+		}
+	}
+}
+
+/*
+ * Capture resources that the caller said not to release ("held") when the
+ * transaction commits.  Caller is responsible for zero-initializing @dres.
+ */
+static int
+xfs_defer_save_resources(
+	struct xfs_defer_resources	*dres,
+	struct xfs_trans		*tp)
+{
+	struct xfs_buf_log_item		*bli;
+	struct xfs_inode_log_item	*ili;
+	struct xfs_log_item		*lip;
+
+	BUILD_BUG_ON(NBBY * sizeof(dres->dr_ordered) < XFS_DEFER_OPS_NR_BUFS);
+
+	list_for_each_entry(lip, &tp->t_items, li_trans) {
+		switch (lip->li_type) {
+		case XFS_LI_BUF:
+			bli = container_of(lip, struct xfs_buf_log_item,
+					   bli_item);
+			if (bli->bli_flags & XFS_BLI_HOLD) {
+				if (dres->dr_bufs >= XFS_DEFER_OPS_NR_BUFS) {
+					ASSERT(0);
+					return -EFSCORRUPTED;
+				}
+				if (bli->bli_flags & XFS_BLI_ORDERED)
+					dres->dr_ordered |=
+							(1U << dres->dr_bufs);
+				else
+					xfs_trans_dirty_buf(tp, bli->bli_buf);
+				dres->dr_bp[dres->dr_bufs++] = bli->bli_buf;
+			}
+			break;
+		case XFS_LI_INODE:
+			ili = container_of(lip, struct xfs_inode_log_item,
+					   ili_item);
+			if (ili->ili_lock_flags == 0) {
+				if (dres->dr_inos >= XFS_DEFER_OPS_NR_INODES) {
+					ASSERT(0);
+					return -EFSCORRUPTED;
+				}
+				xfs_trans_log_inode(tp, ili->ili_inode,
+						    XFS_ILOG_CORE);
+				dres->dr_ip[dres->dr_inos++] = ili->ili_inode;
+			}
+			break;
+		default:
+			break;
+		}
+	}
+
+	return 0;
+}
+
+/* Attach the held resources to the transaction. */
+static void
+xfs_defer_restore_resources(
+	struct xfs_trans		*tp,
+	struct xfs_defer_resources	*dres)
+{
+	unsigned short			i;
+
+	/* Rejoin the joined inodes. */
+	for (i = 0; i < dres->dr_inos; i++)
+		xfs_trans_ijoin(tp, dres->dr_ip[i], 0);
+
+	/* Rejoin the buffers and dirty them so the log moves forward. */
+	for (i = 0; i < dres->dr_bufs; i++) {
+		xfs_trans_bjoin(tp, dres->dr_bp[i]);
+		if (dres->dr_ordered & (1U << i))
+			xfs_trans_ordered_buf(tp, dres->dr_bp[i]);
+		xfs_trans_bhold(tp, dres->dr_bp[i]);
+	}
+}
+
+/* Roll a transaction so we can do some deferred op processing. */
+STATIC int
+xfs_defer_trans_roll(
+	struct xfs_trans		**tpp)
+{
+	struct xfs_defer_resources	dres = { };
+	int				error;
+
+	error = xfs_defer_save_resources(&dres, *tpp);
+	if (error)
+		return error;
+
+	trace_xfs_defer_trans_roll(*tpp, _RET_IP_);
+
+	/*
+	 * Roll the transaction.  Rolling always given a new transaction (even
+	 * if committing the old one fails!) to hand back to the caller, so we
+	 * join the held resources to the new transaction so that we always
+	 * return with the held resources joined to @tpp, no matter what
+	 * happened.
+	 */
+	error = xfs_trans_roll(tpp);
+
+	xfs_defer_restore_resources(*tpp, &dres);
+
+	if (error)
+		trace_xfs_defer_trans_roll_error(*tpp, error);
+	return error;
+}
+
+/*
+ * Free up any items left in the list.
+ */
+static void
+xfs_defer_cancel_list(
+	struct xfs_mount		*mp,
+	struct list_head		*dop_list)
+{
+	struct xfs_defer_pending	*dfp;
+	struct xfs_defer_pending	*pli;
+	struct list_head		*pwi;
+	struct list_head		*n;
+	const struct xfs_defer_op_type	*ops;
+
+	/*
+	 * Free the pending items.  Caller should already have arranged
+	 * for the intent items to be released.
+	 */
+	list_for_each_entry_safe(dfp, pli, dop_list, dfp_list) {
+		ops = defer_op_types[dfp->dfp_type];
+		trace_xfs_defer_cancel_list(mp, dfp);
+		list_del(&dfp->dfp_list);
+		list_for_each_safe(pwi, n, &dfp->dfp_work) {
+			list_del(pwi);
+			dfp->dfp_count--;
+			ops->cancel_item(pwi);
+		}
+		ASSERT(dfp->dfp_count == 0);
+		kmem_cache_free(xfs_defer_pending_cache, dfp);
+	}
+}
+
+/*
+ * Prevent a log intent item from pinning the tail of the log by logging a
+ * done item to release the intent item; and then log a new intent item.
+ * The caller should provide a fresh transaction and roll it after we're done.
+ */
+static int
+xfs_defer_relog(
+	struct xfs_trans		**tpp,
+	struct list_head		*dfops)
+{
+	struct xlog			*log = (*tpp)->t_mountp->m_log;
+	struct xfs_defer_pending	*dfp;
+	xfs_lsn_t			threshold_lsn = NULLCOMMITLSN;
+
+
+	ASSERT((*tpp)->t_flags & XFS_TRANS_PERM_LOG_RES);
+
+	list_for_each_entry(dfp, dfops, dfp_list) {
+		/*
+		 * If the log intent item for this deferred op is not a part of
+		 * the current log checkpoint, relog the intent item to keep
+		 * the log tail moving forward.  We're ok with this being racy
+		 * because an incorrect decision means we'll be a little slower
+		 * at pushing the tail.
+		 */
+		if (dfp->dfp_intent == NULL ||
+		    xfs_log_item_in_current_chkpt(dfp->dfp_intent))
+			continue;
+
+		/*
+		 * Figure out where we need the tail to be in order to maintain
+		 * the minimum required free space in the log.  Only sample
+		 * the log threshold once per call.
+		 */
+		if (threshold_lsn == NULLCOMMITLSN) {
+			threshold_lsn = xlog_grant_push_threshold(log, 0);
+			if (threshold_lsn == NULLCOMMITLSN)
+				break;
+		}
+		if (XFS_LSN_CMP(dfp->dfp_intent->li_lsn, threshold_lsn) >= 0)
+			continue;
+
+		trace_xfs_defer_relog_intent((*tpp)->t_mountp, dfp);
+		XFS_STATS_INC((*tpp)->t_mountp, defer_relog);
+		dfp->dfp_intent = xfs_trans_item_relog(dfp->dfp_intent, *tpp);
+	}
+
+	if ((*tpp)->t_flags & XFS_TRANS_DIRTY)
+		return xfs_defer_trans_roll(tpp);
+	return 0;
+}
+
+/*
+ * Log an intent-done item for the first pending intent, and finish the work
+ * items.
+ */
+static int
+xfs_defer_finish_one(
+	struct xfs_trans		*tp,
+	struct xfs_defer_pending	*dfp)
+{
+	const struct xfs_defer_op_type	*ops = defer_op_types[dfp->dfp_type];
+	struct xfs_btree_cur		*state = NULL;
+	struct list_head		*li, *n;
+	int				error;
+
+	trace_xfs_defer_pending_finish(tp->t_mountp, dfp);
+
+	dfp->dfp_done = ops->create_done(tp, dfp->dfp_intent, dfp->dfp_count);
+	list_for_each_safe(li, n, &dfp->dfp_work) {
+		list_del(li);
+		dfp->dfp_count--;
+		error = ops->finish_item(tp, dfp->dfp_done, li, &state);
+		if (error == -EAGAIN) {
+			int		ret;
+
+			/*
+			 * Caller wants a fresh transaction; put the work item
+			 * back on the list and log a new log intent item to
+			 * replace the old one.  See "Requesting a Fresh
+			 * Transaction while Finishing Deferred Work" above.
+			 */
+			list_add(li, &dfp->dfp_work);
+			dfp->dfp_count++;
+			dfp->dfp_done = NULL;
+			dfp->dfp_intent = NULL;
+			ret = xfs_defer_create_intent(tp, dfp, false);
+			if (ret < 0)
+				error = ret;
+		}
+
+		if (error)
+			goto out;
+	}
+
+	/* Done with the dfp, free it. */
+	list_del(&dfp->dfp_list);
+	kmem_cache_free(xfs_defer_pending_cache, dfp);
+out:
+	if (ops->finish_cleanup)
+		ops->finish_cleanup(tp, state, error);
+	return error;
+}
+
+/*
+ * Finish all the pending work.  This involves logging intent items for
+ * any work items that wandered in since the last transaction roll (if
+ * one has even happened), rolling the transaction, and finishing the
+ * work items in the first item on the logged-and-pending list.
+ *
+ * If an inode is provided, relog it to the new transaction.
+ */
+int
+xfs_defer_finish_noroll(
+	struct xfs_trans		**tp)
+{
+	struct xfs_defer_pending	*dfp = NULL;
+	int				error = 0;
+	LIST_HEAD(dop_pending);
+
+	ASSERT((*tp)->t_flags & XFS_TRANS_PERM_LOG_RES);
+
+	trace_xfs_defer_finish(*tp, _RET_IP_);
+
+	/* Until we run out of pending work to finish... */
+	while (!list_empty(&dop_pending) || !list_empty(&(*tp)->t_dfops)) {
+		/*
+		 * Deferred items that are created in the process of finishing
+		 * other deferred work items should be queued at the head of
+		 * the pending list, which puts them ahead of the deferred work
+		 * that was created by the caller.  This keeps the number of
+		 * pending work items to a minimum, which decreases the amount
+		 * of time that any one intent item can stick around in memory,
+		 * pinning the log tail.
+		 */
+		int has_intents = xfs_defer_create_intents(*tp);
+
+		list_splice_init(&(*tp)->t_dfops, &dop_pending);
+
+		if (has_intents < 0) {
+			error = has_intents;
+			goto out_shutdown;
+		}
+		if (has_intents || dfp) {
+			error = xfs_defer_trans_roll(tp);
+			if (error)
+				goto out_shutdown;
+
+			/* Relog intent items to keep the log moving. */
+			error = xfs_defer_relog(tp, &dop_pending);
+			if (error)
+				goto out_shutdown;
+		}
+
+		dfp = list_first_entry(&dop_pending, struct xfs_defer_pending,
+				       dfp_list);
+		error = xfs_defer_finish_one(*tp, dfp);
+		if (error && error != -EAGAIN)
+			goto out_shutdown;
+	}
+
+	trace_xfs_defer_finish_done(*tp, _RET_IP_);
+	return 0;
+
+out_shutdown:
+	xfs_defer_trans_abort(*tp, &dop_pending);
+	xfs_force_shutdown((*tp)->t_mountp, SHUTDOWN_CORRUPT_INCORE);
+	trace_xfs_defer_finish_error(*tp, error);
+	xfs_defer_cancel_list((*tp)->t_mountp, &dop_pending);
+	xfs_defer_cancel(*tp);
+	return error;
+}
+
+int
+xfs_defer_finish(
+	struct xfs_trans	**tp)
+{
+	int			error;
+
+	/*
+	 * Finish and roll the transaction once more to avoid returning to the
+	 * caller with a dirty transaction.
+	 */
+	error = xfs_defer_finish_noroll(tp);
+	if (error)
+		return error;
+	if ((*tp)->t_flags & XFS_TRANS_DIRTY) {
+		error = xfs_defer_trans_roll(tp);
+		if (error) {
+			xfs_force_shutdown((*tp)->t_mountp,
+					   SHUTDOWN_CORRUPT_INCORE);
+			return error;
+		}
+	}
+
+	/* Reset LOWMODE now that we've finished all the dfops. */
+	ASSERT(list_empty(&(*tp)->t_dfops));
+	(*tp)->t_flags &= ~XFS_TRANS_LOWMODE;
+	return 0;
+}
+
+void
+xfs_defer_cancel(
+	struct xfs_trans	*tp)
+{
+	struct xfs_mount	*mp = tp->t_mountp;
+
+	trace_xfs_defer_cancel(tp, _RET_IP_);
+	xfs_defer_cancel_list(mp, &tp->t_dfops);
+}
+
+/* Add an item for later deferred processing. */
+void
+xfs_defer_add(
+	struct xfs_trans		*tp,
+	enum xfs_defer_ops_type		type,
+	struct list_head		*li)
+{
+	struct xfs_defer_pending	*dfp = NULL;
+	const struct xfs_defer_op_type	*ops;
+
+	ASSERT(tp->t_flags & XFS_TRANS_PERM_LOG_RES);
+	BUILD_BUG_ON(ARRAY_SIZE(defer_op_types) != XFS_DEFER_OPS_TYPE_MAX);
+
+	/*
+	 * Add the item to a pending item at the end of the intake list.
+	 * If the last pending item has the same type, reuse it.  Else,
+	 * create a new pending item at the end of the intake list.
+	 */
+	if (!list_empty(&tp->t_dfops)) {
+		dfp = list_last_entry(&tp->t_dfops,
+				struct xfs_defer_pending, dfp_list);
+		ops = defer_op_types[dfp->dfp_type];
+		if (dfp->dfp_type != type ||
+		    (ops->max_items && dfp->dfp_count >= ops->max_items))
+			dfp = NULL;
+	}
+	if (!dfp) {
+		dfp = kmem_cache_zalloc(xfs_defer_pending_cache,
+				GFP_NOFS | __GFP_NOFAIL);
+		dfp->dfp_type = type;
+		dfp->dfp_intent = NULL;
+		dfp->dfp_done = NULL;
+		dfp->dfp_count = 0;
+		INIT_LIST_HEAD(&dfp->dfp_work);
+		list_add_tail(&dfp->dfp_list, &tp->t_dfops);
+	}
+
+	list_add_tail(li, &dfp->dfp_work);
+	dfp->dfp_count++;
+}
+
+/*
+ * Move deferred ops from one transaction to another and reset the source to
+ * initial state. This is primarily used to carry state forward across
+ * transaction rolls with pending dfops.
+ */
+void
+xfs_defer_move(
+	struct xfs_trans	*dtp,
+	struct xfs_trans	*stp)
+{
+	list_splice_init(&stp->t_dfops, &dtp->t_dfops);
+
+	/*
+	 * Low free space mode was historically controlled by a dfops field.
+	 * This meant that low mode state potentially carried across multiple
+	 * transaction rolls. Transfer low mode on a dfops move to preserve
+	 * that behavior.
+	 */
+	dtp->t_flags |= (stp->t_flags & XFS_TRANS_LOWMODE);
+	stp->t_flags &= ~XFS_TRANS_LOWMODE;
+}
+
+/*
+ * Prepare a chain of fresh deferred ops work items to be completed later.  Log
+ * recovery requires the ability to put off until later the actual finishing
+ * work so that it can process unfinished items recovered from the log in
+ * correct order.
+ *
+ * Create and log intent items for all the work that we're capturing so that we
+ * can be assured that the items will get replayed if the system goes down
+ * before log recovery gets a chance to finish the work it put off.  The entire
+ * deferred ops state is transferred to the capture structure and the
+ * transaction is then ready for the caller to commit it.  If there are no
+ * intent items to capture, this function returns NULL.
+ *
+ * If capture_ip is not NULL, the capture structure will obtain an extra
+ * reference to the inode.
+ */
+static struct xfs_defer_capture *
+xfs_defer_ops_capture(
+	struct xfs_trans		*tp)
+{
+	struct xfs_defer_capture	*dfc;
+	unsigned short			i;
+	int				error;
+
+	if (list_empty(&tp->t_dfops))
+		return NULL;
+
+	error = xfs_defer_create_intents(tp);
+	if (error < 0)
+		return ERR_PTR(error);
+
+	/* Create an object to capture the defer ops. */
+	dfc = kmem_zalloc(sizeof(*dfc), KM_NOFS);
+	INIT_LIST_HEAD(&dfc->dfc_list);
+	INIT_LIST_HEAD(&dfc->dfc_dfops);
+
+	/* Move the dfops chain and transaction state to the capture struct. */
+	list_splice_init(&tp->t_dfops, &dfc->dfc_dfops);
+	dfc->dfc_tpflags = tp->t_flags & XFS_TRANS_LOWMODE;
+	tp->t_flags &= ~XFS_TRANS_LOWMODE;
+
+	/* Capture the remaining block reservations along with the dfops. */
+	dfc->dfc_blkres = tp->t_blk_res - tp->t_blk_res_used;
+	dfc->dfc_rtxres = tp->t_rtx_res - tp->t_rtx_res_used;
+
+	/* Preserve the log reservation size. */
+	dfc->dfc_logres = tp->t_log_res;
+
+	error = xfs_defer_save_resources(&dfc->dfc_held, tp);
+	if (error) {
+		/*
+		 * Resource capture should never fail, but if it does, we
+		 * still have to shut down the log and release things
+		 * properly.
+		 */
+		xfs_force_shutdown(tp->t_mountp, SHUTDOWN_CORRUPT_INCORE);
+	}
+
+	/*
+	 * Grab extra references to the inodes and buffers because callers are
+	 * expected to release their held references after we commit the
+	 * transaction.
+	 */
+	for (i = 0; i < dfc->dfc_held.dr_inos; i++) {
+		ASSERT(xfs_isilocked(dfc->dfc_held.dr_ip[i], XFS_ILOCK_EXCL));
+		ihold(VFS_I(dfc->dfc_held.dr_ip[i]));
+	}
+
+	for (i = 0; i < dfc->dfc_held.dr_bufs; i++)
+		xfs_buf_hold(dfc->dfc_held.dr_bp[i]);
+
+	return dfc;
+}
+
+/* Release all resources that we used to capture deferred ops. */
+void
+xfs_defer_ops_capture_free(
+	struct xfs_mount		*mp,
+	struct xfs_defer_capture	*dfc)
+{
+	unsigned short			i;
+
+	xfs_defer_cancel_list(mp, &dfc->dfc_dfops);
+
+	for (i = 0; i < dfc->dfc_held.dr_bufs; i++)
+		xfs_buf_relse(dfc->dfc_held.dr_bp[i]);
+
+	for (i = 0; i < dfc->dfc_held.dr_inos; i++)
+		xfs_irele(dfc->dfc_held.dr_ip[i]);
+
+	kmem_free(dfc);
+}
+
+/*
+ * Capture any deferred ops and commit the transaction.  This is the last step
+ * needed to finish a log intent item that we recovered from the log.  If any
+ * of the deferred ops operate on an inode, the caller must pass in that inode
+ * so that the reference can be transferred to the capture structure.  The
+ * caller must hold ILOCK_EXCL on the inode, and must unlock it before calling
+ * xfs_defer_ops_continue.
+ */
+int
+xfs_defer_ops_capture_and_commit(
+	struct xfs_trans		*tp,
+	struct list_head		*capture_list)
+{
+	struct xfs_mount		*mp = tp->t_mountp;
+	struct xfs_defer_capture	*dfc;
+	int				error;
+
+	/* If we don't capture anything, commit transaction and exit. */
+	dfc = xfs_defer_ops_capture(tp);
+	if (IS_ERR(dfc)) {
+		xfs_trans_cancel(tp);
+		return PTR_ERR(dfc);
+	}
+	if (!dfc)
+		return xfs_trans_commit(tp);
+
+	/* Commit the transaction and add the capture structure to the list. */
+	error = xfs_trans_commit(tp);
+	if (error) {
+		xfs_defer_ops_capture_free(mp, dfc);
+		return error;
+	}
+
+	list_add_tail(&dfc->dfc_list, capture_list);
+	return 0;
+}
+
+/*
+ * Attach a chain of captured deferred ops to a new transaction and free the
+ * capture structure.  If an inode was captured, it will be passed back to the
+ * caller with ILOCK_EXCL held and joined to the transaction with lockflags==0.
+ * The caller now owns the inode reference.
+ */
+void
+xfs_defer_ops_continue(
+	struct xfs_defer_capture	*dfc,
+	struct xfs_trans		*tp,
+	struct xfs_defer_resources	*dres)
+{
+	unsigned int			i;
+
+	ASSERT(tp->t_flags & XFS_TRANS_PERM_LOG_RES);
+	ASSERT(!(tp->t_flags & XFS_TRANS_DIRTY));
+
+	/* Lock the captured resources to the new transaction. */
+	if (dfc->dfc_held.dr_inos == 2)
+		xfs_lock_two_inodes(dfc->dfc_held.dr_ip[0], XFS_ILOCK_EXCL,
+				    dfc->dfc_held.dr_ip[1], XFS_ILOCK_EXCL);
+	else if (dfc->dfc_held.dr_inos == 1)
+		xfs_ilock(dfc->dfc_held.dr_ip[0], XFS_ILOCK_EXCL);
+
+	for (i = 0; i < dfc->dfc_held.dr_bufs; i++)
+		xfs_buf_lock(dfc->dfc_held.dr_bp[i]);
+
+	/* Join the captured resources to the new transaction. */
+	xfs_defer_restore_resources(tp, &dfc->dfc_held);
+	memcpy(dres, &dfc->dfc_held, sizeof(struct xfs_defer_resources));
+	dres->dr_bufs = 0;
+
+	/* Move captured dfops chain and state to the transaction. */
+	list_splice_init(&dfc->dfc_dfops, &tp->t_dfops);
+	tp->t_flags |= dfc->dfc_tpflags;
+
+	kmem_free(dfc);
+}
+
+/* Release the resources captured and continued during recovery. */
+void
+xfs_defer_resources_rele(
+	struct xfs_defer_resources	*dres)
+{
+	unsigned short			i;
+
+	for (i = 0; i < dres->dr_inos; i++) {
+		xfs_iunlock(dres->dr_ip[i], XFS_ILOCK_EXCL);
+		xfs_irele(dres->dr_ip[i]);
+		dres->dr_ip[i] = NULL;
+	}
+
+	for (i = 0; i < dres->dr_bufs; i++) {
+		xfs_buf_relse(dres->dr_bp[i]);
+		dres->dr_bp[i] = NULL;
+	}
+
+	dres->dr_inos = 0;
+	dres->dr_bufs = 0;
+	dres->dr_ordered = 0;
+}
+
+static inline int __init
+xfs_defer_init_cache(void)
+{
+	xfs_defer_pending_cache = kmem_cache_create("xfs_defer_pending",
+			sizeof(struct xfs_defer_pending),
+			0, 0, NULL);
+
+	return xfs_defer_pending_cache != NULL ? 0 : -ENOMEM;
+}
+
+static inline void
+xfs_defer_destroy_cache(void)
+{
+	kmem_cache_destroy(xfs_defer_pending_cache);
+	xfs_defer_pending_cache = NULL;
+}
+
+/* Set up caches for deferred work items. */
+int __init
+xfs_defer_init_item_caches(void)
+{
+	int				error;
+
+	error = xfs_defer_init_cache();
+	if (error)
+		return error;
+	error = xfs_rmap_intent_init_cache();
+	if (error)
+		goto err;
+	error = xfs_refcount_intent_init_cache();
+	if (error)
+		goto err;
+	error = xfs_bmap_intent_init_cache();
+	if (error)
+		goto err;
+	error = xfs_extfree_intent_init_cache();
+	if (error)
+		goto err;
+	error = xfs_attr_intent_init_cache();
+	if (error)
+		goto err;
+	return 0;
+err:
+	xfs_defer_destroy_item_caches();
+	return error;
+}
+
+/* Destroy all the deferred work item caches, if they've been allocated. */
+void
+xfs_defer_destroy_item_caches(void)
+{
+	xfs_attr_intent_destroy_cache();
+	xfs_extfree_intent_destroy_cache();
+	xfs_bmap_intent_destroy_cache();
+	xfs_refcount_intent_destroy_cache();
+	xfs_rmap_intent_destroy_cache();
+	xfs_defer_destroy_cache();
+}
diff --git a/fs/xfs/libxfs/xfs_defer.h b/fs/xfs/libxfs/xfs_defer.h
new file mode 100644
index 000000000..114a3a493
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_defer.h
@@ -0,0 +1,131 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Copyright (C) 2016 Oracle.  All Rights Reserved.
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ */
+#ifndef __XFS_DEFER_H__
+#define	__XFS_DEFER_H__
+
+struct xfs_btree_cur;
+struct xfs_defer_op_type;
+struct xfs_defer_capture;
+
+/*
+ * Header for deferred operation list.
+ */
+enum xfs_defer_ops_type {
+	XFS_DEFER_OPS_TYPE_BMAP,
+	XFS_DEFER_OPS_TYPE_REFCOUNT,
+	XFS_DEFER_OPS_TYPE_RMAP,
+	XFS_DEFER_OPS_TYPE_FREE,
+	XFS_DEFER_OPS_TYPE_AGFL_FREE,
+	XFS_DEFER_OPS_TYPE_ATTR,
+	XFS_DEFER_OPS_TYPE_MAX,
+};
+
+/*
+ * Save a log intent item and a list of extents, so that we can replay
+ * whatever action had to happen to the extent list and file the log done
+ * item.
+ */
+struct xfs_defer_pending {
+	struct list_head		dfp_list;	/* pending items */
+	struct list_head		dfp_work;	/* work items */
+	struct xfs_log_item		*dfp_intent;	/* log intent item */
+	struct xfs_log_item		*dfp_done;	/* log done item */
+	unsigned int			dfp_count;	/* # extent items */
+	enum xfs_defer_ops_type		dfp_type;
+};
+
+void xfs_defer_add(struct xfs_trans *tp, enum xfs_defer_ops_type type,
+		struct list_head *h);
+int xfs_defer_finish_noroll(struct xfs_trans **tp);
+int xfs_defer_finish(struct xfs_trans **tp);
+void xfs_defer_cancel(struct xfs_trans *);
+void xfs_defer_move(struct xfs_trans *dtp, struct xfs_trans *stp);
+
+/* Description of a deferred type. */
+struct xfs_defer_op_type {
+	struct xfs_log_item *(*create_intent)(struct xfs_trans *tp,
+			struct list_head *items, unsigned int count, bool sort);
+	void (*abort_intent)(struct xfs_log_item *intent);
+	struct xfs_log_item *(*create_done)(struct xfs_trans *tp,
+			struct xfs_log_item *intent, unsigned int count);
+	int (*finish_item)(struct xfs_trans *tp, struct xfs_log_item *done,
+			struct list_head *item, struct xfs_btree_cur **state);
+	void (*finish_cleanup)(struct xfs_trans *tp,
+			struct xfs_btree_cur *state, int error);
+	void (*cancel_item)(struct list_head *item);
+	unsigned int		max_items;
+};
+
+extern const struct xfs_defer_op_type xfs_bmap_update_defer_type;
+extern const struct xfs_defer_op_type xfs_refcount_update_defer_type;
+extern const struct xfs_defer_op_type xfs_rmap_update_defer_type;
+extern const struct xfs_defer_op_type xfs_extent_free_defer_type;
+extern const struct xfs_defer_op_type xfs_agfl_free_defer_type;
+extern const struct xfs_defer_op_type xfs_attr_defer_type;
+
+
+/*
+ * Deferred operation item relogging limits.
+ */
+#define XFS_DEFER_OPS_NR_INODES	2	/* join up to two inodes */
+#define XFS_DEFER_OPS_NR_BUFS	2	/* join up to two buffers */
+
+/* Resources that must be held across a transaction roll. */
+struct xfs_defer_resources {
+	/* held buffers */
+	struct xfs_buf		*dr_bp[XFS_DEFER_OPS_NR_BUFS];
+
+	/* inodes with no unlock flags */
+	struct xfs_inode	*dr_ip[XFS_DEFER_OPS_NR_INODES];
+
+	/* number of held buffers */
+	unsigned short		dr_bufs;
+
+	/* bitmap of ordered buffers */
+	unsigned short		dr_ordered;
+
+	/* number of held inodes */
+	unsigned short		dr_inos;
+};
+
+/*
+ * This structure enables a dfops user to detach the chain of deferred
+ * operations from a transaction so that they can be continued later.
+ */
+struct xfs_defer_capture {
+	/* List of other capture structures. */
+	struct list_head	dfc_list;
+
+	/* Deferred ops state saved from the transaction. */
+	struct list_head	dfc_dfops;
+	unsigned int		dfc_tpflags;
+
+	/* Block reservations for the data and rt devices. */
+	unsigned int		dfc_blkres;
+	unsigned int		dfc_rtxres;
+
+	/* Log reservation saved from the transaction. */
+	unsigned int		dfc_logres;
+
+	struct xfs_defer_resources dfc_held;
+};
+
+/*
+ * Functions to capture a chain of deferred operations and continue them later.
+ * This doesn't normally happen except log recovery.
+ */
+int xfs_defer_ops_capture_and_commit(struct xfs_trans *tp,
+		struct list_head *capture_list);
+void xfs_defer_ops_continue(struct xfs_defer_capture *d, struct xfs_trans *tp,
+		struct xfs_defer_resources *dres);
+void xfs_defer_ops_capture_free(struct xfs_mount *mp,
+		struct xfs_defer_capture *d);
+void xfs_defer_resources_rele(struct xfs_defer_resources *dres);
+
+int __init xfs_defer_init_item_caches(void);
+void xfs_defer_destroy_item_caches(void);
+
+#endif /* __XFS_DEFER_H__ */
diff --git a/fs/xfs/libxfs/xfs_dir2.c b/fs/xfs/libxfs/xfs_dir2.c
new file mode 100644
index 000000000..92bac3373
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_dir2.c
@@ -0,0 +1,769 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_inode.h"
+#include "xfs_trans.h"
+#include "xfs_bmap.h"
+#include "xfs_dir2.h"
+#include "xfs_dir2_priv.h"
+#include "xfs_errortag.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+
+const struct xfs_name xfs_name_dotdot = {
+	.name	= (const unsigned char *)"..",
+	.len	= 2,
+	.type	= XFS_DIR3_FT_DIR,
+};
+
+/*
+ * Convert inode mode to directory entry filetype
+ */
+unsigned char
+xfs_mode_to_ftype(
+	int		mode)
+{
+	switch (mode & S_IFMT) {
+	case S_IFREG:
+		return XFS_DIR3_FT_REG_FILE;
+	case S_IFDIR:
+		return XFS_DIR3_FT_DIR;
+	case S_IFCHR:
+		return XFS_DIR3_FT_CHRDEV;
+	case S_IFBLK:
+		return XFS_DIR3_FT_BLKDEV;
+	case S_IFIFO:
+		return XFS_DIR3_FT_FIFO;
+	case S_IFSOCK:
+		return XFS_DIR3_FT_SOCK;
+	case S_IFLNK:
+		return XFS_DIR3_FT_SYMLINK;
+	default:
+		return XFS_DIR3_FT_UNKNOWN;
+	}
+}
+
+/*
+ * ASCII case-insensitive (ie. A-Z) support for directories that was
+ * used in IRIX.
+ */
+xfs_dahash_t
+xfs_ascii_ci_hashname(
+	const struct xfs_name	*name)
+{
+	xfs_dahash_t		hash;
+	int			i;
+
+	for (i = 0, hash = 0; i < name->len; i++)
+		hash = tolower(name->name[i]) ^ rol32(hash, 7);
+
+	return hash;
+}
+
+enum xfs_dacmp
+xfs_ascii_ci_compname(
+	struct xfs_da_args	*args,
+	const unsigned char	*name,
+	int			len)
+{
+	enum xfs_dacmp		result;
+	int			i;
+
+	if (args->namelen != len)
+		return XFS_CMP_DIFFERENT;
+
+	result = XFS_CMP_EXACT;
+	for (i = 0; i < len; i++) {
+		if (args->name[i] == name[i])
+			continue;
+		if (tolower(args->name[i]) != tolower(name[i]))
+			return XFS_CMP_DIFFERENT;
+		result = XFS_CMP_CASE;
+	}
+
+	return result;
+}
+
+int
+xfs_da_mount(
+	struct xfs_mount	*mp)
+{
+	struct xfs_da_geometry	*dageo;
+
+
+	ASSERT(mp->m_sb.sb_versionnum & XFS_SB_VERSION_DIRV2BIT);
+	ASSERT(xfs_dir2_dirblock_bytes(&mp->m_sb) <= XFS_MAX_BLOCKSIZE);
+
+	mp->m_dir_geo = kmem_zalloc(sizeof(struct xfs_da_geometry),
+				    KM_MAYFAIL);
+	mp->m_attr_geo = kmem_zalloc(sizeof(struct xfs_da_geometry),
+				     KM_MAYFAIL);
+	if (!mp->m_dir_geo || !mp->m_attr_geo) {
+		kmem_free(mp->m_dir_geo);
+		kmem_free(mp->m_attr_geo);
+		return -ENOMEM;
+	}
+
+	/* set up directory geometry */
+	dageo = mp->m_dir_geo;
+	dageo->blklog = mp->m_sb.sb_blocklog + mp->m_sb.sb_dirblklog;
+	dageo->fsblog = mp->m_sb.sb_blocklog;
+	dageo->blksize = xfs_dir2_dirblock_bytes(&mp->m_sb);
+	dageo->fsbcount = 1 << mp->m_sb.sb_dirblklog;
+	if (xfs_has_crc(mp)) {
+		dageo->node_hdr_size = sizeof(struct xfs_da3_node_hdr);
+		dageo->leaf_hdr_size = sizeof(struct xfs_dir3_leaf_hdr);
+		dageo->free_hdr_size = sizeof(struct xfs_dir3_free_hdr);
+		dageo->data_entry_offset =
+				sizeof(struct xfs_dir3_data_hdr);
+	} else {
+		dageo->node_hdr_size = sizeof(struct xfs_da_node_hdr);
+		dageo->leaf_hdr_size = sizeof(struct xfs_dir2_leaf_hdr);
+		dageo->free_hdr_size = sizeof(struct xfs_dir2_free_hdr);
+		dageo->data_entry_offset =
+				sizeof(struct xfs_dir2_data_hdr);
+	}
+	dageo->leaf_max_ents = (dageo->blksize - dageo->leaf_hdr_size) /
+			sizeof(struct xfs_dir2_leaf_entry);
+	dageo->free_max_bests = (dageo->blksize - dageo->free_hdr_size) /
+			sizeof(xfs_dir2_data_off_t);
+
+	dageo->data_first_offset = dageo->data_entry_offset +
+			xfs_dir2_data_entsize(mp, 1) +
+			xfs_dir2_data_entsize(mp, 2);
+
+	/*
+	 * Now we've set up the block conversion variables, we can calculate the
+	 * segment block constants using the geometry structure.
+	 */
+	dageo->datablk = xfs_dir2_byte_to_da(dageo, XFS_DIR2_DATA_OFFSET);
+	dageo->leafblk = xfs_dir2_byte_to_da(dageo, XFS_DIR2_LEAF_OFFSET);
+	dageo->freeblk = xfs_dir2_byte_to_da(dageo, XFS_DIR2_FREE_OFFSET);
+	dageo->node_ents = (dageo->blksize - dageo->node_hdr_size) /
+				(uint)sizeof(xfs_da_node_entry_t);
+	dageo->max_extents = (XFS_DIR2_MAX_SPACES * XFS_DIR2_SPACE_SIZE) >>
+					mp->m_sb.sb_blocklog;
+	dageo->magicpct = (dageo->blksize * 37) / 100;
+
+	/* set up attribute geometry - single fsb only */
+	dageo = mp->m_attr_geo;
+	dageo->blklog = mp->m_sb.sb_blocklog;
+	dageo->fsblog = mp->m_sb.sb_blocklog;
+	dageo->blksize = 1 << dageo->blklog;
+	dageo->fsbcount = 1;
+	dageo->node_hdr_size = mp->m_dir_geo->node_hdr_size;
+	dageo->node_ents = (dageo->blksize - dageo->node_hdr_size) /
+				(uint)sizeof(xfs_da_node_entry_t);
+
+	if (xfs_has_large_extent_counts(mp))
+		dageo->max_extents = XFS_MAX_EXTCNT_ATTR_FORK_LARGE;
+	else
+		dageo->max_extents = XFS_MAX_EXTCNT_ATTR_FORK_SMALL;
+
+	dageo->magicpct = (dageo->blksize * 37) / 100;
+	return 0;
+}
+
+void
+xfs_da_unmount(
+	struct xfs_mount	*mp)
+{
+	kmem_free(mp->m_dir_geo);
+	kmem_free(mp->m_attr_geo);
+}
+
+/*
+ * Return 1 if directory contains only "." and "..".
+ */
+int
+xfs_dir_isempty(
+	xfs_inode_t	*dp)
+{
+	xfs_dir2_sf_hdr_t	*sfp;
+
+	ASSERT(S_ISDIR(VFS_I(dp)->i_mode));
+	if (dp->i_disk_size == 0)	/* might happen during shutdown. */
+		return 1;
+	if (dp->i_disk_size > xfs_inode_data_fork_size(dp))
+		return 0;
+	sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	return !sfp->count;
+}
+
+/*
+ * Validate a given inode number.
+ */
+int
+xfs_dir_ino_validate(
+	xfs_mount_t	*mp,
+	xfs_ino_t	ino)
+{
+	bool		ino_ok = xfs_verify_dir_ino(mp, ino);
+
+	if (XFS_IS_CORRUPT(mp, !ino_ok) ||
+	    XFS_TEST_ERROR(false, mp, XFS_ERRTAG_DIR_INO_VALIDATE)) {
+		xfs_warn(mp, "Invalid inode number 0x%Lx",
+				(unsigned long long) ino);
+		return -EFSCORRUPTED;
+	}
+	return 0;
+}
+
+/*
+ * Initialize a directory with its "." and ".." entries.
+ */
+int
+xfs_dir_init(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*dp,
+	xfs_inode_t	*pdp)
+{
+	struct xfs_da_args *args;
+	int		error;
+
+	ASSERT(S_ISDIR(VFS_I(dp)->i_mode));
+	error = xfs_dir_ino_validate(tp->t_mountp, pdp->i_ino);
+	if (error)
+		return error;
+
+	args = kmem_zalloc(sizeof(*args), KM_NOFS);
+	if (!args)
+		return -ENOMEM;
+
+	args->geo = dp->i_mount->m_dir_geo;
+	args->dp = dp;
+	args->trans = tp;
+	error = xfs_dir2_sf_create(args, pdp->i_ino);
+	kmem_free(args);
+	return error;
+}
+
+/*
+ * Enter a name in a directory, or check for available space.
+ * If inum is 0, only the available space test is performed.
+ */
+int
+xfs_dir_createname(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*dp,
+	const struct xfs_name	*name,
+	xfs_ino_t		inum,		/* new entry inode number */
+	xfs_extlen_t		total)		/* bmap's total block count */
+{
+	struct xfs_da_args	*args;
+	int			rval;
+	bool			v;
+
+	ASSERT(S_ISDIR(VFS_I(dp)->i_mode));
+
+	if (inum) {
+		rval = xfs_dir_ino_validate(tp->t_mountp, inum);
+		if (rval)
+			return rval;
+		XFS_STATS_INC(dp->i_mount, xs_dir_create);
+	}
+
+	args = kmem_zalloc(sizeof(*args), KM_NOFS);
+	if (!args)
+		return -ENOMEM;
+
+	args->geo = dp->i_mount->m_dir_geo;
+	args->name = name->name;
+	args->namelen = name->len;
+	args->filetype = name->type;
+	args->hashval = xfs_dir2_hashname(dp->i_mount, name);
+	args->inumber = inum;
+	args->dp = dp;
+	args->total = total;
+	args->whichfork = XFS_DATA_FORK;
+	args->trans = tp;
+	args->op_flags = XFS_DA_OP_ADDNAME | XFS_DA_OP_OKNOENT;
+	if (!inum)
+		args->op_flags |= XFS_DA_OP_JUSTCHECK;
+
+	if (dp->i_df.if_format == XFS_DINODE_FMT_LOCAL) {
+		rval = xfs_dir2_sf_addname(args);
+		goto out_free;
+	}
+
+	rval = xfs_dir2_isblock(args, &v);
+	if (rval)
+		goto out_free;
+	if (v) {
+		rval = xfs_dir2_block_addname(args);
+		goto out_free;
+	}
+
+	rval = xfs_dir2_isleaf(args, &v);
+	if (rval)
+		goto out_free;
+	if (v)
+		rval = xfs_dir2_leaf_addname(args);
+	else
+		rval = xfs_dir2_node_addname(args);
+
+out_free:
+	kmem_free(args);
+	return rval;
+}
+
+/*
+ * If doing a CI lookup and case-insensitive match, dup actual name into
+ * args.value. Return EEXIST for success (ie. name found) or an error.
+ */
+int
+xfs_dir_cilookup_result(
+	struct xfs_da_args *args,
+	const unsigned char *name,
+	int		len)
+{
+	if (args->cmpresult == XFS_CMP_DIFFERENT)
+		return -ENOENT;
+	if (args->cmpresult != XFS_CMP_CASE ||
+					!(args->op_flags & XFS_DA_OP_CILOOKUP))
+		return -EEXIST;
+
+	args->value = kmem_alloc(len, KM_NOFS | KM_MAYFAIL);
+	if (!args->value)
+		return -ENOMEM;
+
+	memcpy(args->value, name, len);
+	args->valuelen = len;
+	return -EEXIST;
+}
+
+/*
+ * Lookup a name in a directory, give back the inode number.
+ * If ci_name is not NULL, returns the actual name in ci_name if it differs
+ * to name, or ci_name->name is set to NULL for an exact match.
+ */
+
+int
+xfs_dir_lookup(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*dp,
+	const struct xfs_name	*name,
+	xfs_ino_t		*inum,	  /* out: inode number */
+	struct xfs_name		*ci_name) /* out: actual name if CI match */
+{
+	struct xfs_da_args	*args;
+	int			rval;
+	bool			v;
+	int			lock_mode;
+
+	ASSERT(S_ISDIR(VFS_I(dp)->i_mode));
+	XFS_STATS_INC(dp->i_mount, xs_dir_lookup);
+
+	/*
+	 * We need to use KM_NOFS here so that lockdep will not throw false
+	 * positive deadlock warnings on a non-transactional lookup path. It is
+	 * safe to recurse into inode recalim in that case, but lockdep can't
+	 * easily be taught about it. Hence KM_NOFS avoids having to add more
+	 * lockdep Doing this avoids having to add a bunch of lockdep class
+	 * annotations into the reclaim path for the ilock.
+	 */
+	args = kmem_zalloc(sizeof(*args), KM_NOFS);
+	args->geo = dp->i_mount->m_dir_geo;
+	args->name = name->name;
+	args->namelen = name->len;
+	args->filetype = name->type;
+	args->hashval = xfs_dir2_hashname(dp->i_mount, name);
+	args->dp = dp;
+	args->whichfork = XFS_DATA_FORK;
+	args->trans = tp;
+	args->op_flags = XFS_DA_OP_OKNOENT;
+	if (ci_name)
+		args->op_flags |= XFS_DA_OP_CILOOKUP;
+
+	lock_mode = xfs_ilock_data_map_shared(dp);
+	if (dp->i_df.if_format == XFS_DINODE_FMT_LOCAL) {
+		rval = xfs_dir2_sf_lookup(args);
+		goto out_check_rval;
+	}
+
+	rval = xfs_dir2_isblock(args, &v);
+	if (rval)
+		goto out_free;
+	if (v) {
+		rval = xfs_dir2_block_lookup(args);
+		goto out_check_rval;
+	}
+
+	rval = xfs_dir2_isleaf(args, &v);
+	if (rval)
+		goto out_free;
+	if (v)
+		rval = xfs_dir2_leaf_lookup(args);
+	else
+		rval = xfs_dir2_node_lookup(args);
+
+out_check_rval:
+	if (rval == -EEXIST)
+		rval = 0;
+	if (!rval) {
+		*inum = args->inumber;
+		if (ci_name) {
+			ci_name->name = args->value;
+			ci_name->len = args->valuelen;
+		}
+	}
+out_free:
+	xfs_iunlock(dp, lock_mode);
+	kmem_free(args);
+	return rval;
+}
+
+/*
+ * Remove an entry from a directory.
+ */
+int
+xfs_dir_removename(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*dp,
+	struct xfs_name		*name,
+	xfs_ino_t		ino,
+	xfs_extlen_t		total)		/* bmap's total block count */
+{
+	struct xfs_da_args	*args;
+	int			rval;
+	bool			v;
+
+	ASSERT(S_ISDIR(VFS_I(dp)->i_mode));
+	XFS_STATS_INC(dp->i_mount, xs_dir_remove);
+
+	args = kmem_zalloc(sizeof(*args), KM_NOFS);
+	if (!args)
+		return -ENOMEM;
+
+	args->geo = dp->i_mount->m_dir_geo;
+	args->name = name->name;
+	args->namelen = name->len;
+	args->filetype = name->type;
+	args->hashval = xfs_dir2_hashname(dp->i_mount, name);
+	args->inumber = ino;
+	args->dp = dp;
+	args->total = total;
+	args->whichfork = XFS_DATA_FORK;
+	args->trans = tp;
+
+	if (dp->i_df.if_format == XFS_DINODE_FMT_LOCAL) {
+		rval = xfs_dir2_sf_removename(args);
+		goto out_free;
+	}
+
+	rval = xfs_dir2_isblock(args, &v);
+	if (rval)
+		goto out_free;
+	if (v) {
+		rval = xfs_dir2_block_removename(args);
+		goto out_free;
+	}
+
+	rval = xfs_dir2_isleaf(args, &v);
+	if (rval)
+		goto out_free;
+	if (v)
+		rval = xfs_dir2_leaf_removename(args);
+	else
+		rval = xfs_dir2_node_removename(args);
+out_free:
+	kmem_free(args);
+	return rval;
+}
+
+/*
+ * Replace the inode number of a directory entry.
+ */
+int
+xfs_dir_replace(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*dp,
+	const struct xfs_name	*name,		/* name of entry to replace */
+	xfs_ino_t		inum,		/* new inode number */
+	xfs_extlen_t		total)		/* bmap's total block count */
+{
+	struct xfs_da_args	*args;
+	int			rval;
+	bool			v;
+
+	ASSERT(S_ISDIR(VFS_I(dp)->i_mode));
+
+	rval = xfs_dir_ino_validate(tp->t_mountp, inum);
+	if (rval)
+		return rval;
+
+	args = kmem_zalloc(sizeof(*args), KM_NOFS);
+	if (!args)
+		return -ENOMEM;
+
+	args->geo = dp->i_mount->m_dir_geo;
+	args->name = name->name;
+	args->namelen = name->len;
+	args->filetype = name->type;
+	args->hashval = xfs_dir2_hashname(dp->i_mount, name);
+	args->inumber = inum;
+	args->dp = dp;
+	args->total = total;
+	args->whichfork = XFS_DATA_FORK;
+	args->trans = tp;
+
+	if (dp->i_df.if_format == XFS_DINODE_FMT_LOCAL) {
+		rval = xfs_dir2_sf_replace(args);
+		goto out_free;
+	}
+
+	rval = xfs_dir2_isblock(args, &v);
+	if (rval)
+		goto out_free;
+	if (v) {
+		rval = xfs_dir2_block_replace(args);
+		goto out_free;
+	}
+
+	rval = xfs_dir2_isleaf(args, &v);
+	if (rval)
+		goto out_free;
+	if (v)
+		rval = xfs_dir2_leaf_replace(args);
+	else
+		rval = xfs_dir2_node_replace(args);
+out_free:
+	kmem_free(args);
+	return rval;
+}
+
+/*
+ * See if this entry can be added to the directory without allocating space.
+ */
+int
+xfs_dir_canenter(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*dp,
+	struct xfs_name	*name)		/* name of entry to add */
+{
+	return xfs_dir_createname(tp, dp, name, 0, 0);
+}
+
+/*
+ * Utility routines.
+ */
+
+/*
+ * Add a block to the directory.
+ *
+ * This routine is for data and free blocks, not leaf/node blocks which are
+ * handled by xfs_da_grow_inode.
+ */
+int
+xfs_dir2_grow_inode(
+	struct xfs_da_args	*args,
+	int			space,	/* v2 dir's space XFS_DIR2_xxx_SPACE */
+	xfs_dir2_db_t		*dbp)	/* out: block number added */
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	xfs_fileoff_t		bno;	/* directory offset of new block */
+	int			count;	/* count of filesystem blocks */
+	int			error;
+
+	trace_xfs_dir2_grow_inode(args, space);
+
+	/*
+	 * Set lowest possible block in the space requested.
+	 */
+	bno = XFS_B_TO_FSBT(mp, space * XFS_DIR2_SPACE_SIZE);
+	count = args->geo->fsbcount;
+
+	error = xfs_da_grow_inode_int(args, &bno, count);
+	if (error)
+		return error;
+
+	*dbp = xfs_dir2_da_to_db(args->geo, (xfs_dablk_t)bno);
+
+	/*
+	 * Update file's size if this is the data space and it grew.
+	 */
+	if (space == XFS_DIR2_DATA_SPACE) {
+		xfs_fsize_t	size;		/* directory file (data) size */
+
+		size = XFS_FSB_TO_B(mp, bno + count);
+		if (size > dp->i_disk_size) {
+			dp->i_disk_size = size;
+			xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE);
+		}
+	}
+	return 0;
+}
+
+/*
+ * See if the directory is a single-block form directory.
+ */
+int
+xfs_dir2_isblock(
+	struct xfs_da_args	*args,
+	bool			*isblock)
+{
+	struct xfs_mount	*mp = args->dp->i_mount;
+	xfs_fileoff_t		eof;
+	int			error;
+
+	error = xfs_bmap_last_offset(args->dp, &eof, XFS_DATA_FORK);
+	if (error)
+		return error;
+
+	*isblock = false;
+	if (XFS_FSB_TO_B(mp, eof) != args->geo->blksize)
+		return 0;
+
+	*isblock = true;
+	if (XFS_IS_CORRUPT(mp, args->dp->i_disk_size != args->geo->blksize))
+		return -EFSCORRUPTED;
+	return 0;
+}
+
+/*
+ * See if the directory is a single-leaf form directory.
+ */
+int
+xfs_dir2_isleaf(
+	struct xfs_da_args	*args,
+	bool			*isleaf)
+{
+	xfs_fileoff_t		eof;
+	int			error;
+
+	error = xfs_bmap_last_offset(args->dp, &eof, XFS_DATA_FORK);
+	if (error)
+		return error;
+
+	*isleaf = false;
+	if (eof != args->geo->leafblk + args->geo->fsbcount)
+		return 0;
+
+	*isleaf = true;
+	return 0;
+}
+
+/*
+ * Remove the given block from the directory.
+ * This routine is used for data and free blocks, leaf/node are done
+ * by xfs_da_shrink_inode.
+ */
+int
+xfs_dir2_shrink_inode(
+	struct xfs_da_args	*args,
+	xfs_dir2_db_t		db,
+	struct xfs_buf		*bp)
+{
+	xfs_fileoff_t		bno;		/* directory file offset */
+	xfs_dablk_t		da;		/* directory file offset */
+	int			done;		/* bunmap is finished */
+	struct xfs_inode	*dp;
+	int			error;
+	struct xfs_mount	*mp;
+	struct xfs_trans	*tp;
+
+	trace_xfs_dir2_shrink_inode(args, db);
+
+	dp = args->dp;
+	mp = dp->i_mount;
+	tp = args->trans;
+	da = xfs_dir2_db_to_da(args->geo, db);
+
+	/* Unmap the fsblock(s). */
+	error = xfs_bunmapi(tp, dp, da, args->geo->fsbcount, 0, 0, &done);
+	if (error) {
+		/*
+		 * ENOSPC actually can happen if we're in a removename with no
+		 * space reservation, and the resulting block removal would
+		 * cause a bmap btree split or conversion from extents to btree.
+		 * This can only happen for un-fragmented directory blocks,
+		 * since you need to be punching out the middle of an extent.
+		 * In this case we need to leave the block in the file, and not
+		 * binval it.  So the block has to be in a consistent empty
+		 * state and appropriately logged.  We don't free up the buffer,
+		 * the caller can tell it hasn't happened since it got an error
+		 * back.
+		 */
+		return error;
+	}
+	ASSERT(done);
+	/*
+	 * Invalidate the buffer from the transaction.
+	 */
+	xfs_trans_binval(tp, bp);
+	/*
+	 * If it's not a data block, we're done.
+	 */
+	if (db >= xfs_dir2_byte_to_db(args->geo, XFS_DIR2_LEAF_OFFSET))
+		return 0;
+	/*
+	 * If the block isn't the last one in the directory, we're done.
+	 */
+	if (dp->i_disk_size > xfs_dir2_db_off_to_byte(args->geo, db + 1, 0))
+		return 0;
+	bno = da;
+	if ((error = xfs_bmap_last_before(tp, dp, &bno, XFS_DATA_FORK))) {
+		/*
+		 * This can't really happen unless there's kernel corruption.
+		 */
+		return error;
+	}
+	if (db == args->geo->datablk)
+		ASSERT(bno == 0);
+	else
+		ASSERT(bno > 0);
+	/*
+	 * Set the size to the new last block.
+	 */
+	dp->i_disk_size = XFS_FSB_TO_B(mp, bno);
+	xfs_trans_log_inode(tp, dp, XFS_ILOG_CORE);
+	return 0;
+}
+
+/* Returns true if the directory entry name is valid. */
+bool
+xfs_dir2_namecheck(
+	const void	*name,
+	size_t		length)
+{
+	/*
+	 * MAXNAMELEN includes the trailing null, but (name/length) leave it
+	 * out, so use >= for the length check.
+	 */
+	if (length >= MAXNAMELEN)
+		return false;
+
+	/* There shouldn't be any slashes or nulls here */
+	return !memchr(name, '/', length) && !memchr(name, 0, length);
+}
+
+xfs_dahash_t
+xfs_dir2_hashname(
+	struct xfs_mount	*mp,
+	const struct xfs_name	*name)
+{
+	if (unlikely(xfs_has_asciici(mp)))
+		return xfs_ascii_ci_hashname(name);
+	return xfs_da_hashname(name->name, name->len);
+}
+
+enum xfs_dacmp
+xfs_dir2_compname(
+	struct xfs_da_args	*args,
+	const unsigned char	*name,
+	int			len)
+{
+	if (unlikely(xfs_has_asciici(args->dp->i_mount)))
+		return xfs_ascii_ci_compname(args, name, len);
+	return xfs_da_compname(args, name, len);
+}
diff --git a/fs/xfs/libxfs/xfs_dir2.h b/fs/xfs/libxfs/xfs_dir2.h
new file mode 100644
index 000000000..dd39f17dd
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_dir2.h
@@ -0,0 +1,251 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_DIR2_H__
+#define __XFS_DIR2_H__
+
+#include "xfs_da_format.h"
+#include "xfs_da_btree.h"
+
+struct xfs_da_args;
+struct xfs_inode;
+struct xfs_mount;
+struct xfs_trans;
+struct xfs_dir2_sf_hdr;
+struct xfs_dir2_sf_entry;
+struct xfs_dir2_data_hdr;
+struct xfs_dir2_data_entry;
+struct xfs_dir2_data_unused;
+struct xfs_dir3_icfree_hdr;
+struct xfs_dir3_icleaf_hdr;
+
+extern const struct xfs_name	xfs_name_dotdot;
+
+/*
+ * Convert inode mode to directory entry filetype
+ */
+extern unsigned char xfs_mode_to_ftype(int mode);
+
+/*
+ * Generic directory interface routines
+ */
+extern void xfs_dir_startup(void);
+extern int xfs_da_mount(struct xfs_mount *mp);
+extern void xfs_da_unmount(struct xfs_mount *mp);
+
+extern int xfs_dir_isempty(struct xfs_inode *dp);
+extern int xfs_dir_init(struct xfs_trans *tp, struct xfs_inode *dp,
+				struct xfs_inode *pdp);
+extern int xfs_dir_createname(struct xfs_trans *tp, struct xfs_inode *dp,
+				const struct xfs_name *name, xfs_ino_t inum,
+				xfs_extlen_t tot);
+extern int xfs_dir_lookup(struct xfs_trans *tp, struct xfs_inode *dp,
+				const struct xfs_name *name, xfs_ino_t *inum,
+				struct xfs_name *ci_name);
+extern int xfs_dir_removename(struct xfs_trans *tp, struct xfs_inode *dp,
+				struct xfs_name *name, xfs_ino_t ino,
+				xfs_extlen_t tot);
+extern int xfs_dir_replace(struct xfs_trans *tp, struct xfs_inode *dp,
+				const struct xfs_name *name, xfs_ino_t inum,
+				xfs_extlen_t tot);
+extern int xfs_dir_canenter(struct xfs_trans *tp, struct xfs_inode *dp,
+				struct xfs_name *name);
+
+/*
+ * Direct call from the bmap code, bypassing the generic directory layer.
+ */
+extern int xfs_dir2_sf_to_block(struct xfs_da_args *args);
+
+/*
+ * Interface routines used by userspace utilities
+ */
+extern int xfs_dir2_isblock(struct xfs_da_args *args, bool *isblock);
+extern int xfs_dir2_isleaf(struct xfs_da_args *args, bool *isleaf);
+extern int xfs_dir2_shrink_inode(struct xfs_da_args *args, xfs_dir2_db_t db,
+				struct xfs_buf *bp);
+
+extern void xfs_dir2_data_freescan(struct xfs_mount *mp,
+		struct xfs_dir2_data_hdr *hdr, int *loghead);
+extern void xfs_dir2_data_log_entry(struct xfs_da_args *args,
+		struct xfs_buf *bp, struct xfs_dir2_data_entry *dep);
+extern void xfs_dir2_data_log_header(struct xfs_da_args *args,
+		struct xfs_buf *bp);
+extern void xfs_dir2_data_log_unused(struct xfs_da_args *args,
+		struct xfs_buf *bp, struct xfs_dir2_data_unused *dup);
+extern void xfs_dir2_data_make_free(struct xfs_da_args *args,
+		struct xfs_buf *bp, xfs_dir2_data_aoff_t offset,
+		xfs_dir2_data_aoff_t len, int *needlogp, int *needscanp);
+extern int xfs_dir2_data_use_free(struct xfs_da_args *args,
+		struct xfs_buf *bp, struct xfs_dir2_data_unused *dup,
+		xfs_dir2_data_aoff_t offset, xfs_dir2_data_aoff_t len,
+		int *needlogp, int *needscanp);
+
+extern struct xfs_dir2_data_free *xfs_dir2_data_freefind(
+		struct xfs_dir2_data_hdr *hdr, struct xfs_dir2_data_free *bf,
+		struct xfs_dir2_data_unused *dup);
+
+extern int xfs_dir_ino_validate(struct xfs_mount *mp, xfs_ino_t ino);
+
+extern const struct xfs_buf_ops xfs_dir3_block_buf_ops;
+extern const struct xfs_buf_ops xfs_dir3_leafn_buf_ops;
+extern const struct xfs_buf_ops xfs_dir3_leaf1_buf_ops;
+extern const struct xfs_buf_ops xfs_dir3_free_buf_ops;
+extern const struct xfs_buf_ops xfs_dir3_data_buf_ops;
+
+/*
+ * Directory offset/block conversion functions.
+ *
+ * DB blocks here are logical directory block numbers, not filesystem blocks.
+ */
+
+/*
+ * Convert dataptr to byte in file space
+ */
+static inline xfs_dir2_off_t
+xfs_dir2_dataptr_to_byte(xfs_dir2_dataptr_t dp)
+{
+	return (xfs_dir2_off_t)dp << XFS_DIR2_DATA_ALIGN_LOG;
+}
+
+/*
+ * Convert byte in file space to dataptr.  It had better be aligned.
+ */
+static inline xfs_dir2_dataptr_t
+xfs_dir2_byte_to_dataptr(xfs_dir2_off_t by)
+{
+	return (xfs_dir2_dataptr_t)(by >> XFS_DIR2_DATA_ALIGN_LOG);
+}
+
+/*
+ * Convert byte in space to (DB) block
+ */
+static inline xfs_dir2_db_t
+xfs_dir2_byte_to_db(struct xfs_da_geometry *geo, xfs_dir2_off_t by)
+{
+	return (xfs_dir2_db_t)(by >> geo->blklog);
+}
+
+/*
+ * Convert dataptr to a block number
+ */
+static inline xfs_dir2_db_t
+xfs_dir2_dataptr_to_db(struct xfs_da_geometry *geo, xfs_dir2_dataptr_t dp)
+{
+	return xfs_dir2_byte_to_db(geo, xfs_dir2_dataptr_to_byte(dp));
+}
+
+/*
+ * Convert byte in space to offset in a block
+ */
+static inline xfs_dir2_data_aoff_t
+xfs_dir2_byte_to_off(struct xfs_da_geometry *geo, xfs_dir2_off_t by)
+{
+	return (xfs_dir2_data_aoff_t)(by & (geo->blksize - 1));
+}
+
+/*
+ * Convert dataptr to a byte offset in a block
+ */
+static inline xfs_dir2_data_aoff_t
+xfs_dir2_dataptr_to_off(struct xfs_da_geometry *geo, xfs_dir2_dataptr_t dp)
+{
+	return xfs_dir2_byte_to_off(geo, xfs_dir2_dataptr_to_byte(dp));
+}
+
+/*
+ * Convert block and offset to byte in space
+ */
+static inline xfs_dir2_off_t
+xfs_dir2_db_off_to_byte(struct xfs_da_geometry *geo, xfs_dir2_db_t db,
+			xfs_dir2_data_aoff_t o)
+{
+	return ((xfs_dir2_off_t)db << geo->blklog) + o;
+}
+
+/*
+ * Convert block (DB) to block (dablk)
+ */
+static inline xfs_dablk_t
+xfs_dir2_db_to_da(struct xfs_da_geometry *geo, xfs_dir2_db_t db)
+{
+	return (xfs_dablk_t)(db << (geo->blklog - geo->fsblog));
+}
+
+/*
+ * Convert byte in space to (DA) block
+ */
+static inline xfs_dablk_t
+xfs_dir2_byte_to_da(struct xfs_da_geometry *geo, xfs_dir2_off_t by)
+{
+	return xfs_dir2_db_to_da(geo, xfs_dir2_byte_to_db(geo, by));
+}
+
+/*
+ * Convert block and offset to dataptr
+ */
+static inline xfs_dir2_dataptr_t
+xfs_dir2_db_off_to_dataptr(struct xfs_da_geometry *geo, xfs_dir2_db_t db,
+			   xfs_dir2_data_aoff_t o)
+{
+	return xfs_dir2_byte_to_dataptr(xfs_dir2_db_off_to_byte(geo, db, o));
+}
+
+/*
+ * Convert block (dablk) to block (DB)
+ */
+static inline xfs_dir2_db_t
+xfs_dir2_da_to_db(struct xfs_da_geometry *geo, xfs_dablk_t da)
+{
+	return (xfs_dir2_db_t)(da >> (geo->blklog - geo->fsblog));
+}
+
+/*
+ * Convert block (dablk) to byte offset in space
+ */
+static inline xfs_dir2_off_t
+xfs_dir2_da_to_byte(struct xfs_da_geometry *geo, xfs_dablk_t da)
+{
+	return xfs_dir2_db_off_to_byte(geo, xfs_dir2_da_to_db(geo, da), 0);
+}
+
+/*
+ * Directory tail pointer accessor functions. Based on block geometry.
+ */
+static inline struct xfs_dir2_block_tail *
+xfs_dir2_block_tail_p(struct xfs_da_geometry *geo, struct xfs_dir2_data_hdr *hdr)
+{
+	return ((struct xfs_dir2_block_tail *)
+		((char *)hdr + geo->blksize)) - 1;
+}
+
+static inline struct xfs_dir2_leaf_tail *
+xfs_dir2_leaf_tail_p(struct xfs_da_geometry *geo, struct xfs_dir2_leaf *lp)
+{
+	return (struct xfs_dir2_leaf_tail *)
+		((char *)lp + geo->blksize -
+		  sizeof(struct xfs_dir2_leaf_tail));
+}
+
+/*
+ * The Linux API doesn't pass down the total size of the buffer
+ * we read into down to the filesystem.  With the filldir concept
+ * it's not needed for correct information, but the XFS dir2 leaf
+ * code wants an estimate of the buffer size to calculate it's
+ * readahead window and size the buffers used for mapping to
+ * physical blocks.
+ *
+ * Try to give it an estimate that's good enough, maybe at some
+ * point we can change the ->readdir prototype to include the
+ * buffer size.  For now we use the current glibc buffer size.
+ * musl libc hardcodes 2k and dietlibc uses PAGE_SIZE.
+ */
+#define XFS_READDIR_BUFSIZE	(32768)
+
+unsigned char xfs_dir3_get_dtype(struct xfs_mount *mp, uint8_t filetype);
+unsigned int xfs_dir3_data_end_offset(struct xfs_da_geometry *geo,
+		struct xfs_dir2_data_hdr *hdr);
+bool xfs_dir2_namecheck(const void *name, size_t length);
+
+#endif	/* __XFS_DIR2_H__ */
diff --git a/fs/xfs/libxfs/xfs_dir2_block.c b/fs/xfs/libxfs/xfs_dir2_block.c
new file mode 100644
index 000000000..00f960a70
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_dir2_block.c
@@ -0,0 +1,1275 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_inode.h"
+#include "xfs_trans.h"
+#include "xfs_bmap.h"
+#include "xfs_buf_item.h"
+#include "xfs_dir2.h"
+#include "xfs_dir2_priv.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+#include "xfs_log.h"
+
+/*
+ * Local function prototypes.
+ */
+static void xfs_dir2_block_log_leaf(xfs_trans_t *tp, struct xfs_buf *bp,
+				    int first, int last);
+static void xfs_dir2_block_log_tail(xfs_trans_t *tp, struct xfs_buf *bp);
+static int xfs_dir2_block_lookup_int(xfs_da_args_t *args, struct xfs_buf **bpp,
+				     int *entno);
+static int xfs_dir2_block_sort(const void *a, const void *b);
+
+static xfs_dahash_t xfs_dir_hash_dot, xfs_dir_hash_dotdot;
+
+/*
+ * One-time startup routine called from xfs_init().
+ */
+void
+xfs_dir_startup(void)
+{
+	xfs_dir_hash_dot = xfs_da_hashname((unsigned char *)".", 1);
+	xfs_dir_hash_dotdot = xfs_da_hashname((unsigned char *)"..", 2);
+}
+
+static xfs_failaddr_t
+xfs_dir3_block_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_dir3_blk_hdr	*hdr3 = bp->b_addr;
+
+	if (!xfs_verify_magic(bp, hdr3->magic))
+		return __this_address;
+
+	if (xfs_has_crc(mp)) {
+		if (!uuid_equal(&hdr3->uuid, &mp->m_sb.sb_meta_uuid))
+			return __this_address;
+		if (be64_to_cpu(hdr3->blkno) != xfs_buf_daddr(bp))
+			return __this_address;
+		if (!xfs_log_check_lsn(mp, be64_to_cpu(hdr3->lsn)))
+			return __this_address;
+	}
+	return __xfs_dir3_data_check(NULL, bp);
+}
+
+static void
+xfs_dir3_block_read_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	xfs_failaddr_t		fa;
+
+	if (xfs_has_crc(mp) &&
+	     !xfs_buf_verify_cksum(bp, XFS_DIR3_DATA_CRC_OFF))
+		xfs_verifier_error(bp, -EFSBADCRC, __this_address);
+	else {
+		fa = xfs_dir3_block_verify(bp);
+		if (fa)
+			xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+	}
+}
+
+static void
+xfs_dir3_block_write_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_buf_log_item	*bip = bp->b_log_item;
+	struct xfs_dir3_blk_hdr	*hdr3 = bp->b_addr;
+	xfs_failaddr_t		fa;
+
+	fa = xfs_dir3_block_verify(bp);
+	if (fa) {
+		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+		return;
+	}
+
+	if (!xfs_has_crc(mp))
+		return;
+
+	if (bip)
+		hdr3->lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+	xfs_buf_update_cksum(bp, XFS_DIR3_DATA_CRC_OFF);
+}
+
+const struct xfs_buf_ops xfs_dir3_block_buf_ops = {
+	.name = "xfs_dir3_block",
+	.magic = { cpu_to_be32(XFS_DIR2_BLOCK_MAGIC),
+		   cpu_to_be32(XFS_DIR3_BLOCK_MAGIC) },
+	.verify_read = xfs_dir3_block_read_verify,
+	.verify_write = xfs_dir3_block_write_verify,
+	.verify_struct = xfs_dir3_block_verify,
+};
+
+static xfs_failaddr_t
+xfs_dir3_block_header_check(
+	struct xfs_inode	*dp,
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = dp->i_mount;
+
+	if (xfs_has_crc(mp)) {
+		struct xfs_dir3_blk_hdr *hdr3 = bp->b_addr;
+
+		if (be64_to_cpu(hdr3->owner) != dp->i_ino)
+			return __this_address;
+	}
+
+	return NULL;
+}
+
+int
+xfs_dir3_block_read(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*dp,
+	struct xfs_buf		**bpp)
+{
+	struct xfs_mount	*mp = dp->i_mount;
+	xfs_failaddr_t		fa;
+	int			err;
+
+	err = xfs_da_read_buf(tp, dp, mp->m_dir_geo->datablk, 0, bpp,
+				XFS_DATA_FORK, &xfs_dir3_block_buf_ops);
+	if (err || !*bpp)
+		return err;
+
+	/* Check things that we can't do in the verifier. */
+	fa = xfs_dir3_block_header_check(dp, *bpp);
+	if (fa) {
+		__xfs_buf_mark_corrupt(*bpp, fa);
+		xfs_trans_brelse(tp, *bpp);
+		*bpp = NULL;
+		return -EFSCORRUPTED;
+	}
+
+	xfs_trans_buf_set_type(tp, *bpp, XFS_BLFT_DIR_BLOCK_BUF);
+	return err;
+}
+
+static void
+xfs_dir3_block_init(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_buf		*bp,
+	struct xfs_inode	*dp)
+{
+	struct xfs_dir3_blk_hdr *hdr3 = bp->b_addr;
+
+	bp->b_ops = &xfs_dir3_block_buf_ops;
+	xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DIR_BLOCK_BUF);
+
+	if (xfs_has_crc(mp)) {
+		memset(hdr3, 0, sizeof(*hdr3));
+		hdr3->magic = cpu_to_be32(XFS_DIR3_BLOCK_MAGIC);
+		hdr3->blkno = cpu_to_be64(xfs_buf_daddr(bp));
+		hdr3->owner = cpu_to_be64(dp->i_ino);
+		uuid_copy(&hdr3->uuid, &mp->m_sb.sb_meta_uuid);
+		return;
+
+	}
+	hdr3->magic = cpu_to_be32(XFS_DIR2_BLOCK_MAGIC);
+}
+
+static void
+xfs_dir2_block_need_space(
+	struct xfs_inode		*dp,
+	struct xfs_dir2_data_hdr	*hdr,
+	struct xfs_dir2_block_tail	*btp,
+	struct xfs_dir2_leaf_entry	*blp,
+	__be16				**tagpp,
+	struct xfs_dir2_data_unused	**dupp,
+	struct xfs_dir2_data_unused	**enddupp,
+	int				*compact,
+	int				len)
+{
+	struct xfs_dir2_data_free	*bf;
+	__be16				*tagp = NULL;
+	struct xfs_dir2_data_unused	*dup = NULL;
+	struct xfs_dir2_data_unused	*enddup = NULL;
+
+	*compact = 0;
+	bf = xfs_dir2_data_bestfree_p(dp->i_mount, hdr);
+
+	/*
+	 * If there are stale entries we'll use one for the leaf.
+	 */
+	if (btp->stale) {
+		if (be16_to_cpu(bf[0].length) >= len) {
+			/*
+			 * The biggest entry enough to avoid compaction.
+			 */
+			dup = (xfs_dir2_data_unused_t *)
+			      ((char *)hdr + be16_to_cpu(bf[0].offset));
+			goto out;
+		}
+
+		/*
+		 * Will need to compact to make this work.
+		 * Tag just before the first leaf entry.
+		 */
+		*compact = 1;
+		tagp = (__be16 *)blp - 1;
+
+		/* Data object just before the first leaf entry.  */
+		dup = (xfs_dir2_data_unused_t *)((char *)hdr + be16_to_cpu(*tagp));
+
+		/*
+		 * If it's not free then the data will go where the
+		 * leaf data starts now, if it works at all.
+		 */
+		if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG) {
+			if (be16_to_cpu(dup->length) + (be32_to_cpu(btp->stale) - 1) *
+			    (uint)sizeof(*blp) < len)
+				dup = NULL;
+		} else if ((be32_to_cpu(btp->stale) - 1) * (uint)sizeof(*blp) < len)
+			dup = NULL;
+		else
+			dup = (xfs_dir2_data_unused_t *)blp;
+		goto out;
+	}
+
+	/*
+	 * no stale entries, so just use free space.
+	 * Tag just before the first leaf entry.
+	 */
+	tagp = (__be16 *)blp - 1;
+
+	/* Data object just before the first leaf entry.  */
+	enddup = (xfs_dir2_data_unused_t *)((char *)hdr + be16_to_cpu(*tagp));
+
+	/*
+	 * If it's not free then can't do this add without cleaning up:
+	 * the space before the first leaf entry needs to be free so it
+	 * can be expanded to hold the pointer to the new entry.
+	 */
+	if (be16_to_cpu(enddup->freetag) == XFS_DIR2_DATA_FREE_TAG) {
+		/*
+		 * Check out the biggest freespace and see if it's the same one.
+		 */
+		dup = (xfs_dir2_data_unused_t *)
+		      ((char *)hdr + be16_to_cpu(bf[0].offset));
+		if (dup != enddup) {
+			/*
+			 * Not the same free entry, just check its length.
+			 */
+			if (be16_to_cpu(dup->length) < len)
+				dup = NULL;
+			goto out;
+		}
+
+		/*
+		 * It is the biggest freespace, can it hold the leaf too?
+		 */
+		if (be16_to_cpu(dup->length) < len + (uint)sizeof(*blp)) {
+			/*
+			 * Yes, use the second-largest entry instead if it works.
+			 */
+			if (be16_to_cpu(bf[1].length) >= len)
+				dup = (xfs_dir2_data_unused_t *)
+				      ((char *)hdr + be16_to_cpu(bf[1].offset));
+			else
+				dup = NULL;
+		}
+	}
+out:
+	*tagpp = tagp;
+	*dupp = dup;
+	*enddupp = enddup;
+}
+
+/*
+ * compact the leaf entries.
+ * Leave the highest-numbered stale entry stale.
+ * XXX should be the one closest to mid but mid is not yet computed.
+ */
+static void
+xfs_dir2_block_compact(
+	struct xfs_da_args		*args,
+	struct xfs_buf			*bp,
+	struct xfs_dir2_data_hdr	*hdr,
+	struct xfs_dir2_block_tail	*btp,
+	struct xfs_dir2_leaf_entry	*blp,
+	int				*needlog,
+	int				*lfloghigh,
+	int				*lfloglow)
+{
+	int			fromidx;	/* source leaf index */
+	int			toidx;		/* target leaf index */
+	int			needscan = 0;
+	int			highstale;	/* high stale index */
+
+	fromidx = toidx = be32_to_cpu(btp->count) - 1;
+	highstale = *lfloghigh = -1;
+	for (; fromidx >= 0; fromidx--) {
+		if (blp[fromidx].address == cpu_to_be32(XFS_DIR2_NULL_DATAPTR)) {
+			if (highstale == -1)
+				highstale = toidx;
+			else {
+				if (*lfloghigh == -1)
+					*lfloghigh = toidx;
+				continue;
+			}
+		}
+		if (fromidx < toidx)
+			blp[toidx] = blp[fromidx];
+		toidx--;
+	}
+	*lfloglow = toidx + 1 - (be32_to_cpu(btp->stale) - 1);
+	*lfloghigh -= be32_to_cpu(btp->stale) - 1;
+	be32_add_cpu(&btp->count, -(be32_to_cpu(btp->stale) - 1));
+	xfs_dir2_data_make_free(args, bp,
+		(xfs_dir2_data_aoff_t)((char *)blp - (char *)hdr),
+		(xfs_dir2_data_aoff_t)((be32_to_cpu(btp->stale) - 1) * sizeof(*blp)),
+		needlog, &needscan);
+	btp->stale = cpu_to_be32(1);
+	/*
+	 * If we now need to rebuild the bestfree map, do so.
+	 * This needs to happen before the next call to use_free.
+	 */
+	if (needscan)
+		xfs_dir2_data_freescan(args->dp->i_mount, hdr, needlog);
+}
+
+/*
+ * Add an entry to a block directory.
+ */
+int						/* error */
+xfs_dir2_block_addname(
+	xfs_da_args_t		*args)		/* directory op arguments */
+{
+	xfs_dir2_data_hdr_t	*hdr;		/* block header */
+	xfs_dir2_leaf_entry_t	*blp;		/* block leaf entries */
+	struct xfs_buf		*bp;		/* buffer for block */
+	xfs_dir2_block_tail_t	*btp;		/* block tail */
+	int			compact;	/* need to compact leaf ents */
+	xfs_dir2_data_entry_t	*dep;		/* block data entry */
+	xfs_inode_t		*dp;		/* directory inode */
+	xfs_dir2_data_unused_t	*dup;		/* block unused entry */
+	int			error;		/* error return value */
+	xfs_dir2_data_unused_t	*enddup=NULL;	/* unused at end of data */
+	xfs_dahash_t		hash;		/* hash value of found entry */
+	int			high;		/* high index for binary srch */
+	int			highstale;	/* high stale index */
+	int			lfloghigh=0;	/* last final leaf to log */
+	int			lfloglow=0;	/* first final leaf to log */
+	int			len;		/* length of the new entry */
+	int			low;		/* low index for binary srch */
+	int			lowstale;	/* low stale index */
+	int			mid=0;		/* midpoint for binary srch */
+	int			needlog;	/* need to log header */
+	int			needscan;	/* need to rescan freespace */
+	__be16			*tagp;		/* pointer to tag value */
+	xfs_trans_t		*tp;		/* transaction structure */
+
+	trace_xfs_dir2_block_addname(args);
+
+	dp = args->dp;
+	tp = args->trans;
+
+	/* Read the (one and only) directory block into bp. */
+	error = xfs_dir3_block_read(tp, dp, &bp);
+	if (error)
+		return error;
+
+	len = xfs_dir2_data_entsize(dp->i_mount, args->namelen);
+
+	/*
+	 * Set up pointers to parts of the block.
+	 */
+	hdr = bp->b_addr;
+	btp = xfs_dir2_block_tail_p(args->geo, hdr);
+	blp = xfs_dir2_block_leaf_p(btp);
+
+	/*
+	 * Find out if we can reuse stale entries or whether we need extra
+	 * space for entry and new leaf.
+	 */
+	xfs_dir2_block_need_space(dp, hdr, btp, blp, &tagp, &dup,
+				  &enddup, &compact, len);
+
+	/*
+	 * Done everything we need for a space check now.
+	 */
+	if (args->op_flags & XFS_DA_OP_JUSTCHECK) {
+		xfs_trans_brelse(tp, bp);
+		if (!dup)
+			return -ENOSPC;
+		return 0;
+	}
+
+	/*
+	 * If we don't have space for the new entry & leaf ...
+	 */
+	if (!dup) {
+		/* Don't have a space reservation: return no-space.  */
+		if (args->total == 0)
+			return -ENOSPC;
+		/*
+		 * Convert to the next larger format.
+		 * Then add the new entry in that format.
+		 */
+		error = xfs_dir2_block_to_leaf(args, bp);
+		if (error)
+			return error;
+		return xfs_dir2_leaf_addname(args);
+	}
+
+	needlog = needscan = 0;
+
+	/*
+	 * If need to compact the leaf entries, do it now.
+	 */
+	if (compact) {
+		xfs_dir2_block_compact(args, bp, hdr, btp, blp, &needlog,
+				      &lfloghigh, &lfloglow);
+		/* recalculate blp post-compaction */
+		blp = xfs_dir2_block_leaf_p(btp);
+	} else if (btp->stale) {
+		/*
+		 * Set leaf logging boundaries to impossible state.
+		 * For the no-stale case they're set explicitly.
+		 */
+		lfloglow = be32_to_cpu(btp->count);
+		lfloghigh = -1;
+	}
+
+	/*
+	 * Find the slot that's first lower than our hash value, -1 if none.
+	 */
+	for (low = 0, high = be32_to_cpu(btp->count) - 1; low <= high; ) {
+		mid = (low + high) >> 1;
+		if ((hash = be32_to_cpu(blp[mid].hashval)) == args->hashval)
+			break;
+		if (hash < args->hashval)
+			low = mid + 1;
+		else
+			high = mid - 1;
+	}
+	while (mid >= 0 && be32_to_cpu(blp[mid].hashval) >= args->hashval) {
+		mid--;
+	}
+	/*
+	 * No stale entries, will use enddup space to hold new leaf.
+	 */
+	if (!btp->stale) {
+		xfs_dir2_data_aoff_t	aoff;
+
+		/*
+		 * Mark the space needed for the new leaf entry, now in use.
+		 */
+		aoff = (xfs_dir2_data_aoff_t)((char *)enddup - (char *)hdr +
+				be16_to_cpu(enddup->length) - sizeof(*blp));
+		error = xfs_dir2_data_use_free(args, bp, enddup, aoff,
+				(xfs_dir2_data_aoff_t)sizeof(*blp), &needlog,
+				&needscan);
+		if (error)
+			return error;
+
+		/*
+		 * Update the tail (entry count).
+		 */
+		be32_add_cpu(&btp->count, 1);
+		/*
+		 * If we now need to rebuild the bestfree map, do so.
+		 * This needs to happen before the next call to use_free.
+		 */
+		if (needscan) {
+			xfs_dir2_data_freescan(dp->i_mount, hdr, &needlog);
+			needscan = 0;
+		}
+		/*
+		 * Adjust pointer to the first leaf entry, we're about to move
+		 * the table up one to open up space for the new leaf entry.
+		 * Then adjust our index to match.
+		 */
+		blp--;
+		mid++;
+		if (mid)
+			memmove(blp, &blp[1], mid * sizeof(*blp));
+		lfloglow = 0;
+		lfloghigh = mid;
+	}
+	/*
+	 * Use a stale leaf for our new entry.
+	 */
+	else {
+		for (lowstale = mid;
+		     lowstale >= 0 &&
+			blp[lowstale].address !=
+			cpu_to_be32(XFS_DIR2_NULL_DATAPTR);
+		     lowstale--)
+			continue;
+		for (highstale = mid + 1;
+		     highstale < be32_to_cpu(btp->count) &&
+			blp[highstale].address !=
+			cpu_to_be32(XFS_DIR2_NULL_DATAPTR) &&
+			(lowstale < 0 || mid - lowstale > highstale - mid);
+		     highstale++)
+			continue;
+		/*
+		 * Move entries toward the low-numbered stale entry.
+		 */
+		if (lowstale >= 0 &&
+		    (highstale == be32_to_cpu(btp->count) ||
+		     mid - lowstale <= highstale - mid)) {
+			if (mid - lowstale)
+				memmove(&blp[lowstale], &blp[lowstale + 1],
+					(mid - lowstale) * sizeof(*blp));
+			lfloglow = min(lowstale, lfloglow);
+			lfloghigh = max(mid, lfloghigh);
+		}
+		/*
+		 * Move entries toward the high-numbered stale entry.
+		 */
+		else {
+			ASSERT(highstale < be32_to_cpu(btp->count));
+			mid++;
+			if (highstale - mid)
+				memmove(&blp[mid + 1], &blp[mid],
+					(highstale - mid) * sizeof(*blp));
+			lfloglow = min(mid, lfloglow);
+			lfloghigh = max(highstale, lfloghigh);
+		}
+		be32_add_cpu(&btp->stale, -1);
+	}
+	/*
+	 * Point to the new data entry.
+	 */
+	dep = (xfs_dir2_data_entry_t *)dup;
+	/*
+	 * Fill in the leaf entry.
+	 */
+	blp[mid].hashval = cpu_to_be32(args->hashval);
+	blp[mid].address = cpu_to_be32(xfs_dir2_byte_to_dataptr(
+				(char *)dep - (char *)hdr));
+	xfs_dir2_block_log_leaf(tp, bp, lfloglow, lfloghigh);
+	/*
+	 * Mark space for the data entry used.
+	 */
+	error = xfs_dir2_data_use_free(args, bp, dup,
+			(xfs_dir2_data_aoff_t)((char *)dup - (char *)hdr),
+			(xfs_dir2_data_aoff_t)len, &needlog, &needscan);
+	if (error)
+		return error;
+	/*
+	 * Create the new data entry.
+	 */
+	dep->inumber = cpu_to_be64(args->inumber);
+	dep->namelen = args->namelen;
+	memcpy(dep->name, args->name, args->namelen);
+	xfs_dir2_data_put_ftype(dp->i_mount, dep, args->filetype);
+	tagp = xfs_dir2_data_entry_tag_p(dp->i_mount, dep);
+	*tagp = cpu_to_be16((char *)dep - (char *)hdr);
+	/*
+	 * Clean up the bestfree array and log the header, tail, and entry.
+	 */
+	if (needscan)
+		xfs_dir2_data_freescan(dp->i_mount, hdr, &needlog);
+	if (needlog)
+		xfs_dir2_data_log_header(args, bp);
+	xfs_dir2_block_log_tail(tp, bp);
+	xfs_dir2_data_log_entry(args, bp, dep);
+	xfs_dir3_data_check(dp, bp);
+	return 0;
+}
+
+/*
+ * Log leaf entries from the block.
+ */
+static void
+xfs_dir2_block_log_leaf(
+	xfs_trans_t		*tp,		/* transaction structure */
+	struct xfs_buf		*bp,		/* block buffer */
+	int			first,		/* index of first logged leaf */
+	int			last)		/* index of last logged leaf */
+{
+	xfs_dir2_data_hdr_t	*hdr = bp->b_addr;
+	xfs_dir2_leaf_entry_t	*blp;
+	xfs_dir2_block_tail_t	*btp;
+
+	btp = xfs_dir2_block_tail_p(tp->t_mountp->m_dir_geo, hdr);
+	blp = xfs_dir2_block_leaf_p(btp);
+	xfs_trans_log_buf(tp, bp, (uint)((char *)&blp[first] - (char *)hdr),
+		(uint)((char *)&blp[last + 1] - (char *)hdr - 1));
+}
+
+/*
+ * Log the block tail.
+ */
+static void
+xfs_dir2_block_log_tail(
+	xfs_trans_t		*tp,		/* transaction structure */
+	struct xfs_buf		*bp)		/* block buffer */
+{
+	xfs_dir2_data_hdr_t	*hdr = bp->b_addr;
+	xfs_dir2_block_tail_t	*btp;
+
+	btp = xfs_dir2_block_tail_p(tp->t_mountp->m_dir_geo, hdr);
+	xfs_trans_log_buf(tp, bp, (uint)((char *)btp - (char *)hdr),
+		(uint)((char *)(btp + 1) - (char *)hdr - 1));
+}
+
+/*
+ * Look up an entry in the block.  This is the external routine,
+ * xfs_dir2_block_lookup_int does the real work.
+ */
+int						/* error */
+xfs_dir2_block_lookup(
+	xfs_da_args_t		*args)		/* dir lookup arguments */
+{
+	xfs_dir2_data_hdr_t	*hdr;		/* block header */
+	xfs_dir2_leaf_entry_t	*blp;		/* block leaf entries */
+	struct xfs_buf		*bp;		/* block buffer */
+	xfs_dir2_block_tail_t	*btp;		/* block tail */
+	xfs_dir2_data_entry_t	*dep;		/* block data entry */
+	xfs_inode_t		*dp;		/* incore inode */
+	int			ent;		/* entry index */
+	int			error;		/* error return value */
+
+	trace_xfs_dir2_block_lookup(args);
+
+	/*
+	 * Get the buffer, look up the entry.
+	 * If not found (ENOENT) then return, have no buffer.
+	 */
+	if ((error = xfs_dir2_block_lookup_int(args, &bp, &ent)))
+		return error;
+	dp = args->dp;
+	hdr = bp->b_addr;
+	xfs_dir3_data_check(dp, bp);
+	btp = xfs_dir2_block_tail_p(args->geo, hdr);
+	blp = xfs_dir2_block_leaf_p(btp);
+	/*
+	 * Get the offset from the leaf entry, to point to the data.
+	 */
+	dep = (xfs_dir2_data_entry_t *)((char *)hdr +
+			xfs_dir2_dataptr_to_off(args->geo,
+						be32_to_cpu(blp[ent].address)));
+	/*
+	 * Fill in inode number, CI name if appropriate, release the block.
+	 */
+	args->inumber = be64_to_cpu(dep->inumber);
+	args->filetype = xfs_dir2_data_get_ftype(dp->i_mount, dep);
+	error = xfs_dir_cilookup_result(args, dep->name, dep->namelen);
+	xfs_trans_brelse(args->trans, bp);
+	return error;
+}
+
+/*
+ * Internal block lookup routine.
+ */
+static int					/* error */
+xfs_dir2_block_lookup_int(
+	xfs_da_args_t		*args,		/* dir lookup arguments */
+	struct xfs_buf		**bpp,		/* returned block buffer */
+	int			*entno)		/* returned entry number */
+{
+	xfs_dir2_dataptr_t	addr;		/* data entry address */
+	xfs_dir2_data_hdr_t	*hdr;		/* block header */
+	xfs_dir2_leaf_entry_t	*blp;		/* block leaf entries */
+	struct xfs_buf		*bp;		/* block buffer */
+	xfs_dir2_block_tail_t	*btp;		/* block tail */
+	xfs_dir2_data_entry_t	*dep;		/* block data entry */
+	xfs_inode_t		*dp;		/* incore inode */
+	int			error;		/* error return value */
+	xfs_dahash_t		hash;		/* found hash value */
+	int			high;		/* binary search high index */
+	int			low;		/* binary search low index */
+	int			mid;		/* binary search current idx */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	enum xfs_dacmp		cmp;		/* comparison result */
+
+	dp = args->dp;
+	tp = args->trans;
+
+	error = xfs_dir3_block_read(tp, dp, &bp);
+	if (error)
+		return error;
+
+	hdr = bp->b_addr;
+	xfs_dir3_data_check(dp, bp);
+	btp = xfs_dir2_block_tail_p(args->geo, hdr);
+	blp = xfs_dir2_block_leaf_p(btp);
+	/*
+	 * Loop doing a binary search for our hash value.
+	 * Find our entry, ENOENT if it's not there.
+	 */
+	for (low = 0, high = be32_to_cpu(btp->count) - 1; ; ) {
+		ASSERT(low <= high);
+		mid = (low + high) >> 1;
+		if ((hash = be32_to_cpu(blp[mid].hashval)) == args->hashval)
+			break;
+		if (hash < args->hashval)
+			low = mid + 1;
+		else
+			high = mid - 1;
+		if (low > high) {
+			ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
+			xfs_trans_brelse(tp, bp);
+			return -ENOENT;
+		}
+	}
+	/*
+	 * Back up to the first one with the right hash value.
+	 */
+	while (mid > 0 && be32_to_cpu(blp[mid - 1].hashval) == args->hashval) {
+		mid--;
+	}
+	/*
+	 * Now loop forward through all the entries with the
+	 * right hash value looking for our name.
+	 */
+	do {
+		if ((addr = be32_to_cpu(blp[mid].address)) == XFS_DIR2_NULL_DATAPTR)
+			continue;
+		/*
+		 * Get pointer to the entry from the leaf.
+		 */
+		dep = (xfs_dir2_data_entry_t *)
+			((char *)hdr + xfs_dir2_dataptr_to_off(args->geo, addr));
+		/*
+		 * Compare name and if it's an exact match, return the index
+		 * and buffer. If it's the first case-insensitive match, store
+		 * the index and buffer and continue looking for an exact match.
+		 */
+		cmp = xfs_dir2_compname(args, dep->name, dep->namelen);
+		if (cmp != XFS_CMP_DIFFERENT && cmp != args->cmpresult) {
+			args->cmpresult = cmp;
+			*bpp = bp;
+			*entno = mid;
+			if (cmp == XFS_CMP_EXACT)
+				return 0;
+		}
+	} while (++mid < be32_to_cpu(btp->count) &&
+			be32_to_cpu(blp[mid].hashval) == hash);
+
+	ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
+	/*
+	 * Here, we can only be doing a lookup (not a rename or replace).
+	 * If a case-insensitive match was found earlier, return success.
+	 */
+	if (args->cmpresult == XFS_CMP_CASE)
+		return 0;
+	/*
+	 * No match, release the buffer and return ENOENT.
+	 */
+	xfs_trans_brelse(tp, bp);
+	return -ENOENT;
+}
+
+/*
+ * Remove an entry from a block format directory.
+ * If that makes the block small enough to fit in shortform, transform it.
+ */
+int						/* error */
+xfs_dir2_block_removename(
+	xfs_da_args_t		*args)		/* directory operation args */
+{
+	xfs_dir2_data_hdr_t	*hdr;		/* block header */
+	xfs_dir2_leaf_entry_t	*blp;		/* block leaf pointer */
+	struct xfs_buf		*bp;		/* block buffer */
+	xfs_dir2_block_tail_t	*btp;		/* block tail */
+	xfs_dir2_data_entry_t	*dep;		/* block data entry */
+	xfs_inode_t		*dp;		/* incore inode */
+	int			ent;		/* block leaf entry index */
+	int			error;		/* error return value */
+	int			needlog;	/* need to log block header */
+	int			needscan;	/* need to fixup bestfree */
+	xfs_dir2_sf_hdr_t	sfh;		/* shortform header */
+	int			size;		/* shortform size */
+	xfs_trans_t		*tp;		/* transaction pointer */
+
+	trace_xfs_dir2_block_removename(args);
+
+	/*
+	 * Look up the entry in the block.  Gets the buffer and entry index.
+	 * It will always be there, the vnodeops level does a lookup first.
+	 */
+	if ((error = xfs_dir2_block_lookup_int(args, &bp, &ent))) {
+		return error;
+	}
+	dp = args->dp;
+	tp = args->trans;
+	hdr = bp->b_addr;
+	btp = xfs_dir2_block_tail_p(args->geo, hdr);
+	blp = xfs_dir2_block_leaf_p(btp);
+	/*
+	 * Point to the data entry using the leaf entry.
+	 */
+	dep = (xfs_dir2_data_entry_t *)((char *)hdr +
+			xfs_dir2_dataptr_to_off(args->geo,
+						be32_to_cpu(blp[ent].address)));
+	/*
+	 * Mark the data entry's space free.
+	 */
+	needlog = needscan = 0;
+	xfs_dir2_data_make_free(args, bp,
+		(xfs_dir2_data_aoff_t)((char *)dep - (char *)hdr),
+		xfs_dir2_data_entsize(dp->i_mount, dep->namelen), &needlog,
+		&needscan);
+	/*
+	 * Fix up the block tail.
+	 */
+	be32_add_cpu(&btp->stale, 1);
+	xfs_dir2_block_log_tail(tp, bp);
+	/*
+	 * Remove the leaf entry by marking it stale.
+	 */
+	blp[ent].address = cpu_to_be32(XFS_DIR2_NULL_DATAPTR);
+	xfs_dir2_block_log_leaf(tp, bp, ent, ent);
+	/*
+	 * Fix up bestfree, log the header if necessary.
+	 */
+	if (needscan)
+		xfs_dir2_data_freescan(dp->i_mount, hdr, &needlog);
+	if (needlog)
+		xfs_dir2_data_log_header(args, bp);
+	xfs_dir3_data_check(dp, bp);
+	/*
+	 * See if the size as a shortform is good enough.
+	 */
+	size = xfs_dir2_block_sfsize(dp, hdr, &sfh);
+	if (size > xfs_inode_data_fork_size(dp))
+		return 0;
+
+	/*
+	 * If it works, do the conversion.
+	 */
+	return xfs_dir2_block_to_sf(args, bp, size, &sfh);
+}
+
+/*
+ * Replace an entry in a V2 block directory.
+ * Change the inode number to the new value.
+ */
+int						/* error */
+xfs_dir2_block_replace(
+	xfs_da_args_t		*args)		/* directory operation args */
+{
+	xfs_dir2_data_hdr_t	*hdr;		/* block header */
+	xfs_dir2_leaf_entry_t	*blp;		/* block leaf entries */
+	struct xfs_buf		*bp;		/* block buffer */
+	xfs_dir2_block_tail_t	*btp;		/* block tail */
+	xfs_dir2_data_entry_t	*dep;		/* block data entry */
+	xfs_inode_t		*dp;		/* incore inode */
+	int			ent;		/* leaf entry index */
+	int			error;		/* error return value */
+
+	trace_xfs_dir2_block_replace(args);
+
+	/*
+	 * Lookup the entry in the directory.  Get buffer and entry index.
+	 * This will always succeed since the caller has already done a lookup.
+	 */
+	if ((error = xfs_dir2_block_lookup_int(args, &bp, &ent))) {
+		return error;
+	}
+	dp = args->dp;
+	hdr = bp->b_addr;
+	btp = xfs_dir2_block_tail_p(args->geo, hdr);
+	blp = xfs_dir2_block_leaf_p(btp);
+	/*
+	 * Point to the data entry we need to change.
+	 */
+	dep = (xfs_dir2_data_entry_t *)((char *)hdr +
+			xfs_dir2_dataptr_to_off(args->geo,
+						be32_to_cpu(blp[ent].address)));
+	ASSERT(be64_to_cpu(dep->inumber) != args->inumber);
+	/*
+	 * Change the inode number to the new value.
+	 */
+	dep->inumber = cpu_to_be64(args->inumber);
+	xfs_dir2_data_put_ftype(dp->i_mount, dep, args->filetype);
+	xfs_dir2_data_log_entry(args, bp, dep);
+	xfs_dir3_data_check(dp, bp);
+	return 0;
+}
+
+/*
+ * Qsort comparison routine for the block leaf entries.
+ */
+static int					/* sort order */
+xfs_dir2_block_sort(
+	const void			*a,	/* first leaf entry */
+	const void			*b)	/* second leaf entry */
+{
+	const xfs_dir2_leaf_entry_t	*la;	/* first leaf entry */
+	const xfs_dir2_leaf_entry_t	*lb;	/* second leaf entry */
+
+	la = a;
+	lb = b;
+	return be32_to_cpu(la->hashval) < be32_to_cpu(lb->hashval) ? -1 :
+		(be32_to_cpu(la->hashval) > be32_to_cpu(lb->hashval) ? 1 : 0);
+}
+
+/*
+ * Convert a V2 leaf directory to a V2 block directory if possible.
+ */
+int						/* error */
+xfs_dir2_leaf_to_block(
+	xfs_da_args_t		*args,		/* operation arguments */
+	struct xfs_buf		*lbp,		/* leaf buffer */
+	struct xfs_buf		*dbp)		/* data buffer */
+{
+	__be16			*bestsp;	/* leaf bests table */
+	xfs_dir2_data_hdr_t	*hdr;		/* block header */
+	xfs_dir2_block_tail_t	*btp;		/* block tail */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	xfs_dir2_data_unused_t	*dup;		/* unused data entry */
+	int			error;		/* error return value */
+	int			from;		/* leaf from index */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	xfs_dir2_leaf_tail_t	*ltp;		/* leaf tail structure */
+	xfs_mount_t		*mp;		/* file system mount point */
+	int			needlog;	/* need to log data header */
+	int			needscan;	/* need to scan for bestfree */
+	xfs_dir2_sf_hdr_t	sfh;		/* shortform header */
+	int			size;		/* bytes used */
+	__be16			*tagp;		/* end of entry (tag) */
+	int			to;		/* block/leaf to index */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	struct xfs_dir3_icleaf_hdr leafhdr;
+
+	trace_xfs_dir2_leaf_to_block(args);
+
+	dp = args->dp;
+	tp = args->trans;
+	mp = dp->i_mount;
+	leaf = lbp->b_addr;
+	xfs_dir2_leaf_hdr_from_disk(mp, &leafhdr, leaf);
+	ltp = xfs_dir2_leaf_tail_p(args->geo, leaf);
+
+	ASSERT(leafhdr.magic == XFS_DIR2_LEAF1_MAGIC ||
+	       leafhdr.magic == XFS_DIR3_LEAF1_MAGIC);
+	/*
+	 * If there are data blocks other than the first one, take this
+	 * opportunity to remove trailing empty data blocks that may have
+	 * been left behind during no-space-reservation operations.
+	 * These will show up in the leaf bests table.
+	 */
+	while (dp->i_disk_size > args->geo->blksize) {
+		int hdrsz;
+
+		hdrsz = args->geo->data_entry_offset;
+		bestsp = xfs_dir2_leaf_bests_p(ltp);
+		if (be16_to_cpu(bestsp[be32_to_cpu(ltp->bestcount) - 1]) ==
+					    args->geo->blksize - hdrsz) {
+			if ((error =
+			    xfs_dir2_leaf_trim_data(args, lbp,
+				    (xfs_dir2_db_t)(be32_to_cpu(ltp->bestcount) - 1))))
+				return error;
+		} else
+			return 0;
+	}
+	/*
+	 * Read the data block if we don't already have it, give up if it fails.
+	 */
+	if (!dbp) {
+		error = xfs_dir3_data_read(tp, dp, args->geo->datablk, 0, &dbp);
+		if (error)
+			return error;
+	}
+	hdr = dbp->b_addr;
+	ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC));
+
+	/*
+	 * Size of the "leaf" area in the block.
+	 */
+	size = (uint)sizeof(xfs_dir2_block_tail_t) +
+	       (uint)sizeof(*lep) * (leafhdr.count - leafhdr.stale);
+	/*
+	 * Look at the last data entry.
+	 */
+	tagp = (__be16 *)((char *)hdr + args->geo->blksize) - 1;
+	dup = (xfs_dir2_data_unused_t *)((char *)hdr + be16_to_cpu(*tagp));
+	/*
+	 * If it's not free or is too short we can't do it.
+	 */
+	if (be16_to_cpu(dup->freetag) != XFS_DIR2_DATA_FREE_TAG ||
+	    be16_to_cpu(dup->length) < size)
+		return 0;
+
+	/*
+	 * Start converting it to block form.
+	 */
+	xfs_dir3_block_init(mp, tp, dbp, dp);
+
+	needlog = 1;
+	needscan = 0;
+	/*
+	 * Use up the space at the end of the block (blp/btp).
+	 */
+	error = xfs_dir2_data_use_free(args, dbp, dup,
+			args->geo->blksize - size, size, &needlog, &needscan);
+	if (error)
+		return error;
+	/*
+	 * Initialize the block tail.
+	 */
+	btp = xfs_dir2_block_tail_p(args->geo, hdr);
+	btp->count = cpu_to_be32(leafhdr.count - leafhdr.stale);
+	btp->stale = 0;
+	xfs_dir2_block_log_tail(tp, dbp);
+	/*
+	 * Initialize the block leaf area.  We compact out stale entries.
+	 */
+	lep = xfs_dir2_block_leaf_p(btp);
+	for (from = to = 0; from < leafhdr.count; from++) {
+		if (leafhdr.ents[from].address ==
+		    cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
+			continue;
+		lep[to++] = leafhdr.ents[from];
+	}
+	ASSERT(to == be32_to_cpu(btp->count));
+	xfs_dir2_block_log_leaf(tp, dbp, 0, be32_to_cpu(btp->count) - 1);
+	/*
+	 * Scan the bestfree if we need it and log the data block header.
+	 */
+	if (needscan)
+		xfs_dir2_data_freescan(dp->i_mount, hdr, &needlog);
+	if (needlog)
+		xfs_dir2_data_log_header(args, dbp);
+	/*
+	 * Pitch the old leaf block.
+	 */
+	error = xfs_da_shrink_inode(args, args->geo->leafblk, lbp);
+	if (error)
+		return error;
+
+	/*
+	 * Now see if the resulting block can be shrunken to shortform.
+	 */
+	size = xfs_dir2_block_sfsize(dp, hdr, &sfh);
+	if (size > xfs_inode_data_fork_size(dp))
+		return 0;
+
+	return xfs_dir2_block_to_sf(args, dbp, size, &sfh);
+}
+
+/*
+ * Convert the shortform directory to block form.
+ */
+int						/* error */
+xfs_dir2_sf_to_block(
+	struct xfs_da_args	*args)
+{
+	struct xfs_trans	*tp = args->trans;
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(dp, XFS_DATA_FORK);
+	struct xfs_da_geometry	*geo = args->geo;
+	xfs_dir2_db_t		blkno;		/* dir-relative block # (0) */
+	xfs_dir2_data_hdr_t	*hdr;		/* block header */
+	xfs_dir2_leaf_entry_t	*blp;		/* block leaf entries */
+	struct xfs_buf		*bp;		/* block buffer */
+	xfs_dir2_block_tail_t	*btp;		/* block tail pointer */
+	xfs_dir2_data_entry_t	*dep;		/* data entry pointer */
+	int			dummy;		/* trash */
+	xfs_dir2_data_unused_t	*dup;		/* unused entry pointer */
+	int			endoffset;	/* end of data objects */
+	int			error;		/* error return value */
+	int			i;		/* index */
+	int			needlog;	/* need to log block header */
+	int			needscan;	/* need to scan block freespc */
+	int			newoffset;	/* offset from current entry */
+	unsigned int		offset = geo->data_entry_offset;
+	xfs_dir2_sf_entry_t	*sfep;		/* sf entry pointer */
+	xfs_dir2_sf_hdr_t	*oldsfp;	/* old shortform header  */
+	xfs_dir2_sf_hdr_t	*sfp;		/* shortform header  */
+	__be16			*tagp;		/* end of data entry */
+	struct xfs_name		name;
+
+	trace_xfs_dir2_sf_to_block(args);
+
+	ASSERT(ifp->if_format == XFS_DINODE_FMT_LOCAL);
+	ASSERT(dp->i_disk_size >= offsetof(struct xfs_dir2_sf_hdr, parent));
+
+	oldsfp = (xfs_dir2_sf_hdr_t *)ifp->if_u1.if_data;
+
+	ASSERT(ifp->if_bytes == dp->i_disk_size);
+	ASSERT(ifp->if_u1.if_data != NULL);
+	ASSERT(dp->i_disk_size >= xfs_dir2_sf_hdr_size(oldsfp->i8count));
+	ASSERT(dp->i_df.if_nextents == 0);
+
+	/*
+	 * Copy the directory into a temporary buffer.
+	 * Then pitch the incore inode data so we can make extents.
+	 */
+	sfp = kmem_alloc(ifp->if_bytes, 0);
+	memcpy(sfp, oldsfp, ifp->if_bytes);
+
+	xfs_idata_realloc(dp, -ifp->if_bytes, XFS_DATA_FORK);
+	xfs_bmap_local_to_extents_empty(tp, dp, XFS_DATA_FORK);
+	dp->i_disk_size = 0;
+
+	/*
+	 * Add block 0 to the inode.
+	 */
+	error = xfs_dir2_grow_inode(args, XFS_DIR2_DATA_SPACE, &blkno);
+	if (error)
+		goto out_free;
+	/*
+	 * Initialize the data block, then convert it to block format.
+	 */
+	error = xfs_dir3_data_init(args, blkno, &bp);
+	if (error)
+		goto out_free;
+	xfs_dir3_block_init(mp, tp, bp, dp);
+	hdr = bp->b_addr;
+
+	/*
+	 * Compute size of block "tail" area.
+	 */
+	i = (uint)sizeof(*btp) +
+	    (sfp->count + 2) * (uint)sizeof(xfs_dir2_leaf_entry_t);
+	/*
+	 * The whole thing is initialized to free by the init routine.
+	 * Say we're using the leaf and tail area.
+	 */
+	dup = bp->b_addr + offset;
+	needlog = needscan = 0;
+	error = xfs_dir2_data_use_free(args, bp, dup, args->geo->blksize - i,
+			i, &needlog, &needscan);
+	if (error)
+		goto out_free;
+	ASSERT(needscan == 0);
+	/*
+	 * Fill in the tail.
+	 */
+	btp = xfs_dir2_block_tail_p(args->geo, hdr);
+	btp->count = cpu_to_be32(sfp->count + 2);	/* ., .. */
+	btp->stale = 0;
+	blp = xfs_dir2_block_leaf_p(btp);
+	endoffset = (uint)((char *)blp - (char *)hdr);
+	/*
+	 * Remove the freespace, we'll manage it.
+	 */
+	error = xfs_dir2_data_use_free(args, bp, dup,
+			(xfs_dir2_data_aoff_t)((char *)dup - (char *)hdr),
+			be16_to_cpu(dup->length), &needlog, &needscan);
+	if (error)
+		goto out_free;
+
+	/*
+	 * Create entry for .
+	 */
+	dep = bp->b_addr + offset;
+	dep->inumber = cpu_to_be64(dp->i_ino);
+	dep->namelen = 1;
+	dep->name[0] = '.';
+	xfs_dir2_data_put_ftype(mp, dep, XFS_DIR3_FT_DIR);
+	tagp = xfs_dir2_data_entry_tag_p(mp, dep);
+	*tagp = cpu_to_be16(offset);
+	xfs_dir2_data_log_entry(args, bp, dep);
+	blp[0].hashval = cpu_to_be32(xfs_dir_hash_dot);
+	blp[0].address = cpu_to_be32(xfs_dir2_byte_to_dataptr(offset));
+	offset += xfs_dir2_data_entsize(mp, dep->namelen);
+
+	/*
+	 * Create entry for ..
+	 */
+	dep = bp->b_addr + offset;
+	dep->inumber = cpu_to_be64(xfs_dir2_sf_get_parent_ino(sfp));
+	dep->namelen = 2;
+	dep->name[0] = dep->name[1] = '.';
+	xfs_dir2_data_put_ftype(mp, dep, XFS_DIR3_FT_DIR);
+	tagp = xfs_dir2_data_entry_tag_p(mp, dep);
+	*tagp = cpu_to_be16(offset);
+	xfs_dir2_data_log_entry(args, bp, dep);
+	blp[1].hashval = cpu_to_be32(xfs_dir_hash_dotdot);
+	blp[1].address = cpu_to_be32(xfs_dir2_byte_to_dataptr(offset));
+	offset += xfs_dir2_data_entsize(mp, dep->namelen);
+
+	/*
+	 * Loop over existing entries, stuff them in.
+	 */
+	i = 0;
+	if (!sfp->count)
+		sfep = NULL;
+	else
+		sfep = xfs_dir2_sf_firstentry(sfp);
+
+	/*
+	 * Need to preserve the existing offset values in the sf directory.
+	 * Insert holes (unused entries) where necessary.
+	 */
+	while (offset < endoffset) {
+		/*
+		 * sfep is null when we reach the end of the list.
+		 */
+		if (sfep == NULL)
+			newoffset = endoffset;
+		else
+			newoffset = xfs_dir2_sf_get_offset(sfep);
+		/*
+		 * There should be a hole here, make one.
+		 */
+		if (offset < newoffset) {
+			dup = bp->b_addr + offset;
+			dup->freetag = cpu_to_be16(XFS_DIR2_DATA_FREE_TAG);
+			dup->length = cpu_to_be16(newoffset - offset);
+			*xfs_dir2_data_unused_tag_p(dup) = cpu_to_be16(offset);
+			xfs_dir2_data_log_unused(args, bp, dup);
+			xfs_dir2_data_freeinsert(hdr,
+					xfs_dir2_data_bestfree_p(mp, hdr),
+					dup, &dummy);
+			offset += be16_to_cpu(dup->length);
+			continue;
+		}
+		/*
+		 * Copy a real entry.
+		 */
+		dep = bp->b_addr + newoffset;
+		dep->inumber = cpu_to_be64(xfs_dir2_sf_get_ino(mp, sfp, sfep));
+		dep->namelen = sfep->namelen;
+		xfs_dir2_data_put_ftype(mp, dep,
+				xfs_dir2_sf_get_ftype(mp, sfep));
+		memcpy(dep->name, sfep->name, dep->namelen);
+		tagp = xfs_dir2_data_entry_tag_p(mp, dep);
+		*tagp = cpu_to_be16(newoffset);
+		xfs_dir2_data_log_entry(args, bp, dep);
+		name.name = sfep->name;
+		name.len = sfep->namelen;
+		blp[2 + i].hashval = cpu_to_be32(xfs_dir2_hashname(mp, &name));
+		blp[2 + i].address =
+			cpu_to_be32(xfs_dir2_byte_to_dataptr(newoffset));
+		offset = (int)((char *)(tagp + 1) - (char *)hdr);
+		if (++i == sfp->count)
+			sfep = NULL;
+		else
+			sfep = xfs_dir2_sf_nextentry(mp, sfp, sfep);
+	}
+	/* Done with the temporary buffer */
+	kmem_free(sfp);
+	/*
+	 * Sort the leaf entries by hash value.
+	 */
+	xfs_sort(blp, be32_to_cpu(btp->count), sizeof(*blp), xfs_dir2_block_sort);
+	/*
+	 * Log the leaf entry area and tail.
+	 * Already logged the header in data_init, ignore needlog.
+	 */
+	ASSERT(needscan == 0);
+	xfs_dir2_block_log_leaf(tp, bp, 0, be32_to_cpu(btp->count) - 1);
+	xfs_dir2_block_log_tail(tp, bp);
+	xfs_dir3_data_check(dp, bp);
+	return 0;
+out_free:
+	kmem_free(sfp);
+	return error;
+}
diff --git a/fs/xfs/libxfs/xfs_dir2_data.c b/fs/xfs/libxfs/xfs_dir2_data.c
new file mode 100644
index 000000000..dbcf58979
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_dir2_data.c
@@ -0,0 +1,1223 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_inode.h"
+#include "xfs_dir2.h"
+#include "xfs_dir2_priv.h"
+#include "xfs_error.h"
+#include "xfs_trans.h"
+#include "xfs_buf_item.h"
+#include "xfs_log.h"
+
+static xfs_failaddr_t xfs_dir2_data_freefind_verify(
+		struct xfs_dir2_data_hdr *hdr, struct xfs_dir2_data_free *bf,
+		struct xfs_dir2_data_unused *dup,
+		struct xfs_dir2_data_free **bf_ent);
+
+struct xfs_dir2_data_free *
+xfs_dir2_data_bestfree_p(
+	struct xfs_mount		*mp,
+	struct xfs_dir2_data_hdr	*hdr)
+{
+	if (xfs_has_crc(mp))
+		return ((struct xfs_dir3_data_hdr *)hdr)->best_free;
+	return hdr->bestfree;
+}
+
+/*
+ * Pointer to an entry's tag word.
+ */
+__be16 *
+xfs_dir2_data_entry_tag_p(
+	struct xfs_mount		*mp,
+	struct xfs_dir2_data_entry	*dep)
+{
+	return (__be16 *)((char *)dep +
+		xfs_dir2_data_entsize(mp, dep->namelen) - sizeof(__be16));
+}
+
+uint8_t
+xfs_dir2_data_get_ftype(
+	struct xfs_mount		*mp,
+	struct xfs_dir2_data_entry	*dep)
+{
+	if (xfs_has_ftype(mp)) {
+		uint8_t			ftype = dep->name[dep->namelen];
+
+		if (likely(ftype < XFS_DIR3_FT_MAX))
+			return ftype;
+	}
+
+	return XFS_DIR3_FT_UNKNOWN;
+}
+
+void
+xfs_dir2_data_put_ftype(
+	struct xfs_mount		*mp,
+	struct xfs_dir2_data_entry	*dep,
+	uint8_t				ftype)
+{
+	ASSERT(ftype < XFS_DIR3_FT_MAX);
+	ASSERT(dep->namelen != 0);
+
+	if (xfs_has_ftype(mp))
+		dep->name[dep->namelen] = ftype;
+}
+
+/*
+ * The number of leaf entries is limited by the size of the block and the amount
+ * of space used by the data entries.  We don't know how much space is used by
+ * the data entries yet, so just ensure that the count falls somewhere inside
+ * the block right now.
+ */
+static inline unsigned int
+xfs_dir2_data_max_leaf_entries(
+	struct xfs_da_geometry		*geo)
+{
+	return (geo->blksize - sizeof(struct xfs_dir2_block_tail) -
+		geo->data_entry_offset) /
+			sizeof(struct xfs_dir2_leaf_entry);
+}
+
+/*
+ * Check the consistency of the data block.
+ * The input can also be a block-format directory.
+ * Return NULL if the buffer is good, otherwise the address of the error.
+ */
+xfs_failaddr_t
+__xfs_dir3_data_check(
+	struct xfs_inode	*dp,		/* incore inode pointer */
+	struct xfs_buf		*bp)		/* data block's buffer */
+{
+	xfs_dir2_dataptr_t	addr;		/* addr for leaf lookup */
+	xfs_dir2_data_free_t	*bf;		/* bestfree table */
+	xfs_dir2_block_tail_t	*btp=NULL;	/* block tail */
+	int			count;		/* count of entries found */
+	xfs_dir2_data_hdr_t	*hdr;		/* data block header */
+	xfs_dir2_data_free_t	*dfp;		/* bestfree entry */
+	int			freeseen;	/* mask of bestfrees seen */
+	xfs_dahash_t		hash;		/* hash of current name */
+	int			i;		/* leaf index */
+	int			lastfree;	/* last entry was unused */
+	xfs_dir2_leaf_entry_t	*lep=NULL;	/* block leaf entries */
+	struct xfs_mount	*mp = bp->b_mount;
+	int			stale;		/* count of stale leaves */
+	struct xfs_name		name;
+	unsigned int		offset;
+	unsigned int		end;
+	struct xfs_da_geometry	*geo = mp->m_dir_geo;
+
+	/*
+	 * If this isn't a directory, something is seriously wrong.  Bail out.
+	 */
+	if (dp && !S_ISDIR(VFS_I(dp)->i_mode))
+		return __this_address;
+
+	hdr = bp->b_addr;
+	offset = geo->data_entry_offset;
+
+	switch (hdr->magic) {
+	case cpu_to_be32(XFS_DIR3_BLOCK_MAGIC):
+	case cpu_to_be32(XFS_DIR2_BLOCK_MAGIC):
+		btp = xfs_dir2_block_tail_p(geo, hdr);
+		lep = xfs_dir2_block_leaf_p(btp);
+
+		if (be32_to_cpu(btp->count) >=
+		    xfs_dir2_data_max_leaf_entries(geo))
+			return __this_address;
+		break;
+	case cpu_to_be32(XFS_DIR3_DATA_MAGIC):
+	case cpu_to_be32(XFS_DIR2_DATA_MAGIC):
+		break;
+	default:
+		return __this_address;
+	}
+	end = xfs_dir3_data_end_offset(geo, hdr);
+	if (!end)
+		return __this_address;
+
+	/*
+	 * Account for zero bestfree entries.
+	 */
+	bf = xfs_dir2_data_bestfree_p(mp, hdr);
+	count = lastfree = freeseen = 0;
+	if (!bf[0].length) {
+		if (bf[0].offset)
+			return __this_address;
+		freeseen |= 1 << 0;
+	}
+	if (!bf[1].length) {
+		if (bf[1].offset)
+			return __this_address;
+		freeseen |= 1 << 1;
+	}
+	if (!bf[2].length) {
+		if (bf[2].offset)
+			return __this_address;
+		freeseen |= 1 << 2;
+	}
+
+	if (be16_to_cpu(bf[0].length) < be16_to_cpu(bf[1].length))
+		return __this_address;
+	if (be16_to_cpu(bf[1].length) < be16_to_cpu(bf[2].length))
+		return __this_address;
+	/*
+	 * Loop over the data/unused entries.
+	 */
+	while (offset < end) {
+		struct xfs_dir2_data_unused	*dup = bp->b_addr + offset;
+		struct xfs_dir2_data_entry	*dep = bp->b_addr + offset;
+
+		/*
+		 * If it's unused, look for the space in the bestfree table.
+		 * If we find it, account for that, else make sure it
+		 * doesn't need to be there.
+		 */
+		if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG) {
+			xfs_failaddr_t	fa;
+
+			if (lastfree != 0)
+				return __this_address;
+			if (offset + be16_to_cpu(dup->length) > end)
+				return __this_address;
+			if (be16_to_cpu(*xfs_dir2_data_unused_tag_p(dup)) !=
+			    offset)
+				return __this_address;
+			fa = xfs_dir2_data_freefind_verify(hdr, bf, dup, &dfp);
+			if (fa)
+				return fa;
+			if (dfp) {
+				i = (int)(dfp - bf);
+				if ((freeseen & (1 << i)) != 0)
+					return __this_address;
+				freeseen |= 1 << i;
+			} else {
+				if (be16_to_cpu(dup->length) >
+				    be16_to_cpu(bf[2].length))
+					return __this_address;
+			}
+			offset += be16_to_cpu(dup->length);
+			lastfree = 1;
+			continue;
+		}
+		/*
+		 * It's a real entry.  Validate the fields.
+		 * If this is a block directory then make sure it's
+		 * in the leaf section of the block.
+		 * The linear search is crude but this is DEBUG code.
+		 */
+		if (dep->namelen == 0)
+			return __this_address;
+		if (!xfs_verify_dir_ino(mp, be64_to_cpu(dep->inumber)))
+			return __this_address;
+		if (offset + xfs_dir2_data_entsize(mp, dep->namelen) > end)
+			return __this_address;
+		if (be16_to_cpu(*xfs_dir2_data_entry_tag_p(mp, dep)) != offset)
+			return __this_address;
+		if (xfs_dir2_data_get_ftype(mp, dep) >= XFS_DIR3_FT_MAX)
+			return __this_address;
+		count++;
+		lastfree = 0;
+		if (hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+		    hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC)) {
+			addr = xfs_dir2_db_off_to_dataptr(geo, geo->datablk,
+						(xfs_dir2_data_aoff_t)
+						((char *)dep - (char *)hdr));
+			name.name = dep->name;
+			name.len = dep->namelen;
+			hash = xfs_dir2_hashname(mp, &name);
+			for (i = 0; i < be32_to_cpu(btp->count); i++) {
+				if (be32_to_cpu(lep[i].address) == addr &&
+				    be32_to_cpu(lep[i].hashval) == hash)
+					break;
+			}
+			if (i >= be32_to_cpu(btp->count))
+				return __this_address;
+		}
+		offset += xfs_dir2_data_entsize(mp, dep->namelen);
+	}
+	/*
+	 * Need to have seen all the entries and all the bestfree slots.
+	 */
+	if (freeseen != 7)
+		return __this_address;
+	if (hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+	    hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC)) {
+		for (i = stale = 0; i < be32_to_cpu(btp->count); i++) {
+			if (lep[i].address ==
+			    cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
+				stale++;
+			if (i > 0 && be32_to_cpu(lep[i].hashval) <
+				     be32_to_cpu(lep[i - 1].hashval))
+				return __this_address;
+		}
+		if (count != be32_to_cpu(btp->count) - be32_to_cpu(btp->stale))
+			return __this_address;
+		if (stale != be32_to_cpu(btp->stale))
+			return __this_address;
+	}
+	return NULL;
+}
+
+#ifdef DEBUG
+void
+xfs_dir3_data_check(
+	struct xfs_inode	*dp,
+	struct xfs_buf		*bp)
+{
+	xfs_failaddr_t		fa;
+
+	fa = __xfs_dir3_data_check(dp, bp);
+	if (!fa)
+		return;
+	xfs_corruption_error(__func__, XFS_ERRLEVEL_LOW, dp->i_mount,
+			bp->b_addr, BBTOB(bp->b_length), __FILE__, __LINE__,
+			fa);
+	ASSERT(0);
+}
+#endif
+
+static xfs_failaddr_t
+xfs_dir3_data_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_dir3_blk_hdr	*hdr3 = bp->b_addr;
+
+	if (!xfs_verify_magic(bp, hdr3->magic))
+		return __this_address;
+
+	if (xfs_has_crc(mp)) {
+		if (!uuid_equal(&hdr3->uuid, &mp->m_sb.sb_meta_uuid))
+			return __this_address;
+		if (be64_to_cpu(hdr3->blkno) != xfs_buf_daddr(bp))
+			return __this_address;
+		if (!xfs_log_check_lsn(mp, be64_to_cpu(hdr3->lsn)))
+			return __this_address;
+	}
+	return __xfs_dir3_data_check(NULL, bp);
+}
+
+/*
+ * Readahead of the first block of the directory when it is opened is completely
+ * oblivious to the format of the directory. Hence we can either get a block
+ * format buffer or a data format buffer on readahead.
+ */
+static void
+xfs_dir3_data_reada_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_dir2_data_hdr *hdr = bp->b_addr;
+
+	switch (hdr->magic) {
+	case cpu_to_be32(XFS_DIR2_BLOCK_MAGIC):
+	case cpu_to_be32(XFS_DIR3_BLOCK_MAGIC):
+		bp->b_ops = &xfs_dir3_block_buf_ops;
+		bp->b_ops->verify_read(bp);
+		return;
+	case cpu_to_be32(XFS_DIR2_DATA_MAGIC):
+	case cpu_to_be32(XFS_DIR3_DATA_MAGIC):
+		bp->b_ops = &xfs_dir3_data_buf_ops;
+		bp->b_ops->verify_read(bp);
+		return;
+	default:
+		xfs_verifier_error(bp, -EFSCORRUPTED, __this_address);
+		break;
+	}
+}
+
+static void
+xfs_dir3_data_read_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	xfs_failaddr_t		fa;
+
+	if (xfs_has_crc(mp) &&
+	    !xfs_buf_verify_cksum(bp, XFS_DIR3_DATA_CRC_OFF))
+		xfs_verifier_error(bp, -EFSBADCRC, __this_address);
+	else {
+		fa = xfs_dir3_data_verify(bp);
+		if (fa)
+			xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+	}
+}
+
+static void
+xfs_dir3_data_write_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_buf_log_item	*bip = bp->b_log_item;
+	struct xfs_dir3_blk_hdr	*hdr3 = bp->b_addr;
+	xfs_failaddr_t		fa;
+
+	fa = xfs_dir3_data_verify(bp);
+	if (fa) {
+		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+		return;
+	}
+
+	if (!xfs_has_crc(mp))
+		return;
+
+	if (bip)
+		hdr3->lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+	xfs_buf_update_cksum(bp, XFS_DIR3_DATA_CRC_OFF);
+}
+
+const struct xfs_buf_ops xfs_dir3_data_buf_ops = {
+	.name = "xfs_dir3_data",
+	.magic = { cpu_to_be32(XFS_DIR2_DATA_MAGIC),
+		   cpu_to_be32(XFS_DIR3_DATA_MAGIC) },
+	.verify_read = xfs_dir3_data_read_verify,
+	.verify_write = xfs_dir3_data_write_verify,
+	.verify_struct = xfs_dir3_data_verify,
+};
+
+static const struct xfs_buf_ops xfs_dir3_data_reada_buf_ops = {
+	.name = "xfs_dir3_data_reada",
+	.magic = { cpu_to_be32(XFS_DIR2_DATA_MAGIC),
+		   cpu_to_be32(XFS_DIR3_DATA_MAGIC) },
+	.verify_read = xfs_dir3_data_reada_verify,
+	.verify_write = xfs_dir3_data_write_verify,
+};
+
+static xfs_failaddr_t
+xfs_dir3_data_header_check(
+	struct xfs_inode	*dp,
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = dp->i_mount;
+
+	if (xfs_has_crc(mp)) {
+		struct xfs_dir3_data_hdr *hdr3 = bp->b_addr;
+
+		if (be64_to_cpu(hdr3->hdr.owner) != dp->i_ino)
+			return __this_address;
+	}
+
+	return NULL;
+}
+
+int
+xfs_dir3_data_read(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*dp,
+	xfs_dablk_t		bno,
+	unsigned int		flags,
+	struct xfs_buf		**bpp)
+{
+	xfs_failaddr_t		fa;
+	int			err;
+
+	err = xfs_da_read_buf(tp, dp, bno, flags, bpp, XFS_DATA_FORK,
+			&xfs_dir3_data_buf_ops);
+	if (err || !*bpp)
+		return err;
+
+	/* Check things that we can't do in the verifier. */
+	fa = xfs_dir3_data_header_check(dp, *bpp);
+	if (fa) {
+		__xfs_buf_mark_corrupt(*bpp, fa);
+		xfs_trans_brelse(tp, *bpp);
+		*bpp = NULL;
+		return -EFSCORRUPTED;
+	}
+
+	xfs_trans_buf_set_type(tp, *bpp, XFS_BLFT_DIR_DATA_BUF);
+	return err;
+}
+
+int
+xfs_dir3_data_readahead(
+	struct xfs_inode	*dp,
+	xfs_dablk_t		bno,
+	unsigned int		flags)
+{
+	return xfs_da_reada_buf(dp, bno, flags, XFS_DATA_FORK,
+				&xfs_dir3_data_reada_buf_ops);
+}
+
+/*
+ * Find the bestfree entry that exactly coincides with unused directory space
+ * or a verifier error because the bestfree data are bad.
+ */
+static xfs_failaddr_t
+xfs_dir2_data_freefind_verify(
+	struct xfs_dir2_data_hdr	*hdr,
+	struct xfs_dir2_data_free	*bf,
+	struct xfs_dir2_data_unused	*dup,
+	struct xfs_dir2_data_free	**bf_ent)
+{
+	struct xfs_dir2_data_free	*dfp;
+	xfs_dir2_data_aoff_t		off;
+	bool				matched = false;
+	bool				seenzero = false;
+
+	*bf_ent = NULL;
+	off = (xfs_dir2_data_aoff_t)((char *)dup - (char *)hdr);
+
+	/*
+	 * Validate some consistency in the bestfree table.
+	 * Check order, non-overlapping entries, and if we find the
+	 * one we're looking for it has to be exact.
+	 */
+	for (dfp = &bf[0]; dfp < &bf[XFS_DIR2_DATA_FD_COUNT]; dfp++) {
+		if (!dfp->offset) {
+			if (dfp->length)
+				return __this_address;
+			seenzero = true;
+			continue;
+		}
+		if (seenzero)
+			return __this_address;
+		if (be16_to_cpu(dfp->offset) == off) {
+			matched = true;
+			if (dfp->length != dup->length)
+				return __this_address;
+		} else if (be16_to_cpu(dfp->offset) > off) {
+			if (off + be16_to_cpu(dup->length) >
+					be16_to_cpu(dfp->offset))
+				return __this_address;
+		} else {
+			if (be16_to_cpu(dfp->offset) +
+					be16_to_cpu(dfp->length) > off)
+				return __this_address;
+		}
+		if (!matched &&
+		    be16_to_cpu(dfp->length) < be16_to_cpu(dup->length))
+			return __this_address;
+		if (dfp > &bf[0] &&
+		    be16_to_cpu(dfp[-1].length) < be16_to_cpu(dfp[0].length))
+			return __this_address;
+	}
+
+	/* Looks ok so far; now try to match up with a bestfree entry. */
+	*bf_ent = xfs_dir2_data_freefind(hdr, bf, dup);
+	return NULL;
+}
+
+/*
+ * Given a data block and an unused entry from that block,
+ * return the bestfree entry if any that corresponds to it.
+ */
+xfs_dir2_data_free_t *
+xfs_dir2_data_freefind(
+	struct xfs_dir2_data_hdr *hdr,		/* data block header */
+	struct xfs_dir2_data_free *bf,		/* bestfree table pointer */
+	struct xfs_dir2_data_unused *dup)	/* unused space */
+{
+	xfs_dir2_data_free_t	*dfp;		/* bestfree entry */
+	xfs_dir2_data_aoff_t	off;		/* offset value needed */
+
+	off = (xfs_dir2_data_aoff_t)((char *)dup - (char *)hdr);
+
+	/*
+	 * If this is smaller than the smallest bestfree entry,
+	 * it can't be there since they're sorted.
+	 */
+	if (be16_to_cpu(dup->length) <
+	    be16_to_cpu(bf[XFS_DIR2_DATA_FD_COUNT - 1].length))
+		return NULL;
+	/*
+	 * Look at the three bestfree entries for our guy.
+	 */
+	for (dfp = &bf[0]; dfp < &bf[XFS_DIR2_DATA_FD_COUNT]; dfp++) {
+		if (!dfp->offset)
+			return NULL;
+		if (be16_to_cpu(dfp->offset) == off)
+			return dfp;
+	}
+	/*
+	 * Didn't find it.  This only happens if there are duplicate lengths.
+	 */
+	return NULL;
+}
+
+/*
+ * Insert an unused-space entry into the bestfree table.
+ */
+xfs_dir2_data_free_t *				/* entry inserted */
+xfs_dir2_data_freeinsert(
+	struct xfs_dir2_data_hdr *hdr,		/* data block pointer */
+	struct xfs_dir2_data_free *dfp,		/* bestfree table pointer */
+	struct xfs_dir2_data_unused *dup,	/* unused space */
+	int			*loghead)	/* log the data header (out) */
+{
+	xfs_dir2_data_free_t	new;		/* new bestfree entry */
+
+	ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
+
+	new.length = dup->length;
+	new.offset = cpu_to_be16((char *)dup - (char *)hdr);
+
+	/*
+	 * Insert at position 0, 1, or 2; or not at all.
+	 */
+	if (be16_to_cpu(new.length) > be16_to_cpu(dfp[0].length)) {
+		dfp[2] = dfp[1];
+		dfp[1] = dfp[0];
+		dfp[0] = new;
+		*loghead = 1;
+		return &dfp[0];
+	}
+	if (be16_to_cpu(new.length) > be16_to_cpu(dfp[1].length)) {
+		dfp[2] = dfp[1];
+		dfp[1] = new;
+		*loghead = 1;
+		return &dfp[1];
+	}
+	if (be16_to_cpu(new.length) > be16_to_cpu(dfp[2].length)) {
+		dfp[2] = new;
+		*loghead = 1;
+		return &dfp[2];
+	}
+	return NULL;
+}
+
+/*
+ * Remove a bestfree entry from the table.
+ */
+STATIC void
+xfs_dir2_data_freeremove(
+	struct xfs_dir2_data_hdr *hdr,		/* data block header */
+	struct xfs_dir2_data_free *bf,		/* bestfree table pointer */
+	struct xfs_dir2_data_free *dfp,		/* bestfree entry pointer */
+	int			*loghead)	/* out: log data header */
+{
+
+	ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
+
+	/*
+	 * It's the first entry, slide the next 2 up.
+	 */
+	if (dfp == &bf[0]) {
+		bf[0] = bf[1];
+		bf[1] = bf[2];
+	}
+	/*
+	 * It's the second entry, slide the 3rd entry up.
+	 */
+	else if (dfp == &bf[1])
+		bf[1] = bf[2];
+	/*
+	 * Must be the last entry.
+	 */
+	else
+		ASSERT(dfp == &bf[2]);
+	/*
+	 * Clear the 3rd entry, must be zero now.
+	 */
+	bf[2].length = 0;
+	bf[2].offset = 0;
+	*loghead = 1;
+}
+
+/*
+ * Given a data block, reconstruct its bestfree map.
+ */
+void
+xfs_dir2_data_freescan(
+	struct xfs_mount		*mp,
+	struct xfs_dir2_data_hdr	*hdr,
+	int				*loghead)
+{
+	struct xfs_da_geometry		*geo = mp->m_dir_geo;
+	struct xfs_dir2_data_free	*bf = xfs_dir2_data_bestfree_p(mp, hdr);
+	void				*addr = hdr;
+	unsigned int			offset = geo->data_entry_offset;
+	unsigned int			end;
+
+	ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
+
+	/*
+	 * Start by clearing the table.
+	 */
+	memset(bf, 0, sizeof(*bf) * XFS_DIR2_DATA_FD_COUNT);
+	*loghead = 1;
+
+	end = xfs_dir3_data_end_offset(geo, addr);
+	while (offset < end) {
+		struct xfs_dir2_data_unused	*dup = addr + offset;
+		struct xfs_dir2_data_entry	*dep = addr + offset;
+
+		/*
+		 * If it's a free entry, insert it.
+		 */
+		if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG) {
+			ASSERT(offset ==
+			       be16_to_cpu(*xfs_dir2_data_unused_tag_p(dup)));
+			xfs_dir2_data_freeinsert(hdr, bf, dup, loghead);
+			offset += be16_to_cpu(dup->length);
+			continue;
+		}
+
+		/*
+		 * For active entries, check their tags and skip them.
+		 */
+		ASSERT(offset ==
+		       be16_to_cpu(*xfs_dir2_data_entry_tag_p(mp, dep)));
+		offset += xfs_dir2_data_entsize(mp, dep->namelen);
+	}
+}
+
+/*
+ * Initialize a data block at the given block number in the directory.
+ * Give back the buffer for the created block.
+ */
+int						/* error */
+xfs_dir3_data_init(
+	struct xfs_da_args		*args,	/* directory operation args */
+	xfs_dir2_db_t			blkno,	/* logical dir block number */
+	struct xfs_buf			**bpp)	/* output block buffer */
+{
+	struct xfs_trans		*tp = args->trans;
+	struct xfs_inode		*dp = args->dp;
+	struct xfs_mount		*mp = dp->i_mount;
+	struct xfs_da_geometry		*geo = args->geo;
+	struct xfs_buf			*bp;
+	struct xfs_dir2_data_hdr	*hdr;
+	struct xfs_dir2_data_unused	*dup;
+	struct xfs_dir2_data_free 	*bf;
+	int				error;
+	int				i;
+
+	/*
+	 * Get the buffer set up for the block.
+	 */
+	error = xfs_da_get_buf(tp, dp, xfs_dir2_db_to_da(args->geo, blkno),
+			       &bp, XFS_DATA_FORK);
+	if (error)
+		return error;
+	bp->b_ops = &xfs_dir3_data_buf_ops;
+	xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DIR_DATA_BUF);
+
+	/*
+	 * Initialize the header.
+	 */
+	hdr = bp->b_addr;
+	if (xfs_has_crc(mp)) {
+		struct xfs_dir3_blk_hdr *hdr3 = bp->b_addr;
+
+		memset(hdr3, 0, sizeof(*hdr3));
+		hdr3->magic = cpu_to_be32(XFS_DIR3_DATA_MAGIC);
+		hdr3->blkno = cpu_to_be64(xfs_buf_daddr(bp));
+		hdr3->owner = cpu_to_be64(dp->i_ino);
+		uuid_copy(&hdr3->uuid, &mp->m_sb.sb_meta_uuid);
+
+	} else
+		hdr->magic = cpu_to_be32(XFS_DIR2_DATA_MAGIC);
+
+	bf = xfs_dir2_data_bestfree_p(mp, hdr);
+	bf[0].offset = cpu_to_be16(geo->data_entry_offset);
+	bf[0].length = cpu_to_be16(geo->blksize - geo->data_entry_offset);
+	for (i = 1; i < XFS_DIR2_DATA_FD_COUNT; i++) {
+		bf[i].length = 0;
+		bf[i].offset = 0;
+	}
+
+	/*
+	 * Set up an unused entry for the block's body.
+	 */
+	dup = bp->b_addr + geo->data_entry_offset;
+	dup->freetag = cpu_to_be16(XFS_DIR2_DATA_FREE_TAG);
+	dup->length = bf[0].length;
+	*xfs_dir2_data_unused_tag_p(dup) = cpu_to_be16((char *)dup - (char *)hdr);
+
+	/*
+	 * Log it and return it.
+	 */
+	xfs_dir2_data_log_header(args, bp);
+	xfs_dir2_data_log_unused(args, bp, dup);
+	*bpp = bp;
+	return 0;
+}
+
+/*
+ * Log an active data entry from the block.
+ */
+void
+xfs_dir2_data_log_entry(
+	struct xfs_da_args	*args,
+	struct xfs_buf		*bp,
+	xfs_dir2_data_entry_t	*dep)		/* data entry pointer */
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_dir2_data_hdr *hdr = bp->b_addr;
+
+	ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
+
+	xfs_trans_log_buf(args->trans, bp, (uint)((char *)dep - (char *)hdr),
+		(uint)((char *)(xfs_dir2_data_entry_tag_p(mp, dep) + 1) -
+		       (char *)hdr - 1));
+}
+
+/*
+ * Log a data block header.
+ */
+void
+xfs_dir2_data_log_header(
+	struct xfs_da_args	*args,
+	struct xfs_buf		*bp)
+{
+#ifdef DEBUG
+	struct xfs_dir2_data_hdr *hdr = bp->b_addr;
+
+	ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
+#endif
+
+	xfs_trans_log_buf(args->trans, bp, 0, args->geo->data_entry_offset - 1);
+}
+
+/*
+ * Log a data unused entry.
+ */
+void
+xfs_dir2_data_log_unused(
+	struct xfs_da_args	*args,
+	struct xfs_buf		*bp,
+	xfs_dir2_data_unused_t	*dup)		/* data unused pointer */
+{
+	xfs_dir2_data_hdr_t	*hdr = bp->b_addr;
+
+	ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
+
+	/*
+	 * Log the first part of the unused entry.
+	 */
+	xfs_trans_log_buf(args->trans, bp, (uint)((char *)dup - (char *)hdr),
+		(uint)((char *)&dup->length + sizeof(dup->length) -
+		       1 - (char *)hdr));
+	/*
+	 * Log the end (tag) of the unused entry.
+	 */
+	xfs_trans_log_buf(args->trans, bp,
+		(uint)((char *)xfs_dir2_data_unused_tag_p(dup) - (char *)hdr),
+		(uint)((char *)xfs_dir2_data_unused_tag_p(dup) - (char *)hdr +
+		       sizeof(xfs_dir2_data_off_t) - 1));
+}
+
+/*
+ * Make a byte range in the data block unused.
+ * Its current contents are unimportant.
+ */
+void
+xfs_dir2_data_make_free(
+	struct xfs_da_args	*args,
+	struct xfs_buf		*bp,
+	xfs_dir2_data_aoff_t	offset,		/* starting byte offset */
+	xfs_dir2_data_aoff_t	len,		/* length in bytes */
+	int			*needlogp,	/* out: log header */
+	int			*needscanp)	/* out: regen bestfree */
+{
+	xfs_dir2_data_hdr_t	*hdr;		/* data block pointer */
+	xfs_dir2_data_free_t	*dfp;		/* bestfree pointer */
+	int			needscan;	/* need to regen bestfree */
+	xfs_dir2_data_unused_t	*newdup;	/* new unused entry */
+	xfs_dir2_data_unused_t	*postdup;	/* unused entry after us */
+	xfs_dir2_data_unused_t	*prevdup;	/* unused entry before us */
+	unsigned int		end;
+	struct xfs_dir2_data_free *bf;
+
+	hdr = bp->b_addr;
+
+	/*
+	 * Figure out where the end of the data area is.
+	 */
+	end = xfs_dir3_data_end_offset(args->geo, hdr);
+	ASSERT(end != 0);
+
+	/*
+	 * If this isn't the start of the block, then back up to
+	 * the previous entry and see if it's free.
+	 */
+	if (offset > args->geo->data_entry_offset) {
+		__be16			*tagp;	/* tag just before us */
+
+		tagp = (__be16 *)((char *)hdr + offset) - 1;
+		prevdup = (xfs_dir2_data_unused_t *)((char *)hdr + be16_to_cpu(*tagp));
+		if (be16_to_cpu(prevdup->freetag) != XFS_DIR2_DATA_FREE_TAG)
+			prevdup = NULL;
+	} else
+		prevdup = NULL;
+	/*
+	 * If this isn't the end of the block, see if the entry after
+	 * us is free.
+	 */
+	if (offset + len < end) {
+		postdup =
+			(xfs_dir2_data_unused_t *)((char *)hdr + offset + len);
+		if (be16_to_cpu(postdup->freetag) != XFS_DIR2_DATA_FREE_TAG)
+			postdup = NULL;
+	} else
+		postdup = NULL;
+	ASSERT(*needscanp == 0);
+	needscan = 0;
+	/*
+	 * Previous and following entries are both free,
+	 * merge everything into a single free entry.
+	 */
+	bf = xfs_dir2_data_bestfree_p(args->dp->i_mount, hdr);
+	if (prevdup && postdup) {
+		xfs_dir2_data_free_t	*dfp2;	/* another bestfree pointer */
+
+		/*
+		 * See if prevdup and/or postdup are in bestfree table.
+		 */
+		dfp = xfs_dir2_data_freefind(hdr, bf, prevdup);
+		dfp2 = xfs_dir2_data_freefind(hdr, bf, postdup);
+		/*
+		 * We need a rescan unless there are exactly 2 free entries
+		 * namely our two.  Then we know what's happening, otherwise
+		 * since the third bestfree is there, there might be more
+		 * entries.
+		 */
+		needscan = (bf[2].length != 0);
+		/*
+		 * Fix up the new big freespace.
+		 */
+		be16_add_cpu(&prevdup->length, len + be16_to_cpu(postdup->length));
+		*xfs_dir2_data_unused_tag_p(prevdup) =
+			cpu_to_be16((char *)prevdup - (char *)hdr);
+		xfs_dir2_data_log_unused(args, bp, prevdup);
+		if (!needscan) {
+			/*
+			 * Has to be the case that entries 0 and 1 are
+			 * dfp and dfp2 (don't know which is which), and
+			 * entry 2 is empty.
+			 * Remove entry 1 first then entry 0.
+			 */
+			ASSERT(dfp && dfp2);
+			if (dfp == &bf[1]) {
+				dfp = &bf[0];
+				ASSERT(dfp2 == dfp);
+				dfp2 = &bf[1];
+			}
+			xfs_dir2_data_freeremove(hdr, bf, dfp2, needlogp);
+			xfs_dir2_data_freeremove(hdr, bf, dfp, needlogp);
+			/*
+			 * Now insert the new entry.
+			 */
+			dfp = xfs_dir2_data_freeinsert(hdr, bf, prevdup,
+						       needlogp);
+			ASSERT(dfp == &bf[0]);
+			ASSERT(dfp->length == prevdup->length);
+			ASSERT(!dfp[1].length);
+			ASSERT(!dfp[2].length);
+		}
+	}
+	/*
+	 * The entry before us is free, merge with it.
+	 */
+	else if (prevdup) {
+		dfp = xfs_dir2_data_freefind(hdr, bf, prevdup);
+		be16_add_cpu(&prevdup->length, len);
+		*xfs_dir2_data_unused_tag_p(prevdup) =
+			cpu_to_be16((char *)prevdup - (char *)hdr);
+		xfs_dir2_data_log_unused(args, bp, prevdup);
+		/*
+		 * If the previous entry was in the table, the new entry
+		 * is longer, so it will be in the table too.  Remove
+		 * the old one and add the new one.
+		 */
+		if (dfp) {
+			xfs_dir2_data_freeremove(hdr, bf, dfp, needlogp);
+			xfs_dir2_data_freeinsert(hdr, bf, prevdup, needlogp);
+		}
+		/*
+		 * Otherwise we need a scan if the new entry is big enough.
+		 */
+		else {
+			needscan = be16_to_cpu(prevdup->length) >
+				   be16_to_cpu(bf[2].length);
+		}
+	}
+	/*
+	 * The following entry is free, merge with it.
+	 */
+	else if (postdup) {
+		dfp = xfs_dir2_data_freefind(hdr, bf, postdup);
+		newdup = (xfs_dir2_data_unused_t *)((char *)hdr + offset);
+		newdup->freetag = cpu_to_be16(XFS_DIR2_DATA_FREE_TAG);
+		newdup->length = cpu_to_be16(len + be16_to_cpu(postdup->length));
+		*xfs_dir2_data_unused_tag_p(newdup) =
+			cpu_to_be16((char *)newdup - (char *)hdr);
+		xfs_dir2_data_log_unused(args, bp, newdup);
+		/*
+		 * If the following entry was in the table, the new entry
+		 * is longer, so it will be in the table too.  Remove
+		 * the old one and add the new one.
+		 */
+		if (dfp) {
+			xfs_dir2_data_freeremove(hdr, bf, dfp, needlogp);
+			xfs_dir2_data_freeinsert(hdr, bf, newdup, needlogp);
+		}
+		/*
+		 * Otherwise we need a scan if the new entry is big enough.
+		 */
+		else {
+			needscan = be16_to_cpu(newdup->length) >
+				   be16_to_cpu(bf[2].length);
+		}
+	}
+	/*
+	 * Neither neighbor is free.  Make a new entry.
+	 */
+	else {
+		newdup = (xfs_dir2_data_unused_t *)((char *)hdr + offset);
+		newdup->freetag = cpu_to_be16(XFS_DIR2_DATA_FREE_TAG);
+		newdup->length = cpu_to_be16(len);
+		*xfs_dir2_data_unused_tag_p(newdup) =
+			cpu_to_be16((char *)newdup - (char *)hdr);
+		xfs_dir2_data_log_unused(args, bp, newdup);
+		xfs_dir2_data_freeinsert(hdr, bf, newdup, needlogp);
+	}
+	*needscanp = needscan;
+}
+
+/* Check our free data for obvious signs of corruption. */
+static inline xfs_failaddr_t
+xfs_dir2_data_check_free(
+	struct xfs_dir2_data_hdr	*hdr,
+	struct xfs_dir2_data_unused	*dup,
+	xfs_dir2_data_aoff_t		offset,
+	xfs_dir2_data_aoff_t		len)
+{
+	if (hdr->magic != cpu_to_be32(XFS_DIR2_DATA_MAGIC) &&
+	    hdr->magic != cpu_to_be32(XFS_DIR3_DATA_MAGIC) &&
+	    hdr->magic != cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) &&
+	    hdr->magic != cpu_to_be32(XFS_DIR3_BLOCK_MAGIC))
+		return __this_address;
+	if (be16_to_cpu(dup->freetag) != XFS_DIR2_DATA_FREE_TAG)
+		return __this_address;
+	if (offset < (char *)dup - (char *)hdr)
+		return __this_address;
+	if (offset + len > (char *)dup + be16_to_cpu(dup->length) - (char *)hdr)
+		return __this_address;
+	if ((char *)dup - (char *)hdr !=
+			be16_to_cpu(*xfs_dir2_data_unused_tag_p(dup)))
+		return __this_address;
+	return NULL;
+}
+
+/* Sanity-check a new bestfree entry. */
+static inline xfs_failaddr_t
+xfs_dir2_data_check_new_free(
+	struct xfs_dir2_data_hdr	*hdr,
+	struct xfs_dir2_data_free	*dfp,
+	struct xfs_dir2_data_unused	*newdup)
+{
+	if (dfp == NULL)
+		return __this_address;
+	if (dfp->length != newdup->length)
+		return __this_address;
+	if (be16_to_cpu(dfp->offset) != (char *)newdup - (char *)hdr)
+		return __this_address;
+	return NULL;
+}
+
+/*
+ * Take a byte range out of an existing unused space and make it un-free.
+ */
+int
+xfs_dir2_data_use_free(
+	struct xfs_da_args	*args,
+	struct xfs_buf		*bp,
+	xfs_dir2_data_unused_t	*dup,		/* unused entry */
+	xfs_dir2_data_aoff_t	offset,		/* starting offset to use */
+	xfs_dir2_data_aoff_t	len,		/* length to use */
+	int			*needlogp,	/* out: need to log header */
+	int			*needscanp)	/* out: need regen bestfree */
+{
+	xfs_dir2_data_hdr_t	*hdr;		/* data block header */
+	xfs_dir2_data_free_t	*dfp;		/* bestfree pointer */
+	xfs_dir2_data_unused_t	*newdup;	/* new unused entry */
+	xfs_dir2_data_unused_t	*newdup2;	/* another new unused entry */
+	struct xfs_dir2_data_free *bf;
+	xfs_failaddr_t		fa;
+	int			matchback;	/* matches end of freespace */
+	int			matchfront;	/* matches start of freespace */
+	int			needscan;	/* need to regen bestfree */
+	int			oldlen;		/* old unused entry's length */
+
+	hdr = bp->b_addr;
+	fa = xfs_dir2_data_check_free(hdr, dup, offset, len);
+	if (fa)
+		goto corrupt;
+	/*
+	 * Look up the entry in the bestfree table.
+	 */
+	oldlen = be16_to_cpu(dup->length);
+	bf = xfs_dir2_data_bestfree_p(args->dp->i_mount, hdr);
+	dfp = xfs_dir2_data_freefind(hdr, bf, dup);
+	ASSERT(dfp || oldlen <= be16_to_cpu(bf[2].length));
+	/*
+	 * Check for alignment with front and back of the entry.
+	 */
+	matchfront = (char *)dup - (char *)hdr == offset;
+	matchback = (char *)dup + oldlen - (char *)hdr == offset + len;
+	ASSERT(*needscanp == 0);
+	needscan = 0;
+	/*
+	 * If we matched it exactly we just need to get rid of it from
+	 * the bestfree table.
+	 */
+	if (matchfront && matchback) {
+		if (dfp) {
+			needscan = (bf[2].offset != 0);
+			if (!needscan)
+				xfs_dir2_data_freeremove(hdr, bf, dfp,
+							 needlogp);
+		}
+	}
+	/*
+	 * We match the first part of the entry.
+	 * Make a new entry with the remaining freespace.
+	 */
+	else if (matchfront) {
+		newdup = (xfs_dir2_data_unused_t *)((char *)hdr + offset + len);
+		newdup->freetag = cpu_to_be16(XFS_DIR2_DATA_FREE_TAG);
+		newdup->length = cpu_to_be16(oldlen - len);
+		*xfs_dir2_data_unused_tag_p(newdup) =
+			cpu_to_be16((char *)newdup - (char *)hdr);
+		xfs_dir2_data_log_unused(args, bp, newdup);
+		/*
+		 * If it was in the table, remove it and add the new one.
+		 */
+		if (dfp) {
+			xfs_dir2_data_freeremove(hdr, bf, dfp, needlogp);
+			dfp = xfs_dir2_data_freeinsert(hdr, bf, newdup,
+						       needlogp);
+			fa = xfs_dir2_data_check_new_free(hdr, dfp, newdup);
+			if (fa)
+				goto corrupt;
+			/*
+			 * If we got inserted at the last slot,
+			 * that means we don't know if there was a better
+			 * choice for the last slot, or not.  Rescan.
+			 */
+			needscan = dfp == &bf[2];
+		}
+	}
+	/*
+	 * We match the last part of the entry.
+	 * Trim the allocated space off the tail of the entry.
+	 */
+	else if (matchback) {
+		newdup = dup;
+		newdup->length = cpu_to_be16(((char *)hdr + offset) - (char *)newdup);
+		*xfs_dir2_data_unused_tag_p(newdup) =
+			cpu_to_be16((char *)newdup - (char *)hdr);
+		xfs_dir2_data_log_unused(args, bp, newdup);
+		/*
+		 * If it was in the table, remove it and add the new one.
+		 */
+		if (dfp) {
+			xfs_dir2_data_freeremove(hdr, bf, dfp, needlogp);
+			dfp = xfs_dir2_data_freeinsert(hdr, bf, newdup,
+						       needlogp);
+			fa = xfs_dir2_data_check_new_free(hdr, dfp, newdup);
+			if (fa)
+				goto corrupt;
+			/*
+			 * If we got inserted at the last slot,
+			 * that means we don't know if there was a better
+			 * choice for the last slot, or not.  Rescan.
+			 */
+			needscan = dfp == &bf[2];
+		}
+	}
+	/*
+	 * Poking out the middle of an entry.
+	 * Make two new entries.
+	 */
+	else {
+		newdup = dup;
+		newdup->length = cpu_to_be16(((char *)hdr + offset) - (char *)newdup);
+		*xfs_dir2_data_unused_tag_p(newdup) =
+			cpu_to_be16((char *)newdup - (char *)hdr);
+		xfs_dir2_data_log_unused(args, bp, newdup);
+		newdup2 = (xfs_dir2_data_unused_t *)((char *)hdr + offset + len);
+		newdup2->freetag = cpu_to_be16(XFS_DIR2_DATA_FREE_TAG);
+		newdup2->length = cpu_to_be16(oldlen - len - be16_to_cpu(newdup->length));
+		*xfs_dir2_data_unused_tag_p(newdup2) =
+			cpu_to_be16((char *)newdup2 - (char *)hdr);
+		xfs_dir2_data_log_unused(args, bp, newdup2);
+		/*
+		 * If the old entry was in the table, we need to scan
+		 * if the 3rd entry was valid, since these entries
+		 * are smaller than the old one.
+		 * If we don't need to scan that means there were 1 or 2
+		 * entries in the table, and removing the old and adding
+		 * the 2 new will work.
+		 */
+		if (dfp) {
+			needscan = (bf[2].length != 0);
+			if (!needscan) {
+				xfs_dir2_data_freeremove(hdr, bf, dfp,
+							 needlogp);
+				xfs_dir2_data_freeinsert(hdr, bf, newdup,
+							 needlogp);
+				xfs_dir2_data_freeinsert(hdr, bf, newdup2,
+							 needlogp);
+			}
+		}
+	}
+	*needscanp = needscan;
+	return 0;
+corrupt:
+	xfs_corruption_error(__func__, XFS_ERRLEVEL_LOW, args->dp->i_mount,
+			hdr, sizeof(*hdr), __FILE__, __LINE__, fa);
+	return -EFSCORRUPTED;
+}
+
+/* Find the end of the entry data in a data/block format dir block. */
+unsigned int
+xfs_dir3_data_end_offset(
+	struct xfs_da_geometry		*geo,
+	struct xfs_dir2_data_hdr	*hdr)
+{
+	void				*p;
+
+	switch (hdr->magic) {
+	case cpu_to_be32(XFS_DIR3_BLOCK_MAGIC):
+	case cpu_to_be32(XFS_DIR2_BLOCK_MAGIC):
+		p = xfs_dir2_block_leaf_p(xfs_dir2_block_tail_p(geo, hdr));
+		return p - (void *)hdr;
+	case cpu_to_be32(XFS_DIR3_DATA_MAGIC):
+	case cpu_to_be32(XFS_DIR2_DATA_MAGIC):
+		return geo->blksize;
+	default:
+		return 0;
+	}
+}
diff --git a/fs/xfs/libxfs/xfs_dir2_leaf.c b/fs/xfs/libxfs/xfs_dir2_leaf.c
new file mode 100644
index 000000000..cb9e950a9
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_dir2_leaf.c
@@ -0,0 +1,1824 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_inode.h"
+#include "xfs_bmap.h"
+#include "xfs_dir2.h"
+#include "xfs_dir2_priv.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+#include "xfs_trans.h"
+#include "xfs_buf_item.h"
+
+/*
+ * Local function declarations.
+ */
+static int xfs_dir2_leaf_lookup_int(xfs_da_args_t *args, struct xfs_buf **lbpp,
+				    int *indexp, struct xfs_buf **dbpp,
+				    struct xfs_dir3_icleaf_hdr *leafhdr);
+static void xfs_dir3_leaf_log_bests(struct xfs_da_args *args,
+				    struct xfs_buf *bp, int first, int last);
+static void xfs_dir3_leaf_log_tail(struct xfs_da_args *args,
+				   struct xfs_buf *bp);
+
+void
+xfs_dir2_leaf_hdr_from_disk(
+	struct xfs_mount		*mp,
+	struct xfs_dir3_icleaf_hdr	*to,
+	struct xfs_dir2_leaf		*from)
+{
+	if (xfs_has_crc(mp)) {
+		struct xfs_dir3_leaf *from3 = (struct xfs_dir3_leaf *)from;
+
+		to->forw = be32_to_cpu(from3->hdr.info.hdr.forw);
+		to->back = be32_to_cpu(from3->hdr.info.hdr.back);
+		to->magic = be16_to_cpu(from3->hdr.info.hdr.magic);
+		to->count = be16_to_cpu(from3->hdr.count);
+		to->stale = be16_to_cpu(from3->hdr.stale);
+		to->ents = from3->__ents;
+
+		ASSERT(to->magic == XFS_DIR3_LEAF1_MAGIC ||
+		       to->magic == XFS_DIR3_LEAFN_MAGIC);
+	} else {
+		to->forw = be32_to_cpu(from->hdr.info.forw);
+		to->back = be32_to_cpu(from->hdr.info.back);
+		to->magic = be16_to_cpu(from->hdr.info.magic);
+		to->count = be16_to_cpu(from->hdr.count);
+		to->stale = be16_to_cpu(from->hdr.stale);
+		to->ents = from->__ents;
+
+		ASSERT(to->magic == XFS_DIR2_LEAF1_MAGIC ||
+		       to->magic == XFS_DIR2_LEAFN_MAGIC);
+	}
+}
+
+void
+xfs_dir2_leaf_hdr_to_disk(
+	struct xfs_mount		*mp,
+	struct xfs_dir2_leaf		*to,
+	struct xfs_dir3_icleaf_hdr	*from)
+{
+	if (xfs_has_crc(mp)) {
+		struct xfs_dir3_leaf *to3 = (struct xfs_dir3_leaf *)to;
+
+		ASSERT(from->magic == XFS_DIR3_LEAF1_MAGIC ||
+		       from->magic == XFS_DIR3_LEAFN_MAGIC);
+
+		to3->hdr.info.hdr.forw = cpu_to_be32(from->forw);
+		to3->hdr.info.hdr.back = cpu_to_be32(from->back);
+		to3->hdr.info.hdr.magic = cpu_to_be16(from->magic);
+		to3->hdr.count = cpu_to_be16(from->count);
+		to3->hdr.stale = cpu_to_be16(from->stale);
+	} else {
+		ASSERT(from->magic == XFS_DIR2_LEAF1_MAGIC ||
+		       from->magic == XFS_DIR2_LEAFN_MAGIC);
+
+		to->hdr.info.forw = cpu_to_be32(from->forw);
+		to->hdr.info.back = cpu_to_be32(from->back);
+		to->hdr.info.magic = cpu_to_be16(from->magic);
+		to->hdr.count = cpu_to_be16(from->count);
+		to->hdr.stale = cpu_to_be16(from->stale);
+	}
+}
+
+/*
+ * Check the internal consistency of a leaf1 block.
+ * Pop an assert if something is wrong.
+ */
+#ifdef DEBUG
+static xfs_failaddr_t
+xfs_dir3_leaf1_check(
+	struct xfs_inode	*dp,
+	struct xfs_buf		*bp)
+{
+	struct xfs_dir2_leaf	*leaf = bp->b_addr;
+	struct xfs_dir3_icleaf_hdr leafhdr;
+
+	xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &leafhdr, leaf);
+
+	if (leafhdr.magic == XFS_DIR3_LEAF1_MAGIC) {
+		struct xfs_dir3_leaf_hdr *leaf3 = bp->b_addr;
+		if (be64_to_cpu(leaf3->info.blkno) != xfs_buf_daddr(bp))
+			return __this_address;
+	} else if (leafhdr.magic != XFS_DIR2_LEAF1_MAGIC)
+		return __this_address;
+
+	return xfs_dir3_leaf_check_int(dp->i_mount, &leafhdr, leaf, false);
+}
+
+static inline void
+xfs_dir3_leaf_check(
+	struct xfs_inode	*dp,
+	struct xfs_buf		*bp)
+{
+	xfs_failaddr_t		fa;
+
+	fa = xfs_dir3_leaf1_check(dp, bp);
+	if (!fa)
+		return;
+	xfs_corruption_error(__func__, XFS_ERRLEVEL_LOW, dp->i_mount,
+			bp->b_addr, BBTOB(bp->b_length), __FILE__, __LINE__,
+			fa);
+	ASSERT(0);
+}
+#else
+#define	xfs_dir3_leaf_check(dp, bp)
+#endif
+
+xfs_failaddr_t
+xfs_dir3_leaf_check_int(
+	struct xfs_mount		*mp,
+	struct xfs_dir3_icleaf_hdr	*hdr,
+	struct xfs_dir2_leaf		*leaf,
+	bool				expensive_checking)
+{
+	struct xfs_da_geometry		*geo = mp->m_dir_geo;
+	xfs_dir2_leaf_tail_t		*ltp;
+	int				stale;
+	int				i;
+	bool				isleaf1 = (hdr->magic == XFS_DIR2_LEAF1_MAGIC ||
+						   hdr->magic == XFS_DIR3_LEAF1_MAGIC);
+
+	ltp = xfs_dir2_leaf_tail_p(geo, leaf);
+
+	/*
+	 * XXX (dgc): This value is not restrictive enough.
+	 * Should factor in the size of the bests table as well.
+	 * We can deduce a value for that from i_disk_size.
+	 */
+	if (hdr->count > geo->leaf_max_ents)
+		return __this_address;
+
+	/* Leaves and bests don't overlap in leaf format. */
+	if (isleaf1 &&
+	    (char *)&hdr->ents[hdr->count] > (char *)xfs_dir2_leaf_bests_p(ltp))
+		return __this_address;
+
+	if (!expensive_checking)
+		return NULL;
+
+	/* Check hash value order, count stale entries.  */
+	for (i = stale = 0; i < hdr->count; i++) {
+		if (i + 1 < hdr->count) {
+			if (be32_to_cpu(hdr->ents[i].hashval) >
+					be32_to_cpu(hdr->ents[i + 1].hashval))
+				return __this_address;
+		}
+		if (hdr->ents[i].address == cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
+			stale++;
+		if (isleaf1 && xfs_dir2_dataptr_to_db(geo,
+				be32_to_cpu(hdr->ents[i].address)) >=
+				be32_to_cpu(ltp->bestcount))
+			return __this_address;
+	}
+	if (hdr->stale != stale)
+		return __this_address;
+	return NULL;
+}
+
+/*
+ * We verify the magic numbers before decoding the leaf header so that on debug
+ * kernels we don't get assertion failures in xfs_dir3_leaf_hdr_from_disk() due
+ * to incorrect magic numbers.
+ */
+static xfs_failaddr_t
+xfs_dir3_leaf_verify(
+	struct xfs_buf			*bp)
+{
+	struct xfs_mount		*mp = bp->b_mount;
+	struct xfs_dir3_icleaf_hdr	leafhdr;
+	xfs_failaddr_t			fa;
+
+	fa = xfs_da3_blkinfo_verify(bp, bp->b_addr);
+	if (fa)
+		return fa;
+
+	xfs_dir2_leaf_hdr_from_disk(mp, &leafhdr, bp->b_addr);
+	return xfs_dir3_leaf_check_int(mp, &leafhdr, bp->b_addr, true);
+}
+
+static void
+xfs_dir3_leaf_read_verify(
+	struct xfs_buf  *bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	xfs_failaddr_t		fa;
+
+	if (xfs_has_crc(mp) &&
+	     !xfs_buf_verify_cksum(bp, XFS_DIR3_LEAF_CRC_OFF))
+		xfs_verifier_error(bp, -EFSBADCRC, __this_address);
+	else {
+		fa = xfs_dir3_leaf_verify(bp);
+		if (fa)
+			xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+	}
+}
+
+static void
+xfs_dir3_leaf_write_verify(
+	struct xfs_buf  *bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_buf_log_item	*bip = bp->b_log_item;
+	struct xfs_dir3_leaf_hdr *hdr3 = bp->b_addr;
+	xfs_failaddr_t		fa;
+
+	fa = xfs_dir3_leaf_verify(bp);
+	if (fa) {
+		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+		return;
+	}
+
+	if (!xfs_has_crc(mp))
+		return;
+
+	if (bip)
+		hdr3->info.lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+	xfs_buf_update_cksum(bp, XFS_DIR3_LEAF_CRC_OFF);
+}
+
+const struct xfs_buf_ops xfs_dir3_leaf1_buf_ops = {
+	.name = "xfs_dir3_leaf1",
+	.magic16 = { cpu_to_be16(XFS_DIR2_LEAF1_MAGIC),
+		     cpu_to_be16(XFS_DIR3_LEAF1_MAGIC) },
+	.verify_read = xfs_dir3_leaf_read_verify,
+	.verify_write = xfs_dir3_leaf_write_verify,
+	.verify_struct = xfs_dir3_leaf_verify,
+};
+
+const struct xfs_buf_ops xfs_dir3_leafn_buf_ops = {
+	.name = "xfs_dir3_leafn",
+	.magic16 = { cpu_to_be16(XFS_DIR2_LEAFN_MAGIC),
+		     cpu_to_be16(XFS_DIR3_LEAFN_MAGIC) },
+	.verify_read = xfs_dir3_leaf_read_verify,
+	.verify_write = xfs_dir3_leaf_write_verify,
+	.verify_struct = xfs_dir3_leaf_verify,
+};
+
+int
+xfs_dir3_leaf_read(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*dp,
+	xfs_dablk_t		fbno,
+	struct xfs_buf		**bpp)
+{
+	int			err;
+
+	err = xfs_da_read_buf(tp, dp, fbno, 0, bpp, XFS_DATA_FORK,
+			&xfs_dir3_leaf1_buf_ops);
+	if (!err && tp && *bpp)
+		xfs_trans_buf_set_type(tp, *bpp, XFS_BLFT_DIR_LEAF1_BUF);
+	return err;
+}
+
+int
+xfs_dir3_leafn_read(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*dp,
+	xfs_dablk_t		fbno,
+	struct xfs_buf		**bpp)
+{
+	int			err;
+
+	err = xfs_da_read_buf(tp, dp, fbno, 0, bpp, XFS_DATA_FORK,
+			&xfs_dir3_leafn_buf_ops);
+	if (!err && tp && *bpp)
+		xfs_trans_buf_set_type(tp, *bpp, XFS_BLFT_DIR_LEAFN_BUF);
+	return err;
+}
+
+/*
+ * Initialize a new leaf block, leaf1 or leafn magic accepted.
+ */
+static void
+xfs_dir3_leaf_init(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_buf		*bp,
+	xfs_ino_t		owner,
+	uint16_t		type)
+{
+	struct xfs_dir2_leaf	*leaf = bp->b_addr;
+
+	ASSERT(type == XFS_DIR2_LEAF1_MAGIC || type == XFS_DIR2_LEAFN_MAGIC);
+
+	if (xfs_has_crc(mp)) {
+		struct xfs_dir3_leaf_hdr *leaf3 = bp->b_addr;
+
+		memset(leaf3, 0, sizeof(*leaf3));
+
+		leaf3->info.hdr.magic = (type == XFS_DIR2_LEAF1_MAGIC)
+					 ? cpu_to_be16(XFS_DIR3_LEAF1_MAGIC)
+					 : cpu_to_be16(XFS_DIR3_LEAFN_MAGIC);
+		leaf3->info.blkno = cpu_to_be64(xfs_buf_daddr(bp));
+		leaf3->info.owner = cpu_to_be64(owner);
+		uuid_copy(&leaf3->info.uuid, &mp->m_sb.sb_meta_uuid);
+	} else {
+		memset(leaf, 0, sizeof(*leaf));
+		leaf->hdr.info.magic = cpu_to_be16(type);
+	}
+
+	/*
+	 * If it's a leaf-format directory initialize the tail.
+	 * Caller is responsible for initialising the bests table.
+	 */
+	if (type == XFS_DIR2_LEAF1_MAGIC) {
+		struct xfs_dir2_leaf_tail *ltp;
+
+		ltp = xfs_dir2_leaf_tail_p(mp->m_dir_geo, leaf);
+		ltp->bestcount = 0;
+		bp->b_ops = &xfs_dir3_leaf1_buf_ops;
+		xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DIR_LEAF1_BUF);
+	} else {
+		bp->b_ops = &xfs_dir3_leafn_buf_ops;
+		xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DIR_LEAFN_BUF);
+	}
+}
+
+int
+xfs_dir3_leaf_get_buf(
+	xfs_da_args_t		*args,
+	xfs_dir2_db_t		bno,
+	struct xfs_buf		**bpp,
+	uint16_t		magic)
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_trans	*tp = args->trans;
+	struct xfs_mount	*mp = dp->i_mount;
+	struct xfs_buf		*bp;
+	int			error;
+
+	ASSERT(magic == XFS_DIR2_LEAF1_MAGIC || magic == XFS_DIR2_LEAFN_MAGIC);
+	ASSERT(bno >= xfs_dir2_byte_to_db(args->geo, XFS_DIR2_LEAF_OFFSET) &&
+	       bno < xfs_dir2_byte_to_db(args->geo, XFS_DIR2_FREE_OFFSET));
+
+	error = xfs_da_get_buf(tp, dp, xfs_dir2_db_to_da(args->geo, bno),
+			       &bp, XFS_DATA_FORK);
+	if (error)
+		return error;
+
+	xfs_dir3_leaf_init(mp, tp, bp, dp->i_ino, magic);
+	xfs_dir3_leaf_log_header(args, bp);
+	if (magic == XFS_DIR2_LEAF1_MAGIC)
+		xfs_dir3_leaf_log_tail(args, bp);
+	*bpp = bp;
+	return 0;
+}
+
+/*
+ * Convert a block form directory to a leaf form directory.
+ */
+int						/* error */
+xfs_dir2_block_to_leaf(
+	xfs_da_args_t		*args,		/* operation arguments */
+	struct xfs_buf		*dbp)		/* input block's buffer */
+{
+	__be16			*bestsp;	/* leaf's bestsp entries */
+	xfs_dablk_t		blkno;		/* leaf block's bno */
+	xfs_dir2_data_hdr_t	*hdr;		/* block header */
+	xfs_dir2_leaf_entry_t	*blp;		/* block's leaf entries */
+	xfs_dir2_block_tail_t	*btp;		/* block's tail */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return code */
+	struct xfs_buf		*lbp;		/* leaf block's buffer */
+	xfs_dir2_db_t		ldb;		/* leaf block's bno */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_tail_t	*ltp;		/* leaf's tail */
+	int			needlog;	/* need to log block header */
+	int			needscan;	/* need to rescan bestfree */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	struct xfs_dir2_data_free *bf;
+	struct xfs_dir3_icleaf_hdr leafhdr;
+
+	trace_xfs_dir2_block_to_leaf(args);
+
+	dp = args->dp;
+	tp = args->trans;
+	/*
+	 * Add the leaf block to the inode.
+	 * This interface will only put blocks in the leaf/node range.
+	 * Since that's empty now, we'll get the root (block 0 in range).
+	 */
+	if ((error = xfs_da_grow_inode(args, &blkno))) {
+		return error;
+	}
+	ldb = xfs_dir2_da_to_db(args->geo, blkno);
+	ASSERT(ldb == xfs_dir2_byte_to_db(args->geo, XFS_DIR2_LEAF_OFFSET));
+	/*
+	 * Initialize the leaf block, get a buffer for it.
+	 */
+	error = xfs_dir3_leaf_get_buf(args, ldb, &lbp, XFS_DIR2_LEAF1_MAGIC);
+	if (error)
+		return error;
+
+	leaf = lbp->b_addr;
+	hdr = dbp->b_addr;
+	xfs_dir3_data_check(dp, dbp);
+	btp = xfs_dir2_block_tail_p(args->geo, hdr);
+	blp = xfs_dir2_block_leaf_p(btp);
+	bf = xfs_dir2_data_bestfree_p(dp->i_mount, hdr);
+
+	/*
+	 * Set the counts in the leaf header.
+	 */
+	xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &leafhdr, leaf);
+	leafhdr.count = be32_to_cpu(btp->count);
+	leafhdr.stale = be32_to_cpu(btp->stale);
+	xfs_dir2_leaf_hdr_to_disk(dp->i_mount, leaf, &leafhdr);
+	xfs_dir3_leaf_log_header(args, lbp);
+
+	/*
+	 * Could compact these but I think we always do the conversion
+	 * after squeezing out stale entries.
+	 */
+	memcpy(leafhdr.ents, blp,
+		be32_to_cpu(btp->count) * sizeof(struct xfs_dir2_leaf_entry));
+	xfs_dir3_leaf_log_ents(args, &leafhdr, lbp, 0, leafhdr.count - 1);
+	needscan = 0;
+	needlog = 1;
+	/*
+	 * Make the space formerly occupied by the leaf entries and block
+	 * tail be free.
+	 */
+	xfs_dir2_data_make_free(args, dbp,
+		(xfs_dir2_data_aoff_t)((char *)blp - (char *)hdr),
+		(xfs_dir2_data_aoff_t)((char *)hdr + args->geo->blksize -
+				       (char *)blp),
+		&needlog, &needscan);
+	/*
+	 * Fix up the block header, make it a data block.
+	 */
+	dbp->b_ops = &xfs_dir3_data_buf_ops;
+	xfs_trans_buf_set_type(tp, dbp, XFS_BLFT_DIR_DATA_BUF);
+	if (hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC))
+		hdr->magic = cpu_to_be32(XFS_DIR2_DATA_MAGIC);
+	else
+		hdr->magic = cpu_to_be32(XFS_DIR3_DATA_MAGIC);
+
+	if (needscan)
+		xfs_dir2_data_freescan(dp->i_mount, hdr, &needlog);
+	/*
+	 * Set up leaf tail and bests table.
+	 */
+	ltp = xfs_dir2_leaf_tail_p(args->geo, leaf);
+	ltp->bestcount = cpu_to_be32(1);
+	bestsp = xfs_dir2_leaf_bests_p(ltp);
+	bestsp[0] =  bf[0].length;
+	/*
+	 * Log the data header and leaf bests table.
+	 */
+	if (needlog)
+		xfs_dir2_data_log_header(args, dbp);
+	xfs_dir3_leaf_check(dp, lbp);
+	xfs_dir3_data_check(dp, dbp);
+	xfs_dir3_leaf_log_bests(args, lbp, 0, 0);
+	return 0;
+}
+
+STATIC void
+xfs_dir3_leaf_find_stale(
+	struct xfs_dir3_icleaf_hdr *leafhdr,
+	struct xfs_dir2_leaf_entry *ents,
+	int			index,
+	int			*lowstale,
+	int			*highstale)
+{
+	/*
+	 * Find the first stale entry before our index, if any.
+	 */
+	for (*lowstale = index - 1; *lowstale >= 0; --*lowstale) {
+		if (ents[*lowstale].address ==
+		    cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
+			break;
+	}
+
+	/*
+	 * Find the first stale entry at or after our index, if any.
+	 * Stop if the result would require moving more entries than using
+	 * lowstale.
+	 */
+	for (*highstale = index; *highstale < leafhdr->count; ++*highstale) {
+		if (ents[*highstale].address ==
+		    cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
+			break;
+		if (*lowstale >= 0 && index - *lowstale <= *highstale - index)
+			break;
+	}
+}
+
+struct xfs_dir2_leaf_entry *
+xfs_dir3_leaf_find_entry(
+	struct xfs_dir3_icleaf_hdr *leafhdr,
+	struct xfs_dir2_leaf_entry *ents,
+	int			index,		/* leaf table position */
+	int			compact,	/* need to compact leaves */
+	int			lowstale,	/* index of prev stale leaf */
+	int			highstale,	/* index of next stale leaf */
+	int			*lfloglow,	/* low leaf logging index */
+	int			*lfloghigh)	/* high leaf logging index */
+{
+	if (!leafhdr->stale) {
+		xfs_dir2_leaf_entry_t	*lep;	/* leaf entry table pointer */
+
+		/*
+		 * Now we need to make room to insert the leaf entry.
+		 *
+		 * If there are no stale entries, just insert a hole at index.
+		 */
+		lep = &ents[index];
+		if (index < leafhdr->count)
+			memmove(lep + 1, lep,
+				(leafhdr->count - index) * sizeof(*lep));
+
+		/*
+		 * Record low and high logging indices for the leaf.
+		 */
+		*lfloglow = index;
+		*lfloghigh = leafhdr->count++;
+		return lep;
+	}
+
+	/*
+	 * There are stale entries.
+	 *
+	 * We will use one of them for the new entry.  It's probably not at
+	 * the right location, so we'll have to shift some up or down first.
+	 *
+	 * If we didn't compact before, we need to find the nearest stale
+	 * entries before and after our insertion point.
+	 */
+	if (compact == 0)
+		xfs_dir3_leaf_find_stale(leafhdr, ents, index,
+					 &lowstale, &highstale);
+
+	/*
+	 * If the low one is better, use it.
+	 */
+	if (lowstale >= 0 &&
+	    (highstale == leafhdr->count ||
+	     index - lowstale - 1 < highstale - index)) {
+		ASSERT(index - lowstale - 1 >= 0);
+		ASSERT(ents[lowstale].address ==
+		       cpu_to_be32(XFS_DIR2_NULL_DATAPTR));
+
+		/*
+		 * Copy entries up to cover the stale entry and make room
+		 * for the new entry.
+		 */
+		if (index - lowstale - 1 > 0) {
+			memmove(&ents[lowstale], &ents[lowstale + 1],
+				(index - lowstale - 1) *
+					sizeof(xfs_dir2_leaf_entry_t));
+		}
+		*lfloglow = min(lowstale, *lfloglow);
+		*lfloghigh = max(index - 1, *lfloghigh);
+		leafhdr->stale--;
+		return &ents[index - 1];
+	}
+
+	/*
+	 * The high one is better, so use that one.
+	 */
+	ASSERT(highstale - index >= 0);
+	ASSERT(ents[highstale].address == cpu_to_be32(XFS_DIR2_NULL_DATAPTR));
+
+	/*
+	 * Copy entries down to cover the stale entry and make room for the
+	 * new entry.
+	 */
+	if (highstale - index > 0) {
+		memmove(&ents[index + 1], &ents[index],
+			(highstale - index) * sizeof(xfs_dir2_leaf_entry_t));
+	}
+	*lfloglow = min(index, *lfloglow);
+	*lfloghigh = max(highstale, *lfloghigh);
+	leafhdr->stale--;
+	return &ents[index];
+}
+
+/*
+ * Add an entry to a leaf form directory.
+ */
+int						/* error */
+xfs_dir2_leaf_addname(
+	struct xfs_da_args	*args)		/* operation arguments */
+{
+	struct xfs_dir3_icleaf_hdr leafhdr;
+	struct xfs_trans	*tp = args->trans;
+	__be16			*bestsp;	/* freespace table in leaf */
+	__be16			*tagp;		/* end of data entry */
+	struct xfs_buf		*dbp;		/* data block buffer */
+	struct xfs_buf		*lbp;		/* leaf's buffer */
+	struct xfs_dir2_leaf	*leaf;		/* leaf structure */
+	struct xfs_inode	*dp = args->dp;	/* incore directory inode */
+	struct xfs_dir2_data_hdr *hdr;		/* data block header */
+	struct xfs_dir2_data_entry *dep;	/* data block entry */
+	struct xfs_dir2_leaf_entry *lep;	/* leaf entry table pointer */
+	struct xfs_dir2_leaf_entry *ents;
+	struct xfs_dir2_data_unused *dup;	/* data unused entry */
+	struct xfs_dir2_leaf_tail *ltp;		/* leaf tail pointer */
+	struct xfs_dir2_data_free *bf;		/* bestfree table */
+	int			compact;	/* need to compact leaves */
+	int			error;		/* error return value */
+	int			grown;		/* allocated new data block */
+	int			highstale = 0;	/* index of next stale leaf */
+	int			i;		/* temporary, index */
+	int			index;		/* leaf table position */
+	int			length;		/* length of new entry */
+	int			lfloglow;	/* low leaf logging index */
+	int			lfloghigh;	/* high leaf logging index */
+	int			lowstale = 0;	/* index of prev stale leaf */
+	int			needbytes;	/* leaf block bytes needed */
+	int			needlog;	/* need to log data header */
+	int			needscan;	/* need to rescan data free */
+	xfs_dir2_db_t		use_block;	/* data block number */
+
+	trace_xfs_dir2_leaf_addname(args);
+
+	error = xfs_dir3_leaf_read(tp, dp, args->geo->leafblk, &lbp);
+	if (error)
+		return error;
+
+	/*
+	 * Look up the entry by hash value and name.
+	 * We know it's not there, our caller has already done a lookup.
+	 * So the index is of the entry to insert in front of.
+	 * But if there are dup hash values the index is of the first of those.
+	 */
+	index = xfs_dir2_leaf_search_hash(args, lbp);
+	leaf = lbp->b_addr;
+	ltp = xfs_dir2_leaf_tail_p(args->geo, leaf);
+	xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &leafhdr, leaf);
+	ents = leafhdr.ents;
+	bestsp = xfs_dir2_leaf_bests_p(ltp);
+	length = xfs_dir2_data_entsize(dp->i_mount, args->namelen);
+
+	/*
+	 * See if there are any entries with the same hash value
+	 * and space in their block for the new entry.
+	 * This is good because it puts multiple same-hash value entries
+	 * in a data block, improving the lookup of those entries.
+	 */
+	for (use_block = -1, lep = &ents[index];
+	     index < leafhdr.count && be32_to_cpu(lep->hashval) == args->hashval;
+	     index++, lep++) {
+		if (be32_to_cpu(lep->address) == XFS_DIR2_NULL_DATAPTR)
+			continue;
+		i = xfs_dir2_dataptr_to_db(args->geo, be32_to_cpu(lep->address));
+		ASSERT(i < be32_to_cpu(ltp->bestcount));
+		ASSERT(bestsp[i] != cpu_to_be16(NULLDATAOFF));
+		if (be16_to_cpu(bestsp[i]) >= length) {
+			use_block = i;
+			break;
+		}
+	}
+	/*
+	 * Didn't find a block yet, linear search all the data blocks.
+	 */
+	if (use_block == -1) {
+		for (i = 0; i < be32_to_cpu(ltp->bestcount); i++) {
+			/*
+			 * Remember a block we see that's missing.
+			 */
+			if (bestsp[i] == cpu_to_be16(NULLDATAOFF) &&
+			    use_block == -1)
+				use_block = i;
+			else if (be16_to_cpu(bestsp[i]) >= length) {
+				use_block = i;
+				break;
+			}
+		}
+	}
+	/*
+	 * How many bytes do we need in the leaf block?
+	 */
+	needbytes = 0;
+	if (!leafhdr.stale)
+		needbytes += sizeof(xfs_dir2_leaf_entry_t);
+	if (use_block == -1)
+		needbytes += sizeof(xfs_dir2_data_off_t);
+
+	/*
+	 * Now kill use_block if it refers to a missing block, so we
+	 * can use it as an indication of allocation needed.
+	 */
+	if (use_block != -1 && bestsp[use_block] == cpu_to_be16(NULLDATAOFF))
+		use_block = -1;
+	/*
+	 * If we don't have enough free bytes but we can make enough
+	 * by compacting out stale entries, we'll do that.
+	 */
+	if ((char *)bestsp - (char *)&ents[leafhdr.count] < needbytes &&
+	    leafhdr.stale > 1)
+		compact = 1;
+
+	/*
+	 * Otherwise if we don't have enough free bytes we need to
+	 * convert to node form.
+	 */
+	else if ((char *)bestsp - (char *)&ents[leafhdr.count] < needbytes) {
+		/*
+		 * Just checking or no space reservation, give up.
+		 */
+		if ((args->op_flags & XFS_DA_OP_JUSTCHECK) ||
+							args->total == 0) {
+			xfs_trans_brelse(tp, lbp);
+			return -ENOSPC;
+		}
+		/*
+		 * Convert to node form.
+		 */
+		error = xfs_dir2_leaf_to_node(args, lbp);
+		if (error)
+			return error;
+		/*
+		 * Then add the new entry.
+		 */
+		return xfs_dir2_node_addname(args);
+	}
+	/*
+	 * Otherwise it will fit without compaction.
+	 */
+	else
+		compact = 0;
+	/*
+	 * If just checking, then it will fit unless we needed to allocate
+	 * a new data block.
+	 */
+	if (args->op_flags & XFS_DA_OP_JUSTCHECK) {
+		xfs_trans_brelse(tp, lbp);
+		return use_block == -1 ? -ENOSPC : 0;
+	}
+	/*
+	 * If no allocations are allowed, return now before we've
+	 * changed anything.
+	 */
+	if (args->total == 0 && use_block == -1) {
+		xfs_trans_brelse(tp, lbp);
+		return -ENOSPC;
+	}
+	/*
+	 * Need to compact the leaf entries, removing stale ones.
+	 * Leave one stale entry behind - the one closest to our
+	 * insertion index - and we'll shift that one to our insertion
+	 * point later.
+	 */
+	if (compact) {
+		xfs_dir3_leaf_compact_x1(&leafhdr, ents, &index, &lowstale,
+			&highstale, &lfloglow, &lfloghigh);
+	}
+	/*
+	 * There are stale entries, so we'll need log-low and log-high
+	 * impossibly bad values later.
+	 */
+	else if (leafhdr.stale) {
+		lfloglow = leafhdr.count;
+		lfloghigh = -1;
+	}
+	/*
+	 * If there was no data block space found, we need to allocate
+	 * a new one.
+	 */
+	if (use_block == -1) {
+		/*
+		 * Add the new data block.
+		 */
+		if ((error = xfs_dir2_grow_inode(args, XFS_DIR2_DATA_SPACE,
+				&use_block))) {
+			xfs_trans_brelse(tp, lbp);
+			return error;
+		}
+		/*
+		 * Initialize the block.
+		 */
+		if ((error = xfs_dir3_data_init(args, use_block, &dbp))) {
+			xfs_trans_brelse(tp, lbp);
+			return error;
+		}
+		/*
+		 * If we're adding a new data block on the end we need to
+		 * extend the bests table.  Copy it up one entry.
+		 */
+		if (use_block >= be32_to_cpu(ltp->bestcount)) {
+			bestsp--;
+			memmove(&bestsp[0], &bestsp[1],
+				be32_to_cpu(ltp->bestcount) * sizeof(bestsp[0]));
+			be32_add_cpu(&ltp->bestcount, 1);
+			xfs_dir3_leaf_log_tail(args, lbp);
+			xfs_dir3_leaf_log_bests(args, lbp, 0,
+						be32_to_cpu(ltp->bestcount) - 1);
+		}
+		/*
+		 * If we're filling in a previously empty block just log it.
+		 */
+		else
+			xfs_dir3_leaf_log_bests(args, lbp, use_block, use_block);
+		hdr = dbp->b_addr;
+		bf = xfs_dir2_data_bestfree_p(dp->i_mount, hdr);
+		bestsp[use_block] = bf[0].length;
+		grown = 1;
+	} else {
+		/*
+		 * Already had space in some data block.
+		 * Just read that one in.
+		 */
+		error = xfs_dir3_data_read(tp, dp,
+				   xfs_dir2_db_to_da(args->geo, use_block),
+				   0, &dbp);
+		if (error) {
+			xfs_trans_brelse(tp, lbp);
+			return error;
+		}
+		hdr = dbp->b_addr;
+		bf = xfs_dir2_data_bestfree_p(dp->i_mount, hdr);
+		grown = 0;
+	}
+	/*
+	 * Point to the biggest freespace in our data block.
+	 */
+	dup = (xfs_dir2_data_unused_t *)
+	      ((char *)hdr + be16_to_cpu(bf[0].offset));
+	needscan = needlog = 0;
+	/*
+	 * Mark the initial part of our freespace in use for the new entry.
+	 */
+	error = xfs_dir2_data_use_free(args, dbp, dup,
+			(xfs_dir2_data_aoff_t)((char *)dup - (char *)hdr),
+			length, &needlog, &needscan);
+	if (error) {
+		xfs_trans_brelse(tp, lbp);
+		return error;
+	}
+	/*
+	 * Initialize our new entry (at last).
+	 */
+	dep = (xfs_dir2_data_entry_t *)dup;
+	dep->inumber = cpu_to_be64(args->inumber);
+	dep->namelen = args->namelen;
+	memcpy(dep->name, args->name, dep->namelen);
+	xfs_dir2_data_put_ftype(dp->i_mount, dep, args->filetype);
+	tagp = xfs_dir2_data_entry_tag_p(dp->i_mount, dep);
+	*tagp = cpu_to_be16((char *)dep - (char *)hdr);
+	/*
+	 * Need to scan fix up the bestfree table.
+	 */
+	if (needscan)
+		xfs_dir2_data_freescan(dp->i_mount, hdr, &needlog);
+	/*
+	 * Need to log the data block's header.
+	 */
+	if (needlog)
+		xfs_dir2_data_log_header(args, dbp);
+	xfs_dir2_data_log_entry(args, dbp, dep);
+	/*
+	 * If the bests table needs to be changed, do it.
+	 * Log the change unless we've already done that.
+	 */
+	if (be16_to_cpu(bestsp[use_block]) != be16_to_cpu(bf[0].length)) {
+		bestsp[use_block] = bf[0].length;
+		if (!grown)
+			xfs_dir3_leaf_log_bests(args, lbp, use_block, use_block);
+	}
+
+	lep = xfs_dir3_leaf_find_entry(&leafhdr, ents, index, compact, lowstale,
+				       highstale, &lfloglow, &lfloghigh);
+
+	/*
+	 * Fill in the new leaf entry.
+	 */
+	lep->hashval = cpu_to_be32(args->hashval);
+	lep->address = cpu_to_be32(
+				xfs_dir2_db_off_to_dataptr(args->geo, use_block,
+				be16_to_cpu(*tagp)));
+	/*
+	 * Log the leaf fields and give up the buffers.
+	 */
+	xfs_dir2_leaf_hdr_to_disk(dp->i_mount, leaf, &leafhdr);
+	xfs_dir3_leaf_log_header(args, lbp);
+	xfs_dir3_leaf_log_ents(args, &leafhdr, lbp, lfloglow, lfloghigh);
+	xfs_dir3_leaf_check(dp, lbp);
+	xfs_dir3_data_check(dp, dbp);
+	return 0;
+}
+
+/*
+ * Compact out any stale entries in the leaf.
+ * Log the header and changed leaf entries, if any.
+ */
+void
+xfs_dir3_leaf_compact(
+	xfs_da_args_t	*args,		/* operation arguments */
+	struct xfs_dir3_icleaf_hdr *leafhdr,
+	struct xfs_buf	*bp)		/* leaf buffer */
+{
+	int		from;		/* source leaf index */
+	xfs_dir2_leaf_t	*leaf;		/* leaf structure */
+	int		loglow;		/* first leaf entry to log */
+	int		to;		/* target leaf index */
+	struct xfs_inode *dp = args->dp;
+
+	leaf = bp->b_addr;
+	if (!leafhdr->stale)
+		return;
+
+	/*
+	 * Compress out the stale entries in place.
+	 */
+	for (from = to = 0, loglow = -1; from < leafhdr->count; from++) {
+		if (leafhdr->ents[from].address ==
+		    cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
+			continue;
+		/*
+		 * Only actually copy the entries that are different.
+		 */
+		if (from > to) {
+			if (loglow == -1)
+				loglow = to;
+			leafhdr->ents[to] = leafhdr->ents[from];
+		}
+		to++;
+	}
+	/*
+	 * Update and log the header, log the leaf entries.
+	 */
+	ASSERT(leafhdr->stale == from - to);
+	leafhdr->count -= leafhdr->stale;
+	leafhdr->stale = 0;
+
+	xfs_dir2_leaf_hdr_to_disk(dp->i_mount, leaf, leafhdr);
+	xfs_dir3_leaf_log_header(args, bp);
+	if (loglow != -1)
+		xfs_dir3_leaf_log_ents(args, leafhdr, bp, loglow, to - 1);
+}
+
+/*
+ * Compact the leaf entries, removing stale ones.
+ * Leave one stale entry behind - the one closest to our
+ * insertion index - and the caller will shift that one to our insertion
+ * point later.
+ * Return new insertion index, where the remaining stale entry is,
+ * and leaf logging indices.
+ */
+void
+xfs_dir3_leaf_compact_x1(
+	struct xfs_dir3_icleaf_hdr *leafhdr,
+	struct xfs_dir2_leaf_entry *ents,
+	int		*indexp,	/* insertion index */
+	int		*lowstalep,	/* out: stale entry before us */
+	int		*highstalep,	/* out: stale entry after us */
+	int		*lowlogp,	/* out: low log index */
+	int		*highlogp)	/* out: high log index */
+{
+	int		from;		/* source copy index */
+	int		highstale;	/* stale entry at/after index */
+	int		index;		/* insertion index */
+	int		keepstale;	/* source index of kept stale */
+	int		lowstale;	/* stale entry before index */
+	int		newindex=0;	/* new insertion index */
+	int		to;		/* destination copy index */
+
+	ASSERT(leafhdr->stale > 1);
+	index = *indexp;
+
+	xfs_dir3_leaf_find_stale(leafhdr, ents, index, &lowstale, &highstale);
+
+	/*
+	 * Pick the better of lowstale and highstale.
+	 */
+	if (lowstale >= 0 &&
+	    (highstale == leafhdr->count ||
+	     index - lowstale <= highstale - index))
+		keepstale = lowstale;
+	else
+		keepstale = highstale;
+	/*
+	 * Copy the entries in place, removing all the stale entries
+	 * except keepstale.
+	 */
+	for (from = to = 0; from < leafhdr->count; from++) {
+		/*
+		 * Notice the new value of index.
+		 */
+		if (index == from)
+			newindex = to;
+		if (from != keepstale &&
+		    ents[from].address == cpu_to_be32(XFS_DIR2_NULL_DATAPTR)) {
+			if (from == to)
+				*lowlogp = to;
+			continue;
+		}
+		/*
+		 * Record the new keepstale value for the insertion.
+		 */
+		if (from == keepstale)
+			lowstale = highstale = to;
+		/*
+		 * Copy only the entries that have moved.
+		 */
+		if (from > to)
+			ents[to] = ents[from];
+		to++;
+	}
+	ASSERT(from > to);
+	/*
+	 * If the insertion point was past the last entry,
+	 * set the new insertion point accordingly.
+	 */
+	if (index == from)
+		newindex = to;
+	*indexp = newindex;
+	/*
+	 * Adjust the leaf header values.
+	 */
+	leafhdr->count -= from - to;
+	leafhdr->stale = 1;
+	/*
+	 * Remember the low/high stale value only in the "right"
+	 * direction.
+	 */
+	if (lowstale >= newindex)
+		lowstale = -1;
+	else
+		highstale = leafhdr->count;
+	*highlogp = leafhdr->count - 1;
+	*lowstalep = lowstale;
+	*highstalep = highstale;
+}
+
+/*
+ * Log the bests entries indicated from a leaf1 block.
+ */
+static void
+xfs_dir3_leaf_log_bests(
+	struct xfs_da_args	*args,
+	struct xfs_buf		*bp,		/* leaf buffer */
+	int			first,		/* first entry to log */
+	int			last)		/* last entry to log */
+{
+	__be16			*firstb;	/* pointer to first entry */
+	__be16			*lastb;		/* pointer to last entry */
+	struct xfs_dir2_leaf	*leaf = bp->b_addr;
+	xfs_dir2_leaf_tail_t	*ltp;		/* leaf tail structure */
+
+	ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAF1_MAGIC) ||
+	       leaf->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAF1_MAGIC));
+
+	ltp = xfs_dir2_leaf_tail_p(args->geo, leaf);
+	firstb = xfs_dir2_leaf_bests_p(ltp) + first;
+	lastb = xfs_dir2_leaf_bests_p(ltp) + last;
+	xfs_trans_log_buf(args->trans, bp,
+		(uint)((char *)firstb - (char *)leaf),
+		(uint)((char *)lastb - (char *)leaf + sizeof(*lastb) - 1));
+}
+
+/*
+ * Log the leaf entries indicated from a leaf1 or leafn block.
+ */
+void
+xfs_dir3_leaf_log_ents(
+	struct xfs_da_args	*args,
+	struct xfs_dir3_icleaf_hdr *hdr,
+	struct xfs_buf		*bp,
+	int			first,
+	int			last)
+{
+	xfs_dir2_leaf_entry_t	*firstlep;	/* pointer to first entry */
+	xfs_dir2_leaf_entry_t	*lastlep;	/* pointer to last entry */
+	struct xfs_dir2_leaf	*leaf = bp->b_addr;
+
+	ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAF1_MAGIC) ||
+	       leaf->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAF1_MAGIC) ||
+	       leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
+	       leaf->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC));
+
+	firstlep = &hdr->ents[first];
+	lastlep = &hdr->ents[last];
+	xfs_trans_log_buf(args->trans, bp,
+		(uint)((char *)firstlep - (char *)leaf),
+		(uint)((char *)lastlep - (char *)leaf + sizeof(*lastlep) - 1));
+}
+
+/*
+ * Log the header of the leaf1 or leafn block.
+ */
+void
+xfs_dir3_leaf_log_header(
+	struct xfs_da_args	*args,
+	struct xfs_buf		*bp)
+{
+	struct xfs_dir2_leaf	*leaf = bp->b_addr;
+
+	ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAF1_MAGIC) ||
+	       leaf->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAF1_MAGIC) ||
+	       leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
+	       leaf->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC));
+
+	xfs_trans_log_buf(args->trans, bp,
+			  (uint)((char *)&leaf->hdr - (char *)leaf),
+			  args->geo->leaf_hdr_size - 1);
+}
+
+/*
+ * Log the tail of the leaf1 block.
+ */
+STATIC void
+xfs_dir3_leaf_log_tail(
+	struct xfs_da_args	*args,
+	struct xfs_buf		*bp)
+{
+	struct xfs_dir2_leaf	*leaf = bp->b_addr;
+	xfs_dir2_leaf_tail_t	*ltp;		/* leaf tail structure */
+
+	ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAF1_MAGIC) ||
+	       leaf->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAF1_MAGIC) ||
+	       leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
+	       leaf->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC));
+
+	ltp = xfs_dir2_leaf_tail_p(args->geo, leaf);
+	xfs_trans_log_buf(args->trans, bp, (uint)((char *)ltp - (char *)leaf),
+		(uint)(args->geo->blksize - 1));
+}
+
+/*
+ * Look up the entry referred to by args in the leaf format directory.
+ * Most of the work is done by the xfs_dir2_leaf_lookup_int routine which
+ * is also used by the node-format code.
+ */
+int
+xfs_dir2_leaf_lookup(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	struct xfs_buf		*dbp;		/* data block buffer */
+	xfs_dir2_data_entry_t	*dep;		/* data block entry */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return code */
+	int			index;		/* found entry index */
+	struct xfs_buf		*lbp;		/* leaf buffer */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	struct xfs_dir3_icleaf_hdr leafhdr;
+
+	trace_xfs_dir2_leaf_lookup(args);
+
+	/*
+	 * Look up name in the leaf block, returning both buffers and index.
+	 */
+	error = xfs_dir2_leaf_lookup_int(args, &lbp, &index, &dbp, &leafhdr);
+	if (error)
+		return error;
+
+	tp = args->trans;
+	dp = args->dp;
+	xfs_dir3_leaf_check(dp, lbp);
+
+	/*
+	 * Get to the leaf entry and contained data entry address.
+	 */
+	lep = &leafhdr.ents[index];
+
+	/*
+	 * Point to the data entry.
+	 */
+	dep = (xfs_dir2_data_entry_t *)
+	      ((char *)dbp->b_addr +
+	       xfs_dir2_dataptr_to_off(args->geo, be32_to_cpu(lep->address)));
+	/*
+	 * Return the found inode number & CI name if appropriate
+	 */
+	args->inumber = be64_to_cpu(dep->inumber);
+	args->filetype = xfs_dir2_data_get_ftype(dp->i_mount, dep);
+	error = xfs_dir_cilookup_result(args, dep->name, dep->namelen);
+	xfs_trans_brelse(tp, dbp);
+	xfs_trans_brelse(tp, lbp);
+	return error;
+}
+
+/*
+ * Look up name/hash in the leaf block.
+ * Fill in indexp with the found index, and dbpp with the data buffer.
+ * If not found dbpp will be NULL, and ENOENT comes back.
+ * lbpp will always be filled in with the leaf buffer unless there's an error.
+ */
+static int					/* error */
+xfs_dir2_leaf_lookup_int(
+	xfs_da_args_t		*args,		/* operation arguments */
+	struct xfs_buf		**lbpp,		/* out: leaf buffer */
+	int			*indexp,	/* out: index in leaf block */
+	struct xfs_buf		**dbpp,		/* out: data buffer */
+	struct xfs_dir3_icleaf_hdr *leafhdr)
+{
+	xfs_dir2_db_t		curdb = -1;	/* current data block number */
+	struct xfs_buf		*dbp = NULL;	/* data buffer */
+	xfs_dir2_data_entry_t	*dep;		/* data entry */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return code */
+	int			index;		/* index in leaf block */
+	struct xfs_buf		*lbp;		/* leaf buffer */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	xfs_dir2_db_t		newdb;		/* new data block number */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	xfs_dir2_db_t		cidb = -1;	/* case match data block no. */
+	enum xfs_dacmp		cmp;		/* name compare result */
+
+	dp = args->dp;
+	tp = args->trans;
+	mp = dp->i_mount;
+
+	error = xfs_dir3_leaf_read(tp, dp, args->geo->leafblk, &lbp);
+	if (error)
+		return error;
+
+	*lbpp = lbp;
+	leaf = lbp->b_addr;
+	xfs_dir3_leaf_check(dp, lbp);
+	xfs_dir2_leaf_hdr_from_disk(mp, leafhdr, leaf);
+
+	/*
+	 * Look for the first leaf entry with our hash value.
+	 */
+	index = xfs_dir2_leaf_search_hash(args, lbp);
+	/*
+	 * Loop over all the entries with the right hash value
+	 * looking to match the name.
+	 */
+	for (lep = &leafhdr->ents[index];
+	     index < leafhdr->count &&
+			be32_to_cpu(lep->hashval) == args->hashval;
+	     lep++, index++) {
+		/*
+		 * Skip over stale leaf entries.
+		 */
+		if (be32_to_cpu(lep->address) == XFS_DIR2_NULL_DATAPTR)
+			continue;
+		/*
+		 * Get the new data block number.
+		 */
+		newdb = xfs_dir2_dataptr_to_db(args->geo,
+					       be32_to_cpu(lep->address));
+		/*
+		 * If it's not the same as the old data block number,
+		 * need to pitch the old one and read the new one.
+		 */
+		if (newdb != curdb) {
+			if (dbp)
+				xfs_trans_brelse(tp, dbp);
+			error = xfs_dir3_data_read(tp, dp,
+					   xfs_dir2_db_to_da(args->geo, newdb),
+					   0, &dbp);
+			if (error) {
+				xfs_trans_brelse(tp, lbp);
+				return error;
+			}
+			curdb = newdb;
+		}
+		/*
+		 * Point to the data entry.
+		 */
+		dep = (xfs_dir2_data_entry_t *)((char *)dbp->b_addr +
+			xfs_dir2_dataptr_to_off(args->geo,
+						be32_to_cpu(lep->address)));
+		/*
+		 * Compare name and if it's an exact match, return the index
+		 * and buffer. If it's the first case-insensitive match, store
+		 * the index and buffer and continue looking for an exact match.
+		 */
+		cmp = xfs_dir2_compname(args, dep->name, dep->namelen);
+		if (cmp != XFS_CMP_DIFFERENT && cmp != args->cmpresult) {
+			args->cmpresult = cmp;
+			*indexp = index;
+			/* case exact match: return the current buffer. */
+			if (cmp == XFS_CMP_EXACT) {
+				*dbpp = dbp;
+				return 0;
+			}
+			cidb = curdb;
+		}
+	}
+	ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
+	/*
+	 * Here, we can only be doing a lookup (not a rename or remove).
+	 * If a case-insensitive match was found earlier, re-read the
+	 * appropriate data block if required and return it.
+	 */
+	if (args->cmpresult == XFS_CMP_CASE) {
+		ASSERT(cidb != -1);
+		if (cidb != curdb) {
+			xfs_trans_brelse(tp, dbp);
+			error = xfs_dir3_data_read(tp, dp,
+					   xfs_dir2_db_to_da(args->geo, cidb),
+					   0, &dbp);
+			if (error) {
+				xfs_trans_brelse(tp, lbp);
+				return error;
+			}
+		}
+		*dbpp = dbp;
+		return 0;
+	}
+	/*
+	 * No match found, return -ENOENT.
+	 */
+	ASSERT(cidb == -1);
+	if (dbp)
+		xfs_trans_brelse(tp, dbp);
+	xfs_trans_brelse(tp, lbp);
+	return -ENOENT;
+}
+
+/*
+ * Remove an entry from a leaf format directory.
+ */
+int						/* error */
+xfs_dir2_leaf_removename(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	struct xfs_da_geometry	*geo = args->geo;
+	__be16			*bestsp;	/* leaf block best freespace */
+	xfs_dir2_data_hdr_t	*hdr;		/* data block header */
+	xfs_dir2_db_t		db;		/* data block number */
+	struct xfs_buf		*dbp;		/* data block buffer */
+	xfs_dir2_data_entry_t	*dep;		/* data entry structure */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return code */
+	xfs_dir2_db_t		i;		/* temporary data block # */
+	int			index;		/* index into leaf entries */
+	struct xfs_buf		*lbp;		/* leaf buffer */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	xfs_dir2_leaf_tail_t	*ltp;		/* leaf tail structure */
+	int			needlog;	/* need to log data header */
+	int			needscan;	/* need to rescan data frees */
+	xfs_dir2_data_off_t	oldbest;	/* old value of best free */
+	struct xfs_dir2_data_free *bf;		/* bestfree table */
+	struct xfs_dir3_icleaf_hdr leafhdr;
+
+	trace_xfs_dir2_leaf_removename(args);
+
+	/*
+	 * Lookup the leaf entry, get the leaf and data blocks read in.
+	 */
+	error = xfs_dir2_leaf_lookup_int(args, &lbp, &index, &dbp, &leafhdr);
+	if (error)
+		return error;
+
+	dp = args->dp;
+	leaf = lbp->b_addr;
+	hdr = dbp->b_addr;
+	xfs_dir3_data_check(dp, dbp);
+	bf = xfs_dir2_data_bestfree_p(dp->i_mount, hdr);
+
+	/*
+	 * Point to the leaf entry, use that to point to the data entry.
+	 */
+	lep = &leafhdr.ents[index];
+	db = xfs_dir2_dataptr_to_db(geo, be32_to_cpu(lep->address));
+	dep = (xfs_dir2_data_entry_t *)((char *)hdr +
+		xfs_dir2_dataptr_to_off(geo, be32_to_cpu(lep->address)));
+	needscan = needlog = 0;
+	oldbest = be16_to_cpu(bf[0].length);
+	ltp = xfs_dir2_leaf_tail_p(geo, leaf);
+	bestsp = xfs_dir2_leaf_bests_p(ltp);
+	if (be16_to_cpu(bestsp[db]) != oldbest) {
+		xfs_buf_mark_corrupt(lbp);
+		return -EFSCORRUPTED;
+	}
+	/*
+	 * Mark the former data entry unused.
+	 */
+	xfs_dir2_data_make_free(args, dbp,
+		(xfs_dir2_data_aoff_t)((char *)dep - (char *)hdr),
+		xfs_dir2_data_entsize(dp->i_mount, dep->namelen), &needlog,
+		&needscan);
+	/*
+	 * We just mark the leaf entry stale by putting a null in it.
+	 */
+	leafhdr.stale++;
+	xfs_dir2_leaf_hdr_to_disk(dp->i_mount, leaf, &leafhdr);
+	xfs_dir3_leaf_log_header(args, lbp);
+
+	lep->address = cpu_to_be32(XFS_DIR2_NULL_DATAPTR);
+	xfs_dir3_leaf_log_ents(args, &leafhdr, lbp, index, index);
+
+	/*
+	 * Scan the freespace in the data block again if necessary,
+	 * log the data block header if necessary.
+	 */
+	if (needscan)
+		xfs_dir2_data_freescan(dp->i_mount, hdr, &needlog);
+	if (needlog)
+		xfs_dir2_data_log_header(args, dbp);
+	/*
+	 * If the longest freespace in the data block has changed,
+	 * put the new value in the bests table and log that.
+	 */
+	if (be16_to_cpu(bf[0].length) != oldbest) {
+		bestsp[db] = bf[0].length;
+		xfs_dir3_leaf_log_bests(args, lbp, db, db);
+	}
+	xfs_dir3_data_check(dp, dbp);
+	/*
+	 * If the data block is now empty then get rid of the data block.
+	 */
+	if (be16_to_cpu(bf[0].length) ==
+	    geo->blksize - geo->data_entry_offset) {
+		ASSERT(db != geo->datablk);
+		if ((error = xfs_dir2_shrink_inode(args, db, dbp))) {
+			/*
+			 * Nope, can't get rid of it because it caused
+			 * allocation of a bmap btree block to do so.
+			 * Just go on, returning success, leaving the
+			 * empty block in place.
+			 */
+			if (error == -ENOSPC && args->total == 0)
+				error = 0;
+			xfs_dir3_leaf_check(dp, lbp);
+			return error;
+		}
+		dbp = NULL;
+		/*
+		 * If this is the last data block then compact the
+		 * bests table by getting rid of entries.
+		 */
+		if (db == be32_to_cpu(ltp->bestcount) - 1) {
+			/*
+			 * Look for the last active entry (i).
+			 */
+			for (i = db - 1; i > 0; i--) {
+				if (bestsp[i] != cpu_to_be16(NULLDATAOFF))
+					break;
+			}
+			/*
+			 * Copy the table down so inactive entries at the
+			 * end are removed.
+			 */
+			memmove(&bestsp[db - i], bestsp,
+				(be32_to_cpu(ltp->bestcount) - (db - i)) * sizeof(*bestsp));
+			be32_add_cpu(&ltp->bestcount, -(db - i));
+			xfs_dir3_leaf_log_tail(args, lbp);
+			xfs_dir3_leaf_log_bests(args, lbp, 0,
+						be32_to_cpu(ltp->bestcount) - 1);
+		} else
+			bestsp[db] = cpu_to_be16(NULLDATAOFF);
+	}
+	/*
+	 * If the data block was not the first one, drop it.
+	 */
+	else if (db != geo->datablk)
+		dbp = NULL;
+
+	xfs_dir3_leaf_check(dp, lbp);
+	/*
+	 * See if we can convert to block form.
+	 */
+	return xfs_dir2_leaf_to_block(args, lbp, dbp);
+}
+
+/*
+ * Replace the inode number in a leaf format directory entry.
+ */
+int						/* error */
+xfs_dir2_leaf_replace(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	struct xfs_buf		*dbp;		/* data block buffer */
+	xfs_dir2_data_entry_t	*dep;		/* data block entry */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return code */
+	int			index;		/* index of leaf entry */
+	struct xfs_buf		*lbp;		/* leaf buffer */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	struct xfs_dir3_icleaf_hdr leafhdr;
+
+	trace_xfs_dir2_leaf_replace(args);
+
+	/*
+	 * Look up the entry.
+	 */
+	error = xfs_dir2_leaf_lookup_int(args, &lbp, &index, &dbp, &leafhdr);
+	if (error)
+		return error;
+
+	dp = args->dp;
+	/*
+	 * Point to the leaf entry, get data address from it.
+	 */
+	lep = &leafhdr.ents[index];
+	/*
+	 * Point to the data entry.
+	 */
+	dep = (xfs_dir2_data_entry_t *)
+	      ((char *)dbp->b_addr +
+	       xfs_dir2_dataptr_to_off(args->geo, be32_to_cpu(lep->address)));
+	ASSERT(args->inumber != be64_to_cpu(dep->inumber));
+	/*
+	 * Put the new inode number in, log it.
+	 */
+	dep->inumber = cpu_to_be64(args->inumber);
+	xfs_dir2_data_put_ftype(dp->i_mount, dep, args->filetype);
+	tp = args->trans;
+	xfs_dir2_data_log_entry(args, dbp, dep);
+	xfs_dir3_leaf_check(dp, lbp);
+	xfs_trans_brelse(tp, lbp);
+	return 0;
+}
+
+/*
+ * Return index in the leaf block (lbp) which is either the first
+ * one with this hash value, or if there are none, the insert point
+ * for that hash value.
+ */
+int						/* index value */
+xfs_dir2_leaf_search_hash(
+	xfs_da_args_t		*args,		/* operation arguments */
+	struct xfs_buf		*lbp)		/* leaf buffer */
+{
+	xfs_dahash_t		hash=0;		/* hash from this entry */
+	xfs_dahash_t		hashwant;	/* hash value looking for */
+	int			high;		/* high leaf index */
+	int			low;		/* low leaf index */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	int			mid=0;		/* current leaf index */
+	struct xfs_dir3_icleaf_hdr leafhdr;
+
+	xfs_dir2_leaf_hdr_from_disk(args->dp->i_mount, &leafhdr, lbp->b_addr);
+
+	/*
+	 * Note, the table cannot be empty, so we have to go through the loop.
+	 * Binary search the leaf entries looking for our hash value.
+	 */
+	for (lep = leafhdr.ents, low = 0, high = leafhdr.count - 1,
+		hashwant = args->hashval;
+	     low <= high; ) {
+		mid = (low + high) >> 1;
+		if ((hash = be32_to_cpu(lep[mid].hashval)) == hashwant)
+			break;
+		if (hash < hashwant)
+			low = mid + 1;
+		else
+			high = mid - 1;
+	}
+	/*
+	 * Found one, back up through all the equal hash values.
+	 */
+	if (hash == hashwant) {
+		while (mid > 0 && be32_to_cpu(lep[mid - 1].hashval) == hashwant) {
+			mid--;
+		}
+	}
+	/*
+	 * Need to point to an entry higher than ours.
+	 */
+	else if (hash < hashwant)
+		mid++;
+	return mid;
+}
+
+/*
+ * Trim off a trailing data block.  We know it's empty since the leaf
+ * freespace table says so.
+ */
+int						/* error */
+xfs_dir2_leaf_trim_data(
+	xfs_da_args_t		*args,		/* operation arguments */
+	struct xfs_buf		*lbp,		/* leaf buffer */
+	xfs_dir2_db_t		db)		/* data block number */
+{
+	struct xfs_da_geometry	*geo = args->geo;
+	__be16			*bestsp;	/* leaf bests table */
+	struct xfs_buf		*dbp;		/* data block buffer */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return value */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_tail_t	*ltp;		/* leaf tail structure */
+	xfs_trans_t		*tp;		/* transaction pointer */
+
+	dp = args->dp;
+	tp = args->trans;
+	/*
+	 * Read the offending data block.  We need its buffer.
+	 */
+	error = xfs_dir3_data_read(tp, dp, xfs_dir2_db_to_da(geo, db), 0, &dbp);
+	if (error)
+		return error;
+
+	leaf = lbp->b_addr;
+	ltp = xfs_dir2_leaf_tail_p(geo, leaf);
+
+#ifdef DEBUG
+{
+	struct xfs_dir2_data_hdr *hdr = dbp->b_addr;
+	struct xfs_dir2_data_free *bf =
+		xfs_dir2_data_bestfree_p(dp->i_mount, hdr);
+
+	ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+	       hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC));
+	ASSERT(be16_to_cpu(bf[0].length) ==
+	       geo->blksize - geo->data_entry_offset);
+	ASSERT(db == be32_to_cpu(ltp->bestcount) - 1);
+}
+#endif
+
+	/*
+	 * Get rid of the data block.
+	 */
+	if ((error = xfs_dir2_shrink_inode(args, db, dbp))) {
+		ASSERT(error != -ENOSPC);
+		xfs_trans_brelse(tp, dbp);
+		return error;
+	}
+	/*
+	 * Eliminate the last bests entry from the table.
+	 */
+	bestsp = xfs_dir2_leaf_bests_p(ltp);
+	be32_add_cpu(&ltp->bestcount, -1);
+	memmove(&bestsp[1], &bestsp[0], be32_to_cpu(ltp->bestcount) * sizeof(*bestsp));
+	xfs_dir3_leaf_log_tail(args, lbp);
+	xfs_dir3_leaf_log_bests(args, lbp, 0, be32_to_cpu(ltp->bestcount) - 1);
+	return 0;
+}
+
+static inline size_t
+xfs_dir3_leaf_size(
+	struct xfs_dir3_icleaf_hdr	*hdr,
+	int				counts)
+{
+	int	entries;
+	int	hdrsize;
+
+	entries = hdr->count - hdr->stale;
+	if (hdr->magic == XFS_DIR2_LEAF1_MAGIC ||
+	    hdr->magic == XFS_DIR2_LEAFN_MAGIC)
+		hdrsize = sizeof(struct xfs_dir2_leaf_hdr);
+	else
+		hdrsize = sizeof(struct xfs_dir3_leaf_hdr);
+
+	return hdrsize + entries * sizeof(xfs_dir2_leaf_entry_t)
+	               + counts * sizeof(xfs_dir2_data_off_t)
+		       + sizeof(xfs_dir2_leaf_tail_t);
+}
+
+/*
+ * Convert node form directory to leaf form directory.
+ * The root of the node form dir needs to already be a LEAFN block.
+ * Just return if we can't do anything.
+ */
+int						/* error */
+xfs_dir2_node_to_leaf(
+	xfs_da_state_t		*state)		/* directory operation state */
+{
+	xfs_da_args_t		*args;		/* operation arguments */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return code */
+	struct xfs_buf		*fbp;		/* buffer for freespace block */
+	xfs_fileoff_t		fo;		/* freespace file offset */
+	struct xfs_buf		*lbp;		/* buffer for leaf block */
+	xfs_dir2_leaf_tail_t	*ltp;		/* tail of leaf structure */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	int			rval;		/* successful free trim? */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	struct xfs_dir3_icleaf_hdr leafhdr;
+	struct xfs_dir3_icfree_hdr freehdr;
+
+	/*
+	 * There's more than a leaf level in the btree, so there must
+	 * be multiple leafn blocks.  Give up.
+	 */
+	if (state->path.active > 1)
+		return 0;
+	args = state->args;
+
+	trace_xfs_dir2_node_to_leaf(args);
+
+	mp = state->mp;
+	dp = args->dp;
+	tp = args->trans;
+	/*
+	 * Get the last offset in the file.
+	 */
+	if ((error = xfs_bmap_last_offset(dp, &fo, XFS_DATA_FORK))) {
+		return error;
+	}
+	fo -= args->geo->fsbcount;
+	/*
+	 * If there are freespace blocks other than the first one,
+	 * take this opportunity to remove trailing empty freespace blocks
+	 * that may have been left behind during no-space-reservation
+	 * operations.
+	 */
+	while (fo > args->geo->freeblk) {
+		if ((error = xfs_dir2_node_trim_free(args, fo, &rval))) {
+			return error;
+		}
+		if (rval)
+			fo -= args->geo->fsbcount;
+		else
+			return 0;
+	}
+	/*
+	 * Now find the block just before the freespace block.
+	 */
+	if ((error = xfs_bmap_last_before(tp, dp, &fo, XFS_DATA_FORK))) {
+		return error;
+	}
+	/*
+	 * If it's not the single leaf block, give up.
+	 */
+	if (XFS_FSB_TO_B(mp, fo) > XFS_DIR2_LEAF_OFFSET + args->geo->blksize)
+		return 0;
+	lbp = state->path.blk[0].bp;
+	leaf = lbp->b_addr;
+	xfs_dir2_leaf_hdr_from_disk(mp, &leafhdr, leaf);
+
+	ASSERT(leafhdr.magic == XFS_DIR2_LEAFN_MAGIC ||
+	       leafhdr.magic == XFS_DIR3_LEAFN_MAGIC);
+
+	/*
+	 * Read the freespace block.
+	 */
+	error = xfs_dir2_free_read(tp, dp,  args->geo->freeblk, &fbp);
+	if (error)
+		return error;
+	xfs_dir2_free_hdr_from_disk(mp, &freehdr, fbp->b_addr);
+
+	ASSERT(!freehdr.firstdb);
+
+	/*
+	 * Now see if the leafn and free data will fit in a leaf1.
+	 * If not, release the buffer and give up.
+	 */
+	if (xfs_dir3_leaf_size(&leafhdr, freehdr.nvalid) > args->geo->blksize) {
+		xfs_trans_brelse(tp, fbp);
+		return 0;
+	}
+
+	/*
+	 * If the leaf has any stale entries in it, compress them out.
+	 */
+	if (leafhdr.stale)
+		xfs_dir3_leaf_compact(args, &leafhdr, lbp);
+
+	lbp->b_ops = &xfs_dir3_leaf1_buf_ops;
+	xfs_trans_buf_set_type(tp, lbp, XFS_BLFT_DIR_LEAF1_BUF);
+	leafhdr.magic = (leafhdr.magic == XFS_DIR2_LEAFN_MAGIC)
+					? XFS_DIR2_LEAF1_MAGIC
+					: XFS_DIR3_LEAF1_MAGIC;
+
+	/*
+	 * Set up the leaf tail from the freespace block.
+	 */
+	ltp = xfs_dir2_leaf_tail_p(args->geo, leaf);
+	ltp->bestcount = cpu_to_be32(freehdr.nvalid);
+
+	/*
+	 * Set up the leaf bests table.
+	 */
+	memcpy(xfs_dir2_leaf_bests_p(ltp), freehdr.bests,
+		freehdr.nvalid * sizeof(xfs_dir2_data_off_t));
+
+	xfs_dir2_leaf_hdr_to_disk(mp, leaf, &leafhdr);
+	xfs_dir3_leaf_log_header(args, lbp);
+	xfs_dir3_leaf_log_bests(args, lbp, 0, be32_to_cpu(ltp->bestcount) - 1);
+	xfs_dir3_leaf_log_tail(args, lbp);
+	xfs_dir3_leaf_check(dp, lbp);
+
+	/*
+	 * Get rid of the freespace block.
+	 */
+	error = xfs_dir2_shrink_inode(args,
+			xfs_dir2_byte_to_db(args->geo, XFS_DIR2_FREE_OFFSET),
+			fbp);
+	if (error) {
+		/*
+		 * This can't fail here because it can only happen when
+		 * punching out the middle of an extent, and this is an
+		 * isolated block.
+		 */
+		ASSERT(error != -ENOSPC);
+		return error;
+	}
+	fbp = NULL;
+	/*
+	 * Now see if we can convert the single-leaf directory
+	 * down to a block form directory.
+	 * This routine always kills the dabuf for the leaf, so
+	 * eliminate it from the path.
+	 */
+	error = xfs_dir2_leaf_to_block(args, lbp, NULL);
+	state->path.blk[0].bp = NULL;
+	return error;
+}
diff --git a/fs/xfs/libxfs/xfs_dir2_node.c b/fs/xfs/libxfs/xfs_dir2_node.c
new file mode 100644
index 000000000..7a03aeb9f
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_dir2_node.c
@@ -0,0 +1,2337 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_inode.h"
+#include "xfs_bmap.h"
+#include "xfs_dir2.h"
+#include "xfs_dir2_priv.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+#include "xfs_trans.h"
+#include "xfs_buf_item.h"
+#include "xfs_log.h"
+
+/*
+ * Function declarations.
+ */
+static int xfs_dir2_leafn_add(struct xfs_buf *bp, xfs_da_args_t *args,
+			      int index);
+static void xfs_dir2_leafn_rebalance(xfs_da_state_t *state,
+				     xfs_da_state_blk_t *blk1,
+				     xfs_da_state_blk_t *blk2);
+static int xfs_dir2_leafn_remove(xfs_da_args_t *args, struct xfs_buf *bp,
+				 int index, xfs_da_state_blk_t *dblk,
+				 int *rval);
+
+/*
+ * Convert data space db to the corresponding free db.
+ */
+static xfs_dir2_db_t
+xfs_dir2_db_to_fdb(struct xfs_da_geometry *geo, xfs_dir2_db_t db)
+{
+	return xfs_dir2_byte_to_db(geo, XFS_DIR2_FREE_OFFSET) +
+			(db / geo->free_max_bests);
+}
+
+/*
+ * Convert data space db to the corresponding index in a free db.
+ */
+static int
+xfs_dir2_db_to_fdindex(struct xfs_da_geometry *geo, xfs_dir2_db_t db)
+{
+	return db % geo->free_max_bests;
+}
+
+/*
+ * Check internal consistency of a leafn block.
+ */
+#ifdef DEBUG
+static xfs_failaddr_t
+xfs_dir3_leafn_check(
+	struct xfs_inode	*dp,
+	struct xfs_buf		*bp)
+{
+	struct xfs_dir2_leaf	*leaf = bp->b_addr;
+	struct xfs_dir3_icleaf_hdr leafhdr;
+
+	xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &leafhdr, leaf);
+
+	if (leafhdr.magic == XFS_DIR3_LEAFN_MAGIC) {
+		struct xfs_dir3_leaf_hdr *leaf3 = bp->b_addr;
+		if (be64_to_cpu(leaf3->info.blkno) != xfs_buf_daddr(bp))
+			return __this_address;
+	} else if (leafhdr.magic != XFS_DIR2_LEAFN_MAGIC)
+		return __this_address;
+
+	return xfs_dir3_leaf_check_int(dp->i_mount, &leafhdr, leaf, false);
+}
+
+static inline void
+xfs_dir3_leaf_check(
+	struct xfs_inode	*dp,
+	struct xfs_buf		*bp)
+{
+	xfs_failaddr_t		fa;
+
+	fa = xfs_dir3_leafn_check(dp, bp);
+	if (!fa)
+		return;
+	xfs_corruption_error(__func__, XFS_ERRLEVEL_LOW, dp->i_mount,
+			bp->b_addr, BBTOB(bp->b_length), __FILE__, __LINE__,
+			fa);
+	ASSERT(0);
+}
+#else
+#define	xfs_dir3_leaf_check(dp, bp)
+#endif
+
+static xfs_failaddr_t
+xfs_dir3_free_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_dir2_free_hdr *hdr = bp->b_addr;
+
+	if (!xfs_verify_magic(bp, hdr->magic))
+		return __this_address;
+
+	if (xfs_has_crc(mp)) {
+		struct xfs_dir3_blk_hdr *hdr3 = bp->b_addr;
+
+		if (!uuid_equal(&hdr3->uuid, &mp->m_sb.sb_meta_uuid))
+			return __this_address;
+		if (be64_to_cpu(hdr3->blkno) != xfs_buf_daddr(bp))
+			return __this_address;
+		if (!xfs_log_check_lsn(mp, be64_to_cpu(hdr3->lsn)))
+			return __this_address;
+	}
+
+	/* XXX: should bounds check the xfs_dir3_icfree_hdr here */
+
+	return NULL;
+}
+
+static void
+xfs_dir3_free_read_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	xfs_failaddr_t		fa;
+
+	if (xfs_has_crc(mp) &&
+	    !xfs_buf_verify_cksum(bp, XFS_DIR3_FREE_CRC_OFF))
+		xfs_verifier_error(bp, -EFSBADCRC, __this_address);
+	else {
+		fa = xfs_dir3_free_verify(bp);
+		if (fa)
+			xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+	}
+}
+
+static void
+xfs_dir3_free_write_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_buf_log_item	*bip = bp->b_log_item;
+	struct xfs_dir3_blk_hdr	*hdr3 = bp->b_addr;
+	xfs_failaddr_t		fa;
+
+	fa = xfs_dir3_free_verify(bp);
+	if (fa) {
+		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+		return;
+	}
+
+	if (!xfs_has_crc(mp))
+		return;
+
+	if (bip)
+		hdr3->lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+	xfs_buf_update_cksum(bp, XFS_DIR3_FREE_CRC_OFF);
+}
+
+const struct xfs_buf_ops xfs_dir3_free_buf_ops = {
+	.name = "xfs_dir3_free",
+	.magic = { cpu_to_be32(XFS_DIR2_FREE_MAGIC),
+		   cpu_to_be32(XFS_DIR3_FREE_MAGIC) },
+	.verify_read = xfs_dir3_free_read_verify,
+	.verify_write = xfs_dir3_free_write_verify,
+	.verify_struct = xfs_dir3_free_verify,
+};
+
+/* Everything ok in the free block header? */
+static xfs_failaddr_t
+xfs_dir3_free_header_check(
+	struct xfs_inode	*dp,
+	xfs_dablk_t		fbno,
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = dp->i_mount;
+	int			maxbests = mp->m_dir_geo->free_max_bests;
+	unsigned int		firstdb;
+
+	firstdb = (xfs_dir2_da_to_db(mp->m_dir_geo, fbno) -
+		   xfs_dir2_byte_to_db(mp->m_dir_geo, XFS_DIR2_FREE_OFFSET)) *
+			maxbests;
+	if (xfs_has_crc(mp)) {
+		struct xfs_dir3_free_hdr *hdr3 = bp->b_addr;
+
+		if (be32_to_cpu(hdr3->firstdb) != firstdb)
+			return __this_address;
+		if (be32_to_cpu(hdr3->nvalid) > maxbests)
+			return __this_address;
+		if (be32_to_cpu(hdr3->nvalid) < be32_to_cpu(hdr3->nused))
+			return __this_address;
+		if (be64_to_cpu(hdr3->hdr.owner) != dp->i_ino)
+			return __this_address;
+	} else {
+		struct xfs_dir2_free_hdr *hdr = bp->b_addr;
+
+		if (be32_to_cpu(hdr->firstdb) != firstdb)
+			return __this_address;
+		if (be32_to_cpu(hdr->nvalid) > maxbests)
+			return __this_address;
+		if (be32_to_cpu(hdr->nvalid) < be32_to_cpu(hdr->nused))
+			return __this_address;
+	}
+	return NULL;
+}
+
+static int
+__xfs_dir3_free_read(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*dp,
+	xfs_dablk_t		fbno,
+	unsigned int		flags,
+	struct xfs_buf		**bpp)
+{
+	xfs_failaddr_t		fa;
+	int			err;
+
+	err = xfs_da_read_buf(tp, dp, fbno, flags, bpp, XFS_DATA_FORK,
+			&xfs_dir3_free_buf_ops);
+	if (err || !*bpp)
+		return err;
+
+	/* Check things that we can't do in the verifier. */
+	fa = xfs_dir3_free_header_check(dp, fbno, *bpp);
+	if (fa) {
+		__xfs_buf_mark_corrupt(*bpp, fa);
+		xfs_trans_brelse(tp, *bpp);
+		*bpp = NULL;
+		return -EFSCORRUPTED;
+	}
+
+	/* try read returns without an error or *bpp if it lands in a hole */
+	if (tp)
+		xfs_trans_buf_set_type(tp, *bpp, XFS_BLFT_DIR_FREE_BUF);
+
+	return 0;
+}
+
+void
+xfs_dir2_free_hdr_from_disk(
+	struct xfs_mount		*mp,
+	struct xfs_dir3_icfree_hdr	*to,
+	struct xfs_dir2_free		*from)
+{
+	if (xfs_has_crc(mp)) {
+		struct xfs_dir3_free	*from3 = (struct xfs_dir3_free *)from;
+
+		to->magic = be32_to_cpu(from3->hdr.hdr.magic);
+		to->firstdb = be32_to_cpu(from3->hdr.firstdb);
+		to->nvalid = be32_to_cpu(from3->hdr.nvalid);
+		to->nused = be32_to_cpu(from3->hdr.nused);
+		to->bests = from3->bests;
+
+		ASSERT(to->magic == XFS_DIR3_FREE_MAGIC);
+	} else {
+		to->magic = be32_to_cpu(from->hdr.magic);
+		to->firstdb = be32_to_cpu(from->hdr.firstdb);
+		to->nvalid = be32_to_cpu(from->hdr.nvalid);
+		to->nused = be32_to_cpu(from->hdr.nused);
+		to->bests = from->bests;
+
+		ASSERT(to->magic == XFS_DIR2_FREE_MAGIC);
+	}
+}
+
+static void
+xfs_dir2_free_hdr_to_disk(
+	struct xfs_mount		*mp,
+	struct xfs_dir2_free		*to,
+	struct xfs_dir3_icfree_hdr	*from)
+{
+	if (xfs_has_crc(mp)) {
+		struct xfs_dir3_free	*to3 = (struct xfs_dir3_free *)to;
+
+		ASSERT(from->magic == XFS_DIR3_FREE_MAGIC);
+
+		to3->hdr.hdr.magic = cpu_to_be32(from->magic);
+		to3->hdr.firstdb = cpu_to_be32(from->firstdb);
+		to3->hdr.nvalid = cpu_to_be32(from->nvalid);
+		to3->hdr.nused = cpu_to_be32(from->nused);
+	} else {
+		ASSERT(from->magic == XFS_DIR2_FREE_MAGIC);
+
+		to->hdr.magic = cpu_to_be32(from->magic);
+		to->hdr.firstdb = cpu_to_be32(from->firstdb);
+		to->hdr.nvalid = cpu_to_be32(from->nvalid);
+		to->hdr.nused = cpu_to_be32(from->nused);
+	}
+}
+
+int
+xfs_dir2_free_read(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*dp,
+	xfs_dablk_t		fbno,
+	struct xfs_buf		**bpp)
+{
+	return __xfs_dir3_free_read(tp, dp, fbno, 0, bpp);
+}
+
+static int
+xfs_dir2_free_try_read(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*dp,
+	xfs_dablk_t		fbno,
+	struct xfs_buf		**bpp)
+{
+	return __xfs_dir3_free_read(tp, dp, fbno, XFS_DABUF_MAP_HOLE_OK, bpp);
+}
+
+static int
+xfs_dir3_free_get_buf(
+	xfs_da_args_t		*args,
+	xfs_dir2_db_t		fbno,
+	struct xfs_buf		**bpp)
+{
+	struct xfs_trans	*tp = args->trans;
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	struct xfs_buf		*bp;
+	int			error;
+	struct xfs_dir3_icfree_hdr hdr;
+
+	error = xfs_da_get_buf(tp, dp, xfs_dir2_db_to_da(args->geo, fbno),
+			&bp, XFS_DATA_FORK);
+	if (error)
+		return error;
+
+	xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DIR_FREE_BUF);
+	bp->b_ops = &xfs_dir3_free_buf_ops;
+
+	/*
+	 * Initialize the new block to be empty, and remember
+	 * its first slot as our empty slot.
+	 */
+	memset(bp->b_addr, 0, sizeof(struct xfs_dir3_free_hdr));
+	memset(&hdr, 0, sizeof(hdr));
+
+	if (xfs_has_crc(mp)) {
+		struct xfs_dir3_free_hdr *hdr3 = bp->b_addr;
+
+		hdr.magic = XFS_DIR3_FREE_MAGIC;
+
+		hdr3->hdr.blkno = cpu_to_be64(xfs_buf_daddr(bp));
+		hdr3->hdr.owner = cpu_to_be64(dp->i_ino);
+		uuid_copy(&hdr3->hdr.uuid, &mp->m_sb.sb_meta_uuid);
+	} else
+		hdr.magic = XFS_DIR2_FREE_MAGIC;
+	xfs_dir2_free_hdr_to_disk(mp, bp->b_addr, &hdr);
+	*bpp = bp;
+	return 0;
+}
+
+/*
+ * Log entries from a freespace block.
+ */
+STATIC void
+xfs_dir2_free_log_bests(
+	struct xfs_da_args	*args,
+	struct xfs_dir3_icfree_hdr *hdr,
+	struct xfs_buf		*bp,
+	int			first,		/* first entry to log */
+	int			last)		/* last entry to log */
+{
+	struct xfs_dir2_free	*free = bp->b_addr;
+
+	ASSERT(free->hdr.magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC) ||
+	       free->hdr.magic == cpu_to_be32(XFS_DIR3_FREE_MAGIC));
+	xfs_trans_log_buf(args->trans, bp,
+			  (char *)&hdr->bests[first] - (char *)free,
+			  (char *)&hdr->bests[last] - (char *)free +
+			   sizeof(hdr->bests[0]) - 1);
+}
+
+/*
+ * Log header from a freespace block.
+ */
+static void
+xfs_dir2_free_log_header(
+	struct xfs_da_args	*args,
+	struct xfs_buf		*bp)
+{
+#ifdef DEBUG
+	xfs_dir2_free_t		*free;		/* freespace structure */
+
+	free = bp->b_addr;
+	ASSERT(free->hdr.magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC) ||
+	       free->hdr.magic == cpu_to_be32(XFS_DIR3_FREE_MAGIC));
+#endif
+	xfs_trans_log_buf(args->trans, bp, 0,
+			  args->geo->free_hdr_size - 1);
+}
+
+/*
+ * Convert a leaf-format directory to a node-format directory.
+ * We need to change the magic number of the leaf block, and copy
+ * the freespace table out of the leaf block into its own block.
+ */
+int						/* error */
+xfs_dir2_leaf_to_node(
+	xfs_da_args_t		*args,		/* operation arguments */
+	struct xfs_buf		*lbp)		/* leaf buffer */
+{
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return value */
+	struct xfs_buf		*fbp;		/* freespace buffer */
+	xfs_dir2_db_t		fdb;		/* freespace block number */
+	__be16			*from;		/* pointer to freespace entry */
+	int			i;		/* leaf freespace index */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_tail_t	*ltp;		/* leaf tail structure */
+	int			n;		/* count of live freespc ents */
+	xfs_dir2_data_off_t	off;		/* freespace entry value */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	struct xfs_dir3_icfree_hdr freehdr;
+
+	trace_xfs_dir2_leaf_to_node(args);
+
+	dp = args->dp;
+	tp = args->trans;
+	/*
+	 * Add a freespace block to the directory.
+	 */
+	if ((error = xfs_dir2_grow_inode(args, XFS_DIR2_FREE_SPACE, &fdb))) {
+		return error;
+	}
+	ASSERT(fdb == xfs_dir2_byte_to_db(args->geo, XFS_DIR2_FREE_OFFSET));
+	/*
+	 * Get the buffer for the new freespace block.
+	 */
+	error = xfs_dir3_free_get_buf(args, fdb, &fbp);
+	if (error)
+		return error;
+
+	xfs_dir2_free_hdr_from_disk(dp->i_mount, &freehdr, fbp->b_addr);
+	leaf = lbp->b_addr;
+	ltp = xfs_dir2_leaf_tail_p(args->geo, leaf);
+	if (be32_to_cpu(ltp->bestcount) >
+				(uint)dp->i_disk_size / args->geo->blksize) {
+		xfs_buf_mark_corrupt(lbp);
+		return -EFSCORRUPTED;
+	}
+
+	/*
+	 * Copy freespace entries from the leaf block to the new block.
+	 * Count active entries.
+	 */
+	from = xfs_dir2_leaf_bests_p(ltp);
+	for (i = n = 0; i < be32_to_cpu(ltp->bestcount); i++, from++) {
+		off = be16_to_cpu(*from);
+		if (off != NULLDATAOFF)
+			n++;
+		freehdr.bests[i] = cpu_to_be16(off);
+	}
+
+	/*
+	 * Now initialize the freespace block header.
+	 */
+	freehdr.nused = n;
+	freehdr.nvalid = be32_to_cpu(ltp->bestcount);
+
+	xfs_dir2_free_hdr_to_disk(dp->i_mount, fbp->b_addr, &freehdr);
+	xfs_dir2_free_log_bests(args, &freehdr, fbp, 0, freehdr.nvalid - 1);
+	xfs_dir2_free_log_header(args, fbp);
+
+	/*
+	 * Converting the leaf to a leafnode is just a matter of changing the
+	 * magic number and the ops. Do the change directly to the buffer as
+	 * it's less work (and less code) than decoding the header to host
+	 * format and back again.
+	 */
+	if (leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAF1_MAGIC))
+		leaf->hdr.info.magic = cpu_to_be16(XFS_DIR2_LEAFN_MAGIC);
+	else
+		leaf->hdr.info.magic = cpu_to_be16(XFS_DIR3_LEAFN_MAGIC);
+	lbp->b_ops = &xfs_dir3_leafn_buf_ops;
+	xfs_trans_buf_set_type(tp, lbp, XFS_BLFT_DIR_LEAFN_BUF);
+	xfs_dir3_leaf_log_header(args, lbp);
+	xfs_dir3_leaf_check(dp, lbp);
+	return 0;
+}
+
+/*
+ * Add a leaf entry to a leaf block in a node-form directory.
+ * The other work necessary is done from the caller.
+ */
+static int					/* error */
+xfs_dir2_leafn_add(
+	struct xfs_buf		*bp,		/* leaf buffer */
+	struct xfs_da_args	*args,		/* operation arguments */
+	int			index)		/* insertion pt for new entry */
+{
+	struct xfs_dir3_icleaf_hdr leafhdr;
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_dir2_leaf	*leaf = bp->b_addr;
+	struct xfs_dir2_leaf_entry *lep;
+	struct xfs_dir2_leaf_entry *ents;
+	int			compact;	/* compacting stale leaves */
+	int			highstale = 0;	/* next stale entry */
+	int			lfloghigh;	/* high leaf entry logging */
+	int			lfloglow;	/* low leaf entry logging */
+	int			lowstale = 0;	/* previous stale entry */
+
+	trace_xfs_dir2_leafn_add(args, index);
+
+	xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &leafhdr, leaf);
+	ents = leafhdr.ents;
+
+	/*
+	 * Quick check just to make sure we are not going to index
+	 * into other peoples memory
+	 */
+	if (index < 0) {
+		xfs_buf_mark_corrupt(bp);
+		return -EFSCORRUPTED;
+	}
+
+	/*
+	 * If there are already the maximum number of leaf entries in
+	 * the block, if there are no stale entries it won't fit.
+	 * Caller will do a split.  If there are stale entries we'll do
+	 * a compact.
+	 */
+
+	if (leafhdr.count == args->geo->leaf_max_ents) {
+		if (!leafhdr.stale)
+			return -ENOSPC;
+		compact = leafhdr.stale > 1;
+	} else
+		compact = 0;
+	ASSERT(index == 0 || be32_to_cpu(ents[index - 1].hashval) <= args->hashval);
+	ASSERT(index == leafhdr.count ||
+	       be32_to_cpu(ents[index].hashval) >= args->hashval);
+
+	if (args->op_flags & XFS_DA_OP_JUSTCHECK)
+		return 0;
+
+	/*
+	 * Compact out all but one stale leaf entry.  Leaves behind
+	 * the entry closest to index.
+	 */
+	if (compact)
+		xfs_dir3_leaf_compact_x1(&leafhdr, ents, &index, &lowstale,
+					 &highstale, &lfloglow, &lfloghigh);
+	else if (leafhdr.stale) {
+		/*
+		 * Set impossible logging indices for this case.
+		 */
+		lfloglow = leafhdr.count;
+		lfloghigh = -1;
+	}
+
+	/*
+	 * Insert the new entry, log everything.
+	 */
+	lep = xfs_dir3_leaf_find_entry(&leafhdr, ents, index, compact, lowstale,
+				       highstale, &lfloglow, &lfloghigh);
+
+	lep->hashval = cpu_to_be32(args->hashval);
+	lep->address = cpu_to_be32(xfs_dir2_db_off_to_dataptr(args->geo,
+				args->blkno, args->index));
+
+	xfs_dir2_leaf_hdr_to_disk(dp->i_mount, leaf, &leafhdr);
+	xfs_dir3_leaf_log_header(args, bp);
+	xfs_dir3_leaf_log_ents(args, &leafhdr, bp, lfloglow, lfloghigh);
+	xfs_dir3_leaf_check(dp, bp);
+	return 0;
+}
+
+#ifdef DEBUG
+static void
+xfs_dir2_free_hdr_check(
+	struct xfs_inode *dp,
+	struct xfs_buf	*bp,
+	xfs_dir2_db_t	db)
+{
+	struct xfs_dir3_icfree_hdr hdr;
+
+	xfs_dir2_free_hdr_from_disk(dp->i_mount, &hdr, bp->b_addr);
+
+	ASSERT((hdr.firstdb % dp->i_mount->m_dir_geo->free_max_bests) == 0);
+	ASSERT(hdr.firstdb <= db);
+	ASSERT(db < hdr.firstdb + hdr.nvalid);
+}
+#else
+#define xfs_dir2_free_hdr_check(dp, bp, db)
+#endif	/* DEBUG */
+
+/*
+ * Return the last hash value in the leaf.
+ * Stale entries are ok.
+ */
+xfs_dahash_t					/* hash value */
+xfs_dir2_leaf_lasthash(
+	struct xfs_inode *dp,
+	struct xfs_buf	*bp,			/* leaf buffer */
+	int		*count)			/* count of entries in leaf */
+{
+	struct xfs_dir3_icleaf_hdr leafhdr;
+
+	xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &leafhdr, bp->b_addr);
+
+	ASSERT(leafhdr.magic == XFS_DIR2_LEAFN_MAGIC ||
+	       leafhdr.magic == XFS_DIR3_LEAFN_MAGIC ||
+	       leafhdr.magic == XFS_DIR2_LEAF1_MAGIC ||
+	       leafhdr.magic == XFS_DIR3_LEAF1_MAGIC);
+
+	if (count)
+		*count = leafhdr.count;
+	if (!leafhdr.count)
+		return 0;
+	return be32_to_cpu(leafhdr.ents[leafhdr.count - 1].hashval);
+}
+
+/*
+ * Look up a leaf entry for space to add a name in a node-format leaf block.
+ * The extrablk in state is a freespace block.
+ */
+STATIC int
+xfs_dir2_leafn_lookup_for_addname(
+	struct xfs_buf		*bp,		/* leaf buffer */
+	xfs_da_args_t		*args,		/* operation arguments */
+	int			*indexp,	/* out: leaf entry index */
+	xfs_da_state_t		*state)		/* state to fill in */
+{
+	struct xfs_buf		*curbp = NULL;	/* current data/free buffer */
+	xfs_dir2_db_t		curdb = -1;	/* current data block number */
+	xfs_dir2_db_t		curfdb = -1;	/* current free block number */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return value */
+	int			fi;		/* free entry index */
+	xfs_dir2_free_t		*free = NULL;	/* free block structure */
+	int			index;		/* leaf entry index */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	int			length;		/* length of new data entry */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	xfs_dir2_db_t		newdb;		/* new data block number */
+	xfs_dir2_db_t		newfdb;		/* new free block number */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	struct xfs_dir3_icleaf_hdr leafhdr;
+
+	dp = args->dp;
+	tp = args->trans;
+	mp = dp->i_mount;
+	leaf = bp->b_addr;
+	xfs_dir2_leaf_hdr_from_disk(mp, &leafhdr, leaf);
+
+	xfs_dir3_leaf_check(dp, bp);
+	ASSERT(leafhdr.count > 0);
+
+	/*
+	 * Look up the hash value in the leaf entries.
+	 */
+	index = xfs_dir2_leaf_search_hash(args, bp);
+	/*
+	 * Do we have a buffer coming in?
+	 */
+	if (state->extravalid) {
+		/* If so, it's a free block buffer, get the block number. */
+		curbp = state->extrablk.bp;
+		curfdb = state->extrablk.blkno;
+		free = curbp->b_addr;
+		ASSERT(free->hdr.magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC) ||
+		       free->hdr.magic == cpu_to_be32(XFS_DIR3_FREE_MAGIC));
+	}
+	length = xfs_dir2_data_entsize(mp, args->namelen);
+	/*
+	 * Loop over leaf entries with the right hash value.
+	 */
+	for (lep = &leafhdr.ents[index];
+	     index < leafhdr.count && be32_to_cpu(lep->hashval) == args->hashval;
+	     lep++, index++) {
+		/*
+		 * Skip stale leaf entries.
+		 */
+		if (be32_to_cpu(lep->address) == XFS_DIR2_NULL_DATAPTR)
+			continue;
+		/*
+		 * Pull the data block number from the entry.
+		 */
+		newdb = xfs_dir2_dataptr_to_db(args->geo,
+					       be32_to_cpu(lep->address));
+		/*
+		 * For addname, we're looking for a place to put the new entry.
+		 * We want to use a data block with an entry of equal
+		 * hash value to ours if there is one with room.
+		 *
+		 * If this block isn't the data block we already have
+		 * in hand, take a look at it.
+		 */
+		if (newdb != curdb) {
+			struct xfs_dir3_icfree_hdr freehdr;
+
+			curdb = newdb;
+			/*
+			 * Convert the data block to the free block
+			 * holding its freespace information.
+			 */
+			newfdb = xfs_dir2_db_to_fdb(args->geo, newdb);
+			/*
+			 * If it's not the one we have in hand, read it in.
+			 */
+			if (newfdb != curfdb) {
+				/*
+				 * If we had one before, drop it.
+				 */
+				if (curbp)
+					xfs_trans_brelse(tp, curbp);
+
+				error = xfs_dir2_free_read(tp, dp,
+						xfs_dir2_db_to_da(args->geo,
+								  newfdb),
+						&curbp);
+				if (error)
+					return error;
+				free = curbp->b_addr;
+
+				xfs_dir2_free_hdr_check(dp, curbp, curdb);
+			}
+			/*
+			 * Get the index for our entry.
+			 */
+			fi = xfs_dir2_db_to_fdindex(args->geo, curdb);
+			/*
+			 * If it has room, return it.
+			 */
+			xfs_dir2_free_hdr_from_disk(mp, &freehdr, free);
+			if (XFS_IS_CORRUPT(mp,
+					   freehdr.bests[fi] ==
+					   cpu_to_be16(NULLDATAOFF))) {
+				if (curfdb != newfdb)
+					xfs_trans_brelse(tp, curbp);
+				return -EFSCORRUPTED;
+			}
+			curfdb = newfdb;
+			if (be16_to_cpu(freehdr.bests[fi]) >= length)
+				goto out;
+		}
+	}
+	/* Didn't find any space */
+	fi = -1;
+out:
+	ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
+	if (curbp) {
+		/* Giving back a free block. */
+		state->extravalid = 1;
+		state->extrablk.bp = curbp;
+		state->extrablk.index = fi;
+		state->extrablk.blkno = curfdb;
+
+		/*
+		 * Important: this magic number is not in the buffer - it's for
+		 * buffer type information and therefore only the free/data type
+		 * matters here, not whether CRCs are enabled or not.
+		 */
+		state->extrablk.magic = XFS_DIR2_FREE_MAGIC;
+	} else {
+		state->extravalid = 0;
+	}
+	/*
+	 * Return the index, that will be the insertion point.
+	 */
+	*indexp = index;
+	return -ENOENT;
+}
+
+/*
+ * Look up a leaf entry in a node-format leaf block.
+ * The extrablk in state a data block.
+ */
+STATIC int
+xfs_dir2_leafn_lookup_for_entry(
+	struct xfs_buf		*bp,		/* leaf buffer */
+	xfs_da_args_t		*args,		/* operation arguments */
+	int			*indexp,	/* out: leaf entry index */
+	xfs_da_state_t		*state)		/* state to fill in */
+{
+	struct xfs_buf		*curbp = NULL;	/* current data/free buffer */
+	xfs_dir2_db_t		curdb = -1;	/* current data block number */
+	xfs_dir2_data_entry_t	*dep;		/* data block entry */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return value */
+	int			index;		/* leaf entry index */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	xfs_dir2_db_t		newdb;		/* new data block number */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	enum xfs_dacmp		cmp;		/* comparison result */
+	struct xfs_dir3_icleaf_hdr leafhdr;
+
+	dp = args->dp;
+	tp = args->trans;
+	mp = dp->i_mount;
+	leaf = bp->b_addr;
+	xfs_dir2_leaf_hdr_from_disk(mp, &leafhdr, leaf);
+
+	xfs_dir3_leaf_check(dp, bp);
+	if (leafhdr.count <= 0) {
+		xfs_buf_mark_corrupt(bp);
+		return -EFSCORRUPTED;
+	}
+
+	/*
+	 * Look up the hash value in the leaf entries.
+	 */
+	index = xfs_dir2_leaf_search_hash(args, bp);
+	/*
+	 * Do we have a buffer coming in?
+	 */
+	if (state->extravalid) {
+		curbp = state->extrablk.bp;
+		curdb = state->extrablk.blkno;
+	}
+	/*
+	 * Loop over leaf entries with the right hash value.
+	 */
+	for (lep = &leafhdr.ents[index];
+	     index < leafhdr.count && be32_to_cpu(lep->hashval) == args->hashval;
+	     lep++, index++) {
+		/*
+		 * Skip stale leaf entries.
+		 */
+		if (be32_to_cpu(lep->address) == XFS_DIR2_NULL_DATAPTR)
+			continue;
+		/*
+		 * Pull the data block number from the entry.
+		 */
+		newdb = xfs_dir2_dataptr_to_db(args->geo,
+					       be32_to_cpu(lep->address));
+		/*
+		 * Not adding a new entry, so we really want to find
+		 * the name given to us.
+		 *
+		 * If it's a different data block, go get it.
+		 */
+		if (newdb != curdb) {
+			/*
+			 * If we had a block before that we aren't saving
+			 * for a CI name, drop it
+			 */
+			if (curbp && (args->cmpresult == XFS_CMP_DIFFERENT ||
+						curdb != state->extrablk.blkno))
+				xfs_trans_brelse(tp, curbp);
+			/*
+			 * If needing the block that is saved with a CI match,
+			 * use it otherwise read in the new data block.
+			 */
+			if (args->cmpresult != XFS_CMP_DIFFERENT &&
+					newdb == state->extrablk.blkno) {
+				ASSERT(state->extravalid);
+				curbp = state->extrablk.bp;
+			} else {
+				error = xfs_dir3_data_read(tp, dp,
+						xfs_dir2_db_to_da(args->geo,
+								  newdb),
+						0, &curbp);
+				if (error)
+					return error;
+			}
+			xfs_dir3_data_check(dp, curbp);
+			curdb = newdb;
+		}
+		/*
+		 * Point to the data entry.
+		 */
+		dep = (xfs_dir2_data_entry_t *)((char *)curbp->b_addr +
+			xfs_dir2_dataptr_to_off(args->geo,
+						be32_to_cpu(lep->address)));
+		/*
+		 * Compare the entry and if it's an exact match, return
+		 * EEXIST immediately. If it's the first case-insensitive
+		 * match, store the block & inode number and continue looking.
+		 */
+		cmp = xfs_dir2_compname(args, dep->name, dep->namelen);
+		if (cmp != XFS_CMP_DIFFERENT && cmp != args->cmpresult) {
+			/* If there is a CI match block, drop it */
+			if (args->cmpresult != XFS_CMP_DIFFERENT &&
+						curdb != state->extrablk.blkno)
+				xfs_trans_brelse(tp, state->extrablk.bp);
+			args->cmpresult = cmp;
+			args->inumber = be64_to_cpu(dep->inumber);
+			args->filetype = xfs_dir2_data_get_ftype(mp, dep);
+			*indexp = index;
+			state->extravalid = 1;
+			state->extrablk.bp = curbp;
+			state->extrablk.blkno = curdb;
+			state->extrablk.index = (int)((char *)dep -
+							(char *)curbp->b_addr);
+			state->extrablk.magic = XFS_DIR2_DATA_MAGIC;
+			curbp->b_ops = &xfs_dir3_data_buf_ops;
+			xfs_trans_buf_set_type(tp, curbp, XFS_BLFT_DIR_DATA_BUF);
+			if (cmp == XFS_CMP_EXACT)
+				return -EEXIST;
+		}
+	}
+	ASSERT(index == leafhdr.count || (args->op_flags & XFS_DA_OP_OKNOENT));
+	if (curbp) {
+		if (args->cmpresult == XFS_CMP_DIFFERENT) {
+			/* Giving back last used data block. */
+			state->extravalid = 1;
+			state->extrablk.bp = curbp;
+			state->extrablk.index = -1;
+			state->extrablk.blkno = curdb;
+			state->extrablk.magic = XFS_DIR2_DATA_MAGIC;
+			curbp->b_ops = &xfs_dir3_data_buf_ops;
+			xfs_trans_buf_set_type(tp, curbp, XFS_BLFT_DIR_DATA_BUF);
+		} else {
+			/* If the curbp is not the CI match block, drop it */
+			if (state->extrablk.bp != curbp)
+				xfs_trans_brelse(tp, curbp);
+		}
+	} else {
+		state->extravalid = 0;
+	}
+	*indexp = index;
+	return -ENOENT;
+}
+
+/*
+ * Look up a leaf entry in a node-format leaf block.
+ * If this is an addname then the extrablk in state is a freespace block,
+ * otherwise it's a data block.
+ */
+int
+xfs_dir2_leafn_lookup_int(
+	struct xfs_buf		*bp,		/* leaf buffer */
+	xfs_da_args_t		*args,		/* operation arguments */
+	int			*indexp,	/* out: leaf entry index */
+	xfs_da_state_t		*state)		/* state to fill in */
+{
+	if (args->op_flags & XFS_DA_OP_ADDNAME)
+		return xfs_dir2_leafn_lookup_for_addname(bp, args, indexp,
+							state);
+	return xfs_dir2_leafn_lookup_for_entry(bp, args, indexp, state);
+}
+
+/*
+ * Move count leaf entries from source to destination leaf.
+ * Log entries and headers.  Stale entries are preserved.
+ */
+static void
+xfs_dir3_leafn_moveents(
+	xfs_da_args_t			*args,	/* operation arguments */
+	struct xfs_buf			*bp_s,	/* source */
+	struct xfs_dir3_icleaf_hdr	*shdr,
+	struct xfs_dir2_leaf_entry	*sents,
+	int				start_s,/* source leaf index */
+	struct xfs_buf			*bp_d,	/* destination */
+	struct xfs_dir3_icleaf_hdr	*dhdr,
+	struct xfs_dir2_leaf_entry	*dents,
+	int				start_d,/* destination leaf index */
+	int				count)	/* count of leaves to copy */
+{
+	int				stale;	/* count stale leaves copied */
+
+	trace_xfs_dir2_leafn_moveents(args, start_s, start_d, count);
+
+	/*
+	 * Silently return if nothing to do.
+	 */
+	if (count == 0)
+		return;
+
+	/*
+	 * If the destination index is not the end of the current
+	 * destination leaf entries, open up a hole in the destination
+	 * to hold the new entries.
+	 */
+	if (start_d < dhdr->count) {
+		memmove(&dents[start_d + count], &dents[start_d],
+			(dhdr->count - start_d) * sizeof(xfs_dir2_leaf_entry_t));
+		xfs_dir3_leaf_log_ents(args, dhdr, bp_d, start_d + count,
+				       count + dhdr->count - 1);
+	}
+	/*
+	 * If the source has stale leaves, count the ones in the copy range
+	 * so we can update the header correctly.
+	 */
+	if (shdr->stale) {
+		int	i;			/* temp leaf index */
+
+		for (i = start_s, stale = 0; i < start_s + count; i++) {
+			if (sents[i].address ==
+					cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
+				stale++;
+		}
+	} else
+		stale = 0;
+	/*
+	 * Copy the leaf entries from source to destination.
+	 */
+	memcpy(&dents[start_d], &sents[start_s],
+		count * sizeof(xfs_dir2_leaf_entry_t));
+	xfs_dir3_leaf_log_ents(args, dhdr, bp_d, start_d, start_d + count - 1);
+
+	/*
+	 * If there are source entries after the ones we copied,
+	 * delete the ones we copied by sliding the next ones down.
+	 */
+	if (start_s + count < shdr->count) {
+		memmove(&sents[start_s], &sents[start_s + count],
+			count * sizeof(xfs_dir2_leaf_entry_t));
+		xfs_dir3_leaf_log_ents(args, shdr, bp_s, start_s,
+				       start_s + count - 1);
+	}
+
+	/*
+	 * Update the headers and log them.
+	 */
+	shdr->count -= count;
+	shdr->stale -= stale;
+	dhdr->count += count;
+	dhdr->stale += stale;
+}
+
+/*
+ * Determine the sort order of two leaf blocks.
+ * Returns 1 if both are valid and leaf2 should be before leaf1, else 0.
+ */
+int						/* sort order */
+xfs_dir2_leafn_order(
+	struct xfs_inode	*dp,
+	struct xfs_buf		*leaf1_bp,		/* leaf1 buffer */
+	struct xfs_buf		*leaf2_bp)		/* leaf2 buffer */
+{
+	struct xfs_dir2_leaf	*leaf1 = leaf1_bp->b_addr;
+	struct xfs_dir2_leaf	*leaf2 = leaf2_bp->b_addr;
+	struct xfs_dir2_leaf_entry *ents1;
+	struct xfs_dir2_leaf_entry *ents2;
+	struct xfs_dir3_icleaf_hdr hdr1;
+	struct xfs_dir3_icleaf_hdr hdr2;
+
+	xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &hdr1, leaf1);
+	xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &hdr2, leaf2);
+	ents1 = hdr1.ents;
+	ents2 = hdr2.ents;
+
+	if (hdr1.count > 0 && hdr2.count > 0 &&
+	    (be32_to_cpu(ents2[0].hashval) < be32_to_cpu(ents1[0].hashval) ||
+	     be32_to_cpu(ents2[hdr2.count - 1].hashval) <
+				be32_to_cpu(ents1[hdr1.count - 1].hashval)))
+		return 1;
+	return 0;
+}
+
+/*
+ * Rebalance leaf entries between two leaf blocks.
+ * This is actually only called when the second block is new,
+ * though the code deals with the general case.
+ * A new entry will be inserted in one of the blocks, and that
+ * entry is taken into account when balancing.
+ */
+static void
+xfs_dir2_leafn_rebalance(
+	xfs_da_state_t		*state,		/* btree cursor */
+	xfs_da_state_blk_t	*blk1,		/* first btree block */
+	xfs_da_state_blk_t	*blk2)		/* second btree block */
+{
+	xfs_da_args_t		*args;		/* operation arguments */
+	int			count;		/* count (& direction) leaves */
+	int			isleft;		/* new goes in left leaf */
+	xfs_dir2_leaf_t		*leaf1;		/* first leaf structure */
+	xfs_dir2_leaf_t		*leaf2;		/* second leaf structure */
+	int			mid;		/* midpoint leaf index */
+#if defined(DEBUG) || defined(XFS_WARN)
+	int			oldstale;	/* old count of stale leaves */
+#endif
+	int			oldsum;		/* old total leaf count */
+	int			swap_blocks;	/* swapped leaf blocks */
+	struct xfs_dir2_leaf_entry *ents1;
+	struct xfs_dir2_leaf_entry *ents2;
+	struct xfs_dir3_icleaf_hdr hdr1;
+	struct xfs_dir3_icleaf_hdr hdr2;
+	struct xfs_inode	*dp = state->args->dp;
+
+	args = state->args;
+	/*
+	 * If the block order is wrong, swap the arguments.
+	 */
+	swap_blocks = xfs_dir2_leafn_order(dp, blk1->bp, blk2->bp);
+	if (swap_blocks)
+		swap(blk1, blk2);
+
+	leaf1 = blk1->bp->b_addr;
+	leaf2 = blk2->bp->b_addr;
+	xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &hdr1, leaf1);
+	xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &hdr2, leaf2);
+	ents1 = hdr1.ents;
+	ents2 = hdr2.ents;
+
+	oldsum = hdr1.count + hdr2.count;
+#if defined(DEBUG) || defined(XFS_WARN)
+	oldstale = hdr1.stale + hdr2.stale;
+#endif
+	mid = oldsum >> 1;
+
+	/*
+	 * If the old leaf count was odd then the new one will be even,
+	 * so we need to divide the new count evenly.
+	 */
+	if (oldsum & 1) {
+		xfs_dahash_t	midhash;	/* middle entry hash value */
+
+		if (mid >= hdr1.count)
+			midhash = be32_to_cpu(ents2[mid - hdr1.count].hashval);
+		else
+			midhash = be32_to_cpu(ents1[mid].hashval);
+		isleft = args->hashval <= midhash;
+	}
+	/*
+	 * If the old count is even then the new count is odd, so there's
+	 * no preferred side for the new entry.
+	 * Pick the left one.
+	 */
+	else
+		isleft = 1;
+	/*
+	 * Calculate moved entry count.  Positive means left-to-right,
+	 * negative means right-to-left.  Then move the entries.
+	 */
+	count = hdr1.count - mid + (isleft == 0);
+	if (count > 0)
+		xfs_dir3_leafn_moveents(args, blk1->bp, &hdr1, ents1,
+					hdr1.count - count, blk2->bp,
+					&hdr2, ents2, 0, count);
+	else if (count < 0)
+		xfs_dir3_leafn_moveents(args, blk2->bp, &hdr2, ents2, 0,
+					blk1->bp, &hdr1, ents1,
+					hdr1.count, count);
+
+	ASSERT(hdr1.count + hdr2.count == oldsum);
+	ASSERT(hdr1.stale + hdr2.stale == oldstale);
+
+	/* log the changes made when moving the entries */
+	xfs_dir2_leaf_hdr_to_disk(dp->i_mount, leaf1, &hdr1);
+	xfs_dir2_leaf_hdr_to_disk(dp->i_mount, leaf2, &hdr2);
+	xfs_dir3_leaf_log_header(args, blk1->bp);
+	xfs_dir3_leaf_log_header(args, blk2->bp);
+
+	xfs_dir3_leaf_check(dp, blk1->bp);
+	xfs_dir3_leaf_check(dp, blk2->bp);
+
+	/*
+	 * Mark whether we're inserting into the old or new leaf.
+	 */
+	if (hdr1.count < hdr2.count)
+		state->inleaf = swap_blocks;
+	else if (hdr1.count > hdr2.count)
+		state->inleaf = !swap_blocks;
+	else
+		state->inleaf = swap_blocks ^ (blk1->index <= hdr1.count);
+	/*
+	 * Adjust the expected index for insertion.
+	 */
+	if (!state->inleaf)
+		blk2->index = blk1->index - hdr1.count;
+
+	/*
+	 * Finally sanity check just to make sure we are not returning a
+	 * negative index
+	 */
+	if (blk2->index < 0) {
+		state->inleaf = 1;
+		blk2->index = 0;
+		xfs_alert(dp->i_mount,
+	"%s: picked the wrong leaf? reverting original leaf: blk1->index %d",
+			__func__, blk1->index);
+	}
+}
+
+static int
+xfs_dir3_data_block_free(
+	xfs_da_args_t		*args,
+	struct xfs_dir2_data_hdr *hdr,
+	struct xfs_dir2_free	*free,
+	xfs_dir2_db_t		fdb,
+	int			findex,
+	struct xfs_buf		*fbp,
+	int			longest)
+{
+	int			logfree = 0;
+	struct xfs_dir3_icfree_hdr freehdr;
+	struct xfs_inode	*dp = args->dp;
+
+	xfs_dir2_free_hdr_from_disk(dp->i_mount, &freehdr, free);
+	if (hdr) {
+		/*
+		 * Data block is not empty, just set the free entry to the new
+		 * value.
+		 */
+		freehdr.bests[findex] = cpu_to_be16(longest);
+		xfs_dir2_free_log_bests(args, &freehdr, fbp, findex, findex);
+		return 0;
+	}
+
+	/* One less used entry in the free table. */
+	freehdr.nused--;
+
+	/*
+	 * If this was the last entry in the table, we can trim the table size
+	 * back.  There might be other entries at the end referring to
+	 * non-existent data blocks, get those too.
+	 */
+	if (findex == freehdr.nvalid - 1) {
+		int	i;		/* free entry index */
+
+		for (i = findex - 1; i >= 0; i--) {
+			if (freehdr.bests[i] != cpu_to_be16(NULLDATAOFF))
+				break;
+		}
+		freehdr.nvalid = i + 1;
+		logfree = 0;
+	} else {
+		/* Not the last entry, just punch it out.  */
+		freehdr.bests[findex] = cpu_to_be16(NULLDATAOFF);
+		logfree = 1;
+	}
+
+	xfs_dir2_free_hdr_to_disk(dp->i_mount, free, &freehdr);
+	xfs_dir2_free_log_header(args, fbp);
+
+	/*
+	 * If there are no useful entries left in the block, get rid of the
+	 * block if we can.
+	 */
+	if (!freehdr.nused) {
+		int error;
+
+		error = xfs_dir2_shrink_inode(args, fdb, fbp);
+		if (error == 0) {
+			fbp = NULL;
+			logfree = 0;
+		} else if (error != -ENOSPC || args->total != 0)
+			return error;
+		/*
+		 * It's possible to get ENOSPC if there is no
+		 * space reservation.  In this case some one
+		 * else will eventually get rid of this block.
+		 */
+	}
+
+	/* Log the free entry that changed, unless we got rid of it.  */
+	if (logfree)
+		xfs_dir2_free_log_bests(args, &freehdr, fbp, findex, findex);
+	return 0;
+}
+
+/*
+ * Remove an entry from a node directory.
+ * This removes the leaf entry and the data entry,
+ * and updates the free block if necessary.
+ */
+static int					/* error */
+xfs_dir2_leafn_remove(
+	xfs_da_args_t		*args,		/* operation arguments */
+	struct xfs_buf		*bp,		/* leaf buffer */
+	int			index,		/* leaf entry index */
+	xfs_da_state_blk_t	*dblk,		/* data block */
+	int			*rval)		/* resulting block needs join */
+{
+	struct xfs_da_geometry	*geo = args->geo;
+	xfs_dir2_data_hdr_t	*hdr;		/* data block header */
+	xfs_dir2_db_t		db;		/* data block number */
+	struct xfs_buf		*dbp;		/* data block buffer */
+	xfs_dir2_data_entry_t	*dep;		/* data block entry */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	int			longest;	/* longest data free entry */
+	int			off;		/* data block entry offset */
+	int			needlog;	/* need to log data header */
+	int			needscan;	/* need to rescan data frees */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	struct xfs_dir2_data_free *bf;		/* bestfree table */
+	struct xfs_dir3_icleaf_hdr leafhdr;
+
+	trace_xfs_dir2_leafn_remove(args, index);
+
+	dp = args->dp;
+	tp = args->trans;
+	leaf = bp->b_addr;
+	xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &leafhdr, leaf);
+
+	/*
+	 * Point to the entry we're removing.
+	 */
+	lep = &leafhdr.ents[index];
+
+	/*
+	 * Extract the data block and offset from the entry.
+	 */
+	db = xfs_dir2_dataptr_to_db(geo, be32_to_cpu(lep->address));
+	ASSERT(dblk->blkno == db);
+	off = xfs_dir2_dataptr_to_off(geo, be32_to_cpu(lep->address));
+	ASSERT(dblk->index == off);
+
+	/*
+	 * Kill the leaf entry by marking it stale.
+	 * Log the leaf block changes.
+	 */
+	leafhdr.stale++;
+	xfs_dir2_leaf_hdr_to_disk(dp->i_mount, leaf, &leafhdr);
+	xfs_dir3_leaf_log_header(args, bp);
+
+	lep->address = cpu_to_be32(XFS_DIR2_NULL_DATAPTR);
+	xfs_dir3_leaf_log_ents(args, &leafhdr, bp, index, index);
+
+	/*
+	 * Make the data entry free.  Keep track of the longest freespace
+	 * in the data block in case it changes.
+	 */
+	dbp = dblk->bp;
+	hdr = dbp->b_addr;
+	dep = (xfs_dir2_data_entry_t *)((char *)hdr + off);
+	bf = xfs_dir2_data_bestfree_p(dp->i_mount, hdr);
+	longest = be16_to_cpu(bf[0].length);
+	needlog = needscan = 0;
+	xfs_dir2_data_make_free(args, dbp, off,
+		xfs_dir2_data_entsize(dp->i_mount, dep->namelen), &needlog,
+		&needscan);
+	/*
+	 * Rescan the data block freespaces for bestfree.
+	 * Log the data block header if needed.
+	 */
+	if (needscan)
+		xfs_dir2_data_freescan(dp->i_mount, hdr, &needlog);
+	if (needlog)
+		xfs_dir2_data_log_header(args, dbp);
+	xfs_dir3_data_check(dp, dbp);
+	/*
+	 * If the longest data block freespace changes, need to update
+	 * the corresponding freeblock entry.
+	 */
+	if (longest < be16_to_cpu(bf[0].length)) {
+		int		error;		/* error return value */
+		struct xfs_buf	*fbp;		/* freeblock buffer */
+		xfs_dir2_db_t	fdb;		/* freeblock block number */
+		int		findex;		/* index in freeblock entries */
+		xfs_dir2_free_t	*free;		/* freeblock structure */
+
+		/*
+		 * Convert the data block number to a free block,
+		 * read in the free block.
+		 */
+		fdb = xfs_dir2_db_to_fdb(geo, db);
+		error = xfs_dir2_free_read(tp, dp, xfs_dir2_db_to_da(geo, fdb),
+					   &fbp);
+		if (error)
+			return error;
+		free = fbp->b_addr;
+#ifdef DEBUG
+	{
+		struct xfs_dir3_icfree_hdr freehdr;
+
+		xfs_dir2_free_hdr_from_disk(dp->i_mount, &freehdr, free);
+		ASSERT(freehdr.firstdb == geo->free_max_bests *
+			(fdb - xfs_dir2_byte_to_db(geo, XFS_DIR2_FREE_OFFSET)));
+	}
+#endif
+		/*
+		 * Calculate which entry we need to fix.
+		 */
+		findex = xfs_dir2_db_to_fdindex(geo, db);
+		longest = be16_to_cpu(bf[0].length);
+		/*
+		 * If the data block is now empty we can get rid of it
+		 * (usually).
+		 */
+		if (longest == geo->blksize - geo->data_entry_offset) {
+			/*
+			 * Try to punch out the data block.
+			 */
+			error = xfs_dir2_shrink_inode(args, db, dbp);
+			if (error == 0) {
+				dblk->bp = NULL;
+				hdr = NULL;
+			}
+			/*
+			 * We can get ENOSPC if there's no space reservation.
+			 * In this case just drop the buffer and some one else
+			 * will eventually get rid of the empty block.
+			 */
+			else if (!(error == -ENOSPC && args->total == 0))
+				return error;
+		}
+		/*
+		 * If we got rid of the data block, we can eliminate that entry
+		 * in the free block.
+		 */
+		error = xfs_dir3_data_block_free(args, hdr, free,
+						 fdb, findex, fbp, longest);
+		if (error)
+			return error;
+	}
+
+	xfs_dir3_leaf_check(dp, bp);
+	/*
+	 * Return indication of whether this leaf block is empty enough
+	 * to justify trying to join it with a neighbor.
+	 */
+	*rval = (geo->leaf_hdr_size +
+		 (uint)sizeof(leafhdr.ents) * (leafhdr.count - leafhdr.stale)) <
+		geo->magicpct;
+	return 0;
+}
+
+/*
+ * Split the leaf entries in the old block into old and new blocks.
+ */
+int						/* error */
+xfs_dir2_leafn_split(
+	xfs_da_state_t		*state,		/* btree cursor */
+	xfs_da_state_blk_t	*oldblk,	/* original block */
+	xfs_da_state_blk_t	*newblk)	/* newly created block */
+{
+	xfs_da_args_t		*args;		/* operation arguments */
+	xfs_dablk_t		blkno;		/* new leaf block number */
+	int			error;		/* error return value */
+	struct xfs_inode	*dp;
+
+	/*
+	 * Allocate space for a new leaf node.
+	 */
+	args = state->args;
+	dp = args->dp;
+	ASSERT(oldblk->magic == XFS_DIR2_LEAFN_MAGIC);
+	error = xfs_da_grow_inode(args, &blkno);
+	if (error) {
+		return error;
+	}
+	/*
+	 * Initialize the new leaf block.
+	 */
+	error = xfs_dir3_leaf_get_buf(args, xfs_dir2_da_to_db(args->geo, blkno),
+				      &newblk->bp, XFS_DIR2_LEAFN_MAGIC);
+	if (error)
+		return error;
+
+	newblk->blkno = blkno;
+	newblk->magic = XFS_DIR2_LEAFN_MAGIC;
+	/*
+	 * Rebalance the entries across the two leaves, link the new
+	 * block into the leaves.
+	 */
+	xfs_dir2_leafn_rebalance(state, oldblk, newblk);
+	error = xfs_da3_blk_link(state, oldblk, newblk);
+	if (error) {
+		return error;
+	}
+	/*
+	 * Insert the new entry in the correct block.
+	 */
+	if (state->inleaf)
+		error = xfs_dir2_leafn_add(oldblk->bp, args, oldblk->index);
+	else
+		error = xfs_dir2_leafn_add(newblk->bp, args, newblk->index);
+	/*
+	 * Update last hashval in each block since we added the name.
+	 */
+	oldblk->hashval = xfs_dir2_leaf_lasthash(dp, oldblk->bp, NULL);
+	newblk->hashval = xfs_dir2_leaf_lasthash(dp, newblk->bp, NULL);
+	xfs_dir3_leaf_check(dp, oldblk->bp);
+	xfs_dir3_leaf_check(dp, newblk->bp);
+	return error;
+}
+
+/*
+ * Check a leaf block and its neighbors to see if the block should be
+ * collapsed into one or the other neighbor.  Always keep the block
+ * with the smaller block number.
+ * If the current block is over 50% full, don't try to join it, return 0.
+ * If the block is empty, fill in the state structure and return 2.
+ * If it can be collapsed, fill in the state structure and return 1.
+ * If nothing can be done, return 0.
+ */
+int						/* error */
+xfs_dir2_leafn_toosmall(
+	xfs_da_state_t		*state,		/* btree cursor */
+	int			*action)	/* resulting action to take */
+{
+	xfs_da_state_blk_t	*blk;		/* leaf block */
+	xfs_dablk_t		blkno;		/* leaf block number */
+	struct xfs_buf		*bp;		/* leaf buffer */
+	int			bytes;		/* bytes in use */
+	int			count;		/* leaf live entry count */
+	int			error;		/* error return value */
+	int			forward;	/* sibling block direction */
+	int			i;		/* sibling counter */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	int			rval;		/* result from path_shift */
+	struct xfs_dir3_icleaf_hdr leafhdr;
+	struct xfs_dir2_leaf_entry *ents;
+	struct xfs_inode	*dp = state->args->dp;
+
+	/*
+	 * Check for the degenerate case of the block being over 50% full.
+	 * If so, it's not worth even looking to see if we might be able
+	 * to coalesce with a sibling.
+	 */
+	blk = &state->path.blk[state->path.active - 1];
+	leaf = blk->bp->b_addr;
+	xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &leafhdr, leaf);
+	ents = leafhdr.ents;
+	xfs_dir3_leaf_check(dp, blk->bp);
+
+	count = leafhdr.count - leafhdr.stale;
+	bytes = state->args->geo->leaf_hdr_size + count * sizeof(ents[0]);
+	if (bytes > (state->args->geo->blksize >> 1)) {
+		/*
+		 * Blk over 50%, don't try to join.
+		 */
+		*action = 0;
+		return 0;
+	}
+	/*
+	 * Check for the degenerate case of the block being empty.
+	 * If the block is empty, we'll simply delete it, no need to
+	 * coalesce it with a sibling block.  We choose (arbitrarily)
+	 * to merge with the forward block unless it is NULL.
+	 */
+	if (count == 0) {
+		/*
+		 * Make altpath point to the block we want to keep and
+		 * path point to the block we want to drop (this one).
+		 */
+		forward = (leafhdr.forw != 0);
+		memcpy(&state->altpath, &state->path, sizeof(state->path));
+		error = xfs_da3_path_shift(state, &state->altpath, forward, 0,
+			&rval);
+		if (error)
+			return error;
+		*action = rval ? 2 : 0;
+		return 0;
+	}
+	/*
+	 * Examine each sibling block to see if we can coalesce with
+	 * at least 25% free space to spare.  We need to figure out
+	 * whether to merge with the forward or the backward block.
+	 * We prefer coalescing with the lower numbered sibling so as
+	 * to shrink a directory over time.
+	 */
+	forward = leafhdr.forw < leafhdr.back;
+	for (i = 0, bp = NULL; i < 2; forward = !forward, i++) {
+		struct xfs_dir3_icleaf_hdr hdr2;
+
+		blkno = forward ? leafhdr.forw : leafhdr.back;
+		if (blkno == 0)
+			continue;
+		/*
+		 * Read the sibling leaf block.
+		 */
+		error = xfs_dir3_leafn_read(state->args->trans, dp, blkno, &bp);
+		if (error)
+			return error;
+
+		/*
+		 * Count bytes in the two blocks combined.
+		 */
+		count = leafhdr.count - leafhdr.stale;
+		bytes = state->args->geo->blksize -
+			(state->args->geo->blksize >> 2);
+
+		leaf = bp->b_addr;
+		xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &hdr2, leaf);
+		ents = hdr2.ents;
+		count += hdr2.count - hdr2.stale;
+		bytes -= count * sizeof(ents[0]);
+
+		/*
+		 * Fits with at least 25% to spare.
+		 */
+		if (bytes >= 0)
+			break;
+		xfs_trans_brelse(state->args->trans, bp);
+	}
+	/*
+	 * Didn't like either block, give up.
+	 */
+	if (i >= 2) {
+		*action = 0;
+		return 0;
+	}
+
+	/*
+	 * Make altpath point to the block we want to keep (the lower
+	 * numbered block) and path point to the block we want to drop.
+	 */
+	memcpy(&state->altpath, &state->path, sizeof(state->path));
+	if (blkno < blk->blkno)
+		error = xfs_da3_path_shift(state, &state->altpath, forward, 0,
+			&rval);
+	else
+		error = xfs_da3_path_shift(state, &state->path, forward, 0,
+			&rval);
+	if (error) {
+		return error;
+	}
+	*action = rval ? 0 : 1;
+	return 0;
+}
+
+/*
+ * Move all the leaf entries from drop_blk to save_blk.
+ * This is done as part of a join operation.
+ */
+void
+xfs_dir2_leafn_unbalance(
+	xfs_da_state_t		*state,		/* cursor */
+	xfs_da_state_blk_t	*drop_blk,	/* dead block */
+	xfs_da_state_blk_t	*save_blk)	/* surviving block */
+{
+	xfs_da_args_t		*args;		/* operation arguments */
+	xfs_dir2_leaf_t		*drop_leaf;	/* dead leaf structure */
+	xfs_dir2_leaf_t		*save_leaf;	/* surviving leaf structure */
+	struct xfs_dir3_icleaf_hdr savehdr;
+	struct xfs_dir3_icleaf_hdr drophdr;
+	struct xfs_dir2_leaf_entry *sents;
+	struct xfs_dir2_leaf_entry *dents;
+	struct xfs_inode	*dp = state->args->dp;
+
+	args = state->args;
+	ASSERT(drop_blk->magic == XFS_DIR2_LEAFN_MAGIC);
+	ASSERT(save_blk->magic == XFS_DIR2_LEAFN_MAGIC);
+	drop_leaf = drop_blk->bp->b_addr;
+	save_leaf = save_blk->bp->b_addr;
+
+	xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &savehdr, save_leaf);
+	xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &drophdr, drop_leaf);
+	sents = savehdr.ents;
+	dents = drophdr.ents;
+
+	/*
+	 * If there are any stale leaf entries, take this opportunity
+	 * to purge them.
+	 */
+	if (drophdr.stale)
+		xfs_dir3_leaf_compact(args, &drophdr, drop_blk->bp);
+	if (savehdr.stale)
+		xfs_dir3_leaf_compact(args, &savehdr, save_blk->bp);
+
+	/*
+	 * Move the entries from drop to the appropriate end of save.
+	 */
+	drop_blk->hashval = be32_to_cpu(dents[drophdr.count - 1].hashval);
+	if (xfs_dir2_leafn_order(dp, save_blk->bp, drop_blk->bp))
+		xfs_dir3_leafn_moveents(args, drop_blk->bp, &drophdr, dents, 0,
+					save_blk->bp, &savehdr, sents, 0,
+					drophdr.count);
+	else
+		xfs_dir3_leafn_moveents(args, drop_blk->bp, &drophdr, dents, 0,
+					save_blk->bp, &savehdr, sents,
+					savehdr.count, drophdr.count);
+	save_blk->hashval = be32_to_cpu(sents[savehdr.count - 1].hashval);
+
+	/* log the changes made when moving the entries */
+	xfs_dir2_leaf_hdr_to_disk(dp->i_mount, save_leaf, &savehdr);
+	xfs_dir2_leaf_hdr_to_disk(dp->i_mount, drop_leaf, &drophdr);
+	xfs_dir3_leaf_log_header(args, save_blk->bp);
+	xfs_dir3_leaf_log_header(args, drop_blk->bp);
+
+	xfs_dir3_leaf_check(dp, save_blk->bp);
+	xfs_dir3_leaf_check(dp, drop_blk->bp);
+}
+
+/*
+ * Add a new data block to the directory at the free space index that the caller
+ * has specified.
+ */
+static int
+xfs_dir2_node_add_datablk(
+	struct xfs_da_args	*args,
+	struct xfs_da_state_blk	*fblk,
+	xfs_dir2_db_t		*dbno,
+	struct xfs_buf		**dbpp,
+	struct xfs_buf		**fbpp,
+	struct xfs_dir3_icfree_hdr *hdr,
+	int			*findex)
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_trans	*tp = args->trans;
+	struct xfs_mount	*mp = dp->i_mount;
+	struct xfs_dir2_data_free *bf;
+	xfs_dir2_db_t		fbno;
+	struct xfs_buf		*fbp;
+	struct xfs_buf		*dbp;
+	int			error;
+
+	/* Not allowed to allocate, return failure. */
+	if (args->total == 0)
+		return -ENOSPC;
+
+	/* Allocate and initialize the new data block.  */
+	error = xfs_dir2_grow_inode(args, XFS_DIR2_DATA_SPACE, dbno);
+	if (error)
+		return error;
+	error = xfs_dir3_data_init(args, *dbno, &dbp);
+	if (error)
+		return error;
+
+	/*
+	 * Get the freespace block corresponding to the data block
+	 * that was just allocated.
+	 */
+	fbno = xfs_dir2_db_to_fdb(args->geo, *dbno);
+	error = xfs_dir2_free_try_read(tp, dp,
+			       xfs_dir2_db_to_da(args->geo, fbno), &fbp);
+	if (error)
+		return error;
+
+	/*
+	 * If there wasn't a freespace block, the read will
+	 * return a NULL fbp.  Allocate and initialize a new one.
+	 */
+	if (!fbp) {
+		error = xfs_dir2_grow_inode(args, XFS_DIR2_FREE_SPACE, &fbno);
+		if (error)
+			return error;
+
+		if (XFS_IS_CORRUPT(mp,
+				   xfs_dir2_db_to_fdb(args->geo, *dbno) !=
+				   fbno)) {
+			xfs_alert(mp,
+"%s: dir ino %llu needed freesp block %lld for data block %lld, got %lld",
+				__func__, (unsigned long long)dp->i_ino,
+				(long long)xfs_dir2_db_to_fdb(args->geo, *dbno),
+				(long long)*dbno, (long long)fbno);
+			if (fblk) {
+				xfs_alert(mp,
+			" fblk "PTR_FMT" blkno %llu index %d magic 0x%x",
+					fblk, (unsigned long long)fblk->blkno,
+					fblk->index, fblk->magic);
+			} else {
+				xfs_alert(mp, " ... fblk is NULL");
+			}
+			return -EFSCORRUPTED;
+		}
+
+		/* Get a buffer for the new block. */
+		error = xfs_dir3_free_get_buf(args, fbno, &fbp);
+		if (error)
+			return error;
+		xfs_dir2_free_hdr_from_disk(mp, hdr, fbp->b_addr);
+
+		/* Remember the first slot as our empty slot. */
+		hdr->firstdb = (fbno - xfs_dir2_byte_to_db(args->geo,
+							XFS_DIR2_FREE_OFFSET)) *
+				args->geo->free_max_bests;
+	} else {
+		xfs_dir2_free_hdr_from_disk(mp, hdr, fbp->b_addr);
+	}
+
+	/* Set the freespace block index from the data block number. */
+	*findex = xfs_dir2_db_to_fdindex(args->geo, *dbno);
+
+	/* Extend the freespace table if the new data block is off the end. */
+	if (*findex >= hdr->nvalid) {
+		ASSERT(*findex < args->geo->free_max_bests);
+		hdr->nvalid = *findex + 1;
+		hdr->bests[*findex] = cpu_to_be16(NULLDATAOFF);
+	}
+
+	/*
+	 * If this entry was for an empty data block (this should always be
+	 * true) then update the header.
+	 */
+	if (hdr->bests[*findex] == cpu_to_be16(NULLDATAOFF)) {
+		hdr->nused++;
+		xfs_dir2_free_hdr_to_disk(mp, fbp->b_addr, hdr);
+		xfs_dir2_free_log_header(args, fbp);
+	}
+
+	/* Update the freespace value for the new block in the table. */
+	bf = xfs_dir2_data_bestfree_p(mp, dbp->b_addr);
+	hdr->bests[*findex] = bf[0].length;
+
+	*dbpp = dbp;
+	*fbpp = fbp;
+	return 0;
+}
+
+static int
+xfs_dir2_node_find_freeblk(
+	struct xfs_da_args	*args,
+	struct xfs_da_state_blk	*fblk,
+	xfs_dir2_db_t		*dbnop,
+	struct xfs_buf		**fbpp,
+	struct xfs_dir3_icfree_hdr *hdr,
+	int			*findexp,
+	int			length)
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_trans	*tp = args->trans;
+	struct xfs_buf		*fbp = NULL;
+	xfs_dir2_db_t		firstfbno;
+	xfs_dir2_db_t		lastfbno;
+	xfs_dir2_db_t		ifbno = -1;
+	xfs_dir2_db_t		dbno = -1;
+	xfs_dir2_db_t		fbno;
+	xfs_fileoff_t		fo;
+	int			findex = 0;
+	int			error;
+
+	/*
+	 * If we came in with a freespace block that means that lookup
+	 * found an entry with our hash value.  This is the freespace
+	 * block for that data entry.
+	 */
+	if (fblk) {
+		fbp = fblk->bp;
+		findex = fblk->index;
+		xfs_dir2_free_hdr_from_disk(dp->i_mount, hdr, fbp->b_addr);
+		if (findex >= 0) {
+			/* caller already found the freespace for us. */
+			ASSERT(findex < hdr->nvalid);
+			ASSERT(be16_to_cpu(hdr->bests[findex]) != NULLDATAOFF);
+			ASSERT(be16_to_cpu(hdr->bests[findex]) >= length);
+			dbno = hdr->firstdb + findex;
+			goto found_block;
+		}
+
+		/*
+		 * The data block looked at didn't have enough room.
+		 * We'll start at the beginning of the freespace entries.
+		 */
+		ifbno = fblk->blkno;
+		xfs_trans_brelse(tp, fbp);
+		fbp = NULL;
+		fblk->bp = NULL;
+	}
+
+	/*
+	 * If we don't have a data block yet, we're going to scan the freespace
+	 * data for a data block with enough free space in it.
+	 */
+	error = xfs_bmap_last_offset(dp, &fo, XFS_DATA_FORK);
+	if (error)
+		return error;
+	lastfbno = xfs_dir2_da_to_db(args->geo, (xfs_dablk_t)fo);
+	firstfbno = xfs_dir2_byte_to_db(args->geo, XFS_DIR2_FREE_OFFSET);
+
+	for (fbno = lastfbno - 1; fbno >= firstfbno; fbno--) {
+		/* If it's ifbno we already looked at it. */
+		if (fbno == ifbno)
+			continue;
+
+		/*
+		 * Read the block.  There can be holes in the freespace blocks,
+		 * so this might not succeed.  This should be really rare, so
+		 * there's no reason to avoid it.
+		 */
+		error = xfs_dir2_free_try_read(tp, dp,
+				xfs_dir2_db_to_da(args->geo, fbno),
+				&fbp);
+		if (error)
+			return error;
+		if (!fbp)
+			continue;
+
+		xfs_dir2_free_hdr_from_disk(dp->i_mount, hdr, fbp->b_addr);
+
+		/* Scan the free entry array for a large enough free space. */
+		for (findex = hdr->nvalid - 1; findex >= 0; findex--) {
+			if (be16_to_cpu(hdr->bests[findex]) != NULLDATAOFF &&
+			    be16_to_cpu(hdr->bests[findex]) >= length) {
+				dbno = hdr->firstdb + findex;
+				goto found_block;
+			}
+		}
+
+		/* Didn't find free space, go on to next free block */
+		xfs_trans_brelse(tp, fbp);
+	}
+
+found_block:
+	*dbnop = dbno;
+	*fbpp = fbp;
+	*findexp = findex;
+	return 0;
+}
+
+/*
+ * Add the data entry for a node-format directory name addition.
+ * The leaf entry is added in xfs_dir2_leafn_add.
+ * We may enter with a freespace block that the lookup found.
+ */
+static int
+xfs_dir2_node_addname_int(
+	struct xfs_da_args	*args,		/* operation arguments */
+	struct xfs_da_state_blk	*fblk)		/* optional freespace block */
+{
+	struct xfs_dir2_data_unused *dup;	/* data unused entry pointer */
+	struct xfs_dir2_data_entry *dep;	/* data entry pointer */
+	struct xfs_dir2_data_hdr *hdr;		/* data block header */
+	struct xfs_dir2_data_free *bf;
+	struct xfs_trans	*tp = args->trans;
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_dir3_icfree_hdr freehdr;
+	struct xfs_buf		*dbp;		/* data block buffer */
+	struct xfs_buf		*fbp;		/* freespace buffer */
+	xfs_dir2_data_aoff_t	aoff;
+	xfs_dir2_db_t		dbno;		/* data block number */
+	int			error;		/* error return value */
+	int			findex;		/* freespace entry index */
+	int			length;		/* length of the new entry */
+	int			logfree = 0;	/* need to log free entry */
+	int			needlog = 0;	/* need to log data header */
+	int			needscan = 0;	/* need to rescan data frees */
+	__be16			*tagp;		/* data entry tag pointer */
+
+	length = xfs_dir2_data_entsize(dp->i_mount, args->namelen);
+	error = xfs_dir2_node_find_freeblk(args, fblk, &dbno, &fbp, &freehdr,
+					   &findex, length);
+	if (error)
+		return error;
+
+	/*
+	 * Now we know if we must allocate blocks, so if we are checking whether
+	 * we can insert without allocation then we can return now.
+	 */
+	if (args->op_flags & XFS_DA_OP_JUSTCHECK) {
+		if (dbno == -1)
+			return -ENOSPC;
+		return 0;
+	}
+
+	/*
+	 * If we don't have a data block, we need to allocate one and make
+	 * the freespace entries refer to it.
+	 */
+	if (dbno == -1) {
+		/* we're going to have to log the free block index later */
+		logfree = 1;
+		error = xfs_dir2_node_add_datablk(args, fblk, &dbno, &dbp, &fbp,
+						  &freehdr, &findex);
+	} else {
+		/* Read the data block in. */
+		error = xfs_dir3_data_read(tp, dp,
+					   xfs_dir2_db_to_da(args->geo, dbno),
+					   0, &dbp);
+	}
+	if (error)
+		return error;
+
+	/* setup for data block up now */
+	hdr = dbp->b_addr;
+	bf = xfs_dir2_data_bestfree_p(dp->i_mount, hdr);
+	ASSERT(be16_to_cpu(bf[0].length) >= length);
+
+	/* Point to the existing unused space. */
+	dup = (xfs_dir2_data_unused_t *)
+	      ((char *)hdr + be16_to_cpu(bf[0].offset));
+
+	/* Mark the first part of the unused space, inuse for us. */
+	aoff = (xfs_dir2_data_aoff_t)((char *)dup - (char *)hdr);
+	error = xfs_dir2_data_use_free(args, dbp, dup, aoff, length,
+			&needlog, &needscan);
+	if (error) {
+		xfs_trans_brelse(tp, dbp);
+		return error;
+	}
+
+	/* Fill in the new entry and log it. */
+	dep = (xfs_dir2_data_entry_t *)dup;
+	dep->inumber = cpu_to_be64(args->inumber);
+	dep->namelen = args->namelen;
+	memcpy(dep->name, args->name, dep->namelen);
+	xfs_dir2_data_put_ftype(dp->i_mount, dep, args->filetype);
+	tagp = xfs_dir2_data_entry_tag_p(dp->i_mount, dep);
+	*tagp = cpu_to_be16((char *)dep - (char *)hdr);
+	xfs_dir2_data_log_entry(args, dbp, dep);
+
+	/* Rescan the freespace and log the data block if needed. */
+	if (needscan)
+		xfs_dir2_data_freescan(dp->i_mount, hdr, &needlog);
+	if (needlog)
+		xfs_dir2_data_log_header(args, dbp);
+
+	/* If the freespace block entry is now wrong, update it. */
+	if (freehdr.bests[findex] != bf[0].length) {
+		freehdr.bests[findex] = bf[0].length;
+		logfree = 1;
+	}
+
+	/* Log the freespace entry if needed. */
+	if (logfree)
+		xfs_dir2_free_log_bests(args, &freehdr, fbp, findex, findex);
+
+	/* Return the data block and offset in args. */
+	args->blkno = (xfs_dablk_t)dbno;
+	args->index = be16_to_cpu(*tagp);
+	return 0;
+}
+
+/*
+ * Top-level node form directory addname routine.
+ */
+int						/* error */
+xfs_dir2_node_addname(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	xfs_da_state_blk_t	*blk;		/* leaf block for insert */
+	int			error;		/* error return value */
+	int			rval;		/* sub-return value */
+	xfs_da_state_t		*state;		/* btree cursor */
+
+	trace_xfs_dir2_node_addname(args);
+
+	/*
+	 * Allocate and initialize the state (btree cursor).
+	 */
+	state = xfs_da_state_alloc(args);
+	/*
+	 * Look up the name.  We're not supposed to find it, but
+	 * this gives us the insertion point.
+	 */
+	error = xfs_da3_node_lookup_int(state, &rval);
+	if (error)
+		rval = error;
+	if (rval != -ENOENT) {
+		goto done;
+	}
+	/*
+	 * Add the data entry to a data block.
+	 * Extravalid is set to a freeblock found by lookup.
+	 */
+	rval = xfs_dir2_node_addname_int(args,
+		state->extravalid ? &state->extrablk : NULL);
+	if (rval) {
+		goto done;
+	}
+	blk = &state->path.blk[state->path.active - 1];
+	ASSERT(blk->magic == XFS_DIR2_LEAFN_MAGIC);
+	/*
+	 * Add the new leaf entry.
+	 */
+	rval = xfs_dir2_leafn_add(blk->bp, args, blk->index);
+	if (rval == 0) {
+		/*
+		 * It worked, fix the hash values up the btree.
+		 */
+		if (!(args->op_flags & XFS_DA_OP_JUSTCHECK))
+			xfs_da3_fixhashpath(state, &state->path);
+	} else {
+		/*
+		 * It didn't work, we need to split the leaf block.
+		 */
+		if (args->total == 0) {
+			ASSERT(rval == -ENOSPC);
+			goto done;
+		}
+		/*
+		 * Split the leaf block and insert the new entry.
+		 */
+		rval = xfs_da3_split(state);
+	}
+done:
+	xfs_da_state_free(state);
+	return rval;
+}
+
+/*
+ * Lookup an entry in a node-format directory.
+ * All the real work happens in xfs_da3_node_lookup_int.
+ * The only real output is the inode number of the entry.
+ */
+int						/* error */
+xfs_dir2_node_lookup(
+	xfs_da_args_t	*args)			/* operation arguments */
+{
+	int		error;			/* error return value */
+	int		i;			/* btree level */
+	int		rval;			/* operation return value */
+	xfs_da_state_t	*state;			/* btree cursor */
+
+	trace_xfs_dir2_node_lookup(args);
+
+	/*
+	 * Allocate and initialize the btree cursor.
+	 */
+	state = xfs_da_state_alloc(args);
+
+	/*
+	 * Fill in the path to the entry in the cursor.
+	 */
+	error = xfs_da3_node_lookup_int(state, &rval);
+	if (error)
+		rval = error;
+	else if (rval == -ENOENT && args->cmpresult == XFS_CMP_CASE) {
+		/* If a CI match, dup the actual name and return -EEXIST */
+		xfs_dir2_data_entry_t	*dep;
+
+		dep = (xfs_dir2_data_entry_t *)
+			((char *)state->extrablk.bp->b_addr +
+						 state->extrablk.index);
+		rval = xfs_dir_cilookup_result(args, dep->name, dep->namelen);
+	}
+	/*
+	 * Release the btree blocks and leaf block.
+	 */
+	for (i = 0; i < state->path.active; i++) {
+		xfs_trans_brelse(args->trans, state->path.blk[i].bp);
+		state->path.blk[i].bp = NULL;
+	}
+	/*
+	 * Release the data block if we have it.
+	 */
+	if (state->extravalid && state->extrablk.bp) {
+		xfs_trans_brelse(args->trans, state->extrablk.bp);
+		state->extrablk.bp = NULL;
+	}
+	xfs_da_state_free(state);
+	return rval;
+}
+
+/*
+ * Remove an entry from a node-format directory.
+ */
+int						/* error */
+xfs_dir2_node_removename(
+	struct xfs_da_args	*args)		/* operation arguments */
+{
+	struct xfs_da_state_blk	*blk;		/* leaf block */
+	int			error;		/* error return value */
+	int			rval;		/* operation return value */
+	struct xfs_da_state	*state;		/* btree cursor */
+
+	trace_xfs_dir2_node_removename(args);
+
+	/*
+	 * Allocate and initialize the btree cursor.
+	 */
+	state = xfs_da_state_alloc(args);
+
+	/* Look up the entry we're deleting, set up the cursor. */
+	error = xfs_da3_node_lookup_int(state, &rval);
+	if (error)
+		goto out_free;
+
+	/* Didn't find it, upper layer screwed up. */
+	if (rval != -EEXIST) {
+		error = rval;
+		goto out_free;
+	}
+
+	blk = &state->path.blk[state->path.active - 1];
+	ASSERT(blk->magic == XFS_DIR2_LEAFN_MAGIC);
+	ASSERT(state->extravalid);
+	/*
+	 * Remove the leaf and data entries.
+	 * Extrablk refers to the data block.
+	 */
+	error = xfs_dir2_leafn_remove(args, blk->bp, blk->index,
+		&state->extrablk, &rval);
+	if (error)
+		goto out_free;
+	/*
+	 * Fix the hash values up the btree.
+	 */
+	xfs_da3_fixhashpath(state, &state->path);
+	/*
+	 * If we need to join leaf blocks, do it.
+	 */
+	if (rval && state->path.active > 1)
+		error = xfs_da3_join(state);
+	/*
+	 * If no errors so far, try conversion to leaf format.
+	 */
+	if (!error)
+		error = xfs_dir2_node_to_leaf(state);
+out_free:
+	xfs_da_state_free(state);
+	return error;
+}
+
+/*
+ * Replace an entry's inode number in a node-format directory.
+ */
+int						/* error */
+xfs_dir2_node_replace(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	xfs_da_state_blk_t	*blk;		/* leaf block */
+	xfs_dir2_data_hdr_t	*hdr;		/* data block header */
+	xfs_dir2_data_entry_t	*dep;		/* data entry changed */
+	int			error;		/* error return value */
+	int			i;		/* btree level */
+	xfs_ino_t		inum;		/* new inode number */
+	int			ftype;		/* new file type */
+	int			rval;		/* internal return value */
+	xfs_da_state_t		*state;		/* btree cursor */
+
+	trace_xfs_dir2_node_replace(args);
+
+	/*
+	 * Allocate and initialize the btree cursor.
+	 */
+	state = xfs_da_state_alloc(args);
+
+	/*
+	 * We have to save new inode number and ftype since
+	 * xfs_da3_node_lookup_int() is going to overwrite them
+	 */
+	inum = args->inumber;
+	ftype = args->filetype;
+
+	/*
+	 * Lookup the entry to change in the btree.
+	 */
+	error = xfs_da3_node_lookup_int(state, &rval);
+	if (error) {
+		rval = error;
+	}
+	/*
+	 * It should be found, since the vnodeops layer has looked it up
+	 * and locked it.  But paranoia is good.
+	 */
+	if (rval == -EEXIST) {
+		struct xfs_dir3_icleaf_hdr	leafhdr;
+
+		/*
+		 * Find the leaf entry.
+		 */
+		blk = &state->path.blk[state->path.active - 1];
+		ASSERT(blk->magic == XFS_DIR2_LEAFN_MAGIC);
+		ASSERT(state->extravalid);
+
+		xfs_dir2_leaf_hdr_from_disk(state->mp, &leafhdr,
+					    blk->bp->b_addr);
+		/*
+		 * Point to the data entry.
+		 */
+		hdr = state->extrablk.bp->b_addr;
+		ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+		       hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC));
+		dep = (xfs_dir2_data_entry_t *)
+		      ((char *)hdr +
+		       xfs_dir2_dataptr_to_off(args->geo,
+				be32_to_cpu(leafhdr.ents[blk->index].address)));
+		ASSERT(inum != be64_to_cpu(dep->inumber));
+		/*
+		 * Fill in the new inode number and log the entry.
+		 */
+		dep->inumber = cpu_to_be64(inum);
+		xfs_dir2_data_put_ftype(state->mp, dep, ftype);
+		xfs_dir2_data_log_entry(args, state->extrablk.bp, dep);
+		rval = 0;
+	}
+	/*
+	 * Didn't find it, and we're holding a data block.  Drop it.
+	 */
+	else if (state->extravalid) {
+		xfs_trans_brelse(args->trans, state->extrablk.bp);
+		state->extrablk.bp = NULL;
+	}
+	/*
+	 * Release all the buffers in the cursor.
+	 */
+	for (i = 0; i < state->path.active; i++) {
+		xfs_trans_brelse(args->trans, state->path.blk[i].bp);
+		state->path.blk[i].bp = NULL;
+	}
+	xfs_da_state_free(state);
+	return rval;
+}
+
+/*
+ * Trim off a trailing empty freespace block.
+ * Return (in rvalp) 1 if we did it, 0 if not.
+ */
+int						/* error */
+xfs_dir2_node_trim_free(
+	xfs_da_args_t		*args,		/* operation arguments */
+	xfs_fileoff_t		fo,		/* free block number */
+	int			*rvalp)		/* out: did something */
+{
+	struct xfs_buf		*bp;		/* freespace buffer */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return code */
+	xfs_dir2_free_t		*free;		/* freespace structure */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	struct xfs_dir3_icfree_hdr freehdr;
+
+	dp = args->dp;
+	tp = args->trans;
+
+	*rvalp = 0;
+
+	/*
+	 * Read the freespace block.
+	 */
+	error = xfs_dir2_free_try_read(tp, dp, fo, &bp);
+	if (error)
+		return error;
+	/*
+	 * There can be holes in freespace.  If fo is a hole, there's
+	 * nothing to do.
+	 */
+	if (!bp)
+		return 0;
+	free = bp->b_addr;
+	xfs_dir2_free_hdr_from_disk(dp->i_mount, &freehdr, free);
+
+	/*
+	 * If there are used entries, there's nothing to do.
+	 */
+	if (freehdr.nused > 0) {
+		xfs_trans_brelse(tp, bp);
+		return 0;
+	}
+	/*
+	 * Blow the block away.
+	 */
+	error = xfs_dir2_shrink_inode(args,
+			xfs_dir2_da_to_db(args->geo, (xfs_dablk_t)fo), bp);
+	if (error) {
+		/*
+		 * Can't fail with ENOSPC since that only happens with no
+		 * space reservation, when breaking up an extent into two
+		 * pieces.  This is the last block of an extent.
+		 */
+		ASSERT(error != -ENOSPC);
+		xfs_trans_brelse(tp, bp);
+		return error;
+	}
+	/*
+	 * Return that we succeeded.
+	 */
+	*rvalp = 1;
+	return 0;
+}
diff --git a/fs/xfs/libxfs/xfs_dir2_priv.h b/fs/xfs/libxfs/xfs_dir2_priv.h
new file mode 100644
index 000000000..7404a9ff1
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_dir2_priv.h
@@ -0,0 +1,209 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_DIR2_PRIV_H__
+#define __XFS_DIR2_PRIV_H__
+
+struct dir_context;
+
+/*
+ * In-core version of the leaf and free block headers to abstract the
+ * differences in the v2 and v3 disk format of the headers.
+ */
+struct xfs_dir3_icleaf_hdr {
+	uint32_t		forw;
+	uint32_t		back;
+	uint16_t		magic;
+	uint16_t		count;
+	uint16_t		stale;
+
+	/*
+	 * Pointer to the on-disk format entries, which are behind the
+	 * variable size (v4 vs v5) header in the on-disk block.
+	 */
+	struct xfs_dir2_leaf_entry *ents;
+};
+
+struct xfs_dir3_icfree_hdr {
+	uint32_t		magic;
+	uint32_t		firstdb;
+	uint32_t		nvalid;
+	uint32_t		nused;
+
+	/*
+	 * Pointer to the on-disk format entries, which are behind the
+	 * variable size (v4 vs v5) header in the on-disk block.
+	 */
+	__be16			*bests;
+};
+
+/* xfs_dir2.c */
+xfs_dahash_t xfs_ascii_ci_hashname(const struct xfs_name *name);
+enum xfs_dacmp xfs_ascii_ci_compname(struct xfs_da_args *args,
+		const unsigned char *name, int len);
+extern int xfs_dir2_grow_inode(struct xfs_da_args *args, int space,
+				xfs_dir2_db_t *dbp);
+extern int xfs_dir_cilookup_result(struct xfs_da_args *args,
+				const unsigned char *name, int len);
+
+
+/* xfs_dir2_block.c */
+extern int xfs_dir3_block_read(struct xfs_trans *tp, struct xfs_inode *dp,
+			       struct xfs_buf **bpp);
+extern int xfs_dir2_block_addname(struct xfs_da_args *args);
+extern int xfs_dir2_block_lookup(struct xfs_da_args *args);
+extern int xfs_dir2_block_removename(struct xfs_da_args *args);
+extern int xfs_dir2_block_replace(struct xfs_da_args *args);
+extern int xfs_dir2_leaf_to_block(struct xfs_da_args *args,
+		struct xfs_buf *lbp, struct xfs_buf *dbp);
+
+/* xfs_dir2_data.c */
+struct xfs_dir2_data_free *xfs_dir2_data_bestfree_p(struct xfs_mount *mp,
+		struct xfs_dir2_data_hdr *hdr);
+__be16 *xfs_dir2_data_entry_tag_p(struct xfs_mount *mp,
+		struct xfs_dir2_data_entry *dep);
+uint8_t xfs_dir2_data_get_ftype(struct xfs_mount *mp,
+		struct xfs_dir2_data_entry *dep);
+void xfs_dir2_data_put_ftype(struct xfs_mount *mp,
+		struct xfs_dir2_data_entry *dep, uint8_t ftype);
+
+#ifdef DEBUG
+extern void xfs_dir3_data_check(struct xfs_inode *dp, struct xfs_buf *bp);
+#else
+#define	xfs_dir3_data_check(dp,bp)
+#endif
+
+extern xfs_failaddr_t __xfs_dir3_data_check(struct xfs_inode *dp,
+		struct xfs_buf *bp);
+int xfs_dir3_data_read(struct xfs_trans *tp, struct xfs_inode *dp,
+		xfs_dablk_t bno, unsigned int flags, struct xfs_buf **bpp);
+int xfs_dir3_data_readahead(struct xfs_inode *dp, xfs_dablk_t bno,
+		unsigned int flags);
+
+extern struct xfs_dir2_data_free *
+xfs_dir2_data_freeinsert(struct xfs_dir2_data_hdr *hdr,
+		struct xfs_dir2_data_free *bf, struct xfs_dir2_data_unused *dup,
+		int *loghead);
+extern int xfs_dir3_data_init(struct xfs_da_args *args, xfs_dir2_db_t blkno,
+		struct xfs_buf **bpp);
+
+/* xfs_dir2_leaf.c */
+void xfs_dir2_leaf_hdr_from_disk(struct xfs_mount *mp,
+		struct xfs_dir3_icleaf_hdr *to, struct xfs_dir2_leaf *from);
+void xfs_dir2_leaf_hdr_to_disk(struct xfs_mount *mp, struct xfs_dir2_leaf *to,
+		struct xfs_dir3_icleaf_hdr *from);
+int xfs_dir3_leaf_read(struct xfs_trans *tp, struct xfs_inode *dp,
+		xfs_dablk_t fbno, struct xfs_buf **bpp);
+int xfs_dir3_leafn_read(struct xfs_trans *tp, struct xfs_inode *dp,
+		xfs_dablk_t fbno, struct xfs_buf **bpp);
+extern int xfs_dir2_block_to_leaf(struct xfs_da_args *args,
+		struct xfs_buf *dbp);
+extern int xfs_dir2_leaf_addname(struct xfs_da_args *args);
+extern void xfs_dir3_leaf_compact(struct xfs_da_args *args,
+		struct xfs_dir3_icleaf_hdr *leafhdr, struct xfs_buf *bp);
+extern void xfs_dir3_leaf_compact_x1(struct xfs_dir3_icleaf_hdr *leafhdr,
+		struct xfs_dir2_leaf_entry *ents, int *indexp,
+		int *lowstalep, int *highstalep, int *lowlogp, int *highlogp);
+extern int xfs_dir3_leaf_get_buf(struct xfs_da_args *args, xfs_dir2_db_t bno,
+		struct xfs_buf **bpp, uint16_t magic);
+extern void xfs_dir3_leaf_log_ents(struct xfs_da_args *args,
+		struct xfs_dir3_icleaf_hdr *hdr, struct xfs_buf *bp, int first,
+		int last);
+extern void xfs_dir3_leaf_log_header(struct xfs_da_args *args,
+		struct xfs_buf *bp);
+extern int xfs_dir2_leaf_lookup(struct xfs_da_args *args);
+extern int xfs_dir2_leaf_removename(struct xfs_da_args *args);
+extern int xfs_dir2_leaf_replace(struct xfs_da_args *args);
+extern int xfs_dir2_leaf_search_hash(struct xfs_da_args *args,
+		struct xfs_buf *lbp);
+extern int xfs_dir2_leaf_trim_data(struct xfs_da_args *args,
+		struct xfs_buf *lbp, xfs_dir2_db_t db);
+extern struct xfs_dir2_leaf_entry *
+xfs_dir3_leaf_find_entry(struct xfs_dir3_icleaf_hdr *leafhdr,
+		struct xfs_dir2_leaf_entry *ents, int index, int compact,
+		int lowstale, int highstale, int *lfloglow, int *lfloghigh);
+extern int xfs_dir2_node_to_leaf(struct xfs_da_state *state);
+
+extern xfs_failaddr_t xfs_dir3_leaf_check_int(struct xfs_mount *mp,
+		struct xfs_dir3_icleaf_hdr *hdr, struct xfs_dir2_leaf *leaf,
+		bool expensive_checks);
+
+/* xfs_dir2_node.c */
+void xfs_dir2_free_hdr_from_disk(struct xfs_mount *mp,
+		struct xfs_dir3_icfree_hdr *to, struct xfs_dir2_free *from);
+extern int xfs_dir2_leaf_to_node(struct xfs_da_args *args,
+		struct xfs_buf *lbp);
+extern xfs_dahash_t xfs_dir2_leaf_lasthash(struct xfs_inode *dp,
+		struct xfs_buf *bp, int *count);
+extern int xfs_dir2_leafn_lookup_int(struct xfs_buf *bp,
+		struct xfs_da_args *args, int *indexp,
+		struct xfs_da_state *state);
+extern int xfs_dir2_leafn_order(struct xfs_inode *dp, struct xfs_buf *leaf1_bp,
+		struct xfs_buf *leaf2_bp);
+extern int xfs_dir2_leafn_split(struct xfs_da_state *state,
+	struct xfs_da_state_blk *oldblk, struct xfs_da_state_blk *newblk);
+extern int xfs_dir2_leafn_toosmall(struct xfs_da_state *state, int *action);
+extern void xfs_dir2_leafn_unbalance(struct xfs_da_state *state,
+		struct xfs_da_state_blk *drop_blk,
+		struct xfs_da_state_blk *save_blk);
+extern int xfs_dir2_node_addname(struct xfs_da_args *args);
+extern int xfs_dir2_node_lookup(struct xfs_da_args *args);
+extern int xfs_dir2_node_removename(struct xfs_da_args *args);
+extern int xfs_dir2_node_replace(struct xfs_da_args *args);
+extern int xfs_dir2_node_trim_free(struct xfs_da_args *args, xfs_fileoff_t fo,
+		int *rvalp);
+extern int xfs_dir2_free_read(struct xfs_trans *tp, struct xfs_inode *dp,
+		xfs_dablk_t fbno, struct xfs_buf **bpp);
+
+/* xfs_dir2_sf.c */
+xfs_ino_t xfs_dir2_sf_get_ino(struct xfs_mount *mp, struct xfs_dir2_sf_hdr *hdr,
+		struct xfs_dir2_sf_entry *sfep);
+xfs_ino_t xfs_dir2_sf_get_parent_ino(struct xfs_dir2_sf_hdr *hdr);
+void xfs_dir2_sf_put_parent_ino(struct xfs_dir2_sf_hdr *hdr, xfs_ino_t ino);
+uint8_t xfs_dir2_sf_get_ftype(struct xfs_mount *mp,
+		struct xfs_dir2_sf_entry *sfep);
+struct xfs_dir2_sf_entry *xfs_dir2_sf_nextentry(struct xfs_mount *mp,
+		struct xfs_dir2_sf_hdr *hdr, struct xfs_dir2_sf_entry *sfep);
+extern int xfs_dir2_block_sfsize(struct xfs_inode *dp,
+		struct xfs_dir2_data_hdr *block, struct xfs_dir2_sf_hdr *sfhp);
+extern int xfs_dir2_block_to_sf(struct xfs_da_args *args, struct xfs_buf *bp,
+		int size, xfs_dir2_sf_hdr_t *sfhp);
+extern int xfs_dir2_sf_addname(struct xfs_da_args *args);
+extern int xfs_dir2_sf_create(struct xfs_da_args *args, xfs_ino_t pino);
+extern int xfs_dir2_sf_lookup(struct xfs_da_args *args);
+extern int xfs_dir2_sf_removename(struct xfs_da_args *args);
+extern int xfs_dir2_sf_replace(struct xfs_da_args *args);
+extern xfs_failaddr_t xfs_dir2_sf_verify(struct xfs_inode *ip);
+int xfs_dir2_sf_entsize(struct xfs_mount *mp,
+		struct xfs_dir2_sf_hdr *hdr, int len);
+void xfs_dir2_sf_put_ino(struct xfs_mount *mp, struct xfs_dir2_sf_hdr *hdr,
+		struct xfs_dir2_sf_entry *sfep, xfs_ino_t ino);
+void xfs_dir2_sf_put_ftype(struct xfs_mount *mp,
+		struct xfs_dir2_sf_entry *sfep, uint8_t ftype);
+
+/* xfs_dir2_readdir.c */
+extern int xfs_readdir(struct xfs_trans *tp, struct xfs_inode *dp,
+		       struct dir_context *ctx, size_t bufsize);
+
+static inline unsigned int
+xfs_dir2_data_entsize(
+	struct xfs_mount	*mp,
+	unsigned int		namelen)
+{
+	unsigned int		len;
+
+	len = offsetof(struct xfs_dir2_data_entry, name[0]) + namelen +
+			sizeof(xfs_dir2_data_off_t) /* tag */;
+	if (xfs_has_ftype(mp))
+		len += sizeof(uint8_t);
+	return round_up(len, XFS_DIR2_DATA_ALIGN);
+}
+
+xfs_dahash_t xfs_dir2_hashname(struct xfs_mount *mp,
+		const struct xfs_name *name);
+enum xfs_dacmp xfs_dir2_compname(struct xfs_da_args *args,
+		const unsigned char *name, int len);
+
+#endif /* __XFS_DIR2_PRIV_H__ */
diff --git a/fs/xfs/libxfs/xfs_dir2_sf.c b/fs/xfs/libxfs/xfs_dir2_sf.c
new file mode 100644
index 000000000..8cd37e6e9
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_dir2_sf.c
@@ -0,0 +1,1293 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_inode.h"
+#include "xfs_trans.h"
+#include "xfs_dir2.h"
+#include "xfs_dir2_priv.h"
+#include "xfs_trace.h"
+
+/*
+ * Prototypes for internal functions.
+ */
+static void xfs_dir2_sf_addname_easy(xfs_da_args_t *args,
+				     xfs_dir2_sf_entry_t *sfep,
+				     xfs_dir2_data_aoff_t offset,
+				     int new_isize);
+static void xfs_dir2_sf_addname_hard(xfs_da_args_t *args, int objchange,
+				     int new_isize);
+static int xfs_dir2_sf_addname_pick(xfs_da_args_t *args, int objchange,
+				    xfs_dir2_sf_entry_t **sfepp,
+				    xfs_dir2_data_aoff_t *offsetp);
+#ifdef DEBUG
+static void xfs_dir2_sf_check(xfs_da_args_t *args);
+#else
+#define	xfs_dir2_sf_check(args)
+#endif /* DEBUG */
+
+static void xfs_dir2_sf_toino4(xfs_da_args_t *args);
+static void xfs_dir2_sf_toino8(xfs_da_args_t *args);
+
+int
+xfs_dir2_sf_entsize(
+	struct xfs_mount	*mp,
+	struct xfs_dir2_sf_hdr	*hdr,
+	int			len)
+{
+	int			count = len;
+
+	count += sizeof(struct xfs_dir2_sf_entry);	/* namelen + offset */
+	count += hdr->i8count ? XFS_INO64_SIZE : XFS_INO32_SIZE; /* ino # */
+
+	if (xfs_has_ftype(mp))
+		count += sizeof(uint8_t);
+	return count;
+}
+
+struct xfs_dir2_sf_entry *
+xfs_dir2_sf_nextentry(
+	struct xfs_mount	*mp,
+	struct xfs_dir2_sf_hdr	*hdr,
+	struct xfs_dir2_sf_entry *sfep)
+{
+	return (void *)sfep + xfs_dir2_sf_entsize(mp, hdr, sfep->namelen);
+}
+
+/*
+ * In short-form directory entries the inode numbers are stored at variable
+ * offset behind the entry name. If the entry stores a filetype value, then it
+ * sits between the name and the inode number.  The actual inode numbers can
+ * come in two formats as well, either 4 bytes or 8 bytes wide.
+ */
+xfs_ino_t
+xfs_dir2_sf_get_ino(
+	struct xfs_mount		*mp,
+	struct xfs_dir2_sf_hdr		*hdr,
+	struct xfs_dir2_sf_entry	*sfep)
+{
+	uint8_t				*from = sfep->name + sfep->namelen;
+
+	if (xfs_has_ftype(mp))
+		from++;
+
+	if (!hdr->i8count)
+		return get_unaligned_be32(from);
+	return get_unaligned_be64(from) & XFS_MAXINUMBER;
+}
+
+void
+xfs_dir2_sf_put_ino(
+	struct xfs_mount		*mp,
+	struct xfs_dir2_sf_hdr		*hdr,
+	struct xfs_dir2_sf_entry	*sfep,
+	xfs_ino_t			ino)
+{
+	uint8_t				*to = sfep->name + sfep->namelen;
+
+	ASSERT(ino <= XFS_MAXINUMBER);
+
+	if (xfs_has_ftype(mp))
+		to++;
+
+	if (hdr->i8count)
+		put_unaligned_be64(ino, to);
+	else
+		put_unaligned_be32(ino, to);
+}
+
+xfs_ino_t
+xfs_dir2_sf_get_parent_ino(
+	struct xfs_dir2_sf_hdr	*hdr)
+{
+	if (!hdr->i8count)
+		return get_unaligned_be32(hdr->parent);
+	return get_unaligned_be64(hdr->parent) & XFS_MAXINUMBER;
+}
+
+void
+xfs_dir2_sf_put_parent_ino(
+	struct xfs_dir2_sf_hdr		*hdr,
+	xfs_ino_t			ino)
+{
+	ASSERT(ino <= XFS_MAXINUMBER);
+
+	if (hdr->i8count)
+		put_unaligned_be64(ino, hdr->parent);
+	else
+		put_unaligned_be32(ino, hdr->parent);
+}
+
+/*
+ * The file type field is stored at the end of the name for filetype enabled
+ * shortform directories, or not at all otherwise.
+ */
+uint8_t
+xfs_dir2_sf_get_ftype(
+	struct xfs_mount		*mp,
+	struct xfs_dir2_sf_entry	*sfep)
+{
+	if (xfs_has_ftype(mp)) {
+		uint8_t			ftype = sfep->name[sfep->namelen];
+
+		if (ftype < XFS_DIR3_FT_MAX)
+			return ftype;
+	}
+
+	return XFS_DIR3_FT_UNKNOWN;
+}
+
+void
+xfs_dir2_sf_put_ftype(
+	struct xfs_mount	*mp,
+	struct xfs_dir2_sf_entry *sfep,
+	uint8_t			ftype)
+{
+	ASSERT(ftype < XFS_DIR3_FT_MAX);
+
+	if (xfs_has_ftype(mp))
+		sfep->name[sfep->namelen] = ftype;
+}
+
+/*
+ * Given a block directory (dp/block), calculate its size as a shortform (sf)
+ * directory and a header for the sf directory, if it will fit it the
+ * space currently present in the inode.  If it won't fit, the output
+ * size is too big (but not accurate).
+ */
+int						/* size for sf form */
+xfs_dir2_block_sfsize(
+	xfs_inode_t		*dp,		/* incore inode pointer */
+	xfs_dir2_data_hdr_t	*hdr,		/* block directory data */
+	xfs_dir2_sf_hdr_t	*sfhp)		/* output: header for sf form */
+{
+	xfs_dir2_dataptr_t	addr;		/* data entry address */
+	xfs_dir2_leaf_entry_t	*blp;		/* leaf area of the block */
+	xfs_dir2_block_tail_t	*btp;		/* tail area of the block */
+	int			count;		/* shortform entry count */
+	xfs_dir2_data_entry_t	*dep;		/* data entry in the block */
+	int			i;		/* block entry index */
+	int			i8count;	/* count of big-inode entries */
+	int			isdot;		/* entry is "." */
+	int			isdotdot;	/* entry is ".." */
+	xfs_mount_t		*mp;		/* mount structure pointer */
+	int			namelen;	/* total name bytes */
+	xfs_ino_t		parent = 0;	/* parent inode number */
+	int			size=0;		/* total computed size */
+	int			has_ftype;
+	struct xfs_da_geometry	*geo;
+
+	mp = dp->i_mount;
+	geo = mp->m_dir_geo;
+
+	/*
+	 * if there is a filetype field, add the extra byte to the namelen
+	 * for each entry that we see.
+	 */
+	has_ftype = xfs_has_ftype(mp) ? 1 : 0;
+
+	count = i8count = namelen = 0;
+	btp = xfs_dir2_block_tail_p(geo, hdr);
+	blp = xfs_dir2_block_leaf_p(btp);
+
+	/*
+	 * Iterate over the block's data entries by using the leaf pointers.
+	 */
+	for (i = 0; i < be32_to_cpu(btp->count); i++) {
+		if ((addr = be32_to_cpu(blp[i].address)) == XFS_DIR2_NULL_DATAPTR)
+			continue;
+		/*
+		 * Calculate the pointer to the entry at hand.
+		 */
+		dep = (xfs_dir2_data_entry_t *)((char *)hdr +
+				xfs_dir2_dataptr_to_off(geo, addr));
+		/*
+		 * Detect . and .., so we can special-case them.
+		 * . is not included in sf directories.
+		 * .. is included by just the parent inode number.
+		 */
+		isdot = dep->namelen == 1 && dep->name[0] == '.';
+		isdotdot =
+			dep->namelen == 2 &&
+			dep->name[0] == '.' && dep->name[1] == '.';
+
+		if (!isdot)
+			i8count += be64_to_cpu(dep->inumber) > XFS_DIR2_MAX_SHORT_INUM;
+
+		/* take into account the file type field */
+		if (!isdot && !isdotdot) {
+			count++;
+			namelen += dep->namelen + has_ftype;
+		} else if (isdotdot)
+			parent = be64_to_cpu(dep->inumber);
+		/*
+		 * Calculate the new size, see if we should give up yet.
+		 */
+		size = xfs_dir2_sf_hdr_size(i8count) +	/* header */
+		       count * 3 * sizeof(u8) +		/* namelen + offset */
+		       namelen +			/* name */
+		       (i8count ?			/* inumber */
+				count * XFS_INO64_SIZE :
+				count * XFS_INO32_SIZE);
+		if (size > xfs_inode_data_fork_size(dp))
+			return size;		/* size value is a failure */
+	}
+	/*
+	 * Create the output header, if it worked.
+	 */
+	sfhp->count = count;
+	sfhp->i8count = i8count;
+	xfs_dir2_sf_put_parent_ino(sfhp, parent);
+	return size;
+}
+
+/*
+ * Convert a block format directory to shortform.
+ * Caller has already checked that it will fit, and built us a header.
+ */
+int						/* error */
+xfs_dir2_block_to_sf(
+	struct xfs_da_args	*args,		/* operation arguments */
+	struct xfs_buf		*bp,
+	int			size,		/* shortform directory size */
+	struct xfs_dir2_sf_hdr	*sfhp)		/* shortform directory hdr */
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	int			error;		/* error return value */
+	int			logflags;	/* inode logging flags */
+	struct xfs_dir2_sf_entry *sfep;		/* shortform entry */
+	struct xfs_dir2_sf_hdr	*sfp;		/* shortform directory header */
+	unsigned int		offset = args->geo->data_entry_offset;
+	unsigned int		end;
+
+	trace_xfs_dir2_block_to_sf(args);
+
+	/*
+	 * Allocate a temporary destination buffer the size of the inode to
+	 * format the data into.  Once we have formatted the data, we can free
+	 * the block and copy the formatted data into the inode literal area.
+	 */
+	sfp = kmem_alloc(mp->m_sb.sb_inodesize, 0);
+	memcpy(sfp, sfhp, xfs_dir2_sf_hdr_size(sfhp->i8count));
+
+	/*
+	 * Loop over the active and unused entries.  Stop when we reach the
+	 * leaf/tail portion of the block.
+	 */
+	end = xfs_dir3_data_end_offset(args->geo, bp->b_addr);
+	sfep = xfs_dir2_sf_firstentry(sfp);
+	while (offset < end) {
+		struct xfs_dir2_data_unused	*dup = bp->b_addr + offset;
+		struct xfs_dir2_data_entry	*dep = bp->b_addr + offset;
+
+		/*
+		 * If it's unused, just skip over it.
+		 */
+		if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG) {
+			offset += be16_to_cpu(dup->length);
+			continue;
+		}
+
+		/*
+		 * Skip .
+		 */
+		if (dep->namelen == 1 && dep->name[0] == '.')
+			ASSERT(be64_to_cpu(dep->inumber) == dp->i_ino);
+		/*
+		 * Skip .., but make sure the inode number is right.
+		 */
+		else if (dep->namelen == 2 &&
+			 dep->name[0] == '.' && dep->name[1] == '.')
+			ASSERT(be64_to_cpu(dep->inumber) ==
+			       xfs_dir2_sf_get_parent_ino(sfp));
+		/*
+		 * Normal entry, copy it into shortform.
+		 */
+		else {
+			sfep->namelen = dep->namelen;
+			xfs_dir2_sf_put_offset(sfep, offset);
+			memcpy(sfep->name, dep->name, dep->namelen);
+			xfs_dir2_sf_put_ino(mp, sfp, sfep,
+					      be64_to_cpu(dep->inumber));
+			xfs_dir2_sf_put_ftype(mp, sfep,
+					xfs_dir2_data_get_ftype(mp, dep));
+
+			sfep = xfs_dir2_sf_nextentry(mp, sfp, sfep);
+		}
+		offset += xfs_dir2_data_entsize(mp, dep->namelen);
+	}
+	ASSERT((char *)sfep - (char *)sfp == size);
+
+	/* now we are done with the block, we can shrink the inode */
+	logflags = XFS_ILOG_CORE;
+	error = xfs_dir2_shrink_inode(args, args->geo->datablk, bp);
+	if (error) {
+		ASSERT(error != -ENOSPC);
+		goto out;
+	}
+
+	/*
+	 * The buffer is now unconditionally gone, whether
+	 * xfs_dir2_shrink_inode worked or not.
+	 *
+	 * Convert the inode to local format and copy the data in.
+	 */
+	ASSERT(dp->i_df.if_bytes == 0);
+	xfs_init_local_fork(dp, XFS_DATA_FORK, sfp, size);
+	dp->i_df.if_format = XFS_DINODE_FMT_LOCAL;
+	dp->i_disk_size = size;
+
+	logflags |= XFS_ILOG_DDATA;
+	xfs_dir2_sf_check(args);
+out:
+	xfs_trans_log_inode(args->trans, dp, logflags);
+	kmem_free(sfp);
+	return error;
+}
+
+/*
+ * Add a name to a shortform directory.
+ * There are two algorithms, "easy" and "hard" which we decide on
+ * before changing anything.
+ * Convert to block form if necessary, if the new entry won't fit.
+ */
+int						/* error */
+xfs_dir2_sf_addname(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return value */
+	int			incr_isize;	/* total change in size */
+	int			new_isize;	/* size after adding name */
+	int			objchange;	/* changing to 8-byte inodes */
+	xfs_dir2_data_aoff_t	offset = 0;	/* offset for new entry */
+	int			pick;		/* which algorithm to use */
+	xfs_dir2_sf_hdr_t	*sfp;		/* shortform structure */
+	xfs_dir2_sf_entry_t	*sfep = NULL;	/* shortform entry */
+
+	trace_xfs_dir2_sf_addname(args);
+
+	ASSERT(xfs_dir2_sf_lookup(args) == -ENOENT);
+	dp = args->dp;
+	ASSERT(dp->i_df.if_format == XFS_DINODE_FMT_LOCAL);
+	ASSERT(dp->i_disk_size >= offsetof(struct xfs_dir2_sf_hdr, parent));
+	ASSERT(dp->i_df.if_bytes == dp->i_disk_size);
+	ASSERT(dp->i_df.if_u1.if_data != NULL);
+	sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	ASSERT(dp->i_disk_size >= xfs_dir2_sf_hdr_size(sfp->i8count));
+	/*
+	 * Compute entry (and change in) size.
+	 */
+	incr_isize = xfs_dir2_sf_entsize(dp->i_mount, sfp, args->namelen);
+	objchange = 0;
+
+	/*
+	 * Do we have to change to 8 byte inodes?
+	 */
+	if (args->inumber > XFS_DIR2_MAX_SHORT_INUM && sfp->i8count == 0) {
+		/*
+		 * Yes, adjust the inode size.  old count + (parent + new)
+		 */
+		incr_isize += (sfp->count + 2) * XFS_INO64_DIFF;
+		objchange = 1;
+	}
+
+	new_isize = (int)dp->i_disk_size + incr_isize;
+	/*
+	 * Won't fit as shortform any more (due to size),
+	 * or the pick routine says it won't (due to offset values).
+	 */
+	if (new_isize > xfs_inode_data_fork_size(dp) ||
+	    (pick =
+	     xfs_dir2_sf_addname_pick(args, objchange, &sfep, &offset)) == 0) {
+		/*
+		 * Just checking or no space reservation, it doesn't fit.
+		 */
+		if ((args->op_flags & XFS_DA_OP_JUSTCHECK) || args->total == 0)
+			return -ENOSPC;
+		/*
+		 * Convert to block form then add the name.
+		 */
+		error = xfs_dir2_sf_to_block(args);
+		if (error)
+			return error;
+		return xfs_dir2_block_addname(args);
+	}
+	/*
+	 * Just checking, it fits.
+	 */
+	if (args->op_flags & XFS_DA_OP_JUSTCHECK)
+		return 0;
+	/*
+	 * Do it the easy way - just add it at the end.
+	 */
+	if (pick == 1)
+		xfs_dir2_sf_addname_easy(args, sfep, offset, new_isize);
+	/*
+	 * Do it the hard way - look for a place to insert the new entry.
+	 * Convert to 8 byte inode numbers first if necessary.
+	 */
+	else {
+		ASSERT(pick == 2);
+		if (objchange)
+			xfs_dir2_sf_toino8(args);
+		xfs_dir2_sf_addname_hard(args, objchange, new_isize);
+	}
+	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_DDATA);
+	return 0;
+}
+
+/*
+ * Add the new entry the "easy" way.
+ * This is copying the old directory and adding the new entry at the end.
+ * Since it's sorted by "offset" we need room after the last offset
+ * that's already there, and then room to convert to a block directory.
+ * This is already checked by the pick routine.
+ */
+static void
+xfs_dir2_sf_addname_easy(
+	xfs_da_args_t		*args,		/* operation arguments */
+	xfs_dir2_sf_entry_t	*sfep,		/* pointer to new entry */
+	xfs_dir2_data_aoff_t	offset,		/* offset to use for new ent */
+	int			new_isize)	/* new directory size */
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	int			byteoff;	/* byte offset in sf dir */
+	xfs_dir2_sf_hdr_t	*sfp;		/* shortform structure */
+
+	sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	byteoff = (int)((char *)sfep - (char *)sfp);
+	/*
+	 * Grow the in-inode space.
+	 */
+	xfs_idata_realloc(dp, xfs_dir2_sf_entsize(mp, sfp, args->namelen),
+			  XFS_DATA_FORK);
+	/*
+	 * Need to set up again due to realloc of the inode data.
+	 */
+	sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	sfep = (xfs_dir2_sf_entry_t *)((char *)sfp + byteoff);
+	/*
+	 * Fill in the new entry.
+	 */
+	sfep->namelen = args->namelen;
+	xfs_dir2_sf_put_offset(sfep, offset);
+	memcpy(sfep->name, args->name, sfep->namelen);
+	xfs_dir2_sf_put_ino(mp, sfp, sfep, args->inumber);
+	xfs_dir2_sf_put_ftype(mp, sfep, args->filetype);
+
+	/*
+	 * Update the header and inode.
+	 */
+	sfp->count++;
+	if (args->inumber > XFS_DIR2_MAX_SHORT_INUM)
+		sfp->i8count++;
+	dp->i_disk_size = new_isize;
+	xfs_dir2_sf_check(args);
+}
+
+/*
+ * Add the new entry the "hard" way.
+ * The caller has already converted to 8 byte inode numbers if necessary,
+ * in which case we need to leave the i8count at 1.
+ * Find a hole that the new entry will fit into, and copy
+ * the first part of the entries, the new entry, and the last part of
+ * the entries.
+ */
+/* ARGSUSED */
+static void
+xfs_dir2_sf_addname_hard(
+	xfs_da_args_t		*args,		/* operation arguments */
+	int			objchange,	/* changing inode number size */
+	int			new_isize)	/* new directory size */
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	int			add_datasize;	/* data size need for new ent */
+	char			*buf;		/* buffer for old */
+	int			eof;		/* reached end of old dir */
+	int			nbytes;		/* temp for byte copies */
+	xfs_dir2_data_aoff_t	new_offset;	/* next offset value */
+	xfs_dir2_data_aoff_t	offset;		/* current offset value */
+	int			old_isize;	/* previous size */
+	xfs_dir2_sf_entry_t	*oldsfep;	/* entry in original dir */
+	xfs_dir2_sf_hdr_t	*oldsfp;	/* original shortform dir */
+	xfs_dir2_sf_entry_t	*sfep;		/* entry in new dir */
+	xfs_dir2_sf_hdr_t	*sfp;		/* new shortform dir */
+
+	/*
+	 * Copy the old directory to the stack buffer.
+	 */
+	sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	old_isize = (int)dp->i_disk_size;
+	buf = kmem_alloc(old_isize, 0);
+	oldsfp = (xfs_dir2_sf_hdr_t *)buf;
+	memcpy(oldsfp, sfp, old_isize);
+	/*
+	 * Loop over the old directory finding the place we're going
+	 * to insert the new entry.
+	 * If it's going to end up at the end then oldsfep will point there.
+	 */
+	for (offset = args->geo->data_first_offset,
+	      oldsfep = xfs_dir2_sf_firstentry(oldsfp),
+	      add_datasize = xfs_dir2_data_entsize(mp, args->namelen),
+	      eof = (char *)oldsfep == &buf[old_isize];
+	     !eof;
+	     offset = new_offset + xfs_dir2_data_entsize(mp, oldsfep->namelen),
+	      oldsfep = xfs_dir2_sf_nextentry(mp, oldsfp, oldsfep),
+	      eof = (char *)oldsfep == &buf[old_isize]) {
+		new_offset = xfs_dir2_sf_get_offset(oldsfep);
+		if (offset + add_datasize <= new_offset)
+			break;
+	}
+	/*
+	 * Get rid of the old directory, then allocate space for
+	 * the new one.  We do this so xfs_idata_realloc won't copy
+	 * the data.
+	 */
+	xfs_idata_realloc(dp, -old_isize, XFS_DATA_FORK);
+	xfs_idata_realloc(dp, new_isize, XFS_DATA_FORK);
+	/*
+	 * Reset the pointer since the buffer was reallocated.
+	 */
+	sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	/*
+	 * Copy the first part of the directory, including the header.
+	 */
+	nbytes = (int)((char *)oldsfep - (char *)oldsfp);
+	memcpy(sfp, oldsfp, nbytes);
+	sfep = (xfs_dir2_sf_entry_t *)((char *)sfp + nbytes);
+	/*
+	 * Fill in the new entry, and update the header counts.
+	 */
+	sfep->namelen = args->namelen;
+	xfs_dir2_sf_put_offset(sfep, offset);
+	memcpy(sfep->name, args->name, sfep->namelen);
+	xfs_dir2_sf_put_ino(mp, sfp, sfep, args->inumber);
+	xfs_dir2_sf_put_ftype(mp, sfep, args->filetype);
+	sfp->count++;
+	if (args->inumber > XFS_DIR2_MAX_SHORT_INUM && !objchange)
+		sfp->i8count++;
+	/*
+	 * If there's more left to copy, do that.
+	 */
+	if (!eof) {
+		sfep = xfs_dir2_sf_nextentry(mp, sfp, sfep);
+		memcpy(sfep, oldsfep, old_isize - nbytes);
+	}
+	kmem_free(buf);
+	dp->i_disk_size = new_isize;
+	xfs_dir2_sf_check(args);
+}
+
+/*
+ * Decide if the new entry will fit at all.
+ * If it will fit, pick between adding the new entry to the end (easy)
+ * or somewhere else (hard).
+ * Return 0 (won't fit), 1 (easy), 2 (hard).
+ */
+/*ARGSUSED*/
+static int					/* pick result */
+xfs_dir2_sf_addname_pick(
+	xfs_da_args_t		*args,		/* operation arguments */
+	int			objchange,	/* inode # size changes */
+	xfs_dir2_sf_entry_t	**sfepp,	/* out(1): new entry ptr */
+	xfs_dir2_data_aoff_t	*offsetp)	/* out(1): new offset */
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	int			holefit;	/* found hole it will fit in */
+	int			i;		/* entry number */
+	xfs_dir2_data_aoff_t	offset;		/* data block offset */
+	xfs_dir2_sf_entry_t	*sfep;		/* shortform entry */
+	xfs_dir2_sf_hdr_t	*sfp;		/* shortform structure */
+	int			size;		/* entry's data size */
+	int			used;		/* data bytes used */
+
+	sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	size = xfs_dir2_data_entsize(mp, args->namelen);
+	offset = args->geo->data_first_offset;
+	sfep = xfs_dir2_sf_firstentry(sfp);
+	holefit = 0;
+	/*
+	 * Loop over sf entries.
+	 * Keep track of data offset and whether we've seen a place
+	 * to insert the new entry.
+	 */
+	for (i = 0; i < sfp->count; i++) {
+		if (!holefit)
+			holefit = offset + size <= xfs_dir2_sf_get_offset(sfep);
+		offset = xfs_dir2_sf_get_offset(sfep) +
+			 xfs_dir2_data_entsize(mp, sfep->namelen);
+		sfep = xfs_dir2_sf_nextentry(mp, sfp, sfep);
+	}
+	/*
+	 * Calculate data bytes used excluding the new entry, if this
+	 * was a data block (block form directory).
+	 */
+	used = offset +
+	       (sfp->count + 3) * (uint)sizeof(xfs_dir2_leaf_entry_t) +
+	       (uint)sizeof(xfs_dir2_block_tail_t);
+	/*
+	 * If it won't fit in a block form then we can't insert it,
+	 * we'll go back, convert to block, then try the insert and convert
+	 * to leaf.
+	 */
+	if (used + (holefit ? 0 : size) > args->geo->blksize)
+		return 0;
+	/*
+	 * If changing the inode number size, do it the hard way.
+	 */
+	if (objchange)
+		return 2;
+	/*
+	 * If it won't fit at the end then do it the hard way (use the hole).
+	 */
+	if (used + size > args->geo->blksize)
+		return 2;
+	/*
+	 * Do it the easy way.
+	 */
+	*sfepp = sfep;
+	*offsetp = offset;
+	return 1;
+}
+
+#ifdef DEBUG
+/*
+ * Check consistency of shortform directory, assert if bad.
+ */
+static void
+xfs_dir2_sf_check(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	int			i;		/* entry number */
+	int			i8count;	/* number of big inode#s */
+	xfs_ino_t		ino;		/* entry inode number */
+	int			offset;		/* data offset */
+	xfs_dir2_sf_entry_t	*sfep;		/* shortform dir entry */
+	xfs_dir2_sf_hdr_t	*sfp;		/* shortform structure */
+
+	sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	offset = args->geo->data_first_offset;
+	ino = xfs_dir2_sf_get_parent_ino(sfp);
+	i8count = ino > XFS_DIR2_MAX_SHORT_INUM;
+
+	for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp);
+	     i < sfp->count;
+	     i++, sfep = xfs_dir2_sf_nextentry(mp, sfp, sfep)) {
+		ASSERT(xfs_dir2_sf_get_offset(sfep) >= offset);
+		ino = xfs_dir2_sf_get_ino(mp, sfp, sfep);
+		i8count += ino > XFS_DIR2_MAX_SHORT_INUM;
+		offset =
+			xfs_dir2_sf_get_offset(sfep) +
+			xfs_dir2_data_entsize(mp, sfep->namelen);
+		ASSERT(xfs_dir2_sf_get_ftype(mp, sfep) < XFS_DIR3_FT_MAX);
+	}
+	ASSERT(i8count == sfp->i8count);
+	ASSERT((char *)sfep - (char *)sfp == dp->i_disk_size);
+	ASSERT(offset +
+	       (sfp->count + 2) * (uint)sizeof(xfs_dir2_leaf_entry_t) +
+	       (uint)sizeof(xfs_dir2_block_tail_t) <= args->geo->blksize);
+}
+#endif	/* DEBUG */
+
+/* Verify the consistency of an inline directory. */
+xfs_failaddr_t
+xfs_dir2_sf_verify(
+	struct xfs_inode		*ip)
+{
+	struct xfs_mount		*mp = ip->i_mount;
+	struct xfs_ifork		*ifp = xfs_ifork_ptr(ip, XFS_DATA_FORK);
+	struct xfs_dir2_sf_hdr		*sfp;
+	struct xfs_dir2_sf_entry	*sfep;
+	struct xfs_dir2_sf_entry	*next_sfep;
+	char				*endp;
+	xfs_ino_t			ino;
+	int				i;
+	int				i8count;
+	int				offset;
+	int64_t				size;
+	int				error;
+	uint8_t				filetype;
+
+	ASSERT(ifp->if_format == XFS_DINODE_FMT_LOCAL);
+
+	sfp = (struct xfs_dir2_sf_hdr *)ifp->if_u1.if_data;
+	size = ifp->if_bytes;
+
+	/*
+	 * Give up if the directory is way too short.
+	 */
+	if (size <= offsetof(struct xfs_dir2_sf_hdr, parent) ||
+	    size < xfs_dir2_sf_hdr_size(sfp->i8count))
+		return __this_address;
+
+	endp = (char *)sfp + size;
+
+	/* Check .. entry */
+	ino = xfs_dir2_sf_get_parent_ino(sfp);
+	i8count = ino > XFS_DIR2_MAX_SHORT_INUM;
+	error = xfs_dir_ino_validate(mp, ino);
+	if (error)
+		return __this_address;
+	offset = mp->m_dir_geo->data_first_offset;
+
+	/* Check all reported entries */
+	sfep = xfs_dir2_sf_firstentry(sfp);
+	for (i = 0; i < sfp->count; i++) {
+		/*
+		 * struct xfs_dir2_sf_entry has a variable length.
+		 * Check the fixed-offset parts of the structure are
+		 * within the data buffer.
+		 */
+		if (((char *)sfep + sizeof(*sfep)) >= endp)
+			return __this_address;
+
+		/* Don't allow names with known bad length. */
+		if (sfep->namelen == 0)
+			return __this_address;
+
+		/*
+		 * Check that the variable-length part of the structure is
+		 * within the data buffer.  The next entry starts after the
+		 * name component, so nextentry is an acceptable test.
+		 */
+		next_sfep = xfs_dir2_sf_nextentry(mp, sfp, sfep);
+		if (endp < (char *)next_sfep)
+			return __this_address;
+
+		/* Check that the offsets always increase. */
+		if (xfs_dir2_sf_get_offset(sfep) < offset)
+			return __this_address;
+
+		/* Check the inode number. */
+		ino = xfs_dir2_sf_get_ino(mp, sfp, sfep);
+		i8count += ino > XFS_DIR2_MAX_SHORT_INUM;
+		error = xfs_dir_ino_validate(mp, ino);
+		if (error)
+			return __this_address;
+
+		/* Check the file type. */
+		filetype = xfs_dir2_sf_get_ftype(mp, sfep);
+		if (filetype >= XFS_DIR3_FT_MAX)
+			return __this_address;
+
+		offset = xfs_dir2_sf_get_offset(sfep) +
+				xfs_dir2_data_entsize(mp, sfep->namelen);
+
+		sfep = next_sfep;
+	}
+	if (i8count != sfp->i8count)
+		return __this_address;
+	if ((void *)sfep != (void *)endp)
+		return __this_address;
+
+	/* Make sure this whole thing ought to be in local format. */
+	if (offset + (sfp->count + 2) * (uint)sizeof(xfs_dir2_leaf_entry_t) +
+	    (uint)sizeof(xfs_dir2_block_tail_t) > mp->m_dir_geo->blksize)
+		return __this_address;
+
+	return NULL;
+}
+
+/*
+ * Create a new (shortform) directory.
+ */
+int					/* error, always 0 */
+xfs_dir2_sf_create(
+	xfs_da_args_t	*args,		/* operation arguments */
+	xfs_ino_t	pino)		/* parent inode number */
+{
+	xfs_inode_t	*dp;		/* incore directory inode */
+	int		i8count;	/* parent inode is an 8-byte number */
+	xfs_dir2_sf_hdr_t *sfp;		/* shortform structure */
+	int		size;		/* directory size */
+
+	trace_xfs_dir2_sf_create(args);
+
+	dp = args->dp;
+
+	ASSERT(dp != NULL);
+	ASSERT(dp->i_disk_size == 0);
+	/*
+	 * If it's currently a zero-length extent file,
+	 * convert it to local format.
+	 */
+	if (dp->i_df.if_format == XFS_DINODE_FMT_EXTENTS) {
+		dp->i_df.if_format = XFS_DINODE_FMT_LOCAL;
+		xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE);
+	}
+	ASSERT(dp->i_df.if_format == XFS_DINODE_FMT_LOCAL);
+	ASSERT(dp->i_df.if_bytes == 0);
+	i8count = pino > XFS_DIR2_MAX_SHORT_INUM;
+	size = xfs_dir2_sf_hdr_size(i8count);
+	/*
+	 * Make a buffer for the data.
+	 */
+	xfs_idata_realloc(dp, size, XFS_DATA_FORK);
+	/*
+	 * Fill in the header,
+	 */
+	sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	sfp->i8count = i8count;
+	/*
+	 * Now can put in the inode number, since i8count is set.
+	 */
+	xfs_dir2_sf_put_parent_ino(sfp, pino);
+	sfp->count = 0;
+	dp->i_disk_size = size;
+	xfs_dir2_sf_check(args);
+	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_DDATA);
+	return 0;
+}
+
+/*
+ * Lookup an entry in a shortform directory.
+ * Returns EEXIST if found, ENOENT if not found.
+ */
+int						/* error */
+xfs_dir2_sf_lookup(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	int			i;		/* entry index */
+	xfs_dir2_sf_entry_t	*sfep;		/* shortform directory entry */
+	xfs_dir2_sf_hdr_t	*sfp;		/* shortform structure */
+	enum xfs_dacmp		cmp;		/* comparison result */
+	xfs_dir2_sf_entry_t	*ci_sfep;	/* case-insens. entry */
+
+	trace_xfs_dir2_sf_lookup(args);
+
+	xfs_dir2_sf_check(args);
+
+	ASSERT(dp->i_df.if_format == XFS_DINODE_FMT_LOCAL);
+	ASSERT(dp->i_disk_size >= offsetof(struct xfs_dir2_sf_hdr, parent));
+	ASSERT(dp->i_df.if_bytes == dp->i_disk_size);
+	ASSERT(dp->i_df.if_u1.if_data != NULL);
+	sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	ASSERT(dp->i_disk_size >= xfs_dir2_sf_hdr_size(sfp->i8count));
+	/*
+	 * Special case for .
+	 */
+	if (args->namelen == 1 && args->name[0] == '.') {
+		args->inumber = dp->i_ino;
+		args->cmpresult = XFS_CMP_EXACT;
+		args->filetype = XFS_DIR3_FT_DIR;
+		return -EEXIST;
+	}
+	/*
+	 * Special case for ..
+	 */
+	if (args->namelen == 2 &&
+	    args->name[0] == '.' && args->name[1] == '.') {
+		args->inumber = xfs_dir2_sf_get_parent_ino(sfp);
+		args->cmpresult = XFS_CMP_EXACT;
+		args->filetype = XFS_DIR3_FT_DIR;
+		return -EEXIST;
+	}
+	/*
+	 * Loop over all the entries trying to match ours.
+	 */
+	ci_sfep = NULL;
+	for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp); i < sfp->count;
+	     i++, sfep = xfs_dir2_sf_nextentry(mp, sfp, sfep)) {
+		/*
+		 * Compare name and if it's an exact match, return the inode
+		 * number. If it's the first case-insensitive match, store the
+		 * inode number and continue looking for an exact match.
+		 */
+		cmp = xfs_dir2_compname(args, sfep->name, sfep->namelen);
+		if (cmp != XFS_CMP_DIFFERENT && cmp != args->cmpresult) {
+			args->cmpresult = cmp;
+			args->inumber = xfs_dir2_sf_get_ino(mp, sfp, sfep);
+			args->filetype = xfs_dir2_sf_get_ftype(mp, sfep);
+			if (cmp == XFS_CMP_EXACT)
+				return -EEXIST;
+			ci_sfep = sfep;
+		}
+	}
+	ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
+	/*
+	 * Here, we can only be doing a lookup (not a rename or replace).
+	 * If a case-insensitive match was not found, return -ENOENT.
+	 */
+	if (!ci_sfep)
+		return -ENOENT;
+	/* otherwise process the CI match as required by the caller */
+	return xfs_dir_cilookup_result(args, ci_sfep->name, ci_sfep->namelen);
+}
+
+/*
+ * Remove an entry from a shortform directory.
+ */
+int						/* error */
+xfs_dir2_sf_removename(
+	xfs_da_args_t		*args)
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	int			byteoff;	/* offset of removed entry */
+	int			entsize;	/* this entry's size */
+	int			i;		/* shortform entry index */
+	int			newsize;	/* new inode size */
+	int			oldsize;	/* old inode size */
+	xfs_dir2_sf_entry_t	*sfep;		/* shortform directory entry */
+	xfs_dir2_sf_hdr_t	*sfp;		/* shortform structure */
+
+	trace_xfs_dir2_sf_removename(args);
+
+	ASSERT(dp->i_df.if_format == XFS_DINODE_FMT_LOCAL);
+	oldsize = (int)dp->i_disk_size;
+	ASSERT(oldsize >= offsetof(struct xfs_dir2_sf_hdr, parent));
+	ASSERT(dp->i_df.if_bytes == oldsize);
+	ASSERT(dp->i_df.if_u1.if_data != NULL);
+	sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	ASSERT(oldsize >= xfs_dir2_sf_hdr_size(sfp->i8count));
+	/*
+	 * Loop over the old directory entries.
+	 * Find the one we're deleting.
+	 */
+	for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp); i < sfp->count;
+	     i++, sfep = xfs_dir2_sf_nextentry(mp, sfp, sfep)) {
+		if (xfs_da_compname(args, sfep->name, sfep->namelen) ==
+								XFS_CMP_EXACT) {
+			ASSERT(xfs_dir2_sf_get_ino(mp, sfp, sfep) ==
+			       args->inumber);
+			break;
+		}
+	}
+	/*
+	 * Didn't find it.
+	 */
+	if (i == sfp->count)
+		return -ENOENT;
+	/*
+	 * Calculate sizes.
+	 */
+	byteoff = (int)((char *)sfep - (char *)sfp);
+	entsize = xfs_dir2_sf_entsize(mp, sfp, args->namelen);
+	newsize = oldsize - entsize;
+	/*
+	 * Copy the part if any after the removed entry, sliding it down.
+	 */
+	if (byteoff + entsize < oldsize)
+		memmove((char *)sfp + byteoff, (char *)sfp + byteoff + entsize,
+			oldsize - (byteoff + entsize));
+	/*
+	 * Fix up the header and file size.
+	 */
+	sfp->count--;
+	dp->i_disk_size = newsize;
+	/*
+	 * Reallocate, making it smaller.
+	 */
+	xfs_idata_realloc(dp, newsize - oldsize, XFS_DATA_FORK);
+	sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	/*
+	 * Are we changing inode number size?
+	 */
+	if (args->inumber > XFS_DIR2_MAX_SHORT_INUM) {
+		if (sfp->i8count == 1)
+			xfs_dir2_sf_toino4(args);
+		else
+			sfp->i8count--;
+	}
+	xfs_dir2_sf_check(args);
+	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_DDATA);
+	return 0;
+}
+
+/*
+ * Check whether the sf dir replace operation need more blocks.
+ */
+static bool
+xfs_dir2_sf_replace_needblock(
+	struct xfs_inode	*dp,
+	xfs_ino_t		inum)
+{
+	int			newsize;
+	struct xfs_dir2_sf_hdr	*sfp;
+
+	if (dp->i_df.if_format != XFS_DINODE_FMT_LOCAL)
+		return false;
+
+	sfp = (struct xfs_dir2_sf_hdr *)dp->i_df.if_u1.if_data;
+	newsize = dp->i_df.if_bytes + (sfp->count + 1) * XFS_INO64_DIFF;
+
+	return inum > XFS_DIR2_MAX_SHORT_INUM &&
+	       sfp->i8count == 0 && newsize > xfs_inode_data_fork_size(dp);
+}
+
+/*
+ * Replace the inode number of an entry in a shortform directory.
+ */
+int						/* error */
+xfs_dir2_sf_replace(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	int			i;		/* entry index */
+	xfs_ino_t		ino=0;		/* entry old inode number */
+	int			i8elevated;	/* sf_toino8 set i8count=1 */
+	xfs_dir2_sf_entry_t	*sfep;		/* shortform directory entry */
+	xfs_dir2_sf_hdr_t	*sfp;		/* shortform structure */
+
+	trace_xfs_dir2_sf_replace(args);
+
+	ASSERT(dp->i_df.if_format == XFS_DINODE_FMT_LOCAL);
+	ASSERT(dp->i_disk_size >= offsetof(struct xfs_dir2_sf_hdr, parent));
+	ASSERT(dp->i_df.if_bytes == dp->i_disk_size);
+	ASSERT(dp->i_df.if_u1.if_data != NULL);
+	sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	ASSERT(dp->i_disk_size >= xfs_dir2_sf_hdr_size(sfp->i8count));
+
+	/*
+	 * New inode number is large, and need to convert to 8-byte inodes.
+	 */
+	if (args->inumber > XFS_DIR2_MAX_SHORT_INUM && sfp->i8count == 0) {
+		int	error;			/* error return value */
+
+		/*
+		 * Won't fit as shortform, convert to block then do replace.
+		 */
+		if (xfs_dir2_sf_replace_needblock(dp, args->inumber)) {
+			error = xfs_dir2_sf_to_block(args);
+			if (error)
+				return error;
+			return xfs_dir2_block_replace(args);
+		}
+		/*
+		 * Still fits, convert to 8-byte now.
+		 */
+		xfs_dir2_sf_toino8(args);
+		i8elevated = 1;
+		sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	} else
+		i8elevated = 0;
+
+	ASSERT(args->namelen != 1 || args->name[0] != '.');
+	/*
+	 * Replace ..'s entry.
+	 */
+	if (args->namelen == 2 &&
+	    args->name[0] == '.' && args->name[1] == '.') {
+		ino = xfs_dir2_sf_get_parent_ino(sfp);
+		ASSERT(args->inumber != ino);
+		xfs_dir2_sf_put_parent_ino(sfp, args->inumber);
+	}
+	/*
+	 * Normal entry, look for the name.
+	 */
+	else {
+		for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp); i < sfp->count;
+		     i++, sfep = xfs_dir2_sf_nextentry(mp, sfp, sfep)) {
+			if (xfs_da_compname(args, sfep->name, sfep->namelen) ==
+								XFS_CMP_EXACT) {
+				ino = xfs_dir2_sf_get_ino(mp, sfp, sfep);
+				ASSERT(args->inumber != ino);
+				xfs_dir2_sf_put_ino(mp, sfp, sfep,
+						args->inumber);
+				xfs_dir2_sf_put_ftype(mp, sfep, args->filetype);
+				break;
+			}
+		}
+		/*
+		 * Didn't find it.
+		 */
+		if (i == sfp->count) {
+			ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
+			if (i8elevated)
+				xfs_dir2_sf_toino4(args);
+			return -ENOENT;
+		}
+	}
+	/*
+	 * See if the old number was large, the new number is small.
+	 */
+	if (ino > XFS_DIR2_MAX_SHORT_INUM &&
+	    args->inumber <= XFS_DIR2_MAX_SHORT_INUM) {
+		/*
+		 * And the old count was one, so need to convert to small.
+		 */
+		if (sfp->i8count == 1)
+			xfs_dir2_sf_toino4(args);
+		else
+			sfp->i8count--;
+	}
+	/*
+	 * See if the old number was small, the new number is large.
+	 */
+	if (ino <= XFS_DIR2_MAX_SHORT_INUM &&
+	    args->inumber > XFS_DIR2_MAX_SHORT_INUM) {
+		/*
+		 * add to the i8count unless we just converted to 8-byte
+		 * inodes (which does an implied i8count = 1)
+		 */
+		ASSERT(sfp->i8count != 0);
+		if (!i8elevated)
+			sfp->i8count++;
+	}
+	xfs_dir2_sf_check(args);
+	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_DDATA);
+	return 0;
+}
+
+/*
+ * Convert from 8-byte inode numbers to 4-byte inode numbers.
+ * The last 8-byte inode number is gone, but the count is still 1.
+ */
+static void
+xfs_dir2_sf_toino4(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	char			*buf;		/* old dir's buffer */
+	int			i;		/* entry index */
+	int			newsize;	/* new inode size */
+	xfs_dir2_sf_entry_t	*oldsfep;	/* old sf entry */
+	xfs_dir2_sf_hdr_t	*oldsfp;	/* old sf directory */
+	int			oldsize;	/* old inode size */
+	xfs_dir2_sf_entry_t	*sfep;		/* new sf entry */
+	xfs_dir2_sf_hdr_t	*sfp;		/* new sf directory */
+
+	trace_xfs_dir2_sf_toino4(args);
+
+	/*
+	 * Copy the old directory to the buffer.
+	 * Then nuke it from the inode, and add the new buffer to the inode.
+	 * Don't want xfs_idata_realloc copying the data here.
+	 */
+	oldsize = dp->i_df.if_bytes;
+	buf = kmem_alloc(oldsize, 0);
+	oldsfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	ASSERT(oldsfp->i8count == 1);
+	memcpy(buf, oldsfp, oldsize);
+	/*
+	 * Compute the new inode size.
+	 */
+	newsize = oldsize - (oldsfp->count + 1) * XFS_INO64_DIFF;
+	xfs_idata_realloc(dp, -oldsize, XFS_DATA_FORK);
+	xfs_idata_realloc(dp, newsize, XFS_DATA_FORK);
+	/*
+	 * Reset our pointers, the data has moved.
+	 */
+	oldsfp = (xfs_dir2_sf_hdr_t *)buf;
+	sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	/*
+	 * Fill in the new header.
+	 */
+	sfp->count = oldsfp->count;
+	sfp->i8count = 0;
+	xfs_dir2_sf_put_parent_ino(sfp, xfs_dir2_sf_get_parent_ino(oldsfp));
+	/*
+	 * Copy the entries field by field.
+	 */
+	for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp),
+		    oldsfep = xfs_dir2_sf_firstentry(oldsfp);
+	     i < sfp->count;
+	     i++, sfep = xfs_dir2_sf_nextentry(mp, sfp, sfep),
+		  oldsfep = xfs_dir2_sf_nextentry(mp, oldsfp, oldsfep)) {
+		sfep->namelen = oldsfep->namelen;
+		memcpy(sfep->offset, oldsfep->offset, sizeof(sfep->offset));
+		memcpy(sfep->name, oldsfep->name, sfep->namelen);
+		xfs_dir2_sf_put_ino(mp, sfp, sfep,
+				xfs_dir2_sf_get_ino(mp, oldsfp, oldsfep));
+		xfs_dir2_sf_put_ftype(mp, sfep,
+				xfs_dir2_sf_get_ftype(mp, oldsfep));
+	}
+	/*
+	 * Clean up the inode.
+	 */
+	kmem_free(buf);
+	dp->i_disk_size = newsize;
+	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_DDATA);
+}
+
+/*
+ * Convert existing entries from 4-byte inode numbers to 8-byte inode numbers.
+ * The new entry w/ an 8-byte inode number is not there yet; we leave with
+ * i8count set to 1, but no corresponding 8-byte entry.
+ */
+static void
+xfs_dir2_sf_toino8(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	struct xfs_inode	*dp = args->dp;
+	struct xfs_mount	*mp = dp->i_mount;
+	char			*buf;		/* old dir's buffer */
+	int			i;		/* entry index */
+	int			newsize;	/* new inode size */
+	xfs_dir2_sf_entry_t	*oldsfep;	/* old sf entry */
+	xfs_dir2_sf_hdr_t	*oldsfp;	/* old sf directory */
+	int			oldsize;	/* old inode size */
+	xfs_dir2_sf_entry_t	*sfep;		/* new sf entry */
+	xfs_dir2_sf_hdr_t	*sfp;		/* new sf directory */
+
+	trace_xfs_dir2_sf_toino8(args);
+
+	/*
+	 * Copy the old directory to the buffer.
+	 * Then nuke it from the inode, and add the new buffer to the inode.
+	 * Don't want xfs_idata_realloc copying the data here.
+	 */
+	oldsize = dp->i_df.if_bytes;
+	buf = kmem_alloc(oldsize, 0);
+	oldsfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	ASSERT(oldsfp->i8count == 0);
+	memcpy(buf, oldsfp, oldsize);
+	/*
+	 * Compute the new inode size (nb: entry count + 1 for parent)
+	 */
+	newsize = oldsize + (oldsfp->count + 1) * XFS_INO64_DIFF;
+	xfs_idata_realloc(dp, -oldsize, XFS_DATA_FORK);
+	xfs_idata_realloc(dp, newsize, XFS_DATA_FORK);
+	/*
+	 * Reset our pointers, the data has moved.
+	 */
+	oldsfp = (xfs_dir2_sf_hdr_t *)buf;
+	sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
+	/*
+	 * Fill in the new header.
+	 */
+	sfp->count = oldsfp->count;
+	sfp->i8count = 1;
+	xfs_dir2_sf_put_parent_ino(sfp, xfs_dir2_sf_get_parent_ino(oldsfp));
+	/*
+	 * Copy the entries field by field.
+	 */
+	for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp),
+		    oldsfep = xfs_dir2_sf_firstentry(oldsfp);
+	     i < sfp->count;
+	     i++, sfep = xfs_dir2_sf_nextentry(mp, sfp, sfep),
+		  oldsfep = xfs_dir2_sf_nextentry(mp, oldsfp, oldsfep)) {
+		sfep->namelen = oldsfep->namelen;
+		memcpy(sfep->offset, oldsfep->offset, sizeof(sfep->offset));
+		memcpy(sfep->name, oldsfep->name, sfep->namelen);
+		xfs_dir2_sf_put_ino(mp, sfp, sfep,
+				xfs_dir2_sf_get_ino(mp, oldsfp, oldsfep));
+		xfs_dir2_sf_put_ftype(mp, sfep,
+				xfs_dir2_sf_get_ftype(mp, oldsfep));
+	}
+	/*
+	 * Clean up the inode.
+	 */
+	kmem_free(buf);
+	dp->i_disk_size = newsize;
+	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_DDATA);
+}
diff --git a/fs/xfs/libxfs/xfs_dquot_buf.c b/fs/xfs/libxfs/xfs_dquot_buf.c
new file mode 100644
index 000000000..15a362e2f
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_dquot_buf.c
@@ -0,0 +1,325 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_inode.h"
+#include "xfs_quota.h"
+#include "xfs_trans.h"
+#include "xfs_qm.h"
+#include "xfs_error.h"
+
+int
+xfs_calc_dquots_per_chunk(
+	unsigned int		nbblks)	/* basic block units */
+{
+	ASSERT(nbblks > 0);
+	return BBTOB(nbblks) / sizeof(struct xfs_dqblk);
+}
+
+/*
+ * Do some primitive error checking on ondisk dquot data structures.
+ *
+ * The xfs_dqblk structure /contains/ the xfs_disk_dquot structure;
+ * we verify them separately because at some points we have only the
+ * smaller xfs_disk_dquot structure available.
+ */
+
+xfs_failaddr_t
+xfs_dquot_verify(
+	struct xfs_mount	*mp,
+	struct xfs_disk_dquot	*ddq,
+	xfs_dqid_t		id)	/* used only during quotacheck */
+{
+	__u8			ddq_type;
+
+	/*
+	 * We can encounter an uninitialized dquot buffer for 2 reasons:
+	 * 1. If we crash while deleting the quotainode(s), and those blks got
+	 *    used for user data. This is because we take the path of regular
+	 *    file deletion; however, the size field of quotainodes is never
+	 *    updated, so all the tricks that we play in itruncate_finish
+	 *    don't quite matter.
+	 *
+	 * 2. We don't play the quota buffers when there's a quotaoff logitem.
+	 *    But the allocation will be replayed so we'll end up with an
+	 *    uninitialized quota block.
+	 *
+	 * This is all fine; things are still consistent, and we haven't lost
+	 * any quota information. Just don't complain about bad dquot blks.
+	 */
+	if (ddq->d_magic != cpu_to_be16(XFS_DQUOT_MAGIC))
+		return __this_address;
+	if (ddq->d_version != XFS_DQUOT_VERSION)
+		return __this_address;
+
+	if (ddq->d_type & ~XFS_DQTYPE_ANY)
+		return __this_address;
+	ddq_type = ddq->d_type & XFS_DQTYPE_REC_MASK;
+	if (ddq_type != XFS_DQTYPE_USER &&
+	    ddq_type != XFS_DQTYPE_PROJ &&
+	    ddq_type != XFS_DQTYPE_GROUP)
+		return __this_address;
+
+	if ((ddq->d_type & XFS_DQTYPE_BIGTIME) &&
+	    !xfs_has_bigtime(mp))
+		return __this_address;
+
+	if ((ddq->d_type & XFS_DQTYPE_BIGTIME) && !ddq->d_id)
+		return __this_address;
+
+	if (id != -1 && id != be32_to_cpu(ddq->d_id))
+		return __this_address;
+
+	if (!ddq->d_id)
+		return NULL;
+
+	if (ddq->d_blk_softlimit &&
+	    be64_to_cpu(ddq->d_bcount) > be64_to_cpu(ddq->d_blk_softlimit) &&
+	    !ddq->d_btimer)
+		return __this_address;
+
+	if (ddq->d_ino_softlimit &&
+	    be64_to_cpu(ddq->d_icount) > be64_to_cpu(ddq->d_ino_softlimit) &&
+	    !ddq->d_itimer)
+		return __this_address;
+
+	if (ddq->d_rtb_softlimit &&
+	    be64_to_cpu(ddq->d_rtbcount) > be64_to_cpu(ddq->d_rtb_softlimit) &&
+	    !ddq->d_rtbtimer)
+		return __this_address;
+
+	return NULL;
+}
+
+xfs_failaddr_t
+xfs_dqblk_verify(
+	struct xfs_mount	*mp,
+	struct xfs_dqblk	*dqb,
+	xfs_dqid_t		id)	/* used only during quotacheck */
+{
+	if (xfs_has_crc(mp) &&
+	    !uuid_equal(&dqb->dd_uuid, &mp->m_sb.sb_meta_uuid))
+		return __this_address;
+
+	return xfs_dquot_verify(mp, &dqb->dd_diskdq, id);
+}
+
+/*
+ * Do some primitive error checking on ondisk dquot data structures.
+ */
+void
+xfs_dqblk_repair(
+	struct xfs_mount	*mp,
+	struct xfs_dqblk	*dqb,
+	xfs_dqid_t		id,
+	xfs_dqtype_t		type)
+{
+	/*
+	 * Typically, a repair is only requested by quotacheck.
+	 */
+	ASSERT(id != -1);
+	memset(dqb, 0, sizeof(struct xfs_dqblk));
+
+	dqb->dd_diskdq.d_magic = cpu_to_be16(XFS_DQUOT_MAGIC);
+	dqb->dd_diskdq.d_version = XFS_DQUOT_VERSION;
+	dqb->dd_diskdq.d_type = type;
+	dqb->dd_diskdq.d_id = cpu_to_be32(id);
+
+	if (xfs_has_crc(mp)) {
+		uuid_copy(&dqb->dd_uuid, &mp->m_sb.sb_meta_uuid);
+		xfs_update_cksum((char *)dqb, sizeof(struct xfs_dqblk),
+				 XFS_DQUOT_CRC_OFF);
+	}
+}
+
+STATIC bool
+xfs_dquot_buf_verify_crc(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp,
+	bool			readahead)
+{
+	struct xfs_dqblk	*d = (struct xfs_dqblk *)bp->b_addr;
+	int			ndquots;
+	int			i;
+
+	if (!xfs_has_crc(mp))
+		return true;
+
+	/*
+	 * if we are in log recovery, the quota subsystem has not been
+	 * initialised so we have no quotainfo structure. In that case, we need
+	 * to manually calculate the number of dquots in the buffer.
+	 */
+	if (mp->m_quotainfo)
+		ndquots = mp->m_quotainfo->qi_dqperchunk;
+	else
+		ndquots = xfs_calc_dquots_per_chunk(bp->b_length);
+
+	for (i = 0; i < ndquots; i++, d++) {
+		if (!xfs_verify_cksum((char *)d, sizeof(struct xfs_dqblk),
+				 XFS_DQUOT_CRC_OFF)) {
+			if (!readahead)
+				xfs_buf_verifier_error(bp, -EFSBADCRC, __func__,
+					d, sizeof(*d), __this_address);
+			return false;
+		}
+	}
+	return true;
+}
+
+STATIC xfs_failaddr_t
+xfs_dquot_buf_verify(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp,
+	bool			readahead)
+{
+	struct xfs_dqblk	*dqb = bp->b_addr;
+	xfs_failaddr_t		fa;
+	xfs_dqid_t		id = 0;
+	int			ndquots;
+	int			i;
+
+	/*
+	 * if we are in log recovery, the quota subsystem has not been
+	 * initialised so we have no quotainfo structure. In that case, we need
+	 * to manually calculate the number of dquots in the buffer.
+	 */
+	if (mp->m_quotainfo)
+		ndquots = mp->m_quotainfo->qi_dqperchunk;
+	else
+		ndquots = xfs_calc_dquots_per_chunk(bp->b_length);
+
+	/*
+	 * On the first read of the buffer, verify that each dquot is valid.
+	 * We don't know what the id of the dquot is supposed to be, just that
+	 * they should be increasing monotonically within the buffer. If the
+	 * first id is corrupt, then it will fail on the second dquot in the
+	 * buffer so corruptions could point to the wrong dquot in this case.
+	 */
+	for (i = 0; i < ndquots; i++) {
+		struct xfs_disk_dquot	*ddq;
+
+		ddq = &dqb[i].dd_diskdq;
+
+		if (i == 0)
+			id = be32_to_cpu(ddq->d_id);
+
+		fa = xfs_dqblk_verify(mp, &dqb[i], id + i);
+		if (fa) {
+			if (!readahead)
+				xfs_buf_verifier_error(bp, -EFSCORRUPTED,
+					__func__, &dqb[i],
+					sizeof(struct xfs_dqblk), fa);
+			return fa;
+		}
+	}
+
+	return NULL;
+}
+
+static xfs_failaddr_t
+xfs_dquot_buf_verify_struct(
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+
+	return xfs_dquot_buf_verify(mp, bp, false);
+}
+
+static void
+xfs_dquot_buf_read_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+
+	if (!xfs_dquot_buf_verify_crc(mp, bp, false))
+		return;
+	xfs_dquot_buf_verify(mp, bp, false);
+}
+
+/*
+ * readahead errors are silent and simply leave the buffer as !done so a real
+ * read will then be run with the xfs_dquot_buf_ops verifier. See
+ * xfs_inode_buf_verify() for why we use EIO and ~XBF_DONE here rather than
+ * reporting the failure.
+ */
+static void
+xfs_dquot_buf_readahead_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+
+	if (!xfs_dquot_buf_verify_crc(mp, bp, true) ||
+	    xfs_dquot_buf_verify(mp, bp, true) != NULL) {
+		xfs_buf_ioerror(bp, -EIO);
+		bp->b_flags &= ~XBF_DONE;
+	}
+}
+
+/*
+ * we don't calculate the CRC here as that is done when the dquot is flushed to
+ * the buffer after the update is done. This ensures that the dquot in the
+ * buffer always has an up-to-date CRC value.
+ */
+static void
+xfs_dquot_buf_write_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+
+	xfs_dquot_buf_verify(mp, bp, false);
+}
+
+const struct xfs_buf_ops xfs_dquot_buf_ops = {
+	.name = "xfs_dquot",
+	.magic16 = { cpu_to_be16(XFS_DQUOT_MAGIC),
+		     cpu_to_be16(XFS_DQUOT_MAGIC) },
+	.verify_read = xfs_dquot_buf_read_verify,
+	.verify_write = xfs_dquot_buf_write_verify,
+	.verify_struct = xfs_dquot_buf_verify_struct,
+};
+
+const struct xfs_buf_ops xfs_dquot_buf_ra_ops = {
+	.name = "xfs_dquot_ra",
+	.magic16 = { cpu_to_be16(XFS_DQUOT_MAGIC),
+		     cpu_to_be16(XFS_DQUOT_MAGIC) },
+	.verify_read = xfs_dquot_buf_readahead_verify,
+	.verify_write = xfs_dquot_buf_write_verify,
+};
+
+/* Convert an on-disk timer value into an incore timer value. */
+time64_t
+xfs_dquot_from_disk_ts(
+	struct xfs_disk_dquot	*ddq,
+	__be32			dtimer)
+{
+	uint32_t		t = be32_to_cpu(dtimer);
+
+	if (t != 0 && (ddq->d_type & XFS_DQTYPE_BIGTIME))
+		return xfs_dq_bigtime_to_unix(t);
+
+	return t;
+}
+
+/* Convert an incore timer value into an on-disk timer value. */
+__be32
+xfs_dquot_to_disk_ts(
+	struct xfs_dquot	*dqp,
+	time64_t		timer)
+{
+	uint32_t		t = timer;
+
+	if (timer != 0 && (dqp->q_type & XFS_DQTYPE_BIGTIME))
+		t = xfs_dq_unix_to_bigtime(timer);
+
+	return cpu_to_be32(t);
+}
diff --git a/fs/xfs/libxfs/xfs_errortag.h b/fs/xfs/libxfs/xfs_errortag.h
new file mode 100644
index 000000000..536290816
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_errortag.h
@@ -0,0 +1,113 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * Copyright (C) 2017 Oracle.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_ERRORTAG_H_
+#define __XFS_ERRORTAG_H_
+
+/*
+ * error injection tags - the labels can be anything you want
+ * but each tag should have its own unique number
+ */
+
+#define XFS_ERRTAG_NOERROR				0
+#define XFS_ERRTAG_IFLUSH_1				1
+#define XFS_ERRTAG_IFLUSH_2				2
+#define XFS_ERRTAG_IFLUSH_3				3
+#define XFS_ERRTAG_IFLUSH_4				4
+#define XFS_ERRTAG_IFLUSH_5				5
+#define XFS_ERRTAG_IFLUSH_6				6
+#define XFS_ERRTAG_DA_READ_BUF				7
+#define XFS_ERRTAG_BTREE_CHECK_LBLOCK			8
+#define XFS_ERRTAG_BTREE_CHECK_SBLOCK			9
+#define XFS_ERRTAG_ALLOC_READ_AGF			10
+#define XFS_ERRTAG_IALLOC_READ_AGI			11
+#define XFS_ERRTAG_ITOBP_INOTOBP			12
+#define XFS_ERRTAG_IUNLINK				13
+#define XFS_ERRTAG_IUNLINK_REMOVE			14
+#define XFS_ERRTAG_DIR_INO_VALIDATE			15
+#define XFS_ERRTAG_BULKSTAT_READ_CHUNK			16
+#define XFS_ERRTAG_IODONE_IOERR				17
+#define XFS_ERRTAG_STRATREAD_IOERR			18
+#define XFS_ERRTAG_STRATCMPL_IOERR			19
+#define XFS_ERRTAG_DIOWRITE_IOERR			20
+#define XFS_ERRTAG_BMAPIFORMAT				21
+#define XFS_ERRTAG_FREE_EXTENT				22
+#define XFS_ERRTAG_RMAP_FINISH_ONE			23
+#define XFS_ERRTAG_REFCOUNT_CONTINUE_UPDATE		24
+#define XFS_ERRTAG_REFCOUNT_FINISH_ONE			25
+#define XFS_ERRTAG_BMAP_FINISH_ONE			26
+#define XFS_ERRTAG_AG_RESV_CRITICAL			27
+/*
+ * DEBUG mode instrumentation to test and/or trigger delayed allocation
+ * block killing in the event of failed writes. When enabled, all
+ * buffered writes are silenty dropped and handled as if they failed.
+ * All delalloc blocks in the range of the write (including pre-existing
+ * delalloc blocks!) are tossed as part of the write failure error
+ * handling sequence.
+ */
+#define XFS_ERRTAG_DROP_WRITES				28
+#define XFS_ERRTAG_LOG_BAD_CRC				29
+#define XFS_ERRTAG_LOG_ITEM_PIN				30
+#define XFS_ERRTAG_BUF_LRU_REF				31
+#define XFS_ERRTAG_FORCE_SCRUB_REPAIR			32
+#define XFS_ERRTAG_FORCE_SUMMARY_RECALC			33
+#define XFS_ERRTAG_IUNLINK_FALLBACK			34
+#define XFS_ERRTAG_BUF_IOERROR				35
+#define XFS_ERRTAG_REDUCE_MAX_IEXTENTS			36
+#define XFS_ERRTAG_BMAP_ALLOC_MINLEN_EXTENT		37
+#define XFS_ERRTAG_AG_RESV_FAIL				38
+#define XFS_ERRTAG_LARP					39
+#define XFS_ERRTAG_DA_LEAF_SPLIT			40
+#define XFS_ERRTAG_ATTR_LEAF_TO_NODE			41
+#define XFS_ERRTAG_MAX					42
+
+/*
+ * Random factors for above tags, 1 means always, 2 means 1/2 time, etc.
+ */
+#define XFS_RANDOM_DEFAULT				100
+#define XFS_RANDOM_IFLUSH_1				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IFLUSH_2				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IFLUSH_3				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IFLUSH_4				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IFLUSH_5				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IFLUSH_6				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_DA_READ_BUF				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_BTREE_CHECK_LBLOCK			(XFS_RANDOM_DEFAULT/4)
+#define XFS_RANDOM_BTREE_CHECK_SBLOCK			XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_ALLOC_READ_AGF			XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IALLOC_READ_AGI			XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_ITOBP_INOTOBP			XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IUNLINK				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IUNLINK_REMOVE			XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_DIR_INO_VALIDATE			XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_BULKSTAT_READ_CHUNK			XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IODONE_IOERR				(XFS_RANDOM_DEFAULT/10)
+#define XFS_RANDOM_STRATREAD_IOERR			(XFS_RANDOM_DEFAULT/10)
+#define XFS_RANDOM_STRATCMPL_IOERR			(XFS_RANDOM_DEFAULT/10)
+#define XFS_RANDOM_DIOWRITE_IOERR			(XFS_RANDOM_DEFAULT/10)
+#define XFS_RANDOM_BMAPIFORMAT				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_FREE_EXTENT				1
+#define XFS_RANDOM_RMAP_FINISH_ONE			1
+#define XFS_RANDOM_REFCOUNT_CONTINUE_UPDATE		1
+#define XFS_RANDOM_REFCOUNT_FINISH_ONE			1
+#define XFS_RANDOM_BMAP_FINISH_ONE			1
+#define XFS_RANDOM_AG_RESV_CRITICAL			4
+#define XFS_RANDOM_DROP_WRITES				1
+#define XFS_RANDOM_LOG_BAD_CRC				1
+#define XFS_RANDOM_LOG_ITEM_PIN				1
+#define XFS_RANDOM_BUF_LRU_REF				2
+#define XFS_RANDOM_FORCE_SCRUB_REPAIR			1
+#define XFS_RANDOM_FORCE_SUMMARY_RECALC			1
+#define XFS_RANDOM_IUNLINK_FALLBACK			(XFS_RANDOM_DEFAULT/10)
+#define XFS_RANDOM_BUF_IOERROR				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_REDUCE_MAX_IEXTENTS			1
+#define XFS_RANDOM_BMAP_ALLOC_MINLEN_EXTENT		1
+#define XFS_RANDOM_AG_RESV_FAIL				1
+#define XFS_RANDOM_LARP					1
+#define XFS_RANDOM_DA_LEAF_SPLIT			1
+#define XFS_RANDOM_ATTR_LEAF_TO_NODE			1
+
+#endif /* __XFS_ERRORTAG_H_ */
diff --git a/fs/xfs/libxfs/xfs_format.h b/fs/xfs/libxfs/xfs_format.h
new file mode 100644
index 000000000..371dc0723
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_format.h
@@ -0,0 +1,1829 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_FORMAT_H__
+#define __XFS_FORMAT_H__
+
+/*
+ * XFS On Disk Format Definitions
+ *
+ * This header file defines all the on-disk format definitions for
+ * general XFS objects. Directory and attribute related objects are defined in
+ * xfs_da_format.h, which log and log item formats are defined in
+ * xfs_log_format.h. Everything else goes here.
+ */
+
+struct xfs_mount;
+struct xfs_trans;
+struct xfs_inode;
+struct xfs_buf;
+struct xfs_ifork;
+
+/*
+ * Super block
+ * Fits into a sector-sized buffer at address 0 of each allocation group.
+ * Only the first of these is ever updated except during growfs.
+ */
+#define	XFS_SB_MAGIC		0x58465342	/* 'XFSB' */
+#define	XFS_SB_VERSION_1	1		/* 5.3, 6.0.1, 6.1 */
+#define	XFS_SB_VERSION_2	2		/* 6.2 - attributes */
+#define	XFS_SB_VERSION_3	3		/* 6.2 - new inode version */
+#define	XFS_SB_VERSION_4	4		/* 6.2+ - bitmask version */
+#define	XFS_SB_VERSION_5	5		/* CRC enabled filesystem */
+#define	XFS_SB_VERSION_NUMBITS		0x000f
+#define	XFS_SB_VERSION_ALLFBITS		0xfff0
+#define	XFS_SB_VERSION_ATTRBIT		0x0010
+#define	XFS_SB_VERSION_NLINKBIT		0x0020
+#define	XFS_SB_VERSION_QUOTABIT		0x0040
+#define	XFS_SB_VERSION_ALIGNBIT		0x0080
+#define	XFS_SB_VERSION_DALIGNBIT	0x0100
+#define	XFS_SB_VERSION_SHAREDBIT	0x0200
+#define XFS_SB_VERSION_LOGV2BIT		0x0400
+#define XFS_SB_VERSION_SECTORBIT	0x0800
+#define	XFS_SB_VERSION_EXTFLGBIT	0x1000
+#define	XFS_SB_VERSION_DIRV2BIT		0x2000
+#define	XFS_SB_VERSION_BORGBIT		0x4000	/* ASCII only case-insens. */
+#define	XFS_SB_VERSION_MOREBITSBIT	0x8000
+
+/*
+ * The size of a single extended attribute on disk is limited by
+ * the size of index values within the attribute entries themselves.
+ * These are be16 fields, so we can only support attribute data
+ * sizes up to 2^16 bytes in length.
+ */
+#define XFS_XATTR_SIZE_MAX (1 << 16)
+
+/*
+ * Supported feature bit list is just all bits in the versionnum field because
+ * we've used them all up and understand them all. Except, of course, for the
+ * shared superblock bit, which nobody knows what it does and so is unsupported.
+ */
+#define	XFS_SB_VERSION_OKBITS		\
+	((XFS_SB_VERSION_NUMBITS | XFS_SB_VERSION_ALLFBITS) & \
+		~XFS_SB_VERSION_SHAREDBIT)
+
+/*
+ * There are two words to hold XFS "feature" bits: the original
+ * word, sb_versionnum, and sb_features2.  Whenever a bit is set in
+ * sb_features2, the feature bit XFS_SB_VERSION_MOREBITSBIT must be set.
+ *
+ * These defines represent bits in sb_features2.
+ */
+#define XFS_SB_VERSION2_RESERVED1BIT	0x00000001
+#define XFS_SB_VERSION2_LAZYSBCOUNTBIT	0x00000002	/* Superblk counters */
+#define XFS_SB_VERSION2_RESERVED4BIT	0x00000004
+#define XFS_SB_VERSION2_ATTR2BIT	0x00000008	/* Inline attr rework */
+#define XFS_SB_VERSION2_PARENTBIT	0x00000010	/* parent pointers */
+#define XFS_SB_VERSION2_PROJID32BIT	0x00000080	/* 32 bit project id */
+#define XFS_SB_VERSION2_CRCBIT		0x00000100	/* metadata CRCs */
+#define XFS_SB_VERSION2_FTYPE		0x00000200	/* inode type in dir */
+
+#define	XFS_SB_VERSION2_OKBITS		\
+	(XFS_SB_VERSION2_LAZYSBCOUNTBIT	| \
+	 XFS_SB_VERSION2_ATTR2BIT	| \
+	 XFS_SB_VERSION2_PROJID32BIT	| \
+	 XFS_SB_VERSION2_FTYPE)
+
+/* Maximum size of the xfs filesystem label, no terminating NULL */
+#define XFSLABEL_MAX			12
+
+/*
+ * Superblock - in core version.  Must match the ondisk version below.
+ * Must be padded to 64 bit alignment.
+ */
+typedef struct xfs_sb {
+	uint32_t	sb_magicnum;	/* magic number == XFS_SB_MAGIC */
+	uint32_t	sb_blocksize;	/* logical block size, bytes */
+	xfs_rfsblock_t	sb_dblocks;	/* number of data blocks */
+	xfs_rfsblock_t	sb_rblocks;	/* number of realtime blocks */
+	xfs_rtblock_t	sb_rextents;	/* number of realtime extents */
+	uuid_t		sb_uuid;	/* user-visible file system unique id */
+	xfs_fsblock_t	sb_logstart;	/* starting block of log if internal */
+	xfs_ino_t	sb_rootino;	/* root inode number */
+	xfs_ino_t	sb_rbmino;	/* bitmap inode for realtime extents */
+	xfs_ino_t	sb_rsumino;	/* summary inode for rt bitmap */
+	xfs_agblock_t	sb_rextsize;	/* realtime extent size, blocks */
+	xfs_agblock_t	sb_agblocks;	/* size of an allocation group */
+	xfs_agnumber_t	sb_agcount;	/* number of allocation groups */
+	xfs_extlen_t	sb_rbmblocks;	/* number of rt bitmap blocks */
+	xfs_extlen_t	sb_logblocks;	/* number of log blocks */
+	uint16_t	sb_versionnum;	/* header version == XFS_SB_VERSION */
+	uint16_t	sb_sectsize;	/* volume sector size, bytes */
+	uint16_t	sb_inodesize;	/* inode size, bytes */
+	uint16_t	sb_inopblock;	/* inodes per block */
+	char		sb_fname[XFSLABEL_MAX]; /* file system name */
+	uint8_t		sb_blocklog;	/* log2 of sb_blocksize */
+	uint8_t		sb_sectlog;	/* log2 of sb_sectsize */
+	uint8_t		sb_inodelog;	/* log2 of sb_inodesize */
+	uint8_t		sb_inopblog;	/* log2 of sb_inopblock */
+	uint8_t		sb_agblklog;	/* log2 of sb_agblocks (rounded up) */
+	uint8_t		sb_rextslog;	/* log2 of sb_rextents */
+	uint8_t		sb_inprogress;	/* mkfs is in progress, don't mount */
+	uint8_t		sb_imax_pct;	/* max % of fs for inode space */
+					/* statistics */
+	/*
+	 * These fields must remain contiguous.  If you really
+	 * want to change their layout, make sure you fix the
+	 * code in xfs_trans_apply_sb_deltas().
+	 */
+	uint64_t	sb_icount;	/* allocated inodes */
+	uint64_t	sb_ifree;	/* free inodes */
+	uint64_t	sb_fdblocks;	/* free data blocks */
+	uint64_t	sb_frextents;	/* free realtime extents */
+	/*
+	 * End contiguous fields.
+	 */
+	xfs_ino_t	sb_uquotino;	/* user quota inode */
+	xfs_ino_t	sb_gquotino;	/* group quota inode */
+	uint16_t	sb_qflags;	/* quota flags */
+	uint8_t		sb_flags;	/* misc. flags */
+	uint8_t		sb_shared_vn;	/* shared version number */
+	xfs_extlen_t	sb_inoalignmt;	/* inode chunk alignment, fsblocks */
+	uint32_t	sb_unit;	/* stripe or raid unit */
+	uint32_t	sb_width;	/* stripe or raid width */
+	uint8_t		sb_dirblklog;	/* log2 of dir block size (fsbs) */
+	uint8_t		sb_logsectlog;	/* log2 of the log sector size */
+	uint16_t	sb_logsectsize;	/* sector size for the log, bytes */
+	uint32_t	sb_logsunit;	/* stripe unit size for the log */
+	uint32_t	sb_features2;	/* additional feature bits */
+
+	/*
+	 * bad features2 field as a result of failing to pad the sb structure to
+	 * 64 bits. Some machines will be using this field for features2 bits.
+	 * Easiest just to mark it bad and not use it for anything else.
+	 *
+	 * This is not kept up to date in memory; it is always overwritten by
+	 * the value in sb_features2 when formatting the incore superblock to
+	 * the disk buffer.
+	 */
+	uint32_t	sb_bad_features2;
+
+	/* version 5 superblock fields start here */
+
+	/* feature masks */
+	uint32_t	sb_features_compat;
+	uint32_t	sb_features_ro_compat;
+	uint32_t	sb_features_incompat;
+	uint32_t	sb_features_log_incompat;
+
+	uint32_t	sb_crc;		/* superblock crc */
+	xfs_extlen_t	sb_spino_align;	/* sparse inode chunk alignment */
+
+	xfs_ino_t	sb_pquotino;	/* project quota inode */
+	xfs_lsn_t	sb_lsn;		/* last write sequence */
+	uuid_t		sb_meta_uuid;	/* metadata file system unique id */
+
+	/* must be padded to 64 bit alignment */
+} xfs_sb_t;
+
+#define XFS_SB_CRC_OFF		offsetof(struct xfs_sb, sb_crc)
+
+/*
+ * Superblock - on disk version.  Must match the in core version above.
+ * Must be padded to 64 bit alignment.
+ */
+struct xfs_dsb {
+	__be32		sb_magicnum;	/* magic number == XFS_SB_MAGIC */
+	__be32		sb_blocksize;	/* logical block size, bytes */
+	__be64		sb_dblocks;	/* number of data blocks */
+	__be64		sb_rblocks;	/* number of realtime blocks */
+	__be64		sb_rextents;	/* number of realtime extents */
+	uuid_t		sb_uuid;	/* user-visible file system unique id */
+	__be64		sb_logstart;	/* starting block of log if internal */
+	__be64		sb_rootino;	/* root inode number */
+	__be64		sb_rbmino;	/* bitmap inode for realtime extents */
+	__be64		sb_rsumino;	/* summary inode for rt bitmap */
+	__be32		sb_rextsize;	/* realtime extent size, blocks */
+	__be32		sb_agblocks;	/* size of an allocation group */
+	__be32		sb_agcount;	/* number of allocation groups */
+	__be32		sb_rbmblocks;	/* number of rt bitmap blocks */
+	__be32		sb_logblocks;	/* number of log blocks */
+	__be16		sb_versionnum;	/* header version == XFS_SB_VERSION */
+	__be16		sb_sectsize;	/* volume sector size, bytes */
+	__be16		sb_inodesize;	/* inode size, bytes */
+	__be16		sb_inopblock;	/* inodes per block */
+	char		sb_fname[XFSLABEL_MAX]; /* file system name */
+	__u8		sb_blocklog;	/* log2 of sb_blocksize */
+	__u8		sb_sectlog;	/* log2 of sb_sectsize */
+	__u8		sb_inodelog;	/* log2 of sb_inodesize */
+	__u8		sb_inopblog;	/* log2 of sb_inopblock */
+	__u8		sb_agblklog;	/* log2 of sb_agblocks (rounded up) */
+	__u8		sb_rextslog;	/* log2 of sb_rextents */
+	__u8		sb_inprogress;	/* mkfs is in progress, don't mount */
+	__u8		sb_imax_pct;	/* max % of fs for inode space */
+					/* statistics */
+	/*
+	 * These fields must remain contiguous.  If you really
+	 * want to change their layout, make sure you fix the
+	 * code in xfs_trans_apply_sb_deltas().
+	 */
+	__be64		sb_icount;	/* allocated inodes */
+	__be64		sb_ifree;	/* free inodes */
+	__be64		sb_fdblocks;	/* free data blocks */
+	__be64		sb_frextents;	/* free realtime extents */
+	/*
+	 * End contiguous fields.
+	 */
+	__be64		sb_uquotino;	/* user quota inode */
+	__be64		sb_gquotino;	/* group quota inode */
+	__be16		sb_qflags;	/* quota flags */
+	__u8		sb_flags;	/* misc. flags */
+	__u8		sb_shared_vn;	/* shared version number */
+	__be32		sb_inoalignmt;	/* inode chunk alignment, fsblocks */
+	__be32		sb_unit;	/* stripe or raid unit */
+	__be32		sb_width;	/* stripe or raid width */
+	__u8		sb_dirblklog;	/* log2 of dir block size (fsbs) */
+	__u8		sb_logsectlog;	/* log2 of the log sector size */
+	__be16		sb_logsectsize;	/* sector size for the log, bytes */
+	__be32		sb_logsunit;	/* stripe unit size for the log */
+	__be32		sb_features2;	/* additional feature bits */
+	/*
+	 * bad features2 field as a result of failing to pad the sb
+	 * structure to 64 bits. Some machines will be using this field
+	 * for features2 bits. Easiest just to mark it bad and not use
+	 * it for anything else.
+	 */
+	__be32		sb_bad_features2;
+
+	/* version 5 superblock fields start here */
+
+	/* feature masks */
+	__be32		sb_features_compat;
+	__be32		sb_features_ro_compat;
+	__be32		sb_features_incompat;
+	__be32		sb_features_log_incompat;
+
+	__le32		sb_crc;		/* superblock crc */
+	__be32		sb_spino_align;	/* sparse inode chunk alignment */
+
+	__be64		sb_pquotino;	/* project quota inode */
+	__be64		sb_lsn;		/* last write sequence */
+	uuid_t		sb_meta_uuid;	/* metadata file system unique id */
+
+	/* must be padded to 64 bit alignment */
+};
+
+/*
+ * Misc. Flags - warning - these will be cleared by xfs_repair unless
+ * a feature bit is set when the flag is used.
+ */
+#define XFS_SBF_NOFLAGS		0x00	/* no flags set */
+#define XFS_SBF_READONLY	0x01	/* only read-only mounts allowed */
+
+/*
+ * define max. shared version we can interoperate with
+ */
+#define XFS_SB_MAX_SHARED_VN	0
+
+#define	XFS_SB_VERSION_NUM(sbp)	((sbp)->sb_versionnum & XFS_SB_VERSION_NUMBITS)
+
+static inline bool xfs_sb_is_v5(struct xfs_sb *sbp)
+{
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5;
+}
+
+/*
+ * Detect a mismatched features2 field.  Older kernels read/wrote
+ * this into the wrong slot, so to be safe we keep them in sync.
+ */
+static inline bool xfs_sb_has_mismatched_features2(struct xfs_sb *sbp)
+{
+	return sbp->sb_bad_features2 != sbp->sb_features2;
+}
+
+static inline bool xfs_sb_version_hasmorebits(struct xfs_sb *sbp)
+{
+	return xfs_sb_is_v5(sbp) ||
+	       (sbp->sb_versionnum & XFS_SB_VERSION_MOREBITSBIT);
+}
+
+static inline void xfs_sb_version_addattr(struct xfs_sb *sbp)
+{
+	sbp->sb_versionnum |= XFS_SB_VERSION_ATTRBIT;
+}
+
+static inline void xfs_sb_version_addquota(struct xfs_sb *sbp)
+{
+	sbp->sb_versionnum |= XFS_SB_VERSION_QUOTABIT;
+}
+
+static inline void xfs_sb_version_addattr2(struct xfs_sb *sbp)
+{
+	sbp->sb_versionnum |= XFS_SB_VERSION_MOREBITSBIT;
+	sbp->sb_features2 |= XFS_SB_VERSION2_ATTR2BIT;
+}
+
+static inline void xfs_sb_version_addprojid32(struct xfs_sb *sbp)
+{
+	sbp->sb_versionnum |= XFS_SB_VERSION_MOREBITSBIT;
+	sbp->sb_features2 |= XFS_SB_VERSION2_PROJID32BIT;
+}
+
+/*
+ * Extended v5 superblock feature masks. These are to be used for new v5
+ * superblock features only.
+ *
+ * Compat features are new features that old kernels will not notice or affect
+ * and so can mount read-write without issues.
+ *
+ * RO-Compat (read only) are features that old kernels can read but will break
+ * if they write. Hence only read-only mounts of such filesystems are allowed on
+ * kernels that don't support the feature bit.
+ *
+ * InCompat features are features which old kernels will not understand and so
+ * must not mount.
+ *
+ * Log-InCompat features are for changes to log formats or new transactions that
+ * can't be replayed on older kernels. The fields are set when the filesystem is
+ * mounted, and a clean unmount clears the fields.
+ */
+#define XFS_SB_FEAT_COMPAT_ALL 0
+#define XFS_SB_FEAT_COMPAT_UNKNOWN	~XFS_SB_FEAT_COMPAT_ALL
+static inline bool
+xfs_sb_has_compat_feature(
+	struct xfs_sb	*sbp,
+	uint32_t	feature)
+{
+	return (sbp->sb_features_compat & feature) != 0;
+}
+
+#define XFS_SB_FEAT_RO_COMPAT_FINOBT   (1 << 0)		/* free inode btree */
+#define XFS_SB_FEAT_RO_COMPAT_RMAPBT   (1 << 1)		/* reverse map btree */
+#define XFS_SB_FEAT_RO_COMPAT_REFLINK  (1 << 2)		/* reflinked files */
+#define XFS_SB_FEAT_RO_COMPAT_INOBTCNT (1 << 3)		/* inobt block counts */
+#define XFS_SB_FEAT_RO_COMPAT_ALL \
+		(XFS_SB_FEAT_RO_COMPAT_FINOBT | \
+		 XFS_SB_FEAT_RO_COMPAT_RMAPBT | \
+		 XFS_SB_FEAT_RO_COMPAT_REFLINK| \
+		 XFS_SB_FEAT_RO_COMPAT_INOBTCNT)
+#define XFS_SB_FEAT_RO_COMPAT_UNKNOWN	~XFS_SB_FEAT_RO_COMPAT_ALL
+static inline bool
+xfs_sb_has_ro_compat_feature(
+	struct xfs_sb	*sbp,
+	uint32_t	feature)
+{
+	return (sbp->sb_features_ro_compat & feature) != 0;
+}
+
+#define XFS_SB_FEAT_INCOMPAT_FTYPE	(1 << 0)	/* filetype in dirent */
+#define XFS_SB_FEAT_INCOMPAT_SPINODES	(1 << 1)	/* sparse inode chunks */
+#define XFS_SB_FEAT_INCOMPAT_META_UUID	(1 << 2)	/* metadata UUID */
+#define XFS_SB_FEAT_INCOMPAT_BIGTIME	(1 << 3)	/* large timestamps */
+#define XFS_SB_FEAT_INCOMPAT_NEEDSREPAIR (1 << 4)	/* needs xfs_repair */
+#define XFS_SB_FEAT_INCOMPAT_NREXT64	(1 << 5)	/* large extent counters */
+#define XFS_SB_FEAT_INCOMPAT_ALL \
+		(XFS_SB_FEAT_INCOMPAT_FTYPE|	\
+		 XFS_SB_FEAT_INCOMPAT_SPINODES|	\
+		 XFS_SB_FEAT_INCOMPAT_META_UUID| \
+		 XFS_SB_FEAT_INCOMPAT_BIGTIME| \
+		 XFS_SB_FEAT_INCOMPAT_NEEDSREPAIR| \
+		 XFS_SB_FEAT_INCOMPAT_NREXT64)
+
+#define XFS_SB_FEAT_INCOMPAT_UNKNOWN	~XFS_SB_FEAT_INCOMPAT_ALL
+static inline bool
+xfs_sb_has_incompat_feature(
+	struct xfs_sb	*sbp,
+	uint32_t	feature)
+{
+	return (sbp->sb_features_incompat & feature) != 0;
+}
+
+#define XFS_SB_FEAT_INCOMPAT_LOG_XATTRS   (1 << 0)	/* Delayed Attributes */
+#define XFS_SB_FEAT_INCOMPAT_LOG_ALL \
+	(XFS_SB_FEAT_INCOMPAT_LOG_XATTRS)
+#define XFS_SB_FEAT_INCOMPAT_LOG_UNKNOWN	~XFS_SB_FEAT_INCOMPAT_LOG_ALL
+static inline bool
+xfs_sb_has_incompat_log_feature(
+	struct xfs_sb	*sbp,
+	uint32_t	feature)
+{
+	return (sbp->sb_features_log_incompat & feature) != 0;
+}
+
+static inline void
+xfs_sb_remove_incompat_log_features(
+	struct xfs_sb	*sbp)
+{
+	sbp->sb_features_log_incompat &= ~XFS_SB_FEAT_INCOMPAT_LOG_ALL;
+}
+
+static inline void
+xfs_sb_add_incompat_log_features(
+	struct xfs_sb	*sbp,
+	unsigned int	features)
+{
+	sbp->sb_features_log_incompat |= features;
+}
+
+static inline bool xfs_sb_version_haslogxattrs(struct xfs_sb *sbp)
+{
+	return xfs_sb_is_v5(sbp) && (sbp->sb_features_log_incompat &
+		 XFS_SB_FEAT_INCOMPAT_LOG_XATTRS);
+}
+
+static inline bool
+xfs_is_quota_inode(struct xfs_sb *sbp, xfs_ino_t ino)
+{
+	return (ino == sbp->sb_uquotino ||
+		ino == sbp->sb_gquotino ||
+		ino == sbp->sb_pquotino);
+}
+
+#define XFS_SB_DADDR		((xfs_daddr_t)0) /* daddr in filesystem/ag */
+#define	XFS_SB_BLOCK(mp)	XFS_HDR_BLOCK(mp, XFS_SB_DADDR)
+
+#define	XFS_HDR_BLOCK(mp,d)	((xfs_agblock_t)XFS_BB_TO_FSBT(mp,d))
+#define	XFS_DADDR_TO_FSB(mp,d)	XFS_AGB_TO_FSB(mp, \
+			xfs_daddr_to_agno(mp,d), xfs_daddr_to_agbno(mp,d))
+#define	XFS_FSB_TO_DADDR(mp,fsbno)	XFS_AGB_TO_DADDR(mp, \
+			XFS_FSB_TO_AGNO(mp,fsbno), XFS_FSB_TO_AGBNO(mp,fsbno))
+
+/*
+ * File system sector to basic block conversions.
+ */
+#define XFS_FSS_TO_BB(mp,sec)	((sec) << (mp)->m_sectbb_log)
+
+/*
+ * File system block to basic block conversions.
+ */
+#define	XFS_FSB_TO_BB(mp,fsbno)	((fsbno) << (mp)->m_blkbb_log)
+#define	XFS_BB_TO_FSB(mp,bb)	\
+	(((bb) + (XFS_FSB_TO_BB(mp,1) - 1)) >> (mp)->m_blkbb_log)
+#define	XFS_BB_TO_FSBT(mp,bb)	((bb) >> (mp)->m_blkbb_log)
+
+/*
+ * File system block to byte conversions.
+ */
+#define XFS_FSB_TO_B(mp,fsbno)	((xfs_fsize_t)(fsbno) << (mp)->m_sb.sb_blocklog)
+#define XFS_B_TO_FSB(mp,b)	\
+	((((uint64_t)(b)) + (mp)->m_blockmask) >> (mp)->m_sb.sb_blocklog)
+#define XFS_B_TO_FSBT(mp,b)	(((uint64_t)(b)) >> (mp)->m_sb.sb_blocklog)
+
+/*
+ * Allocation group header
+ *
+ * This is divided into three structures, placed in sequential 512-byte
+ * buffers after a copy of the superblock (also in a 512-byte buffer).
+ */
+#define	XFS_AGF_MAGIC	0x58414746	/* 'XAGF' */
+#define	XFS_AGI_MAGIC	0x58414749	/* 'XAGI' */
+#define	XFS_AGFL_MAGIC	0x5841464c	/* 'XAFL' */
+#define	XFS_AGF_VERSION	1
+#define	XFS_AGI_VERSION	1
+
+#define	XFS_AGF_GOOD_VERSION(v)	((v) == XFS_AGF_VERSION)
+#define	XFS_AGI_GOOD_VERSION(v)	((v) == XFS_AGI_VERSION)
+
+/*
+ * Btree number 0 is bno, 1 is cnt, 2 is rmap. This value gives the size of the
+ * arrays below.
+ */
+#define	XFS_BTNUM_AGF	((int)XFS_BTNUM_RMAPi + 1)
+
+/*
+ * The second word of agf_levels in the first a.g. overlaps the EFS
+ * superblock's magic number.  Since the magic numbers valid for EFS
+ * are > 64k, our value cannot be confused for an EFS superblock's.
+ */
+
+typedef struct xfs_agf {
+	/*
+	 * Common allocation group header information
+	 */
+	__be32		agf_magicnum;	/* magic number == XFS_AGF_MAGIC */
+	__be32		agf_versionnum;	/* header version == XFS_AGF_VERSION */
+	__be32		agf_seqno;	/* sequence # starting from 0 */
+	__be32		agf_length;	/* size in blocks of a.g. */
+	/*
+	 * Freespace and rmap information
+	 */
+	__be32		agf_roots[XFS_BTNUM_AGF];	/* root blocks */
+	__be32		agf_levels[XFS_BTNUM_AGF];	/* btree levels */
+
+	__be32		agf_flfirst;	/* first freelist block's index */
+	__be32		agf_fllast;	/* last freelist block's index */
+	__be32		agf_flcount;	/* count of blocks in freelist */
+	__be32		agf_freeblks;	/* total free blocks */
+
+	__be32		agf_longest;	/* longest free space */
+	__be32		agf_btreeblks;	/* # of blocks held in AGF btrees */
+	uuid_t		agf_uuid;	/* uuid of filesystem */
+
+	__be32		agf_rmap_blocks;	/* rmapbt blocks used */
+	__be32		agf_refcount_blocks;	/* refcountbt blocks used */
+
+	__be32		agf_refcount_root;	/* refcount tree root block */
+	__be32		agf_refcount_level;	/* refcount btree levels */
+
+	/*
+	 * reserve some contiguous space for future logged fields before we add
+	 * the unlogged fields. This makes the range logging via flags and
+	 * structure offsets much simpler.
+	 */
+	__be64		agf_spare64[14];
+
+	/* unlogged fields, written during buffer writeback. */
+	__be64		agf_lsn;	/* last write sequence */
+	__be32		agf_crc;	/* crc of agf sector */
+	__be32		agf_spare2;
+
+	/* structure must be padded to 64 bit alignment */
+} xfs_agf_t;
+
+#define XFS_AGF_CRC_OFF		offsetof(struct xfs_agf, agf_crc)
+
+#define	XFS_AGF_MAGICNUM	(1u << 0)
+#define	XFS_AGF_VERSIONNUM	(1u << 1)
+#define	XFS_AGF_SEQNO		(1u << 2)
+#define	XFS_AGF_LENGTH		(1u << 3)
+#define	XFS_AGF_ROOTS		(1u << 4)
+#define	XFS_AGF_LEVELS		(1u << 5)
+#define	XFS_AGF_FLFIRST		(1u << 6)
+#define	XFS_AGF_FLLAST		(1u << 7)
+#define	XFS_AGF_FLCOUNT		(1u << 8)
+#define	XFS_AGF_FREEBLKS	(1u << 9)
+#define	XFS_AGF_LONGEST		(1u << 10)
+#define	XFS_AGF_BTREEBLKS	(1u << 11)
+#define	XFS_AGF_UUID		(1u << 12)
+#define	XFS_AGF_RMAP_BLOCKS	(1u << 13)
+#define	XFS_AGF_REFCOUNT_BLOCKS	(1u << 14)
+#define	XFS_AGF_REFCOUNT_ROOT	(1u << 15)
+#define	XFS_AGF_REFCOUNT_LEVEL	(1u << 16)
+#define	XFS_AGF_SPARE64		(1u << 17)
+#define	XFS_AGF_NUM_BITS	18
+#define	XFS_AGF_ALL_BITS	((1u << XFS_AGF_NUM_BITS) - 1)
+
+#define XFS_AGF_FLAGS \
+	{ XFS_AGF_MAGICNUM,	"MAGICNUM" }, \
+	{ XFS_AGF_VERSIONNUM,	"VERSIONNUM" }, \
+	{ XFS_AGF_SEQNO,	"SEQNO" }, \
+	{ XFS_AGF_LENGTH,	"LENGTH" }, \
+	{ XFS_AGF_ROOTS,	"ROOTS" }, \
+	{ XFS_AGF_LEVELS,	"LEVELS" }, \
+	{ XFS_AGF_FLFIRST,	"FLFIRST" }, \
+	{ XFS_AGF_FLLAST,	"FLLAST" }, \
+	{ XFS_AGF_FLCOUNT,	"FLCOUNT" }, \
+	{ XFS_AGF_FREEBLKS,	"FREEBLKS" }, \
+	{ XFS_AGF_LONGEST,	"LONGEST" }, \
+	{ XFS_AGF_BTREEBLKS,	"BTREEBLKS" }, \
+	{ XFS_AGF_UUID,		"UUID" }, \
+	{ XFS_AGF_RMAP_BLOCKS,	"RMAP_BLOCKS" }, \
+	{ XFS_AGF_REFCOUNT_BLOCKS,	"REFCOUNT_BLOCKS" }, \
+	{ XFS_AGF_REFCOUNT_ROOT,	"REFCOUNT_ROOT" }, \
+	{ XFS_AGF_REFCOUNT_LEVEL,	"REFCOUNT_LEVEL" }, \
+	{ XFS_AGF_SPARE64,	"SPARE64" }
+
+/* disk block (xfs_daddr_t) in the AG */
+#define XFS_AGF_DADDR(mp)	((xfs_daddr_t)(1 << (mp)->m_sectbb_log))
+#define	XFS_AGF_BLOCK(mp)	XFS_HDR_BLOCK(mp, XFS_AGF_DADDR(mp))
+
+/*
+ * Size of the unlinked inode hash table in the agi.
+ */
+#define	XFS_AGI_UNLINKED_BUCKETS	64
+
+typedef struct xfs_agi {
+	/*
+	 * Common allocation group header information
+	 */
+	__be32		agi_magicnum;	/* magic number == XFS_AGI_MAGIC */
+	__be32		agi_versionnum;	/* header version == XFS_AGI_VERSION */
+	__be32		agi_seqno;	/* sequence # starting from 0 */
+	__be32		agi_length;	/* size in blocks of a.g. */
+	/*
+	 * Inode information
+	 * Inodes are mapped by interpreting the inode number, so no
+	 * mapping data is needed here.
+	 */
+	__be32		agi_count;	/* count of allocated inodes */
+	__be32		agi_root;	/* root of inode btree */
+	__be32		agi_level;	/* levels in inode btree */
+	__be32		agi_freecount;	/* number of free inodes */
+
+	__be32		agi_newino;	/* new inode just allocated */
+	__be32		agi_dirino;	/* last directory inode chunk */
+	/*
+	 * Hash table of inodes which have been unlinked but are
+	 * still being referenced.
+	 */
+	__be32		agi_unlinked[XFS_AGI_UNLINKED_BUCKETS];
+	/*
+	 * This marks the end of logging region 1 and start of logging region 2.
+	 */
+	uuid_t		agi_uuid;	/* uuid of filesystem */
+	__be32		agi_crc;	/* crc of agi sector */
+	__be32		agi_pad32;
+	__be64		agi_lsn;	/* last write sequence */
+
+	__be32		agi_free_root; /* root of the free inode btree */
+	__be32		agi_free_level;/* levels in free inode btree */
+
+	__be32		agi_iblocks;	/* inobt blocks used */
+	__be32		agi_fblocks;	/* finobt blocks used */
+
+	/* structure must be padded to 64 bit alignment */
+} xfs_agi_t;
+
+#define XFS_AGI_CRC_OFF		offsetof(struct xfs_agi, agi_crc)
+
+#define	XFS_AGI_MAGICNUM	(1u << 0)
+#define	XFS_AGI_VERSIONNUM	(1u << 1)
+#define	XFS_AGI_SEQNO		(1u << 2)
+#define	XFS_AGI_LENGTH		(1u << 3)
+#define	XFS_AGI_COUNT		(1u << 4)
+#define	XFS_AGI_ROOT		(1u << 5)
+#define	XFS_AGI_LEVEL		(1u << 6)
+#define	XFS_AGI_FREECOUNT	(1u << 7)
+#define	XFS_AGI_NEWINO		(1u << 8)
+#define	XFS_AGI_DIRINO		(1u << 9)
+#define	XFS_AGI_UNLINKED	(1u << 10)
+#define	XFS_AGI_NUM_BITS_R1	11	/* end of the 1st agi logging region */
+#define	XFS_AGI_ALL_BITS_R1	((1u << XFS_AGI_NUM_BITS_R1) - 1)
+#define	XFS_AGI_FREE_ROOT	(1u << 11)
+#define	XFS_AGI_FREE_LEVEL	(1u << 12)
+#define	XFS_AGI_IBLOCKS		(1u << 13) /* both inobt/finobt block counters */
+#define	XFS_AGI_NUM_BITS_R2	14
+
+/* disk block (xfs_daddr_t) in the AG */
+#define XFS_AGI_DADDR(mp)	((xfs_daddr_t)(2 << (mp)->m_sectbb_log))
+#define	XFS_AGI_BLOCK(mp)	XFS_HDR_BLOCK(mp, XFS_AGI_DADDR(mp))
+
+/*
+ * The third a.g. block contains the a.g. freelist, an array
+ * of block pointers to blocks owned by the allocation btree code.
+ */
+#define XFS_AGFL_DADDR(mp)	((xfs_daddr_t)(3 << (mp)->m_sectbb_log))
+#define	XFS_AGFL_BLOCK(mp)	XFS_HDR_BLOCK(mp, XFS_AGFL_DADDR(mp))
+#define	XFS_BUF_TO_AGFL(bp)	((struct xfs_agfl *)((bp)->b_addr))
+
+struct xfs_agfl {
+	__be32		agfl_magicnum;
+	__be32		agfl_seqno;
+	uuid_t		agfl_uuid;
+	__be64		agfl_lsn;
+	__be32		agfl_crc;
+} __attribute__((packed));
+
+#define XFS_AGFL_CRC_OFF	offsetof(struct xfs_agfl, agfl_crc)
+
+#define XFS_AGB_TO_FSB(mp,agno,agbno)	\
+	(((xfs_fsblock_t)(agno) << (mp)->m_sb.sb_agblklog) | (agbno))
+#define	XFS_FSB_TO_AGNO(mp,fsbno)	\
+	((xfs_agnumber_t)((fsbno) >> (mp)->m_sb.sb_agblklog))
+#define	XFS_FSB_TO_AGBNO(mp,fsbno)	\
+	((xfs_agblock_t)((fsbno) & xfs_mask32lo((mp)->m_sb.sb_agblklog)))
+#define	XFS_AGB_TO_DADDR(mp,agno,agbno)	\
+	((xfs_daddr_t)XFS_FSB_TO_BB(mp, \
+		(xfs_fsblock_t)(agno) * (mp)->m_sb.sb_agblocks + (agbno)))
+#define	XFS_AG_DADDR(mp,agno,d)		(XFS_AGB_TO_DADDR(mp, agno, 0) + (d))
+
+/*
+ * For checking for bad ranges of xfs_daddr_t's, covering multiple
+ * allocation groups or a single xfs_daddr_t that's a superblock copy.
+ */
+#define	XFS_AG_CHECK_DADDR(mp,d,len)	\
+	((len) == 1 ? \
+	    ASSERT((d) == XFS_SB_DADDR || \
+		   xfs_daddr_to_agbno(mp, d) != XFS_SB_DADDR) : \
+	    ASSERT(xfs_daddr_to_agno(mp, d) == \
+		   xfs_daddr_to_agno(mp, (d) + (len) - 1)))
+
+/*
+ * XFS Timestamps
+ * ==============
+ *
+ * Traditional ondisk inode timestamps consist of signed 32-bit counters for
+ * seconds and nanoseconds; time zero is the Unix epoch, Jan  1 00:00:00 UTC
+ * 1970, which means that the timestamp epoch is the same as the Unix epoch.
+ * Therefore, the ondisk min and max defined here can be used directly to
+ * constrain the incore timestamps on a Unix system.  Note that we actually
+ * encode a __be64 value on disk.
+ *
+ * When the bigtime feature is enabled, ondisk inode timestamps become an
+ * unsigned 64-bit nanoseconds counter.  This means that the bigtime inode
+ * timestamp epoch is the start of the classic timestamp range, which is
+ * Dec 13 20:45:52 UTC 1901.  Because the epochs are not the same, callers
+ * /must/ use the bigtime conversion functions when encoding and decoding raw
+ * timestamps.
+ */
+typedef __be64 xfs_timestamp_t;
+
+/* Legacy timestamp encoding format. */
+struct xfs_legacy_timestamp {
+	__be32		t_sec;		/* timestamp seconds */
+	__be32		t_nsec;		/* timestamp nanoseconds */
+};
+
+/*
+ * Smallest possible ondisk seconds value with traditional timestamps.  This
+ * corresponds exactly with the incore timestamp Dec 13 20:45:52 UTC 1901.
+ */
+#define XFS_LEGACY_TIME_MIN	((int64_t)S32_MIN)
+
+/*
+ * Largest possible ondisk seconds value with traditional timestamps.  This
+ * corresponds exactly with the incore timestamp Jan 19 03:14:07 UTC 2038.
+ */
+#define XFS_LEGACY_TIME_MAX	((int64_t)S32_MAX)
+
+/*
+ * Smallest possible ondisk seconds value with bigtime timestamps.  This
+ * corresponds (after conversion to a Unix timestamp) with the traditional
+ * minimum timestamp of Dec 13 20:45:52 UTC 1901.
+ */
+#define XFS_BIGTIME_TIME_MIN	((int64_t)0)
+
+/*
+ * Largest supported ondisk seconds value with bigtime timestamps.  This
+ * corresponds (after conversion to a Unix timestamp) with an incore timestamp
+ * of Jul  2 20:20:24 UTC 2486.
+ *
+ * We round down the ondisk limit so that the bigtime quota and inode max
+ * timestamps will be the same.
+ */
+#define XFS_BIGTIME_TIME_MAX	((int64_t)((-1ULL / NSEC_PER_SEC) & ~0x3ULL))
+
+/*
+ * Bigtime epoch is set exactly to the minimum time value that a traditional
+ * 32-bit timestamp can represent when using the Unix epoch as a reference.
+ * Hence the Unix epoch is at a fixed offset into the supported bigtime
+ * timestamp range.
+ *
+ * The bigtime epoch also matches the minimum value an on-disk 32-bit XFS
+ * timestamp can represent so we will not lose any fidelity in converting
+ * to/from unix and bigtime timestamps.
+ *
+ * The following conversion factor converts a seconds counter from the Unix
+ * epoch to the bigtime epoch.
+ */
+#define XFS_BIGTIME_EPOCH_OFFSET	(-(int64_t)S32_MIN)
+
+/* Convert a timestamp from the Unix epoch to the bigtime epoch. */
+static inline uint64_t xfs_unix_to_bigtime(time64_t unix_seconds)
+{
+	return (uint64_t)unix_seconds + XFS_BIGTIME_EPOCH_OFFSET;
+}
+
+/* Convert a timestamp from the bigtime epoch to the Unix epoch. */
+static inline time64_t xfs_bigtime_to_unix(uint64_t ondisk_seconds)
+{
+	return (time64_t)ondisk_seconds - XFS_BIGTIME_EPOCH_OFFSET;
+}
+
+/*
+ * On-disk inode structure.
+ *
+ * This is just the header or "dinode core", the inode is expanded to fill a
+ * variable size the leftover area split into a data and an attribute fork.
+ * The format of the data and attribute fork depends on the format of the
+ * inode as indicated by di_format and di_aformat.  To access the data and
+ * attribute use the XFS_DFORK_DPTR, XFS_DFORK_APTR, and XFS_DFORK_PTR macros
+ * below.
+ *
+ * There is a very similar struct xfs_log_dinode which matches the layout of
+ * this structure, but is kept in native format instead of big endian.
+ *
+ * Note: di_flushiter is only used by v1/2 inodes - it's effectively a zeroed
+ * padding field for v3 inodes.
+ */
+#define	XFS_DINODE_MAGIC		0x494e	/* 'IN' */
+struct xfs_dinode {
+	__be16		di_magic;	/* inode magic # = XFS_DINODE_MAGIC */
+	__be16		di_mode;	/* mode and type of file */
+	__u8		di_version;	/* inode version */
+	__u8		di_format;	/* format of di_c data */
+	__be16		di_onlink;	/* old number of links to file */
+	__be32		di_uid;		/* owner's user id */
+	__be32		di_gid;		/* owner's group id */
+	__be32		di_nlink;	/* number of links to file */
+	__be16		di_projid_lo;	/* lower part of owner's project id */
+	__be16		di_projid_hi;	/* higher part owner's project id */
+	union {
+		/* Number of data fork extents if NREXT64 is set */
+		__be64	di_big_nextents;
+
+		/* Padding for V3 inodes without NREXT64 set. */
+		__be64	di_v3_pad;
+
+		/* Padding and inode flush counter for V2 inodes. */
+		struct {
+			__u8	di_v2_pad[6];
+			__be16	di_flushiter;
+		};
+	};
+	xfs_timestamp_t	di_atime;	/* time last accessed */
+	xfs_timestamp_t	di_mtime;	/* time last modified */
+	xfs_timestamp_t	di_ctime;	/* time created/inode modified */
+	__be64		di_size;	/* number of bytes in file */
+	__be64		di_nblocks;	/* # of direct & btree blocks used */
+	__be32		di_extsize;	/* basic/minimum extent size for file */
+	union {
+		/*
+		 * For V2 inodes and V3 inodes without NREXT64 set, this
+		 * is the number of data and attr fork extents.
+		 */
+		struct {
+			__be32	di_nextents;
+			__be16	di_anextents;
+		} __packed;
+
+		/* Number of attr fork extents if NREXT64 is set. */
+		struct {
+			__be32	di_big_anextents;
+			__be16	di_nrext64_pad;
+		} __packed;
+	} __packed;
+	__u8		di_forkoff;	/* attr fork offs, <<3 for 64b align */
+	__s8		di_aformat;	/* format of attr fork's data */
+	__be32		di_dmevmask;	/* DMIG event mask */
+	__be16		di_dmstate;	/* DMIG state info */
+	__be16		di_flags;	/* random flags, XFS_DIFLAG_... */
+	__be32		di_gen;		/* generation number */
+
+	/* di_next_unlinked is the only non-core field in the old dinode */
+	__be32		di_next_unlinked;/* agi unlinked list ptr */
+
+	/* start of the extended dinode, writable fields */
+	__le32		di_crc;		/* CRC of the inode */
+	__be64		di_changecount;	/* number of attribute changes */
+	__be64		di_lsn;		/* flush sequence */
+	__be64		di_flags2;	/* more random flags */
+	__be32		di_cowextsize;	/* basic cow extent size for file */
+	__u8		di_pad2[12];	/* more padding for future expansion */
+
+	/* fields only written to during inode creation */
+	xfs_timestamp_t	di_crtime;	/* time created */
+	__be64		di_ino;		/* inode number */
+	uuid_t		di_uuid;	/* UUID of the filesystem */
+
+	/* structure must be padded to 64 bit alignment */
+};
+
+#define XFS_DINODE_CRC_OFF	offsetof(struct xfs_dinode, di_crc)
+
+#define DI_MAX_FLUSH 0xffff
+
+/*
+ * Size of the core inode on disk.  Version 1 and 2 inodes have
+ * the same size, but version 3 has grown a few additional fields.
+ */
+static inline uint xfs_dinode_size(int version)
+{
+	if (version == 3)
+		return sizeof(struct xfs_dinode);
+	return offsetof(struct xfs_dinode, di_crc);
+}
+
+/*
+ * The 32 bit link count in the inode theoretically maxes out at UINT_MAX.
+ * Since the pathconf interface is signed, we use 2^31 - 1 instead.
+ */
+#define	XFS_MAXLINK		((1U << 31) - 1U)
+
+/*
+ * Values for di_format
+ *
+ * This enum is used in string mapping in xfs_trace.h; please keep the
+ * TRACE_DEFINE_ENUMs for it up to date.
+ */
+enum xfs_dinode_fmt {
+	XFS_DINODE_FMT_DEV,		/* xfs_dev_t */
+	XFS_DINODE_FMT_LOCAL,		/* bulk data */
+	XFS_DINODE_FMT_EXTENTS,		/* struct xfs_bmbt_rec */
+	XFS_DINODE_FMT_BTREE,		/* struct xfs_bmdr_block */
+	XFS_DINODE_FMT_UUID		/* added long ago, but never used */
+};
+
+#define XFS_INODE_FORMAT_STR \
+	{ XFS_DINODE_FMT_DEV,		"dev" }, \
+	{ XFS_DINODE_FMT_LOCAL,		"local" }, \
+	{ XFS_DINODE_FMT_EXTENTS,	"extent" }, \
+	{ XFS_DINODE_FMT_BTREE,		"btree" }, \
+	{ XFS_DINODE_FMT_UUID,		"uuid" }
+
+/*
+ * Max values for extnum and aextnum.
+ *
+ * The original on-disk extent counts were held in signed fields, resulting in
+ * maximum extent counts of 2^31 and 2^15 for the data and attr forks
+ * respectively. Similarly the maximum extent length is limited to 2^21 blocks
+ * by the 21-bit wide blockcount field of a BMBT extent record.
+ *
+ * The newly introduced data fork extent counter can hold a 64-bit value,
+ * however the maximum number of extents in a file is also limited to 2^54
+ * extents by the 54-bit wide startoff field of a BMBT extent record.
+ *
+ * It is further limited by the maximum supported file size of 2^63
+ * *bytes*. This leads to a maximum extent count for maximally sized filesystem
+ * blocks (64kB) of:
+ *
+ * 2^63 bytes / 2^16 bytes per block = 2^47 blocks
+ *
+ * Rounding up 47 to the nearest multiple of bits-per-byte results in 48. Hence
+ * 2^48 was chosen as the maximum data fork extent count.
+ *
+ * The maximum file size that can be represented by the data fork extent counter
+ * in the worst case occurs when all extents are 1 block in length and each
+ * block is 1KB in size.
+ *
+ * With XFS_MAX_EXTCNT_DATA_FORK_SMALL representing maximum extent count and
+ * with 1KB sized blocks, a file can reach upto,
+ * 1KB * (2^31) = 2TB
+ *
+ * This is much larger than the theoretical maximum size of a directory
+ * i.e. XFS_DIR2_SPACE_SIZE * XFS_DIR2_MAX_SPACES = ~96GB.
+ *
+ * Hence, a directory inode can never overflow its data fork extent counter.
+ */
+#define XFS_MAX_EXTCNT_DATA_FORK_LARGE	((xfs_extnum_t)((1ULL << 48) - 1))
+#define XFS_MAX_EXTCNT_ATTR_FORK_LARGE	((xfs_extnum_t)((1ULL << 32) - 1))
+#define XFS_MAX_EXTCNT_DATA_FORK_SMALL	((xfs_extnum_t)((1ULL << 31) - 1))
+#define XFS_MAX_EXTCNT_ATTR_FORK_SMALL	((xfs_extnum_t)((1ULL << 15) - 1))
+
+/*
+ * When we upgrade an inode to the large extent counts, the maximum value by
+ * which the extent count can increase is bound by the change in size of the
+ * on-disk field. No upgrade operation should ever be adding more than a few
+ * tens of extents, so if we get a really large value it is a sign of a code bug
+ * or corruption.
+ */
+#define XFS_MAX_EXTCNT_UPGRADE_NR	\
+	min(XFS_MAX_EXTCNT_ATTR_FORK_LARGE - XFS_MAX_EXTCNT_ATTR_FORK_SMALL,	\
+	    XFS_MAX_EXTCNT_DATA_FORK_LARGE - XFS_MAX_EXTCNT_DATA_FORK_SMALL)
+
+/*
+ * Inode minimum and maximum sizes.
+ */
+#define	XFS_DINODE_MIN_LOG	8
+#define	XFS_DINODE_MAX_LOG	11
+#define	XFS_DINODE_MIN_SIZE	(1 << XFS_DINODE_MIN_LOG)
+#define	XFS_DINODE_MAX_SIZE	(1 << XFS_DINODE_MAX_LOG)
+
+/*
+ * Inode size for given fs.
+ */
+#define XFS_DINODE_SIZE(mp) \
+	(xfs_has_v3inodes(mp) ? \
+		sizeof(struct xfs_dinode) : \
+		offsetof(struct xfs_dinode, di_crc))
+#define XFS_LITINO(mp) \
+	((mp)->m_sb.sb_inodesize - XFS_DINODE_SIZE(mp))
+
+/*
+ * Inode data & attribute fork sizes, per inode.
+ */
+#define XFS_DFORK_BOFF(dip)		((int)((dip)->di_forkoff << 3))
+
+#define XFS_DFORK_DSIZE(dip,mp) \
+	((dip)->di_forkoff ? XFS_DFORK_BOFF(dip) : XFS_LITINO(mp))
+#define XFS_DFORK_ASIZE(dip,mp) \
+	((dip)->di_forkoff ? XFS_LITINO(mp) - XFS_DFORK_BOFF(dip) : 0)
+#define XFS_DFORK_SIZE(dip,mp,w) \
+	((w) == XFS_DATA_FORK ? \
+		XFS_DFORK_DSIZE(dip, mp) : \
+		XFS_DFORK_ASIZE(dip, mp))
+
+#define XFS_DFORK_MAXEXT(dip, mp, w) \
+	(XFS_DFORK_SIZE(dip, mp, w) / sizeof(struct xfs_bmbt_rec))
+
+/*
+ * Return pointers to the data or attribute forks.
+ */
+#define XFS_DFORK_DPTR(dip) \
+	((char *)dip + xfs_dinode_size(dip->di_version))
+#define XFS_DFORK_APTR(dip)	\
+	(XFS_DFORK_DPTR(dip) + XFS_DFORK_BOFF(dip))
+#define XFS_DFORK_PTR(dip,w)	\
+	((w) == XFS_DATA_FORK ? XFS_DFORK_DPTR(dip) : XFS_DFORK_APTR(dip))
+
+#define XFS_DFORK_FORMAT(dip,w) \
+	((w) == XFS_DATA_FORK ? \
+		(dip)->di_format : \
+		(dip)->di_aformat)
+
+/*
+ * For block and character special files the 32bit dev_t is stored at the
+ * beginning of the data fork.
+ */
+static inline xfs_dev_t xfs_dinode_get_rdev(struct xfs_dinode *dip)
+{
+	return be32_to_cpu(*(__be32 *)XFS_DFORK_DPTR(dip));
+}
+
+static inline void xfs_dinode_put_rdev(struct xfs_dinode *dip, xfs_dev_t rdev)
+{
+	*(__be32 *)XFS_DFORK_DPTR(dip) = cpu_to_be32(rdev);
+}
+
+/*
+ * Values for di_flags
+ */
+#define XFS_DIFLAG_REALTIME_BIT  0	/* file's blocks come from rt area */
+#define XFS_DIFLAG_PREALLOC_BIT  1	/* file space has been preallocated */
+#define XFS_DIFLAG_NEWRTBM_BIT   2	/* for rtbitmap inode, new format */
+#define XFS_DIFLAG_IMMUTABLE_BIT 3	/* inode is immutable */
+#define XFS_DIFLAG_APPEND_BIT    4	/* inode is append-only */
+#define XFS_DIFLAG_SYNC_BIT      5	/* inode is written synchronously */
+#define XFS_DIFLAG_NOATIME_BIT   6	/* do not update atime */
+#define XFS_DIFLAG_NODUMP_BIT    7	/* do not dump */
+#define XFS_DIFLAG_RTINHERIT_BIT 8	/* create with realtime bit set */
+#define XFS_DIFLAG_PROJINHERIT_BIT   9	/* create with parents projid */
+#define XFS_DIFLAG_NOSYMLINKS_BIT   10	/* disallow symlink creation */
+#define XFS_DIFLAG_EXTSIZE_BIT      11	/* inode extent size allocator hint */
+#define XFS_DIFLAG_EXTSZINHERIT_BIT 12	/* inherit inode extent size */
+#define XFS_DIFLAG_NODEFRAG_BIT     13	/* do not reorganize/defragment */
+#define XFS_DIFLAG_FILESTREAM_BIT   14  /* use filestream allocator */
+/* Do not use bit 15, di_flags is legacy and unchanging now */
+
+#define XFS_DIFLAG_REALTIME      (1 << XFS_DIFLAG_REALTIME_BIT)
+#define XFS_DIFLAG_PREALLOC      (1 << XFS_DIFLAG_PREALLOC_BIT)
+#define XFS_DIFLAG_NEWRTBM       (1 << XFS_DIFLAG_NEWRTBM_BIT)
+#define XFS_DIFLAG_IMMUTABLE     (1 << XFS_DIFLAG_IMMUTABLE_BIT)
+#define XFS_DIFLAG_APPEND        (1 << XFS_DIFLAG_APPEND_BIT)
+#define XFS_DIFLAG_SYNC          (1 << XFS_DIFLAG_SYNC_BIT)
+#define XFS_DIFLAG_NOATIME       (1 << XFS_DIFLAG_NOATIME_BIT)
+#define XFS_DIFLAG_NODUMP        (1 << XFS_DIFLAG_NODUMP_BIT)
+#define XFS_DIFLAG_RTINHERIT     (1 << XFS_DIFLAG_RTINHERIT_BIT)
+#define XFS_DIFLAG_PROJINHERIT   (1 << XFS_DIFLAG_PROJINHERIT_BIT)
+#define XFS_DIFLAG_NOSYMLINKS    (1 << XFS_DIFLAG_NOSYMLINKS_BIT)
+#define XFS_DIFLAG_EXTSIZE       (1 << XFS_DIFLAG_EXTSIZE_BIT)
+#define XFS_DIFLAG_EXTSZINHERIT  (1 << XFS_DIFLAG_EXTSZINHERIT_BIT)
+#define XFS_DIFLAG_NODEFRAG      (1 << XFS_DIFLAG_NODEFRAG_BIT)
+#define XFS_DIFLAG_FILESTREAM    (1 << XFS_DIFLAG_FILESTREAM_BIT)
+
+#define XFS_DIFLAG_ANY \
+	(XFS_DIFLAG_REALTIME | XFS_DIFLAG_PREALLOC | XFS_DIFLAG_NEWRTBM | \
+	 XFS_DIFLAG_IMMUTABLE | XFS_DIFLAG_APPEND | XFS_DIFLAG_SYNC | \
+	 XFS_DIFLAG_NOATIME | XFS_DIFLAG_NODUMP | XFS_DIFLAG_RTINHERIT | \
+	 XFS_DIFLAG_PROJINHERIT | XFS_DIFLAG_NOSYMLINKS | XFS_DIFLAG_EXTSIZE | \
+	 XFS_DIFLAG_EXTSZINHERIT | XFS_DIFLAG_NODEFRAG | XFS_DIFLAG_FILESTREAM)
+
+/*
+ * Values for di_flags2 These start by being exposed to userspace in the upper
+ * 16 bits of the XFS_XFLAG_s range.
+ */
+#define XFS_DIFLAG2_DAX_BIT	0	/* use DAX for this inode */
+#define XFS_DIFLAG2_REFLINK_BIT	1	/* file's blocks may be shared */
+#define XFS_DIFLAG2_COWEXTSIZE_BIT   2  /* copy on write extent size hint */
+#define XFS_DIFLAG2_BIGTIME_BIT	3	/* big timestamps */
+#define XFS_DIFLAG2_NREXT64_BIT 4	/* large extent counters */
+
+#define XFS_DIFLAG2_DAX		(1 << XFS_DIFLAG2_DAX_BIT)
+#define XFS_DIFLAG2_REFLINK     (1 << XFS_DIFLAG2_REFLINK_BIT)
+#define XFS_DIFLAG2_COWEXTSIZE  (1 << XFS_DIFLAG2_COWEXTSIZE_BIT)
+#define XFS_DIFLAG2_BIGTIME	(1 << XFS_DIFLAG2_BIGTIME_BIT)
+#define XFS_DIFLAG2_NREXT64	(1 << XFS_DIFLAG2_NREXT64_BIT)
+
+#define XFS_DIFLAG2_ANY \
+	(XFS_DIFLAG2_DAX | XFS_DIFLAG2_REFLINK | XFS_DIFLAG2_COWEXTSIZE | \
+	 XFS_DIFLAG2_BIGTIME | XFS_DIFLAG2_NREXT64)
+
+static inline bool xfs_dinode_has_bigtime(const struct xfs_dinode *dip)
+{
+	return dip->di_version >= 3 &&
+	       (dip->di_flags2 & cpu_to_be64(XFS_DIFLAG2_BIGTIME));
+}
+
+static inline bool xfs_dinode_has_large_extent_counts(
+	const struct xfs_dinode *dip)
+{
+	return dip->di_version >= 3 &&
+	       (dip->di_flags2 & cpu_to_be64(XFS_DIFLAG2_NREXT64));
+}
+
+/*
+ * Inode number format:
+ * low inopblog bits - offset in block
+ * next agblklog bits - block number in ag
+ * next agno_log bits - ag number
+ * high agno_log-agblklog-inopblog bits - 0
+ */
+#define	XFS_INO_MASK(k)			(uint32_t)((1ULL << (k)) - 1)
+#define	XFS_INO_OFFSET_BITS(mp)		(mp)->m_sb.sb_inopblog
+#define	XFS_INO_AGBNO_BITS(mp)		(mp)->m_sb.sb_agblklog
+#define	XFS_INO_AGINO_BITS(mp)		((mp)->m_ino_geo.agino_log)
+#define	XFS_INO_AGNO_BITS(mp)		(mp)->m_agno_log
+#define	XFS_INO_BITS(mp)		\
+	XFS_INO_AGNO_BITS(mp) + XFS_INO_AGINO_BITS(mp)
+#define	XFS_INO_TO_AGNO(mp,i)		\
+	((xfs_agnumber_t)((i) >> XFS_INO_AGINO_BITS(mp)))
+#define	XFS_INO_TO_AGINO(mp,i)		\
+	((xfs_agino_t)(i) & XFS_INO_MASK(XFS_INO_AGINO_BITS(mp)))
+#define	XFS_INO_TO_AGBNO(mp,i)		\
+	(((xfs_agblock_t)(i) >> XFS_INO_OFFSET_BITS(mp)) & \
+		XFS_INO_MASK(XFS_INO_AGBNO_BITS(mp)))
+#define	XFS_INO_TO_OFFSET(mp,i)		\
+	((int)(i) & XFS_INO_MASK(XFS_INO_OFFSET_BITS(mp)))
+#define	XFS_INO_TO_FSB(mp,i)		\
+	XFS_AGB_TO_FSB(mp, XFS_INO_TO_AGNO(mp,i), XFS_INO_TO_AGBNO(mp,i))
+#define	XFS_AGINO_TO_INO(mp,a,i)	\
+	(((xfs_ino_t)(a) << XFS_INO_AGINO_BITS(mp)) | (i))
+#define	XFS_AGINO_TO_AGBNO(mp,i)	((i) >> XFS_INO_OFFSET_BITS(mp))
+#define	XFS_AGINO_TO_OFFSET(mp,i)	\
+	((i) & XFS_INO_MASK(XFS_INO_OFFSET_BITS(mp)))
+#define	XFS_OFFBNO_TO_AGINO(mp,b,o)	\
+	((xfs_agino_t)(((b) << XFS_INO_OFFSET_BITS(mp)) | (o)))
+#define	XFS_FSB_TO_INO(mp, b)	((xfs_ino_t)((b) << XFS_INO_OFFSET_BITS(mp)))
+#define	XFS_AGB_TO_AGINO(mp, b)	((xfs_agino_t)((b) << XFS_INO_OFFSET_BITS(mp)))
+
+#define	XFS_MAXINUMBER		((xfs_ino_t)((1ULL << 56) - 1ULL))
+#define	XFS_MAXINUMBER_32	((xfs_ino_t)((1ULL << 32) - 1ULL))
+
+/*
+ * RealTime Device format definitions
+ */
+
+/* Min and max rt extent sizes, specified in bytes */
+#define	XFS_MAX_RTEXTSIZE	(1024 * 1024 * 1024)	/* 1GB */
+#define	XFS_DFL_RTEXTSIZE	(64 * 1024)	        /* 64kB */
+#define	XFS_MIN_RTEXTSIZE	(4 * 1024)		/* 4kB */
+
+#define	XFS_BLOCKSIZE(mp)	((mp)->m_sb.sb_blocksize)
+#define	XFS_BLOCKMASK(mp)	((mp)->m_blockmask)
+#define	XFS_BLOCKWSIZE(mp)	((mp)->m_blockwsize)
+#define	XFS_BLOCKWMASK(mp)	((mp)->m_blockwmask)
+
+/*
+ * RT Summary and bit manipulation macros.
+ */
+#define	XFS_SUMOFFS(mp,ls,bb)	((int)((ls) * (mp)->m_sb.sb_rbmblocks + (bb)))
+#define	XFS_SUMOFFSTOBLOCK(mp,s)	\
+	(((s) * (uint)sizeof(xfs_suminfo_t)) >> (mp)->m_sb.sb_blocklog)
+#define	XFS_SUMPTR(mp,bp,so)	\
+	((xfs_suminfo_t *)((bp)->b_addr + \
+		(((so) * (uint)sizeof(xfs_suminfo_t)) & XFS_BLOCKMASK(mp))))
+
+#define	XFS_BITTOBLOCK(mp,bi)	((bi) >> (mp)->m_blkbit_log)
+#define	XFS_BLOCKTOBIT(mp,bb)	((bb) << (mp)->m_blkbit_log)
+#define	XFS_BITTOWORD(mp,bi)	\
+	((int)(((bi) >> XFS_NBWORDLOG) & XFS_BLOCKWMASK(mp)))
+
+#define	XFS_RTMIN(a,b)	((a) < (b) ? (a) : (b))
+#define	XFS_RTMAX(a,b)	((a) > (b) ? (a) : (b))
+
+#define	XFS_RTLOBIT(w)	xfs_lowbit32(w)
+#define	XFS_RTHIBIT(w)	xfs_highbit32(w)
+
+#define	XFS_RTBLOCKLOG(b)	xfs_highbit64(b)
+
+/*
+ * Dquot and dquot block format definitions
+ */
+#define XFS_DQUOT_MAGIC		0x4451		/* 'DQ' */
+#define XFS_DQUOT_VERSION	(uint8_t)0x01	/* latest version number */
+
+#define XFS_DQTYPE_USER		(1u << 0)	/* user dquot record */
+#define XFS_DQTYPE_PROJ		(1u << 1)	/* project dquot record */
+#define XFS_DQTYPE_GROUP	(1u << 2)	/* group dquot record */
+#define XFS_DQTYPE_BIGTIME	(1u << 7)	/* large expiry timestamps */
+
+/* bitmask to determine if this is a user/group/project dquot */
+#define XFS_DQTYPE_REC_MASK	(XFS_DQTYPE_USER | \
+				 XFS_DQTYPE_PROJ | \
+				 XFS_DQTYPE_GROUP)
+
+#define XFS_DQTYPE_ANY		(XFS_DQTYPE_REC_MASK | \
+				 XFS_DQTYPE_BIGTIME)
+
+/*
+ * XFS Quota Timers
+ * ================
+ *
+ * Traditional quota grace period expiration timers are an unsigned 32-bit
+ * seconds counter; time zero is the Unix epoch, Jan  1 00:00:01 UTC 1970.
+ * Note that an expiration value of zero means that the quota limit has not
+ * been reached, and therefore no expiration has been set.  Therefore, the
+ * ondisk min and max defined here can be used directly to constrain the incore
+ * quota expiration timestamps on a Unix system.
+ *
+ * When bigtime is enabled, we trade two bits of precision to expand the
+ * expiration timeout range to match that of big inode timestamps.  The min and
+ * max recorded here are the on-disk limits, not a Unix timestamp.
+ *
+ * The grace period for each quota type is stored in the root dquot (id = 0)
+ * and is applied to a non-root dquot when it exceeds the soft or hard limits.
+ * The length of quota grace periods are unsigned 32-bit quantities measured in
+ * units of seconds.  A value of zero means to use the default period.
+ */
+
+/*
+ * Smallest possible ondisk quota expiration value with traditional timestamps.
+ * This corresponds exactly with the incore expiration Jan  1 00:00:01 UTC 1970.
+ */
+#define XFS_DQ_LEGACY_EXPIRY_MIN	((int64_t)1)
+
+/*
+ * Largest possible ondisk quota expiration value with traditional timestamps.
+ * This corresponds exactly with the incore expiration Feb  7 06:28:15 UTC 2106.
+ */
+#define XFS_DQ_LEGACY_EXPIRY_MAX	((int64_t)U32_MAX)
+
+/*
+ * Smallest possible ondisk quota expiration value with bigtime timestamps.
+ * This corresponds (after conversion to a Unix timestamp) with the incore
+ * expiration of Jan  1 00:00:04 UTC 1970.
+ */
+#define XFS_DQ_BIGTIME_EXPIRY_MIN	(XFS_DQ_LEGACY_EXPIRY_MIN)
+
+/*
+ * Largest supported ondisk quota expiration value with bigtime timestamps.
+ * This corresponds (after conversion to a Unix timestamp) with an incore
+ * expiration of Jul  2 20:20:24 UTC 2486.
+ *
+ * The ondisk field supports values up to -1U, which corresponds to an incore
+ * expiration in 2514.  This is beyond the maximum the bigtime inode timestamp,
+ * so we cap the maximum bigtime quota expiration to the max inode timestamp.
+ */
+#define XFS_DQ_BIGTIME_EXPIRY_MAX	((int64_t)4074815106U)
+
+/*
+ * The following conversion factors assist in converting a quota expiration
+ * timestamp between the incore and ondisk formats.
+ */
+#define XFS_DQ_BIGTIME_SHIFT	(2)
+#define XFS_DQ_BIGTIME_SLACK	((int64_t)(1ULL << XFS_DQ_BIGTIME_SHIFT) - 1)
+
+/* Convert an incore quota expiration timestamp to an ondisk bigtime value. */
+static inline uint32_t xfs_dq_unix_to_bigtime(time64_t unix_seconds)
+{
+	/*
+	 * Round the expiration timestamp up to the nearest bigtime timestamp
+	 * that we can store, to give users the most time to fix problems.
+	 */
+	return ((uint64_t)unix_seconds + XFS_DQ_BIGTIME_SLACK) >>
+			XFS_DQ_BIGTIME_SHIFT;
+}
+
+/* Convert an ondisk bigtime quota expiration value to an incore timestamp. */
+static inline time64_t xfs_dq_bigtime_to_unix(uint32_t ondisk_seconds)
+{
+	return (time64_t)ondisk_seconds << XFS_DQ_BIGTIME_SHIFT;
+}
+
+/*
+ * Default quota grace periods, ranging from zero (use the compiled defaults)
+ * to ~136 years.  These are applied to a non-root dquot that has exceeded
+ * either limit.
+ */
+#define XFS_DQ_GRACE_MIN		((int64_t)0)
+#define XFS_DQ_GRACE_MAX		((int64_t)U32_MAX)
+
+/*
+ * This is the main portion of the on-disk representation of quota information
+ * for a user.  We pad this with some more expansion room to construct the on
+ * disk structure.
+ */
+struct xfs_disk_dquot {
+	__be16		d_magic;	/* dquot magic = XFS_DQUOT_MAGIC */
+	__u8		d_version;	/* dquot version */
+	__u8		d_type;		/* XFS_DQTYPE_USER/PROJ/GROUP */
+	__be32		d_id;		/* user,project,group id */
+	__be64		d_blk_hardlimit;/* absolute limit on disk blks */
+	__be64		d_blk_softlimit;/* preferred limit on disk blks */
+	__be64		d_ino_hardlimit;/* maximum # allocated inodes */
+	__be64		d_ino_softlimit;/* preferred inode limit */
+	__be64		d_bcount;	/* disk blocks owned by the user */
+	__be64		d_icount;	/* inodes owned by the user */
+	__be32		d_itimer;	/* zero if within inode limits if not,
+					   this is when we refuse service */
+	__be32		d_btimer;	/* similar to above; for disk blocks */
+	__be16		d_iwarns;	/* warnings issued wrt num inodes */
+	__be16		d_bwarns;	/* warnings issued wrt disk blocks */
+	__be32		d_pad0;		/* 64 bit align */
+	__be64		d_rtb_hardlimit;/* absolute limit on realtime blks */
+	__be64		d_rtb_softlimit;/* preferred limit on RT disk blks */
+	__be64		d_rtbcount;	/* realtime blocks owned */
+	__be32		d_rtbtimer;	/* similar to above; for RT disk blocks */
+	__be16		d_rtbwarns;	/* warnings issued wrt RT disk blocks */
+	__be16		d_pad;
+};
+
+/*
+ * This is what goes on disk. This is separated from the xfs_disk_dquot because
+ * carrying the unnecessary padding would be a waste of memory.
+ */
+struct xfs_dqblk {
+	struct xfs_disk_dquot	dd_diskdq; /* portion living incore as well */
+	char			dd_fill[4];/* filling for posterity */
+
+	/*
+	 * These two are only present on filesystems with the CRC bits set.
+	 */
+	__be32		  dd_crc;	/* checksum */
+	__be64		  dd_lsn;	/* last modification in log */
+	uuid_t		  dd_uuid;	/* location information */
+};
+
+#define XFS_DQUOT_CRC_OFF	offsetof(struct xfs_dqblk, dd_crc)
+
+/*
+ * This defines the unit of allocation of dquots.
+ *
+ * Currently, it is just one file system block, and a 4K blk contains 30
+ * (136 * 30 = 4080) dquots. It's probably not worth trying to make
+ * this more dynamic.
+ *
+ * However, if this number is changed, we have to make sure that we don't
+ * implicitly assume that we do allocations in chunks of a single filesystem
+ * block in the dquot/xqm code.
+ *
+ * This is part of the ondisk format because the structure size is not a power
+ * of two, which leaves slack at the end of the disk block.
+ */
+#define XFS_DQUOT_CLUSTER_SIZE_FSB	(xfs_filblks_t)1
+
+/*
+ * Remote symlink format and access functions.
+ */
+#define XFS_SYMLINK_MAGIC	0x58534c4d	/* XSLM */
+
+struct xfs_dsymlink_hdr {
+	__be32	sl_magic;
+	__be32	sl_offset;
+	__be32	sl_bytes;
+	__be32	sl_crc;
+	uuid_t	sl_uuid;
+	__be64	sl_owner;
+	__be64	sl_blkno;
+	__be64	sl_lsn;
+};
+
+#define XFS_SYMLINK_CRC_OFF	offsetof(struct xfs_dsymlink_hdr, sl_crc)
+
+#define XFS_SYMLINK_MAXLEN	1024
+/*
+ * The maximum pathlen is 1024 bytes. Since the minimum file system
+ * blocksize is 512 bytes, we can get a max of 3 extents back from
+ * bmapi when crc headers are taken into account.
+ */
+#define XFS_SYMLINK_MAPS 3
+
+#define XFS_SYMLINK_BUF_SPACE(mp, bufsize)	\
+	((bufsize) - (xfs_has_crc((mp)) ? \
+			sizeof(struct xfs_dsymlink_hdr) : 0))
+
+
+/*
+ * Allocation Btree format definitions
+ *
+ * There are two on-disk btrees, one sorted by blockno and one sorted
+ * by blockcount and blockno.  All blocks look the same to make the code
+ * simpler; if we have time later, we'll make the optimizations.
+ */
+#define	XFS_ABTB_MAGIC		0x41425442	/* 'ABTB' for bno tree */
+#define	XFS_ABTB_CRC_MAGIC	0x41423342	/* 'AB3B' */
+#define	XFS_ABTC_MAGIC		0x41425443	/* 'ABTC' for cnt tree */
+#define	XFS_ABTC_CRC_MAGIC	0x41423343	/* 'AB3C' */
+
+/*
+ * Data record/key structure
+ */
+typedef struct xfs_alloc_rec {
+	__be32		ar_startblock;	/* starting block number */
+	__be32		ar_blockcount;	/* count of free blocks */
+} xfs_alloc_rec_t, xfs_alloc_key_t;
+
+typedef struct xfs_alloc_rec_incore {
+	xfs_agblock_t	ar_startblock;	/* starting block number */
+	xfs_extlen_t	ar_blockcount;	/* count of free blocks */
+} xfs_alloc_rec_incore_t;
+
+/* btree pointer type */
+typedef __be32 xfs_alloc_ptr_t;
+
+/*
+ * Block numbers in the AG:
+ * SB is sector 0, AGF is sector 1, AGI is sector 2, AGFL is sector 3.
+ */
+#define	XFS_BNO_BLOCK(mp)	((xfs_agblock_t)(XFS_AGFL_BLOCK(mp) + 1))
+#define	XFS_CNT_BLOCK(mp)	((xfs_agblock_t)(XFS_BNO_BLOCK(mp) + 1))
+
+
+/*
+ * Inode Allocation Btree format definitions
+ *
+ * There is a btree for the inode map per allocation group.
+ */
+#define	XFS_IBT_MAGIC		0x49414254	/* 'IABT' */
+#define	XFS_IBT_CRC_MAGIC	0x49414233	/* 'IAB3' */
+#define	XFS_FIBT_MAGIC		0x46494254	/* 'FIBT' */
+#define	XFS_FIBT_CRC_MAGIC	0x46494233	/* 'FIB3' */
+
+typedef uint64_t	xfs_inofree_t;
+#define	XFS_INODES_PER_CHUNK		(NBBY * sizeof(xfs_inofree_t))
+#define	XFS_INODES_PER_CHUNK_LOG	(XFS_NBBYLOG + 3)
+#define	XFS_INOBT_ALL_FREE		((xfs_inofree_t)-1)
+#define	XFS_INOBT_MASK(i)		((xfs_inofree_t)1 << (i))
+
+#define XFS_INOBT_HOLEMASK_FULL		0	/* holemask for full chunk */
+#define XFS_INOBT_HOLEMASK_BITS		(NBBY * sizeof(uint16_t))
+#define XFS_INODES_PER_HOLEMASK_BIT	\
+	(XFS_INODES_PER_CHUNK / (NBBY * sizeof(uint16_t)))
+
+static inline xfs_inofree_t xfs_inobt_maskn(int i, int n)
+{
+	return ((n >= XFS_INODES_PER_CHUNK ? 0 : XFS_INOBT_MASK(n)) - 1) << i;
+}
+
+/*
+ * The on-disk inode record structure has two formats. The original "full"
+ * format uses a 4-byte freecount. The "sparse" format uses a 1-byte freecount
+ * and replaces the 3 high-order freecount bytes wth the holemask and inode
+ * count.
+ *
+ * The holemask of the sparse record format allows an inode chunk to have holes
+ * that refer to blocks not owned by the inode record. This facilitates inode
+ * allocation in the event of severe free space fragmentation.
+ */
+typedef struct xfs_inobt_rec {
+	__be32		ir_startino;	/* starting inode number */
+	union {
+		struct {
+			__be32	ir_freecount;	/* count of free inodes */
+		} f;
+		struct {
+			__be16	ir_holemask;/* hole mask for sparse chunks */
+			__u8	ir_count;	/* total inode count */
+			__u8	ir_freecount;	/* count of free inodes */
+		} sp;
+	} ir_u;
+	__be64		ir_free;	/* free inode mask */
+} xfs_inobt_rec_t;
+
+typedef struct xfs_inobt_rec_incore {
+	xfs_agino_t	ir_startino;	/* starting inode number */
+	uint16_t	ir_holemask;	/* hole mask for sparse chunks */
+	uint8_t		ir_count;	/* total inode count */
+	uint8_t		ir_freecount;	/* count of free inodes (set bits) */
+	xfs_inofree_t	ir_free;	/* free inode mask */
+} xfs_inobt_rec_incore_t;
+
+static inline bool xfs_inobt_issparse(uint16_t holemask)
+{
+	/* non-zero holemask represents a sparse rec. */
+	return holemask;
+}
+
+/*
+ * Key structure
+ */
+typedef struct xfs_inobt_key {
+	__be32		ir_startino;	/* starting inode number */
+} xfs_inobt_key_t;
+
+/* btree pointer type */
+typedef __be32 xfs_inobt_ptr_t;
+
+/*
+ * block numbers in the AG.
+ */
+#define	XFS_IBT_BLOCK(mp)		((xfs_agblock_t)(XFS_CNT_BLOCK(mp) + 1))
+#define	XFS_FIBT_BLOCK(mp)		((xfs_agblock_t)(XFS_IBT_BLOCK(mp) + 1))
+
+/*
+ * Reverse mapping btree format definitions
+ *
+ * There is a btree for the reverse map per allocation group
+ */
+#define	XFS_RMAP_CRC_MAGIC	0x524d4233	/* 'RMB3' */
+
+/*
+ * Ownership info for an extent.  This is used to create reverse-mapping
+ * entries.
+ */
+#define XFS_OWNER_INFO_ATTR_FORK	(1 << 0)
+#define XFS_OWNER_INFO_BMBT_BLOCK	(1 << 1)
+struct xfs_owner_info {
+	uint64_t		oi_owner;
+	xfs_fileoff_t		oi_offset;
+	unsigned int		oi_flags;
+};
+
+/*
+ * Special owner types.
+ *
+ * Seeing as we only support up to 8EB, we have the upper bit of the owner field
+ * to tell us we have a special owner value. We use these for static metadata
+ * allocated at mkfs/growfs time, as well as for freespace management metadata.
+ */
+#define XFS_RMAP_OWN_NULL	(-1ULL)	/* No owner, for growfs */
+#define XFS_RMAP_OWN_UNKNOWN	(-2ULL)	/* Unknown owner, for EFI recovery */
+#define XFS_RMAP_OWN_FS		(-3ULL)	/* static fs metadata */
+#define XFS_RMAP_OWN_LOG	(-4ULL)	/* static fs metadata */
+#define XFS_RMAP_OWN_AG		(-5ULL)	/* AG freespace btree blocks */
+#define XFS_RMAP_OWN_INOBT	(-6ULL)	/* Inode btree blocks */
+#define XFS_RMAP_OWN_INODES	(-7ULL)	/* Inode chunk */
+#define XFS_RMAP_OWN_REFC	(-8ULL) /* refcount tree */
+#define XFS_RMAP_OWN_COW	(-9ULL) /* cow allocations */
+#define XFS_RMAP_OWN_MIN	(-10ULL) /* guard */
+
+#define XFS_RMAP_NON_INODE_OWNER(owner)	(!!((owner) & (1ULL << 63)))
+
+/*
+ * Data record structure
+ */
+struct xfs_rmap_rec {
+	__be32		rm_startblock;	/* extent start block */
+	__be32		rm_blockcount;	/* extent length */
+	__be64		rm_owner;	/* extent owner */
+	__be64		rm_offset;	/* offset within the owner */
+};
+
+/*
+ * rmap btree record
+ *  rm_offset:63 is the attribute fork flag
+ *  rm_offset:62 is the bmbt block flag
+ *  rm_offset:61 is the unwritten extent flag (same as l0:63 in bmbt)
+ *  rm_offset:54-60 aren't used and should be zero
+ *  rm_offset:0-53 is the block offset within the inode
+ */
+#define XFS_RMAP_OFF_ATTR_FORK	((uint64_t)1ULL << 63)
+#define XFS_RMAP_OFF_BMBT_BLOCK	((uint64_t)1ULL << 62)
+#define XFS_RMAP_OFF_UNWRITTEN	((uint64_t)1ULL << 61)
+
+#define XFS_RMAP_LEN_MAX	((uint32_t)~0U)
+#define XFS_RMAP_OFF_FLAGS	(XFS_RMAP_OFF_ATTR_FORK | \
+				 XFS_RMAP_OFF_BMBT_BLOCK | \
+				 XFS_RMAP_OFF_UNWRITTEN)
+#define XFS_RMAP_OFF_MASK	((uint64_t)0x3FFFFFFFFFFFFFULL)
+
+#define XFS_RMAP_OFF(off)		((off) & XFS_RMAP_OFF_MASK)
+
+#define XFS_RMAP_IS_BMBT_BLOCK(off)	(!!((off) & XFS_RMAP_OFF_BMBT_BLOCK))
+#define XFS_RMAP_IS_ATTR_FORK(off)	(!!((off) & XFS_RMAP_OFF_ATTR_FORK))
+#define XFS_RMAP_IS_UNWRITTEN(len)	(!!((off) & XFS_RMAP_OFF_UNWRITTEN))
+
+#define RMAPBT_STARTBLOCK_BITLEN	32
+#define RMAPBT_BLOCKCOUNT_BITLEN	32
+#define RMAPBT_OWNER_BITLEN		64
+#define RMAPBT_ATTRFLAG_BITLEN		1
+#define RMAPBT_BMBTFLAG_BITLEN		1
+#define RMAPBT_EXNTFLAG_BITLEN		1
+#define RMAPBT_UNUSED_OFFSET_BITLEN	7
+#define RMAPBT_OFFSET_BITLEN		54
+
+/*
+ * Key structure
+ *
+ * We don't use the length for lookups
+ */
+struct xfs_rmap_key {
+	__be32		rm_startblock;	/* extent start block */
+	__be64		rm_owner;	/* extent owner */
+	__be64		rm_offset;	/* offset within the owner */
+} __attribute__((packed));
+
+/* btree pointer type */
+typedef __be32 xfs_rmap_ptr_t;
+
+#define	XFS_RMAP_BLOCK(mp) \
+	(xfs_has_finobt(((mp))) ? \
+	 XFS_FIBT_BLOCK(mp) + 1 : \
+	 XFS_IBT_BLOCK(mp) + 1)
+
+/*
+ * Reference Count Btree format definitions
+ *
+ */
+#define	XFS_REFC_CRC_MAGIC	0x52334643	/* 'R3FC' */
+
+unsigned int xfs_refc_block(struct xfs_mount *mp);
+
+/*
+ * Data record/key structure
+ *
+ * Each record associates a range of physical blocks (starting at
+ * rc_startblock and ending rc_blockcount blocks later) with a reference
+ * count (rc_refcount).  Extents that are being used to stage a copy on
+ * write (CoW) operation are recorded in the refcount btree with a
+ * refcount of 1.  All other records must have a refcount > 1 and must
+ * track an extent mapped only by file data forks.
+ *
+ * Extents with a single owner (attributes, metadata, non-shared file
+ * data) are not tracked here.  Free space is also not tracked here.
+ * This is consistent with pre-reflink XFS.
+ */
+
+/*
+ * Extents that are being used to stage a copy on write are stored
+ * in the refcount btree with a refcount of 1 and the upper bit set
+ * on the startblock.  This speeds up mount time deletion of stale
+ * staging extents because they're all at the right side of the tree.
+ */
+#define XFS_REFC_COWFLAG		(1U << 31)
+#define REFCNTBT_COWFLAG_BITLEN		1
+#define REFCNTBT_AGBLOCK_BITLEN		31
+
+struct xfs_refcount_rec {
+	__be32		rc_startblock;	/* starting block number */
+	__be32		rc_blockcount;	/* count of blocks */
+	__be32		rc_refcount;	/* number of inodes linked here */
+};
+
+struct xfs_refcount_key {
+	__be32		rc_startblock;	/* starting block number */
+};
+
+#define MAXREFCOUNT	((xfs_nlink_t)~0U)
+#define MAXREFCEXTLEN	((xfs_extlen_t)~0U)
+
+/* btree pointer type */
+typedef __be32 xfs_refcount_ptr_t;
+
+
+/*
+ * BMAP Btree format definitions
+ *
+ * This includes both the root block definition that sits inside an inode fork
+ * and the record/pointer formats for the leaf/node in the blocks.
+ */
+#define XFS_BMAP_MAGIC		0x424d4150	/* 'BMAP' */
+#define XFS_BMAP_CRC_MAGIC	0x424d4133	/* 'BMA3' */
+
+/*
+ * Bmap root header, on-disk form only.
+ */
+typedef struct xfs_bmdr_block {
+	__be16		bb_level;	/* 0 is a leaf */
+	__be16		bb_numrecs;	/* current # of data records */
+} xfs_bmdr_block_t;
+
+/*
+ * Bmap btree record and extent descriptor.
+ *  l0:63 is an extent flag (value 1 indicates non-normal).
+ *  l0:9-62 are startoff.
+ *  l0:0-8 and l1:21-63 are startblock.
+ *  l1:0-20 are blockcount.
+ */
+#define BMBT_EXNTFLAG_BITLEN	1
+#define BMBT_STARTOFF_BITLEN	54
+#define BMBT_STARTBLOCK_BITLEN	52
+#define BMBT_BLOCKCOUNT_BITLEN	21
+
+#define BMBT_STARTOFF_MASK	((1ULL << BMBT_STARTOFF_BITLEN) - 1)
+#define BMBT_BLOCKCOUNT_MASK	((1ULL << BMBT_BLOCKCOUNT_BITLEN) - 1)
+
+#define XFS_MAX_BMBT_EXTLEN	((xfs_extlen_t)(BMBT_BLOCKCOUNT_MASK))
+
+/*
+ * bmbt records have a file offset (block) field that is 54 bits wide, so this
+ * is the largest xfs_fileoff_t that we ever expect to see.
+ */
+#define XFS_MAX_FILEOFF		(BMBT_STARTOFF_MASK + BMBT_BLOCKCOUNT_MASK)
+
+typedef struct xfs_bmbt_rec {
+	__be64			l0, l1;
+} xfs_bmbt_rec_t;
+
+typedef uint64_t	xfs_bmbt_rec_base_t;	/* use this for casts */
+typedef xfs_bmbt_rec_t xfs_bmdr_rec_t;
+
+/*
+ * Values and macros for delayed-allocation startblock fields.
+ */
+#define STARTBLOCKVALBITS	17
+#define STARTBLOCKMASKBITS	(15 + 20)
+#define STARTBLOCKMASK		\
+	(((((xfs_fsblock_t)1) << STARTBLOCKMASKBITS) - 1) << STARTBLOCKVALBITS)
+
+static inline int isnullstartblock(xfs_fsblock_t x)
+{
+	return ((x) & STARTBLOCKMASK) == STARTBLOCKMASK;
+}
+
+static inline xfs_fsblock_t nullstartblock(int k)
+{
+	ASSERT(k < (1 << STARTBLOCKVALBITS));
+	return STARTBLOCKMASK | (k);
+}
+
+static inline xfs_filblks_t startblockval(xfs_fsblock_t x)
+{
+	return (xfs_filblks_t)((x) & ~STARTBLOCKMASK);
+}
+
+/*
+ * Key structure for non-leaf levels of the tree.
+ */
+typedef struct xfs_bmbt_key {
+	__be64		br_startoff;	/* starting file offset */
+} xfs_bmbt_key_t, xfs_bmdr_key_t;
+
+/* btree pointer type */
+typedef __be64 xfs_bmbt_ptr_t, xfs_bmdr_ptr_t;
+
+
+/*
+ * Generic Btree block format definitions
+ *
+ * This is a combination of the actual format used on disk for short and long
+ * format btrees.  The first three fields are shared by both format, but the
+ * pointers are different and should be used with care.
+ *
+ * To get the size of the actual short or long form headers please use the size
+ * macros below.  Never use sizeof(xfs_btree_block).
+ *
+ * The blkno, crc, lsn, owner and uuid fields are only available in filesystems
+ * with the crc feature bit, and all accesses to them must be conditional on
+ * that flag.
+ */
+/* short form block header */
+struct xfs_btree_block_shdr {
+	__be32		bb_leftsib;
+	__be32		bb_rightsib;
+
+	__be64		bb_blkno;
+	__be64		bb_lsn;
+	uuid_t		bb_uuid;
+	__be32		bb_owner;
+	__le32		bb_crc;
+};
+
+/* long form block header */
+struct xfs_btree_block_lhdr {
+	__be64		bb_leftsib;
+	__be64		bb_rightsib;
+
+	__be64		bb_blkno;
+	__be64		bb_lsn;
+	uuid_t		bb_uuid;
+	__be64		bb_owner;
+	__le32		bb_crc;
+	__be32		bb_pad; /* padding for alignment */
+};
+
+struct xfs_btree_block {
+	__be32		bb_magic;	/* magic number for block type */
+	__be16		bb_level;	/* 0 is a leaf */
+	__be16		bb_numrecs;	/* current # of data records */
+	union {
+		struct xfs_btree_block_shdr s;
+		struct xfs_btree_block_lhdr l;
+	} bb_u;				/* rest */
+};
+
+/* size of a short form block */
+#define XFS_BTREE_SBLOCK_LEN \
+	(offsetof(struct xfs_btree_block, bb_u) + \
+	 offsetof(struct xfs_btree_block_shdr, bb_blkno))
+/* size of a long form block */
+#define XFS_BTREE_LBLOCK_LEN \
+	(offsetof(struct xfs_btree_block, bb_u) + \
+	 offsetof(struct xfs_btree_block_lhdr, bb_blkno))
+
+/* sizes of CRC enabled btree blocks */
+#define XFS_BTREE_SBLOCK_CRC_LEN \
+	(offsetof(struct xfs_btree_block, bb_u) + \
+	 sizeof(struct xfs_btree_block_shdr))
+#define XFS_BTREE_LBLOCK_CRC_LEN \
+	(offsetof(struct xfs_btree_block, bb_u) + \
+	 sizeof(struct xfs_btree_block_lhdr))
+
+#define XFS_BTREE_SBLOCK_CRC_OFF \
+	offsetof(struct xfs_btree_block, bb_u.s.bb_crc)
+#define XFS_BTREE_LBLOCK_CRC_OFF \
+	offsetof(struct xfs_btree_block, bb_u.l.bb_crc)
+
+/*
+ * On-disk XFS access control list structure.
+ */
+struct xfs_acl_entry {
+	__be32	ae_tag;
+	__be32	ae_id;
+	__be16	ae_perm;
+	__be16	ae_pad;		/* fill the implicit hole in the structure */
+};
+
+struct xfs_acl {
+	__be32			acl_cnt;
+	struct xfs_acl_entry	acl_entry[];
+};
+
+/*
+ * The number of ACL entries allowed is defined by the on-disk format.
+ * For v4 superblocks, that is limited to 25 entries. For v5 superblocks, it is
+ * limited only by the maximum size of the xattr that stores the information.
+ */
+#define XFS_ACL_MAX_ENTRIES(mp)	\
+	(xfs_has_crc(mp) \
+		?  (XFS_XATTR_SIZE_MAX - sizeof(struct xfs_acl)) / \
+						sizeof(struct xfs_acl_entry) \
+		: 25)
+
+#define XFS_ACL_SIZE(cnt) \
+	(sizeof(struct xfs_acl) + \
+		sizeof(struct xfs_acl_entry) * cnt)
+
+#define XFS_ACL_MAX_SIZE(mp) \
+	XFS_ACL_SIZE(XFS_ACL_MAX_ENTRIES((mp)))
+
+
+/* On-disk XFS extended attribute names */
+#define SGI_ACL_FILE		"SGI_ACL_FILE"
+#define SGI_ACL_DEFAULT		"SGI_ACL_DEFAULT"
+#define SGI_ACL_FILE_SIZE	(sizeof(SGI_ACL_FILE)-1)
+#define SGI_ACL_DEFAULT_SIZE	(sizeof(SGI_ACL_DEFAULT)-1)
+
+#endif /* __XFS_FORMAT_H__ */
diff --git a/fs/xfs/libxfs/xfs_fs.h b/fs/xfs/libxfs/xfs_fs.h
new file mode 100644
index 000000000..1cfd5bc65
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_fs.h
@@ -0,0 +1,851 @@
+/* SPDX-License-Identifier: LGPL-2.1 */
+/*
+ * Copyright (c) 1995-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_FS_H__
+#define __XFS_FS_H__
+
+/*
+ * SGI's XFS filesystem's major stuff (constants, structures)
+ */
+
+/*
+ * Direct I/O attribute record used with XFS_IOC_DIOINFO
+ * d_miniosz is the min xfer size, xfer size multiple and file seek offset
+ * alignment.
+ */
+#ifndef HAVE_DIOATTR
+struct dioattr {
+	__u32		d_mem;		/* data buffer memory alignment */
+	__u32		d_miniosz;	/* min xfer size		*/
+	__u32		d_maxiosz;	/* max xfer size		*/
+};
+#endif
+
+/*
+ * Structure for XFS_IOC_GETBMAP.
+ * On input, fill in bmv_offset and bmv_length of the first structure
+ * to indicate the area of interest in the file, and bmv_entries with
+ * the number of array elements given back.  The first structure is
+ * updated on return to give the offset and length for the next call.
+ */
+#ifndef HAVE_GETBMAP
+struct getbmap {
+	__s64		bmv_offset;	/* file offset of segment in blocks */
+	__s64		bmv_block;	/* starting block (64-bit daddr_t)  */
+	__s64		bmv_length;	/* length of segment, blocks	    */
+	__s32		bmv_count;	/* # of entries in array incl. 1st  */
+	__s32		bmv_entries;	/* # of entries filled in (output)  */
+};
+#endif
+
+/*
+ *	Structure for XFS_IOC_GETBMAPX.	 Fields bmv_offset through bmv_entries
+ *	are used exactly as in the getbmap structure.  The getbmapx structure
+ *	has additional bmv_iflags and bmv_oflags fields. The bmv_iflags field
+ *	is only used for the first structure.  It contains input flags
+ *	specifying XFS_IOC_GETBMAPX actions.  The bmv_oflags field is filled
+ *	in by the XFS_IOC_GETBMAPX command for each returned structure after
+ *	the first.
+ */
+#ifndef HAVE_GETBMAPX
+struct getbmapx {
+	__s64		bmv_offset;	/* file offset of segment in blocks */
+	__s64		bmv_block;	/* starting block (64-bit daddr_t)  */
+	__s64		bmv_length;	/* length of segment, blocks	    */
+	__s32		bmv_count;	/* # of entries in array incl. 1st  */
+	__s32		bmv_entries;	/* # of entries filled in (output). */
+	__s32		bmv_iflags;	/* input flags (1st structure)	    */
+	__s32		bmv_oflags;	/* output flags (after 1st structure)*/
+	__s32		bmv_unused1;	/* future use			    */
+	__s32		bmv_unused2;	/* future use			    */
+};
+#endif
+
+/*	bmv_iflags values - set by XFS_IOC_GETBMAPX caller.	*/
+#define BMV_IF_ATTRFORK		0x1	/* return attr fork rather than data */
+#define BMV_IF_NO_DMAPI_READ	0x2	/* Deprecated */
+#define BMV_IF_PREALLOC		0x4	/* rtn status BMV_OF_PREALLOC if req */
+#define BMV_IF_DELALLOC		0x8	/* rtn status BMV_OF_DELALLOC if req */
+#define BMV_IF_NO_HOLES		0x10	/* Do not return holes */
+#define BMV_IF_COWFORK		0x20	/* return CoW fork rather than data */
+#define BMV_IF_VALID	\
+	(BMV_IF_ATTRFORK|BMV_IF_NO_DMAPI_READ|BMV_IF_PREALLOC|	\
+	 BMV_IF_DELALLOC|BMV_IF_NO_HOLES|BMV_IF_COWFORK)
+
+/*	bmv_oflags values - returned for each non-header segment */
+#define BMV_OF_PREALLOC		0x1	/* segment = unwritten pre-allocation */
+#define BMV_OF_DELALLOC		0x2	/* segment = delayed allocation */
+#define BMV_OF_LAST		0x4	/* segment is the last in the file */
+#define BMV_OF_SHARED		0x8	/* segment shared with another file */
+
+/*	fmr_owner special values for FS_IOC_GETFSMAP */
+#define XFS_FMR_OWN_FREE	FMR_OWN_FREE      /* free space */
+#define XFS_FMR_OWN_UNKNOWN	FMR_OWN_UNKNOWN   /* unknown owner */
+#define XFS_FMR_OWN_FS		FMR_OWNER('X', 1) /* static fs metadata */
+#define XFS_FMR_OWN_LOG		FMR_OWNER('X', 2) /* journalling log */
+#define XFS_FMR_OWN_AG		FMR_OWNER('X', 3) /* per-AG metadata */
+#define XFS_FMR_OWN_INOBT	FMR_OWNER('X', 4) /* inode btree blocks */
+#define XFS_FMR_OWN_INODES	FMR_OWNER('X', 5) /* inodes */
+#define XFS_FMR_OWN_REFC	FMR_OWNER('X', 6) /* refcount tree */
+#define XFS_FMR_OWN_COW		FMR_OWNER('X', 7) /* cow staging */
+#define XFS_FMR_OWN_DEFECTIVE	FMR_OWNER('X', 8) /* bad blocks */
+
+/*
+ * File segment locking set data type for 64 bit access.
+ * Also used for all the RESV/FREE interfaces.
+ */
+typedef struct xfs_flock64 {
+	__s16		l_type;
+	__s16		l_whence;
+	__s64		l_start;
+	__s64		l_len;		/* len == 0 means until end of file */
+	__s32		l_sysid;
+	__u32		l_pid;
+	__s32		l_pad[4];	/* reserve area			    */
+} xfs_flock64_t;
+
+/*
+ * Output for XFS_IOC_FSGEOMETRY_V1
+ */
+struct xfs_fsop_geom_v1 {
+	__u32		blocksize;	/* filesystem (data) block size */
+	__u32		rtextsize;	/* realtime extent size		*/
+	__u32		agblocks;	/* fsblocks in an AG		*/
+	__u32		agcount;	/* number of allocation groups	*/
+	__u32		logblocks;	/* fsblocks in the log		*/
+	__u32		sectsize;	/* (data) sector size, bytes	*/
+	__u32		inodesize;	/* inode size in bytes		*/
+	__u32		imaxpct;	/* max allowed inode space(%)	*/
+	__u64		datablocks;	/* fsblocks in data subvolume	*/
+	__u64		rtblocks;	/* fsblocks in realtime subvol	*/
+	__u64		rtextents;	/* rt extents in realtime subvol*/
+	__u64		logstart;	/* starting fsblock of the log	*/
+	unsigned char	uuid[16];	/* unique id of the filesystem	*/
+	__u32		sunit;		/* stripe unit, fsblocks	*/
+	__u32		swidth;		/* stripe width, fsblocks	*/
+	__s32		version;	/* structure version		*/
+	__u32		flags;		/* superblock version flags	*/
+	__u32		logsectsize;	/* log sector size, bytes	*/
+	__u32		rtsectsize;	/* realtime sector size, bytes	*/
+	__u32		dirblocksize;	/* directory block size, bytes	*/
+};
+
+/*
+ * Output for XFS_IOC_FSGEOMETRY_V4
+ */
+struct xfs_fsop_geom_v4 {
+	__u32		blocksize;	/* filesystem (data) block size */
+	__u32		rtextsize;	/* realtime extent size		*/
+	__u32		agblocks;	/* fsblocks in an AG		*/
+	__u32		agcount;	/* number of allocation groups	*/
+	__u32		logblocks;	/* fsblocks in the log		*/
+	__u32		sectsize;	/* (data) sector size, bytes	*/
+	__u32		inodesize;	/* inode size in bytes		*/
+	__u32		imaxpct;	/* max allowed inode space(%)	*/
+	__u64		datablocks;	/* fsblocks in data subvolume	*/
+	__u64		rtblocks;	/* fsblocks in realtime subvol	*/
+	__u64		rtextents;	/* rt extents in realtime subvol*/
+	__u64		logstart;	/* starting fsblock of the log	*/
+	unsigned char	uuid[16];	/* unique id of the filesystem	*/
+	__u32		sunit;		/* stripe unit, fsblocks	*/
+	__u32		swidth;		/* stripe width, fsblocks	*/
+	__s32		version;	/* structure version		*/
+	__u32		flags;		/* superblock version flags	*/
+	__u32		logsectsize;	/* log sector size, bytes	*/
+	__u32		rtsectsize;	/* realtime sector size, bytes	*/
+	__u32		dirblocksize;	/* directory block size, bytes	*/
+	__u32		logsunit;	/* log stripe unit, bytes	*/
+};
+
+/*
+ * Output for XFS_IOC_FSGEOMETRY
+ */
+struct xfs_fsop_geom {
+	__u32		blocksize;	/* filesystem (data) block size */
+	__u32		rtextsize;	/* realtime extent size		*/
+	__u32		agblocks;	/* fsblocks in an AG		*/
+	__u32		agcount;	/* number of allocation groups	*/
+	__u32		logblocks;	/* fsblocks in the log		*/
+	__u32		sectsize;	/* (data) sector size, bytes	*/
+	__u32		inodesize;	/* inode size in bytes		*/
+	__u32		imaxpct;	/* max allowed inode space(%)	*/
+	__u64		datablocks;	/* fsblocks in data subvolume	*/
+	__u64		rtblocks;	/* fsblocks in realtime subvol	*/
+	__u64		rtextents;	/* rt extents in realtime subvol*/
+	__u64		logstart;	/* starting fsblock of the log	*/
+	unsigned char	uuid[16];	/* unique id of the filesystem	*/
+	__u32		sunit;		/* stripe unit, fsblocks	*/
+	__u32		swidth;		/* stripe width, fsblocks	*/
+	__s32		version;	/* structure version		*/
+	__u32		flags;		/* superblock version flags	*/
+	__u32		logsectsize;	/* log sector size, bytes	*/
+	__u32		rtsectsize;	/* realtime sector size, bytes	*/
+	__u32		dirblocksize;	/* directory block size, bytes	*/
+	__u32		logsunit;	/* log stripe unit, bytes	*/
+	uint32_t	sick;		/* o: unhealthy fs & rt metadata */
+	uint32_t	checked;	/* o: checked fs & rt metadata	*/
+	__u64		reserved[17];	/* reserved space		*/
+};
+
+#define XFS_FSOP_GEOM_SICK_COUNTERS	(1 << 0)  /* summary counters */
+#define XFS_FSOP_GEOM_SICK_UQUOTA	(1 << 1)  /* user quota */
+#define XFS_FSOP_GEOM_SICK_GQUOTA	(1 << 2)  /* group quota */
+#define XFS_FSOP_GEOM_SICK_PQUOTA	(1 << 3)  /* project quota */
+#define XFS_FSOP_GEOM_SICK_RT_BITMAP	(1 << 4)  /* realtime bitmap */
+#define XFS_FSOP_GEOM_SICK_RT_SUMMARY	(1 << 5)  /* realtime summary */
+
+/* Output for XFS_FS_COUNTS */
+typedef struct xfs_fsop_counts {
+	__u64	freedata;	/* free data section blocks */
+	__u64	freertx;	/* free rt extents */
+	__u64	freeino;	/* free inodes */
+	__u64	allocino;	/* total allocated inodes */
+} xfs_fsop_counts_t;
+
+/* Input/Output for XFS_GET_RESBLKS and XFS_SET_RESBLKS */
+typedef struct xfs_fsop_resblks {
+	__u64  resblks;
+	__u64  resblks_avail;
+} xfs_fsop_resblks_t;
+
+#define XFS_FSOP_GEOM_VERSION		0
+#define XFS_FSOP_GEOM_VERSION_V5	5
+
+#define XFS_FSOP_GEOM_FLAGS_ATTR	(1 << 0)  /* attributes in use	   */
+#define XFS_FSOP_GEOM_FLAGS_NLINK	(1 << 1)  /* 32-bit nlink values   */
+#define XFS_FSOP_GEOM_FLAGS_QUOTA	(1 << 2)  /* quotas enabled	   */
+#define XFS_FSOP_GEOM_FLAGS_IALIGN	(1 << 3)  /* inode alignment	   */
+#define XFS_FSOP_GEOM_FLAGS_DALIGN	(1 << 4)  /* large data alignment  */
+#define XFS_FSOP_GEOM_FLAGS_SHARED	(1 << 5)  /* read-only shared	   */
+#define XFS_FSOP_GEOM_FLAGS_EXTFLG	(1 << 6)  /* special extent flag   */
+#define XFS_FSOP_GEOM_FLAGS_DIRV2	(1 << 7)  /* directory version 2   */
+#define XFS_FSOP_GEOM_FLAGS_LOGV2	(1 << 8)  /* log format version 2  */
+#define XFS_FSOP_GEOM_FLAGS_SECTOR	(1 << 9)  /* sector sizes >1BB	   */
+#define XFS_FSOP_GEOM_FLAGS_ATTR2	(1 << 10) /* inline attributes rework */
+#define XFS_FSOP_GEOM_FLAGS_PROJID32	(1 << 11) /* 32-bit project IDs	   */
+#define XFS_FSOP_GEOM_FLAGS_DIRV2CI	(1 << 12) /* ASCII only CI names   */
+	/*  -- Do not use --		(1 << 13)    SGI parent pointers   */
+#define XFS_FSOP_GEOM_FLAGS_LAZYSB	(1 << 14) /* lazy superblock counters */
+#define XFS_FSOP_GEOM_FLAGS_V5SB	(1 << 15) /* version 5 superblock  */
+#define XFS_FSOP_GEOM_FLAGS_FTYPE	(1 << 16) /* inode directory types */
+#define XFS_FSOP_GEOM_FLAGS_FINOBT	(1 << 17) /* free inode btree	   */
+#define XFS_FSOP_GEOM_FLAGS_SPINODES	(1 << 18) /* sparse inode chunks   */
+#define XFS_FSOP_GEOM_FLAGS_RMAPBT	(1 << 19) /* reverse mapping btree */
+#define XFS_FSOP_GEOM_FLAGS_REFLINK	(1 << 20) /* files can share blocks */
+#define XFS_FSOP_GEOM_FLAGS_BIGTIME	(1 << 21) /* 64-bit nsec timestamps */
+#define XFS_FSOP_GEOM_FLAGS_INOBTCNT	(1 << 22) /* inobt btree counter */
+#define XFS_FSOP_GEOM_FLAGS_NREXT64	(1 << 23) /* large extent counters */
+
+/*
+ * Minimum and maximum sizes need for growth checks.
+ *
+ * Block counts are in units of filesystem blocks, not basic blocks.
+ */
+#define XFS_MIN_AG_BLOCKS	64
+#define XFS_MIN_LOG_BLOCKS	512ULL
+#define XFS_MAX_LOG_BLOCKS	(1024 * 1024ULL)
+#define XFS_MIN_LOG_BYTES	(10 * 1024 * 1024ULL)
+
+/*
+ * Limits on sb_agblocks/sb_agblklog -- mkfs won't format AGs smaller than
+ * 16MB or larger than 1TB.
+ */
+#define XFS_MIN_AG_BYTES	(1ULL << 24)	/* 16 MB */
+#define XFS_MAX_AG_BYTES	(1ULL << 40)	/* 1 TB */
+#define XFS_MAX_AG_BLOCKS	(XFS_MAX_AG_BYTES / XFS_MIN_BLOCKSIZE)
+#define XFS_MAX_CRC_AG_BLOCKS	(XFS_MAX_AG_BYTES / XFS_MIN_CRC_BLOCKSIZE)
+
+/* keep the maximum size under 2^31 by a small amount */
+#define XFS_MAX_LOG_BYTES \
+	((2 * 1024 * 1024 * 1024ULL) - XFS_MIN_LOG_BYTES)
+
+/* Used for sanity checks on superblock */
+#define XFS_MAX_DBLOCKS(s) ((xfs_rfsblock_t)(s)->sb_agcount * (s)->sb_agblocks)
+#define XFS_MIN_DBLOCKS(s) ((xfs_rfsblock_t)((s)->sb_agcount - 1) *	\
+			 (s)->sb_agblocks + XFS_MIN_AG_BLOCKS)
+
+/*
+ * Output for XFS_IOC_AG_GEOMETRY
+ */
+struct xfs_ag_geometry {
+	uint32_t	ag_number;	/* i/o: AG number */
+	uint32_t	ag_length;	/* o: length in blocks */
+	uint32_t	ag_freeblks;	/* o: free space */
+	uint32_t	ag_icount;	/* o: inodes allocated */
+	uint32_t	ag_ifree;	/* o: inodes free */
+	uint32_t	ag_sick;	/* o: sick things in ag */
+	uint32_t	ag_checked;	/* o: checked metadata in ag */
+	uint32_t	ag_flags;	/* i/o: flags for this ag */
+	uint64_t	ag_reserved[12];/* o: zero */
+};
+#define XFS_AG_GEOM_SICK_SB	(1 << 0)  /* superblock */
+#define XFS_AG_GEOM_SICK_AGF	(1 << 1)  /* AGF header */
+#define XFS_AG_GEOM_SICK_AGFL	(1 << 2)  /* AGFL header */
+#define XFS_AG_GEOM_SICK_AGI	(1 << 3)  /* AGI header */
+#define XFS_AG_GEOM_SICK_BNOBT	(1 << 4)  /* free space by block */
+#define XFS_AG_GEOM_SICK_CNTBT	(1 << 5)  /* free space by length */
+#define XFS_AG_GEOM_SICK_INOBT	(1 << 6)  /* inode index */
+#define XFS_AG_GEOM_SICK_FINOBT	(1 << 7)  /* free inode index */
+#define XFS_AG_GEOM_SICK_RMAPBT	(1 << 8)  /* reverse mappings */
+#define XFS_AG_GEOM_SICK_REFCNTBT (1 << 9)  /* reference counts */
+
+/*
+ * Structures for XFS_IOC_FSGROWFSDATA, XFS_IOC_FSGROWFSLOG & XFS_IOC_FSGROWFSRT
+ */
+typedef struct xfs_growfs_data {
+	__u64		newblocks;	/* new data subvol size, fsblocks */
+	__u32		imaxpct;	/* new inode space percentage limit */
+} xfs_growfs_data_t;
+
+typedef struct xfs_growfs_log {
+	__u32		newblocks;	/* new log size, fsblocks */
+	__u32		isint;		/* 1 if new log is internal */
+} xfs_growfs_log_t;
+
+typedef struct xfs_growfs_rt {
+	__u64		newblocks;	/* new realtime size, fsblocks */
+	__u32		extsize;	/* new realtime extent size, fsblocks */
+} xfs_growfs_rt_t;
+
+
+/*
+ * Structures returned from ioctl XFS_IOC_FSBULKSTAT & XFS_IOC_FSBULKSTAT_SINGLE
+ */
+typedef struct xfs_bstime {
+	__kernel_long_t tv_sec;		/* seconds		*/
+	__s32		tv_nsec;	/* and nanoseconds	*/
+} xfs_bstime_t;
+
+struct xfs_bstat {
+	__u64		bs_ino;		/* inode number			*/
+	__u16		bs_mode;	/* type and mode		*/
+	__u16		bs_nlink;	/* number of links		*/
+	__u32		bs_uid;		/* user id			*/
+	__u32		bs_gid;		/* group id			*/
+	__u32		bs_rdev;	/* device value			*/
+	__s32		bs_blksize;	/* block size			*/
+	__s64		bs_size;	/* file size			*/
+	xfs_bstime_t	bs_atime;	/* access time			*/
+	xfs_bstime_t	bs_mtime;	/* modify time			*/
+	xfs_bstime_t	bs_ctime;	/* inode change time		*/
+	int64_t		bs_blocks;	/* number of blocks		*/
+	__u32		bs_xflags;	/* extended flags		*/
+	__s32		bs_extsize;	/* extent size			*/
+	__s32		bs_extents;	/* number of extents		*/
+	__u32		bs_gen;		/* generation count		*/
+	__u16		bs_projid_lo;	/* lower part of project id	*/
+#define	bs_projid	bs_projid_lo	/* (previously just bs_projid)	*/
+	__u16		bs_forkoff;	/* inode fork offset in bytes	*/
+	__u16		bs_projid_hi;	/* higher part of project id	*/
+	uint16_t	bs_sick;	/* sick inode metadata		*/
+	uint16_t	bs_checked;	/* checked inode metadata	*/
+	unsigned char	bs_pad[2];	/* pad space, unused		*/
+	__u32		bs_cowextsize;	/* cow extent size		*/
+	__u32		bs_dmevmask;	/* DMIG event mask		*/
+	__u16		bs_dmstate;	/* DMIG state info		*/
+	__u16		bs_aextents;	/* attribute number of extents	*/
+};
+
+/* New bulkstat structure that reports v5 features and fixes padding issues */
+struct xfs_bulkstat {
+	uint64_t	bs_ino;		/* inode number			*/
+	uint64_t	bs_size;	/* file size			*/
+
+	uint64_t	bs_blocks;	/* number of blocks		*/
+	uint64_t	bs_xflags;	/* extended flags		*/
+
+	int64_t		bs_atime;	/* access time, seconds		*/
+	int64_t		bs_mtime;	/* modify time, seconds		*/
+
+	int64_t		bs_ctime;	/* inode change time, seconds	*/
+	int64_t		bs_btime;	/* creation time, seconds	*/
+
+	uint32_t	bs_gen;		/* generation count		*/
+	uint32_t	bs_uid;		/* user id			*/
+	uint32_t	bs_gid;		/* group id			*/
+	uint32_t	bs_projectid;	/* project id			*/
+
+	uint32_t	bs_atime_nsec;	/* access time, nanoseconds	*/
+	uint32_t	bs_mtime_nsec;	/* modify time, nanoseconds	*/
+	uint32_t	bs_ctime_nsec;	/* inode change time, nanoseconds */
+	uint32_t	bs_btime_nsec;	/* creation time, nanoseconds	*/
+
+	uint32_t	bs_blksize;	/* block size			*/
+	uint32_t	bs_rdev;	/* device value			*/
+	uint32_t	bs_cowextsize_blks; /* cow extent size hint, blocks */
+	uint32_t	bs_extsize_blks; /* extent size hint, blocks	*/
+
+	uint32_t	bs_nlink;	/* number of links		*/
+	uint32_t	bs_extents;	/* 32-bit data fork extent counter */
+	uint32_t	bs_aextents;	/* attribute number of extents	*/
+	uint16_t	bs_version;	/* structure version		*/
+	uint16_t	bs_forkoff;	/* inode fork offset in bytes	*/
+
+	uint16_t	bs_sick;	/* sick inode metadata		*/
+	uint16_t	bs_checked;	/* checked inode metadata	*/
+	uint16_t	bs_mode;	/* type and mode		*/
+	uint16_t	bs_pad2;	/* zeroed			*/
+	uint64_t	bs_extents64;	/* 64-bit data fork extent counter */
+
+	uint64_t	bs_pad[6];	/* zeroed			*/
+};
+
+#define XFS_BULKSTAT_VERSION_V1	(1)
+#define XFS_BULKSTAT_VERSION_V5	(5)
+
+/* bs_sick flags */
+#define XFS_BS_SICK_INODE	(1 << 0)  /* inode core */
+#define XFS_BS_SICK_BMBTD	(1 << 1)  /* data fork */
+#define XFS_BS_SICK_BMBTA	(1 << 2)  /* attr fork */
+#define XFS_BS_SICK_BMBTC	(1 << 3)  /* cow fork */
+#define XFS_BS_SICK_DIR		(1 << 4)  /* directory */
+#define XFS_BS_SICK_XATTR	(1 << 5)  /* extended attributes */
+#define XFS_BS_SICK_SYMLINK	(1 << 6)  /* symbolic link remote target */
+#define XFS_BS_SICK_PARENT	(1 << 7)  /* parent pointers */
+
+/*
+ * Project quota id helpers (previously projid was 16bit only
+ * and using two 16bit values to hold new 32bit projid was chosen
+ * to retain compatibility with "old" filesystems).
+ */
+static inline uint32_t
+bstat_get_projid(const struct xfs_bstat *bs)
+{
+	return (uint32_t)bs->bs_projid_hi << 16 | bs->bs_projid_lo;
+}
+
+/*
+ * The user-level BulkStat Request interface structure.
+ */
+struct xfs_fsop_bulkreq {
+	__u64		__user *lastip;	/* last inode # pointer		*/
+	__s32		icount;		/* count of entries in buffer	*/
+	void		__user *ubuffer;/* user buffer for inode desc.	*/
+	__s32		__user *ocount;	/* output count pointer		*/
+};
+
+/*
+ * Structures returned from xfs_inumbers routine (XFS_IOC_FSINUMBERS).
+ */
+struct xfs_inogrp {
+	__u64		xi_startino;	/* starting inode number	*/
+	__s32		xi_alloccount;	/* # bits set in allocmask	*/
+	__u64		xi_allocmask;	/* mask of allocated inodes	*/
+};
+
+/* New inumbers structure that reports v5 features and fixes padding issues */
+struct xfs_inumbers {
+	uint64_t	xi_startino;	/* starting inode number	*/
+	uint64_t	xi_allocmask;	/* mask of allocated inodes	*/
+	uint8_t		xi_alloccount;	/* # bits set in allocmask	*/
+	uint8_t		xi_version;	/* version			*/
+	uint8_t		xi_padding[6];	/* zero				*/
+};
+
+#define XFS_INUMBERS_VERSION_V1	(1)
+#define XFS_INUMBERS_VERSION_V5	(5)
+
+/* Header for bulk inode requests. */
+struct xfs_bulk_ireq {
+	uint64_t	ino;		/* I/O: start with this inode	*/
+	uint32_t	flags;		/* I/O: operation flags		*/
+	uint32_t	icount;		/* I: count of entries in buffer */
+	uint32_t	ocount;		/* O: count of entries filled out */
+	uint32_t	agno;		/* I: see comment for IREQ_AGNO	*/
+	uint64_t	reserved[5];	/* must be zero			*/
+};
+
+/*
+ * Only return results from the specified @agno.  If @ino is zero, start
+ * with the first inode of @agno.
+ */
+#define XFS_BULK_IREQ_AGNO	(1U << 0)
+
+/*
+ * Return bulkstat information for a single inode, where @ino value is a
+ * special value, not a literal inode number.  See the XFS_BULK_IREQ_SPECIAL_*
+ * values below.  Not compatible with XFS_BULK_IREQ_AGNO.
+ */
+#define XFS_BULK_IREQ_SPECIAL	(1U << 1)
+
+/*
+ * Return data fork extent count via xfs_bulkstat->bs_extents64 field and assign
+ * 0 to xfs_bulkstat->bs_extents when the flag is set.  Otherwise, use
+ * xfs_bulkstat->bs_extents for returning data fork extent count and set
+ * xfs_bulkstat->bs_extents64 to 0. In the second case, return -EOVERFLOW and
+ * assign 0 to xfs_bulkstat->bs_extents if data fork extent count is larger than
+ * XFS_MAX_EXTCNT_DATA_FORK_OLD.
+ */
+#define XFS_BULK_IREQ_NREXT64	(1U << 2)
+
+#define XFS_BULK_IREQ_FLAGS_ALL	(XFS_BULK_IREQ_AGNO |	 \
+				 XFS_BULK_IREQ_SPECIAL | \
+				 XFS_BULK_IREQ_NREXT64)
+
+/* Operate on the root directory inode. */
+#define XFS_BULK_IREQ_SPECIAL_ROOT	(1)
+
+/*
+ * ioctl structures for v5 bulkstat and inumbers requests
+ */
+struct xfs_bulkstat_req {
+	struct xfs_bulk_ireq	hdr;
+	struct xfs_bulkstat	bulkstat[];
+};
+#define XFS_BULKSTAT_REQ_SIZE(nr)	(sizeof(struct xfs_bulkstat_req) + \
+					 (nr) * sizeof(struct xfs_bulkstat))
+
+struct xfs_inumbers_req {
+	struct xfs_bulk_ireq	hdr;
+	struct xfs_inumbers	inumbers[];
+};
+#define XFS_INUMBERS_REQ_SIZE(nr)	(sizeof(struct xfs_inumbers_req) + \
+					 (nr) * sizeof(struct xfs_inumbers))
+
+/*
+ * Error injection.
+ */
+typedef struct xfs_error_injection {
+	__s32		fd;
+	__s32		errtag;
+} xfs_error_injection_t;
+
+
+/*
+ * Speculative preallocation trimming.
+ */
+#define XFS_EOFBLOCKS_VERSION		1
+struct xfs_fs_eofblocks {
+	__u32		eof_version;
+	__u32		eof_flags;
+	uid_t		eof_uid;
+	gid_t		eof_gid;
+	prid_t		eof_prid;
+	__u32		pad32;
+	__u64		eof_min_file_size;
+	__u64		pad64[12];
+};
+
+/* eof_flags values */
+#define XFS_EOF_FLAGS_SYNC		(1 << 0) /* sync/wait mode scan */
+#define XFS_EOF_FLAGS_UID		(1 << 1) /* filter by uid */
+#define XFS_EOF_FLAGS_GID		(1 << 2) /* filter by gid */
+#define XFS_EOF_FLAGS_PRID		(1 << 3) /* filter by project id */
+#define XFS_EOF_FLAGS_MINFILESIZE	(1 << 4) /* filter by min file size */
+#define XFS_EOF_FLAGS_UNION		(1 << 5) /* union filter algorithm;
+						  * kernel only, not included in
+						  * valid mask */
+#define XFS_EOF_FLAGS_VALID	\
+	(XFS_EOF_FLAGS_SYNC |	\
+	 XFS_EOF_FLAGS_UID |	\
+	 XFS_EOF_FLAGS_GID |	\
+	 XFS_EOF_FLAGS_PRID |	\
+	 XFS_EOF_FLAGS_MINFILESIZE)
+
+
+/*
+ * The user-level Handle Request interface structure.
+ */
+typedef struct xfs_fsop_handlereq {
+	__u32		fd;		/* fd for FD_TO_HANDLE		*/
+	void		__user *path;	/* user pathname		*/
+	__u32		oflags;		/* open flags			*/
+	void		__user *ihandle;/* user supplied handle		*/
+	__u32		ihandlen;	/* user supplied length		*/
+	void		__user *ohandle;/* user buffer for handle	*/
+	__u32		__user *ohandlen;/* user buffer length		*/
+} xfs_fsop_handlereq_t;
+
+/*
+ * Compound structures for passing args through Handle Request interfaces
+ * xfs_attrlist_by_handle, xfs_attrmulti_by_handle
+ * - ioctls: XFS_IOC_ATTRLIST_BY_HANDLE, and XFS_IOC_ATTRMULTI_BY_HANDLE
+ */
+
+/*
+ * Flags passed in xfs_attr_multiop.am_flags for the attr ioctl interface.
+ *
+ * NOTE: Must match the values declared in libattr without the XFS_IOC_ prefix.
+ */
+#define XFS_IOC_ATTR_ROOT	0x0002	/* use attrs in root namespace */
+#define XFS_IOC_ATTR_SECURE	0x0008	/* use attrs in security namespace */
+#define XFS_IOC_ATTR_CREATE	0x0010	/* fail if attr already exists */
+#define XFS_IOC_ATTR_REPLACE	0x0020	/* fail if attr does not exist */
+
+typedef struct xfs_attrlist_cursor {
+	__u32		opaque[4];
+} xfs_attrlist_cursor_t;
+
+/*
+ * Define how lists of attribute names are returned to userspace from the
+ * XFS_IOC_ATTRLIST_BY_HANDLE ioctl.  struct xfs_attrlist is the header at the
+ * beginning of the returned buffer, and a each entry in al_offset contains the
+ * relative offset of an xfs_attrlist_ent containing the actual entry.
+ *
+ * NOTE: struct xfs_attrlist must match struct attrlist defined in libattr, and
+ * struct xfs_attrlist_ent must match struct attrlist_ent defined in libattr.
+ */
+struct xfs_attrlist {
+	__s32	al_count;	/* number of entries in attrlist */
+	__s32	al_more;	/* T/F: more attrs (do call again) */
+	__s32	al_offset[1];	/* byte offsets of attrs [var-sized] */
+};
+
+struct xfs_attrlist_ent {	/* data from attr_list() */
+	__u32	a_valuelen;	/* number bytes in value of attr */
+	char	a_name[1];	/* attr name (NULL terminated) */
+};
+
+typedef struct xfs_fsop_attrlist_handlereq {
+	struct xfs_fsop_handlereq	hreq; /* handle interface structure */
+	struct xfs_attrlist_cursor	pos; /* opaque cookie, list offset */
+	__u32				flags;	/* which namespace to use */
+	__u32				buflen;	/* length of buffer supplied */
+	void				__user *buffer;	/* returned names */
+} xfs_fsop_attrlist_handlereq_t;
+
+typedef struct xfs_attr_multiop {
+	__u32		am_opcode;
+#define ATTR_OP_GET	1	/* return the indicated attr's value */
+#define ATTR_OP_SET	2	/* set/create the indicated attr/value pair */
+#define ATTR_OP_REMOVE	3	/* remove the indicated attr */
+	__s32		am_error;
+	void		__user *am_attrname;
+	void		__user *am_attrvalue;
+	__u32		am_length;
+	__u32		am_flags; /* XFS_IOC_ATTR_* */
+} xfs_attr_multiop_t;
+
+typedef struct xfs_fsop_attrmulti_handlereq {
+	struct xfs_fsop_handlereq	hreq; /* handle interface structure */
+	__u32				opcount;/* count of following multiop */
+	struct xfs_attr_multiop		__user *ops; /* attr_multi data */
+} xfs_fsop_attrmulti_handlereq_t;
+
+/*
+ * per machine unique filesystem identifier types.
+ */
+typedef struct { __u32 val[2]; } xfs_fsid_t; /* file system id type */
+
+typedef struct xfs_fid {
+	__u16	fid_len;		/* length of remainder	*/
+	__u16	fid_pad;
+	__u32	fid_gen;		/* generation number	*/
+	__u64	fid_ino;		/* 64 bits inode number */
+} xfs_fid_t;
+
+typedef struct xfs_handle {
+	union {
+		__s64	    align;	/* force alignment of ha_fid	 */
+		xfs_fsid_t  _ha_fsid;	/* unique file system identifier */
+	} ha_u;
+	xfs_fid_t	ha_fid;		/* file system specific file ID	 */
+} xfs_handle_t;
+#define ha_fsid ha_u._ha_fsid
+
+/*
+ * Structure passed to XFS_IOC_SWAPEXT
+ */
+typedef struct xfs_swapext
+{
+	int64_t		sx_version;	/* version */
+#define XFS_SX_VERSION		0
+	int64_t		sx_fdtarget;	/* fd of target file */
+	int64_t		sx_fdtmp;	/* fd of tmp file */
+	xfs_off_t	sx_offset;	/* offset into file */
+	xfs_off_t	sx_length;	/* leng from offset */
+	char		sx_pad[16];	/* pad space, unused */
+	struct xfs_bstat sx_stat;	/* stat of target b4 copy */
+} xfs_swapext_t;
+
+/*
+ * Flags for going down operation
+ */
+#define XFS_FSOP_GOING_FLAGS_DEFAULT		0x0	/* going down */
+#define XFS_FSOP_GOING_FLAGS_LOGFLUSH		0x1	/* flush log but not data */
+#define XFS_FSOP_GOING_FLAGS_NOLOGFLUSH		0x2	/* don't flush log nor data */
+
+/* metadata scrubbing */
+struct xfs_scrub_metadata {
+	__u32 sm_type;		/* What to check? */
+	__u32 sm_flags;		/* flags; see below. */
+	__u64 sm_ino;		/* inode number. */
+	__u32 sm_gen;		/* inode generation. */
+	__u32 sm_agno;		/* ag number. */
+	__u64 sm_reserved[5];	/* pad to 64 bytes */
+};
+
+/*
+ * Metadata types and flags for scrub operation.
+ */
+
+/* Scrub subcommands. */
+#define XFS_SCRUB_TYPE_PROBE	0	/* presence test ioctl */
+#define XFS_SCRUB_TYPE_SB	1	/* superblock */
+#define XFS_SCRUB_TYPE_AGF	2	/* AG free header */
+#define XFS_SCRUB_TYPE_AGFL	3	/* AG free list */
+#define XFS_SCRUB_TYPE_AGI	4	/* AG inode header */
+#define XFS_SCRUB_TYPE_BNOBT	5	/* freesp by block btree */
+#define XFS_SCRUB_TYPE_CNTBT	6	/* freesp by length btree */
+#define XFS_SCRUB_TYPE_INOBT	7	/* inode btree */
+#define XFS_SCRUB_TYPE_FINOBT	8	/* free inode btree */
+#define XFS_SCRUB_TYPE_RMAPBT	9	/* reverse mapping btree */
+#define XFS_SCRUB_TYPE_REFCNTBT	10	/* reference count btree */
+#define XFS_SCRUB_TYPE_INODE	11	/* inode record */
+#define XFS_SCRUB_TYPE_BMBTD	12	/* data fork block mapping */
+#define XFS_SCRUB_TYPE_BMBTA	13	/* attr fork block mapping */
+#define XFS_SCRUB_TYPE_BMBTC	14	/* CoW fork block mapping */
+#define XFS_SCRUB_TYPE_DIR	15	/* directory */
+#define XFS_SCRUB_TYPE_XATTR	16	/* extended attribute */
+#define XFS_SCRUB_TYPE_SYMLINK	17	/* symbolic link */
+#define XFS_SCRUB_TYPE_PARENT	18	/* parent pointers */
+#define XFS_SCRUB_TYPE_RTBITMAP	19	/* realtime bitmap */
+#define XFS_SCRUB_TYPE_RTSUM	20	/* realtime summary */
+#define XFS_SCRUB_TYPE_UQUOTA	21	/* user quotas */
+#define XFS_SCRUB_TYPE_GQUOTA	22	/* group quotas */
+#define XFS_SCRUB_TYPE_PQUOTA	23	/* project quotas */
+#define XFS_SCRUB_TYPE_FSCOUNTERS 24	/* fs summary counters */
+
+/* Number of scrub subcommands. */
+#define XFS_SCRUB_TYPE_NR	25
+
+/* i: Repair this metadata. */
+#define XFS_SCRUB_IFLAG_REPAIR		(1u << 0)
+
+/* o: Metadata object needs repair. */
+#define XFS_SCRUB_OFLAG_CORRUPT		(1u << 1)
+
+/*
+ * o: Metadata object could be optimized.  It's not corrupt, but
+ *    we could improve on it somehow.
+ */
+#define XFS_SCRUB_OFLAG_PREEN		(1u << 2)
+
+/* o: Cross-referencing failed. */
+#define XFS_SCRUB_OFLAG_XFAIL		(1u << 3)
+
+/* o: Metadata object disagrees with cross-referenced metadata. */
+#define XFS_SCRUB_OFLAG_XCORRUPT	(1u << 4)
+
+/* o: Scan was not complete. */
+#define XFS_SCRUB_OFLAG_INCOMPLETE	(1u << 5)
+
+/* o: Metadata object looked funny but isn't corrupt. */
+#define XFS_SCRUB_OFLAG_WARNING		(1u << 6)
+
+/*
+ * o: IFLAG_REPAIR was set but metadata object did not need fixing or
+ *    optimization and has therefore not been altered.
+ */
+#define XFS_SCRUB_OFLAG_NO_REPAIR_NEEDED (1u << 7)
+
+#define XFS_SCRUB_FLAGS_IN	(XFS_SCRUB_IFLAG_REPAIR)
+#define XFS_SCRUB_FLAGS_OUT	(XFS_SCRUB_OFLAG_CORRUPT | \
+				 XFS_SCRUB_OFLAG_PREEN | \
+				 XFS_SCRUB_OFLAG_XFAIL | \
+				 XFS_SCRUB_OFLAG_XCORRUPT | \
+				 XFS_SCRUB_OFLAG_INCOMPLETE | \
+				 XFS_SCRUB_OFLAG_WARNING | \
+				 XFS_SCRUB_OFLAG_NO_REPAIR_NEEDED)
+#define XFS_SCRUB_FLAGS_ALL	(XFS_SCRUB_FLAGS_IN | XFS_SCRUB_FLAGS_OUT)
+
+/*
+ * ioctl limits
+ */
+#ifdef XATTR_LIST_MAX
+#  define XFS_XATTR_LIST_MAX XATTR_LIST_MAX
+#else
+#  define XFS_XATTR_LIST_MAX 65536
+#endif
+
+
+/*
+ * ioctl commands that are used by Linux filesystems
+ */
+#define XFS_IOC_GETXFLAGS	FS_IOC_GETFLAGS
+#define XFS_IOC_SETXFLAGS	FS_IOC_SETFLAGS
+#define XFS_IOC_GETVERSION	FS_IOC_GETVERSION
+
+/*
+ * ioctl commands that replace IRIX fcntl()'s
+ * For 'documentation' purposed more than anything else,
+ * the "cmd #" field reflects the IRIX fcntl number.
+ */
+/*	XFS_IOC_ALLOCSP ------- deprecated 10	 */
+/*	XFS_IOC_FREESP -------- deprecated 11	 */
+#define XFS_IOC_DIOINFO		_IOR ('X', 30, struct dioattr)
+#define XFS_IOC_FSGETXATTR	FS_IOC_FSGETXATTR
+#define XFS_IOC_FSSETXATTR	FS_IOC_FSSETXATTR
+/*	XFS_IOC_ALLOCSP64 ----- deprecated 36	 */
+/*	XFS_IOC_FREESP64 ------ deprecated 37	 */
+#define XFS_IOC_GETBMAP		_IOWR('X', 38, struct getbmap)
+/*      XFS_IOC_FSSETDM ------- deprecated 39    */
+#define XFS_IOC_RESVSP		_IOW ('X', 40, struct xfs_flock64)
+#define XFS_IOC_UNRESVSP	_IOW ('X', 41, struct xfs_flock64)
+#define XFS_IOC_RESVSP64	_IOW ('X', 42, struct xfs_flock64)
+#define XFS_IOC_UNRESVSP64	_IOW ('X', 43, struct xfs_flock64)
+#define XFS_IOC_GETBMAPA	_IOWR('X', 44, struct getbmap)
+#define XFS_IOC_FSGETXATTRA	_IOR ('X', 45, struct fsxattr)
+/*	XFS_IOC_SETBIOSIZE ---- deprecated 46	   */
+/*	XFS_IOC_GETBIOSIZE ---- deprecated 47	   */
+#define XFS_IOC_GETBMAPX	_IOWR('X', 56, struct getbmap)
+#define XFS_IOC_ZERO_RANGE	_IOW ('X', 57, struct xfs_flock64)
+#define XFS_IOC_FREE_EOFBLOCKS	_IOR ('X', 58, struct xfs_fs_eofblocks)
+/*	XFS_IOC_GETFSMAP ------ hoisted 59         */
+#define XFS_IOC_SCRUB_METADATA	_IOWR('X', 60, struct xfs_scrub_metadata)
+#define XFS_IOC_AG_GEOMETRY	_IOWR('X', 61, struct xfs_ag_geometry)
+
+/*
+ * ioctl commands that replace IRIX syssgi()'s
+ */
+#define XFS_IOC_FSGEOMETRY_V1	     _IOR ('X', 100, struct xfs_fsop_geom_v1)
+#define XFS_IOC_FSBULKSTAT	     _IOWR('X', 101, struct xfs_fsop_bulkreq)
+#define XFS_IOC_FSBULKSTAT_SINGLE    _IOWR('X', 102, struct xfs_fsop_bulkreq)
+#define XFS_IOC_FSINUMBERS	     _IOWR('X', 103, struct xfs_fsop_bulkreq)
+#define XFS_IOC_PATH_TO_FSHANDLE     _IOWR('X', 104, struct xfs_fsop_handlereq)
+#define XFS_IOC_PATH_TO_HANDLE	     _IOWR('X', 105, struct xfs_fsop_handlereq)
+#define XFS_IOC_FD_TO_HANDLE	     _IOWR('X', 106, struct xfs_fsop_handlereq)
+#define XFS_IOC_OPEN_BY_HANDLE	     _IOWR('X', 107, struct xfs_fsop_handlereq)
+#define XFS_IOC_READLINK_BY_HANDLE   _IOWR('X', 108, struct xfs_fsop_handlereq)
+#define XFS_IOC_SWAPEXT		     _IOWR('X', 109, struct xfs_swapext)
+#define XFS_IOC_FSGROWFSDATA	     _IOW ('X', 110, struct xfs_growfs_data)
+#define XFS_IOC_FSGROWFSLOG	     _IOW ('X', 111, struct xfs_growfs_log)
+#define XFS_IOC_FSGROWFSRT	     _IOW ('X', 112, struct xfs_growfs_rt)
+#define XFS_IOC_FSCOUNTS	     _IOR ('X', 113, struct xfs_fsop_counts)
+#define XFS_IOC_SET_RESBLKS	     _IOWR('X', 114, struct xfs_fsop_resblks)
+#define XFS_IOC_GET_RESBLKS	     _IOR ('X', 115, struct xfs_fsop_resblks)
+#define XFS_IOC_ERROR_INJECTION	     _IOW ('X', 116, struct xfs_error_injection)
+#define XFS_IOC_ERROR_CLEARALL	     _IOW ('X', 117, struct xfs_error_injection)
+/*	XFS_IOC_ATTRCTL_BY_HANDLE -- deprecated 118	 */
+
+#define XFS_IOC_FREEZE		     _IOWR('X', 119, int)	/* aka FIFREEZE */
+#define XFS_IOC_THAW		     _IOWR('X', 120, int)	/* aka FITHAW */
+
+/*      XFS_IOC_FSSETDM_BY_HANDLE -- deprecated 121      */
+#define XFS_IOC_ATTRLIST_BY_HANDLE   _IOW ('X', 122, struct xfs_fsop_attrlist_handlereq)
+#define XFS_IOC_ATTRMULTI_BY_HANDLE  _IOW ('X', 123, struct xfs_fsop_attrmulti_handlereq)
+#define XFS_IOC_FSGEOMETRY_V4	     _IOR ('X', 124, struct xfs_fsop_geom_v4)
+#define XFS_IOC_GOINGDOWN	     _IOR ('X', 125, uint32_t)
+#define XFS_IOC_FSGEOMETRY	     _IOR ('X', 126, struct xfs_fsop_geom)
+#define XFS_IOC_BULKSTAT	     _IOR ('X', 127, struct xfs_bulkstat_req)
+#define XFS_IOC_INUMBERS	     _IOR ('X', 128, struct xfs_inumbers_req)
+/*	XFS_IOC_GETFSUUID ---------- deprecated 140	 */
+
+
+#ifndef HAVE_BBMACROS
+/*
+ * Block I/O parameterization.	A basic block (BB) is the lowest size of
+ * filesystem allocation, and must equal 512.  Length units given to bio
+ * routines are in BB's.
+ */
+#define BBSHIFT		9
+#define BBSIZE		(1<<BBSHIFT)
+#define BBMASK		(BBSIZE-1)
+#define BTOBB(bytes)	(((__u64)(bytes) + BBSIZE - 1) >> BBSHIFT)
+#define BTOBBT(bytes)	((__u64)(bytes) >> BBSHIFT)
+#define BBTOB(bbs)	((bbs) << BBSHIFT)
+#endif
+
+#endif	/* __XFS_FS_H__ */
diff --git a/fs/xfs/libxfs/xfs_health.h b/fs/xfs/libxfs/xfs_health.h
new file mode 100644
index 000000000..99e796256
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_health.h
@@ -0,0 +1,190 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Copyright (C) 2019 Oracle.  All Rights Reserved.
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ */
+#ifndef __XFS_HEALTH_H__
+#define __XFS_HEALTH_H__
+
+/*
+ * In-Core Filesystem Health Assessments
+ * =====================================
+ *
+ * We'd like to be able to summarize the current health status of the
+ * filesystem so that the administrator knows when it's necessary to schedule
+ * some downtime for repairs.  Until then, we would also like to avoid abrupt
+ * shutdowns due to corrupt metadata.
+ *
+ * The online scrub feature evaluates the health of all filesystem metadata.
+ * When scrub detects corruption in a piece of metadata it will set the
+ * corresponding sickness flag, and repair will clear it if successful.  If
+ * problems remain at unmount time, we can also request manual intervention by
+ * logging a notice to run xfs_repair.
+ *
+ * Each health tracking group uses a pair of fields for reporting.  The
+ * "checked" field tell us if a given piece of metadata has ever been examined,
+ * and the "sick" field tells us if that piece was found to need repairs.
+ * Therefore we can conclude that for a given sick flag value:
+ *
+ *  - checked && sick  => metadata needs repair
+ *  - checked && !sick => metadata is ok
+ *  - !checked         => has not been examined since mount
+ */
+
+struct xfs_mount;
+struct xfs_perag;
+struct xfs_inode;
+struct xfs_fsop_geom;
+
+/* Observable health issues for metadata spanning the entire filesystem. */
+#define XFS_SICK_FS_COUNTERS	(1 << 0)  /* summary counters */
+#define XFS_SICK_FS_UQUOTA	(1 << 1)  /* user quota */
+#define XFS_SICK_FS_GQUOTA	(1 << 2)  /* group quota */
+#define XFS_SICK_FS_PQUOTA	(1 << 3)  /* project quota */
+
+/* Observable health issues for realtime volume metadata. */
+#define XFS_SICK_RT_BITMAP	(1 << 0)  /* realtime bitmap */
+#define XFS_SICK_RT_SUMMARY	(1 << 1)  /* realtime summary */
+
+/* Observable health issues for AG metadata. */
+#define XFS_SICK_AG_SB		(1 << 0)  /* superblock */
+#define XFS_SICK_AG_AGF		(1 << 1)  /* AGF header */
+#define XFS_SICK_AG_AGFL	(1 << 2)  /* AGFL header */
+#define XFS_SICK_AG_AGI		(1 << 3)  /* AGI header */
+#define XFS_SICK_AG_BNOBT	(1 << 4)  /* free space by block */
+#define XFS_SICK_AG_CNTBT	(1 << 5)  /* free space by length */
+#define XFS_SICK_AG_INOBT	(1 << 6)  /* inode index */
+#define XFS_SICK_AG_FINOBT	(1 << 7)  /* free inode index */
+#define XFS_SICK_AG_RMAPBT	(1 << 8)  /* reverse mappings */
+#define XFS_SICK_AG_REFCNTBT	(1 << 9)  /* reference counts */
+
+/* Observable health issues for inode metadata. */
+#define XFS_SICK_INO_CORE	(1 << 0)  /* inode core */
+#define XFS_SICK_INO_BMBTD	(1 << 1)  /* data fork */
+#define XFS_SICK_INO_BMBTA	(1 << 2)  /* attr fork */
+#define XFS_SICK_INO_BMBTC	(1 << 3)  /* cow fork */
+#define XFS_SICK_INO_DIR	(1 << 4)  /* directory */
+#define XFS_SICK_INO_XATTR	(1 << 5)  /* extended attributes */
+#define XFS_SICK_INO_SYMLINK	(1 << 6)  /* symbolic link remote target */
+#define XFS_SICK_INO_PARENT	(1 << 7)  /* parent pointers */
+
+/* Primary evidence of health problems in a given group. */
+#define XFS_SICK_FS_PRIMARY	(XFS_SICK_FS_COUNTERS | \
+				 XFS_SICK_FS_UQUOTA | \
+				 XFS_SICK_FS_GQUOTA | \
+				 XFS_SICK_FS_PQUOTA)
+
+#define XFS_SICK_RT_PRIMARY	(XFS_SICK_RT_BITMAP | \
+				 XFS_SICK_RT_SUMMARY)
+
+#define XFS_SICK_AG_PRIMARY	(XFS_SICK_AG_SB | \
+				 XFS_SICK_AG_AGF | \
+				 XFS_SICK_AG_AGFL | \
+				 XFS_SICK_AG_AGI | \
+				 XFS_SICK_AG_BNOBT | \
+				 XFS_SICK_AG_CNTBT | \
+				 XFS_SICK_AG_INOBT | \
+				 XFS_SICK_AG_FINOBT | \
+				 XFS_SICK_AG_RMAPBT | \
+				 XFS_SICK_AG_REFCNTBT)
+
+#define XFS_SICK_INO_PRIMARY	(XFS_SICK_INO_CORE | \
+				 XFS_SICK_INO_BMBTD | \
+				 XFS_SICK_INO_BMBTA | \
+				 XFS_SICK_INO_BMBTC | \
+				 XFS_SICK_INO_DIR | \
+				 XFS_SICK_INO_XATTR | \
+				 XFS_SICK_INO_SYMLINK | \
+				 XFS_SICK_INO_PARENT)
+
+/* These functions must be provided by the xfs implementation. */
+
+void xfs_fs_mark_sick(struct xfs_mount *mp, unsigned int mask);
+void xfs_fs_mark_healthy(struct xfs_mount *mp, unsigned int mask);
+void xfs_fs_measure_sickness(struct xfs_mount *mp, unsigned int *sick,
+		unsigned int *checked);
+
+void xfs_rt_mark_sick(struct xfs_mount *mp, unsigned int mask);
+void xfs_rt_mark_healthy(struct xfs_mount *mp, unsigned int mask);
+void xfs_rt_measure_sickness(struct xfs_mount *mp, unsigned int *sick,
+		unsigned int *checked);
+
+void xfs_ag_mark_sick(struct xfs_perag *pag, unsigned int mask);
+void xfs_ag_mark_healthy(struct xfs_perag *pag, unsigned int mask);
+void xfs_ag_measure_sickness(struct xfs_perag *pag, unsigned int *sick,
+		unsigned int *checked);
+
+void xfs_inode_mark_sick(struct xfs_inode *ip, unsigned int mask);
+void xfs_inode_mark_healthy(struct xfs_inode *ip, unsigned int mask);
+void xfs_inode_measure_sickness(struct xfs_inode *ip, unsigned int *sick,
+		unsigned int *checked);
+
+void xfs_health_unmount(struct xfs_mount *mp);
+
+/* Now some helpers. */
+
+static inline bool
+xfs_fs_has_sickness(struct xfs_mount *mp, unsigned int mask)
+{
+	unsigned int	sick, checked;
+
+	xfs_fs_measure_sickness(mp, &sick, &checked);
+	return sick & mask;
+}
+
+static inline bool
+xfs_rt_has_sickness(struct xfs_mount *mp, unsigned int mask)
+{
+	unsigned int	sick, checked;
+
+	xfs_rt_measure_sickness(mp, &sick, &checked);
+	return sick & mask;
+}
+
+static inline bool
+xfs_ag_has_sickness(struct xfs_perag *pag, unsigned int mask)
+{
+	unsigned int	sick, checked;
+
+	xfs_ag_measure_sickness(pag, &sick, &checked);
+	return sick & mask;
+}
+
+static inline bool
+xfs_inode_has_sickness(struct xfs_inode *ip, unsigned int mask)
+{
+	unsigned int	sick, checked;
+
+	xfs_inode_measure_sickness(ip, &sick, &checked);
+	return sick & mask;
+}
+
+static inline bool
+xfs_fs_is_healthy(struct xfs_mount *mp)
+{
+	return !xfs_fs_has_sickness(mp, -1U);
+}
+
+static inline bool
+xfs_rt_is_healthy(struct xfs_mount *mp)
+{
+	return !xfs_rt_has_sickness(mp, -1U);
+}
+
+static inline bool
+xfs_ag_is_healthy(struct xfs_perag *pag)
+{
+	return !xfs_ag_has_sickness(pag, -1U);
+}
+
+static inline bool
+xfs_inode_is_healthy(struct xfs_inode *ip)
+{
+	return !xfs_inode_has_sickness(ip, -1U);
+}
+
+void xfs_fsop_geom_health(struct xfs_mount *mp, struct xfs_fsop_geom *geo);
+void xfs_ag_geom_health(struct xfs_perag *pag, struct xfs_ag_geometry *ageo);
+void xfs_bulkstat_health(struct xfs_inode *ip, struct xfs_bulkstat *bs);
+
+#endif	/* __XFS_HEALTH_H__ */
diff --git a/fs/xfs/libxfs/xfs_ialloc.c b/fs/xfs/libxfs/xfs_ialloc.c
new file mode 100644
index 000000000..94db50eb7
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_ialloc.c
@@ -0,0 +1,2969 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_bit.h"
+#include "xfs_mount.h"
+#include "xfs_inode.h"
+#include "xfs_btree.h"
+#include "xfs_ialloc.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_alloc.h"
+#include "xfs_errortag.h"
+#include "xfs_error.h"
+#include "xfs_bmap.h"
+#include "xfs_trans.h"
+#include "xfs_buf_item.h"
+#include "xfs_icreate_item.h"
+#include "xfs_icache.h"
+#include "xfs_trace.h"
+#include "xfs_log.h"
+#include "xfs_rmap.h"
+#include "xfs_ag.h"
+
+/*
+ * Lookup a record by ino in the btree given by cur.
+ */
+int					/* error */
+xfs_inobt_lookup(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agino_t		ino,	/* starting inode of chunk */
+	xfs_lookup_t		dir,	/* <=, >=, == */
+	int			*stat)	/* success/failure */
+{
+	cur->bc_rec.i.ir_startino = ino;
+	cur->bc_rec.i.ir_holemask = 0;
+	cur->bc_rec.i.ir_count = 0;
+	cur->bc_rec.i.ir_freecount = 0;
+	cur->bc_rec.i.ir_free = 0;
+	return xfs_btree_lookup(cur, dir, stat);
+}
+
+/*
+ * Update the record referred to by cur to the value given.
+ * This either works (return 0) or gets an EFSCORRUPTED error.
+ */
+STATIC int				/* error */
+xfs_inobt_update(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_inobt_rec_incore_t	*irec)	/* btree record */
+{
+	union xfs_btree_rec	rec;
+
+	rec.inobt.ir_startino = cpu_to_be32(irec->ir_startino);
+	if (xfs_has_sparseinodes(cur->bc_mp)) {
+		rec.inobt.ir_u.sp.ir_holemask = cpu_to_be16(irec->ir_holemask);
+		rec.inobt.ir_u.sp.ir_count = irec->ir_count;
+		rec.inobt.ir_u.sp.ir_freecount = irec->ir_freecount;
+	} else {
+		/* ir_holemask/ir_count not supported on-disk */
+		rec.inobt.ir_u.f.ir_freecount = cpu_to_be32(irec->ir_freecount);
+	}
+	rec.inobt.ir_free = cpu_to_be64(irec->ir_free);
+	return xfs_btree_update(cur, &rec);
+}
+
+/* Convert on-disk btree record to incore inobt record. */
+void
+xfs_inobt_btrec_to_irec(
+	struct xfs_mount		*mp,
+	const union xfs_btree_rec	*rec,
+	struct xfs_inobt_rec_incore	*irec)
+{
+	irec->ir_startino = be32_to_cpu(rec->inobt.ir_startino);
+	if (xfs_has_sparseinodes(mp)) {
+		irec->ir_holemask = be16_to_cpu(rec->inobt.ir_u.sp.ir_holemask);
+		irec->ir_count = rec->inobt.ir_u.sp.ir_count;
+		irec->ir_freecount = rec->inobt.ir_u.sp.ir_freecount;
+	} else {
+		/*
+		 * ir_holemask/ir_count not supported on-disk. Fill in hardcoded
+		 * values for full inode chunks.
+		 */
+		irec->ir_holemask = XFS_INOBT_HOLEMASK_FULL;
+		irec->ir_count = XFS_INODES_PER_CHUNK;
+		irec->ir_freecount =
+				be32_to_cpu(rec->inobt.ir_u.f.ir_freecount);
+	}
+	irec->ir_free = be64_to_cpu(rec->inobt.ir_free);
+}
+
+/*
+ * Get the data from the pointed-to record.
+ */
+int
+xfs_inobt_get_rec(
+	struct xfs_btree_cur		*cur,
+	struct xfs_inobt_rec_incore	*irec,
+	int				*stat)
+{
+	struct xfs_mount		*mp = cur->bc_mp;
+	union xfs_btree_rec		*rec;
+	int				error;
+	uint64_t			realfree;
+
+	error = xfs_btree_get_rec(cur, &rec, stat);
+	if (error || *stat == 0)
+		return error;
+
+	xfs_inobt_btrec_to_irec(mp, rec, irec);
+
+	if (!xfs_verify_agino(cur->bc_ag.pag, irec->ir_startino))
+		goto out_bad_rec;
+	if (irec->ir_count < XFS_INODES_PER_HOLEMASK_BIT ||
+	    irec->ir_count > XFS_INODES_PER_CHUNK)
+		goto out_bad_rec;
+	if (irec->ir_freecount > XFS_INODES_PER_CHUNK)
+		goto out_bad_rec;
+
+	/* if there are no holes, return the first available offset */
+	if (!xfs_inobt_issparse(irec->ir_holemask))
+		realfree = irec->ir_free;
+	else
+		realfree = irec->ir_free & xfs_inobt_irec_to_allocmask(irec);
+	if (hweight64(realfree) != irec->ir_freecount)
+		goto out_bad_rec;
+
+	return 0;
+
+out_bad_rec:
+	xfs_warn(mp,
+		"%s Inode BTree record corruption in AG %d detected!",
+		cur->bc_btnum == XFS_BTNUM_INO ? "Used" : "Free",
+		cur->bc_ag.pag->pag_agno);
+	xfs_warn(mp,
+"start inode 0x%x, count 0x%x, free 0x%x freemask 0x%llx, holemask 0x%x",
+		irec->ir_startino, irec->ir_count, irec->ir_freecount,
+		irec->ir_free, irec->ir_holemask);
+	return -EFSCORRUPTED;
+}
+
+/*
+ * Insert a single inobt record. Cursor must already point to desired location.
+ */
+int
+xfs_inobt_insert_rec(
+	struct xfs_btree_cur	*cur,
+	uint16_t		holemask,
+	uint8_t			count,
+	int32_t			freecount,
+	xfs_inofree_t		free,
+	int			*stat)
+{
+	cur->bc_rec.i.ir_holemask = holemask;
+	cur->bc_rec.i.ir_count = count;
+	cur->bc_rec.i.ir_freecount = freecount;
+	cur->bc_rec.i.ir_free = free;
+	return xfs_btree_insert(cur, stat);
+}
+
+/*
+ * Insert records describing a newly allocated inode chunk into the inobt.
+ */
+STATIC int
+xfs_inobt_insert(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp,
+	struct xfs_perag	*pag,
+	xfs_agino_t		newino,
+	xfs_agino_t		newlen,
+	xfs_btnum_t		btnum)
+{
+	struct xfs_btree_cur	*cur;
+	xfs_agino_t		thisino;
+	int			i;
+	int			error;
+
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, pag, btnum);
+
+	for (thisino = newino;
+	     thisino < newino + newlen;
+	     thisino += XFS_INODES_PER_CHUNK) {
+		error = xfs_inobt_lookup(cur, thisino, XFS_LOOKUP_EQ, &i);
+		if (error) {
+			xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+			return error;
+		}
+		ASSERT(i == 0);
+
+		error = xfs_inobt_insert_rec(cur, XFS_INOBT_HOLEMASK_FULL,
+					     XFS_INODES_PER_CHUNK,
+					     XFS_INODES_PER_CHUNK,
+					     XFS_INOBT_ALL_FREE, &i);
+		if (error) {
+			xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+			return error;
+		}
+		ASSERT(i == 1);
+	}
+
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+
+	return 0;
+}
+
+/*
+ * Verify that the number of free inodes in the AGI is correct.
+ */
+#ifdef DEBUG
+static int
+xfs_check_agi_freecount(
+	struct xfs_btree_cur	*cur)
+{
+	if (cur->bc_nlevels == 1) {
+		xfs_inobt_rec_incore_t rec;
+		int		freecount = 0;
+		int		error;
+		int		i;
+
+		error = xfs_inobt_lookup(cur, 0, XFS_LOOKUP_GE, &i);
+		if (error)
+			return error;
+
+		do {
+			error = xfs_inobt_get_rec(cur, &rec, &i);
+			if (error)
+				return error;
+
+			if (i) {
+				freecount += rec.ir_freecount;
+				error = xfs_btree_increment(cur, 0, &i);
+				if (error)
+					return error;
+			}
+		} while (i == 1);
+
+		if (!xfs_is_shutdown(cur->bc_mp))
+			ASSERT(freecount == cur->bc_ag.pag->pagi_freecount);
+	}
+	return 0;
+}
+#else
+#define xfs_check_agi_freecount(cur)	0
+#endif
+
+/*
+ * Initialise a new set of inodes. When called without a transaction context
+ * (e.g. from recovery) we initiate a delayed write of the inode buffers rather
+ * than logging them (which in a transaction context puts them into the AIL
+ * for writeback rather than the xfsbufd queue).
+ */
+int
+xfs_ialloc_inode_init(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct list_head	*buffer_list,
+	int			icount,
+	xfs_agnumber_t		agno,
+	xfs_agblock_t		agbno,
+	xfs_agblock_t		length,
+	unsigned int		gen)
+{
+	struct xfs_buf		*fbuf;
+	struct xfs_dinode	*free;
+	int			nbufs;
+	int			version;
+	int			i, j;
+	xfs_daddr_t		d;
+	xfs_ino_t		ino = 0;
+	int			error;
+
+	/*
+	 * Loop over the new block(s), filling in the inodes.  For small block
+	 * sizes, manipulate the inodes in buffers  which are multiples of the
+	 * blocks size.
+	 */
+	nbufs = length / M_IGEO(mp)->blocks_per_cluster;
+
+	/*
+	 * Figure out what version number to use in the inodes we create.  If
+	 * the superblock version has caught up to the one that supports the new
+	 * inode format, then use the new inode version.  Otherwise use the old
+	 * version so that old kernels will continue to be able to use the file
+	 * system.
+	 *
+	 * For v3 inodes, we also need to write the inode number into the inode,
+	 * so calculate the first inode number of the chunk here as
+	 * XFS_AGB_TO_AGINO() only works within a filesystem block, not
+	 * across multiple filesystem blocks (such as a cluster) and so cannot
+	 * be used in the cluster buffer loop below.
+	 *
+	 * Further, because we are writing the inode directly into the buffer
+	 * and calculating a CRC on the entire inode, we have ot log the entire
+	 * inode so that the entire range the CRC covers is present in the log.
+	 * That means for v3 inode we log the entire buffer rather than just the
+	 * inode cores.
+	 */
+	if (xfs_has_v3inodes(mp)) {
+		version = 3;
+		ino = XFS_AGINO_TO_INO(mp, agno, XFS_AGB_TO_AGINO(mp, agbno));
+
+		/*
+		 * log the initialisation that is about to take place as an
+		 * logical operation. This means the transaction does not
+		 * need to log the physical changes to the inode buffers as log
+		 * recovery will know what initialisation is actually needed.
+		 * Hence we only need to log the buffers as "ordered" buffers so
+		 * they track in the AIL as if they were physically logged.
+		 */
+		if (tp)
+			xfs_icreate_log(tp, agno, agbno, icount,
+					mp->m_sb.sb_inodesize, length, gen);
+	} else
+		version = 2;
+
+	for (j = 0; j < nbufs; j++) {
+		/*
+		 * Get the block.
+		 */
+		d = XFS_AGB_TO_DADDR(mp, agno, agbno +
+				(j * M_IGEO(mp)->blocks_per_cluster));
+		error = xfs_trans_get_buf(tp, mp->m_ddev_targp, d,
+				mp->m_bsize * M_IGEO(mp)->blocks_per_cluster,
+				XBF_UNMAPPED, &fbuf);
+		if (error)
+			return error;
+
+		/* Initialize the inode buffers and log them appropriately. */
+		fbuf->b_ops = &xfs_inode_buf_ops;
+		xfs_buf_zero(fbuf, 0, BBTOB(fbuf->b_length));
+		for (i = 0; i < M_IGEO(mp)->inodes_per_cluster; i++) {
+			int	ioffset = i << mp->m_sb.sb_inodelog;
+
+			free = xfs_make_iptr(mp, fbuf, i);
+			free->di_magic = cpu_to_be16(XFS_DINODE_MAGIC);
+			free->di_version = version;
+			free->di_gen = cpu_to_be32(gen);
+			free->di_next_unlinked = cpu_to_be32(NULLAGINO);
+
+			if (version == 3) {
+				free->di_ino = cpu_to_be64(ino);
+				ino++;
+				uuid_copy(&free->di_uuid,
+					  &mp->m_sb.sb_meta_uuid);
+				xfs_dinode_calc_crc(mp, free);
+			} else if (tp) {
+				/* just log the inode core */
+				xfs_trans_log_buf(tp, fbuf, ioffset,
+					  ioffset + XFS_DINODE_SIZE(mp) - 1);
+			}
+		}
+
+		if (tp) {
+			/*
+			 * Mark the buffer as an inode allocation buffer so it
+			 * sticks in AIL at the point of this allocation
+			 * transaction. This ensures the they are on disk before
+			 * the tail of the log can be moved past this
+			 * transaction (i.e. by preventing relogging from moving
+			 * it forward in the log).
+			 */
+			xfs_trans_inode_alloc_buf(tp, fbuf);
+			if (version == 3) {
+				/*
+				 * Mark the buffer as ordered so that they are
+				 * not physically logged in the transaction but
+				 * still tracked in the AIL as part of the
+				 * transaction and pin the log appropriately.
+				 */
+				xfs_trans_ordered_buf(tp, fbuf);
+			}
+		} else {
+			fbuf->b_flags |= XBF_DONE;
+			xfs_buf_delwri_queue(fbuf, buffer_list);
+			xfs_buf_relse(fbuf);
+		}
+	}
+	return 0;
+}
+
+/*
+ * Align startino and allocmask for a recently allocated sparse chunk such that
+ * they are fit for insertion (or merge) into the on-disk inode btrees.
+ *
+ * Background:
+ *
+ * When enabled, sparse inode support increases the inode alignment from cluster
+ * size to inode chunk size. This means that the minimum range between two
+ * non-adjacent inode records in the inobt is large enough for a full inode
+ * record. This allows for cluster sized, cluster aligned block allocation
+ * without need to worry about whether the resulting inode record overlaps with
+ * another record in the tree. Without this basic rule, we would have to deal
+ * with the consequences of overlap by potentially undoing recent allocations in
+ * the inode allocation codepath.
+ *
+ * Because of this alignment rule (which is enforced on mount), there are two
+ * inobt possibilities for newly allocated sparse chunks. One is that the
+ * aligned inode record for the chunk covers a range of inodes not already
+ * covered in the inobt (i.e., it is safe to insert a new sparse record). The
+ * other is that a record already exists at the aligned startino that considers
+ * the newly allocated range as sparse. In the latter case, record content is
+ * merged in hope that sparse inode chunks fill to full chunks over time.
+ */
+STATIC void
+xfs_align_sparse_ino(
+	struct xfs_mount		*mp,
+	xfs_agino_t			*startino,
+	uint16_t			*allocmask)
+{
+	xfs_agblock_t			agbno;
+	xfs_agblock_t			mod;
+	int				offset;
+
+	agbno = XFS_AGINO_TO_AGBNO(mp, *startino);
+	mod = agbno % mp->m_sb.sb_inoalignmt;
+	if (!mod)
+		return;
+
+	/* calculate the inode offset and align startino */
+	offset = XFS_AGB_TO_AGINO(mp, mod);
+	*startino -= offset;
+
+	/*
+	 * Since startino has been aligned down, left shift allocmask such that
+	 * it continues to represent the same physical inodes relative to the
+	 * new startino.
+	 */
+	*allocmask <<= offset / XFS_INODES_PER_HOLEMASK_BIT;
+}
+
+/*
+ * Determine whether the source inode record can merge into the target. Both
+ * records must be sparse, the inode ranges must match and there must be no
+ * allocation overlap between the records.
+ */
+STATIC bool
+__xfs_inobt_can_merge(
+	struct xfs_inobt_rec_incore	*trec,	/* tgt record */
+	struct xfs_inobt_rec_incore	*srec)	/* src record */
+{
+	uint64_t			talloc;
+	uint64_t			salloc;
+
+	/* records must cover the same inode range */
+	if (trec->ir_startino != srec->ir_startino)
+		return false;
+
+	/* both records must be sparse */
+	if (!xfs_inobt_issparse(trec->ir_holemask) ||
+	    !xfs_inobt_issparse(srec->ir_holemask))
+		return false;
+
+	/* both records must track some inodes */
+	if (!trec->ir_count || !srec->ir_count)
+		return false;
+
+	/* can't exceed capacity of a full record */
+	if (trec->ir_count + srec->ir_count > XFS_INODES_PER_CHUNK)
+		return false;
+
+	/* verify there is no allocation overlap */
+	talloc = xfs_inobt_irec_to_allocmask(trec);
+	salloc = xfs_inobt_irec_to_allocmask(srec);
+	if (talloc & salloc)
+		return false;
+
+	return true;
+}
+
+/*
+ * Merge the source inode record into the target. The caller must call
+ * __xfs_inobt_can_merge() to ensure the merge is valid.
+ */
+STATIC void
+__xfs_inobt_rec_merge(
+	struct xfs_inobt_rec_incore	*trec,	/* target */
+	struct xfs_inobt_rec_incore	*srec)	/* src */
+{
+	ASSERT(trec->ir_startino == srec->ir_startino);
+
+	/* combine the counts */
+	trec->ir_count += srec->ir_count;
+	trec->ir_freecount += srec->ir_freecount;
+
+	/*
+	 * Merge the holemask and free mask. For both fields, 0 bits refer to
+	 * allocated inodes. We combine the allocated ranges with bitwise AND.
+	 */
+	trec->ir_holemask &= srec->ir_holemask;
+	trec->ir_free &= srec->ir_free;
+}
+
+/*
+ * Insert a new sparse inode chunk into the associated inode btree. The inode
+ * record for the sparse chunk is pre-aligned to a startino that should match
+ * any pre-existing sparse inode record in the tree. This allows sparse chunks
+ * to fill over time.
+ *
+ * This function supports two modes of handling preexisting records depending on
+ * the merge flag. If merge is true, the provided record is merged with the
+ * existing record and updated in place. The merged record is returned in nrec.
+ * If merge is false, an existing record is replaced with the provided record.
+ * If no preexisting record exists, the provided record is always inserted.
+ *
+ * It is considered corruption if a merge is requested and not possible. Given
+ * the sparse inode alignment constraints, this should never happen.
+ */
+STATIC int
+xfs_inobt_insert_sprec(
+	struct xfs_mount		*mp,
+	struct xfs_trans		*tp,
+	struct xfs_buf			*agbp,
+	struct xfs_perag		*pag,
+	int				btnum,
+	struct xfs_inobt_rec_incore	*nrec,	/* in/out: new/merged rec. */
+	bool				merge)	/* merge or replace */
+{
+	struct xfs_btree_cur		*cur;
+	int				error;
+	int				i;
+	struct xfs_inobt_rec_incore	rec;
+
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, pag, btnum);
+
+	/* the new record is pre-aligned so we know where to look */
+	error = xfs_inobt_lookup(cur, nrec->ir_startino, XFS_LOOKUP_EQ, &i);
+	if (error)
+		goto error;
+	/* if nothing there, insert a new record and return */
+	if (i == 0) {
+		error = xfs_inobt_insert_rec(cur, nrec->ir_holemask,
+					     nrec->ir_count, nrec->ir_freecount,
+					     nrec->ir_free, &i);
+		if (error)
+			goto error;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error;
+		}
+
+		goto out;
+	}
+
+	/*
+	 * A record exists at this startino. Merge or replace the record
+	 * depending on what we've been asked to do.
+	 */
+	if (merge) {
+		error = xfs_inobt_get_rec(cur, &rec, &i);
+		if (error)
+			goto error;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error;
+		}
+		if (XFS_IS_CORRUPT(mp, rec.ir_startino != nrec->ir_startino)) {
+			error = -EFSCORRUPTED;
+			goto error;
+		}
+
+		/*
+		 * This should never fail. If we have coexisting records that
+		 * cannot merge, something is seriously wrong.
+		 */
+		if (XFS_IS_CORRUPT(mp, !__xfs_inobt_can_merge(nrec, &rec))) {
+			error = -EFSCORRUPTED;
+			goto error;
+		}
+
+		trace_xfs_irec_merge_pre(mp, pag->pag_agno, rec.ir_startino,
+					 rec.ir_holemask, nrec->ir_startino,
+					 nrec->ir_holemask);
+
+		/* merge to nrec to output the updated record */
+		__xfs_inobt_rec_merge(nrec, &rec);
+
+		trace_xfs_irec_merge_post(mp, pag->pag_agno, nrec->ir_startino,
+					  nrec->ir_holemask);
+
+		error = xfs_inobt_rec_check_count(mp, nrec);
+		if (error)
+			goto error;
+	}
+
+	error = xfs_inobt_update(cur, nrec);
+	if (error)
+		goto error;
+
+out:
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	return 0;
+error:
+	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/*
+ * Allocate new inodes in the allocation group specified by agbp.  Returns 0 if
+ * inodes were allocated in this AG; -EAGAIN if there was no space in this AG so
+ * the caller knows it can try another AG, a hard -ENOSPC when over the maximum
+ * inode count threshold, or the usual negative error code for other errors.
+ */
+STATIC int
+xfs_ialloc_ag_alloc(
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp,
+	struct xfs_perag	*pag)
+{
+	struct xfs_agi		*agi;
+	struct xfs_alloc_arg	args;
+	int			error;
+	xfs_agino_t		newino;		/* new first inode's number */
+	xfs_agino_t		newlen;		/* new number of inodes */
+	int			isaligned = 0;	/* inode allocation at stripe */
+						/* unit boundary */
+	/* init. to full chunk */
+	struct xfs_inobt_rec_incore rec;
+	struct xfs_ino_geometry	*igeo = M_IGEO(tp->t_mountp);
+	uint16_t		allocmask = (uint16_t) -1;
+	int			do_sparse = 0;
+
+	memset(&args, 0, sizeof(args));
+	args.tp = tp;
+	args.mp = tp->t_mountp;
+	args.fsbno = NULLFSBLOCK;
+	args.oinfo = XFS_RMAP_OINFO_INODES;
+
+#ifdef DEBUG
+	/* randomly do sparse inode allocations */
+	if (xfs_has_sparseinodes(tp->t_mountp) &&
+	    igeo->ialloc_min_blks < igeo->ialloc_blks)
+		do_sparse = prandom_u32_max(2);
+#endif
+
+	/*
+	 * Locking will ensure that we don't have two callers in here
+	 * at one time.
+	 */
+	newlen = igeo->ialloc_inos;
+	if (igeo->maxicount &&
+	    percpu_counter_read_positive(&args.mp->m_icount) + newlen >
+							igeo->maxicount)
+		return -ENOSPC;
+	args.minlen = args.maxlen = igeo->ialloc_blks;
+	/*
+	 * First try to allocate inodes contiguous with the last-allocated
+	 * chunk of inodes.  If the filesystem is striped, this will fill
+	 * an entire stripe unit with inodes.
+	 */
+	agi = agbp->b_addr;
+	newino = be32_to_cpu(agi->agi_newino);
+	args.agbno = XFS_AGINO_TO_AGBNO(args.mp, newino) +
+		     igeo->ialloc_blks;
+	if (do_sparse)
+		goto sparse_alloc;
+	if (likely(newino != NULLAGINO &&
+		  (args.agbno < be32_to_cpu(agi->agi_length)))) {
+		args.fsbno = XFS_AGB_TO_FSB(args.mp, pag->pag_agno, args.agbno);
+		args.type = XFS_ALLOCTYPE_THIS_BNO;
+		args.prod = 1;
+
+		/*
+		 * We need to take into account alignment here to ensure that
+		 * we don't modify the free list if we fail to have an exact
+		 * block. If we don't have an exact match, and every oher
+		 * attempt allocation attempt fails, we'll end up cancelling
+		 * a dirty transaction and shutting down.
+		 *
+		 * For an exact allocation, alignment must be 1,
+		 * however we need to take cluster alignment into account when
+		 * fixing up the freelist. Use the minalignslop field to
+		 * indicate that extra blocks might be required for alignment,
+		 * but not to use them in the actual exact allocation.
+		 */
+		args.alignment = 1;
+		args.minalignslop = igeo->cluster_align - 1;
+
+		/* Allow space for the inode btree to split. */
+		args.minleft = igeo->inobt_maxlevels;
+		if ((error = xfs_alloc_vextent(&args)))
+			return error;
+
+		/*
+		 * This request might have dirtied the transaction if the AG can
+		 * satisfy the request, but the exact block was not available.
+		 * If the allocation did fail, subsequent requests will relax
+		 * the exact agbno requirement and increase the alignment
+		 * instead. It is critical that the total size of the request
+		 * (len + alignment + slop) does not increase from this point
+		 * on, so reset minalignslop to ensure it is not included in
+		 * subsequent requests.
+		 */
+		args.minalignslop = 0;
+	}
+
+	if (unlikely(args.fsbno == NULLFSBLOCK)) {
+		/*
+		 * Set the alignment for the allocation.
+		 * If stripe alignment is turned on then align at stripe unit
+		 * boundary.
+		 * If the cluster size is smaller than a filesystem block
+		 * then we're doing I/O for inodes in filesystem block size
+		 * pieces, so don't need alignment anyway.
+		 */
+		isaligned = 0;
+		if (igeo->ialloc_align) {
+			ASSERT(!xfs_has_noalign(args.mp));
+			args.alignment = args.mp->m_dalign;
+			isaligned = 1;
+		} else
+			args.alignment = igeo->cluster_align;
+		/*
+		 * Need to figure out where to allocate the inode blocks.
+		 * Ideally they should be spaced out through the a.g.
+		 * For now, just allocate blocks up front.
+		 */
+		args.agbno = be32_to_cpu(agi->agi_root);
+		args.fsbno = XFS_AGB_TO_FSB(args.mp, pag->pag_agno, args.agbno);
+		/*
+		 * Allocate a fixed-size extent of inodes.
+		 */
+		args.type = XFS_ALLOCTYPE_NEAR_BNO;
+		args.prod = 1;
+		/*
+		 * Allow space for the inode btree to split.
+		 */
+		args.minleft = igeo->inobt_maxlevels;
+		if ((error = xfs_alloc_vextent(&args)))
+			return error;
+	}
+
+	/*
+	 * If stripe alignment is turned on, then try again with cluster
+	 * alignment.
+	 */
+	if (isaligned && args.fsbno == NULLFSBLOCK) {
+		args.type = XFS_ALLOCTYPE_NEAR_BNO;
+		args.agbno = be32_to_cpu(agi->agi_root);
+		args.fsbno = XFS_AGB_TO_FSB(args.mp, pag->pag_agno, args.agbno);
+		args.alignment = igeo->cluster_align;
+		if ((error = xfs_alloc_vextent(&args)))
+			return error;
+	}
+
+	/*
+	 * Finally, try a sparse allocation if the filesystem supports it and
+	 * the sparse allocation length is smaller than a full chunk.
+	 */
+	if (xfs_has_sparseinodes(args.mp) &&
+	    igeo->ialloc_min_blks < igeo->ialloc_blks &&
+	    args.fsbno == NULLFSBLOCK) {
+sparse_alloc:
+		args.type = XFS_ALLOCTYPE_NEAR_BNO;
+		args.agbno = be32_to_cpu(agi->agi_root);
+		args.fsbno = XFS_AGB_TO_FSB(args.mp, pag->pag_agno, args.agbno);
+		args.alignment = args.mp->m_sb.sb_spino_align;
+		args.prod = 1;
+
+		args.minlen = igeo->ialloc_min_blks;
+		args.maxlen = args.minlen;
+
+		/*
+		 * The inode record will be aligned to full chunk size. We must
+		 * prevent sparse allocation from AG boundaries that result in
+		 * invalid inode records, such as records that start at agbno 0
+		 * or extend beyond the AG.
+		 *
+		 * Set min agbno to the first aligned, non-zero agbno and max to
+		 * the last aligned agbno that is at least one full chunk from
+		 * the end of the AG.
+		 */
+		args.min_agbno = args.mp->m_sb.sb_inoalignmt;
+		args.max_agbno = round_down(args.mp->m_sb.sb_agblocks,
+					    args.mp->m_sb.sb_inoalignmt) -
+				 igeo->ialloc_blks;
+
+		error = xfs_alloc_vextent(&args);
+		if (error)
+			return error;
+
+		newlen = XFS_AGB_TO_AGINO(args.mp, args.len);
+		ASSERT(newlen <= XFS_INODES_PER_CHUNK);
+		allocmask = (1 << (newlen / XFS_INODES_PER_HOLEMASK_BIT)) - 1;
+	}
+
+	if (args.fsbno == NULLFSBLOCK)
+		return -EAGAIN;
+
+	ASSERT(args.len == args.minlen);
+
+	/*
+	 * Stamp and write the inode buffers.
+	 *
+	 * Seed the new inode cluster with a random generation number. This
+	 * prevents short-term reuse of generation numbers if a chunk is
+	 * freed and then immediately reallocated. We use random numbers
+	 * rather than a linear progression to prevent the next generation
+	 * number from being easily guessable.
+	 */
+	error = xfs_ialloc_inode_init(args.mp, tp, NULL, newlen, pag->pag_agno,
+			args.agbno, args.len, get_random_u32());
+
+	if (error)
+		return error;
+	/*
+	 * Convert the results.
+	 */
+	newino = XFS_AGB_TO_AGINO(args.mp, args.agbno);
+
+	if (xfs_inobt_issparse(~allocmask)) {
+		/*
+		 * We've allocated a sparse chunk. Align the startino and mask.
+		 */
+		xfs_align_sparse_ino(args.mp, &newino, &allocmask);
+
+		rec.ir_startino = newino;
+		rec.ir_holemask = ~allocmask;
+		rec.ir_count = newlen;
+		rec.ir_freecount = newlen;
+		rec.ir_free = XFS_INOBT_ALL_FREE;
+
+		/*
+		 * Insert the sparse record into the inobt and allow for a merge
+		 * if necessary. If a merge does occur, rec is updated to the
+		 * merged record.
+		 */
+		error = xfs_inobt_insert_sprec(args.mp, tp, agbp, pag,
+				XFS_BTNUM_INO, &rec, true);
+		if (error == -EFSCORRUPTED) {
+			xfs_alert(args.mp,
+	"invalid sparse inode record: ino 0x%llx holemask 0x%x count %u",
+				  XFS_AGINO_TO_INO(args.mp, pag->pag_agno,
+						   rec.ir_startino),
+				  rec.ir_holemask, rec.ir_count);
+			xfs_force_shutdown(args.mp, SHUTDOWN_CORRUPT_INCORE);
+		}
+		if (error)
+			return error;
+
+		/*
+		 * We can't merge the part we've just allocated as for the inobt
+		 * due to finobt semantics. The original record may or may not
+		 * exist independent of whether physical inodes exist in this
+		 * sparse chunk.
+		 *
+		 * We must update the finobt record based on the inobt record.
+		 * rec contains the fully merged and up to date inobt record
+		 * from the previous call. Set merge false to replace any
+		 * existing record with this one.
+		 */
+		if (xfs_has_finobt(args.mp)) {
+			error = xfs_inobt_insert_sprec(args.mp, tp, agbp, pag,
+				       XFS_BTNUM_FINO, &rec, false);
+			if (error)
+				return error;
+		}
+	} else {
+		/* full chunk - insert new records to both btrees */
+		error = xfs_inobt_insert(args.mp, tp, agbp, pag, newino, newlen,
+					 XFS_BTNUM_INO);
+		if (error)
+			return error;
+
+		if (xfs_has_finobt(args.mp)) {
+			error = xfs_inobt_insert(args.mp, tp, agbp, pag, newino,
+						 newlen, XFS_BTNUM_FINO);
+			if (error)
+				return error;
+		}
+	}
+
+	/*
+	 * Update AGI counts and newino.
+	 */
+	be32_add_cpu(&agi->agi_count, newlen);
+	be32_add_cpu(&agi->agi_freecount, newlen);
+	pag->pagi_freecount += newlen;
+	pag->pagi_count += newlen;
+	agi->agi_newino = cpu_to_be32(newino);
+
+	/*
+	 * Log allocation group header fields
+	 */
+	xfs_ialloc_log_agi(tp, agbp,
+		XFS_AGI_COUNT | XFS_AGI_FREECOUNT | XFS_AGI_NEWINO);
+	/*
+	 * Modify/log superblock values for inode count and inode free count.
+	 */
+	xfs_trans_mod_sb(tp, XFS_TRANS_SB_ICOUNT, (long)newlen);
+	xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, (long)newlen);
+	return 0;
+}
+
+/*
+ * Try to retrieve the next record to the left/right from the current one.
+ */
+STATIC int
+xfs_ialloc_next_rec(
+	struct xfs_btree_cur	*cur,
+	xfs_inobt_rec_incore_t	*rec,
+	int			*done,
+	int			left)
+{
+	int                     error;
+	int			i;
+
+	if (left)
+		error = xfs_btree_decrement(cur, 0, &i);
+	else
+		error = xfs_btree_increment(cur, 0, &i);
+
+	if (error)
+		return error;
+	*done = !i;
+	if (i) {
+		error = xfs_inobt_get_rec(cur, rec, &i);
+		if (error)
+			return error;
+		if (XFS_IS_CORRUPT(cur->bc_mp, i != 1))
+			return -EFSCORRUPTED;
+	}
+
+	return 0;
+}
+
+STATIC int
+xfs_ialloc_get_rec(
+	struct xfs_btree_cur	*cur,
+	xfs_agino_t		agino,
+	xfs_inobt_rec_incore_t	*rec,
+	int			*done)
+{
+	int                     error;
+	int			i;
+
+	error = xfs_inobt_lookup(cur, agino, XFS_LOOKUP_EQ, &i);
+	if (error)
+		return error;
+	*done = !i;
+	if (i) {
+		error = xfs_inobt_get_rec(cur, rec, &i);
+		if (error)
+			return error;
+		if (XFS_IS_CORRUPT(cur->bc_mp, i != 1))
+			return -EFSCORRUPTED;
+	}
+
+	return 0;
+}
+
+/*
+ * Return the offset of the first free inode in the record. If the inode chunk
+ * is sparsely allocated, we convert the record holemask to inode granularity
+ * and mask off the unallocated regions from the inode free mask.
+ */
+STATIC int
+xfs_inobt_first_free_inode(
+	struct xfs_inobt_rec_incore	*rec)
+{
+	xfs_inofree_t			realfree;
+
+	/* if there are no holes, return the first available offset */
+	if (!xfs_inobt_issparse(rec->ir_holemask))
+		return xfs_lowbit64(rec->ir_free);
+
+	realfree = xfs_inobt_irec_to_allocmask(rec);
+	realfree &= rec->ir_free;
+
+	return xfs_lowbit64(realfree);
+}
+
+/*
+ * Allocate an inode using the inobt-only algorithm.
+ */
+STATIC int
+xfs_dialloc_ag_inobt(
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp,
+	struct xfs_perag	*pag,
+	xfs_ino_t		parent,
+	xfs_ino_t		*inop)
+{
+	struct xfs_mount	*mp = tp->t_mountp;
+	struct xfs_agi		*agi = agbp->b_addr;
+	xfs_agnumber_t		pagno = XFS_INO_TO_AGNO(mp, parent);
+	xfs_agino_t		pagino = XFS_INO_TO_AGINO(mp, parent);
+	struct xfs_btree_cur	*cur, *tcur;
+	struct xfs_inobt_rec_incore rec, trec;
+	xfs_ino_t		ino;
+	int			error;
+	int			offset;
+	int			i, j;
+	int			searchdistance = 10;
+
+	ASSERT(pag->pagi_init);
+	ASSERT(pag->pagi_inodeok);
+	ASSERT(pag->pagi_freecount > 0);
+
+ restart_pagno:
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, pag, XFS_BTNUM_INO);
+	/*
+	 * If pagino is 0 (this is the root inode allocation) use newino.
+	 * This must work because we've just allocated some.
+	 */
+	if (!pagino)
+		pagino = be32_to_cpu(agi->agi_newino);
+
+	error = xfs_check_agi_freecount(cur);
+	if (error)
+		goto error0;
+
+	/*
+	 * If in the same AG as the parent, try to get near the parent.
+	 */
+	if (pagno == pag->pag_agno) {
+		int		doneleft;	/* done, to the left */
+		int		doneright;	/* done, to the right */
+
+		error = xfs_inobt_lookup(cur, pagino, XFS_LOOKUP_LE, &i);
+		if (error)
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+
+		error = xfs_inobt_get_rec(cur, &rec, &j);
+		if (error)
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, j != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+
+		if (rec.ir_freecount > 0) {
+			/*
+			 * Found a free inode in the same chunk
+			 * as the parent, done.
+			 */
+			goto alloc_inode;
+		}
+
+
+		/*
+		 * In the same AG as parent, but parent's chunk is full.
+		 */
+
+		/* duplicate the cursor, search left & right simultaneously */
+		error = xfs_btree_dup_cursor(cur, &tcur);
+		if (error)
+			goto error0;
+
+		/*
+		 * Skip to last blocks looked up if same parent inode.
+		 */
+		if (pagino != NULLAGINO &&
+		    pag->pagl_pagino == pagino &&
+		    pag->pagl_leftrec != NULLAGINO &&
+		    pag->pagl_rightrec != NULLAGINO) {
+			error = xfs_ialloc_get_rec(tcur, pag->pagl_leftrec,
+						   &trec, &doneleft);
+			if (error)
+				goto error1;
+
+			error = xfs_ialloc_get_rec(cur, pag->pagl_rightrec,
+						   &rec, &doneright);
+			if (error)
+				goto error1;
+		} else {
+			/* search left with tcur, back up 1 record */
+			error = xfs_ialloc_next_rec(tcur, &trec, &doneleft, 1);
+			if (error)
+				goto error1;
+
+			/* search right with cur, go forward 1 record. */
+			error = xfs_ialloc_next_rec(cur, &rec, &doneright, 0);
+			if (error)
+				goto error1;
+		}
+
+		/*
+		 * Loop until we find an inode chunk with a free inode.
+		 */
+		while (--searchdistance > 0 && (!doneleft || !doneright)) {
+			int	useleft;  /* using left inode chunk this time */
+
+			/* figure out the closer block if both are valid. */
+			if (!doneleft && !doneright) {
+				useleft = pagino -
+				 (trec.ir_startino + XFS_INODES_PER_CHUNK - 1) <
+				  rec.ir_startino - pagino;
+			} else {
+				useleft = !doneleft;
+			}
+
+			/* free inodes to the left? */
+			if (useleft && trec.ir_freecount) {
+				xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+				cur = tcur;
+
+				pag->pagl_leftrec = trec.ir_startino;
+				pag->pagl_rightrec = rec.ir_startino;
+				pag->pagl_pagino = pagino;
+				rec = trec;
+				goto alloc_inode;
+			}
+
+			/* free inodes to the right? */
+			if (!useleft && rec.ir_freecount) {
+				xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+
+				pag->pagl_leftrec = trec.ir_startino;
+				pag->pagl_rightrec = rec.ir_startino;
+				pag->pagl_pagino = pagino;
+				goto alloc_inode;
+			}
+
+			/* get next record to check */
+			if (useleft) {
+				error = xfs_ialloc_next_rec(tcur, &trec,
+								 &doneleft, 1);
+			} else {
+				error = xfs_ialloc_next_rec(cur, &rec,
+								 &doneright, 0);
+			}
+			if (error)
+				goto error1;
+		}
+
+		if (searchdistance <= 0) {
+			/*
+			 * Not in range - save last search
+			 * location and allocate a new inode
+			 */
+			xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+			pag->pagl_leftrec = trec.ir_startino;
+			pag->pagl_rightrec = rec.ir_startino;
+			pag->pagl_pagino = pagino;
+
+		} else {
+			/*
+			 * We've reached the end of the btree. because
+			 * we are only searching a small chunk of the
+			 * btree each search, there is obviously free
+			 * inodes closer to the parent inode than we
+			 * are now. restart the search again.
+			 */
+			pag->pagl_pagino = NULLAGINO;
+			pag->pagl_leftrec = NULLAGINO;
+			pag->pagl_rightrec = NULLAGINO;
+			xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+			xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+			goto restart_pagno;
+		}
+	}
+
+	/*
+	 * In a different AG from the parent.
+	 * See if the most recently allocated block has any free.
+	 */
+	if (agi->agi_newino != cpu_to_be32(NULLAGINO)) {
+		error = xfs_inobt_lookup(cur, be32_to_cpu(agi->agi_newino),
+					 XFS_LOOKUP_EQ, &i);
+		if (error)
+			goto error0;
+
+		if (i == 1) {
+			error = xfs_inobt_get_rec(cur, &rec, &j);
+			if (error)
+				goto error0;
+
+			if (j == 1 && rec.ir_freecount > 0) {
+				/*
+				 * The last chunk allocated in the group
+				 * still has a free inode.
+				 */
+				goto alloc_inode;
+			}
+		}
+	}
+
+	/*
+	 * None left in the last group, search the whole AG
+	 */
+	error = xfs_inobt_lookup(cur, 0, XFS_LOOKUP_GE, &i);
+	if (error)
+		goto error0;
+	if (XFS_IS_CORRUPT(mp, i != 1)) {
+		error = -EFSCORRUPTED;
+		goto error0;
+	}
+
+	for (;;) {
+		error = xfs_inobt_get_rec(cur, &rec, &i);
+		if (error)
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+		if (rec.ir_freecount > 0)
+			break;
+		error = xfs_btree_increment(cur, 0, &i);
+		if (error)
+			goto error0;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto error0;
+		}
+	}
+
+alloc_inode:
+	offset = xfs_inobt_first_free_inode(&rec);
+	ASSERT(offset >= 0);
+	ASSERT(offset < XFS_INODES_PER_CHUNK);
+	ASSERT((XFS_AGINO_TO_OFFSET(mp, rec.ir_startino) %
+				   XFS_INODES_PER_CHUNK) == 0);
+	ino = XFS_AGINO_TO_INO(mp, pag->pag_agno, rec.ir_startino + offset);
+	rec.ir_free &= ~XFS_INOBT_MASK(offset);
+	rec.ir_freecount--;
+	error = xfs_inobt_update(cur, &rec);
+	if (error)
+		goto error0;
+	be32_add_cpu(&agi->agi_freecount, -1);
+	xfs_ialloc_log_agi(tp, agbp, XFS_AGI_FREECOUNT);
+	pag->pagi_freecount--;
+
+	error = xfs_check_agi_freecount(cur);
+	if (error)
+		goto error0;
+
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, -1);
+	*inop = ino;
+	return 0;
+error1:
+	xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
+error0:
+	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/*
+ * Use the free inode btree to allocate an inode based on distance from the
+ * parent. Note that the provided cursor may be deleted and replaced.
+ */
+STATIC int
+xfs_dialloc_ag_finobt_near(
+	xfs_agino_t			pagino,
+	struct xfs_btree_cur		**ocur,
+	struct xfs_inobt_rec_incore	*rec)
+{
+	struct xfs_btree_cur		*lcur = *ocur;	/* left search cursor */
+	struct xfs_btree_cur		*rcur;	/* right search cursor */
+	struct xfs_inobt_rec_incore	rrec;
+	int				error;
+	int				i, j;
+
+	error = xfs_inobt_lookup(lcur, pagino, XFS_LOOKUP_LE, &i);
+	if (error)
+		return error;
+
+	if (i == 1) {
+		error = xfs_inobt_get_rec(lcur, rec, &i);
+		if (error)
+			return error;
+		if (XFS_IS_CORRUPT(lcur->bc_mp, i != 1))
+			return -EFSCORRUPTED;
+
+		/*
+		 * See if we've landed in the parent inode record. The finobt
+		 * only tracks chunks with at least one free inode, so record
+		 * existence is enough.
+		 */
+		if (pagino >= rec->ir_startino &&
+		    pagino < (rec->ir_startino + XFS_INODES_PER_CHUNK))
+			return 0;
+	}
+
+	error = xfs_btree_dup_cursor(lcur, &rcur);
+	if (error)
+		return error;
+
+	error = xfs_inobt_lookup(rcur, pagino, XFS_LOOKUP_GE, &j);
+	if (error)
+		goto error_rcur;
+	if (j == 1) {
+		error = xfs_inobt_get_rec(rcur, &rrec, &j);
+		if (error)
+			goto error_rcur;
+		if (XFS_IS_CORRUPT(lcur->bc_mp, j != 1)) {
+			error = -EFSCORRUPTED;
+			goto error_rcur;
+		}
+	}
+
+	if (XFS_IS_CORRUPT(lcur->bc_mp, i != 1 && j != 1)) {
+		error = -EFSCORRUPTED;
+		goto error_rcur;
+	}
+	if (i == 1 && j == 1) {
+		/*
+		 * Both the left and right records are valid. Choose the closer
+		 * inode chunk to the target.
+		 */
+		if ((pagino - rec->ir_startino + XFS_INODES_PER_CHUNK - 1) >
+		    (rrec.ir_startino - pagino)) {
+			*rec = rrec;
+			xfs_btree_del_cursor(lcur, XFS_BTREE_NOERROR);
+			*ocur = rcur;
+		} else {
+			xfs_btree_del_cursor(rcur, XFS_BTREE_NOERROR);
+		}
+	} else if (j == 1) {
+		/* only the right record is valid */
+		*rec = rrec;
+		xfs_btree_del_cursor(lcur, XFS_BTREE_NOERROR);
+		*ocur = rcur;
+	} else if (i == 1) {
+		/* only the left record is valid */
+		xfs_btree_del_cursor(rcur, XFS_BTREE_NOERROR);
+	}
+
+	return 0;
+
+error_rcur:
+	xfs_btree_del_cursor(rcur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/*
+ * Use the free inode btree to find a free inode based on a newino hint. If
+ * the hint is NULL, find the first free inode in the AG.
+ */
+STATIC int
+xfs_dialloc_ag_finobt_newino(
+	struct xfs_agi			*agi,
+	struct xfs_btree_cur		*cur,
+	struct xfs_inobt_rec_incore	*rec)
+{
+	int error;
+	int i;
+
+	if (agi->agi_newino != cpu_to_be32(NULLAGINO)) {
+		error = xfs_inobt_lookup(cur, be32_to_cpu(agi->agi_newino),
+					 XFS_LOOKUP_EQ, &i);
+		if (error)
+			return error;
+		if (i == 1) {
+			error = xfs_inobt_get_rec(cur, rec, &i);
+			if (error)
+				return error;
+			if (XFS_IS_CORRUPT(cur->bc_mp, i != 1))
+				return -EFSCORRUPTED;
+			return 0;
+		}
+	}
+
+	/*
+	 * Find the first inode available in the AG.
+	 */
+	error = xfs_inobt_lookup(cur, 0, XFS_LOOKUP_GE, &i);
+	if (error)
+		return error;
+	if (XFS_IS_CORRUPT(cur->bc_mp, i != 1))
+		return -EFSCORRUPTED;
+
+	error = xfs_inobt_get_rec(cur, rec, &i);
+	if (error)
+		return error;
+	if (XFS_IS_CORRUPT(cur->bc_mp, i != 1))
+		return -EFSCORRUPTED;
+
+	return 0;
+}
+
+/*
+ * Update the inobt based on a modification made to the finobt. Also ensure that
+ * the records from both trees are equivalent post-modification.
+ */
+STATIC int
+xfs_dialloc_ag_update_inobt(
+	struct xfs_btree_cur		*cur,	/* inobt cursor */
+	struct xfs_inobt_rec_incore	*frec,	/* finobt record */
+	int				offset) /* inode offset */
+{
+	struct xfs_inobt_rec_incore	rec;
+	int				error;
+	int				i;
+
+	error = xfs_inobt_lookup(cur, frec->ir_startino, XFS_LOOKUP_EQ, &i);
+	if (error)
+		return error;
+	if (XFS_IS_CORRUPT(cur->bc_mp, i != 1))
+		return -EFSCORRUPTED;
+
+	error = xfs_inobt_get_rec(cur, &rec, &i);
+	if (error)
+		return error;
+	if (XFS_IS_CORRUPT(cur->bc_mp, i != 1))
+		return -EFSCORRUPTED;
+	ASSERT((XFS_AGINO_TO_OFFSET(cur->bc_mp, rec.ir_startino) %
+				   XFS_INODES_PER_CHUNK) == 0);
+
+	rec.ir_free &= ~XFS_INOBT_MASK(offset);
+	rec.ir_freecount--;
+
+	if (XFS_IS_CORRUPT(cur->bc_mp,
+			   rec.ir_free != frec->ir_free ||
+			   rec.ir_freecount != frec->ir_freecount))
+		return -EFSCORRUPTED;
+
+	return xfs_inobt_update(cur, &rec);
+}
+
+/*
+ * Allocate an inode using the free inode btree, if available. Otherwise, fall
+ * back to the inobt search algorithm.
+ *
+ * The caller selected an AG for us, and made sure that free inodes are
+ * available.
+ */
+static int
+xfs_dialloc_ag(
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp,
+	struct xfs_perag	*pag,
+	xfs_ino_t		parent,
+	xfs_ino_t		*inop)
+{
+	struct xfs_mount		*mp = tp->t_mountp;
+	struct xfs_agi			*agi = agbp->b_addr;
+	xfs_agnumber_t			pagno = XFS_INO_TO_AGNO(mp, parent);
+	xfs_agino_t			pagino = XFS_INO_TO_AGINO(mp, parent);
+	struct xfs_btree_cur		*cur;	/* finobt cursor */
+	struct xfs_btree_cur		*icur;	/* inobt cursor */
+	struct xfs_inobt_rec_incore	rec;
+	xfs_ino_t			ino;
+	int				error;
+	int				offset;
+	int				i;
+
+	if (!xfs_has_finobt(mp))
+		return xfs_dialloc_ag_inobt(tp, agbp, pag, parent, inop);
+
+	/*
+	 * If pagino is 0 (this is the root inode allocation) use newino.
+	 * This must work because we've just allocated some.
+	 */
+	if (!pagino)
+		pagino = be32_to_cpu(agi->agi_newino);
+
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, pag, XFS_BTNUM_FINO);
+
+	error = xfs_check_agi_freecount(cur);
+	if (error)
+		goto error_cur;
+
+	/*
+	 * The search algorithm depends on whether we're in the same AG as the
+	 * parent. If so, find the closest available inode to the parent. If
+	 * not, consider the agi hint or find the first free inode in the AG.
+	 */
+	if (pag->pag_agno == pagno)
+		error = xfs_dialloc_ag_finobt_near(pagino, &cur, &rec);
+	else
+		error = xfs_dialloc_ag_finobt_newino(agi, cur, &rec);
+	if (error)
+		goto error_cur;
+
+	offset = xfs_inobt_first_free_inode(&rec);
+	ASSERT(offset >= 0);
+	ASSERT(offset < XFS_INODES_PER_CHUNK);
+	ASSERT((XFS_AGINO_TO_OFFSET(mp, rec.ir_startino) %
+				   XFS_INODES_PER_CHUNK) == 0);
+	ino = XFS_AGINO_TO_INO(mp, pag->pag_agno, rec.ir_startino + offset);
+
+	/*
+	 * Modify or remove the finobt record.
+	 */
+	rec.ir_free &= ~XFS_INOBT_MASK(offset);
+	rec.ir_freecount--;
+	if (rec.ir_freecount)
+		error = xfs_inobt_update(cur, &rec);
+	else
+		error = xfs_btree_delete(cur, &i);
+	if (error)
+		goto error_cur;
+
+	/*
+	 * The finobt has now been updated appropriately. We haven't updated the
+	 * agi and superblock yet, so we can create an inobt cursor and validate
+	 * the original freecount. If all is well, make the equivalent update to
+	 * the inobt using the finobt record and offset information.
+	 */
+	icur = xfs_inobt_init_cursor(mp, tp, agbp, pag, XFS_BTNUM_INO);
+
+	error = xfs_check_agi_freecount(icur);
+	if (error)
+		goto error_icur;
+
+	error = xfs_dialloc_ag_update_inobt(icur, &rec, offset);
+	if (error)
+		goto error_icur;
+
+	/*
+	 * Both trees have now been updated. We must update the perag and
+	 * superblock before we can check the freecount for each btree.
+	 */
+	be32_add_cpu(&agi->agi_freecount, -1);
+	xfs_ialloc_log_agi(tp, agbp, XFS_AGI_FREECOUNT);
+	pag->pagi_freecount--;
+
+	xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, -1);
+
+	error = xfs_check_agi_freecount(icur);
+	if (error)
+		goto error_icur;
+	error = xfs_check_agi_freecount(cur);
+	if (error)
+		goto error_icur;
+
+	xfs_btree_del_cursor(icur, XFS_BTREE_NOERROR);
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	*inop = ino;
+	return 0;
+
+error_icur:
+	xfs_btree_del_cursor(icur, XFS_BTREE_ERROR);
+error_cur:
+	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+	return error;
+}
+
+static int
+xfs_dialloc_roll(
+	struct xfs_trans	**tpp,
+	struct xfs_buf		*agibp)
+{
+	struct xfs_trans	*tp = *tpp;
+	struct xfs_dquot_acct	*dqinfo;
+	int			error;
+
+	/*
+	 * Hold to on to the agibp across the commit so no other allocation can
+	 * come in and take the free inodes we just allocated for our caller.
+	 */
+	xfs_trans_bhold(tp, agibp);
+
+	/*
+	 * We want the quota changes to be associated with the next transaction,
+	 * NOT this one. So, detach the dqinfo from this and attach it to the
+	 * next transaction.
+	 */
+	dqinfo = tp->t_dqinfo;
+	tp->t_dqinfo = NULL;
+
+	error = xfs_trans_roll(&tp);
+
+	/* Re-attach the quota info that we detached from prev trx. */
+	tp->t_dqinfo = dqinfo;
+
+	/*
+	 * Join the buffer even on commit error so that the buffer is released
+	 * when the caller cancels the transaction and doesn't have to handle
+	 * this error case specially.
+	 */
+	xfs_trans_bjoin(tp, agibp);
+	*tpp = tp;
+	return error;
+}
+
+static xfs_agnumber_t
+xfs_ialloc_next_ag(
+	xfs_mount_t	*mp)
+{
+	xfs_agnumber_t	agno;
+
+	spin_lock(&mp->m_agirotor_lock);
+	agno = mp->m_agirotor;
+	if (++mp->m_agirotor >= mp->m_maxagi)
+		mp->m_agirotor = 0;
+	spin_unlock(&mp->m_agirotor_lock);
+
+	return agno;
+}
+
+static bool
+xfs_dialloc_good_ag(
+	struct xfs_trans	*tp,
+	struct xfs_perag	*pag,
+	umode_t			mode,
+	int			flags,
+	bool			ok_alloc)
+{
+	struct xfs_mount	*mp = tp->t_mountp;
+	xfs_extlen_t		ineed;
+	xfs_extlen_t		longest = 0;
+	int			needspace;
+	int			error;
+
+	if (!pag->pagi_inodeok)
+		return false;
+
+	if (!pag->pagi_init) {
+		error = xfs_ialloc_read_agi(pag, tp, NULL);
+		if (error)
+			return false;
+	}
+
+	if (pag->pagi_freecount)
+		return true;
+	if (!ok_alloc)
+		return false;
+
+	if (!pag->pagf_init) {
+		error = xfs_alloc_read_agf(pag, tp, flags, NULL);
+		if (error)
+			return false;
+	}
+
+	/*
+	 * Check that there is enough free space for the file plus a chunk of
+	 * inodes if we need to allocate some. If this is the first pass across
+	 * the AGs, take into account the potential space needed for alignment
+	 * of inode chunks when checking the longest contiguous free space in
+	 * the AG - this prevents us from getting ENOSPC because we have free
+	 * space larger than ialloc_blks but alignment constraints prevent us
+	 * from using it.
+	 *
+	 * If we can't find an AG with space for full alignment slack to be
+	 * taken into account, we must be near ENOSPC in all AGs.  Hence we
+	 * don't include alignment for the second pass and so if we fail
+	 * allocation due to alignment issues then it is most likely a real
+	 * ENOSPC condition.
+	 *
+	 * XXX(dgc): this calculation is now bogus thanks to the per-ag
+	 * reservations that xfs_alloc_fix_freelist() now does via
+	 * xfs_alloc_space_available(). When the AG fills up, pagf_freeblks will
+	 * be more than large enough for the check below to succeed, but
+	 * xfs_alloc_space_available() will fail because of the non-zero
+	 * metadata reservation and hence we won't actually be able to allocate
+	 * more inodes in this AG. We do soooo much unnecessary work near ENOSPC
+	 * because of this.
+	 */
+	ineed = M_IGEO(mp)->ialloc_min_blks;
+	if (flags && ineed > 1)
+		ineed += M_IGEO(mp)->cluster_align;
+	longest = pag->pagf_longest;
+	if (!longest)
+		longest = pag->pagf_flcount > 0;
+	needspace = S_ISDIR(mode) || S_ISREG(mode) || S_ISLNK(mode);
+
+	if (pag->pagf_freeblks < needspace + ineed || longest < ineed)
+		return false;
+	return true;
+}
+
+static int
+xfs_dialloc_try_ag(
+	struct xfs_trans	**tpp,
+	struct xfs_perag	*pag,
+	xfs_ino_t		parent,
+	xfs_ino_t		*new_ino,
+	bool			ok_alloc)
+{
+	struct xfs_buf		*agbp;
+	xfs_ino_t		ino;
+	int			error;
+
+	/*
+	 * Then read in the AGI buffer and recheck with the AGI buffer
+	 * lock held.
+	 */
+	error = xfs_ialloc_read_agi(pag, *tpp, &agbp);
+	if (error)
+		return error;
+
+	if (!pag->pagi_freecount) {
+		if (!ok_alloc) {
+			error = -EAGAIN;
+			goto out_release;
+		}
+
+		error = xfs_ialloc_ag_alloc(*tpp, agbp, pag);
+		if (error < 0)
+			goto out_release;
+
+		/*
+		 * We successfully allocated space for an inode cluster in this
+		 * AG.  Roll the transaction so that we can allocate one of the
+		 * new inodes.
+		 */
+		ASSERT(pag->pagi_freecount > 0);
+		error = xfs_dialloc_roll(tpp, agbp);
+		if (error)
+			goto out_release;
+	}
+
+	/* Allocate an inode in the found AG */
+	error = xfs_dialloc_ag(*tpp, agbp, pag, parent, &ino);
+	if (!error)
+		*new_ino = ino;
+	return error;
+
+out_release:
+	xfs_trans_brelse(*tpp, agbp);
+	return error;
+}
+
+/*
+ * Allocate an on-disk inode.
+ *
+ * Mode is used to tell whether the new inode is a directory and hence where to
+ * locate it. The on-disk inode that is allocated will be returned in @new_ino
+ * on success, otherwise an error will be set to indicate the failure (e.g.
+ * -ENOSPC).
+ */
+int
+xfs_dialloc(
+	struct xfs_trans	**tpp,
+	xfs_ino_t		parent,
+	umode_t			mode,
+	xfs_ino_t		*new_ino)
+{
+	struct xfs_mount	*mp = (*tpp)->t_mountp;
+	xfs_agnumber_t		agno;
+	int			error = 0;
+	xfs_agnumber_t		start_agno;
+	struct xfs_perag	*pag;
+	struct xfs_ino_geometry	*igeo = M_IGEO(mp);
+	bool			ok_alloc = true;
+	int			flags;
+	xfs_ino_t		ino;
+
+	/*
+	 * Directories, symlinks, and regular files frequently allocate at least
+	 * one block, so factor that potential expansion when we examine whether
+	 * an AG has enough space for file creation.
+	 */
+	if (S_ISDIR(mode))
+		start_agno = xfs_ialloc_next_ag(mp);
+	else {
+		start_agno = XFS_INO_TO_AGNO(mp, parent);
+		if (start_agno >= mp->m_maxagi)
+			start_agno = 0;
+	}
+
+	/*
+	 * If we have already hit the ceiling of inode blocks then clear
+	 * ok_alloc so we scan all available agi structures for a free
+	 * inode.
+	 *
+	 * Read rough value of mp->m_icount by percpu_counter_read_positive,
+	 * which will sacrifice the preciseness but improve the performance.
+	 */
+	if (igeo->maxicount &&
+	    percpu_counter_read_positive(&mp->m_icount) + igeo->ialloc_inos
+							> igeo->maxicount) {
+		ok_alloc = false;
+	}
+
+	/*
+	 * Loop until we find an allocation group that either has free inodes
+	 * or in which we can allocate some inodes.  Iterate through the
+	 * allocation groups upward, wrapping at the end.
+	 */
+	agno = start_agno;
+	flags = XFS_ALLOC_FLAG_TRYLOCK;
+	for (;;) {
+		pag = xfs_perag_get(mp, agno);
+		if (xfs_dialloc_good_ag(*tpp, pag, mode, flags, ok_alloc)) {
+			error = xfs_dialloc_try_ag(tpp, pag, parent,
+					&ino, ok_alloc);
+			if (error != -EAGAIN)
+				break;
+		}
+
+		if (xfs_is_shutdown(mp)) {
+			error = -EFSCORRUPTED;
+			break;
+		}
+		if (++agno == mp->m_maxagi)
+			agno = 0;
+		if (agno == start_agno) {
+			if (!flags) {
+				error = -ENOSPC;
+				break;
+			}
+			flags = 0;
+		}
+		xfs_perag_put(pag);
+	}
+
+	if (!error)
+		*new_ino = ino;
+	xfs_perag_put(pag);
+	return error;
+}
+
+/*
+ * Free the blocks of an inode chunk. We must consider that the inode chunk
+ * might be sparse and only free the regions that are allocated as part of the
+ * chunk.
+ */
+STATIC void
+xfs_difree_inode_chunk(
+	struct xfs_trans		*tp,
+	xfs_agnumber_t			agno,
+	struct xfs_inobt_rec_incore	*rec)
+{
+	struct xfs_mount		*mp = tp->t_mountp;
+	xfs_agblock_t			sagbno = XFS_AGINO_TO_AGBNO(mp,
+							rec->ir_startino);
+	int				startidx, endidx;
+	int				nextbit;
+	xfs_agblock_t			agbno;
+	int				contigblk;
+	DECLARE_BITMAP(holemask, XFS_INOBT_HOLEMASK_BITS);
+
+	if (!xfs_inobt_issparse(rec->ir_holemask)) {
+		/* not sparse, calculate extent info directly */
+		xfs_free_extent_later(tp, XFS_AGB_TO_FSB(mp, agno, sagbno),
+				  M_IGEO(mp)->ialloc_blks,
+				  &XFS_RMAP_OINFO_INODES);
+		return;
+	}
+
+	/* holemask is only 16-bits (fits in an unsigned long) */
+	ASSERT(sizeof(rec->ir_holemask) <= sizeof(holemask[0]));
+	holemask[0] = rec->ir_holemask;
+
+	/*
+	 * Find contiguous ranges of zeroes (i.e., allocated regions) in the
+	 * holemask and convert the start/end index of each range to an extent.
+	 * We start with the start and end index both pointing at the first 0 in
+	 * the mask.
+	 */
+	startidx = endidx = find_first_zero_bit(holemask,
+						XFS_INOBT_HOLEMASK_BITS);
+	nextbit = startidx + 1;
+	while (startidx < XFS_INOBT_HOLEMASK_BITS) {
+		nextbit = find_next_zero_bit(holemask, XFS_INOBT_HOLEMASK_BITS,
+					     nextbit);
+		/*
+		 * If the next zero bit is contiguous, update the end index of
+		 * the current range and continue.
+		 */
+		if (nextbit != XFS_INOBT_HOLEMASK_BITS &&
+		    nextbit == endidx + 1) {
+			endidx = nextbit;
+			goto next;
+		}
+
+		/*
+		 * nextbit is not contiguous with the current end index. Convert
+		 * the current start/end to an extent and add it to the free
+		 * list.
+		 */
+		agbno = sagbno + (startidx * XFS_INODES_PER_HOLEMASK_BIT) /
+				  mp->m_sb.sb_inopblock;
+		contigblk = ((endidx - startidx + 1) *
+			     XFS_INODES_PER_HOLEMASK_BIT) /
+			    mp->m_sb.sb_inopblock;
+
+		ASSERT(agbno % mp->m_sb.sb_spino_align == 0);
+		ASSERT(contigblk % mp->m_sb.sb_spino_align == 0);
+		xfs_free_extent_later(tp, XFS_AGB_TO_FSB(mp, agno, agbno),
+				  contigblk, &XFS_RMAP_OINFO_INODES);
+
+		/* reset range to current bit and carry on... */
+		startidx = endidx = nextbit;
+
+next:
+		nextbit++;
+	}
+}
+
+STATIC int
+xfs_difree_inobt(
+	struct xfs_mount		*mp,
+	struct xfs_trans		*tp,
+	struct xfs_buf			*agbp,
+	struct xfs_perag		*pag,
+	xfs_agino_t			agino,
+	struct xfs_icluster		*xic,
+	struct xfs_inobt_rec_incore	*orec)
+{
+	struct xfs_agi			*agi = agbp->b_addr;
+	struct xfs_btree_cur		*cur;
+	struct xfs_inobt_rec_incore	rec;
+	int				ilen;
+	int				error;
+	int				i;
+	int				off;
+
+	ASSERT(agi->agi_magicnum == cpu_to_be32(XFS_AGI_MAGIC));
+	ASSERT(XFS_AGINO_TO_AGBNO(mp, agino) < be32_to_cpu(agi->agi_length));
+
+	/*
+	 * Initialize the cursor.
+	 */
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, pag, XFS_BTNUM_INO);
+
+	error = xfs_check_agi_freecount(cur);
+	if (error)
+		goto error0;
+
+	/*
+	 * Look for the entry describing this inode.
+	 */
+	if ((error = xfs_inobt_lookup(cur, agino, XFS_LOOKUP_LE, &i))) {
+		xfs_warn(mp, "%s: xfs_inobt_lookup() returned error %d.",
+			__func__, error);
+		goto error0;
+	}
+	if (XFS_IS_CORRUPT(mp, i != 1)) {
+		error = -EFSCORRUPTED;
+		goto error0;
+	}
+	error = xfs_inobt_get_rec(cur, &rec, &i);
+	if (error) {
+		xfs_warn(mp, "%s: xfs_inobt_get_rec() returned error %d.",
+			__func__, error);
+		goto error0;
+	}
+	if (XFS_IS_CORRUPT(mp, i != 1)) {
+		error = -EFSCORRUPTED;
+		goto error0;
+	}
+	/*
+	 * Get the offset in the inode chunk.
+	 */
+	off = agino - rec.ir_startino;
+	ASSERT(off >= 0 && off < XFS_INODES_PER_CHUNK);
+	ASSERT(!(rec.ir_free & XFS_INOBT_MASK(off)));
+	/*
+	 * Mark the inode free & increment the count.
+	 */
+	rec.ir_free |= XFS_INOBT_MASK(off);
+	rec.ir_freecount++;
+
+	/*
+	 * When an inode chunk is free, it becomes eligible for removal. Don't
+	 * remove the chunk if the block size is large enough for multiple inode
+	 * chunks (that might not be free).
+	 */
+	if (!xfs_has_ikeep(mp) && rec.ir_free == XFS_INOBT_ALL_FREE &&
+	    mp->m_sb.sb_inopblock <= XFS_INODES_PER_CHUNK) {
+		struct xfs_perag	*pag = agbp->b_pag;
+
+		xic->deleted = true;
+		xic->first_ino = XFS_AGINO_TO_INO(mp, pag->pag_agno,
+				rec.ir_startino);
+		xic->alloc = xfs_inobt_irec_to_allocmask(&rec);
+
+		/*
+		 * Remove the inode cluster from the AGI B+Tree, adjust the
+		 * AGI and Superblock inode counts, and mark the disk space
+		 * to be freed when the transaction is committed.
+		 */
+		ilen = rec.ir_freecount;
+		be32_add_cpu(&agi->agi_count, -ilen);
+		be32_add_cpu(&agi->agi_freecount, -(ilen - 1));
+		xfs_ialloc_log_agi(tp, agbp, XFS_AGI_COUNT | XFS_AGI_FREECOUNT);
+		pag->pagi_freecount -= ilen - 1;
+		pag->pagi_count -= ilen;
+		xfs_trans_mod_sb(tp, XFS_TRANS_SB_ICOUNT, -ilen);
+		xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, -(ilen - 1));
+
+		if ((error = xfs_btree_delete(cur, &i))) {
+			xfs_warn(mp, "%s: xfs_btree_delete returned error %d.",
+				__func__, error);
+			goto error0;
+		}
+
+		xfs_difree_inode_chunk(tp, pag->pag_agno, &rec);
+	} else {
+		xic->deleted = false;
+
+		error = xfs_inobt_update(cur, &rec);
+		if (error) {
+			xfs_warn(mp, "%s: xfs_inobt_update returned error %d.",
+				__func__, error);
+			goto error0;
+		}
+
+		/*
+		 * Change the inode free counts and log the ag/sb changes.
+		 */
+		be32_add_cpu(&agi->agi_freecount, 1);
+		xfs_ialloc_log_agi(tp, agbp, XFS_AGI_FREECOUNT);
+		pag->pagi_freecount++;
+		xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, 1);
+	}
+
+	error = xfs_check_agi_freecount(cur);
+	if (error)
+		goto error0;
+
+	*orec = rec;
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	return 0;
+
+error0:
+	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/*
+ * Free an inode in the free inode btree.
+ */
+STATIC int
+xfs_difree_finobt(
+	struct xfs_mount		*mp,
+	struct xfs_trans		*tp,
+	struct xfs_buf			*agbp,
+	struct xfs_perag		*pag,
+	xfs_agino_t			agino,
+	struct xfs_inobt_rec_incore	*ibtrec) /* inobt record */
+{
+	struct xfs_btree_cur		*cur;
+	struct xfs_inobt_rec_incore	rec;
+	int				offset = agino - ibtrec->ir_startino;
+	int				error;
+	int				i;
+
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, pag, XFS_BTNUM_FINO);
+
+	error = xfs_inobt_lookup(cur, ibtrec->ir_startino, XFS_LOOKUP_EQ, &i);
+	if (error)
+		goto error;
+	if (i == 0) {
+		/*
+		 * If the record does not exist in the finobt, we must have just
+		 * freed an inode in a previously fully allocated chunk. If not,
+		 * something is out of sync.
+		 */
+		if (XFS_IS_CORRUPT(mp, ibtrec->ir_freecount != 1)) {
+			error = -EFSCORRUPTED;
+			goto error;
+		}
+
+		error = xfs_inobt_insert_rec(cur, ibtrec->ir_holemask,
+					     ibtrec->ir_count,
+					     ibtrec->ir_freecount,
+					     ibtrec->ir_free, &i);
+		if (error)
+			goto error;
+		ASSERT(i == 1);
+
+		goto out;
+	}
+
+	/*
+	 * Read and update the existing record. We could just copy the ibtrec
+	 * across here, but that would defeat the purpose of having redundant
+	 * metadata. By making the modifications independently, we can catch
+	 * corruptions that we wouldn't see if we just copied from one record
+	 * to another.
+	 */
+	error = xfs_inobt_get_rec(cur, &rec, &i);
+	if (error)
+		goto error;
+	if (XFS_IS_CORRUPT(mp, i != 1)) {
+		error = -EFSCORRUPTED;
+		goto error;
+	}
+
+	rec.ir_free |= XFS_INOBT_MASK(offset);
+	rec.ir_freecount++;
+
+	if (XFS_IS_CORRUPT(mp,
+			   rec.ir_free != ibtrec->ir_free ||
+			   rec.ir_freecount != ibtrec->ir_freecount)) {
+		error = -EFSCORRUPTED;
+		goto error;
+	}
+
+	/*
+	 * The content of inobt records should always match between the inobt
+	 * and finobt. The lifecycle of records in the finobt is different from
+	 * the inobt in that the finobt only tracks records with at least one
+	 * free inode. Hence, if all of the inodes are free and we aren't
+	 * keeping inode chunks permanently on disk, remove the record.
+	 * Otherwise, update the record with the new information.
+	 *
+	 * Note that we currently can't free chunks when the block size is large
+	 * enough for multiple chunks. Leave the finobt record to remain in sync
+	 * with the inobt.
+	 */
+	if (!xfs_has_ikeep(mp) && rec.ir_free == XFS_INOBT_ALL_FREE &&
+	    mp->m_sb.sb_inopblock <= XFS_INODES_PER_CHUNK) {
+		error = xfs_btree_delete(cur, &i);
+		if (error)
+			goto error;
+		ASSERT(i == 1);
+	} else {
+		error = xfs_inobt_update(cur, &rec);
+		if (error)
+			goto error;
+	}
+
+out:
+	error = xfs_check_agi_freecount(cur);
+	if (error)
+		goto error;
+
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	return 0;
+
+error:
+	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/*
+ * Free disk inode.  Carefully avoids touching the incore inode, all
+ * manipulations incore are the caller's responsibility.
+ * The on-disk inode is not changed by this operation, only the
+ * btree (free inode mask) is changed.
+ */
+int
+xfs_difree(
+	struct xfs_trans	*tp,
+	struct xfs_perag	*pag,
+	xfs_ino_t		inode,
+	struct xfs_icluster	*xic)
+{
+	/* REFERENCED */
+	xfs_agblock_t		agbno;	/* block number containing inode */
+	struct xfs_buf		*agbp;	/* buffer for allocation group header */
+	xfs_agino_t		agino;	/* allocation group inode number */
+	int			error;	/* error return value */
+	struct xfs_mount	*mp = tp->t_mountp;
+	struct xfs_inobt_rec_incore rec;/* btree record */
+
+	/*
+	 * Break up inode number into its components.
+	 */
+	if (pag->pag_agno != XFS_INO_TO_AGNO(mp, inode)) {
+		xfs_warn(mp, "%s: agno != pag->pag_agno (%d != %d).",
+			__func__, XFS_INO_TO_AGNO(mp, inode), pag->pag_agno);
+		ASSERT(0);
+		return -EINVAL;
+	}
+	agino = XFS_INO_TO_AGINO(mp, inode);
+	if (inode != XFS_AGINO_TO_INO(mp, pag->pag_agno, agino))  {
+		xfs_warn(mp, "%s: inode != XFS_AGINO_TO_INO() (%llu != %llu).",
+			__func__, (unsigned long long)inode,
+			(unsigned long long)XFS_AGINO_TO_INO(mp, pag->pag_agno, agino));
+		ASSERT(0);
+		return -EINVAL;
+	}
+	agbno = XFS_AGINO_TO_AGBNO(mp, agino);
+	if (agbno >= mp->m_sb.sb_agblocks)  {
+		xfs_warn(mp, "%s: agbno >= mp->m_sb.sb_agblocks (%d >= %d).",
+			__func__, agbno, mp->m_sb.sb_agblocks);
+		ASSERT(0);
+		return -EINVAL;
+	}
+	/*
+	 * Get the allocation group header.
+	 */
+	error = xfs_ialloc_read_agi(pag, tp, &agbp);
+	if (error) {
+		xfs_warn(mp, "%s: xfs_ialloc_read_agi() returned error %d.",
+			__func__, error);
+		return error;
+	}
+
+	/*
+	 * Fix up the inode allocation btree.
+	 */
+	error = xfs_difree_inobt(mp, tp, agbp, pag, agino, xic, &rec);
+	if (error)
+		goto error0;
+
+	/*
+	 * Fix up the free inode btree.
+	 */
+	if (xfs_has_finobt(mp)) {
+		error = xfs_difree_finobt(mp, tp, agbp, pag, agino, &rec);
+		if (error)
+			goto error0;
+	}
+
+	return 0;
+
+error0:
+	return error;
+}
+
+STATIC int
+xfs_imap_lookup(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_perag	*pag,
+	xfs_agino_t		agino,
+	xfs_agblock_t		agbno,
+	xfs_agblock_t		*chunk_agbno,
+	xfs_agblock_t		*offset_agbno,
+	int			flags)
+{
+	struct xfs_inobt_rec_incore rec;
+	struct xfs_btree_cur	*cur;
+	struct xfs_buf		*agbp;
+	int			error;
+	int			i;
+
+	error = xfs_ialloc_read_agi(pag, tp, &agbp);
+	if (error) {
+		xfs_alert(mp,
+			"%s: xfs_ialloc_read_agi() returned error %d, agno %d",
+			__func__, error, pag->pag_agno);
+		return error;
+	}
+
+	/*
+	 * Lookup the inode record for the given agino. If the record cannot be
+	 * found, then it's an invalid inode number and we should abort. Once
+	 * we have a record, we need to ensure it contains the inode number
+	 * we are looking up.
+	 */
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, pag, XFS_BTNUM_INO);
+	error = xfs_inobt_lookup(cur, agino, XFS_LOOKUP_LE, &i);
+	if (!error) {
+		if (i)
+			error = xfs_inobt_get_rec(cur, &rec, &i);
+		if (!error && i == 0)
+			error = -EINVAL;
+	}
+
+	xfs_trans_brelse(tp, agbp);
+	xfs_btree_del_cursor(cur, error);
+	if (error)
+		return error;
+
+	/* check that the returned record contains the required inode */
+	if (rec.ir_startino > agino ||
+	    rec.ir_startino + M_IGEO(mp)->ialloc_inos <= agino)
+		return -EINVAL;
+
+	/* for untrusted inodes check it is allocated first */
+	if ((flags & XFS_IGET_UNTRUSTED) &&
+	    (rec.ir_free & XFS_INOBT_MASK(agino - rec.ir_startino)))
+		return -EINVAL;
+
+	*chunk_agbno = XFS_AGINO_TO_AGBNO(mp, rec.ir_startino);
+	*offset_agbno = agbno - *chunk_agbno;
+	return 0;
+}
+
+/*
+ * Return the location of the inode in imap, for mapping it into a buffer.
+ */
+int
+xfs_imap(
+	struct xfs_mount	 *mp,	/* file system mount structure */
+	struct xfs_trans	 *tp,	/* transaction pointer */
+	xfs_ino_t		ino,	/* inode to locate */
+	struct xfs_imap		*imap,	/* location map structure */
+	uint			flags)	/* flags for inode btree lookup */
+{
+	xfs_agblock_t		agbno;	/* block number of inode in the alloc group */
+	xfs_agino_t		agino;	/* inode number within alloc group */
+	xfs_agblock_t		chunk_agbno;	/* first block in inode chunk */
+	xfs_agblock_t		cluster_agbno;	/* first block in inode cluster */
+	int			error;	/* error code */
+	int			offset;	/* index of inode in its buffer */
+	xfs_agblock_t		offset_agbno;	/* blks from chunk start to inode */
+	struct xfs_perag	*pag;
+
+	ASSERT(ino != NULLFSINO);
+
+	/*
+	 * Split up the inode number into its parts.
+	 */
+	pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, ino));
+	agino = XFS_INO_TO_AGINO(mp, ino);
+	agbno = XFS_AGINO_TO_AGBNO(mp, agino);
+	if (!pag || agbno >= mp->m_sb.sb_agblocks ||
+	    ino != XFS_AGINO_TO_INO(mp, pag->pag_agno, agino)) {
+		error = -EINVAL;
+#ifdef DEBUG
+		/*
+		 * Don't output diagnostic information for untrusted inodes
+		 * as they can be invalid without implying corruption.
+		 */
+		if (flags & XFS_IGET_UNTRUSTED)
+			goto out_drop;
+		if (!pag) {
+			xfs_alert(mp,
+				"%s: agno (%d) >= mp->m_sb.sb_agcount (%d)",
+				__func__, XFS_INO_TO_AGNO(mp, ino),
+				mp->m_sb.sb_agcount);
+		}
+		if (agbno >= mp->m_sb.sb_agblocks) {
+			xfs_alert(mp,
+		"%s: agbno (0x%llx) >= mp->m_sb.sb_agblocks (0x%lx)",
+				__func__, (unsigned long long)agbno,
+				(unsigned long)mp->m_sb.sb_agblocks);
+		}
+		if (pag && ino != XFS_AGINO_TO_INO(mp, pag->pag_agno, agino)) {
+			xfs_alert(mp,
+		"%s: ino (0x%llx) != XFS_AGINO_TO_INO() (0x%llx)",
+				__func__, ino,
+				XFS_AGINO_TO_INO(mp, pag->pag_agno, agino));
+		}
+		xfs_stack_trace();
+#endif /* DEBUG */
+		goto out_drop;
+	}
+
+	/*
+	 * For bulkstat and handle lookups, we have an untrusted inode number
+	 * that we have to verify is valid. We cannot do this just by reading
+	 * the inode buffer as it may have been unlinked and removed leaving
+	 * inodes in stale state on disk. Hence we have to do a btree lookup
+	 * in all cases where an untrusted inode number is passed.
+	 */
+	if (flags & XFS_IGET_UNTRUSTED) {
+		error = xfs_imap_lookup(mp, tp, pag, agino, agbno,
+					&chunk_agbno, &offset_agbno, flags);
+		if (error)
+			goto out_drop;
+		goto out_map;
+	}
+
+	/*
+	 * If the inode cluster size is the same as the blocksize or
+	 * smaller we get to the buffer by simple arithmetics.
+	 */
+	if (M_IGEO(mp)->blocks_per_cluster == 1) {
+		offset = XFS_INO_TO_OFFSET(mp, ino);
+		ASSERT(offset < mp->m_sb.sb_inopblock);
+
+		imap->im_blkno = XFS_AGB_TO_DADDR(mp, pag->pag_agno, agbno);
+		imap->im_len = XFS_FSB_TO_BB(mp, 1);
+		imap->im_boffset = (unsigned short)(offset <<
+							mp->m_sb.sb_inodelog);
+		error = 0;
+		goto out_drop;
+	}
+
+	/*
+	 * If the inode chunks are aligned then use simple maths to
+	 * find the location. Otherwise we have to do a btree
+	 * lookup to find the location.
+	 */
+	if (M_IGEO(mp)->inoalign_mask) {
+		offset_agbno = agbno & M_IGEO(mp)->inoalign_mask;
+		chunk_agbno = agbno - offset_agbno;
+	} else {
+		error = xfs_imap_lookup(mp, tp, pag, agino, agbno,
+					&chunk_agbno, &offset_agbno, flags);
+		if (error)
+			goto out_drop;
+	}
+
+out_map:
+	ASSERT(agbno >= chunk_agbno);
+	cluster_agbno = chunk_agbno +
+		((offset_agbno / M_IGEO(mp)->blocks_per_cluster) *
+		 M_IGEO(mp)->blocks_per_cluster);
+	offset = ((agbno - cluster_agbno) * mp->m_sb.sb_inopblock) +
+		XFS_INO_TO_OFFSET(mp, ino);
+
+	imap->im_blkno = XFS_AGB_TO_DADDR(mp, pag->pag_agno, cluster_agbno);
+	imap->im_len = XFS_FSB_TO_BB(mp, M_IGEO(mp)->blocks_per_cluster);
+	imap->im_boffset = (unsigned short)(offset << mp->m_sb.sb_inodelog);
+
+	/*
+	 * If the inode number maps to a block outside the bounds
+	 * of the file system then return NULL rather than calling
+	 * read_buf and panicing when we get an error from the
+	 * driver.
+	 */
+	if ((imap->im_blkno + imap->im_len) >
+	    XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks)) {
+		xfs_alert(mp,
+	"%s: (im_blkno (0x%llx) + im_len (0x%llx)) > sb_dblocks (0x%llx)",
+			__func__, (unsigned long long) imap->im_blkno,
+			(unsigned long long) imap->im_len,
+			XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks));
+		error = -EINVAL;
+		goto out_drop;
+	}
+	error = 0;
+out_drop:
+	if (pag)
+		xfs_perag_put(pag);
+	return error;
+}
+
+/*
+ * Log specified fields for the ag hdr (inode section). The growth of the agi
+ * structure over time requires that we interpret the buffer as two logical
+ * regions delineated by the end of the unlinked list. This is due to the size
+ * of the hash table and its location in the middle of the agi.
+ *
+ * For example, a request to log a field before agi_unlinked and a field after
+ * agi_unlinked could cause us to log the entire hash table and use an excessive
+ * amount of log space. To avoid this behavior, log the region up through
+ * agi_unlinked in one call and the region after agi_unlinked through the end of
+ * the structure in another.
+ */
+void
+xfs_ialloc_log_agi(
+	struct xfs_trans	*tp,
+	struct xfs_buf		*bp,
+	uint32_t		fields)
+{
+	int			first;		/* first byte number */
+	int			last;		/* last byte number */
+	static const short	offsets[] = {	/* field starting offsets */
+					/* keep in sync with bit definitions */
+		offsetof(xfs_agi_t, agi_magicnum),
+		offsetof(xfs_agi_t, agi_versionnum),
+		offsetof(xfs_agi_t, agi_seqno),
+		offsetof(xfs_agi_t, agi_length),
+		offsetof(xfs_agi_t, agi_count),
+		offsetof(xfs_agi_t, agi_root),
+		offsetof(xfs_agi_t, agi_level),
+		offsetof(xfs_agi_t, agi_freecount),
+		offsetof(xfs_agi_t, agi_newino),
+		offsetof(xfs_agi_t, agi_dirino),
+		offsetof(xfs_agi_t, agi_unlinked),
+		offsetof(xfs_agi_t, agi_free_root),
+		offsetof(xfs_agi_t, agi_free_level),
+		offsetof(xfs_agi_t, agi_iblocks),
+		sizeof(xfs_agi_t)
+	};
+#ifdef DEBUG
+	struct xfs_agi		*agi = bp->b_addr;
+
+	ASSERT(agi->agi_magicnum == cpu_to_be32(XFS_AGI_MAGIC));
+#endif
+
+	/*
+	 * Compute byte offsets for the first and last fields in the first
+	 * region and log the agi buffer. This only logs up through
+	 * agi_unlinked.
+	 */
+	if (fields & XFS_AGI_ALL_BITS_R1) {
+		xfs_btree_offsets(fields, offsets, XFS_AGI_NUM_BITS_R1,
+				  &first, &last);
+		xfs_trans_log_buf(tp, bp, first, last);
+	}
+
+	/*
+	 * Mask off the bits in the first region and calculate the first and
+	 * last field offsets for any bits in the second region.
+	 */
+	fields &= ~XFS_AGI_ALL_BITS_R1;
+	if (fields) {
+		xfs_btree_offsets(fields, offsets, XFS_AGI_NUM_BITS_R2,
+				  &first, &last);
+		xfs_trans_log_buf(tp, bp, first, last);
+	}
+}
+
+static xfs_failaddr_t
+xfs_agi_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount *mp = bp->b_mount;
+	struct xfs_agi	*agi = bp->b_addr;
+	int		i;
+
+	if (xfs_has_crc(mp)) {
+		if (!uuid_equal(&agi->agi_uuid, &mp->m_sb.sb_meta_uuid))
+			return __this_address;
+		if (!xfs_log_check_lsn(mp, be64_to_cpu(agi->agi_lsn)))
+			return __this_address;
+	}
+
+	/*
+	 * Validate the magic number of the agi block.
+	 */
+	if (!xfs_verify_magic(bp, agi->agi_magicnum))
+		return __this_address;
+	if (!XFS_AGI_GOOD_VERSION(be32_to_cpu(agi->agi_versionnum)))
+		return __this_address;
+
+	if (be32_to_cpu(agi->agi_level) < 1 ||
+	    be32_to_cpu(agi->agi_level) > M_IGEO(mp)->inobt_maxlevels)
+		return __this_address;
+
+	if (xfs_has_finobt(mp) &&
+	    (be32_to_cpu(agi->agi_free_level) < 1 ||
+	     be32_to_cpu(agi->agi_free_level) > M_IGEO(mp)->inobt_maxlevels))
+		return __this_address;
+
+	/*
+	 * during growfs operations, the perag is not fully initialised,
+	 * so we can't use it for any useful checking. growfs ensures we can't
+	 * use it by using uncached buffers that don't have the perag attached
+	 * so we can detect and avoid this problem.
+	 */
+	if (bp->b_pag && be32_to_cpu(agi->agi_seqno) != bp->b_pag->pag_agno)
+		return __this_address;
+
+	for (i = 0; i < XFS_AGI_UNLINKED_BUCKETS; i++) {
+		if (agi->agi_unlinked[i] == cpu_to_be32(NULLAGINO))
+			continue;
+		if (!xfs_verify_ino(mp, be32_to_cpu(agi->agi_unlinked[i])))
+			return __this_address;
+	}
+
+	return NULL;
+}
+
+static void
+xfs_agi_read_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount *mp = bp->b_mount;
+	xfs_failaddr_t	fa;
+
+	if (xfs_has_crc(mp) &&
+	    !xfs_buf_verify_cksum(bp, XFS_AGI_CRC_OFF))
+		xfs_verifier_error(bp, -EFSBADCRC, __this_address);
+	else {
+		fa = xfs_agi_verify(bp);
+		if (XFS_TEST_ERROR(fa, mp, XFS_ERRTAG_IALLOC_READ_AGI))
+			xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+	}
+}
+
+static void
+xfs_agi_write_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_buf_log_item	*bip = bp->b_log_item;
+	struct xfs_agi		*agi = bp->b_addr;
+	xfs_failaddr_t		fa;
+
+	fa = xfs_agi_verify(bp);
+	if (fa) {
+		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+		return;
+	}
+
+	if (!xfs_has_crc(mp))
+		return;
+
+	if (bip)
+		agi->agi_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+	xfs_buf_update_cksum(bp, XFS_AGI_CRC_OFF);
+}
+
+const struct xfs_buf_ops xfs_agi_buf_ops = {
+	.name = "xfs_agi",
+	.magic = { cpu_to_be32(XFS_AGI_MAGIC), cpu_to_be32(XFS_AGI_MAGIC) },
+	.verify_read = xfs_agi_read_verify,
+	.verify_write = xfs_agi_write_verify,
+	.verify_struct = xfs_agi_verify,
+};
+
+/*
+ * Read in the allocation group header (inode allocation section)
+ */
+int
+xfs_read_agi(
+	struct xfs_perag	*pag,
+	struct xfs_trans	*tp,
+	struct xfs_buf		**agibpp)
+{
+	struct xfs_mount	*mp = pag->pag_mount;
+	int			error;
+
+	trace_xfs_read_agi(pag->pag_mount, pag->pag_agno);
+
+	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
+			XFS_AG_DADDR(mp, pag->pag_agno, XFS_AGI_DADDR(mp)),
+			XFS_FSS_TO_BB(mp, 1), 0, agibpp, &xfs_agi_buf_ops);
+	if (error)
+		return error;
+	if (tp)
+		xfs_trans_buf_set_type(tp, *agibpp, XFS_BLFT_AGI_BUF);
+
+	xfs_buf_set_ref(*agibpp, XFS_AGI_REF);
+	return 0;
+}
+
+/*
+ * Read in the agi and initialise the per-ag data. If the caller supplies a
+ * @agibpp, return the locked AGI buffer to them, otherwise release it.
+ */
+int
+xfs_ialloc_read_agi(
+	struct xfs_perag	*pag,
+	struct xfs_trans	*tp,
+	struct xfs_buf		**agibpp)
+{
+	struct xfs_buf		*agibp;
+	struct xfs_agi		*agi;
+	int			error;
+
+	trace_xfs_ialloc_read_agi(pag->pag_mount, pag->pag_agno);
+
+	error = xfs_read_agi(pag, tp, &agibp);
+	if (error)
+		return error;
+
+	agi = agibp->b_addr;
+	if (!pag->pagi_init) {
+		pag->pagi_freecount = be32_to_cpu(agi->agi_freecount);
+		pag->pagi_count = be32_to_cpu(agi->agi_count);
+		pag->pagi_init = 1;
+	}
+
+	/*
+	 * It's possible for these to be out of sync if
+	 * we are in the middle of a forced shutdown.
+	 */
+	ASSERT(pag->pagi_freecount == be32_to_cpu(agi->agi_freecount) ||
+		xfs_is_shutdown(pag->pag_mount));
+	if (agibpp)
+		*agibpp = agibp;
+	else
+		xfs_trans_brelse(tp, agibp);
+	return 0;
+}
+
+/* Is there an inode record covering a given range of inode numbers? */
+int
+xfs_ialloc_has_inode_record(
+	struct xfs_btree_cur	*cur,
+	xfs_agino_t		low,
+	xfs_agino_t		high,
+	bool			*exists)
+{
+	struct xfs_inobt_rec_incore	irec;
+	xfs_agino_t		agino;
+	uint16_t		holemask;
+	int			has_record;
+	int			i;
+	int			error;
+
+	*exists = false;
+	error = xfs_inobt_lookup(cur, low, XFS_LOOKUP_LE, &has_record);
+	while (error == 0 && has_record) {
+		error = xfs_inobt_get_rec(cur, &irec, &has_record);
+		if (error || irec.ir_startino > high)
+			break;
+
+		agino = irec.ir_startino;
+		holemask = irec.ir_holemask;
+		for (i = 0; i < XFS_INOBT_HOLEMASK_BITS; holemask >>= 1,
+				i++, agino += XFS_INODES_PER_HOLEMASK_BIT) {
+			if (holemask & 1)
+				continue;
+			if (agino + XFS_INODES_PER_HOLEMASK_BIT > low &&
+					agino <= high) {
+				*exists = true;
+				return 0;
+			}
+		}
+
+		error = xfs_btree_increment(cur, 0, &has_record);
+	}
+	return error;
+}
+
+/* Is there an inode record covering a given extent? */
+int
+xfs_ialloc_has_inodes_at_extent(
+	struct xfs_btree_cur	*cur,
+	xfs_agblock_t		bno,
+	xfs_extlen_t		len,
+	bool			*exists)
+{
+	xfs_agino_t		low;
+	xfs_agino_t		high;
+
+	low = XFS_AGB_TO_AGINO(cur->bc_mp, bno);
+	high = XFS_AGB_TO_AGINO(cur->bc_mp, bno + len) - 1;
+
+	return xfs_ialloc_has_inode_record(cur, low, high, exists);
+}
+
+struct xfs_ialloc_count_inodes {
+	xfs_agino_t			count;
+	xfs_agino_t			freecount;
+};
+
+/* Record inode counts across all inobt records. */
+STATIC int
+xfs_ialloc_count_inodes_rec(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_rec	*rec,
+	void				*priv)
+{
+	struct xfs_inobt_rec_incore	irec;
+	struct xfs_ialloc_count_inodes	*ci = priv;
+
+	xfs_inobt_btrec_to_irec(cur->bc_mp, rec, &irec);
+	ci->count += irec.ir_count;
+	ci->freecount += irec.ir_freecount;
+
+	return 0;
+}
+
+/* Count allocated and free inodes under an inobt. */
+int
+xfs_ialloc_count_inodes(
+	struct xfs_btree_cur		*cur,
+	xfs_agino_t			*count,
+	xfs_agino_t			*freecount)
+{
+	struct xfs_ialloc_count_inodes	ci = {0};
+	int				error;
+
+	ASSERT(cur->bc_btnum == XFS_BTNUM_INO);
+	error = xfs_btree_query_all(cur, xfs_ialloc_count_inodes_rec, &ci);
+	if (error)
+		return error;
+
+	*count = ci.count;
+	*freecount = ci.freecount;
+	return 0;
+}
+
+/*
+ * Initialize inode-related geometry information.
+ *
+ * Compute the inode btree min and max levels and set maxicount.
+ *
+ * Set the inode cluster size.  This may still be overridden by the file
+ * system block size if it is larger than the chosen cluster size.
+ *
+ * For v5 filesystems, scale the cluster size with the inode size to keep a
+ * constant ratio of inode per cluster buffer, but only if mkfs has set the
+ * inode alignment value appropriately for larger cluster sizes.
+ *
+ * Then compute the inode cluster alignment information.
+ */
+void
+xfs_ialloc_setup_geometry(
+	struct xfs_mount	*mp)
+{
+	struct xfs_sb		*sbp = &mp->m_sb;
+	struct xfs_ino_geometry	*igeo = M_IGEO(mp);
+	uint64_t		icount;
+	uint			inodes;
+
+	igeo->new_diflags2 = 0;
+	if (xfs_has_bigtime(mp))
+		igeo->new_diflags2 |= XFS_DIFLAG2_BIGTIME;
+	if (xfs_has_large_extent_counts(mp))
+		igeo->new_diflags2 |= XFS_DIFLAG2_NREXT64;
+
+	/* Compute inode btree geometry. */
+	igeo->agino_log = sbp->sb_inopblog + sbp->sb_agblklog;
+	igeo->inobt_mxr[0] = xfs_inobt_maxrecs(mp, sbp->sb_blocksize, 1);
+	igeo->inobt_mxr[1] = xfs_inobt_maxrecs(mp, sbp->sb_blocksize, 0);
+	igeo->inobt_mnr[0] = igeo->inobt_mxr[0] / 2;
+	igeo->inobt_mnr[1] = igeo->inobt_mxr[1] / 2;
+
+	igeo->ialloc_inos = max_t(uint16_t, XFS_INODES_PER_CHUNK,
+			sbp->sb_inopblock);
+	igeo->ialloc_blks = igeo->ialloc_inos >> sbp->sb_inopblog;
+
+	if (sbp->sb_spino_align)
+		igeo->ialloc_min_blks = sbp->sb_spino_align;
+	else
+		igeo->ialloc_min_blks = igeo->ialloc_blks;
+
+	/* Compute and fill in value of m_ino_geo.inobt_maxlevels. */
+	inodes = (1LL << XFS_INO_AGINO_BITS(mp)) >> XFS_INODES_PER_CHUNK_LOG;
+	igeo->inobt_maxlevels = xfs_btree_compute_maxlevels(igeo->inobt_mnr,
+			inodes);
+	ASSERT(igeo->inobt_maxlevels <= xfs_iallocbt_maxlevels_ondisk());
+
+	/*
+	 * Set the maximum inode count for this filesystem, being careful not
+	 * to use obviously garbage sb_inopblog/sb_inopblock values.  Regular
+	 * users should never get here due to failing sb verification, but
+	 * certain users (xfs_db) need to be usable even with corrupt metadata.
+	 */
+	if (sbp->sb_imax_pct && igeo->ialloc_blks) {
+		/*
+		 * Make sure the maximum inode count is a multiple
+		 * of the units we allocate inodes in.
+		 */
+		icount = sbp->sb_dblocks * sbp->sb_imax_pct;
+		do_div(icount, 100);
+		do_div(icount, igeo->ialloc_blks);
+		igeo->maxicount = XFS_FSB_TO_INO(mp,
+				icount * igeo->ialloc_blks);
+	} else {
+		igeo->maxicount = 0;
+	}
+
+	/*
+	 * Compute the desired size of an inode cluster buffer size, which
+	 * starts at 8K and (on v5 filesystems) scales up with larger inode
+	 * sizes.
+	 *
+	 * Preserve the desired inode cluster size because the sparse inodes
+	 * feature uses that desired size (not the actual size) to compute the
+	 * sparse inode alignment.  The mount code validates this value, so we
+	 * cannot change the behavior.
+	 */
+	igeo->inode_cluster_size_raw = XFS_INODE_BIG_CLUSTER_SIZE;
+	if (xfs_has_v3inodes(mp)) {
+		int	new_size = igeo->inode_cluster_size_raw;
+
+		new_size *= mp->m_sb.sb_inodesize / XFS_DINODE_MIN_SIZE;
+		if (mp->m_sb.sb_inoalignmt >= XFS_B_TO_FSBT(mp, new_size))
+			igeo->inode_cluster_size_raw = new_size;
+	}
+
+	/* Calculate inode cluster ratios. */
+	if (igeo->inode_cluster_size_raw > mp->m_sb.sb_blocksize)
+		igeo->blocks_per_cluster = XFS_B_TO_FSBT(mp,
+				igeo->inode_cluster_size_raw);
+	else
+		igeo->blocks_per_cluster = 1;
+	igeo->inode_cluster_size = XFS_FSB_TO_B(mp, igeo->blocks_per_cluster);
+	igeo->inodes_per_cluster = XFS_FSB_TO_INO(mp, igeo->blocks_per_cluster);
+
+	/* Calculate inode cluster alignment. */
+	if (xfs_has_align(mp) &&
+	    mp->m_sb.sb_inoalignmt >= igeo->blocks_per_cluster)
+		igeo->cluster_align = mp->m_sb.sb_inoalignmt;
+	else
+		igeo->cluster_align = 1;
+	igeo->inoalign_mask = igeo->cluster_align - 1;
+	igeo->cluster_align_inodes = XFS_FSB_TO_INO(mp, igeo->cluster_align);
+
+	/*
+	 * If we are using stripe alignment, check whether
+	 * the stripe unit is a multiple of the inode alignment
+	 */
+	if (mp->m_dalign && igeo->inoalign_mask &&
+	    !(mp->m_dalign & igeo->inoalign_mask))
+		igeo->ialloc_align = mp->m_dalign;
+	else
+		igeo->ialloc_align = 0;
+}
+
+/* Compute the location of the root directory inode that is laid out by mkfs. */
+xfs_ino_t
+xfs_ialloc_calc_rootino(
+	struct xfs_mount	*mp,
+	int			sunit)
+{
+	struct xfs_ino_geometry	*igeo = M_IGEO(mp);
+	xfs_agblock_t		first_bno;
+
+	/*
+	 * Pre-calculate the geometry of AG 0.  We know what it looks like
+	 * because libxfs knows how to create allocation groups now.
+	 *
+	 * first_bno is the first block in which mkfs could possibly have
+	 * allocated the root directory inode, once we factor in the metadata
+	 * that mkfs formats before it.  Namely, the four AG headers...
+	 */
+	first_bno = howmany(4 * mp->m_sb.sb_sectsize, mp->m_sb.sb_blocksize);
+
+	/* ...the two free space btree roots... */
+	first_bno += 2;
+
+	/* ...the inode btree root... */
+	first_bno += 1;
+
+	/* ...the initial AGFL... */
+	first_bno += xfs_alloc_min_freelist(mp, NULL);
+
+	/* ...the free inode btree root... */
+	if (xfs_has_finobt(mp))
+		first_bno++;
+
+	/* ...the reverse mapping btree root... */
+	if (xfs_has_rmapbt(mp))
+		first_bno++;
+
+	/* ...the reference count btree... */
+	if (xfs_has_reflink(mp))
+		first_bno++;
+
+	/*
+	 * ...and the log, if it is allocated in the first allocation group.
+	 *
+	 * This can happen with filesystems that only have a single
+	 * allocation group, or very odd geometries created by old mkfs
+	 * versions on very small filesystems.
+	 */
+	if (xfs_ag_contains_log(mp, 0))
+		 first_bno += mp->m_sb.sb_logblocks;
+
+	/*
+	 * Now round first_bno up to whatever allocation alignment is given
+	 * by the filesystem or was passed in.
+	 */
+	if (xfs_has_dalign(mp) && igeo->ialloc_align > 0)
+		first_bno = roundup(first_bno, sunit);
+	else if (xfs_has_align(mp) &&
+			mp->m_sb.sb_inoalignmt > 1)
+		first_bno = roundup(first_bno, mp->m_sb.sb_inoalignmt);
+
+	return XFS_AGINO_TO_INO(mp, 0, XFS_AGB_TO_AGINO(mp, first_bno));
+}
+
+/*
+ * Ensure there are not sparse inode clusters that cross the new EOAG.
+ *
+ * This is a no-op for non-spinode filesystems since clusters are always fully
+ * allocated and checking the bnobt suffices.  However, a spinode filesystem
+ * could have a record where the upper inodes are free blocks.  If those blocks
+ * were removed from the filesystem, the inode record would extend beyond EOAG,
+ * which will be flagged as corruption.
+ */
+int
+xfs_ialloc_check_shrink(
+	struct xfs_trans	*tp,
+	xfs_agnumber_t		agno,
+	struct xfs_buf		*agibp,
+	xfs_agblock_t		new_length)
+{
+	struct xfs_inobt_rec_incore rec;
+	struct xfs_btree_cur	*cur;
+	struct xfs_mount	*mp = tp->t_mountp;
+	struct xfs_perag	*pag;
+	xfs_agino_t		agino = XFS_AGB_TO_AGINO(mp, new_length);
+	int			has;
+	int			error;
+
+	if (!xfs_has_sparseinodes(mp))
+		return 0;
+
+	pag = xfs_perag_get(mp, agno);
+	cur = xfs_inobt_init_cursor(mp, tp, agibp, pag, XFS_BTNUM_INO);
+
+	/* Look up the inobt record that would correspond to the new EOFS. */
+	error = xfs_inobt_lookup(cur, agino, XFS_LOOKUP_LE, &has);
+	if (error || !has)
+		goto out;
+
+	error = xfs_inobt_get_rec(cur, &rec, &has);
+	if (error)
+		goto out;
+
+	if (!has) {
+		error = -EFSCORRUPTED;
+		goto out;
+	}
+
+	/* If the record covers inodes that would be beyond EOFS, bail out. */
+	if (rec.ir_startino + XFS_INODES_PER_CHUNK > agino) {
+		error = -ENOSPC;
+		goto out;
+	}
+out:
+	xfs_btree_del_cursor(cur, error);
+	xfs_perag_put(pag);
+	return error;
+}
diff --git a/fs/xfs/libxfs/xfs_ialloc.h b/fs/xfs/libxfs/xfs_ialloc.h
new file mode 100644
index 000000000..9bbbca6ac
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_ialloc.h
@@ -0,0 +1,112 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_IALLOC_H__
+#define	__XFS_IALLOC_H__
+
+struct xfs_buf;
+struct xfs_dinode;
+struct xfs_imap;
+struct xfs_mount;
+struct xfs_trans;
+struct xfs_btree_cur;
+
+/* Move inodes in clusters of this size */
+#define	XFS_INODE_BIG_CLUSTER_SIZE	8192
+
+struct xfs_icluster {
+	bool		deleted;	/* record is deleted */
+	xfs_ino_t	first_ino;	/* first inode number */
+	uint64_t	alloc;		/* inode phys. allocation bitmap for
+					 * sparse chunks */
+};
+
+/*
+ * Make an inode pointer out of the buffer/offset.
+ */
+static inline struct xfs_dinode *
+xfs_make_iptr(struct xfs_mount *mp, struct xfs_buf *b, int o)
+{
+	return xfs_buf_offset(b, o << (mp)->m_sb.sb_inodelog);
+}
+
+/*
+ * Allocate an inode on disk.  Mode is used to tell whether the new inode will
+ * need space, and whether it is a directory.
+ */
+int xfs_dialloc(struct xfs_trans **tpp, xfs_ino_t parent, umode_t mode,
+		xfs_ino_t *new_ino);
+
+int xfs_difree(struct xfs_trans *tp, struct xfs_perag *pag,
+		xfs_ino_t ino, struct xfs_icluster *ifree);
+
+/*
+ * Return the location of the inode in imap, for mapping it into a buffer.
+ */
+int
+xfs_imap(
+	struct xfs_mount *mp,		/* file system mount structure */
+	struct xfs_trans *tp,		/* transaction pointer */
+	xfs_ino_t	ino,		/* inode to locate */
+	struct xfs_imap	*imap,		/* location map structure */
+	uint		flags);		/* flags for inode btree lookup */
+
+/*
+ * Log specified fields for the ag hdr (inode section)
+ */
+void
+xfs_ialloc_log_agi(
+	struct xfs_trans *tp,		/* transaction pointer */
+	struct xfs_buf	*bp,		/* allocation group header buffer */
+	uint32_t	fields);	/* bitmask of fields to log */
+
+int xfs_read_agi(struct xfs_perag *pag, struct xfs_trans *tp,
+		struct xfs_buf **agibpp);
+int xfs_ialloc_read_agi(struct xfs_perag *pag, struct xfs_trans *tp,
+		struct xfs_buf **agibpp);
+
+/*
+ * Lookup a record by ino in the btree given by cur.
+ */
+int xfs_inobt_lookup(struct xfs_btree_cur *cur, xfs_agino_t ino,
+		xfs_lookup_t dir, int *stat);
+
+/*
+ * Get the data from the pointed-to record.
+ */
+int xfs_inobt_get_rec(struct xfs_btree_cur *cur,
+		xfs_inobt_rec_incore_t *rec, int *stat);
+
+/*
+ * Inode chunk initialisation routine
+ */
+int xfs_ialloc_inode_init(struct xfs_mount *mp, struct xfs_trans *tp,
+			  struct list_head *buffer_list, int icount,
+			  xfs_agnumber_t agno, xfs_agblock_t agbno,
+			  xfs_agblock_t length, unsigned int gen);
+
+
+union xfs_btree_rec;
+void xfs_inobt_btrec_to_irec(struct xfs_mount *mp,
+		const union xfs_btree_rec *rec,
+		struct xfs_inobt_rec_incore *irec);
+int xfs_ialloc_has_inodes_at_extent(struct xfs_btree_cur *cur,
+		xfs_agblock_t bno, xfs_extlen_t len, bool *exists);
+int xfs_ialloc_has_inode_record(struct xfs_btree_cur *cur, xfs_agino_t low,
+		xfs_agino_t high, bool *exists);
+int xfs_ialloc_count_inodes(struct xfs_btree_cur *cur, xfs_agino_t *count,
+		xfs_agino_t *freecount);
+int xfs_inobt_insert_rec(struct xfs_btree_cur *cur, uint16_t holemask,
+		uint8_t count, int32_t freecount, xfs_inofree_t free,
+		int *stat);
+
+int xfs_ialloc_cluster_alignment(struct xfs_mount *mp);
+void xfs_ialloc_setup_geometry(struct xfs_mount *mp);
+xfs_ino_t xfs_ialloc_calc_rootino(struct xfs_mount *mp, int sunit);
+
+int xfs_ialloc_check_shrink(struct xfs_trans *tp, xfs_agnumber_t agno,
+		struct xfs_buf *agibp, xfs_agblock_t new_length);
+
+#endif	/* __XFS_IALLOC_H__ */
diff --git a/fs/xfs/libxfs/xfs_ialloc_btree.c b/fs/xfs/libxfs/xfs_ialloc_btree.c
new file mode 100644
index 000000000..8c83e2657
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_ialloc_btree.c
@@ -0,0 +1,833 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_bit.h"
+#include "xfs_mount.h"
+#include "xfs_btree.h"
+#include "xfs_btree_staging.h"
+#include "xfs_ialloc.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_alloc.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+#include "xfs_trans.h"
+#include "xfs_rmap.h"
+#include "xfs_ag.h"
+
+static struct kmem_cache	*xfs_inobt_cur_cache;
+
+STATIC int
+xfs_inobt_get_minrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	return M_IGEO(cur->bc_mp)->inobt_mnr[level != 0];
+}
+
+STATIC struct xfs_btree_cur *
+xfs_inobt_dup_cursor(
+	struct xfs_btree_cur	*cur)
+{
+	return xfs_inobt_init_cursor(cur->bc_mp, cur->bc_tp,
+			cur->bc_ag.agbp, cur->bc_ag.pag, cur->bc_btnum);
+}
+
+STATIC void
+xfs_inobt_set_root(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*nptr,
+	int				inc)	/* level change */
+{
+	struct xfs_buf		*agbp = cur->bc_ag.agbp;
+	struct xfs_agi		*agi = agbp->b_addr;
+
+	agi->agi_root = nptr->s;
+	be32_add_cpu(&agi->agi_level, inc);
+	xfs_ialloc_log_agi(cur->bc_tp, agbp, XFS_AGI_ROOT | XFS_AGI_LEVEL);
+}
+
+STATIC void
+xfs_finobt_set_root(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*nptr,
+	int				inc)	/* level change */
+{
+	struct xfs_buf		*agbp = cur->bc_ag.agbp;
+	struct xfs_agi		*agi = agbp->b_addr;
+
+	agi->agi_free_root = nptr->s;
+	be32_add_cpu(&agi->agi_free_level, inc);
+	xfs_ialloc_log_agi(cur->bc_tp, agbp,
+			   XFS_AGI_FREE_ROOT | XFS_AGI_FREE_LEVEL);
+}
+
+/* Update the inode btree block counter for this btree. */
+static inline void
+xfs_inobt_mod_blockcount(
+	struct xfs_btree_cur	*cur,
+	int			howmuch)
+{
+	struct xfs_buf		*agbp = cur->bc_ag.agbp;
+	struct xfs_agi		*agi = agbp->b_addr;
+
+	if (!xfs_has_inobtcounts(cur->bc_mp))
+		return;
+
+	if (cur->bc_btnum == XFS_BTNUM_FINO)
+		be32_add_cpu(&agi->agi_fblocks, howmuch);
+	else if (cur->bc_btnum == XFS_BTNUM_INO)
+		be32_add_cpu(&agi->agi_iblocks, howmuch);
+	xfs_ialloc_log_agi(cur->bc_tp, agbp, XFS_AGI_IBLOCKS);
+}
+
+STATIC int
+__xfs_inobt_alloc_block(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*start,
+	union xfs_btree_ptr		*new,
+	int				*stat,
+	enum xfs_ag_resv_type		resv)
+{
+	xfs_alloc_arg_t		args;		/* block allocation args */
+	int			error;		/* error return value */
+	xfs_agblock_t		sbno = be32_to_cpu(start->s);
+
+	memset(&args, 0, sizeof(args));
+	args.tp = cur->bc_tp;
+	args.mp = cur->bc_mp;
+	args.oinfo = XFS_RMAP_OINFO_INOBT;
+	args.fsbno = XFS_AGB_TO_FSB(args.mp, cur->bc_ag.pag->pag_agno, sbno);
+	args.minlen = 1;
+	args.maxlen = 1;
+	args.prod = 1;
+	args.type = XFS_ALLOCTYPE_NEAR_BNO;
+	args.resv = resv;
+
+	error = xfs_alloc_vextent(&args);
+	if (error)
+		return error;
+
+	if (args.fsbno == NULLFSBLOCK) {
+		*stat = 0;
+		return 0;
+	}
+	ASSERT(args.len == 1);
+
+	new->s = cpu_to_be32(XFS_FSB_TO_AGBNO(args.mp, args.fsbno));
+	*stat = 1;
+	xfs_inobt_mod_blockcount(cur, 1);
+	return 0;
+}
+
+STATIC int
+xfs_inobt_alloc_block(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*start,
+	union xfs_btree_ptr		*new,
+	int				*stat)
+{
+	return __xfs_inobt_alloc_block(cur, start, new, stat, XFS_AG_RESV_NONE);
+}
+
+STATIC int
+xfs_finobt_alloc_block(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*start,
+	union xfs_btree_ptr		*new,
+	int				*stat)
+{
+	if (cur->bc_mp->m_finobt_nores)
+		return xfs_inobt_alloc_block(cur, start, new, stat);
+	return __xfs_inobt_alloc_block(cur, start, new, stat,
+			XFS_AG_RESV_METADATA);
+}
+
+STATIC int
+__xfs_inobt_free_block(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp,
+	enum xfs_ag_resv_type	resv)
+{
+	xfs_inobt_mod_blockcount(cur, -1);
+	return xfs_free_extent(cur->bc_tp,
+			XFS_DADDR_TO_FSB(cur->bc_mp, xfs_buf_daddr(bp)), 1,
+			&XFS_RMAP_OINFO_INOBT, resv);
+}
+
+STATIC int
+xfs_inobt_free_block(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp)
+{
+	return __xfs_inobt_free_block(cur, bp, XFS_AG_RESV_NONE);
+}
+
+STATIC int
+xfs_finobt_free_block(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp)
+{
+	if (cur->bc_mp->m_finobt_nores)
+		return xfs_inobt_free_block(cur, bp);
+	return __xfs_inobt_free_block(cur, bp, XFS_AG_RESV_METADATA);
+}
+
+STATIC int
+xfs_inobt_get_maxrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	return M_IGEO(cur->bc_mp)->inobt_mxr[level != 0];
+}
+
+STATIC void
+xfs_inobt_init_key_from_rec(
+	union xfs_btree_key		*key,
+	const union xfs_btree_rec	*rec)
+{
+	key->inobt.ir_startino = rec->inobt.ir_startino;
+}
+
+STATIC void
+xfs_inobt_init_high_key_from_rec(
+	union xfs_btree_key		*key,
+	const union xfs_btree_rec	*rec)
+{
+	__u32				x;
+
+	x = be32_to_cpu(rec->inobt.ir_startino);
+	x += XFS_INODES_PER_CHUNK - 1;
+	key->inobt.ir_startino = cpu_to_be32(x);
+}
+
+STATIC void
+xfs_inobt_init_rec_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec)
+{
+	rec->inobt.ir_startino = cpu_to_be32(cur->bc_rec.i.ir_startino);
+	if (xfs_has_sparseinodes(cur->bc_mp)) {
+		rec->inobt.ir_u.sp.ir_holemask =
+					cpu_to_be16(cur->bc_rec.i.ir_holemask);
+		rec->inobt.ir_u.sp.ir_count = cur->bc_rec.i.ir_count;
+		rec->inobt.ir_u.sp.ir_freecount = cur->bc_rec.i.ir_freecount;
+	} else {
+		/* ir_holemask/ir_count not supported on-disk */
+		rec->inobt.ir_u.f.ir_freecount =
+					cpu_to_be32(cur->bc_rec.i.ir_freecount);
+	}
+	rec->inobt.ir_free = cpu_to_be64(cur->bc_rec.i.ir_free);
+}
+
+/*
+ * initial value of ptr for lookup
+ */
+STATIC void
+xfs_inobt_init_ptr_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	struct xfs_agi		*agi = cur->bc_ag.agbp->b_addr;
+
+	ASSERT(cur->bc_ag.pag->pag_agno == be32_to_cpu(agi->agi_seqno));
+
+	ptr->s = agi->agi_root;
+}
+
+STATIC void
+xfs_finobt_init_ptr_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	struct xfs_agi		*agi = cur->bc_ag.agbp->b_addr;
+
+	ASSERT(cur->bc_ag.pag->pag_agno == be32_to_cpu(agi->agi_seqno));
+	ptr->s = agi->agi_free_root;
+}
+
+STATIC int64_t
+xfs_inobt_key_diff(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*key)
+{
+	return (int64_t)be32_to_cpu(key->inobt.ir_startino) -
+			  cur->bc_rec.i.ir_startino;
+}
+
+STATIC int64_t
+xfs_inobt_diff_two_keys(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*k1,
+	const union xfs_btree_key	*k2)
+{
+	return (int64_t)be32_to_cpu(k1->inobt.ir_startino) -
+			  be32_to_cpu(k2->inobt.ir_startino);
+}
+
+static xfs_failaddr_t
+xfs_inobt_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
+	xfs_failaddr_t		fa;
+	unsigned int		level;
+
+	if (!xfs_verify_magic(bp, block->bb_magic))
+		return __this_address;
+
+	/*
+	 * During growfs operations, we can't verify the exact owner as the
+	 * perag is not fully initialised and hence not attached to the buffer.
+	 *
+	 * Similarly, during log recovery we will have a perag structure
+	 * attached, but the agi information will not yet have been initialised
+	 * from the on disk AGI. We don't currently use any of this information,
+	 * but beware of the landmine (i.e. need to check pag->pagi_init) if we
+	 * ever do.
+	 */
+	if (xfs_has_crc(mp)) {
+		fa = xfs_btree_sblock_v5hdr_verify(bp);
+		if (fa)
+			return fa;
+	}
+
+	/* level verification */
+	level = be16_to_cpu(block->bb_level);
+	if (level >= M_IGEO(mp)->inobt_maxlevels)
+		return __this_address;
+
+	return xfs_btree_sblock_verify(bp,
+			M_IGEO(mp)->inobt_mxr[level != 0]);
+}
+
+static void
+xfs_inobt_read_verify(
+	struct xfs_buf	*bp)
+{
+	xfs_failaddr_t	fa;
+
+	if (!xfs_btree_sblock_verify_crc(bp))
+		xfs_verifier_error(bp, -EFSBADCRC, __this_address);
+	else {
+		fa = xfs_inobt_verify(bp);
+		if (fa)
+			xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+	}
+
+	if (bp->b_error)
+		trace_xfs_btree_corrupt(bp, _RET_IP_);
+}
+
+static void
+xfs_inobt_write_verify(
+	struct xfs_buf	*bp)
+{
+	xfs_failaddr_t	fa;
+
+	fa = xfs_inobt_verify(bp);
+	if (fa) {
+		trace_xfs_btree_corrupt(bp, _RET_IP_);
+		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+		return;
+	}
+	xfs_btree_sblock_calc_crc(bp);
+
+}
+
+const struct xfs_buf_ops xfs_inobt_buf_ops = {
+	.name = "xfs_inobt",
+	.magic = { cpu_to_be32(XFS_IBT_MAGIC), cpu_to_be32(XFS_IBT_CRC_MAGIC) },
+	.verify_read = xfs_inobt_read_verify,
+	.verify_write = xfs_inobt_write_verify,
+	.verify_struct = xfs_inobt_verify,
+};
+
+const struct xfs_buf_ops xfs_finobt_buf_ops = {
+	.name = "xfs_finobt",
+	.magic = { cpu_to_be32(XFS_FIBT_MAGIC),
+		   cpu_to_be32(XFS_FIBT_CRC_MAGIC) },
+	.verify_read = xfs_inobt_read_verify,
+	.verify_write = xfs_inobt_write_verify,
+	.verify_struct = xfs_inobt_verify,
+};
+
+STATIC int
+xfs_inobt_keys_inorder(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*k1,
+	const union xfs_btree_key	*k2)
+{
+	return be32_to_cpu(k1->inobt.ir_startino) <
+		be32_to_cpu(k2->inobt.ir_startino);
+}
+
+STATIC int
+xfs_inobt_recs_inorder(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_rec	*r1,
+	const union xfs_btree_rec	*r2)
+{
+	return be32_to_cpu(r1->inobt.ir_startino) + XFS_INODES_PER_CHUNK <=
+		be32_to_cpu(r2->inobt.ir_startino);
+}
+
+static const struct xfs_btree_ops xfs_inobt_ops = {
+	.rec_len		= sizeof(xfs_inobt_rec_t),
+	.key_len		= sizeof(xfs_inobt_key_t),
+
+	.dup_cursor		= xfs_inobt_dup_cursor,
+	.set_root		= xfs_inobt_set_root,
+	.alloc_block		= xfs_inobt_alloc_block,
+	.free_block		= xfs_inobt_free_block,
+	.get_minrecs		= xfs_inobt_get_minrecs,
+	.get_maxrecs		= xfs_inobt_get_maxrecs,
+	.init_key_from_rec	= xfs_inobt_init_key_from_rec,
+	.init_high_key_from_rec	= xfs_inobt_init_high_key_from_rec,
+	.init_rec_from_cur	= xfs_inobt_init_rec_from_cur,
+	.init_ptr_from_cur	= xfs_inobt_init_ptr_from_cur,
+	.key_diff		= xfs_inobt_key_diff,
+	.buf_ops		= &xfs_inobt_buf_ops,
+	.diff_two_keys		= xfs_inobt_diff_two_keys,
+	.keys_inorder		= xfs_inobt_keys_inorder,
+	.recs_inorder		= xfs_inobt_recs_inorder,
+};
+
+static const struct xfs_btree_ops xfs_finobt_ops = {
+	.rec_len		= sizeof(xfs_inobt_rec_t),
+	.key_len		= sizeof(xfs_inobt_key_t),
+
+	.dup_cursor		= xfs_inobt_dup_cursor,
+	.set_root		= xfs_finobt_set_root,
+	.alloc_block		= xfs_finobt_alloc_block,
+	.free_block		= xfs_finobt_free_block,
+	.get_minrecs		= xfs_inobt_get_minrecs,
+	.get_maxrecs		= xfs_inobt_get_maxrecs,
+	.init_key_from_rec	= xfs_inobt_init_key_from_rec,
+	.init_high_key_from_rec	= xfs_inobt_init_high_key_from_rec,
+	.init_rec_from_cur	= xfs_inobt_init_rec_from_cur,
+	.init_ptr_from_cur	= xfs_finobt_init_ptr_from_cur,
+	.key_diff		= xfs_inobt_key_diff,
+	.buf_ops		= &xfs_finobt_buf_ops,
+	.diff_two_keys		= xfs_inobt_diff_two_keys,
+	.keys_inorder		= xfs_inobt_keys_inorder,
+	.recs_inorder		= xfs_inobt_recs_inorder,
+};
+
+/*
+ * Initialize a new inode btree cursor.
+ */
+static struct xfs_btree_cur *
+xfs_inobt_init_common(
+	struct xfs_mount	*mp,		/* file system mount point */
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_perag	*pag,
+	xfs_btnum_t		btnum)		/* ialloc or free ino btree */
+{
+	struct xfs_btree_cur	*cur;
+
+	cur = xfs_btree_alloc_cursor(mp, tp, btnum,
+			M_IGEO(mp)->inobt_maxlevels, xfs_inobt_cur_cache);
+	if (btnum == XFS_BTNUM_INO) {
+		cur->bc_statoff = XFS_STATS_CALC_INDEX(xs_ibt_2);
+		cur->bc_ops = &xfs_inobt_ops;
+	} else {
+		cur->bc_statoff = XFS_STATS_CALC_INDEX(xs_fibt_2);
+		cur->bc_ops = &xfs_finobt_ops;
+	}
+
+	if (xfs_has_crc(mp))
+		cur->bc_flags |= XFS_BTREE_CRC_BLOCKS;
+
+	/* take a reference for the cursor */
+	atomic_inc(&pag->pag_ref);
+	cur->bc_ag.pag = pag;
+	return cur;
+}
+
+/* Create an inode btree cursor. */
+struct xfs_btree_cur *
+xfs_inobt_init_cursor(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp,
+	struct xfs_perag	*pag,
+	xfs_btnum_t		btnum)
+{
+	struct xfs_btree_cur	*cur;
+	struct xfs_agi		*agi = agbp->b_addr;
+
+	cur = xfs_inobt_init_common(mp, tp, pag, btnum);
+	if (btnum == XFS_BTNUM_INO)
+		cur->bc_nlevels = be32_to_cpu(agi->agi_level);
+	else
+		cur->bc_nlevels = be32_to_cpu(agi->agi_free_level);
+	cur->bc_ag.agbp = agbp;
+	return cur;
+}
+
+/* Create an inode btree cursor with a fake root for staging. */
+struct xfs_btree_cur *
+xfs_inobt_stage_cursor(
+	struct xfs_mount	*mp,
+	struct xbtree_afakeroot	*afake,
+	struct xfs_perag	*pag,
+	xfs_btnum_t		btnum)
+{
+	struct xfs_btree_cur	*cur;
+
+	cur = xfs_inobt_init_common(mp, NULL, pag, btnum);
+	xfs_btree_stage_afakeroot(cur, afake);
+	return cur;
+}
+
+/*
+ * Install a new inobt btree root.  Caller is responsible for invalidating
+ * and freeing the old btree blocks.
+ */
+void
+xfs_inobt_commit_staged_btree(
+	struct xfs_btree_cur	*cur,
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp)
+{
+	struct xfs_agi		*agi = agbp->b_addr;
+	struct xbtree_afakeroot	*afake = cur->bc_ag.afake;
+	int			fields;
+
+	ASSERT(cur->bc_flags & XFS_BTREE_STAGING);
+
+	if (cur->bc_btnum == XFS_BTNUM_INO) {
+		fields = XFS_AGI_ROOT | XFS_AGI_LEVEL;
+		agi->agi_root = cpu_to_be32(afake->af_root);
+		agi->agi_level = cpu_to_be32(afake->af_levels);
+		if (xfs_has_inobtcounts(cur->bc_mp)) {
+			agi->agi_iblocks = cpu_to_be32(afake->af_blocks);
+			fields |= XFS_AGI_IBLOCKS;
+		}
+		xfs_ialloc_log_agi(tp, agbp, fields);
+		xfs_btree_commit_afakeroot(cur, tp, agbp, &xfs_inobt_ops);
+	} else {
+		fields = XFS_AGI_FREE_ROOT | XFS_AGI_FREE_LEVEL;
+		agi->agi_free_root = cpu_to_be32(afake->af_root);
+		agi->agi_free_level = cpu_to_be32(afake->af_levels);
+		if (xfs_has_inobtcounts(cur->bc_mp)) {
+			agi->agi_fblocks = cpu_to_be32(afake->af_blocks);
+			fields |= XFS_AGI_IBLOCKS;
+		}
+		xfs_ialloc_log_agi(tp, agbp, fields);
+		xfs_btree_commit_afakeroot(cur, tp, agbp, &xfs_finobt_ops);
+	}
+}
+
+/* Calculate number of records in an inode btree block. */
+static inline unsigned int
+xfs_inobt_block_maxrecs(
+	unsigned int		blocklen,
+	bool			leaf)
+{
+	if (leaf)
+		return blocklen / sizeof(xfs_inobt_rec_t);
+	return blocklen / (sizeof(xfs_inobt_key_t) + sizeof(xfs_inobt_ptr_t));
+}
+
+/*
+ * Calculate number of records in an inobt btree block.
+ */
+int
+xfs_inobt_maxrecs(
+	struct xfs_mount	*mp,
+	int			blocklen,
+	int			leaf)
+{
+	blocklen -= XFS_INOBT_BLOCK_LEN(mp);
+	return xfs_inobt_block_maxrecs(blocklen, leaf);
+}
+
+/*
+ * Maximum number of inode btree records per AG.  Pretend that we can fill an
+ * entire AG completely full of inodes except for the AG headers.
+ */
+#define XFS_MAX_INODE_RECORDS \
+	((XFS_MAX_AG_BYTES - (4 * BBSIZE)) / XFS_DINODE_MIN_SIZE) / \
+			XFS_INODES_PER_CHUNK
+
+/* Compute the max possible height for the inode btree. */
+static inline unsigned int
+xfs_inobt_maxlevels_ondisk(void)
+{
+	unsigned int		minrecs[2];
+	unsigned int		blocklen;
+
+	blocklen = min(XFS_MIN_BLOCKSIZE - XFS_BTREE_SBLOCK_LEN,
+		       XFS_MIN_CRC_BLOCKSIZE - XFS_BTREE_SBLOCK_CRC_LEN);
+
+	minrecs[0] = xfs_inobt_block_maxrecs(blocklen, true) / 2;
+	minrecs[1] = xfs_inobt_block_maxrecs(blocklen, false) / 2;
+
+	return xfs_btree_compute_maxlevels(minrecs, XFS_MAX_INODE_RECORDS);
+}
+
+/* Compute the max possible height for the free inode btree. */
+static inline unsigned int
+xfs_finobt_maxlevels_ondisk(void)
+{
+	unsigned int		minrecs[2];
+	unsigned int		blocklen;
+
+	blocklen = XFS_MIN_CRC_BLOCKSIZE - XFS_BTREE_SBLOCK_CRC_LEN;
+
+	minrecs[0] = xfs_inobt_block_maxrecs(blocklen, true) / 2;
+	minrecs[1] = xfs_inobt_block_maxrecs(blocklen, false) / 2;
+
+	return xfs_btree_compute_maxlevels(minrecs, XFS_MAX_INODE_RECORDS);
+}
+
+/* Compute the max possible height for either inode btree. */
+unsigned int
+xfs_iallocbt_maxlevels_ondisk(void)
+{
+	return max(xfs_inobt_maxlevels_ondisk(),
+		   xfs_finobt_maxlevels_ondisk());
+}
+
+/*
+ * Convert the inode record holemask to an inode allocation bitmap. The inode
+ * allocation bitmap is inode granularity and specifies whether an inode is
+ * physically allocated on disk (not whether the inode is considered allocated
+ * or free by the fs).
+ *
+ * A bit value of 1 means the inode is allocated, a value of 0 means it is free.
+ */
+uint64_t
+xfs_inobt_irec_to_allocmask(
+	struct xfs_inobt_rec_incore	*rec)
+{
+	uint64_t			bitmap = 0;
+	uint64_t			inodespbit;
+	int				nextbit;
+	uint				allocbitmap;
+
+	/*
+	 * The holemask has 16-bits for a 64 inode record. Therefore each
+	 * holemask bit represents multiple inodes. Create a mask of bits to set
+	 * in the allocmask for each holemask bit.
+	 */
+	inodespbit = (1 << XFS_INODES_PER_HOLEMASK_BIT) - 1;
+
+	/*
+	 * Allocated inodes are represented by 0 bits in holemask. Invert the 0
+	 * bits to 1 and convert to a uint so we can use xfs_next_bit(). Mask
+	 * anything beyond the 16 holemask bits since this casts to a larger
+	 * type.
+	 */
+	allocbitmap = ~rec->ir_holemask & ((1 << XFS_INOBT_HOLEMASK_BITS) - 1);
+
+	/*
+	 * allocbitmap is the inverted holemask so every set bit represents
+	 * allocated inodes. To expand from 16-bit holemask granularity to
+	 * 64-bit (e.g., bit-per-inode), set inodespbit bits in the target
+	 * bitmap for every holemask bit.
+	 */
+	nextbit = xfs_next_bit(&allocbitmap, 1, 0);
+	while (nextbit != -1) {
+		ASSERT(nextbit < (sizeof(rec->ir_holemask) * NBBY));
+
+		bitmap |= (inodespbit <<
+			   (nextbit * XFS_INODES_PER_HOLEMASK_BIT));
+
+		nextbit = xfs_next_bit(&allocbitmap, 1, nextbit + 1);
+	}
+
+	return bitmap;
+}
+
+#if defined(DEBUG) || defined(XFS_WARN)
+/*
+ * Verify that an in-core inode record has a valid inode count.
+ */
+int
+xfs_inobt_rec_check_count(
+	struct xfs_mount		*mp,
+	struct xfs_inobt_rec_incore	*rec)
+{
+	int				inocount = 0;
+	int				nextbit = 0;
+	uint64_t			allocbmap;
+	int				wordsz;
+
+	wordsz = sizeof(allocbmap) / sizeof(unsigned int);
+	allocbmap = xfs_inobt_irec_to_allocmask(rec);
+
+	nextbit = xfs_next_bit((uint *) &allocbmap, wordsz, nextbit);
+	while (nextbit != -1) {
+		inocount++;
+		nextbit = xfs_next_bit((uint *) &allocbmap, wordsz,
+				       nextbit + 1);
+	}
+
+	if (inocount != rec->ir_count)
+		return -EFSCORRUPTED;
+
+	return 0;
+}
+#endif	/* DEBUG */
+
+static xfs_extlen_t
+xfs_inobt_max_size(
+	struct xfs_perag	*pag)
+{
+	struct xfs_mount	*mp = pag->pag_mount;
+	xfs_agblock_t		agblocks = pag->block_count;
+
+	/* Bail out if we're uninitialized, which can happen in mkfs. */
+	if (M_IGEO(mp)->inobt_mxr[0] == 0)
+		return 0;
+
+	/*
+	 * The log is permanently allocated, so the space it occupies will
+	 * never be available for the kinds of things that would require btree
+	 * expansion.  We therefore can pretend the space isn't there.
+	 */
+	if (xfs_ag_contains_log(mp, pag->pag_agno))
+		agblocks -= mp->m_sb.sb_logblocks;
+
+	return xfs_btree_calc_size(M_IGEO(mp)->inobt_mnr,
+				(uint64_t)agblocks * mp->m_sb.sb_inopblock /
+					XFS_INODES_PER_CHUNK);
+}
+
+/* Read AGI and create inobt cursor. */
+int
+xfs_inobt_cur(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_perag	*pag,
+	xfs_btnum_t		which,
+	struct xfs_btree_cur	**curpp,
+	struct xfs_buf		**agi_bpp)
+{
+	struct xfs_btree_cur	*cur;
+	int			error;
+
+	ASSERT(*agi_bpp == NULL);
+	ASSERT(*curpp == NULL);
+
+	error = xfs_ialloc_read_agi(pag, tp, agi_bpp);
+	if (error)
+		return error;
+
+	cur = xfs_inobt_init_cursor(mp, tp, *agi_bpp, pag, which);
+	*curpp = cur;
+	return 0;
+}
+
+static int
+xfs_inobt_count_blocks(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_perag	*pag,
+	xfs_btnum_t		btnum,
+	xfs_extlen_t		*tree_blocks)
+{
+	struct xfs_buf		*agbp = NULL;
+	struct xfs_btree_cur	*cur = NULL;
+	int			error;
+
+	error = xfs_inobt_cur(mp, tp, pag, btnum, &cur, &agbp);
+	if (error)
+		return error;
+
+	error = xfs_btree_count_blocks(cur, tree_blocks);
+	xfs_btree_del_cursor(cur, error);
+	xfs_trans_brelse(tp, agbp);
+
+	return error;
+}
+
+/* Read finobt block count from AGI header. */
+static int
+xfs_finobt_read_blocks(
+	struct xfs_perag	*pag,
+	struct xfs_trans	*tp,
+	xfs_extlen_t		*tree_blocks)
+{
+	struct xfs_buf		*agbp;
+	struct xfs_agi		*agi;
+	int			error;
+
+	error = xfs_ialloc_read_agi(pag, tp, &agbp);
+	if (error)
+		return error;
+
+	agi = agbp->b_addr;
+	*tree_blocks = be32_to_cpu(agi->agi_fblocks);
+	xfs_trans_brelse(tp, agbp);
+	return 0;
+}
+
+/*
+ * Figure out how many blocks to reserve and how many are used by this btree.
+ */
+int
+xfs_finobt_calc_reserves(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_perag	*pag,
+	xfs_extlen_t		*ask,
+	xfs_extlen_t		*used)
+{
+	xfs_extlen_t		tree_len = 0;
+	int			error;
+
+	if (!xfs_has_finobt(mp))
+		return 0;
+
+	if (xfs_has_inobtcounts(mp))
+		error = xfs_finobt_read_blocks(pag, tp, &tree_len);
+	else
+		error = xfs_inobt_count_blocks(mp, tp, pag, XFS_BTNUM_FINO,
+				&tree_len);
+	if (error)
+		return error;
+
+	*ask += xfs_inobt_max_size(pag);
+	*used += tree_len;
+	return 0;
+}
+
+/* Calculate the inobt btree size for some records. */
+xfs_extlen_t
+xfs_iallocbt_calc_size(
+	struct xfs_mount	*mp,
+	unsigned long long	len)
+{
+	return xfs_btree_calc_size(M_IGEO(mp)->inobt_mnr, len);
+}
+
+int __init
+xfs_inobt_init_cur_cache(void)
+{
+	xfs_inobt_cur_cache = kmem_cache_create("xfs_inobt_cur",
+			xfs_btree_cur_sizeof(xfs_inobt_maxlevels_ondisk()),
+			0, 0, NULL);
+
+	if (!xfs_inobt_cur_cache)
+		return -ENOMEM;
+	return 0;
+}
+
+void
+xfs_inobt_destroy_cur_cache(void)
+{
+	kmem_cache_destroy(xfs_inobt_cur_cache);
+	xfs_inobt_cur_cache = NULL;
+}
diff --git a/fs/xfs/libxfs/xfs_ialloc_btree.h b/fs/xfs/libxfs/xfs_ialloc_btree.h
new file mode 100644
index 000000000..26451cb76
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_ialloc_btree.h
@@ -0,0 +1,83 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_IALLOC_BTREE_H__
+#define	__XFS_IALLOC_BTREE_H__
+
+/*
+ * Inode map on-disk structures
+ */
+
+struct xfs_buf;
+struct xfs_btree_cur;
+struct xfs_mount;
+struct xfs_perag;
+
+/*
+ * Btree block header size depends on a superblock flag.
+ */
+#define XFS_INOBT_BLOCK_LEN(mp) \
+	(xfs_has_crc(((mp))) ? \
+		XFS_BTREE_SBLOCK_CRC_LEN : XFS_BTREE_SBLOCK_LEN)
+
+/*
+ * Record, key, and pointer address macros for btree blocks.
+ *
+ * (note that some of these may appear unused, but they are used in userspace)
+ */
+#define XFS_INOBT_REC_ADDR(mp, block, index) \
+	((xfs_inobt_rec_t *) \
+		((char *)(block) + \
+		 XFS_INOBT_BLOCK_LEN(mp) + \
+		 (((index) - 1) * sizeof(xfs_inobt_rec_t))))
+
+#define XFS_INOBT_KEY_ADDR(mp, block, index) \
+	((xfs_inobt_key_t *) \
+		((char *)(block) + \
+		 XFS_INOBT_BLOCK_LEN(mp) + \
+		 ((index) - 1) * sizeof(xfs_inobt_key_t)))
+
+#define XFS_INOBT_PTR_ADDR(mp, block, index, maxrecs) \
+	((xfs_inobt_ptr_t *) \
+		((char *)(block) + \
+		 XFS_INOBT_BLOCK_LEN(mp) + \
+		 (maxrecs) * sizeof(xfs_inobt_key_t) + \
+		 ((index) - 1) * sizeof(xfs_inobt_ptr_t)))
+
+extern struct xfs_btree_cur *xfs_inobt_init_cursor(struct xfs_mount *mp,
+		struct xfs_trans *tp, struct xfs_buf *agbp,
+		struct xfs_perag *pag, xfs_btnum_t btnum);
+struct xfs_btree_cur *xfs_inobt_stage_cursor(struct xfs_mount *mp,
+		struct xbtree_afakeroot *afake, struct xfs_perag *pag,
+		xfs_btnum_t btnum);
+extern int xfs_inobt_maxrecs(struct xfs_mount *, int, int);
+
+/* ir_holemask to inode allocation bitmap conversion */
+uint64_t xfs_inobt_irec_to_allocmask(struct xfs_inobt_rec_incore *);
+
+#if defined(DEBUG) || defined(XFS_WARN)
+int xfs_inobt_rec_check_count(struct xfs_mount *,
+			      struct xfs_inobt_rec_incore *);
+#else
+#define xfs_inobt_rec_check_count(mp, rec)	0
+#endif	/* DEBUG */
+
+int xfs_finobt_calc_reserves(struct xfs_mount *mp, struct xfs_trans *tp,
+		struct xfs_perag *pag, xfs_extlen_t *ask, xfs_extlen_t *used);
+extern xfs_extlen_t xfs_iallocbt_calc_size(struct xfs_mount *mp,
+		unsigned long long len);
+int xfs_inobt_cur(struct xfs_mount *mp, struct xfs_trans *tp,
+		struct xfs_perag *pag, xfs_btnum_t btnum,
+		struct xfs_btree_cur **curpp, struct xfs_buf **agi_bpp);
+
+void xfs_inobt_commit_staged_btree(struct xfs_btree_cur *cur,
+		struct xfs_trans *tp, struct xfs_buf *agbp);
+
+unsigned int xfs_iallocbt_maxlevels_ondisk(void);
+
+int __init xfs_inobt_init_cur_cache(void);
+void xfs_inobt_destroy_cur_cache(void);
+
+#endif	/* __XFS_IALLOC_BTREE_H__ */
diff --git a/fs/xfs/libxfs/xfs_iext_tree.c b/fs/xfs/libxfs/xfs_iext_tree.c
new file mode 100644
index 000000000..773cf4349
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_iext_tree.c
@@ -0,0 +1,1050 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2017 Christoph Hellwig.
+ */
+
+#include "xfs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_bit.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_inode.h"
+#include "xfs_trace.h"
+
+/*
+ * In-core extent record layout:
+ *
+ * +-------+----------------------------+
+ * | 00:53 | all 54 bits of startoff    |
+ * | 54:63 | low 10 bits of startblock  |
+ * +-------+----------------------------+
+ * | 00:20 | all 21 bits of length      |
+ * |    21 | unwritten extent bit       |
+ * | 22:63 | high 42 bits of startblock |
+ * +-------+----------------------------+
+ */
+#define XFS_IEXT_STARTOFF_MASK		xfs_mask64lo(BMBT_STARTOFF_BITLEN)
+#define XFS_IEXT_LENGTH_MASK		xfs_mask64lo(BMBT_BLOCKCOUNT_BITLEN)
+#define XFS_IEXT_STARTBLOCK_MASK	xfs_mask64lo(BMBT_STARTBLOCK_BITLEN)
+
+struct xfs_iext_rec {
+	uint64_t			lo;
+	uint64_t			hi;
+};
+
+/*
+ * Given that the length can't be a zero, only an empty hi value indicates an
+ * unused record.
+ */
+static bool xfs_iext_rec_is_empty(struct xfs_iext_rec *rec)
+{
+	return rec->hi == 0;
+}
+
+static inline void xfs_iext_rec_clear(struct xfs_iext_rec *rec)
+{
+	rec->lo = 0;
+	rec->hi = 0;
+}
+
+static void
+xfs_iext_set(
+	struct xfs_iext_rec	*rec,
+	struct xfs_bmbt_irec	*irec)
+{
+	ASSERT((irec->br_startoff & ~XFS_IEXT_STARTOFF_MASK) == 0);
+	ASSERT((irec->br_blockcount & ~XFS_IEXT_LENGTH_MASK) == 0);
+	ASSERT((irec->br_startblock & ~XFS_IEXT_STARTBLOCK_MASK) == 0);
+
+	rec->lo = irec->br_startoff & XFS_IEXT_STARTOFF_MASK;
+	rec->hi = irec->br_blockcount & XFS_IEXT_LENGTH_MASK;
+
+	rec->lo |= (irec->br_startblock << 54);
+	rec->hi |= ((irec->br_startblock & ~xfs_mask64lo(10)) << (22 - 10));
+
+	if (irec->br_state == XFS_EXT_UNWRITTEN)
+		rec->hi |= (1 << 21);
+}
+
+static void
+xfs_iext_get(
+	struct xfs_bmbt_irec	*irec,
+	struct xfs_iext_rec	*rec)
+{
+	irec->br_startoff = rec->lo & XFS_IEXT_STARTOFF_MASK;
+	irec->br_blockcount = rec->hi & XFS_IEXT_LENGTH_MASK;
+
+	irec->br_startblock = rec->lo >> 54;
+	irec->br_startblock |= (rec->hi & xfs_mask64hi(42)) >> (22 - 10);
+
+	if (rec->hi & (1 << 21))
+		irec->br_state = XFS_EXT_UNWRITTEN;
+	else
+		irec->br_state = XFS_EXT_NORM;
+}
+
+enum {
+	NODE_SIZE	= 256,
+	KEYS_PER_NODE	= NODE_SIZE / (sizeof(uint64_t) + sizeof(void *)),
+	RECS_PER_LEAF	= (NODE_SIZE - (2 * sizeof(struct xfs_iext_leaf *))) /
+				sizeof(struct xfs_iext_rec),
+};
+
+/*
+ * In-core extent btree block layout:
+ *
+ * There are two types of blocks in the btree: leaf and inner (non-leaf) blocks.
+ *
+ * The leaf blocks are made up by %KEYS_PER_NODE extent records, which each
+ * contain the startoffset, blockcount, startblock and unwritten extent flag.
+ * See above for the exact format, followed by pointers to the previous and next
+ * leaf blocks (if there are any).
+ *
+ * The inner (non-leaf) blocks first contain KEYS_PER_NODE lookup keys, followed
+ * by an equal number of pointers to the btree blocks at the next lower level.
+ *
+ *		+-------+-------+-------+-------+-------+----------+----------+
+ * Leaf:	| rec 1 | rec 2 | rec 3 | rec 4 | rec N | prev-ptr | next-ptr |
+ *		+-------+-------+-------+-------+-------+----------+----------+
+ *
+ *		+-------+-------+-------+-------+-------+-------+------+-------+
+ * Inner:	| key 1 | key 2 | key 3 | key N | ptr 1 | ptr 2 | ptr3 | ptr N |
+ *		+-------+-------+-------+-------+-------+-------+------+-------+
+ */
+struct xfs_iext_node {
+	uint64_t		keys[KEYS_PER_NODE];
+#define XFS_IEXT_KEY_INVALID	(1ULL << 63)
+	void			*ptrs[KEYS_PER_NODE];
+};
+
+struct xfs_iext_leaf {
+	struct xfs_iext_rec	recs[RECS_PER_LEAF];
+	struct xfs_iext_leaf	*prev;
+	struct xfs_iext_leaf	*next;
+};
+
+inline xfs_extnum_t xfs_iext_count(struct xfs_ifork *ifp)
+{
+	return ifp->if_bytes / sizeof(struct xfs_iext_rec);
+}
+
+static inline int xfs_iext_max_recs(struct xfs_ifork *ifp)
+{
+	if (ifp->if_height == 1)
+		return xfs_iext_count(ifp);
+	return RECS_PER_LEAF;
+}
+
+static inline struct xfs_iext_rec *cur_rec(struct xfs_iext_cursor *cur)
+{
+	return &cur->leaf->recs[cur->pos];
+}
+
+static inline bool xfs_iext_valid(struct xfs_ifork *ifp,
+		struct xfs_iext_cursor *cur)
+{
+	if (!cur->leaf)
+		return false;
+	if (cur->pos < 0 || cur->pos >= xfs_iext_max_recs(ifp))
+		return false;
+	if (xfs_iext_rec_is_empty(cur_rec(cur)))
+		return false;
+	return true;
+}
+
+static void *
+xfs_iext_find_first_leaf(
+	struct xfs_ifork	*ifp)
+{
+	struct xfs_iext_node	*node = ifp->if_u1.if_root;
+	int			height;
+
+	if (!ifp->if_height)
+		return NULL;
+
+	for (height = ifp->if_height; height > 1; height--) {
+		node = node->ptrs[0];
+		ASSERT(node);
+	}
+
+	return node;
+}
+
+static void *
+xfs_iext_find_last_leaf(
+	struct xfs_ifork	*ifp)
+{
+	struct xfs_iext_node	*node = ifp->if_u1.if_root;
+	int			height, i;
+
+	if (!ifp->if_height)
+		return NULL;
+
+	for (height = ifp->if_height; height > 1; height--) {
+		for (i = 1; i < KEYS_PER_NODE; i++)
+			if (!node->ptrs[i])
+				break;
+		node = node->ptrs[i - 1];
+		ASSERT(node);
+	}
+
+	return node;
+}
+
+void
+xfs_iext_first(
+	struct xfs_ifork	*ifp,
+	struct xfs_iext_cursor	*cur)
+{
+	cur->pos = 0;
+	cur->leaf = xfs_iext_find_first_leaf(ifp);
+}
+
+void
+xfs_iext_last(
+	struct xfs_ifork	*ifp,
+	struct xfs_iext_cursor	*cur)
+{
+	int			i;
+
+	cur->leaf = xfs_iext_find_last_leaf(ifp);
+	if (!cur->leaf) {
+		cur->pos = 0;
+		return;
+	}
+
+	for (i = 1; i < xfs_iext_max_recs(ifp); i++) {
+		if (xfs_iext_rec_is_empty(&cur->leaf->recs[i]))
+			break;
+	}
+	cur->pos = i - 1;
+}
+
+void
+xfs_iext_next(
+	struct xfs_ifork	*ifp,
+	struct xfs_iext_cursor	*cur)
+{
+	if (!cur->leaf) {
+		ASSERT(cur->pos <= 0 || cur->pos >= RECS_PER_LEAF);
+		xfs_iext_first(ifp, cur);
+		return;
+	}
+
+	ASSERT(cur->pos >= 0);
+	ASSERT(cur->pos < xfs_iext_max_recs(ifp));
+
+	cur->pos++;
+	if (ifp->if_height > 1 && !xfs_iext_valid(ifp, cur) &&
+	    cur->leaf->next) {
+		cur->leaf = cur->leaf->next;
+		cur->pos = 0;
+	}
+}
+
+void
+xfs_iext_prev(
+	struct xfs_ifork	*ifp,
+	struct xfs_iext_cursor	*cur)
+{
+	if (!cur->leaf) {
+		ASSERT(cur->pos <= 0 || cur->pos >= RECS_PER_LEAF);
+		xfs_iext_last(ifp, cur);
+		return;
+	}
+
+	ASSERT(cur->pos >= 0);
+	ASSERT(cur->pos <= RECS_PER_LEAF);
+
+recurse:
+	do {
+		cur->pos--;
+		if (xfs_iext_valid(ifp, cur))
+			return;
+	} while (cur->pos > 0);
+
+	if (ifp->if_height > 1 && cur->leaf->prev) {
+		cur->leaf = cur->leaf->prev;
+		cur->pos = RECS_PER_LEAF;
+		goto recurse;
+	}
+}
+
+static inline int
+xfs_iext_key_cmp(
+	struct xfs_iext_node	*node,
+	int			n,
+	xfs_fileoff_t		offset)
+{
+	if (node->keys[n] > offset)
+		return 1;
+	if (node->keys[n] < offset)
+		return -1;
+	return 0;
+}
+
+static inline int
+xfs_iext_rec_cmp(
+	struct xfs_iext_rec	*rec,
+	xfs_fileoff_t		offset)
+{
+	uint64_t		rec_offset = rec->lo & XFS_IEXT_STARTOFF_MASK;
+	uint32_t		rec_len = rec->hi & XFS_IEXT_LENGTH_MASK;
+
+	if (rec_offset > offset)
+		return 1;
+	if (rec_offset + rec_len <= offset)
+		return -1;
+	return 0;
+}
+
+static void *
+xfs_iext_find_level(
+	struct xfs_ifork	*ifp,
+	xfs_fileoff_t		offset,
+	int			level)
+{
+	struct xfs_iext_node	*node = ifp->if_u1.if_root;
+	int			height, i;
+
+	if (!ifp->if_height)
+		return NULL;
+
+	for (height = ifp->if_height; height > level; height--) {
+		for (i = 1; i < KEYS_PER_NODE; i++)
+			if (xfs_iext_key_cmp(node, i, offset) > 0)
+				break;
+
+		node = node->ptrs[i - 1];
+		if (!node)
+			break;
+	}
+
+	return node;
+}
+
+static int
+xfs_iext_node_pos(
+	struct xfs_iext_node	*node,
+	xfs_fileoff_t		offset)
+{
+	int			i;
+
+	for (i = 1; i < KEYS_PER_NODE; i++) {
+		if (xfs_iext_key_cmp(node, i, offset) > 0)
+			break;
+	}
+
+	return i - 1;
+}
+
+static int
+xfs_iext_node_insert_pos(
+	struct xfs_iext_node	*node,
+	xfs_fileoff_t		offset)
+{
+	int			i;
+
+	for (i = 0; i < KEYS_PER_NODE; i++) {
+		if (xfs_iext_key_cmp(node, i, offset) > 0)
+			return i;
+	}
+
+	return KEYS_PER_NODE;
+}
+
+static int
+xfs_iext_node_nr_entries(
+	struct xfs_iext_node	*node,
+	int			start)
+{
+	int			i;
+
+	for (i = start; i < KEYS_PER_NODE; i++) {
+		if (node->keys[i] == XFS_IEXT_KEY_INVALID)
+			break;
+	}
+
+	return i;
+}
+
+static int
+xfs_iext_leaf_nr_entries(
+	struct xfs_ifork	*ifp,
+	struct xfs_iext_leaf	*leaf,
+	int			start)
+{
+	int			i;
+
+	for (i = start; i < xfs_iext_max_recs(ifp); i++) {
+		if (xfs_iext_rec_is_empty(&leaf->recs[i]))
+			break;
+	}
+
+	return i;
+}
+
+static inline uint64_t
+xfs_iext_leaf_key(
+	struct xfs_iext_leaf	*leaf,
+	int			n)
+{
+	return leaf->recs[n].lo & XFS_IEXT_STARTOFF_MASK;
+}
+
+static void
+xfs_iext_grow(
+	struct xfs_ifork	*ifp)
+{
+	struct xfs_iext_node	*node = kmem_zalloc(NODE_SIZE, KM_NOFS);
+	int			i;
+
+	if (ifp->if_height == 1) {
+		struct xfs_iext_leaf *prev = ifp->if_u1.if_root;
+
+		node->keys[0] = xfs_iext_leaf_key(prev, 0);
+		node->ptrs[0] = prev;
+	} else  {
+		struct xfs_iext_node *prev = ifp->if_u1.if_root;
+
+		ASSERT(ifp->if_height > 1);
+
+		node->keys[0] = prev->keys[0];
+		node->ptrs[0] = prev;
+	}
+
+	for (i = 1; i < KEYS_PER_NODE; i++)
+		node->keys[i] = XFS_IEXT_KEY_INVALID;
+
+	ifp->if_u1.if_root = node;
+	ifp->if_height++;
+}
+
+static void
+xfs_iext_update_node(
+	struct xfs_ifork	*ifp,
+	xfs_fileoff_t		old_offset,
+	xfs_fileoff_t		new_offset,
+	int			level,
+	void			*ptr)
+{
+	struct xfs_iext_node	*node = ifp->if_u1.if_root;
+	int			height, i;
+
+	for (height = ifp->if_height; height > level; height--) {
+		for (i = 0; i < KEYS_PER_NODE; i++) {
+			if (i > 0 && xfs_iext_key_cmp(node, i, old_offset) > 0)
+				break;
+			if (node->keys[i] == old_offset)
+				node->keys[i] = new_offset;
+		}
+		node = node->ptrs[i - 1];
+		ASSERT(node);
+	}
+
+	ASSERT(node == ptr);
+}
+
+static struct xfs_iext_node *
+xfs_iext_split_node(
+	struct xfs_iext_node	**nodep,
+	int			*pos,
+	int			*nr_entries)
+{
+	struct xfs_iext_node	*node = *nodep;
+	struct xfs_iext_node	*new = kmem_zalloc(NODE_SIZE, KM_NOFS);
+	const int		nr_move = KEYS_PER_NODE / 2;
+	int			nr_keep = nr_move + (KEYS_PER_NODE & 1);
+	int			i = 0;
+
+	/* for sequential append operations just spill over into the new node */
+	if (*pos == KEYS_PER_NODE) {
+		*nodep = new;
+		*pos = 0;
+		*nr_entries = 0;
+		goto done;
+	}
+
+
+	for (i = 0; i < nr_move; i++) {
+		new->keys[i] = node->keys[nr_keep + i];
+		new->ptrs[i] = node->ptrs[nr_keep + i];
+
+		node->keys[nr_keep + i] = XFS_IEXT_KEY_INVALID;
+		node->ptrs[nr_keep + i] = NULL;
+	}
+
+	if (*pos >= nr_keep) {
+		*nodep = new;
+		*pos -= nr_keep;
+		*nr_entries = nr_move;
+	} else {
+		*nr_entries = nr_keep;
+	}
+done:
+	for (; i < KEYS_PER_NODE; i++)
+		new->keys[i] = XFS_IEXT_KEY_INVALID;
+	return new;
+}
+
+static void
+xfs_iext_insert_node(
+	struct xfs_ifork	*ifp,
+	uint64_t		offset,
+	void			*ptr,
+	int			level)
+{
+	struct xfs_iext_node	*node, *new;
+	int			i, pos, nr_entries;
+
+again:
+	if (ifp->if_height < level)
+		xfs_iext_grow(ifp);
+
+	new = NULL;
+	node = xfs_iext_find_level(ifp, offset, level);
+	pos = xfs_iext_node_insert_pos(node, offset);
+	nr_entries = xfs_iext_node_nr_entries(node, pos);
+
+	ASSERT(pos >= nr_entries || xfs_iext_key_cmp(node, pos, offset) != 0);
+	ASSERT(nr_entries <= KEYS_PER_NODE);
+
+	if (nr_entries == KEYS_PER_NODE)
+		new = xfs_iext_split_node(&node, &pos, &nr_entries);
+
+	/*
+	 * Update the pointers in higher levels if the first entry changes
+	 * in an existing node.
+	 */
+	if (node != new && pos == 0 && nr_entries > 0)
+		xfs_iext_update_node(ifp, node->keys[0], offset, level, node);
+
+	for (i = nr_entries; i > pos; i--) {
+		node->keys[i] = node->keys[i - 1];
+		node->ptrs[i] = node->ptrs[i - 1];
+	}
+	node->keys[pos] = offset;
+	node->ptrs[pos] = ptr;
+
+	if (new) {
+		offset = new->keys[0];
+		ptr = new;
+		level++;
+		goto again;
+	}
+}
+
+static struct xfs_iext_leaf *
+xfs_iext_split_leaf(
+	struct xfs_iext_cursor	*cur,
+	int			*nr_entries)
+{
+	struct xfs_iext_leaf	*leaf = cur->leaf;
+	struct xfs_iext_leaf	*new = kmem_zalloc(NODE_SIZE, KM_NOFS);
+	const int		nr_move = RECS_PER_LEAF / 2;
+	int			nr_keep = nr_move + (RECS_PER_LEAF & 1);
+	int			i;
+
+	/* for sequential append operations just spill over into the new node */
+	if (cur->pos == RECS_PER_LEAF) {
+		cur->leaf = new;
+		cur->pos = 0;
+		*nr_entries = 0;
+		goto done;
+	}
+
+	for (i = 0; i < nr_move; i++) {
+		new->recs[i] = leaf->recs[nr_keep + i];
+		xfs_iext_rec_clear(&leaf->recs[nr_keep + i]);
+	}
+
+	if (cur->pos >= nr_keep) {
+		cur->leaf = new;
+		cur->pos -= nr_keep;
+		*nr_entries = nr_move;
+	} else {
+		*nr_entries = nr_keep;
+	}
+done:
+	if (leaf->next)
+		leaf->next->prev = new;
+	new->next = leaf->next;
+	new->prev = leaf;
+	leaf->next = new;
+	return new;
+}
+
+static void
+xfs_iext_alloc_root(
+	struct xfs_ifork	*ifp,
+	struct xfs_iext_cursor	*cur)
+{
+	ASSERT(ifp->if_bytes == 0);
+
+	ifp->if_u1.if_root = kmem_zalloc(sizeof(struct xfs_iext_rec), KM_NOFS);
+	ifp->if_height = 1;
+
+	/* now that we have a node step into it */
+	cur->leaf = ifp->if_u1.if_root;
+	cur->pos = 0;
+}
+
+static void
+xfs_iext_realloc_root(
+	struct xfs_ifork	*ifp,
+	struct xfs_iext_cursor	*cur)
+{
+	int64_t new_size = ifp->if_bytes + sizeof(struct xfs_iext_rec);
+	void *new;
+
+	/* account for the prev/next pointers */
+	if (new_size / sizeof(struct xfs_iext_rec) == RECS_PER_LEAF)
+		new_size = NODE_SIZE;
+
+	new = krealloc(ifp->if_u1.if_root, new_size, GFP_NOFS | __GFP_NOFAIL);
+	memset(new + ifp->if_bytes, 0, new_size - ifp->if_bytes);
+	ifp->if_u1.if_root = new;
+	cur->leaf = new;
+}
+
+/*
+ * Increment the sequence counter on extent tree changes. If we are on a COW
+ * fork, this allows the writeback code to skip looking for a COW extent if the
+ * COW fork hasn't changed. We use WRITE_ONCE here to ensure the update to the
+ * sequence counter is seen before the modifications to the extent tree itself
+ * take effect.
+ */
+static inline void xfs_iext_inc_seq(struct xfs_ifork *ifp)
+{
+	WRITE_ONCE(ifp->if_seq, READ_ONCE(ifp->if_seq) + 1);
+}
+
+void
+xfs_iext_insert(
+	struct xfs_inode	*ip,
+	struct xfs_iext_cursor	*cur,
+	struct xfs_bmbt_irec	*irec,
+	int			state)
+{
+	struct xfs_ifork	*ifp = xfs_iext_state_to_fork(ip, state);
+	xfs_fileoff_t		offset = irec->br_startoff;
+	struct xfs_iext_leaf	*new = NULL;
+	int			nr_entries, i;
+
+	xfs_iext_inc_seq(ifp);
+
+	if (ifp->if_height == 0)
+		xfs_iext_alloc_root(ifp, cur);
+	else if (ifp->if_height == 1)
+		xfs_iext_realloc_root(ifp, cur);
+
+	nr_entries = xfs_iext_leaf_nr_entries(ifp, cur->leaf, cur->pos);
+	ASSERT(nr_entries <= RECS_PER_LEAF);
+	ASSERT(cur->pos >= nr_entries ||
+	       xfs_iext_rec_cmp(cur_rec(cur), irec->br_startoff) != 0);
+
+	if (nr_entries == RECS_PER_LEAF)
+		new = xfs_iext_split_leaf(cur, &nr_entries);
+
+	/*
+	 * Update the pointers in higher levels if the first entry changes
+	 * in an existing node.
+	 */
+	if (cur->leaf != new && cur->pos == 0 && nr_entries > 0) {
+		xfs_iext_update_node(ifp, xfs_iext_leaf_key(cur->leaf, 0),
+				offset, 1, cur->leaf);
+	}
+
+	for (i = nr_entries; i > cur->pos; i--)
+		cur->leaf->recs[i] = cur->leaf->recs[i - 1];
+	xfs_iext_set(cur_rec(cur), irec);
+	ifp->if_bytes += sizeof(struct xfs_iext_rec);
+
+	trace_xfs_iext_insert(ip, cur, state, _RET_IP_);
+
+	if (new)
+		xfs_iext_insert_node(ifp, xfs_iext_leaf_key(new, 0), new, 2);
+}
+
+static struct xfs_iext_node *
+xfs_iext_rebalance_node(
+	struct xfs_iext_node	*parent,
+	int			*pos,
+	struct xfs_iext_node	*node,
+	int			nr_entries)
+{
+	/*
+	 * If the neighbouring nodes are completely full, or have different
+	 * parents, we might never be able to merge our node, and will only
+	 * delete it once the number of entries hits zero.
+	 */
+	if (nr_entries == 0)
+		return node;
+
+	if (*pos > 0) {
+		struct xfs_iext_node *prev = parent->ptrs[*pos - 1];
+		int nr_prev = xfs_iext_node_nr_entries(prev, 0), i;
+
+		if (nr_prev + nr_entries <= KEYS_PER_NODE) {
+			for (i = 0; i < nr_entries; i++) {
+				prev->keys[nr_prev + i] = node->keys[i];
+				prev->ptrs[nr_prev + i] = node->ptrs[i];
+			}
+			return node;
+		}
+	}
+
+	if (*pos + 1 < xfs_iext_node_nr_entries(parent, *pos)) {
+		struct xfs_iext_node *next = parent->ptrs[*pos + 1];
+		int nr_next = xfs_iext_node_nr_entries(next, 0), i;
+
+		if (nr_entries + nr_next <= KEYS_PER_NODE) {
+			/*
+			 * Merge the next node into this node so that we don't
+			 * have to do an additional update of the keys in the
+			 * higher levels.
+			 */
+			for (i = 0; i < nr_next; i++) {
+				node->keys[nr_entries + i] = next->keys[i];
+				node->ptrs[nr_entries + i] = next->ptrs[i];
+			}
+
+			++*pos;
+			return next;
+		}
+	}
+
+	return NULL;
+}
+
+static void
+xfs_iext_remove_node(
+	struct xfs_ifork	*ifp,
+	xfs_fileoff_t		offset,
+	void			*victim)
+{
+	struct xfs_iext_node	*node, *parent;
+	int			level = 2, pos, nr_entries, i;
+
+	ASSERT(level <= ifp->if_height);
+	node = xfs_iext_find_level(ifp, offset, level);
+	pos = xfs_iext_node_pos(node, offset);
+again:
+	ASSERT(node->ptrs[pos]);
+	ASSERT(node->ptrs[pos] == victim);
+	kmem_free(victim);
+
+	nr_entries = xfs_iext_node_nr_entries(node, pos) - 1;
+	offset = node->keys[0];
+	for (i = pos; i < nr_entries; i++) {
+		node->keys[i] = node->keys[i + 1];
+		node->ptrs[i] = node->ptrs[i + 1];
+	}
+	node->keys[nr_entries] = XFS_IEXT_KEY_INVALID;
+	node->ptrs[nr_entries] = NULL;
+
+	if (pos == 0 && nr_entries > 0) {
+		xfs_iext_update_node(ifp, offset, node->keys[0], level, node);
+		offset = node->keys[0];
+	}
+
+	if (nr_entries >= KEYS_PER_NODE / 2)
+		return;
+
+	if (level < ifp->if_height) {
+		/*
+		 * If we aren't at the root yet try to find a neighbour node to
+		 * merge with (or delete the node if it is empty), and then
+		 * recurse up to the next level.
+		 */
+		level++;
+		parent = xfs_iext_find_level(ifp, offset, level);
+		pos = xfs_iext_node_pos(parent, offset);
+
+		ASSERT(pos != KEYS_PER_NODE);
+		ASSERT(parent->ptrs[pos] == node);
+
+		node = xfs_iext_rebalance_node(parent, &pos, node, nr_entries);
+		if (node) {
+			victim = node;
+			node = parent;
+			goto again;
+		}
+	} else if (nr_entries == 1) {
+		/*
+		 * If we are at the root and only one entry is left we can just
+		 * free this node and update the root pointer.
+		 */
+		ASSERT(node == ifp->if_u1.if_root);
+		ifp->if_u1.if_root = node->ptrs[0];
+		ifp->if_height--;
+		kmem_free(node);
+	}
+}
+
+static void
+xfs_iext_rebalance_leaf(
+	struct xfs_ifork	*ifp,
+	struct xfs_iext_cursor	*cur,
+	struct xfs_iext_leaf	*leaf,
+	xfs_fileoff_t		offset,
+	int			nr_entries)
+{
+	/*
+	 * If the neighbouring nodes are completely full we might never be able
+	 * to merge our node, and will only delete it once the number of
+	 * entries hits zero.
+	 */
+	if (nr_entries == 0)
+		goto remove_node;
+
+	if (leaf->prev) {
+		int nr_prev = xfs_iext_leaf_nr_entries(ifp, leaf->prev, 0), i;
+
+		if (nr_prev + nr_entries <= RECS_PER_LEAF) {
+			for (i = 0; i < nr_entries; i++)
+				leaf->prev->recs[nr_prev + i] = leaf->recs[i];
+
+			if (cur->leaf == leaf) {
+				cur->leaf = leaf->prev;
+				cur->pos += nr_prev;
+			}
+			goto remove_node;
+		}
+	}
+
+	if (leaf->next) {
+		int nr_next = xfs_iext_leaf_nr_entries(ifp, leaf->next, 0), i;
+
+		if (nr_entries + nr_next <= RECS_PER_LEAF) {
+			/*
+			 * Merge the next node into this node so that we don't
+			 * have to do an additional update of the keys in the
+			 * higher levels.
+			 */
+			for (i = 0; i < nr_next; i++) {
+				leaf->recs[nr_entries + i] =
+					leaf->next->recs[i];
+			}
+
+			if (cur->leaf == leaf->next) {
+				cur->leaf = leaf;
+				cur->pos += nr_entries;
+			}
+
+			offset = xfs_iext_leaf_key(leaf->next, 0);
+			leaf = leaf->next;
+			goto remove_node;
+		}
+	}
+
+	return;
+remove_node:
+	if (leaf->prev)
+		leaf->prev->next = leaf->next;
+	if (leaf->next)
+		leaf->next->prev = leaf->prev;
+	xfs_iext_remove_node(ifp, offset, leaf);
+}
+
+static void
+xfs_iext_free_last_leaf(
+	struct xfs_ifork	*ifp)
+{
+	ifp->if_height--;
+	kmem_free(ifp->if_u1.if_root);
+	ifp->if_u1.if_root = NULL;
+}
+
+void
+xfs_iext_remove(
+	struct xfs_inode	*ip,
+	struct xfs_iext_cursor	*cur,
+	int			state)
+{
+	struct xfs_ifork	*ifp = xfs_iext_state_to_fork(ip, state);
+	struct xfs_iext_leaf	*leaf = cur->leaf;
+	xfs_fileoff_t		offset = xfs_iext_leaf_key(leaf, 0);
+	int			i, nr_entries;
+
+	trace_xfs_iext_remove(ip, cur, state, _RET_IP_);
+
+	ASSERT(ifp->if_height > 0);
+	ASSERT(ifp->if_u1.if_root != NULL);
+	ASSERT(xfs_iext_valid(ifp, cur));
+
+	xfs_iext_inc_seq(ifp);
+
+	nr_entries = xfs_iext_leaf_nr_entries(ifp, leaf, cur->pos) - 1;
+	for (i = cur->pos; i < nr_entries; i++)
+		leaf->recs[i] = leaf->recs[i + 1];
+	xfs_iext_rec_clear(&leaf->recs[nr_entries]);
+	ifp->if_bytes -= sizeof(struct xfs_iext_rec);
+
+	if (cur->pos == 0 && nr_entries > 0) {
+		xfs_iext_update_node(ifp, offset, xfs_iext_leaf_key(leaf, 0), 1,
+				leaf);
+		offset = xfs_iext_leaf_key(leaf, 0);
+	} else if (cur->pos == nr_entries) {
+		if (ifp->if_height > 1 && leaf->next)
+			cur->leaf = leaf->next;
+		else
+			cur->leaf = NULL;
+		cur->pos = 0;
+	}
+
+	if (nr_entries >= RECS_PER_LEAF / 2)
+		return;
+
+	if (ifp->if_height > 1)
+		xfs_iext_rebalance_leaf(ifp, cur, leaf, offset, nr_entries);
+	else if (nr_entries == 0)
+		xfs_iext_free_last_leaf(ifp);
+}
+
+/*
+ * Lookup the extent covering bno.
+ *
+ * If there is an extent covering bno return the extent index, and store the
+ * expanded extent structure in *gotp, and the extent cursor in *cur.
+ * If there is no extent covering bno, but there is an extent after it (e.g.
+ * it lies in a hole) return that extent in *gotp and its cursor in *cur
+ * instead.
+ * If bno is beyond the last extent return false, and return an invalid
+ * cursor value.
+ */
+bool
+xfs_iext_lookup_extent(
+	struct xfs_inode	*ip,
+	struct xfs_ifork	*ifp,
+	xfs_fileoff_t		offset,
+	struct xfs_iext_cursor	*cur,
+	struct xfs_bmbt_irec	*gotp)
+{
+	XFS_STATS_INC(ip->i_mount, xs_look_exlist);
+
+	cur->leaf = xfs_iext_find_level(ifp, offset, 1);
+	if (!cur->leaf) {
+		cur->pos = 0;
+		return false;
+	}
+
+	for (cur->pos = 0; cur->pos < xfs_iext_max_recs(ifp); cur->pos++) {
+		struct xfs_iext_rec *rec = cur_rec(cur);
+
+		if (xfs_iext_rec_is_empty(rec))
+			break;
+		if (xfs_iext_rec_cmp(rec, offset) >= 0)
+			goto found;
+	}
+
+	/* Try looking in the next node for an entry > offset */
+	if (ifp->if_height == 1 || !cur->leaf->next)
+		return false;
+	cur->leaf = cur->leaf->next;
+	cur->pos = 0;
+	if (!xfs_iext_valid(ifp, cur))
+		return false;
+found:
+	xfs_iext_get(gotp, cur_rec(cur));
+	return true;
+}
+
+/*
+ * Returns the last extent before end, and if this extent doesn't cover
+ * end, update end to the end of the extent.
+ */
+bool
+xfs_iext_lookup_extent_before(
+	struct xfs_inode	*ip,
+	struct xfs_ifork	*ifp,
+	xfs_fileoff_t		*end,
+	struct xfs_iext_cursor	*cur,
+	struct xfs_bmbt_irec	*gotp)
+{
+	/* could be optimized to not even look up the next on a match.. */
+	if (xfs_iext_lookup_extent(ip, ifp, *end - 1, cur, gotp) &&
+	    gotp->br_startoff <= *end - 1)
+		return true;
+	if (!xfs_iext_prev_extent(ifp, cur, gotp))
+		return false;
+	*end = gotp->br_startoff + gotp->br_blockcount;
+	return true;
+}
+
+void
+xfs_iext_update_extent(
+	struct xfs_inode	*ip,
+	int			state,
+	struct xfs_iext_cursor	*cur,
+	struct xfs_bmbt_irec	*new)
+{
+	struct xfs_ifork	*ifp = xfs_iext_state_to_fork(ip, state);
+
+	xfs_iext_inc_seq(ifp);
+
+	if (cur->pos == 0) {
+		struct xfs_bmbt_irec	old;
+
+		xfs_iext_get(&old, cur_rec(cur));
+		if (new->br_startoff != old.br_startoff) {
+			xfs_iext_update_node(ifp, old.br_startoff,
+					new->br_startoff, 1, cur->leaf);
+		}
+	}
+
+	trace_xfs_bmap_pre_update(ip, cur, state, _RET_IP_);
+	xfs_iext_set(cur_rec(cur), new);
+	trace_xfs_bmap_post_update(ip, cur, state, _RET_IP_);
+}
+
+/*
+ * Return true if the cursor points at an extent and return the extent structure
+ * in gotp.  Else return false.
+ */
+bool
+xfs_iext_get_extent(
+	struct xfs_ifork	*ifp,
+	struct xfs_iext_cursor	*cur,
+	struct xfs_bmbt_irec	*gotp)
+{
+	if (!xfs_iext_valid(ifp, cur))
+		return false;
+	xfs_iext_get(gotp, cur_rec(cur));
+	return true;
+}
+
+/*
+ * This is a recursive function, because of that we need to be extremely
+ * careful with stack usage.
+ */
+static void
+xfs_iext_destroy_node(
+	struct xfs_iext_node	*node,
+	int			level)
+{
+	int			i;
+
+	if (level > 1) {
+		for (i = 0; i < KEYS_PER_NODE; i++) {
+			if (node->keys[i] == XFS_IEXT_KEY_INVALID)
+				break;
+			xfs_iext_destroy_node(node->ptrs[i], level - 1);
+		}
+	}
+
+	kmem_free(node);
+}
+
+void
+xfs_iext_destroy(
+	struct xfs_ifork	*ifp)
+{
+	xfs_iext_destroy_node(ifp->if_u1.if_root, ifp->if_height);
+
+	ifp->if_bytes = 0;
+	ifp->if_height = 0;
+	ifp->if_u1.if_root = NULL;
+}
diff --git a/fs/xfs/libxfs/xfs_inode_buf.c b/fs/xfs/libxfs/xfs_inode_buf.c
new file mode 100644
index 000000000..758aacd81
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_inode_buf.c
@@ -0,0 +1,773 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_ag.h"
+#include "xfs_inode.h"
+#include "xfs_errortag.h"
+#include "xfs_error.h"
+#include "xfs_icache.h"
+#include "xfs_trans.h"
+#include "xfs_ialloc.h"
+#include "xfs_dir2.h"
+
+#include <linux/iversion.h>
+
+/*
+ * If we are doing readahead on an inode buffer, we might be in log recovery
+ * reading an inode allocation buffer that hasn't yet been replayed, and hence
+ * has not had the inode cores stamped into it. Hence for readahead, the buffer
+ * may be potentially invalid.
+ *
+ * If the readahead buffer is invalid, we need to mark it with an error and
+ * clear the DONE status of the buffer so that a followup read will re-read it
+ * from disk. We don't report the error otherwise to avoid warnings during log
+ * recovery and we don't get unnecessary panics on debug kernels. We use EIO here
+ * because all we want to do is say readahead failed; there is no-one to report
+ * the error to, so this will distinguish it from a non-ra verifier failure.
+ * Changes to this readahead error behaviour also need to be reflected in
+ * xfs_dquot_buf_readahead_verify().
+ */
+static void
+xfs_inode_buf_verify(
+	struct xfs_buf	*bp,
+	bool		readahead)
+{
+	struct xfs_mount *mp = bp->b_mount;
+	int		i;
+	int		ni;
+
+	/*
+	 * Validate the magic number and version of every inode in the buffer
+	 */
+	ni = XFS_BB_TO_FSB(mp, bp->b_length) * mp->m_sb.sb_inopblock;
+	for (i = 0; i < ni; i++) {
+		struct xfs_dinode	*dip;
+		xfs_agino_t		unlinked_ino;
+		int			di_ok;
+
+		dip = xfs_buf_offset(bp, (i << mp->m_sb.sb_inodelog));
+		unlinked_ino = be32_to_cpu(dip->di_next_unlinked);
+		di_ok = xfs_verify_magic16(bp, dip->di_magic) &&
+			xfs_dinode_good_version(mp, dip->di_version) &&
+			xfs_verify_agino_or_null(bp->b_pag, unlinked_ino);
+		if (unlikely(XFS_TEST_ERROR(!di_ok, mp,
+						XFS_ERRTAG_ITOBP_INOTOBP))) {
+			if (readahead) {
+				bp->b_flags &= ~XBF_DONE;
+				xfs_buf_ioerror(bp, -EIO);
+				return;
+			}
+
+#ifdef DEBUG
+			xfs_alert(mp,
+				"bad inode magic/vsn daddr %lld #%d (magic=%x)",
+				(unsigned long long)xfs_buf_daddr(bp), i,
+				be16_to_cpu(dip->di_magic));
+#endif
+			xfs_buf_verifier_error(bp, -EFSCORRUPTED,
+					__func__, dip, sizeof(*dip),
+					NULL);
+			return;
+		}
+	}
+}
+
+
+static void
+xfs_inode_buf_read_verify(
+	struct xfs_buf	*bp)
+{
+	xfs_inode_buf_verify(bp, false);
+}
+
+static void
+xfs_inode_buf_readahead_verify(
+	struct xfs_buf	*bp)
+{
+	xfs_inode_buf_verify(bp, true);
+}
+
+static void
+xfs_inode_buf_write_verify(
+	struct xfs_buf	*bp)
+{
+	xfs_inode_buf_verify(bp, false);
+}
+
+const struct xfs_buf_ops xfs_inode_buf_ops = {
+	.name = "xfs_inode",
+	.magic16 = { cpu_to_be16(XFS_DINODE_MAGIC),
+		     cpu_to_be16(XFS_DINODE_MAGIC) },
+	.verify_read = xfs_inode_buf_read_verify,
+	.verify_write = xfs_inode_buf_write_verify,
+};
+
+const struct xfs_buf_ops xfs_inode_buf_ra_ops = {
+	.name = "xfs_inode_ra",
+	.magic16 = { cpu_to_be16(XFS_DINODE_MAGIC),
+		     cpu_to_be16(XFS_DINODE_MAGIC) },
+	.verify_read = xfs_inode_buf_readahead_verify,
+	.verify_write = xfs_inode_buf_write_verify,
+};
+
+
+/*
+ * This routine is called to map an inode to the buffer containing the on-disk
+ * version of the inode.  It returns a pointer to the buffer containing the
+ * on-disk inode in the bpp parameter.
+ */
+int
+xfs_imap_to_bp(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_imap		*imap,
+	struct xfs_buf		**bpp)
+{
+	return xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, imap->im_blkno,
+				   imap->im_len, XBF_UNMAPPED, bpp,
+				   &xfs_inode_buf_ops);
+}
+
+static inline struct timespec64 xfs_inode_decode_bigtime(uint64_t ts)
+{
+	struct timespec64	tv;
+	uint32_t		n;
+
+	tv.tv_sec = xfs_bigtime_to_unix(div_u64_rem(ts, NSEC_PER_SEC, &n));
+	tv.tv_nsec = n;
+
+	return tv;
+}
+
+/* Convert an ondisk timestamp to an incore timestamp. */
+struct timespec64
+xfs_inode_from_disk_ts(
+	struct xfs_dinode		*dip,
+	const xfs_timestamp_t		ts)
+{
+	struct timespec64		tv;
+	struct xfs_legacy_timestamp	*lts;
+
+	if (xfs_dinode_has_bigtime(dip))
+		return xfs_inode_decode_bigtime(be64_to_cpu(ts));
+
+	lts = (struct xfs_legacy_timestamp *)&ts;
+	tv.tv_sec = (int)be32_to_cpu(lts->t_sec);
+	tv.tv_nsec = (int)be32_to_cpu(lts->t_nsec);
+
+	return tv;
+}
+
+int
+xfs_inode_from_disk(
+	struct xfs_inode	*ip,
+	struct xfs_dinode	*from)
+{
+	struct inode		*inode = VFS_I(ip);
+	int			error;
+	xfs_failaddr_t		fa;
+
+	ASSERT(ip->i_cowfp == NULL);
+
+	fa = xfs_dinode_verify(ip->i_mount, ip->i_ino, from);
+	if (fa) {
+		xfs_inode_verifier_error(ip, -EFSCORRUPTED, "dinode", from,
+				sizeof(*from), fa);
+		return -EFSCORRUPTED;
+	}
+
+	/*
+	 * First get the permanent information that is needed to allocate an
+	 * inode. If the inode is unused, mode is zero and we shouldn't mess
+	 * with the uninitialized part of it.
+	 */
+	if (!xfs_has_v3inodes(ip->i_mount))
+		ip->i_flushiter = be16_to_cpu(from->di_flushiter);
+	inode->i_generation = be32_to_cpu(from->di_gen);
+	inode->i_mode = be16_to_cpu(from->di_mode);
+	if (!inode->i_mode)
+		return 0;
+
+	/*
+	 * Convert v1 inodes immediately to v2 inode format as this is the
+	 * minimum inode version format we support in the rest of the code.
+	 * They will also be unconditionally written back to disk as v2 inodes.
+	 */
+	if (unlikely(from->di_version == 1)) {
+		set_nlink(inode, be16_to_cpu(from->di_onlink));
+		ip->i_projid = 0;
+	} else {
+		set_nlink(inode, be32_to_cpu(from->di_nlink));
+		ip->i_projid = (prid_t)be16_to_cpu(from->di_projid_hi) << 16 |
+					be16_to_cpu(from->di_projid_lo);
+	}
+
+	i_uid_write(inode, be32_to_cpu(from->di_uid));
+	i_gid_write(inode, be32_to_cpu(from->di_gid));
+
+	/*
+	 * Time is signed, so need to convert to signed 32 bit before
+	 * storing in inode timestamp which may be 64 bit. Otherwise
+	 * a time before epoch is converted to a time long after epoch
+	 * on 64 bit systems.
+	 */
+	inode->i_atime = xfs_inode_from_disk_ts(from, from->di_atime);
+	inode->i_mtime = xfs_inode_from_disk_ts(from, from->di_mtime);
+	inode->i_ctime = xfs_inode_from_disk_ts(from, from->di_ctime);
+
+	ip->i_disk_size = be64_to_cpu(from->di_size);
+	ip->i_nblocks = be64_to_cpu(from->di_nblocks);
+	ip->i_extsize = be32_to_cpu(from->di_extsize);
+	ip->i_forkoff = from->di_forkoff;
+	ip->i_diflags = be16_to_cpu(from->di_flags);
+	ip->i_next_unlinked = be32_to_cpu(from->di_next_unlinked);
+
+	if (from->di_dmevmask || from->di_dmstate)
+		xfs_iflags_set(ip, XFS_IPRESERVE_DM_FIELDS);
+
+	if (xfs_has_v3inodes(ip->i_mount)) {
+		inode_set_iversion_queried(inode,
+					   be64_to_cpu(from->di_changecount));
+		ip->i_crtime = xfs_inode_from_disk_ts(from, from->di_crtime);
+		ip->i_diflags2 = be64_to_cpu(from->di_flags2);
+		ip->i_cowextsize = be32_to_cpu(from->di_cowextsize);
+	}
+
+	error = xfs_iformat_data_fork(ip, from);
+	if (error)
+		return error;
+	if (from->di_forkoff) {
+		error = xfs_iformat_attr_fork(ip, from);
+		if (error)
+			goto out_destroy_data_fork;
+	}
+	if (xfs_is_reflink_inode(ip))
+		xfs_ifork_init_cow(ip);
+	return 0;
+
+out_destroy_data_fork:
+	xfs_idestroy_fork(&ip->i_df);
+	return error;
+}
+
+/* Convert an incore timestamp to an ondisk timestamp. */
+static inline xfs_timestamp_t
+xfs_inode_to_disk_ts(
+	struct xfs_inode		*ip,
+	const struct timespec64		tv)
+{
+	struct xfs_legacy_timestamp	*lts;
+	xfs_timestamp_t			ts;
+
+	if (xfs_inode_has_bigtime(ip))
+		return cpu_to_be64(xfs_inode_encode_bigtime(tv));
+
+	lts = (struct xfs_legacy_timestamp *)&ts;
+	lts->t_sec = cpu_to_be32(tv.tv_sec);
+	lts->t_nsec = cpu_to_be32(tv.tv_nsec);
+
+	return ts;
+}
+
+static inline void
+xfs_inode_to_disk_iext_counters(
+	struct xfs_inode	*ip,
+	struct xfs_dinode	*to)
+{
+	if (xfs_inode_has_large_extent_counts(ip)) {
+		to->di_big_nextents = cpu_to_be64(xfs_ifork_nextents(&ip->i_df));
+		to->di_big_anextents = cpu_to_be32(xfs_ifork_nextents(&ip->i_af));
+		/*
+		 * We might be upgrading the inode to use larger extent counters
+		 * than was previously used. Hence zero the unused field.
+		 */
+		to->di_nrext64_pad = cpu_to_be16(0);
+	} else {
+		to->di_nextents = cpu_to_be32(xfs_ifork_nextents(&ip->i_df));
+		to->di_anextents = cpu_to_be16(xfs_ifork_nextents(&ip->i_af));
+	}
+}
+
+void
+xfs_inode_to_disk(
+	struct xfs_inode	*ip,
+	struct xfs_dinode	*to,
+	xfs_lsn_t		lsn)
+{
+	struct inode		*inode = VFS_I(ip);
+
+	to->di_magic = cpu_to_be16(XFS_DINODE_MAGIC);
+	to->di_onlink = 0;
+
+	to->di_format = xfs_ifork_format(&ip->i_df);
+	to->di_uid = cpu_to_be32(i_uid_read(inode));
+	to->di_gid = cpu_to_be32(i_gid_read(inode));
+	to->di_projid_lo = cpu_to_be16(ip->i_projid & 0xffff);
+	to->di_projid_hi = cpu_to_be16(ip->i_projid >> 16);
+
+	to->di_atime = xfs_inode_to_disk_ts(ip, inode->i_atime);
+	to->di_mtime = xfs_inode_to_disk_ts(ip, inode->i_mtime);
+	to->di_ctime = xfs_inode_to_disk_ts(ip, inode->i_ctime);
+	to->di_nlink = cpu_to_be32(inode->i_nlink);
+	to->di_gen = cpu_to_be32(inode->i_generation);
+	to->di_mode = cpu_to_be16(inode->i_mode);
+
+	to->di_size = cpu_to_be64(ip->i_disk_size);
+	to->di_nblocks = cpu_to_be64(ip->i_nblocks);
+	to->di_extsize = cpu_to_be32(ip->i_extsize);
+	to->di_forkoff = ip->i_forkoff;
+	to->di_aformat = xfs_ifork_format(&ip->i_af);
+	to->di_flags = cpu_to_be16(ip->i_diflags);
+
+	if (xfs_has_v3inodes(ip->i_mount)) {
+		to->di_version = 3;
+		to->di_changecount = cpu_to_be64(inode_peek_iversion(inode));
+		to->di_crtime = xfs_inode_to_disk_ts(ip, ip->i_crtime);
+		to->di_flags2 = cpu_to_be64(ip->i_diflags2);
+		to->di_cowextsize = cpu_to_be32(ip->i_cowextsize);
+		to->di_ino = cpu_to_be64(ip->i_ino);
+		to->di_lsn = cpu_to_be64(lsn);
+		memset(to->di_pad2, 0, sizeof(to->di_pad2));
+		uuid_copy(&to->di_uuid, &ip->i_mount->m_sb.sb_meta_uuid);
+		to->di_v3_pad = 0;
+	} else {
+		to->di_version = 2;
+		to->di_flushiter = cpu_to_be16(ip->i_flushiter);
+		memset(to->di_v2_pad, 0, sizeof(to->di_v2_pad));
+	}
+
+	xfs_inode_to_disk_iext_counters(ip, to);
+}
+
+static xfs_failaddr_t
+xfs_dinode_verify_fork(
+	struct xfs_dinode	*dip,
+	struct xfs_mount	*mp,
+	int			whichfork)
+{
+	xfs_extnum_t		di_nextents;
+	xfs_extnum_t		max_extents;
+	mode_t			mode = be16_to_cpu(dip->di_mode);
+	uint32_t		fork_size = XFS_DFORK_SIZE(dip, mp, whichfork);
+	uint32_t		fork_format = XFS_DFORK_FORMAT(dip, whichfork);
+
+	di_nextents = xfs_dfork_nextents(dip, whichfork);
+
+	/*
+	 * For fork types that can contain local data, check that the fork
+	 * format matches the size of local data contained within the fork.
+	 *
+	 * For all types, check that when the size says the should be in extent
+	 * or btree format, the inode isn't claiming it is in local format.
+	 */
+	if (whichfork == XFS_DATA_FORK) {
+		if (S_ISDIR(mode) || S_ISLNK(mode)) {
+			if (be64_to_cpu(dip->di_size) <= fork_size &&
+			    fork_format != XFS_DINODE_FMT_LOCAL)
+				return __this_address;
+		}
+
+		if (be64_to_cpu(dip->di_size) > fork_size &&
+		    fork_format == XFS_DINODE_FMT_LOCAL)
+			return __this_address;
+	}
+
+	switch (fork_format) {
+	case XFS_DINODE_FMT_LOCAL:
+		/*
+		 * No local regular files yet.
+		 */
+		if (S_ISREG(mode) && whichfork == XFS_DATA_FORK)
+			return __this_address;
+		if (di_nextents)
+			return __this_address;
+		break;
+	case XFS_DINODE_FMT_EXTENTS:
+		if (di_nextents > XFS_DFORK_MAXEXT(dip, mp, whichfork))
+			return __this_address;
+		break;
+	case XFS_DINODE_FMT_BTREE:
+		max_extents = xfs_iext_max_nextents(
+					xfs_dinode_has_large_extent_counts(dip),
+					whichfork);
+		if (di_nextents > max_extents)
+			return __this_address;
+		break;
+	default:
+		return __this_address;
+	}
+	return NULL;
+}
+
+static xfs_failaddr_t
+xfs_dinode_verify_forkoff(
+	struct xfs_dinode	*dip,
+	struct xfs_mount	*mp)
+{
+	if (!dip->di_forkoff)
+		return NULL;
+
+	switch (dip->di_format)  {
+	case XFS_DINODE_FMT_DEV:
+		if (dip->di_forkoff != (roundup(sizeof(xfs_dev_t), 8) >> 3))
+			return __this_address;
+		break;
+	case XFS_DINODE_FMT_LOCAL:	/* fall through ... */
+	case XFS_DINODE_FMT_EXTENTS:    /* fall through ... */
+	case XFS_DINODE_FMT_BTREE:
+		if (dip->di_forkoff >= (XFS_LITINO(mp) >> 3))
+			return __this_address;
+		break;
+	default:
+		return __this_address;
+	}
+	return NULL;
+}
+
+static xfs_failaddr_t
+xfs_dinode_verify_nrext64(
+	struct xfs_mount	*mp,
+	struct xfs_dinode	*dip)
+{
+	if (xfs_dinode_has_large_extent_counts(dip)) {
+		if (!xfs_has_large_extent_counts(mp))
+			return __this_address;
+		if (dip->di_nrext64_pad != 0)
+			return __this_address;
+	} else if (dip->di_version >= 3) {
+		if (dip->di_v3_pad != 0)
+			return __this_address;
+	}
+
+	return NULL;
+}
+
+xfs_failaddr_t
+xfs_dinode_verify(
+	struct xfs_mount	*mp,
+	xfs_ino_t		ino,
+	struct xfs_dinode	*dip)
+{
+	xfs_failaddr_t		fa;
+	uint16_t		mode;
+	uint16_t		flags;
+	uint64_t		flags2;
+	uint64_t		di_size;
+	xfs_extnum_t		nextents;
+	xfs_extnum_t		naextents;
+	xfs_filblks_t		nblocks;
+
+	if (dip->di_magic != cpu_to_be16(XFS_DINODE_MAGIC))
+		return __this_address;
+
+	/* Verify v3 integrity information first */
+	if (dip->di_version >= 3) {
+		if (!xfs_has_v3inodes(mp))
+			return __this_address;
+		if (!xfs_verify_cksum((char *)dip, mp->m_sb.sb_inodesize,
+				      XFS_DINODE_CRC_OFF))
+			return __this_address;
+		if (be64_to_cpu(dip->di_ino) != ino)
+			return __this_address;
+		if (!uuid_equal(&dip->di_uuid, &mp->m_sb.sb_meta_uuid))
+			return __this_address;
+	}
+
+	/* don't allow invalid i_size */
+	di_size = be64_to_cpu(dip->di_size);
+	if (di_size & (1ULL << 63))
+		return __this_address;
+
+	mode = be16_to_cpu(dip->di_mode);
+	if (mode && xfs_mode_to_ftype(mode) == XFS_DIR3_FT_UNKNOWN)
+		return __this_address;
+
+	/* No zero-length symlinks/dirs. */
+	if ((S_ISLNK(mode) || S_ISDIR(mode)) && di_size == 0)
+		return __this_address;
+
+	fa = xfs_dinode_verify_nrext64(mp, dip);
+	if (fa)
+		return fa;
+
+	nextents = xfs_dfork_data_extents(dip);
+	naextents = xfs_dfork_attr_extents(dip);
+	nblocks = be64_to_cpu(dip->di_nblocks);
+
+	/* Fork checks carried over from xfs_iformat_fork */
+	if (mode && nextents + naextents > nblocks)
+		return __this_address;
+
+	if (S_ISDIR(mode) && nextents > mp->m_dir_geo->max_extents)
+		return __this_address;
+
+	if (mode && XFS_DFORK_BOFF(dip) > mp->m_sb.sb_inodesize)
+		return __this_address;
+
+	flags = be16_to_cpu(dip->di_flags);
+
+	if (mode && (flags & XFS_DIFLAG_REALTIME) && !mp->m_rtdev_targp)
+		return __this_address;
+
+	/* check for illegal values of forkoff */
+	fa = xfs_dinode_verify_forkoff(dip, mp);
+	if (fa)
+		return fa;
+
+	/* Do we have appropriate data fork formats for the mode? */
+	switch (mode & S_IFMT) {
+	case S_IFIFO:
+	case S_IFCHR:
+	case S_IFBLK:
+	case S_IFSOCK:
+		if (dip->di_format != XFS_DINODE_FMT_DEV)
+			return __this_address;
+		break;
+	case S_IFREG:
+	case S_IFLNK:
+	case S_IFDIR:
+		fa = xfs_dinode_verify_fork(dip, mp, XFS_DATA_FORK);
+		if (fa)
+			return fa;
+		break;
+	case 0:
+		/* Uninitialized inode ok. */
+		break;
+	default:
+		return __this_address;
+	}
+
+	if (dip->di_forkoff) {
+		fa = xfs_dinode_verify_fork(dip, mp, XFS_ATTR_FORK);
+		if (fa)
+			return fa;
+	} else {
+		/*
+		 * If there is no fork offset, this may be a freshly-made inode
+		 * in a new disk cluster, in which case di_aformat is zeroed.
+		 * Otherwise, such an inode must be in EXTENTS format; this goes
+		 * for freed inodes as well.
+		 */
+		switch (dip->di_aformat) {
+		case 0:
+		case XFS_DINODE_FMT_EXTENTS:
+			break;
+		default:
+			return __this_address;
+		}
+		if (naextents)
+			return __this_address;
+	}
+
+	/* extent size hint validation */
+	fa = xfs_inode_validate_extsize(mp, be32_to_cpu(dip->di_extsize),
+			mode, flags);
+	if (fa)
+		return fa;
+
+	/* only version 3 or greater inodes are extensively verified here */
+	if (dip->di_version < 3)
+		return NULL;
+
+	flags2 = be64_to_cpu(dip->di_flags2);
+
+	/* don't allow reflink/cowextsize if we don't have reflink */
+	if ((flags2 & (XFS_DIFLAG2_REFLINK | XFS_DIFLAG2_COWEXTSIZE)) &&
+	     !xfs_has_reflink(mp))
+		return __this_address;
+
+	/* only regular files get reflink */
+	if ((flags2 & XFS_DIFLAG2_REFLINK) && (mode & S_IFMT) != S_IFREG)
+		return __this_address;
+
+	/* don't let reflink and realtime mix */
+	if ((flags2 & XFS_DIFLAG2_REFLINK) && (flags & XFS_DIFLAG_REALTIME))
+		return __this_address;
+
+	/* COW extent size hint validation */
+	fa = xfs_inode_validate_cowextsize(mp, be32_to_cpu(dip->di_cowextsize),
+			mode, flags, flags2);
+	if (fa)
+		return fa;
+
+	/* bigtime iflag can only happen on bigtime filesystems */
+	if (xfs_dinode_has_bigtime(dip) &&
+	    !xfs_has_bigtime(mp))
+		return __this_address;
+
+	return NULL;
+}
+
+void
+xfs_dinode_calc_crc(
+	struct xfs_mount	*mp,
+	struct xfs_dinode	*dip)
+{
+	uint32_t		crc;
+
+	if (dip->di_version < 3)
+		return;
+
+	ASSERT(xfs_has_crc(mp));
+	crc = xfs_start_cksum_update((char *)dip, mp->m_sb.sb_inodesize,
+			      XFS_DINODE_CRC_OFF);
+	dip->di_crc = xfs_end_cksum(crc);
+}
+
+/*
+ * Validate di_extsize hint.
+ *
+ * 1. Extent size hint is only valid for directories and regular files.
+ * 2. FS_XFLAG_EXTSIZE is only valid for regular files.
+ * 3. FS_XFLAG_EXTSZINHERIT is only valid for directories.
+ * 4. Hint cannot be larger than MAXTEXTLEN.
+ * 5. Can be changed on directories at any time.
+ * 6. Hint value of 0 turns off hints, clears inode flags.
+ * 7. Extent size must be a multiple of the appropriate block size.
+ *    For realtime files, this is the rt extent size.
+ * 8. For non-realtime files, the extent size hint must be limited
+ *    to half the AG size to avoid alignment extending the extent beyond the
+ *    limits of the AG.
+ */
+xfs_failaddr_t
+xfs_inode_validate_extsize(
+	struct xfs_mount		*mp,
+	uint32_t			extsize,
+	uint16_t			mode,
+	uint16_t			flags)
+{
+	bool				rt_flag;
+	bool				hint_flag;
+	bool				inherit_flag;
+	uint32_t			extsize_bytes;
+	uint32_t			blocksize_bytes;
+
+	rt_flag = (flags & XFS_DIFLAG_REALTIME);
+	hint_flag = (flags & XFS_DIFLAG_EXTSIZE);
+	inherit_flag = (flags & XFS_DIFLAG_EXTSZINHERIT);
+	extsize_bytes = XFS_FSB_TO_B(mp, extsize);
+
+	/*
+	 * This comment describes a historic gap in this verifier function.
+	 *
+	 * For a directory with both RTINHERIT and EXTSZINHERIT flags set, this
+	 * function has never checked that the extent size hint is an integer
+	 * multiple of the realtime extent size.  Since we allow users to set
+	 * this combination  on non-rt filesystems /and/ to change the rt
+	 * extent size when adding a rt device to a filesystem, the net effect
+	 * is that users can configure a filesystem anticipating one rt
+	 * geometry and change their minds later.  Directories do not use the
+	 * extent size hint, so this is harmless for them.
+	 *
+	 * If a directory with a misaligned extent size hint is allowed to
+	 * propagate that hint into a new regular realtime file, the result
+	 * is that the inode cluster buffer verifier will trigger a corruption
+	 * shutdown the next time it is run, because the verifier has always
+	 * enforced the alignment rule for regular files.
+	 *
+	 * Because we allow administrators to set a new rt extent size when
+	 * adding a rt section, we cannot add a check to this verifier because
+	 * that will result a new source of directory corruption errors when
+	 * reading an existing filesystem.  Instead, we rely on callers to
+	 * decide when alignment checks are appropriate, and fix things up as
+	 * needed.
+	 */
+
+	if (rt_flag)
+		blocksize_bytes = XFS_FSB_TO_B(mp, mp->m_sb.sb_rextsize);
+	else
+		blocksize_bytes = mp->m_sb.sb_blocksize;
+
+	if ((hint_flag || inherit_flag) && !(S_ISDIR(mode) || S_ISREG(mode)))
+		return __this_address;
+
+	if (hint_flag && !S_ISREG(mode))
+		return __this_address;
+
+	if (inherit_flag && !S_ISDIR(mode))
+		return __this_address;
+
+	if ((hint_flag || inherit_flag) && extsize == 0)
+		return __this_address;
+
+	/* free inodes get flags set to zero but extsize remains */
+	if (mode && !(hint_flag || inherit_flag) && extsize != 0)
+		return __this_address;
+
+	if (extsize_bytes % blocksize_bytes)
+		return __this_address;
+
+	if (extsize > XFS_MAX_BMBT_EXTLEN)
+		return __this_address;
+
+	if (!rt_flag && extsize > mp->m_sb.sb_agblocks / 2)
+		return __this_address;
+
+	return NULL;
+}
+
+/*
+ * Validate di_cowextsize hint.
+ *
+ * 1. CoW extent size hint can only be set if reflink is enabled on the fs.
+ *    The inode does not have to have any shared blocks, but it must be a v3.
+ * 2. FS_XFLAG_COWEXTSIZE is only valid for directories and regular files;
+ *    for a directory, the hint is propagated to new files.
+ * 3. Can be changed on files & directories at any time.
+ * 4. Hint value of 0 turns off hints, clears inode flags.
+ * 5. Extent size must be a multiple of the appropriate block size.
+ * 6. The extent size hint must be limited to half the AG size to avoid
+ *    alignment extending the extent beyond the limits of the AG.
+ */
+xfs_failaddr_t
+xfs_inode_validate_cowextsize(
+	struct xfs_mount		*mp,
+	uint32_t			cowextsize,
+	uint16_t			mode,
+	uint16_t			flags,
+	uint64_t			flags2)
+{
+	bool				rt_flag;
+	bool				hint_flag;
+	uint32_t			cowextsize_bytes;
+
+	rt_flag = (flags & XFS_DIFLAG_REALTIME);
+	hint_flag = (flags2 & XFS_DIFLAG2_COWEXTSIZE);
+	cowextsize_bytes = XFS_FSB_TO_B(mp, cowextsize);
+
+	if (hint_flag && !xfs_has_reflink(mp))
+		return __this_address;
+
+	if (hint_flag && !(S_ISDIR(mode) || S_ISREG(mode)))
+		return __this_address;
+
+	if (hint_flag && cowextsize == 0)
+		return __this_address;
+
+	/* free inodes get flags set to zero but cowextsize remains */
+	if (mode && !hint_flag && cowextsize != 0)
+		return __this_address;
+
+	if (hint_flag && rt_flag)
+		return __this_address;
+
+	if (cowextsize_bytes % mp->m_sb.sb_blocksize)
+		return __this_address;
+
+	if (cowextsize > XFS_MAX_BMBT_EXTLEN)
+		return __this_address;
+
+	if (cowextsize > mp->m_sb.sb_agblocks / 2)
+		return __this_address;
+
+	return NULL;
+}
diff --git a/fs/xfs/libxfs/xfs_inode_buf.h b/fs/xfs/libxfs/xfs_inode_buf.h
new file mode 100644
index 000000000..585ed5a11
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_inode_buf.h
@@ -0,0 +1,54 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef	__XFS_INODE_BUF_H__
+#define	__XFS_INODE_BUF_H__
+
+struct xfs_inode;
+struct xfs_dinode;
+
+/*
+ * Inode location information.  Stored in the inode and passed to
+ * xfs_imap_to_bp() to get a buffer and dinode for a given inode.
+ */
+struct xfs_imap {
+	xfs_daddr_t	im_blkno;	/* starting BB of inode chunk */
+	unsigned short	im_len;		/* length in BBs of inode chunk */
+	unsigned short	im_boffset;	/* inode offset in block in bytes */
+};
+
+int	xfs_imap_to_bp(struct xfs_mount *mp, struct xfs_trans *tp,
+		       struct xfs_imap *imap, struct xfs_buf **bpp);
+void	xfs_dinode_calc_crc(struct xfs_mount *mp, struct xfs_dinode *dip);
+void	xfs_inode_to_disk(struct xfs_inode *ip, struct xfs_dinode *to,
+			  xfs_lsn_t lsn);
+int	xfs_inode_from_disk(struct xfs_inode *ip, struct xfs_dinode *from);
+
+xfs_failaddr_t xfs_dinode_verify(struct xfs_mount *mp, xfs_ino_t ino,
+			   struct xfs_dinode *dip);
+xfs_failaddr_t xfs_inode_validate_extsize(struct xfs_mount *mp,
+		uint32_t extsize, uint16_t mode, uint16_t flags);
+xfs_failaddr_t xfs_inode_validate_cowextsize(struct xfs_mount *mp,
+		uint32_t cowextsize, uint16_t mode, uint16_t flags,
+		uint64_t flags2);
+
+static inline uint64_t xfs_inode_encode_bigtime(struct timespec64 tv)
+{
+	return xfs_unix_to_bigtime(tv.tv_sec) * NSEC_PER_SEC + tv.tv_nsec;
+}
+
+struct timespec64 xfs_inode_from_disk_ts(struct xfs_dinode *dip,
+		const xfs_timestamp_t ts);
+
+static inline bool
+xfs_dinode_good_version(struct xfs_mount *mp, uint8_t version)
+{
+	if (xfs_has_v3inodes(mp))
+		return version == 3;
+	return version == 1 || version == 2;
+}
+
+
+#endif	/* __XFS_INODE_BUF_H__ */
diff --git a/fs/xfs/libxfs/xfs_inode_fork.c b/fs/xfs/libxfs/xfs_inode_fork.c
new file mode 100644
index 000000000..6b2176018
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_inode_fork.c
@@ -0,0 +1,779 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_inode.h"
+#include "xfs_trans.h"
+#include "xfs_inode_item.h"
+#include "xfs_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_bmap.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+#include "xfs_da_format.h"
+#include "xfs_da_btree.h"
+#include "xfs_dir2_priv.h"
+#include "xfs_attr_leaf.h"
+#include "xfs_types.h"
+#include "xfs_errortag.h"
+
+struct kmem_cache *xfs_ifork_cache;
+
+void
+xfs_init_local_fork(
+	struct xfs_inode	*ip,
+	int			whichfork,
+	const void		*data,
+	int64_t			size)
+{
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	int			mem_size = size;
+	bool			zero_terminate;
+
+	/*
+	 * If we are using the local fork to store a symlink body we need to
+	 * zero-terminate it so that we can pass it back to the VFS directly.
+	 * Overallocate the in-memory fork by one for that and add a zero
+	 * to terminate it below.
+	 */
+	zero_terminate = S_ISLNK(VFS_I(ip)->i_mode);
+	if (zero_terminate)
+		mem_size++;
+
+	if (size) {
+		ifp->if_u1.if_data = kmem_alloc(mem_size, KM_NOFS);
+		memcpy(ifp->if_u1.if_data, data, size);
+		if (zero_terminate)
+			ifp->if_u1.if_data[size] = '\0';
+	} else {
+		ifp->if_u1.if_data = NULL;
+	}
+
+	ifp->if_bytes = size;
+}
+
+/*
+ * The file is in-lined in the on-disk inode.
+ */
+STATIC int
+xfs_iformat_local(
+	struct xfs_inode	*ip,
+	struct xfs_dinode	*dip,
+	int			whichfork,
+	int			size)
+{
+	/*
+	 * If the size is unreasonable, then something
+	 * is wrong and we just bail out rather than crash in
+	 * kmem_alloc() or memcpy() below.
+	 */
+	if (unlikely(size > XFS_DFORK_SIZE(dip, ip->i_mount, whichfork))) {
+		xfs_warn(ip->i_mount,
+	"corrupt inode %llu (bad size %d for local fork, size = %zd).",
+			(unsigned long long) ip->i_ino, size,
+			XFS_DFORK_SIZE(dip, ip->i_mount, whichfork));
+		xfs_inode_verifier_error(ip, -EFSCORRUPTED,
+				"xfs_iformat_local", dip, sizeof(*dip),
+				__this_address);
+		return -EFSCORRUPTED;
+	}
+
+	xfs_init_local_fork(ip, whichfork, XFS_DFORK_PTR(dip, whichfork), size);
+	return 0;
+}
+
+/*
+ * The file consists of a set of extents all of which fit into the on-disk
+ * inode.
+ */
+STATIC int
+xfs_iformat_extents(
+	struct xfs_inode	*ip,
+	struct xfs_dinode	*dip,
+	int			whichfork)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	int			state = xfs_bmap_fork_to_state(whichfork);
+	xfs_extnum_t		nex = xfs_dfork_nextents(dip, whichfork);
+	int			size = nex * sizeof(xfs_bmbt_rec_t);
+	struct xfs_iext_cursor	icur;
+	struct xfs_bmbt_rec	*dp;
+	struct xfs_bmbt_irec	new;
+	int			i;
+
+	/*
+	 * If the number of extents is unreasonable, then something is wrong and
+	 * we just bail out rather than crash in kmem_alloc() or memcpy() below.
+	 */
+	if (unlikely(size < 0 || size > XFS_DFORK_SIZE(dip, mp, whichfork))) {
+		xfs_warn(ip->i_mount, "corrupt inode %llu ((a)extents = %llu).",
+			ip->i_ino, nex);
+		xfs_inode_verifier_error(ip, -EFSCORRUPTED,
+				"xfs_iformat_extents(1)", dip, sizeof(*dip),
+				__this_address);
+		return -EFSCORRUPTED;
+	}
+
+	ifp->if_bytes = 0;
+	ifp->if_u1.if_root = NULL;
+	ifp->if_height = 0;
+	if (size) {
+		dp = (xfs_bmbt_rec_t *) XFS_DFORK_PTR(dip, whichfork);
+
+		xfs_iext_first(ifp, &icur);
+		for (i = 0; i < nex; i++, dp++) {
+			xfs_failaddr_t	fa;
+
+			xfs_bmbt_disk_get_all(dp, &new);
+			fa = xfs_bmap_validate_extent(ip, whichfork, &new);
+			if (fa) {
+				xfs_inode_verifier_error(ip, -EFSCORRUPTED,
+						"xfs_iformat_extents(2)",
+						dp, sizeof(*dp), fa);
+				return -EFSCORRUPTED;
+			}
+
+			xfs_iext_insert(ip, &icur, &new, state);
+			trace_xfs_read_extent(ip, &icur, state, _THIS_IP_);
+			xfs_iext_next(ifp, &icur);
+		}
+	}
+	return 0;
+}
+
+/*
+ * The file has too many extents to fit into
+ * the inode, so they are in B-tree format.
+ * Allocate a buffer for the root of the B-tree
+ * and copy the root into it.  The i_extents
+ * field will remain NULL until all of the
+ * extents are read in (when they are needed).
+ */
+STATIC int
+xfs_iformat_btree(
+	struct xfs_inode	*ip,
+	struct xfs_dinode	*dip,
+	int			whichfork)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	xfs_bmdr_block_t	*dfp;
+	struct xfs_ifork	*ifp;
+	/* REFERENCED */
+	int			nrecs;
+	int			size;
+	int			level;
+
+	ifp = xfs_ifork_ptr(ip, whichfork);
+	dfp = (xfs_bmdr_block_t *)XFS_DFORK_PTR(dip, whichfork);
+	size = XFS_BMAP_BROOT_SPACE(mp, dfp);
+	nrecs = be16_to_cpu(dfp->bb_numrecs);
+	level = be16_to_cpu(dfp->bb_level);
+
+	/*
+	 * blow out if -- fork has less extents than can fit in
+	 * fork (fork shouldn't be a btree format), root btree
+	 * block has more records than can fit into the fork,
+	 * or the number of extents is greater than the number of
+	 * blocks.
+	 */
+	if (unlikely(ifp->if_nextents <= XFS_IFORK_MAXEXT(ip, whichfork) ||
+		     nrecs == 0 ||
+		     XFS_BMDR_SPACE_CALC(nrecs) >
+					XFS_DFORK_SIZE(dip, mp, whichfork) ||
+		     ifp->if_nextents > ip->i_nblocks) ||
+		     level == 0 || level > XFS_BM_MAXLEVELS(mp, whichfork)) {
+		xfs_warn(mp, "corrupt inode %llu (btree).",
+					(unsigned long long) ip->i_ino);
+		xfs_inode_verifier_error(ip, -EFSCORRUPTED,
+				"xfs_iformat_btree", dfp, size,
+				__this_address);
+		return -EFSCORRUPTED;
+	}
+
+	ifp->if_broot_bytes = size;
+	ifp->if_broot = kmem_alloc(size, KM_NOFS);
+	ASSERT(ifp->if_broot != NULL);
+	/*
+	 * Copy and convert from the on-disk structure
+	 * to the in-memory structure.
+	 */
+	xfs_bmdr_to_bmbt(ip, dfp, XFS_DFORK_SIZE(dip, ip->i_mount, whichfork),
+			 ifp->if_broot, size);
+
+	ifp->if_bytes = 0;
+	ifp->if_u1.if_root = NULL;
+	ifp->if_height = 0;
+	return 0;
+}
+
+int
+xfs_iformat_data_fork(
+	struct xfs_inode	*ip,
+	struct xfs_dinode	*dip)
+{
+	struct inode		*inode = VFS_I(ip);
+	int			error;
+
+	/*
+	 * Initialize the extent count early, as the per-format routines may
+	 * depend on it.
+	 */
+	ip->i_df.if_format = dip->di_format;
+	ip->i_df.if_nextents = xfs_dfork_data_extents(dip);
+
+	switch (inode->i_mode & S_IFMT) {
+	case S_IFIFO:
+	case S_IFCHR:
+	case S_IFBLK:
+	case S_IFSOCK:
+		ip->i_disk_size = 0;
+		inode->i_rdev = xfs_to_linux_dev_t(xfs_dinode_get_rdev(dip));
+		return 0;
+	case S_IFREG:
+	case S_IFLNK:
+	case S_IFDIR:
+		switch (ip->i_df.if_format) {
+		case XFS_DINODE_FMT_LOCAL:
+			error = xfs_iformat_local(ip, dip, XFS_DATA_FORK,
+					be64_to_cpu(dip->di_size));
+			if (!error)
+				error = xfs_ifork_verify_local_data(ip);
+			return error;
+		case XFS_DINODE_FMT_EXTENTS:
+			return xfs_iformat_extents(ip, dip, XFS_DATA_FORK);
+		case XFS_DINODE_FMT_BTREE:
+			return xfs_iformat_btree(ip, dip, XFS_DATA_FORK);
+		default:
+			xfs_inode_verifier_error(ip, -EFSCORRUPTED, __func__,
+					dip, sizeof(*dip), __this_address);
+			return -EFSCORRUPTED;
+		}
+		break;
+	default:
+		xfs_inode_verifier_error(ip, -EFSCORRUPTED, __func__, dip,
+				sizeof(*dip), __this_address);
+		return -EFSCORRUPTED;
+	}
+}
+
+static uint16_t
+xfs_dfork_attr_shortform_size(
+	struct xfs_dinode		*dip)
+{
+	struct xfs_attr_shortform	*atp =
+		(struct xfs_attr_shortform *)XFS_DFORK_APTR(dip);
+
+	return be16_to_cpu(atp->hdr.totsize);
+}
+
+void
+xfs_ifork_init_attr(
+	struct xfs_inode	*ip,
+	enum xfs_dinode_fmt	format,
+	xfs_extnum_t		nextents)
+{
+	ip->i_af.if_format = format;
+	ip->i_af.if_nextents = nextents;
+}
+
+void
+xfs_ifork_zap_attr(
+	struct xfs_inode	*ip)
+{
+	xfs_idestroy_fork(&ip->i_af);
+	memset(&ip->i_af, 0, sizeof(struct xfs_ifork));
+	ip->i_af.if_format = XFS_DINODE_FMT_EXTENTS;
+}
+
+int
+xfs_iformat_attr_fork(
+	struct xfs_inode	*ip,
+	struct xfs_dinode	*dip)
+{
+	xfs_extnum_t		naextents = xfs_dfork_attr_extents(dip);
+	int			error = 0;
+
+	/*
+	 * Initialize the extent count early, as the per-format routines may
+	 * depend on it.
+	 */
+	xfs_ifork_init_attr(ip, dip->di_aformat, naextents);
+
+	switch (ip->i_af.if_format) {
+	case XFS_DINODE_FMT_LOCAL:
+		error = xfs_iformat_local(ip, dip, XFS_ATTR_FORK,
+				xfs_dfork_attr_shortform_size(dip));
+		if (!error)
+			error = xfs_ifork_verify_local_attr(ip);
+		break;
+	case XFS_DINODE_FMT_EXTENTS:
+		error = xfs_iformat_extents(ip, dip, XFS_ATTR_FORK);
+		break;
+	case XFS_DINODE_FMT_BTREE:
+		error = xfs_iformat_btree(ip, dip, XFS_ATTR_FORK);
+		break;
+	default:
+		xfs_inode_verifier_error(ip, error, __func__, dip,
+				sizeof(*dip), __this_address);
+		error = -EFSCORRUPTED;
+		break;
+	}
+
+	if (error)
+		xfs_ifork_zap_attr(ip);
+	return error;
+}
+
+/*
+ * Reallocate the space for if_broot based on the number of records
+ * being added or deleted as indicated in rec_diff.  Move the records
+ * and pointers in if_broot to fit the new size.  When shrinking this
+ * will eliminate holes between the records and pointers created by
+ * the caller.  When growing this will create holes to be filled in
+ * by the caller.
+ *
+ * The caller must not request to add more records than would fit in
+ * the on-disk inode root.  If the if_broot is currently NULL, then
+ * if we are adding records, one will be allocated.  The caller must also
+ * not request that the number of records go below zero, although
+ * it can go to zero.
+ *
+ * ip -- the inode whose if_broot area is changing
+ * ext_diff -- the change in the number of records, positive or negative,
+ *	 requested for the if_broot array.
+ */
+void
+xfs_iroot_realloc(
+	xfs_inode_t		*ip,
+	int			rec_diff,
+	int			whichfork)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	int			cur_max;
+	struct xfs_ifork	*ifp;
+	struct xfs_btree_block	*new_broot;
+	int			new_max;
+	size_t			new_size;
+	char			*np;
+	char			*op;
+
+	/*
+	 * Handle the degenerate case quietly.
+	 */
+	if (rec_diff == 0) {
+		return;
+	}
+
+	ifp = xfs_ifork_ptr(ip, whichfork);
+	if (rec_diff > 0) {
+		/*
+		 * If there wasn't any memory allocated before, just
+		 * allocate it now and get out.
+		 */
+		if (ifp->if_broot_bytes == 0) {
+			new_size = XFS_BMAP_BROOT_SPACE_CALC(mp, rec_diff);
+			ifp->if_broot = kmem_alloc(new_size, KM_NOFS);
+			ifp->if_broot_bytes = (int)new_size;
+			return;
+		}
+
+		/*
+		 * If there is already an existing if_broot, then we need
+		 * to realloc() it and shift the pointers to their new
+		 * location.  The records don't change location because
+		 * they are kept butted up against the btree block header.
+		 */
+		cur_max = xfs_bmbt_maxrecs(mp, ifp->if_broot_bytes, 0);
+		new_max = cur_max + rec_diff;
+		new_size = XFS_BMAP_BROOT_SPACE_CALC(mp, new_max);
+		ifp->if_broot = krealloc(ifp->if_broot, new_size,
+					 GFP_NOFS | __GFP_NOFAIL);
+		op = (char *)XFS_BMAP_BROOT_PTR_ADDR(mp, ifp->if_broot, 1,
+						     ifp->if_broot_bytes);
+		np = (char *)XFS_BMAP_BROOT_PTR_ADDR(mp, ifp->if_broot, 1,
+						     (int)new_size);
+		ifp->if_broot_bytes = (int)new_size;
+		ASSERT(XFS_BMAP_BMDR_SPACE(ifp->if_broot) <=
+			xfs_inode_fork_size(ip, whichfork));
+		memmove(np, op, cur_max * (uint)sizeof(xfs_fsblock_t));
+		return;
+	}
+
+	/*
+	 * rec_diff is less than 0.  In this case, we are shrinking the
+	 * if_broot buffer.  It must already exist.  If we go to zero
+	 * records, just get rid of the root and clear the status bit.
+	 */
+	ASSERT((ifp->if_broot != NULL) && (ifp->if_broot_bytes > 0));
+	cur_max = xfs_bmbt_maxrecs(mp, ifp->if_broot_bytes, 0);
+	new_max = cur_max + rec_diff;
+	ASSERT(new_max >= 0);
+	if (new_max > 0)
+		new_size = XFS_BMAP_BROOT_SPACE_CALC(mp, new_max);
+	else
+		new_size = 0;
+	if (new_size > 0) {
+		new_broot = kmem_alloc(new_size, KM_NOFS);
+		/*
+		 * First copy over the btree block header.
+		 */
+		memcpy(new_broot, ifp->if_broot,
+			XFS_BMBT_BLOCK_LEN(ip->i_mount));
+	} else {
+		new_broot = NULL;
+	}
+
+	/*
+	 * Only copy the records and pointers if there are any.
+	 */
+	if (new_max > 0) {
+		/*
+		 * First copy the records.
+		 */
+		op = (char *)XFS_BMBT_REC_ADDR(mp, ifp->if_broot, 1);
+		np = (char *)XFS_BMBT_REC_ADDR(mp, new_broot, 1);
+		memcpy(np, op, new_max * (uint)sizeof(xfs_bmbt_rec_t));
+
+		/*
+		 * Then copy the pointers.
+		 */
+		op = (char *)XFS_BMAP_BROOT_PTR_ADDR(mp, ifp->if_broot, 1,
+						     ifp->if_broot_bytes);
+		np = (char *)XFS_BMAP_BROOT_PTR_ADDR(mp, new_broot, 1,
+						     (int)new_size);
+		memcpy(np, op, new_max * (uint)sizeof(xfs_fsblock_t));
+	}
+	kmem_free(ifp->if_broot);
+	ifp->if_broot = new_broot;
+	ifp->if_broot_bytes = (int)new_size;
+	if (ifp->if_broot)
+		ASSERT(XFS_BMAP_BMDR_SPACE(ifp->if_broot) <=
+			xfs_inode_fork_size(ip, whichfork));
+	return;
+}
+
+
+/*
+ * This is called when the amount of space needed for if_data
+ * is increased or decreased.  The change in size is indicated by
+ * the number of bytes that need to be added or deleted in the
+ * byte_diff parameter.
+ *
+ * If the amount of space needed has decreased below the size of the
+ * inline buffer, then switch to using the inline buffer.  Otherwise,
+ * use kmem_realloc() or kmem_alloc() to adjust the size of the buffer
+ * to what is needed.
+ *
+ * ip -- the inode whose if_data area is changing
+ * byte_diff -- the change in the number of bytes, positive or negative,
+ *	 requested for the if_data array.
+ */
+void
+xfs_idata_realloc(
+	struct xfs_inode	*ip,
+	int64_t			byte_diff,
+	int			whichfork)
+{
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	int64_t			new_size = ifp->if_bytes + byte_diff;
+
+	ASSERT(new_size >= 0);
+	ASSERT(new_size <= xfs_inode_fork_size(ip, whichfork));
+
+	if (byte_diff == 0)
+		return;
+
+	if (new_size == 0) {
+		kmem_free(ifp->if_u1.if_data);
+		ifp->if_u1.if_data = NULL;
+		ifp->if_bytes = 0;
+		return;
+	}
+
+	ifp->if_u1.if_data = krealloc(ifp->if_u1.if_data, new_size,
+				      GFP_NOFS | __GFP_NOFAIL);
+	ifp->if_bytes = new_size;
+}
+
+void
+xfs_idestroy_fork(
+	struct xfs_ifork	*ifp)
+{
+	if (ifp->if_broot != NULL) {
+		kmem_free(ifp->if_broot);
+		ifp->if_broot = NULL;
+	}
+
+	switch (ifp->if_format) {
+	case XFS_DINODE_FMT_LOCAL:
+		kmem_free(ifp->if_u1.if_data);
+		ifp->if_u1.if_data = NULL;
+		break;
+	case XFS_DINODE_FMT_EXTENTS:
+	case XFS_DINODE_FMT_BTREE:
+		if (ifp->if_height)
+			xfs_iext_destroy(ifp);
+		break;
+	}
+}
+
+/*
+ * Convert in-core extents to on-disk form
+ *
+ * In the case of the data fork, the in-core and on-disk fork sizes can be
+ * different due to delayed allocation extents. We only copy on-disk extents
+ * here, so callers must always use the physical fork size to determine the
+ * size of the buffer passed to this routine.  We will return the size actually
+ * used.
+ */
+int
+xfs_iextents_copy(
+	struct xfs_inode	*ip,
+	struct xfs_bmbt_rec	*dp,
+	int			whichfork)
+{
+	int			state = xfs_bmap_fork_to_state(whichfork);
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	struct xfs_iext_cursor	icur;
+	struct xfs_bmbt_irec	rec;
+	int64_t			copied = 0;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL | XFS_ILOCK_SHARED));
+	ASSERT(ifp->if_bytes > 0);
+
+	for_each_xfs_iext(ifp, &icur, &rec) {
+		if (isnullstartblock(rec.br_startblock))
+			continue;
+		ASSERT(xfs_bmap_validate_extent(ip, whichfork, &rec) == NULL);
+		xfs_bmbt_disk_set_all(dp, &rec);
+		trace_xfs_write_extent(ip, &icur, state, _RET_IP_);
+		copied += sizeof(struct xfs_bmbt_rec);
+		dp++;
+	}
+
+	ASSERT(copied > 0);
+	ASSERT(copied <= ifp->if_bytes);
+	return copied;
+}
+
+/*
+ * Each of the following cases stores data into the same region
+ * of the on-disk inode, so only one of them can be valid at
+ * any given time. While it is possible to have conflicting formats
+ * and log flags, e.g. having XFS_ILOG_?DATA set when the fork is
+ * in EXTENTS format, this can only happen when the fork has
+ * changed formats after being modified but before being flushed.
+ * In these cases, the format always takes precedence, because the
+ * format indicates the current state of the fork.
+ */
+void
+xfs_iflush_fork(
+	struct xfs_inode	*ip,
+	struct xfs_dinode	*dip,
+	struct xfs_inode_log_item *iip,
+	int			whichfork)
+{
+	char			*cp;
+	struct xfs_ifork	*ifp;
+	xfs_mount_t		*mp;
+	static const short	brootflag[2] =
+		{ XFS_ILOG_DBROOT, XFS_ILOG_ABROOT };
+	static const short	dataflag[2] =
+		{ XFS_ILOG_DDATA, XFS_ILOG_ADATA };
+	static const short	extflag[2] =
+		{ XFS_ILOG_DEXT, XFS_ILOG_AEXT };
+
+	if (!iip)
+		return;
+	ifp = xfs_ifork_ptr(ip, whichfork);
+	/*
+	 * This can happen if we gave up in iformat in an error path,
+	 * for the attribute fork.
+	 */
+	if (!ifp) {
+		ASSERT(whichfork == XFS_ATTR_FORK);
+		return;
+	}
+	cp = XFS_DFORK_PTR(dip, whichfork);
+	mp = ip->i_mount;
+	switch (ifp->if_format) {
+	case XFS_DINODE_FMT_LOCAL:
+		if ((iip->ili_fields & dataflag[whichfork]) &&
+		    (ifp->if_bytes > 0)) {
+			ASSERT(ifp->if_u1.if_data != NULL);
+			ASSERT(ifp->if_bytes <= xfs_inode_fork_size(ip, whichfork));
+			memcpy(cp, ifp->if_u1.if_data, ifp->if_bytes);
+		}
+		break;
+
+	case XFS_DINODE_FMT_EXTENTS:
+		if ((iip->ili_fields & extflag[whichfork]) &&
+		    (ifp->if_bytes > 0)) {
+			ASSERT(ifp->if_nextents > 0);
+			(void)xfs_iextents_copy(ip, (xfs_bmbt_rec_t *)cp,
+				whichfork);
+		}
+		break;
+
+	case XFS_DINODE_FMT_BTREE:
+		if ((iip->ili_fields & brootflag[whichfork]) &&
+		    (ifp->if_broot_bytes > 0)) {
+			ASSERT(ifp->if_broot != NULL);
+			ASSERT(XFS_BMAP_BMDR_SPACE(ifp->if_broot) <=
+			        xfs_inode_fork_size(ip, whichfork));
+			xfs_bmbt_to_bmdr(mp, ifp->if_broot, ifp->if_broot_bytes,
+				(xfs_bmdr_block_t *)cp,
+				XFS_DFORK_SIZE(dip, mp, whichfork));
+		}
+		break;
+
+	case XFS_DINODE_FMT_DEV:
+		if (iip->ili_fields & XFS_ILOG_DEV) {
+			ASSERT(whichfork == XFS_DATA_FORK);
+			xfs_dinode_put_rdev(dip,
+					linux_to_xfs_dev_t(VFS_I(ip)->i_rdev));
+		}
+		break;
+
+	default:
+		ASSERT(0);
+		break;
+	}
+}
+
+/* Convert bmap state flags to an inode fork. */
+struct xfs_ifork *
+xfs_iext_state_to_fork(
+	struct xfs_inode	*ip,
+	int			state)
+{
+	if (state & BMAP_COWFORK)
+		return ip->i_cowfp;
+	else if (state & BMAP_ATTRFORK)
+		return &ip->i_af;
+	return &ip->i_df;
+}
+
+/*
+ * Initialize an inode's copy-on-write fork.
+ */
+void
+xfs_ifork_init_cow(
+	struct xfs_inode	*ip)
+{
+	if (ip->i_cowfp)
+		return;
+
+	ip->i_cowfp = kmem_cache_zalloc(xfs_ifork_cache,
+				       GFP_NOFS | __GFP_NOFAIL);
+	ip->i_cowfp->if_format = XFS_DINODE_FMT_EXTENTS;
+}
+
+/* Verify the inline contents of the data fork of an inode. */
+int
+xfs_ifork_verify_local_data(
+	struct xfs_inode	*ip)
+{
+	xfs_failaddr_t		fa = NULL;
+
+	switch (VFS_I(ip)->i_mode & S_IFMT) {
+	case S_IFDIR:
+		fa = xfs_dir2_sf_verify(ip);
+		break;
+	case S_IFLNK:
+		fa = xfs_symlink_shortform_verify(ip);
+		break;
+	default:
+		break;
+	}
+
+	if (fa) {
+		xfs_inode_verifier_error(ip, -EFSCORRUPTED, "data fork",
+				ip->i_df.if_u1.if_data, ip->i_df.if_bytes, fa);
+		return -EFSCORRUPTED;
+	}
+
+	return 0;
+}
+
+/* Verify the inline contents of the attr fork of an inode. */
+int
+xfs_ifork_verify_local_attr(
+	struct xfs_inode	*ip)
+{
+	struct xfs_ifork	*ifp = &ip->i_af;
+	xfs_failaddr_t		fa;
+
+	if (!xfs_inode_has_attr_fork(ip))
+		fa = __this_address;
+	else
+		fa = xfs_attr_shortform_verify(ip);
+
+	if (fa) {
+		xfs_inode_verifier_error(ip, -EFSCORRUPTED, "attr fork",
+				ifp->if_u1.if_data, ifp->if_bytes, fa);
+		return -EFSCORRUPTED;
+	}
+
+	return 0;
+}
+
+int
+xfs_iext_count_may_overflow(
+	struct xfs_inode	*ip,
+	int			whichfork,
+	int			nr_to_add)
+{
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
+	uint64_t		max_exts;
+	uint64_t		nr_exts;
+
+	if (whichfork == XFS_COW_FORK)
+		return 0;
+
+	max_exts = xfs_iext_max_nextents(xfs_inode_has_large_extent_counts(ip),
+				whichfork);
+
+	if (XFS_TEST_ERROR(false, ip->i_mount, XFS_ERRTAG_REDUCE_MAX_IEXTENTS))
+		max_exts = 10;
+
+	nr_exts = ifp->if_nextents + nr_to_add;
+	if (nr_exts < ifp->if_nextents || nr_exts > max_exts)
+		return -EFBIG;
+
+	return 0;
+}
+
+/*
+ * Upgrade this inode's extent counter fields to be able to handle a potential
+ * increase in the extent count by nr_to_add.  Normally this is the same
+ * quantity that caused xfs_iext_count_may_overflow() to return -EFBIG.
+ */
+int
+xfs_iext_count_upgrade(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	uint			nr_to_add)
+{
+	ASSERT(nr_to_add <= XFS_MAX_EXTCNT_UPGRADE_NR);
+
+	if (!xfs_has_large_extent_counts(ip->i_mount) ||
+	    xfs_inode_has_large_extent_counts(ip) ||
+	    XFS_TEST_ERROR(false, ip->i_mount, XFS_ERRTAG_REDUCE_MAX_IEXTENTS))
+		return -EFBIG;
+
+	ip->i_diflags2 |= XFS_DIFLAG2_NREXT64;
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+
+	return 0;
+}
diff --git a/fs/xfs/libxfs/xfs_inode_fork.h b/fs/xfs/libxfs/xfs_inode_fork.h
new file mode 100644
index 000000000..d3943d6ad
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_inode_fork.h
@@ -0,0 +1,268 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef	__XFS_INODE_FORK_H__
+#define	__XFS_INODE_FORK_H__
+
+struct xfs_inode_log_item;
+struct xfs_dinode;
+
+/*
+ * File incore extent information, present for each of data & attr forks.
+ */
+struct xfs_ifork {
+	int64_t			if_bytes;	/* bytes in if_u1 */
+	struct xfs_btree_block	*if_broot;	/* file's incore btree root */
+	unsigned int		if_seq;		/* fork mod counter */
+	int			if_height;	/* height of the extent tree */
+	union {
+		void		*if_root;	/* extent tree root */
+		char		*if_data;	/* inline file data */
+	} if_u1;
+	xfs_extnum_t		if_nextents;	/* # of extents in this fork */
+	short			if_broot_bytes;	/* bytes allocated for root */
+	int8_t			if_format;	/* format of this fork */
+};
+
+/*
+ * Worst-case increase in the fork extent count when we're adding a single
+ * extent to a fork and there's no possibility of splitting an existing mapping.
+ */
+#define XFS_IEXT_ADD_NOSPLIT_CNT	(1)
+
+/*
+ * Punching out an extent from the middle of an existing extent can cause the
+ * extent count to increase by 1.
+ * i.e. | Old extent | Hole | Old extent |
+ */
+#define XFS_IEXT_PUNCH_HOLE_CNT		(1)
+
+/*
+ * Adding/removing an xattr can cause XFS_DA_NODE_MAXDEPTH extents to
+ * be added. One extra extent for dabtree in case a local attr is
+ * large enough to cause a double split.  It can also cause extent
+ * count to increase proportional to the size of a remote xattr's
+ * value.
+ */
+#define XFS_IEXT_ATTR_MANIP_CNT(rmt_blks) \
+	(XFS_DA_NODE_MAXDEPTH + max(1, rmt_blks))
+
+/*
+ * A write to a sub-interval of an existing unwritten extent causes the original
+ * extent to be split into 3 extents
+ * i.e. | Unwritten | Real | Unwritten |
+ * Hence extent count can increase by 2.
+ */
+#define XFS_IEXT_WRITE_UNWRITTEN_CNT	(2)
+
+
+/*
+ * Moving an extent to data fork can cause a sub-interval of an existing extent
+ * to be unmapped. This will increase extent count by 1. Mapping in the new
+ * extent can increase the extent count by 1 again i.e.
+ * | Old extent | New extent | Old extent |
+ * Hence number of extents increases by 2.
+ */
+#define XFS_IEXT_REFLINK_END_COW_CNT	(2)
+
+/*
+ * Removing an initial range of source/donor file's extent and adding a new
+ * extent (from donor/source file) in its place will cause extent count to
+ * increase by 1.
+ */
+#define XFS_IEXT_SWAP_RMAP_CNT		(1)
+
+/*
+ * Fork handling.
+ */
+#define XFS_IFORK_MAXEXT(ip, w) \
+	(xfs_inode_fork_size(ip, w) / sizeof(xfs_bmbt_rec_t))
+
+static inline bool xfs_ifork_has_extents(struct xfs_ifork *ifp)
+{
+	return ifp->if_format == XFS_DINODE_FMT_EXTENTS ||
+		ifp->if_format == XFS_DINODE_FMT_BTREE;
+}
+
+static inline xfs_extnum_t xfs_ifork_nextents(struct xfs_ifork *ifp)
+{
+	if (!ifp)
+		return 0;
+	return ifp->if_nextents;
+}
+
+static inline int8_t xfs_ifork_format(struct xfs_ifork *ifp)
+{
+	if (!ifp)
+		return XFS_DINODE_FMT_EXTENTS;
+	return ifp->if_format;
+}
+
+static inline xfs_extnum_t xfs_iext_max_nextents(bool has_large_extent_counts,
+				int whichfork)
+{
+	switch (whichfork) {
+	case XFS_DATA_FORK:
+	case XFS_COW_FORK:
+		if (has_large_extent_counts)
+			return XFS_MAX_EXTCNT_DATA_FORK_LARGE;
+		return XFS_MAX_EXTCNT_DATA_FORK_SMALL;
+
+	case XFS_ATTR_FORK:
+		if (has_large_extent_counts)
+			return XFS_MAX_EXTCNT_ATTR_FORK_LARGE;
+		return XFS_MAX_EXTCNT_ATTR_FORK_SMALL;
+
+	default:
+		ASSERT(0);
+		return 0;
+	}
+}
+
+static inline xfs_extnum_t
+xfs_dfork_data_extents(
+	struct xfs_dinode	*dip)
+{
+	if (xfs_dinode_has_large_extent_counts(dip))
+		return be64_to_cpu(dip->di_big_nextents);
+
+	return be32_to_cpu(dip->di_nextents);
+}
+
+static inline xfs_extnum_t
+xfs_dfork_attr_extents(
+	struct xfs_dinode	*dip)
+{
+	if (xfs_dinode_has_large_extent_counts(dip))
+		return be32_to_cpu(dip->di_big_anextents);
+
+	return be16_to_cpu(dip->di_anextents);
+}
+
+static inline xfs_extnum_t
+xfs_dfork_nextents(
+	struct xfs_dinode	*dip,
+	int			whichfork)
+{
+	switch (whichfork) {
+	case XFS_DATA_FORK:
+		return xfs_dfork_data_extents(dip);
+	case XFS_ATTR_FORK:
+		return xfs_dfork_attr_extents(dip);
+	default:
+		ASSERT(0);
+		break;
+	}
+
+	return 0;
+}
+
+void xfs_ifork_zap_attr(struct xfs_inode *ip);
+void xfs_ifork_init_attr(struct xfs_inode *ip, enum xfs_dinode_fmt format,
+		xfs_extnum_t nextents);
+struct xfs_ifork *xfs_iext_state_to_fork(struct xfs_inode *ip, int state);
+
+int		xfs_iformat_data_fork(struct xfs_inode *, struct xfs_dinode *);
+int		xfs_iformat_attr_fork(struct xfs_inode *, struct xfs_dinode *);
+void		xfs_iflush_fork(struct xfs_inode *, struct xfs_dinode *,
+				struct xfs_inode_log_item *, int);
+void		xfs_idestroy_fork(struct xfs_ifork *ifp);
+void		xfs_idata_realloc(struct xfs_inode *ip, int64_t byte_diff,
+				int whichfork);
+void		xfs_iroot_realloc(struct xfs_inode *, int, int);
+int		xfs_iread_extents(struct xfs_trans *, struct xfs_inode *, int);
+int		xfs_iextents_copy(struct xfs_inode *, struct xfs_bmbt_rec *,
+				  int);
+void		xfs_init_local_fork(struct xfs_inode *ip, int whichfork,
+				const void *data, int64_t size);
+
+xfs_extnum_t	xfs_iext_count(struct xfs_ifork *ifp);
+void		xfs_iext_insert(struct xfs_inode *, struct xfs_iext_cursor *cur,
+			struct xfs_bmbt_irec *, int);
+void		xfs_iext_remove(struct xfs_inode *, struct xfs_iext_cursor *,
+			int);
+void		xfs_iext_destroy(struct xfs_ifork *);
+
+bool		xfs_iext_lookup_extent(struct xfs_inode *ip,
+			struct xfs_ifork *ifp, xfs_fileoff_t bno,
+			struct xfs_iext_cursor *cur,
+			struct xfs_bmbt_irec *gotp);
+bool		xfs_iext_lookup_extent_before(struct xfs_inode *ip,
+			struct xfs_ifork *ifp, xfs_fileoff_t *end,
+			struct xfs_iext_cursor *cur,
+			struct xfs_bmbt_irec *gotp);
+bool		xfs_iext_get_extent(struct xfs_ifork *ifp,
+			struct xfs_iext_cursor *cur,
+			struct xfs_bmbt_irec *gotp);
+void		xfs_iext_update_extent(struct xfs_inode *ip, int state,
+			struct xfs_iext_cursor *cur,
+			struct xfs_bmbt_irec *gotp);
+
+void		xfs_iext_first(struct xfs_ifork *, struct xfs_iext_cursor *);
+void		xfs_iext_last(struct xfs_ifork *, struct xfs_iext_cursor *);
+void		xfs_iext_next(struct xfs_ifork *, struct xfs_iext_cursor *);
+void		xfs_iext_prev(struct xfs_ifork *, struct xfs_iext_cursor *);
+
+static inline bool xfs_iext_next_extent(struct xfs_ifork *ifp,
+		struct xfs_iext_cursor *cur, struct xfs_bmbt_irec *gotp)
+{
+	xfs_iext_next(ifp, cur);
+	return xfs_iext_get_extent(ifp, cur, gotp);
+}
+
+static inline bool xfs_iext_prev_extent(struct xfs_ifork *ifp,
+		struct xfs_iext_cursor *cur, struct xfs_bmbt_irec *gotp)
+{
+	xfs_iext_prev(ifp, cur);
+	return xfs_iext_get_extent(ifp, cur, gotp);
+}
+
+/*
+ * Return the extent after cur in gotp without updating the cursor.
+ */
+static inline bool xfs_iext_peek_next_extent(struct xfs_ifork *ifp,
+		struct xfs_iext_cursor *cur, struct xfs_bmbt_irec *gotp)
+{
+	struct xfs_iext_cursor ncur = *cur;
+
+	xfs_iext_next(ifp, &ncur);
+	return xfs_iext_get_extent(ifp, &ncur, gotp);
+}
+
+/*
+ * Return the extent before cur in gotp without updating the cursor.
+ */
+static inline bool xfs_iext_peek_prev_extent(struct xfs_ifork *ifp,
+		struct xfs_iext_cursor *cur, struct xfs_bmbt_irec *gotp)
+{
+	struct xfs_iext_cursor ncur = *cur;
+
+	xfs_iext_prev(ifp, &ncur);
+	return xfs_iext_get_extent(ifp, &ncur, gotp);
+}
+
+#define for_each_xfs_iext(ifp, ext, got)		\
+	for (xfs_iext_first((ifp), (ext));		\
+	     xfs_iext_get_extent((ifp), (ext), (got));	\
+	     xfs_iext_next((ifp), (ext)))
+
+extern struct kmem_cache	*xfs_ifork_cache;
+
+extern void xfs_ifork_init_cow(struct xfs_inode *ip);
+
+int xfs_ifork_verify_local_data(struct xfs_inode *ip);
+int xfs_ifork_verify_local_attr(struct xfs_inode *ip);
+int xfs_iext_count_may_overflow(struct xfs_inode *ip, int whichfork,
+		int nr_to_add);
+int xfs_iext_count_upgrade(struct xfs_trans *tp, struct xfs_inode *ip,
+		uint nr_to_add);
+
+/* returns true if the fork has extents but they are not read in yet. */
+static inline bool xfs_need_iread_extents(struct xfs_ifork *ifp)
+{
+	return ifp->if_format == XFS_DINODE_FMT_BTREE && ifp->if_height == 0;
+}
+
+#endif	/* __XFS_INODE_FORK_H__ */
diff --git a/fs/xfs/libxfs/xfs_log_format.h b/fs/xfs/libxfs/xfs_log_format.h
new file mode 100644
index 000000000..f13e0809d
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_log_format.h
@@ -0,0 +1,993 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef	__XFS_LOG_FORMAT_H__
+#define __XFS_LOG_FORMAT_H__
+
+struct xfs_mount;
+struct xfs_trans_res;
+
+/*
+ * On-disk Log Format definitions.
+ *
+ * This file contains all the on-disk format definitions used within the log. It
+ * includes the physical log structure itself, as well as all the log item
+ * format structures that are written into the log and intepreted by log
+ * recovery. We start with the physical log format definitions, and then work
+ * through all the log items definitions and everything they encode into the
+ * log.
+ */
+typedef uint32_t xlog_tid_t;
+
+#define XLOG_MIN_ICLOGS		2
+#define XLOG_MAX_ICLOGS		8
+#define XLOG_HEADER_MAGIC_NUM	0xFEEDbabe	/* Invalid cycle number */
+#define XLOG_VERSION_1		1
+#define XLOG_VERSION_2		2		/* Large IClogs, Log sunit */
+#define XLOG_VERSION_OKBITS	(XLOG_VERSION_1 | XLOG_VERSION_2)
+#define XLOG_MIN_RECORD_BSIZE	(16*1024)	/* eventually 32k */
+#define XLOG_BIG_RECORD_BSIZE	(32*1024)	/* 32k buffers */
+#define XLOG_MAX_RECORD_BSIZE	(256*1024)
+#define XLOG_HEADER_CYCLE_SIZE	(32*1024)	/* cycle data in header */
+#define XLOG_MIN_RECORD_BSHIFT	14		/* 16384 == 1 << 14 */
+#define XLOG_BIG_RECORD_BSHIFT	15		/* 32k == 1 << 15 */
+#define XLOG_MAX_RECORD_BSHIFT	18		/* 256k == 1 << 18 */
+
+#define XLOG_HEADER_SIZE	512
+
+/* Minimum number of transactions that must fit in the log (defined by mkfs) */
+#define XFS_MIN_LOG_FACTOR	3
+
+#define XLOG_REC_SHIFT(log) \
+	BTOBB(1 << (xfs_has_logv2(log->l_mp) ? \
+	 XLOG_MAX_RECORD_BSHIFT : XLOG_BIG_RECORD_BSHIFT))
+#define XLOG_TOTAL_REC_SHIFT(log) \
+	BTOBB(XLOG_MAX_ICLOGS << (xfs_has_logv2(log->l_mp) ? \
+	 XLOG_MAX_RECORD_BSHIFT : XLOG_BIG_RECORD_BSHIFT))
+
+/* get lsn fields */
+#define CYCLE_LSN(lsn) ((uint)((lsn)>>32))
+#define BLOCK_LSN(lsn) ((uint)(lsn))
+
+/* this is used in a spot where we might otherwise double-endian-flip */
+#define CYCLE_LSN_DISK(lsn) (((__be32 *)&(lsn))[0])
+
+static inline xfs_lsn_t xlog_assign_lsn(uint cycle, uint block)
+{
+	return ((xfs_lsn_t)cycle << 32) | block;
+}
+
+static inline uint xlog_get_cycle(char *ptr)
+{
+	if (be32_to_cpu(*(__be32 *)ptr) == XLOG_HEADER_MAGIC_NUM)
+		return be32_to_cpu(*((__be32 *)ptr + 1));
+	else
+		return be32_to_cpu(*(__be32 *)ptr);
+}
+
+/* Log Clients */
+#define XFS_TRANSACTION		0x69
+#define XFS_LOG			0xaa
+
+#define XLOG_UNMOUNT_TYPE	0x556e	/* Un for Unmount */
+
+/*
+ * Log item for unmount records.
+ *
+ * The unmount record used to have a string "Unmount filesystem--" in the
+ * data section where the "Un" was really a magic number (XLOG_UNMOUNT_TYPE).
+ * We just write the magic number now; see xfs_log_unmount_write.
+ */
+struct xfs_unmount_log_format {
+	uint16_t	magic;	/* XLOG_UNMOUNT_TYPE */
+	uint16_t	pad1;
+	uint32_t	pad2;	/* may as well make it 64 bits */
+};
+
+/* Region types for iovec's i_type */
+#define XLOG_REG_TYPE_BFORMAT		1
+#define XLOG_REG_TYPE_BCHUNK		2
+#define XLOG_REG_TYPE_EFI_FORMAT	3
+#define XLOG_REG_TYPE_EFD_FORMAT	4
+#define XLOG_REG_TYPE_IFORMAT		5
+#define XLOG_REG_TYPE_ICORE		6
+#define XLOG_REG_TYPE_IEXT		7
+#define XLOG_REG_TYPE_IBROOT		8
+#define XLOG_REG_TYPE_ILOCAL		9
+#define XLOG_REG_TYPE_IATTR_EXT		10
+#define XLOG_REG_TYPE_IATTR_BROOT	11
+#define XLOG_REG_TYPE_IATTR_LOCAL	12
+#define XLOG_REG_TYPE_QFORMAT		13
+#define XLOG_REG_TYPE_DQUOT		14
+#define XLOG_REG_TYPE_QUOTAOFF		15
+#define XLOG_REG_TYPE_LRHEADER		16
+#define XLOG_REG_TYPE_UNMOUNT		17
+#define XLOG_REG_TYPE_COMMIT		18
+#define XLOG_REG_TYPE_TRANSHDR		19
+#define XLOG_REG_TYPE_ICREATE		20
+#define XLOG_REG_TYPE_RUI_FORMAT	21
+#define XLOG_REG_TYPE_RUD_FORMAT	22
+#define XLOG_REG_TYPE_CUI_FORMAT	23
+#define XLOG_REG_TYPE_CUD_FORMAT	24
+#define XLOG_REG_TYPE_BUI_FORMAT	25
+#define XLOG_REG_TYPE_BUD_FORMAT	26
+#define XLOG_REG_TYPE_ATTRI_FORMAT	27
+#define XLOG_REG_TYPE_ATTRD_FORMAT	28
+#define XLOG_REG_TYPE_ATTR_NAME	29
+#define XLOG_REG_TYPE_ATTR_VALUE	30
+#define XLOG_REG_TYPE_MAX		30
+
+
+/*
+ * Flags to log operation header
+ *
+ * The first write of a new transaction will be preceded with a start
+ * record, XLOG_START_TRANS.  Once a transaction is committed, a commit
+ * record is written, XLOG_COMMIT_TRANS.  If a single region can not fit into
+ * the remainder of the current active in-core log, it is split up into
+ * multiple regions.  Each partial region will be marked with a
+ * XLOG_CONTINUE_TRANS until the last one, which gets marked with XLOG_END_TRANS.
+ *
+ */
+#define XLOG_START_TRANS	0x01	/* Start a new transaction */
+#define XLOG_COMMIT_TRANS	0x02	/* Commit this transaction */
+#define XLOG_CONTINUE_TRANS	0x04	/* Cont this trans into new region */
+#define XLOG_WAS_CONT_TRANS	0x08	/* Cont this trans into new region */
+#define XLOG_END_TRANS		0x10	/* End a continued transaction */
+#define XLOG_UNMOUNT_TRANS	0x20	/* Unmount a filesystem transaction */
+
+
+typedef struct xlog_op_header {
+	__be32	   oh_tid;	/* transaction id of operation	:  4 b */
+	__be32	   oh_len;	/* bytes in data region		:  4 b */
+	__u8	   oh_clientid;	/* who sent me this		:  1 b */
+	__u8	   oh_flags;	/*				:  1 b */
+	__u16	   oh_res2;	/* 32 bit align			:  2 b */
+} xlog_op_header_t;
+
+/* valid values for h_fmt */
+#define XLOG_FMT_UNKNOWN  0
+#define XLOG_FMT_LINUX_LE 1
+#define XLOG_FMT_LINUX_BE 2
+#define XLOG_FMT_IRIX_BE  3
+
+/* our fmt */
+#ifdef XFS_NATIVE_HOST
+#define XLOG_FMT XLOG_FMT_LINUX_BE
+#else
+#define XLOG_FMT XLOG_FMT_LINUX_LE
+#endif
+
+typedef struct xlog_rec_header {
+	__be32	  h_magicno;	/* log record (LR) identifier		:  4 */
+	__be32	  h_cycle;	/* write cycle of log			:  4 */
+	__be32	  h_version;	/* LR version				:  4 */
+	__be32	  h_len;	/* len in bytes; should be 64-bit aligned: 4 */
+	__be64	  h_lsn;	/* lsn of this LR			:  8 */
+	__be64	  h_tail_lsn;	/* lsn of 1st LR w/ buffers not committed: 8 */
+	__le32	  h_crc;	/* crc of log record                    :  4 */
+	__be32	  h_prev_block; /* block number to previous LR		:  4 */
+	__be32	  h_num_logops;	/* number of log operations in this LR	:  4 */
+	__be32	  h_cycle_data[XLOG_HEADER_CYCLE_SIZE / BBSIZE];
+	/* new fields */
+	__be32    h_fmt;        /* format of log record                 :  4 */
+	uuid_t	  h_fs_uuid;    /* uuid of FS                           : 16 */
+	__be32	  h_size;	/* iclog size				:  4 */
+} xlog_rec_header_t;
+
+typedef struct xlog_rec_ext_header {
+	__be32	  xh_cycle;	/* write cycle of log			: 4 */
+	__be32	  xh_cycle_data[XLOG_HEADER_CYCLE_SIZE / BBSIZE]; /*	: 256 */
+} xlog_rec_ext_header_t;
+
+/*
+ * Quite misnamed, because this union lays out the actual on-disk log buffer.
+ */
+typedef union xlog_in_core2 {
+	xlog_rec_header_t	hic_header;
+	xlog_rec_ext_header_t	hic_xheader;
+	char			hic_sector[XLOG_HEADER_SIZE];
+} xlog_in_core_2_t;
+
+/* not an on-disk structure, but needed by log recovery in userspace */
+typedef struct xfs_log_iovec {
+	void		*i_addr;	/* beginning address of region */
+	int		i_len;		/* length in bytes of region */
+	uint		i_type;		/* type of region */
+} xfs_log_iovec_t;
+
+
+/*
+ * Transaction Header definitions.
+ *
+ * This is the structure written in the log at the head of every transaction. It
+ * identifies the type and id of the transaction, and contains the number of
+ * items logged by the transaction so we know how many to expect during
+ * recovery.
+ *
+ * Do not change the below structure without redoing the code in
+ * xlog_recover_add_to_trans() and xlog_recover_add_to_cont_trans().
+ */
+typedef struct xfs_trans_header {
+	uint		th_magic;		/* magic number */
+	uint		th_type;		/* transaction type */
+	int32_t		th_tid;			/* transaction id (unused) */
+	uint		th_num_items;		/* num items logged by trans */
+} xfs_trans_header_t;
+
+#define	XFS_TRANS_HEADER_MAGIC	0x5452414e	/* TRAN */
+
+/*
+ * The only type valid for th_type in CIL-enabled file system logs:
+ */
+#define XFS_TRANS_CHECKPOINT	40
+
+/*
+ * Log item types.
+ */
+#define	XFS_LI_EFI		0x1236
+#define	XFS_LI_EFD		0x1237
+#define	XFS_LI_IUNLINK		0x1238
+#define	XFS_LI_INODE		0x123b	/* aligned ino chunks, var-size ibufs */
+#define	XFS_LI_BUF		0x123c	/* v2 bufs, variable sized inode bufs */
+#define	XFS_LI_DQUOT		0x123d
+#define	XFS_LI_QUOTAOFF		0x123e
+#define	XFS_LI_ICREATE		0x123f
+#define	XFS_LI_RUI		0x1240	/* rmap update intent */
+#define	XFS_LI_RUD		0x1241
+#define	XFS_LI_CUI		0x1242	/* refcount update intent */
+#define	XFS_LI_CUD		0x1243
+#define	XFS_LI_BUI		0x1244	/* bmbt update intent */
+#define	XFS_LI_BUD		0x1245
+#define	XFS_LI_ATTRI		0x1246  /* attr set/remove intent*/
+#define	XFS_LI_ATTRD		0x1247  /* attr set/remove done */
+
+#define XFS_LI_TYPE_DESC \
+	{ XFS_LI_EFI,		"XFS_LI_EFI" }, \
+	{ XFS_LI_EFD,		"XFS_LI_EFD" }, \
+	{ XFS_LI_IUNLINK,	"XFS_LI_IUNLINK" }, \
+	{ XFS_LI_INODE,		"XFS_LI_INODE" }, \
+	{ XFS_LI_BUF,		"XFS_LI_BUF" }, \
+	{ XFS_LI_DQUOT,		"XFS_LI_DQUOT" }, \
+	{ XFS_LI_QUOTAOFF,	"XFS_LI_QUOTAOFF" }, \
+	{ XFS_LI_ICREATE,	"XFS_LI_ICREATE" }, \
+	{ XFS_LI_RUI,		"XFS_LI_RUI" }, \
+	{ XFS_LI_RUD,		"XFS_LI_RUD" }, \
+	{ XFS_LI_CUI,		"XFS_LI_CUI" }, \
+	{ XFS_LI_CUD,		"XFS_LI_CUD" }, \
+	{ XFS_LI_BUI,		"XFS_LI_BUI" }, \
+	{ XFS_LI_BUD,		"XFS_LI_BUD" }, \
+	{ XFS_LI_ATTRI,		"XFS_LI_ATTRI" }, \
+	{ XFS_LI_ATTRD,		"XFS_LI_ATTRD" }
+
+/*
+ * Inode Log Item Format definitions.
+ *
+ * This is the structure used to lay out an inode log item in the
+ * log.  The size of the inline data/extents/b-tree root to be logged
+ * (if any) is indicated in the ilf_dsize field.  Changes to this structure
+ * must be added on to the end.
+ */
+struct xfs_inode_log_format {
+	uint16_t		ilf_type;	/* inode log item type */
+	uint16_t		ilf_size;	/* size of this item */
+	uint32_t		ilf_fields;	/* flags for fields logged */
+	uint16_t		ilf_asize;	/* size of attr d/ext/root */
+	uint16_t		ilf_dsize;	/* size of data/ext/root */
+	uint32_t		ilf_pad;	/* pad for 64 bit boundary */
+	uint64_t		ilf_ino;	/* inode number */
+	union {
+		uint32_t	ilfu_rdev;	/* rdev value for dev inode*/
+		uint8_t		__pad[16];	/* unused */
+	} ilf_u;
+	int64_t			ilf_blkno;	/* blkno of inode buffer */
+	int32_t			ilf_len;	/* len of inode buffer */
+	int32_t			ilf_boffset;	/* off of inode in buffer */
+};
+
+/*
+ * Old 32 bit systems will log in this format without the 64 bit
+ * alignment padding. Recovery will detect this and convert it to the
+ * correct format.
+ */
+struct xfs_inode_log_format_32 {
+	uint16_t		ilf_type;	/* inode log item type */
+	uint16_t		ilf_size;	/* size of this item */
+	uint32_t		ilf_fields;	/* flags for fields logged */
+	uint16_t		ilf_asize;	/* size of attr d/ext/root */
+	uint16_t		ilf_dsize;	/* size of data/ext/root */
+	uint64_t		ilf_ino;	/* inode number */
+	union {
+		uint32_t	ilfu_rdev;	/* rdev value for dev inode*/
+		uint8_t		__pad[16];	/* unused */
+	} ilf_u;
+	int64_t			ilf_blkno;	/* blkno of inode buffer */
+	int32_t			ilf_len;	/* len of inode buffer */
+	int32_t			ilf_boffset;	/* off of inode in buffer */
+} __attribute__((packed));
+
+
+/*
+ * Flags for xfs_trans_log_inode flags field.
+ */
+#define	XFS_ILOG_CORE	0x001	/* log standard inode fields */
+#define	XFS_ILOG_DDATA	0x002	/* log i_df.if_data */
+#define	XFS_ILOG_DEXT	0x004	/* log i_df.if_extents */
+#define	XFS_ILOG_DBROOT	0x008	/* log i_df.i_broot */
+#define	XFS_ILOG_DEV	0x010	/* log the dev field */
+#define	XFS_ILOG_UUID	0x020	/* added long ago, but never used */
+#define	XFS_ILOG_ADATA	0x040	/* log i_af.if_data */
+#define	XFS_ILOG_AEXT	0x080	/* log i_af.if_extents */
+#define	XFS_ILOG_ABROOT	0x100	/* log i_af.i_broot */
+#define XFS_ILOG_DOWNER	0x200	/* change the data fork owner on replay */
+#define XFS_ILOG_AOWNER	0x400	/* change the attr fork owner on replay */
+
+
+/*
+ * The timestamps are dirty, but not necessarily anything else in the inode
+ * core.  Unlike the other fields above this one must never make it to disk
+ * in the ilf_fields of the inode_log_format, but is purely store in-memory in
+ * ili_fields in the inode_log_item.
+ */
+#define XFS_ILOG_TIMESTAMP	0x4000
+
+#define	XFS_ILOG_NONCORE	(XFS_ILOG_DDATA | XFS_ILOG_DEXT | \
+				 XFS_ILOG_DBROOT | XFS_ILOG_DEV | \
+				 XFS_ILOG_ADATA | XFS_ILOG_AEXT | \
+				 XFS_ILOG_ABROOT | XFS_ILOG_DOWNER | \
+				 XFS_ILOG_AOWNER)
+
+#define	XFS_ILOG_DFORK		(XFS_ILOG_DDATA | XFS_ILOG_DEXT | \
+				 XFS_ILOG_DBROOT)
+
+#define	XFS_ILOG_AFORK		(XFS_ILOG_ADATA | XFS_ILOG_AEXT | \
+				 XFS_ILOG_ABROOT)
+
+#define	XFS_ILOG_ALL		(XFS_ILOG_CORE | XFS_ILOG_DDATA | \
+				 XFS_ILOG_DEXT | XFS_ILOG_DBROOT | \
+				 XFS_ILOG_DEV | XFS_ILOG_ADATA | \
+				 XFS_ILOG_AEXT | XFS_ILOG_ABROOT | \
+				 XFS_ILOG_TIMESTAMP | XFS_ILOG_DOWNER | \
+				 XFS_ILOG_AOWNER)
+
+static inline int xfs_ilog_fbroot(int w)
+{
+	return (w == XFS_DATA_FORK ? XFS_ILOG_DBROOT : XFS_ILOG_ABROOT);
+}
+
+static inline int xfs_ilog_fext(int w)
+{
+	return (w == XFS_DATA_FORK ? XFS_ILOG_DEXT : XFS_ILOG_AEXT);
+}
+
+static inline int xfs_ilog_fdata(int w)
+{
+	return (w == XFS_DATA_FORK ? XFS_ILOG_DDATA : XFS_ILOG_ADATA);
+}
+
+/*
+ * Incore version of the on-disk inode core structures. We log this directly
+ * into the journal in host CPU format (for better or worse) and as such
+ * directly mirrors the xfs_dinode structure as it must contain all the same
+ * information.
+ */
+typedef uint64_t xfs_log_timestamp_t;
+
+/* Legacy timestamp encoding format. */
+struct xfs_log_legacy_timestamp {
+	int32_t		t_sec;		/* timestamp seconds */
+	int32_t		t_nsec;		/* timestamp nanoseconds */
+};
+
+/*
+ * Define the format of the inode core that is logged. This structure must be
+ * kept identical to struct xfs_dinode except for the endianness annotations.
+ */
+struct xfs_log_dinode {
+	uint16_t	di_magic;	/* inode magic # = XFS_DINODE_MAGIC */
+	uint16_t	di_mode;	/* mode and type of file */
+	int8_t		di_version;	/* inode version */
+	int8_t		di_format;	/* format of di_c data */
+	uint8_t		di_pad3[2];	/* unused in v2/3 inodes */
+	uint32_t	di_uid;		/* owner's user id */
+	uint32_t	di_gid;		/* owner's group id */
+	uint32_t	di_nlink;	/* number of links to file */
+	uint16_t	di_projid_lo;	/* lower part of owner's project id */
+	uint16_t	di_projid_hi;	/* higher part of owner's project id */
+	union {
+		/* Number of data fork extents if NREXT64 is set */
+		uint64_t	di_big_nextents;
+
+		/* Padding for V3 inodes without NREXT64 set. */
+		uint64_t	di_v3_pad;
+
+		/* Padding and inode flush counter for V2 inodes. */
+		struct {
+			uint8_t	di_v2_pad[6];	/* V2 inode zeroed space */
+			uint16_t di_flushiter;	/* V2 inode incremented on flush */
+		};
+	};
+	xfs_log_timestamp_t di_atime;	/* time last accessed */
+	xfs_log_timestamp_t di_mtime;	/* time last modified */
+	xfs_log_timestamp_t di_ctime;	/* time created/inode modified */
+	xfs_fsize_t	di_size;	/* number of bytes in file */
+	xfs_rfsblock_t	di_nblocks;	/* # of direct & btree blocks used */
+	xfs_extlen_t	di_extsize;	/* basic/minimum extent size for file */
+	union {
+		/*
+		 * For V2 inodes and V3 inodes without NREXT64 set, this
+		 * is the number of data and attr fork extents.
+		 */
+		struct {
+			uint32_t  di_nextents;
+			uint16_t  di_anextents;
+		} __packed;
+
+		/* Number of attr fork extents if NREXT64 is set. */
+		struct {
+			uint32_t  di_big_anextents;
+			uint16_t  di_nrext64_pad;
+		} __packed;
+	} __packed;
+	uint8_t		di_forkoff;	/* attr fork offs, <<3 for 64b align */
+	int8_t		di_aformat;	/* format of attr fork's data */
+	uint32_t	di_dmevmask;	/* DMIG event mask */
+	uint16_t	di_dmstate;	/* DMIG state info */
+	uint16_t	di_flags;	/* random flags, XFS_DIFLAG_... */
+	uint32_t	di_gen;		/* generation number */
+
+	/* di_next_unlinked is the only non-core field in the old dinode */
+	xfs_agino_t	di_next_unlinked;/* agi unlinked list ptr */
+
+	/* start of the extended dinode, writable fields */
+	uint32_t	di_crc;		/* CRC of the inode */
+	uint64_t	di_changecount;	/* number of attribute changes */
+
+	/*
+	 * The LSN we write to this field during formatting is not a reflection
+	 * of the current on-disk LSN. It should never be used for recovery
+	 * sequencing, nor should it be recovered into the on-disk inode at all.
+	 * See xlog_recover_inode_commit_pass2() and xfs_log_dinode_to_disk()
+	 * for details.
+	 */
+	xfs_lsn_t	di_lsn;
+
+	uint64_t	di_flags2;	/* more random flags */
+	uint32_t	di_cowextsize;	/* basic cow extent size for file */
+	uint8_t		di_pad2[12];	/* more padding for future expansion */
+
+	/* fields only written to during inode creation */
+	xfs_log_timestamp_t di_crtime;	/* time created */
+	xfs_ino_t	di_ino;		/* inode number */
+	uuid_t		di_uuid;	/* UUID of the filesystem */
+
+	/* structure must be padded to 64 bit alignment */
+};
+
+#define xfs_log_dinode_size(mp)						\
+	(xfs_has_v3inodes((mp)) ?					\
+		sizeof(struct xfs_log_dinode) :				\
+		offsetof(struct xfs_log_dinode, di_next_unlinked))
+
+/*
+ * Buffer Log Format definitions
+ *
+ * These are the physical dirty bitmap definitions for the log format structure.
+ */
+#define	XFS_BLF_CHUNK		128
+#define	XFS_BLF_SHIFT		7
+#define	BIT_TO_WORD_SHIFT	5
+#define	NBWORD			(NBBY * sizeof(unsigned int))
+
+/*
+ * This flag indicates that the buffer contains on disk inodes
+ * and requires special recovery handling.
+ */
+#define	XFS_BLF_INODE_BUF	(1<<0)
+
+/*
+ * This flag indicates that the buffer should not be replayed
+ * during recovery because its blocks are being freed.
+ */
+#define	XFS_BLF_CANCEL		(1<<1)
+
+/*
+ * This flag indicates that the buffer contains on disk
+ * user or group dquots and may require special recovery handling.
+ */
+#define	XFS_BLF_UDQUOT_BUF	(1<<2)
+#define XFS_BLF_PDQUOT_BUF	(1<<3)
+#define	XFS_BLF_GDQUOT_BUF	(1<<4)
+
+/*
+ * This is the structure used to lay out a buf log item in the log.  The data
+ * map describes which 128 byte chunks of the buffer have been logged.
+ *
+ * The placement of blf_map_size causes blf_data_map to start at an odd
+ * multiple of sizeof(unsigned int) offset within the struct.  Because the data
+ * bitmap size will always be an even number, the end of the data_map (and
+ * therefore the structure) will also be at an odd multiple of sizeof(unsigned
+ * int).  Some 64-bit compilers will insert padding at the end of the struct to
+ * ensure 64-bit alignment of blf_blkno, but 32-bit ones will not.  Therefore,
+ * XFS_BLF_DATAMAP_SIZE must be an odd number to make the padding explicit and
+ * keep the structure size consistent between 32-bit and 64-bit platforms.
+ */
+#define __XFS_BLF_DATAMAP_SIZE	((XFS_MAX_BLOCKSIZE / XFS_BLF_CHUNK) / NBWORD)
+#define XFS_BLF_DATAMAP_SIZE	(__XFS_BLF_DATAMAP_SIZE + 1)
+
+typedef struct xfs_buf_log_format {
+	unsigned short	blf_type;	/* buf log item type indicator */
+	unsigned short	blf_size;	/* size of this item */
+	unsigned short	blf_flags;	/* misc state */
+	unsigned short	blf_len;	/* number of blocks in this buf */
+	int64_t		blf_blkno;	/* starting blkno of this buf */
+	unsigned int	blf_map_size;	/* used size of data bitmap in words */
+	unsigned int	blf_data_map[XFS_BLF_DATAMAP_SIZE]; /* dirty bitmap */
+} xfs_buf_log_format_t;
+
+/*
+ * All buffers now need to tell recovery where the magic number
+ * is so that it can verify and calculate the CRCs on the buffer correctly
+ * once the changes have been replayed into the buffer.
+ *
+ * The type value is held in the upper 5 bits of the blf_flags field, which is
+ * an unsigned 16 bit field. Hence we need to shift it 11 bits up and down.
+ */
+#define XFS_BLFT_BITS	5
+#define XFS_BLFT_SHIFT	11
+#define XFS_BLFT_MASK	(((1 << XFS_BLFT_BITS) - 1) << XFS_BLFT_SHIFT)
+
+enum xfs_blft {
+	XFS_BLFT_UNKNOWN_BUF = 0,
+	XFS_BLFT_UDQUOT_BUF,
+	XFS_BLFT_PDQUOT_BUF,
+	XFS_BLFT_GDQUOT_BUF,
+	XFS_BLFT_BTREE_BUF,
+	XFS_BLFT_AGF_BUF,
+	XFS_BLFT_AGFL_BUF,
+	XFS_BLFT_AGI_BUF,
+	XFS_BLFT_DINO_BUF,
+	XFS_BLFT_SYMLINK_BUF,
+	XFS_BLFT_DIR_BLOCK_BUF,
+	XFS_BLFT_DIR_DATA_BUF,
+	XFS_BLFT_DIR_FREE_BUF,
+	XFS_BLFT_DIR_LEAF1_BUF,
+	XFS_BLFT_DIR_LEAFN_BUF,
+	XFS_BLFT_DA_NODE_BUF,
+	XFS_BLFT_ATTR_LEAF_BUF,
+	XFS_BLFT_ATTR_RMT_BUF,
+	XFS_BLFT_SB_BUF,
+	XFS_BLFT_RTBITMAP_BUF,
+	XFS_BLFT_RTSUMMARY_BUF,
+	XFS_BLFT_MAX_BUF = (1 << XFS_BLFT_BITS),
+};
+
+static inline void
+xfs_blft_to_flags(struct xfs_buf_log_format *blf, enum xfs_blft type)
+{
+	ASSERT(type > XFS_BLFT_UNKNOWN_BUF && type < XFS_BLFT_MAX_BUF);
+	blf->blf_flags &= ~XFS_BLFT_MASK;
+	blf->blf_flags |= ((type << XFS_BLFT_SHIFT) & XFS_BLFT_MASK);
+}
+
+static inline uint16_t
+xfs_blft_from_flags(struct xfs_buf_log_format *blf)
+{
+	return (blf->blf_flags & XFS_BLFT_MASK) >> XFS_BLFT_SHIFT;
+}
+
+/*
+ * EFI/EFD log format definitions
+ */
+typedef struct xfs_extent {
+	xfs_fsblock_t	ext_start;
+	xfs_extlen_t	ext_len;
+} xfs_extent_t;
+
+/*
+ * Since an xfs_extent_t has types (start:64, len: 32)
+ * there are different alignments on 32 bit and 64 bit kernels.
+ * So we provide the different variants for use by a
+ * conversion routine.
+ */
+typedef struct xfs_extent_32 {
+	uint64_t	ext_start;
+	uint32_t	ext_len;
+} __attribute__((packed)) xfs_extent_32_t;
+
+typedef struct xfs_extent_64 {
+	uint64_t	ext_start;
+	uint32_t	ext_len;
+	uint32_t	ext_pad;
+} xfs_extent_64_t;
+
+/*
+ * This is the structure used to lay out an efi log item in the
+ * log.  The efi_extents field is a variable size array whose
+ * size is given by efi_nextents.
+ */
+typedef struct xfs_efi_log_format {
+	uint16_t		efi_type;	/* efi log item type */
+	uint16_t		efi_size;	/* size of this item */
+	uint32_t		efi_nextents;	/* # extents to free */
+	uint64_t		efi_id;		/* efi identifier */
+	xfs_extent_t		efi_extents[];	/* array of extents to free */
+} xfs_efi_log_format_t;
+
+static inline size_t
+xfs_efi_log_format_sizeof(
+	unsigned int		nr)
+{
+	return sizeof(struct xfs_efi_log_format) +
+			nr * sizeof(struct xfs_extent);
+}
+
+typedef struct xfs_efi_log_format_32 {
+	uint16_t		efi_type;	/* efi log item type */
+	uint16_t		efi_size;	/* size of this item */
+	uint32_t		efi_nextents;	/* # extents to free */
+	uint64_t		efi_id;		/* efi identifier */
+	xfs_extent_32_t		efi_extents[];	/* array of extents to free */
+} __attribute__((packed)) xfs_efi_log_format_32_t;
+
+static inline size_t
+xfs_efi_log_format32_sizeof(
+	unsigned int		nr)
+{
+	return sizeof(struct xfs_efi_log_format_32) +
+			nr * sizeof(struct xfs_extent_32);
+}
+
+typedef struct xfs_efi_log_format_64 {
+	uint16_t		efi_type;	/* efi log item type */
+	uint16_t		efi_size;	/* size of this item */
+	uint32_t		efi_nextents;	/* # extents to free */
+	uint64_t		efi_id;		/* efi identifier */
+	xfs_extent_64_t		efi_extents[];	/* array of extents to free */
+} xfs_efi_log_format_64_t;
+
+static inline size_t
+xfs_efi_log_format64_sizeof(
+	unsigned int		nr)
+{
+	return sizeof(struct xfs_efi_log_format_64) +
+			nr * sizeof(struct xfs_extent_64);
+}
+
+/*
+ * This is the structure used to lay out an efd log item in the
+ * log.  The efd_extents array is a variable size array whose
+ * size is given by efd_nextents;
+ */
+typedef struct xfs_efd_log_format {
+	uint16_t		efd_type;	/* efd log item type */
+	uint16_t		efd_size;	/* size of this item */
+	uint32_t		efd_nextents;	/* # of extents freed */
+	uint64_t		efd_efi_id;	/* id of corresponding efi */
+	xfs_extent_t		efd_extents[];	/* array of extents freed */
+} xfs_efd_log_format_t;
+
+static inline size_t
+xfs_efd_log_format_sizeof(
+	unsigned int		nr)
+{
+	return sizeof(struct xfs_efd_log_format) +
+			nr * sizeof(struct xfs_extent);
+}
+
+typedef struct xfs_efd_log_format_32 {
+	uint16_t		efd_type;	/* efd log item type */
+	uint16_t		efd_size;	/* size of this item */
+	uint32_t		efd_nextents;	/* # of extents freed */
+	uint64_t		efd_efi_id;	/* id of corresponding efi */
+	xfs_extent_32_t		efd_extents[];	/* array of extents freed */
+} __attribute__((packed)) xfs_efd_log_format_32_t;
+
+static inline size_t
+xfs_efd_log_format32_sizeof(
+	unsigned int		nr)
+{
+	return sizeof(struct xfs_efd_log_format_32) +
+			nr * sizeof(struct xfs_extent_32);
+}
+
+typedef struct xfs_efd_log_format_64 {
+	uint16_t		efd_type;	/* efd log item type */
+	uint16_t		efd_size;	/* size of this item */
+	uint32_t		efd_nextents;	/* # of extents freed */
+	uint64_t		efd_efi_id;	/* id of corresponding efi */
+	xfs_extent_64_t		efd_extents[];	/* array of extents freed */
+} xfs_efd_log_format_64_t;
+
+static inline size_t
+xfs_efd_log_format64_sizeof(
+	unsigned int		nr)
+{
+	return sizeof(struct xfs_efd_log_format_64) +
+			nr * sizeof(struct xfs_extent_64);
+}
+
+/*
+ * RUI/RUD (reverse mapping) log format definitions
+ */
+struct xfs_map_extent {
+	uint64_t		me_owner;
+	uint64_t		me_startblock;
+	uint64_t		me_startoff;
+	uint32_t		me_len;
+	uint32_t		me_flags;
+};
+
+/* rmap me_flags: upper bits are flags, lower byte is type code */
+#define XFS_RMAP_EXTENT_MAP		1
+#define XFS_RMAP_EXTENT_MAP_SHARED	2
+#define XFS_RMAP_EXTENT_UNMAP		3
+#define XFS_RMAP_EXTENT_UNMAP_SHARED	4
+#define XFS_RMAP_EXTENT_CONVERT		5
+#define XFS_RMAP_EXTENT_CONVERT_SHARED	6
+#define XFS_RMAP_EXTENT_ALLOC		7
+#define XFS_RMAP_EXTENT_FREE		8
+#define XFS_RMAP_EXTENT_TYPE_MASK	0xFF
+
+#define XFS_RMAP_EXTENT_ATTR_FORK	(1U << 31)
+#define XFS_RMAP_EXTENT_BMBT_BLOCK	(1U << 30)
+#define XFS_RMAP_EXTENT_UNWRITTEN	(1U << 29)
+
+#define XFS_RMAP_EXTENT_FLAGS		(XFS_RMAP_EXTENT_TYPE_MASK | \
+					 XFS_RMAP_EXTENT_ATTR_FORK | \
+					 XFS_RMAP_EXTENT_BMBT_BLOCK | \
+					 XFS_RMAP_EXTENT_UNWRITTEN)
+
+/*
+ * This is the structure used to lay out an rui log item in the
+ * log.  The rui_extents field is a variable size array whose
+ * size is given by rui_nextents.
+ */
+struct xfs_rui_log_format {
+	uint16_t		rui_type;	/* rui log item type */
+	uint16_t		rui_size;	/* size of this item */
+	uint32_t		rui_nextents;	/* # extents to free */
+	uint64_t		rui_id;		/* rui identifier */
+	struct xfs_map_extent	rui_extents[];	/* array of extents to rmap */
+};
+
+static inline size_t
+xfs_rui_log_format_sizeof(
+	unsigned int		nr)
+{
+	return sizeof(struct xfs_rui_log_format) +
+			nr * sizeof(struct xfs_map_extent);
+}
+
+/*
+ * This is the structure used to lay out an rud log item in the
+ * log.  The rud_extents array is a variable size array whose
+ * size is given by rud_nextents;
+ */
+struct xfs_rud_log_format {
+	uint16_t		rud_type;	/* rud log item type */
+	uint16_t		rud_size;	/* size of this item */
+	uint32_t		__pad;
+	uint64_t		rud_rui_id;	/* id of corresponding rui */
+};
+
+/*
+ * CUI/CUD (refcount update) log format definitions
+ */
+struct xfs_phys_extent {
+	uint64_t		pe_startblock;
+	uint32_t		pe_len;
+	uint32_t		pe_flags;
+};
+
+/* refcount pe_flags: upper bits are flags, lower byte is type code */
+/* Type codes are taken directly from enum xfs_refcount_intent_type. */
+#define XFS_REFCOUNT_EXTENT_TYPE_MASK	0xFF
+
+#define XFS_REFCOUNT_EXTENT_FLAGS	(XFS_REFCOUNT_EXTENT_TYPE_MASK)
+
+/*
+ * This is the structure used to lay out a cui log item in the
+ * log.  The cui_extents field is a variable size array whose
+ * size is given by cui_nextents.
+ */
+struct xfs_cui_log_format {
+	uint16_t		cui_type;	/* cui log item type */
+	uint16_t		cui_size;	/* size of this item */
+	uint32_t		cui_nextents;	/* # extents to free */
+	uint64_t		cui_id;		/* cui identifier */
+	struct xfs_phys_extent	cui_extents[];	/* array of extents */
+};
+
+static inline size_t
+xfs_cui_log_format_sizeof(
+	unsigned int		nr)
+{
+	return sizeof(struct xfs_cui_log_format) +
+			nr * sizeof(struct xfs_phys_extent);
+}
+
+/*
+ * This is the structure used to lay out a cud log item in the
+ * log.  The cud_extents array is a variable size array whose
+ * size is given by cud_nextents;
+ */
+struct xfs_cud_log_format {
+	uint16_t		cud_type;	/* cud log item type */
+	uint16_t		cud_size;	/* size of this item */
+	uint32_t		__pad;
+	uint64_t		cud_cui_id;	/* id of corresponding cui */
+};
+
+/*
+ * BUI/BUD (inode block mapping) log format definitions
+ */
+
+/* bmbt me_flags: upper bits are flags, lower byte is type code */
+/* Type codes are taken directly from enum xfs_bmap_intent_type. */
+#define XFS_BMAP_EXTENT_TYPE_MASK	0xFF
+
+#define XFS_BMAP_EXTENT_ATTR_FORK	(1U << 31)
+#define XFS_BMAP_EXTENT_UNWRITTEN	(1U << 30)
+
+#define XFS_BMAP_EXTENT_FLAGS		(XFS_BMAP_EXTENT_TYPE_MASK | \
+					 XFS_BMAP_EXTENT_ATTR_FORK | \
+					 XFS_BMAP_EXTENT_UNWRITTEN)
+
+/*
+ * This is the structure used to lay out an bui log item in the
+ * log.  The bui_extents field is a variable size array whose
+ * size is given by bui_nextents.
+ */
+struct xfs_bui_log_format {
+	uint16_t		bui_type;	/* bui log item type */
+	uint16_t		bui_size;	/* size of this item */
+	uint32_t		bui_nextents;	/* # extents to free */
+	uint64_t		bui_id;		/* bui identifier */
+	struct xfs_map_extent	bui_extents[];	/* array of extents to bmap */
+};
+
+static inline size_t
+xfs_bui_log_format_sizeof(
+	unsigned int		nr)
+{
+	return sizeof(struct xfs_bui_log_format) +
+			nr * sizeof(struct xfs_map_extent);
+}
+
+/*
+ * This is the structure used to lay out an bud log item in the
+ * log.  The bud_extents array is a variable size array whose
+ * size is given by bud_nextents;
+ */
+struct xfs_bud_log_format {
+	uint16_t		bud_type;	/* bud log item type */
+	uint16_t		bud_size;	/* size of this item */
+	uint32_t		__pad;
+	uint64_t		bud_bui_id;	/* id of corresponding bui */
+};
+
+/*
+ * Dquot Log format definitions.
+ *
+ * The first two fields must be the type and size fitting into
+ * 32 bits : log_recovery code assumes that.
+ */
+typedef struct xfs_dq_logformat {
+	uint16_t		qlf_type;      /* dquot log item type */
+	uint16_t		qlf_size;      /* size of this item */
+	xfs_dqid_t		qlf_id;	       /* usr/grp/proj id : 32 bits */
+	int64_t			qlf_blkno;     /* blkno of dquot buffer */
+	int32_t			qlf_len;       /* len of dquot buffer */
+	uint32_t		qlf_boffset;   /* off of dquot in buffer */
+} xfs_dq_logformat_t;
+
+/*
+ * log format struct for QUOTAOFF records.
+ * The first two fields must be the type and size fitting into
+ * 32 bits : log_recovery code assumes that.
+ * We write two LI_QUOTAOFF logitems per quotaoff, the last one keeps a pointer
+ * to the first and ensures that the first logitem is taken out of the AIL
+ * only when the last one is securely committed.
+ */
+typedef struct xfs_qoff_logformat {
+	unsigned short		qf_type;	/* quotaoff log item type */
+	unsigned short		qf_size;	/* size of this item */
+	unsigned int		qf_flags;	/* USR and/or GRP */
+	char			qf_pad[12];	/* padding for future */
+} xfs_qoff_logformat_t;
+
+/*
+ * Disk quotas status in m_qflags, and also sb_qflags. 16 bits.
+ */
+#define XFS_UQUOTA_ACCT	0x0001  /* user quota accounting ON */
+#define XFS_UQUOTA_ENFD	0x0002  /* user quota limits enforced */
+#define XFS_UQUOTA_CHKD	0x0004  /* quotacheck run on usr quotas */
+#define XFS_PQUOTA_ACCT	0x0008  /* project quota accounting ON */
+#define XFS_OQUOTA_ENFD	0x0010  /* other (grp/prj) quota limits enforced */
+#define XFS_OQUOTA_CHKD	0x0020  /* quotacheck run on other (grp/prj) quotas */
+#define XFS_GQUOTA_ACCT	0x0040  /* group quota accounting ON */
+
+/*
+ * Conversion to and from the combined OQUOTA flag (if necessary)
+ * is done only in xfs_sb_qflags_to_disk() and xfs_sb_qflags_from_disk()
+ */
+#define XFS_GQUOTA_ENFD	0x0080  /* group quota limits enforced */
+#define XFS_GQUOTA_CHKD	0x0100  /* quotacheck run on group quotas */
+#define XFS_PQUOTA_ENFD	0x0200  /* project quota limits enforced */
+#define XFS_PQUOTA_CHKD	0x0400  /* quotacheck run on project quotas */
+
+#define XFS_ALL_QUOTA_ACCT	\
+		(XFS_UQUOTA_ACCT | XFS_GQUOTA_ACCT | XFS_PQUOTA_ACCT)
+#define XFS_ALL_QUOTA_ENFD	\
+		(XFS_UQUOTA_ENFD | XFS_GQUOTA_ENFD | XFS_PQUOTA_ENFD)
+#define XFS_ALL_QUOTA_CHKD	\
+		(XFS_UQUOTA_CHKD | XFS_GQUOTA_CHKD | XFS_PQUOTA_CHKD)
+
+#define XFS_MOUNT_QUOTA_ALL	(XFS_UQUOTA_ACCT|XFS_UQUOTA_ENFD|\
+				 XFS_UQUOTA_CHKD|XFS_GQUOTA_ACCT|\
+				 XFS_GQUOTA_ENFD|XFS_GQUOTA_CHKD|\
+				 XFS_PQUOTA_ACCT|XFS_PQUOTA_ENFD|\
+				 XFS_PQUOTA_CHKD)
+
+/*
+ * Inode create log item structure
+ *
+ * Log recovery assumes the first two entries are the type and size and they fit
+ * in 32 bits. Also in host order (ugh) so they have to be 32 bit aligned so
+ * decoding can be done correctly.
+ */
+struct xfs_icreate_log {
+	uint16_t	icl_type;	/* type of log format structure */
+	uint16_t	icl_size;	/* size of log format structure */
+	__be32		icl_ag;		/* ag being allocated in */
+	__be32		icl_agbno;	/* start block of inode range */
+	__be32		icl_count;	/* number of inodes to initialise */
+	__be32		icl_isize;	/* size of inodes */
+	__be32		icl_length;	/* length of extent to initialise */
+	__be32		icl_gen;	/* inode generation number to use */
+};
+
+/*
+ * Flags for deferred attribute operations.
+ * Upper bits are flags, lower byte is type code
+ */
+#define XFS_ATTRI_OP_FLAGS_SET		1	/* Set the attribute */
+#define XFS_ATTRI_OP_FLAGS_REMOVE	2	/* Remove the attribute */
+#define XFS_ATTRI_OP_FLAGS_REPLACE	3	/* Replace the attribute */
+#define XFS_ATTRI_OP_FLAGS_TYPE_MASK	0xFF	/* Flags type mask */
+
+/*
+ * alfi_attr_filter captures the state of xfs_da_args.attr_filter, so it should
+ * never have any other bits set.
+ */
+#define XFS_ATTRI_FILTER_MASK		(XFS_ATTR_ROOT | \
+					 XFS_ATTR_SECURE | \
+					 XFS_ATTR_INCOMPLETE)
+
+/*
+ * This is the structure used to lay out an attr log item in the
+ * log.
+ */
+struct xfs_attri_log_format {
+	uint16_t	alfi_type;	/* attri log item type */
+	uint16_t	alfi_size;	/* size of this item */
+	uint32_t	__pad;		/* pad to 64 bit aligned */
+	uint64_t	alfi_id;	/* attri identifier */
+	uint64_t	alfi_ino;	/* the inode for this attr operation */
+	uint32_t	alfi_op_flags;	/* marks the op as a set or remove */
+	uint32_t	alfi_name_len;	/* attr name length */
+	uint32_t	alfi_value_len;	/* attr value length */
+	uint32_t	alfi_attr_filter;/* attr filter flags */
+};
+
+struct xfs_attrd_log_format {
+	uint16_t	alfd_type;	/* attrd log item type */
+	uint16_t	alfd_size;	/* size of this item */
+	uint32_t	__pad;		/* pad to 64 bit aligned */
+	uint64_t	alfd_alf_id;	/* id of corresponding attri */
+};
+
+#endif /* __XFS_LOG_FORMAT_H__ */
diff --git a/fs/xfs/libxfs/xfs_log_recover.h b/fs/xfs/libxfs/xfs_log_recover.h
new file mode 100644
index 000000000..2420865f3
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_log_recover.h
@@ -0,0 +1,134 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef	__XFS_LOG_RECOVER_H__
+#define __XFS_LOG_RECOVER_H__
+
+/*
+ * Each log item type (XFS_LI_*) gets its own xlog_recover_item_ops to
+ * define how recovery should work for that type of log item.
+ */
+struct xlog_recover_item;
+
+/* Sorting hat for log items as they're read in. */
+enum xlog_recover_reorder {
+	XLOG_REORDER_BUFFER_LIST,
+	XLOG_REORDER_ITEM_LIST,
+	XLOG_REORDER_INODE_BUFFER_LIST,
+	XLOG_REORDER_CANCEL_LIST,
+};
+
+struct xlog_recover_item_ops {
+	uint16_t	item_type;	/* XFS_LI_* type code. */
+
+	/*
+	 * Help sort recovered log items into the order required to replay them
+	 * correctly.  Log item types that always use XLOG_REORDER_ITEM_LIST do
+	 * not have to supply a function here.  See the comment preceding
+	 * xlog_recover_reorder_trans for more details about what the return
+	 * values mean.
+	 */
+	enum xlog_recover_reorder (*reorder)(struct xlog_recover_item *item);
+
+	/* Start readahead for pass2, if provided. */
+	void (*ra_pass2)(struct xlog *log, struct xlog_recover_item *item);
+
+	/* Do whatever work we need to do for pass1, if provided. */
+	int (*commit_pass1)(struct xlog *log, struct xlog_recover_item *item);
+
+	/*
+	 * This function should do whatever work is needed for pass2 of log
+	 * recovery, if provided.
+	 *
+	 * If the recovered item is an intent item, this function should parse
+	 * the recovered item to construct an in-core log intent item and
+	 * insert it into the AIL.  The in-core log intent item should have 1
+	 * refcount so that the item is freed either (a) when we commit the
+	 * recovered log item for the intent-done item; (b) replay the work and
+	 * log a new intent-done item; or (c) recovery fails and we have to
+	 * abort.
+	 *
+	 * If the recovered item is an intent-done item, this function should
+	 * parse the recovered item to find the id of the corresponding intent
+	 * log item.  Next, it should find the in-core log intent item in the
+	 * AIL and release it.
+	 */
+	int (*commit_pass2)(struct xlog *log, struct list_head *buffer_list,
+			    struct xlog_recover_item *item, xfs_lsn_t lsn);
+};
+
+extern const struct xlog_recover_item_ops xlog_icreate_item_ops;
+extern const struct xlog_recover_item_ops xlog_buf_item_ops;
+extern const struct xlog_recover_item_ops xlog_inode_item_ops;
+extern const struct xlog_recover_item_ops xlog_dquot_item_ops;
+extern const struct xlog_recover_item_ops xlog_quotaoff_item_ops;
+extern const struct xlog_recover_item_ops xlog_bui_item_ops;
+extern const struct xlog_recover_item_ops xlog_bud_item_ops;
+extern const struct xlog_recover_item_ops xlog_efi_item_ops;
+extern const struct xlog_recover_item_ops xlog_efd_item_ops;
+extern const struct xlog_recover_item_ops xlog_rui_item_ops;
+extern const struct xlog_recover_item_ops xlog_rud_item_ops;
+extern const struct xlog_recover_item_ops xlog_cui_item_ops;
+extern const struct xlog_recover_item_ops xlog_cud_item_ops;
+extern const struct xlog_recover_item_ops xlog_attri_item_ops;
+extern const struct xlog_recover_item_ops xlog_attrd_item_ops;
+
+/*
+ * Macros, structures, prototypes for internal log manager use.
+ */
+
+#define XLOG_RHASH_BITS  4
+#define XLOG_RHASH_SIZE	16
+#define XLOG_RHASH_SHIFT 2
+#define XLOG_RHASH(tid)	\
+	((((uint32_t)tid)>>XLOG_RHASH_SHIFT) & (XLOG_RHASH_SIZE-1))
+
+#define XLOG_MAX_REGIONS_IN_ITEM   (XFS_MAX_BLOCKSIZE / XFS_BLF_CHUNK / 2 + 1)
+
+
+/*
+ * item headers are in ri_buf[0].  Additional buffers follow.
+ */
+struct xlog_recover_item {
+	struct list_head	ri_list;
+	int			ri_cnt;	/* count of regions found */
+	int			ri_total;	/* total regions */
+	struct xfs_log_iovec	*ri_buf;	/* ptr to regions buffer */
+	const struct xlog_recover_item_ops *ri_ops;
+};
+
+struct xlog_recover {
+	struct hlist_node	r_list;
+	xlog_tid_t		r_log_tid;	/* log's transaction id */
+	xfs_trans_header_t	r_theader;	/* trans header for partial */
+	int			r_state;	/* not needed */
+	xfs_lsn_t		r_lsn;		/* xact lsn */
+	struct list_head	r_itemq;	/* q for items */
+};
+
+#define ITEM_TYPE(i)	(*(unsigned short *)(i)->ri_buf[0].i_addr)
+
+#define	XLOG_RECOVER_CRCPASS	0
+#define	XLOG_RECOVER_PASS1	1
+#define	XLOG_RECOVER_PASS2	2
+
+void xlog_buf_readahead(struct xlog *log, xfs_daddr_t blkno, uint len,
+		const struct xfs_buf_ops *ops);
+bool xlog_is_buffer_cancelled(struct xlog *log, xfs_daddr_t blkno, uint len);
+
+int xlog_recover_iget(struct xfs_mount *mp, xfs_ino_t ino,
+		struct xfs_inode **ipp);
+void xlog_recover_release_intent(struct xlog *log, unsigned short intent_type,
+		uint64_t intent_id);
+int xlog_alloc_buf_cancel_table(struct xlog *log);
+void xlog_free_buf_cancel_table(struct xlog *log);
+
+#ifdef DEBUG
+void xlog_check_buf_cancel_table(struct xlog *log);
+#else
+#define xlog_check_buf_cancel_table(log) do { } while (0)
+#endif
+
+#endif	/* __XFS_LOG_RECOVER_H__ */
diff --git a/fs/xfs/libxfs/xfs_log_rlimit.c b/fs/xfs/libxfs/xfs_log_rlimit.c
new file mode 100644
index 000000000..9975b93a7
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_log_rlimit.c
@@ -0,0 +1,201 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2013 Jie Liu.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_da_format.h"
+#include "xfs_trans_space.h"
+#include "xfs_da_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_trace.h"
+
+/*
+ * Calculate the maximum length in bytes that would be required for a local
+ * attribute value as large attributes out of line are not logged.
+ */
+STATIC int
+xfs_log_calc_max_attrsetm_res(
+	struct xfs_mount	*mp)
+{
+	int			size;
+	int			nblks;
+
+	size = xfs_attr_leaf_entsize_local_max(mp->m_attr_geo->blksize) -
+	       MAXNAMELEN - 1;
+	nblks = XFS_DAENTER_SPACE_RES(mp, XFS_ATTR_FORK);
+	nblks += XFS_B_TO_FSB(mp, size);
+	nblks += XFS_NEXTENTADD_SPACE_RES(mp, size, XFS_ATTR_FORK);
+
+	return  M_RES(mp)->tr_attrsetm.tr_logres +
+		M_RES(mp)->tr_attrsetrt.tr_logres * nblks;
+}
+
+/*
+ * Compute an alternate set of log reservation sizes for use exclusively with
+ * minimum log size calculations.
+ */
+static void
+xfs_log_calc_trans_resv_for_minlogblocks(
+	struct xfs_mount	*mp,
+	struct xfs_trans_resv	*resv)
+{
+	unsigned int		rmap_maxlevels = mp->m_rmap_maxlevels;
+
+	/*
+	 * In the early days of rmap+reflink, we always set the rmap maxlevels
+	 * to 9 even if the AG was small enough that it would never grow to
+	 * that height.  Transaction reservation sizes influence the minimum
+	 * log size calculation, which influences the size of the log that mkfs
+	 * creates.  Use the old value here to ensure that newly formatted
+	 * small filesystems will mount on older kernels.
+	 */
+	if (xfs_has_rmapbt(mp) && xfs_has_reflink(mp))
+		mp->m_rmap_maxlevels = XFS_OLD_REFLINK_RMAP_MAXLEVELS;
+
+	xfs_trans_resv_calc(mp, resv);
+
+	if (xfs_has_reflink(mp)) {
+		/*
+		 * In the early days of reflink, typical log operation counts
+		 * were greatly overestimated.
+		 */
+		resv->tr_write.tr_logcount = XFS_WRITE_LOG_COUNT_REFLINK;
+		resv->tr_itruncate.tr_logcount =
+				XFS_ITRUNCATE_LOG_COUNT_REFLINK;
+		resv->tr_qm_dqalloc.tr_logcount = XFS_WRITE_LOG_COUNT_REFLINK;
+	} else if (xfs_has_rmapbt(mp)) {
+		/*
+		 * In the early days of non-reflink rmap, the impact of rmapbt
+		 * updates on log counts were not taken into account at all.
+		 */
+		resv->tr_write.tr_logcount = XFS_WRITE_LOG_COUNT;
+		resv->tr_itruncate.tr_logcount = XFS_ITRUNCATE_LOG_COUNT;
+		resv->tr_qm_dqalloc.tr_logcount = XFS_WRITE_LOG_COUNT;
+	}
+
+	/*
+	 * In the early days of reflink, we did not use deferred refcount
+	 * update log items, so log reservations must be recomputed using the
+	 * old calculations.
+	 */
+	resv->tr_write.tr_logres =
+			xfs_calc_write_reservation_minlogsize(mp);
+	resv->tr_itruncate.tr_logres =
+			xfs_calc_itruncate_reservation_minlogsize(mp);
+	resv->tr_qm_dqalloc.tr_logres =
+			xfs_calc_qm_dqalloc_reservation_minlogsize(mp);
+
+	/* Put everything back the way it was.  This goes at the end. */
+	mp->m_rmap_maxlevels = rmap_maxlevels;
+}
+
+/*
+ * Iterate over the log space reservation table to figure out and return
+ * the maximum one in terms of the pre-calculated values which were done
+ * at mount time.
+ */
+void
+xfs_log_get_max_trans_res(
+	struct xfs_mount	*mp,
+	struct xfs_trans_res	*max_resp)
+{
+	struct xfs_trans_resv	resv = {};
+	struct xfs_trans_res	*resp;
+	struct xfs_trans_res	*end_resp;
+	unsigned int		i;
+	int			log_space = 0;
+	int			attr_space;
+
+	attr_space = xfs_log_calc_max_attrsetm_res(mp);
+
+	xfs_log_calc_trans_resv_for_minlogblocks(mp, &resv);
+
+	resp = (struct xfs_trans_res *)&resv;
+	end_resp = (struct xfs_trans_res *)(&resv + 1);
+	for (i = 0; resp < end_resp; i++, resp++) {
+		int		tmp = resp->tr_logcount > 1 ?
+				      resp->tr_logres * resp->tr_logcount :
+				      resp->tr_logres;
+
+		trace_xfs_trans_resv_calc_minlogsize(mp, i, resp);
+		if (log_space < tmp) {
+			log_space = tmp;
+			*max_resp = *resp;		/* struct copy */
+		}
+	}
+
+	if (attr_space > log_space) {
+		*max_resp = resv.tr_attrsetm;	/* struct copy */
+		max_resp->tr_logres = attr_space;
+	}
+	trace_xfs_log_get_max_trans_res(mp, max_resp);
+}
+
+/*
+ * Calculate the minimum valid log size for the given superblock configuration.
+ * Used to calculate the minimum log size at mkfs time, and to determine if
+ * the log is large enough or not at mount time. Returns the minimum size in
+ * filesystem block size units.
+ */
+int
+xfs_log_calc_minimum_size(
+	struct xfs_mount	*mp)
+{
+	struct xfs_trans_res	tres = {0};
+	int			max_logres;
+	int			min_logblks = 0;
+	int			lsunit = 0;
+
+	xfs_log_get_max_trans_res(mp, &tres);
+
+	max_logres = xfs_log_calc_unit_res(mp, tres.tr_logres);
+	if (tres.tr_logcount > 1)
+		max_logres *= tres.tr_logcount;
+
+	if (xfs_has_logv2(mp) && mp->m_sb.sb_logsunit > 1)
+		lsunit = BTOBB(mp->m_sb.sb_logsunit);
+
+	/*
+	 * Two factors should be taken into account for calculating the minimum
+	 * log space.
+	 * 1) The fundamental limitation is that no single transaction can be
+	 *    larger than half size of the log.
+	 *
+	 *    From mkfs.xfs, this is considered by the XFS_MIN_LOG_FACTOR
+	 *    define, which is set to 3. That means we can definitely fit
+	 *    maximally sized 2 transactions in the log. We'll use this same
+	 *    value here.
+	 *
+	 * 2) If the lsunit option is specified, a transaction requires 2 LSU
+	 *    for the reservation because there are two log writes that can
+	 *    require padding - the transaction data and the commit record which
+	 *    are written separately and both can require padding to the LSU.
+	 *    Consider that we can have an active CIL reservation holding 2*LSU,
+	 *    but the CIL is not over a push threshold, in this case, if we
+	 *    don't have enough log space for at one new transaction, which
+	 *    includes another 2*LSU in the reservation, we will run into dead
+	 *    loop situation in log space grant procedure. i.e.
+	 *    xlog_grant_head_wait().
+	 *
+	 *    Hence the log size needs to be able to contain two maximally sized
+	 *    and padded transactions, which is (2 * (2 * LSU + maxlres)).
+	 *
+	 * Also, the log size should be a multiple of the log stripe unit, round
+	 * it up to lsunit boundary if lsunit is specified.
+	 */
+	if (lsunit) {
+		min_logblks = roundup_64(BTOBB(max_logres), lsunit) +
+			      2 * lsunit;
+	} else
+		min_logblks = BTOBB(max_logres) + 2 * BBSIZE;
+	min_logblks *= XFS_MIN_LOG_FACTOR;
+
+	return XFS_BB_TO_FSB(mp, min_logblks);
+}
diff --git a/fs/xfs/libxfs/xfs_quota_defs.h b/fs/xfs/libxfs/xfs_quota_defs.h
new file mode 100644
index 000000000..cb035da3f
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_quota_defs.h
@@ -0,0 +1,146 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_QUOTA_DEFS_H__
+#define __XFS_QUOTA_DEFS_H__
+
+/*
+ * Quota definitions shared between user and kernel source trees.
+ */
+
+/*
+ * Even though users may not have quota limits occupying all 64-bits,
+ * they may need 64-bit accounting. Hence, 64-bit quota-counters,
+ * and quota-limits. This is a waste in the common case, but hey ...
+ */
+typedef uint64_t	xfs_qcnt_t;
+
+typedef uint8_t		xfs_dqtype_t;
+
+#define XFS_DQTYPE_STRINGS \
+	{ XFS_DQTYPE_USER,	"USER" }, \
+	{ XFS_DQTYPE_PROJ,	"PROJ" }, \
+	{ XFS_DQTYPE_GROUP,	"GROUP" }, \
+	{ XFS_DQTYPE_BIGTIME,	"BIGTIME" }
+
+/*
+ * flags for q_flags field in the dquot.
+ */
+#define XFS_DQFLAG_DIRTY	(1u << 0)	/* dquot is dirty */
+#define XFS_DQFLAG_FREEING	(1u << 1)	/* dquot is being torn down */
+
+#define XFS_DQFLAG_STRINGS \
+	{ XFS_DQFLAG_DIRTY,	"DIRTY" }, \
+	{ XFS_DQFLAG_FREEING,	"FREEING" }
+
+/*
+ * We have the possibility of all three quota types being active at once, and
+ * hence free space modification requires modification of all three current
+ * dquots in a single transaction. For this case we need to have a reservation
+ * of at least 3 dquots.
+ *
+ * However, a chmod operation can change both UID and GID in a single
+ * transaction, resulting in requiring {old, new} x {uid, gid} dquots to be
+ * modified. Hence for this case we need to reserve space for at least 4 dquots.
+ *
+ * And in the worst case, there's a rename operation that can be modifying up to
+ * 4 inodes with dquots attached to them. In reality, the only inodes that can
+ * have their dquots modified are the source and destination directory inodes
+ * due to directory name creation and removal. That can require space allocation
+ * and/or freeing on both directory inodes, and hence all three dquots on each
+ * inode can be modified. And if the directories are world writeable, all the
+ * dquots can be unique and so 6 dquots can be modified....
+ *
+ * And, of course, we also need to take into account the dquot log format item
+ * used to describe each dquot.
+ */
+#define XFS_DQUOT_LOGRES(mp)	\
+	((sizeof(struct xfs_dq_logformat) + sizeof(struct xfs_disk_dquot)) * 6)
+
+#define XFS_IS_QUOTA_ON(mp)		((mp)->m_qflags & XFS_ALL_QUOTA_ACCT)
+#define XFS_IS_UQUOTA_ON(mp)		((mp)->m_qflags & XFS_UQUOTA_ACCT)
+#define XFS_IS_PQUOTA_ON(mp)		((mp)->m_qflags & XFS_PQUOTA_ACCT)
+#define XFS_IS_GQUOTA_ON(mp)		((mp)->m_qflags & XFS_GQUOTA_ACCT)
+#define XFS_IS_UQUOTA_ENFORCED(mp)	((mp)->m_qflags & XFS_UQUOTA_ENFD)
+#define XFS_IS_GQUOTA_ENFORCED(mp)	((mp)->m_qflags & XFS_GQUOTA_ENFD)
+#define XFS_IS_PQUOTA_ENFORCED(mp)	((mp)->m_qflags & XFS_PQUOTA_ENFD)
+
+/*
+ * Flags to tell various functions what to do. Not all of these are meaningful
+ * to a single function. None of these XFS_QMOPT_* flags are meant to have
+ * persistent values (ie. their values can and will change between versions)
+ */
+#define XFS_QMOPT_UQUOTA	(1u << 0) /* user dquot requested */
+#define XFS_QMOPT_GQUOTA	(1u << 1) /* group dquot requested */
+#define XFS_QMOPT_PQUOTA	(1u << 2) /* project dquot requested */
+#define XFS_QMOPT_FORCE_RES	(1u << 3) /* ignore quota limits */
+#define XFS_QMOPT_SBVERSION	(1u << 4) /* change superblock version num */
+
+/*
+ * flags to xfs_trans_mod_dquot to indicate which field needs to be
+ * modified.
+ */
+#define XFS_QMOPT_RES_REGBLKS	(1u << 7)
+#define XFS_QMOPT_RES_RTBLKS	(1u << 8)
+#define XFS_QMOPT_BCOUNT	(1u << 9)
+#define XFS_QMOPT_ICOUNT	(1u << 10)
+#define XFS_QMOPT_RTBCOUNT	(1u << 11)
+#define XFS_QMOPT_DELBCOUNT	(1u << 12)
+#define XFS_QMOPT_DELRTBCOUNT	(1u << 13)
+#define XFS_QMOPT_RES_INOS	(1u << 14)
+
+/*
+ * flags for dqalloc.
+ */
+#define XFS_QMOPT_INHERIT	(1u << 31)
+
+#define XFS_QMOPT_FLAGS \
+	{ XFS_QMOPT_UQUOTA,		"UQUOTA" }, \
+	{ XFS_QMOPT_PQUOTA,		"PQUOTA" }, \
+	{ XFS_QMOPT_FORCE_RES,		"FORCE_RES" }, \
+	{ XFS_QMOPT_SBVERSION,		"SBVERSION" }, \
+	{ XFS_QMOPT_GQUOTA,		"GQUOTA" }, \
+	{ XFS_QMOPT_INHERIT,		"INHERIT" }, \
+	{ XFS_QMOPT_RES_REGBLKS,	"RES_REGBLKS" }, \
+	{ XFS_QMOPT_RES_RTBLKS,		"RES_RTBLKS" }, \
+	{ XFS_QMOPT_BCOUNT,		"BCOUNT" }, \
+	{ XFS_QMOPT_ICOUNT,		"ICOUNT" }, \
+	{ XFS_QMOPT_RTBCOUNT,		"RTBCOUNT" }, \
+	{ XFS_QMOPT_DELBCOUNT,		"DELBCOUNT" }, \
+	{ XFS_QMOPT_DELRTBCOUNT,	"DELRTBCOUNT" }, \
+	{ XFS_QMOPT_RES_INOS,		"RES_INOS" }
+
+/*
+ * flags to xfs_trans_mod_dquot.
+ */
+#define XFS_TRANS_DQ_RES_BLKS	XFS_QMOPT_RES_REGBLKS
+#define XFS_TRANS_DQ_RES_RTBLKS	XFS_QMOPT_RES_RTBLKS
+#define XFS_TRANS_DQ_RES_INOS	XFS_QMOPT_RES_INOS
+#define XFS_TRANS_DQ_BCOUNT	XFS_QMOPT_BCOUNT
+#define XFS_TRANS_DQ_DELBCOUNT	XFS_QMOPT_DELBCOUNT
+#define XFS_TRANS_DQ_ICOUNT	XFS_QMOPT_ICOUNT
+#define XFS_TRANS_DQ_RTBCOUNT	XFS_QMOPT_RTBCOUNT
+#define XFS_TRANS_DQ_DELRTBCOUNT XFS_QMOPT_DELRTBCOUNT
+
+
+#define XFS_QMOPT_QUOTALL	\
+		(XFS_QMOPT_UQUOTA | XFS_QMOPT_PQUOTA | XFS_QMOPT_GQUOTA)
+#define XFS_QMOPT_RESBLK_MASK	(XFS_QMOPT_RES_REGBLKS | XFS_QMOPT_RES_RTBLKS)
+
+
+extern xfs_failaddr_t xfs_dquot_verify(struct xfs_mount *mp,
+		struct xfs_disk_dquot *ddq, xfs_dqid_t id);
+extern xfs_failaddr_t xfs_dqblk_verify(struct xfs_mount *mp,
+		struct xfs_dqblk *dqb, xfs_dqid_t id);
+extern int xfs_calc_dquots_per_chunk(unsigned int nbblks);
+extern void xfs_dqblk_repair(struct xfs_mount *mp, struct xfs_dqblk *dqb,
+		xfs_dqid_t id, xfs_dqtype_t type);
+
+struct xfs_dquot;
+time64_t xfs_dquot_from_disk_ts(struct xfs_disk_dquot *ddq,
+		__be32 dtimer);
+__be32 xfs_dquot_to_disk_ts(struct xfs_dquot *ddq, time64_t timer);
+
+#endif	/* __XFS_QUOTA_H__ */
diff --git a/fs/xfs/libxfs/xfs_refcount.c b/fs/xfs/libxfs/xfs_refcount.c
new file mode 100644
index 000000000..3f34bafe1
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_refcount.c
@@ -0,0 +1,1915 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2016 Oracle.  All Rights Reserved.
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_defer.h"
+#include "xfs_btree.h"
+#include "xfs_bmap.h"
+#include "xfs_refcount_btree.h"
+#include "xfs_alloc.h"
+#include "xfs_errortag.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+#include "xfs_trans.h"
+#include "xfs_bit.h"
+#include "xfs_refcount.h"
+#include "xfs_rmap.h"
+#include "xfs_ag.h"
+
+struct kmem_cache	*xfs_refcount_intent_cache;
+
+/* Allowable refcount adjustment amounts. */
+enum xfs_refc_adjust_op {
+	XFS_REFCOUNT_ADJUST_INCREASE	= 1,
+	XFS_REFCOUNT_ADJUST_DECREASE	= -1,
+	XFS_REFCOUNT_ADJUST_COW_ALLOC	= 0,
+	XFS_REFCOUNT_ADJUST_COW_FREE	= -1,
+};
+
+STATIC int __xfs_refcount_cow_alloc(struct xfs_btree_cur *rcur,
+		xfs_agblock_t agbno, xfs_extlen_t aglen);
+STATIC int __xfs_refcount_cow_free(struct xfs_btree_cur *rcur,
+		xfs_agblock_t agbno, xfs_extlen_t aglen);
+
+/*
+ * Look up the first record less than or equal to [bno, len] in the btree
+ * given by cur.
+ */
+int
+xfs_refcount_lookup_le(
+	struct xfs_btree_cur	*cur,
+	enum xfs_refc_domain	domain,
+	xfs_agblock_t		bno,
+	int			*stat)
+{
+	trace_xfs_refcount_lookup(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+			xfs_refcount_encode_startblock(bno, domain),
+			XFS_LOOKUP_LE);
+	cur->bc_rec.rc.rc_startblock = bno;
+	cur->bc_rec.rc.rc_blockcount = 0;
+	cur->bc_rec.rc.rc_domain = domain;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_LE, stat);
+}
+
+/*
+ * Look up the first record greater than or equal to [bno, len] in the btree
+ * given by cur.
+ */
+int
+xfs_refcount_lookup_ge(
+	struct xfs_btree_cur	*cur,
+	enum xfs_refc_domain	domain,
+	xfs_agblock_t		bno,
+	int			*stat)
+{
+	trace_xfs_refcount_lookup(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+			xfs_refcount_encode_startblock(bno, domain),
+			XFS_LOOKUP_GE);
+	cur->bc_rec.rc.rc_startblock = bno;
+	cur->bc_rec.rc.rc_blockcount = 0;
+	cur->bc_rec.rc.rc_domain = domain;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_GE, stat);
+}
+
+/*
+ * Look up the first record equal to [bno, len] in the btree
+ * given by cur.
+ */
+int
+xfs_refcount_lookup_eq(
+	struct xfs_btree_cur	*cur,
+	enum xfs_refc_domain	domain,
+	xfs_agblock_t		bno,
+	int			*stat)
+{
+	trace_xfs_refcount_lookup(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+			xfs_refcount_encode_startblock(bno, domain),
+			XFS_LOOKUP_LE);
+	cur->bc_rec.rc.rc_startblock = bno;
+	cur->bc_rec.rc.rc_blockcount = 0;
+	cur->bc_rec.rc.rc_domain = domain;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_EQ, stat);
+}
+
+/* Convert on-disk record to in-core format. */
+void
+xfs_refcount_btrec_to_irec(
+	const union xfs_btree_rec	*rec,
+	struct xfs_refcount_irec	*irec)
+{
+	uint32_t			start;
+
+	start = be32_to_cpu(rec->refc.rc_startblock);
+	if (start & XFS_REFC_COWFLAG) {
+		start &= ~XFS_REFC_COWFLAG;
+		irec->rc_domain = XFS_REFC_DOMAIN_COW;
+	} else {
+		irec->rc_domain = XFS_REFC_DOMAIN_SHARED;
+	}
+
+	irec->rc_startblock = start;
+	irec->rc_blockcount = be32_to_cpu(rec->refc.rc_blockcount);
+	irec->rc_refcount = be32_to_cpu(rec->refc.rc_refcount);
+}
+
+/*
+ * Get the data from the pointed-to record.
+ */
+int
+xfs_refcount_get_rec(
+	struct xfs_btree_cur		*cur,
+	struct xfs_refcount_irec	*irec,
+	int				*stat)
+{
+	struct xfs_mount		*mp = cur->bc_mp;
+	struct xfs_perag		*pag = cur->bc_ag.pag;
+	union xfs_btree_rec		*rec;
+	int				error;
+
+	error = xfs_btree_get_rec(cur, &rec, stat);
+	if (error || !*stat)
+		return error;
+
+	xfs_refcount_btrec_to_irec(rec, irec);
+	if (irec->rc_blockcount == 0 || irec->rc_blockcount > MAXREFCEXTLEN)
+		goto out_bad_rec;
+
+	if (!xfs_refcount_check_domain(irec))
+		goto out_bad_rec;
+
+	/* check for valid extent range, including overflow */
+	if (!xfs_verify_agbext(pag, irec->rc_startblock, irec->rc_blockcount))
+		goto out_bad_rec;
+
+	if (irec->rc_refcount == 0 || irec->rc_refcount > MAXREFCOUNT)
+		goto out_bad_rec;
+
+	trace_xfs_refcount_get(cur->bc_mp, pag->pag_agno, irec);
+	return 0;
+
+out_bad_rec:
+	xfs_warn(mp,
+		"Refcount BTree record corruption in AG %d detected!",
+		pag->pag_agno);
+	xfs_warn(mp,
+		"Start block 0x%x, block count 0x%x, references 0x%x",
+		irec->rc_startblock, irec->rc_blockcount, irec->rc_refcount);
+	return -EFSCORRUPTED;
+}
+
+/*
+ * Update the record referred to by cur to the value given
+ * by [bno, len, refcount].
+ * This either works (return 0) or gets an EFSCORRUPTED error.
+ */
+STATIC int
+xfs_refcount_update(
+	struct xfs_btree_cur		*cur,
+	struct xfs_refcount_irec	*irec)
+{
+	union xfs_btree_rec	rec;
+	uint32_t		start;
+	int			error;
+
+	trace_xfs_refcount_update(cur->bc_mp, cur->bc_ag.pag->pag_agno, irec);
+
+	start = xfs_refcount_encode_startblock(irec->rc_startblock,
+			irec->rc_domain);
+	rec.refc.rc_startblock = cpu_to_be32(start);
+	rec.refc.rc_blockcount = cpu_to_be32(irec->rc_blockcount);
+	rec.refc.rc_refcount = cpu_to_be32(irec->rc_refcount);
+
+	error = xfs_btree_update(cur, &rec);
+	if (error)
+		trace_xfs_refcount_update_error(cur->bc_mp,
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+/*
+ * Insert the record referred to by cur to the value given
+ * by [bno, len, refcount].
+ * This either works (return 0) or gets an EFSCORRUPTED error.
+ */
+int
+xfs_refcount_insert(
+	struct xfs_btree_cur		*cur,
+	struct xfs_refcount_irec	*irec,
+	int				*i)
+{
+	int				error;
+
+	trace_xfs_refcount_insert(cur->bc_mp, cur->bc_ag.pag->pag_agno, irec);
+
+	cur->bc_rec.rc.rc_startblock = irec->rc_startblock;
+	cur->bc_rec.rc.rc_blockcount = irec->rc_blockcount;
+	cur->bc_rec.rc.rc_refcount = irec->rc_refcount;
+	cur->bc_rec.rc.rc_domain = irec->rc_domain;
+
+	error = xfs_btree_insert(cur, i);
+	if (error)
+		goto out_error;
+	if (XFS_IS_CORRUPT(cur->bc_mp, *i != 1)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+
+out_error:
+	if (error)
+		trace_xfs_refcount_insert_error(cur->bc_mp,
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+/*
+ * Remove the record referred to by cur, then set the pointer to the spot
+ * where the record could be re-inserted, in case we want to increment or
+ * decrement the cursor.
+ * This either works (return 0) or gets an EFSCORRUPTED error.
+ */
+STATIC int
+xfs_refcount_delete(
+	struct xfs_btree_cur	*cur,
+	int			*i)
+{
+	struct xfs_refcount_irec	irec;
+	int			found_rec;
+	int			error;
+
+	error = xfs_refcount_get_rec(cur, &irec, &found_rec);
+	if (error)
+		goto out_error;
+	if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+	trace_xfs_refcount_delete(cur->bc_mp, cur->bc_ag.pag->pag_agno, &irec);
+	error = xfs_btree_delete(cur, i);
+	if (XFS_IS_CORRUPT(cur->bc_mp, *i != 1)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+	if (error)
+		goto out_error;
+	error = xfs_refcount_lookup_ge(cur, irec.rc_domain, irec.rc_startblock,
+			&found_rec);
+out_error:
+	if (error)
+		trace_xfs_refcount_delete_error(cur->bc_mp,
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+/*
+ * Adjusting the Reference Count
+ *
+ * As stated elsewhere, the reference count btree (refcbt) stores
+ * >1 reference counts for extents of physical blocks.  In this
+ * operation, we're either raising or lowering the reference count of
+ * some subrange stored in the tree:
+ *
+ *      <------ adjustment range ------>
+ * ----+   +---+-----+ +--+--------+---------
+ *  2  |   | 3 |  4  | |17|   55   |   10
+ * ----+   +---+-----+ +--+--------+---------
+ * X axis is physical blocks number;
+ * reference counts are the numbers inside the rectangles
+ *
+ * The first thing we need to do is to ensure that there are no
+ * refcount extents crossing either boundary of the range to be
+ * adjusted.  For any extent that does cross a boundary, split it into
+ * two extents so that we can increment the refcount of one of the
+ * pieces later:
+ *
+ *      <------ adjustment range ------>
+ * ----+   +---+-----+ +--+--------+----+----
+ *  2  |   | 3 |  2  | |17|   55   | 10 | 10
+ * ----+   +---+-----+ +--+--------+----+----
+ *
+ * For this next step, let's assume that all the physical blocks in
+ * the adjustment range are mapped to a file and are therefore in use
+ * at least once.  Therefore, we can infer that any gap in the
+ * refcount tree within the adjustment range represents a physical
+ * extent with refcount == 1:
+ *
+ *      <------ adjustment range ------>
+ * ----+---+---+-----+-+--+--------+----+----
+ *  2  |"1"| 3 |  2  |1|17|   55   | 10 | 10
+ * ----+---+---+-----+-+--+--------+----+----
+ *      ^
+ *
+ * For each extent that falls within the interval range, figure out
+ * which extent is to the left or the right of that extent.  Now we
+ * have a left, current, and right extent.  If the new reference count
+ * of the center extent enables us to merge left, center, and right
+ * into one record covering all three, do so.  If the center extent is
+ * at the left end of the range, abuts the left extent, and its new
+ * reference count matches the left extent's record, then merge them.
+ * If the center extent is at the right end of the range, abuts the
+ * right extent, and the reference counts match, merge those.  In the
+ * example, we can left merge (assuming an increment operation):
+ *
+ *      <------ adjustment range ------>
+ * --------+---+-----+-+--+--------+----+----
+ *    2    | 3 |  2  |1|17|   55   | 10 | 10
+ * --------+---+-----+-+--+--------+----+----
+ *          ^
+ *
+ * For all other extents within the range, adjust the reference count
+ * or delete it if the refcount falls below 2.  If we were
+ * incrementing, the end result looks like this:
+ *
+ *      <------ adjustment range ------>
+ * --------+---+-----+-+--+--------+----+----
+ *    2    | 4 |  3  |2|18|   56   | 11 | 10
+ * --------+---+-----+-+--+--------+----+----
+ *
+ * The result of a decrement operation looks as such:
+ *
+ *      <------ adjustment range ------>
+ * ----+   +---+       +--+--------+----+----
+ *  2  |   | 2 |       |16|   54   |  9 | 10
+ * ----+   +---+       +--+--------+----+----
+ *      DDDD    111111DD
+ *
+ * The blocks marked "D" are freed; the blocks marked "1" are only
+ * referenced once and therefore the record is removed from the
+ * refcount btree.
+ */
+
+/* Next block after this extent. */
+static inline xfs_agblock_t
+xfs_refc_next(
+	struct xfs_refcount_irec	*rc)
+{
+	return rc->rc_startblock + rc->rc_blockcount;
+}
+
+/*
+ * Split a refcount extent that crosses agbno.
+ */
+STATIC int
+xfs_refcount_split_extent(
+	struct xfs_btree_cur		*cur,
+	enum xfs_refc_domain		domain,
+	xfs_agblock_t			agbno,
+	bool				*shape_changed)
+{
+	struct xfs_refcount_irec	rcext, tmp;
+	int				found_rec;
+	int				error;
+
+	*shape_changed = false;
+	error = xfs_refcount_lookup_le(cur, domain, agbno, &found_rec);
+	if (error)
+		goto out_error;
+	if (!found_rec)
+		return 0;
+
+	error = xfs_refcount_get_rec(cur, &rcext, &found_rec);
+	if (error)
+		goto out_error;
+	if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+	if (rcext.rc_domain != domain)
+		return 0;
+	if (rcext.rc_startblock == agbno || xfs_refc_next(&rcext) <= agbno)
+		return 0;
+
+	*shape_changed = true;
+	trace_xfs_refcount_split_extent(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+			&rcext, agbno);
+
+	/* Establish the right extent. */
+	tmp = rcext;
+	tmp.rc_startblock = agbno;
+	tmp.rc_blockcount -= (agbno - rcext.rc_startblock);
+	error = xfs_refcount_update(cur, &tmp);
+	if (error)
+		goto out_error;
+
+	/* Insert the left extent. */
+	tmp = rcext;
+	tmp.rc_blockcount = agbno - rcext.rc_startblock;
+	error = xfs_refcount_insert(cur, &tmp, &found_rec);
+	if (error)
+		goto out_error;
+	if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+	return error;
+
+out_error:
+	trace_xfs_refcount_split_extent_error(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+/*
+ * Merge the left, center, and right extents.
+ */
+STATIC int
+xfs_refcount_merge_center_extents(
+	struct xfs_btree_cur		*cur,
+	struct xfs_refcount_irec	*left,
+	struct xfs_refcount_irec	*center,
+	struct xfs_refcount_irec	*right,
+	unsigned long long		extlen,
+	xfs_extlen_t			*aglen)
+{
+	int				error;
+	int				found_rec;
+
+	trace_xfs_refcount_merge_center_extents(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, left, center, right);
+
+	ASSERT(left->rc_domain == center->rc_domain);
+	ASSERT(right->rc_domain == center->rc_domain);
+
+	/*
+	 * Make sure the center and right extents are not in the btree.
+	 * If the center extent was synthesized, the first delete call
+	 * removes the right extent and we skip the second deletion.
+	 * If center and right were in the btree, then the first delete
+	 * call removes the center and the second one removes the right
+	 * extent.
+	 */
+	error = xfs_refcount_lookup_ge(cur, center->rc_domain,
+			center->rc_startblock, &found_rec);
+	if (error)
+		goto out_error;
+	if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+
+	error = xfs_refcount_delete(cur, &found_rec);
+	if (error)
+		goto out_error;
+	if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+
+	if (center->rc_refcount > 1) {
+		error = xfs_refcount_delete(cur, &found_rec);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+	}
+
+	/* Enlarge the left extent. */
+	error = xfs_refcount_lookup_le(cur, left->rc_domain,
+			left->rc_startblock, &found_rec);
+	if (error)
+		goto out_error;
+	if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+
+	left->rc_blockcount = extlen;
+	error = xfs_refcount_update(cur, left);
+	if (error)
+		goto out_error;
+
+	*aglen = 0;
+	return error;
+
+out_error:
+	trace_xfs_refcount_merge_center_extents_error(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+/*
+ * Merge with the left extent.
+ */
+STATIC int
+xfs_refcount_merge_left_extent(
+	struct xfs_btree_cur		*cur,
+	struct xfs_refcount_irec	*left,
+	struct xfs_refcount_irec	*cleft,
+	xfs_agblock_t			*agbno,
+	xfs_extlen_t			*aglen)
+{
+	int				error;
+	int				found_rec;
+
+	trace_xfs_refcount_merge_left_extent(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, left, cleft);
+
+	ASSERT(left->rc_domain == cleft->rc_domain);
+
+	/* If the extent at agbno (cleft) wasn't synthesized, remove it. */
+	if (cleft->rc_refcount > 1) {
+		error = xfs_refcount_lookup_le(cur, cleft->rc_domain,
+				cleft->rc_startblock, &found_rec);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+
+		error = xfs_refcount_delete(cur, &found_rec);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+	}
+
+	/* Enlarge the left extent. */
+	error = xfs_refcount_lookup_le(cur, left->rc_domain,
+			left->rc_startblock, &found_rec);
+	if (error)
+		goto out_error;
+	if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+
+	left->rc_blockcount += cleft->rc_blockcount;
+	error = xfs_refcount_update(cur, left);
+	if (error)
+		goto out_error;
+
+	*agbno += cleft->rc_blockcount;
+	*aglen -= cleft->rc_blockcount;
+	return error;
+
+out_error:
+	trace_xfs_refcount_merge_left_extent_error(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+/*
+ * Merge with the right extent.
+ */
+STATIC int
+xfs_refcount_merge_right_extent(
+	struct xfs_btree_cur		*cur,
+	struct xfs_refcount_irec	*right,
+	struct xfs_refcount_irec	*cright,
+	xfs_extlen_t			*aglen)
+{
+	int				error;
+	int				found_rec;
+
+	trace_xfs_refcount_merge_right_extent(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, cright, right);
+
+	ASSERT(right->rc_domain == cright->rc_domain);
+
+	/*
+	 * If the extent ending at agbno+aglen (cright) wasn't synthesized,
+	 * remove it.
+	 */
+	if (cright->rc_refcount > 1) {
+		error = xfs_refcount_lookup_le(cur, cright->rc_domain,
+				cright->rc_startblock, &found_rec);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+
+		error = xfs_refcount_delete(cur, &found_rec);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+	}
+
+	/* Enlarge the right extent. */
+	error = xfs_refcount_lookup_le(cur, right->rc_domain,
+			right->rc_startblock, &found_rec);
+	if (error)
+		goto out_error;
+	if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+
+	right->rc_startblock -= cright->rc_blockcount;
+	right->rc_blockcount += cright->rc_blockcount;
+	error = xfs_refcount_update(cur, right);
+	if (error)
+		goto out_error;
+
+	*aglen -= cright->rc_blockcount;
+	return error;
+
+out_error:
+	trace_xfs_refcount_merge_right_extent_error(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+/*
+ * Find the left extent and the one after it (cleft).  This function assumes
+ * that we've already split any extent crossing agbno.
+ */
+STATIC int
+xfs_refcount_find_left_extents(
+	struct xfs_btree_cur		*cur,
+	struct xfs_refcount_irec	*left,
+	struct xfs_refcount_irec	*cleft,
+	enum xfs_refc_domain		domain,
+	xfs_agblock_t			agbno,
+	xfs_extlen_t			aglen)
+{
+	struct xfs_refcount_irec	tmp;
+	int				error;
+	int				found_rec;
+
+	left->rc_startblock = cleft->rc_startblock = NULLAGBLOCK;
+	error = xfs_refcount_lookup_le(cur, domain, agbno - 1, &found_rec);
+	if (error)
+		goto out_error;
+	if (!found_rec)
+		return 0;
+
+	error = xfs_refcount_get_rec(cur, &tmp, &found_rec);
+	if (error)
+		goto out_error;
+	if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+
+	if (tmp.rc_domain != domain)
+		return 0;
+	if (xfs_refc_next(&tmp) != agbno)
+		return 0;
+	/* We have a left extent; retrieve (or invent) the next right one */
+	*left = tmp;
+
+	error = xfs_btree_increment(cur, 0, &found_rec);
+	if (error)
+		goto out_error;
+	if (found_rec) {
+		error = xfs_refcount_get_rec(cur, &tmp, &found_rec);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+
+		if (tmp.rc_domain != domain)
+			goto not_found;
+
+		/* if tmp starts at the end of our range, just use that */
+		if (tmp.rc_startblock == agbno)
+			*cleft = tmp;
+		else {
+			/*
+			 * There's a gap in the refcntbt at the start of the
+			 * range we're interested in (refcount == 1) so
+			 * synthesize the implied extent and pass it back.
+			 * We assume here that the agbno/aglen range was
+			 * passed in from a data fork extent mapping and
+			 * therefore is allocated to exactly one owner.
+			 */
+			cleft->rc_startblock = agbno;
+			cleft->rc_blockcount = min(aglen,
+					tmp.rc_startblock - agbno);
+			cleft->rc_refcount = 1;
+			cleft->rc_domain = domain;
+		}
+	} else {
+not_found:
+		/*
+		 * No extents, so pretend that there's one covering the whole
+		 * range.
+		 */
+		cleft->rc_startblock = agbno;
+		cleft->rc_blockcount = aglen;
+		cleft->rc_refcount = 1;
+		cleft->rc_domain = domain;
+	}
+	trace_xfs_refcount_find_left_extent(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+			left, cleft, agbno);
+	return error;
+
+out_error:
+	trace_xfs_refcount_find_left_extent_error(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+/*
+ * Find the right extent and the one before it (cright).  This function
+ * assumes that we've already split any extents crossing agbno + aglen.
+ */
+STATIC int
+xfs_refcount_find_right_extents(
+	struct xfs_btree_cur		*cur,
+	struct xfs_refcount_irec	*right,
+	struct xfs_refcount_irec	*cright,
+	enum xfs_refc_domain		domain,
+	xfs_agblock_t			agbno,
+	xfs_extlen_t			aglen)
+{
+	struct xfs_refcount_irec	tmp;
+	int				error;
+	int				found_rec;
+
+	right->rc_startblock = cright->rc_startblock = NULLAGBLOCK;
+	error = xfs_refcount_lookup_ge(cur, domain, agbno + aglen, &found_rec);
+	if (error)
+		goto out_error;
+	if (!found_rec)
+		return 0;
+
+	error = xfs_refcount_get_rec(cur, &tmp, &found_rec);
+	if (error)
+		goto out_error;
+	if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+
+	if (tmp.rc_domain != domain)
+		return 0;
+	if (tmp.rc_startblock != agbno + aglen)
+		return 0;
+	/* We have a right extent; retrieve (or invent) the next left one */
+	*right = tmp;
+
+	error = xfs_btree_decrement(cur, 0, &found_rec);
+	if (error)
+		goto out_error;
+	if (found_rec) {
+		error = xfs_refcount_get_rec(cur, &tmp, &found_rec);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+
+		if (tmp.rc_domain != domain)
+			goto not_found;
+
+		/* if tmp ends at the end of our range, just use that */
+		if (xfs_refc_next(&tmp) == agbno + aglen)
+			*cright = tmp;
+		else {
+			/*
+			 * There's a gap in the refcntbt at the end of the
+			 * range we're interested in (refcount == 1) so
+			 * create the implied extent and pass it back.
+			 * We assume here that the agbno/aglen range was
+			 * passed in from a data fork extent mapping and
+			 * therefore is allocated to exactly one owner.
+			 */
+			cright->rc_startblock = max(agbno, xfs_refc_next(&tmp));
+			cright->rc_blockcount = right->rc_startblock -
+					cright->rc_startblock;
+			cright->rc_refcount = 1;
+			cright->rc_domain = domain;
+		}
+	} else {
+not_found:
+		/*
+		 * No extents, so pretend that there's one covering the whole
+		 * range.
+		 */
+		cright->rc_startblock = agbno;
+		cright->rc_blockcount = aglen;
+		cright->rc_refcount = 1;
+		cright->rc_domain = domain;
+	}
+	trace_xfs_refcount_find_right_extent(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+			cright, right, agbno + aglen);
+	return error;
+
+out_error:
+	trace_xfs_refcount_find_right_extent_error(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+/* Is this extent valid? */
+static inline bool
+xfs_refc_valid(
+	struct xfs_refcount_irec	*rc)
+{
+	return rc->rc_startblock != NULLAGBLOCK;
+}
+
+/*
+ * Try to merge with any extents on the boundaries of the adjustment range.
+ */
+STATIC int
+xfs_refcount_merge_extents(
+	struct xfs_btree_cur	*cur,
+	enum xfs_refc_domain	domain,
+	xfs_agblock_t		*agbno,
+	xfs_extlen_t		*aglen,
+	enum xfs_refc_adjust_op adjust,
+	bool			*shape_changed)
+{
+	struct xfs_refcount_irec	left = {0}, cleft = {0};
+	struct xfs_refcount_irec	cright = {0}, right = {0};
+	int				error;
+	unsigned long long		ulen;
+	bool				cequal;
+
+	*shape_changed = false;
+	/*
+	 * Find the extent just below agbno [left], just above agbno [cleft],
+	 * just below (agbno + aglen) [cright], and just above (agbno + aglen)
+	 * [right].
+	 */
+	error = xfs_refcount_find_left_extents(cur, &left, &cleft, domain,
+			*agbno, *aglen);
+	if (error)
+		return error;
+	error = xfs_refcount_find_right_extents(cur, &right, &cright, domain,
+			*agbno, *aglen);
+	if (error)
+		return error;
+
+	/* No left or right extent to merge; exit. */
+	if (!xfs_refc_valid(&left) && !xfs_refc_valid(&right))
+		return 0;
+
+	cequal = (cleft.rc_startblock == cright.rc_startblock) &&
+		 (cleft.rc_blockcount == cright.rc_blockcount);
+
+	/* Try to merge left, cleft, and right.  cleft must == cright. */
+	ulen = (unsigned long long)left.rc_blockcount + cleft.rc_blockcount +
+			right.rc_blockcount;
+	if (xfs_refc_valid(&left) && xfs_refc_valid(&right) &&
+	    xfs_refc_valid(&cleft) && xfs_refc_valid(&cright) && cequal &&
+	    left.rc_refcount == cleft.rc_refcount + adjust &&
+	    right.rc_refcount == cleft.rc_refcount + adjust &&
+	    ulen < MAXREFCEXTLEN) {
+		*shape_changed = true;
+		return xfs_refcount_merge_center_extents(cur, &left, &cleft,
+				&right, ulen, aglen);
+	}
+
+	/* Try to merge left and cleft. */
+	ulen = (unsigned long long)left.rc_blockcount + cleft.rc_blockcount;
+	if (xfs_refc_valid(&left) && xfs_refc_valid(&cleft) &&
+	    left.rc_refcount == cleft.rc_refcount + adjust &&
+	    ulen < MAXREFCEXTLEN) {
+		*shape_changed = true;
+		error = xfs_refcount_merge_left_extent(cur, &left, &cleft,
+				agbno, aglen);
+		if (error)
+			return error;
+
+		/*
+		 * If we just merged left + cleft and cleft == cright,
+		 * we no longer have a cright to merge with right.  We're done.
+		 */
+		if (cequal)
+			return 0;
+	}
+
+	/* Try to merge cright and right. */
+	ulen = (unsigned long long)right.rc_blockcount + cright.rc_blockcount;
+	if (xfs_refc_valid(&right) && xfs_refc_valid(&cright) &&
+	    right.rc_refcount == cright.rc_refcount + adjust &&
+	    ulen < MAXREFCEXTLEN) {
+		*shape_changed = true;
+		return xfs_refcount_merge_right_extent(cur, &right, &cright,
+				aglen);
+	}
+
+	return 0;
+}
+
+/*
+ * XXX: This is a pretty hand-wavy estimate.  The penalty for guessing
+ * true incorrectly is a shutdown FS; the penalty for guessing false
+ * incorrectly is more transaction rolls than might be necessary.
+ * Be conservative here.
+ */
+static bool
+xfs_refcount_still_have_space(
+	struct xfs_btree_cur		*cur)
+{
+	unsigned long			overhead;
+
+	/*
+	 * Worst case estimate: full splits of the free space and rmap btrees
+	 * to handle each of the shape changes to the refcount btree.
+	 */
+	overhead = xfs_allocfree_block_count(cur->bc_mp,
+				cur->bc_ag.refc.shape_changes);
+	overhead += cur->bc_mp->m_refc_maxlevels;
+	overhead *= cur->bc_mp->m_sb.sb_blocksize;
+
+	/*
+	 * Only allow 2 refcount extent updates per transaction if the
+	 * refcount continue update "error" has been injected.
+	 */
+	if (cur->bc_ag.refc.nr_ops > 2 &&
+	    XFS_TEST_ERROR(false, cur->bc_mp,
+			XFS_ERRTAG_REFCOUNT_CONTINUE_UPDATE))
+		return false;
+
+	if (cur->bc_ag.refc.nr_ops == 0)
+		return true;
+	else if (overhead > cur->bc_tp->t_log_res)
+		return false;
+	return  cur->bc_tp->t_log_res - overhead >
+		cur->bc_ag.refc.nr_ops * XFS_REFCOUNT_ITEM_OVERHEAD;
+}
+
+/*
+ * Adjust the refcounts of middle extents.  At this point we should have
+ * split extents that crossed the adjustment range; merged with adjacent
+ * extents; and updated agbno/aglen to reflect the merges.  Therefore,
+ * all we have to do is update the extents inside [agbno, agbno + aglen].
+ */
+STATIC int
+xfs_refcount_adjust_extents(
+	struct xfs_btree_cur	*cur,
+	xfs_agblock_t		*agbno,
+	xfs_extlen_t		*aglen,
+	enum xfs_refc_adjust_op	adj)
+{
+	struct xfs_refcount_irec	ext, tmp;
+	int				error;
+	int				found_rec, found_tmp;
+	xfs_fsblock_t			fsbno;
+
+	/* Merging did all the work already. */
+	if (*aglen == 0)
+		return 0;
+
+	error = xfs_refcount_lookup_ge(cur, XFS_REFC_DOMAIN_SHARED, *agbno,
+			&found_rec);
+	if (error)
+		goto out_error;
+
+	while (*aglen > 0 && xfs_refcount_still_have_space(cur)) {
+		error = xfs_refcount_get_rec(cur, &ext, &found_rec);
+		if (error)
+			goto out_error;
+		if (!found_rec || ext.rc_domain != XFS_REFC_DOMAIN_SHARED) {
+			ext.rc_startblock = cur->bc_mp->m_sb.sb_agblocks;
+			ext.rc_blockcount = 0;
+			ext.rc_refcount = 0;
+			ext.rc_domain = XFS_REFC_DOMAIN_SHARED;
+		}
+
+		/*
+		 * Deal with a hole in the refcount tree; if a file maps to
+		 * these blocks and there's no refcountbt record, pretend that
+		 * there is one with refcount == 1.
+		 */
+		if (ext.rc_startblock != *agbno) {
+			tmp.rc_startblock = *agbno;
+			tmp.rc_blockcount = min(*aglen,
+					ext.rc_startblock - *agbno);
+			tmp.rc_refcount = 1 + adj;
+			tmp.rc_domain = XFS_REFC_DOMAIN_SHARED;
+
+			trace_xfs_refcount_modify_extent(cur->bc_mp,
+					cur->bc_ag.pag->pag_agno, &tmp);
+
+			/*
+			 * Either cover the hole (increment) or
+			 * delete the range (decrement).
+			 */
+			cur->bc_ag.refc.nr_ops++;
+			if (tmp.rc_refcount) {
+				error = xfs_refcount_insert(cur, &tmp,
+						&found_tmp);
+				if (error)
+					goto out_error;
+				if (XFS_IS_CORRUPT(cur->bc_mp,
+						   found_tmp != 1)) {
+					error = -EFSCORRUPTED;
+					goto out_error;
+				}
+			} else {
+				fsbno = XFS_AGB_TO_FSB(cur->bc_mp,
+						cur->bc_ag.pag->pag_agno,
+						tmp.rc_startblock);
+				xfs_free_extent_later(cur->bc_tp, fsbno,
+						  tmp.rc_blockcount, NULL);
+			}
+
+			(*agbno) += tmp.rc_blockcount;
+			(*aglen) -= tmp.rc_blockcount;
+
+			/* Stop if there's nothing left to modify */
+			if (*aglen == 0 || !xfs_refcount_still_have_space(cur))
+				break;
+
+			/* Move the cursor to the start of ext. */
+			error = xfs_refcount_lookup_ge(cur,
+					XFS_REFC_DOMAIN_SHARED, *agbno,
+					&found_rec);
+			if (error)
+				goto out_error;
+		}
+
+		/*
+		 * A previous step trimmed agbno/aglen such that the end of the
+		 * range would not be in the middle of the record.  If this is
+		 * no longer the case, something is seriously wrong with the
+		 * btree.  Make sure we never feed the synthesized record into
+		 * the processing loop below.
+		 */
+		if (XFS_IS_CORRUPT(cur->bc_mp, ext.rc_blockcount == 0) ||
+		    XFS_IS_CORRUPT(cur->bc_mp, ext.rc_blockcount > *aglen)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+
+		/*
+		 * Adjust the reference count and either update the tree
+		 * (incr) or free the blocks (decr).
+		 */
+		if (ext.rc_refcount == MAXREFCOUNT)
+			goto skip;
+		ext.rc_refcount += adj;
+		trace_xfs_refcount_modify_extent(cur->bc_mp,
+				cur->bc_ag.pag->pag_agno, &ext);
+		cur->bc_ag.refc.nr_ops++;
+		if (ext.rc_refcount > 1) {
+			error = xfs_refcount_update(cur, &ext);
+			if (error)
+				goto out_error;
+		} else if (ext.rc_refcount == 1) {
+			error = xfs_refcount_delete(cur, &found_rec);
+			if (error)
+				goto out_error;
+			if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+				error = -EFSCORRUPTED;
+				goto out_error;
+			}
+			goto advloop;
+		} else {
+			fsbno = XFS_AGB_TO_FSB(cur->bc_mp,
+					cur->bc_ag.pag->pag_agno,
+					ext.rc_startblock);
+			xfs_free_extent_later(cur->bc_tp, fsbno,
+					ext.rc_blockcount, NULL);
+		}
+
+skip:
+		error = xfs_btree_increment(cur, 0, &found_rec);
+		if (error)
+			goto out_error;
+
+advloop:
+		(*agbno) += ext.rc_blockcount;
+		(*aglen) -= ext.rc_blockcount;
+	}
+
+	return error;
+out_error:
+	trace_xfs_refcount_modify_extent_error(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+/* Adjust the reference count of a range of AG blocks. */
+STATIC int
+xfs_refcount_adjust(
+	struct xfs_btree_cur	*cur,
+	xfs_agblock_t		agbno,
+	xfs_extlen_t		aglen,
+	xfs_agblock_t		*new_agbno,
+	xfs_extlen_t		*new_aglen,
+	enum xfs_refc_adjust_op	adj)
+{
+	bool			shape_changed;
+	int			shape_changes = 0;
+	int			error;
+
+	*new_agbno = agbno;
+	*new_aglen = aglen;
+	if (adj == XFS_REFCOUNT_ADJUST_INCREASE)
+		trace_xfs_refcount_increase(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+				agbno, aglen);
+	else
+		trace_xfs_refcount_decrease(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+				agbno, aglen);
+
+	/*
+	 * Ensure that no rcextents cross the boundary of the adjustment range.
+	 */
+	error = xfs_refcount_split_extent(cur, XFS_REFC_DOMAIN_SHARED,
+			agbno, &shape_changed);
+	if (error)
+		goto out_error;
+	if (shape_changed)
+		shape_changes++;
+
+	error = xfs_refcount_split_extent(cur, XFS_REFC_DOMAIN_SHARED,
+			agbno + aglen, &shape_changed);
+	if (error)
+		goto out_error;
+	if (shape_changed)
+		shape_changes++;
+
+	/*
+	 * Try to merge with the left or right extents of the range.
+	 */
+	error = xfs_refcount_merge_extents(cur, XFS_REFC_DOMAIN_SHARED,
+			new_agbno, new_aglen, adj, &shape_changed);
+	if (error)
+		goto out_error;
+	if (shape_changed)
+		shape_changes++;
+	if (shape_changes)
+		cur->bc_ag.refc.shape_changes++;
+
+	/* Now that we've taken care of the ends, adjust the middle extents */
+	error = xfs_refcount_adjust_extents(cur, new_agbno, new_aglen, adj);
+	if (error)
+		goto out_error;
+
+	return 0;
+
+out_error:
+	trace_xfs_refcount_adjust_error(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+			error, _RET_IP_);
+	return error;
+}
+
+/* Clean up after calling xfs_refcount_finish_one. */
+void
+xfs_refcount_finish_one_cleanup(
+	struct xfs_trans	*tp,
+	struct xfs_btree_cur	*rcur,
+	int			error)
+{
+	struct xfs_buf		*agbp;
+
+	if (rcur == NULL)
+		return;
+	agbp = rcur->bc_ag.agbp;
+	xfs_btree_del_cursor(rcur, error);
+	if (error)
+		xfs_trans_brelse(tp, agbp);
+}
+
+/*
+ * Set up a continuation a deferred refcount operation by updating the intent.
+ * Checks to make sure we're not going to run off the end of the AG.
+ */
+static inline int
+xfs_refcount_continue_op(
+	struct xfs_btree_cur		*cur,
+	xfs_fsblock_t			startblock,
+	xfs_agblock_t			new_agbno,
+	xfs_extlen_t			new_len,
+	xfs_fsblock_t			*new_fsbno)
+{
+	struct xfs_mount		*mp = cur->bc_mp;
+	struct xfs_perag		*pag = cur->bc_ag.pag;
+
+	if (XFS_IS_CORRUPT(mp, !xfs_verify_agbext(pag, new_agbno, new_len)))
+		return -EFSCORRUPTED;
+
+	*new_fsbno = XFS_AGB_TO_FSB(mp, pag->pag_agno, new_agbno);
+
+	ASSERT(xfs_verify_fsbext(mp, *new_fsbno, new_len));
+	ASSERT(pag->pag_agno == XFS_FSB_TO_AGNO(mp, *new_fsbno));
+
+	return 0;
+}
+
+/*
+ * Process one of the deferred refcount operations.  We pass back the
+ * btree cursor to maintain our lock on the btree between calls.
+ * This saves time and eliminates a buffer deadlock between the
+ * superblock and the AGF because we'll always grab them in the same
+ * order.
+ */
+int
+xfs_refcount_finish_one(
+	struct xfs_trans		*tp,
+	enum xfs_refcount_intent_type	type,
+	xfs_fsblock_t			startblock,
+	xfs_extlen_t			blockcount,
+	xfs_fsblock_t			*new_fsb,
+	xfs_extlen_t			*new_len,
+	struct xfs_btree_cur		**pcur)
+{
+	struct xfs_mount		*mp = tp->t_mountp;
+	struct xfs_btree_cur		*rcur;
+	struct xfs_buf			*agbp = NULL;
+	int				error = 0;
+	xfs_agblock_t			bno;
+	xfs_agblock_t			new_agbno;
+	unsigned long			nr_ops = 0;
+	int				shape_changes = 0;
+	struct xfs_perag		*pag;
+
+	pag = xfs_perag_get(mp, XFS_FSB_TO_AGNO(mp, startblock));
+	bno = XFS_FSB_TO_AGBNO(mp, startblock);
+
+	trace_xfs_refcount_deferred(mp, XFS_FSB_TO_AGNO(mp, startblock),
+			type, XFS_FSB_TO_AGBNO(mp, startblock),
+			blockcount);
+
+	if (XFS_TEST_ERROR(false, mp, XFS_ERRTAG_REFCOUNT_FINISH_ONE)) {
+		error = -EIO;
+		goto out_drop;
+	}
+
+	/*
+	 * If we haven't gotten a cursor or the cursor AG doesn't match
+	 * the startblock, get one now.
+	 */
+	rcur = *pcur;
+	if (rcur != NULL && rcur->bc_ag.pag != pag) {
+		nr_ops = rcur->bc_ag.refc.nr_ops;
+		shape_changes = rcur->bc_ag.refc.shape_changes;
+		xfs_refcount_finish_one_cleanup(tp, rcur, 0);
+		rcur = NULL;
+		*pcur = NULL;
+	}
+	if (rcur == NULL) {
+		error = xfs_alloc_read_agf(pag, tp, XFS_ALLOC_FLAG_FREEING,
+				&agbp);
+		if (error)
+			goto out_drop;
+
+		rcur = xfs_refcountbt_init_cursor(mp, tp, agbp, pag);
+		rcur->bc_ag.refc.nr_ops = nr_ops;
+		rcur->bc_ag.refc.shape_changes = shape_changes;
+	}
+	*pcur = rcur;
+
+	switch (type) {
+	case XFS_REFCOUNT_INCREASE:
+		error = xfs_refcount_adjust(rcur, bno, blockcount, &new_agbno,
+				new_len, XFS_REFCOUNT_ADJUST_INCREASE);
+		if (error)
+			goto out_drop;
+		if (*new_len > 0)
+			error = xfs_refcount_continue_op(rcur, startblock,
+					new_agbno, *new_len, new_fsb);
+		break;
+	case XFS_REFCOUNT_DECREASE:
+		error = xfs_refcount_adjust(rcur, bno, blockcount, &new_agbno,
+				new_len, XFS_REFCOUNT_ADJUST_DECREASE);
+		if (error)
+			goto out_drop;
+		if (*new_len > 0)
+			error = xfs_refcount_continue_op(rcur, startblock,
+					new_agbno, *new_len, new_fsb);
+		break;
+	case XFS_REFCOUNT_ALLOC_COW:
+		*new_fsb = startblock + blockcount;
+		*new_len = 0;
+		error = __xfs_refcount_cow_alloc(rcur, bno, blockcount);
+		break;
+	case XFS_REFCOUNT_FREE_COW:
+		*new_fsb = startblock + blockcount;
+		*new_len = 0;
+		error = __xfs_refcount_cow_free(rcur, bno, blockcount);
+		break;
+	default:
+		ASSERT(0);
+		error = -EFSCORRUPTED;
+	}
+	if (!error && *new_len > 0)
+		trace_xfs_refcount_finish_one_leftover(mp, pag->pag_agno, type,
+				bno, blockcount, new_agbno, *new_len);
+out_drop:
+	xfs_perag_put(pag);
+	return error;
+}
+
+/*
+ * Record a refcount intent for later processing.
+ */
+static void
+__xfs_refcount_add(
+	struct xfs_trans		*tp,
+	enum xfs_refcount_intent_type	type,
+	xfs_fsblock_t			startblock,
+	xfs_extlen_t			blockcount)
+{
+	struct xfs_refcount_intent	*ri;
+
+	trace_xfs_refcount_defer(tp->t_mountp,
+			XFS_FSB_TO_AGNO(tp->t_mountp, startblock),
+			type, XFS_FSB_TO_AGBNO(tp->t_mountp, startblock),
+			blockcount);
+
+	ri = kmem_cache_alloc(xfs_refcount_intent_cache,
+			GFP_NOFS | __GFP_NOFAIL);
+	INIT_LIST_HEAD(&ri->ri_list);
+	ri->ri_type = type;
+	ri->ri_startblock = startblock;
+	ri->ri_blockcount = blockcount;
+
+	xfs_defer_add(tp, XFS_DEFER_OPS_TYPE_REFCOUNT, &ri->ri_list);
+}
+
+/*
+ * Increase the reference count of the blocks backing a file's extent.
+ */
+void
+xfs_refcount_increase_extent(
+	struct xfs_trans		*tp,
+	struct xfs_bmbt_irec		*PREV)
+{
+	if (!xfs_has_reflink(tp->t_mountp))
+		return;
+
+	__xfs_refcount_add(tp, XFS_REFCOUNT_INCREASE, PREV->br_startblock,
+			PREV->br_blockcount);
+}
+
+/*
+ * Decrease the reference count of the blocks backing a file's extent.
+ */
+void
+xfs_refcount_decrease_extent(
+	struct xfs_trans		*tp,
+	struct xfs_bmbt_irec		*PREV)
+{
+	if (!xfs_has_reflink(tp->t_mountp))
+		return;
+
+	__xfs_refcount_add(tp, XFS_REFCOUNT_DECREASE, PREV->br_startblock,
+			PREV->br_blockcount);
+}
+
+/*
+ * Given an AG extent, find the lowest-numbered run of shared blocks
+ * within that range and return the range in fbno/flen.  If
+ * find_end_of_shared is set, return the longest contiguous extent of
+ * shared blocks; if not, just return the first extent we find.  If no
+ * shared blocks are found, fbno and flen will be set to NULLAGBLOCK
+ * and 0, respectively.
+ */
+int
+xfs_refcount_find_shared(
+	struct xfs_btree_cur		*cur,
+	xfs_agblock_t			agbno,
+	xfs_extlen_t			aglen,
+	xfs_agblock_t			*fbno,
+	xfs_extlen_t			*flen,
+	bool				find_end_of_shared)
+{
+	struct xfs_refcount_irec	tmp;
+	int				i;
+	int				have;
+	int				error;
+
+	trace_xfs_refcount_find_shared(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+			agbno, aglen);
+
+	/* By default, skip the whole range */
+	*fbno = NULLAGBLOCK;
+	*flen = 0;
+
+	/* Try to find a refcount extent that crosses the start */
+	error = xfs_refcount_lookup_le(cur, XFS_REFC_DOMAIN_SHARED, agbno,
+			&have);
+	if (error)
+		goto out_error;
+	if (!have) {
+		/* No left extent, look at the next one */
+		error = xfs_btree_increment(cur, 0, &have);
+		if (error)
+			goto out_error;
+		if (!have)
+			goto done;
+	}
+	error = xfs_refcount_get_rec(cur, &tmp, &i);
+	if (error)
+		goto out_error;
+	if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+	if (tmp.rc_domain != XFS_REFC_DOMAIN_SHARED)
+		goto done;
+
+	/* If the extent ends before the start, look at the next one */
+	if (tmp.rc_startblock + tmp.rc_blockcount <= agbno) {
+		error = xfs_btree_increment(cur, 0, &have);
+		if (error)
+			goto out_error;
+		if (!have)
+			goto done;
+		error = xfs_refcount_get_rec(cur, &tmp, &i);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+		if (tmp.rc_domain != XFS_REFC_DOMAIN_SHARED)
+			goto done;
+	}
+
+	/* If the extent starts after the range we want, bail out */
+	if (tmp.rc_startblock >= agbno + aglen)
+		goto done;
+
+	/* We found the start of a shared extent! */
+	if (tmp.rc_startblock < agbno) {
+		tmp.rc_blockcount -= (agbno - tmp.rc_startblock);
+		tmp.rc_startblock = agbno;
+	}
+
+	*fbno = tmp.rc_startblock;
+	*flen = min(tmp.rc_blockcount, agbno + aglen - *fbno);
+	if (!find_end_of_shared)
+		goto done;
+
+	/* Otherwise, find the end of this shared extent */
+	while (*fbno + *flen < agbno + aglen) {
+		error = xfs_btree_increment(cur, 0, &have);
+		if (error)
+			goto out_error;
+		if (!have)
+			break;
+		error = xfs_refcount_get_rec(cur, &tmp, &i);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+		if (tmp.rc_domain != XFS_REFC_DOMAIN_SHARED ||
+		    tmp.rc_startblock >= agbno + aglen ||
+		    tmp.rc_startblock != *fbno + *flen)
+			break;
+		*flen = min(*flen + tmp.rc_blockcount, agbno + aglen - *fbno);
+	}
+
+done:
+	trace_xfs_refcount_find_shared_result(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, *fbno, *flen);
+
+out_error:
+	if (error)
+		trace_xfs_refcount_find_shared_error(cur->bc_mp,
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+/*
+ * Recovering CoW Blocks After a Crash
+ *
+ * Due to the way that the copy on write mechanism works, there's a window of
+ * opportunity in which we can lose track of allocated blocks during a crash.
+ * Because CoW uses delayed allocation in the in-core CoW fork, writeback
+ * causes blocks to be allocated and stored in the CoW fork.  The blocks are
+ * no longer in the free space btree but are not otherwise recorded anywhere
+ * until the write completes and the blocks are mapped into the file.  A crash
+ * in between allocation and remapping results in the replacement blocks being
+ * lost.  This situation is exacerbated by the CoW extent size hint because
+ * allocations can hang around for long time.
+ *
+ * However, there is a place where we can record these allocations before they
+ * become mappings -- the reference count btree.  The btree does not record
+ * extents with refcount == 1, so we can record allocations with a refcount of
+ * 1.  Blocks being used for CoW writeout cannot be shared, so there should be
+ * no conflict with shared block records.  These mappings should be created
+ * when we allocate blocks to the CoW fork and deleted when they're removed
+ * from the CoW fork.
+ *
+ * Minor nit: records for in-progress CoW allocations and records for shared
+ * extents must never be merged, to preserve the property that (except for CoW
+ * allocations) there are no refcount btree entries with refcount == 1.  The
+ * only time this could potentially happen is when unsharing a block that's
+ * adjacent to CoW allocations, so we must be careful to avoid this.
+ *
+ * At mount time we recover lost CoW allocations by searching the refcount
+ * btree for these refcount == 1 mappings.  These represent CoW allocations
+ * that were in progress at the time the filesystem went down, so we can free
+ * them to get the space back.
+ *
+ * This mechanism is superior to creating EFIs for unmapped CoW extents for
+ * several reasons -- first, EFIs pin the tail of the log and would have to be
+ * periodically relogged to avoid filling up the log.  Second, CoW completions
+ * will have to file an EFD and create new EFIs for whatever remains in the
+ * CoW fork; this partially takes care of (1) but extent-size reservations
+ * will have to periodically relog even if there's no writeout in progress.
+ * This can happen if the CoW extent size hint is set, which you really want.
+ * Third, EFIs cannot currently be automatically relogged into newer
+ * transactions to advance the log tail.  Fourth, stuffing the log full of
+ * EFIs places an upper bound on the number of CoW allocations that can be
+ * held filesystem-wide at any given time.  Recording them in the refcount
+ * btree doesn't require us to maintain any state in memory and doesn't pin
+ * the log.
+ */
+/*
+ * Adjust the refcounts of CoW allocations.  These allocations are "magic"
+ * in that they're not referenced anywhere else in the filesystem, so we
+ * stash them in the refcount btree with a refcount of 1 until either file
+ * remapping (or CoW cancellation) happens.
+ */
+STATIC int
+xfs_refcount_adjust_cow_extents(
+	struct xfs_btree_cur	*cur,
+	xfs_agblock_t		agbno,
+	xfs_extlen_t		aglen,
+	enum xfs_refc_adjust_op	adj)
+{
+	struct xfs_refcount_irec	ext, tmp;
+	int				error;
+	int				found_rec, found_tmp;
+
+	if (aglen == 0)
+		return 0;
+
+	/* Find any overlapping refcount records */
+	error = xfs_refcount_lookup_ge(cur, XFS_REFC_DOMAIN_COW, agbno,
+			&found_rec);
+	if (error)
+		goto out_error;
+	error = xfs_refcount_get_rec(cur, &ext, &found_rec);
+	if (error)
+		goto out_error;
+	if (XFS_IS_CORRUPT(cur->bc_mp, found_rec &&
+				ext.rc_domain != XFS_REFC_DOMAIN_COW)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+	if (!found_rec) {
+		ext.rc_startblock = cur->bc_mp->m_sb.sb_agblocks;
+		ext.rc_blockcount = 0;
+		ext.rc_refcount = 0;
+		ext.rc_domain = XFS_REFC_DOMAIN_COW;
+	}
+
+	switch (adj) {
+	case XFS_REFCOUNT_ADJUST_COW_ALLOC:
+		/* Adding a CoW reservation, there should be nothing here. */
+		if (XFS_IS_CORRUPT(cur->bc_mp,
+				   agbno + aglen > ext.rc_startblock)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+
+		tmp.rc_startblock = agbno;
+		tmp.rc_blockcount = aglen;
+		tmp.rc_refcount = 1;
+		tmp.rc_domain = XFS_REFC_DOMAIN_COW;
+
+		trace_xfs_refcount_modify_extent(cur->bc_mp,
+				cur->bc_ag.pag->pag_agno, &tmp);
+
+		error = xfs_refcount_insert(cur, &tmp,
+				&found_tmp);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(cur->bc_mp, found_tmp != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+		break;
+	case XFS_REFCOUNT_ADJUST_COW_FREE:
+		/* Removing a CoW reservation, there should be one extent. */
+		if (XFS_IS_CORRUPT(cur->bc_mp, ext.rc_startblock != agbno)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+		if (XFS_IS_CORRUPT(cur->bc_mp, ext.rc_blockcount != aglen)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+		if (XFS_IS_CORRUPT(cur->bc_mp, ext.rc_refcount != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+
+		ext.rc_refcount = 0;
+		trace_xfs_refcount_modify_extent(cur->bc_mp,
+				cur->bc_ag.pag->pag_agno, &ext);
+		error = xfs_refcount_delete(cur, &found_rec);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(cur->bc_mp, found_rec != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+		break;
+	default:
+		ASSERT(0);
+	}
+
+	return error;
+out_error:
+	trace_xfs_refcount_modify_extent_error(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+/*
+ * Add or remove refcount btree entries for CoW reservations.
+ */
+STATIC int
+xfs_refcount_adjust_cow(
+	struct xfs_btree_cur	*cur,
+	xfs_agblock_t		agbno,
+	xfs_extlen_t		aglen,
+	enum xfs_refc_adjust_op	adj)
+{
+	bool			shape_changed;
+	int			error;
+
+	/*
+	 * Ensure that no rcextents cross the boundary of the adjustment range.
+	 */
+	error = xfs_refcount_split_extent(cur, XFS_REFC_DOMAIN_COW,
+			agbno, &shape_changed);
+	if (error)
+		goto out_error;
+
+	error = xfs_refcount_split_extent(cur, XFS_REFC_DOMAIN_COW,
+			agbno + aglen, &shape_changed);
+	if (error)
+		goto out_error;
+
+	/*
+	 * Try to merge with the left or right extents of the range.
+	 */
+	error = xfs_refcount_merge_extents(cur, XFS_REFC_DOMAIN_COW, &agbno,
+			&aglen, adj, &shape_changed);
+	if (error)
+		goto out_error;
+
+	/* Now that we've taken care of the ends, adjust the middle extents */
+	error = xfs_refcount_adjust_cow_extents(cur, agbno, aglen, adj);
+	if (error)
+		goto out_error;
+
+	return 0;
+
+out_error:
+	trace_xfs_refcount_adjust_cow_error(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+			error, _RET_IP_);
+	return error;
+}
+
+/*
+ * Record a CoW allocation in the refcount btree.
+ */
+STATIC int
+__xfs_refcount_cow_alloc(
+	struct xfs_btree_cur	*rcur,
+	xfs_agblock_t		agbno,
+	xfs_extlen_t		aglen)
+{
+	trace_xfs_refcount_cow_increase(rcur->bc_mp, rcur->bc_ag.pag->pag_agno,
+			agbno, aglen);
+
+	/* Add refcount btree reservation */
+	return xfs_refcount_adjust_cow(rcur, agbno, aglen,
+			XFS_REFCOUNT_ADJUST_COW_ALLOC);
+}
+
+/*
+ * Remove a CoW allocation from the refcount btree.
+ */
+STATIC int
+__xfs_refcount_cow_free(
+	struct xfs_btree_cur	*rcur,
+	xfs_agblock_t		agbno,
+	xfs_extlen_t		aglen)
+{
+	trace_xfs_refcount_cow_decrease(rcur->bc_mp, rcur->bc_ag.pag->pag_agno,
+			agbno, aglen);
+
+	/* Remove refcount btree reservation */
+	return xfs_refcount_adjust_cow(rcur, agbno, aglen,
+			XFS_REFCOUNT_ADJUST_COW_FREE);
+}
+
+/* Record a CoW staging extent in the refcount btree. */
+void
+xfs_refcount_alloc_cow_extent(
+	struct xfs_trans		*tp,
+	xfs_fsblock_t			fsb,
+	xfs_extlen_t			len)
+{
+	struct xfs_mount		*mp = tp->t_mountp;
+
+	if (!xfs_has_reflink(mp))
+		return;
+
+	__xfs_refcount_add(tp, XFS_REFCOUNT_ALLOC_COW, fsb, len);
+
+	/* Add rmap entry */
+	xfs_rmap_alloc_extent(tp, XFS_FSB_TO_AGNO(mp, fsb),
+			XFS_FSB_TO_AGBNO(mp, fsb), len, XFS_RMAP_OWN_COW);
+}
+
+/* Forget a CoW staging event in the refcount btree. */
+void
+xfs_refcount_free_cow_extent(
+	struct xfs_trans		*tp,
+	xfs_fsblock_t			fsb,
+	xfs_extlen_t			len)
+{
+	struct xfs_mount		*mp = tp->t_mountp;
+
+	if (!xfs_has_reflink(mp))
+		return;
+
+	/* Remove rmap entry */
+	xfs_rmap_free_extent(tp, XFS_FSB_TO_AGNO(mp, fsb),
+			XFS_FSB_TO_AGBNO(mp, fsb), len, XFS_RMAP_OWN_COW);
+	__xfs_refcount_add(tp, XFS_REFCOUNT_FREE_COW, fsb, len);
+}
+
+struct xfs_refcount_recovery {
+	struct list_head		rr_list;
+	struct xfs_refcount_irec	rr_rrec;
+};
+
+/* Stuff an extent on the recovery list. */
+STATIC int
+xfs_refcount_recover_extent(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_rec	*rec,
+	void				*priv)
+{
+	struct list_head		*debris = priv;
+	struct xfs_refcount_recovery	*rr;
+
+	if (XFS_IS_CORRUPT(cur->bc_mp,
+			   be32_to_cpu(rec->refc.rc_refcount) != 1))
+		return -EFSCORRUPTED;
+
+	rr = kmalloc(sizeof(struct xfs_refcount_recovery),
+			GFP_KERNEL | __GFP_NOFAIL);
+	INIT_LIST_HEAD(&rr->rr_list);
+	xfs_refcount_btrec_to_irec(rec, &rr->rr_rrec);
+
+	if (XFS_IS_CORRUPT(cur->bc_mp,
+			   rr->rr_rrec.rc_domain != XFS_REFC_DOMAIN_COW)) {
+		kfree(rr);
+		return -EFSCORRUPTED;
+	}
+
+	list_add_tail(&rr->rr_list, debris);
+	return 0;
+}
+
+/* Find and remove leftover CoW reservations. */
+int
+xfs_refcount_recover_cow_leftovers(
+	struct xfs_mount		*mp,
+	struct xfs_perag		*pag)
+{
+	struct xfs_trans		*tp;
+	struct xfs_btree_cur		*cur;
+	struct xfs_buf			*agbp;
+	struct xfs_refcount_recovery	*rr, *n;
+	struct list_head		debris;
+	union xfs_btree_irec		low;
+	union xfs_btree_irec		high;
+	xfs_fsblock_t			fsb;
+	int				error;
+
+	/* reflink filesystems mustn't have AGs larger than 2^31-1 blocks */
+	BUILD_BUG_ON(XFS_MAX_CRC_AG_BLOCKS >= XFS_REFC_COWFLAG);
+	if (mp->m_sb.sb_agblocks > XFS_MAX_CRC_AG_BLOCKS)
+		return -EOPNOTSUPP;
+
+	INIT_LIST_HEAD(&debris);
+
+	/*
+	 * In this first part, we use an empty transaction to gather up
+	 * all the leftover CoW extents so that we can subsequently
+	 * delete them.  The empty transaction is used to avoid
+	 * a buffer lock deadlock if there happens to be a loop in the
+	 * refcountbt because we're allowed to re-grab a buffer that is
+	 * already attached to our transaction.  When we're done
+	 * recording the CoW debris we cancel the (empty) transaction
+	 * and everything goes away cleanly.
+	 */
+	error = xfs_trans_alloc_empty(mp, &tp);
+	if (error)
+		return error;
+
+	error = xfs_alloc_read_agf(pag, tp, 0, &agbp);
+	if (error)
+		goto out_trans;
+	cur = xfs_refcountbt_init_cursor(mp, tp, agbp, pag);
+
+	/* Find all the leftover CoW staging extents. */
+	memset(&low, 0, sizeof(low));
+	memset(&high, 0, sizeof(high));
+	low.rc.rc_domain = high.rc.rc_domain = XFS_REFC_DOMAIN_COW;
+	high.rc.rc_startblock = -1U;
+	error = xfs_btree_query_range(cur, &low, &high,
+			xfs_refcount_recover_extent, &debris);
+	xfs_btree_del_cursor(cur, error);
+	xfs_trans_brelse(tp, agbp);
+	xfs_trans_cancel(tp);
+	if (error)
+		goto out_free;
+
+	/* Now iterate the list to free the leftovers */
+	list_for_each_entry_safe(rr, n, &debris, rr_list) {
+		/* Set up transaction. */
+		error = xfs_trans_alloc(mp, &M_RES(mp)->tr_write, 0, 0, 0, &tp);
+		if (error)
+			goto out_free;
+
+		trace_xfs_refcount_recover_extent(mp, pag->pag_agno,
+				&rr->rr_rrec);
+
+		/* Free the orphan record */
+		fsb = XFS_AGB_TO_FSB(mp, pag->pag_agno,
+				rr->rr_rrec.rc_startblock);
+		xfs_refcount_free_cow_extent(tp, fsb,
+				rr->rr_rrec.rc_blockcount);
+
+		/* Free the block. */
+		xfs_free_extent_later(tp, fsb, rr->rr_rrec.rc_blockcount, NULL);
+
+		error = xfs_trans_commit(tp);
+		if (error)
+			goto out_free;
+
+		list_del(&rr->rr_list);
+		kfree(rr);
+	}
+
+	return error;
+out_trans:
+	xfs_trans_cancel(tp);
+out_free:
+	/* Free the leftover list */
+	list_for_each_entry_safe(rr, n, &debris, rr_list) {
+		list_del(&rr->rr_list);
+		kfree(rr);
+	}
+	return error;
+}
+
+/* Is there a record covering a given extent? */
+int
+xfs_refcount_has_record(
+	struct xfs_btree_cur	*cur,
+	enum xfs_refc_domain	domain,
+	xfs_agblock_t		bno,
+	xfs_extlen_t		len,
+	bool			*exists)
+{
+	union xfs_btree_irec	low;
+	union xfs_btree_irec	high;
+
+	memset(&low, 0, sizeof(low));
+	low.rc.rc_startblock = bno;
+	memset(&high, 0xFF, sizeof(high));
+	high.rc.rc_startblock = bno + len - 1;
+	low.rc.rc_domain = high.rc.rc_domain = domain;
+
+	return xfs_btree_has_record(cur, &low, &high, exists);
+}
+
+int __init
+xfs_refcount_intent_init_cache(void)
+{
+	xfs_refcount_intent_cache = kmem_cache_create("xfs_refc_intent",
+			sizeof(struct xfs_refcount_intent),
+			0, 0, NULL);
+
+	return xfs_refcount_intent_cache != NULL ? 0 : -ENOMEM;
+}
+
+void
+xfs_refcount_intent_destroy_cache(void)
+{
+	kmem_cache_destroy(xfs_refcount_intent_cache);
+	xfs_refcount_intent_cache = NULL;
+}
diff --git a/fs/xfs/libxfs/xfs_refcount.h b/fs/xfs/libxfs/xfs_refcount.h
new file mode 100644
index 000000000..452f30556
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_refcount.h
@@ -0,0 +1,126 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2016 Oracle.  All Rights Reserved.
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ */
+#ifndef __XFS_REFCOUNT_H__
+#define __XFS_REFCOUNT_H__
+
+struct xfs_trans;
+struct xfs_mount;
+struct xfs_perag;
+struct xfs_btree_cur;
+struct xfs_bmbt_irec;
+struct xfs_refcount_irec;
+
+extern int xfs_refcount_lookup_le(struct xfs_btree_cur *cur,
+		enum xfs_refc_domain domain, xfs_agblock_t bno, int *stat);
+extern int xfs_refcount_lookup_ge(struct xfs_btree_cur *cur,
+		enum xfs_refc_domain domain, xfs_agblock_t bno, int *stat);
+extern int xfs_refcount_lookup_eq(struct xfs_btree_cur *cur,
+		enum xfs_refc_domain domain, xfs_agblock_t bno, int *stat);
+extern int xfs_refcount_get_rec(struct xfs_btree_cur *cur,
+		struct xfs_refcount_irec *irec, int *stat);
+
+static inline uint32_t
+xfs_refcount_encode_startblock(
+	xfs_agblock_t		startblock,
+	enum xfs_refc_domain	domain)
+{
+	uint32_t		start;
+
+	/*
+	 * low level btree operations need to handle the generic btree range
+	 * query functions (which set rc_domain == -1U), so we check that the
+	 * domain is /not/ shared.
+	 */
+	start = startblock & ~XFS_REFC_COWFLAG;
+	if (domain != XFS_REFC_DOMAIN_SHARED)
+		start |= XFS_REFC_COWFLAG;
+
+	return start;
+}
+
+enum xfs_refcount_intent_type {
+	XFS_REFCOUNT_INCREASE = 1,
+	XFS_REFCOUNT_DECREASE,
+	XFS_REFCOUNT_ALLOC_COW,
+	XFS_REFCOUNT_FREE_COW,
+};
+
+struct xfs_refcount_intent {
+	struct list_head			ri_list;
+	enum xfs_refcount_intent_type		ri_type;
+	xfs_extlen_t				ri_blockcount;
+	xfs_fsblock_t				ri_startblock;
+};
+
+/* Check that the refcount is appropriate for the record domain. */
+static inline bool
+xfs_refcount_check_domain(
+	const struct xfs_refcount_irec	*irec)
+{
+	if (irec->rc_domain == XFS_REFC_DOMAIN_COW && irec->rc_refcount != 1)
+		return false;
+	if (irec->rc_domain == XFS_REFC_DOMAIN_SHARED && irec->rc_refcount < 2)
+		return false;
+	return true;
+}
+
+void xfs_refcount_increase_extent(struct xfs_trans *tp,
+		struct xfs_bmbt_irec *irec);
+void xfs_refcount_decrease_extent(struct xfs_trans *tp,
+		struct xfs_bmbt_irec *irec);
+
+extern void xfs_refcount_finish_one_cleanup(struct xfs_trans *tp,
+		struct xfs_btree_cur *rcur, int error);
+extern int xfs_refcount_finish_one(struct xfs_trans *tp,
+		enum xfs_refcount_intent_type type, xfs_fsblock_t startblock,
+		xfs_extlen_t blockcount, xfs_fsblock_t *new_fsb,
+		xfs_extlen_t *new_len, struct xfs_btree_cur **pcur);
+
+extern int xfs_refcount_find_shared(struct xfs_btree_cur *cur,
+		xfs_agblock_t agbno, xfs_extlen_t aglen, xfs_agblock_t *fbno,
+		xfs_extlen_t *flen, bool find_end_of_shared);
+
+void xfs_refcount_alloc_cow_extent(struct xfs_trans *tp, xfs_fsblock_t fsb,
+		xfs_extlen_t len);
+void xfs_refcount_free_cow_extent(struct xfs_trans *tp, xfs_fsblock_t fsb,
+		xfs_extlen_t len);
+extern int xfs_refcount_recover_cow_leftovers(struct xfs_mount *mp,
+		struct xfs_perag *pag);
+
+/*
+ * While we're adjusting the refcounts records of an extent, we have
+ * to keep an eye on the number of extents we're dirtying -- run too
+ * many in a single transaction and we'll exceed the transaction's
+ * reservation and crash the fs.  Each record adds 12 bytes to the
+ * log (plus any key updates) so we'll conservatively assume 32 bytes
+ * per record.  We must also leave space for btree splits on both ends
+ * of the range and space for the CUD and a new CUI.
+ *
+ * Each EFI that we attach to the transaction is assumed to consume ~32 bytes.
+ * This is a low estimate for an EFI tracking a single extent (16 bytes for the
+ * EFI header, 16 for the extent, and 12 for the xlog op header), but the
+ * estimate is acceptable if there's more than one extent being freed.
+ * In the worst case of freeing every other block during a refcount decrease
+ * operation, we amortize the space used for one EFI log item across 16
+ * extents.
+ */
+#define XFS_REFCOUNT_ITEM_OVERHEAD	32
+
+extern int xfs_refcount_has_record(struct xfs_btree_cur *cur,
+		enum xfs_refc_domain domain, xfs_agblock_t bno,
+		xfs_extlen_t len, bool *exists);
+union xfs_btree_rec;
+extern void xfs_refcount_btrec_to_irec(const union xfs_btree_rec *rec,
+		struct xfs_refcount_irec *irec);
+extern int xfs_refcount_insert(struct xfs_btree_cur *cur,
+		struct xfs_refcount_irec *irec, int *stat);
+
+extern struct kmem_cache	*xfs_refcount_intent_cache;
+
+int __init xfs_refcount_intent_init_cache(void);
+void xfs_refcount_intent_destroy_cache(void);
+
+#endif	/* __XFS_REFCOUNT_H__ */
diff --git a/fs/xfs/libxfs/xfs_refcount_btree.c b/fs/xfs/libxfs/xfs_refcount_btree.c
new file mode 100644
index 000000000..e1f789866
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_refcount_btree.c
@@ -0,0 +1,545 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2016 Oracle.  All Rights Reserved.
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_btree.h"
+#include "xfs_btree_staging.h"
+#include "xfs_refcount_btree.h"
+#include "xfs_refcount.h"
+#include "xfs_alloc.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+#include "xfs_trans.h"
+#include "xfs_bit.h"
+#include "xfs_rmap.h"
+#include "xfs_ag.h"
+
+static struct kmem_cache	*xfs_refcountbt_cur_cache;
+
+static struct xfs_btree_cur *
+xfs_refcountbt_dup_cursor(
+	struct xfs_btree_cur	*cur)
+{
+	return xfs_refcountbt_init_cursor(cur->bc_mp, cur->bc_tp,
+			cur->bc_ag.agbp, cur->bc_ag.pag);
+}
+
+STATIC void
+xfs_refcountbt_set_root(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*ptr,
+	int				inc)
+{
+	struct xfs_buf		*agbp = cur->bc_ag.agbp;
+	struct xfs_agf		*agf = agbp->b_addr;
+	struct xfs_perag	*pag = agbp->b_pag;
+
+	ASSERT(ptr->s != 0);
+
+	agf->agf_refcount_root = ptr->s;
+	be32_add_cpu(&agf->agf_refcount_level, inc);
+	pag->pagf_refcount_level += inc;
+
+	xfs_alloc_log_agf(cur->bc_tp, agbp,
+			XFS_AGF_REFCOUNT_ROOT | XFS_AGF_REFCOUNT_LEVEL);
+}
+
+STATIC int
+xfs_refcountbt_alloc_block(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*start,
+	union xfs_btree_ptr		*new,
+	int				*stat)
+{
+	struct xfs_buf		*agbp = cur->bc_ag.agbp;
+	struct xfs_agf		*agf = agbp->b_addr;
+	struct xfs_alloc_arg	args;		/* block allocation args */
+	int			error;		/* error return value */
+
+	memset(&args, 0, sizeof(args));
+	args.tp = cur->bc_tp;
+	args.mp = cur->bc_mp;
+	args.type = XFS_ALLOCTYPE_NEAR_BNO;
+	args.fsbno = XFS_AGB_TO_FSB(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+			xfs_refc_block(args.mp));
+	args.oinfo = XFS_RMAP_OINFO_REFC;
+	args.minlen = args.maxlen = args.prod = 1;
+	args.resv = XFS_AG_RESV_METADATA;
+
+	error = xfs_alloc_vextent(&args);
+	if (error)
+		goto out_error;
+	trace_xfs_refcountbt_alloc_block(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+			args.agbno, 1);
+	if (args.fsbno == NULLFSBLOCK) {
+		*stat = 0;
+		return 0;
+	}
+	ASSERT(args.agno == cur->bc_ag.pag->pag_agno);
+	ASSERT(args.len == 1);
+
+	new->s = cpu_to_be32(args.agbno);
+	be32_add_cpu(&agf->agf_refcount_blocks, 1);
+	xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_REFCOUNT_BLOCKS);
+
+	*stat = 1;
+	return 0;
+
+out_error:
+	return error;
+}
+
+STATIC int
+xfs_refcountbt_free_block(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = cur->bc_mp;
+	struct xfs_buf		*agbp = cur->bc_ag.agbp;
+	struct xfs_agf		*agf = agbp->b_addr;
+	xfs_fsblock_t		fsbno = XFS_DADDR_TO_FSB(mp, xfs_buf_daddr(bp));
+	int			error;
+
+	trace_xfs_refcountbt_free_block(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+			XFS_FSB_TO_AGBNO(cur->bc_mp, fsbno), 1);
+	be32_add_cpu(&agf->agf_refcount_blocks, -1);
+	xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_REFCOUNT_BLOCKS);
+	error = xfs_free_extent(cur->bc_tp, fsbno, 1, &XFS_RMAP_OINFO_REFC,
+			XFS_AG_RESV_METADATA);
+	if (error)
+		return error;
+
+	return error;
+}
+
+STATIC int
+xfs_refcountbt_get_minrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	return cur->bc_mp->m_refc_mnr[level != 0];
+}
+
+STATIC int
+xfs_refcountbt_get_maxrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	return cur->bc_mp->m_refc_mxr[level != 0];
+}
+
+STATIC void
+xfs_refcountbt_init_key_from_rec(
+	union xfs_btree_key		*key,
+	const union xfs_btree_rec	*rec)
+{
+	key->refc.rc_startblock = rec->refc.rc_startblock;
+}
+
+STATIC void
+xfs_refcountbt_init_high_key_from_rec(
+	union xfs_btree_key		*key,
+	const union xfs_btree_rec	*rec)
+{
+	__u32				x;
+
+	x = be32_to_cpu(rec->refc.rc_startblock);
+	x += be32_to_cpu(rec->refc.rc_blockcount) - 1;
+	key->refc.rc_startblock = cpu_to_be32(x);
+}
+
+STATIC void
+xfs_refcountbt_init_rec_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec)
+{
+	const struct xfs_refcount_irec *irec = &cur->bc_rec.rc;
+	uint32_t		start;
+
+	start = xfs_refcount_encode_startblock(irec->rc_startblock,
+			irec->rc_domain);
+	rec->refc.rc_startblock = cpu_to_be32(start);
+	rec->refc.rc_blockcount = cpu_to_be32(cur->bc_rec.rc.rc_blockcount);
+	rec->refc.rc_refcount = cpu_to_be32(cur->bc_rec.rc.rc_refcount);
+}
+
+STATIC void
+xfs_refcountbt_init_ptr_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	struct xfs_agf		*agf = cur->bc_ag.agbp->b_addr;
+
+	ASSERT(cur->bc_ag.pag->pag_agno == be32_to_cpu(agf->agf_seqno));
+
+	ptr->s = agf->agf_refcount_root;
+}
+
+STATIC int64_t
+xfs_refcountbt_key_diff(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*key)
+{
+	const struct xfs_refcount_key	*kp = &key->refc;
+	const struct xfs_refcount_irec	*irec = &cur->bc_rec.rc;
+	uint32_t			start;
+
+	start = xfs_refcount_encode_startblock(irec->rc_startblock,
+			irec->rc_domain);
+	return (int64_t)be32_to_cpu(kp->rc_startblock) - start;
+}
+
+STATIC int64_t
+xfs_refcountbt_diff_two_keys(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*k1,
+	const union xfs_btree_key	*k2)
+{
+	return (int64_t)be32_to_cpu(k1->refc.rc_startblock) -
+			  be32_to_cpu(k2->refc.rc_startblock);
+}
+
+STATIC xfs_failaddr_t
+xfs_refcountbt_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
+	struct xfs_perag	*pag = bp->b_pag;
+	xfs_failaddr_t		fa;
+	unsigned int		level;
+
+	if (!xfs_verify_magic(bp, block->bb_magic))
+		return __this_address;
+
+	if (!xfs_has_reflink(mp))
+		return __this_address;
+	fa = xfs_btree_sblock_v5hdr_verify(bp);
+	if (fa)
+		return fa;
+
+	level = be16_to_cpu(block->bb_level);
+	if (pag && pag->pagf_init) {
+		if (level >= pag->pagf_refcount_level)
+			return __this_address;
+	} else if (level >= mp->m_refc_maxlevels)
+		return __this_address;
+
+	return xfs_btree_sblock_verify(bp, mp->m_refc_mxr[level != 0]);
+}
+
+STATIC void
+xfs_refcountbt_read_verify(
+	struct xfs_buf	*bp)
+{
+	xfs_failaddr_t	fa;
+
+	if (!xfs_btree_sblock_verify_crc(bp))
+		xfs_verifier_error(bp, -EFSBADCRC, __this_address);
+	else {
+		fa = xfs_refcountbt_verify(bp);
+		if (fa)
+			xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+	}
+
+	if (bp->b_error)
+		trace_xfs_btree_corrupt(bp, _RET_IP_);
+}
+
+STATIC void
+xfs_refcountbt_write_verify(
+	struct xfs_buf	*bp)
+{
+	xfs_failaddr_t	fa;
+
+	fa = xfs_refcountbt_verify(bp);
+	if (fa) {
+		trace_xfs_btree_corrupt(bp, _RET_IP_);
+		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+		return;
+	}
+	xfs_btree_sblock_calc_crc(bp);
+
+}
+
+const struct xfs_buf_ops xfs_refcountbt_buf_ops = {
+	.name			= "xfs_refcountbt",
+	.magic			= { 0, cpu_to_be32(XFS_REFC_CRC_MAGIC) },
+	.verify_read		= xfs_refcountbt_read_verify,
+	.verify_write		= xfs_refcountbt_write_verify,
+	.verify_struct		= xfs_refcountbt_verify,
+};
+
+STATIC int
+xfs_refcountbt_keys_inorder(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*k1,
+	const union xfs_btree_key	*k2)
+{
+	return be32_to_cpu(k1->refc.rc_startblock) <
+	       be32_to_cpu(k2->refc.rc_startblock);
+}
+
+STATIC int
+xfs_refcountbt_recs_inorder(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_rec	*r1,
+	const union xfs_btree_rec	*r2)
+{
+	return  be32_to_cpu(r1->refc.rc_startblock) +
+		be32_to_cpu(r1->refc.rc_blockcount) <=
+		be32_to_cpu(r2->refc.rc_startblock);
+}
+
+static const struct xfs_btree_ops xfs_refcountbt_ops = {
+	.rec_len		= sizeof(struct xfs_refcount_rec),
+	.key_len		= sizeof(struct xfs_refcount_key),
+
+	.dup_cursor		= xfs_refcountbt_dup_cursor,
+	.set_root		= xfs_refcountbt_set_root,
+	.alloc_block		= xfs_refcountbt_alloc_block,
+	.free_block		= xfs_refcountbt_free_block,
+	.get_minrecs		= xfs_refcountbt_get_minrecs,
+	.get_maxrecs		= xfs_refcountbt_get_maxrecs,
+	.init_key_from_rec	= xfs_refcountbt_init_key_from_rec,
+	.init_high_key_from_rec	= xfs_refcountbt_init_high_key_from_rec,
+	.init_rec_from_cur	= xfs_refcountbt_init_rec_from_cur,
+	.init_ptr_from_cur	= xfs_refcountbt_init_ptr_from_cur,
+	.key_diff		= xfs_refcountbt_key_diff,
+	.buf_ops		= &xfs_refcountbt_buf_ops,
+	.diff_two_keys		= xfs_refcountbt_diff_two_keys,
+	.keys_inorder		= xfs_refcountbt_keys_inorder,
+	.recs_inorder		= xfs_refcountbt_recs_inorder,
+};
+
+/*
+ * Initialize a new refcount btree cursor.
+ */
+static struct xfs_btree_cur *
+xfs_refcountbt_init_common(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_perag	*pag)
+{
+	struct xfs_btree_cur	*cur;
+
+	ASSERT(pag->pag_agno < mp->m_sb.sb_agcount);
+
+	cur = xfs_btree_alloc_cursor(mp, tp, XFS_BTNUM_REFC,
+			mp->m_refc_maxlevels, xfs_refcountbt_cur_cache);
+	cur->bc_statoff = XFS_STATS_CALC_INDEX(xs_refcbt_2);
+
+	cur->bc_flags |= XFS_BTREE_CRC_BLOCKS;
+
+	/* take a reference for the cursor */
+	atomic_inc(&pag->pag_ref);
+	cur->bc_ag.pag = pag;
+
+	cur->bc_ag.refc.nr_ops = 0;
+	cur->bc_ag.refc.shape_changes = 0;
+	cur->bc_ops = &xfs_refcountbt_ops;
+	return cur;
+}
+
+/* Create a btree cursor. */
+struct xfs_btree_cur *
+xfs_refcountbt_init_cursor(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp,
+	struct xfs_perag	*pag)
+{
+	struct xfs_agf		*agf = agbp->b_addr;
+	struct xfs_btree_cur	*cur;
+
+	cur = xfs_refcountbt_init_common(mp, tp, pag);
+	cur->bc_nlevels = be32_to_cpu(agf->agf_refcount_level);
+	cur->bc_ag.agbp = agbp;
+	return cur;
+}
+
+/* Create a btree cursor with a fake root for staging. */
+struct xfs_btree_cur *
+xfs_refcountbt_stage_cursor(
+	struct xfs_mount	*mp,
+	struct xbtree_afakeroot	*afake,
+	struct xfs_perag	*pag)
+{
+	struct xfs_btree_cur	*cur;
+
+	cur = xfs_refcountbt_init_common(mp, NULL, pag);
+	xfs_btree_stage_afakeroot(cur, afake);
+	return cur;
+}
+
+/*
+ * Swap in the new btree root.  Once we pass this point the newly rebuilt btree
+ * is in place and we have to kill off all the old btree blocks.
+ */
+void
+xfs_refcountbt_commit_staged_btree(
+	struct xfs_btree_cur	*cur,
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp)
+{
+	struct xfs_agf		*agf = agbp->b_addr;
+	struct xbtree_afakeroot	*afake = cur->bc_ag.afake;
+
+	ASSERT(cur->bc_flags & XFS_BTREE_STAGING);
+
+	agf->agf_refcount_root = cpu_to_be32(afake->af_root);
+	agf->agf_refcount_level = cpu_to_be32(afake->af_levels);
+	agf->agf_refcount_blocks = cpu_to_be32(afake->af_blocks);
+	xfs_alloc_log_agf(tp, agbp, XFS_AGF_REFCOUNT_BLOCKS |
+				    XFS_AGF_REFCOUNT_ROOT |
+				    XFS_AGF_REFCOUNT_LEVEL);
+	xfs_btree_commit_afakeroot(cur, tp, agbp, &xfs_refcountbt_ops);
+}
+
+/* Calculate number of records in a refcount btree block. */
+static inline unsigned int
+xfs_refcountbt_block_maxrecs(
+	unsigned int		blocklen,
+	bool			leaf)
+{
+	if (leaf)
+		return blocklen / sizeof(struct xfs_refcount_rec);
+	return blocklen / (sizeof(struct xfs_refcount_key) +
+			   sizeof(xfs_refcount_ptr_t));
+}
+
+/*
+ * Calculate the number of records in a refcount btree block.
+ */
+int
+xfs_refcountbt_maxrecs(
+	int			blocklen,
+	bool			leaf)
+{
+	blocklen -= XFS_REFCOUNT_BLOCK_LEN;
+	return xfs_refcountbt_block_maxrecs(blocklen, leaf);
+}
+
+/* Compute the max possible height of the maximally sized refcount btree. */
+unsigned int
+xfs_refcountbt_maxlevels_ondisk(void)
+{
+	unsigned int		minrecs[2];
+	unsigned int		blocklen;
+
+	blocklen = XFS_MIN_CRC_BLOCKSIZE - XFS_BTREE_SBLOCK_CRC_LEN;
+
+	minrecs[0] = xfs_refcountbt_block_maxrecs(blocklen, true) / 2;
+	minrecs[1] = xfs_refcountbt_block_maxrecs(blocklen, false) / 2;
+
+	return xfs_btree_compute_maxlevels(minrecs, XFS_MAX_CRC_AG_BLOCKS);
+}
+
+/* Compute the maximum height of a refcount btree. */
+void
+xfs_refcountbt_compute_maxlevels(
+	struct xfs_mount		*mp)
+{
+	if (!xfs_has_reflink(mp)) {
+		mp->m_refc_maxlevels = 0;
+		return;
+	}
+
+	mp->m_refc_maxlevels = xfs_btree_compute_maxlevels(
+			mp->m_refc_mnr, mp->m_sb.sb_agblocks);
+	ASSERT(mp->m_refc_maxlevels <= xfs_refcountbt_maxlevels_ondisk());
+}
+
+/* Calculate the refcount btree size for some records. */
+xfs_extlen_t
+xfs_refcountbt_calc_size(
+	struct xfs_mount	*mp,
+	unsigned long long	len)
+{
+	return xfs_btree_calc_size(mp->m_refc_mnr, len);
+}
+
+/*
+ * Calculate the maximum refcount btree size.
+ */
+xfs_extlen_t
+xfs_refcountbt_max_size(
+	struct xfs_mount	*mp,
+	xfs_agblock_t		agblocks)
+{
+	/* Bail out if we're uninitialized, which can happen in mkfs. */
+	if (mp->m_refc_mxr[0] == 0)
+		return 0;
+
+	return xfs_refcountbt_calc_size(mp, agblocks);
+}
+
+/*
+ * Figure out how many blocks to reserve and how many are used by this btree.
+ */
+int
+xfs_refcountbt_calc_reserves(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_perag	*pag,
+	xfs_extlen_t		*ask,
+	xfs_extlen_t		*used)
+{
+	struct xfs_buf		*agbp;
+	struct xfs_agf		*agf;
+	xfs_agblock_t		agblocks;
+	xfs_extlen_t		tree_len;
+	int			error;
+
+	if (!xfs_has_reflink(mp))
+		return 0;
+
+	error = xfs_alloc_read_agf(pag, tp, 0, &agbp);
+	if (error)
+		return error;
+
+	agf = agbp->b_addr;
+	agblocks = be32_to_cpu(agf->agf_length);
+	tree_len = be32_to_cpu(agf->agf_refcount_blocks);
+	xfs_trans_brelse(tp, agbp);
+
+	/*
+	 * The log is permanently allocated, so the space it occupies will
+	 * never be available for the kinds of things that would require btree
+	 * expansion.  We therefore can pretend the space isn't there.
+	 */
+	if (xfs_ag_contains_log(mp, pag->pag_agno))
+		agblocks -= mp->m_sb.sb_logblocks;
+
+	*ask += xfs_refcountbt_max_size(mp, agblocks);
+	*used += tree_len;
+
+	return error;
+}
+
+int __init
+xfs_refcountbt_init_cur_cache(void)
+{
+	xfs_refcountbt_cur_cache = kmem_cache_create("xfs_refcbt_cur",
+			xfs_btree_cur_sizeof(xfs_refcountbt_maxlevels_ondisk()),
+			0, 0, NULL);
+
+	if (!xfs_refcountbt_cur_cache)
+		return -ENOMEM;
+	return 0;
+}
+
+void
+xfs_refcountbt_destroy_cur_cache(void)
+{
+	kmem_cache_destroy(xfs_refcountbt_cur_cache);
+	xfs_refcountbt_cur_cache = NULL;
+}
diff --git a/fs/xfs/libxfs/xfs_refcount_btree.h b/fs/xfs/libxfs/xfs_refcount_btree.h
new file mode 100644
index 000000000..d66b37259
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_refcount_btree.h
@@ -0,0 +1,73 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2016 Oracle.  All Rights Reserved.
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ */
+#ifndef __XFS_REFCOUNT_BTREE_H__
+#define	__XFS_REFCOUNT_BTREE_H__
+
+/*
+ * Reference Count Btree on-disk structures
+ */
+
+struct xfs_buf;
+struct xfs_btree_cur;
+struct xfs_mount;
+struct xfs_perag;
+struct xbtree_afakeroot;
+
+/*
+ * Btree block header size
+ */
+#define XFS_REFCOUNT_BLOCK_LEN	XFS_BTREE_SBLOCK_CRC_LEN
+
+/*
+ * Record, key, and pointer address macros for btree blocks.
+ *
+ * (note that some of these may appear unused, but they are used in userspace)
+ */
+#define XFS_REFCOUNT_REC_ADDR(block, index) \
+	((struct xfs_refcount_rec *) \
+		((char *)(block) + \
+		 XFS_REFCOUNT_BLOCK_LEN + \
+		 (((index) - 1) * sizeof(struct xfs_refcount_rec))))
+
+#define XFS_REFCOUNT_KEY_ADDR(block, index) \
+	((struct xfs_refcount_key *) \
+		((char *)(block) + \
+		 XFS_REFCOUNT_BLOCK_LEN + \
+		 ((index) - 1) * sizeof(struct xfs_refcount_key)))
+
+#define XFS_REFCOUNT_PTR_ADDR(block, index, maxrecs) \
+	((xfs_refcount_ptr_t *) \
+		((char *)(block) + \
+		 XFS_REFCOUNT_BLOCK_LEN + \
+		 (maxrecs) * sizeof(struct xfs_refcount_key) + \
+		 ((index) - 1) * sizeof(xfs_refcount_ptr_t)))
+
+extern struct xfs_btree_cur *xfs_refcountbt_init_cursor(struct xfs_mount *mp,
+		struct xfs_trans *tp, struct xfs_buf *agbp,
+		struct xfs_perag *pag);
+struct xfs_btree_cur *xfs_refcountbt_stage_cursor(struct xfs_mount *mp,
+		struct xbtree_afakeroot *afake, struct xfs_perag *pag);
+extern int xfs_refcountbt_maxrecs(int blocklen, bool leaf);
+extern void xfs_refcountbt_compute_maxlevels(struct xfs_mount *mp);
+
+extern xfs_extlen_t xfs_refcountbt_calc_size(struct xfs_mount *mp,
+		unsigned long long len);
+extern xfs_extlen_t xfs_refcountbt_max_size(struct xfs_mount *mp,
+		xfs_agblock_t agblocks);
+
+extern int xfs_refcountbt_calc_reserves(struct xfs_mount *mp,
+		struct xfs_trans *tp, struct xfs_perag *pag, xfs_extlen_t *ask,
+		xfs_extlen_t *used);
+
+void xfs_refcountbt_commit_staged_btree(struct xfs_btree_cur *cur,
+		struct xfs_trans *tp, struct xfs_buf *agbp);
+
+unsigned int xfs_refcountbt_maxlevels_ondisk(void);
+
+int __init xfs_refcountbt_init_cur_cache(void);
+void xfs_refcountbt_destroy_cur_cache(void);
+
+#endif	/* __XFS_REFCOUNT_BTREE_H__ */
diff --git a/fs/xfs/libxfs/xfs_rmap.c b/fs/xfs/libxfs/xfs_rmap.c
new file mode 100644
index 000000000..b56aca1e7
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_rmap.c
@@ -0,0 +1,2826 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2014 Red Hat, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_bit.h"
+#include "xfs_mount.h"
+#include "xfs_sb.h"
+#include "xfs_defer.h"
+#include "xfs_btree.h"
+#include "xfs_trans.h"
+#include "xfs_alloc.h"
+#include "xfs_rmap.h"
+#include "xfs_rmap_btree.h"
+#include "xfs_trace.h"
+#include "xfs_errortag.h"
+#include "xfs_error.h"
+#include "xfs_inode.h"
+#include "xfs_ag.h"
+
+struct kmem_cache	*xfs_rmap_intent_cache;
+
+/*
+ * Lookup the first record less than or equal to [bno, len, owner, offset]
+ * in the btree given by cur.
+ */
+int
+xfs_rmap_lookup_le(
+	struct xfs_btree_cur	*cur,
+	xfs_agblock_t		bno,
+	uint64_t		owner,
+	uint64_t		offset,
+	unsigned int		flags,
+	struct xfs_rmap_irec	*irec,
+	int			*stat)
+{
+	int			get_stat = 0;
+	int			error;
+
+	cur->bc_rec.r.rm_startblock = bno;
+	cur->bc_rec.r.rm_blockcount = 0;
+	cur->bc_rec.r.rm_owner = owner;
+	cur->bc_rec.r.rm_offset = offset;
+	cur->bc_rec.r.rm_flags = flags;
+
+	error = xfs_btree_lookup(cur, XFS_LOOKUP_LE, stat);
+	if (error || !(*stat) || !irec)
+		return error;
+
+	error = xfs_rmap_get_rec(cur, irec, &get_stat);
+	if (error)
+		return error;
+	if (!get_stat)
+		return -EFSCORRUPTED;
+
+	return 0;
+}
+
+/*
+ * Lookup the record exactly matching [bno, len, owner, offset]
+ * in the btree given by cur.
+ */
+int
+xfs_rmap_lookup_eq(
+	struct xfs_btree_cur	*cur,
+	xfs_agblock_t		bno,
+	xfs_extlen_t		len,
+	uint64_t		owner,
+	uint64_t		offset,
+	unsigned int		flags,
+	int			*stat)
+{
+	cur->bc_rec.r.rm_startblock = bno;
+	cur->bc_rec.r.rm_blockcount = len;
+	cur->bc_rec.r.rm_owner = owner;
+	cur->bc_rec.r.rm_offset = offset;
+	cur->bc_rec.r.rm_flags = flags;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_EQ, stat);
+}
+
+/*
+ * Update the record referred to by cur to the value given
+ * by [bno, len, owner, offset].
+ * This either works (return 0) or gets an EFSCORRUPTED error.
+ */
+STATIC int
+xfs_rmap_update(
+	struct xfs_btree_cur	*cur,
+	struct xfs_rmap_irec	*irec)
+{
+	union xfs_btree_rec	rec;
+	int			error;
+
+	trace_xfs_rmap_update(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+			irec->rm_startblock, irec->rm_blockcount,
+			irec->rm_owner, irec->rm_offset, irec->rm_flags);
+
+	rec.rmap.rm_startblock = cpu_to_be32(irec->rm_startblock);
+	rec.rmap.rm_blockcount = cpu_to_be32(irec->rm_blockcount);
+	rec.rmap.rm_owner = cpu_to_be64(irec->rm_owner);
+	rec.rmap.rm_offset = cpu_to_be64(
+			xfs_rmap_irec_offset_pack(irec));
+	error = xfs_btree_update(cur, &rec);
+	if (error)
+		trace_xfs_rmap_update_error(cur->bc_mp,
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+int
+xfs_rmap_insert(
+	struct xfs_btree_cur	*rcur,
+	xfs_agblock_t		agbno,
+	xfs_extlen_t		len,
+	uint64_t		owner,
+	uint64_t		offset,
+	unsigned int		flags)
+{
+	int			i;
+	int			error;
+
+	trace_xfs_rmap_insert(rcur->bc_mp, rcur->bc_ag.pag->pag_agno, agbno,
+			len, owner, offset, flags);
+
+	error = xfs_rmap_lookup_eq(rcur, agbno, len, owner, offset, flags, &i);
+	if (error)
+		goto done;
+	if (XFS_IS_CORRUPT(rcur->bc_mp, i != 0)) {
+		error = -EFSCORRUPTED;
+		goto done;
+	}
+
+	rcur->bc_rec.r.rm_startblock = agbno;
+	rcur->bc_rec.r.rm_blockcount = len;
+	rcur->bc_rec.r.rm_owner = owner;
+	rcur->bc_rec.r.rm_offset = offset;
+	rcur->bc_rec.r.rm_flags = flags;
+	error = xfs_btree_insert(rcur, &i);
+	if (error)
+		goto done;
+	if (XFS_IS_CORRUPT(rcur->bc_mp, i != 1)) {
+		error = -EFSCORRUPTED;
+		goto done;
+	}
+done:
+	if (error)
+		trace_xfs_rmap_insert_error(rcur->bc_mp,
+				rcur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+STATIC int
+xfs_rmap_delete(
+	struct xfs_btree_cur	*rcur,
+	xfs_agblock_t		agbno,
+	xfs_extlen_t		len,
+	uint64_t		owner,
+	uint64_t		offset,
+	unsigned int		flags)
+{
+	int			i;
+	int			error;
+
+	trace_xfs_rmap_delete(rcur->bc_mp, rcur->bc_ag.pag->pag_agno, agbno,
+			len, owner, offset, flags);
+
+	error = xfs_rmap_lookup_eq(rcur, agbno, len, owner, offset, flags, &i);
+	if (error)
+		goto done;
+	if (XFS_IS_CORRUPT(rcur->bc_mp, i != 1)) {
+		error = -EFSCORRUPTED;
+		goto done;
+	}
+
+	error = xfs_btree_delete(rcur, &i);
+	if (error)
+		goto done;
+	if (XFS_IS_CORRUPT(rcur->bc_mp, i != 1)) {
+		error = -EFSCORRUPTED;
+		goto done;
+	}
+done:
+	if (error)
+		trace_xfs_rmap_delete_error(rcur->bc_mp,
+				rcur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+/* Convert an internal btree record to an rmap record. */
+int
+xfs_rmap_btrec_to_irec(
+	const union xfs_btree_rec	*rec,
+	struct xfs_rmap_irec		*irec)
+{
+	irec->rm_startblock = be32_to_cpu(rec->rmap.rm_startblock);
+	irec->rm_blockcount = be32_to_cpu(rec->rmap.rm_blockcount);
+	irec->rm_owner = be64_to_cpu(rec->rmap.rm_owner);
+	return xfs_rmap_irec_offset_unpack(be64_to_cpu(rec->rmap.rm_offset),
+			irec);
+}
+
+/*
+ * Get the data from the pointed-to record.
+ */
+int
+xfs_rmap_get_rec(
+	struct xfs_btree_cur	*cur,
+	struct xfs_rmap_irec	*irec,
+	int			*stat)
+{
+	struct xfs_mount	*mp = cur->bc_mp;
+	struct xfs_perag	*pag = cur->bc_ag.pag;
+	union xfs_btree_rec	*rec;
+	int			error;
+
+	error = xfs_btree_get_rec(cur, &rec, stat);
+	if (error || !*stat)
+		return error;
+
+	if (xfs_rmap_btrec_to_irec(rec, irec))
+		goto out_bad_rec;
+
+	if (irec->rm_blockcount == 0)
+		goto out_bad_rec;
+	if (irec->rm_startblock <= XFS_AGFL_BLOCK(mp)) {
+		if (irec->rm_owner != XFS_RMAP_OWN_FS)
+			goto out_bad_rec;
+		if (irec->rm_blockcount != XFS_AGFL_BLOCK(mp) + 1)
+			goto out_bad_rec;
+	} else {
+		/* check for valid extent range, including overflow */
+		if (!xfs_verify_agbext(pag, irec->rm_startblock,
+					    irec->rm_blockcount))
+			goto out_bad_rec;
+	}
+
+	if (!(xfs_verify_ino(mp, irec->rm_owner) ||
+	      (irec->rm_owner <= XFS_RMAP_OWN_FS &&
+	       irec->rm_owner >= XFS_RMAP_OWN_MIN)))
+		goto out_bad_rec;
+
+	return 0;
+out_bad_rec:
+	xfs_warn(mp,
+		"Reverse Mapping BTree record corruption in AG %d detected!",
+		pag->pag_agno);
+	xfs_warn(mp,
+		"Owner 0x%llx, flags 0x%x, start block 0x%x block count 0x%x",
+		irec->rm_owner, irec->rm_flags, irec->rm_startblock,
+		irec->rm_blockcount);
+	return -EFSCORRUPTED;
+}
+
+struct xfs_find_left_neighbor_info {
+	struct xfs_rmap_irec	high;
+	struct xfs_rmap_irec	*irec;
+};
+
+/* For each rmap given, figure out if it matches the key we want. */
+STATIC int
+xfs_rmap_find_left_neighbor_helper(
+	struct xfs_btree_cur		*cur,
+	const struct xfs_rmap_irec	*rec,
+	void				*priv)
+{
+	struct xfs_find_left_neighbor_info	*info = priv;
+
+	trace_xfs_rmap_find_left_neighbor_candidate(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, rec->rm_startblock,
+			rec->rm_blockcount, rec->rm_owner, rec->rm_offset,
+			rec->rm_flags);
+
+	if (rec->rm_owner != info->high.rm_owner)
+		return 0;
+	if (!XFS_RMAP_NON_INODE_OWNER(rec->rm_owner) &&
+	    !(rec->rm_flags & XFS_RMAP_BMBT_BLOCK) &&
+	    rec->rm_offset + rec->rm_blockcount - 1 != info->high.rm_offset)
+		return 0;
+
+	*info->irec = *rec;
+	return -ECANCELED;
+}
+
+/*
+ * Find the record to the left of the given extent, being careful only to
+ * return a match with the same owner and adjacent physical and logical
+ * block ranges.
+ */
+STATIC int
+xfs_rmap_find_left_neighbor(
+	struct xfs_btree_cur	*cur,
+	xfs_agblock_t		bno,
+	uint64_t		owner,
+	uint64_t		offset,
+	unsigned int		flags,
+	struct xfs_rmap_irec	*irec,
+	int			*stat)
+{
+	struct xfs_find_left_neighbor_info	info;
+	int			found = 0;
+	int			error;
+
+	*stat = 0;
+	if (bno == 0)
+		return 0;
+	info.high.rm_startblock = bno - 1;
+	info.high.rm_owner = owner;
+	if (!XFS_RMAP_NON_INODE_OWNER(owner) &&
+	    !(flags & XFS_RMAP_BMBT_BLOCK)) {
+		if (offset == 0)
+			return 0;
+		info.high.rm_offset = offset - 1;
+	} else
+		info.high.rm_offset = 0;
+	info.high.rm_flags = flags;
+	info.high.rm_blockcount = 0;
+	info.irec = irec;
+
+	trace_xfs_rmap_find_left_neighbor_query(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, bno, 0, owner, offset, flags);
+
+	/*
+	 * Historically, we always used the range query to walk every reverse
+	 * mapping that could possibly overlap the key that the caller asked
+	 * for, and filter out the ones that don't.  That is very slow when
+	 * there are a lot of records.
+	 *
+	 * However, there are two scenarios where the classic btree search can
+	 * produce correct results -- if the index contains a record that is an
+	 * exact match for the lookup key; and if there are no other records
+	 * between the record we want and the key we supplied.
+	 *
+	 * As an optimization, try a non-overlapped lookup first.  This makes
+	 * extent conversion and remap operations run a bit faster if the
+	 * physical extents aren't being shared.  If we don't find what we
+	 * want, we fall back to the overlapped query.
+	 */
+	error = xfs_rmap_lookup_le(cur, bno, owner, offset, flags, irec,
+			&found);
+	if (error)
+		return error;
+	if (found)
+		error = xfs_rmap_find_left_neighbor_helper(cur, irec, &info);
+	if (!error)
+		error = xfs_rmap_query_range(cur, &info.high, &info.high,
+				xfs_rmap_find_left_neighbor_helper, &info);
+	if (error != -ECANCELED)
+		return error;
+
+	*stat = 1;
+	trace_xfs_rmap_find_left_neighbor_result(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, irec->rm_startblock,
+			irec->rm_blockcount, irec->rm_owner, irec->rm_offset,
+			irec->rm_flags);
+	return 0;
+}
+
+/* For each rmap given, figure out if it matches the key we want. */
+STATIC int
+xfs_rmap_lookup_le_range_helper(
+	struct xfs_btree_cur		*cur,
+	const struct xfs_rmap_irec	*rec,
+	void				*priv)
+{
+	struct xfs_find_left_neighbor_info	*info = priv;
+
+	trace_xfs_rmap_lookup_le_range_candidate(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, rec->rm_startblock,
+			rec->rm_blockcount, rec->rm_owner, rec->rm_offset,
+			rec->rm_flags);
+
+	if (rec->rm_owner != info->high.rm_owner)
+		return 0;
+	if (!XFS_RMAP_NON_INODE_OWNER(rec->rm_owner) &&
+	    !(rec->rm_flags & XFS_RMAP_BMBT_BLOCK) &&
+	    (rec->rm_offset > info->high.rm_offset ||
+	     rec->rm_offset + rec->rm_blockcount <= info->high.rm_offset))
+		return 0;
+
+	*info->irec = *rec;
+	return -ECANCELED;
+}
+
+/*
+ * Find the record to the left of the given extent, being careful only to
+ * return a match with the same owner and overlapping physical and logical
+ * block ranges.  This is the overlapping-interval version of
+ * xfs_rmap_lookup_le.
+ */
+int
+xfs_rmap_lookup_le_range(
+	struct xfs_btree_cur	*cur,
+	xfs_agblock_t		bno,
+	uint64_t		owner,
+	uint64_t		offset,
+	unsigned int		flags,
+	struct xfs_rmap_irec	*irec,
+	int			*stat)
+{
+	struct xfs_find_left_neighbor_info	info;
+	int			found = 0;
+	int			error;
+
+	info.high.rm_startblock = bno;
+	info.high.rm_owner = owner;
+	if (!XFS_RMAP_NON_INODE_OWNER(owner) && !(flags & XFS_RMAP_BMBT_BLOCK))
+		info.high.rm_offset = offset;
+	else
+		info.high.rm_offset = 0;
+	info.high.rm_flags = flags;
+	info.high.rm_blockcount = 0;
+	*stat = 0;
+	info.irec = irec;
+
+	trace_xfs_rmap_lookup_le_range(cur->bc_mp, cur->bc_ag.pag->pag_agno,
+			bno, 0, owner, offset, flags);
+
+	/*
+	 * Historically, we always used the range query to walk every reverse
+	 * mapping that could possibly overlap the key that the caller asked
+	 * for, and filter out the ones that don't.  That is very slow when
+	 * there are a lot of records.
+	 *
+	 * However, there are two scenarios where the classic btree search can
+	 * produce correct results -- if the index contains a record that is an
+	 * exact match for the lookup key; and if there are no other records
+	 * between the record we want and the key we supplied.
+	 *
+	 * As an optimization, try a non-overlapped lookup first.  This makes
+	 * scrub run much faster on most filesystems because bmbt records are
+	 * usually an exact match for rmap records.  If we don't find what we
+	 * want, we fall back to the overlapped query.
+	 */
+	error = xfs_rmap_lookup_le(cur, bno, owner, offset, flags, irec,
+			&found);
+	if (error)
+		return error;
+	if (found)
+		error = xfs_rmap_lookup_le_range_helper(cur, irec, &info);
+	if (!error)
+		error = xfs_rmap_query_range(cur, &info.high, &info.high,
+				xfs_rmap_lookup_le_range_helper, &info);
+	if (error != -ECANCELED)
+		return error;
+
+	*stat = 1;
+	trace_xfs_rmap_lookup_le_range_result(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, irec->rm_startblock,
+			irec->rm_blockcount, irec->rm_owner, irec->rm_offset,
+			irec->rm_flags);
+	return 0;
+}
+
+/*
+ * Perform all the relevant owner checks for a removal op.  If we're doing an
+ * unknown-owner removal then we have no owner information to check.
+ */
+static int
+xfs_rmap_free_check_owner(
+	struct xfs_mount	*mp,
+	uint64_t		ltoff,
+	struct xfs_rmap_irec	*rec,
+	xfs_filblks_t		len,
+	uint64_t		owner,
+	uint64_t		offset,
+	unsigned int		flags)
+{
+	int			error = 0;
+
+	if (owner == XFS_RMAP_OWN_UNKNOWN)
+		return 0;
+
+	/* Make sure the unwritten flag matches. */
+	if (XFS_IS_CORRUPT(mp,
+			   (flags & XFS_RMAP_UNWRITTEN) !=
+			   (rec->rm_flags & XFS_RMAP_UNWRITTEN))) {
+		error = -EFSCORRUPTED;
+		goto out;
+	}
+
+	/* Make sure the owner matches what we expect to find in the tree. */
+	if (XFS_IS_CORRUPT(mp, owner != rec->rm_owner)) {
+		error = -EFSCORRUPTED;
+		goto out;
+	}
+
+	/* Check the offset, if necessary. */
+	if (XFS_RMAP_NON_INODE_OWNER(owner))
+		goto out;
+
+	if (flags & XFS_RMAP_BMBT_BLOCK) {
+		if (XFS_IS_CORRUPT(mp,
+				   !(rec->rm_flags & XFS_RMAP_BMBT_BLOCK))) {
+			error = -EFSCORRUPTED;
+			goto out;
+		}
+	} else {
+		if (XFS_IS_CORRUPT(mp, rec->rm_offset > offset)) {
+			error = -EFSCORRUPTED;
+			goto out;
+		}
+		if (XFS_IS_CORRUPT(mp,
+				   offset + len > ltoff + rec->rm_blockcount)) {
+			error = -EFSCORRUPTED;
+			goto out;
+		}
+	}
+
+out:
+	return error;
+}
+
+/*
+ * Find the extent in the rmap btree and remove it.
+ *
+ * The record we find should always be an exact match for the extent that we're
+ * looking for, since we insert them into the btree without modification.
+ *
+ * Special Case #1: when growing the filesystem, we "free" an extent when
+ * growing the last AG. This extent is new space and so it is not tracked as
+ * used space in the btree. The growfs code will pass in an owner of
+ * XFS_RMAP_OWN_NULL to indicate that it expected that there is no owner of this
+ * extent. We verify that - the extent lookup result in a record that does not
+ * overlap.
+ *
+ * Special Case #2: EFIs do not record the owner of the extent, so when
+ * recovering EFIs from the log we pass in XFS_RMAP_OWN_UNKNOWN to tell the rmap
+ * btree to ignore the owner (i.e. wildcard match) so we don't trigger
+ * corruption checks during log recovery.
+ */
+STATIC int
+xfs_rmap_unmap(
+	struct xfs_btree_cur		*cur,
+	xfs_agblock_t			bno,
+	xfs_extlen_t			len,
+	bool				unwritten,
+	const struct xfs_owner_info	*oinfo)
+{
+	struct xfs_mount		*mp = cur->bc_mp;
+	struct xfs_rmap_irec		ltrec;
+	uint64_t			ltoff;
+	int				error = 0;
+	int				i;
+	uint64_t			owner;
+	uint64_t			offset;
+	unsigned int			flags;
+	bool				ignore_off;
+
+	xfs_owner_info_unpack(oinfo, &owner, &offset, &flags);
+	ignore_off = XFS_RMAP_NON_INODE_OWNER(owner) ||
+			(flags & XFS_RMAP_BMBT_BLOCK);
+	if (unwritten)
+		flags |= XFS_RMAP_UNWRITTEN;
+	trace_xfs_rmap_unmap(mp, cur->bc_ag.pag->pag_agno, bno, len,
+			unwritten, oinfo);
+
+	/*
+	 * We should always have a left record because there's a static record
+	 * for the AG headers at rm_startblock == 0 created by mkfs/growfs that
+	 * will not ever be removed from the tree.
+	 */
+	error = xfs_rmap_lookup_le(cur, bno, owner, offset, flags, &ltrec, &i);
+	if (error)
+		goto out_error;
+	if (XFS_IS_CORRUPT(mp, i != 1)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+
+	trace_xfs_rmap_lookup_le_range_result(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, ltrec.rm_startblock,
+			ltrec.rm_blockcount, ltrec.rm_owner,
+			ltrec.rm_offset, ltrec.rm_flags);
+	ltoff = ltrec.rm_offset;
+
+	/*
+	 * For growfs, the incoming extent must be beyond the left record we
+	 * just found as it is new space and won't be used by anyone. This is
+	 * just a corruption check as we don't actually do anything with this
+	 * extent.  Note that we need to use >= instead of > because it might
+	 * be the case that the "left" extent goes all the way to EOFS.
+	 */
+	if (owner == XFS_RMAP_OWN_NULL) {
+		if (XFS_IS_CORRUPT(mp,
+				   bno <
+				   ltrec.rm_startblock + ltrec.rm_blockcount)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+		goto out_done;
+	}
+
+	/*
+	 * If we're doing an unknown-owner removal for EFI recovery, we expect
+	 * to find the full range in the rmapbt or nothing at all.  If we
+	 * don't find any rmaps overlapping either end of the range, we're
+	 * done.  Hopefully this means that the EFI creator already queued
+	 * (and finished) a RUI to remove the rmap.
+	 */
+	if (owner == XFS_RMAP_OWN_UNKNOWN &&
+	    ltrec.rm_startblock + ltrec.rm_blockcount <= bno) {
+		struct xfs_rmap_irec    rtrec;
+
+		error = xfs_btree_increment(cur, 0, &i);
+		if (error)
+			goto out_error;
+		if (i == 0)
+			goto out_done;
+		error = xfs_rmap_get_rec(cur, &rtrec, &i);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+		if (rtrec.rm_startblock >= bno + len)
+			goto out_done;
+	}
+
+	/* Make sure the extent we found covers the entire freeing range. */
+	if (XFS_IS_CORRUPT(mp,
+			   ltrec.rm_startblock > bno ||
+			   ltrec.rm_startblock + ltrec.rm_blockcount <
+			   bno + len)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+
+	/* Check owner information. */
+	error = xfs_rmap_free_check_owner(mp, ltoff, &ltrec, len, owner,
+			offset, flags);
+	if (error)
+		goto out_error;
+
+	if (ltrec.rm_startblock == bno && ltrec.rm_blockcount == len) {
+		/* exact match, simply remove the record from rmap tree */
+		trace_xfs_rmap_delete(mp, cur->bc_ag.pag->pag_agno,
+				ltrec.rm_startblock, ltrec.rm_blockcount,
+				ltrec.rm_owner, ltrec.rm_offset,
+				ltrec.rm_flags);
+		error = xfs_btree_delete(cur, &i);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+	} else if (ltrec.rm_startblock == bno) {
+		/*
+		 * overlap left hand side of extent: move the start, trim the
+		 * length and update the current record.
+		 *
+		 *       ltbno                ltlen
+		 * Orig:    |oooooooooooooooooooo|
+		 * Freeing: |fffffffff|
+		 * Result:            |rrrrrrrrrr|
+		 *         bno       len
+		 */
+		ltrec.rm_startblock += len;
+		ltrec.rm_blockcount -= len;
+		if (!ignore_off)
+			ltrec.rm_offset += len;
+		error = xfs_rmap_update(cur, &ltrec);
+		if (error)
+			goto out_error;
+	} else if (ltrec.rm_startblock + ltrec.rm_blockcount == bno + len) {
+		/*
+		 * overlap right hand side of extent: trim the length and update
+		 * the current record.
+		 *
+		 *       ltbno                ltlen
+		 * Orig:    |oooooooooooooooooooo|
+		 * Freeing:            |fffffffff|
+		 * Result:  |rrrrrrrrrr|
+		 *                    bno       len
+		 */
+		ltrec.rm_blockcount -= len;
+		error = xfs_rmap_update(cur, &ltrec);
+		if (error)
+			goto out_error;
+	} else {
+
+		/*
+		 * overlap middle of extent: trim the length of the existing
+		 * record to the length of the new left-extent size, increment
+		 * the insertion position so we can insert a new record
+		 * containing the remaining right-extent space.
+		 *
+		 *       ltbno                ltlen
+		 * Orig:    |oooooooooooooooooooo|
+		 * Freeing:       |fffffffff|
+		 * Result:  |rrrrr|         |rrrr|
+		 *               bno       len
+		 */
+		xfs_extlen_t	orig_len = ltrec.rm_blockcount;
+
+		ltrec.rm_blockcount = bno - ltrec.rm_startblock;
+		error = xfs_rmap_update(cur, &ltrec);
+		if (error)
+			goto out_error;
+
+		error = xfs_btree_increment(cur, 0, &i);
+		if (error)
+			goto out_error;
+
+		cur->bc_rec.r.rm_startblock = bno + len;
+		cur->bc_rec.r.rm_blockcount = orig_len - len -
+						     ltrec.rm_blockcount;
+		cur->bc_rec.r.rm_owner = ltrec.rm_owner;
+		if (ignore_off)
+			cur->bc_rec.r.rm_offset = 0;
+		else
+			cur->bc_rec.r.rm_offset = offset + len;
+		cur->bc_rec.r.rm_flags = flags;
+		trace_xfs_rmap_insert(mp, cur->bc_ag.pag->pag_agno,
+				cur->bc_rec.r.rm_startblock,
+				cur->bc_rec.r.rm_blockcount,
+				cur->bc_rec.r.rm_owner,
+				cur->bc_rec.r.rm_offset,
+				cur->bc_rec.r.rm_flags);
+		error = xfs_btree_insert(cur, &i);
+		if (error)
+			goto out_error;
+	}
+
+out_done:
+	trace_xfs_rmap_unmap_done(mp, cur->bc_ag.pag->pag_agno, bno, len,
+			unwritten, oinfo);
+out_error:
+	if (error)
+		trace_xfs_rmap_unmap_error(mp, cur->bc_ag.pag->pag_agno,
+				error, _RET_IP_);
+	return error;
+}
+
+/*
+ * Remove a reference to an extent in the rmap btree.
+ */
+int
+xfs_rmap_free(
+	struct xfs_trans		*tp,
+	struct xfs_buf			*agbp,
+	struct xfs_perag		*pag,
+	xfs_agblock_t			bno,
+	xfs_extlen_t			len,
+	const struct xfs_owner_info	*oinfo)
+{
+	struct xfs_mount		*mp = tp->t_mountp;
+	struct xfs_btree_cur		*cur;
+	int				error;
+
+	if (!xfs_has_rmapbt(mp))
+		return 0;
+
+	cur = xfs_rmapbt_init_cursor(mp, tp, agbp, pag);
+
+	error = xfs_rmap_unmap(cur, bno, len, false, oinfo);
+
+	xfs_btree_del_cursor(cur, error);
+	return error;
+}
+
+/*
+ * A mergeable rmap must have the same owner and the same values for
+ * the unwritten, attr_fork, and bmbt flags.  The startblock and
+ * offset are checked separately.
+ */
+static bool
+xfs_rmap_is_mergeable(
+	struct xfs_rmap_irec	*irec,
+	uint64_t		owner,
+	unsigned int		flags)
+{
+	if (irec->rm_owner == XFS_RMAP_OWN_NULL)
+		return false;
+	if (irec->rm_owner != owner)
+		return false;
+	if ((flags & XFS_RMAP_UNWRITTEN) ^
+	    (irec->rm_flags & XFS_RMAP_UNWRITTEN))
+		return false;
+	if ((flags & XFS_RMAP_ATTR_FORK) ^
+	    (irec->rm_flags & XFS_RMAP_ATTR_FORK))
+		return false;
+	if ((flags & XFS_RMAP_BMBT_BLOCK) ^
+	    (irec->rm_flags & XFS_RMAP_BMBT_BLOCK))
+		return false;
+	return true;
+}
+
+/*
+ * When we allocate a new block, the first thing we do is add a reference to
+ * the extent in the rmap btree. This takes the form of a [agbno, length,
+ * owner, offset] record.  Flags are encoded in the high bits of the offset
+ * field.
+ */
+STATIC int
+xfs_rmap_map(
+	struct xfs_btree_cur		*cur,
+	xfs_agblock_t			bno,
+	xfs_extlen_t			len,
+	bool				unwritten,
+	const struct xfs_owner_info	*oinfo)
+{
+	struct xfs_mount		*mp = cur->bc_mp;
+	struct xfs_rmap_irec		ltrec;
+	struct xfs_rmap_irec		gtrec;
+	int				have_gt;
+	int				have_lt;
+	int				error = 0;
+	int				i;
+	uint64_t			owner;
+	uint64_t			offset;
+	unsigned int			flags = 0;
+	bool				ignore_off;
+
+	xfs_owner_info_unpack(oinfo, &owner, &offset, &flags);
+	ASSERT(owner != 0);
+	ignore_off = XFS_RMAP_NON_INODE_OWNER(owner) ||
+			(flags & XFS_RMAP_BMBT_BLOCK);
+	if (unwritten)
+		flags |= XFS_RMAP_UNWRITTEN;
+	trace_xfs_rmap_map(mp, cur->bc_ag.pag->pag_agno, bno, len,
+			unwritten, oinfo);
+	ASSERT(!xfs_rmap_should_skip_owner_update(oinfo));
+
+	/*
+	 * For the initial lookup, look for an exact match or the left-adjacent
+	 * record for our insertion point. This will also give us the record for
+	 * start block contiguity tests.
+	 */
+	error = xfs_rmap_lookup_le(cur, bno, owner, offset, flags, &ltrec,
+			&have_lt);
+	if (error)
+		goto out_error;
+	if (have_lt) {
+		trace_xfs_rmap_lookup_le_range_result(cur->bc_mp,
+				cur->bc_ag.pag->pag_agno, ltrec.rm_startblock,
+				ltrec.rm_blockcount, ltrec.rm_owner,
+				ltrec.rm_offset, ltrec.rm_flags);
+
+		if (!xfs_rmap_is_mergeable(&ltrec, owner, flags))
+			have_lt = 0;
+	}
+
+	if (XFS_IS_CORRUPT(mp,
+			   have_lt != 0 &&
+			   ltrec.rm_startblock + ltrec.rm_blockcount > bno)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+
+	/*
+	 * Increment the cursor to see if we have a right-adjacent record to our
+	 * insertion point. This will give us the record for end block
+	 * contiguity tests.
+	 */
+	error = xfs_btree_increment(cur, 0, &have_gt);
+	if (error)
+		goto out_error;
+	if (have_gt) {
+		error = xfs_rmap_get_rec(cur, &gtrec, &have_gt);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(mp, have_gt != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+		if (XFS_IS_CORRUPT(mp, bno + len > gtrec.rm_startblock)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+		trace_xfs_rmap_find_right_neighbor_result(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, gtrec.rm_startblock,
+			gtrec.rm_blockcount, gtrec.rm_owner,
+			gtrec.rm_offset, gtrec.rm_flags);
+		if (!xfs_rmap_is_mergeable(&gtrec, owner, flags))
+			have_gt = 0;
+	}
+
+	/*
+	 * Note: cursor currently points one record to the right of ltrec, even
+	 * if there is no record in the tree to the right.
+	 */
+	if (have_lt &&
+	    ltrec.rm_startblock + ltrec.rm_blockcount == bno &&
+	    (ignore_off || ltrec.rm_offset + ltrec.rm_blockcount == offset)) {
+		/*
+		 * left edge contiguous, merge into left record.
+		 *
+		 *       ltbno     ltlen
+		 * orig:   |ooooooooo|
+		 * adding:           |aaaaaaaaa|
+		 * result: |rrrrrrrrrrrrrrrrrrr|
+		 *                  bno       len
+		 */
+		ltrec.rm_blockcount += len;
+		if (have_gt &&
+		    bno + len == gtrec.rm_startblock &&
+		    (ignore_off || offset + len == gtrec.rm_offset) &&
+		    (unsigned long)ltrec.rm_blockcount + len +
+				gtrec.rm_blockcount <= XFS_RMAP_LEN_MAX) {
+			/*
+			 * right edge also contiguous, delete right record
+			 * and merge into left record.
+			 *
+			 *       ltbno     ltlen    gtbno     gtlen
+			 * orig:   |ooooooooo|         |ooooooooo|
+			 * adding:           |aaaaaaaaa|
+			 * result: |rrrrrrrrrrrrrrrrrrrrrrrrrrrrr|
+			 */
+			ltrec.rm_blockcount += gtrec.rm_blockcount;
+			trace_xfs_rmap_delete(mp, cur->bc_ag.pag->pag_agno,
+					gtrec.rm_startblock,
+					gtrec.rm_blockcount,
+					gtrec.rm_owner,
+					gtrec.rm_offset,
+					gtrec.rm_flags);
+			error = xfs_btree_delete(cur, &i);
+			if (error)
+				goto out_error;
+			if (XFS_IS_CORRUPT(mp, i != 1)) {
+				error = -EFSCORRUPTED;
+				goto out_error;
+			}
+		}
+
+		/* point the cursor back to the left record and update */
+		error = xfs_btree_decrement(cur, 0, &have_gt);
+		if (error)
+			goto out_error;
+		error = xfs_rmap_update(cur, &ltrec);
+		if (error)
+			goto out_error;
+	} else if (have_gt &&
+		   bno + len == gtrec.rm_startblock &&
+		   (ignore_off || offset + len == gtrec.rm_offset)) {
+		/*
+		 * right edge contiguous, merge into right record.
+		 *
+		 *                 gtbno     gtlen
+		 * Orig:             |ooooooooo|
+		 * adding: |aaaaaaaaa|
+		 * Result: |rrrrrrrrrrrrrrrrrrr|
+		 *        bno       len
+		 */
+		gtrec.rm_startblock = bno;
+		gtrec.rm_blockcount += len;
+		if (!ignore_off)
+			gtrec.rm_offset = offset;
+		error = xfs_rmap_update(cur, &gtrec);
+		if (error)
+			goto out_error;
+	} else {
+		/*
+		 * no contiguous edge with identical owner, insert
+		 * new record at current cursor position.
+		 */
+		cur->bc_rec.r.rm_startblock = bno;
+		cur->bc_rec.r.rm_blockcount = len;
+		cur->bc_rec.r.rm_owner = owner;
+		cur->bc_rec.r.rm_offset = offset;
+		cur->bc_rec.r.rm_flags = flags;
+		trace_xfs_rmap_insert(mp, cur->bc_ag.pag->pag_agno, bno, len,
+			owner, offset, flags);
+		error = xfs_btree_insert(cur, &i);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+	}
+
+	trace_xfs_rmap_map_done(mp, cur->bc_ag.pag->pag_agno, bno, len,
+			unwritten, oinfo);
+out_error:
+	if (error)
+		trace_xfs_rmap_map_error(mp, cur->bc_ag.pag->pag_agno,
+				error, _RET_IP_);
+	return error;
+}
+
+/*
+ * Add a reference to an extent in the rmap btree.
+ */
+int
+xfs_rmap_alloc(
+	struct xfs_trans		*tp,
+	struct xfs_buf			*agbp,
+	struct xfs_perag		*pag,
+	xfs_agblock_t			bno,
+	xfs_extlen_t			len,
+	const struct xfs_owner_info	*oinfo)
+{
+	struct xfs_mount		*mp = tp->t_mountp;
+	struct xfs_btree_cur		*cur;
+	int				error;
+
+	if (!xfs_has_rmapbt(mp))
+		return 0;
+
+	cur = xfs_rmapbt_init_cursor(mp, tp, agbp, pag);
+	error = xfs_rmap_map(cur, bno, len, false, oinfo);
+
+	xfs_btree_del_cursor(cur, error);
+	return error;
+}
+
+#define RMAP_LEFT_CONTIG	(1 << 0)
+#define RMAP_RIGHT_CONTIG	(1 << 1)
+#define RMAP_LEFT_FILLING	(1 << 2)
+#define RMAP_RIGHT_FILLING	(1 << 3)
+#define RMAP_LEFT_VALID		(1 << 6)
+#define RMAP_RIGHT_VALID	(1 << 7)
+
+#define LEFT		r[0]
+#define RIGHT		r[1]
+#define PREV		r[2]
+#define NEW		r[3]
+
+/*
+ * Convert an unwritten extent to a real extent or vice versa.
+ * Does not handle overlapping extents.
+ */
+STATIC int
+xfs_rmap_convert(
+	struct xfs_btree_cur		*cur,
+	xfs_agblock_t			bno,
+	xfs_extlen_t			len,
+	bool				unwritten,
+	const struct xfs_owner_info	*oinfo)
+{
+	struct xfs_mount		*mp = cur->bc_mp;
+	struct xfs_rmap_irec		r[4];	/* neighbor extent entries */
+						/* left is 0, right is 1, */
+						/* prev is 2, new is 3 */
+	uint64_t		owner;
+	uint64_t		offset;
+	uint64_t		new_endoff;
+	unsigned int		oldext;
+	unsigned int		newext;
+	unsigned int		flags = 0;
+	int			i;
+	int			state = 0;
+	int			error;
+
+	xfs_owner_info_unpack(oinfo, &owner, &offset, &flags);
+	ASSERT(!(XFS_RMAP_NON_INODE_OWNER(owner) ||
+			(flags & (XFS_RMAP_ATTR_FORK | XFS_RMAP_BMBT_BLOCK))));
+	oldext = unwritten ? XFS_RMAP_UNWRITTEN : 0;
+	new_endoff = offset + len;
+	trace_xfs_rmap_convert(mp, cur->bc_ag.pag->pag_agno, bno, len,
+			unwritten, oinfo);
+
+	/*
+	 * For the initial lookup, look for an exact match or the left-adjacent
+	 * record for our insertion point. This will also give us the record for
+	 * start block contiguity tests.
+	 */
+	error = xfs_rmap_lookup_le(cur, bno, owner, offset, oldext, &PREV, &i);
+	if (error)
+		goto done;
+	if (XFS_IS_CORRUPT(mp, i != 1)) {
+		error = -EFSCORRUPTED;
+		goto done;
+	}
+
+	trace_xfs_rmap_lookup_le_range_result(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, PREV.rm_startblock,
+			PREV.rm_blockcount, PREV.rm_owner,
+			PREV.rm_offset, PREV.rm_flags);
+
+	ASSERT(PREV.rm_offset <= offset);
+	ASSERT(PREV.rm_offset + PREV.rm_blockcount >= new_endoff);
+	ASSERT((PREV.rm_flags & XFS_RMAP_UNWRITTEN) == oldext);
+	newext = ~oldext & XFS_RMAP_UNWRITTEN;
+
+	/*
+	 * Set flags determining what part of the previous oldext allocation
+	 * extent is being replaced by a newext allocation.
+	 */
+	if (PREV.rm_offset == offset)
+		state |= RMAP_LEFT_FILLING;
+	if (PREV.rm_offset + PREV.rm_blockcount == new_endoff)
+		state |= RMAP_RIGHT_FILLING;
+
+	/*
+	 * Decrement the cursor to see if we have a left-adjacent record to our
+	 * insertion point. This will give us the record for end block
+	 * contiguity tests.
+	 */
+	error = xfs_btree_decrement(cur, 0, &i);
+	if (error)
+		goto done;
+	if (i) {
+		state |= RMAP_LEFT_VALID;
+		error = xfs_rmap_get_rec(cur, &LEFT, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		if (XFS_IS_CORRUPT(mp,
+				   LEFT.rm_startblock + LEFT.rm_blockcount >
+				   bno)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		trace_xfs_rmap_find_left_neighbor_result(cur->bc_mp,
+				cur->bc_ag.pag->pag_agno, LEFT.rm_startblock,
+				LEFT.rm_blockcount, LEFT.rm_owner,
+				LEFT.rm_offset, LEFT.rm_flags);
+		if (LEFT.rm_startblock + LEFT.rm_blockcount == bno &&
+		    LEFT.rm_offset + LEFT.rm_blockcount == offset &&
+		    xfs_rmap_is_mergeable(&LEFT, owner, newext))
+			state |= RMAP_LEFT_CONTIG;
+	}
+
+	/*
+	 * Increment the cursor to see if we have a right-adjacent record to our
+	 * insertion point. This will give us the record for end block
+	 * contiguity tests.
+	 */
+	error = xfs_btree_increment(cur, 0, &i);
+	if (error)
+		goto done;
+	if (XFS_IS_CORRUPT(mp, i != 1)) {
+		error = -EFSCORRUPTED;
+		goto done;
+	}
+	error = xfs_btree_increment(cur, 0, &i);
+	if (error)
+		goto done;
+	if (i) {
+		state |= RMAP_RIGHT_VALID;
+		error = xfs_rmap_get_rec(cur, &RIGHT, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		if (XFS_IS_CORRUPT(mp, bno + len > RIGHT.rm_startblock)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		trace_xfs_rmap_find_right_neighbor_result(cur->bc_mp,
+				cur->bc_ag.pag->pag_agno, RIGHT.rm_startblock,
+				RIGHT.rm_blockcount, RIGHT.rm_owner,
+				RIGHT.rm_offset, RIGHT.rm_flags);
+		if (bno + len == RIGHT.rm_startblock &&
+		    offset + len == RIGHT.rm_offset &&
+		    xfs_rmap_is_mergeable(&RIGHT, owner, newext))
+			state |= RMAP_RIGHT_CONTIG;
+	}
+
+	/* check that left + prev + right is not too long */
+	if ((state & (RMAP_LEFT_FILLING | RMAP_LEFT_CONTIG |
+			 RMAP_RIGHT_FILLING | RMAP_RIGHT_CONTIG)) ==
+	    (RMAP_LEFT_FILLING | RMAP_LEFT_CONTIG |
+	     RMAP_RIGHT_FILLING | RMAP_RIGHT_CONTIG) &&
+	    (unsigned long)LEFT.rm_blockcount + len +
+	     RIGHT.rm_blockcount > XFS_RMAP_LEN_MAX)
+		state &= ~RMAP_RIGHT_CONTIG;
+
+	trace_xfs_rmap_convert_state(mp, cur->bc_ag.pag->pag_agno, state,
+			_RET_IP_);
+
+	/* reset the cursor back to PREV */
+	error = xfs_rmap_lookup_le(cur, bno, owner, offset, oldext, NULL, &i);
+	if (error)
+		goto done;
+	if (XFS_IS_CORRUPT(mp, i != 1)) {
+		error = -EFSCORRUPTED;
+		goto done;
+	}
+
+	/*
+	 * Switch out based on the FILLING and CONTIG state bits.
+	 */
+	switch (state & (RMAP_LEFT_FILLING | RMAP_LEFT_CONTIG |
+			 RMAP_RIGHT_FILLING | RMAP_RIGHT_CONTIG)) {
+	case RMAP_LEFT_FILLING | RMAP_LEFT_CONTIG |
+	     RMAP_RIGHT_FILLING | RMAP_RIGHT_CONTIG:
+		/*
+		 * Setting all of a previous oldext extent to newext.
+		 * The left and right neighbors are both contiguous with new.
+		 */
+		error = xfs_btree_increment(cur, 0, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		trace_xfs_rmap_delete(mp, cur->bc_ag.pag->pag_agno,
+				RIGHT.rm_startblock, RIGHT.rm_blockcount,
+				RIGHT.rm_owner, RIGHT.rm_offset,
+				RIGHT.rm_flags);
+		error = xfs_btree_delete(cur, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		error = xfs_btree_decrement(cur, 0, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		trace_xfs_rmap_delete(mp, cur->bc_ag.pag->pag_agno,
+				PREV.rm_startblock, PREV.rm_blockcount,
+				PREV.rm_owner, PREV.rm_offset,
+				PREV.rm_flags);
+		error = xfs_btree_delete(cur, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		error = xfs_btree_decrement(cur, 0, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		NEW = LEFT;
+		NEW.rm_blockcount += PREV.rm_blockcount + RIGHT.rm_blockcount;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		break;
+
+	case RMAP_LEFT_FILLING | RMAP_RIGHT_FILLING | RMAP_LEFT_CONTIG:
+		/*
+		 * Setting all of a previous oldext extent to newext.
+		 * The left neighbor is contiguous, the right is not.
+		 */
+		trace_xfs_rmap_delete(mp, cur->bc_ag.pag->pag_agno,
+				PREV.rm_startblock, PREV.rm_blockcount,
+				PREV.rm_owner, PREV.rm_offset,
+				PREV.rm_flags);
+		error = xfs_btree_delete(cur, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		error = xfs_btree_decrement(cur, 0, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		NEW = LEFT;
+		NEW.rm_blockcount += PREV.rm_blockcount;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		break;
+
+	case RMAP_LEFT_FILLING | RMAP_RIGHT_FILLING | RMAP_RIGHT_CONTIG:
+		/*
+		 * Setting all of a previous oldext extent to newext.
+		 * The right neighbor is contiguous, the left is not.
+		 */
+		error = xfs_btree_increment(cur, 0, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		trace_xfs_rmap_delete(mp, cur->bc_ag.pag->pag_agno,
+				RIGHT.rm_startblock, RIGHT.rm_blockcount,
+				RIGHT.rm_owner, RIGHT.rm_offset,
+				RIGHT.rm_flags);
+		error = xfs_btree_delete(cur, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		error = xfs_btree_decrement(cur, 0, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		NEW = PREV;
+		NEW.rm_blockcount = len + RIGHT.rm_blockcount;
+		NEW.rm_flags = newext;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		break;
+
+	case RMAP_LEFT_FILLING | RMAP_RIGHT_FILLING:
+		/*
+		 * Setting all of a previous oldext extent to newext.
+		 * Neither the left nor right neighbors are contiguous with
+		 * the new one.
+		 */
+		NEW = PREV;
+		NEW.rm_flags = newext;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		break;
+
+	case RMAP_LEFT_FILLING | RMAP_LEFT_CONTIG:
+		/*
+		 * Setting the first part of a previous oldext extent to newext.
+		 * The left neighbor is contiguous.
+		 */
+		NEW = PREV;
+		NEW.rm_offset += len;
+		NEW.rm_startblock += len;
+		NEW.rm_blockcount -= len;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		error = xfs_btree_decrement(cur, 0, &i);
+		if (error)
+			goto done;
+		NEW = LEFT;
+		NEW.rm_blockcount += len;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		break;
+
+	case RMAP_LEFT_FILLING:
+		/*
+		 * Setting the first part of a previous oldext extent to newext.
+		 * The left neighbor is not contiguous.
+		 */
+		NEW = PREV;
+		NEW.rm_startblock += len;
+		NEW.rm_offset += len;
+		NEW.rm_blockcount -= len;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		NEW.rm_startblock = bno;
+		NEW.rm_owner = owner;
+		NEW.rm_offset = offset;
+		NEW.rm_blockcount = len;
+		NEW.rm_flags = newext;
+		cur->bc_rec.r = NEW;
+		trace_xfs_rmap_insert(mp, cur->bc_ag.pag->pag_agno, bno,
+				len, owner, offset, newext);
+		error = xfs_btree_insert(cur, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		break;
+
+	case RMAP_RIGHT_FILLING | RMAP_RIGHT_CONTIG:
+		/*
+		 * Setting the last part of a previous oldext extent to newext.
+		 * The right neighbor is contiguous with the new allocation.
+		 */
+		NEW = PREV;
+		NEW.rm_blockcount -= len;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		error = xfs_btree_increment(cur, 0, &i);
+		if (error)
+			goto done;
+		NEW = RIGHT;
+		NEW.rm_offset = offset;
+		NEW.rm_startblock = bno;
+		NEW.rm_blockcount += len;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		break;
+
+	case RMAP_RIGHT_FILLING:
+		/*
+		 * Setting the last part of a previous oldext extent to newext.
+		 * The right neighbor is not contiguous.
+		 */
+		NEW = PREV;
+		NEW.rm_blockcount -= len;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		error = xfs_rmap_lookup_eq(cur, bno, len, owner, offset,
+				oldext, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 0)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		NEW.rm_startblock = bno;
+		NEW.rm_owner = owner;
+		NEW.rm_offset = offset;
+		NEW.rm_blockcount = len;
+		NEW.rm_flags = newext;
+		cur->bc_rec.r = NEW;
+		trace_xfs_rmap_insert(mp, cur->bc_ag.pag->pag_agno, bno,
+				len, owner, offset, newext);
+		error = xfs_btree_insert(cur, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		break;
+
+	case 0:
+		/*
+		 * Setting the middle part of a previous oldext extent to
+		 * newext.  Contiguity is impossible here.
+		 * One extent becomes three extents.
+		 */
+		/* new right extent - oldext */
+		NEW.rm_startblock = bno + len;
+		NEW.rm_owner = owner;
+		NEW.rm_offset = new_endoff;
+		NEW.rm_blockcount = PREV.rm_offset + PREV.rm_blockcount -
+				new_endoff;
+		NEW.rm_flags = PREV.rm_flags;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		/* new left extent - oldext */
+		NEW = PREV;
+		NEW.rm_blockcount = offset - PREV.rm_offset;
+		cur->bc_rec.r = NEW;
+		trace_xfs_rmap_insert(mp, cur->bc_ag.pag->pag_agno,
+				NEW.rm_startblock, NEW.rm_blockcount,
+				NEW.rm_owner, NEW.rm_offset,
+				NEW.rm_flags);
+		error = xfs_btree_insert(cur, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		/*
+		 * Reset the cursor to the position of the new extent
+		 * we are about to insert as we can't trust it after
+		 * the previous insert.
+		 */
+		error = xfs_rmap_lookup_eq(cur, bno, len, owner, offset,
+				oldext, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 0)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		/* new middle extent - newext */
+		cur->bc_rec.r.rm_flags &= ~XFS_RMAP_UNWRITTEN;
+		cur->bc_rec.r.rm_flags |= newext;
+		trace_xfs_rmap_insert(mp, cur->bc_ag.pag->pag_agno, bno, len,
+				owner, offset, newext);
+		error = xfs_btree_insert(cur, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		break;
+
+	case RMAP_LEFT_FILLING | RMAP_LEFT_CONTIG | RMAP_RIGHT_CONTIG:
+	case RMAP_RIGHT_FILLING | RMAP_LEFT_CONTIG | RMAP_RIGHT_CONTIG:
+	case RMAP_LEFT_FILLING | RMAP_RIGHT_CONTIG:
+	case RMAP_RIGHT_FILLING | RMAP_LEFT_CONTIG:
+	case RMAP_LEFT_CONTIG | RMAP_RIGHT_CONTIG:
+	case RMAP_LEFT_CONTIG:
+	case RMAP_RIGHT_CONTIG:
+		/*
+		 * These cases are all impossible.
+		 */
+		ASSERT(0);
+	}
+
+	trace_xfs_rmap_convert_done(mp, cur->bc_ag.pag->pag_agno, bno, len,
+			unwritten, oinfo);
+done:
+	if (error)
+		trace_xfs_rmap_convert_error(cur->bc_mp,
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+/*
+ * Convert an unwritten extent to a real extent or vice versa.  If there is no
+ * possibility of overlapping extents, delegate to the simpler convert
+ * function.
+ */
+STATIC int
+xfs_rmap_convert_shared(
+	struct xfs_btree_cur		*cur,
+	xfs_agblock_t			bno,
+	xfs_extlen_t			len,
+	bool				unwritten,
+	const struct xfs_owner_info	*oinfo)
+{
+	struct xfs_mount		*mp = cur->bc_mp;
+	struct xfs_rmap_irec		r[4];	/* neighbor extent entries */
+						/* left is 0, right is 1, */
+						/* prev is 2, new is 3 */
+	uint64_t		owner;
+	uint64_t		offset;
+	uint64_t		new_endoff;
+	unsigned int		oldext;
+	unsigned int		newext;
+	unsigned int		flags = 0;
+	int			i;
+	int			state = 0;
+	int			error;
+
+	xfs_owner_info_unpack(oinfo, &owner, &offset, &flags);
+	ASSERT(!(XFS_RMAP_NON_INODE_OWNER(owner) ||
+			(flags & (XFS_RMAP_ATTR_FORK | XFS_RMAP_BMBT_BLOCK))));
+	oldext = unwritten ? XFS_RMAP_UNWRITTEN : 0;
+	new_endoff = offset + len;
+	trace_xfs_rmap_convert(mp, cur->bc_ag.pag->pag_agno, bno, len,
+			unwritten, oinfo);
+
+	/*
+	 * For the initial lookup, look for and exact match or the left-adjacent
+	 * record for our insertion point. This will also give us the record for
+	 * start block contiguity tests.
+	 */
+	error = xfs_rmap_lookup_le_range(cur, bno, owner, offset, oldext,
+			&PREV, &i);
+	if (error)
+		goto done;
+	if (XFS_IS_CORRUPT(mp, i != 1)) {
+		error = -EFSCORRUPTED;
+		goto done;
+	}
+
+	ASSERT(PREV.rm_offset <= offset);
+	ASSERT(PREV.rm_offset + PREV.rm_blockcount >= new_endoff);
+	ASSERT((PREV.rm_flags & XFS_RMAP_UNWRITTEN) == oldext);
+	newext = ~oldext & XFS_RMAP_UNWRITTEN;
+
+	/*
+	 * Set flags determining what part of the previous oldext allocation
+	 * extent is being replaced by a newext allocation.
+	 */
+	if (PREV.rm_offset == offset)
+		state |= RMAP_LEFT_FILLING;
+	if (PREV.rm_offset + PREV.rm_blockcount == new_endoff)
+		state |= RMAP_RIGHT_FILLING;
+
+	/* Is there a left record that abuts our range? */
+	error = xfs_rmap_find_left_neighbor(cur, bno, owner, offset, newext,
+			&LEFT, &i);
+	if (error)
+		goto done;
+	if (i) {
+		state |= RMAP_LEFT_VALID;
+		if (XFS_IS_CORRUPT(mp,
+				   LEFT.rm_startblock + LEFT.rm_blockcount >
+				   bno)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		if (xfs_rmap_is_mergeable(&LEFT, owner, newext))
+			state |= RMAP_LEFT_CONTIG;
+	}
+
+	/* Is there a right record that abuts our range? */
+	error = xfs_rmap_lookup_eq(cur, bno + len, len, owner, offset + len,
+			newext, &i);
+	if (error)
+		goto done;
+	if (i) {
+		state |= RMAP_RIGHT_VALID;
+		error = xfs_rmap_get_rec(cur, &RIGHT, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		if (XFS_IS_CORRUPT(mp, bno + len > RIGHT.rm_startblock)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		trace_xfs_rmap_find_right_neighbor_result(cur->bc_mp,
+				cur->bc_ag.pag->pag_agno, RIGHT.rm_startblock,
+				RIGHT.rm_blockcount, RIGHT.rm_owner,
+				RIGHT.rm_offset, RIGHT.rm_flags);
+		if (xfs_rmap_is_mergeable(&RIGHT, owner, newext))
+			state |= RMAP_RIGHT_CONTIG;
+	}
+
+	/* check that left + prev + right is not too long */
+	if ((state & (RMAP_LEFT_FILLING | RMAP_LEFT_CONTIG |
+			 RMAP_RIGHT_FILLING | RMAP_RIGHT_CONTIG)) ==
+	    (RMAP_LEFT_FILLING | RMAP_LEFT_CONTIG |
+	     RMAP_RIGHT_FILLING | RMAP_RIGHT_CONTIG) &&
+	    (unsigned long)LEFT.rm_blockcount + len +
+	     RIGHT.rm_blockcount > XFS_RMAP_LEN_MAX)
+		state &= ~RMAP_RIGHT_CONTIG;
+
+	trace_xfs_rmap_convert_state(mp, cur->bc_ag.pag->pag_agno, state,
+			_RET_IP_);
+	/*
+	 * Switch out based on the FILLING and CONTIG state bits.
+	 */
+	switch (state & (RMAP_LEFT_FILLING | RMAP_LEFT_CONTIG |
+			 RMAP_RIGHT_FILLING | RMAP_RIGHT_CONTIG)) {
+	case RMAP_LEFT_FILLING | RMAP_LEFT_CONTIG |
+	     RMAP_RIGHT_FILLING | RMAP_RIGHT_CONTIG:
+		/*
+		 * Setting all of a previous oldext extent to newext.
+		 * The left and right neighbors are both contiguous with new.
+		 */
+		error = xfs_rmap_delete(cur, RIGHT.rm_startblock,
+				RIGHT.rm_blockcount, RIGHT.rm_owner,
+				RIGHT.rm_offset, RIGHT.rm_flags);
+		if (error)
+			goto done;
+		error = xfs_rmap_delete(cur, PREV.rm_startblock,
+				PREV.rm_blockcount, PREV.rm_owner,
+				PREV.rm_offset, PREV.rm_flags);
+		if (error)
+			goto done;
+		NEW = LEFT;
+		error = xfs_rmap_lookup_eq(cur, NEW.rm_startblock,
+				NEW.rm_blockcount, NEW.rm_owner,
+				NEW.rm_offset, NEW.rm_flags, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		NEW.rm_blockcount += PREV.rm_blockcount + RIGHT.rm_blockcount;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		break;
+
+	case RMAP_LEFT_FILLING | RMAP_RIGHT_FILLING | RMAP_LEFT_CONTIG:
+		/*
+		 * Setting all of a previous oldext extent to newext.
+		 * The left neighbor is contiguous, the right is not.
+		 */
+		error = xfs_rmap_delete(cur, PREV.rm_startblock,
+				PREV.rm_blockcount, PREV.rm_owner,
+				PREV.rm_offset, PREV.rm_flags);
+		if (error)
+			goto done;
+		NEW = LEFT;
+		error = xfs_rmap_lookup_eq(cur, NEW.rm_startblock,
+				NEW.rm_blockcount, NEW.rm_owner,
+				NEW.rm_offset, NEW.rm_flags, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		NEW.rm_blockcount += PREV.rm_blockcount;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		break;
+
+	case RMAP_LEFT_FILLING | RMAP_RIGHT_FILLING | RMAP_RIGHT_CONTIG:
+		/*
+		 * Setting all of a previous oldext extent to newext.
+		 * The right neighbor is contiguous, the left is not.
+		 */
+		error = xfs_rmap_delete(cur, RIGHT.rm_startblock,
+				RIGHT.rm_blockcount, RIGHT.rm_owner,
+				RIGHT.rm_offset, RIGHT.rm_flags);
+		if (error)
+			goto done;
+		NEW = PREV;
+		error = xfs_rmap_lookup_eq(cur, NEW.rm_startblock,
+				NEW.rm_blockcount, NEW.rm_owner,
+				NEW.rm_offset, NEW.rm_flags, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		NEW.rm_blockcount += RIGHT.rm_blockcount;
+		NEW.rm_flags = RIGHT.rm_flags;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		break;
+
+	case RMAP_LEFT_FILLING | RMAP_RIGHT_FILLING:
+		/*
+		 * Setting all of a previous oldext extent to newext.
+		 * Neither the left nor right neighbors are contiguous with
+		 * the new one.
+		 */
+		NEW = PREV;
+		error = xfs_rmap_lookup_eq(cur, NEW.rm_startblock,
+				NEW.rm_blockcount, NEW.rm_owner,
+				NEW.rm_offset, NEW.rm_flags, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		NEW.rm_flags = newext;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		break;
+
+	case RMAP_LEFT_FILLING | RMAP_LEFT_CONTIG:
+		/*
+		 * Setting the first part of a previous oldext extent to newext.
+		 * The left neighbor is contiguous.
+		 */
+		NEW = PREV;
+		error = xfs_rmap_delete(cur, NEW.rm_startblock,
+				NEW.rm_blockcount, NEW.rm_owner,
+				NEW.rm_offset, NEW.rm_flags);
+		if (error)
+			goto done;
+		NEW.rm_offset += len;
+		NEW.rm_startblock += len;
+		NEW.rm_blockcount -= len;
+		error = xfs_rmap_insert(cur, NEW.rm_startblock,
+				NEW.rm_blockcount, NEW.rm_owner,
+				NEW.rm_offset, NEW.rm_flags);
+		if (error)
+			goto done;
+		NEW = LEFT;
+		error = xfs_rmap_lookup_eq(cur, NEW.rm_startblock,
+				NEW.rm_blockcount, NEW.rm_owner,
+				NEW.rm_offset, NEW.rm_flags, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		NEW.rm_blockcount += len;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		break;
+
+	case RMAP_LEFT_FILLING:
+		/*
+		 * Setting the first part of a previous oldext extent to newext.
+		 * The left neighbor is not contiguous.
+		 */
+		NEW = PREV;
+		error = xfs_rmap_delete(cur, NEW.rm_startblock,
+				NEW.rm_blockcount, NEW.rm_owner,
+				NEW.rm_offset, NEW.rm_flags);
+		if (error)
+			goto done;
+		NEW.rm_offset += len;
+		NEW.rm_startblock += len;
+		NEW.rm_blockcount -= len;
+		error = xfs_rmap_insert(cur, NEW.rm_startblock,
+				NEW.rm_blockcount, NEW.rm_owner,
+				NEW.rm_offset, NEW.rm_flags);
+		if (error)
+			goto done;
+		error = xfs_rmap_insert(cur, bno, len, owner, offset, newext);
+		if (error)
+			goto done;
+		break;
+
+	case RMAP_RIGHT_FILLING | RMAP_RIGHT_CONTIG:
+		/*
+		 * Setting the last part of a previous oldext extent to newext.
+		 * The right neighbor is contiguous with the new allocation.
+		 */
+		NEW = PREV;
+		error = xfs_rmap_lookup_eq(cur, NEW.rm_startblock,
+				NEW.rm_blockcount, NEW.rm_owner,
+				NEW.rm_offset, NEW.rm_flags, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		NEW.rm_blockcount = offset - NEW.rm_offset;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		NEW = RIGHT;
+		error = xfs_rmap_delete(cur, NEW.rm_startblock,
+				NEW.rm_blockcount, NEW.rm_owner,
+				NEW.rm_offset, NEW.rm_flags);
+		if (error)
+			goto done;
+		NEW.rm_offset = offset;
+		NEW.rm_startblock = bno;
+		NEW.rm_blockcount += len;
+		error = xfs_rmap_insert(cur, NEW.rm_startblock,
+				NEW.rm_blockcount, NEW.rm_owner,
+				NEW.rm_offset, NEW.rm_flags);
+		if (error)
+			goto done;
+		break;
+
+	case RMAP_RIGHT_FILLING:
+		/*
+		 * Setting the last part of a previous oldext extent to newext.
+		 * The right neighbor is not contiguous.
+		 */
+		NEW = PREV;
+		error = xfs_rmap_lookup_eq(cur, NEW.rm_startblock,
+				NEW.rm_blockcount, NEW.rm_owner,
+				NEW.rm_offset, NEW.rm_flags, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		NEW.rm_blockcount -= len;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		error = xfs_rmap_insert(cur, bno, len, owner, offset, newext);
+		if (error)
+			goto done;
+		break;
+
+	case 0:
+		/*
+		 * Setting the middle part of a previous oldext extent to
+		 * newext.  Contiguity is impossible here.
+		 * One extent becomes three extents.
+		 */
+		/* new right extent - oldext */
+		NEW.rm_startblock = bno + len;
+		NEW.rm_owner = owner;
+		NEW.rm_offset = new_endoff;
+		NEW.rm_blockcount = PREV.rm_offset + PREV.rm_blockcount -
+				new_endoff;
+		NEW.rm_flags = PREV.rm_flags;
+		error = xfs_rmap_insert(cur, NEW.rm_startblock,
+				NEW.rm_blockcount, NEW.rm_owner, NEW.rm_offset,
+				NEW.rm_flags);
+		if (error)
+			goto done;
+		/* new left extent - oldext */
+		NEW = PREV;
+		error = xfs_rmap_lookup_eq(cur, NEW.rm_startblock,
+				NEW.rm_blockcount, NEW.rm_owner,
+				NEW.rm_offset, NEW.rm_flags, &i);
+		if (error)
+			goto done;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto done;
+		}
+		NEW.rm_blockcount = offset - NEW.rm_offset;
+		error = xfs_rmap_update(cur, &NEW);
+		if (error)
+			goto done;
+		/* new middle extent - newext */
+		NEW.rm_startblock = bno;
+		NEW.rm_blockcount = len;
+		NEW.rm_owner = owner;
+		NEW.rm_offset = offset;
+		NEW.rm_flags = newext;
+		error = xfs_rmap_insert(cur, NEW.rm_startblock,
+				NEW.rm_blockcount, NEW.rm_owner, NEW.rm_offset,
+				NEW.rm_flags);
+		if (error)
+			goto done;
+		break;
+
+	case RMAP_LEFT_FILLING | RMAP_LEFT_CONTIG | RMAP_RIGHT_CONTIG:
+	case RMAP_RIGHT_FILLING | RMAP_LEFT_CONTIG | RMAP_RIGHT_CONTIG:
+	case RMAP_LEFT_FILLING | RMAP_RIGHT_CONTIG:
+	case RMAP_RIGHT_FILLING | RMAP_LEFT_CONTIG:
+	case RMAP_LEFT_CONTIG | RMAP_RIGHT_CONTIG:
+	case RMAP_LEFT_CONTIG:
+	case RMAP_RIGHT_CONTIG:
+		/*
+		 * These cases are all impossible.
+		 */
+		ASSERT(0);
+	}
+
+	trace_xfs_rmap_convert_done(mp, cur->bc_ag.pag->pag_agno, bno, len,
+			unwritten, oinfo);
+done:
+	if (error)
+		trace_xfs_rmap_convert_error(cur->bc_mp,
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+#undef	NEW
+#undef	LEFT
+#undef	RIGHT
+#undef	PREV
+
+/*
+ * Find an extent in the rmap btree and unmap it.  For rmap extent types that
+ * can overlap (data fork rmaps on reflink filesystems) we must be careful
+ * that the prev/next records in the btree might belong to another owner.
+ * Therefore we must use delete+insert to alter any of the key fields.
+ *
+ * For every other situation there can only be one owner for a given extent,
+ * so we can call the regular _free function.
+ */
+STATIC int
+xfs_rmap_unmap_shared(
+	struct xfs_btree_cur		*cur,
+	xfs_agblock_t			bno,
+	xfs_extlen_t			len,
+	bool				unwritten,
+	const struct xfs_owner_info	*oinfo)
+{
+	struct xfs_mount		*mp = cur->bc_mp;
+	struct xfs_rmap_irec		ltrec;
+	uint64_t			ltoff;
+	int				error = 0;
+	int				i;
+	uint64_t			owner;
+	uint64_t			offset;
+	unsigned int			flags;
+
+	xfs_owner_info_unpack(oinfo, &owner, &offset, &flags);
+	if (unwritten)
+		flags |= XFS_RMAP_UNWRITTEN;
+	trace_xfs_rmap_unmap(mp, cur->bc_ag.pag->pag_agno, bno, len,
+			unwritten, oinfo);
+
+	/*
+	 * We should always have a left record because there's a static record
+	 * for the AG headers at rm_startblock == 0 created by mkfs/growfs that
+	 * will not ever be removed from the tree.
+	 */
+	error = xfs_rmap_lookup_le_range(cur, bno, owner, offset, flags,
+			&ltrec, &i);
+	if (error)
+		goto out_error;
+	if (XFS_IS_CORRUPT(mp, i != 1)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+	ltoff = ltrec.rm_offset;
+
+	/* Make sure the extent we found covers the entire freeing range. */
+	if (XFS_IS_CORRUPT(mp,
+			   ltrec.rm_startblock > bno ||
+			   ltrec.rm_startblock + ltrec.rm_blockcount <
+			   bno + len)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+
+	/* Make sure the owner matches what we expect to find in the tree. */
+	if (XFS_IS_CORRUPT(mp, owner != ltrec.rm_owner)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+
+	/* Make sure the unwritten flag matches. */
+	if (XFS_IS_CORRUPT(mp,
+			   (flags & XFS_RMAP_UNWRITTEN) !=
+			   (ltrec.rm_flags & XFS_RMAP_UNWRITTEN))) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+
+	/* Check the offset. */
+	if (XFS_IS_CORRUPT(mp, ltrec.rm_offset > offset)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+	if (XFS_IS_CORRUPT(mp, offset > ltoff + ltrec.rm_blockcount)) {
+		error = -EFSCORRUPTED;
+		goto out_error;
+	}
+
+	if (ltrec.rm_startblock == bno && ltrec.rm_blockcount == len) {
+		/* Exact match, simply remove the record from rmap tree. */
+		error = xfs_rmap_delete(cur, ltrec.rm_startblock,
+				ltrec.rm_blockcount, ltrec.rm_owner,
+				ltrec.rm_offset, ltrec.rm_flags);
+		if (error)
+			goto out_error;
+	} else if (ltrec.rm_startblock == bno) {
+		/*
+		 * Overlap left hand side of extent: move the start, trim the
+		 * length and update the current record.
+		 *
+		 *       ltbno                ltlen
+		 * Orig:    |oooooooooooooooooooo|
+		 * Freeing: |fffffffff|
+		 * Result:            |rrrrrrrrrr|
+		 *         bno       len
+		 */
+
+		/* Delete prev rmap. */
+		error = xfs_rmap_delete(cur, ltrec.rm_startblock,
+				ltrec.rm_blockcount, ltrec.rm_owner,
+				ltrec.rm_offset, ltrec.rm_flags);
+		if (error)
+			goto out_error;
+
+		/* Add an rmap at the new offset. */
+		ltrec.rm_startblock += len;
+		ltrec.rm_blockcount -= len;
+		ltrec.rm_offset += len;
+		error = xfs_rmap_insert(cur, ltrec.rm_startblock,
+				ltrec.rm_blockcount, ltrec.rm_owner,
+				ltrec.rm_offset, ltrec.rm_flags);
+		if (error)
+			goto out_error;
+	} else if (ltrec.rm_startblock + ltrec.rm_blockcount == bno + len) {
+		/*
+		 * Overlap right hand side of extent: trim the length and
+		 * update the current record.
+		 *
+		 *       ltbno                ltlen
+		 * Orig:    |oooooooooooooooooooo|
+		 * Freeing:            |fffffffff|
+		 * Result:  |rrrrrrrrrr|
+		 *                    bno       len
+		 */
+		error = xfs_rmap_lookup_eq(cur, ltrec.rm_startblock,
+				ltrec.rm_blockcount, ltrec.rm_owner,
+				ltrec.rm_offset, ltrec.rm_flags, &i);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+		ltrec.rm_blockcount -= len;
+		error = xfs_rmap_update(cur, &ltrec);
+		if (error)
+			goto out_error;
+	} else {
+		/*
+		 * Overlap middle of extent: trim the length of the existing
+		 * record to the length of the new left-extent size, increment
+		 * the insertion position so we can insert a new record
+		 * containing the remaining right-extent space.
+		 *
+		 *       ltbno                ltlen
+		 * Orig:    |oooooooooooooooooooo|
+		 * Freeing:       |fffffffff|
+		 * Result:  |rrrrr|         |rrrr|
+		 *               bno       len
+		 */
+		xfs_extlen_t	orig_len = ltrec.rm_blockcount;
+
+		/* Shrink the left side of the rmap */
+		error = xfs_rmap_lookup_eq(cur, ltrec.rm_startblock,
+				ltrec.rm_blockcount, ltrec.rm_owner,
+				ltrec.rm_offset, ltrec.rm_flags, &i);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+		ltrec.rm_blockcount = bno - ltrec.rm_startblock;
+		error = xfs_rmap_update(cur, &ltrec);
+		if (error)
+			goto out_error;
+
+		/* Add an rmap at the new offset */
+		error = xfs_rmap_insert(cur, bno + len,
+				orig_len - len - ltrec.rm_blockcount,
+				ltrec.rm_owner, offset + len,
+				ltrec.rm_flags);
+		if (error)
+			goto out_error;
+	}
+
+	trace_xfs_rmap_unmap_done(mp, cur->bc_ag.pag->pag_agno, bno, len,
+			unwritten, oinfo);
+out_error:
+	if (error)
+		trace_xfs_rmap_unmap_error(cur->bc_mp,
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+/*
+ * Find an extent in the rmap btree and map it.  For rmap extent types that
+ * can overlap (data fork rmaps on reflink filesystems) we must be careful
+ * that the prev/next records in the btree might belong to another owner.
+ * Therefore we must use delete+insert to alter any of the key fields.
+ *
+ * For every other situation there can only be one owner for a given extent,
+ * so we can call the regular _alloc function.
+ */
+STATIC int
+xfs_rmap_map_shared(
+	struct xfs_btree_cur		*cur,
+	xfs_agblock_t			bno,
+	xfs_extlen_t			len,
+	bool				unwritten,
+	const struct xfs_owner_info	*oinfo)
+{
+	struct xfs_mount		*mp = cur->bc_mp;
+	struct xfs_rmap_irec		ltrec;
+	struct xfs_rmap_irec		gtrec;
+	int				have_gt;
+	int				have_lt;
+	int				error = 0;
+	int				i;
+	uint64_t			owner;
+	uint64_t			offset;
+	unsigned int			flags = 0;
+
+	xfs_owner_info_unpack(oinfo, &owner, &offset, &flags);
+	if (unwritten)
+		flags |= XFS_RMAP_UNWRITTEN;
+	trace_xfs_rmap_map(mp, cur->bc_ag.pag->pag_agno, bno, len,
+			unwritten, oinfo);
+
+	/* Is there a left record that abuts our range? */
+	error = xfs_rmap_find_left_neighbor(cur, bno, owner, offset, flags,
+			&ltrec, &have_lt);
+	if (error)
+		goto out_error;
+	if (have_lt &&
+	    !xfs_rmap_is_mergeable(&ltrec, owner, flags))
+		have_lt = 0;
+
+	/* Is there a right record that abuts our range? */
+	error = xfs_rmap_lookup_eq(cur, bno + len, len, owner, offset + len,
+			flags, &have_gt);
+	if (error)
+		goto out_error;
+	if (have_gt) {
+		error = xfs_rmap_get_rec(cur, &gtrec, &have_gt);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(mp, have_gt != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+		trace_xfs_rmap_find_right_neighbor_result(cur->bc_mp,
+			cur->bc_ag.pag->pag_agno, gtrec.rm_startblock,
+			gtrec.rm_blockcount, gtrec.rm_owner,
+			gtrec.rm_offset, gtrec.rm_flags);
+
+		if (!xfs_rmap_is_mergeable(&gtrec, owner, flags))
+			have_gt = 0;
+	}
+
+	if (have_lt &&
+	    ltrec.rm_startblock + ltrec.rm_blockcount == bno &&
+	    ltrec.rm_offset + ltrec.rm_blockcount == offset) {
+		/*
+		 * Left edge contiguous, merge into left record.
+		 *
+		 *       ltbno     ltlen
+		 * orig:   |ooooooooo|
+		 * adding:           |aaaaaaaaa|
+		 * result: |rrrrrrrrrrrrrrrrrrr|
+		 *                  bno       len
+		 */
+		ltrec.rm_blockcount += len;
+		if (have_gt &&
+		    bno + len == gtrec.rm_startblock &&
+		    offset + len == gtrec.rm_offset) {
+			/*
+			 * Right edge also contiguous, delete right record
+			 * and merge into left record.
+			 *
+			 *       ltbno     ltlen    gtbno     gtlen
+			 * orig:   |ooooooooo|         |ooooooooo|
+			 * adding:           |aaaaaaaaa|
+			 * result: |rrrrrrrrrrrrrrrrrrrrrrrrrrrrr|
+			 */
+			ltrec.rm_blockcount += gtrec.rm_blockcount;
+			error = xfs_rmap_delete(cur, gtrec.rm_startblock,
+					gtrec.rm_blockcount, gtrec.rm_owner,
+					gtrec.rm_offset, gtrec.rm_flags);
+			if (error)
+				goto out_error;
+		}
+
+		/* Point the cursor back to the left record and update. */
+		error = xfs_rmap_lookup_eq(cur, ltrec.rm_startblock,
+				ltrec.rm_blockcount, ltrec.rm_owner,
+				ltrec.rm_offset, ltrec.rm_flags, &i);
+		if (error)
+			goto out_error;
+		if (XFS_IS_CORRUPT(mp, i != 1)) {
+			error = -EFSCORRUPTED;
+			goto out_error;
+		}
+
+		error = xfs_rmap_update(cur, &ltrec);
+		if (error)
+			goto out_error;
+	} else if (have_gt &&
+		   bno + len == gtrec.rm_startblock &&
+		   offset + len == gtrec.rm_offset) {
+		/*
+		 * Right edge contiguous, merge into right record.
+		 *
+		 *                 gtbno     gtlen
+		 * Orig:             |ooooooooo|
+		 * adding: |aaaaaaaaa|
+		 * Result: |rrrrrrrrrrrrrrrrrrr|
+		 *        bno       len
+		 */
+		/* Delete the old record. */
+		error = xfs_rmap_delete(cur, gtrec.rm_startblock,
+				gtrec.rm_blockcount, gtrec.rm_owner,
+				gtrec.rm_offset, gtrec.rm_flags);
+		if (error)
+			goto out_error;
+
+		/* Move the start and re-add it. */
+		gtrec.rm_startblock = bno;
+		gtrec.rm_blockcount += len;
+		gtrec.rm_offset = offset;
+		error = xfs_rmap_insert(cur, gtrec.rm_startblock,
+				gtrec.rm_blockcount, gtrec.rm_owner,
+				gtrec.rm_offset, gtrec.rm_flags);
+		if (error)
+			goto out_error;
+	} else {
+		/*
+		 * No contiguous edge with identical owner, insert
+		 * new record at current cursor position.
+		 */
+		error = xfs_rmap_insert(cur, bno, len, owner, offset, flags);
+		if (error)
+			goto out_error;
+	}
+
+	trace_xfs_rmap_map_done(mp, cur->bc_ag.pag->pag_agno, bno, len,
+			unwritten, oinfo);
+out_error:
+	if (error)
+		trace_xfs_rmap_map_error(cur->bc_mp,
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
+	return error;
+}
+
+/* Insert a raw rmap into the rmapbt. */
+int
+xfs_rmap_map_raw(
+	struct xfs_btree_cur	*cur,
+	struct xfs_rmap_irec	*rmap)
+{
+	struct xfs_owner_info	oinfo;
+
+	oinfo.oi_owner = rmap->rm_owner;
+	oinfo.oi_offset = rmap->rm_offset;
+	oinfo.oi_flags = 0;
+	if (rmap->rm_flags & XFS_RMAP_ATTR_FORK)
+		oinfo.oi_flags |= XFS_OWNER_INFO_ATTR_FORK;
+	if (rmap->rm_flags & XFS_RMAP_BMBT_BLOCK)
+		oinfo.oi_flags |= XFS_OWNER_INFO_BMBT_BLOCK;
+
+	if (rmap->rm_flags || XFS_RMAP_NON_INODE_OWNER(rmap->rm_owner))
+		return xfs_rmap_map(cur, rmap->rm_startblock,
+				rmap->rm_blockcount,
+				rmap->rm_flags & XFS_RMAP_UNWRITTEN,
+				&oinfo);
+
+	return xfs_rmap_map_shared(cur, rmap->rm_startblock,
+			rmap->rm_blockcount,
+			rmap->rm_flags & XFS_RMAP_UNWRITTEN,
+			&oinfo);
+}
+
+struct xfs_rmap_query_range_info {
+	xfs_rmap_query_range_fn	fn;
+	void				*priv;
+};
+
+/* Format btree record and pass to our callback. */
+STATIC int
+xfs_rmap_query_range_helper(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_rec	*rec,
+	void				*priv)
+{
+	struct xfs_rmap_query_range_info	*query = priv;
+	struct xfs_rmap_irec			irec;
+	int					error;
+
+	error = xfs_rmap_btrec_to_irec(rec, &irec);
+	if (error)
+		return error;
+	return query->fn(cur, &irec, query->priv);
+}
+
+/* Find all rmaps between two keys. */
+int
+xfs_rmap_query_range(
+	struct xfs_btree_cur			*cur,
+	const struct xfs_rmap_irec		*low_rec,
+	const struct xfs_rmap_irec		*high_rec,
+	xfs_rmap_query_range_fn			fn,
+	void					*priv)
+{
+	union xfs_btree_irec			low_brec;
+	union xfs_btree_irec			high_brec;
+	struct xfs_rmap_query_range_info	query;
+
+	low_brec.r = *low_rec;
+	high_brec.r = *high_rec;
+	query.priv = priv;
+	query.fn = fn;
+	return xfs_btree_query_range(cur, &low_brec, &high_brec,
+			xfs_rmap_query_range_helper, &query);
+}
+
+/* Find all rmaps. */
+int
+xfs_rmap_query_all(
+	struct xfs_btree_cur			*cur,
+	xfs_rmap_query_range_fn			fn,
+	void					*priv)
+{
+	struct xfs_rmap_query_range_info	query;
+
+	query.priv = priv;
+	query.fn = fn;
+	return xfs_btree_query_all(cur, xfs_rmap_query_range_helper, &query);
+}
+
+/* Clean up after calling xfs_rmap_finish_one. */
+void
+xfs_rmap_finish_one_cleanup(
+	struct xfs_trans	*tp,
+	struct xfs_btree_cur	*rcur,
+	int			error)
+{
+	struct xfs_buf		*agbp;
+
+	if (rcur == NULL)
+		return;
+	agbp = rcur->bc_ag.agbp;
+	xfs_btree_del_cursor(rcur, error);
+	if (error)
+		xfs_trans_brelse(tp, agbp);
+}
+
+/*
+ * Process one of the deferred rmap operations.  We pass back the
+ * btree cursor to maintain our lock on the rmapbt between calls.
+ * This saves time and eliminates a buffer deadlock between the
+ * superblock and the AGF because we'll always grab them in the same
+ * order.
+ */
+int
+xfs_rmap_finish_one(
+	struct xfs_trans		*tp,
+	enum xfs_rmap_intent_type	type,
+	uint64_t			owner,
+	int				whichfork,
+	xfs_fileoff_t			startoff,
+	xfs_fsblock_t			startblock,
+	xfs_filblks_t			blockcount,
+	xfs_exntst_t			state,
+	struct xfs_btree_cur		**pcur)
+{
+	struct xfs_mount		*mp = tp->t_mountp;
+	struct xfs_perag		*pag;
+	struct xfs_btree_cur		*rcur;
+	struct xfs_buf			*agbp = NULL;
+	int				error = 0;
+	struct xfs_owner_info		oinfo;
+	xfs_agblock_t			bno;
+	bool				unwritten;
+
+	pag = xfs_perag_get(mp, XFS_FSB_TO_AGNO(mp, startblock));
+	bno = XFS_FSB_TO_AGBNO(mp, startblock);
+
+	trace_xfs_rmap_deferred(mp, pag->pag_agno, type, bno, owner, whichfork,
+			startoff, blockcount, state);
+
+	if (XFS_TEST_ERROR(false, mp, XFS_ERRTAG_RMAP_FINISH_ONE)) {
+		error = -EIO;
+		goto out_drop;
+	}
+
+
+	/*
+	 * If we haven't gotten a cursor or the cursor AG doesn't match
+	 * the startblock, get one now.
+	 */
+	rcur = *pcur;
+	if (rcur != NULL && rcur->bc_ag.pag != pag) {
+		xfs_rmap_finish_one_cleanup(tp, rcur, 0);
+		rcur = NULL;
+		*pcur = NULL;
+	}
+	if (rcur == NULL) {
+		/*
+		 * Refresh the freelist before we start changing the
+		 * rmapbt, because a shape change could cause us to
+		 * allocate blocks.
+		 */
+		error = xfs_free_extent_fix_freelist(tp, pag, &agbp);
+		if (error)
+			goto out_drop;
+		if (XFS_IS_CORRUPT(tp->t_mountp, !agbp)) {
+			error = -EFSCORRUPTED;
+			goto out_drop;
+		}
+
+		rcur = xfs_rmapbt_init_cursor(mp, tp, agbp, pag);
+	}
+	*pcur = rcur;
+
+	xfs_rmap_ino_owner(&oinfo, owner, whichfork, startoff);
+	unwritten = state == XFS_EXT_UNWRITTEN;
+	bno = XFS_FSB_TO_AGBNO(rcur->bc_mp, startblock);
+
+	switch (type) {
+	case XFS_RMAP_ALLOC:
+	case XFS_RMAP_MAP:
+		error = xfs_rmap_map(rcur, bno, blockcount, unwritten, &oinfo);
+		break;
+	case XFS_RMAP_MAP_SHARED:
+		error = xfs_rmap_map_shared(rcur, bno, blockcount, unwritten,
+				&oinfo);
+		break;
+	case XFS_RMAP_FREE:
+	case XFS_RMAP_UNMAP:
+		error = xfs_rmap_unmap(rcur, bno, blockcount, unwritten,
+				&oinfo);
+		break;
+	case XFS_RMAP_UNMAP_SHARED:
+		error = xfs_rmap_unmap_shared(rcur, bno, blockcount, unwritten,
+				&oinfo);
+		break;
+	case XFS_RMAP_CONVERT:
+		error = xfs_rmap_convert(rcur, bno, blockcount, !unwritten,
+				&oinfo);
+		break;
+	case XFS_RMAP_CONVERT_SHARED:
+		error = xfs_rmap_convert_shared(rcur, bno, blockcount,
+				!unwritten, &oinfo);
+		break;
+	default:
+		ASSERT(0);
+		error = -EFSCORRUPTED;
+	}
+out_drop:
+	xfs_perag_put(pag);
+	return error;
+}
+
+/*
+ * Don't defer an rmap if we aren't an rmap filesystem.
+ */
+static bool
+xfs_rmap_update_is_needed(
+	struct xfs_mount	*mp,
+	int			whichfork)
+{
+	return xfs_has_rmapbt(mp) && whichfork != XFS_COW_FORK;
+}
+
+/*
+ * Record a rmap intent; the list is kept sorted first by AG and then by
+ * increasing age.
+ */
+static void
+__xfs_rmap_add(
+	struct xfs_trans		*tp,
+	enum xfs_rmap_intent_type	type,
+	uint64_t			owner,
+	int				whichfork,
+	struct xfs_bmbt_irec		*bmap)
+{
+	struct xfs_rmap_intent		*ri;
+
+	trace_xfs_rmap_defer(tp->t_mountp,
+			XFS_FSB_TO_AGNO(tp->t_mountp, bmap->br_startblock),
+			type,
+			XFS_FSB_TO_AGBNO(tp->t_mountp, bmap->br_startblock),
+			owner, whichfork,
+			bmap->br_startoff,
+			bmap->br_blockcount,
+			bmap->br_state);
+
+	ri = kmem_cache_alloc(xfs_rmap_intent_cache, GFP_NOFS | __GFP_NOFAIL);
+	INIT_LIST_HEAD(&ri->ri_list);
+	ri->ri_type = type;
+	ri->ri_owner = owner;
+	ri->ri_whichfork = whichfork;
+	ri->ri_bmap = *bmap;
+
+	xfs_defer_add(tp, XFS_DEFER_OPS_TYPE_RMAP, &ri->ri_list);
+}
+
+/* Map an extent into a file. */
+void
+xfs_rmap_map_extent(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	int			whichfork,
+	struct xfs_bmbt_irec	*PREV)
+{
+	enum xfs_rmap_intent_type type = XFS_RMAP_MAP;
+
+	if (!xfs_rmap_update_is_needed(tp->t_mountp, whichfork))
+		return;
+
+	if (whichfork != XFS_ATTR_FORK && xfs_is_reflink_inode(ip))
+		type = XFS_RMAP_MAP_SHARED;
+
+	__xfs_rmap_add(tp, type, ip->i_ino, whichfork, PREV);
+}
+
+/* Unmap an extent out of a file. */
+void
+xfs_rmap_unmap_extent(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	int			whichfork,
+	struct xfs_bmbt_irec	*PREV)
+{
+	enum xfs_rmap_intent_type type = XFS_RMAP_UNMAP;
+
+	if (!xfs_rmap_update_is_needed(tp->t_mountp, whichfork))
+		return;
+
+	if (whichfork != XFS_ATTR_FORK && xfs_is_reflink_inode(ip))
+		type = XFS_RMAP_UNMAP_SHARED;
+
+	__xfs_rmap_add(tp, type, ip->i_ino, whichfork, PREV);
+}
+
+/*
+ * Convert a data fork extent from unwritten to real or vice versa.
+ *
+ * Note that tp can be NULL here as no transaction is used for COW fork
+ * unwritten conversion.
+ */
+void
+xfs_rmap_convert_extent(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	int			whichfork,
+	struct xfs_bmbt_irec	*PREV)
+{
+	enum xfs_rmap_intent_type type = XFS_RMAP_CONVERT;
+
+	if (!xfs_rmap_update_is_needed(mp, whichfork))
+		return;
+
+	if (whichfork != XFS_ATTR_FORK && xfs_is_reflink_inode(ip))
+		type = XFS_RMAP_CONVERT_SHARED;
+
+	__xfs_rmap_add(tp, type, ip->i_ino, whichfork, PREV);
+}
+
+/* Schedule the creation of an rmap for non-file data. */
+void
+xfs_rmap_alloc_extent(
+	struct xfs_trans	*tp,
+	xfs_agnumber_t		agno,
+	xfs_agblock_t		bno,
+	xfs_extlen_t		len,
+	uint64_t		owner)
+{
+	struct xfs_bmbt_irec	bmap;
+
+	if (!xfs_rmap_update_is_needed(tp->t_mountp, XFS_DATA_FORK))
+		return;
+
+	bmap.br_startblock = XFS_AGB_TO_FSB(tp->t_mountp, agno, bno);
+	bmap.br_blockcount = len;
+	bmap.br_startoff = 0;
+	bmap.br_state = XFS_EXT_NORM;
+
+	__xfs_rmap_add(tp, XFS_RMAP_ALLOC, owner, XFS_DATA_FORK, &bmap);
+}
+
+/* Schedule the deletion of an rmap for non-file data. */
+void
+xfs_rmap_free_extent(
+	struct xfs_trans	*tp,
+	xfs_agnumber_t		agno,
+	xfs_agblock_t		bno,
+	xfs_extlen_t		len,
+	uint64_t		owner)
+{
+	struct xfs_bmbt_irec	bmap;
+
+	if (!xfs_rmap_update_is_needed(tp->t_mountp, XFS_DATA_FORK))
+		return;
+
+	bmap.br_startblock = XFS_AGB_TO_FSB(tp->t_mountp, agno, bno);
+	bmap.br_blockcount = len;
+	bmap.br_startoff = 0;
+	bmap.br_state = XFS_EXT_NORM;
+
+	__xfs_rmap_add(tp, XFS_RMAP_FREE, owner, XFS_DATA_FORK, &bmap);
+}
+
+/* Compare rmap records.  Returns -1 if a < b, 1 if a > b, and 0 if equal. */
+int
+xfs_rmap_compare(
+	const struct xfs_rmap_irec	*a,
+	const struct xfs_rmap_irec	*b)
+{
+	__u64				oa;
+	__u64				ob;
+
+	oa = xfs_rmap_irec_offset_pack(a);
+	ob = xfs_rmap_irec_offset_pack(b);
+
+	if (a->rm_startblock < b->rm_startblock)
+		return -1;
+	else if (a->rm_startblock > b->rm_startblock)
+		return 1;
+	else if (a->rm_owner < b->rm_owner)
+		return -1;
+	else if (a->rm_owner > b->rm_owner)
+		return 1;
+	else if (oa < ob)
+		return -1;
+	else if (oa > ob)
+		return 1;
+	else
+		return 0;
+}
+
+/* Is there a record covering a given extent? */
+int
+xfs_rmap_has_record(
+	struct xfs_btree_cur	*cur,
+	xfs_agblock_t		bno,
+	xfs_extlen_t		len,
+	bool			*exists)
+{
+	union xfs_btree_irec	low;
+	union xfs_btree_irec	high;
+
+	memset(&low, 0, sizeof(low));
+	low.r.rm_startblock = bno;
+	memset(&high, 0xFF, sizeof(high));
+	high.r.rm_startblock = bno + len - 1;
+
+	return xfs_btree_has_record(cur, &low, &high, exists);
+}
+
+/*
+ * Is there a record for this owner completely covering a given physical
+ * extent?  If so, *has_rmap will be set to true.  If there is no record
+ * or the record only covers part of the range, we set *has_rmap to false.
+ * This function doesn't perform range lookups or offset checks, so it is
+ * not suitable for checking data fork blocks.
+ */
+int
+xfs_rmap_record_exists(
+	struct xfs_btree_cur		*cur,
+	xfs_agblock_t			bno,
+	xfs_extlen_t			len,
+	const struct xfs_owner_info	*oinfo,
+	bool				*has_rmap)
+{
+	uint64_t			owner;
+	uint64_t			offset;
+	unsigned int			flags;
+	int				has_record;
+	struct xfs_rmap_irec		irec;
+	int				error;
+
+	xfs_owner_info_unpack(oinfo, &owner, &offset, &flags);
+	ASSERT(XFS_RMAP_NON_INODE_OWNER(owner) ||
+	       (flags & XFS_RMAP_BMBT_BLOCK));
+
+	error = xfs_rmap_lookup_le(cur, bno, owner, offset, flags, &irec,
+			&has_record);
+	if (error)
+		return error;
+	if (!has_record) {
+		*has_rmap = false;
+		return 0;
+	}
+
+	*has_rmap = (irec.rm_owner == owner && irec.rm_startblock <= bno &&
+		     irec.rm_startblock + irec.rm_blockcount >= bno + len);
+	return 0;
+}
+
+struct xfs_rmap_key_state {
+	uint64_t			owner;
+	uint64_t			offset;
+	unsigned int			flags;
+};
+
+/* For each rmap given, figure out if it doesn't match the key we want. */
+STATIC int
+xfs_rmap_has_other_keys_helper(
+	struct xfs_btree_cur		*cur,
+	const struct xfs_rmap_irec	*rec,
+	void				*priv)
+{
+	struct xfs_rmap_key_state	*rks = priv;
+
+	if (rks->owner == rec->rm_owner && rks->offset == rec->rm_offset &&
+	    ((rks->flags & rec->rm_flags) & XFS_RMAP_KEY_FLAGS) == rks->flags)
+		return 0;
+	return -ECANCELED;
+}
+
+/*
+ * Given an extent and some owner info, can we find records overlapping
+ * the extent whose owner info does not match the given owner?
+ */
+int
+xfs_rmap_has_other_keys(
+	struct xfs_btree_cur		*cur,
+	xfs_agblock_t			bno,
+	xfs_extlen_t			len,
+	const struct xfs_owner_info	*oinfo,
+	bool				*has_rmap)
+{
+	struct xfs_rmap_irec		low = {0};
+	struct xfs_rmap_irec		high;
+	struct xfs_rmap_key_state	rks;
+	int				error;
+
+	xfs_owner_info_unpack(oinfo, &rks.owner, &rks.offset, &rks.flags);
+	*has_rmap = false;
+
+	low.rm_startblock = bno;
+	memset(&high, 0xFF, sizeof(high));
+	high.rm_startblock = bno + len - 1;
+
+	error = xfs_rmap_query_range(cur, &low, &high,
+			xfs_rmap_has_other_keys_helper, &rks);
+	if (error == -ECANCELED) {
+		*has_rmap = true;
+		return 0;
+	}
+
+	return error;
+}
+
+const struct xfs_owner_info XFS_RMAP_OINFO_SKIP_UPDATE = {
+	.oi_owner = XFS_RMAP_OWN_NULL,
+};
+const struct xfs_owner_info XFS_RMAP_OINFO_ANY_OWNER = {
+	.oi_owner = XFS_RMAP_OWN_UNKNOWN,
+};
+const struct xfs_owner_info XFS_RMAP_OINFO_FS = {
+	.oi_owner = XFS_RMAP_OWN_FS,
+};
+const struct xfs_owner_info XFS_RMAP_OINFO_LOG = {
+	.oi_owner = XFS_RMAP_OWN_LOG,
+};
+const struct xfs_owner_info XFS_RMAP_OINFO_AG = {
+	.oi_owner = XFS_RMAP_OWN_AG,
+};
+const struct xfs_owner_info XFS_RMAP_OINFO_INOBT = {
+	.oi_owner = XFS_RMAP_OWN_INOBT,
+};
+const struct xfs_owner_info XFS_RMAP_OINFO_INODES = {
+	.oi_owner = XFS_RMAP_OWN_INODES,
+};
+const struct xfs_owner_info XFS_RMAP_OINFO_REFC = {
+	.oi_owner = XFS_RMAP_OWN_REFC,
+};
+const struct xfs_owner_info XFS_RMAP_OINFO_COW = {
+	.oi_owner = XFS_RMAP_OWN_COW,
+};
+
+int __init
+xfs_rmap_intent_init_cache(void)
+{
+	xfs_rmap_intent_cache = kmem_cache_create("xfs_rmap_intent",
+			sizeof(struct xfs_rmap_intent),
+			0, 0, NULL);
+
+	return xfs_rmap_intent_cache != NULL ? 0 : -ENOMEM;
+}
+
+void
+xfs_rmap_intent_destroy_cache(void)
+{
+	kmem_cache_destroy(xfs_rmap_intent_cache);
+	xfs_rmap_intent_cache = NULL;
+}
diff --git a/fs/xfs/libxfs/xfs_rmap.h b/fs/xfs/libxfs/xfs_rmap.h
new file mode 100644
index 000000000..54741a591
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_rmap.h
@@ -0,0 +1,220 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2016 Oracle.  All Rights Reserved.
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ */
+#ifndef __XFS_RMAP_H__
+#define __XFS_RMAP_H__
+
+struct xfs_perag;
+
+static inline void
+xfs_rmap_ino_bmbt_owner(
+	struct xfs_owner_info	*oi,
+	xfs_ino_t		ino,
+	int			whichfork)
+{
+	oi->oi_owner = ino;
+	oi->oi_offset = 0;
+	oi->oi_flags = XFS_OWNER_INFO_BMBT_BLOCK;
+	if (whichfork == XFS_ATTR_FORK)
+		oi->oi_flags |= XFS_OWNER_INFO_ATTR_FORK;
+}
+
+static inline void
+xfs_rmap_ino_owner(
+	struct xfs_owner_info	*oi,
+	xfs_ino_t		ino,
+	int			whichfork,
+	xfs_fileoff_t		offset)
+{
+	oi->oi_owner = ino;
+	oi->oi_offset = offset;
+	oi->oi_flags = 0;
+	if (whichfork == XFS_ATTR_FORK)
+		oi->oi_flags |= XFS_OWNER_INFO_ATTR_FORK;
+}
+
+static inline bool
+xfs_rmap_should_skip_owner_update(
+	const struct xfs_owner_info	*oi)
+{
+	return oi->oi_owner == XFS_RMAP_OWN_NULL;
+}
+
+/* Reverse mapping functions. */
+
+struct xfs_buf;
+
+static inline __u64
+xfs_rmap_irec_offset_pack(
+	const struct xfs_rmap_irec	*irec)
+{
+	__u64			x;
+
+	x = XFS_RMAP_OFF(irec->rm_offset);
+	if (irec->rm_flags & XFS_RMAP_ATTR_FORK)
+		x |= XFS_RMAP_OFF_ATTR_FORK;
+	if (irec->rm_flags & XFS_RMAP_BMBT_BLOCK)
+		x |= XFS_RMAP_OFF_BMBT_BLOCK;
+	if (irec->rm_flags & XFS_RMAP_UNWRITTEN)
+		x |= XFS_RMAP_OFF_UNWRITTEN;
+	return x;
+}
+
+static inline int
+xfs_rmap_irec_offset_unpack(
+	__u64			offset,
+	struct xfs_rmap_irec	*irec)
+{
+	if (offset & ~(XFS_RMAP_OFF_MASK | XFS_RMAP_OFF_FLAGS))
+		return -EFSCORRUPTED;
+	irec->rm_offset = XFS_RMAP_OFF(offset);
+	irec->rm_flags = 0;
+	if (offset & XFS_RMAP_OFF_ATTR_FORK)
+		irec->rm_flags |= XFS_RMAP_ATTR_FORK;
+	if (offset & XFS_RMAP_OFF_BMBT_BLOCK)
+		irec->rm_flags |= XFS_RMAP_BMBT_BLOCK;
+	if (offset & XFS_RMAP_OFF_UNWRITTEN)
+		irec->rm_flags |= XFS_RMAP_UNWRITTEN;
+	return 0;
+}
+
+static inline void
+xfs_owner_info_unpack(
+	const struct xfs_owner_info	*oinfo,
+	uint64_t			*owner,
+	uint64_t			*offset,
+	unsigned int			*flags)
+{
+	unsigned int			r = 0;
+
+	*owner = oinfo->oi_owner;
+	*offset = oinfo->oi_offset;
+	if (oinfo->oi_flags & XFS_OWNER_INFO_ATTR_FORK)
+		r |= XFS_RMAP_ATTR_FORK;
+	if (oinfo->oi_flags & XFS_OWNER_INFO_BMBT_BLOCK)
+		r |= XFS_RMAP_BMBT_BLOCK;
+	*flags = r;
+}
+
+static inline void
+xfs_owner_info_pack(
+	struct xfs_owner_info	*oinfo,
+	uint64_t		owner,
+	uint64_t		offset,
+	unsigned int		flags)
+{
+	oinfo->oi_owner = owner;
+	oinfo->oi_offset = XFS_RMAP_OFF(offset);
+	oinfo->oi_flags = 0;
+	if (flags & XFS_RMAP_ATTR_FORK)
+		oinfo->oi_flags |= XFS_OWNER_INFO_ATTR_FORK;
+	if (flags & XFS_RMAP_BMBT_BLOCK)
+		oinfo->oi_flags |= XFS_OWNER_INFO_BMBT_BLOCK;
+}
+
+int xfs_rmap_alloc(struct xfs_trans *tp, struct xfs_buf *agbp,
+		   struct xfs_perag *pag, xfs_agblock_t bno, xfs_extlen_t len,
+		   const struct xfs_owner_info *oinfo);
+int xfs_rmap_free(struct xfs_trans *tp, struct xfs_buf *agbp,
+		  struct xfs_perag *pag, xfs_agblock_t bno, xfs_extlen_t len,
+		  const struct xfs_owner_info *oinfo);
+
+int xfs_rmap_lookup_le(struct xfs_btree_cur *cur, xfs_agblock_t bno,
+		uint64_t owner, uint64_t offset, unsigned int flags,
+		struct xfs_rmap_irec *irec, int *stat);
+int xfs_rmap_lookup_eq(struct xfs_btree_cur *cur, xfs_agblock_t bno,
+		xfs_extlen_t len, uint64_t owner, uint64_t offset,
+		unsigned int flags, int *stat);
+int xfs_rmap_insert(struct xfs_btree_cur *rcur, xfs_agblock_t agbno,
+		xfs_extlen_t len, uint64_t owner, uint64_t offset,
+		unsigned int flags);
+int xfs_rmap_get_rec(struct xfs_btree_cur *cur, struct xfs_rmap_irec *irec,
+		int *stat);
+
+typedef int (*xfs_rmap_query_range_fn)(
+	struct xfs_btree_cur		*cur,
+	const struct xfs_rmap_irec	*rec,
+	void				*priv);
+
+int xfs_rmap_query_range(struct xfs_btree_cur *cur,
+		const struct xfs_rmap_irec *low_rec,
+		const struct xfs_rmap_irec *high_rec,
+		xfs_rmap_query_range_fn fn, void *priv);
+int xfs_rmap_query_all(struct xfs_btree_cur *cur, xfs_rmap_query_range_fn fn,
+		void *priv);
+
+enum xfs_rmap_intent_type {
+	XFS_RMAP_MAP,
+	XFS_RMAP_MAP_SHARED,
+	XFS_RMAP_UNMAP,
+	XFS_RMAP_UNMAP_SHARED,
+	XFS_RMAP_CONVERT,
+	XFS_RMAP_CONVERT_SHARED,
+	XFS_RMAP_ALLOC,
+	XFS_RMAP_FREE,
+};
+
+struct xfs_rmap_intent {
+	struct list_head			ri_list;
+	enum xfs_rmap_intent_type		ri_type;
+	int					ri_whichfork;
+	uint64_t				ri_owner;
+	struct xfs_bmbt_irec			ri_bmap;
+};
+
+/* functions for updating the rmapbt based on bmbt map/unmap operations */
+void xfs_rmap_map_extent(struct xfs_trans *tp, struct xfs_inode *ip,
+		int whichfork, struct xfs_bmbt_irec *imap);
+void xfs_rmap_unmap_extent(struct xfs_trans *tp, struct xfs_inode *ip,
+		int whichfork, struct xfs_bmbt_irec *imap);
+void xfs_rmap_convert_extent(struct xfs_mount *mp, struct xfs_trans *tp,
+		struct xfs_inode *ip, int whichfork,
+		struct xfs_bmbt_irec *imap);
+void xfs_rmap_alloc_extent(struct xfs_trans *tp, xfs_agnumber_t agno,
+		xfs_agblock_t bno, xfs_extlen_t len, uint64_t owner);
+void xfs_rmap_free_extent(struct xfs_trans *tp, xfs_agnumber_t agno,
+		xfs_agblock_t bno, xfs_extlen_t len, uint64_t owner);
+
+void xfs_rmap_finish_one_cleanup(struct xfs_trans *tp,
+		struct xfs_btree_cur *rcur, int error);
+int xfs_rmap_finish_one(struct xfs_trans *tp, enum xfs_rmap_intent_type type,
+		uint64_t owner, int whichfork, xfs_fileoff_t startoff,
+		xfs_fsblock_t startblock, xfs_filblks_t blockcount,
+		xfs_exntst_t state, struct xfs_btree_cur **pcur);
+
+int xfs_rmap_lookup_le_range(struct xfs_btree_cur *cur, xfs_agblock_t bno,
+		uint64_t owner, uint64_t offset, unsigned int flags,
+		struct xfs_rmap_irec *irec, int	*stat);
+int xfs_rmap_compare(const struct xfs_rmap_irec *a,
+		const struct xfs_rmap_irec *b);
+union xfs_btree_rec;
+int xfs_rmap_btrec_to_irec(const union xfs_btree_rec *rec,
+		struct xfs_rmap_irec *irec);
+int xfs_rmap_has_record(struct xfs_btree_cur *cur, xfs_agblock_t bno,
+		xfs_extlen_t len, bool *exists);
+int xfs_rmap_record_exists(struct xfs_btree_cur *cur, xfs_agblock_t bno,
+		xfs_extlen_t len, const struct xfs_owner_info *oinfo,
+		bool *has_rmap);
+int xfs_rmap_has_other_keys(struct xfs_btree_cur *cur, xfs_agblock_t bno,
+		xfs_extlen_t len, const struct xfs_owner_info *oinfo,
+		bool *has_rmap);
+int xfs_rmap_map_raw(struct xfs_btree_cur *cur, struct xfs_rmap_irec *rmap);
+
+extern const struct xfs_owner_info XFS_RMAP_OINFO_SKIP_UPDATE;
+extern const struct xfs_owner_info XFS_RMAP_OINFO_ANY_OWNER;
+extern const struct xfs_owner_info XFS_RMAP_OINFO_FS;
+extern const struct xfs_owner_info XFS_RMAP_OINFO_LOG;
+extern const struct xfs_owner_info XFS_RMAP_OINFO_AG;
+extern const struct xfs_owner_info XFS_RMAP_OINFO_INOBT;
+extern const struct xfs_owner_info XFS_RMAP_OINFO_INODES;
+extern const struct xfs_owner_info XFS_RMAP_OINFO_REFC;
+extern const struct xfs_owner_info XFS_RMAP_OINFO_COW;
+
+extern struct kmem_cache	*xfs_rmap_intent_cache;
+
+int __init xfs_rmap_intent_init_cache(void);
+void xfs_rmap_intent_destroy_cache(void);
+
+#endif	/* __XFS_RMAP_H__ */
diff --git a/fs/xfs/libxfs/xfs_rmap_btree.c b/fs/xfs/libxfs/xfs_rmap_btree.c
new file mode 100644
index 000000000..7f83f62e5
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_rmap_btree.c
@@ -0,0 +1,696 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2014 Red Hat, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_trans.h"
+#include "xfs_alloc.h"
+#include "xfs_btree.h"
+#include "xfs_btree_staging.h"
+#include "xfs_rmap.h"
+#include "xfs_rmap_btree.h"
+#include "xfs_trace.h"
+#include "xfs_error.h"
+#include "xfs_extent_busy.h"
+#include "xfs_ag.h"
+#include "xfs_ag_resv.h"
+
+static struct kmem_cache	*xfs_rmapbt_cur_cache;
+
+/*
+ * Reverse map btree.
+ *
+ * This is a per-ag tree used to track the owner(s) of a given extent. With
+ * reflink it is possible for there to be multiple owners, which is a departure
+ * from classic XFS. Owner records for data extents are inserted when the
+ * extent is mapped and removed when an extent is unmapped.  Owner records for
+ * all other block types (i.e. metadata) are inserted when an extent is
+ * allocated and removed when an extent is freed. There can only be one owner
+ * of a metadata extent, usually an inode or some other metadata structure like
+ * an AG btree.
+ *
+ * The rmap btree is part of the free space management, so blocks for the tree
+ * are sourced from the agfl. Hence we need transaction reservation support for
+ * this tree so that the freelist is always large enough. This also impacts on
+ * the minimum space we need to leave free in the AG.
+ *
+ * The tree is ordered by [ag block, owner, offset]. This is a large key size,
+ * but it is the only way to enforce unique keys when a block can be owned by
+ * multiple files at any offset. There's no need to order/search by extent
+ * size for online updating/management of the tree. It is intended that most
+ * reverse lookups will be to find the owner(s) of a particular block, or to
+ * try to recover tree and file data from corrupt primary metadata.
+ */
+
+static struct xfs_btree_cur *
+xfs_rmapbt_dup_cursor(
+	struct xfs_btree_cur	*cur)
+{
+	return xfs_rmapbt_init_cursor(cur->bc_mp, cur->bc_tp,
+				cur->bc_ag.agbp, cur->bc_ag.pag);
+}
+
+STATIC void
+xfs_rmapbt_set_root(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*ptr,
+	int				inc)
+{
+	struct xfs_buf		*agbp = cur->bc_ag.agbp;
+	struct xfs_agf		*agf = agbp->b_addr;
+	int			btnum = cur->bc_btnum;
+
+	ASSERT(ptr->s != 0);
+
+	agf->agf_roots[btnum] = ptr->s;
+	be32_add_cpu(&agf->agf_levels[btnum], inc);
+	cur->bc_ag.pag->pagf_levels[btnum] += inc;
+
+	xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_ROOTS | XFS_AGF_LEVELS);
+}
+
+STATIC int
+xfs_rmapbt_alloc_block(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*start,
+	union xfs_btree_ptr		*new,
+	int				*stat)
+{
+	struct xfs_buf		*agbp = cur->bc_ag.agbp;
+	struct xfs_agf		*agf = agbp->b_addr;
+	struct xfs_perag	*pag = cur->bc_ag.pag;
+	int			error;
+	xfs_agblock_t		bno;
+
+	/* Allocate the new block from the freelist. If we can't, give up.  */
+	error = xfs_alloc_get_freelist(pag, cur->bc_tp, cur->bc_ag.agbp,
+				       &bno, 1);
+	if (error)
+		return error;
+
+	trace_xfs_rmapbt_alloc_block(cur->bc_mp, pag->pag_agno, bno, 1);
+	if (bno == NULLAGBLOCK) {
+		*stat = 0;
+		return 0;
+	}
+
+	xfs_extent_busy_reuse(cur->bc_mp, pag, bno, 1, false);
+
+	new->s = cpu_to_be32(bno);
+	be32_add_cpu(&agf->agf_rmap_blocks, 1);
+	xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_RMAP_BLOCKS);
+
+	xfs_ag_resv_rmapbt_alloc(cur->bc_mp, pag->pag_agno);
+
+	*stat = 1;
+	return 0;
+}
+
+STATIC int
+xfs_rmapbt_free_block(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp)
+{
+	struct xfs_buf		*agbp = cur->bc_ag.agbp;
+	struct xfs_agf		*agf = agbp->b_addr;
+	struct xfs_perag	*pag = cur->bc_ag.pag;
+	xfs_agblock_t		bno;
+	int			error;
+
+	bno = xfs_daddr_to_agbno(cur->bc_mp, xfs_buf_daddr(bp));
+	trace_xfs_rmapbt_free_block(cur->bc_mp, pag->pag_agno,
+			bno, 1);
+	be32_add_cpu(&agf->agf_rmap_blocks, -1);
+	xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_RMAP_BLOCKS);
+	error = xfs_alloc_put_freelist(pag, cur->bc_tp, agbp, NULL, bno, 1);
+	if (error)
+		return error;
+
+	xfs_extent_busy_insert(cur->bc_tp, pag, bno, 1,
+			      XFS_EXTENT_BUSY_SKIP_DISCARD);
+
+	xfs_ag_resv_free_extent(pag, XFS_AG_RESV_RMAPBT, NULL, 1);
+	return 0;
+}
+
+STATIC int
+xfs_rmapbt_get_minrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	return cur->bc_mp->m_rmap_mnr[level != 0];
+}
+
+STATIC int
+xfs_rmapbt_get_maxrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	return cur->bc_mp->m_rmap_mxr[level != 0];
+}
+
+STATIC void
+xfs_rmapbt_init_key_from_rec(
+	union xfs_btree_key		*key,
+	const union xfs_btree_rec	*rec)
+{
+	key->rmap.rm_startblock = rec->rmap.rm_startblock;
+	key->rmap.rm_owner = rec->rmap.rm_owner;
+	key->rmap.rm_offset = rec->rmap.rm_offset;
+}
+
+/*
+ * The high key for a reverse mapping record can be computed by shifting
+ * the startblock and offset to the highest value that would still map
+ * to that record.  In practice this means that we add blockcount-1 to
+ * the startblock for all records, and if the record is for a data/attr
+ * fork mapping, we add blockcount-1 to the offset too.
+ */
+STATIC void
+xfs_rmapbt_init_high_key_from_rec(
+	union xfs_btree_key		*key,
+	const union xfs_btree_rec	*rec)
+{
+	uint64_t			off;
+	int				adj;
+
+	adj = be32_to_cpu(rec->rmap.rm_blockcount) - 1;
+
+	key->rmap.rm_startblock = rec->rmap.rm_startblock;
+	be32_add_cpu(&key->rmap.rm_startblock, adj);
+	key->rmap.rm_owner = rec->rmap.rm_owner;
+	key->rmap.rm_offset = rec->rmap.rm_offset;
+	if (XFS_RMAP_NON_INODE_OWNER(be64_to_cpu(rec->rmap.rm_owner)) ||
+	    XFS_RMAP_IS_BMBT_BLOCK(be64_to_cpu(rec->rmap.rm_offset)))
+		return;
+	off = be64_to_cpu(key->rmap.rm_offset);
+	off = (XFS_RMAP_OFF(off) + adj) | (off & ~XFS_RMAP_OFF_MASK);
+	key->rmap.rm_offset = cpu_to_be64(off);
+}
+
+STATIC void
+xfs_rmapbt_init_rec_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec)
+{
+	rec->rmap.rm_startblock = cpu_to_be32(cur->bc_rec.r.rm_startblock);
+	rec->rmap.rm_blockcount = cpu_to_be32(cur->bc_rec.r.rm_blockcount);
+	rec->rmap.rm_owner = cpu_to_be64(cur->bc_rec.r.rm_owner);
+	rec->rmap.rm_offset = cpu_to_be64(
+			xfs_rmap_irec_offset_pack(&cur->bc_rec.r));
+}
+
+STATIC void
+xfs_rmapbt_init_ptr_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	struct xfs_agf		*agf = cur->bc_ag.agbp->b_addr;
+
+	ASSERT(cur->bc_ag.pag->pag_agno == be32_to_cpu(agf->agf_seqno));
+
+	ptr->s = agf->agf_roots[cur->bc_btnum];
+}
+
+STATIC int64_t
+xfs_rmapbt_key_diff(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*key)
+{
+	struct xfs_rmap_irec		*rec = &cur->bc_rec.r;
+	const struct xfs_rmap_key	*kp = &key->rmap;
+	__u64				x, y;
+	int64_t				d;
+
+	d = (int64_t)be32_to_cpu(kp->rm_startblock) - rec->rm_startblock;
+	if (d)
+		return d;
+
+	x = be64_to_cpu(kp->rm_owner);
+	y = rec->rm_owner;
+	if (x > y)
+		return 1;
+	else if (y > x)
+		return -1;
+
+	x = XFS_RMAP_OFF(be64_to_cpu(kp->rm_offset));
+	y = rec->rm_offset;
+	if (x > y)
+		return 1;
+	else if (y > x)
+		return -1;
+	return 0;
+}
+
+STATIC int64_t
+xfs_rmapbt_diff_two_keys(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*k1,
+	const union xfs_btree_key	*k2)
+{
+	const struct xfs_rmap_key	*kp1 = &k1->rmap;
+	const struct xfs_rmap_key	*kp2 = &k2->rmap;
+	int64_t				d;
+	__u64				x, y;
+
+	d = (int64_t)be32_to_cpu(kp1->rm_startblock) -
+		       be32_to_cpu(kp2->rm_startblock);
+	if (d)
+		return d;
+
+	x = be64_to_cpu(kp1->rm_owner);
+	y = be64_to_cpu(kp2->rm_owner);
+	if (x > y)
+		return 1;
+	else if (y > x)
+		return -1;
+
+	x = XFS_RMAP_OFF(be64_to_cpu(kp1->rm_offset));
+	y = XFS_RMAP_OFF(be64_to_cpu(kp2->rm_offset));
+	if (x > y)
+		return 1;
+	else if (y > x)
+		return -1;
+	return 0;
+}
+
+static xfs_failaddr_t
+xfs_rmapbt_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
+	struct xfs_perag	*pag = bp->b_pag;
+	xfs_failaddr_t		fa;
+	unsigned int		level;
+
+	/*
+	 * magic number and level verification
+	 *
+	 * During growfs operations, we can't verify the exact level or owner as
+	 * the perag is not fully initialised and hence not attached to the
+	 * buffer.  In this case, check against the maximum tree depth.
+	 *
+	 * Similarly, during log recovery we will have a perag structure
+	 * attached, but the agf information will not yet have been initialised
+	 * from the on disk AGF. Again, we can only check against maximum limits
+	 * in this case.
+	 */
+	if (!xfs_verify_magic(bp, block->bb_magic))
+		return __this_address;
+
+	if (!xfs_has_rmapbt(mp))
+		return __this_address;
+	fa = xfs_btree_sblock_v5hdr_verify(bp);
+	if (fa)
+		return fa;
+
+	level = be16_to_cpu(block->bb_level);
+	if (pag && pag->pagf_init) {
+		if (level >= pag->pagf_levels[XFS_BTNUM_RMAPi])
+			return __this_address;
+	} else if (level >= mp->m_rmap_maxlevels)
+		return __this_address;
+
+	return xfs_btree_sblock_verify(bp, mp->m_rmap_mxr[level != 0]);
+}
+
+static void
+xfs_rmapbt_read_verify(
+	struct xfs_buf	*bp)
+{
+	xfs_failaddr_t	fa;
+
+	if (!xfs_btree_sblock_verify_crc(bp))
+		xfs_verifier_error(bp, -EFSBADCRC, __this_address);
+	else {
+		fa = xfs_rmapbt_verify(bp);
+		if (fa)
+			xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+	}
+
+	if (bp->b_error)
+		trace_xfs_btree_corrupt(bp, _RET_IP_);
+}
+
+static void
+xfs_rmapbt_write_verify(
+	struct xfs_buf	*bp)
+{
+	xfs_failaddr_t	fa;
+
+	fa = xfs_rmapbt_verify(bp);
+	if (fa) {
+		trace_xfs_btree_corrupt(bp, _RET_IP_);
+		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+		return;
+	}
+	xfs_btree_sblock_calc_crc(bp);
+
+}
+
+const struct xfs_buf_ops xfs_rmapbt_buf_ops = {
+	.name			= "xfs_rmapbt",
+	.magic			= { 0, cpu_to_be32(XFS_RMAP_CRC_MAGIC) },
+	.verify_read		= xfs_rmapbt_read_verify,
+	.verify_write		= xfs_rmapbt_write_verify,
+	.verify_struct		= xfs_rmapbt_verify,
+};
+
+STATIC int
+xfs_rmapbt_keys_inorder(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*k1,
+	const union xfs_btree_key	*k2)
+{
+	uint32_t		x;
+	uint32_t		y;
+	uint64_t		a;
+	uint64_t		b;
+
+	x = be32_to_cpu(k1->rmap.rm_startblock);
+	y = be32_to_cpu(k2->rmap.rm_startblock);
+	if (x < y)
+		return 1;
+	else if (x > y)
+		return 0;
+	a = be64_to_cpu(k1->rmap.rm_owner);
+	b = be64_to_cpu(k2->rmap.rm_owner);
+	if (a < b)
+		return 1;
+	else if (a > b)
+		return 0;
+	a = XFS_RMAP_OFF(be64_to_cpu(k1->rmap.rm_offset));
+	b = XFS_RMAP_OFF(be64_to_cpu(k2->rmap.rm_offset));
+	if (a <= b)
+		return 1;
+	return 0;
+}
+
+STATIC int
+xfs_rmapbt_recs_inorder(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_rec	*r1,
+	const union xfs_btree_rec	*r2)
+{
+	uint32_t		x;
+	uint32_t		y;
+	uint64_t		a;
+	uint64_t		b;
+
+	x = be32_to_cpu(r1->rmap.rm_startblock);
+	y = be32_to_cpu(r2->rmap.rm_startblock);
+	if (x < y)
+		return 1;
+	else if (x > y)
+		return 0;
+	a = be64_to_cpu(r1->rmap.rm_owner);
+	b = be64_to_cpu(r2->rmap.rm_owner);
+	if (a < b)
+		return 1;
+	else if (a > b)
+		return 0;
+	a = XFS_RMAP_OFF(be64_to_cpu(r1->rmap.rm_offset));
+	b = XFS_RMAP_OFF(be64_to_cpu(r2->rmap.rm_offset));
+	if (a <= b)
+		return 1;
+	return 0;
+}
+
+static const struct xfs_btree_ops xfs_rmapbt_ops = {
+	.rec_len		= sizeof(struct xfs_rmap_rec),
+	.key_len		= 2 * sizeof(struct xfs_rmap_key),
+
+	.dup_cursor		= xfs_rmapbt_dup_cursor,
+	.set_root		= xfs_rmapbt_set_root,
+	.alloc_block		= xfs_rmapbt_alloc_block,
+	.free_block		= xfs_rmapbt_free_block,
+	.get_minrecs		= xfs_rmapbt_get_minrecs,
+	.get_maxrecs		= xfs_rmapbt_get_maxrecs,
+	.init_key_from_rec	= xfs_rmapbt_init_key_from_rec,
+	.init_high_key_from_rec	= xfs_rmapbt_init_high_key_from_rec,
+	.init_rec_from_cur	= xfs_rmapbt_init_rec_from_cur,
+	.init_ptr_from_cur	= xfs_rmapbt_init_ptr_from_cur,
+	.key_diff		= xfs_rmapbt_key_diff,
+	.buf_ops		= &xfs_rmapbt_buf_ops,
+	.diff_two_keys		= xfs_rmapbt_diff_two_keys,
+	.keys_inorder		= xfs_rmapbt_keys_inorder,
+	.recs_inorder		= xfs_rmapbt_recs_inorder,
+};
+
+static struct xfs_btree_cur *
+xfs_rmapbt_init_common(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_perag	*pag)
+{
+	struct xfs_btree_cur	*cur;
+
+	/* Overlapping btree; 2 keys per pointer. */
+	cur = xfs_btree_alloc_cursor(mp, tp, XFS_BTNUM_RMAP,
+			mp->m_rmap_maxlevels, xfs_rmapbt_cur_cache);
+	cur->bc_flags = XFS_BTREE_CRC_BLOCKS | XFS_BTREE_OVERLAPPING;
+	cur->bc_statoff = XFS_STATS_CALC_INDEX(xs_rmap_2);
+	cur->bc_ops = &xfs_rmapbt_ops;
+
+	/* take a reference for the cursor */
+	atomic_inc(&pag->pag_ref);
+	cur->bc_ag.pag = pag;
+
+	return cur;
+}
+
+/* Create a new reverse mapping btree cursor. */
+struct xfs_btree_cur *
+xfs_rmapbt_init_cursor(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp,
+	struct xfs_perag	*pag)
+{
+	struct xfs_agf		*agf = agbp->b_addr;
+	struct xfs_btree_cur	*cur;
+
+	cur = xfs_rmapbt_init_common(mp, tp, pag);
+	cur->bc_nlevels = be32_to_cpu(agf->agf_levels[XFS_BTNUM_RMAP]);
+	cur->bc_ag.agbp = agbp;
+	return cur;
+}
+
+/* Create a new reverse mapping btree cursor with a fake root for staging. */
+struct xfs_btree_cur *
+xfs_rmapbt_stage_cursor(
+	struct xfs_mount	*mp,
+	struct xbtree_afakeroot	*afake,
+	struct xfs_perag	*pag)
+{
+	struct xfs_btree_cur	*cur;
+
+	cur = xfs_rmapbt_init_common(mp, NULL, pag);
+	xfs_btree_stage_afakeroot(cur, afake);
+	return cur;
+}
+
+/*
+ * Install a new reverse mapping btree root.  Caller is responsible for
+ * invalidating and freeing the old btree blocks.
+ */
+void
+xfs_rmapbt_commit_staged_btree(
+	struct xfs_btree_cur	*cur,
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp)
+{
+	struct xfs_agf		*agf = agbp->b_addr;
+	struct xbtree_afakeroot	*afake = cur->bc_ag.afake;
+
+	ASSERT(cur->bc_flags & XFS_BTREE_STAGING);
+
+	agf->agf_roots[cur->bc_btnum] = cpu_to_be32(afake->af_root);
+	agf->agf_levels[cur->bc_btnum] = cpu_to_be32(afake->af_levels);
+	agf->agf_rmap_blocks = cpu_to_be32(afake->af_blocks);
+	xfs_alloc_log_agf(tp, agbp, XFS_AGF_ROOTS | XFS_AGF_LEVELS |
+				    XFS_AGF_RMAP_BLOCKS);
+	xfs_btree_commit_afakeroot(cur, tp, agbp, &xfs_rmapbt_ops);
+}
+
+/* Calculate number of records in a reverse mapping btree block. */
+static inline unsigned int
+xfs_rmapbt_block_maxrecs(
+	unsigned int		blocklen,
+	bool			leaf)
+{
+	if (leaf)
+		return blocklen / sizeof(struct xfs_rmap_rec);
+	return blocklen /
+		(2 * sizeof(struct xfs_rmap_key) + sizeof(xfs_rmap_ptr_t));
+}
+
+/*
+ * Calculate number of records in an rmap btree block.
+ */
+int
+xfs_rmapbt_maxrecs(
+	int			blocklen,
+	int			leaf)
+{
+	blocklen -= XFS_RMAP_BLOCK_LEN;
+	return xfs_rmapbt_block_maxrecs(blocklen, leaf);
+}
+
+/* Compute the max possible height for reverse mapping btrees. */
+unsigned int
+xfs_rmapbt_maxlevels_ondisk(void)
+{
+	unsigned int		minrecs[2];
+	unsigned int		blocklen;
+
+	blocklen = XFS_MIN_CRC_BLOCKSIZE - XFS_BTREE_SBLOCK_CRC_LEN;
+
+	minrecs[0] = xfs_rmapbt_block_maxrecs(blocklen, true) / 2;
+	minrecs[1] = xfs_rmapbt_block_maxrecs(blocklen, false) / 2;
+
+	/*
+	 * Compute the asymptotic maxlevels for an rmapbt on any reflink fs.
+	 *
+	 * On a reflink filesystem, each AG block can have up to 2^32 (per the
+	 * refcount record format) owners, which means that theoretically we
+	 * could face up to 2^64 rmap records.  However, we're likely to run
+	 * out of blocks in the AG long before that happens, which means that
+	 * we must compute the max height based on what the btree will look
+	 * like if it consumes almost all the blocks in the AG due to maximal
+	 * sharing factor.
+	 */
+	return xfs_btree_space_to_height(minrecs, XFS_MAX_CRC_AG_BLOCKS);
+}
+
+/* Compute the maximum height of an rmap btree. */
+void
+xfs_rmapbt_compute_maxlevels(
+	struct xfs_mount		*mp)
+{
+	if (!xfs_has_rmapbt(mp)) {
+		mp->m_rmap_maxlevels = 0;
+		return;
+	}
+
+	if (xfs_has_reflink(mp)) {
+		/*
+		 * Compute the asymptotic maxlevels for an rmap btree on a
+		 * filesystem that supports reflink.
+		 *
+		 * On a reflink filesystem, each AG block can have up to 2^32
+		 * (per the refcount record format) owners, which means that
+		 * theoretically we could face up to 2^64 rmap records.
+		 * However, we're likely to run out of blocks in the AG long
+		 * before that happens, which means that we must compute the
+		 * max height based on what the btree will look like if it
+		 * consumes almost all the blocks in the AG due to maximal
+		 * sharing factor.
+		 */
+		mp->m_rmap_maxlevels = xfs_btree_space_to_height(mp->m_rmap_mnr,
+				mp->m_sb.sb_agblocks);
+	} else {
+		/*
+		 * If there's no block sharing, compute the maximum rmapbt
+		 * height assuming one rmap record per AG block.
+		 */
+		mp->m_rmap_maxlevels = xfs_btree_compute_maxlevels(
+				mp->m_rmap_mnr, mp->m_sb.sb_agblocks);
+	}
+	ASSERT(mp->m_rmap_maxlevels <= xfs_rmapbt_maxlevels_ondisk());
+}
+
+/* Calculate the refcount btree size for some records. */
+xfs_extlen_t
+xfs_rmapbt_calc_size(
+	struct xfs_mount	*mp,
+	unsigned long long	len)
+{
+	return xfs_btree_calc_size(mp->m_rmap_mnr, len);
+}
+
+/*
+ * Calculate the maximum refcount btree size.
+ */
+xfs_extlen_t
+xfs_rmapbt_max_size(
+	struct xfs_mount	*mp,
+	xfs_agblock_t		agblocks)
+{
+	/* Bail out if we're uninitialized, which can happen in mkfs. */
+	if (mp->m_rmap_mxr[0] == 0)
+		return 0;
+
+	return xfs_rmapbt_calc_size(mp, agblocks);
+}
+
+/*
+ * Figure out how many blocks to reserve and how many are used by this btree.
+ */
+int
+xfs_rmapbt_calc_reserves(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_perag	*pag,
+	xfs_extlen_t		*ask,
+	xfs_extlen_t		*used)
+{
+	struct xfs_buf		*agbp;
+	struct xfs_agf		*agf;
+	xfs_agblock_t		agblocks;
+	xfs_extlen_t		tree_len;
+	int			error;
+
+	if (!xfs_has_rmapbt(mp))
+		return 0;
+
+	error = xfs_alloc_read_agf(pag, tp, 0, &agbp);
+	if (error)
+		return error;
+
+	agf = agbp->b_addr;
+	agblocks = be32_to_cpu(agf->agf_length);
+	tree_len = be32_to_cpu(agf->agf_rmap_blocks);
+	xfs_trans_brelse(tp, agbp);
+
+	/*
+	 * The log is permanently allocated, so the space it occupies will
+	 * never be available for the kinds of things that would require btree
+	 * expansion.  We therefore can pretend the space isn't there.
+	 */
+	if (xfs_ag_contains_log(mp, pag->pag_agno))
+		agblocks -= mp->m_sb.sb_logblocks;
+
+	/* Reserve 1% of the AG or enough for 1 block per record. */
+	*ask += max(agblocks / 100, xfs_rmapbt_max_size(mp, agblocks));
+	*used += tree_len;
+
+	return error;
+}
+
+int __init
+xfs_rmapbt_init_cur_cache(void)
+{
+	xfs_rmapbt_cur_cache = kmem_cache_create("xfs_rmapbt_cur",
+			xfs_btree_cur_sizeof(xfs_rmapbt_maxlevels_ondisk()),
+			0, 0, NULL);
+
+	if (!xfs_rmapbt_cur_cache)
+		return -ENOMEM;
+	return 0;
+}
+
+void
+xfs_rmapbt_destroy_cur_cache(void)
+{
+	kmem_cache_destroy(xfs_rmapbt_cur_cache);
+	xfs_rmapbt_cur_cache = NULL;
+}
diff --git a/fs/xfs/libxfs/xfs_rmap_btree.h b/fs/xfs/libxfs/xfs_rmap_btree.h
new file mode 100644
index 000000000..3244715dd
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_rmap_btree.h
@@ -0,0 +1,67 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2014 Red Hat, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_RMAP_BTREE_H__
+#define __XFS_RMAP_BTREE_H__
+
+struct xfs_buf;
+struct xfs_btree_cur;
+struct xfs_mount;
+struct xbtree_afakeroot;
+
+/* rmaps only exist on crc enabled filesystems */
+#define XFS_RMAP_BLOCK_LEN	XFS_BTREE_SBLOCK_CRC_LEN
+
+/*
+ * Record, key, and pointer address macros for btree blocks.
+ *
+ * (note that some of these may appear unused, but they are used in userspace)
+ */
+#define XFS_RMAP_REC_ADDR(block, index) \
+	((struct xfs_rmap_rec *) \
+		((char *)(block) + XFS_RMAP_BLOCK_LEN + \
+		 (((index) - 1) * sizeof(struct xfs_rmap_rec))))
+
+#define XFS_RMAP_KEY_ADDR(block, index) \
+	((struct xfs_rmap_key *) \
+		((char *)(block) + XFS_RMAP_BLOCK_LEN + \
+		 ((index) - 1) * 2 * sizeof(struct xfs_rmap_key)))
+
+#define XFS_RMAP_HIGH_KEY_ADDR(block, index) \
+	((struct xfs_rmap_key *) \
+		((char *)(block) + XFS_RMAP_BLOCK_LEN + \
+		 sizeof(struct xfs_rmap_key) + \
+		 ((index) - 1) * 2 * sizeof(struct xfs_rmap_key)))
+
+#define XFS_RMAP_PTR_ADDR(block, index, maxrecs) \
+	((xfs_rmap_ptr_t *) \
+		((char *)(block) + XFS_RMAP_BLOCK_LEN + \
+		 (maxrecs) * 2 * sizeof(struct xfs_rmap_key) + \
+		 ((index) - 1) * sizeof(xfs_rmap_ptr_t)))
+
+struct xfs_btree_cur *xfs_rmapbt_init_cursor(struct xfs_mount *mp,
+				struct xfs_trans *tp, struct xfs_buf *bp,
+				struct xfs_perag *pag);
+struct xfs_btree_cur *xfs_rmapbt_stage_cursor(struct xfs_mount *mp,
+		struct xbtree_afakeroot *afake, struct xfs_perag *pag);
+void xfs_rmapbt_commit_staged_btree(struct xfs_btree_cur *cur,
+		struct xfs_trans *tp, struct xfs_buf *agbp);
+int xfs_rmapbt_maxrecs(int blocklen, int leaf);
+extern void xfs_rmapbt_compute_maxlevels(struct xfs_mount *mp);
+
+extern xfs_extlen_t xfs_rmapbt_calc_size(struct xfs_mount *mp,
+		unsigned long long len);
+extern xfs_extlen_t xfs_rmapbt_max_size(struct xfs_mount *mp,
+		xfs_agblock_t agblocks);
+
+extern int xfs_rmapbt_calc_reserves(struct xfs_mount *mp, struct xfs_trans *tp,
+		struct xfs_perag *pag, xfs_extlen_t *ask, xfs_extlen_t *used);
+
+unsigned int xfs_rmapbt_maxlevels_ondisk(void);
+
+int __init xfs_rmapbt_init_cur_cache(void);
+void xfs_rmapbt_destroy_cur_cache(void);
+
+#endif /* __XFS_RMAP_BTREE_H__ */
diff --git a/fs/xfs/libxfs/xfs_rtbitmap.c b/fs/xfs/libxfs/xfs_rtbitmap.c
new file mode 100644
index 000000000..fa180ab66
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_rtbitmap.c
@@ -0,0 +1,1098 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_bit.h"
+#include "xfs_mount.h"
+#include "xfs_inode.h"
+#include "xfs_bmap.h"
+#include "xfs_trans.h"
+#include "xfs_rtalloc.h"
+#include "xfs_error.h"
+
+/*
+ * Realtime allocator bitmap functions shared with userspace.
+ */
+
+/*
+ * Real time buffers need verifiers to avoid runtime warnings during IO.
+ * We don't have anything to verify, however, so these are just dummy
+ * operations.
+ */
+static void
+xfs_rtbuf_verify_read(
+	struct xfs_buf	*bp)
+{
+	return;
+}
+
+static void
+xfs_rtbuf_verify_write(
+	struct xfs_buf	*bp)
+{
+	return;
+}
+
+const struct xfs_buf_ops xfs_rtbuf_ops = {
+	.name = "rtbuf",
+	.verify_read = xfs_rtbuf_verify_read,
+	.verify_write = xfs_rtbuf_verify_write,
+};
+
+/*
+ * Get a buffer for the bitmap or summary file block specified.
+ * The buffer is returned read and locked.
+ */
+int
+xfs_rtbuf_get(
+	xfs_mount_t	*mp,		/* file system mount structure */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	block,		/* block number in bitmap or summary */
+	int		issum,		/* is summary not bitmap */
+	struct xfs_buf	**bpp)		/* output: buffer for the block */
+{
+	struct xfs_buf	*bp;		/* block buffer, result */
+	xfs_inode_t	*ip;		/* bitmap or summary inode */
+	xfs_bmbt_irec_t	map;
+	int		nmap = 1;
+	int		error;		/* error value */
+
+	ip = issum ? mp->m_rsumip : mp->m_rbmip;
+
+	error = xfs_bmapi_read(ip, block, 1, &map, &nmap, 0);
+	if (error)
+		return error;
+
+	if (XFS_IS_CORRUPT(mp, nmap == 0 || !xfs_bmap_is_written_extent(&map)))
+		return -EFSCORRUPTED;
+
+	ASSERT(map.br_startblock != NULLFSBLOCK);
+	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
+				   XFS_FSB_TO_DADDR(mp, map.br_startblock),
+				   mp->m_bsize, 0, &bp, &xfs_rtbuf_ops);
+	if (error)
+		return error;
+
+	xfs_trans_buf_set_type(tp, bp, issum ? XFS_BLFT_RTSUMMARY_BUF
+					     : XFS_BLFT_RTBITMAP_BUF);
+	*bpp = bp;
+	return 0;
+}
+
+/*
+ * Searching backward from start to limit, find the first block whose
+ * allocated/free state is different from start's.
+ */
+int
+xfs_rtfind_back(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	start,		/* starting block to look at */
+	xfs_rtblock_t	limit,		/* last block to look at */
+	xfs_rtblock_t	*rtblock)	/* out: start block found */
+{
+	xfs_rtword_t	*b;		/* current word in buffer */
+	int		bit;		/* bit number in the word */
+	xfs_rtblock_t	block;		/* bitmap block number */
+	struct xfs_buf	*bp;		/* buf for the block */
+	xfs_rtword_t	*bufp;		/* starting word in buffer */
+	int		error;		/* error value */
+	xfs_rtblock_t	firstbit;	/* first useful bit in the word */
+	xfs_rtblock_t	i;		/* current bit number rel. to start */
+	xfs_rtblock_t	len;		/* length of inspected area */
+	xfs_rtword_t	mask;		/* mask of relevant bits for value */
+	xfs_rtword_t	want;		/* mask for "good" values */
+	xfs_rtword_t	wdiff;		/* difference from wanted value */
+	int		word;		/* word number in the buffer */
+
+	/*
+	 * Compute and read in starting bitmap block for starting block.
+	 */
+	block = XFS_BITTOBLOCK(mp, start);
+	error = xfs_rtbuf_get(mp, tp, block, 0, &bp);
+	if (error) {
+		return error;
+	}
+	bufp = bp->b_addr;
+	/*
+	 * Get the first word's index & point to it.
+	 */
+	word = XFS_BITTOWORD(mp, start);
+	b = &bufp[word];
+	bit = (int)(start & (XFS_NBWORD - 1));
+	len = start - limit + 1;
+	/*
+	 * Compute match value, based on the bit at start: if 1 (free)
+	 * then all-ones, else all-zeroes.
+	 */
+	want = (*b & ((xfs_rtword_t)1 << bit)) ? -1 : 0;
+	/*
+	 * If the starting position is not word-aligned, deal with the
+	 * partial word.
+	 */
+	if (bit < XFS_NBWORD - 1) {
+		/*
+		 * Calculate first (leftmost) bit number to look at,
+		 * and mask for all the relevant bits in this word.
+		 */
+		firstbit = XFS_RTMAX((xfs_srtblock_t)(bit - len + 1), 0);
+		mask = (((xfs_rtword_t)1 << (bit - firstbit + 1)) - 1) <<
+			firstbit;
+		/*
+		 * Calculate the difference between the value there
+		 * and what we're looking for.
+		 */
+		if ((wdiff = (*b ^ want) & mask)) {
+			/*
+			 * Different.  Mark where we are and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i = bit - XFS_RTHIBIT(wdiff);
+			*rtblock = start - i + 1;
+			return 0;
+		}
+		i = bit - firstbit + 1;
+		/*
+		 * Go on to previous block if that's where the previous word is
+		 * and we need the previous word.
+		 */
+		if (--word == -1 && i < len) {
+			/*
+			 * If done with this block, get the previous one.
+			 */
+			xfs_trans_brelse(tp, bp);
+			error = xfs_rtbuf_get(mp, tp, --block, 0, &bp);
+			if (error) {
+				return error;
+			}
+			bufp = bp->b_addr;
+			word = XFS_BLOCKWMASK(mp);
+			b = &bufp[word];
+		} else {
+			/*
+			 * Go on to the previous word in the buffer.
+			 */
+			b--;
+		}
+	} else {
+		/*
+		 * Starting on a word boundary, no partial word.
+		 */
+		i = 0;
+	}
+	/*
+	 * Loop over whole words in buffers.  When we use up one buffer
+	 * we move on to the previous one.
+	 */
+	while (len - i >= XFS_NBWORD) {
+		/*
+		 * Compute difference between actual and desired value.
+		 */
+		if ((wdiff = *b ^ want)) {
+			/*
+			 * Different, mark where we are and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i += XFS_NBWORD - 1 - XFS_RTHIBIT(wdiff);
+			*rtblock = start - i + 1;
+			return 0;
+		}
+		i += XFS_NBWORD;
+		/*
+		 * Go on to previous block if that's where the previous word is
+		 * and we need the previous word.
+		 */
+		if (--word == -1 && i < len) {
+			/*
+			 * If done with this block, get the previous one.
+			 */
+			xfs_trans_brelse(tp, bp);
+			error = xfs_rtbuf_get(mp, tp, --block, 0, &bp);
+			if (error) {
+				return error;
+			}
+			bufp = bp->b_addr;
+			word = XFS_BLOCKWMASK(mp);
+			b = &bufp[word];
+		} else {
+			/*
+			 * Go on to the previous word in the buffer.
+			 */
+			b--;
+		}
+	}
+	/*
+	 * If not ending on a word boundary, deal with the last
+	 * (partial) word.
+	 */
+	if (len - i) {
+		/*
+		 * Calculate first (leftmost) bit number to look at,
+		 * and mask for all the relevant bits in this word.
+		 */
+		firstbit = XFS_NBWORD - (len - i);
+		mask = (((xfs_rtword_t)1 << (len - i)) - 1) << firstbit;
+		/*
+		 * Compute difference between actual and desired value.
+		 */
+		if ((wdiff = (*b ^ want) & mask)) {
+			/*
+			 * Different, mark where we are and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i += XFS_NBWORD - 1 - XFS_RTHIBIT(wdiff);
+			*rtblock = start - i + 1;
+			return 0;
+		} else
+			i = len;
+	}
+	/*
+	 * No match, return that we scanned the whole area.
+	 */
+	xfs_trans_brelse(tp, bp);
+	*rtblock = start - i + 1;
+	return 0;
+}
+
+/*
+ * Searching forward from start to limit, find the first block whose
+ * allocated/free state is different from start's.
+ */
+int
+xfs_rtfind_forw(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	start,		/* starting block to look at */
+	xfs_rtblock_t	limit,		/* last block to look at */
+	xfs_rtblock_t	*rtblock)	/* out: start block found */
+{
+	xfs_rtword_t	*b;		/* current word in buffer */
+	int		bit;		/* bit number in the word */
+	xfs_rtblock_t	block;		/* bitmap block number */
+	struct xfs_buf	*bp;		/* buf for the block */
+	xfs_rtword_t	*bufp;		/* starting word in buffer */
+	int		error;		/* error value */
+	xfs_rtblock_t	i;		/* current bit number rel. to start */
+	xfs_rtblock_t	lastbit;	/* last useful bit in the word */
+	xfs_rtblock_t	len;		/* length of inspected area */
+	xfs_rtword_t	mask;		/* mask of relevant bits for value */
+	xfs_rtword_t	want;		/* mask for "good" values */
+	xfs_rtword_t	wdiff;		/* difference from wanted value */
+	int		word;		/* word number in the buffer */
+
+	/*
+	 * Compute and read in starting bitmap block for starting block.
+	 */
+	block = XFS_BITTOBLOCK(mp, start);
+	error = xfs_rtbuf_get(mp, tp, block, 0, &bp);
+	if (error) {
+		return error;
+	}
+	bufp = bp->b_addr;
+	/*
+	 * Get the first word's index & point to it.
+	 */
+	word = XFS_BITTOWORD(mp, start);
+	b = &bufp[word];
+	bit = (int)(start & (XFS_NBWORD - 1));
+	len = limit - start + 1;
+	/*
+	 * Compute match value, based on the bit at start: if 1 (free)
+	 * then all-ones, else all-zeroes.
+	 */
+	want = (*b & ((xfs_rtword_t)1 << bit)) ? -1 : 0;
+	/*
+	 * If the starting position is not word-aligned, deal with the
+	 * partial word.
+	 */
+	if (bit) {
+		/*
+		 * Calculate last (rightmost) bit number to look at,
+		 * and mask for all the relevant bits in this word.
+		 */
+		lastbit = XFS_RTMIN(bit + len, XFS_NBWORD);
+		mask = (((xfs_rtword_t)1 << (lastbit - bit)) - 1) << bit;
+		/*
+		 * Calculate the difference between the value there
+		 * and what we're looking for.
+		 */
+		if ((wdiff = (*b ^ want) & mask)) {
+			/*
+			 * Different.  Mark where we are and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i = XFS_RTLOBIT(wdiff) - bit;
+			*rtblock = start + i - 1;
+			return 0;
+		}
+		i = lastbit - bit;
+		/*
+		 * Go on to next block if that's where the next word is
+		 * and we need the next word.
+		 */
+		if (++word == XFS_BLOCKWSIZE(mp) && i < len) {
+			/*
+			 * If done with this block, get the previous one.
+			 */
+			xfs_trans_brelse(tp, bp);
+			error = xfs_rtbuf_get(mp, tp, ++block, 0, &bp);
+			if (error) {
+				return error;
+			}
+			b = bufp = bp->b_addr;
+			word = 0;
+		} else {
+			/*
+			 * Go on to the previous word in the buffer.
+			 */
+			b++;
+		}
+	} else {
+		/*
+		 * Starting on a word boundary, no partial word.
+		 */
+		i = 0;
+	}
+	/*
+	 * Loop over whole words in buffers.  When we use up one buffer
+	 * we move on to the next one.
+	 */
+	while (len - i >= XFS_NBWORD) {
+		/*
+		 * Compute difference between actual and desired value.
+		 */
+		if ((wdiff = *b ^ want)) {
+			/*
+			 * Different, mark where we are and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i += XFS_RTLOBIT(wdiff);
+			*rtblock = start + i - 1;
+			return 0;
+		}
+		i += XFS_NBWORD;
+		/*
+		 * Go on to next block if that's where the next word is
+		 * and we need the next word.
+		 */
+		if (++word == XFS_BLOCKWSIZE(mp) && i < len) {
+			/*
+			 * If done with this block, get the next one.
+			 */
+			xfs_trans_brelse(tp, bp);
+			error = xfs_rtbuf_get(mp, tp, ++block, 0, &bp);
+			if (error) {
+				return error;
+			}
+			b = bufp = bp->b_addr;
+			word = 0;
+		} else {
+			/*
+			 * Go on to the next word in the buffer.
+			 */
+			b++;
+		}
+	}
+	/*
+	 * If not ending on a word boundary, deal with the last
+	 * (partial) word.
+	 */
+	if ((lastbit = len - i)) {
+		/*
+		 * Calculate mask for all the relevant bits in this word.
+		 */
+		mask = ((xfs_rtword_t)1 << lastbit) - 1;
+		/*
+		 * Compute difference between actual and desired value.
+		 */
+		if ((wdiff = (*b ^ want) & mask)) {
+			/*
+			 * Different, mark where we are and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i += XFS_RTLOBIT(wdiff);
+			*rtblock = start + i - 1;
+			return 0;
+		} else
+			i = len;
+	}
+	/*
+	 * No match, return that we scanned the whole area.
+	 */
+	xfs_trans_brelse(tp, bp);
+	*rtblock = start + i - 1;
+	return 0;
+}
+
+/*
+ * Read and/or modify the summary information for a given extent size,
+ * bitmap block combination.
+ * Keeps track of a current summary block, so we don't keep reading
+ * it from the buffer cache.
+ *
+ * Summary information is returned in *sum if specified.
+ * If no delta is specified, returns summary only.
+ */
+int
+xfs_rtmodify_summary_int(
+	xfs_mount_t	*mp,		/* file system mount structure */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	int		log,		/* log2 of extent size */
+	xfs_rtblock_t	bbno,		/* bitmap block number */
+	int		delta,		/* change to make to summary info */
+	struct xfs_buf	**rbpp,		/* in/out: summary block buffer */
+	xfs_fsblock_t	*rsb,		/* in/out: summary block number */
+	xfs_suminfo_t	*sum)		/* out: summary info for this block */
+{
+	struct xfs_buf	*bp;		/* buffer for the summary block */
+	int		error;		/* error value */
+	xfs_fsblock_t	sb;		/* summary fsblock */
+	int		so;		/* index into the summary file */
+	xfs_suminfo_t	*sp;		/* pointer to returned data */
+
+	/*
+	 * Compute entry number in the summary file.
+	 */
+	so = XFS_SUMOFFS(mp, log, bbno);
+	/*
+	 * Compute the block number in the summary file.
+	 */
+	sb = XFS_SUMOFFSTOBLOCK(mp, so);
+	/*
+	 * If we have an old buffer, and the block number matches, use that.
+	 */
+	if (*rbpp && *rsb == sb)
+		bp = *rbpp;
+	/*
+	 * Otherwise we have to get the buffer.
+	 */
+	else {
+		/*
+		 * If there was an old one, get rid of it first.
+		 */
+		if (*rbpp)
+			xfs_trans_brelse(tp, *rbpp);
+		error = xfs_rtbuf_get(mp, tp, sb, 1, &bp);
+		if (error) {
+			return error;
+		}
+		/*
+		 * Remember this buffer and block for the next call.
+		 */
+		*rbpp = bp;
+		*rsb = sb;
+	}
+	/*
+	 * Point to the summary information, modify/log it, and/or copy it out.
+	 */
+	sp = XFS_SUMPTR(mp, bp, so);
+	if (delta) {
+		uint first = (uint)((char *)sp - (char *)bp->b_addr);
+
+		*sp += delta;
+		if (mp->m_rsum_cache) {
+			if (*sp == 0 && log == mp->m_rsum_cache[bbno])
+				mp->m_rsum_cache[bbno]++;
+			if (*sp != 0 && log < mp->m_rsum_cache[bbno])
+				mp->m_rsum_cache[bbno] = log;
+		}
+		xfs_trans_log_buf(tp, bp, first, first + sizeof(*sp) - 1);
+	}
+	if (sum)
+		*sum = *sp;
+	return 0;
+}
+
+int
+xfs_rtmodify_summary(
+	xfs_mount_t	*mp,		/* file system mount structure */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	int		log,		/* log2 of extent size */
+	xfs_rtblock_t	bbno,		/* bitmap block number */
+	int		delta,		/* change to make to summary info */
+	struct xfs_buf	**rbpp,		/* in/out: summary block buffer */
+	xfs_fsblock_t	*rsb)		/* in/out: summary block number */
+{
+	return xfs_rtmodify_summary_int(mp, tp, log, bbno,
+					delta, rbpp, rsb, NULL);
+}
+
+/*
+ * Set the given range of bitmap bits to the given value.
+ * Do whatever I/O and logging is required.
+ */
+int
+xfs_rtmodify_range(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	start,		/* starting block to modify */
+	xfs_extlen_t	len,		/* length of extent to modify */
+	int		val)		/* 1 for free, 0 for allocated */
+{
+	xfs_rtword_t	*b;		/* current word in buffer */
+	int		bit;		/* bit number in the word */
+	xfs_rtblock_t	block;		/* bitmap block number */
+	struct xfs_buf	*bp;		/* buf for the block */
+	xfs_rtword_t	*bufp;		/* starting word in buffer */
+	int		error;		/* error value */
+	xfs_rtword_t	*first;		/* first used word in the buffer */
+	int		i;		/* current bit number rel. to start */
+	int		lastbit;	/* last useful bit in word */
+	xfs_rtword_t	mask;		/* mask o frelevant bits for value */
+	int		word;		/* word number in the buffer */
+
+	/*
+	 * Compute starting bitmap block number.
+	 */
+	block = XFS_BITTOBLOCK(mp, start);
+	/*
+	 * Read the bitmap block, and point to its data.
+	 */
+	error = xfs_rtbuf_get(mp, tp, block, 0, &bp);
+	if (error) {
+		return error;
+	}
+	bufp = bp->b_addr;
+	/*
+	 * Compute the starting word's address, and starting bit.
+	 */
+	word = XFS_BITTOWORD(mp, start);
+	first = b = &bufp[word];
+	bit = (int)(start & (XFS_NBWORD - 1));
+	/*
+	 * 0 (allocated) => all zeroes; 1 (free) => all ones.
+	 */
+	val = -val;
+	/*
+	 * If not starting on a word boundary, deal with the first
+	 * (partial) word.
+	 */
+	if (bit) {
+		/*
+		 * Compute first bit not changed and mask of relevant bits.
+		 */
+		lastbit = XFS_RTMIN(bit + len, XFS_NBWORD);
+		mask = (((xfs_rtword_t)1 << (lastbit - bit)) - 1) << bit;
+		/*
+		 * Set/clear the active bits.
+		 */
+		if (val)
+			*b |= mask;
+		else
+			*b &= ~mask;
+		i = lastbit - bit;
+		/*
+		 * Go on to the next block if that's where the next word is
+		 * and we need the next word.
+		 */
+		if (++word == XFS_BLOCKWSIZE(mp) && i < len) {
+			/*
+			 * Log the changed part of this block.
+			 * Get the next one.
+			 */
+			xfs_trans_log_buf(tp, bp,
+				(uint)((char *)first - (char *)bufp),
+				(uint)((char *)b - (char *)bufp));
+			error = xfs_rtbuf_get(mp, tp, ++block, 0, &bp);
+			if (error) {
+				return error;
+			}
+			first = b = bufp = bp->b_addr;
+			word = 0;
+		} else {
+			/*
+			 * Go on to the next word in the buffer
+			 */
+			b++;
+		}
+	} else {
+		/*
+		 * Starting on a word boundary, no partial word.
+		 */
+		i = 0;
+	}
+	/*
+	 * Loop over whole words in buffers.  When we use up one buffer
+	 * we move on to the next one.
+	 */
+	while (len - i >= XFS_NBWORD) {
+		/*
+		 * Set the word value correctly.
+		 */
+		*b = val;
+		i += XFS_NBWORD;
+		/*
+		 * Go on to the next block if that's where the next word is
+		 * and we need the next word.
+		 */
+		if (++word == XFS_BLOCKWSIZE(mp) && i < len) {
+			/*
+			 * Log the changed part of this block.
+			 * Get the next one.
+			 */
+			xfs_trans_log_buf(tp, bp,
+				(uint)((char *)first - (char *)bufp),
+				(uint)((char *)b - (char *)bufp));
+			error = xfs_rtbuf_get(mp, tp, ++block, 0, &bp);
+			if (error) {
+				return error;
+			}
+			first = b = bufp = bp->b_addr;
+			word = 0;
+		} else {
+			/*
+			 * Go on to the next word in the buffer
+			 */
+			b++;
+		}
+	}
+	/*
+	 * If not ending on a word boundary, deal with the last
+	 * (partial) word.
+	 */
+	if ((lastbit = len - i)) {
+		/*
+		 * Compute a mask of relevant bits.
+		 */
+		mask = ((xfs_rtword_t)1 << lastbit) - 1;
+		/*
+		 * Set/clear the active bits.
+		 */
+		if (val)
+			*b |= mask;
+		else
+			*b &= ~mask;
+		b++;
+	}
+	/*
+	 * Log any remaining changed bytes.
+	 */
+	if (b > first)
+		xfs_trans_log_buf(tp, bp, (uint)((char *)first - (char *)bufp),
+			(uint)((char *)b - (char *)bufp - 1));
+	return 0;
+}
+
+/*
+ * Mark an extent specified by start and len freed.
+ * Updates all the summary information as well as the bitmap.
+ */
+int
+xfs_rtfree_range(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	start,		/* starting block to free */
+	xfs_extlen_t	len,		/* length to free */
+	struct xfs_buf	**rbpp,		/* in/out: summary block buffer */
+	xfs_fsblock_t	*rsb)		/* in/out: summary block number */
+{
+	xfs_rtblock_t	end;		/* end of the freed extent */
+	int		error;		/* error value */
+	xfs_rtblock_t	postblock;	/* first block freed > end */
+	xfs_rtblock_t	preblock;	/* first block freed < start */
+
+	end = start + len - 1;
+	/*
+	 * Modify the bitmap to mark this extent freed.
+	 */
+	error = xfs_rtmodify_range(mp, tp, start, len, 1);
+	if (error) {
+		return error;
+	}
+	/*
+	 * Assume we're freeing out of the middle of an allocated extent.
+	 * We need to find the beginning and end of the extent so we can
+	 * properly update the summary.
+	 */
+	error = xfs_rtfind_back(mp, tp, start, 0, &preblock);
+	if (error) {
+		return error;
+	}
+	/*
+	 * Find the next allocated block (end of allocated extent).
+	 */
+	error = xfs_rtfind_forw(mp, tp, end, mp->m_sb.sb_rextents - 1,
+		&postblock);
+	if (error)
+		return error;
+	/*
+	 * If there are blocks not being freed at the front of the
+	 * old extent, add summary data for them to be allocated.
+	 */
+	if (preblock < start) {
+		error = xfs_rtmodify_summary(mp, tp,
+			XFS_RTBLOCKLOG(start - preblock),
+			XFS_BITTOBLOCK(mp, preblock), -1, rbpp, rsb);
+		if (error) {
+			return error;
+		}
+	}
+	/*
+	 * If there are blocks not being freed at the end of the
+	 * old extent, add summary data for them to be allocated.
+	 */
+	if (postblock > end) {
+		error = xfs_rtmodify_summary(mp, tp,
+			XFS_RTBLOCKLOG(postblock - end),
+			XFS_BITTOBLOCK(mp, end + 1), -1, rbpp, rsb);
+		if (error) {
+			return error;
+		}
+	}
+	/*
+	 * Increment the summary information corresponding to the entire
+	 * (new) free extent.
+	 */
+	error = xfs_rtmodify_summary(mp, tp,
+		XFS_RTBLOCKLOG(postblock + 1 - preblock),
+		XFS_BITTOBLOCK(mp, preblock), 1, rbpp, rsb);
+	return error;
+}
+
+/*
+ * Check that the given range is either all allocated (val = 0) or
+ * all free (val = 1).
+ */
+int
+xfs_rtcheck_range(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	start,		/* starting block number of extent */
+	xfs_extlen_t	len,		/* length of extent */
+	int		val,		/* 1 for free, 0 for allocated */
+	xfs_rtblock_t	*new,		/* out: first block not matching */
+	int		*stat)		/* out: 1 for matches, 0 for not */
+{
+	xfs_rtword_t	*b;		/* current word in buffer */
+	int		bit;		/* bit number in the word */
+	xfs_rtblock_t	block;		/* bitmap block number */
+	struct xfs_buf	*bp;		/* buf for the block */
+	xfs_rtword_t	*bufp;		/* starting word in buffer */
+	int		error;		/* error value */
+	xfs_rtblock_t	i;		/* current bit number rel. to start */
+	xfs_rtblock_t	lastbit;	/* last useful bit in word */
+	xfs_rtword_t	mask;		/* mask of relevant bits for value */
+	xfs_rtword_t	wdiff;		/* difference from wanted value */
+	int		word;		/* word number in the buffer */
+
+	/*
+	 * Compute starting bitmap block number
+	 */
+	block = XFS_BITTOBLOCK(mp, start);
+	/*
+	 * Read the bitmap block.
+	 */
+	error = xfs_rtbuf_get(mp, tp, block, 0, &bp);
+	if (error) {
+		return error;
+	}
+	bufp = bp->b_addr;
+	/*
+	 * Compute the starting word's address, and starting bit.
+	 */
+	word = XFS_BITTOWORD(mp, start);
+	b = &bufp[word];
+	bit = (int)(start & (XFS_NBWORD - 1));
+	/*
+	 * 0 (allocated) => all zero's; 1 (free) => all one's.
+	 */
+	val = -val;
+	/*
+	 * If not starting on a word boundary, deal with the first
+	 * (partial) word.
+	 */
+	if (bit) {
+		/*
+		 * Compute first bit not examined.
+		 */
+		lastbit = XFS_RTMIN(bit + len, XFS_NBWORD);
+		/*
+		 * Mask of relevant bits.
+		 */
+		mask = (((xfs_rtword_t)1 << (lastbit - bit)) - 1) << bit;
+		/*
+		 * Compute difference between actual and desired value.
+		 */
+		if ((wdiff = (*b ^ val) & mask)) {
+			/*
+			 * Different, compute first wrong bit and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i = XFS_RTLOBIT(wdiff) - bit;
+			*new = start + i;
+			*stat = 0;
+			return 0;
+		}
+		i = lastbit - bit;
+		/*
+		 * Go on to next block if that's where the next word is
+		 * and we need the next word.
+		 */
+		if (++word == XFS_BLOCKWSIZE(mp) && i < len) {
+			/*
+			 * If done with this block, get the next one.
+			 */
+			xfs_trans_brelse(tp, bp);
+			error = xfs_rtbuf_get(mp, tp, ++block, 0, &bp);
+			if (error) {
+				return error;
+			}
+			b = bufp = bp->b_addr;
+			word = 0;
+		} else {
+			/*
+			 * Go on to the next word in the buffer.
+			 */
+			b++;
+		}
+	} else {
+		/*
+		 * Starting on a word boundary, no partial word.
+		 */
+		i = 0;
+	}
+	/*
+	 * Loop over whole words in buffers.  When we use up one buffer
+	 * we move on to the next one.
+	 */
+	while (len - i >= XFS_NBWORD) {
+		/*
+		 * Compute difference between actual and desired value.
+		 */
+		if ((wdiff = *b ^ val)) {
+			/*
+			 * Different, compute first wrong bit and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i += XFS_RTLOBIT(wdiff);
+			*new = start + i;
+			*stat = 0;
+			return 0;
+		}
+		i += XFS_NBWORD;
+		/*
+		 * Go on to next block if that's where the next word is
+		 * and we need the next word.
+		 */
+		if (++word == XFS_BLOCKWSIZE(mp) && i < len) {
+			/*
+			 * If done with this block, get the next one.
+			 */
+			xfs_trans_brelse(tp, bp);
+			error = xfs_rtbuf_get(mp, tp, ++block, 0, &bp);
+			if (error) {
+				return error;
+			}
+			b = bufp = bp->b_addr;
+			word = 0;
+		} else {
+			/*
+			 * Go on to the next word in the buffer.
+			 */
+			b++;
+		}
+	}
+	/*
+	 * If not ending on a word boundary, deal with the last
+	 * (partial) word.
+	 */
+	if ((lastbit = len - i)) {
+		/*
+		 * Mask of relevant bits.
+		 */
+		mask = ((xfs_rtword_t)1 << lastbit) - 1;
+		/*
+		 * Compute difference between actual and desired value.
+		 */
+		if ((wdiff = (*b ^ val) & mask)) {
+			/*
+			 * Different, compute first wrong bit and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i += XFS_RTLOBIT(wdiff);
+			*new = start + i;
+			*stat = 0;
+			return 0;
+		} else
+			i = len;
+	}
+	/*
+	 * Successful, return.
+	 */
+	xfs_trans_brelse(tp, bp);
+	*new = start + i;
+	*stat = 1;
+	return 0;
+}
+
+#ifdef DEBUG
+/*
+ * Check that the given extent (block range) is allocated already.
+ */
+STATIC int				/* error */
+xfs_rtcheck_alloc_range(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	bno,		/* starting block number of extent */
+	xfs_extlen_t	len)		/* length of extent */
+{
+	xfs_rtblock_t	new;		/* dummy for xfs_rtcheck_range */
+	int		stat;
+	int		error;
+
+	error = xfs_rtcheck_range(mp, tp, bno, len, 0, &new, &stat);
+	if (error)
+		return error;
+	ASSERT(stat);
+	return 0;
+}
+#else
+#define xfs_rtcheck_alloc_range(m,t,b,l)	(0)
+#endif
+/*
+ * Free an extent in the realtime subvolume.  Length is expressed in
+ * realtime extents, as is the block number.
+ */
+int					/* error */
+xfs_rtfree_extent(
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	bno,		/* starting block number to free */
+	xfs_extlen_t	len)		/* length of extent freed */
+{
+	int		error;		/* error value */
+	xfs_mount_t	*mp;		/* file system mount structure */
+	xfs_fsblock_t	sb;		/* summary file block number */
+	struct xfs_buf	*sumbp = NULL;	/* summary file block buffer */
+
+	mp = tp->t_mountp;
+
+	ASSERT(mp->m_rbmip->i_itemp != NULL);
+	ASSERT(xfs_isilocked(mp->m_rbmip, XFS_ILOCK_EXCL));
+
+	error = xfs_rtcheck_alloc_range(mp, tp, bno, len);
+	if (error)
+		return error;
+
+	/*
+	 * Free the range of realtime blocks.
+	 */
+	error = xfs_rtfree_range(mp, tp, bno, len, &sumbp, &sb);
+	if (error) {
+		return error;
+	}
+	/*
+	 * Mark more blocks free in the superblock.
+	 */
+	xfs_trans_mod_sb(tp, XFS_TRANS_SB_FREXTENTS, (long)len);
+	/*
+	 * If we've now freed all the blocks, reset the file sequence
+	 * number to 0.
+	 */
+	if (tp->t_frextents_delta + mp->m_sb.sb_frextents ==
+	    mp->m_sb.sb_rextents) {
+		if (!(mp->m_rbmip->i_diflags & XFS_DIFLAG_NEWRTBM))
+			mp->m_rbmip->i_diflags |= XFS_DIFLAG_NEWRTBM;
+		*(uint64_t *)&VFS_I(mp->m_rbmip)->i_atime = 0;
+		xfs_trans_log_inode(tp, mp->m_rbmip, XFS_ILOG_CORE);
+	}
+	return 0;
+}
+
+/* Find all the free records within a given range. */
+int
+xfs_rtalloc_query_range(
+	struct xfs_mount		*mp,
+	struct xfs_trans		*tp,
+	const struct xfs_rtalloc_rec	*low_rec,
+	const struct xfs_rtalloc_rec	*high_rec,
+	xfs_rtalloc_query_range_fn	fn,
+	void				*priv)
+{
+	struct xfs_rtalloc_rec		rec;
+	xfs_rtblock_t			rtstart;
+	xfs_rtblock_t			rtend;
+	xfs_rtblock_t			high_key;
+	int				is_free;
+	int				error = 0;
+
+	if (low_rec->ar_startext > high_rec->ar_startext)
+		return -EINVAL;
+	if (low_rec->ar_startext >= mp->m_sb.sb_rextents ||
+	    low_rec->ar_startext == high_rec->ar_startext)
+		return 0;
+
+	high_key = min(high_rec->ar_startext, mp->m_sb.sb_rextents - 1);
+
+	/* Iterate the bitmap, looking for discrepancies. */
+	rtstart = low_rec->ar_startext;
+	while (rtstart <= high_key) {
+		/* Is the first block free? */
+		error = xfs_rtcheck_range(mp, tp, rtstart, 1, 1, &rtend,
+				&is_free);
+		if (error)
+			break;
+
+		/* How long does the extent go for? */
+		error = xfs_rtfind_forw(mp, tp, rtstart, high_key, &rtend);
+		if (error)
+			break;
+
+		if (is_free) {
+			rec.ar_startext = rtstart;
+			rec.ar_extcount = rtend - rtstart + 1;
+
+			error = fn(mp, tp, &rec, priv);
+			if (error)
+				break;
+		}
+
+		rtstart = rtend + 1;
+	}
+
+	return error;
+}
+
+/* Find all the free records. */
+int
+xfs_rtalloc_query_all(
+	struct xfs_mount		*mp,
+	struct xfs_trans		*tp,
+	xfs_rtalloc_query_range_fn	fn,
+	void				*priv)
+{
+	struct xfs_rtalloc_rec		keys[2];
+
+	keys[0].ar_startext = 0;
+	keys[1].ar_startext = mp->m_sb.sb_rextents - 1;
+	keys[0].ar_extcount = keys[1].ar_extcount = 0;
+
+	return xfs_rtalloc_query_range(mp, tp, &keys[0], &keys[1], fn, priv);
+}
+
+/* Is the given extent all free? */
+int
+xfs_rtalloc_extent_is_free(
+	struct xfs_mount		*mp,
+	struct xfs_trans		*tp,
+	xfs_rtblock_t			start,
+	xfs_extlen_t			len,
+	bool				*is_free)
+{
+	xfs_rtblock_t			end;
+	int				matches;
+	int				error;
+
+	error = xfs_rtcheck_range(mp, tp, start, len, 1, &end, &matches);
+	if (error)
+		return error;
+
+	*is_free = matches;
+	return 0;
+}
diff --git a/fs/xfs/libxfs/xfs_sb.c b/fs/xfs/libxfs/xfs_sb.c
new file mode 100644
index 000000000..b6a584e04
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_sb.c
@@ -0,0 +1,1317 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_bit.h"
+#include "xfs_sb.h"
+#include "xfs_mount.h"
+#include "xfs_ialloc.h"
+#include "xfs_alloc.h"
+#include "xfs_error.h"
+#include "xfs_trans.h"
+#include "xfs_buf_item.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_log.h"
+#include "xfs_rmap_btree.h"
+#include "xfs_refcount_btree.h"
+#include "xfs_da_format.h"
+#include "xfs_health.h"
+#include "xfs_ag.h"
+
+/*
+ * Physical superblock buffer manipulations. Shared with libxfs in userspace.
+ */
+
+/*
+ * Check that all the V4 feature bits that the V5 filesystem format requires are
+ * correctly set.
+ */
+static bool
+xfs_sb_validate_v5_features(
+	struct xfs_sb	*sbp)
+{
+	/* We must not have any unknown V4 feature bits set */
+	if (sbp->sb_versionnum & ~XFS_SB_VERSION_OKBITS)
+		return false;
+
+	/*
+	 * The CRC bit is considered an invalid V4 flag, so we have to add it
+	 * manually to the OKBITS mask.
+	 */
+	if (sbp->sb_features2 & ~(XFS_SB_VERSION2_OKBITS |
+				  XFS_SB_VERSION2_CRCBIT))
+		return false;
+
+	/* Now check all the required V4 feature flags are set. */
+
+#define V5_VERS_FLAGS	(XFS_SB_VERSION_NLINKBIT	| \
+			XFS_SB_VERSION_ALIGNBIT		| \
+			XFS_SB_VERSION_LOGV2BIT		| \
+			XFS_SB_VERSION_EXTFLGBIT	| \
+			XFS_SB_VERSION_DIRV2BIT		| \
+			XFS_SB_VERSION_MOREBITSBIT)
+
+#define V5_FEAT_FLAGS	(XFS_SB_VERSION2_LAZYSBCOUNTBIT	| \
+			XFS_SB_VERSION2_ATTR2BIT	| \
+			XFS_SB_VERSION2_PROJID32BIT	| \
+			XFS_SB_VERSION2_CRCBIT)
+
+	if ((sbp->sb_versionnum & V5_VERS_FLAGS) != V5_VERS_FLAGS)
+		return false;
+	if ((sbp->sb_features2 & V5_FEAT_FLAGS) != V5_FEAT_FLAGS)
+		return false;
+	return true;
+}
+
+/*
+ * We current support XFS v5 formats with known features and v4 superblocks with
+ * at least V2 directories.
+ */
+bool
+xfs_sb_good_version(
+	struct xfs_sb	*sbp)
+{
+	/*
+	 * All v5 filesystems are supported, but we must check that all the
+	 * required v4 feature flags are enabled correctly as the code checks
+	 * those flags and not for v5 support.
+	 */
+	if (xfs_sb_is_v5(sbp))
+		return xfs_sb_validate_v5_features(sbp);
+
+	/* versions prior to v4 are not supported */
+	if (XFS_SB_VERSION_NUM(sbp) != XFS_SB_VERSION_4)
+		return false;
+
+	/* We must not have any unknown v4 feature bits set */
+	if ((sbp->sb_versionnum & ~XFS_SB_VERSION_OKBITS) ||
+	    ((sbp->sb_versionnum & XFS_SB_VERSION_MOREBITSBIT) &&
+	     (sbp->sb_features2 & ~XFS_SB_VERSION2_OKBITS)))
+		return false;
+
+	/* V4 filesystems need v2 directories and unwritten extents */
+	if (!(sbp->sb_versionnum & XFS_SB_VERSION_DIRV2BIT))
+		return false;
+	if (!(sbp->sb_versionnum & XFS_SB_VERSION_EXTFLGBIT))
+		return false;
+
+	/* It's a supported v4 filesystem */
+	return true;
+}
+
+uint64_t
+xfs_sb_version_to_features(
+	struct xfs_sb	*sbp)
+{
+	uint64_t	features = 0;
+
+	/* optional V4 features */
+	if (sbp->sb_rblocks > 0)
+		features |= XFS_FEAT_REALTIME;
+	if (sbp->sb_versionnum & XFS_SB_VERSION_NLINKBIT)
+		features |= XFS_FEAT_NLINK;
+	if (sbp->sb_versionnum & XFS_SB_VERSION_ATTRBIT)
+		features |= XFS_FEAT_ATTR;
+	if (sbp->sb_versionnum & XFS_SB_VERSION_QUOTABIT)
+		features |= XFS_FEAT_QUOTA;
+	if (sbp->sb_versionnum & XFS_SB_VERSION_ALIGNBIT)
+		features |= XFS_FEAT_ALIGN;
+	if (sbp->sb_versionnum & XFS_SB_VERSION_LOGV2BIT)
+		features |= XFS_FEAT_LOGV2;
+	if (sbp->sb_versionnum & XFS_SB_VERSION_DALIGNBIT)
+		features |= XFS_FEAT_DALIGN;
+	if (sbp->sb_versionnum & XFS_SB_VERSION_EXTFLGBIT)
+		features |= XFS_FEAT_EXTFLG;
+	if (sbp->sb_versionnum & XFS_SB_VERSION_SECTORBIT)
+		features |= XFS_FEAT_SECTOR;
+	if (sbp->sb_versionnum & XFS_SB_VERSION_BORGBIT)
+		features |= XFS_FEAT_ASCIICI;
+	if (sbp->sb_versionnum & XFS_SB_VERSION_MOREBITSBIT) {
+		if (sbp->sb_features2 & XFS_SB_VERSION2_LAZYSBCOUNTBIT)
+			features |= XFS_FEAT_LAZYSBCOUNT;
+		if (sbp->sb_features2 & XFS_SB_VERSION2_ATTR2BIT)
+			features |= XFS_FEAT_ATTR2;
+		if (sbp->sb_features2 & XFS_SB_VERSION2_PROJID32BIT)
+			features |= XFS_FEAT_PROJID32;
+		if (sbp->sb_features2 & XFS_SB_VERSION2_FTYPE)
+			features |= XFS_FEAT_FTYPE;
+	}
+
+	if (!xfs_sb_is_v5(sbp))
+		return features;
+
+	/* Always on V5 features */
+	features |= XFS_FEAT_ALIGN | XFS_FEAT_LOGV2 | XFS_FEAT_EXTFLG |
+		    XFS_FEAT_LAZYSBCOUNT | XFS_FEAT_ATTR2 | XFS_FEAT_PROJID32 |
+		    XFS_FEAT_V3INODES | XFS_FEAT_CRC | XFS_FEAT_PQUOTINO;
+
+	/* Optional V5 features */
+	if (sbp->sb_features_ro_compat & XFS_SB_FEAT_RO_COMPAT_FINOBT)
+		features |= XFS_FEAT_FINOBT;
+	if (sbp->sb_features_ro_compat & XFS_SB_FEAT_RO_COMPAT_RMAPBT)
+		features |= XFS_FEAT_RMAPBT;
+	if (sbp->sb_features_ro_compat & XFS_SB_FEAT_RO_COMPAT_REFLINK)
+		features |= XFS_FEAT_REFLINK;
+	if (sbp->sb_features_ro_compat & XFS_SB_FEAT_RO_COMPAT_INOBTCNT)
+		features |= XFS_FEAT_INOBTCNT;
+	if (sbp->sb_features_incompat & XFS_SB_FEAT_INCOMPAT_FTYPE)
+		features |= XFS_FEAT_FTYPE;
+	if (sbp->sb_features_incompat & XFS_SB_FEAT_INCOMPAT_SPINODES)
+		features |= XFS_FEAT_SPINODES;
+	if (sbp->sb_features_incompat & XFS_SB_FEAT_INCOMPAT_META_UUID)
+		features |= XFS_FEAT_META_UUID;
+	if (sbp->sb_features_incompat & XFS_SB_FEAT_INCOMPAT_BIGTIME)
+		features |= XFS_FEAT_BIGTIME;
+	if (sbp->sb_features_incompat & XFS_SB_FEAT_INCOMPAT_NEEDSREPAIR)
+		features |= XFS_FEAT_NEEDSREPAIR;
+	if (sbp->sb_features_incompat & XFS_SB_FEAT_INCOMPAT_NREXT64)
+		features |= XFS_FEAT_NREXT64;
+
+	return features;
+}
+
+/* Check all the superblock fields we care about when reading one in. */
+STATIC int
+xfs_validate_sb_read(
+	struct xfs_mount	*mp,
+	struct xfs_sb		*sbp)
+{
+	if (!xfs_sb_is_v5(sbp))
+		return 0;
+
+	/*
+	 * Version 5 superblock feature mask validation. Reject combinations
+	 * the kernel cannot support up front before checking anything else.
+	 */
+	if (xfs_sb_has_compat_feature(sbp, XFS_SB_FEAT_COMPAT_UNKNOWN)) {
+		xfs_warn(mp,
+"Superblock has unknown compatible features (0x%x) enabled.",
+			(sbp->sb_features_compat & XFS_SB_FEAT_COMPAT_UNKNOWN));
+		xfs_warn(mp,
+"Using a more recent kernel is recommended.");
+	}
+
+	if (xfs_sb_has_ro_compat_feature(sbp, XFS_SB_FEAT_RO_COMPAT_UNKNOWN)) {
+		xfs_alert(mp,
+"Superblock has unknown read-only compatible features (0x%x) enabled.",
+			(sbp->sb_features_ro_compat &
+					XFS_SB_FEAT_RO_COMPAT_UNKNOWN));
+		if (!xfs_is_readonly(mp)) {
+			xfs_warn(mp,
+"Attempted to mount read-only compatible filesystem read-write.");
+			xfs_warn(mp,
+"Filesystem can only be safely mounted read only.");
+
+			return -EINVAL;
+		}
+	}
+	if (xfs_sb_has_incompat_feature(sbp, XFS_SB_FEAT_INCOMPAT_UNKNOWN)) {
+		xfs_warn(mp,
+"Superblock has unknown incompatible features (0x%x) enabled.",
+			(sbp->sb_features_incompat &
+					XFS_SB_FEAT_INCOMPAT_UNKNOWN));
+		xfs_warn(mp,
+"Filesystem cannot be safely mounted by this kernel.");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+/* Check all the superblock fields we care about when writing one out. */
+STATIC int
+xfs_validate_sb_write(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp,
+	struct xfs_sb		*sbp)
+{
+	/*
+	 * Carry out additional sb summary counter sanity checks when we write
+	 * the superblock.  We skip this in the read validator because there
+	 * could be newer superblocks in the log and if the values are garbage
+	 * even after replay we'll recalculate them at the end of log mount.
+	 *
+	 * mkfs has traditionally written zeroed counters to inprogress and
+	 * secondary superblocks, so allow this usage to continue because
+	 * we never read counters from such superblocks.
+	 */
+	if (xfs_buf_daddr(bp) == XFS_SB_DADDR && !sbp->sb_inprogress &&
+	    (sbp->sb_fdblocks > sbp->sb_dblocks ||
+	     !xfs_verify_icount(mp, sbp->sb_icount) ||
+	     sbp->sb_ifree > sbp->sb_icount)) {
+		xfs_warn(mp, "SB summary counter sanity check failed");
+		return -EFSCORRUPTED;
+	}
+
+	if (!xfs_sb_is_v5(sbp))
+		return 0;
+
+	/*
+	 * Version 5 superblock feature mask validation. Reject combinations
+	 * the kernel cannot support since we checked for unsupported bits in
+	 * the read verifier, which means that memory is corrupt.
+	 */
+	if (xfs_sb_has_compat_feature(sbp, XFS_SB_FEAT_COMPAT_UNKNOWN)) {
+		xfs_warn(mp,
+"Corruption detected in superblock compatible features (0x%x)!",
+			(sbp->sb_features_compat & XFS_SB_FEAT_COMPAT_UNKNOWN));
+		return -EFSCORRUPTED;
+	}
+
+	if (xfs_sb_has_ro_compat_feature(sbp, XFS_SB_FEAT_RO_COMPAT_UNKNOWN)) {
+		xfs_alert(mp,
+"Corruption detected in superblock read-only compatible features (0x%x)!",
+			(sbp->sb_features_ro_compat &
+					XFS_SB_FEAT_RO_COMPAT_UNKNOWN));
+		return -EFSCORRUPTED;
+	}
+	if (xfs_sb_has_incompat_feature(sbp, XFS_SB_FEAT_INCOMPAT_UNKNOWN)) {
+		xfs_warn(mp,
+"Corruption detected in superblock incompatible features (0x%x)!",
+			(sbp->sb_features_incompat &
+					XFS_SB_FEAT_INCOMPAT_UNKNOWN));
+		return -EFSCORRUPTED;
+	}
+	if (xfs_sb_has_incompat_log_feature(sbp,
+			XFS_SB_FEAT_INCOMPAT_LOG_UNKNOWN)) {
+		xfs_warn(mp,
+"Corruption detected in superblock incompatible log features (0x%x)!",
+			(sbp->sb_features_log_incompat &
+					XFS_SB_FEAT_INCOMPAT_LOG_UNKNOWN));
+		return -EFSCORRUPTED;
+	}
+
+	/*
+	 * We can't read verify the sb LSN because the read verifier is called
+	 * before the log is allocated and processed. We know the log is set up
+	 * before write verifier calls, so check it here.
+	 */
+	if (!xfs_log_check_lsn(mp, sbp->sb_lsn))
+		return -EFSCORRUPTED;
+
+	return 0;
+}
+
+/* Check the validity of the SB. */
+STATIC int
+xfs_validate_sb_common(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp,
+	struct xfs_sb		*sbp)
+{
+	struct xfs_dsb		*dsb = bp->b_addr;
+	uint32_t		agcount = 0;
+	uint32_t		rem;
+	bool			has_dalign;
+
+	if (!xfs_verify_magic(bp, dsb->sb_magicnum)) {
+		xfs_warn(mp,
+"Superblock has bad magic number 0x%x. Not an XFS filesystem?",
+			be32_to_cpu(dsb->sb_magicnum));
+		return -EWRONGFS;
+	}
+
+	if (!xfs_sb_good_version(sbp)) {
+		xfs_warn(mp,
+"Superblock has unknown features enabled or corrupted feature masks.");
+		return -EWRONGFS;
+	}
+
+	/*
+	 * Validate feature flags and state
+	 */
+	if (xfs_sb_is_v5(sbp)) {
+		if (sbp->sb_blocksize < XFS_MIN_CRC_BLOCKSIZE) {
+			xfs_notice(mp,
+"Block size (%u bytes) too small for Version 5 superblock (minimum %d bytes)",
+				sbp->sb_blocksize, XFS_MIN_CRC_BLOCKSIZE);
+			return -EFSCORRUPTED;
+		}
+
+		/* V5 has a separate project quota inode */
+		if (sbp->sb_qflags & (XFS_OQUOTA_ENFD | XFS_OQUOTA_CHKD)) {
+			xfs_notice(mp,
+			   "Version 5 of Super block has XFS_OQUOTA bits.");
+			return -EFSCORRUPTED;
+		}
+
+		/*
+		 * Full inode chunks must be aligned to inode chunk size when
+		 * sparse inodes are enabled to support the sparse chunk
+		 * allocation algorithm and prevent overlapping inode records.
+		 */
+		if (sbp->sb_features_incompat & XFS_SB_FEAT_INCOMPAT_SPINODES) {
+			uint32_t	align;
+
+			align = XFS_INODES_PER_CHUNK * sbp->sb_inodesize
+					>> sbp->sb_blocklog;
+			if (sbp->sb_inoalignmt != align) {
+				xfs_warn(mp,
+"Inode block alignment (%u) must match chunk size (%u) for sparse inodes.",
+					 sbp->sb_inoalignmt, align);
+				return -EINVAL;
+			}
+		}
+	} else if (sbp->sb_qflags & (XFS_PQUOTA_ENFD | XFS_GQUOTA_ENFD |
+				XFS_PQUOTA_CHKD | XFS_GQUOTA_CHKD)) {
+			xfs_notice(mp,
+"Superblock earlier than Version 5 has XFS_{P|G}QUOTA_{ENFD|CHKD} bits.");
+			return -EFSCORRUPTED;
+	}
+
+	if (unlikely(
+	    sbp->sb_logstart == 0 && mp->m_logdev_targp == mp->m_ddev_targp)) {
+		xfs_warn(mp,
+		"filesystem is marked as having an external log; "
+		"specify logdev on the mount command line.");
+		return -EINVAL;
+	}
+
+	if (unlikely(
+	    sbp->sb_logstart != 0 && mp->m_logdev_targp != mp->m_ddev_targp)) {
+		xfs_warn(mp,
+		"filesystem is marked as having an internal log; "
+		"do not specify logdev on the mount command line.");
+		return -EINVAL;
+	}
+
+	/* Compute agcount for this number of dblocks and agblocks */
+	if (sbp->sb_agblocks) {
+		agcount = div_u64_rem(sbp->sb_dblocks, sbp->sb_agblocks, &rem);
+		if (rem)
+			agcount++;
+	}
+
+	/*
+	 * More sanity checking.  Most of these were stolen directly from
+	 * xfs_repair.
+	 */
+	if (unlikely(
+	    sbp->sb_agcount <= 0					||
+	    sbp->sb_sectsize < XFS_MIN_SECTORSIZE			||
+	    sbp->sb_sectsize > XFS_MAX_SECTORSIZE			||
+	    sbp->sb_sectlog < XFS_MIN_SECTORSIZE_LOG			||
+	    sbp->sb_sectlog > XFS_MAX_SECTORSIZE_LOG			||
+	    sbp->sb_sectsize != (1 << sbp->sb_sectlog)			||
+	    sbp->sb_blocksize < XFS_MIN_BLOCKSIZE			||
+	    sbp->sb_blocksize > XFS_MAX_BLOCKSIZE			||
+	    sbp->sb_blocklog < XFS_MIN_BLOCKSIZE_LOG			||
+	    sbp->sb_blocklog > XFS_MAX_BLOCKSIZE_LOG			||
+	    sbp->sb_blocksize != (1 << sbp->sb_blocklog)		||
+	    sbp->sb_dirblklog + sbp->sb_blocklog > XFS_MAX_BLOCKSIZE_LOG ||
+	    sbp->sb_inodesize < XFS_DINODE_MIN_SIZE			||
+	    sbp->sb_inodesize > XFS_DINODE_MAX_SIZE			||
+	    sbp->sb_inodelog < XFS_DINODE_MIN_LOG			||
+	    sbp->sb_inodelog > XFS_DINODE_MAX_LOG			||
+	    sbp->sb_inodesize != (1 << sbp->sb_inodelog)		||
+	    sbp->sb_logsunit > XLOG_MAX_RECORD_BSIZE			||
+	    sbp->sb_inopblock != howmany(sbp->sb_blocksize,sbp->sb_inodesize) ||
+	    XFS_FSB_TO_B(mp, sbp->sb_agblocks) < XFS_MIN_AG_BYTES	||
+	    XFS_FSB_TO_B(mp, sbp->sb_agblocks) > XFS_MAX_AG_BYTES	||
+	    sbp->sb_agblklog != xfs_highbit32(sbp->sb_agblocks - 1) + 1	||
+	    agcount == 0 || agcount != sbp->sb_agcount			||
+	    (sbp->sb_blocklog - sbp->sb_inodelog != sbp->sb_inopblog)	||
+	    (sbp->sb_rextsize * sbp->sb_blocksize > XFS_MAX_RTEXTSIZE)	||
+	    (sbp->sb_rextsize * sbp->sb_blocksize < XFS_MIN_RTEXTSIZE)	||
+	    (sbp->sb_imax_pct > 100 /* zero sb_imax_pct is valid */)	||
+	    sbp->sb_dblocks == 0					||
+	    sbp->sb_dblocks > XFS_MAX_DBLOCKS(sbp)			||
+	    sbp->sb_dblocks < XFS_MIN_DBLOCKS(sbp)			||
+	    sbp->sb_shared_vn != 0)) {
+		xfs_notice(mp, "SB sanity check failed");
+		return -EFSCORRUPTED;
+	}
+
+	/* Validate the realtime geometry; stolen from xfs_repair */
+	if (sbp->sb_rextsize * sbp->sb_blocksize > XFS_MAX_RTEXTSIZE ||
+	    sbp->sb_rextsize * sbp->sb_blocksize < XFS_MIN_RTEXTSIZE) {
+		xfs_notice(mp,
+			"realtime extent sanity check failed");
+		return -EFSCORRUPTED;
+	}
+
+	if (sbp->sb_rblocks == 0) {
+		if (sbp->sb_rextents != 0 || sbp->sb_rbmblocks != 0 ||
+		    sbp->sb_rextslog != 0 || sbp->sb_frextents != 0) {
+			xfs_notice(mp,
+				"realtime zeroed geometry check failed");
+			return -EFSCORRUPTED;
+		}
+	} else {
+		uint64_t	rexts;
+		uint64_t	rbmblocks;
+
+		rexts = div_u64(sbp->sb_rblocks, sbp->sb_rextsize);
+		rbmblocks = howmany_64(sbp->sb_rextents,
+				       NBBY * sbp->sb_blocksize);
+
+		if (sbp->sb_rextents != rexts ||
+		    sbp->sb_rextslog != xfs_highbit32(sbp->sb_rextents) ||
+		    sbp->sb_rbmblocks != rbmblocks) {
+			xfs_notice(mp,
+				"realtime geometry sanity check failed");
+			return -EFSCORRUPTED;
+		}
+	}
+
+	/*
+	 * Either (sb_unit and !hasdalign) or (!sb_unit and hasdalign)
+	 * would imply the image is corrupted.
+	 */
+	has_dalign = sbp->sb_versionnum & XFS_SB_VERSION_DALIGNBIT;
+	if (!!sbp->sb_unit ^ has_dalign) {
+		xfs_notice(mp, "SB stripe alignment sanity check failed");
+		return -EFSCORRUPTED;
+	}
+
+	if (!xfs_validate_stripe_geometry(mp, XFS_FSB_TO_B(mp, sbp->sb_unit),
+			XFS_FSB_TO_B(mp, sbp->sb_width), 0, false))
+		return -EFSCORRUPTED;
+
+	/*
+	 * Currently only very few inode sizes are supported.
+	 */
+	switch (sbp->sb_inodesize) {
+	case 256:
+	case 512:
+	case 1024:
+	case 2048:
+		break;
+	default:
+		xfs_warn(mp, "inode size of %d bytes not supported",
+				sbp->sb_inodesize);
+		return -ENOSYS;
+	}
+
+	return 0;
+}
+
+void
+xfs_sb_quota_from_disk(struct xfs_sb *sbp)
+{
+	/*
+	 * older mkfs doesn't initialize quota inodes to NULLFSINO. This
+	 * leads to in-core values having two different values for a quota
+	 * inode to be invalid: 0 and NULLFSINO. Change it to a single value
+	 * NULLFSINO.
+	 *
+	 * Note that this change affect only the in-core values. These
+	 * values are not written back to disk unless any quota information
+	 * is written to the disk. Even in that case, sb_pquotino field is
+	 * not written to disk unless the superblock supports pquotino.
+	 */
+	if (sbp->sb_uquotino == 0)
+		sbp->sb_uquotino = NULLFSINO;
+	if (sbp->sb_gquotino == 0)
+		sbp->sb_gquotino = NULLFSINO;
+	if (sbp->sb_pquotino == 0)
+		sbp->sb_pquotino = NULLFSINO;
+
+	/*
+	 * We need to do these manipilations only if we are working
+	 * with an older version of on-disk superblock.
+	 */
+	if (xfs_sb_is_v5(sbp))
+		return;
+
+	if (sbp->sb_qflags & XFS_OQUOTA_ENFD)
+		sbp->sb_qflags |= (sbp->sb_qflags & XFS_PQUOTA_ACCT) ?
+					XFS_PQUOTA_ENFD : XFS_GQUOTA_ENFD;
+	if (sbp->sb_qflags & XFS_OQUOTA_CHKD)
+		sbp->sb_qflags |= (sbp->sb_qflags & XFS_PQUOTA_ACCT) ?
+					XFS_PQUOTA_CHKD : XFS_GQUOTA_CHKD;
+	sbp->sb_qflags &= ~(XFS_OQUOTA_ENFD | XFS_OQUOTA_CHKD);
+
+	if (sbp->sb_qflags & XFS_PQUOTA_ACCT &&
+	    sbp->sb_gquotino != NULLFSINO)  {
+		/*
+		 * In older version of superblock, on-disk superblock only
+		 * has sb_gquotino, and in-core superblock has both sb_gquotino
+		 * and sb_pquotino. But, only one of them is supported at any
+		 * point of time. So, if PQUOTA is set in disk superblock,
+		 * copy over sb_gquotino to sb_pquotino.  The NULLFSINO test
+		 * above is to make sure we don't do this twice and wipe them
+		 * both out!
+		 */
+		sbp->sb_pquotino = sbp->sb_gquotino;
+		sbp->sb_gquotino = NULLFSINO;
+	}
+}
+
+static void
+__xfs_sb_from_disk(
+	struct xfs_sb	*to,
+	struct xfs_dsb	*from,
+	bool		convert_xquota)
+{
+	to->sb_magicnum = be32_to_cpu(from->sb_magicnum);
+	to->sb_blocksize = be32_to_cpu(from->sb_blocksize);
+	to->sb_dblocks = be64_to_cpu(from->sb_dblocks);
+	to->sb_rblocks = be64_to_cpu(from->sb_rblocks);
+	to->sb_rextents = be64_to_cpu(from->sb_rextents);
+	memcpy(&to->sb_uuid, &from->sb_uuid, sizeof(to->sb_uuid));
+	to->sb_logstart = be64_to_cpu(from->sb_logstart);
+	to->sb_rootino = be64_to_cpu(from->sb_rootino);
+	to->sb_rbmino = be64_to_cpu(from->sb_rbmino);
+	to->sb_rsumino = be64_to_cpu(from->sb_rsumino);
+	to->sb_rextsize = be32_to_cpu(from->sb_rextsize);
+	to->sb_agblocks = be32_to_cpu(from->sb_agblocks);
+	to->sb_agcount = be32_to_cpu(from->sb_agcount);
+	to->sb_rbmblocks = be32_to_cpu(from->sb_rbmblocks);
+	to->sb_logblocks = be32_to_cpu(from->sb_logblocks);
+	to->sb_versionnum = be16_to_cpu(from->sb_versionnum);
+	to->sb_sectsize = be16_to_cpu(from->sb_sectsize);
+	to->sb_inodesize = be16_to_cpu(from->sb_inodesize);
+	to->sb_inopblock = be16_to_cpu(from->sb_inopblock);
+	memcpy(&to->sb_fname, &from->sb_fname, sizeof(to->sb_fname));
+	to->sb_blocklog = from->sb_blocklog;
+	to->sb_sectlog = from->sb_sectlog;
+	to->sb_inodelog = from->sb_inodelog;
+	to->sb_inopblog = from->sb_inopblog;
+	to->sb_agblklog = from->sb_agblklog;
+	to->sb_rextslog = from->sb_rextslog;
+	to->sb_inprogress = from->sb_inprogress;
+	to->sb_imax_pct = from->sb_imax_pct;
+	to->sb_icount = be64_to_cpu(from->sb_icount);
+	to->sb_ifree = be64_to_cpu(from->sb_ifree);
+	to->sb_fdblocks = be64_to_cpu(from->sb_fdblocks);
+	to->sb_frextents = be64_to_cpu(from->sb_frextents);
+	to->sb_uquotino = be64_to_cpu(from->sb_uquotino);
+	to->sb_gquotino = be64_to_cpu(from->sb_gquotino);
+	to->sb_qflags = be16_to_cpu(from->sb_qflags);
+	to->sb_flags = from->sb_flags;
+	to->sb_shared_vn = from->sb_shared_vn;
+	to->sb_inoalignmt = be32_to_cpu(from->sb_inoalignmt);
+	to->sb_unit = be32_to_cpu(from->sb_unit);
+	to->sb_width = be32_to_cpu(from->sb_width);
+	to->sb_dirblklog = from->sb_dirblklog;
+	to->sb_logsectlog = from->sb_logsectlog;
+	to->sb_logsectsize = be16_to_cpu(from->sb_logsectsize);
+	to->sb_logsunit = be32_to_cpu(from->sb_logsunit);
+	to->sb_features2 = be32_to_cpu(from->sb_features2);
+	to->sb_bad_features2 = be32_to_cpu(from->sb_bad_features2);
+	to->sb_features_compat = be32_to_cpu(from->sb_features_compat);
+	to->sb_features_ro_compat = be32_to_cpu(from->sb_features_ro_compat);
+	to->sb_features_incompat = be32_to_cpu(from->sb_features_incompat);
+	to->sb_features_log_incompat =
+				be32_to_cpu(from->sb_features_log_incompat);
+	/* crc is only used on disk, not in memory; just init to 0 here. */
+	to->sb_crc = 0;
+	to->sb_spino_align = be32_to_cpu(from->sb_spino_align);
+	to->sb_pquotino = be64_to_cpu(from->sb_pquotino);
+	to->sb_lsn = be64_to_cpu(from->sb_lsn);
+	/*
+	 * sb_meta_uuid is only on disk if it differs from sb_uuid and the
+	 * feature flag is set; if not set we keep it only in memory.
+	 */
+	if (xfs_sb_is_v5(to) &&
+	    (to->sb_features_incompat & XFS_SB_FEAT_INCOMPAT_META_UUID))
+		uuid_copy(&to->sb_meta_uuid, &from->sb_meta_uuid);
+	else
+		uuid_copy(&to->sb_meta_uuid, &from->sb_uuid);
+	/* Convert on-disk flags to in-memory flags? */
+	if (convert_xquota)
+		xfs_sb_quota_from_disk(to);
+}
+
+void
+xfs_sb_from_disk(
+	struct xfs_sb	*to,
+	struct xfs_dsb	*from)
+{
+	__xfs_sb_from_disk(to, from, true);
+}
+
+static void
+xfs_sb_quota_to_disk(
+	struct xfs_dsb	*to,
+	struct xfs_sb	*from)
+{
+	uint16_t	qflags = from->sb_qflags;
+
+	to->sb_uquotino = cpu_to_be64(from->sb_uquotino);
+
+	/*
+	 * The in-memory superblock quota state matches the v5 on-disk format so
+	 * just write them out and return
+	 */
+	if (xfs_sb_is_v5(from)) {
+		to->sb_qflags = cpu_to_be16(from->sb_qflags);
+		to->sb_gquotino = cpu_to_be64(from->sb_gquotino);
+		to->sb_pquotino = cpu_to_be64(from->sb_pquotino);
+		return;
+	}
+
+	/*
+	 * For older superblocks (v4), the in-core version of sb_qflags do not
+	 * have XFS_OQUOTA_* flags, whereas the on-disk version does.  So,
+	 * convert incore XFS_{PG}QUOTA_* flags to on-disk XFS_OQUOTA_* flags.
+	 */
+	qflags &= ~(XFS_PQUOTA_ENFD | XFS_PQUOTA_CHKD |
+			XFS_GQUOTA_ENFD | XFS_GQUOTA_CHKD);
+
+	if (from->sb_qflags &
+			(XFS_PQUOTA_ENFD | XFS_GQUOTA_ENFD))
+		qflags |= XFS_OQUOTA_ENFD;
+	if (from->sb_qflags &
+			(XFS_PQUOTA_CHKD | XFS_GQUOTA_CHKD))
+		qflags |= XFS_OQUOTA_CHKD;
+	to->sb_qflags = cpu_to_be16(qflags);
+
+	/*
+	 * GQUOTINO and PQUOTINO cannot be used together in versions
+	 * of superblock that do not have pquotino. from->sb_flags
+	 * tells us which quota is active and should be copied to
+	 * disk. If neither are active, we should NULL the inode.
+	 *
+	 * In all cases, the separate pquotino must remain 0 because it
+	 * is beyond the "end" of the valid non-pquotino superblock.
+	 */
+	if (from->sb_qflags & XFS_GQUOTA_ACCT)
+		to->sb_gquotino = cpu_to_be64(from->sb_gquotino);
+	else if (from->sb_qflags & XFS_PQUOTA_ACCT)
+		to->sb_gquotino = cpu_to_be64(from->sb_pquotino);
+	else {
+		/*
+		 * We can't rely on just the fields being logged to tell us
+		 * that it is safe to write NULLFSINO - we should only do that
+		 * if quotas are not actually enabled. Hence only write
+		 * NULLFSINO if both in-core quota inodes are NULL.
+		 */
+		if (from->sb_gquotino == NULLFSINO &&
+		    from->sb_pquotino == NULLFSINO)
+			to->sb_gquotino = cpu_to_be64(NULLFSINO);
+	}
+
+	to->sb_pquotino = 0;
+}
+
+void
+xfs_sb_to_disk(
+	struct xfs_dsb	*to,
+	struct xfs_sb	*from)
+{
+	xfs_sb_quota_to_disk(to, from);
+
+	to->sb_magicnum = cpu_to_be32(from->sb_magicnum);
+	to->sb_blocksize = cpu_to_be32(from->sb_blocksize);
+	to->sb_dblocks = cpu_to_be64(from->sb_dblocks);
+	to->sb_rblocks = cpu_to_be64(from->sb_rblocks);
+	to->sb_rextents = cpu_to_be64(from->sb_rextents);
+	memcpy(&to->sb_uuid, &from->sb_uuid, sizeof(to->sb_uuid));
+	to->sb_logstart = cpu_to_be64(from->sb_logstart);
+	to->sb_rootino = cpu_to_be64(from->sb_rootino);
+	to->sb_rbmino = cpu_to_be64(from->sb_rbmino);
+	to->sb_rsumino = cpu_to_be64(from->sb_rsumino);
+	to->sb_rextsize = cpu_to_be32(from->sb_rextsize);
+	to->sb_agblocks = cpu_to_be32(from->sb_agblocks);
+	to->sb_agcount = cpu_to_be32(from->sb_agcount);
+	to->sb_rbmblocks = cpu_to_be32(from->sb_rbmblocks);
+	to->sb_logblocks = cpu_to_be32(from->sb_logblocks);
+	to->sb_versionnum = cpu_to_be16(from->sb_versionnum);
+	to->sb_sectsize = cpu_to_be16(from->sb_sectsize);
+	to->sb_inodesize = cpu_to_be16(from->sb_inodesize);
+	to->sb_inopblock = cpu_to_be16(from->sb_inopblock);
+	memcpy(&to->sb_fname, &from->sb_fname, sizeof(to->sb_fname));
+	to->sb_blocklog = from->sb_blocklog;
+	to->sb_sectlog = from->sb_sectlog;
+	to->sb_inodelog = from->sb_inodelog;
+	to->sb_inopblog = from->sb_inopblog;
+	to->sb_agblklog = from->sb_agblklog;
+	to->sb_rextslog = from->sb_rextslog;
+	to->sb_inprogress = from->sb_inprogress;
+	to->sb_imax_pct = from->sb_imax_pct;
+	to->sb_icount = cpu_to_be64(from->sb_icount);
+	to->sb_ifree = cpu_to_be64(from->sb_ifree);
+	to->sb_fdblocks = cpu_to_be64(from->sb_fdblocks);
+	to->sb_frextents = cpu_to_be64(from->sb_frextents);
+
+	to->sb_flags = from->sb_flags;
+	to->sb_shared_vn = from->sb_shared_vn;
+	to->sb_inoalignmt = cpu_to_be32(from->sb_inoalignmt);
+	to->sb_unit = cpu_to_be32(from->sb_unit);
+	to->sb_width = cpu_to_be32(from->sb_width);
+	to->sb_dirblklog = from->sb_dirblklog;
+	to->sb_logsectlog = from->sb_logsectlog;
+	to->sb_logsectsize = cpu_to_be16(from->sb_logsectsize);
+	to->sb_logsunit = cpu_to_be32(from->sb_logsunit);
+
+	/*
+	 * We need to ensure that bad_features2 always matches features2.
+	 * Hence we enforce that here rather than having to remember to do it
+	 * everywhere else that updates features2.
+	 */
+	from->sb_bad_features2 = from->sb_features2;
+	to->sb_features2 = cpu_to_be32(from->sb_features2);
+	to->sb_bad_features2 = cpu_to_be32(from->sb_bad_features2);
+
+	if (!xfs_sb_is_v5(from))
+		return;
+
+	to->sb_features_compat = cpu_to_be32(from->sb_features_compat);
+	to->sb_features_ro_compat =
+			cpu_to_be32(from->sb_features_ro_compat);
+	to->sb_features_incompat =
+			cpu_to_be32(from->sb_features_incompat);
+	to->sb_features_log_incompat =
+			cpu_to_be32(from->sb_features_log_incompat);
+	to->sb_spino_align = cpu_to_be32(from->sb_spino_align);
+	to->sb_lsn = cpu_to_be64(from->sb_lsn);
+	if (from->sb_features_incompat & XFS_SB_FEAT_INCOMPAT_META_UUID)
+		uuid_copy(&to->sb_meta_uuid, &from->sb_meta_uuid);
+}
+
+/*
+ * If the superblock has the CRC feature bit set or the CRC field is non-null,
+ * check that the CRC is valid.  We check the CRC field is non-null because a
+ * single bit error could clear the feature bit and unused parts of the
+ * superblock are supposed to be zero. Hence a non-null crc field indicates that
+ * we've potentially lost a feature bit and we should check it anyway.
+ *
+ * However, past bugs (i.e. in growfs) left non-zeroed regions beyond the
+ * last field in V4 secondary superblocks.  So for secondary superblocks,
+ * we are more forgiving, and ignore CRC failures if the primary doesn't
+ * indicate that the fs version is V5.
+ */
+static void
+xfs_sb_read_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_sb		sb;
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_dsb		*dsb = bp->b_addr;
+	int			error;
+
+	/*
+	 * open code the version check to avoid needing to convert the entire
+	 * superblock from disk order just to check the version number
+	 */
+	if (dsb->sb_magicnum == cpu_to_be32(XFS_SB_MAGIC) &&
+	    (((be16_to_cpu(dsb->sb_versionnum) & XFS_SB_VERSION_NUMBITS) ==
+						XFS_SB_VERSION_5) ||
+	     dsb->sb_crc != 0)) {
+
+		if (!xfs_buf_verify_cksum(bp, XFS_SB_CRC_OFF)) {
+			/* Only fail bad secondaries on a known V5 filesystem */
+			if (xfs_buf_daddr(bp) == XFS_SB_DADDR ||
+			    xfs_has_crc(mp)) {
+				error = -EFSBADCRC;
+				goto out_error;
+			}
+		}
+	}
+
+	/*
+	 * Check all the superblock fields.  Don't byteswap the xquota flags
+	 * because _verify_common checks the on-disk values.
+	 */
+	__xfs_sb_from_disk(&sb, dsb, false);
+	error = xfs_validate_sb_common(mp, bp, &sb);
+	if (error)
+		goto out_error;
+	error = xfs_validate_sb_read(mp, &sb);
+
+out_error:
+	if (error == -EFSCORRUPTED || error == -EFSBADCRC)
+		xfs_verifier_error(bp, error, __this_address);
+	else if (error)
+		xfs_buf_ioerror(bp, error);
+}
+
+/*
+ * We may be probed for a filesystem match, so we may not want to emit
+ * messages when the superblock buffer is not actually an XFS superblock.
+ * If we find an XFS superblock, then run a normal, noisy mount because we are
+ * really going to mount it and want to know about errors.
+ */
+static void
+xfs_sb_quiet_read_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_dsb	*dsb = bp->b_addr;
+
+	if (dsb->sb_magicnum == cpu_to_be32(XFS_SB_MAGIC)) {
+		/* XFS filesystem, verify noisily! */
+		xfs_sb_read_verify(bp);
+		return;
+	}
+	/* quietly fail */
+	xfs_buf_ioerror(bp, -EWRONGFS);
+}
+
+static void
+xfs_sb_write_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_sb		sb;
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_buf_log_item	*bip = bp->b_log_item;
+	struct xfs_dsb		*dsb = bp->b_addr;
+	int			error;
+
+	/*
+	 * Check all the superblock fields.  Don't byteswap the xquota flags
+	 * because _verify_common checks the on-disk values.
+	 */
+	__xfs_sb_from_disk(&sb, dsb, false);
+	error = xfs_validate_sb_common(mp, bp, &sb);
+	if (error)
+		goto out_error;
+	error = xfs_validate_sb_write(mp, bp, &sb);
+	if (error)
+		goto out_error;
+
+	if (!xfs_sb_is_v5(&sb))
+		return;
+
+	if (bip)
+		dsb->sb_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+	xfs_buf_update_cksum(bp, XFS_SB_CRC_OFF);
+	return;
+
+out_error:
+	xfs_verifier_error(bp, error, __this_address);
+}
+
+const struct xfs_buf_ops xfs_sb_buf_ops = {
+	.name = "xfs_sb",
+	.magic = { cpu_to_be32(XFS_SB_MAGIC), cpu_to_be32(XFS_SB_MAGIC) },
+	.verify_read = xfs_sb_read_verify,
+	.verify_write = xfs_sb_write_verify,
+};
+
+const struct xfs_buf_ops xfs_sb_quiet_buf_ops = {
+	.name = "xfs_sb_quiet",
+	.magic = { cpu_to_be32(XFS_SB_MAGIC), cpu_to_be32(XFS_SB_MAGIC) },
+	.verify_read = xfs_sb_quiet_read_verify,
+	.verify_write = xfs_sb_write_verify,
+};
+
+/*
+ * xfs_mount_common
+ *
+ * Mount initialization code establishing various mount
+ * fields from the superblock associated with the given
+ * mount structure.
+ *
+ * Inode geometry are calculated in xfs_ialloc_setup_geometry.
+ */
+void
+xfs_sb_mount_common(
+	struct xfs_mount	*mp,
+	struct xfs_sb		*sbp)
+{
+	mp->m_agfrotor = mp->m_agirotor = 0;
+	mp->m_maxagi = mp->m_sb.sb_agcount;
+	mp->m_blkbit_log = sbp->sb_blocklog + XFS_NBBYLOG;
+	mp->m_blkbb_log = sbp->sb_blocklog - BBSHIFT;
+	mp->m_sectbb_log = sbp->sb_sectlog - BBSHIFT;
+	mp->m_agno_log = xfs_highbit32(sbp->sb_agcount - 1) + 1;
+	mp->m_blockmask = sbp->sb_blocksize - 1;
+	mp->m_blockwsize = sbp->sb_blocksize >> XFS_WORDLOG;
+	mp->m_blockwmask = mp->m_blockwsize - 1;
+
+	mp->m_alloc_mxr[0] = xfs_allocbt_maxrecs(mp, sbp->sb_blocksize, 1);
+	mp->m_alloc_mxr[1] = xfs_allocbt_maxrecs(mp, sbp->sb_blocksize, 0);
+	mp->m_alloc_mnr[0] = mp->m_alloc_mxr[0] / 2;
+	mp->m_alloc_mnr[1] = mp->m_alloc_mxr[1] / 2;
+
+	mp->m_bmap_dmxr[0] = xfs_bmbt_maxrecs(mp, sbp->sb_blocksize, 1);
+	mp->m_bmap_dmxr[1] = xfs_bmbt_maxrecs(mp, sbp->sb_blocksize, 0);
+	mp->m_bmap_dmnr[0] = mp->m_bmap_dmxr[0] / 2;
+	mp->m_bmap_dmnr[1] = mp->m_bmap_dmxr[1] / 2;
+
+	mp->m_rmap_mxr[0] = xfs_rmapbt_maxrecs(sbp->sb_blocksize, 1);
+	mp->m_rmap_mxr[1] = xfs_rmapbt_maxrecs(sbp->sb_blocksize, 0);
+	mp->m_rmap_mnr[0] = mp->m_rmap_mxr[0] / 2;
+	mp->m_rmap_mnr[1] = mp->m_rmap_mxr[1] / 2;
+
+	mp->m_refc_mxr[0] = xfs_refcountbt_maxrecs(sbp->sb_blocksize, true);
+	mp->m_refc_mxr[1] = xfs_refcountbt_maxrecs(sbp->sb_blocksize, false);
+	mp->m_refc_mnr[0] = mp->m_refc_mxr[0] / 2;
+	mp->m_refc_mnr[1] = mp->m_refc_mxr[1] / 2;
+
+	mp->m_bsize = XFS_FSB_TO_BB(mp, 1);
+	mp->m_alloc_set_aside = xfs_alloc_set_aside(mp);
+	mp->m_ag_max_usable = xfs_alloc_ag_max_usable(mp);
+}
+
+/*
+ * xfs_log_sb() can be used to copy arbitrary changes to the in-core superblock
+ * into the superblock buffer to be logged.  It does not provide the higher
+ * level of locking that is needed to protect the in-core superblock from
+ * concurrent access.
+ */
+void
+xfs_log_sb(
+	struct xfs_trans	*tp)
+{
+	struct xfs_mount	*mp = tp->t_mountp;
+	struct xfs_buf		*bp = xfs_trans_getsb(tp);
+
+	/*
+	 * Lazy sb counters don't update the in-core superblock so do that now.
+	 * If this is at unmount, the counters will be exactly correct, but at
+	 * any other time they will only be ballpark correct because of
+	 * reservations that have been taken out percpu counters. If we have an
+	 * unclean shutdown, this will be corrected by log recovery rebuilding
+	 * the counters from the AGF block counts.
+	 *
+	 * Do not update sb_frextents here because it is not part of the lazy
+	 * sb counters, despite having a percpu counter. It is always kept
+	 * consistent with the ondisk rtbitmap by xfs_trans_apply_sb_deltas()
+	 * and hence we don't need have to update it here.
+	 */
+	if (xfs_has_lazysbcount(mp)) {
+		mp->m_sb.sb_icount = percpu_counter_sum(&mp->m_icount);
+		mp->m_sb.sb_ifree = percpu_counter_sum(&mp->m_ifree);
+		mp->m_sb.sb_fdblocks = percpu_counter_sum(&mp->m_fdblocks);
+	}
+
+	xfs_sb_to_disk(bp->b_addr, &mp->m_sb);
+	xfs_trans_buf_set_type(tp, bp, XFS_BLFT_SB_BUF);
+	xfs_trans_log_buf(tp, bp, 0, sizeof(struct xfs_dsb) - 1);
+}
+
+/*
+ * xfs_sync_sb
+ *
+ * Sync the superblock to disk.
+ *
+ * Note that the caller is responsible for checking the frozen state of the
+ * filesystem. This procedure uses the non-blocking transaction allocator and
+ * thus will allow modifications to a frozen fs. This is required because this
+ * code can be called during the process of freezing where use of the high-level
+ * allocator would deadlock.
+ */
+int
+xfs_sync_sb(
+	struct xfs_mount	*mp,
+	bool			wait)
+{
+	struct xfs_trans	*tp;
+	int			error;
+
+	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_sb, 0, 0,
+			XFS_TRANS_NO_WRITECOUNT, &tp);
+	if (error)
+		return error;
+
+	xfs_log_sb(tp);
+	if (wait)
+		xfs_trans_set_sync(tp);
+	return xfs_trans_commit(tp);
+}
+
+/*
+ * Update all the secondary superblocks to match the new state of the primary.
+ * Because we are completely overwriting all the existing fields in the
+ * secondary superblock buffers, there is no need to read them in from disk.
+ * Just get a new buffer, stamp it and write it.
+ *
+ * The sb buffers need to be cached here so that we serialise against other
+ * operations that access the secondary superblocks, but we don't want to keep
+ * them in memory once it is written so we mark it as a one-shot buffer.
+ */
+int
+xfs_update_secondary_sbs(
+	struct xfs_mount	*mp)
+{
+	struct xfs_perag	*pag;
+	xfs_agnumber_t		agno = 1;
+	int			saved_error = 0;
+	int			error = 0;
+	LIST_HEAD		(buffer_list);
+
+	/* update secondary superblocks. */
+	for_each_perag_from(mp, agno, pag) {
+		struct xfs_buf		*bp;
+
+		error = xfs_buf_get(mp->m_ddev_targp,
+				 XFS_AG_DADDR(mp, pag->pag_agno, XFS_SB_DADDR),
+				 XFS_FSS_TO_BB(mp, 1), &bp);
+		/*
+		 * If we get an error reading or writing alternate superblocks,
+		 * continue.  xfs_repair chooses the "best" superblock based
+		 * on most matches; if we break early, we'll leave more
+		 * superblocks un-updated than updated, and xfs_repair may
+		 * pick them over the properly-updated primary.
+		 */
+		if (error) {
+			xfs_warn(mp,
+		"error allocating secondary superblock for ag %d",
+				pag->pag_agno);
+			if (!saved_error)
+				saved_error = error;
+			continue;
+		}
+
+		bp->b_ops = &xfs_sb_buf_ops;
+		xfs_buf_oneshot(bp);
+		xfs_buf_zero(bp, 0, BBTOB(bp->b_length));
+		xfs_sb_to_disk(bp->b_addr, &mp->m_sb);
+		xfs_buf_delwri_queue(bp, &buffer_list);
+		xfs_buf_relse(bp);
+
+		/* don't hold too many buffers at once */
+		if (agno % 16)
+			continue;
+
+		error = xfs_buf_delwri_submit(&buffer_list);
+		if (error) {
+			xfs_warn(mp,
+		"write error %d updating a secondary superblock near ag %d",
+				error, pag->pag_agno);
+			if (!saved_error)
+				saved_error = error;
+			continue;
+		}
+	}
+	error = xfs_buf_delwri_submit(&buffer_list);
+	if (error) {
+		xfs_warn(mp,
+		"write error %d updating a secondary superblock near ag %d",
+			error, agno);
+	}
+
+	return saved_error ? saved_error : error;
+}
+
+/*
+ * Same behavior as xfs_sync_sb, except that it is always synchronous and it
+ * also writes the superblock buffer to disk sector 0 immediately.
+ */
+int
+xfs_sync_sb_buf(
+	struct xfs_mount	*mp)
+{
+	struct xfs_trans	*tp;
+	struct xfs_buf		*bp;
+	int			error;
+
+	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_sb, 0, 0, 0, &tp);
+	if (error)
+		return error;
+
+	bp = xfs_trans_getsb(tp);
+	xfs_log_sb(tp);
+	xfs_trans_bhold(tp, bp);
+	xfs_trans_set_sync(tp);
+	error = xfs_trans_commit(tp);
+	if (error)
+		goto out;
+	/*
+	 * write out the sb buffer to get the changes to disk
+	 */
+	error = xfs_bwrite(bp);
+out:
+	xfs_buf_relse(bp);
+	return error;
+}
+
+void
+xfs_fs_geometry(
+	struct xfs_mount	*mp,
+	struct xfs_fsop_geom	*geo,
+	int			struct_version)
+{
+	struct xfs_sb		*sbp = &mp->m_sb;
+
+	memset(geo, 0, sizeof(struct xfs_fsop_geom));
+
+	geo->blocksize = sbp->sb_blocksize;
+	geo->rtextsize = sbp->sb_rextsize;
+	geo->agblocks = sbp->sb_agblocks;
+	geo->agcount = sbp->sb_agcount;
+	geo->logblocks = sbp->sb_logblocks;
+	geo->sectsize = sbp->sb_sectsize;
+	geo->inodesize = sbp->sb_inodesize;
+	geo->imaxpct = sbp->sb_imax_pct;
+	geo->datablocks = sbp->sb_dblocks;
+	geo->rtblocks = sbp->sb_rblocks;
+	geo->rtextents = sbp->sb_rextents;
+	geo->logstart = sbp->sb_logstart;
+	BUILD_BUG_ON(sizeof(geo->uuid) != sizeof(sbp->sb_uuid));
+	memcpy(geo->uuid, &sbp->sb_uuid, sizeof(sbp->sb_uuid));
+
+	if (struct_version < 2)
+		return;
+
+	geo->sunit = sbp->sb_unit;
+	geo->swidth = sbp->sb_width;
+
+	if (struct_version < 3)
+		return;
+
+	geo->version = XFS_FSOP_GEOM_VERSION;
+	geo->flags = XFS_FSOP_GEOM_FLAGS_NLINK |
+		     XFS_FSOP_GEOM_FLAGS_DIRV2 |
+		     XFS_FSOP_GEOM_FLAGS_EXTFLG;
+	if (xfs_has_attr(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_ATTR;
+	if (xfs_has_quota(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_QUOTA;
+	if (xfs_has_align(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_IALIGN;
+	if (xfs_has_dalign(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_DALIGN;
+	if (xfs_has_asciici(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_DIRV2CI;
+	if (xfs_has_lazysbcount(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_LAZYSB;
+	if (xfs_has_attr2(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_ATTR2;
+	if (xfs_has_projid32(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_PROJID32;
+	if (xfs_has_crc(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_V5SB;
+	if (xfs_has_ftype(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_FTYPE;
+	if (xfs_has_finobt(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_FINOBT;
+	if (xfs_has_sparseinodes(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_SPINODES;
+	if (xfs_has_rmapbt(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_RMAPBT;
+	if (xfs_has_reflink(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_REFLINK;
+	if (xfs_has_bigtime(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_BIGTIME;
+	if (xfs_has_inobtcounts(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_INOBTCNT;
+	if (xfs_has_sector(mp)) {
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_SECTOR;
+		geo->logsectsize = sbp->sb_logsectsize;
+	} else {
+		geo->logsectsize = BBSIZE;
+	}
+	if (xfs_has_large_extent_counts(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_NREXT64;
+	geo->rtsectsize = sbp->sb_blocksize;
+	geo->dirblocksize = xfs_dir2_dirblock_bytes(sbp);
+
+	if (struct_version < 4)
+		return;
+
+	if (xfs_has_logv2(mp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_LOGV2;
+
+	geo->logsunit = sbp->sb_logsunit;
+
+	if (struct_version < 5)
+		return;
+
+	geo->version = XFS_FSOP_GEOM_VERSION_V5;
+}
+
+/* Read a secondary superblock. */
+int
+xfs_sb_read_secondary(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	xfs_agnumber_t		agno,
+	struct xfs_buf		**bpp)
+{
+	struct xfs_buf		*bp;
+	int			error;
+
+	ASSERT(agno != 0 && agno != NULLAGNUMBER);
+	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
+			XFS_AG_DADDR(mp, agno, XFS_SB_BLOCK(mp)),
+			XFS_FSS_TO_BB(mp, 1), 0, &bp, &xfs_sb_buf_ops);
+	if (error)
+		return error;
+	xfs_buf_set_ref(bp, XFS_SSB_REF);
+	*bpp = bp;
+	return 0;
+}
+
+/* Get an uninitialised secondary superblock buffer. */
+int
+xfs_sb_get_secondary(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	xfs_agnumber_t		agno,
+	struct xfs_buf		**bpp)
+{
+	struct xfs_buf		*bp;
+	int			error;
+
+	ASSERT(agno != 0 && agno != NULLAGNUMBER);
+	error = xfs_trans_get_buf(tp, mp->m_ddev_targp,
+			XFS_AG_DADDR(mp, agno, XFS_SB_BLOCK(mp)),
+			XFS_FSS_TO_BB(mp, 1), 0, &bp);
+	if (error)
+		return error;
+	bp->b_ops = &xfs_sb_buf_ops;
+	xfs_buf_oneshot(bp);
+	*bpp = bp;
+	return 0;
+}
+
+/*
+ * sunit, swidth, sectorsize(optional with 0) should be all in bytes,
+ * so users won't be confused by values in error messages.
+ */
+bool
+xfs_validate_stripe_geometry(
+	struct xfs_mount	*mp,
+	__s64			sunit,
+	__s64			swidth,
+	int			sectorsize,
+	bool			silent)
+{
+	if (swidth > INT_MAX) {
+		if (!silent)
+			xfs_notice(mp,
+"stripe width (%lld) is too large", swidth);
+		return false;
+	}
+
+	if (sunit > swidth) {
+		if (!silent)
+			xfs_notice(mp,
+"stripe unit (%lld) is larger than the stripe width (%lld)", sunit, swidth);
+		return false;
+	}
+
+	if (sectorsize && (int)sunit % sectorsize) {
+		if (!silent)
+			xfs_notice(mp,
+"stripe unit (%lld) must be a multiple of the sector size (%d)",
+				   sunit, sectorsize);
+		return false;
+	}
+
+	if (sunit && !swidth) {
+		if (!silent)
+			xfs_notice(mp,
+"invalid stripe unit (%lld) and stripe width of 0", sunit);
+		return false;
+	}
+
+	if (!sunit && swidth) {
+		if (!silent)
+			xfs_notice(mp,
+"invalid stripe width (%lld) and stripe unit of 0", swidth);
+		return false;
+	}
+
+	if (sunit && (int)swidth % (int)sunit) {
+		if (!silent)
+			xfs_notice(mp,
+"stripe width (%lld) must be a multiple of the stripe unit (%lld)",
+				   swidth, sunit);
+		return false;
+	}
+	return true;
+}
diff --git a/fs/xfs/libxfs/xfs_sb.h b/fs/xfs/libxfs/xfs_sb.h
new file mode 100644
index 000000000..a5e14740e
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_sb.h
@@ -0,0 +1,41 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_SB_H__
+#define	__XFS_SB_H__
+
+struct xfs_mount;
+struct xfs_sb;
+struct xfs_dsb;
+struct xfs_trans;
+struct xfs_fsop_geom;
+struct xfs_perag;
+
+extern void	xfs_log_sb(struct xfs_trans *tp);
+extern int	xfs_sync_sb(struct xfs_mount *mp, bool wait);
+extern int	xfs_sync_sb_buf(struct xfs_mount *mp);
+extern void	xfs_sb_mount_common(struct xfs_mount *mp, struct xfs_sb *sbp);
+extern void	xfs_sb_from_disk(struct xfs_sb *to, struct xfs_dsb *from);
+extern void	xfs_sb_to_disk(struct xfs_dsb *to, struct xfs_sb *from);
+extern void	xfs_sb_quota_from_disk(struct xfs_sb *sbp);
+extern bool	xfs_sb_good_version(struct xfs_sb *sbp);
+extern uint64_t	xfs_sb_version_to_features(struct xfs_sb *sbp);
+
+extern int	xfs_update_secondary_sbs(struct xfs_mount *mp);
+
+#define XFS_FS_GEOM_MAX_STRUCT_VER	(4)
+extern void	xfs_fs_geometry(struct xfs_mount *mp, struct xfs_fsop_geom *geo,
+				int struct_version);
+extern int	xfs_sb_read_secondary(struct xfs_mount *mp,
+				struct xfs_trans *tp, xfs_agnumber_t agno,
+				struct xfs_buf **bpp);
+extern int	xfs_sb_get_secondary(struct xfs_mount *mp,
+				struct xfs_trans *tp, xfs_agnumber_t agno,
+				struct xfs_buf **bpp);
+
+extern bool	xfs_validate_stripe_geometry(struct xfs_mount *mp,
+		__s64 sunit, __s64 swidth, int sectorsize, bool silent);
+
+#endif	/* __XFS_SB_H__ */
diff --git a/fs/xfs/libxfs/xfs_shared.h b/fs/xfs/libxfs/xfs_shared.h
new file mode 100644
index 000000000..c4381388c
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_shared.h
@@ -0,0 +1,193 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_SHARED_H__
+#define __XFS_SHARED_H__
+
+/*
+ * Definitions shared between kernel and userspace that don't fit into any other
+ * header file that is shared with userspace.
+ */
+struct xfs_ifork;
+struct xfs_buf;
+struct xfs_buf_ops;
+struct xfs_mount;
+struct xfs_trans;
+struct xfs_inode;
+
+/*
+ * Buffer verifier operations are widely used, including userspace tools
+ */
+extern const struct xfs_buf_ops xfs_agf_buf_ops;
+extern const struct xfs_buf_ops xfs_agfl_buf_ops;
+extern const struct xfs_buf_ops xfs_agi_buf_ops;
+extern const struct xfs_buf_ops xfs_attr3_leaf_buf_ops;
+extern const struct xfs_buf_ops xfs_attr3_rmt_buf_ops;
+extern const struct xfs_buf_ops xfs_bmbt_buf_ops;
+extern const struct xfs_buf_ops xfs_bnobt_buf_ops;
+extern const struct xfs_buf_ops xfs_cntbt_buf_ops;
+extern const struct xfs_buf_ops xfs_da3_node_buf_ops;
+extern const struct xfs_buf_ops xfs_dquot_buf_ops;
+extern const struct xfs_buf_ops xfs_dquot_buf_ra_ops;
+extern const struct xfs_buf_ops xfs_finobt_buf_ops;
+extern const struct xfs_buf_ops xfs_inobt_buf_ops;
+extern const struct xfs_buf_ops xfs_inode_buf_ops;
+extern const struct xfs_buf_ops xfs_inode_buf_ra_ops;
+extern const struct xfs_buf_ops xfs_refcountbt_buf_ops;
+extern const struct xfs_buf_ops xfs_rmapbt_buf_ops;
+extern const struct xfs_buf_ops xfs_rtbuf_ops;
+extern const struct xfs_buf_ops xfs_sb_buf_ops;
+extern const struct xfs_buf_ops xfs_sb_quiet_buf_ops;
+extern const struct xfs_buf_ops xfs_symlink_buf_ops;
+
+/* log size calculation functions */
+int	xfs_log_calc_unit_res(struct xfs_mount *mp, int unit_bytes);
+int	xfs_log_calc_minimum_size(struct xfs_mount *);
+
+struct xfs_trans_res;
+void	xfs_log_get_max_trans_res(struct xfs_mount *mp,
+				  struct xfs_trans_res *max_resp);
+
+/*
+ * Values for t_flags.
+ */
+/* Transaction needs to be logged */
+#define XFS_TRANS_DIRTY			(1u << 0)
+/* Superblock is dirty and needs to be logged */
+#define XFS_TRANS_SB_DIRTY		(1u << 1)
+/* Transaction took a permanent log reservation */
+#define XFS_TRANS_PERM_LOG_RES		(1u << 2)
+/* Synchronous transaction commit needed */
+#define XFS_TRANS_SYNC			(1u << 3)
+/* Transaction can use reserve block pool */
+#define XFS_TRANS_RESERVE		(1u << 4)
+/* Transaction should avoid VFS level superblock write accounting */
+#define XFS_TRANS_NO_WRITECOUNT		(1u << 5)
+/* Transaction has freed blocks returned to it's reservation */
+#define XFS_TRANS_RES_FDBLKS		(1u << 6)
+/* Transaction contains an intent done log item */
+#define XFS_TRANS_HAS_INTENT_DONE	(1u << 7)
+
+/*
+ * LOWMODE is used by the allocator to activate the lowspace algorithm - when
+ * free space is running low the extent allocator may choose to allocate an
+ * extent from an AG without leaving sufficient space for a btree split when
+ * inserting the new extent. In this case the allocator will enable the
+ * lowspace algorithm which is supposed to allow further allocations (such as
+ * btree splits and newroots) to allocate from sequential AGs. In order to
+ * avoid locking AGs out of order the lowspace algorithm will start searching
+ * for free space from AG 0. If the correct transaction reservations have been
+ * made then this algorithm will eventually find all the space it needs.
+ */
+#define XFS_TRANS_LOWMODE	0x100	/* allocate in low space mode */
+
+/*
+ * Field values for xfs_trans_mod_sb.
+ */
+#define	XFS_TRANS_SB_ICOUNT		0x00000001
+#define	XFS_TRANS_SB_IFREE		0x00000002
+#define	XFS_TRANS_SB_FDBLOCKS		0x00000004
+#define	XFS_TRANS_SB_RES_FDBLOCKS	0x00000008
+#define	XFS_TRANS_SB_FREXTENTS		0x00000010
+#define	XFS_TRANS_SB_RES_FREXTENTS	0x00000020
+#define	XFS_TRANS_SB_DBLOCKS		0x00000040
+#define	XFS_TRANS_SB_AGCOUNT		0x00000080
+#define	XFS_TRANS_SB_IMAXPCT		0x00000100
+#define	XFS_TRANS_SB_REXTSIZE		0x00000200
+#define	XFS_TRANS_SB_RBMBLOCKS		0x00000400
+#define	XFS_TRANS_SB_RBLOCKS		0x00000800
+#define	XFS_TRANS_SB_REXTENTS		0x00001000
+#define	XFS_TRANS_SB_REXTSLOG		0x00002000
+
+/*
+ * Here we centralize the specification of XFS meta-data buffer reference count
+ * values.  This determines how hard the buffer cache tries to hold onto the
+ * buffer.
+ */
+#define	XFS_AGF_REF		4
+#define	XFS_AGI_REF		4
+#define	XFS_AGFL_REF		3
+#define	XFS_INO_BTREE_REF	3
+#define	XFS_ALLOC_BTREE_REF	2
+#define	XFS_BMAP_BTREE_REF	2
+#define	XFS_RMAP_BTREE_REF	2
+#define	XFS_DIR_BTREE_REF	2
+#define	XFS_INO_REF		2
+#define	XFS_ATTR_BTREE_REF	1
+#define	XFS_DQUOT_REF		1
+#define	XFS_REFC_BTREE_REF	1
+#define	XFS_SSB_REF		0
+
+/*
+ * Flags for xfs_trans_ichgtime().
+ */
+#define	XFS_ICHGTIME_MOD	0x1	/* data fork modification timestamp */
+#define	XFS_ICHGTIME_CHG	0x2	/* inode field change timestamp */
+#define	XFS_ICHGTIME_CREATE	0x4	/* inode create timestamp */
+
+
+/*
+ * Symlink decoding/encoding functions
+ */
+int xfs_symlink_blocks(struct xfs_mount *mp, int pathlen);
+int xfs_symlink_hdr_set(struct xfs_mount *mp, xfs_ino_t ino, uint32_t offset,
+			uint32_t size, struct xfs_buf *bp);
+bool xfs_symlink_hdr_ok(xfs_ino_t ino, uint32_t offset,
+			uint32_t size, struct xfs_buf *bp);
+void xfs_symlink_local_to_remote(struct xfs_trans *tp, struct xfs_buf *bp,
+				 struct xfs_inode *ip, struct xfs_ifork *ifp);
+xfs_failaddr_t xfs_symlink_shortform_verify(struct xfs_inode *ip);
+
+/* Computed inode geometry for the filesystem. */
+struct xfs_ino_geometry {
+	/* Maximum inode count in this filesystem. */
+	uint64_t	maxicount;
+
+	/* Actual inode cluster buffer size, in bytes. */
+	unsigned int	inode_cluster_size;
+
+	/*
+	 * Desired inode cluster buffer size, in bytes.  This value is not
+	 * rounded up to at least one filesystem block, which is necessary for
+	 * the sole purpose of validating sb_spino_align.  Runtime code must
+	 * only ever use inode_cluster_size.
+	 */
+	unsigned int	inode_cluster_size_raw;
+
+	/* Inode cluster sizes, adjusted to be at least 1 fsb. */
+	unsigned int	inodes_per_cluster;
+	unsigned int	blocks_per_cluster;
+
+	/* Inode cluster alignment. */
+	unsigned int	cluster_align;
+	unsigned int	cluster_align_inodes;
+	unsigned int	inoalign_mask;	/* mask sb_inoalignmt if used */
+
+	unsigned int	inobt_mxr[2]; /* max inobt btree records */
+	unsigned int	inobt_mnr[2]; /* min inobt btree records */
+	unsigned int	inobt_maxlevels; /* max inobt btree levels. */
+
+	/* Size of inode allocations under normal operation. */
+	unsigned int	ialloc_inos;
+	unsigned int	ialloc_blks;
+
+	/* Minimum inode blocks for a sparse allocation. */
+	unsigned int	ialloc_min_blks;
+
+	/* stripe unit inode alignment */
+	unsigned int	ialloc_align;
+
+	unsigned int	agino_log;	/* #bits for agino in inum */
+
+	/* precomputed default inode attribute fork offset */
+	unsigned int	attr_fork_offset;
+
+	/* precomputed value for di_flags2 */
+	uint64_t	new_diflags2;
+
+};
+
+#endif /* __XFS_SHARED_H__ */
diff --git a/fs/xfs/libxfs/xfs_symlink_remote.c b/fs/xfs/libxfs/xfs_symlink_remote.c
new file mode 100644
index 000000000..bdc777b9e
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_symlink_remote.c
@@ -0,0 +1,233 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * Copyright (c) 2012-2013 Red Hat, Inc.
+ * All rights reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_shared.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_inode.h"
+#include "xfs_error.h"
+#include "xfs_trans.h"
+#include "xfs_buf_item.h"
+#include "xfs_log.h"
+
+
+/*
+ * Each contiguous block has a header, so it is not just a simple pathlen
+ * to FSB conversion.
+ */
+int
+xfs_symlink_blocks(
+	struct xfs_mount *mp,
+	int		pathlen)
+{
+	int buflen = XFS_SYMLINK_BUF_SPACE(mp, mp->m_sb.sb_blocksize);
+
+	return (pathlen + buflen - 1) / buflen;
+}
+
+int
+xfs_symlink_hdr_set(
+	struct xfs_mount	*mp,
+	xfs_ino_t		ino,
+	uint32_t		offset,
+	uint32_t		size,
+	struct xfs_buf		*bp)
+{
+	struct xfs_dsymlink_hdr	*dsl = bp->b_addr;
+
+	if (!xfs_has_crc(mp))
+		return 0;
+
+	memset(dsl, 0, sizeof(struct xfs_dsymlink_hdr));
+	dsl->sl_magic = cpu_to_be32(XFS_SYMLINK_MAGIC);
+	dsl->sl_offset = cpu_to_be32(offset);
+	dsl->sl_bytes = cpu_to_be32(size);
+	uuid_copy(&dsl->sl_uuid, &mp->m_sb.sb_meta_uuid);
+	dsl->sl_owner = cpu_to_be64(ino);
+	dsl->sl_blkno = cpu_to_be64(xfs_buf_daddr(bp));
+	bp->b_ops = &xfs_symlink_buf_ops;
+
+	return sizeof(struct xfs_dsymlink_hdr);
+}
+
+/*
+ * Checking of the symlink header is split into two parts. the verifier does
+ * CRC, location and bounds checking, the unpacking function checks the path
+ * parameters and owner.
+ */
+bool
+xfs_symlink_hdr_ok(
+	xfs_ino_t		ino,
+	uint32_t		offset,
+	uint32_t		size,
+	struct xfs_buf		*bp)
+{
+	struct xfs_dsymlink_hdr *dsl = bp->b_addr;
+
+	if (offset != be32_to_cpu(dsl->sl_offset))
+		return false;
+	if (size != be32_to_cpu(dsl->sl_bytes))
+		return false;
+	if (ino != be64_to_cpu(dsl->sl_owner))
+		return false;
+
+	/* ok */
+	return true;
+}
+
+static xfs_failaddr_t
+xfs_symlink_verify(
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = bp->b_mount;
+	struct xfs_dsymlink_hdr	*dsl = bp->b_addr;
+
+	if (!xfs_has_crc(mp))
+		return __this_address;
+	if (!xfs_verify_magic(bp, dsl->sl_magic))
+		return __this_address;
+	if (!uuid_equal(&dsl->sl_uuid, &mp->m_sb.sb_meta_uuid))
+		return __this_address;
+	if (xfs_buf_daddr(bp) != be64_to_cpu(dsl->sl_blkno))
+		return __this_address;
+	if (be32_to_cpu(dsl->sl_offset) +
+				be32_to_cpu(dsl->sl_bytes) >= XFS_SYMLINK_MAXLEN)
+		return __this_address;
+	if (dsl->sl_owner == 0)
+		return __this_address;
+	if (!xfs_log_check_lsn(mp, be64_to_cpu(dsl->sl_lsn)))
+		return __this_address;
+
+	return NULL;
+}
+
+static void
+xfs_symlink_read_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount *mp = bp->b_mount;
+	xfs_failaddr_t	fa;
+
+	/* no verification of non-crc buffers */
+	if (!xfs_has_crc(mp))
+		return;
+
+	if (!xfs_buf_verify_cksum(bp, XFS_SYMLINK_CRC_OFF))
+		xfs_verifier_error(bp, -EFSBADCRC, __this_address);
+	else {
+		fa = xfs_symlink_verify(bp);
+		if (fa)
+			xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+	}
+}
+
+static void
+xfs_symlink_write_verify(
+	struct xfs_buf	*bp)
+{
+	struct xfs_mount *mp = bp->b_mount;
+	struct xfs_buf_log_item	*bip = bp->b_log_item;
+	xfs_failaddr_t		fa;
+
+	/* no verification of non-crc buffers */
+	if (!xfs_has_crc(mp))
+		return;
+
+	fa = xfs_symlink_verify(bp);
+	if (fa) {
+		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
+		return;
+	}
+
+	if (bip) {
+		struct xfs_dsymlink_hdr *dsl = bp->b_addr;
+		dsl->sl_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+	}
+	xfs_buf_update_cksum(bp, XFS_SYMLINK_CRC_OFF);
+}
+
+const struct xfs_buf_ops xfs_symlink_buf_ops = {
+	.name = "xfs_symlink",
+	.magic = { 0, cpu_to_be32(XFS_SYMLINK_MAGIC) },
+	.verify_read = xfs_symlink_read_verify,
+	.verify_write = xfs_symlink_write_verify,
+	.verify_struct = xfs_symlink_verify,
+};
+
+void
+xfs_symlink_local_to_remote(
+	struct xfs_trans	*tp,
+	struct xfs_buf		*bp,
+	struct xfs_inode	*ip,
+	struct xfs_ifork	*ifp)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	char			*buf;
+
+	xfs_trans_buf_set_type(tp, bp, XFS_BLFT_SYMLINK_BUF);
+
+	if (!xfs_has_crc(mp)) {
+		bp->b_ops = NULL;
+		memcpy(bp->b_addr, ifp->if_u1.if_data, ifp->if_bytes);
+		xfs_trans_log_buf(tp, bp, 0, ifp->if_bytes - 1);
+		return;
+	}
+
+	/*
+	 * As this symlink fits in an inode literal area, it must also fit in
+	 * the smallest buffer the filesystem supports.
+	 */
+	ASSERT(BBTOB(bp->b_length) >=
+			ifp->if_bytes + sizeof(struct xfs_dsymlink_hdr));
+
+	bp->b_ops = &xfs_symlink_buf_ops;
+
+	buf = bp->b_addr;
+	buf += xfs_symlink_hdr_set(mp, ip->i_ino, 0, ifp->if_bytes, bp);
+	memcpy(buf, ifp->if_u1.if_data, ifp->if_bytes);
+	xfs_trans_log_buf(tp, bp, 0, sizeof(struct xfs_dsymlink_hdr) +
+					ifp->if_bytes - 1);
+}
+
+/*
+ * Verify the in-memory consistency of an inline symlink data fork. This
+ * does not do on-disk format checks.
+ */
+xfs_failaddr_t
+xfs_symlink_shortform_verify(
+	struct xfs_inode	*ip)
+{
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, XFS_DATA_FORK);
+	char			*sfp = (char *)ifp->if_u1.if_data;
+	int			size = ifp->if_bytes;
+	char			*endp = sfp + size;
+
+	ASSERT(ifp->if_format == XFS_DINODE_FMT_LOCAL);
+
+	/*
+	 * Zero length symlinks should never occur in memory as they are
+	 * never allowed to exist on disk.
+	 */
+	if (!size)
+		return __this_address;
+
+	/* No negative sizes or overly long symlink targets. */
+	if (size < 0 || size > XFS_SYMLINK_MAXLEN)
+		return __this_address;
+
+	/* No NULLs in the target either. */
+	if (memchr(sfp, 0, size - 1))
+		return __this_address;
+
+	/* We /did/ null-terminate the buffer, right? */
+	if (*endp != 0)
+		return __this_address;
+	return NULL;
+}
diff --git a/fs/xfs/libxfs/xfs_trans_inode.c b/fs/xfs/libxfs/xfs_trans_inode.c
new file mode 100644
index 000000000..8b5547073
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_trans_inode.c
@@ -0,0 +1,225 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_inode.h"
+#include "xfs_trans.h"
+#include "xfs_trans_priv.h"
+#include "xfs_inode_item.h"
+
+#include <linux/iversion.h>
+
+/*
+ * Add a locked inode to the transaction.
+ *
+ * The inode must be locked, and it cannot be associated with any transaction.
+ * If lock_flags is non-zero the inode will be unlocked on transaction commit.
+ */
+void
+xfs_trans_ijoin(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	uint			lock_flags)
+{
+	struct xfs_inode_log_item *iip;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+	if (ip->i_itemp == NULL)
+		xfs_inode_item_init(ip, ip->i_mount);
+	iip = ip->i_itemp;
+
+	ASSERT(iip->ili_lock_flags == 0);
+	iip->ili_lock_flags = lock_flags;
+	ASSERT(!xfs_iflags_test(ip, XFS_ISTALE));
+
+	/*
+	 * Get a log_item_desc to point at the new item.
+	 */
+	xfs_trans_add_item(tp, &iip->ili_item);
+}
+
+/*
+ * Transactional inode timestamp update. Requires the inode to be locked and
+ * joined to the transaction supplied. Relies on the transaction subsystem to
+ * track dirty state and update/writeback the inode accordingly.
+ */
+void
+xfs_trans_ichgtime(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	int			flags)
+{
+	struct inode		*inode = VFS_I(ip);
+	struct timespec64	tv;
+
+	ASSERT(tp);
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+
+	tv = current_time(inode);
+
+	if (flags & XFS_ICHGTIME_MOD)
+		inode->i_mtime = tv;
+	if (flags & XFS_ICHGTIME_CHG)
+		inode->i_ctime = tv;
+	if (flags & XFS_ICHGTIME_CREATE)
+		ip->i_crtime = tv;
+}
+
+/*
+ * This is called to mark the fields indicated in fieldmask as needing to be
+ * logged when the transaction is committed.  The inode must already be
+ * associated with the given transaction.
+ *
+ * The values for fieldmask are defined in xfs_inode_item.h.  We always log all
+ * of the core inode if any of it has changed, and we always log all of the
+ * inline data/extents/b-tree root if any of them has changed.
+ *
+ * Grab and pin the cluster buffer associated with this inode to avoid RMW
+ * cycles at inode writeback time. Avoid the need to add error handling to every
+ * xfs_trans_log_inode() call by shutting down on read error.  This will cause
+ * transactions to fail and everything to error out, just like if we return a
+ * read error in a dirty transaction and cancel it.
+ */
+void
+xfs_trans_log_inode(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	uint			flags)
+{
+	struct xfs_inode_log_item *iip = ip->i_itemp;
+	struct inode		*inode = VFS_I(ip);
+	uint			iversion_flags = 0;
+
+	ASSERT(iip);
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+	ASSERT(!xfs_iflags_test(ip, XFS_ISTALE));
+
+	tp->t_flags |= XFS_TRANS_DIRTY;
+
+	/*
+	 * Don't bother with i_lock for the I_DIRTY_TIME check here, as races
+	 * don't matter - we either will need an extra transaction in 24 hours
+	 * to log the timestamps, or will clear already cleared fields in the
+	 * worst case.
+	 */
+	if (inode->i_state & I_DIRTY_TIME) {
+		spin_lock(&inode->i_lock);
+		inode->i_state &= ~I_DIRTY_TIME;
+		spin_unlock(&inode->i_lock);
+	}
+
+	/*
+	 * First time we log the inode in a transaction, bump the inode change
+	 * counter if it is configured for this to occur. While we have the
+	 * inode locked exclusively for metadata modification, we can usually
+	 * avoid setting XFS_ILOG_CORE if no one has queried the value since
+	 * the last time it was incremented. If we have XFS_ILOG_CORE already
+	 * set however, then go ahead and bump the i_version counter
+	 * unconditionally.
+	 */
+	if (!test_and_set_bit(XFS_LI_DIRTY, &iip->ili_item.li_flags)) {
+		if (IS_I_VERSION(inode) &&
+		    inode_maybe_inc_iversion(inode, flags & XFS_ILOG_CORE))
+			iversion_flags = XFS_ILOG_CORE;
+	}
+
+	/*
+	 * If we're updating the inode core or the timestamps and it's possible
+	 * to upgrade this inode to bigtime format, do so now.
+	 */
+	if ((flags & (XFS_ILOG_CORE | XFS_ILOG_TIMESTAMP)) &&
+	    xfs_has_bigtime(ip->i_mount) &&
+	    !xfs_inode_has_bigtime(ip)) {
+		ip->i_diflags2 |= XFS_DIFLAG2_BIGTIME;
+		flags |= XFS_ILOG_CORE;
+	}
+
+	/*
+	 * Inode verifiers do not check that the extent size hint is an integer
+	 * multiple of the rt extent size on a directory with both rtinherit
+	 * and extszinherit flags set.  If we're logging a directory that is
+	 * misconfigured in this way, clear the hint.
+	 */
+	if ((ip->i_diflags & XFS_DIFLAG_RTINHERIT) &&
+	    (ip->i_diflags & XFS_DIFLAG_EXTSZINHERIT) &&
+	    (ip->i_extsize % ip->i_mount->m_sb.sb_rextsize) > 0) {
+		ip->i_diflags &= ~(XFS_DIFLAG_EXTSIZE |
+				   XFS_DIFLAG_EXTSZINHERIT);
+		ip->i_extsize = 0;
+		flags |= XFS_ILOG_CORE;
+	}
+
+	/*
+	 * Record the specific change for fdatasync optimisation. This allows
+	 * fdatasync to skip log forces for inodes that are only timestamp
+	 * dirty.
+	 */
+	spin_lock(&iip->ili_lock);
+	iip->ili_fsync_fields |= flags;
+
+	if (!iip->ili_item.li_buf) {
+		struct xfs_buf	*bp;
+		int		error;
+
+		/*
+		 * We hold the ILOCK here, so this inode is not going to be
+		 * flushed while we are here. Further, because there is no
+		 * buffer attached to the item, we know that there is no IO in
+		 * progress, so nothing will clear the ili_fields while we read
+		 * in the buffer. Hence we can safely drop the spin lock and
+		 * read the buffer knowing that the state will not change from
+		 * here.
+		 */
+		spin_unlock(&iip->ili_lock);
+		error = xfs_imap_to_bp(ip->i_mount, tp, &ip->i_imap, &bp);
+		if (error) {
+			xfs_force_shutdown(ip->i_mount, SHUTDOWN_META_IO_ERROR);
+			return;
+		}
+
+		/*
+		 * We need an explicit buffer reference for the log item but
+		 * don't want the buffer to remain attached to the transaction.
+		 * Hold the buffer but release the transaction reference once
+		 * we've attached the inode log item to the buffer log item
+		 * list.
+		 */
+		xfs_buf_hold(bp);
+		spin_lock(&iip->ili_lock);
+		iip->ili_item.li_buf = bp;
+		bp->b_flags |= _XBF_INODES;
+		list_add_tail(&iip->ili_item.li_bio_list, &bp->b_li_list);
+		xfs_trans_brelse(tp, bp);
+	}
+
+	/*
+	 * Always OR in the bits from the ili_last_fields field.  This is to
+	 * coordinate with the xfs_iflush() and xfs_buf_inode_iodone() routines
+	 * in the eventual clearing of the ili_fields bits.  See the big comment
+	 * in xfs_iflush() for an explanation of this coordination mechanism.
+	 */
+	iip->ili_fields |= (flags | iip->ili_last_fields | iversion_flags);
+	spin_unlock(&iip->ili_lock);
+}
+
+int
+xfs_trans_roll_inode(
+	struct xfs_trans	**tpp,
+	struct xfs_inode	*ip)
+{
+	int			error;
+
+	xfs_trans_log_inode(*tpp, ip, XFS_ILOG_CORE);
+	error = xfs_trans_roll(tpp);
+	if (!error)
+		xfs_trans_ijoin(*tpp, ip, 0);
+	return error;
+}
diff --git a/fs/xfs/libxfs/xfs_trans_resv.c b/fs/xfs/libxfs/xfs_trans_resv.c
new file mode 100644
index 000000000..5b2f27cbd
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_trans_resv.c
@@ -0,0 +1,1028 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * Copyright (C) 2010 Red Hat, Inc.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_da_format.h"
+#include "xfs_da_btree.h"
+#include "xfs_inode.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_quota.h"
+#include "xfs_trans.h"
+#include "xfs_qm.h"
+#include "xfs_trans_space.h"
+
+#define _ALLOC	true
+#define _FREE	false
+
+/*
+ * A buffer has a format structure overhead in the log in addition
+ * to the data, so we need to take this into account when reserving
+ * space in a transaction for a buffer.  Round the space required up
+ * to a multiple of 128 bytes so that we don't change the historical
+ * reservation that has been used for this overhead.
+ */
+STATIC uint
+xfs_buf_log_overhead(void)
+{
+	return round_up(sizeof(struct xlog_op_header) +
+			sizeof(struct xfs_buf_log_format), 128);
+}
+
+/*
+ * Calculate out transaction log reservation per item in bytes.
+ *
+ * The nbufs argument is used to indicate the number of items that
+ * will be changed in a transaction.  size is used to tell how many
+ * bytes should be reserved per item.
+ */
+STATIC uint
+xfs_calc_buf_res(
+	uint		nbufs,
+	uint		size)
+{
+	return nbufs * (size + xfs_buf_log_overhead());
+}
+
+/*
+ * Per-extent log reservation for the btree changes involved in freeing or
+ * allocating an extent.  In classic XFS there were two trees that will be
+ * modified (bnobt + cntbt).  With rmap enabled, there are three trees
+ * (rmapbt).  The number of blocks reserved is based on the formula:
+ *
+ * num trees * ((2 blocks/level * max depth) - 1)
+ *
+ * Keep in mind that max depth is calculated separately for each type of tree.
+ */
+uint
+xfs_allocfree_block_count(
+	struct xfs_mount *mp,
+	uint		num_ops)
+{
+	uint		blocks;
+
+	blocks = num_ops * 2 * (2 * mp->m_alloc_maxlevels - 1);
+	if (xfs_has_rmapbt(mp))
+		blocks += num_ops * (2 * mp->m_rmap_maxlevels - 1);
+
+	return blocks;
+}
+
+/*
+ * Per-extent log reservation for refcount btree changes.  These are never done
+ * in the same transaction as an allocation or a free, so we compute them
+ * separately.
+ */
+static unsigned int
+xfs_refcountbt_block_count(
+	struct xfs_mount	*mp,
+	unsigned int		num_ops)
+{
+	return num_ops * (2 * mp->m_refc_maxlevels - 1);
+}
+
+/*
+ * Logging inodes is really tricksy. They are logged in memory format,
+ * which means that what we write into the log doesn't directly translate into
+ * the amount of space they use on disk.
+ *
+ * Case in point - btree format forks in memory format use more space than the
+ * on-disk format. In memory, the buffer contains a normal btree block header so
+ * the btree code can treat it as though it is just another generic buffer.
+ * However, when we write it to the inode fork, we don't write all of this
+ * header as it isn't needed. e.g. the root is only ever in the inode, so
+ * there's no need for sibling pointers which would waste 16 bytes of space.
+ *
+ * Hence when we have an inode with a maximally sized btree format fork, then
+ * amount of information we actually log is greater than the size of the inode
+ * on disk. Hence we need an inode reservation function that calculates all this
+ * correctly. So, we log:
+ *
+ * - 4 log op headers for object
+ *	- for the ilf, the inode core and 2 forks
+ * - inode log format object
+ * - the inode core
+ * - two inode forks containing bmap btree root blocks.
+ *	- the btree data contained by both forks will fit into the inode size,
+ *	  hence when combined with the inode core above, we have a total of the
+ *	  actual inode size.
+ *	- the BMBT headers need to be accounted separately, as they are
+ *	  additional to the records and pointers that fit inside the inode
+ *	  forks.
+ */
+STATIC uint
+xfs_calc_inode_res(
+	struct xfs_mount	*mp,
+	uint			ninodes)
+{
+	return ninodes *
+		(4 * sizeof(struct xlog_op_header) +
+		 sizeof(struct xfs_inode_log_format) +
+		 mp->m_sb.sb_inodesize +
+		 2 * XFS_BMBT_BLOCK_LEN(mp));
+}
+
+/*
+ * Inode btree record insertion/removal modifies the inode btree and free space
+ * btrees (since the inobt does not use the agfl). This requires the following
+ * reservation:
+ *
+ * the inode btree: max depth * blocksize
+ * the allocation btrees: 2 trees * (max depth - 1) * block size
+ *
+ * The caller must account for SB and AG header modifications, etc.
+ */
+STATIC uint
+xfs_calc_inobt_res(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_buf_res(M_IGEO(mp)->inobt_maxlevels,
+			XFS_FSB_TO_B(mp, 1)) +
+				xfs_calc_buf_res(xfs_allocfree_block_count(mp, 1),
+			XFS_FSB_TO_B(mp, 1));
+}
+
+/*
+ * The free inode btree is a conditional feature. The behavior differs slightly
+ * from that of the traditional inode btree in that the finobt tracks records
+ * for inode chunks with at least one free inode. A record can be removed from
+ * the tree during individual inode allocation. Therefore the finobt
+ * reservation is unconditional for both the inode chunk allocation and
+ * individual inode allocation (modify) cases.
+ *
+ * Behavior aside, the reservation for finobt modification is equivalent to the
+ * traditional inobt: cover a full finobt shape change plus block allocation.
+ */
+STATIC uint
+xfs_calc_finobt_res(
+	struct xfs_mount	*mp)
+{
+	if (!xfs_has_finobt(mp))
+		return 0;
+
+	return xfs_calc_inobt_res(mp);
+}
+
+/*
+ * Calculate the reservation required to allocate or free an inode chunk. This
+ * includes:
+ *
+ * the allocation btrees: 2 trees * (max depth - 1) * block size
+ * the inode chunk: m_ino_geo.ialloc_blks * N
+ *
+ * The size N of the inode chunk reservation depends on whether it is for
+ * allocation or free and which type of create transaction is in use. An inode
+ * chunk free always invalidates the buffers and only requires reservation for
+ * headers (N == 0). An inode chunk allocation requires a chunk sized
+ * reservation on v4 and older superblocks to initialize the chunk. No chunk
+ * reservation is required for allocation on v5 supers, which use ordered
+ * buffers to initialize.
+ */
+STATIC uint
+xfs_calc_inode_chunk_res(
+	struct xfs_mount	*mp,
+	bool			alloc)
+{
+	uint			res, size = 0;
+
+	res = xfs_calc_buf_res(xfs_allocfree_block_count(mp, 1),
+			       XFS_FSB_TO_B(mp, 1));
+	if (alloc) {
+		/* icreate tx uses ordered buffers */
+		if (xfs_has_v3inodes(mp))
+			return res;
+		size = XFS_FSB_TO_B(mp, 1);
+	}
+
+	res += xfs_calc_buf_res(M_IGEO(mp)->ialloc_blks, size);
+	return res;
+}
+
+/*
+ * Per-extent log reservation for the btree changes involved in freeing or
+ * allocating a realtime extent.  We have to be able to log as many rtbitmap
+ * blocks as needed to mark inuse XFS_BMBT_MAX_EXTLEN blocks' worth of realtime
+ * extents, as well as the realtime summary block.
+ */
+static unsigned int
+xfs_rtalloc_block_count(
+	struct xfs_mount	*mp,
+	unsigned int		num_ops)
+{
+	unsigned int		blksz = XFS_FSB_TO_B(mp, 1);
+	unsigned int		rtbmp_bytes;
+
+	rtbmp_bytes = (XFS_MAX_BMBT_EXTLEN / mp->m_sb.sb_rextsize) / NBBY;
+	return (howmany(rtbmp_bytes, blksz) + 1) * num_ops;
+}
+
+/*
+ * Various log reservation values.
+ *
+ * These are based on the size of the file system block because that is what
+ * most transactions manipulate.  Each adds in an additional 128 bytes per
+ * item logged to try to account for the overhead of the transaction mechanism.
+ *
+ * Note:  Most of the reservations underestimate the number of allocation
+ * groups into which they could free extents in the xfs_defer_finish() call.
+ * This is because the number in the worst case is quite high and quite
+ * unusual.  In order to fix this we need to change xfs_defer_finish() to free
+ * extents in only a single AG at a time.  This will require changes to the
+ * EFI code as well, however, so that the EFI for the extents not freed is
+ * logged again in each transaction.  See SGI PV #261917.
+ *
+ * Reservation functions here avoid a huge stack in xfs_trans_init due to
+ * register overflow from temporaries in the calculations.
+ */
+
+/*
+ * Compute the log reservation required to handle the refcount update
+ * transaction.  Refcount updates are always done via deferred log items.
+ *
+ * This is calculated as:
+ * Data device refcount updates (t1):
+ *    the agfs of the ags containing the blocks: nr_ops * sector size
+ *    the refcount btrees: nr_ops * 1 trees * (2 * max depth - 1) * block size
+ */
+static unsigned int
+xfs_calc_refcountbt_reservation(
+	struct xfs_mount	*mp,
+	unsigned int		nr_ops)
+{
+	unsigned int		blksz = XFS_FSB_TO_B(mp, 1);
+
+	if (!xfs_has_reflink(mp))
+		return 0;
+
+	return xfs_calc_buf_res(nr_ops, mp->m_sb.sb_sectsize) +
+	       xfs_calc_buf_res(xfs_refcountbt_block_count(mp, nr_ops), blksz);
+}
+
+/*
+ * In a write transaction we can allocate a maximum of 2
+ * extents.  This gives (t1):
+ *    the inode getting the new extents: inode size
+ *    the inode's bmap btree: max depth * block size
+ *    the agfs of the ags from which the extents are allocated: 2 * sector
+ *    the superblock free block counter: sector size
+ *    the allocation btrees: 2 exts * 2 trees * (2 * max depth - 1) * block size
+ * Or, if we're writing to a realtime file (t2):
+ *    the inode getting the new extents: inode size
+ *    the inode's bmap btree: max depth * block size
+ *    the agfs of the ags from which the extents are allocated: 2 * sector
+ *    the superblock free block counter: sector size
+ *    the realtime bitmap: ((XFS_BMBT_MAX_EXTLEN / rtextsize) / NBBY) bytes
+ *    the realtime summary: 1 block
+ *    the allocation btrees: 2 trees * (2 * max depth - 1) * block size
+ * And the bmap_finish transaction can free bmap blocks in a join (t3):
+ *    the agfs of the ags containing the blocks: 2 * sector size
+ *    the agfls of the ags containing the blocks: 2 * sector size
+ *    the super block free block counter: sector size
+ *    the allocation btrees: 2 exts * 2 trees * (2 * max depth - 1) * block size
+ * And any refcount updates that happen in a separate transaction (t4).
+ */
+STATIC uint
+xfs_calc_write_reservation(
+	struct xfs_mount	*mp,
+	bool			for_minlogsize)
+{
+	unsigned int		t1, t2, t3, t4;
+	unsigned int		blksz = XFS_FSB_TO_B(mp, 1);
+
+	t1 = xfs_calc_inode_res(mp, 1) +
+	     xfs_calc_buf_res(XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK), blksz) +
+	     xfs_calc_buf_res(3, mp->m_sb.sb_sectsize) +
+	     xfs_calc_buf_res(xfs_allocfree_block_count(mp, 2), blksz);
+
+	if (xfs_has_realtime(mp)) {
+		t2 = xfs_calc_inode_res(mp, 1) +
+		     xfs_calc_buf_res(XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK),
+				     blksz) +
+		     xfs_calc_buf_res(3, mp->m_sb.sb_sectsize) +
+		     xfs_calc_buf_res(xfs_rtalloc_block_count(mp, 1), blksz) +
+		     xfs_calc_buf_res(xfs_allocfree_block_count(mp, 1), blksz);
+	} else {
+		t2 = 0;
+	}
+
+	t3 = xfs_calc_buf_res(5, mp->m_sb.sb_sectsize) +
+	     xfs_calc_buf_res(xfs_allocfree_block_count(mp, 2), blksz);
+
+	/*
+	 * In the early days of reflink, we included enough reservation to log
+	 * two refcountbt splits for each transaction.  The codebase runs
+	 * refcountbt updates in separate transactions now, so to compute the
+	 * minimum log size, add the refcountbtree splits back to t1 and t3 and
+	 * do not account them separately as t4.  Reflink did not support
+	 * realtime when the reservations were established, so no adjustment to
+	 * t2 is needed.
+	 */
+	if (for_minlogsize) {
+		unsigned int	adj = 0;
+
+		if (xfs_has_reflink(mp))
+			adj = xfs_calc_buf_res(
+					xfs_refcountbt_block_count(mp, 2),
+					blksz);
+		t1 += adj;
+		t3 += adj;
+		return XFS_DQUOT_LOGRES(mp) + max3(t1, t2, t3);
+	}
+
+	t4 = xfs_calc_refcountbt_reservation(mp, 1);
+	return XFS_DQUOT_LOGRES(mp) + max(t4, max3(t1, t2, t3));
+}
+
+unsigned int
+xfs_calc_write_reservation_minlogsize(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_write_reservation(mp, true);
+}
+
+/*
+ * In truncating a file we free up to two extents at once.  We can modify (t1):
+ *    the inode being truncated: inode size
+ *    the inode's bmap btree: (max depth + 1) * block size
+ * And the bmap_finish transaction can free the blocks and bmap blocks (t2):
+ *    the agf for each of the ags: 4 * sector size
+ *    the agfl for each of the ags: 4 * sector size
+ *    the super block to reflect the freed blocks: sector size
+ *    worst case split in allocation btrees per extent assuming 4 extents:
+ *		4 exts * 2 trees * (2 * max depth - 1) * block size
+ * Or, if it's a realtime file (t3):
+ *    the agf for each of the ags: 2 * sector size
+ *    the agfl for each of the ags: 2 * sector size
+ *    the super block to reflect the freed blocks: sector size
+ *    the realtime bitmap:
+ *		2 exts * ((XFS_BMBT_MAX_EXTLEN / rtextsize) / NBBY) bytes
+ *    the realtime summary: 2 exts * 1 block
+ *    worst case split in allocation btrees per extent assuming 2 extents:
+ *		2 exts * 2 trees * (2 * max depth - 1) * block size
+ * And any refcount updates that happen in a separate transaction (t4).
+ */
+STATIC uint
+xfs_calc_itruncate_reservation(
+	struct xfs_mount	*mp,
+	bool			for_minlogsize)
+{
+	unsigned int		t1, t2, t3, t4;
+	unsigned int		blksz = XFS_FSB_TO_B(mp, 1);
+
+	t1 = xfs_calc_inode_res(mp, 1) +
+	     xfs_calc_buf_res(XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK) + 1, blksz);
+
+	t2 = xfs_calc_buf_res(9, mp->m_sb.sb_sectsize) +
+	     xfs_calc_buf_res(xfs_allocfree_block_count(mp, 4), blksz);
+
+	if (xfs_has_realtime(mp)) {
+		t3 = xfs_calc_buf_res(5, mp->m_sb.sb_sectsize) +
+		     xfs_calc_buf_res(xfs_rtalloc_block_count(mp, 2), blksz) +
+		     xfs_calc_buf_res(xfs_allocfree_block_count(mp, 2), blksz);
+	} else {
+		t3 = 0;
+	}
+
+	/*
+	 * In the early days of reflink, we included enough reservation to log
+	 * four refcountbt splits in the same transaction as bnobt/cntbt
+	 * updates.  The codebase runs refcountbt updates in separate
+	 * transactions now, so to compute the minimum log size, add the
+	 * refcount btree splits back here and do not compute them separately
+	 * as t4.  Reflink did not support realtime when the reservations were
+	 * established, so do not adjust t3.
+	 */
+	if (for_minlogsize) {
+		if (xfs_has_reflink(mp))
+			t2 += xfs_calc_buf_res(
+					xfs_refcountbt_block_count(mp, 4),
+					blksz);
+
+		return XFS_DQUOT_LOGRES(mp) + max3(t1, t2, t3);
+	}
+
+	t4 = xfs_calc_refcountbt_reservation(mp, 2);
+	return XFS_DQUOT_LOGRES(mp) + max(t4, max3(t1, t2, t3));
+}
+
+unsigned int
+xfs_calc_itruncate_reservation_minlogsize(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_itruncate_reservation(mp, true);
+}
+
+/*
+ * In renaming a files we can modify:
+ *    the five inodes involved: 5 * inode size
+ *    the two directory btrees: 2 * (max depth + v2) * dir block size
+ *    the two directory bmap btrees: 2 * max depth * block size
+ * And the bmap_finish transaction can free dir and bmap blocks (two sets
+ *	of bmap blocks) giving:
+ *    the agf for the ags in which the blocks live: 3 * sector size
+ *    the agfl for the ags in which the blocks live: 3 * sector size
+ *    the superblock for the free block count: sector size
+ *    the allocation btrees: 3 exts * 2 trees * (2 * max depth - 1) * block size
+ */
+STATIC uint
+xfs_calc_rename_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		max((xfs_calc_inode_res(mp, 5) +
+		     xfs_calc_buf_res(2 * XFS_DIROP_LOG_COUNT(mp),
+				      XFS_FSB_TO_B(mp, 1))),
+		    (xfs_calc_buf_res(7, mp->m_sb.sb_sectsize) +
+		     xfs_calc_buf_res(xfs_allocfree_block_count(mp, 3),
+				      XFS_FSB_TO_B(mp, 1))));
+}
+
+/*
+ * For removing an inode from unlinked list at first, we can modify:
+ *    the agi hash list and counters: sector size
+ *    the on disk inode before ours in the agi hash list: inode cluster size
+ *    the on disk inode in the agi hash list: inode cluster size
+ */
+STATIC uint
+xfs_calc_iunlink_remove_reservation(
+	struct xfs_mount        *mp)
+{
+	return xfs_calc_buf_res(1, mp->m_sb.sb_sectsize) +
+	       2 * M_IGEO(mp)->inode_cluster_size;
+}
+
+/*
+ * For creating a link to an inode:
+ *    the parent directory inode: inode size
+ *    the linked inode: inode size
+ *    the directory btree could split: (max depth + v2) * dir block size
+ *    the directory bmap btree could join or split: (max depth + v2) * blocksize
+ * And the bmap_finish transaction can free some bmap blocks giving:
+ *    the agf for the ag in which the blocks live: sector size
+ *    the agfl for the ag in which the blocks live: sector size
+ *    the superblock for the free block count: sector size
+ *    the allocation btrees: 2 trees * (2 * max depth - 1) * block size
+ */
+STATIC uint
+xfs_calc_link_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		xfs_calc_iunlink_remove_reservation(mp) +
+		max((xfs_calc_inode_res(mp, 2) +
+		     xfs_calc_buf_res(XFS_DIROP_LOG_COUNT(mp),
+				      XFS_FSB_TO_B(mp, 1))),
+		    (xfs_calc_buf_res(3, mp->m_sb.sb_sectsize) +
+		     xfs_calc_buf_res(xfs_allocfree_block_count(mp, 1),
+				      XFS_FSB_TO_B(mp, 1))));
+}
+
+/*
+ * For adding an inode to unlinked list we can modify:
+ *    the agi hash list: sector size
+ *    the on disk inode: inode cluster size
+ */
+STATIC uint
+xfs_calc_iunlink_add_reservation(xfs_mount_t *mp)
+{
+	return xfs_calc_buf_res(1, mp->m_sb.sb_sectsize) +
+			M_IGEO(mp)->inode_cluster_size;
+}
+
+/*
+ * For removing a directory entry we can modify:
+ *    the parent directory inode: inode size
+ *    the removed inode: inode size
+ *    the directory btree could join: (max depth + v2) * dir block size
+ *    the directory bmap btree could join or split: (max depth + v2) * blocksize
+ * And the bmap_finish transaction can free the dir and bmap blocks giving:
+ *    the agf for the ag in which the blocks live: 2 * sector size
+ *    the agfl for the ag in which the blocks live: 2 * sector size
+ *    the superblock for the free block count: sector size
+ *    the allocation btrees: 2 exts * 2 trees * (2 * max depth - 1) * block size
+ */
+STATIC uint
+xfs_calc_remove_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		xfs_calc_iunlink_add_reservation(mp) +
+		max((xfs_calc_inode_res(mp, 2) +
+		     xfs_calc_buf_res(XFS_DIROP_LOG_COUNT(mp),
+				      XFS_FSB_TO_B(mp, 1))),
+		    (xfs_calc_buf_res(4, mp->m_sb.sb_sectsize) +
+		     xfs_calc_buf_res(xfs_allocfree_block_count(mp, 2),
+				      XFS_FSB_TO_B(mp, 1))));
+}
+
+/*
+ * For create, break it in to the two cases that the transaction
+ * covers. We start with the modify case - allocation done by modification
+ * of the state of existing inodes - and the allocation case.
+ */
+
+/*
+ * For create we can modify:
+ *    the parent directory inode: inode size
+ *    the new inode: inode size
+ *    the inode btree entry: block size
+ *    the superblock for the nlink flag: sector size
+ *    the directory btree: (max depth + v2) * dir block size
+ *    the directory inode's bmap btree: (max depth + v2) * block size
+ *    the finobt (record modification and allocation btrees)
+ */
+STATIC uint
+xfs_calc_create_resv_modify(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_inode_res(mp, 2) +
+		xfs_calc_buf_res(1, mp->m_sb.sb_sectsize) +
+		(uint)XFS_FSB_TO_B(mp, 1) +
+		xfs_calc_buf_res(XFS_DIROP_LOG_COUNT(mp), XFS_FSB_TO_B(mp, 1)) +
+		xfs_calc_finobt_res(mp);
+}
+
+/*
+ * For icreate we can allocate some inodes giving:
+ *    the agi and agf of the ag getting the new inodes: 2 * sectorsize
+ *    the superblock for the nlink flag: sector size
+ *    the inode chunk (allocation, optional init)
+ *    the inobt (record insertion)
+ *    the finobt (optional, record insertion)
+ */
+STATIC uint
+xfs_calc_icreate_resv_alloc(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_buf_res(2, mp->m_sb.sb_sectsize) +
+		mp->m_sb.sb_sectsize +
+		xfs_calc_inode_chunk_res(mp, _ALLOC) +
+		xfs_calc_inobt_res(mp) +
+		xfs_calc_finobt_res(mp);
+}
+
+STATIC uint
+xfs_calc_icreate_reservation(xfs_mount_t *mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		max(xfs_calc_icreate_resv_alloc(mp),
+		    xfs_calc_create_resv_modify(mp));
+}
+
+STATIC uint
+xfs_calc_create_tmpfile_reservation(
+	struct xfs_mount        *mp)
+{
+	uint	res = XFS_DQUOT_LOGRES(mp);
+
+	res += xfs_calc_icreate_resv_alloc(mp);
+	return res + xfs_calc_iunlink_add_reservation(mp);
+}
+
+/*
+ * Making a new directory is the same as creating a new file.
+ */
+STATIC uint
+xfs_calc_mkdir_reservation(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_icreate_reservation(mp);
+}
+
+
+/*
+ * Making a new symplink is the same as creating a new file, but
+ * with the added blocks for remote symlink data which can be up to 1kB in
+ * length (XFS_SYMLINK_MAXLEN).
+ */
+STATIC uint
+xfs_calc_symlink_reservation(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_icreate_reservation(mp) +
+	       xfs_calc_buf_res(1, XFS_SYMLINK_MAXLEN);
+}
+
+/*
+ * In freeing an inode we can modify:
+ *    the inode being freed: inode size
+ *    the super block free inode counter, AGF and AGFL: sector size
+ *    the on disk inode (agi unlinked list removal)
+ *    the inode chunk (invalidated, headers only)
+ *    the inode btree
+ *    the finobt (record insertion, removal or modification)
+ *
+ * Note that the inode chunk res. includes an allocfree res. for freeing of the
+ * inode chunk. This is technically extraneous because the inode chunk free is
+ * deferred (it occurs after a transaction roll). Include the extra reservation
+ * anyways since we've had reports of ifree transaction overruns due to too many
+ * agfl fixups during inode chunk frees.
+ */
+STATIC uint
+xfs_calc_ifree_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		xfs_calc_inode_res(mp, 1) +
+		xfs_calc_buf_res(3, mp->m_sb.sb_sectsize) +
+		xfs_calc_iunlink_remove_reservation(mp) +
+		xfs_calc_inode_chunk_res(mp, _FREE) +
+		xfs_calc_inobt_res(mp) +
+		xfs_calc_finobt_res(mp);
+}
+
+/*
+ * When only changing the inode we log the inode and possibly the superblock
+ * We also add a bit of slop for the transaction stuff.
+ */
+STATIC uint
+xfs_calc_ichange_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		xfs_calc_inode_res(mp, 1) +
+		xfs_calc_buf_res(1, mp->m_sb.sb_sectsize);
+
+}
+
+/*
+ * Growing the data section of the filesystem.
+ *	superblock
+ *	agi and agf
+ *	allocation btrees
+ */
+STATIC uint
+xfs_calc_growdata_reservation(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_buf_res(3, mp->m_sb.sb_sectsize) +
+		xfs_calc_buf_res(xfs_allocfree_block_count(mp, 1),
+				 XFS_FSB_TO_B(mp, 1));
+}
+
+/*
+ * Growing the rt section of the filesystem.
+ * In the first set of transactions (ALLOC) we allocate space to the
+ * bitmap or summary files.
+ *	superblock: sector size
+ *	agf of the ag from which the extent is allocated: sector size
+ *	bmap btree for bitmap/summary inode: max depth * blocksize
+ *	bitmap/summary inode: inode size
+ *	allocation btrees for 1 block alloc: 2 * (2 * maxdepth - 1) * blocksize
+ */
+STATIC uint
+xfs_calc_growrtalloc_reservation(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_buf_res(2, mp->m_sb.sb_sectsize) +
+		xfs_calc_buf_res(XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK),
+				 XFS_FSB_TO_B(mp, 1)) +
+		xfs_calc_inode_res(mp, 1) +
+		xfs_calc_buf_res(xfs_allocfree_block_count(mp, 1),
+				 XFS_FSB_TO_B(mp, 1));
+}
+
+/*
+ * Growing the rt section of the filesystem.
+ * In the second set of transactions (ZERO) we zero the new metadata blocks.
+ *	one bitmap/summary block: blocksize
+ */
+STATIC uint
+xfs_calc_growrtzero_reservation(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_buf_res(1, mp->m_sb.sb_blocksize);
+}
+
+/*
+ * Growing the rt section of the filesystem.
+ * In the third set of transactions (FREE) we update metadata without
+ * allocating any new blocks.
+ *	superblock: sector size
+ *	bitmap inode: inode size
+ *	summary inode: inode size
+ *	one bitmap block: blocksize
+ *	summary blocks: new summary size
+ */
+STATIC uint
+xfs_calc_growrtfree_reservation(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_buf_res(1, mp->m_sb.sb_sectsize) +
+		xfs_calc_inode_res(mp, 2) +
+		xfs_calc_buf_res(1, mp->m_sb.sb_blocksize) +
+		xfs_calc_buf_res(1, mp->m_rsumsize);
+}
+
+/*
+ * Logging the inode modification timestamp on a synchronous write.
+ *	inode
+ */
+STATIC uint
+xfs_calc_swrite_reservation(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_inode_res(mp, 1);
+}
+
+/*
+ * Logging the inode mode bits when writing a setuid/setgid file
+ *	inode
+ */
+STATIC uint
+xfs_calc_writeid_reservation(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_inode_res(mp, 1);
+}
+
+/*
+ * Converting the inode from non-attributed to attributed.
+ *	the inode being converted: inode size
+ *	agf block and superblock (for block allocation)
+ *	the new block (directory sized)
+ *	bmap blocks for the new directory block
+ *	allocation btrees
+ */
+STATIC uint
+xfs_calc_addafork_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		xfs_calc_inode_res(mp, 1) +
+		xfs_calc_buf_res(2, mp->m_sb.sb_sectsize) +
+		xfs_calc_buf_res(1, mp->m_dir_geo->blksize) +
+		xfs_calc_buf_res(XFS_DAENTER_BMAP1B(mp, XFS_DATA_FORK) + 1,
+				 XFS_FSB_TO_B(mp, 1)) +
+		xfs_calc_buf_res(xfs_allocfree_block_count(mp, 1),
+				 XFS_FSB_TO_B(mp, 1));
+}
+
+/*
+ * Removing the attribute fork of a file
+ *    the inode being truncated: inode size
+ *    the inode's bmap btree: max depth * block size
+ * And the bmap_finish transaction can free the blocks and bmap blocks:
+ *    the agf for each of the ags: 4 * sector size
+ *    the agfl for each of the ags: 4 * sector size
+ *    the super block to reflect the freed blocks: sector size
+ *    worst case split in allocation btrees per extent assuming 4 extents:
+ *		4 exts * 2 trees * (2 * max depth - 1) * block size
+ */
+STATIC uint
+xfs_calc_attrinval_reservation(
+	struct xfs_mount	*mp)
+{
+	return max((xfs_calc_inode_res(mp, 1) +
+		    xfs_calc_buf_res(XFS_BM_MAXLEVELS(mp, XFS_ATTR_FORK),
+				     XFS_FSB_TO_B(mp, 1))),
+		   (xfs_calc_buf_res(9, mp->m_sb.sb_sectsize) +
+		    xfs_calc_buf_res(xfs_allocfree_block_count(mp, 4),
+				     XFS_FSB_TO_B(mp, 1))));
+}
+
+/*
+ * Setting an attribute at mount time.
+ *	the inode getting the attribute
+ *	the superblock for allocations
+ *	the agfs extents are allocated from
+ *	the attribute btree * max depth
+ *	the inode allocation btree
+ * Since attribute transaction space is dependent on the size of the attribute,
+ * the calculation is done partially at mount time and partially at runtime(see
+ * below).
+ */
+STATIC uint
+xfs_calc_attrsetm_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		xfs_calc_inode_res(mp, 1) +
+		xfs_calc_buf_res(1, mp->m_sb.sb_sectsize) +
+		xfs_calc_buf_res(XFS_DA_NODE_MAXDEPTH, XFS_FSB_TO_B(mp, 1));
+}
+
+/*
+ * Setting an attribute at runtime, transaction space unit per block.
+ * 	the superblock for allocations: sector size
+ *	the inode bmap btree could join or split: max depth * block size
+ * Since the runtime attribute transaction space is dependent on the total
+ * blocks needed for the 1st bmap, here we calculate out the space unit for
+ * one block so that the caller could figure out the total space according
+ * to the attibute extent length in blocks by:
+ *	ext * M_RES(mp)->tr_attrsetrt.tr_logres
+ */
+STATIC uint
+xfs_calc_attrsetrt_reservation(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_buf_res(1, mp->m_sb.sb_sectsize) +
+		xfs_calc_buf_res(XFS_BM_MAXLEVELS(mp, XFS_ATTR_FORK),
+				 XFS_FSB_TO_B(mp, 1));
+}
+
+/*
+ * Removing an attribute.
+ *    the inode: inode size
+ *    the attribute btree could join: max depth * block size
+ *    the inode bmap btree could join or split: max depth * block size
+ * And the bmap_finish transaction can free the attr blocks freed giving:
+ *    the agf for the ag in which the blocks live: 2 * sector size
+ *    the agfl for the ag in which the blocks live: 2 * sector size
+ *    the superblock for the free block count: sector size
+ *    the allocation btrees: 2 exts * 2 trees * (2 * max depth - 1) * block size
+ */
+STATIC uint
+xfs_calc_attrrm_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		max((xfs_calc_inode_res(mp, 1) +
+		     xfs_calc_buf_res(XFS_DA_NODE_MAXDEPTH,
+				      XFS_FSB_TO_B(mp, 1)) +
+		     (uint)XFS_FSB_TO_B(mp,
+					XFS_BM_MAXLEVELS(mp, XFS_ATTR_FORK)) +
+		     xfs_calc_buf_res(XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK), 0)),
+		    (xfs_calc_buf_res(5, mp->m_sb.sb_sectsize) +
+		     xfs_calc_buf_res(xfs_allocfree_block_count(mp, 2),
+				      XFS_FSB_TO_B(mp, 1))));
+}
+
+/*
+ * Clearing a bad agino number in an agi hash bucket.
+ */
+STATIC uint
+xfs_calc_clear_agi_bucket_reservation(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_buf_res(1, mp->m_sb.sb_sectsize);
+}
+
+/*
+ * Adjusting quota limits.
+ *    the disk quota buffer: sizeof(struct xfs_disk_dquot)
+ */
+STATIC uint
+xfs_calc_qm_setqlim_reservation(void)
+{
+	return xfs_calc_buf_res(1, sizeof(struct xfs_disk_dquot));
+}
+
+/*
+ * Allocating quota on disk if needed.
+ *	the write transaction log space for quota file extent allocation
+ *	the unit of quota allocation: one system block size
+ */
+STATIC uint
+xfs_calc_qm_dqalloc_reservation(
+	struct xfs_mount	*mp,
+	bool			for_minlogsize)
+{
+	return xfs_calc_write_reservation(mp, for_minlogsize) +
+		xfs_calc_buf_res(1,
+			XFS_FSB_TO_B(mp, XFS_DQUOT_CLUSTER_SIZE_FSB) - 1);
+}
+
+unsigned int
+xfs_calc_qm_dqalloc_reservation_minlogsize(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_qm_dqalloc_reservation(mp, true);
+}
+
+/*
+ * Syncing the incore super block changes to disk.
+ *     the super block to reflect the changes: sector size
+ */
+STATIC uint
+xfs_calc_sb_reservation(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_buf_res(1, mp->m_sb.sb_sectsize);
+}
+
+void
+xfs_trans_resv_calc(
+	struct xfs_mount	*mp,
+	struct xfs_trans_resv	*resp)
+{
+	int			logcount_adj = 0;
+
+	/*
+	 * The following transactions are logged in physical format and
+	 * require a permanent reservation on space.
+	 */
+	resp->tr_write.tr_logres = xfs_calc_write_reservation(mp, false);
+	resp->tr_write.tr_logcount = XFS_WRITE_LOG_COUNT;
+	resp->tr_write.tr_logflags |= XFS_TRANS_PERM_LOG_RES;
+
+	resp->tr_itruncate.tr_logres = xfs_calc_itruncate_reservation(mp, false);
+	resp->tr_itruncate.tr_logcount = XFS_ITRUNCATE_LOG_COUNT;
+	resp->tr_itruncate.tr_logflags |= XFS_TRANS_PERM_LOG_RES;
+
+	resp->tr_rename.tr_logres = xfs_calc_rename_reservation(mp);
+	resp->tr_rename.tr_logcount = XFS_RENAME_LOG_COUNT;
+	resp->tr_rename.tr_logflags |= XFS_TRANS_PERM_LOG_RES;
+
+	resp->tr_link.tr_logres = xfs_calc_link_reservation(mp);
+	resp->tr_link.tr_logcount = XFS_LINK_LOG_COUNT;
+	resp->tr_link.tr_logflags |= XFS_TRANS_PERM_LOG_RES;
+
+	resp->tr_remove.tr_logres = xfs_calc_remove_reservation(mp);
+	resp->tr_remove.tr_logcount = XFS_REMOVE_LOG_COUNT;
+	resp->tr_remove.tr_logflags |= XFS_TRANS_PERM_LOG_RES;
+
+	resp->tr_symlink.tr_logres = xfs_calc_symlink_reservation(mp);
+	resp->tr_symlink.tr_logcount = XFS_SYMLINK_LOG_COUNT;
+	resp->tr_symlink.tr_logflags |= XFS_TRANS_PERM_LOG_RES;
+
+	resp->tr_create.tr_logres = xfs_calc_icreate_reservation(mp);
+	resp->tr_create.tr_logcount = XFS_CREATE_LOG_COUNT;
+	resp->tr_create.tr_logflags |= XFS_TRANS_PERM_LOG_RES;
+
+	resp->tr_create_tmpfile.tr_logres =
+			xfs_calc_create_tmpfile_reservation(mp);
+	resp->tr_create_tmpfile.tr_logcount = XFS_CREATE_TMPFILE_LOG_COUNT;
+	resp->tr_create_tmpfile.tr_logflags |= XFS_TRANS_PERM_LOG_RES;
+
+	resp->tr_mkdir.tr_logres = xfs_calc_mkdir_reservation(mp);
+	resp->tr_mkdir.tr_logcount = XFS_MKDIR_LOG_COUNT;
+	resp->tr_mkdir.tr_logflags |= XFS_TRANS_PERM_LOG_RES;
+
+	resp->tr_ifree.tr_logres = xfs_calc_ifree_reservation(mp);
+	resp->tr_ifree.tr_logcount = XFS_INACTIVE_LOG_COUNT;
+	resp->tr_ifree.tr_logflags |= XFS_TRANS_PERM_LOG_RES;
+
+	resp->tr_addafork.tr_logres = xfs_calc_addafork_reservation(mp);
+	resp->tr_addafork.tr_logcount = XFS_ADDAFORK_LOG_COUNT;
+	resp->tr_addafork.tr_logflags |= XFS_TRANS_PERM_LOG_RES;
+
+	resp->tr_attrinval.tr_logres = xfs_calc_attrinval_reservation(mp);
+	resp->tr_attrinval.tr_logcount = XFS_ATTRINVAL_LOG_COUNT;
+	resp->tr_attrinval.tr_logflags |= XFS_TRANS_PERM_LOG_RES;
+
+	resp->tr_attrsetm.tr_logres = xfs_calc_attrsetm_reservation(mp);
+	resp->tr_attrsetm.tr_logcount = XFS_ATTRSET_LOG_COUNT;
+	resp->tr_attrsetm.tr_logflags |= XFS_TRANS_PERM_LOG_RES;
+
+	resp->tr_attrrm.tr_logres = xfs_calc_attrrm_reservation(mp);
+	resp->tr_attrrm.tr_logcount = XFS_ATTRRM_LOG_COUNT;
+	resp->tr_attrrm.tr_logflags |= XFS_TRANS_PERM_LOG_RES;
+
+	resp->tr_growrtalloc.tr_logres = xfs_calc_growrtalloc_reservation(mp);
+	resp->tr_growrtalloc.tr_logcount = XFS_DEFAULT_PERM_LOG_COUNT;
+	resp->tr_growrtalloc.tr_logflags |= XFS_TRANS_PERM_LOG_RES;
+
+	resp->tr_qm_dqalloc.tr_logres = xfs_calc_qm_dqalloc_reservation(mp,
+			false);
+	resp->tr_qm_dqalloc.tr_logcount = XFS_WRITE_LOG_COUNT;
+	resp->tr_qm_dqalloc.tr_logflags |= XFS_TRANS_PERM_LOG_RES;
+
+	/*
+	 * The following transactions are logged in logical format with
+	 * a default log count.
+	 */
+	resp->tr_qm_setqlim.tr_logres = xfs_calc_qm_setqlim_reservation();
+	resp->tr_qm_setqlim.tr_logcount = XFS_DEFAULT_LOG_COUNT;
+
+	resp->tr_sb.tr_logres = xfs_calc_sb_reservation(mp);
+	resp->tr_sb.tr_logcount = XFS_DEFAULT_LOG_COUNT;
+
+	/* growdata requires permanent res; it can free space to the last AG */
+	resp->tr_growdata.tr_logres = xfs_calc_growdata_reservation(mp);
+	resp->tr_growdata.tr_logcount = XFS_DEFAULT_PERM_LOG_COUNT;
+	resp->tr_growdata.tr_logflags |= XFS_TRANS_PERM_LOG_RES;
+
+	/* The following transaction are logged in logical format */
+	resp->tr_ichange.tr_logres = xfs_calc_ichange_reservation(mp);
+	resp->tr_fsyncts.tr_logres = xfs_calc_swrite_reservation(mp);
+	resp->tr_writeid.tr_logres = xfs_calc_writeid_reservation(mp);
+	resp->tr_attrsetrt.tr_logres = xfs_calc_attrsetrt_reservation(mp);
+	resp->tr_clearagi.tr_logres = xfs_calc_clear_agi_bucket_reservation(mp);
+	resp->tr_growrtzero.tr_logres = xfs_calc_growrtzero_reservation(mp);
+	resp->tr_growrtfree.tr_logres = xfs_calc_growrtfree_reservation(mp);
+
+	/*
+	 * Add one logcount for BUI items that appear with rmap or reflink,
+	 * one logcount for refcount intent items, and one logcount for rmap
+	 * intent items.
+	 */
+	if (xfs_has_reflink(mp) || xfs_has_rmapbt(mp))
+		logcount_adj++;
+	if (xfs_has_reflink(mp))
+		logcount_adj++;
+	if (xfs_has_rmapbt(mp))
+		logcount_adj++;
+
+	resp->tr_itruncate.tr_logcount += logcount_adj;
+	resp->tr_write.tr_logcount += logcount_adj;
+	resp->tr_qm_dqalloc.tr_logcount += logcount_adj;
+}
diff --git a/fs/xfs/libxfs/xfs_trans_resv.h b/fs/xfs/libxfs/xfs_trans_resv.h
new file mode 100644
index 000000000..0554b9d77
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_trans_resv.h
@@ -0,0 +1,105 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef	__XFS_TRANS_RESV_H__
+#define	__XFS_TRANS_RESV_H__
+
+struct xfs_mount;
+
+/*
+ * structure for maintaining pre-calculated transaction reservations.
+ */
+struct xfs_trans_res {
+	uint	tr_logres;	/* log space unit in bytes per log ticket */
+	int	tr_logcount;	/* number of log operations per log ticket */
+	int	tr_logflags;	/* log flags, currently only used for indicating
+				 * a reservation request is permanent or not */
+};
+
+struct xfs_trans_resv {
+	struct xfs_trans_res	tr_write;	/* extent alloc trans */
+	struct xfs_trans_res	tr_itruncate;	/* truncate trans */
+	struct xfs_trans_res	tr_rename;	/* rename trans */
+	struct xfs_trans_res	tr_link;	/* link trans */
+	struct xfs_trans_res	tr_remove;	/* unlink trans */
+	struct xfs_trans_res	tr_symlink;	/* symlink trans */
+	struct xfs_trans_res	tr_create;	/* create trans */
+	struct xfs_trans_res	tr_create_tmpfile; /* create O_TMPFILE trans */
+	struct xfs_trans_res	tr_mkdir;	/* mkdir trans */
+	struct xfs_trans_res	tr_ifree;	/* inode free trans */
+	struct xfs_trans_res	tr_ichange;	/* inode update trans */
+	struct xfs_trans_res	tr_growdata;	/* fs data section grow trans */
+	struct xfs_trans_res	tr_addafork;	/* add inode attr fork trans */
+	struct xfs_trans_res	tr_writeid;	/* write setuid/setgid file */
+	struct xfs_trans_res	tr_attrinval;	/* attr fork buffer
+						 * invalidation */
+	struct xfs_trans_res	tr_attrsetm;	/* set/create an attribute at
+						 * mount time */
+	struct xfs_trans_res	tr_attrsetrt;	/* set/create an attribute at
+						 * runtime */
+	struct xfs_trans_res	tr_attrrm;	/* remove an attribute */
+	struct xfs_trans_res	tr_clearagi;	/* clear agi unlinked bucket */
+	struct xfs_trans_res	tr_growrtalloc;	/* grow realtime allocations */
+	struct xfs_trans_res	tr_growrtzero;	/* grow realtime zeroing */
+	struct xfs_trans_res	tr_growrtfree;	/* grow realtime freeing */
+	struct xfs_trans_res	tr_qm_setqlim;	/* adjust quota limits */
+	struct xfs_trans_res	tr_qm_dqalloc;	/* allocate quota on disk */
+	struct xfs_trans_res	tr_sb;		/* modify superblock */
+	struct xfs_trans_res	tr_fsyncts;	/* update timestamps on fsync */
+};
+
+/* shorthand way of accessing reservation structure */
+#define M_RES(mp)	(&(mp)->m_resv)
+
+/*
+ * Per-directory log reservation for any directory change.
+ * dir blocks: (1 btree block per level + data block + free block) * dblock size
+ * bmap btree: (levels + 2) * max depth * block size
+ * v2 directory blocks can be fragmented below the dirblksize down to the fsb
+ * size, so account for that in the DAENTER macros.
+ */
+#define	XFS_DIROP_LOG_RES(mp)	\
+	(XFS_FSB_TO_B(mp, XFS_DAENTER_BLOCKS(mp, XFS_DATA_FORK)) + \
+	 (XFS_FSB_TO_B(mp, XFS_DAENTER_BMAPS(mp, XFS_DATA_FORK) + 1)))
+#define	XFS_DIROP_LOG_COUNT(mp)	\
+	(XFS_DAENTER_BLOCKS(mp, XFS_DATA_FORK) + \
+	 XFS_DAENTER_BMAPS(mp, XFS_DATA_FORK) + 1)
+
+/*
+ * Various log count values.
+ */
+#define	XFS_DEFAULT_LOG_COUNT		1
+#define	XFS_DEFAULT_PERM_LOG_COUNT	2
+#define	XFS_ITRUNCATE_LOG_COUNT		2
+#define XFS_INACTIVE_LOG_COUNT		2
+#define	XFS_CREATE_LOG_COUNT		2
+#define	XFS_CREATE_TMPFILE_LOG_COUNT	2
+#define	XFS_MKDIR_LOG_COUNT		3
+#define	XFS_SYMLINK_LOG_COUNT		3
+#define	XFS_REMOVE_LOG_COUNT		2
+#define	XFS_LINK_LOG_COUNT		2
+#define	XFS_RENAME_LOG_COUNT		2
+#define	XFS_WRITE_LOG_COUNT		2
+#define	XFS_ADDAFORK_LOG_COUNT		2
+#define	XFS_ATTRINVAL_LOG_COUNT		1
+#define	XFS_ATTRSET_LOG_COUNT		3
+#define	XFS_ATTRRM_LOG_COUNT		3
+
+/*
+ * Original log operation counts were overestimated in the early days of
+ * reflink.  These are retained here purely for minimum log size calculations
+ * and must not be used for runtime reservations.
+ */
+#define	XFS_ITRUNCATE_LOG_COUNT_REFLINK	8
+#define	XFS_WRITE_LOG_COUNT_REFLINK	8
+
+void xfs_trans_resv_calc(struct xfs_mount *mp, struct xfs_trans_resv *resp);
+uint xfs_allocfree_block_count(struct xfs_mount *mp, uint num_ops);
+
+unsigned int xfs_calc_itruncate_reservation_minlogsize(struct xfs_mount *mp);
+unsigned int xfs_calc_write_reservation_minlogsize(struct xfs_mount *mp);
+unsigned int xfs_calc_qm_dqalloc_reservation_minlogsize(struct xfs_mount *mp);
+
+#endif	/* __XFS_TRANS_RESV_H__ */
diff --git a/fs/xfs/libxfs/xfs_trans_space.h b/fs/xfs/libxfs/xfs_trans_space.h
new file mode 100644
index 000000000..87b31c69a
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_trans_space.h
@@ -0,0 +1,106 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_TRANS_SPACE_H__
+#define __XFS_TRANS_SPACE_H__
+
+/*
+ * Components of space reservations.
+ */
+
+/* Worst case number of rmaps that can be held in a block. */
+#define XFS_MAX_CONTIG_RMAPS_PER_BLOCK(mp)    \
+		(((mp)->m_rmap_mxr[0]) - ((mp)->m_rmap_mnr[0]))
+
+/* Adding one rmap could split every level up to the top of the tree. */
+#define XFS_RMAPADD_SPACE_RES(mp) ((mp)->m_rmap_maxlevels)
+
+/*
+ * Note that we historically set m_rmap_maxlevels to 9 when reflink is enabled,
+ * so we must preserve this behavior to avoid changing the transaction space
+ * reservations and minimum log size calculations for existing filesystems.
+ */
+#define XFS_OLD_REFLINK_RMAP_MAXLEVELS		9
+
+/* Blocks we might need to add "b" rmaps to a tree. */
+#define XFS_NRMAPADD_SPACE_RES(mp, b)\
+	(((b + XFS_MAX_CONTIG_RMAPS_PER_BLOCK(mp) - 1) / \
+	  XFS_MAX_CONTIG_RMAPS_PER_BLOCK(mp)) * \
+	  XFS_RMAPADD_SPACE_RES(mp))
+
+#define XFS_MAX_CONTIG_EXTENTS_PER_BLOCK(mp)    \
+		(((mp)->m_alloc_mxr[0]) - ((mp)->m_alloc_mnr[0]))
+#define	XFS_EXTENTADD_SPACE_RES(mp,w)	(XFS_BM_MAXLEVELS(mp,w) - 1)
+#define XFS_NEXTENTADD_SPACE_RES(mp,b,w)\
+	(((b + XFS_MAX_CONTIG_EXTENTS_PER_BLOCK(mp) - 1) / \
+	  XFS_MAX_CONTIG_EXTENTS_PER_BLOCK(mp)) * \
+	  XFS_EXTENTADD_SPACE_RES(mp,w))
+
+/* Blocks we might need to add "b" mappings & rmappings to a file. */
+#define XFS_SWAP_RMAP_SPACE_RES(mp,b,w)\
+	(XFS_NEXTENTADD_SPACE_RES((mp), (b), (w)) + \
+	 XFS_NRMAPADD_SPACE_RES((mp), (b)))
+
+#define	XFS_DAENTER_1B(mp,w)	\
+	((w) == XFS_DATA_FORK ? (mp)->m_dir_geo->fsbcount : 1)
+#define	XFS_DAENTER_DBS(mp,w)	\
+	(XFS_DA_NODE_MAXDEPTH + (((w) == XFS_DATA_FORK) ? 2 : 0))
+#define	XFS_DAENTER_BLOCKS(mp,w)	\
+	(XFS_DAENTER_1B(mp,w) * XFS_DAENTER_DBS(mp,w))
+#define	XFS_DAENTER_BMAP1B(mp,w)	\
+	XFS_NEXTENTADD_SPACE_RES(mp, XFS_DAENTER_1B(mp, w), w)
+#define	XFS_DAENTER_BMAPS(mp,w)		\
+	(XFS_DAENTER_DBS(mp,w) * XFS_DAENTER_BMAP1B(mp,w))
+#define	XFS_DAENTER_SPACE_RES(mp,w)	\
+	(XFS_DAENTER_BLOCKS(mp,w) + XFS_DAENTER_BMAPS(mp,w))
+#define	XFS_DAREMOVE_SPACE_RES(mp,w)	XFS_DAENTER_BMAPS(mp,w)
+#define	XFS_DIRENTER_MAX_SPLIT(mp,nl)	1
+#define	XFS_DIRENTER_SPACE_RES(mp,nl)	\
+	(XFS_DAENTER_SPACE_RES(mp, XFS_DATA_FORK) * \
+	 XFS_DIRENTER_MAX_SPLIT(mp,nl))
+#define	XFS_DIRREMOVE_SPACE_RES(mp)	\
+	XFS_DAREMOVE_SPACE_RES(mp, XFS_DATA_FORK)
+#define	XFS_IALLOC_SPACE_RES(mp)	\
+	(M_IGEO(mp)->ialloc_blks + \
+	 ((xfs_has_finobt(mp) ? 2 : 1) * M_IGEO(mp)->inobt_maxlevels))
+
+/*
+ * Space reservation values for various transactions.
+ */
+#define	XFS_ADDAFORK_SPACE_RES(mp)	\
+	((mp)->m_dir_geo->fsbcount + XFS_DAENTER_BMAP1B(mp, XFS_DATA_FORK))
+#define	XFS_ATTRRM_SPACE_RES(mp)	\
+	XFS_DAREMOVE_SPACE_RES(mp, XFS_ATTR_FORK)
+/* This macro is not used - see inline code in xfs_attr_set */
+#define	XFS_ATTRSET_SPACE_RES(mp, v)	\
+	(XFS_DAENTER_SPACE_RES(mp, XFS_ATTR_FORK) + XFS_B_TO_FSB(mp, v))
+#define	XFS_CREATE_SPACE_RES(mp,nl)	\
+	(XFS_IALLOC_SPACE_RES(mp) + XFS_DIRENTER_SPACE_RES(mp,nl))
+#define	XFS_DIOSTRAT_SPACE_RES(mp, v)	\
+	(XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK) + (v))
+#define	XFS_GROWFS_SPACE_RES(mp)	\
+	(2 * (mp)->m_alloc_maxlevels)
+#define	XFS_GROWFSRT_SPACE_RES(mp,b)	\
+	((b) + XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK))
+#define	XFS_LINK_SPACE_RES(mp,nl)	\
+	XFS_DIRENTER_SPACE_RES(mp,nl)
+#define	XFS_MKDIR_SPACE_RES(mp,nl)	\
+	(XFS_IALLOC_SPACE_RES(mp) + XFS_DIRENTER_SPACE_RES(mp,nl))
+#define	XFS_QM_DQALLOC_SPACE_RES(mp)	\
+	(XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK) + \
+	 XFS_DQUOT_CLUSTER_SIZE_FSB)
+#define	XFS_QM_QINOCREATE_SPACE_RES(mp)	\
+	XFS_IALLOC_SPACE_RES(mp)
+#define	XFS_REMOVE_SPACE_RES(mp)	\
+	XFS_DIRREMOVE_SPACE_RES(mp)
+#define	XFS_RENAME_SPACE_RES(mp,nl)	\
+	(XFS_DIRREMOVE_SPACE_RES(mp) + XFS_DIRENTER_SPACE_RES(mp,nl))
+#define	XFS_SYMLINK_SPACE_RES(mp,nl,b)	\
+	(XFS_IALLOC_SPACE_RES(mp) + XFS_DIRENTER_SPACE_RES(mp,nl) + (b))
+#define XFS_IFREE_SPACE_RES(mp)		\
+	(xfs_has_finobt(mp) ? M_IGEO(mp)->inobt_maxlevels : 0)
+
+
+#endif	/* __XFS_TRANS_SPACE_H__ */
diff --git a/fs/xfs/libxfs/xfs_types.c b/fs/xfs/libxfs/xfs_types.c
new file mode 100644
index 000000000..5c2765934
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_types.c
@@ -0,0 +1,230 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * Copyright (C) 2017 Oracle.
+ * All Rights Reserved.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_format.h"
+#include "xfs_shared.h"
+#include "xfs_trans_resv.h"
+#include "xfs_bit.h"
+#include "xfs_mount.h"
+#include "xfs_ag.h"
+
+
+/*
+ * Verify that an AG block number pointer neither points outside the AG
+ * nor points at static metadata.
+ */
+static inline bool
+xfs_verify_agno_agbno(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		agno,
+	xfs_agblock_t		agbno)
+{
+	xfs_agblock_t		eoag;
+
+	eoag = xfs_ag_block_count(mp, agno);
+	if (agbno >= eoag)
+		return false;
+	if (agbno <= XFS_AGFL_BLOCK(mp))
+		return false;
+	return true;
+}
+
+/*
+ * Verify that an FS block number pointer neither points outside the
+ * filesystem nor points at static AG metadata.
+ */
+inline bool
+xfs_verify_fsbno(
+	struct xfs_mount	*mp,
+	xfs_fsblock_t		fsbno)
+{
+	xfs_agnumber_t		agno = XFS_FSB_TO_AGNO(mp, fsbno);
+
+	if (agno >= mp->m_sb.sb_agcount)
+		return false;
+	return xfs_verify_agno_agbno(mp, agno, XFS_FSB_TO_AGBNO(mp, fsbno));
+}
+
+/*
+ * Verify that a data device extent is fully contained inside the filesystem,
+ * does not cross an AG boundary, and does not point at static metadata.
+ */
+bool
+xfs_verify_fsbext(
+	struct xfs_mount	*mp,
+	xfs_fsblock_t		fsbno,
+	xfs_fsblock_t		len)
+{
+	if (fsbno + len <= fsbno)
+		return false;
+
+	if (!xfs_verify_fsbno(mp, fsbno))
+		return false;
+
+	if (!xfs_verify_fsbno(mp, fsbno + len - 1))
+		return false;
+
+	return  XFS_FSB_TO_AGNO(mp, fsbno) ==
+		XFS_FSB_TO_AGNO(mp, fsbno + len - 1);
+}
+
+/*
+ * Verify that an AG inode number pointer neither points outside the AG
+ * nor points at static metadata.
+ */
+static inline bool
+xfs_verify_agno_agino(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		agno,
+	xfs_agino_t		agino)
+{
+	xfs_agino_t		first;
+	xfs_agino_t		last;
+
+	xfs_agino_range(mp, agno, &first, &last);
+	return agino >= first && agino <= last;
+}
+
+/*
+ * Verify that an FS inode number pointer neither points outside the
+ * filesystem nor points at static AG metadata.
+ */
+inline bool
+xfs_verify_ino(
+	struct xfs_mount	*mp,
+	xfs_ino_t		ino)
+{
+	xfs_agnumber_t		agno = XFS_INO_TO_AGNO(mp, ino);
+	xfs_agino_t		agino = XFS_INO_TO_AGINO(mp, ino);
+
+	if (agno >= mp->m_sb.sb_agcount)
+		return false;
+	if (XFS_AGINO_TO_INO(mp, agno, agino) != ino)
+		return false;
+	return xfs_verify_agno_agino(mp, agno, agino);
+}
+
+/* Is this an internal inode number? */
+inline bool
+xfs_internal_inum(
+	struct xfs_mount	*mp,
+	xfs_ino_t		ino)
+{
+	return ino == mp->m_sb.sb_rbmino || ino == mp->m_sb.sb_rsumino ||
+		(xfs_has_quota(mp) &&
+		 xfs_is_quota_inode(&mp->m_sb, ino));
+}
+
+/*
+ * Verify that a directory entry's inode number doesn't point at an internal
+ * inode, empty space, or static AG metadata.
+ */
+bool
+xfs_verify_dir_ino(
+	struct xfs_mount	*mp,
+	xfs_ino_t		ino)
+{
+	if (xfs_internal_inum(mp, ino))
+		return false;
+	return xfs_verify_ino(mp, ino);
+}
+
+/*
+ * Verify that an realtime block number pointer doesn't point off the
+ * end of the realtime device.
+ */
+inline bool
+xfs_verify_rtbno(
+	struct xfs_mount	*mp,
+	xfs_rtblock_t		rtbno)
+{
+	return rtbno < mp->m_sb.sb_rblocks;
+}
+
+/* Verify that a realtime device extent is fully contained inside the volume. */
+bool
+xfs_verify_rtext(
+	struct xfs_mount	*mp,
+	xfs_rtblock_t		rtbno,
+	xfs_rtblock_t		len)
+{
+	if (rtbno + len <= rtbno)
+		return false;
+
+	if (!xfs_verify_rtbno(mp, rtbno))
+		return false;
+
+	return xfs_verify_rtbno(mp, rtbno + len - 1);
+}
+
+/* Calculate the range of valid icount values. */
+inline void
+xfs_icount_range(
+	struct xfs_mount	*mp,
+	unsigned long long	*min,
+	unsigned long long	*max)
+{
+	unsigned long long	nr_inos = 0;
+	struct xfs_perag	*pag;
+	xfs_agnumber_t		agno;
+
+	/* root, rtbitmap, rtsum all live in the first chunk */
+	*min = XFS_INODES_PER_CHUNK;
+
+	for_each_perag(mp, agno, pag)
+		nr_inos += pag->agino_max - pag->agino_min + 1;
+	*max = nr_inos;
+}
+
+/* Sanity-checking of inode counts. */
+bool
+xfs_verify_icount(
+	struct xfs_mount	*mp,
+	unsigned long long	icount)
+{
+	unsigned long long	min, max;
+
+	xfs_icount_range(mp, &min, &max);
+	return icount >= min && icount <= max;
+}
+
+/* Sanity-checking of dir/attr block offsets. */
+bool
+xfs_verify_dablk(
+	struct xfs_mount	*mp,
+	xfs_fileoff_t		dabno)
+{
+	xfs_dablk_t		max_dablk = -1U;
+
+	return dabno <= max_dablk;
+}
+
+/* Check that a file block offset does not exceed the maximum. */
+bool
+xfs_verify_fileoff(
+	struct xfs_mount	*mp,
+	xfs_fileoff_t		off)
+{
+	return off <= XFS_MAX_FILEOFF;
+}
+
+/* Check that a range of file block offsets do not exceed the maximum. */
+bool
+xfs_verify_fileext(
+	struct xfs_mount	*mp,
+	xfs_fileoff_t		off,
+	xfs_fileoff_t		len)
+{
+	if (off + len <= off)
+		return false;
+
+	if (!xfs_verify_fileoff(mp, off))
+		return false;
+
+	return xfs_verify_fileoff(mp, off + len - 1);
+}
diff --git a/fs/xfs/libxfs/xfs_types.h b/fs/xfs/libxfs/xfs_types.h
new file mode 100644
index 000000000..5ebdda7e1
--- /dev/null
+++ b/fs/xfs/libxfs/xfs_types.h
@@ -0,0 +1,230 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ */
+#ifndef __XFS_TYPES_H__
+#define	__XFS_TYPES_H__
+
+typedef uint32_t	prid_t;		/* project ID */
+
+typedef uint32_t	xfs_agblock_t;	/* blockno in alloc. group */
+typedef uint32_t	xfs_agino_t;	/* inode # within allocation grp */
+typedef uint32_t	xfs_extlen_t;	/* extent length in blocks */
+typedef uint32_t	xfs_agnumber_t;	/* allocation group number */
+typedef uint64_t	xfs_extnum_t;	/* # of extents in a file */
+typedef uint32_t	xfs_aextnum_t;	/* # extents in an attribute fork */
+typedef int64_t		xfs_fsize_t;	/* bytes in a file */
+typedef uint64_t	xfs_ufsize_t;	/* unsigned bytes in a file */
+
+typedef int32_t		xfs_suminfo_t;	/* type of bitmap summary info */
+typedef uint32_t	xfs_rtword_t;	/* word type for bitmap manipulations */
+
+typedef int64_t		xfs_lsn_t;	/* log sequence number */
+typedef int64_t		xfs_csn_t;	/* CIL sequence number */
+
+typedef uint32_t	xfs_dablk_t;	/* dir/attr block number (in file) */
+typedef uint32_t	xfs_dahash_t;	/* dir/attr hash value */
+
+typedef uint64_t	xfs_fsblock_t;	/* blockno in filesystem (agno|agbno) */
+typedef uint64_t	xfs_rfsblock_t;	/* blockno in filesystem (raw) */
+typedef uint64_t	xfs_rtblock_t;	/* extent (block) in realtime area */
+typedef uint64_t	xfs_fileoff_t;	/* block number in a file */
+typedef uint64_t	xfs_filblks_t;	/* number of blocks in a file */
+
+typedef int64_t		xfs_srtblock_t;	/* signed version of xfs_rtblock_t */
+
+/*
+ * New verifiers will return the instruction address of the failing check.
+ * NULL means everything is ok.
+ */
+typedef void *		xfs_failaddr_t;
+
+/*
+ * Null values for the types.
+ */
+#define	NULLFSBLOCK	((xfs_fsblock_t)-1)
+#define	NULLRFSBLOCK	((xfs_rfsblock_t)-1)
+#define	NULLRTBLOCK	((xfs_rtblock_t)-1)
+#define	NULLFILEOFF	((xfs_fileoff_t)-1)
+
+#define	NULLAGBLOCK	((xfs_agblock_t)-1)
+#define	NULLAGNUMBER	((xfs_agnumber_t)-1)
+
+#define NULLCOMMITLSN	((xfs_lsn_t)-1)
+
+#define	NULLFSINO	((xfs_ino_t)-1)
+#define	NULLAGINO	((xfs_agino_t)-1)
+
+/*
+ * Minimum and maximum blocksize and sectorsize.
+ * The blocksize upper limit is pretty much arbitrary.
+ * The sectorsize upper limit is due to sizeof(sb_sectsize).
+ * CRC enable filesystems use 512 byte inodes, meaning 512 byte block sizes
+ * cannot be used.
+ */
+#define XFS_MIN_BLOCKSIZE_LOG	9	/* i.e. 512 bytes */
+#define XFS_MAX_BLOCKSIZE_LOG	16	/* i.e. 65536 bytes */
+#define XFS_MIN_BLOCKSIZE	(1 << XFS_MIN_BLOCKSIZE_LOG)
+#define XFS_MAX_BLOCKSIZE	(1 << XFS_MAX_BLOCKSIZE_LOG)
+#define XFS_MIN_CRC_BLOCKSIZE	(1 << (XFS_MIN_BLOCKSIZE_LOG + 1))
+#define XFS_MIN_SECTORSIZE_LOG	9	/* i.e. 512 bytes */
+#define XFS_MAX_SECTORSIZE_LOG	15	/* i.e. 32768 bytes */
+#define XFS_MIN_SECTORSIZE	(1 << XFS_MIN_SECTORSIZE_LOG)
+#define XFS_MAX_SECTORSIZE	(1 << XFS_MAX_SECTORSIZE_LOG)
+
+/*
+ * Inode fork identifiers.
+ */
+#define	XFS_DATA_FORK	0
+#define	XFS_ATTR_FORK	1
+#define	XFS_COW_FORK	2
+
+#define XFS_WHICHFORK_STRINGS \
+	{ XFS_DATA_FORK, 	"data" }, \
+	{ XFS_ATTR_FORK,	"attr" }, \
+	{ XFS_COW_FORK,		"cow" }
+
+/*
+ * Min numbers of data/attr fork btree root pointers.
+ */
+#define MINDBTPTRS	3
+#define MINABTPTRS	2
+
+/*
+ * MAXNAMELEN is the length (including the terminating null) of
+ * the longest permissible file (component) name.
+ */
+#define MAXNAMELEN	256
+
+/*
+ * This enum is used in string mapping in xfs_trace.h; please keep the
+ * TRACE_DEFINE_ENUMs for it up to date.
+ */
+typedef enum {
+	XFS_LOOKUP_EQi, XFS_LOOKUP_LEi, XFS_LOOKUP_GEi
+} xfs_lookup_t;
+
+#define XFS_AG_BTREE_CMP_FORMAT_STR \
+	{ XFS_LOOKUP_EQi,	"eq" }, \
+	{ XFS_LOOKUP_LEi,	"le" }, \
+	{ XFS_LOOKUP_GEi,	"ge" }
+
+/*
+ * This enum is used in string mapping in xfs_trace.h and scrub/trace.h;
+ * please keep the TRACE_DEFINE_ENUMs for it up to date.
+ */
+typedef enum {
+	XFS_BTNUM_BNOi, XFS_BTNUM_CNTi, XFS_BTNUM_RMAPi, XFS_BTNUM_BMAPi,
+	XFS_BTNUM_INOi, XFS_BTNUM_FINOi, XFS_BTNUM_REFCi, XFS_BTNUM_MAX
+} xfs_btnum_t;
+
+#define XFS_BTNUM_STRINGS \
+	{ XFS_BTNUM_BNOi,	"bnobt" }, \
+	{ XFS_BTNUM_CNTi,	"cntbt" }, \
+	{ XFS_BTNUM_RMAPi,	"rmapbt" }, \
+	{ XFS_BTNUM_BMAPi,	"bmbt" }, \
+	{ XFS_BTNUM_INOi,	"inobt" }, \
+	{ XFS_BTNUM_FINOi,	"finobt" }, \
+	{ XFS_BTNUM_REFCi,	"refcbt" }
+
+struct xfs_name {
+	const unsigned char	*name;
+	int			len;
+	int			type;
+};
+
+/*
+ * uid_t and gid_t are hard-coded to 32 bits in the inode.
+ * Hence, an 'id' in a dquot is 32 bits..
+ */
+typedef uint32_t	xfs_dqid_t;
+
+/*
+ * Constants for bit manipulations.
+ */
+#define	XFS_NBBYLOG	3		/* log2(NBBY) */
+#define	XFS_WORDLOG	2		/* log2(sizeof(xfs_rtword_t)) */
+#define	XFS_NBWORDLOG	(XFS_NBBYLOG + XFS_WORDLOG)
+#define	XFS_NBWORD	(1 << XFS_NBWORDLOG)
+#define	XFS_WORDMASK	((1 << XFS_WORDLOG) - 1)
+
+struct xfs_iext_cursor {
+	struct xfs_iext_leaf	*leaf;
+	int			pos;
+};
+
+typedef enum {
+	XFS_EXT_NORM, XFS_EXT_UNWRITTEN,
+} xfs_exntst_t;
+
+typedef struct xfs_bmbt_irec
+{
+	xfs_fileoff_t	br_startoff;	/* starting file offset */
+	xfs_fsblock_t	br_startblock;	/* starting block number */
+	xfs_filblks_t	br_blockcount;	/* number of blocks */
+	xfs_exntst_t	br_state;	/* extent state */
+} xfs_bmbt_irec_t;
+
+enum xfs_refc_domain {
+	XFS_REFC_DOMAIN_SHARED = 0,
+	XFS_REFC_DOMAIN_COW,
+};
+
+#define XFS_REFC_DOMAIN_STRINGS \
+	{ XFS_REFC_DOMAIN_SHARED,	"shared" }, \
+	{ XFS_REFC_DOMAIN_COW,		"cow" }
+
+struct xfs_refcount_irec {
+	xfs_agblock_t	rc_startblock;	/* starting block number */
+	xfs_extlen_t	rc_blockcount;	/* count of free blocks */
+	xfs_nlink_t	rc_refcount;	/* number of inodes linked here */
+	enum xfs_refc_domain	rc_domain; /* shared or cow staging extent? */
+};
+
+#define XFS_RMAP_ATTR_FORK		(1 << 0)
+#define XFS_RMAP_BMBT_BLOCK		(1 << 1)
+#define XFS_RMAP_UNWRITTEN		(1 << 2)
+#define XFS_RMAP_KEY_FLAGS		(XFS_RMAP_ATTR_FORK | \
+					 XFS_RMAP_BMBT_BLOCK)
+#define XFS_RMAP_REC_FLAGS		(XFS_RMAP_UNWRITTEN)
+struct xfs_rmap_irec {
+	xfs_agblock_t	rm_startblock;	/* extent start block */
+	xfs_extlen_t	rm_blockcount;	/* extent length */
+	uint64_t	rm_owner;	/* extent owner */
+	uint64_t	rm_offset;	/* offset within the owner */
+	unsigned int	rm_flags;	/* state flags */
+};
+
+/* per-AG block reservation types */
+enum xfs_ag_resv_type {
+	XFS_AG_RESV_NONE = 0,
+	XFS_AG_RESV_AGFL,
+	XFS_AG_RESV_METADATA,
+	XFS_AG_RESV_RMAPBT,
+};
+
+/*
+ * Type verifier functions
+ */
+struct xfs_mount;
+
+bool xfs_verify_fsbno(struct xfs_mount *mp, xfs_fsblock_t fsbno);
+bool xfs_verify_fsbext(struct xfs_mount *mp, xfs_fsblock_t fsbno,
+		xfs_fsblock_t len);
+
+bool xfs_verify_ino(struct xfs_mount *mp, xfs_ino_t ino);
+bool xfs_internal_inum(struct xfs_mount *mp, xfs_ino_t ino);
+bool xfs_verify_dir_ino(struct xfs_mount *mp, xfs_ino_t ino);
+bool xfs_verify_rtbno(struct xfs_mount *mp, xfs_rtblock_t rtbno);
+bool xfs_verify_rtext(struct xfs_mount *mp, xfs_rtblock_t rtbno,
+		xfs_rtblock_t len);
+bool xfs_verify_icount(struct xfs_mount *mp, unsigned long long icount);
+bool xfs_verify_dablk(struct xfs_mount *mp, xfs_fileoff_t off);
+void xfs_icount_range(struct xfs_mount *mp, unsigned long long *min,
+		unsigned long long *max);
+bool xfs_verify_fileoff(struct xfs_mount *mp, xfs_fileoff_t off);
+bool xfs_verify_fileext(struct xfs_mount *mp, xfs_fileoff_t off,
+		xfs_fileoff_t len);
+
+#endif	/* __XFS_TYPES_H__ */