Adding upstream version 6.1.76.upstream/6.1.76upstream

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

30 files changed, 30505 insertions, 0 deletions

diff --git a/fs/ceph/Kconfig b/fs/ceph/Kconfig
new file mode 100644
index 000000000..94df85414
--- /dev/null
+++ b/fs/ceph/Kconfig
@@ -0,0 +1,51 @@
+# SPDX-License-Identifier: GPL-2.0-only
+config CEPH_FS
+	tristate "Ceph distributed file system"
+	depends on INET
+	select CEPH_LIB
+	select LIBCRC32C
+	select CRYPTO_AES
+	select CRYPTO
+	select NETFS_SUPPORT
+	default n
+	help
+	  Choose Y or M here to include support for mounting the
+	  experimental Ceph distributed file system.  Ceph is an extremely
+	  scalable file system designed to provide high performance,
+	  reliable access to petabytes of storage.
+
+	  More information at https://ceph.io/.
+
+	  If unsure, say N.
+
+if CEPH_FS
+config CEPH_FSCACHE
+	bool "Enable Ceph client caching support"
+	depends on CEPH_FS=m && FSCACHE || CEPH_FS=y && FSCACHE=y
+	help
+	  Choose Y here to enable persistent, read-only local
+	  caching support for Ceph clients using FS-Cache
+
+endif
+
+config CEPH_FS_POSIX_ACL
+	bool "Ceph POSIX Access Control Lists"
+	depends on CEPH_FS
+	select FS_POSIX_ACL
+	help
+	  POSIX Access Control Lists (ACLs) support permissions for users and
+	  groups beyond the owner/group/world scheme.
+
+	  If you don't know what Access Control Lists are, say N
+
+config CEPH_FS_SECURITY_LABEL
+	bool "CephFS Security Labels"
+	depends on CEPH_FS && SECURITY
+	help
+	  Security labels support alternative access control models
+	  implemented by security modules like SELinux. This option
+	  enables an extended attribute handler for file security
+	  labels in the Ceph filesystem.
+
+	  If you are not using a security module that requires using
+	  extended attributes for file security labels, say N.
diff --git a/fs/ceph/Makefile b/fs/ceph/Makefile
new file mode 100644
index 000000000..50c635dc7
--- /dev/null
+++ b/fs/ceph/Makefile
@@ -0,0 +1,14 @@
+# SPDX-License-Identifier: GPL-2.0
+#
+# Makefile for CEPH filesystem.
+#
+
+obj-$(CONFIG_CEPH_FS) += ceph.o
+
+ceph-y := super.o inode.o dir.o file.o locks.o addr.o ioctl.o \
+	export.o caps.o snap.o xattr.o quota.o io.o \
+	mds_client.o mdsmap.o strings.o ceph_frag.o \
+	debugfs.o util.o metric.o
+
+ceph-$(CONFIG_CEPH_FSCACHE) += cache.o
+ceph-$(CONFIG_CEPH_FS_POSIX_ACL) += acl.o
diff --git a/fs/ceph/acl.c b/fs/ceph/acl.c
new file mode 100644
index 000000000..f4fc8e0b8
--- /dev/null
+++ b/fs/ceph/acl.c
@@ -0,0 +1,262 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * linux/fs/ceph/acl.c
+ *
+ * Copyright (C) 2013 Guangliang Zhao, <lucienchao@gmail.com>
+ */
+
+#include <linux/ceph/ceph_debug.h>
+#include <linux/fs.h>
+#include <linux/string.h>
+#include <linux/xattr.h>
+#include <linux/posix_acl_xattr.h>
+#include <linux/posix_acl.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+
+#include "super.h"
+
+static inline void ceph_set_cached_acl(struct inode *inode,
+					int type, struct posix_acl *acl)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+
+	spin_lock(&ci->i_ceph_lock);
+	if (__ceph_caps_issued_mask_metric(ci, CEPH_CAP_XATTR_SHARED, 0))
+		set_cached_acl(inode, type, acl);
+	else
+		forget_cached_acl(inode, type);
+	spin_unlock(&ci->i_ceph_lock);
+}
+
+struct posix_acl *ceph_get_acl(struct inode *inode, int type, bool rcu)
+{
+	int size;
+	unsigned int retry_cnt = 0;
+	const char *name;
+	char *value = NULL;
+	struct posix_acl *acl;
+
+	if (rcu)
+		return ERR_PTR(-ECHILD);
+
+	switch (type) {
+	case ACL_TYPE_ACCESS:
+		name = XATTR_NAME_POSIX_ACL_ACCESS;
+		break;
+	case ACL_TYPE_DEFAULT:
+		name = XATTR_NAME_POSIX_ACL_DEFAULT;
+		break;
+	default:
+		BUG();
+	}
+
+retry:
+	size = __ceph_getxattr(inode, name, "", 0);
+	if (size > 0) {
+		value = kzalloc(size, GFP_NOFS);
+		if (!value)
+			return ERR_PTR(-ENOMEM);
+		size = __ceph_getxattr(inode, name, value, size);
+	}
+
+	if (size == -ERANGE && retry_cnt < 10) {
+		retry_cnt++;
+		kfree(value);
+		value = NULL;
+		goto retry;
+	}
+
+	if (size > 0) {
+		acl = posix_acl_from_xattr(&init_user_ns, value, size);
+	} else if (size == -ENODATA || size == 0) {
+		acl = NULL;
+	} else {
+		pr_err_ratelimited("get acl %llx.%llx failed, err=%d\n",
+				   ceph_vinop(inode), size);
+		acl = ERR_PTR(-EIO);
+	}
+
+	kfree(value);
+
+	if (!IS_ERR(acl))
+		ceph_set_cached_acl(inode, type, acl);
+
+	return acl;
+}
+
+int ceph_set_acl(struct user_namespace *mnt_userns, struct inode *inode,
+		 struct posix_acl *acl, int type)
+{
+	int ret = 0, size = 0;
+	const char *name = NULL;
+	char *value = NULL;
+	struct iattr newattrs;
+	struct timespec64 old_ctime = inode->i_ctime;
+	umode_t new_mode = inode->i_mode, old_mode = inode->i_mode;
+
+	if (ceph_snap(inode) != CEPH_NOSNAP) {
+		ret = -EROFS;
+		goto out;
+	}
+
+	switch (type) {
+	case ACL_TYPE_ACCESS:
+		name = XATTR_NAME_POSIX_ACL_ACCESS;
+		if (acl) {
+			ret = posix_acl_update_mode(&init_user_ns, inode,
+						    &new_mode, &acl);
+			if (ret)
+				goto out;
+		}
+		break;
+	case ACL_TYPE_DEFAULT:
+		if (!S_ISDIR(inode->i_mode)) {
+			ret = acl ? -EINVAL : 0;
+			goto out;
+		}
+		name = XATTR_NAME_POSIX_ACL_DEFAULT;
+		break;
+	default:
+		ret = -EINVAL;
+		goto out;
+	}
+
+	if (acl) {
+		size = posix_acl_xattr_size(acl->a_count);
+		value = kmalloc(size, GFP_NOFS);
+		if (!value) {
+			ret = -ENOMEM;
+			goto out;
+		}
+
+		ret = posix_acl_to_xattr(&init_user_ns, acl, value, size);
+		if (ret < 0)
+			goto out_free;
+	}
+
+	if (new_mode != old_mode) {
+		newattrs.ia_ctime = current_time(inode);
+		newattrs.ia_mode = new_mode;
+		newattrs.ia_valid = ATTR_MODE | ATTR_CTIME;
+		ret = __ceph_setattr(inode, &newattrs);
+		if (ret)
+			goto out_free;
+	}
+
+	ret = __ceph_setxattr(inode, name, value, size, 0);
+	if (ret) {
+		if (new_mode != old_mode) {
+			newattrs.ia_ctime = old_ctime;
+			newattrs.ia_mode = old_mode;
+			newattrs.ia_valid = ATTR_MODE | ATTR_CTIME;
+			__ceph_setattr(inode, &newattrs);
+		}
+		goto out_free;
+	}
+
+	ceph_set_cached_acl(inode, type, acl);
+
+out_free:
+	kfree(value);
+out:
+	return ret;
+}
+
+int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
+		       struct ceph_acl_sec_ctx *as_ctx)
+{
+	struct posix_acl *acl, *default_acl;
+	size_t val_size1 = 0, val_size2 = 0;
+	struct ceph_pagelist *pagelist = NULL;
+	void *tmp_buf = NULL;
+	int err;
+
+	err = posix_acl_create(dir, mode, &default_acl, &acl);
+	if (err)
+		return err;
+
+	if (acl) {
+		err = posix_acl_equiv_mode(acl, mode);
+		if (err < 0)
+			goto out_err;
+		if (err == 0) {
+			posix_acl_release(acl);
+			acl = NULL;
+		}
+	}
+
+	if (!default_acl && !acl)
+		return 0;
+
+	if (acl)
+		val_size1 = posix_acl_xattr_size(acl->a_count);
+	if (default_acl)
+		val_size2 = posix_acl_xattr_size(default_acl->a_count);
+
+	err = -ENOMEM;
+	tmp_buf = kmalloc(max(val_size1, val_size2), GFP_KERNEL);
+	if (!tmp_buf)
+		goto out_err;
+	pagelist = ceph_pagelist_alloc(GFP_KERNEL);
+	if (!pagelist)
+		goto out_err;
+
+	err = ceph_pagelist_reserve(pagelist, PAGE_SIZE);
+	if (err)
+		goto out_err;
+
+	ceph_pagelist_encode_32(pagelist, acl && default_acl ? 2 : 1);
+
+	if (acl) {
+		size_t len = strlen(XATTR_NAME_POSIX_ACL_ACCESS);
+		err = ceph_pagelist_reserve(pagelist, len + val_size1 + 8);
+		if (err)
+			goto out_err;
+		ceph_pagelist_encode_string(pagelist, XATTR_NAME_POSIX_ACL_ACCESS,
+					    len);
+		err = posix_acl_to_xattr(&init_user_ns, acl,
+					 tmp_buf, val_size1);
+		if (err < 0)
+			goto out_err;
+		ceph_pagelist_encode_32(pagelist, val_size1);
+		ceph_pagelist_append(pagelist, tmp_buf, val_size1);
+	}
+	if (default_acl) {
+		size_t len = strlen(XATTR_NAME_POSIX_ACL_DEFAULT);
+		err = ceph_pagelist_reserve(pagelist, len + val_size2 + 8);
+		if (err)
+			goto out_err;
+		ceph_pagelist_encode_string(pagelist,
+					  XATTR_NAME_POSIX_ACL_DEFAULT, len);
+		err = posix_acl_to_xattr(&init_user_ns, default_acl,
+					 tmp_buf, val_size2);
+		if (err < 0)
+			goto out_err;
+		ceph_pagelist_encode_32(pagelist, val_size2);
+		ceph_pagelist_append(pagelist, tmp_buf, val_size2);
+	}
+
+	kfree(tmp_buf);
+
+	as_ctx->acl = acl;
+	as_ctx->default_acl = default_acl;
+	as_ctx->pagelist = pagelist;
+	return 0;
+
+out_err:
+	posix_acl_release(acl);
+	posix_acl_release(default_acl);
+	kfree(tmp_buf);
+	if (pagelist)
+		ceph_pagelist_release(pagelist);
+	return err;
+}
+
+void ceph_init_inode_acls(struct inode *inode, struct ceph_acl_sec_ctx *as_ctx)
+{
+	if (!inode)
+		return;
+	ceph_set_cached_acl(inode, ACL_TYPE_ACCESS, as_ctx->acl);
+	ceph_set_cached_acl(inode, ACL_TYPE_DEFAULT, as_ctx->default_acl);
+}
diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
new file mode 100644
index 000000000..dd2ce7fab
--- /dev/null
+++ b/fs/ceph/addr.c
@@ -0,0 +1,2104 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/ceph/ceph_debug.h>
+
+#include <linux/backing-dev.h>
+#include <linux/fs.h>
+#include <linux/mm.h>
+#include <linux/swap.h>
+#include <linux/pagemap.h>
+#include <linux/slab.h>
+#include <linux/pagevec.h>
+#include <linux/task_io_accounting_ops.h>
+#include <linux/signal.h>
+#include <linux/iversion.h>
+#include <linux/ktime.h>
+#include <linux/netfs.h>
+
+#include "super.h"
+#include "mds_client.h"
+#include "cache.h"
+#include "metric.h"
+#include <linux/ceph/osd_client.h>
+#include <linux/ceph/striper.h>
+
+/*
+ * Ceph address space ops.
+ *
+ * There are a few funny things going on here.
+ *
+ * The page->private field is used to reference a struct
+ * ceph_snap_context for _every_ dirty page.  This indicates which
+ * snapshot the page was logically dirtied in, and thus which snap
+ * context needs to be associated with the osd write during writeback.
+ *
+ * Similarly, struct ceph_inode_info maintains a set of counters to
+ * count dirty pages on the inode.  In the absence of snapshots,
+ * i_wrbuffer_ref == i_wrbuffer_ref_head == the dirty page count.
+ *
+ * When a snapshot is taken (that is, when the client receives
+ * notification that a snapshot was taken), each inode with caps and
+ * with dirty pages (dirty pages implies there is a cap) gets a new
+ * ceph_cap_snap in the i_cap_snaps list (which is sorted in ascending
+ * order, new snaps go to the tail).  The i_wrbuffer_ref_head count is
+ * moved to capsnap->dirty. (Unless a sync write is currently in
+ * progress.  In that case, the capsnap is said to be "pending", new
+ * writes cannot start, and the capsnap isn't "finalized" until the
+ * write completes (or fails) and a final size/mtime for the inode for
+ * that snap can be settled upon.)  i_wrbuffer_ref_head is reset to 0.
+ *
+ * On writeback, we must submit writes to the osd IN SNAP ORDER.  So,
+ * we look for the first capsnap in i_cap_snaps and write out pages in
+ * that snap context _only_.  Then we move on to the next capsnap,
+ * eventually reaching the "live" or "head" context (i.e., pages that
+ * are not yet snapped) and are writing the most recently dirtied
+ * pages.
+ *
+ * Invalidate and so forth must take care to ensure the dirty page
+ * accounting is preserved.
+ */
+
+#define CONGESTION_ON_THRESH(congestion_kb) (congestion_kb >> (PAGE_SHIFT-10))
+#define CONGESTION_OFF_THRESH(congestion_kb)				\
+	(CONGESTION_ON_THRESH(congestion_kb) -				\
+	 (CONGESTION_ON_THRESH(congestion_kb) >> 2))
+
+static int ceph_netfs_check_write_begin(struct file *file, loff_t pos, unsigned int len,
+					struct folio **foliop, void **_fsdata);
+
+static inline struct ceph_snap_context *page_snap_context(struct page *page)
+{
+	if (PagePrivate(page))
+		return (void *)page->private;
+	return NULL;
+}
+
+/*
+ * Dirty a page.  Optimistically adjust accounting, on the assumption
+ * that we won't race with invalidate.  If we do, readjust.
+ */
+static bool ceph_dirty_folio(struct address_space *mapping, struct folio *folio)
+{
+	struct inode *inode;
+	struct ceph_inode_info *ci;
+	struct ceph_snap_context *snapc;
+
+	if (folio_test_dirty(folio)) {
+		dout("%p dirty_folio %p idx %lu -- already dirty\n",
+		     mapping->host, folio, folio->index);
+		VM_BUG_ON_FOLIO(!folio_test_private(folio), folio);
+		return false;
+	}
+
+	inode = mapping->host;
+	ci = ceph_inode(inode);
+
+	/* dirty the head */
+	spin_lock(&ci->i_ceph_lock);
+	BUG_ON(ci->i_wr_ref == 0); // caller should hold Fw reference
+	if (__ceph_have_pending_cap_snap(ci)) {
+		struct ceph_cap_snap *capsnap =
+				list_last_entry(&ci->i_cap_snaps,
+						struct ceph_cap_snap,
+						ci_item);
+		snapc = ceph_get_snap_context(capsnap->context);
+		capsnap->dirty_pages++;
+	} else {
+		BUG_ON(!ci->i_head_snapc);
+		snapc = ceph_get_snap_context(ci->i_head_snapc);
+		++ci->i_wrbuffer_ref_head;
+	}
+	if (ci->i_wrbuffer_ref == 0)
+		ihold(inode);
+	++ci->i_wrbuffer_ref;
+	dout("%p dirty_folio %p idx %lu head %d/%d -> %d/%d "
+	     "snapc %p seq %lld (%d snaps)\n",
+	     mapping->host, folio, folio->index,
+	     ci->i_wrbuffer_ref-1, ci->i_wrbuffer_ref_head-1,
+	     ci->i_wrbuffer_ref, ci->i_wrbuffer_ref_head,
+	     snapc, snapc->seq, snapc->num_snaps);
+	spin_unlock(&ci->i_ceph_lock);
+
+	/*
+	 * Reference snap context in folio->private.  Also set
+	 * PagePrivate so that we get invalidate_folio callback.
+	 */
+	VM_WARN_ON_FOLIO(folio->private, folio);
+	folio_attach_private(folio, snapc);
+
+	return ceph_fscache_dirty_folio(mapping, folio);
+}
+
+/*
+ * If we are truncating the full folio (i.e. offset == 0), adjust the
+ * dirty folio counters appropriately.  Only called if there is private
+ * data on the folio.
+ */
+static void ceph_invalidate_folio(struct folio *folio, size_t offset,
+				size_t length)
+{
+	struct inode *inode;
+	struct ceph_inode_info *ci;
+	struct ceph_snap_context *snapc;
+
+	inode = folio->mapping->host;
+	ci = ceph_inode(inode);
+
+	if (offset != 0 || length != folio_size(folio)) {
+		dout("%p invalidate_folio idx %lu partial dirty page %zu~%zu\n",
+		     inode, folio->index, offset, length);
+		return;
+	}
+
+	WARN_ON(!folio_test_locked(folio));
+	if (folio_test_private(folio)) {
+		dout("%p invalidate_folio idx %lu full dirty page\n",
+		     inode, folio->index);
+
+		snapc = folio_detach_private(folio);
+		ceph_put_wrbuffer_cap_refs(ci, 1, snapc);
+		ceph_put_snap_context(snapc);
+	}
+
+	folio_wait_fscache(folio);
+}
+
+static bool ceph_release_folio(struct folio *folio, gfp_t gfp)
+{
+	struct inode *inode = folio->mapping->host;
+
+	dout("%llx:%llx release_folio idx %lu (%sdirty)\n",
+	     ceph_vinop(inode),
+	     folio->index, folio_test_dirty(folio) ? "" : "not ");
+
+	if (folio_test_private(folio))
+		return false;
+
+	if (folio_test_fscache(folio)) {
+		if (current_is_kswapd() || !(gfp & __GFP_FS))
+			return false;
+		folio_wait_fscache(folio);
+	}
+	ceph_fscache_note_page_release(inode);
+	return true;
+}
+
+static void ceph_netfs_expand_readahead(struct netfs_io_request *rreq)
+{
+	struct inode *inode = rreq->inode;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_file_layout *lo = &ci->i_layout;
+	unsigned long max_pages = inode->i_sb->s_bdi->ra_pages;
+	loff_t end = rreq->start + rreq->len, new_end;
+	struct ceph_netfs_request_data *priv = rreq->netfs_priv;
+	unsigned long max_len;
+	u32 blockoff;
+
+	if (priv) {
+		/* Readahead is disabled by posix_fadvise POSIX_FADV_RANDOM */
+		if (priv->file_ra_disabled)
+			max_pages = 0;
+		else
+			max_pages = priv->file_ra_pages;
+
+	}
+
+	/* Readahead is disabled */
+	if (!max_pages)
+		return;
+
+	max_len = max_pages << PAGE_SHIFT;
+
+	/*
+	 * Try to expand the length forward by rounding up it to the next
+	 * block, but do not exceed the file size, unless the original
+	 * request already exceeds it.
+	 */
+	new_end = min(round_up(end, lo->stripe_unit), rreq->i_size);
+	if (new_end > end && new_end <= rreq->start + max_len)
+		rreq->len = new_end - rreq->start;
+
+	/* Try to expand the start downward */
+	div_u64_rem(rreq->start, lo->stripe_unit, &blockoff);
+	if (rreq->len + blockoff <= max_len) {
+		rreq->start -= blockoff;
+		rreq->len += blockoff;
+	}
+}
+
+static bool ceph_netfs_clamp_length(struct netfs_io_subrequest *subreq)
+{
+	struct inode *inode = subreq->rreq->inode;
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	u64 objno, objoff;
+	u32 xlen;
+
+	/* Truncate the extent at the end of the current block */
+	ceph_calc_file_object_mapping(&ci->i_layout, subreq->start, subreq->len,
+				      &objno, &objoff, &xlen);
+	subreq->len = min(xlen, fsc->mount_options->rsize);
+	return true;
+}
+
+static void finish_netfs_read(struct ceph_osd_request *req)
+{
+	struct ceph_fs_client *fsc = ceph_inode_to_client(req->r_inode);
+	struct ceph_osd_data *osd_data = osd_req_op_extent_osd_data(req, 0);
+	struct netfs_io_subrequest *subreq = req->r_priv;
+	int num_pages;
+	int err = req->r_result;
+
+	ceph_update_read_metrics(&fsc->mdsc->metric, req->r_start_latency,
+				 req->r_end_latency, osd_data->length, err);
+
+	dout("%s: result %d subreq->len=%zu i_size=%lld\n", __func__, req->r_result,
+	     subreq->len, i_size_read(req->r_inode));
+
+	/* no object means success but no data */
+	if (err == -ENOENT)
+		err = 0;
+	else if (err == -EBLOCKLISTED)
+		fsc->blocklisted = true;
+
+	if (err >= 0 && err < subreq->len)
+		__set_bit(NETFS_SREQ_CLEAR_TAIL, &subreq->flags);
+
+	netfs_subreq_terminated(subreq, err, false);
+
+	num_pages = calc_pages_for(osd_data->alignment, osd_data->length);
+	ceph_put_page_vector(osd_data->pages, num_pages, false);
+	iput(req->r_inode);
+}
+
+static bool ceph_netfs_issue_op_inline(struct netfs_io_subrequest *subreq)
+{
+	struct netfs_io_request *rreq = subreq->rreq;
+	struct inode *inode = rreq->inode;
+	struct ceph_mds_reply_info_parsed *rinfo;
+	struct ceph_mds_reply_info_in *iinfo;
+	struct ceph_mds_request *req;
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct iov_iter iter;
+	ssize_t err = 0;
+	size_t len;
+	int mode;
+
+	__set_bit(NETFS_SREQ_CLEAR_TAIL, &subreq->flags);
+	__clear_bit(NETFS_SREQ_COPY_TO_CACHE, &subreq->flags);
+
+	if (subreq->start >= inode->i_size)
+		goto out;
+
+	/* We need to fetch the inline data. */
+	mode = ceph_try_to_choose_auth_mds(inode, CEPH_STAT_CAP_INLINE_DATA);
+	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_GETATTR, mode);
+	if (IS_ERR(req)) {
+		err = PTR_ERR(req);
+		goto out;
+	}
+	req->r_ino1 = ci->i_vino;
+	req->r_args.getattr.mask = cpu_to_le32(CEPH_STAT_CAP_INLINE_DATA);
+	req->r_num_caps = 2;
+
+	err = ceph_mdsc_do_request(mdsc, NULL, req);
+	if (err < 0)
+		goto out;
+
+	rinfo = &req->r_reply_info;
+	iinfo = &rinfo->targeti;
+	if (iinfo->inline_version == CEPH_INLINE_NONE) {
+		/* The data got uninlined */
+		ceph_mdsc_put_request(req);
+		return false;
+	}
+
+	len = min_t(size_t, iinfo->inline_len - subreq->start, subreq->len);
+	iov_iter_xarray(&iter, ITER_DEST, &rreq->mapping->i_pages, subreq->start, len);
+	err = copy_to_iter(iinfo->inline_data + subreq->start, len, &iter);
+	if (err == 0)
+		err = -EFAULT;
+
+	ceph_mdsc_put_request(req);
+out:
+	netfs_subreq_terminated(subreq, err, false);
+	return true;
+}
+
+static void ceph_netfs_issue_read(struct netfs_io_subrequest *subreq)
+{
+	struct netfs_io_request *rreq = subreq->rreq;
+	struct inode *inode = rreq->inode;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	struct ceph_osd_request *req = NULL;
+	struct ceph_vino vino = ceph_vino(inode);
+	struct iov_iter iter;
+	struct page **pages;
+	size_t page_off;
+	int err = 0;
+	u64 len = subreq->len;
+
+	if (ceph_inode_is_shutdown(inode)) {
+		err = -EIO;
+		goto out;
+	}
+
+	if (ceph_has_inline_data(ci) && ceph_netfs_issue_op_inline(subreq))
+		return;
+
+	req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout, vino, subreq->start, &len,
+			0, 1, CEPH_OSD_OP_READ,
+			CEPH_OSD_FLAG_READ | fsc->client->osdc.client->options->read_from_replica,
+			NULL, ci->i_truncate_seq, ci->i_truncate_size, false);
+	if (IS_ERR(req)) {
+		err = PTR_ERR(req);
+		req = NULL;
+		goto out;
+	}
+
+	dout("%s: pos=%llu orig_len=%zu len=%llu\n", __func__, subreq->start, subreq->len, len);
+	iov_iter_xarray(&iter, ITER_DEST, &rreq->mapping->i_pages, subreq->start, len);
+	err = iov_iter_get_pages_alloc2(&iter, &pages, len, &page_off);
+	if (err < 0) {
+		dout("%s: iov_ter_get_pages_alloc returned %d\n", __func__, err);
+		goto out;
+	}
+
+	/* should always give us a page-aligned read */
+	WARN_ON_ONCE(page_off);
+	len = err;
+	err = 0;
+
+	osd_req_op_extent_osd_data_pages(req, 0, pages, len, 0, false, false);
+	req->r_callback = finish_netfs_read;
+	req->r_priv = subreq;
+	req->r_inode = inode;
+	ihold(inode);
+
+	ceph_osdc_start_request(req->r_osdc, req);
+out:
+	ceph_osdc_put_request(req);
+	if (err)
+		netfs_subreq_terminated(subreq, err, false);
+	dout("%s: result %d\n", __func__, err);
+}
+
+static int ceph_init_request(struct netfs_io_request *rreq, struct file *file)
+{
+	struct inode *inode = rreq->inode;
+	int got = 0, want = CEPH_CAP_FILE_CACHE;
+	struct ceph_netfs_request_data *priv;
+	int ret = 0;
+
+	if (rreq->origin != NETFS_READAHEAD)
+		return 0;
+
+	priv = kzalloc(sizeof(*priv), GFP_NOFS);
+	if (!priv)
+		return -ENOMEM;
+
+	if (file) {
+		struct ceph_rw_context *rw_ctx;
+		struct ceph_file_info *fi = file->private_data;
+
+		priv->file_ra_pages = file->f_ra.ra_pages;
+		priv->file_ra_disabled = file->f_mode & FMODE_RANDOM;
+
+		rw_ctx = ceph_find_rw_context(fi);
+		if (rw_ctx) {
+			rreq->netfs_priv = priv;
+			return 0;
+		}
+	}
+
+	/*
+	 * readahead callers do not necessarily hold Fcb caps
+	 * (e.g. fadvise, madvise).
+	 */
+	ret = ceph_try_get_caps(inode, CEPH_CAP_FILE_RD, want, true, &got);
+	if (ret < 0) {
+		dout("start_read %p, error getting cap\n", inode);
+		goto out;
+	}
+
+	if (!(got & want)) {
+		dout("start_read %p, no cache cap\n", inode);
+		ret = -EACCES;
+		goto out;
+	}
+	if (ret == 0) {
+		ret = -EACCES;
+		goto out;
+	}
+
+	priv->caps = got;
+	rreq->netfs_priv = priv;
+
+out:
+	if (ret < 0)
+		kfree(priv);
+
+	return ret;
+}
+
+static void ceph_netfs_free_request(struct netfs_io_request *rreq)
+{
+	struct ceph_netfs_request_data *priv = rreq->netfs_priv;
+
+	if (!priv)
+		return;
+
+	if (priv->caps)
+		ceph_put_cap_refs(ceph_inode(rreq->inode), priv->caps);
+	kfree(priv);
+	rreq->netfs_priv = NULL;
+}
+
+const struct netfs_request_ops ceph_netfs_ops = {
+	.init_request		= ceph_init_request,
+	.free_request		= ceph_netfs_free_request,
+	.begin_cache_operation	= ceph_begin_cache_operation,
+	.issue_read		= ceph_netfs_issue_read,
+	.expand_readahead	= ceph_netfs_expand_readahead,
+	.clamp_length		= ceph_netfs_clamp_length,
+	.check_write_begin	= ceph_netfs_check_write_begin,
+};
+
+#ifdef CONFIG_CEPH_FSCACHE
+static void ceph_set_page_fscache(struct page *page)
+{
+	set_page_fscache(page);
+}
+
+static void ceph_fscache_write_terminated(void *priv, ssize_t error, bool was_async)
+{
+	struct inode *inode = priv;
+
+	if (IS_ERR_VALUE(error) && error != -ENOBUFS)
+		ceph_fscache_invalidate(inode, false);
+}
+
+static void ceph_fscache_write_to_cache(struct inode *inode, u64 off, u64 len, bool caching)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct fscache_cookie *cookie = ceph_fscache_cookie(ci);
+
+	fscache_write_to_cache(cookie, inode->i_mapping, off, len, i_size_read(inode),
+			       ceph_fscache_write_terminated, inode, caching);
+}
+#else
+static inline void ceph_set_page_fscache(struct page *page)
+{
+}
+
+static inline void ceph_fscache_write_to_cache(struct inode *inode, u64 off, u64 len, bool caching)
+{
+}
+#endif /* CONFIG_CEPH_FSCACHE */
+
+struct ceph_writeback_ctl
+{
+	loff_t i_size;
+	u64 truncate_size;
+	u32 truncate_seq;
+	bool size_stable;
+	bool head_snapc;
+};
+
+/*
+ * Get ref for the oldest snapc for an inode with dirty data... that is, the
+ * only snap context we are allowed to write back.
+ */
+static struct ceph_snap_context *
+get_oldest_context(struct inode *inode, struct ceph_writeback_ctl *ctl,
+		   struct ceph_snap_context *page_snapc)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_snap_context *snapc = NULL;
+	struct ceph_cap_snap *capsnap = NULL;
+
+	spin_lock(&ci->i_ceph_lock);
+	list_for_each_entry(capsnap, &ci->i_cap_snaps, ci_item) {
+		dout(" cap_snap %p snapc %p has %d dirty pages\n", capsnap,
+		     capsnap->context, capsnap->dirty_pages);
+		if (!capsnap->dirty_pages)
+			continue;
+
+		/* get i_size, truncate_{seq,size} for page_snapc? */
+		if (snapc && capsnap->context != page_snapc)
+			continue;
+
+		if (ctl) {
+			if (capsnap->writing) {
+				ctl->i_size = i_size_read(inode);
+				ctl->size_stable = false;
+			} else {
+				ctl->i_size = capsnap->size;
+				ctl->size_stable = true;
+			}
+			ctl->truncate_size = capsnap->truncate_size;
+			ctl->truncate_seq = capsnap->truncate_seq;
+			ctl->head_snapc = false;
+		}
+
+		if (snapc)
+			break;
+
+		snapc = ceph_get_snap_context(capsnap->context);
+		if (!page_snapc ||
+		    page_snapc == snapc ||
+		    page_snapc->seq > snapc->seq)
+			break;
+	}
+	if (!snapc && ci->i_wrbuffer_ref_head) {
+		snapc = ceph_get_snap_context(ci->i_head_snapc);
+		dout(" head snapc %p has %d dirty pages\n",
+		     snapc, ci->i_wrbuffer_ref_head);
+		if (ctl) {
+			ctl->i_size = i_size_read(inode);
+			ctl->truncate_size = ci->i_truncate_size;
+			ctl->truncate_seq = ci->i_truncate_seq;
+			ctl->size_stable = false;
+			ctl->head_snapc = true;
+		}
+	}
+	spin_unlock(&ci->i_ceph_lock);
+	return snapc;
+}
+
+static u64 get_writepages_data_length(struct inode *inode,
+				      struct page *page, u64 start)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_snap_context *snapc = page_snap_context(page);
+	struct ceph_cap_snap *capsnap = NULL;
+	u64 end = i_size_read(inode);
+
+	if (snapc != ci->i_head_snapc) {
+		bool found = false;
+		spin_lock(&ci->i_ceph_lock);
+		list_for_each_entry(capsnap, &ci->i_cap_snaps, ci_item) {
+			if (capsnap->context == snapc) {
+				if (!capsnap->writing)
+					end = capsnap->size;
+				found = true;
+				break;
+			}
+		}
+		spin_unlock(&ci->i_ceph_lock);
+		WARN_ON(!found);
+	}
+	if (end > page_offset(page) + thp_size(page))
+		end = page_offset(page) + thp_size(page);
+	return end > start ? end - start : 0;
+}
+
+/*
+ * Write a single page, but leave the page locked.
+ *
+ * If we get a write error, mark the mapping for error, but still adjust the
+ * dirty page accounting (i.e., page is no longer dirty).
+ */
+static int writepage_nounlock(struct page *page, struct writeback_control *wbc)
+{
+	struct folio *folio = page_folio(page);
+	struct inode *inode = page->mapping->host;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	struct ceph_snap_context *snapc, *oldest;
+	loff_t page_off = page_offset(page);
+	int err;
+	loff_t len = thp_size(page);
+	struct ceph_writeback_ctl ceph_wbc;
+	struct ceph_osd_client *osdc = &fsc->client->osdc;
+	struct ceph_osd_request *req;
+	bool caching = ceph_is_cache_enabled(inode);
+
+	dout("writepage %p idx %lu\n", page, page->index);
+
+	if (ceph_inode_is_shutdown(inode))
+		return -EIO;
+
+	/* verify this is a writeable snap context */
+	snapc = page_snap_context(page);
+	if (!snapc) {
+		dout("writepage %p page %p not dirty?\n", inode, page);
+		return 0;
+	}
+	oldest = get_oldest_context(inode, &ceph_wbc, snapc);
+	if (snapc->seq > oldest->seq) {
+		dout("writepage %p page %p snapc %p not writeable - noop\n",
+		     inode, page, snapc);
+		/* we should only noop if called by kswapd */
+		WARN_ON(!(current->flags & PF_MEMALLOC));
+		ceph_put_snap_context(oldest);
+		redirty_page_for_writepage(wbc, page);
+		return 0;
+	}
+	ceph_put_snap_context(oldest);
+
+	/* is this a partial page at end of file? */
+	if (page_off >= ceph_wbc.i_size) {
+		dout("folio at %lu beyond eof %llu\n", folio->index,
+				ceph_wbc.i_size);
+		folio_invalidate(folio, 0, folio_size(folio));
+		return 0;
+	}
+
+	if (ceph_wbc.i_size < page_off + len)
+		len = ceph_wbc.i_size - page_off;
+
+	dout("writepage %p page %p index %lu on %llu~%llu snapc %p seq %lld\n",
+	     inode, page, page->index, page_off, len, snapc, snapc->seq);
+
+	if (atomic_long_inc_return(&fsc->writeback_count) >
+	    CONGESTION_ON_THRESH(fsc->mount_options->congestion_kb))
+		fsc->write_congested = true;
+
+	req = ceph_osdc_new_request(osdc, &ci->i_layout, ceph_vino(inode), page_off, &len, 0, 1,
+				    CEPH_OSD_OP_WRITE, CEPH_OSD_FLAG_WRITE, snapc,
+				    ceph_wbc.truncate_seq, ceph_wbc.truncate_size,
+				    true);
+	if (IS_ERR(req)) {
+		redirty_page_for_writepage(wbc, page);
+		return PTR_ERR(req);
+	}
+
+	set_page_writeback(page);
+	if (caching)
+		ceph_set_page_fscache(page);
+	ceph_fscache_write_to_cache(inode, page_off, len, caching);
+
+	/* it may be a short write due to an object boundary */
+	WARN_ON_ONCE(len > thp_size(page));
+	osd_req_op_extent_osd_data_pages(req, 0, &page, len, 0, false, false);
+	dout("writepage %llu~%llu (%llu bytes)\n", page_off, len, len);
+
+	req->r_mtime = inode->i_mtime;
+	ceph_osdc_start_request(osdc, req);
+	err = ceph_osdc_wait_request(osdc, req);
+
+	ceph_update_write_metrics(&fsc->mdsc->metric, req->r_start_latency,
+				  req->r_end_latency, len, err);
+
+	ceph_osdc_put_request(req);
+	if (err == 0)
+		err = len;
+
+	if (err < 0) {
+		struct writeback_control tmp_wbc;
+		if (!wbc)
+			wbc = &tmp_wbc;
+		if (err == -ERESTARTSYS) {
+			/* killed by SIGKILL */
+			dout("writepage interrupted page %p\n", page);
+			redirty_page_for_writepage(wbc, page);
+			end_page_writeback(page);
+			return err;
+		}
+		if (err == -EBLOCKLISTED)
+			fsc->blocklisted = true;
+		dout("writepage setting page/mapping error %d %p\n",
+		     err, page);
+		mapping_set_error(&inode->i_data, err);
+		wbc->pages_skipped++;
+	} else {
+		dout("writepage cleaned page %p\n", page);
+		err = 0;  /* vfs expects us to return 0 */
+	}
+	oldest = detach_page_private(page);
+	WARN_ON_ONCE(oldest != snapc);
+	end_page_writeback(page);
+	ceph_put_wrbuffer_cap_refs(ci, 1, snapc);
+	ceph_put_snap_context(snapc);  /* page's reference */
+
+	if (atomic_long_dec_return(&fsc->writeback_count) <
+	    CONGESTION_OFF_THRESH(fsc->mount_options->congestion_kb))
+		fsc->write_congested = false;
+
+	return err;
+}
+
+static int ceph_writepage(struct page *page, struct writeback_control *wbc)
+{
+	int err;
+	struct inode *inode = page->mapping->host;
+	BUG_ON(!inode);
+	ihold(inode);
+
+	if (wbc->sync_mode == WB_SYNC_NONE &&
+	    ceph_inode_to_client(inode)->write_congested)
+		return AOP_WRITEPAGE_ACTIVATE;
+
+	wait_on_page_fscache(page);
+
+	err = writepage_nounlock(page, wbc);
+	if (err == -ERESTARTSYS) {
+		/* direct memory reclaimer was killed by SIGKILL. return 0
+		 * to prevent caller from setting mapping/page error */
+		err = 0;
+	}
+	unlock_page(page);
+	iput(inode);
+	return err;
+}
+
+/*
+ * async writeback completion handler.
+ *
+ * If we get an error, set the mapping error bit, but not the individual
+ * page error bits.
+ */
+static void writepages_finish(struct ceph_osd_request *req)
+{
+	struct inode *inode = req->r_inode;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_osd_data *osd_data;
+	struct page *page;
+	int num_pages, total_pages = 0;
+	int i, j;
+	int rc = req->r_result;
+	struct ceph_snap_context *snapc = req->r_snapc;
+	struct address_space *mapping = inode->i_mapping;
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	unsigned int len = 0;
+	bool remove_page;
+
+	dout("writepages_finish %p rc %d\n", inode, rc);
+	if (rc < 0) {
+		mapping_set_error(mapping, rc);
+		ceph_set_error_write(ci);
+		if (rc == -EBLOCKLISTED)
+			fsc->blocklisted = true;
+	} else {
+		ceph_clear_error_write(ci);
+	}
+
+	/*
+	 * We lost the cache cap, need to truncate the page before
+	 * it is unlocked, otherwise we'd truncate it later in the
+	 * page truncation thread, possibly losing some data that
+	 * raced its way in
+	 */
+	remove_page = !(ceph_caps_issued(ci) &
+			(CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO));
+
+	/* clean all pages */
+	for (i = 0; i < req->r_num_ops; i++) {
+		if (req->r_ops[i].op != CEPH_OSD_OP_WRITE) {
+			pr_warn("%s incorrect op %d req %p index %d tid %llu\n",
+				__func__, req->r_ops[i].op, req, i, req->r_tid);
+			break;
+		}
+
+		osd_data = osd_req_op_extent_osd_data(req, i);
+		BUG_ON(osd_data->type != CEPH_OSD_DATA_TYPE_PAGES);
+		len += osd_data->length;
+		num_pages = calc_pages_for((u64)osd_data->alignment,
+					   (u64)osd_data->length);
+		total_pages += num_pages;
+		for (j = 0; j < num_pages; j++) {
+			page = osd_data->pages[j];
+			BUG_ON(!page);
+			WARN_ON(!PageUptodate(page));
+
+			if (atomic_long_dec_return(&fsc->writeback_count) <
+			     CONGESTION_OFF_THRESH(
+					fsc->mount_options->congestion_kb))
+				fsc->write_congested = false;
+
+			ceph_put_snap_context(detach_page_private(page));
+			end_page_writeback(page);
+			dout("unlocking %p\n", page);
+
+			if (remove_page)
+				generic_error_remove_page(inode->i_mapping,
+							  page);
+
+			unlock_page(page);
+		}
+		dout("writepages_finish %p wrote %llu bytes cleaned %d pages\n",
+		     inode, osd_data->length, rc >= 0 ? num_pages : 0);
+
+		release_pages(osd_data->pages, num_pages);
+	}
+
+	ceph_update_write_metrics(&fsc->mdsc->metric, req->r_start_latency,
+				  req->r_end_latency, len, rc);
+
+	ceph_put_wrbuffer_cap_refs(ci, total_pages, snapc);
+
+	osd_data = osd_req_op_extent_osd_data(req, 0);
+	if (osd_data->pages_from_pool)
+		mempool_free(osd_data->pages, ceph_wb_pagevec_pool);
+	else
+		kfree(osd_data->pages);
+	ceph_osdc_put_request(req);
+}
+
+/*
+ * initiate async writeback
+ */
+static int ceph_writepages_start(struct address_space *mapping,
+				 struct writeback_control *wbc)
+{
+	struct inode *inode = mapping->host;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	struct ceph_vino vino = ceph_vino(inode);
+	pgoff_t index, start_index, end = -1;
+	struct ceph_snap_context *snapc = NULL, *last_snapc = NULL, *pgsnapc;
+	struct pagevec pvec;
+	int rc = 0;
+	unsigned int wsize = i_blocksize(inode);
+	struct ceph_osd_request *req = NULL;
+	struct ceph_writeback_ctl ceph_wbc;
+	bool should_loop, range_whole = false;
+	bool done = false;
+	bool caching = ceph_is_cache_enabled(inode);
+
+	if (wbc->sync_mode == WB_SYNC_NONE &&
+	    fsc->write_congested)
+		return 0;
+
+	dout("writepages_start %p (mode=%s)\n", inode,
+	     wbc->sync_mode == WB_SYNC_NONE ? "NONE" :
+	     (wbc->sync_mode == WB_SYNC_ALL ? "ALL" : "HOLD"));
+
+	if (ceph_inode_is_shutdown(inode)) {
+		if (ci->i_wrbuffer_ref > 0) {
+			pr_warn_ratelimited(
+				"writepage_start %p %lld forced umount\n",
+				inode, ceph_ino(inode));
+		}
+		mapping_set_error(mapping, -EIO);
+		return -EIO; /* we're in a forced umount, don't write! */
+	}
+	if (fsc->mount_options->wsize < wsize)
+		wsize = fsc->mount_options->wsize;
+
+	pagevec_init(&pvec);
+
+	start_index = wbc->range_cyclic ? mapping->writeback_index : 0;
+	index = start_index;
+
+retry:
+	/* find oldest snap context with dirty data */
+	snapc = get_oldest_context(inode, &ceph_wbc, NULL);
+	if (!snapc) {
+		/* hmm, why does writepages get called when there
+		   is no dirty data? */
+		dout(" no snap context with dirty data?\n");
+		goto out;
+	}
+	dout(" oldest snapc is %p seq %lld (%d snaps)\n",
+	     snapc, snapc->seq, snapc->num_snaps);
+
+	should_loop = false;
+	if (ceph_wbc.head_snapc && snapc != last_snapc) {
+		/* where to start/end? */
+		if (wbc->range_cyclic) {
+			index = start_index;
+			end = -1;
+			if (index > 0)
+				should_loop = true;
+			dout(" cyclic, start at %lu\n", index);
+		} else {
+			index = wbc->range_start >> PAGE_SHIFT;
+			end = wbc->range_end >> PAGE_SHIFT;
+			if (wbc->range_start == 0 && wbc->range_end == LLONG_MAX)
+				range_whole = true;
+			dout(" not cyclic, %lu to %lu\n", index, end);
+		}
+	} else if (!ceph_wbc.head_snapc) {
+		/* Do not respect wbc->range_{start,end}. Dirty pages
+		 * in that range can be associated with newer snapc.
+		 * They are not writeable until we write all dirty pages
+		 * associated with 'snapc' get written */
+		if (index > 0)
+			should_loop = true;
+		dout(" non-head snapc, range whole\n");
+	}
+
+	ceph_put_snap_context(last_snapc);
+	last_snapc = snapc;
+
+	while (!done && index <= end) {
+		int num_ops = 0, op_idx;
+		unsigned i, pvec_pages, max_pages, locked_pages = 0;
+		struct page **pages = NULL, **data_pages;
+		struct page *page;
+		pgoff_t strip_unit_end = 0;
+		u64 offset = 0, len = 0;
+		bool from_pool = false;
+
+		max_pages = wsize >> PAGE_SHIFT;
+
+get_more_pages:
+		pvec_pages = pagevec_lookup_range_tag(&pvec, mapping, &index,
+						end, PAGECACHE_TAG_DIRTY);
+		dout("pagevec_lookup_range_tag got %d\n", pvec_pages);
+		if (!pvec_pages && !locked_pages)
+			break;
+		for (i = 0; i < pvec_pages && locked_pages < max_pages; i++) {
+			page = pvec.pages[i];
+			dout("? %p idx %lu\n", page, page->index);
+			if (locked_pages == 0)
+				lock_page(page);  /* first page */
+			else if (!trylock_page(page))
+				break;
+
+			/* only dirty pages, or our accounting breaks */
+			if (unlikely(!PageDirty(page)) ||
+			    unlikely(page->mapping != mapping)) {
+				dout("!dirty or !mapping %p\n", page);
+				unlock_page(page);
+				continue;
+			}
+			/* only if matching snap context */
+			pgsnapc = page_snap_context(page);
+			if (pgsnapc != snapc) {
+				dout("page snapc %p %lld != oldest %p %lld\n",
+				     pgsnapc, pgsnapc->seq, snapc, snapc->seq);
+				if (!should_loop &&
+				    !ceph_wbc.head_snapc &&
+				    wbc->sync_mode != WB_SYNC_NONE)
+					should_loop = true;
+				unlock_page(page);
+				continue;
+			}
+			if (page_offset(page) >= ceph_wbc.i_size) {
+				struct folio *folio = page_folio(page);
+
+				dout("folio at %lu beyond eof %llu\n",
+				     folio->index, ceph_wbc.i_size);
+				if ((ceph_wbc.size_stable ||
+				    folio_pos(folio) >= i_size_read(inode)) &&
+				    folio_clear_dirty_for_io(folio))
+					folio_invalidate(folio, 0,
+							folio_size(folio));
+				folio_unlock(folio);
+				continue;
+			}
+			if (strip_unit_end && (page->index > strip_unit_end)) {
+				dout("end of strip unit %p\n", page);
+				unlock_page(page);
+				break;
+			}
+			if (PageWriteback(page) || PageFsCache(page)) {
+				if (wbc->sync_mode == WB_SYNC_NONE) {
+					dout("%p under writeback\n", page);
+					unlock_page(page);
+					continue;
+				}
+				dout("waiting on writeback %p\n", page);
+				wait_on_page_writeback(page);
+				wait_on_page_fscache(page);
+			}
+
+			if (!clear_page_dirty_for_io(page)) {
+				dout("%p !clear_page_dirty_for_io\n", page);
+				unlock_page(page);
+				continue;
+			}
+
+			/*
+			 * We have something to write.  If this is
+			 * the first locked page this time through,
+			 * calculate max possinle write size and
+			 * allocate a page array
+			 */
+			if (locked_pages == 0) {
+				u64 objnum;
+				u64 objoff;
+				u32 xlen;
+
+				/* prepare async write request */
+				offset = (u64)page_offset(page);
+				ceph_calc_file_object_mapping(&ci->i_layout,
+							      offset, wsize,
+							      &objnum, &objoff,
+							      &xlen);
+				len = xlen;
+
+				num_ops = 1;
+				strip_unit_end = page->index +
+					((len - 1) >> PAGE_SHIFT);
+
+				BUG_ON(pages);
+				max_pages = calc_pages_for(0, (u64)len);
+				pages = kmalloc_array(max_pages,
+						      sizeof(*pages),
+						      GFP_NOFS);
+				if (!pages) {
+					from_pool = true;
+					pages = mempool_alloc(ceph_wb_pagevec_pool, GFP_NOFS);
+					BUG_ON(!pages);
+				}
+
+				len = 0;
+			} else if (page->index !=
+				   (offset + len) >> PAGE_SHIFT) {
+				if (num_ops >= (from_pool ?  CEPH_OSD_SLAB_OPS :
+							     CEPH_OSD_MAX_OPS)) {
+					redirty_page_for_writepage(wbc, page);
+					unlock_page(page);
+					break;
+				}
+
+				num_ops++;
+				offset = (u64)page_offset(page);
+				len = 0;
+			}
+
+			/* note position of first page in pvec */
+			dout("%p will write page %p idx %lu\n",
+			     inode, page, page->index);
+
+			if (atomic_long_inc_return(&fsc->writeback_count) >
+			    CONGESTION_ON_THRESH(
+				    fsc->mount_options->congestion_kb))
+				fsc->write_congested = true;
+
+			pages[locked_pages++] = page;
+			pvec.pages[i] = NULL;
+
+			len += thp_size(page);
+		}
+
+		/* did we get anything? */
+		if (!locked_pages)
+			goto release_pvec_pages;
+		if (i) {
+			unsigned j, n = 0;
+			/* shift unused page to beginning of pvec */
+			for (j = 0; j < pvec_pages; j++) {
+				if (!pvec.pages[j])
+					continue;
+				if (n < j)
+					pvec.pages[n] = pvec.pages[j];
+				n++;
+			}
+			pvec.nr = n;
+
+			if (pvec_pages && i == pvec_pages &&
+			    locked_pages < max_pages) {
+				dout("reached end pvec, trying for more\n");
+				pagevec_release(&pvec);
+				goto get_more_pages;
+			}
+		}
+
+new_request:
+		offset = page_offset(pages[0]);
+		len = wsize;
+
+		req = ceph_osdc_new_request(&fsc->client->osdc,
+					&ci->i_layout, vino,
+					offset, &len, 0, num_ops,
+					CEPH_OSD_OP_WRITE, CEPH_OSD_FLAG_WRITE,
+					snapc, ceph_wbc.truncate_seq,
+					ceph_wbc.truncate_size, false);
+		if (IS_ERR(req)) {
+			req = ceph_osdc_new_request(&fsc->client->osdc,
+						&ci->i_layout, vino,
+						offset, &len, 0,
+						min(num_ops,
+						    CEPH_OSD_SLAB_OPS),
+						CEPH_OSD_OP_WRITE,
+						CEPH_OSD_FLAG_WRITE,
+						snapc, ceph_wbc.truncate_seq,
+						ceph_wbc.truncate_size, true);
+			BUG_ON(IS_ERR(req));
+		}
+		BUG_ON(len < page_offset(pages[locked_pages - 1]) +
+			     thp_size(page) - offset);
+
+		req->r_callback = writepages_finish;
+		req->r_inode = inode;
+
+		/* Format the osd request message and submit the write */
+		len = 0;
+		data_pages = pages;
+		op_idx = 0;
+		for (i = 0; i < locked_pages; i++) {
+			u64 cur_offset = page_offset(pages[i]);
+			/*
+			 * Discontinuity in page range? Ceph can handle that by just passing
+			 * multiple extents in the write op.
+			 */
+			if (offset + len != cur_offset) {
+				/* If it's full, stop here */
+				if (op_idx + 1 == req->r_num_ops)
+					break;
+
+				/* Kick off an fscache write with what we have so far. */
+				ceph_fscache_write_to_cache(inode, offset, len, caching);
+
+				/* Start a new extent */
+				osd_req_op_extent_dup_last(req, op_idx,
+							   cur_offset - offset);
+				dout("writepages got pages at %llu~%llu\n",
+				     offset, len);
+				osd_req_op_extent_osd_data_pages(req, op_idx,
+							data_pages, len, 0,
+							from_pool, false);
+				osd_req_op_extent_update(req, op_idx, len);
+
+				len = 0;
+				offset = cur_offset;
+				data_pages = pages + i;
+				op_idx++;
+			}
+
+			set_page_writeback(pages[i]);
+			if (caching)
+				ceph_set_page_fscache(pages[i]);
+			len += thp_size(page);
+		}
+		ceph_fscache_write_to_cache(inode, offset, len, caching);
+
+		if (ceph_wbc.size_stable) {
+			len = min(len, ceph_wbc.i_size - offset);
+		} else if (i == locked_pages) {
+			/* writepages_finish() clears writeback pages
+			 * according to the data length, so make sure
+			 * data length covers all locked pages */
+			u64 min_len = len + 1 - thp_size(page);
+			len = get_writepages_data_length(inode, pages[i - 1],
+							 offset);
+			len = max(len, min_len);
+		}
+		dout("writepages got pages at %llu~%llu\n", offset, len);
+
+		osd_req_op_extent_osd_data_pages(req, op_idx, data_pages, len,
+						 0, from_pool, false);
+		osd_req_op_extent_update(req, op_idx, len);
+
+		BUG_ON(op_idx + 1 != req->r_num_ops);
+
+		from_pool = false;
+		if (i < locked_pages) {
+			BUG_ON(num_ops <= req->r_num_ops);
+			num_ops -= req->r_num_ops;
+			locked_pages -= i;
+
+			/* allocate new pages array for next request */
+			data_pages = pages;
+			pages = kmalloc_array(locked_pages, sizeof(*pages),
+					      GFP_NOFS);
+			if (!pages) {
+				from_pool = true;
+				pages = mempool_alloc(ceph_wb_pagevec_pool, GFP_NOFS);
+				BUG_ON(!pages);
+			}
+			memcpy(pages, data_pages + i,
+			       locked_pages * sizeof(*pages));
+			memset(data_pages + i, 0,
+			       locked_pages * sizeof(*pages));
+		} else {
+			BUG_ON(num_ops != req->r_num_ops);
+			index = pages[i - 1]->index + 1;
+			/* request message now owns the pages array */
+			pages = NULL;
+		}
+
+		req->r_mtime = inode->i_mtime;
+		ceph_osdc_start_request(&fsc->client->osdc, req);
+		req = NULL;
+
+		wbc->nr_to_write -= i;
+		if (pages)
+			goto new_request;
+
+		/*
+		 * We stop writing back only if we are not doing
+		 * integrity sync. In case of integrity sync we have to
+		 * keep going until we have written all the pages
+		 * we tagged for writeback prior to entering this loop.
+		 */
+		if (wbc->nr_to_write <= 0 && wbc->sync_mode == WB_SYNC_NONE)
+			done = true;
+
+release_pvec_pages:
+		dout("pagevec_release on %d pages (%p)\n", (int)pvec.nr,
+		     pvec.nr ? pvec.pages[0] : NULL);
+		pagevec_release(&pvec);
+	}
+
+	if (should_loop && !done) {
+		/* more to do; loop back to beginning of file */
+		dout("writepages looping back to beginning of file\n");
+		end = start_index - 1; /* OK even when start_index == 0 */
+
+		/* to write dirty pages associated with next snapc,
+		 * we need to wait until current writes complete */
+		if (wbc->sync_mode != WB_SYNC_NONE &&
+		    start_index == 0 && /* all dirty pages were checked */
+		    !ceph_wbc.head_snapc) {
+			struct page *page;
+			unsigned i, nr;
+			index = 0;
+			while ((index <= end) &&
+			       (nr = pagevec_lookup_tag(&pvec, mapping, &index,
+						PAGECACHE_TAG_WRITEBACK))) {
+				for (i = 0; i < nr; i++) {
+					page = pvec.pages[i];
+					if (page_snap_context(page) != snapc)
+						continue;
+					wait_on_page_writeback(page);
+				}
+				pagevec_release(&pvec);
+				cond_resched();
+			}
+		}
+
+		start_index = 0;
+		index = 0;
+		goto retry;
+	}
+
+	if (wbc->range_cyclic || (range_whole && wbc->nr_to_write > 0))
+		mapping->writeback_index = index;
+
+out:
+	ceph_osdc_put_request(req);
+	ceph_put_snap_context(last_snapc);
+	dout("writepages dend - startone, rc = %d\n", rc);
+	return rc;
+}
+
+
+
+/*
+ * See if a given @snapc is either writeable, or already written.
+ */
+static int context_is_writeable_or_written(struct inode *inode,
+					   struct ceph_snap_context *snapc)
+{
+	struct ceph_snap_context *oldest = get_oldest_context(inode, NULL, NULL);
+	int ret = !oldest || snapc->seq <= oldest->seq;
+
+	ceph_put_snap_context(oldest);
+	return ret;
+}
+
+/**
+ * ceph_find_incompatible - find an incompatible context and return it
+ * @page: page being dirtied
+ *
+ * We are only allowed to write into/dirty a page if the page is
+ * clean, or already dirty within the same snap context. Returns a
+ * conflicting context if there is one, NULL if there isn't, or a
+ * negative error code on other errors.
+ *
+ * Must be called with page lock held.
+ */
+static struct ceph_snap_context *
+ceph_find_incompatible(struct page *page)
+{
+	struct inode *inode = page->mapping->host;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+
+	if (ceph_inode_is_shutdown(inode)) {
+		dout(" page %p %llx:%llx is shutdown\n", page,
+		     ceph_vinop(inode));
+		return ERR_PTR(-ESTALE);
+	}
+
+	for (;;) {
+		struct ceph_snap_context *snapc, *oldest;
+
+		wait_on_page_writeback(page);
+
+		snapc = page_snap_context(page);
+		if (!snapc || snapc == ci->i_head_snapc)
+			break;
+
+		/*
+		 * this page is already dirty in another (older) snap
+		 * context!  is it writeable now?
+		 */
+		oldest = get_oldest_context(inode, NULL, NULL);
+		if (snapc->seq > oldest->seq) {
+			/* not writeable -- return it for the caller to deal with */
+			ceph_put_snap_context(oldest);
+			dout(" page %p snapc %p not current or oldest\n", page, snapc);
+			return ceph_get_snap_context(snapc);
+		}
+		ceph_put_snap_context(oldest);
+
+		/* yay, writeable, do it now (without dropping page lock) */
+		dout(" page %p snapc %p not current, but oldest\n", page, snapc);
+		if (clear_page_dirty_for_io(page)) {
+			int r = writepage_nounlock(page, NULL);
+			if (r < 0)
+				return ERR_PTR(r);
+		}
+	}
+	return NULL;
+}
+
+static int ceph_netfs_check_write_begin(struct file *file, loff_t pos, unsigned int len,
+					struct folio **foliop, void **_fsdata)
+{
+	struct inode *inode = file_inode(file);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_snap_context *snapc;
+
+	snapc = ceph_find_incompatible(folio_page(*foliop, 0));
+	if (snapc) {
+		int r;
+
+		folio_unlock(*foliop);
+		folio_put(*foliop);
+		*foliop = NULL;
+		if (IS_ERR(snapc))
+			return PTR_ERR(snapc);
+
+		ceph_queue_writeback(inode);
+		r = wait_event_killable(ci->i_cap_wq,
+					context_is_writeable_or_written(inode, snapc));
+		ceph_put_snap_context(snapc);
+		return r == 0 ? -EAGAIN : r;
+	}
+	return 0;
+}
+
+/*
+ * We are only allowed to write into/dirty the page if the page is
+ * clean, or already dirty within the same snap context.
+ */
+static int ceph_write_begin(struct file *file, struct address_space *mapping,
+			    loff_t pos, unsigned len,
+			    struct page **pagep, void **fsdata)
+{
+	struct inode *inode = file_inode(file);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct folio *folio = NULL;
+	int r;
+
+	r = netfs_write_begin(&ci->netfs, file, inode->i_mapping, pos, len, &folio, NULL);
+	if (r < 0)
+		return r;
+
+	folio_wait_fscache(folio);
+	WARN_ON_ONCE(!folio_test_locked(folio));
+	*pagep = &folio->page;
+	return 0;
+}
+
+/*
+ * we don't do anything in here that simple_write_end doesn't do
+ * except adjust dirty page accounting
+ */
+static int ceph_write_end(struct file *file, struct address_space *mapping,
+			  loff_t pos, unsigned len, unsigned copied,
+			  struct page *subpage, void *fsdata)
+{
+	struct folio *folio = page_folio(subpage);
+	struct inode *inode = file_inode(file);
+	bool check_cap = false;
+
+	dout("write_end file %p inode %p folio %p %d~%d (%d)\n", file,
+	     inode, folio, (int)pos, (int)copied, (int)len);
+
+	if (!folio_test_uptodate(folio)) {
+		/* just return that nothing was copied on a short copy */
+		if (copied < len) {
+			copied = 0;
+			goto out;
+		}
+		folio_mark_uptodate(folio);
+	}
+
+	/* did file size increase? */
+	if (pos+copied > i_size_read(inode))
+		check_cap = ceph_inode_set_size(inode, pos+copied);
+
+	folio_mark_dirty(folio);
+
+out:
+	folio_unlock(folio);
+	folio_put(folio);
+
+	if (check_cap)
+		ceph_check_caps(ceph_inode(inode), CHECK_CAPS_AUTHONLY, NULL);
+
+	return copied;
+}
+
+const struct address_space_operations ceph_aops = {
+	.read_folio = netfs_read_folio,
+	.readahead = netfs_readahead,
+	.writepage = ceph_writepage,
+	.writepages = ceph_writepages_start,
+	.write_begin = ceph_write_begin,
+	.write_end = ceph_write_end,
+	.dirty_folio = ceph_dirty_folio,
+	.invalidate_folio = ceph_invalidate_folio,
+	.release_folio = ceph_release_folio,
+	.direct_IO = noop_direct_IO,
+};
+
+static void ceph_block_sigs(sigset_t *oldset)
+{
+	sigset_t mask;
+	siginitsetinv(&mask, sigmask(SIGKILL));
+	sigprocmask(SIG_BLOCK, &mask, oldset);
+}
+
+static void ceph_restore_sigs(sigset_t *oldset)
+{
+	sigprocmask(SIG_SETMASK, oldset, NULL);
+}
+
+/*
+ * vm ops
+ */
+static vm_fault_t ceph_filemap_fault(struct vm_fault *vmf)
+{
+	struct vm_area_struct *vma = vmf->vma;
+	struct inode *inode = file_inode(vma->vm_file);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_file_info *fi = vma->vm_file->private_data;
+	loff_t off = (loff_t)vmf->pgoff << PAGE_SHIFT;
+	int want, got, err;
+	sigset_t oldset;
+	vm_fault_t ret = VM_FAULT_SIGBUS;
+
+	if (ceph_inode_is_shutdown(inode))
+		return ret;
+
+	ceph_block_sigs(&oldset);
+
+	dout("filemap_fault %p %llx.%llx %llu trying to get caps\n",
+	     inode, ceph_vinop(inode), off);
+	if (fi->fmode & CEPH_FILE_MODE_LAZY)
+		want = CEPH_CAP_FILE_CACHE | CEPH_CAP_FILE_LAZYIO;
+	else
+		want = CEPH_CAP_FILE_CACHE;
+
+	got = 0;
+	err = ceph_get_caps(vma->vm_file, CEPH_CAP_FILE_RD, want, -1, &got);
+	if (err < 0)
+		goto out_restore;
+
+	dout("filemap_fault %p %llu got cap refs on %s\n",
+	     inode, off, ceph_cap_string(got));
+
+	if ((got & (CEPH_CAP_FILE_CACHE | CEPH_CAP_FILE_LAZYIO)) ||
+	    !ceph_has_inline_data(ci)) {
+		CEPH_DEFINE_RW_CONTEXT(rw_ctx, got);
+		ceph_add_rw_context(fi, &rw_ctx);
+		ret = filemap_fault(vmf);
+		ceph_del_rw_context(fi, &rw_ctx);
+		dout("filemap_fault %p %llu drop cap refs %s ret %x\n",
+		     inode, off, ceph_cap_string(got), ret);
+	} else
+		err = -EAGAIN;
+
+	ceph_put_cap_refs(ci, got);
+
+	if (err != -EAGAIN)
+		goto out_restore;
+
+	/* read inline data */
+	if (off >= PAGE_SIZE) {
+		/* does not support inline data > PAGE_SIZE */
+		ret = VM_FAULT_SIGBUS;
+	} else {
+		struct address_space *mapping = inode->i_mapping;
+		struct page *page;
+
+		filemap_invalidate_lock_shared(mapping);
+		page = find_or_create_page(mapping, 0,
+				mapping_gfp_constraint(mapping, ~__GFP_FS));
+		if (!page) {
+			ret = VM_FAULT_OOM;
+			goto out_inline;
+		}
+		err = __ceph_do_getattr(inode, page,
+					 CEPH_STAT_CAP_INLINE_DATA, true);
+		if (err < 0 || off >= i_size_read(inode)) {
+			unlock_page(page);
+			put_page(page);
+			ret = vmf_error(err);
+			goto out_inline;
+		}
+		if (err < PAGE_SIZE)
+			zero_user_segment(page, err, PAGE_SIZE);
+		else
+			flush_dcache_page(page);
+		SetPageUptodate(page);
+		vmf->page = page;
+		ret = VM_FAULT_MAJOR | VM_FAULT_LOCKED;
+out_inline:
+		filemap_invalidate_unlock_shared(mapping);
+		dout("filemap_fault %p %llu read inline data ret %x\n",
+		     inode, off, ret);
+	}
+out_restore:
+	ceph_restore_sigs(&oldset);
+	if (err < 0)
+		ret = vmf_error(err);
+
+	return ret;
+}
+
+static vm_fault_t ceph_page_mkwrite(struct vm_fault *vmf)
+{
+	struct vm_area_struct *vma = vmf->vma;
+	struct inode *inode = file_inode(vma->vm_file);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_file_info *fi = vma->vm_file->private_data;
+	struct ceph_cap_flush *prealloc_cf;
+	struct page *page = vmf->page;
+	loff_t off = page_offset(page);
+	loff_t size = i_size_read(inode);
+	size_t len;
+	int want, got, err;
+	sigset_t oldset;
+	vm_fault_t ret = VM_FAULT_SIGBUS;
+
+	if (ceph_inode_is_shutdown(inode))
+		return ret;
+
+	prealloc_cf = ceph_alloc_cap_flush();
+	if (!prealloc_cf)
+		return VM_FAULT_OOM;
+
+	sb_start_pagefault(inode->i_sb);
+	ceph_block_sigs(&oldset);
+
+	if (off + thp_size(page) <= size)
+		len = thp_size(page);
+	else
+		len = offset_in_thp(page, size);
+
+	dout("page_mkwrite %p %llx.%llx %llu~%zd getting caps i_size %llu\n",
+	     inode, ceph_vinop(inode), off, len, size);
+	if (fi->fmode & CEPH_FILE_MODE_LAZY)
+		want = CEPH_CAP_FILE_BUFFER | CEPH_CAP_FILE_LAZYIO;
+	else
+		want = CEPH_CAP_FILE_BUFFER;
+
+	got = 0;
+	err = ceph_get_caps(vma->vm_file, CEPH_CAP_FILE_WR, want, off + len, &got);
+	if (err < 0)
+		goto out_free;
+
+	dout("page_mkwrite %p %llu~%zd got cap refs on %s\n",
+	     inode, off, len, ceph_cap_string(got));
+
+	/* Update time before taking page lock */
+	file_update_time(vma->vm_file);
+	inode_inc_iversion_raw(inode);
+
+	do {
+		struct ceph_snap_context *snapc;
+
+		lock_page(page);
+
+		if (page_mkwrite_check_truncate(page, inode) < 0) {
+			unlock_page(page);
+			ret = VM_FAULT_NOPAGE;
+			break;
+		}
+
+		snapc = ceph_find_incompatible(page);
+		if (!snapc) {
+			/* success.  we'll keep the page locked. */
+			set_page_dirty(page);
+			ret = VM_FAULT_LOCKED;
+			break;
+		}
+
+		unlock_page(page);
+
+		if (IS_ERR(snapc)) {
+			ret = VM_FAULT_SIGBUS;
+			break;
+		}
+
+		ceph_queue_writeback(inode);
+		err = wait_event_killable(ci->i_cap_wq,
+				context_is_writeable_or_written(inode, snapc));
+		ceph_put_snap_context(snapc);
+	} while (err == 0);
+
+	if (ret == VM_FAULT_LOCKED) {
+		int dirty;
+		spin_lock(&ci->i_ceph_lock);
+		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
+					       &prealloc_cf);
+		spin_unlock(&ci->i_ceph_lock);
+		if (dirty)
+			__mark_inode_dirty(inode, dirty);
+	}
+
+	dout("page_mkwrite %p %llu~%zd dropping cap refs on %s ret %x\n",
+	     inode, off, len, ceph_cap_string(got), ret);
+	ceph_put_cap_refs_async(ci, got);
+out_free:
+	ceph_restore_sigs(&oldset);
+	sb_end_pagefault(inode->i_sb);
+	ceph_free_cap_flush(prealloc_cf);
+	if (err < 0)
+		ret = vmf_error(err);
+	return ret;
+}
+
+void ceph_fill_inline_data(struct inode *inode, struct page *locked_page,
+			   char	*data, size_t len)
+{
+	struct address_space *mapping = inode->i_mapping;
+	struct page *page;
+
+	if (locked_page) {
+		page = locked_page;
+	} else {
+		if (i_size_read(inode) == 0)
+			return;
+		page = find_or_create_page(mapping, 0,
+					   mapping_gfp_constraint(mapping,
+					   ~__GFP_FS));
+		if (!page)
+			return;
+		if (PageUptodate(page)) {
+			unlock_page(page);
+			put_page(page);
+			return;
+		}
+	}
+
+	dout("fill_inline_data %p %llx.%llx len %zu locked_page %p\n",
+	     inode, ceph_vinop(inode), len, locked_page);
+
+	if (len > 0) {
+		void *kaddr = kmap_atomic(page);
+		memcpy(kaddr, data, len);
+		kunmap_atomic(kaddr);
+	}
+
+	if (page != locked_page) {
+		if (len < PAGE_SIZE)
+			zero_user_segment(page, len, PAGE_SIZE);
+		else
+			flush_dcache_page(page);
+
+		SetPageUptodate(page);
+		unlock_page(page);
+		put_page(page);
+	}
+}
+
+int ceph_uninline_data(struct file *file)
+{
+	struct inode *inode = file_inode(file);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	struct ceph_osd_request *req = NULL;
+	struct ceph_cap_flush *prealloc_cf = NULL;
+	struct folio *folio = NULL;
+	u64 inline_version = CEPH_INLINE_NONE;
+	struct page *pages[1];
+	int err = 0;
+	u64 len;
+
+	spin_lock(&ci->i_ceph_lock);
+	inline_version = ci->i_inline_version;
+	spin_unlock(&ci->i_ceph_lock);
+
+	dout("uninline_data %p %llx.%llx inline_version %llu\n",
+	     inode, ceph_vinop(inode), inline_version);
+
+	if (ceph_inode_is_shutdown(inode)) {
+		err = -EIO;
+		goto out;
+	}
+
+	if (inline_version == CEPH_INLINE_NONE)
+		return 0;
+
+	prealloc_cf = ceph_alloc_cap_flush();
+	if (!prealloc_cf)
+		return -ENOMEM;
+
+	if (inline_version == 1) /* initial version, no data */
+		goto out_uninline;
+
+	folio = read_mapping_folio(inode->i_mapping, 0, file);
+	if (IS_ERR(folio)) {
+		err = PTR_ERR(folio);
+		goto out;
+	}
+
+	folio_lock(folio);
+
+	len = i_size_read(inode);
+	if (len > folio_size(folio))
+		len = folio_size(folio);
+
+	req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
+				    ceph_vino(inode), 0, &len, 0, 1,
+				    CEPH_OSD_OP_CREATE, CEPH_OSD_FLAG_WRITE,
+				    NULL, 0, 0, false);
+	if (IS_ERR(req)) {
+		err = PTR_ERR(req);
+		goto out_unlock;
+	}
+
+	req->r_mtime = inode->i_mtime;
+	ceph_osdc_start_request(&fsc->client->osdc, req);
+	err = ceph_osdc_wait_request(&fsc->client->osdc, req);
+	ceph_osdc_put_request(req);
+	if (err < 0)
+		goto out_unlock;
+
+	req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
+				    ceph_vino(inode), 0, &len, 1, 3,
+				    CEPH_OSD_OP_WRITE, CEPH_OSD_FLAG_WRITE,
+				    NULL, ci->i_truncate_seq,
+				    ci->i_truncate_size, false);
+	if (IS_ERR(req)) {
+		err = PTR_ERR(req);
+		goto out_unlock;
+	}
+
+	pages[0] = folio_page(folio, 0);
+	osd_req_op_extent_osd_data_pages(req, 1, pages, len, 0, false, false);
+
+	{
+		__le64 xattr_buf = cpu_to_le64(inline_version);
+		err = osd_req_op_xattr_init(req, 0, CEPH_OSD_OP_CMPXATTR,
+					    "inline_version", &xattr_buf,
+					    sizeof(xattr_buf),
+					    CEPH_OSD_CMPXATTR_OP_GT,
+					    CEPH_OSD_CMPXATTR_MODE_U64);
+		if (err)
+			goto out_put_req;
+	}
+
+	{
+		char xattr_buf[32];
+		int xattr_len = snprintf(xattr_buf, sizeof(xattr_buf),
+					 "%llu", inline_version);
+		err = osd_req_op_xattr_init(req, 2, CEPH_OSD_OP_SETXATTR,
+					    "inline_version",
+					    xattr_buf, xattr_len, 0, 0);
+		if (err)
+			goto out_put_req;
+	}
+
+	req->r_mtime = inode->i_mtime;
+	ceph_osdc_start_request(&fsc->client->osdc, req);
+	err = ceph_osdc_wait_request(&fsc->client->osdc, req);
+
+	ceph_update_write_metrics(&fsc->mdsc->metric, req->r_start_latency,
+				  req->r_end_latency, len, err);
+
+out_uninline:
+	if (!err) {
+		int dirty;
+
+		/* Set to CAP_INLINE_NONE and dirty the caps */
+		down_read(&fsc->mdsc->snap_rwsem);
+		spin_lock(&ci->i_ceph_lock);
+		ci->i_inline_version = CEPH_INLINE_NONE;
+		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR, &prealloc_cf);
+		spin_unlock(&ci->i_ceph_lock);
+		up_read(&fsc->mdsc->snap_rwsem);
+		if (dirty)
+			__mark_inode_dirty(inode, dirty);
+	}
+out_put_req:
+	ceph_osdc_put_request(req);
+	if (err == -ECANCELED)
+		err = 0;
+out_unlock:
+	if (folio) {
+		folio_unlock(folio);
+		folio_put(folio);
+	}
+out:
+	ceph_free_cap_flush(prealloc_cf);
+	dout("uninline_data %p %llx.%llx inline_version %llu = %d\n",
+	     inode, ceph_vinop(inode), inline_version, err);
+	return err;
+}
+
+static const struct vm_operations_struct ceph_vmops = {
+	.fault		= ceph_filemap_fault,
+	.page_mkwrite	= ceph_page_mkwrite,
+};
+
+int ceph_mmap(struct file *file, struct vm_area_struct *vma)
+{
+	struct address_space *mapping = file->f_mapping;
+
+	if (!mapping->a_ops->read_folio)
+		return -ENOEXEC;
+	vma->vm_ops = &ceph_vmops;
+	return 0;
+}
+
+enum {
+	POOL_READ	= 1,
+	POOL_WRITE	= 2,
+};
+
+static int __ceph_pool_perm_get(struct ceph_inode_info *ci,
+				s64 pool, struct ceph_string *pool_ns)
+{
+	struct ceph_fs_client *fsc = ceph_inode_to_client(&ci->netfs.inode);
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	struct ceph_osd_request *rd_req = NULL, *wr_req = NULL;
+	struct rb_node **p, *parent;
+	struct ceph_pool_perm *perm;
+	struct page **pages;
+	size_t pool_ns_len;
+	int err = 0, err2 = 0, have = 0;
+
+	down_read(&mdsc->pool_perm_rwsem);
+	p = &mdsc->pool_perm_tree.rb_node;
+	while (*p) {
+		perm = rb_entry(*p, struct ceph_pool_perm, node);
+		if (pool < perm->pool)
+			p = &(*p)->rb_left;
+		else if (pool > perm->pool)
+			p = &(*p)->rb_right;
+		else {
+			int ret = ceph_compare_string(pool_ns,
+						perm->pool_ns,
+						perm->pool_ns_len);
+			if (ret < 0)
+				p = &(*p)->rb_left;
+			else if (ret > 0)
+				p = &(*p)->rb_right;
+			else {
+				have = perm->perm;
+				break;
+			}
+		}
+	}
+	up_read(&mdsc->pool_perm_rwsem);
+	if (*p)
+		goto out;
+
+	if (pool_ns)
+		dout("__ceph_pool_perm_get pool %lld ns %.*s no perm cached\n",
+		     pool, (int)pool_ns->len, pool_ns->str);
+	else
+		dout("__ceph_pool_perm_get pool %lld no perm cached\n", pool);
+
+	down_write(&mdsc->pool_perm_rwsem);
+	p = &mdsc->pool_perm_tree.rb_node;
+	parent = NULL;
+	while (*p) {
+		parent = *p;
+		perm = rb_entry(parent, struct ceph_pool_perm, node);
+		if (pool < perm->pool)
+			p = &(*p)->rb_left;
+		else if (pool > perm->pool)
+			p = &(*p)->rb_right;
+		else {
+			int ret = ceph_compare_string(pool_ns,
+						perm->pool_ns,
+						perm->pool_ns_len);
+			if (ret < 0)
+				p = &(*p)->rb_left;
+			else if (ret > 0)
+				p = &(*p)->rb_right;
+			else {
+				have = perm->perm;
+				break;
+			}
+		}
+	}
+	if (*p) {
+		up_write(&mdsc->pool_perm_rwsem);
+		goto out;
+	}
+
+	rd_req = ceph_osdc_alloc_request(&fsc->client->osdc, NULL,
+					 1, false, GFP_NOFS);
+	if (!rd_req) {
+		err = -ENOMEM;
+		goto out_unlock;
+	}
+
+	rd_req->r_flags = CEPH_OSD_FLAG_READ;
+	osd_req_op_init(rd_req, 0, CEPH_OSD_OP_STAT, 0);
+	rd_req->r_base_oloc.pool = pool;
+	if (pool_ns)
+		rd_req->r_base_oloc.pool_ns = ceph_get_string(pool_ns);
+	ceph_oid_printf(&rd_req->r_base_oid, "%llx.00000000", ci->i_vino.ino);
+
+	err = ceph_osdc_alloc_messages(rd_req, GFP_NOFS);
+	if (err)
+		goto out_unlock;
+
+	wr_req = ceph_osdc_alloc_request(&fsc->client->osdc, NULL,
+					 1, false, GFP_NOFS);
+	if (!wr_req) {
+		err = -ENOMEM;
+		goto out_unlock;
+	}
+
+	wr_req->r_flags = CEPH_OSD_FLAG_WRITE;
+	osd_req_op_init(wr_req, 0, CEPH_OSD_OP_CREATE, CEPH_OSD_OP_FLAG_EXCL);
+	ceph_oloc_copy(&wr_req->r_base_oloc, &rd_req->r_base_oloc);
+	ceph_oid_copy(&wr_req->r_base_oid, &rd_req->r_base_oid);
+
+	err = ceph_osdc_alloc_messages(wr_req, GFP_NOFS);
+	if (err)
+		goto out_unlock;
+
+	/* one page should be large enough for STAT data */
+	pages = ceph_alloc_page_vector(1, GFP_KERNEL);
+	if (IS_ERR(pages)) {
+		err = PTR_ERR(pages);
+		goto out_unlock;
+	}
+
+	osd_req_op_raw_data_in_pages(rd_req, 0, pages, PAGE_SIZE,
+				     0, false, true);
+	ceph_osdc_start_request(&fsc->client->osdc, rd_req);
+
+	wr_req->r_mtime = ci->netfs.inode.i_mtime;
+	ceph_osdc_start_request(&fsc->client->osdc, wr_req);
+
+	err = ceph_osdc_wait_request(&fsc->client->osdc, rd_req);
+	err2 = ceph_osdc_wait_request(&fsc->client->osdc, wr_req);
+
+	if (err >= 0 || err == -ENOENT)
+		have |= POOL_READ;
+	else if (err != -EPERM) {
+		if (err == -EBLOCKLISTED)
+			fsc->blocklisted = true;
+		goto out_unlock;
+	}
+
+	if (err2 == 0 || err2 == -EEXIST)
+		have |= POOL_WRITE;
+	else if (err2 != -EPERM) {
+		if (err2 == -EBLOCKLISTED)
+			fsc->blocklisted = true;
+		err = err2;
+		goto out_unlock;
+	}
+
+	pool_ns_len = pool_ns ? pool_ns->len : 0;
+	perm = kmalloc(sizeof(*perm) + pool_ns_len + 1, GFP_NOFS);
+	if (!perm) {
+		err = -ENOMEM;
+		goto out_unlock;
+	}
+
+	perm->pool = pool;
+	perm->perm = have;
+	perm->pool_ns_len = pool_ns_len;
+	if (pool_ns_len > 0)
+		memcpy(perm->pool_ns, pool_ns->str, pool_ns_len);
+	perm->pool_ns[pool_ns_len] = 0;
+
+	rb_link_node(&perm->node, parent, p);
+	rb_insert_color(&perm->node, &mdsc->pool_perm_tree);
+	err = 0;
+out_unlock:
+	up_write(&mdsc->pool_perm_rwsem);
+
+	ceph_osdc_put_request(rd_req);
+	ceph_osdc_put_request(wr_req);
+out:
+	if (!err)
+		err = have;
+	if (pool_ns)
+		dout("__ceph_pool_perm_get pool %lld ns %.*s result = %d\n",
+		     pool, (int)pool_ns->len, pool_ns->str, err);
+	else
+		dout("__ceph_pool_perm_get pool %lld result = %d\n", pool, err);
+	return err;
+}
+
+int ceph_pool_perm_check(struct inode *inode, int need)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_string *pool_ns;
+	s64 pool;
+	int ret, flags;
+
+	/* Only need to do this for regular files */
+	if (!S_ISREG(inode->i_mode))
+		return 0;
+
+	if (ci->i_vino.snap != CEPH_NOSNAP) {
+		/*
+		 * Pool permission check needs to write to the first object.
+		 * But for snapshot, head of the first object may have alread
+		 * been deleted. Skip check to avoid creating orphan object.
+		 */
+		return 0;
+	}
+
+	if (ceph_test_mount_opt(ceph_inode_to_client(inode),
+				NOPOOLPERM))
+		return 0;
+
+	spin_lock(&ci->i_ceph_lock);
+	flags = ci->i_ceph_flags;
+	pool = ci->i_layout.pool_id;
+	spin_unlock(&ci->i_ceph_lock);
+check:
+	if (flags & CEPH_I_POOL_PERM) {
+		if ((need & CEPH_CAP_FILE_RD) && !(flags & CEPH_I_POOL_RD)) {
+			dout("ceph_pool_perm_check pool %lld no read perm\n",
+			     pool);
+			return -EPERM;
+		}
+		if ((need & CEPH_CAP_FILE_WR) && !(flags & CEPH_I_POOL_WR)) {
+			dout("ceph_pool_perm_check pool %lld no write perm\n",
+			     pool);
+			return -EPERM;
+		}
+		return 0;
+	}
+
+	pool_ns = ceph_try_get_string(ci->i_layout.pool_ns);
+	ret = __ceph_pool_perm_get(ci, pool, pool_ns);
+	ceph_put_string(pool_ns);
+	if (ret < 0)
+		return ret;
+
+	flags = CEPH_I_POOL_PERM;
+	if (ret & POOL_READ)
+		flags |= CEPH_I_POOL_RD;
+	if (ret & POOL_WRITE)
+		flags |= CEPH_I_POOL_WR;
+
+	spin_lock(&ci->i_ceph_lock);
+	if (pool == ci->i_layout.pool_id &&
+	    pool_ns == rcu_dereference_raw(ci->i_layout.pool_ns)) {
+		ci->i_ceph_flags |= flags;
+        } else {
+		pool = ci->i_layout.pool_id;
+		flags = ci->i_ceph_flags;
+	}
+	spin_unlock(&ci->i_ceph_lock);
+	goto check;
+}
+
+void ceph_pool_perm_destroy(struct ceph_mds_client *mdsc)
+{
+	struct ceph_pool_perm *perm;
+	struct rb_node *n;
+
+	while (!RB_EMPTY_ROOT(&mdsc->pool_perm_tree)) {
+		n = rb_first(&mdsc->pool_perm_tree);
+		perm = rb_entry(n, struct ceph_pool_perm, node);
+		rb_erase(n, &mdsc->pool_perm_tree);
+		kfree(perm);
+	}
+}
diff --git a/fs/ceph/cache.c b/fs/ceph/cache.c
new file mode 100644
index 000000000..de1dee46d
--- /dev/null
+++ b/fs/ceph/cache.c
@@ -0,0 +1,112 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Ceph cache definitions.
+ *
+ *  Copyright (C) 2013 by Adfin Solutions, Inc. All Rights Reserved.
+ *  Written by Milosz Tanski (milosz@adfin.com)
+ */
+
+#include <linux/ceph/ceph_debug.h>
+
+#include <linux/fs_context.h>
+#include "super.h"
+#include "cache.h"
+
+void ceph_fscache_register_inode_cookie(struct inode *inode)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+
+	/* No caching for filesystem? */
+	if (!fsc->fscache)
+		return;
+
+	/* Regular files only */
+	if (!S_ISREG(inode->i_mode))
+		return;
+
+	/* Only new inodes! */
+	if (!(inode->i_state & I_NEW))
+		return;
+
+	WARN_ON_ONCE(ci->netfs.cache);
+
+	ci->netfs.cache =
+		fscache_acquire_cookie(fsc->fscache, 0,
+				       &ci->i_vino, sizeof(ci->i_vino),
+				       &ci->i_version, sizeof(ci->i_version),
+				       i_size_read(inode));
+	if (ci->netfs.cache)
+		mapping_set_release_always(inode->i_mapping);
+}
+
+void ceph_fscache_unregister_inode_cookie(struct ceph_inode_info *ci)
+{
+	fscache_relinquish_cookie(ceph_fscache_cookie(ci), false);
+}
+
+void ceph_fscache_use_cookie(struct inode *inode, bool will_modify)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+
+	fscache_use_cookie(ceph_fscache_cookie(ci), will_modify);
+}
+
+void ceph_fscache_unuse_cookie(struct inode *inode, bool update)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+
+	if (update) {
+		loff_t i_size = i_size_read(inode);
+
+		fscache_unuse_cookie(ceph_fscache_cookie(ci),
+				     &ci->i_version, &i_size);
+	} else {
+		fscache_unuse_cookie(ceph_fscache_cookie(ci), NULL, NULL);
+	}
+}
+
+void ceph_fscache_update(struct inode *inode)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	loff_t i_size = i_size_read(inode);
+
+	fscache_update_cookie(ceph_fscache_cookie(ci), &ci->i_version, &i_size);
+}
+
+void ceph_fscache_invalidate(struct inode *inode, bool dio_write)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+
+	fscache_invalidate(ceph_fscache_cookie(ci),
+			   &ci->i_version, i_size_read(inode),
+			   dio_write ? FSCACHE_INVAL_DIO_WRITE : 0);
+}
+
+int ceph_fscache_register_fs(struct ceph_fs_client* fsc, struct fs_context *fc)
+{
+	const struct ceph_fsid *fsid = &fsc->client->fsid;
+	const char *fscache_uniq = fsc->mount_options->fscache_uniq;
+	size_t uniq_len = fscache_uniq ? strlen(fscache_uniq) : 0;
+	char *name;
+	int err = 0;
+
+	name = kasprintf(GFP_KERNEL, "ceph,%pU%s%s", fsid, uniq_len ? "," : "",
+			 uniq_len ? fscache_uniq : "");
+	if (!name)
+		return -ENOMEM;
+
+	fsc->fscache = fscache_acquire_volume(name, NULL, NULL, 0);
+	if (IS_ERR_OR_NULL(fsc->fscache)) {
+		errorfc(fc, "Unable to register fscache cookie for %s", name);
+		err = fsc->fscache ? PTR_ERR(fsc->fscache) : -EOPNOTSUPP;
+		fsc->fscache = NULL;
+	}
+	kfree(name);
+	return err;
+}
+
+void ceph_fscache_unregister_fs(struct ceph_fs_client* fsc)
+{
+	fscache_relinquish_volume(fsc->fscache, NULL, false);
+}
diff --git a/fs/ceph/cache.h b/fs/ceph/cache.h
new file mode 100644
index 000000000..dc502daac
--- /dev/null
+++ b/fs/ceph/cache.h
@@ -0,0 +1,148 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Ceph cache definitions.
+ *
+ *  Copyright (C) 2013 by Adfin Solutions, Inc. All Rights Reserved.
+ *  Written by Milosz Tanski (milosz@adfin.com)
+ */
+
+#ifndef _CEPH_CACHE_H
+#define _CEPH_CACHE_H
+
+#include <linux/netfs.h>
+
+#ifdef CONFIG_CEPH_FSCACHE
+#include <linux/fscache.h>
+
+int ceph_fscache_register_fs(struct ceph_fs_client* fsc, struct fs_context *fc);
+void ceph_fscache_unregister_fs(struct ceph_fs_client* fsc);
+
+void ceph_fscache_register_inode_cookie(struct inode *inode);
+void ceph_fscache_unregister_inode_cookie(struct ceph_inode_info* ci);
+
+void ceph_fscache_use_cookie(struct inode *inode, bool will_modify);
+void ceph_fscache_unuse_cookie(struct inode *inode, bool update);
+
+void ceph_fscache_update(struct inode *inode);
+void ceph_fscache_invalidate(struct inode *inode, bool dio_write);
+
+static inline struct fscache_cookie *ceph_fscache_cookie(struct ceph_inode_info *ci)
+{
+	return netfs_i_cookie(&ci->netfs);
+}
+
+static inline void ceph_fscache_resize(struct inode *inode, loff_t to)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct fscache_cookie *cookie = ceph_fscache_cookie(ci);
+
+	if (cookie) {
+		ceph_fscache_use_cookie(inode, true);
+		fscache_resize_cookie(cookie, to);
+		ceph_fscache_unuse_cookie(inode, true);
+	}
+}
+
+static inline void ceph_fscache_unpin_writeback(struct inode *inode,
+						struct writeback_control *wbc)
+{
+	fscache_unpin_writeback(wbc, ceph_fscache_cookie(ceph_inode(inode)));
+}
+
+static inline int ceph_fscache_dirty_folio(struct address_space *mapping,
+		struct folio *folio)
+{
+	struct ceph_inode_info *ci = ceph_inode(mapping->host);
+
+	return fscache_dirty_folio(mapping, folio, ceph_fscache_cookie(ci));
+}
+
+static inline int ceph_begin_cache_operation(struct netfs_io_request *rreq)
+{
+	struct fscache_cookie *cookie = ceph_fscache_cookie(ceph_inode(rreq->inode));
+
+	return fscache_begin_read_operation(&rreq->cache_resources, cookie);
+}
+
+static inline bool ceph_is_cache_enabled(struct inode *inode)
+{
+	return fscache_cookie_enabled(ceph_fscache_cookie(ceph_inode(inode)));
+}
+
+static inline void ceph_fscache_note_page_release(struct inode *inode)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+
+	fscache_note_page_release(ceph_fscache_cookie(ci));
+}
+#else /* CONFIG_CEPH_FSCACHE */
+static inline int ceph_fscache_register_fs(struct ceph_fs_client* fsc,
+					   struct fs_context *fc)
+{
+	return 0;
+}
+
+static inline void ceph_fscache_unregister_fs(struct ceph_fs_client* fsc)
+{
+}
+
+static inline void ceph_fscache_register_inode_cookie(struct inode *inode)
+{
+}
+
+static inline void ceph_fscache_unregister_inode_cookie(struct ceph_inode_info* ci)
+{
+}
+
+static inline void ceph_fscache_use_cookie(struct inode *inode, bool will_modify)
+{
+}
+
+static inline void ceph_fscache_unuse_cookie(struct inode *inode, bool update)
+{
+}
+
+static inline void ceph_fscache_update(struct inode *inode)
+{
+}
+
+static inline void ceph_fscache_invalidate(struct inode *inode, bool dio_write)
+{
+}
+
+static inline struct fscache_cookie *ceph_fscache_cookie(struct ceph_inode_info *ci)
+{
+	return NULL;
+}
+
+static inline void ceph_fscache_resize(struct inode *inode, loff_t to)
+{
+}
+
+static inline void ceph_fscache_unpin_writeback(struct inode *inode,
+						struct writeback_control *wbc)
+{
+}
+
+static inline int ceph_fscache_dirty_folio(struct address_space *mapping,
+		struct folio *folio)
+{
+	return filemap_dirty_folio(mapping, folio);
+}
+
+static inline bool ceph_is_cache_enabled(struct inode *inode)
+{
+	return false;
+}
+
+static inline int ceph_begin_cache_operation(struct netfs_io_request *rreq)
+{
+	return -ENOBUFS;
+}
+
+static inline void ceph_fscache_note_page_release(struct inode *inode)
+{
+}
+#endif /* CONFIG_CEPH_FSCACHE */
+
+#endif
diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
new file mode 100644
index 000000000..36052a362
--- /dev/null
+++ b/fs/ceph/caps.c
@@ -0,0 +1,4752 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/ceph/ceph_debug.h>
+
+#include <linux/fs.h>
+#include <linux/kernel.h>
+#include <linux/sched/signal.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+#include <linux/wait.h>
+#include <linux/writeback.h>
+#include <linux/iversion.h>
+
+#include "super.h"
+#include "mds_client.h"
+#include "cache.h"
+#include <linux/ceph/decode.h>
+#include <linux/ceph/messenger.h>
+
+/*
+ * Capability management
+ *
+ * The Ceph metadata servers control client access to inode metadata
+ * and file data by issuing capabilities, granting clients permission
+ * to read and/or write both inode field and file data to OSDs
+ * (storage nodes).  Each capability consists of a set of bits
+ * indicating which operations are allowed.
+ *
+ * If the client holds a *_SHARED cap, the client has a coherent value
+ * that can be safely read from the cached inode.
+ *
+ * In the case of a *_EXCL (exclusive) or FILE_WR capabilities, the
+ * client is allowed to change inode attributes (e.g., file size,
+ * mtime), note its dirty state in the ceph_cap, and asynchronously
+ * flush that metadata change to the MDS.
+ *
+ * In the event of a conflicting operation (perhaps by another
+ * client), the MDS will revoke the conflicting client capabilities.
+ *
+ * In order for a client to cache an inode, it must hold a capability
+ * with at least one MDS server.  When inodes are released, release
+ * notifications are batched and periodically sent en masse to the MDS
+ * cluster to release server state.
+ */
+
+static u64 __get_oldest_flush_tid(struct ceph_mds_client *mdsc);
+static void __kick_flushing_caps(struct ceph_mds_client *mdsc,
+				 struct ceph_mds_session *session,
+				 struct ceph_inode_info *ci,
+				 u64 oldest_flush_tid);
+
+/*
+ * Generate readable cap strings for debugging output.
+ */
+#define MAX_CAP_STR 20
+static char cap_str[MAX_CAP_STR][40];
+static DEFINE_SPINLOCK(cap_str_lock);
+static int last_cap_str;
+
+static char *gcap_string(char *s, int c)
+{
+	if (c & CEPH_CAP_GSHARED)
+		*s++ = 's';
+	if (c & CEPH_CAP_GEXCL)
+		*s++ = 'x';
+	if (c & CEPH_CAP_GCACHE)
+		*s++ = 'c';
+	if (c & CEPH_CAP_GRD)
+		*s++ = 'r';
+	if (c & CEPH_CAP_GWR)
+		*s++ = 'w';
+	if (c & CEPH_CAP_GBUFFER)
+		*s++ = 'b';
+	if (c & CEPH_CAP_GWREXTEND)
+		*s++ = 'a';
+	if (c & CEPH_CAP_GLAZYIO)
+		*s++ = 'l';
+	return s;
+}
+
+const char *ceph_cap_string(int caps)
+{
+	int i;
+	char *s;
+	int c;
+
+	spin_lock(&cap_str_lock);
+	i = last_cap_str++;
+	if (last_cap_str == MAX_CAP_STR)
+		last_cap_str = 0;
+	spin_unlock(&cap_str_lock);
+
+	s = cap_str[i];
+
+	if (caps & CEPH_CAP_PIN)
+		*s++ = 'p';
+
+	c = (caps >> CEPH_CAP_SAUTH) & 3;
+	if (c) {
+		*s++ = 'A';
+		s = gcap_string(s, c);
+	}
+
+	c = (caps >> CEPH_CAP_SLINK) & 3;
+	if (c) {
+		*s++ = 'L';
+		s = gcap_string(s, c);
+	}
+
+	c = (caps >> CEPH_CAP_SXATTR) & 3;
+	if (c) {
+		*s++ = 'X';
+		s = gcap_string(s, c);
+	}
+
+	c = caps >> CEPH_CAP_SFILE;
+	if (c) {
+		*s++ = 'F';
+		s = gcap_string(s, c);
+	}
+
+	if (s == cap_str[i])
+		*s++ = '-';
+	*s = 0;
+	return cap_str[i];
+}
+
+void ceph_caps_init(struct ceph_mds_client *mdsc)
+{
+	INIT_LIST_HEAD(&mdsc->caps_list);
+	spin_lock_init(&mdsc->caps_list_lock);
+}
+
+void ceph_caps_finalize(struct ceph_mds_client *mdsc)
+{
+	struct ceph_cap *cap;
+
+	spin_lock(&mdsc->caps_list_lock);
+	while (!list_empty(&mdsc->caps_list)) {
+		cap = list_first_entry(&mdsc->caps_list,
+				       struct ceph_cap, caps_item);
+		list_del(&cap->caps_item);
+		kmem_cache_free(ceph_cap_cachep, cap);
+	}
+	mdsc->caps_total_count = 0;
+	mdsc->caps_avail_count = 0;
+	mdsc->caps_use_count = 0;
+	mdsc->caps_reserve_count = 0;
+	mdsc->caps_min_count = 0;
+	spin_unlock(&mdsc->caps_list_lock);
+}
+
+void ceph_adjust_caps_max_min(struct ceph_mds_client *mdsc,
+			      struct ceph_mount_options *fsopt)
+{
+	spin_lock(&mdsc->caps_list_lock);
+	mdsc->caps_min_count = fsopt->max_readdir;
+	if (mdsc->caps_min_count < 1024)
+		mdsc->caps_min_count = 1024;
+	mdsc->caps_use_max = fsopt->caps_max;
+	if (mdsc->caps_use_max > 0 &&
+	    mdsc->caps_use_max < mdsc->caps_min_count)
+		mdsc->caps_use_max = mdsc->caps_min_count;
+	spin_unlock(&mdsc->caps_list_lock);
+}
+
+static void __ceph_unreserve_caps(struct ceph_mds_client *mdsc, int nr_caps)
+{
+	struct ceph_cap *cap;
+	int i;
+
+	if (nr_caps) {
+		BUG_ON(mdsc->caps_reserve_count < nr_caps);
+		mdsc->caps_reserve_count -= nr_caps;
+		if (mdsc->caps_avail_count >=
+		    mdsc->caps_reserve_count + mdsc->caps_min_count) {
+			mdsc->caps_total_count -= nr_caps;
+			for (i = 0; i < nr_caps; i++) {
+				cap = list_first_entry(&mdsc->caps_list,
+					struct ceph_cap, caps_item);
+				list_del(&cap->caps_item);
+				kmem_cache_free(ceph_cap_cachep, cap);
+			}
+		} else {
+			mdsc->caps_avail_count += nr_caps;
+		}
+
+		dout("%s: caps %d = %d used + %d resv + %d avail\n",
+		     __func__,
+		     mdsc->caps_total_count, mdsc->caps_use_count,
+		     mdsc->caps_reserve_count, mdsc->caps_avail_count);
+		BUG_ON(mdsc->caps_total_count != mdsc->caps_use_count +
+						 mdsc->caps_reserve_count +
+						 mdsc->caps_avail_count);
+	}
+}
+
+/*
+ * Called under mdsc->mutex.
+ */
+int ceph_reserve_caps(struct ceph_mds_client *mdsc,
+		      struct ceph_cap_reservation *ctx, int need)
+{
+	int i, j;
+	struct ceph_cap *cap;
+	int have;
+	int alloc = 0;
+	int max_caps;
+	int err = 0;
+	bool trimmed = false;
+	struct ceph_mds_session *s;
+	LIST_HEAD(newcaps);
+
+	dout("reserve caps ctx=%p need=%d\n", ctx, need);
+
+	/* first reserve any caps that are already allocated */
+	spin_lock(&mdsc->caps_list_lock);
+	if (mdsc->caps_avail_count >= need)
+		have = need;
+	else
+		have = mdsc->caps_avail_count;
+	mdsc->caps_avail_count -= have;
+	mdsc->caps_reserve_count += have;
+	BUG_ON(mdsc->caps_total_count != mdsc->caps_use_count +
+					 mdsc->caps_reserve_count +
+					 mdsc->caps_avail_count);
+	spin_unlock(&mdsc->caps_list_lock);
+
+	for (i = have; i < need; ) {
+		cap = kmem_cache_alloc(ceph_cap_cachep, GFP_NOFS);
+		if (cap) {
+			list_add(&cap->caps_item, &newcaps);
+			alloc++;
+			i++;
+			continue;
+		}
+
+		if (!trimmed) {
+			for (j = 0; j < mdsc->max_sessions; j++) {
+				s = __ceph_lookup_mds_session(mdsc, j);
+				if (!s)
+					continue;
+				mutex_unlock(&mdsc->mutex);
+
+				mutex_lock(&s->s_mutex);
+				max_caps = s->s_nr_caps - (need - i);
+				ceph_trim_caps(mdsc, s, max_caps);
+				mutex_unlock(&s->s_mutex);
+
+				ceph_put_mds_session(s);
+				mutex_lock(&mdsc->mutex);
+			}
+			trimmed = true;
+
+			spin_lock(&mdsc->caps_list_lock);
+			if (mdsc->caps_avail_count) {
+				int more_have;
+				if (mdsc->caps_avail_count >= need - i)
+					more_have = need - i;
+				else
+					more_have = mdsc->caps_avail_count;
+
+				i += more_have;
+				have += more_have;
+				mdsc->caps_avail_count -= more_have;
+				mdsc->caps_reserve_count += more_have;
+
+			}
+			spin_unlock(&mdsc->caps_list_lock);
+
+			continue;
+		}
+
+		pr_warn("reserve caps ctx=%p ENOMEM need=%d got=%d\n",
+			ctx, need, have + alloc);
+		err = -ENOMEM;
+		break;
+	}
+
+	if (!err) {
+		BUG_ON(have + alloc != need);
+		ctx->count = need;
+		ctx->used = 0;
+	}
+
+	spin_lock(&mdsc->caps_list_lock);
+	mdsc->caps_total_count += alloc;
+	mdsc->caps_reserve_count += alloc;
+	list_splice(&newcaps, &mdsc->caps_list);
+
+	BUG_ON(mdsc->caps_total_count != mdsc->caps_use_count +
+					 mdsc->caps_reserve_count +
+					 mdsc->caps_avail_count);
+
+	if (err)
+		__ceph_unreserve_caps(mdsc, have + alloc);
+
+	spin_unlock(&mdsc->caps_list_lock);
+
+	dout("reserve caps ctx=%p %d = %d used + %d resv + %d avail\n",
+	     ctx, mdsc->caps_total_count, mdsc->caps_use_count,
+	     mdsc->caps_reserve_count, mdsc->caps_avail_count);
+	return err;
+}
+
+void ceph_unreserve_caps(struct ceph_mds_client *mdsc,
+			 struct ceph_cap_reservation *ctx)
+{
+	bool reclaim = false;
+	if (!ctx->count)
+		return;
+
+	dout("unreserve caps ctx=%p count=%d\n", ctx, ctx->count);
+	spin_lock(&mdsc->caps_list_lock);
+	__ceph_unreserve_caps(mdsc, ctx->count);
+	ctx->count = 0;
+
+	if (mdsc->caps_use_max > 0 &&
+	    mdsc->caps_use_count > mdsc->caps_use_max)
+		reclaim = true;
+	spin_unlock(&mdsc->caps_list_lock);
+
+	if (reclaim)
+		ceph_reclaim_caps_nr(mdsc, ctx->used);
+}
+
+struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc,
+			      struct ceph_cap_reservation *ctx)
+{
+	struct ceph_cap *cap = NULL;
+
+	/* temporary, until we do something about cap import/export */
+	if (!ctx) {
+		cap = kmem_cache_alloc(ceph_cap_cachep, GFP_NOFS);
+		if (cap) {
+			spin_lock(&mdsc->caps_list_lock);
+			mdsc->caps_use_count++;
+			mdsc->caps_total_count++;
+			spin_unlock(&mdsc->caps_list_lock);
+		} else {
+			spin_lock(&mdsc->caps_list_lock);
+			if (mdsc->caps_avail_count) {
+				BUG_ON(list_empty(&mdsc->caps_list));
+
+				mdsc->caps_avail_count--;
+				mdsc->caps_use_count++;
+				cap = list_first_entry(&mdsc->caps_list,
+						struct ceph_cap, caps_item);
+				list_del(&cap->caps_item);
+
+				BUG_ON(mdsc->caps_total_count != mdsc->caps_use_count +
+				       mdsc->caps_reserve_count + mdsc->caps_avail_count);
+			}
+			spin_unlock(&mdsc->caps_list_lock);
+		}
+
+		return cap;
+	}
+
+	spin_lock(&mdsc->caps_list_lock);
+	dout("get_cap ctx=%p (%d) %d = %d used + %d resv + %d avail\n",
+	     ctx, ctx->count, mdsc->caps_total_count, mdsc->caps_use_count,
+	     mdsc->caps_reserve_count, mdsc->caps_avail_count);
+	BUG_ON(!ctx->count);
+	BUG_ON(ctx->count > mdsc->caps_reserve_count);
+	BUG_ON(list_empty(&mdsc->caps_list));
+
+	ctx->count--;
+	ctx->used++;
+	mdsc->caps_reserve_count--;
+	mdsc->caps_use_count++;
+
+	cap = list_first_entry(&mdsc->caps_list, struct ceph_cap, caps_item);
+	list_del(&cap->caps_item);
+
+	BUG_ON(mdsc->caps_total_count != mdsc->caps_use_count +
+	       mdsc->caps_reserve_count + mdsc->caps_avail_count);
+	spin_unlock(&mdsc->caps_list_lock);
+	return cap;
+}
+
+void ceph_put_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap)
+{
+	spin_lock(&mdsc->caps_list_lock);
+	dout("put_cap %p %d = %d used + %d resv + %d avail\n",
+	     cap, mdsc->caps_total_count, mdsc->caps_use_count,
+	     mdsc->caps_reserve_count, mdsc->caps_avail_count);
+	mdsc->caps_use_count--;
+	/*
+	 * Keep some preallocated caps around (ceph_min_count), to
+	 * avoid lots of free/alloc churn.
+	 */
+	if (mdsc->caps_avail_count >= mdsc->caps_reserve_count +
+				      mdsc->caps_min_count) {
+		mdsc->caps_total_count--;
+		kmem_cache_free(ceph_cap_cachep, cap);
+	} else {
+		mdsc->caps_avail_count++;
+		list_add(&cap->caps_item, &mdsc->caps_list);
+	}
+
+	BUG_ON(mdsc->caps_total_count != mdsc->caps_use_count +
+	       mdsc->caps_reserve_count + mdsc->caps_avail_count);
+	spin_unlock(&mdsc->caps_list_lock);
+}
+
+void ceph_reservation_status(struct ceph_fs_client *fsc,
+			     int *total, int *avail, int *used, int *reserved,
+			     int *min)
+{
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+
+	spin_lock(&mdsc->caps_list_lock);
+
+	if (total)
+		*total = mdsc->caps_total_count;
+	if (avail)
+		*avail = mdsc->caps_avail_count;
+	if (used)
+		*used = mdsc->caps_use_count;
+	if (reserved)
+		*reserved = mdsc->caps_reserve_count;
+	if (min)
+		*min = mdsc->caps_min_count;
+
+	spin_unlock(&mdsc->caps_list_lock);
+}
+
+/*
+ * Find ceph_cap for given mds, if any.
+ *
+ * Called with i_ceph_lock held.
+ */
+struct ceph_cap *__get_cap_for_mds(struct ceph_inode_info *ci, int mds)
+{
+	struct ceph_cap *cap;
+	struct rb_node *n = ci->i_caps.rb_node;
+
+	while (n) {
+		cap = rb_entry(n, struct ceph_cap, ci_node);
+		if (mds < cap->mds)
+			n = n->rb_left;
+		else if (mds > cap->mds)
+			n = n->rb_right;
+		else
+			return cap;
+	}
+	return NULL;
+}
+
+struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, int mds)
+{
+	struct ceph_cap *cap;
+
+	spin_lock(&ci->i_ceph_lock);
+	cap = __get_cap_for_mds(ci, mds);
+	spin_unlock(&ci->i_ceph_lock);
+	return cap;
+}
+
+/*
+ * Called under i_ceph_lock.
+ */
+static void __insert_cap_node(struct ceph_inode_info *ci,
+			      struct ceph_cap *new)
+{
+	struct rb_node **p = &ci->i_caps.rb_node;
+	struct rb_node *parent = NULL;
+	struct ceph_cap *cap = NULL;
+
+	while (*p) {
+		parent = *p;
+		cap = rb_entry(parent, struct ceph_cap, ci_node);
+		if (new->mds < cap->mds)
+			p = &(*p)->rb_left;
+		else if (new->mds > cap->mds)
+			p = &(*p)->rb_right;
+		else
+			BUG();
+	}
+
+	rb_link_node(&new->ci_node, parent, p);
+	rb_insert_color(&new->ci_node, &ci->i_caps);
+}
+
+/*
+ * (re)set cap hold timeouts, which control the delayed release
+ * of unused caps back to the MDS.  Should be called on cap use.
+ */
+static void __cap_set_timeouts(struct ceph_mds_client *mdsc,
+			       struct ceph_inode_info *ci)
+{
+	struct ceph_mount_options *opt = mdsc->fsc->mount_options;
+	ci->i_hold_caps_max = round_jiffies(jiffies +
+					    opt->caps_wanted_delay_max * HZ);
+	dout("__cap_set_timeouts %p %lu\n", &ci->netfs.inode,
+	     ci->i_hold_caps_max - jiffies);
+}
+
+/*
+ * (Re)queue cap at the end of the delayed cap release list.
+ *
+ * If I_FLUSH is set, leave the inode at the front of the list.
+ *
+ * Caller holds i_ceph_lock
+ *    -> we take mdsc->cap_delay_lock
+ */
+static void __cap_delay_requeue(struct ceph_mds_client *mdsc,
+				struct ceph_inode_info *ci)
+{
+	dout("__cap_delay_requeue %p flags 0x%lx at %lu\n", &ci->netfs.inode,
+	     ci->i_ceph_flags, ci->i_hold_caps_max);
+	if (!mdsc->stopping) {
+		spin_lock(&mdsc->cap_delay_lock);
+		if (!list_empty(&ci->i_cap_delay_list)) {
+			if (ci->i_ceph_flags & CEPH_I_FLUSH)
+				goto no_change;
+			list_del_init(&ci->i_cap_delay_list);
+		}
+		__cap_set_timeouts(mdsc, ci);
+		list_add_tail(&ci->i_cap_delay_list, &mdsc->cap_delay_list);
+no_change:
+		spin_unlock(&mdsc->cap_delay_lock);
+	}
+}
+
+/*
+ * Queue an inode for immediate writeback.  Mark inode with I_FLUSH,
+ * indicating we should send a cap message to flush dirty metadata
+ * asap, and move to the front of the delayed cap list.
+ */
+static void __cap_delay_requeue_front(struct ceph_mds_client *mdsc,
+				      struct ceph_inode_info *ci)
+{
+	dout("__cap_delay_requeue_front %p\n", &ci->netfs.inode);
+	spin_lock(&mdsc->cap_delay_lock);
+	ci->i_ceph_flags |= CEPH_I_FLUSH;
+	if (!list_empty(&ci->i_cap_delay_list))
+		list_del_init(&ci->i_cap_delay_list);
+	list_add(&ci->i_cap_delay_list, &mdsc->cap_delay_list);
+	spin_unlock(&mdsc->cap_delay_lock);
+}
+
+/*
+ * Cancel delayed work on cap.
+ *
+ * Caller must hold i_ceph_lock.
+ */
+static void __cap_delay_cancel(struct ceph_mds_client *mdsc,
+			       struct ceph_inode_info *ci)
+{
+	dout("__cap_delay_cancel %p\n", &ci->netfs.inode);
+	if (list_empty(&ci->i_cap_delay_list))
+		return;
+	spin_lock(&mdsc->cap_delay_lock);
+	list_del_init(&ci->i_cap_delay_list);
+	spin_unlock(&mdsc->cap_delay_lock);
+}
+
+/* Common issue checks for add_cap, handle_cap_grant. */
+static void __check_cap_issue(struct ceph_inode_info *ci, struct ceph_cap *cap,
+			      unsigned issued)
+{
+	unsigned had = __ceph_caps_issued(ci, NULL);
+
+	lockdep_assert_held(&ci->i_ceph_lock);
+
+	/*
+	 * Each time we receive FILE_CACHE anew, we increment
+	 * i_rdcache_gen.
+	 */
+	if (S_ISREG(ci->netfs.inode.i_mode) &&
+	    (issued & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) &&
+	    (had & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) == 0) {
+		ci->i_rdcache_gen++;
+	}
+
+	/*
+	 * If FILE_SHARED is newly issued, mark dir not complete. We don't
+	 * know what happened to this directory while we didn't have the cap.
+	 * If FILE_SHARED is being revoked, also mark dir not complete. It
+	 * stops on-going cached readdir.
+	 */
+	if ((issued & CEPH_CAP_FILE_SHARED) != (had & CEPH_CAP_FILE_SHARED)) {
+		if (issued & CEPH_CAP_FILE_SHARED)
+			atomic_inc(&ci->i_shared_gen);
+		if (S_ISDIR(ci->netfs.inode.i_mode)) {
+			dout(" marking %p NOT complete\n", &ci->netfs.inode);
+			__ceph_dir_clear_complete(ci);
+		}
+	}
+
+	/* Wipe saved layout if we're losing DIR_CREATE caps */
+	if (S_ISDIR(ci->netfs.inode.i_mode) && (had & CEPH_CAP_DIR_CREATE) &&
+		!(issued & CEPH_CAP_DIR_CREATE)) {
+	     ceph_put_string(rcu_dereference_raw(ci->i_cached_layout.pool_ns));
+	     memset(&ci->i_cached_layout, 0, sizeof(ci->i_cached_layout));
+	}
+}
+
+/**
+ * change_auth_cap_ses - move inode to appropriate lists when auth caps change
+ * @ci: inode to be moved
+ * @session: new auth caps session
+ */
+void change_auth_cap_ses(struct ceph_inode_info *ci,
+			 struct ceph_mds_session *session)
+{
+	lockdep_assert_held(&ci->i_ceph_lock);
+
+	if (list_empty(&ci->i_dirty_item) && list_empty(&ci->i_flushing_item))
+		return;
+
+	spin_lock(&session->s_mdsc->cap_dirty_lock);
+	if (!list_empty(&ci->i_dirty_item))
+		list_move(&ci->i_dirty_item, &session->s_cap_dirty);
+	if (!list_empty(&ci->i_flushing_item))
+		list_move_tail(&ci->i_flushing_item, &session->s_cap_flushing);
+	spin_unlock(&session->s_mdsc->cap_dirty_lock);
+}
+
+/*
+ * Add a capability under the given MDS session.
+ *
+ * Caller should hold session snap_rwsem (read) and ci->i_ceph_lock
+ *
+ * @fmode is the open file mode, if we are opening a file, otherwise
+ * it is < 0.  (This is so we can atomically add the cap and add an
+ * open file reference to it.)
+ */
+void ceph_add_cap(struct inode *inode,
+		  struct ceph_mds_session *session, u64 cap_id,
+		  unsigned issued, unsigned wanted,
+		  unsigned seq, unsigned mseq, u64 realmino, int flags,
+		  struct ceph_cap **new_cap)
+{
+	struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_cap *cap;
+	int mds = session->s_mds;
+	int actual_wanted;
+	u32 gen;
+
+	lockdep_assert_held(&ci->i_ceph_lock);
+
+	dout("add_cap %p mds%d cap %llx %s seq %d\n", inode,
+	     session->s_mds, cap_id, ceph_cap_string(issued), seq);
+
+	gen = atomic_read(&session->s_cap_gen);
+
+	cap = __get_cap_for_mds(ci, mds);
+	if (!cap) {
+		cap = *new_cap;
+		*new_cap = NULL;
+
+		cap->issued = 0;
+		cap->implemented = 0;
+		cap->mds = mds;
+		cap->mds_wanted = 0;
+		cap->mseq = 0;
+
+		cap->ci = ci;
+		__insert_cap_node(ci, cap);
+
+		/* add to session cap list */
+		cap->session = session;
+		spin_lock(&session->s_cap_lock);
+		list_add_tail(&cap->session_caps, &session->s_caps);
+		session->s_nr_caps++;
+		atomic64_inc(&mdsc->metric.total_caps);
+		spin_unlock(&session->s_cap_lock);
+	} else {
+		spin_lock(&session->s_cap_lock);
+		list_move_tail(&cap->session_caps, &session->s_caps);
+		spin_unlock(&session->s_cap_lock);
+
+		if (cap->cap_gen < gen)
+			cap->issued = cap->implemented = CEPH_CAP_PIN;
+
+		/*
+		 * auth mds of the inode changed. we received the cap export
+		 * message, but still haven't received the cap import message.
+		 * handle_cap_export() updated the new auth MDS' cap.
+		 *
+		 * "ceph_seq_cmp(seq, cap->seq) <= 0" means we are processing
+		 * a message that was send before the cap import message. So
+		 * don't remove caps.
+		 */
+		if (ceph_seq_cmp(seq, cap->seq) <= 0) {
+			WARN_ON(cap != ci->i_auth_cap);
+			WARN_ON(cap->cap_id != cap_id);
+			seq = cap->seq;
+			mseq = cap->mseq;
+			issued |= cap->issued;
+			flags |= CEPH_CAP_FLAG_AUTH;
+		}
+	}
+
+	if (!ci->i_snap_realm ||
+	    ((flags & CEPH_CAP_FLAG_AUTH) &&
+	     realmino != (u64)-1 && ci->i_snap_realm->ino != realmino)) {
+		/*
+		 * add this inode to the appropriate snap realm
+		 */
+		struct ceph_snap_realm *realm = ceph_lookup_snap_realm(mdsc,
+							       realmino);
+		if (realm)
+			ceph_change_snap_realm(inode, realm);
+		else
+			WARN(1, "%s: couldn't find snap realm 0x%llx (ino 0x%llx oldrealm 0x%llx)\n",
+			     __func__, realmino, ci->i_vino.ino,
+			     ci->i_snap_realm ? ci->i_snap_realm->ino : 0);
+	}
+
+	__check_cap_issue(ci, cap, issued);
+
+	/*
+	 * If we are issued caps we don't want, or the mds' wanted
+	 * value appears to be off, queue a check so we'll release
+	 * later and/or update the mds wanted value.
+	 */
+	actual_wanted = __ceph_caps_wanted(ci);
+	if ((wanted & ~actual_wanted) ||
+	    (issued & ~actual_wanted & CEPH_CAP_ANY_WR)) {
+		dout(" issued %s, mds wanted %s, actual %s, queueing\n",
+		     ceph_cap_string(issued), ceph_cap_string(wanted),
+		     ceph_cap_string(actual_wanted));
+		__cap_delay_requeue(mdsc, ci);
+	}
+
+	if (flags & CEPH_CAP_FLAG_AUTH) {
+		if (!ci->i_auth_cap ||
+		    ceph_seq_cmp(ci->i_auth_cap->mseq, mseq) < 0) {
+			if (ci->i_auth_cap &&
+			    ci->i_auth_cap->session != cap->session)
+				change_auth_cap_ses(ci, cap->session);
+			ci->i_auth_cap = cap;
+			cap->mds_wanted = wanted;
+		}
+	} else {
+		WARN_ON(ci->i_auth_cap == cap);
+	}
+
+	dout("add_cap inode %p (%llx.%llx) cap %p %s now %s seq %d mds%d\n",
+	     inode, ceph_vinop(inode), cap, ceph_cap_string(issued),
+	     ceph_cap_string(issued|cap->issued), seq, mds);
+	cap->cap_id = cap_id;
+	cap->issued = issued;
+	cap->implemented |= issued;
+	if (ceph_seq_cmp(mseq, cap->mseq) > 0)
+		cap->mds_wanted = wanted;
+	else
+		cap->mds_wanted |= wanted;
+	cap->seq = seq;
+	cap->issue_seq = seq;
+	cap->mseq = mseq;
+	cap->cap_gen = gen;
+	wake_up_all(&ci->i_cap_wq);
+}
+
+/*
+ * Return true if cap has not timed out and belongs to the current
+ * generation of the MDS session (i.e. has not gone 'stale' due to
+ * us losing touch with the mds).
+ */
+static int __cap_is_valid(struct ceph_cap *cap)
+{
+	unsigned long ttl;
+	u32 gen;
+
+	gen = atomic_read(&cap->session->s_cap_gen);
+	ttl = cap->session->s_cap_ttl;
+
+	if (cap->cap_gen < gen || time_after_eq(jiffies, ttl)) {
+		dout("__cap_is_valid %p cap %p issued %s "
+		     "but STALE (gen %u vs %u)\n", &cap->ci->netfs.inode,
+		     cap, ceph_cap_string(cap->issued), cap->cap_gen, gen);
+		return 0;
+	}
+
+	return 1;
+}
+
+/*
+ * Return set of valid cap bits issued to us.  Note that caps time
+ * out, and may be invalidated in bulk if the client session times out
+ * and session->s_cap_gen is bumped.
+ */
+int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented)
+{
+	int have = ci->i_snap_caps;
+	struct ceph_cap *cap;
+	struct rb_node *p;
+
+	if (implemented)
+		*implemented = 0;
+	for (p = rb_first(&ci->i_caps); p; p = rb_next(p)) {
+		cap = rb_entry(p, struct ceph_cap, ci_node);
+		if (!__cap_is_valid(cap))
+			continue;
+		dout("__ceph_caps_issued %p cap %p issued %s\n",
+		     &ci->netfs.inode, cap, ceph_cap_string(cap->issued));
+		have |= cap->issued;
+		if (implemented)
+			*implemented |= cap->implemented;
+	}
+	/*
+	 * exclude caps issued by non-auth MDS, but are been revoking
+	 * by the auth MDS. The non-auth MDS should be revoking/exporting
+	 * these caps, but the message is delayed.
+	 */
+	if (ci->i_auth_cap) {
+		cap = ci->i_auth_cap;
+		have &= ~cap->implemented | cap->issued;
+	}
+	return have;
+}
+
+/*
+ * Get cap bits issued by caps other than @ocap
+ */
+int __ceph_caps_issued_other(struct ceph_inode_info *ci, struct ceph_cap *ocap)
+{
+	int have = ci->i_snap_caps;
+	struct ceph_cap *cap;
+	struct rb_node *p;
+
+	for (p = rb_first(&ci->i_caps); p; p = rb_next(p)) {
+		cap = rb_entry(p, struct ceph_cap, ci_node);
+		if (cap == ocap)
+			continue;
+		if (!__cap_is_valid(cap))
+			continue;
+		have |= cap->issued;
+	}
+	return have;
+}
+
+/*
+ * Move a cap to the end of the LRU (oldest caps at list head, newest
+ * at list tail).
+ */
+static void __touch_cap(struct ceph_cap *cap)
+{
+	struct ceph_mds_session *s = cap->session;
+
+	spin_lock(&s->s_cap_lock);
+	if (!s->s_cap_iterator) {
+		dout("__touch_cap %p cap %p mds%d\n", &cap->ci->netfs.inode, cap,
+		     s->s_mds);
+		list_move_tail(&cap->session_caps, &s->s_caps);
+	} else {
+		dout("__touch_cap %p cap %p mds%d NOP, iterating over caps\n",
+		     &cap->ci->netfs.inode, cap, s->s_mds);
+	}
+	spin_unlock(&s->s_cap_lock);
+}
+
+/*
+ * Check if we hold the given mask.  If so, move the cap(s) to the
+ * front of their respective LRUs.  (This is the preferred way for
+ * callers to check for caps they want.)
+ */
+int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int touch)
+{
+	struct ceph_cap *cap;
+	struct rb_node *p;
+	int have = ci->i_snap_caps;
+
+	if ((have & mask) == mask) {
+		dout("__ceph_caps_issued_mask ino 0x%llx snap issued %s"
+		     " (mask %s)\n", ceph_ino(&ci->netfs.inode),
+		     ceph_cap_string(have),
+		     ceph_cap_string(mask));
+		return 1;
+	}
+
+	for (p = rb_first(&ci->i_caps); p; p = rb_next(p)) {
+		cap = rb_entry(p, struct ceph_cap, ci_node);
+		if (!__cap_is_valid(cap))
+			continue;
+		if ((cap->issued & mask) == mask) {
+			dout("__ceph_caps_issued_mask ino 0x%llx cap %p issued %s"
+			     " (mask %s)\n", ceph_ino(&ci->netfs.inode), cap,
+			     ceph_cap_string(cap->issued),
+			     ceph_cap_string(mask));
+			if (touch)
+				__touch_cap(cap);
+			return 1;
+		}
+
+		/* does a combination of caps satisfy mask? */
+		have |= cap->issued;
+		if ((have & mask) == mask) {
+			dout("__ceph_caps_issued_mask ino 0x%llx combo issued %s"
+			     " (mask %s)\n", ceph_ino(&ci->netfs.inode),
+			     ceph_cap_string(cap->issued),
+			     ceph_cap_string(mask));
+			if (touch) {
+				struct rb_node *q;
+
+				/* touch this + preceding caps */
+				__touch_cap(cap);
+				for (q = rb_first(&ci->i_caps); q != p;
+				     q = rb_next(q)) {
+					cap = rb_entry(q, struct ceph_cap,
+						       ci_node);
+					if (!__cap_is_valid(cap))
+						continue;
+					if (cap->issued & mask)
+						__touch_cap(cap);
+				}
+			}
+			return 1;
+		}
+	}
+
+	return 0;
+}
+
+int __ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, int mask,
+				   int touch)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(ci->netfs.inode.i_sb);
+	int r;
+
+	r = __ceph_caps_issued_mask(ci, mask, touch);
+	if (r)
+		ceph_update_cap_hit(&fsc->mdsc->metric);
+	else
+		ceph_update_cap_mis(&fsc->mdsc->metric);
+	return r;
+}
+
+/*
+ * Return true if mask caps are currently being revoked by an MDS.
+ */
+int __ceph_caps_revoking_other(struct ceph_inode_info *ci,
+			       struct ceph_cap *ocap, int mask)
+{
+	struct ceph_cap *cap;
+	struct rb_node *p;
+
+	for (p = rb_first(&ci->i_caps); p; p = rb_next(p)) {
+		cap = rb_entry(p, struct ceph_cap, ci_node);
+		if (cap != ocap &&
+		    (cap->implemented & ~cap->issued & mask))
+			return 1;
+	}
+	return 0;
+}
+
+int ceph_caps_revoking(struct ceph_inode_info *ci, int mask)
+{
+	struct inode *inode = &ci->netfs.inode;
+	int ret;
+
+	spin_lock(&ci->i_ceph_lock);
+	ret = __ceph_caps_revoking_other(ci, NULL, mask);
+	spin_unlock(&ci->i_ceph_lock);
+	dout("ceph_caps_revoking %p %s = %d\n", inode,
+	     ceph_cap_string(mask), ret);
+	return ret;
+}
+
+int __ceph_caps_used(struct ceph_inode_info *ci)
+{
+	int used = 0;
+	if (ci->i_pin_ref)
+		used |= CEPH_CAP_PIN;
+	if (ci->i_rd_ref)
+		used |= CEPH_CAP_FILE_RD;
+	if (ci->i_rdcache_ref ||
+	    (S_ISREG(ci->netfs.inode.i_mode) &&
+	     ci->netfs.inode.i_data.nrpages))
+		used |= CEPH_CAP_FILE_CACHE;
+	if (ci->i_wr_ref)
+		used |= CEPH_CAP_FILE_WR;
+	if (ci->i_wb_ref || ci->i_wrbuffer_ref)
+		used |= CEPH_CAP_FILE_BUFFER;
+	if (ci->i_fx_ref)
+		used |= CEPH_CAP_FILE_EXCL;
+	return used;
+}
+
+#define FMODE_WAIT_BIAS 1000
+
+/*
+ * wanted, by virtue of open file modes
+ */
+int __ceph_caps_file_wanted(struct ceph_inode_info *ci)
+{
+	const int PIN_SHIFT = ffs(CEPH_FILE_MODE_PIN);
+	const int RD_SHIFT = ffs(CEPH_FILE_MODE_RD);
+	const int WR_SHIFT = ffs(CEPH_FILE_MODE_WR);
+	const int LAZY_SHIFT = ffs(CEPH_FILE_MODE_LAZY);
+	struct ceph_mount_options *opt =
+		ceph_inode_to_client(&ci->netfs.inode)->mount_options;
+	unsigned long used_cutoff = jiffies - opt->caps_wanted_delay_max * HZ;
+	unsigned long idle_cutoff = jiffies - opt->caps_wanted_delay_min * HZ;
+
+	if (S_ISDIR(ci->netfs.inode.i_mode)) {
+		int want = 0;
+
+		/* use used_cutoff here, to keep dir's wanted caps longer */
+		if (ci->i_nr_by_mode[RD_SHIFT] > 0 ||
+		    time_after(ci->i_last_rd, used_cutoff))
+			want |= CEPH_CAP_ANY_SHARED;
+
+		if (ci->i_nr_by_mode[WR_SHIFT] > 0 ||
+		    time_after(ci->i_last_wr, used_cutoff)) {
+			want |= CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_EXCL;
+			if (opt->flags & CEPH_MOUNT_OPT_ASYNC_DIROPS)
+				want |= CEPH_CAP_ANY_DIR_OPS;
+		}
+
+		if (want || ci->i_nr_by_mode[PIN_SHIFT] > 0)
+			want |= CEPH_CAP_PIN;
+
+		return want;
+	} else {
+		int bits = 0;
+
+		if (ci->i_nr_by_mode[RD_SHIFT] > 0) {
+			if (ci->i_nr_by_mode[RD_SHIFT] >= FMODE_WAIT_BIAS ||
+			    time_after(ci->i_last_rd, used_cutoff))
+				bits |= 1 << RD_SHIFT;
+		} else if (time_after(ci->i_last_rd, idle_cutoff)) {
+			bits |= 1 << RD_SHIFT;
+		}
+
+		if (ci->i_nr_by_mode[WR_SHIFT] > 0) {
+			if (ci->i_nr_by_mode[WR_SHIFT] >= FMODE_WAIT_BIAS ||
+			    time_after(ci->i_last_wr, used_cutoff))
+				bits |= 1 << WR_SHIFT;
+		} else if (time_after(ci->i_last_wr, idle_cutoff)) {
+			bits |= 1 << WR_SHIFT;
+		}
+
+		/* check lazyio only when read/write is wanted */
+		if ((bits & (CEPH_FILE_MODE_RDWR << 1)) &&
+		    ci->i_nr_by_mode[LAZY_SHIFT] > 0)
+			bits |= 1 << LAZY_SHIFT;
+
+		return bits ? ceph_caps_for_mode(bits >> 1) : 0;
+	}
+}
+
+/*
+ * wanted, by virtue of open file modes AND cap refs (buffered/cached data)
+ */
+int __ceph_caps_wanted(struct ceph_inode_info *ci)
+{
+	int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci);
+	if (S_ISDIR(ci->netfs.inode.i_mode)) {
+		/* we want EXCL if holding caps of dir ops */
+		if (w & CEPH_CAP_ANY_DIR_OPS)
+			w |= CEPH_CAP_FILE_EXCL;
+	} else {
+		/* we want EXCL if dirty data */
+		if (w & CEPH_CAP_FILE_BUFFER)
+			w |= CEPH_CAP_FILE_EXCL;
+	}
+	return w;
+}
+
+/*
+ * Return caps we have registered with the MDS(s) as 'wanted'.
+ */
+int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check)
+{
+	struct ceph_cap *cap;
+	struct rb_node *p;
+	int mds_wanted = 0;
+
+	for (p = rb_first(&ci->i_caps); p; p = rb_next(p)) {
+		cap = rb_entry(p, struct ceph_cap, ci_node);
+		if (check && !__cap_is_valid(cap))
+			continue;
+		if (cap == ci->i_auth_cap)
+			mds_wanted |= cap->mds_wanted;
+		else
+			mds_wanted |= (cap->mds_wanted & ~CEPH_CAP_ANY_FILE_WR);
+	}
+	return mds_wanted;
+}
+
+int ceph_is_any_caps(struct inode *inode)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	int ret;
+
+	spin_lock(&ci->i_ceph_lock);
+	ret = __ceph_is_any_real_caps(ci);
+	spin_unlock(&ci->i_ceph_lock);
+
+	return ret;
+}
+
+/*
+ * Remove a cap.  Take steps to deal with a racing iterate_session_caps.
+ *
+ * caller should hold i_ceph_lock.
+ * caller will not hold session s_mutex if called from destroy_inode.
+ */
+void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release)
+{
+	struct ceph_mds_session *session = cap->session;
+	struct ceph_inode_info *ci = cap->ci;
+	struct ceph_mds_client *mdsc;
+	int removed = 0;
+
+	/* 'ci' being NULL means the remove have already occurred */
+	if (!ci) {
+		dout("%s: cap inode is NULL\n", __func__);
+		return;
+	}
+
+	lockdep_assert_held(&ci->i_ceph_lock);
+
+	dout("__ceph_remove_cap %p from %p\n", cap, &ci->netfs.inode);
+
+	mdsc = ceph_inode_to_client(&ci->netfs.inode)->mdsc;
+
+	/* remove from inode's cap rbtree, and clear auth cap */
+	rb_erase(&cap->ci_node, &ci->i_caps);
+	if (ci->i_auth_cap == cap)
+		ci->i_auth_cap = NULL;
+
+	/* remove from session list */
+	spin_lock(&session->s_cap_lock);
+	if (session->s_cap_iterator == cap) {
+		/* not yet, we are iterating over this very cap */
+		dout("__ceph_remove_cap  delaying %p removal from session %p\n",
+		     cap, cap->session);
+	} else {
+		list_del_init(&cap->session_caps);
+		session->s_nr_caps--;
+		atomic64_dec(&mdsc->metric.total_caps);
+		cap->session = NULL;
+		removed = 1;
+	}
+	/* protect backpointer with s_cap_lock: see iterate_session_caps */
+	cap->ci = NULL;
+
+	/*
+	 * s_cap_reconnect is protected by s_cap_lock. no one changes
+	 * s_cap_gen while session is in the reconnect state.
+	 */
+	if (queue_release &&
+	    (!session->s_cap_reconnect ||
+	     cap->cap_gen == atomic_read(&session->s_cap_gen))) {
+		cap->queue_release = 1;
+		if (removed) {
+			__ceph_queue_cap_release(session, cap);
+			removed = 0;
+		}
+	} else {
+		cap->queue_release = 0;
+	}
+	cap->cap_ino = ci->i_vino.ino;
+
+	spin_unlock(&session->s_cap_lock);
+
+	if (removed)
+		ceph_put_cap(mdsc, cap);
+
+	if (!__ceph_is_any_real_caps(ci)) {
+		/* when reconnect denied, we remove session caps forcibly,
+		 * i_wr_ref can be non-zero. If there are ongoing write,
+		 * keep i_snap_realm.
+		 */
+		if (ci->i_wr_ref == 0 && ci->i_snap_realm)
+			ceph_change_snap_realm(&ci->netfs.inode, NULL);
+
+		__cap_delay_cancel(mdsc, ci);
+	}
+}
+
+void ceph_remove_cap(struct ceph_cap *cap, bool queue_release)
+{
+	struct ceph_inode_info *ci = cap->ci;
+	struct ceph_fs_client *fsc;
+
+	/* 'ci' being NULL means the remove have already occurred */
+	if (!ci) {
+		dout("%s: cap inode is NULL\n", __func__);
+		return;
+	}
+
+	lockdep_assert_held(&ci->i_ceph_lock);
+
+	fsc = ceph_inode_to_client(&ci->netfs.inode);
+	WARN_ON_ONCE(ci->i_auth_cap == cap &&
+		     !list_empty(&ci->i_dirty_item) &&
+		     !fsc->blocklisted &&
+		     !ceph_inode_is_shutdown(&ci->netfs.inode));
+
+	__ceph_remove_cap(cap, queue_release);
+}
+
+struct cap_msg_args {
+	struct ceph_mds_session	*session;
+	u64			ino, cid, follows;
+	u64			flush_tid, oldest_flush_tid, size, max_size;
+	u64			xattr_version;
+	u64			change_attr;
+	struct ceph_buffer	*xattr_buf;
+	struct ceph_buffer	*old_xattr_buf;
+	struct timespec64	atime, mtime, ctime, btime;
+	int			op, caps, wanted, dirty;
+	u32			seq, issue_seq, mseq, time_warp_seq;
+	u32			flags;
+	kuid_t			uid;
+	kgid_t			gid;
+	umode_t			mode;
+	bool			inline_data;
+	bool			wake;
+};
+
+/*
+ * cap struct size + flock buffer size + inline version + inline data size +
+ * osd_epoch_barrier + oldest_flush_tid
+ */
+#define CAP_MSG_SIZE (sizeof(struct ceph_mds_caps) + \
+		      4 + 8 + 4 + 4 + 8 + 4 + 4 + 4 + 8 + 8 + 4)
+
+/* Marshal up the cap msg to the MDS */
+static void encode_cap_msg(struct ceph_msg *msg, struct cap_msg_args *arg)
+{
+	struct ceph_mds_caps *fc;
+	void *p;
+	struct ceph_osd_client *osdc = &arg->session->s_mdsc->fsc->client->osdc;
+
+	dout("%s %s %llx %llx caps %s wanted %s dirty %s seq %u/%u tid %llu/%llu mseq %u follows %lld size %llu/%llu xattr_ver %llu xattr_len %d\n",
+	     __func__, ceph_cap_op_name(arg->op), arg->cid, arg->ino,
+	     ceph_cap_string(arg->caps), ceph_cap_string(arg->wanted),
+	     ceph_cap_string(arg->dirty), arg->seq, arg->issue_seq,
+	     arg->flush_tid, arg->oldest_flush_tid, arg->mseq, arg->follows,
+	     arg->size, arg->max_size, arg->xattr_version,
+	     arg->xattr_buf ? (int)arg->xattr_buf->vec.iov_len : 0);
+
+	msg->hdr.version = cpu_to_le16(10);
+	msg->hdr.tid = cpu_to_le64(arg->flush_tid);
+
+	fc = msg->front.iov_base;
+	memset(fc, 0, sizeof(*fc));
+
+	fc->cap_id = cpu_to_le64(arg->cid);
+	fc->op = cpu_to_le32(arg->op);
+	fc->seq = cpu_to_le32(arg->seq);
+	fc->issue_seq = cpu_to_le32(arg->issue_seq);
+	fc->migrate_seq = cpu_to_le32(arg->mseq);
+	fc->caps = cpu_to_le32(arg->caps);
+	fc->wanted = cpu_to_le32(arg->wanted);
+	fc->dirty = cpu_to_le32(arg->dirty);
+	fc->ino = cpu_to_le64(arg->ino);
+	fc->snap_follows = cpu_to_le64(arg->follows);
+
+	fc->size = cpu_to_le64(arg->size);
+	fc->max_size = cpu_to_le64(arg->max_size);
+	ceph_encode_timespec64(&fc->mtime, &arg->mtime);
+	ceph_encode_timespec64(&fc->atime, &arg->atime);
+	ceph_encode_timespec64(&fc->ctime, &arg->ctime);
+	fc->time_warp_seq = cpu_to_le32(arg->time_warp_seq);
+
+	fc->uid = cpu_to_le32(from_kuid(&init_user_ns, arg->uid));
+	fc->gid = cpu_to_le32(from_kgid(&init_user_ns, arg->gid));
+	fc->mode = cpu_to_le32(arg->mode);
+
+	fc->xattr_version = cpu_to_le64(arg->xattr_version);
+	if (arg->xattr_buf) {
+		msg->middle = ceph_buffer_get(arg->xattr_buf);
+		fc->xattr_len = cpu_to_le32(arg->xattr_buf->vec.iov_len);
+		msg->hdr.middle_len = cpu_to_le32(arg->xattr_buf->vec.iov_len);
+	}
+
+	p = fc + 1;
+	/* flock buffer size (version 2) */
+	ceph_encode_32(&p, 0);
+	/* inline version (version 4) */
+	ceph_encode_64(&p, arg->inline_data ? 0 : CEPH_INLINE_NONE);
+	/* inline data size */
+	ceph_encode_32(&p, 0);
+	/*
+	 * osd_epoch_barrier (version 5)
+	 * The epoch_barrier is protected osdc->lock, so READ_ONCE here in
+	 * case it was recently changed
+	 */
+	ceph_encode_32(&p, READ_ONCE(osdc->epoch_barrier));
+	/* oldest_flush_tid (version 6) */
+	ceph_encode_64(&p, arg->oldest_flush_tid);
+
+	/*
+	 * caller_uid/caller_gid (version 7)
+	 *
+	 * Currently, we don't properly track which caller dirtied the caps
+	 * last, and force a flush of them when there is a conflict. For now,
+	 * just set this to 0:0, to emulate how the MDS has worked up to now.
+	 */
+	ceph_encode_32(&p, 0);
+	ceph_encode_32(&p, 0);
+
+	/* pool namespace (version 8) (mds always ignores this) */
+	ceph_encode_32(&p, 0);
+
+	/* btime and change_attr (version 9) */
+	ceph_encode_timespec64(p, &arg->btime);
+	p += sizeof(struct ceph_timespec);
+	ceph_encode_64(&p, arg->change_attr);
+
+	/* Advisory flags (version 10) */
+	ceph_encode_32(&p, arg->flags);
+}
+
+/*
+ * Queue cap releases when an inode is dropped from our cache.
+ */
+void __ceph_remove_caps(struct ceph_inode_info *ci)
+{
+	struct rb_node *p;
+
+	/* lock i_ceph_lock, because ceph_d_revalidate(..., LOOKUP_RCU)
+	 * may call __ceph_caps_issued_mask() on a freeing inode. */
+	spin_lock(&ci->i_ceph_lock);
+	p = rb_first(&ci->i_caps);
+	while (p) {
+		struct ceph_cap *cap = rb_entry(p, struct ceph_cap, ci_node);
+		p = rb_next(p);
+		ceph_remove_cap(cap, true);
+	}
+	spin_unlock(&ci->i_ceph_lock);
+}
+
+/*
+ * Prepare to send a cap message to an MDS. Update the cap state, and populate
+ * the arg struct with the parameters that will need to be sent. This should
+ * be done under the i_ceph_lock to guard against changes to cap state.
+ *
+ * Make note of max_size reported/requested from mds, revoked caps
+ * that have now been implemented.
+ */
+static void __prep_cap(struct cap_msg_args *arg, struct ceph_cap *cap,
+		       int op, int flags, int used, int want, int retain,
+		       int flushing, u64 flush_tid, u64 oldest_flush_tid)
+{
+	struct ceph_inode_info *ci = cap->ci;
+	struct inode *inode = &ci->netfs.inode;
+	int held, revoking;
+
+	lockdep_assert_held(&ci->i_ceph_lock);
+
+	held = cap->issued | cap->implemented;
+	revoking = cap->implemented & ~cap->issued;
+	retain &= ~revoking;
+
+	dout("%s %p cap %p session %p %s -> %s (revoking %s)\n",
+	     __func__, inode, cap, cap->session,
+	     ceph_cap_string(held), ceph_cap_string(held & retain),
+	     ceph_cap_string(revoking));
+	BUG_ON((retain & CEPH_CAP_PIN) == 0);
+
+	ci->i_ceph_flags &= ~CEPH_I_FLUSH;
+
+	cap->issued &= retain;  /* drop bits we don't want */
+	/*
+	 * Wake up any waiters on wanted -> needed transition. This is due to
+	 * the weird transition from buffered to sync IO... we need to flush
+	 * dirty pages _before_ allowing sync writes to avoid reordering.
+	 */
+	arg->wake = cap->implemented & ~cap->issued;
+	cap->implemented &= cap->issued | used;
+	cap->mds_wanted = want;
+
+	arg->session = cap->session;
+	arg->ino = ceph_vino(inode).ino;
+	arg->cid = cap->cap_id;
+	arg->follows = flushing ? ci->i_head_snapc->seq : 0;
+	arg->flush_tid = flush_tid;
+	arg->oldest_flush_tid = oldest_flush_tid;
+
+	arg->size = i_size_read(inode);
+	ci->i_reported_size = arg->size;
+	arg->max_size = ci->i_wanted_max_size;
+	if (cap == ci->i_auth_cap) {
+		if (want & CEPH_CAP_ANY_FILE_WR)
+			ci->i_requested_max_size = arg->max_size;
+		else
+			ci->i_requested_max_size = 0;
+	}
+
+	if (flushing & CEPH_CAP_XATTR_EXCL) {
+		arg->old_xattr_buf = __ceph_build_xattrs_blob(ci);
+		arg->xattr_version = ci->i_xattrs.version;
+		arg->xattr_buf = ci->i_xattrs.blob;
+	} else {
+		arg->xattr_buf = NULL;
+		arg->old_xattr_buf = NULL;
+	}
+
+	arg->mtime = inode->i_mtime;
+	arg->atime = inode->i_atime;
+	arg->ctime = inode->i_ctime;
+	arg->btime = ci->i_btime;
+	arg->change_attr = inode_peek_iversion_raw(inode);
+
+	arg->op = op;
+	arg->caps = cap->implemented;
+	arg->wanted = want;
+	arg->dirty = flushing;
+
+	arg->seq = cap->seq;
+	arg->issue_seq = cap->issue_seq;
+	arg->mseq = cap->mseq;
+	arg->time_warp_seq = ci->i_time_warp_seq;
+
+	arg->uid = inode->i_uid;
+	arg->gid = inode->i_gid;
+	arg->mode = inode->i_mode;
+
+	arg->inline_data = ci->i_inline_version != CEPH_INLINE_NONE;
+	if (!(flags & CEPH_CLIENT_CAPS_PENDING_CAPSNAP) &&
+	    !list_empty(&ci->i_cap_snaps)) {
+		struct ceph_cap_snap *capsnap;
+		list_for_each_entry_reverse(capsnap, &ci->i_cap_snaps, ci_item) {
+			if (capsnap->cap_flush.tid)
+				break;
+			if (capsnap->need_flush) {
+				flags |= CEPH_CLIENT_CAPS_PENDING_CAPSNAP;
+				break;
+			}
+		}
+	}
+	arg->flags = flags;
+}
+
+/*
+ * Send a cap msg on the given inode.
+ *
+ * Caller should hold snap_rwsem (read), s_mutex.
+ */
+static void __send_cap(struct cap_msg_args *arg, struct ceph_inode_info *ci)
+{
+	struct ceph_msg *msg;
+	struct inode *inode = &ci->netfs.inode;
+
+	msg = ceph_msg_new(CEPH_MSG_CLIENT_CAPS, CAP_MSG_SIZE, GFP_NOFS, false);
+	if (!msg) {
+		pr_err("error allocating cap msg: ino (%llx.%llx) flushing %s tid %llu, requeuing cap.\n",
+		       ceph_vinop(inode), ceph_cap_string(arg->dirty),
+		       arg->flush_tid);
+		spin_lock(&ci->i_ceph_lock);
+		__cap_delay_requeue(arg->session->s_mdsc, ci);
+		spin_unlock(&ci->i_ceph_lock);
+		return;
+	}
+
+	encode_cap_msg(msg, arg);
+	ceph_con_send(&arg->session->s_con, msg);
+	ceph_buffer_put(arg->old_xattr_buf);
+	if (arg->wake)
+		wake_up_all(&ci->i_cap_wq);
+}
+
+static inline int __send_flush_snap(struct inode *inode,
+				    struct ceph_mds_session *session,
+				    struct ceph_cap_snap *capsnap,
+				    u32 mseq, u64 oldest_flush_tid)
+{
+	struct cap_msg_args	arg;
+	struct ceph_msg		*msg;
+
+	msg = ceph_msg_new(CEPH_MSG_CLIENT_CAPS, CAP_MSG_SIZE, GFP_NOFS, false);
+	if (!msg)
+		return -ENOMEM;
+
+	arg.session = session;
+	arg.ino = ceph_vino(inode).ino;
+	arg.cid = 0;
+	arg.follows = capsnap->follows;
+	arg.flush_tid = capsnap->cap_flush.tid;
+	arg.oldest_flush_tid = oldest_flush_tid;
+
+	arg.size = capsnap->size;
+	arg.max_size = 0;
+	arg.xattr_version = capsnap->xattr_version;
+	arg.xattr_buf = capsnap->xattr_blob;
+	arg.old_xattr_buf = NULL;
+
+	arg.atime = capsnap->atime;
+	arg.mtime = capsnap->mtime;
+	arg.ctime = capsnap->ctime;
+	arg.btime = capsnap->btime;
+	arg.change_attr = capsnap->change_attr;
+
+	arg.op = CEPH_CAP_OP_FLUSHSNAP;
+	arg.caps = capsnap->issued;
+	arg.wanted = 0;
+	arg.dirty = capsnap->dirty;
+
+	arg.seq = 0;
+	arg.issue_seq = 0;
+	arg.mseq = mseq;
+	arg.time_warp_seq = capsnap->time_warp_seq;
+
+	arg.uid = capsnap->uid;
+	arg.gid = capsnap->gid;
+	arg.mode = capsnap->mode;
+
+	arg.inline_data = capsnap->inline_data;
+	arg.flags = 0;
+	arg.wake = false;
+
+	encode_cap_msg(msg, &arg);
+	ceph_con_send(&arg.session->s_con, msg);
+	return 0;
+}
+
+/*
+ * When a snapshot is taken, clients accumulate dirty metadata on
+ * inodes with capabilities in ceph_cap_snaps to describe the file
+ * state at the time the snapshot was taken.  This must be flushed
+ * asynchronously back to the MDS once sync writes complete and dirty
+ * data is written out.
+ *
+ * Called under i_ceph_lock.
+ */
+static void __ceph_flush_snaps(struct ceph_inode_info *ci,
+			       struct ceph_mds_session *session)
+		__releases(ci->i_ceph_lock)
+		__acquires(ci->i_ceph_lock)
+{
+	struct inode *inode = &ci->netfs.inode;
+	struct ceph_mds_client *mdsc = session->s_mdsc;
+	struct ceph_cap_snap *capsnap;
+	u64 oldest_flush_tid = 0;
+	u64 first_tid = 1, last_tid = 0;
+
+	dout("__flush_snaps %p session %p\n", inode, session);
+
+	list_for_each_entry(capsnap, &ci->i_cap_snaps, ci_item) {
+		/*
+		 * we need to wait for sync writes to complete and for dirty
+		 * pages to be written out.
+		 */
+		if (capsnap->dirty_pages || capsnap->writing)
+			break;
+
+		/* should be removed by ceph_try_drop_cap_snap() */
+		BUG_ON(!capsnap->need_flush);
+
+		/* only flush each capsnap once */
+		if (capsnap->cap_flush.tid > 0) {
+			dout(" already flushed %p, skipping\n", capsnap);
+			continue;
+		}
+
+		spin_lock(&mdsc->cap_dirty_lock);
+		capsnap->cap_flush.tid = ++mdsc->last_cap_flush_tid;
+		list_add_tail(&capsnap->cap_flush.g_list,
+			      &mdsc->cap_flush_list);
+		if (oldest_flush_tid == 0)
+			oldest_flush_tid = __get_oldest_flush_tid(mdsc);
+		if (list_empty(&ci->i_flushing_item)) {
+			list_add_tail(&ci->i_flushing_item,
+				      &session->s_cap_flushing);
+		}
+		spin_unlock(&mdsc->cap_dirty_lock);
+
+		list_add_tail(&capsnap->cap_flush.i_list,
+			      &ci->i_cap_flush_list);
+
+		if (first_tid == 1)
+			first_tid = capsnap->cap_flush.tid;
+		last_tid = capsnap->cap_flush.tid;
+	}
+
+	ci->i_ceph_flags &= ~CEPH_I_FLUSH_SNAPS;
+
+	while (first_tid <= last_tid) {
+		struct ceph_cap *cap = ci->i_auth_cap;
+		struct ceph_cap_flush *cf = NULL, *iter;
+		int ret;
+
+		if (!(cap && cap->session == session)) {
+			dout("__flush_snaps %p auth cap %p not mds%d, "
+			     "stop\n", inode, cap, session->s_mds);
+			break;
+		}
+
+		ret = -ENOENT;
+		list_for_each_entry(iter, &ci->i_cap_flush_list, i_list) {
+			if (iter->tid >= first_tid) {
+				cf = iter;
+				ret = 0;
+				break;
+			}
+		}
+		if (ret < 0)
+			break;
+
+		first_tid = cf->tid + 1;
+
+		capsnap = container_of(cf, struct ceph_cap_snap, cap_flush);
+		refcount_inc(&capsnap->nref);
+		spin_unlock(&ci->i_ceph_lock);
+
+		dout("__flush_snaps %p capsnap %p tid %llu %s\n",
+		     inode, capsnap, cf->tid, ceph_cap_string(capsnap->dirty));
+
+		ret = __send_flush_snap(inode, session, capsnap, cap->mseq,
+					oldest_flush_tid);
+		if (ret < 0) {
+			pr_err("__flush_snaps: error sending cap flushsnap, "
+			       "ino (%llx.%llx) tid %llu follows %llu\n",
+				ceph_vinop(inode), cf->tid, capsnap->follows);
+		}
+
+		ceph_put_cap_snap(capsnap);
+		spin_lock(&ci->i_ceph_lock);
+	}
+}
+
+void ceph_flush_snaps(struct ceph_inode_info *ci,
+		      struct ceph_mds_session **psession)
+{
+	struct inode *inode = &ci->netfs.inode;
+	struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;
+	struct ceph_mds_session *session = NULL;
+	bool need_put = false;
+	int mds;
+
+	dout("ceph_flush_snaps %p\n", inode);
+	if (psession)
+		session = *psession;
+retry:
+	spin_lock(&ci->i_ceph_lock);
+	if (!(ci->i_ceph_flags & CEPH_I_FLUSH_SNAPS)) {
+		dout(" no capsnap needs flush, doing nothing\n");
+		goto out;
+	}
+	if (!ci->i_auth_cap) {
+		dout(" no auth cap (migrating?), doing nothing\n");
+		goto out;
+	}
+
+	mds = ci->i_auth_cap->session->s_mds;
+	if (session && session->s_mds != mds) {
+		dout(" oops, wrong session %p mutex\n", session);
+		ceph_put_mds_session(session);
+		session = NULL;
+	}
+	if (!session) {
+		spin_unlock(&ci->i_ceph_lock);
+		mutex_lock(&mdsc->mutex);
+		session = __ceph_lookup_mds_session(mdsc, mds);
+		mutex_unlock(&mdsc->mutex);
+		goto retry;
+	}
+
+	// make sure flushsnap messages are sent in proper order.
+	if (ci->i_ceph_flags & CEPH_I_KICK_FLUSH)
+		__kick_flushing_caps(mdsc, session, ci, 0);
+
+	__ceph_flush_snaps(ci, session);
+out:
+	spin_unlock(&ci->i_ceph_lock);
+
+	if (psession)
+		*psession = session;
+	else
+		ceph_put_mds_session(session);
+	/* we flushed them all; remove this inode from the queue */
+	spin_lock(&mdsc->snap_flush_lock);
+	if (!list_empty(&ci->i_snap_flush_item))
+		need_put = true;
+	list_del_init(&ci->i_snap_flush_item);
+	spin_unlock(&mdsc->snap_flush_lock);
+
+	if (need_put)
+		iput(inode);
+}
+
+/*
+ * Mark caps dirty.  If inode is newly dirty, return the dirty flags.
+ * Caller is then responsible for calling __mark_inode_dirty with the
+ * returned flags value.
+ */
+int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask,
+			   struct ceph_cap_flush **pcf)
+{
+	struct ceph_mds_client *mdsc =
+		ceph_sb_to_client(ci->netfs.inode.i_sb)->mdsc;
+	struct inode *inode = &ci->netfs.inode;
+	int was = ci->i_dirty_caps;
+	int dirty = 0;
+
+	lockdep_assert_held(&ci->i_ceph_lock);
+
+	if (!ci->i_auth_cap) {
+		pr_warn("__mark_dirty_caps %p %llx mask %s, "
+			"but no auth cap (session was closed?)\n",
+			inode, ceph_ino(inode), ceph_cap_string(mask));
+		return 0;
+	}
+
+	dout("__mark_dirty_caps %p %s dirty %s -> %s\n", &ci->netfs.inode,
+	     ceph_cap_string(mask), ceph_cap_string(was),
+	     ceph_cap_string(was | mask));
+	ci->i_dirty_caps |= mask;
+	if (was == 0) {
+		struct ceph_mds_session *session = ci->i_auth_cap->session;
+
+		WARN_ON_ONCE(ci->i_prealloc_cap_flush);
+		swap(ci->i_prealloc_cap_flush, *pcf);
+
+		if (!ci->i_head_snapc) {
+			WARN_ON_ONCE(!rwsem_is_locked(&mdsc->snap_rwsem));
+			ci->i_head_snapc = ceph_get_snap_context(
+				ci->i_snap_realm->cached_context);
+		}
+		dout(" inode %p now dirty snapc %p auth cap %p\n",
+		     &ci->netfs.inode, ci->i_head_snapc, ci->i_auth_cap);
+		BUG_ON(!list_empty(&ci->i_dirty_item));
+		spin_lock(&mdsc->cap_dirty_lock);
+		list_add(&ci->i_dirty_item, &session->s_cap_dirty);
+		spin_unlock(&mdsc->cap_dirty_lock);
+		if (ci->i_flushing_caps == 0) {
+			ihold(inode);
+			dirty |= I_DIRTY_SYNC;
+		}
+	} else {
+		WARN_ON_ONCE(!ci->i_prealloc_cap_flush);
+	}
+	BUG_ON(list_empty(&ci->i_dirty_item));
+	if (((was | ci->i_flushing_caps) & CEPH_CAP_FILE_BUFFER) &&
+	    (mask & CEPH_CAP_FILE_BUFFER))
+		dirty |= I_DIRTY_DATASYNC;
+	__cap_delay_requeue(mdsc, ci);
+	return dirty;
+}
+
+struct ceph_cap_flush *ceph_alloc_cap_flush(void)
+{
+	struct ceph_cap_flush *cf;
+
+	cf = kmem_cache_alloc(ceph_cap_flush_cachep, GFP_KERNEL);
+	if (!cf)
+		return NULL;
+
+	cf->is_capsnap = false;
+	return cf;
+}
+
+void ceph_free_cap_flush(struct ceph_cap_flush *cf)
+{
+	if (cf)
+		kmem_cache_free(ceph_cap_flush_cachep, cf);
+}
+
+static u64 __get_oldest_flush_tid(struct ceph_mds_client *mdsc)
+{
+	if (!list_empty(&mdsc->cap_flush_list)) {
+		struct ceph_cap_flush *cf =
+			list_first_entry(&mdsc->cap_flush_list,
+					 struct ceph_cap_flush, g_list);
+		return cf->tid;
+	}
+	return 0;
+}
+
+/*
+ * Remove cap_flush from the mdsc's or inode's flushing cap list.
+ * Return true if caller needs to wake up flush waiters.
+ */
+static bool __detach_cap_flush_from_mdsc(struct ceph_mds_client *mdsc,
+					 struct ceph_cap_flush *cf)
+{
+	struct ceph_cap_flush *prev;
+	bool wake = cf->wake;
+
+	if (wake && cf->g_list.prev != &mdsc->cap_flush_list) {
+		prev = list_prev_entry(cf, g_list);
+		prev->wake = true;
+		wake = false;
+	}
+	list_del_init(&cf->g_list);
+	return wake;
+}
+
+static bool __detach_cap_flush_from_ci(struct ceph_inode_info *ci,
+				       struct ceph_cap_flush *cf)
+{
+	struct ceph_cap_flush *prev;
+	bool wake = cf->wake;
+
+	if (wake && cf->i_list.prev != &ci->i_cap_flush_list) {
+		prev = list_prev_entry(cf, i_list);
+		prev->wake = true;
+		wake = false;
+	}
+	list_del_init(&cf->i_list);
+	return wake;
+}
+
+/*
+ * Add dirty inode to the flushing list.  Assigned a seq number so we
+ * can wait for caps to flush without starving.
+ *
+ * Called under i_ceph_lock. Returns the flush tid.
+ */
+static u64 __mark_caps_flushing(struct inode *inode,
+				struct ceph_mds_session *session, bool wake,
+				u64 *oldest_flush_tid)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_cap_flush *cf = NULL;
+	int flushing;
+
+	lockdep_assert_held(&ci->i_ceph_lock);
+	BUG_ON(ci->i_dirty_caps == 0);
+	BUG_ON(list_empty(&ci->i_dirty_item));
+	BUG_ON(!ci->i_prealloc_cap_flush);
+
+	flushing = ci->i_dirty_caps;
+	dout("__mark_caps_flushing flushing %s, flushing_caps %s -> %s\n",
+	     ceph_cap_string(flushing),
+	     ceph_cap_string(ci->i_flushing_caps),
+	     ceph_cap_string(ci->i_flushing_caps | flushing));
+	ci->i_flushing_caps |= flushing;
+	ci->i_dirty_caps = 0;
+	dout(" inode %p now !dirty\n", inode);
+
+	swap(cf, ci->i_prealloc_cap_flush);
+	cf->caps = flushing;
+	cf->wake = wake;
+
+	spin_lock(&mdsc->cap_dirty_lock);
+	list_del_init(&ci->i_dirty_item);
+
+	cf->tid = ++mdsc->last_cap_flush_tid;
+	list_add_tail(&cf->g_list, &mdsc->cap_flush_list);
+	*oldest_flush_tid = __get_oldest_flush_tid(mdsc);
+
+	if (list_empty(&ci->i_flushing_item)) {
+		list_add_tail(&ci->i_flushing_item, &session->s_cap_flushing);
+		mdsc->num_cap_flushing++;
+	}
+	spin_unlock(&mdsc->cap_dirty_lock);
+
+	list_add_tail(&cf->i_list, &ci->i_cap_flush_list);
+
+	return cf->tid;
+}
+
+/*
+ * try to invalidate mapping pages without blocking.
+ */
+static int try_nonblocking_invalidate(struct inode *inode)
+	__releases(ci->i_ceph_lock)
+	__acquires(ci->i_ceph_lock)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	u32 invalidating_gen = ci->i_rdcache_gen;
+
+	spin_unlock(&ci->i_ceph_lock);
+	ceph_fscache_invalidate(inode, false);
+	invalidate_mapping_pages(&inode->i_data, 0, -1);
+	spin_lock(&ci->i_ceph_lock);
+
+	if (inode->i_data.nrpages == 0 &&
+	    invalidating_gen == ci->i_rdcache_gen) {
+		/* success. */
+		dout("try_nonblocking_invalidate %p success\n", inode);
+		/* save any racing async invalidate some trouble */
+		ci->i_rdcache_revoking = ci->i_rdcache_gen - 1;
+		return 0;
+	}
+	dout("try_nonblocking_invalidate %p failed\n", inode);
+	return -1;
+}
+
+bool __ceph_should_report_size(struct ceph_inode_info *ci)
+{
+	loff_t size = i_size_read(&ci->netfs.inode);
+	/* mds will adjust max size according to the reported size */
+	if (ci->i_flushing_caps & CEPH_CAP_FILE_WR)
+		return false;
+	if (size >= ci->i_max_size)
+		return true;
+	/* half of previous max_size increment has been used */
+	if (ci->i_max_size > ci->i_reported_size &&
+	    (size << 1) >= ci->i_max_size + ci->i_reported_size)
+		return true;
+	return false;
+}
+
+/*
+ * Swiss army knife function to examine currently used and wanted
+ * versus held caps.  Release, flush, ack revoked caps to mds as
+ * appropriate.
+ *
+ *  CHECK_CAPS_AUTHONLY - we should only check the auth cap
+ *  CHECK_CAPS_FLUSH - we should flush any dirty caps immediately, without
+ *    further delay.
+ */
+void ceph_check_caps(struct ceph_inode_info *ci, int flags,
+		     struct ceph_mds_session *session)
+{
+	struct inode *inode = &ci->netfs.inode;
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
+	struct ceph_cap *cap;
+	u64 flush_tid, oldest_flush_tid;
+	int file_wanted, used, cap_used;
+	int issued, implemented, want, retain, revoking, flushing = 0;
+	int mds = -1;   /* keep track of how far we've gone through i_caps list
+			   to avoid an infinite loop on retry */
+	struct rb_node *p;
+	bool queue_invalidate = false;
+	bool tried_invalidate = false;
+	bool queue_writeback = false;
+
+	if (session)
+		ceph_get_mds_session(session);
+
+	spin_lock(&ci->i_ceph_lock);
+	if (ci->i_ceph_flags & CEPH_I_ASYNC_CREATE) {
+		/* Don't send messages until we get async create reply */
+		spin_unlock(&ci->i_ceph_lock);
+		ceph_put_mds_session(session);
+		return;
+	}
+
+	if (ci->i_ceph_flags & CEPH_I_FLUSH)
+		flags |= CHECK_CAPS_FLUSH;
+retry:
+	/* Caps wanted by virtue of active open files. */
+	file_wanted = __ceph_caps_file_wanted(ci);
+
+	/* Caps which have active references against them */
+	used = __ceph_caps_used(ci);
+
+	/*
+	 * "issued" represents the current caps that the MDS wants us to have.
+	 * "implemented" is the set that we have been granted, and includes the
+	 * ones that have not yet been returned to the MDS (the "revoking" set,
+	 * usually because they have outstanding references).
+	 */
+	issued = __ceph_caps_issued(ci, &implemented);
+	revoking = implemented & ~issued;
+
+	want = file_wanted;
+
+	/* The ones we currently want to retain (may be adjusted below) */
+	retain = file_wanted | used | CEPH_CAP_PIN;
+	if (!mdsc->stopping && inode->i_nlink > 0) {
+		if (file_wanted) {
+			retain |= CEPH_CAP_ANY;       /* be greedy */
+		} else if (S_ISDIR(inode->i_mode) &&
+			   (issued & CEPH_CAP_FILE_SHARED) &&
+			   __ceph_dir_is_complete(ci)) {
+			/*
+			 * If a directory is complete, we want to keep
+			 * the exclusive cap. So that MDS does not end up
+			 * revoking the shared cap on every create/unlink
+			 * operation.
+			 */
+			if (IS_RDONLY(inode)) {
+				want = CEPH_CAP_ANY_SHARED;
+			} else {
+				want |= CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_EXCL;
+			}
+			retain |= want;
+		} else {
+
+			retain |= CEPH_CAP_ANY_SHARED;
+			/*
+			 * keep RD only if we didn't have the file open RW,
+			 * because then the mds would revoke it anyway to
+			 * journal max_size=0.
+			 */
+			if (ci->i_max_size == 0)
+				retain |= CEPH_CAP_ANY_RD;
+		}
+	}
+
+	dout("check_caps %llx.%llx file_want %s used %s dirty %s flushing %s"
+	     " issued %s revoking %s retain %s %s%s%s\n", ceph_vinop(inode),
+	     ceph_cap_string(file_wanted),
+	     ceph_cap_string(used), ceph_cap_string(ci->i_dirty_caps),
+	     ceph_cap_string(ci->i_flushing_caps),
+	     ceph_cap_string(issued), ceph_cap_string(revoking),
+	     ceph_cap_string(retain),
+	     (flags & CHECK_CAPS_AUTHONLY) ? " AUTHONLY" : "",
+	     (flags & CHECK_CAPS_FLUSH) ? " FLUSH" : "",
+	     (flags & CHECK_CAPS_NOINVAL) ? " NOINVAL" : "");
+
+	/*
+	 * If we no longer need to hold onto old our caps, and we may
+	 * have cached pages, but don't want them, then try to invalidate.
+	 * If we fail, it's because pages are locked.... try again later.
+	 */
+	if ((!(flags & CHECK_CAPS_NOINVAL) || mdsc->stopping) &&
+	    S_ISREG(inode->i_mode) &&
+	    !(ci->i_wb_ref || ci->i_wrbuffer_ref) &&   /* no dirty pages... */
+	    inode->i_data.nrpages &&		/* have cached pages */
+	    (revoking & (CEPH_CAP_FILE_CACHE|
+			 CEPH_CAP_FILE_LAZYIO)) && /*  or revoking cache */
+	    !tried_invalidate) {
+		dout("check_caps trying to invalidate on %llx.%llx\n",
+		     ceph_vinop(inode));
+		if (try_nonblocking_invalidate(inode) < 0) {
+			dout("check_caps queuing invalidate\n");
+			queue_invalidate = true;
+			ci->i_rdcache_revoking = ci->i_rdcache_gen;
+		}
+		tried_invalidate = true;
+		goto retry;
+	}
+
+	for (p = rb_first(&ci->i_caps); p; p = rb_next(p)) {
+		int mflags = 0;
+		struct cap_msg_args arg;
+
+		cap = rb_entry(p, struct ceph_cap, ci_node);
+
+		/* avoid looping forever */
+		if (mds >= cap->mds ||
+		    ((flags & CHECK_CAPS_AUTHONLY) && cap != ci->i_auth_cap))
+			continue;
+
+		/*
+		 * If we have an auth cap, we don't need to consider any
+		 * overlapping caps as used.
+		 */
+		cap_used = used;
+		if (ci->i_auth_cap && cap != ci->i_auth_cap)
+			cap_used &= ~ci->i_auth_cap->issued;
+
+		revoking = cap->implemented & ~cap->issued;
+		dout(" mds%d cap %p used %s issued %s implemented %s revoking %s\n",
+		     cap->mds, cap, ceph_cap_string(cap_used),
+		     ceph_cap_string(cap->issued),
+		     ceph_cap_string(cap->implemented),
+		     ceph_cap_string(revoking));
+
+		if (cap == ci->i_auth_cap &&
+		    (cap->issued & CEPH_CAP_FILE_WR)) {
+			/* request larger max_size from MDS? */
+			if (ci->i_wanted_max_size > ci->i_max_size &&
+			    ci->i_wanted_max_size > ci->i_requested_max_size) {
+				dout("requesting new max_size\n");
+				goto ack;
+			}
+
+			/* approaching file_max? */
+			if (__ceph_should_report_size(ci)) {
+				dout("i_size approaching max_size\n");
+				goto ack;
+			}
+		}
+		/* flush anything dirty? */
+		if (cap == ci->i_auth_cap) {
+			if ((flags & CHECK_CAPS_FLUSH) && ci->i_dirty_caps) {
+				dout("flushing dirty caps\n");
+				goto ack;
+			}
+			if (ci->i_ceph_flags & CEPH_I_FLUSH_SNAPS) {
+				dout("flushing snap caps\n");
+				goto ack;
+			}
+		}
+
+		/* completed revocation? going down and there are no caps? */
+		if (revoking) {
+			if ((revoking & cap_used) == 0) {
+				dout("completed revocation of %s\n",
+				      ceph_cap_string(cap->implemented & ~cap->issued));
+				goto ack;
+			}
+
+			/*
+			 * If the "i_wrbuffer_ref" was increased by mmap or generic
+			 * cache write just before the ceph_check_caps() is called,
+			 * the Fb capability revoking will fail this time. Then we
+			 * must wait for the BDI's delayed work to flush the dirty
+			 * pages and to release the "i_wrbuffer_ref", which will cost
+			 * at most 5 seconds. That means the MDS needs to wait at
+			 * most 5 seconds to finished the Fb capability's revocation.
+			 *
+			 * Let's queue a writeback for it.
+			 */
+			if (S_ISREG(inode->i_mode) && ci->i_wrbuffer_ref &&
+			    (revoking & CEPH_CAP_FILE_BUFFER))
+				queue_writeback = true;
+		}
+
+		/* want more caps from mds? */
+		if (want & ~cap->mds_wanted) {
+			if (want & ~(cap->mds_wanted | cap->issued))
+				goto ack;
+			if (!__cap_is_valid(cap))
+				goto ack;
+		}
+
+		/* things we might delay */
+		if ((cap->issued & ~retain) == 0)
+			continue;     /* nope, all good */
+
+ack:
+		ceph_put_mds_session(session);
+		session = ceph_get_mds_session(cap->session);
+
+		/* kick flushing and flush snaps before sending normal
+		 * cap message */
+		if (cap == ci->i_auth_cap &&
+		    (ci->i_ceph_flags &
+		     (CEPH_I_KICK_FLUSH | CEPH_I_FLUSH_SNAPS))) {
+			if (ci->i_ceph_flags & CEPH_I_KICK_FLUSH)
+				__kick_flushing_caps(mdsc, session, ci, 0);
+			if (ci->i_ceph_flags & CEPH_I_FLUSH_SNAPS)
+				__ceph_flush_snaps(ci, session);
+
+			goto retry;
+		}
+
+		if (cap == ci->i_auth_cap && ci->i_dirty_caps) {
+			flushing = ci->i_dirty_caps;
+			flush_tid = __mark_caps_flushing(inode, session, false,
+							 &oldest_flush_tid);
+			if (flags & CHECK_CAPS_FLUSH &&
+			    list_empty(&session->s_cap_dirty))
+				mflags |= CEPH_CLIENT_CAPS_SYNC;
+		} else {
+			flushing = 0;
+			flush_tid = 0;
+			spin_lock(&mdsc->cap_dirty_lock);
+			oldest_flush_tid = __get_oldest_flush_tid(mdsc);
+			spin_unlock(&mdsc->cap_dirty_lock);
+		}
+
+		mds = cap->mds;  /* remember mds, so we don't repeat */
+
+		__prep_cap(&arg, cap, CEPH_CAP_OP_UPDATE, mflags, cap_used,
+			   want, retain, flushing, flush_tid, oldest_flush_tid);
+
+		spin_unlock(&ci->i_ceph_lock);
+		__send_cap(&arg, ci);
+		spin_lock(&ci->i_ceph_lock);
+
+		goto retry; /* retake i_ceph_lock and restart our cap scan. */
+	}
+
+	/* periodically re-calculate caps wanted by open files */
+	if (__ceph_is_any_real_caps(ci) &&
+	    list_empty(&ci->i_cap_delay_list) &&
+	    (file_wanted & ~CEPH_CAP_PIN) &&
+	    !(used & (CEPH_CAP_FILE_RD | CEPH_CAP_ANY_FILE_WR))) {
+		__cap_delay_requeue(mdsc, ci);
+	}
+
+	spin_unlock(&ci->i_ceph_lock);
+
+	ceph_put_mds_session(session);
+	if (queue_writeback)
+		ceph_queue_writeback(inode);
+	if (queue_invalidate)
+		ceph_queue_invalidate(inode);
+}
+
+/*
+ * Try to flush dirty caps back to the auth mds.
+ */
+static int try_flush_caps(struct inode *inode, u64 *ptid)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	int flushing = 0;
+	u64 flush_tid = 0, oldest_flush_tid = 0;
+
+	spin_lock(&ci->i_ceph_lock);
+retry_locked:
+	if (ci->i_dirty_caps && ci->i_auth_cap) {
+		struct ceph_cap *cap = ci->i_auth_cap;
+		struct cap_msg_args arg;
+		struct ceph_mds_session *session = cap->session;
+
+		if (session->s_state < CEPH_MDS_SESSION_OPEN) {
+			spin_unlock(&ci->i_ceph_lock);
+			goto out;
+		}
+
+		if (ci->i_ceph_flags &
+		    (CEPH_I_KICK_FLUSH | CEPH_I_FLUSH_SNAPS)) {
+			if (ci->i_ceph_flags & CEPH_I_KICK_FLUSH)
+				__kick_flushing_caps(mdsc, session, ci, 0);
+			if (ci->i_ceph_flags & CEPH_I_FLUSH_SNAPS)
+				__ceph_flush_snaps(ci, session);
+			goto retry_locked;
+		}
+
+		flushing = ci->i_dirty_caps;
+		flush_tid = __mark_caps_flushing(inode, session, true,
+						 &oldest_flush_tid);
+
+		__prep_cap(&arg, cap, CEPH_CAP_OP_FLUSH, CEPH_CLIENT_CAPS_SYNC,
+			   __ceph_caps_used(ci), __ceph_caps_wanted(ci),
+			   (cap->issued | cap->implemented),
+			   flushing, flush_tid, oldest_flush_tid);
+		spin_unlock(&ci->i_ceph_lock);
+
+		__send_cap(&arg, ci);
+	} else {
+		if (!list_empty(&ci->i_cap_flush_list)) {
+			struct ceph_cap_flush *cf =
+				list_last_entry(&ci->i_cap_flush_list,
+						struct ceph_cap_flush, i_list);
+			cf->wake = true;
+			flush_tid = cf->tid;
+		}
+		flushing = ci->i_flushing_caps;
+		spin_unlock(&ci->i_ceph_lock);
+	}
+out:
+	*ptid = flush_tid;
+	return flushing;
+}
+
+/*
+ * Return true if we've flushed caps through the given flush_tid.
+ */
+static int caps_are_flushed(struct inode *inode, u64 flush_tid)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	int ret = 1;
+
+	spin_lock(&ci->i_ceph_lock);
+	if (!list_empty(&ci->i_cap_flush_list)) {
+		struct ceph_cap_flush * cf =
+			list_first_entry(&ci->i_cap_flush_list,
+					 struct ceph_cap_flush, i_list);
+		if (cf->tid <= flush_tid)
+			ret = 0;
+	}
+	spin_unlock(&ci->i_ceph_lock);
+	return ret;
+}
+
+/*
+ * flush the mdlog and wait for any unsafe requests to complete.
+ */
+static int flush_mdlog_and_wait_inode_unsafe_requests(struct inode *inode)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_mds_request *req1 = NULL, *req2 = NULL;
+	int ret, err = 0;
+
+	spin_lock(&ci->i_unsafe_lock);
+	if (S_ISDIR(inode->i_mode) && !list_empty(&ci->i_unsafe_dirops)) {
+		req1 = list_last_entry(&ci->i_unsafe_dirops,
+					struct ceph_mds_request,
+					r_unsafe_dir_item);
+		ceph_mdsc_get_request(req1);
+	}
+	if (!list_empty(&ci->i_unsafe_iops)) {
+		req2 = list_last_entry(&ci->i_unsafe_iops,
+					struct ceph_mds_request,
+					r_unsafe_target_item);
+		ceph_mdsc_get_request(req2);
+	}
+	spin_unlock(&ci->i_unsafe_lock);
+
+	/*
+	 * Trigger to flush the journal logs in all the relevant MDSes
+	 * manually, or in the worst case we must wait at most 5 seconds
+	 * to wait the journal logs to be flushed by the MDSes periodically.
+	 */
+	if (req1 || req2) {
+		struct ceph_mds_request *req;
+		struct ceph_mds_session **sessions;
+		struct ceph_mds_session *s;
+		unsigned int max_sessions;
+		int i;
+
+		mutex_lock(&mdsc->mutex);
+		max_sessions = mdsc->max_sessions;
+
+		sessions = kcalloc(max_sessions, sizeof(s), GFP_KERNEL);
+		if (!sessions) {
+			mutex_unlock(&mdsc->mutex);
+			err = -ENOMEM;
+			goto out;
+		}
+
+		spin_lock(&ci->i_unsafe_lock);
+		if (req1) {
+			list_for_each_entry(req, &ci->i_unsafe_dirops,
+					    r_unsafe_dir_item) {
+				s = req->r_session;
+				if (!s)
+					continue;
+				if (!sessions[s->s_mds]) {
+					s = ceph_get_mds_session(s);
+					sessions[s->s_mds] = s;
+				}
+			}
+		}
+		if (req2) {
+			list_for_each_entry(req, &ci->i_unsafe_iops,
+					    r_unsafe_target_item) {
+				s = req->r_session;
+				if (!s)
+					continue;
+				if (!sessions[s->s_mds]) {
+					s = ceph_get_mds_session(s);
+					sessions[s->s_mds] = s;
+				}
+			}
+		}
+		spin_unlock(&ci->i_unsafe_lock);
+
+		/* the auth MDS */
+		spin_lock(&ci->i_ceph_lock);
+		if (ci->i_auth_cap) {
+			s = ci->i_auth_cap->session;
+			if (!sessions[s->s_mds])
+				sessions[s->s_mds] = ceph_get_mds_session(s);
+		}
+		spin_unlock(&ci->i_ceph_lock);
+		mutex_unlock(&mdsc->mutex);
+
+		/* send flush mdlog request to MDSes */
+		for (i = 0; i < max_sessions; i++) {
+			s = sessions[i];
+			if (s) {
+				send_flush_mdlog(s);
+				ceph_put_mds_session(s);
+			}
+		}
+		kfree(sessions);
+	}
+
+	dout("%s %p wait on tid %llu %llu\n", __func__,
+	     inode, req1 ? req1->r_tid : 0ULL, req2 ? req2->r_tid : 0ULL);
+	if (req1) {
+		ret = !wait_for_completion_timeout(&req1->r_safe_completion,
+					ceph_timeout_jiffies(req1->r_timeout));
+		if (ret)
+			err = -EIO;
+	}
+	if (req2) {
+		ret = !wait_for_completion_timeout(&req2->r_safe_completion,
+					ceph_timeout_jiffies(req2->r_timeout));
+		if (ret)
+			err = -EIO;
+	}
+
+out:
+	if (req1)
+		ceph_mdsc_put_request(req1);
+	if (req2)
+		ceph_mdsc_put_request(req2);
+	return err;
+}
+
+int ceph_fsync(struct file *file, loff_t start, loff_t end, int datasync)
+{
+	struct inode *inode = file->f_mapping->host;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	u64 flush_tid;
+	int ret, err;
+	int dirty;
+
+	dout("fsync %p%s\n", inode, datasync ? " datasync" : "");
+
+	ret = file_write_and_wait_range(file, start, end);
+	if (datasync)
+		goto out;
+
+	ret = ceph_wait_on_async_create(inode);
+	if (ret)
+		goto out;
+
+	dirty = try_flush_caps(inode, &flush_tid);
+	dout("fsync dirty caps are %s\n", ceph_cap_string(dirty));
+
+	err = flush_mdlog_and_wait_inode_unsafe_requests(inode);
+
+	/*
+	 * only wait on non-file metadata writeback (the mds
+	 * can recover size and mtime, so we don't need to
+	 * wait for that)
+	 */
+	if (!err && (dirty & ~CEPH_CAP_ANY_FILE_WR)) {
+		err = wait_event_interruptible(ci->i_cap_wq,
+					caps_are_flushed(inode, flush_tid));
+	}
+
+	if (err < 0)
+		ret = err;
+
+	err = file_check_and_advance_wb_err(file);
+	if (err < 0)
+		ret = err;
+out:
+	dout("fsync %p%s result=%d\n", inode, datasync ? " datasync" : "", ret);
+	return ret;
+}
+
+/*
+ * Flush any dirty caps back to the mds.  If we aren't asked to wait,
+ * queue inode for flush but don't do so immediately, because we can
+ * get by with fewer MDS messages if we wait for data writeback to
+ * complete first.
+ */
+int ceph_write_inode(struct inode *inode, struct writeback_control *wbc)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	u64 flush_tid;
+	int err = 0;
+	int dirty;
+	int wait = (wbc->sync_mode == WB_SYNC_ALL && !wbc->for_sync);
+
+	dout("write_inode %p wait=%d\n", inode, wait);
+	ceph_fscache_unpin_writeback(inode, wbc);
+	if (wait) {
+		err = ceph_wait_on_async_create(inode);
+		if (err)
+			return err;
+		dirty = try_flush_caps(inode, &flush_tid);
+		if (dirty)
+			err = wait_event_interruptible(ci->i_cap_wq,
+				       caps_are_flushed(inode, flush_tid));
+	} else {
+		struct ceph_mds_client *mdsc =
+			ceph_sb_to_client(inode->i_sb)->mdsc;
+
+		spin_lock(&ci->i_ceph_lock);
+		if (__ceph_caps_dirty(ci))
+			__cap_delay_requeue_front(mdsc, ci);
+		spin_unlock(&ci->i_ceph_lock);
+	}
+	return err;
+}
+
+static void __kick_flushing_caps(struct ceph_mds_client *mdsc,
+				 struct ceph_mds_session *session,
+				 struct ceph_inode_info *ci,
+				 u64 oldest_flush_tid)
+	__releases(ci->i_ceph_lock)
+	__acquires(ci->i_ceph_lock)
+{
+	struct inode *inode = &ci->netfs.inode;
+	struct ceph_cap *cap;
+	struct ceph_cap_flush *cf;
+	int ret;
+	u64 first_tid = 0;
+	u64 last_snap_flush = 0;
+
+	/* Don't do anything until create reply comes in */
+	if (ci->i_ceph_flags & CEPH_I_ASYNC_CREATE)
+		return;
+
+	ci->i_ceph_flags &= ~CEPH_I_KICK_FLUSH;
+
+	list_for_each_entry_reverse(cf, &ci->i_cap_flush_list, i_list) {
+		if (cf->is_capsnap) {
+			last_snap_flush = cf->tid;
+			break;
+		}
+	}
+
+	list_for_each_entry(cf, &ci->i_cap_flush_list, i_list) {
+		if (cf->tid < first_tid)
+			continue;
+
+		cap = ci->i_auth_cap;
+		if (!(cap && cap->session == session)) {
+			pr_err("%p auth cap %p not mds%d ???\n",
+			       inode, cap, session->s_mds);
+			break;
+		}
+
+		first_tid = cf->tid + 1;
+
+		if (!cf->is_capsnap) {
+			struct cap_msg_args arg;
+
+			dout("kick_flushing_caps %p cap %p tid %llu %s\n",
+			     inode, cap, cf->tid, ceph_cap_string(cf->caps));
+			__prep_cap(&arg, cap, CEPH_CAP_OP_FLUSH,
+					 (cf->tid < last_snap_flush ?
+					  CEPH_CLIENT_CAPS_PENDING_CAPSNAP : 0),
+					  __ceph_caps_used(ci),
+					  __ceph_caps_wanted(ci),
+					  (cap->issued | cap->implemented),
+					  cf->caps, cf->tid, oldest_flush_tid);
+			spin_unlock(&ci->i_ceph_lock);
+			__send_cap(&arg, ci);
+		} else {
+			struct ceph_cap_snap *capsnap =
+					container_of(cf, struct ceph_cap_snap,
+						    cap_flush);
+			dout("kick_flushing_caps %p capsnap %p tid %llu %s\n",
+			     inode, capsnap, cf->tid,
+			     ceph_cap_string(capsnap->dirty));
+
+			refcount_inc(&capsnap->nref);
+			spin_unlock(&ci->i_ceph_lock);
+
+			ret = __send_flush_snap(inode, session, capsnap, cap->mseq,
+						oldest_flush_tid);
+			if (ret < 0) {
+				pr_err("kick_flushing_caps: error sending "
+					"cap flushsnap, ino (%llx.%llx) "
+					"tid %llu follows %llu\n",
+					ceph_vinop(inode), cf->tid,
+					capsnap->follows);
+			}
+
+			ceph_put_cap_snap(capsnap);
+		}
+
+		spin_lock(&ci->i_ceph_lock);
+	}
+}
+
+void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc,
+				   struct ceph_mds_session *session)
+{
+	struct ceph_inode_info *ci;
+	struct ceph_cap *cap;
+	u64 oldest_flush_tid;
+
+	dout("early_kick_flushing_caps mds%d\n", session->s_mds);
+
+	spin_lock(&mdsc->cap_dirty_lock);
+	oldest_flush_tid = __get_oldest_flush_tid(mdsc);
+	spin_unlock(&mdsc->cap_dirty_lock);
+
+	list_for_each_entry(ci, &session->s_cap_flushing, i_flushing_item) {
+		spin_lock(&ci->i_ceph_lock);
+		cap = ci->i_auth_cap;
+		if (!(cap && cap->session == session)) {
+			pr_err("%p auth cap %p not mds%d ???\n",
+				&ci->netfs.inode, cap, session->s_mds);
+			spin_unlock(&ci->i_ceph_lock);
+			continue;
+		}
+
+
+		/*
+		 * if flushing caps were revoked, we re-send the cap flush
+		 * in client reconnect stage. This guarantees MDS * processes
+		 * the cap flush message before issuing the flushing caps to
+		 * other client.
+		 */
+		if ((cap->issued & ci->i_flushing_caps) !=
+		    ci->i_flushing_caps) {
+			/* encode_caps_cb() also will reset these sequence
+			 * numbers. make sure sequence numbers in cap flush
+			 * message match later reconnect message */
+			cap->seq = 0;
+			cap->issue_seq = 0;
+			cap->mseq = 0;
+			__kick_flushing_caps(mdsc, session, ci,
+					     oldest_flush_tid);
+		} else {
+			ci->i_ceph_flags |= CEPH_I_KICK_FLUSH;
+		}
+
+		spin_unlock(&ci->i_ceph_lock);
+	}
+}
+
+void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
+			     struct ceph_mds_session *session)
+{
+	struct ceph_inode_info *ci;
+	struct ceph_cap *cap;
+	u64 oldest_flush_tid;
+
+	lockdep_assert_held(&session->s_mutex);
+
+	dout("kick_flushing_caps mds%d\n", session->s_mds);
+
+	spin_lock(&mdsc->cap_dirty_lock);
+	oldest_flush_tid = __get_oldest_flush_tid(mdsc);
+	spin_unlock(&mdsc->cap_dirty_lock);
+
+	list_for_each_entry(ci, &session->s_cap_flushing, i_flushing_item) {
+		spin_lock(&ci->i_ceph_lock);
+		cap = ci->i_auth_cap;
+		if (!(cap && cap->session == session)) {
+			pr_err("%p auth cap %p not mds%d ???\n",
+				&ci->netfs.inode, cap, session->s_mds);
+			spin_unlock(&ci->i_ceph_lock);
+			continue;
+		}
+		if (ci->i_ceph_flags & CEPH_I_KICK_FLUSH) {
+			__kick_flushing_caps(mdsc, session, ci,
+					     oldest_flush_tid);
+		}
+		spin_unlock(&ci->i_ceph_lock);
+	}
+}
+
+void ceph_kick_flushing_inode_caps(struct ceph_mds_session *session,
+				   struct ceph_inode_info *ci)
+{
+	struct ceph_mds_client *mdsc = session->s_mdsc;
+	struct ceph_cap *cap = ci->i_auth_cap;
+
+	lockdep_assert_held(&ci->i_ceph_lock);
+
+	dout("%s %p flushing %s\n", __func__, &ci->netfs.inode,
+	     ceph_cap_string(ci->i_flushing_caps));
+
+	if (!list_empty(&ci->i_cap_flush_list)) {
+		u64 oldest_flush_tid;
+		spin_lock(&mdsc->cap_dirty_lock);
+		list_move_tail(&ci->i_flushing_item,
+			       &cap->session->s_cap_flushing);
+		oldest_flush_tid = __get_oldest_flush_tid(mdsc);
+		spin_unlock(&mdsc->cap_dirty_lock);
+
+		__kick_flushing_caps(mdsc, session, ci, oldest_flush_tid);
+	}
+}
+
+
+/*
+ * Take references to capabilities we hold, so that we don't release
+ * them to the MDS prematurely.
+ */
+void ceph_take_cap_refs(struct ceph_inode_info *ci, int got,
+			    bool snap_rwsem_locked)
+{
+	lockdep_assert_held(&ci->i_ceph_lock);
+
+	if (got & CEPH_CAP_PIN)
+		ci->i_pin_ref++;
+	if (got & CEPH_CAP_FILE_RD)
+		ci->i_rd_ref++;
+	if (got & CEPH_CAP_FILE_CACHE)
+		ci->i_rdcache_ref++;
+	if (got & CEPH_CAP_FILE_EXCL)
+		ci->i_fx_ref++;
+	if (got & CEPH_CAP_FILE_WR) {
+		if (ci->i_wr_ref == 0 && !ci->i_head_snapc) {
+			BUG_ON(!snap_rwsem_locked);
+			ci->i_head_snapc = ceph_get_snap_context(
+					ci->i_snap_realm->cached_context);
+		}
+		ci->i_wr_ref++;
+	}
+	if (got & CEPH_CAP_FILE_BUFFER) {
+		if (ci->i_wb_ref == 0)
+			ihold(&ci->netfs.inode);
+		ci->i_wb_ref++;
+		dout("%s %p wb %d -> %d (?)\n", __func__,
+		     &ci->netfs.inode, ci->i_wb_ref-1, ci->i_wb_ref);
+	}
+}
+
+/*
+ * Try to grab cap references.  Specify those refs we @want, and the
+ * minimal set we @need.  Also include the larger offset we are writing
+ * to (when applicable), and check against max_size here as well.
+ * Note that caller is responsible for ensuring max_size increases are
+ * requested from the MDS.
+ *
+ * Returns 0 if caps were not able to be acquired (yet), 1 if succeed,
+ * or a negative error code. There are 3 speical error codes:
+ *  -EAGAIN:  need to sleep but non-blocking is specified
+ *  -EFBIG:   ask caller to call check_max_size() and try again.
+ *  -EUCLEAN: ask caller to call ceph_renew_caps() and try again.
+ */
+enum {
+	/* first 8 bits are reserved for CEPH_FILE_MODE_FOO */
+	NON_BLOCKING	= (1 << 8),
+	CHECK_FILELOCK	= (1 << 9),
+};
+
+static int try_get_cap_refs(struct inode *inode, int need, int want,
+			    loff_t endoff, int flags, int *got)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;
+	int ret = 0;
+	int have, implemented;
+	bool snap_rwsem_locked = false;
+
+	dout("get_cap_refs %p need %s want %s\n", inode,
+	     ceph_cap_string(need), ceph_cap_string(want));
+
+again:
+	spin_lock(&ci->i_ceph_lock);
+
+	if ((flags & CHECK_FILELOCK) &&
+	    (ci->i_ceph_flags & CEPH_I_ERROR_FILELOCK)) {
+		dout("try_get_cap_refs %p error filelock\n", inode);
+		ret = -EIO;
+		goto out_unlock;
+	}
+
+	/* finish pending truncate */
+	while (ci->i_truncate_pending) {
+		spin_unlock(&ci->i_ceph_lock);
+		if (snap_rwsem_locked) {
+			up_read(&mdsc->snap_rwsem);
+			snap_rwsem_locked = false;
+		}
+		__ceph_do_pending_vmtruncate(inode);
+		spin_lock(&ci->i_ceph_lock);
+	}
+
+	have = __ceph_caps_issued(ci, &implemented);
+
+	if (have & need & CEPH_CAP_FILE_WR) {
+		if (endoff >= 0 && endoff > (loff_t)ci->i_max_size) {
+			dout("get_cap_refs %p endoff %llu > maxsize %llu\n",
+			     inode, endoff, ci->i_max_size);
+			if (endoff > ci->i_requested_max_size)
+				ret = ci->i_auth_cap ? -EFBIG : -EUCLEAN;
+			goto out_unlock;
+		}
+		/*
+		 * If a sync write is in progress, we must wait, so that we
+		 * can get a final snapshot value for size+mtime.
+		 */
+		if (__ceph_have_pending_cap_snap(ci)) {
+			dout("get_cap_refs %p cap_snap_pending\n", inode);
+			goto out_unlock;
+		}
+	}
+
+	if ((have & need) == need) {
+		/*
+		 * Look at (implemented & ~have & not) so that we keep waiting
+		 * on transition from wanted -> needed caps.  This is needed
+		 * for WRBUFFER|WR -> WR to avoid a new WR sync write from
+		 * going before a prior buffered writeback happens.
+		 *
+		 * For RDCACHE|RD -> RD, there is not need to wait and we can
+		 * just exclude the revoking caps and force to sync read.
+		 */
+		int not = want & ~(have & need);
+		int revoking = implemented & ~have;
+		int exclude = revoking & not;
+		dout("get_cap_refs %p have %s but not %s (revoking %s)\n",
+		     inode, ceph_cap_string(have), ceph_cap_string(not),
+		     ceph_cap_string(revoking));
+		if (!exclude || !(exclude & CEPH_CAP_FILE_BUFFER)) {
+			if (!snap_rwsem_locked &&
+			    !ci->i_head_snapc &&
+			    (need & CEPH_CAP_FILE_WR)) {
+				if (!down_read_trylock(&mdsc->snap_rwsem)) {
+					/*
+					 * we can not call down_read() when
+					 * task isn't in TASK_RUNNING state
+					 */
+					if (flags & NON_BLOCKING) {
+						ret = -EAGAIN;
+						goto out_unlock;
+					}
+
+					spin_unlock(&ci->i_ceph_lock);
+					down_read(&mdsc->snap_rwsem);
+					snap_rwsem_locked = true;
+					goto again;
+				}
+				snap_rwsem_locked = true;
+			}
+			if ((have & want) == want)
+				*got = need | (want & ~exclude);
+			else
+				*got = need;
+			ceph_take_cap_refs(ci, *got, true);
+			ret = 1;
+		}
+	} else {
+		int session_readonly = false;
+		int mds_wanted;
+		if (ci->i_auth_cap &&
+		    (need & (CEPH_CAP_FILE_WR | CEPH_CAP_FILE_EXCL))) {
+			struct ceph_mds_session *s = ci->i_auth_cap->session;
+			spin_lock(&s->s_cap_lock);
+			session_readonly = s->s_readonly;
+			spin_unlock(&s->s_cap_lock);
+		}
+		if (session_readonly) {
+			dout("get_cap_refs %p need %s but mds%d readonly\n",
+			     inode, ceph_cap_string(need), ci->i_auth_cap->mds);
+			ret = -EROFS;
+			goto out_unlock;
+		}
+
+		if (ceph_inode_is_shutdown(inode)) {
+			dout("get_cap_refs %p inode is shutdown\n", inode);
+			ret = -ESTALE;
+			goto out_unlock;
+		}
+		mds_wanted = __ceph_caps_mds_wanted(ci, false);
+		if (need & ~mds_wanted) {
+			dout("get_cap_refs %p need %s > mds_wanted %s\n",
+			     inode, ceph_cap_string(need),
+			     ceph_cap_string(mds_wanted));
+			ret = -EUCLEAN;
+			goto out_unlock;
+		}
+
+		dout("get_cap_refs %p have %s need %s\n", inode,
+		     ceph_cap_string(have), ceph_cap_string(need));
+	}
+out_unlock:
+
+	__ceph_touch_fmode(ci, mdsc, flags);
+
+	spin_unlock(&ci->i_ceph_lock);
+	if (snap_rwsem_locked)
+		up_read(&mdsc->snap_rwsem);
+
+	if (!ret)
+		ceph_update_cap_mis(&mdsc->metric);
+	else if (ret == 1)
+		ceph_update_cap_hit(&mdsc->metric);
+
+	dout("get_cap_refs %p ret %d got %s\n", inode,
+	     ret, ceph_cap_string(*got));
+	return ret;
+}
+
+/*
+ * Check the offset we are writing up to against our current
+ * max_size.  If necessary, tell the MDS we want to write to
+ * a larger offset.
+ */
+static void check_max_size(struct inode *inode, loff_t endoff)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	int check = 0;
+
+	/* do we need to explicitly request a larger max_size? */
+	spin_lock(&ci->i_ceph_lock);
+	if (endoff >= ci->i_max_size && endoff > ci->i_wanted_max_size) {
+		dout("write %p at large endoff %llu, req max_size\n",
+		     inode, endoff);
+		ci->i_wanted_max_size = endoff;
+	}
+	/* duplicate ceph_check_caps()'s logic */
+	if (ci->i_auth_cap &&
+	    (ci->i_auth_cap->issued & CEPH_CAP_FILE_WR) &&
+	    ci->i_wanted_max_size > ci->i_max_size &&
+	    ci->i_wanted_max_size > ci->i_requested_max_size)
+		check = 1;
+	spin_unlock(&ci->i_ceph_lock);
+	if (check)
+		ceph_check_caps(ci, CHECK_CAPS_AUTHONLY, NULL);
+}
+
+static inline int get_used_fmode(int caps)
+{
+	int fmode = 0;
+	if (caps & CEPH_CAP_FILE_RD)
+		fmode |= CEPH_FILE_MODE_RD;
+	if (caps & CEPH_CAP_FILE_WR)
+		fmode |= CEPH_FILE_MODE_WR;
+	return fmode;
+}
+
+int ceph_try_get_caps(struct inode *inode, int need, int want,
+		      bool nonblock, int *got)
+{
+	int ret, flags;
+
+	BUG_ON(need & ~CEPH_CAP_FILE_RD);
+	BUG_ON(want & ~(CEPH_CAP_FILE_CACHE | CEPH_CAP_FILE_LAZYIO |
+			CEPH_CAP_FILE_SHARED | CEPH_CAP_FILE_EXCL |
+			CEPH_CAP_ANY_DIR_OPS));
+	if (need) {
+		ret = ceph_pool_perm_check(inode, need);
+		if (ret < 0)
+			return ret;
+	}
+
+	flags = get_used_fmode(need | want);
+	if (nonblock)
+		flags |= NON_BLOCKING;
+
+	ret = try_get_cap_refs(inode, need, want, 0, flags, got);
+	/* three special error codes */
+	if (ret == -EAGAIN || ret == -EFBIG || ret == -EUCLEAN)
+		ret = 0;
+	return ret;
+}
+
+/*
+ * Wait for caps, and take cap references.  If we can't get a WR cap
+ * due to a small max_size, make sure we check_max_size (and possibly
+ * ask the mds) so we don't get hung up indefinitely.
+ */
+int ceph_get_caps(struct file *filp, int need, int want, loff_t endoff, int *got)
+{
+	struct ceph_file_info *fi = filp->private_data;
+	struct inode *inode = file_inode(filp);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	int ret, _got, flags;
+
+	ret = ceph_pool_perm_check(inode, need);
+	if (ret < 0)
+		return ret;
+
+	if ((fi->fmode & CEPH_FILE_MODE_WR) &&
+	    fi->filp_gen != READ_ONCE(fsc->filp_gen))
+		return -EBADF;
+
+	flags = get_used_fmode(need | want);
+
+	while (true) {
+		flags &= CEPH_FILE_MODE_MASK;
+		if (vfs_inode_has_locks(inode))
+			flags |= CHECK_FILELOCK;
+		_got = 0;
+		ret = try_get_cap_refs(inode, need, want, endoff,
+				       flags, &_got);
+		WARN_ON_ONCE(ret == -EAGAIN);
+		if (!ret) {
+			struct ceph_mds_client *mdsc = fsc->mdsc;
+			struct cap_wait cw;
+			DEFINE_WAIT_FUNC(wait, woken_wake_function);
+
+			cw.ino = ceph_ino(inode);
+			cw.tgid = current->tgid;
+			cw.need = need;
+			cw.want = want;
+
+			spin_lock(&mdsc->caps_list_lock);
+			list_add(&cw.list, &mdsc->cap_wait_list);
+			spin_unlock(&mdsc->caps_list_lock);
+
+			/* make sure used fmode not timeout */
+			ceph_get_fmode(ci, flags, FMODE_WAIT_BIAS);
+			add_wait_queue(&ci->i_cap_wq, &wait);
+
+			flags |= NON_BLOCKING;
+			while (!(ret = try_get_cap_refs(inode, need, want,
+							endoff, flags, &_got))) {
+				if (signal_pending(current)) {
+					ret = -ERESTARTSYS;
+					break;
+				}
+				wait_woken(&wait, TASK_INTERRUPTIBLE, MAX_SCHEDULE_TIMEOUT);
+			}
+
+			remove_wait_queue(&ci->i_cap_wq, &wait);
+			ceph_put_fmode(ci, flags, FMODE_WAIT_BIAS);
+
+			spin_lock(&mdsc->caps_list_lock);
+			list_del(&cw.list);
+			spin_unlock(&mdsc->caps_list_lock);
+
+			if (ret == -EAGAIN)
+				continue;
+		}
+
+		if ((fi->fmode & CEPH_FILE_MODE_WR) &&
+		    fi->filp_gen != READ_ONCE(fsc->filp_gen)) {
+			if (ret >= 0 && _got)
+				ceph_put_cap_refs(ci, _got);
+			return -EBADF;
+		}
+
+		if (ret < 0) {
+			if (ret == -EFBIG || ret == -EUCLEAN) {
+				int ret2 = ceph_wait_on_async_create(inode);
+				if (ret2 < 0)
+					return ret2;
+			}
+			if (ret == -EFBIG) {
+				check_max_size(inode, endoff);
+				continue;
+			}
+			if (ret == -EUCLEAN) {
+				/* session was killed, try renew caps */
+				ret = ceph_renew_caps(inode, flags);
+				if (ret == 0)
+					continue;
+			}
+			return ret;
+		}
+
+		if (S_ISREG(ci->netfs.inode.i_mode) &&
+		    ceph_has_inline_data(ci) &&
+		    (_got & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) &&
+		    i_size_read(inode) > 0) {
+			struct page *page =
+				find_get_page(inode->i_mapping, 0);
+			if (page) {
+				bool uptodate = PageUptodate(page);
+
+				put_page(page);
+				if (uptodate)
+					break;
+			}
+			/*
+			 * drop cap refs first because getattr while
+			 * holding * caps refs can cause deadlock.
+			 */
+			ceph_put_cap_refs(ci, _got);
+			_got = 0;
+
+			/*
+			 * getattr request will bring inline data into
+			 * page cache
+			 */
+			ret = __ceph_do_getattr(inode, NULL,
+						CEPH_STAT_CAP_INLINE_DATA,
+						true);
+			if (ret < 0)
+				return ret;
+			continue;
+		}
+		break;
+	}
+	*got = _got;
+	return 0;
+}
+
+/*
+ * Take cap refs.  Caller must already know we hold at least one ref
+ * on the caps in question or we don't know this is safe.
+ */
+void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps)
+{
+	spin_lock(&ci->i_ceph_lock);
+	ceph_take_cap_refs(ci, caps, false);
+	spin_unlock(&ci->i_ceph_lock);
+}
+
+
+/*
+ * drop cap_snap that is not associated with any snapshot.
+ * we don't need to send FLUSHSNAP message for it.
+ */
+static int ceph_try_drop_cap_snap(struct ceph_inode_info *ci,
+				  struct ceph_cap_snap *capsnap)
+{
+	if (!capsnap->need_flush &&
+	    !capsnap->writing && !capsnap->dirty_pages) {
+		dout("dropping cap_snap %p follows %llu\n",
+		     capsnap, capsnap->follows);
+		BUG_ON(capsnap->cap_flush.tid > 0);
+		ceph_put_snap_context(capsnap->context);
+		if (!list_is_last(&capsnap->ci_item, &ci->i_cap_snaps))
+			ci->i_ceph_flags |= CEPH_I_FLUSH_SNAPS;
+
+		list_del(&capsnap->ci_item);
+		ceph_put_cap_snap(capsnap);
+		return 1;
+	}
+	return 0;
+}
+
+enum put_cap_refs_mode {
+	PUT_CAP_REFS_SYNC = 0,
+	PUT_CAP_REFS_NO_CHECK,
+	PUT_CAP_REFS_ASYNC,
+};
+
+/*
+ * Release cap refs.
+ *
+ * If we released the last ref on any given cap, call ceph_check_caps
+ * to release (or schedule a release).
+ *
+ * If we are releasing a WR cap (from a sync write), finalize any affected
+ * cap_snap, and wake up any waiters.
+ */
+static void __ceph_put_cap_refs(struct ceph_inode_info *ci, int had,
+				enum put_cap_refs_mode mode)
+{
+	struct inode *inode = &ci->netfs.inode;
+	int last = 0, put = 0, flushsnaps = 0, wake = 0;
+	bool check_flushsnaps = false;
+
+	spin_lock(&ci->i_ceph_lock);
+	if (had & CEPH_CAP_PIN)
+		--ci->i_pin_ref;
+	if (had & CEPH_CAP_FILE_RD)
+		if (--ci->i_rd_ref == 0)
+			last++;
+	if (had & CEPH_CAP_FILE_CACHE)
+		if (--ci->i_rdcache_ref == 0)
+			last++;
+	if (had & CEPH_CAP_FILE_EXCL)
+		if (--ci->i_fx_ref == 0)
+			last++;
+	if (had & CEPH_CAP_FILE_BUFFER) {
+		if (--ci->i_wb_ref == 0) {
+			last++;
+			/* put the ref held by ceph_take_cap_refs() */
+			put++;
+			check_flushsnaps = true;
+		}
+		dout("put_cap_refs %p wb %d -> %d (?)\n",
+		     inode, ci->i_wb_ref+1, ci->i_wb_ref);
+	}
+	if (had & CEPH_CAP_FILE_WR) {
+		if (--ci->i_wr_ref == 0) {
+			last++;
+			check_flushsnaps = true;
+			if (ci->i_wrbuffer_ref_head == 0 &&
+			    ci->i_dirty_caps == 0 &&
+			    ci->i_flushing_caps == 0) {
+				BUG_ON(!ci->i_head_snapc);
+				ceph_put_snap_context(ci->i_head_snapc);
+				ci->i_head_snapc = NULL;
+			}
+			/* see comment in __ceph_remove_cap() */
+			if (!__ceph_is_any_real_caps(ci) && ci->i_snap_realm)
+				ceph_change_snap_realm(inode, NULL);
+		}
+	}
+	if (check_flushsnaps && __ceph_have_pending_cap_snap(ci)) {
+		struct ceph_cap_snap *capsnap =
+			list_last_entry(&ci->i_cap_snaps,
+					struct ceph_cap_snap,
+					ci_item);
+
+		capsnap->writing = 0;
+		if (ceph_try_drop_cap_snap(ci, capsnap))
+			/* put the ref held by ceph_queue_cap_snap() */
+			put++;
+		else if (__ceph_finish_cap_snap(ci, capsnap))
+			flushsnaps = 1;
+		wake = 1;
+	}
+	spin_unlock(&ci->i_ceph_lock);
+
+	dout("put_cap_refs %p had %s%s%s\n", inode, ceph_cap_string(had),
+	     last ? " last" : "", put ? " put" : "");
+
+	switch (mode) {
+	case PUT_CAP_REFS_SYNC:
+		if (last)
+			ceph_check_caps(ci, 0, NULL);
+		else if (flushsnaps)
+			ceph_flush_snaps(ci, NULL);
+		break;
+	case PUT_CAP_REFS_ASYNC:
+		if (last)
+			ceph_queue_check_caps(inode);
+		else if (flushsnaps)
+			ceph_queue_flush_snaps(inode);
+		break;
+	default:
+		break;
+	}
+	if (wake)
+		wake_up_all(&ci->i_cap_wq);
+	while (put-- > 0)
+		iput(inode);
+}
+
+void ceph_put_cap_refs(struct ceph_inode_info *ci, int had)
+{
+	__ceph_put_cap_refs(ci, had, PUT_CAP_REFS_SYNC);
+}
+
+void ceph_put_cap_refs_async(struct ceph_inode_info *ci, int had)
+{
+	__ceph_put_cap_refs(ci, had, PUT_CAP_REFS_ASYNC);
+}
+
+void ceph_put_cap_refs_no_check_caps(struct ceph_inode_info *ci, int had)
+{
+	__ceph_put_cap_refs(ci, had, PUT_CAP_REFS_NO_CHECK);
+}
+
+/*
+ * Release @nr WRBUFFER refs on dirty pages for the given @snapc snap
+ * context.  Adjust per-snap dirty page accounting as appropriate.
+ * Once all dirty data for a cap_snap is flushed, flush snapped file
+ * metadata back to the MDS.  If we dropped the last ref, call
+ * ceph_check_caps.
+ */
+void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
+				struct ceph_snap_context *snapc)
+{
+	struct inode *inode = &ci->netfs.inode;
+	struct ceph_cap_snap *capsnap = NULL, *iter;
+	int put = 0;
+	bool last = false;
+	bool flush_snaps = false;
+	bool complete_capsnap = false;
+
+	spin_lock(&ci->i_ceph_lock);
+	ci->i_wrbuffer_ref -= nr;
+	if (ci->i_wrbuffer_ref == 0) {
+		last = true;
+		put++;
+	}
+
+	if (ci->i_head_snapc == snapc) {
+		ci->i_wrbuffer_ref_head -= nr;
+		if (ci->i_wrbuffer_ref_head == 0 &&
+		    ci->i_wr_ref == 0 &&
+		    ci->i_dirty_caps == 0 &&
+		    ci->i_flushing_caps == 0) {
+			BUG_ON(!ci->i_head_snapc);
+			ceph_put_snap_context(ci->i_head_snapc);
+			ci->i_head_snapc = NULL;
+		}
+		dout("put_wrbuffer_cap_refs on %p head %d/%d -> %d/%d %s\n",
+		     inode,
+		     ci->i_wrbuffer_ref+nr, ci->i_wrbuffer_ref_head+nr,
+		     ci->i_wrbuffer_ref, ci->i_wrbuffer_ref_head,
+		     last ? " LAST" : "");
+	} else {
+		list_for_each_entry(iter, &ci->i_cap_snaps, ci_item) {
+			if (iter->context == snapc) {
+				capsnap = iter;
+				break;
+			}
+		}
+
+		if (!capsnap) {
+			/*
+			 * The capsnap should already be removed when removing
+			 * auth cap in the case of a forced unmount.
+			 */
+			WARN_ON_ONCE(ci->i_auth_cap);
+			goto unlock;
+		}
+
+		capsnap->dirty_pages -= nr;
+		if (capsnap->dirty_pages == 0) {
+			complete_capsnap = true;
+			if (!capsnap->writing) {
+				if (ceph_try_drop_cap_snap(ci, capsnap)) {
+					put++;
+				} else {
+					ci->i_ceph_flags |= CEPH_I_FLUSH_SNAPS;
+					flush_snaps = true;
+				}
+			}
+		}
+		dout("put_wrbuffer_cap_refs on %p cap_snap %p "
+		     " snap %lld %d/%d -> %d/%d %s%s\n",
+		     inode, capsnap, capsnap->context->seq,
+		     ci->i_wrbuffer_ref+nr, capsnap->dirty_pages + nr,
+		     ci->i_wrbuffer_ref, capsnap->dirty_pages,
+		     last ? " (wrbuffer last)" : "",
+		     complete_capsnap ? " (complete capsnap)" : "");
+	}
+
+unlock:
+	spin_unlock(&ci->i_ceph_lock);
+
+	if (last) {
+		ceph_check_caps(ci, 0, NULL);
+	} else if (flush_snaps) {
+		ceph_flush_snaps(ci, NULL);
+	}
+	if (complete_capsnap)
+		wake_up_all(&ci->i_cap_wq);
+	while (put-- > 0) {
+		iput(inode);
+	}
+}
+
+/*
+ * Invalidate unlinked inode's aliases, so we can drop the inode ASAP.
+ */
+static void invalidate_aliases(struct inode *inode)
+{
+	struct dentry *dn, *prev = NULL;
+
+	dout("invalidate_aliases inode %p\n", inode);
+	d_prune_aliases(inode);
+	/*
+	 * For non-directory inode, d_find_alias() only returns
+	 * hashed dentry. After calling d_invalidate(), the
+	 * dentry becomes unhashed.
+	 *
+	 * For directory inode, d_find_alias() can return
+	 * unhashed dentry. But directory inode should have
+	 * one alias at most.
+	 */
+	while ((dn = d_find_alias(inode))) {
+		if (dn == prev) {
+			dput(dn);
+			break;
+		}
+		d_invalidate(dn);
+		if (prev)
+			dput(prev);
+		prev = dn;
+	}
+	if (prev)
+		dput(prev);
+}
+
+struct cap_extra_info {
+	struct ceph_string *pool_ns;
+	/* inline data */
+	u64 inline_version;
+	void *inline_data;
+	u32 inline_len;
+	/* dirstat */
+	bool dirstat_valid;
+	u64 nfiles;
+	u64 nsubdirs;
+	u64 change_attr;
+	/* currently issued */
+	int issued;
+	struct timespec64 btime;
+};
+
+/*
+ * Handle a cap GRANT message from the MDS.  (Note that a GRANT may
+ * actually be a revocation if it specifies a smaller cap set.)
+ *
+ * caller holds s_mutex and i_ceph_lock, we drop both.
+ */
+static void handle_cap_grant(struct inode *inode,
+			     struct ceph_mds_session *session,
+			     struct ceph_cap *cap,
+			     struct ceph_mds_caps *grant,
+			     struct ceph_buffer *xattr_buf,
+			     struct cap_extra_info *extra_info)
+	__releases(ci->i_ceph_lock)
+	__releases(session->s_mdsc->snap_rwsem)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	int seq = le32_to_cpu(grant->seq);
+	int newcaps = le32_to_cpu(grant->caps);
+	int used, wanted, dirty;
+	u64 size = le64_to_cpu(grant->size);
+	u64 max_size = le64_to_cpu(grant->max_size);
+	unsigned char check_caps = 0;
+	bool was_stale = cap->cap_gen < atomic_read(&session->s_cap_gen);
+	bool wake = false;
+	bool writeback = false;
+	bool queue_trunc = false;
+	bool queue_invalidate = false;
+	bool deleted_inode = false;
+	bool fill_inline = false;
+
+	dout("handle_cap_grant inode %p cap %p mds%d seq %d %s\n",
+	     inode, cap, session->s_mds, seq, ceph_cap_string(newcaps));
+	dout(" size %llu max_size %llu, i_size %llu\n", size, max_size,
+		i_size_read(inode));
+
+
+	/*
+	 * If CACHE is being revoked, and we have no dirty buffers,
+	 * try to invalidate (once).  (If there are dirty buffers, we
+	 * will invalidate _after_ writeback.)
+	 */
+	if (S_ISREG(inode->i_mode) && /* don't invalidate readdir cache */
+	    ((cap->issued & ~newcaps) & CEPH_CAP_FILE_CACHE) &&
+	    (newcaps & CEPH_CAP_FILE_LAZYIO) == 0 &&
+	    !(ci->i_wrbuffer_ref || ci->i_wb_ref)) {
+		if (try_nonblocking_invalidate(inode)) {
+			/* there were locked pages.. invalidate later
+			   in a separate thread. */
+			if (ci->i_rdcache_revoking != ci->i_rdcache_gen) {
+				queue_invalidate = true;
+				ci->i_rdcache_revoking = ci->i_rdcache_gen;
+			}
+		}
+	}
+
+	if (was_stale)
+		cap->issued = cap->implemented = CEPH_CAP_PIN;
+
+	/*
+	 * auth mds of the inode changed. we received the cap export message,
+	 * but still haven't received the cap import message. handle_cap_export
+	 * updated the new auth MDS' cap.
+	 *
+	 * "ceph_seq_cmp(seq, cap->seq) <= 0" means we are processing a message
+	 * that was sent before the cap import message. So don't remove caps.
+	 */
+	if (ceph_seq_cmp(seq, cap->seq) <= 0) {
+		WARN_ON(cap != ci->i_auth_cap);
+		WARN_ON(cap->cap_id != le64_to_cpu(grant->cap_id));
+		seq = cap->seq;
+		newcaps |= cap->issued;
+	}
+
+	/* side effects now are allowed */
+	cap->cap_gen = atomic_read(&session->s_cap_gen);
+	cap->seq = seq;
+
+	__check_cap_issue(ci, cap, newcaps);
+
+	inode_set_max_iversion_raw(inode, extra_info->change_attr);
+
+	if ((newcaps & CEPH_CAP_AUTH_SHARED) &&
+	    (extra_info->issued & CEPH_CAP_AUTH_EXCL) == 0) {
+		umode_t mode = le32_to_cpu(grant->mode);
+
+		if (inode_wrong_type(inode, mode))
+			pr_warn_once("inode type changed! (ino %llx.%llx is 0%o, mds says 0%o)\n",
+				     ceph_vinop(inode), inode->i_mode, mode);
+		else
+			inode->i_mode = mode;
+		inode->i_uid = make_kuid(&init_user_ns, le32_to_cpu(grant->uid));
+		inode->i_gid = make_kgid(&init_user_ns, le32_to_cpu(grant->gid));
+		ci->i_btime = extra_info->btime;
+		dout("%p mode 0%o uid.gid %d.%d\n", inode, inode->i_mode,
+		     from_kuid(&init_user_ns, inode->i_uid),
+		     from_kgid(&init_user_ns, inode->i_gid));
+	}
+
+	if ((newcaps & CEPH_CAP_LINK_SHARED) &&
+	    (extra_info->issued & CEPH_CAP_LINK_EXCL) == 0) {
+		set_nlink(inode, le32_to_cpu(grant->nlink));
+		if (inode->i_nlink == 0)
+			deleted_inode = true;
+	}
+
+	if ((extra_info->issued & CEPH_CAP_XATTR_EXCL) == 0 &&
+	    grant->xattr_len) {
+		int len = le32_to_cpu(grant->xattr_len);
+		u64 version = le64_to_cpu(grant->xattr_version);
+
+		if (version > ci->i_xattrs.version) {
+			dout(" got new xattrs v%llu on %p len %d\n",
+			     version, inode, len);
+			if (ci->i_xattrs.blob)
+				ceph_buffer_put(ci->i_xattrs.blob);
+			ci->i_xattrs.blob = ceph_buffer_get(xattr_buf);
+			ci->i_xattrs.version = version;
+			ceph_forget_all_cached_acls(inode);
+			ceph_security_invalidate_secctx(inode);
+		}
+	}
+
+	if (newcaps & CEPH_CAP_ANY_RD) {
+		struct timespec64 mtime, atime, ctime;
+		/* ctime/mtime/atime? */
+		ceph_decode_timespec64(&mtime, &grant->mtime);
+		ceph_decode_timespec64(&atime, &grant->atime);
+		ceph_decode_timespec64(&ctime, &grant->ctime);
+		ceph_fill_file_time(inode, extra_info->issued,
+				    le32_to_cpu(grant->time_warp_seq),
+				    &ctime, &mtime, &atime);
+	}
+
+	if ((newcaps & CEPH_CAP_FILE_SHARED) && extra_info->dirstat_valid) {
+		ci->i_files = extra_info->nfiles;
+		ci->i_subdirs = extra_info->nsubdirs;
+	}
+
+	if (newcaps & (CEPH_CAP_ANY_FILE_RD | CEPH_CAP_ANY_FILE_WR)) {
+		/* file layout may have changed */
+		s64 old_pool = ci->i_layout.pool_id;
+		struct ceph_string *old_ns;
+
+		ceph_file_layout_from_legacy(&ci->i_layout, &grant->layout);
+		old_ns = rcu_dereference_protected(ci->i_layout.pool_ns,
+					lockdep_is_held(&ci->i_ceph_lock));
+		rcu_assign_pointer(ci->i_layout.pool_ns, extra_info->pool_ns);
+
+		if (ci->i_layout.pool_id != old_pool ||
+		    extra_info->pool_ns != old_ns)
+			ci->i_ceph_flags &= ~CEPH_I_POOL_PERM;
+
+		extra_info->pool_ns = old_ns;
+
+		/* size/truncate_seq? */
+		queue_trunc = ceph_fill_file_size(inode, extra_info->issued,
+					le32_to_cpu(grant->truncate_seq),
+					le64_to_cpu(grant->truncate_size),
+					size);
+	}
+
+	if (ci->i_auth_cap == cap && (newcaps & CEPH_CAP_ANY_FILE_WR)) {
+		if (max_size != ci->i_max_size) {
+			dout("max_size %lld -> %llu\n",
+			     ci->i_max_size, max_size);
+			ci->i_max_size = max_size;
+			if (max_size >= ci->i_wanted_max_size) {
+				ci->i_wanted_max_size = 0;  /* reset */
+				ci->i_requested_max_size = 0;
+			}
+			wake = true;
+		}
+	}
+
+	/* check cap bits */
+	wanted = __ceph_caps_wanted(ci);
+	used = __ceph_caps_used(ci);
+	dirty = __ceph_caps_dirty(ci);
+	dout(" my wanted = %s, used = %s, dirty %s\n",
+	     ceph_cap_string(wanted),
+	     ceph_cap_string(used),
+	     ceph_cap_string(dirty));
+
+	if ((was_stale || le32_to_cpu(grant->op) == CEPH_CAP_OP_IMPORT) &&
+	    (wanted & ~(cap->mds_wanted | newcaps))) {
+		/*
+		 * If mds is importing cap, prior cap messages that update
+		 * 'wanted' may get dropped by mds (migrate seq mismatch).
+		 *
+		 * We don't send cap message to update 'wanted' if what we
+		 * want are already issued. If mds revokes caps, cap message
+		 * that releases caps also tells mds what we want. But if
+		 * caps got revoked by mds forcedly (session stale). We may
+		 * haven't told mds what we want.
+		 */
+		check_caps = 1;
+	}
+
+	/* revocation, grant, or no-op? */
+	if (cap->issued & ~newcaps) {
+		int revoking = cap->issued & ~newcaps;
+
+		dout("revocation: %s -> %s (revoking %s)\n",
+		     ceph_cap_string(cap->issued),
+		     ceph_cap_string(newcaps),
+		     ceph_cap_string(revoking));
+		if (S_ISREG(inode->i_mode) &&
+		    (revoking & used & CEPH_CAP_FILE_BUFFER))
+			writeback = true;  /* initiate writeback; will delay ack */
+		else if (queue_invalidate &&
+			 revoking == CEPH_CAP_FILE_CACHE &&
+			 (newcaps & CEPH_CAP_FILE_LAZYIO) == 0)
+			; /* do nothing yet, invalidation will be queued */
+		else if (cap == ci->i_auth_cap)
+			check_caps = 1; /* check auth cap only */
+		else
+			check_caps = 2; /* check all caps */
+		/* If there is new caps, try to wake up the waiters */
+		if (~cap->issued & newcaps)
+			wake = true;
+		cap->issued = newcaps;
+		cap->implemented |= newcaps;
+	} else if (cap->issued == newcaps) {
+		dout("caps unchanged: %s -> %s\n",
+		     ceph_cap_string(cap->issued), ceph_cap_string(newcaps));
+	} else {
+		dout("grant: %s -> %s\n", ceph_cap_string(cap->issued),
+		     ceph_cap_string(newcaps));
+		/* non-auth MDS is revoking the newly grant caps ? */
+		if (cap == ci->i_auth_cap &&
+		    __ceph_caps_revoking_other(ci, cap, newcaps))
+		    check_caps = 2;
+
+		cap->issued = newcaps;
+		cap->implemented |= newcaps; /* add bits only, to
+					      * avoid stepping on a
+					      * pending revocation */
+		wake = true;
+	}
+	BUG_ON(cap->issued & ~cap->implemented);
+
+	/* don't let check_caps skip sending a response to MDS for revoke msgs */
+	if (le32_to_cpu(grant->op) == CEPH_CAP_OP_REVOKE) {
+		cap->mds_wanted = 0;
+		if (cap == ci->i_auth_cap)
+			check_caps = 1; /* check auth cap only */
+		else
+			check_caps = 2; /* check all caps */
+	}
+
+	if (extra_info->inline_version > 0 &&
+	    extra_info->inline_version >= ci->i_inline_version) {
+		ci->i_inline_version = extra_info->inline_version;
+		if (ci->i_inline_version != CEPH_INLINE_NONE &&
+		    (newcaps & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)))
+			fill_inline = true;
+	}
+
+	if (le32_to_cpu(grant->op) == CEPH_CAP_OP_IMPORT) {
+		if (ci->i_auth_cap == cap) {
+			if (newcaps & ~extra_info->issued)
+				wake = true;
+
+			if (ci->i_requested_max_size > max_size ||
+			    !(le32_to_cpu(grant->wanted) & CEPH_CAP_ANY_FILE_WR)) {
+				/* re-request max_size if necessary */
+				ci->i_requested_max_size = 0;
+				wake = true;
+			}
+
+			ceph_kick_flushing_inode_caps(session, ci);
+		}
+		up_read(&session->s_mdsc->snap_rwsem);
+	}
+	spin_unlock(&ci->i_ceph_lock);
+
+	if (fill_inline)
+		ceph_fill_inline_data(inode, NULL, extra_info->inline_data,
+				      extra_info->inline_len);
+
+	if (queue_trunc)
+		ceph_queue_vmtruncate(inode);
+
+	if (writeback)
+		/*
+		 * queue inode for writeback: we can't actually call
+		 * filemap_write_and_wait, etc. from message handler
+		 * context.
+		 */
+		ceph_queue_writeback(inode);
+	if (queue_invalidate)
+		ceph_queue_invalidate(inode);
+	if (deleted_inode)
+		invalidate_aliases(inode);
+	if (wake)
+		wake_up_all(&ci->i_cap_wq);
+
+	mutex_unlock(&session->s_mutex);
+	if (check_caps == 1)
+		ceph_check_caps(ci, CHECK_CAPS_AUTHONLY | CHECK_CAPS_NOINVAL,
+				session);
+	else if (check_caps == 2)
+		ceph_check_caps(ci, CHECK_CAPS_NOINVAL, session);
+}
+
+/*
+ * Handle FLUSH_ACK from MDS, indicating that metadata we sent to the
+ * MDS has been safely committed.
+ */
+static void handle_cap_flush_ack(struct inode *inode, u64 flush_tid,
+				 struct ceph_mds_caps *m,
+				 struct ceph_mds_session *session,
+				 struct ceph_cap *cap)
+	__releases(ci->i_ceph_lock)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
+	struct ceph_cap_flush *cf, *tmp_cf;
+	LIST_HEAD(to_remove);
+	unsigned seq = le32_to_cpu(m->seq);
+	int dirty = le32_to_cpu(m->dirty);
+	int cleaned = 0;
+	bool drop = false;
+	bool wake_ci = false;
+	bool wake_mdsc = false;
+
+	list_for_each_entry_safe(cf, tmp_cf, &ci->i_cap_flush_list, i_list) {
+		/* Is this the one that was flushed? */
+		if (cf->tid == flush_tid)
+			cleaned = cf->caps;
+
+		/* Is this a capsnap? */
+		if (cf->is_capsnap)
+			continue;
+
+		if (cf->tid <= flush_tid) {
+			/*
+			 * An earlier or current tid. The FLUSH_ACK should
+			 * represent a superset of this flush's caps.
+			 */
+			wake_ci |= __detach_cap_flush_from_ci(ci, cf);
+			list_add_tail(&cf->i_list, &to_remove);
+		} else {
+			/*
+			 * This is a later one. Any caps in it are still dirty
+			 * so don't count them as cleaned.
+			 */
+			cleaned &= ~cf->caps;
+			if (!cleaned)
+				break;
+		}
+	}
+
+	dout("handle_cap_flush_ack inode %p mds%d seq %d on %s cleaned %s,"
+	     " flushing %s -> %s\n",
+	     inode, session->s_mds, seq, ceph_cap_string(dirty),
+	     ceph_cap_string(cleaned), ceph_cap_string(ci->i_flushing_caps),
+	     ceph_cap_string(ci->i_flushing_caps & ~cleaned));
+
+	if (list_empty(&to_remove) && !cleaned)
+		goto out;
+
+	ci->i_flushing_caps &= ~cleaned;
+
+	spin_lock(&mdsc->cap_dirty_lock);
+
+	list_for_each_entry(cf, &to_remove, i_list)
+		wake_mdsc |= __detach_cap_flush_from_mdsc(mdsc, cf);
+
+	if (ci->i_flushing_caps == 0) {
+		if (list_empty(&ci->i_cap_flush_list)) {
+			list_del_init(&ci->i_flushing_item);
+			if (!list_empty(&session->s_cap_flushing)) {
+				dout(" mds%d still flushing cap on %p\n",
+				     session->s_mds,
+				     &list_first_entry(&session->s_cap_flushing,
+						struct ceph_inode_info,
+						i_flushing_item)->netfs.inode);
+			}
+		}
+		mdsc->num_cap_flushing--;
+		dout(" inode %p now !flushing\n", inode);
+
+		if (ci->i_dirty_caps == 0) {
+			dout(" inode %p now clean\n", inode);
+			BUG_ON(!list_empty(&ci->i_dirty_item));
+			drop = true;
+			if (ci->i_wr_ref == 0 &&
+			    ci->i_wrbuffer_ref_head == 0) {
+				BUG_ON(!ci->i_head_snapc);
+				ceph_put_snap_context(ci->i_head_snapc);
+				ci->i_head_snapc = NULL;
+			}
+		} else {
+			BUG_ON(list_empty(&ci->i_dirty_item));
+		}
+	}
+	spin_unlock(&mdsc->cap_dirty_lock);
+
+out:
+	spin_unlock(&ci->i_ceph_lock);
+
+	while (!list_empty(&to_remove)) {
+		cf = list_first_entry(&to_remove,
+				      struct ceph_cap_flush, i_list);
+		list_del_init(&cf->i_list);
+		if (!cf->is_capsnap)
+			ceph_free_cap_flush(cf);
+	}
+
+	if (wake_ci)
+		wake_up_all(&ci->i_cap_wq);
+	if (wake_mdsc)
+		wake_up_all(&mdsc->cap_flushing_wq);
+	if (drop)
+		iput(inode);
+}
+
+void __ceph_remove_capsnap(struct inode *inode, struct ceph_cap_snap *capsnap,
+			   bool *wake_ci, bool *wake_mdsc)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
+	bool ret;
+
+	lockdep_assert_held(&ci->i_ceph_lock);
+
+	dout("removing capsnap %p, inode %p ci %p\n", capsnap, inode, ci);
+
+	list_del_init(&capsnap->ci_item);
+	ret = __detach_cap_flush_from_ci(ci, &capsnap->cap_flush);
+	if (wake_ci)
+		*wake_ci = ret;
+
+	spin_lock(&mdsc->cap_dirty_lock);
+	if (list_empty(&ci->i_cap_flush_list))
+		list_del_init(&ci->i_flushing_item);
+
+	ret = __detach_cap_flush_from_mdsc(mdsc, &capsnap->cap_flush);
+	if (wake_mdsc)
+		*wake_mdsc = ret;
+	spin_unlock(&mdsc->cap_dirty_lock);
+}
+
+void ceph_remove_capsnap(struct inode *inode, struct ceph_cap_snap *capsnap,
+			 bool *wake_ci, bool *wake_mdsc)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+
+	lockdep_assert_held(&ci->i_ceph_lock);
+
+	WARN_ON_ONCE(capsnap->dirty_pages || capsnap->writing);
+	__ceph_remove_capsnap(inode, capsnap, wake_ci, wake_mdsc);
+}
+
+/*
+ * Handle FLUSHSNAP_ACK.  MDS has flushed snap data to disk and we can
+ * throw away our cap_snap.
+ *
+ * Caller hold s_mutex.
+ */
+static void handle_cap_flushsnap_ack(struct inode *inode, u64 flush_tid,
+				     struct ceph_mds_caps *m,
+				     struct ceph_mds_session *session)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
+	u64 follows = le64_to_cpu(m->snap_follows);
+	struct ceph_cap_snap *capsnap = NULL, *iter;
+	bool wake_ci = false;
+	bool wake_mdsc = false;
+
+	dout("handle_cap_flushsnap_ack inode %p ci %p mds%d follows %lld\n",
+	     inode, ci, session->s_mds, follows);
+
+	spin_lock(&ci->i_ceph_lock);
+	list_for_each_entry(iter, &ci->i_cap_snaps, ci_item) {
+		if (iter->follows == follows) {
+			if (iter->cap_flush.tid != flush_tid) {
+				dout(" cap_snap %p follows %lld tid %lld !="
+				     " %lld\n", iter, follows,
+				     flush_tid, iter->cap_flush.tid);
+				break;
+			}
+			capsnap = iter;
+			break;
+		} else {
+			dout(" skipping cap_snap %p follows %lld\n",
+			     iter, iter->follows);
+		}
+	}
+	if (capsnap)
+		ceph_remove_capsnap(inode, capsnap, &wake_ci, &wake_mdsc);
+	spin_unlock(&ci->i_ceph_lock);
+
+	if (capsnap) {
+		ceph_put_snap_context(capsnap->context);
+		ceph_put_cap_snap(capsnap);
+		if (wake_ci)
+			wake_up_all(&ci->i_cap_wq);
+		if (wake_mdsc)
+			wake_up_all(&mdsc->cap_flushing_wq);
+		iput(inode);
+	}
+}
+
+/*
+ * Handle TRUNC from MDS, indicating file truncation.
+ *
+ * caller hold s_mutex.
+ */
+static bool handle_cap_trunc(struct inode *inode,
+			     struct ceph_mds_caps *trunc,
+			     struct ceph_mds_session *session)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	int mds = session->s_mds;
+	int seq = le32_to_cpu(trunc->seq);
+	u32 truncate_seq = le32_to_cpu(trunc->truncate_seq);
+	u64 truncate_size = le64_to_cpu(trunc->truncate_size);
+	u64 size = le64_to_cpu(trunc->size);
+	int implemented = 0;
+	int dirty = __ceph_caps_dirty(ci);
+	int issued = __ceph_caps_issued(ceph_inode(inode), &implemented);
+	bool queue_trunc = false;
+
+	lockdep_assert_held(&ci->i_ceph_lock);
+
+	issued |= implemented | dirty;
+
+	dout("handle_cap_trunc inode %p mds%d seq %d to %lld seq %d\n",
+	     inode, mds, seq, truncate_size, truncate_seq);
+	queue_trunc = ceph_fill_file_size(inode, issued,
+					  truncate_seq, truncate_size, size);
+	return queue_trunc;
+}
+
+/*
+ * Handle EXPORT from MDS.  Cap is being migrated _from_ this mds to a
+ * different one.  If we are the most recent migration we've seen (as
+ * indicated by mseq), make note of the migrating cap bits for the
+ * duration (until we see the corresponding IMPORT).
+ *
+ * caller holds s_mutex
+ */
+static void handle_cap_export(struct inode *inode, struct ceph_mds_caps *ex,
+			      struct ceph_mds_cap_peer *ph,
+			      struct ceph_mds_session *session)
+{
+	struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;
+	struct ceph_mds_session *tsession = NULL;
+	struct ceph_cap *cap, *tcap, *new_cap = NULL;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	u64 t_cap_id;
+	unsigned mseq = le32_to_cpu(ex->migrate_seq);
+	unsigned t_seq, t_mseq;
+	int target, issued;
+	int mds = session->s_mds;
+
+	if (ph) {
+		t_cap_id = le64_to_cpu(ph->cap_id);
+		t_seq = le32_to_cpu(ph->seq);
+		t_mseq = le32_to_cpu(ph->mseq);
+		target = le32_to_cpu(ph->mds);
+	} else {
+		t_cap_id = t_seq = t_mseq = 0;
+		target = -1;
+	}
+
+	dout("handle_cap_export inode %p ci %p mds%d mseq %d target %d\n",
+	     inode, ci, mds, mseq, target);
+retry:
+	down_read(&mdsc->snap_rwsem);
+	spin_lock(&ci->i_ceph_lock);
+	cap = __get_cap_for_mds(ci, mds);
+	if (!cap || cap->cap_id != le64_to_cpu(ex->cap_id))
+		goto out_unlock;
+
+	if (target < 0) {
+		ceph_remove_cap(cap, false);
+		goto out_unlock;
+	}
+
+	/*
+	 * now we know we haven't received the cap import message yet
+	 * because the exported cap still exist.
+	 */
+
+	issued = cap->issued;
+	if (issued != cap->implemented)
+		pr_err_ratelimited("handle_cap_export: issued != implemented: "
+				"ino (%llx.%llx) mds%d seq %d mseq %d "
+				"issued %s implemented %s\n",
+				ceph_vinop(inode), mds, cap->seq, cap->mseq,
+				ceph_cap_string(issued),
+				ceph_cap_string(cap->implemented));
+
+
+	tcap = __get_cap_for_mds(ci, target);
+	if (tcap) {
+		/* already have caps from the target */
+		if (tcap->cap_id == t_cap_id &&
+		    ceph_seq_cmp(tcap->seq, t_seq) < 0) {
+			dout(" updating import cap %p mds%d\n", tcap, target);
+			tcap->cap_id = t_cap_id;
+			tcap->seq = t_seq - 1;
+			tcap->issue_seq = t_seq - 1;
+			tcap->issued |= issued;
+			tcap->implemented |= issued;
+			if (cap == ci->i_auth_cap) {
+				ci->i_auth_cap = tcap;
+				change_auth_cap_ses(ci, tcap->session);
+			}
+		}
+		ceph_remove_cap(cap, false);
+		goto out_unlock;
+	} else if (tsession) {
+		/* add placeholder for the export tagert */
+		int flag = (cap == ci->i_auth_cap) ? CEPH_CAP_FLAG_AUTH : 0;
+		tcap = new_cap;
+		ceph_add_cap(inode, tsession, t_cap_id, issued, 0,
+			     t_seq - 1, t_mseq, (u64)-1, flag, &new_cap);
+
+		if (!list_empty(&ci->i_cap_flush_list) &&
+		    ci->i_auth_cap == tcap) {
+			spin_lock(&mdsc->cap_dirty_lock);
+			list_move_tail(&ci->i_flushing_item,
+				       &tcap->session->s_cap_flushing);
+			spin_unlock(&mdsc->cap_dirty_lock);
+		}
+
+		ceph_remove_cap(cap, false);
+		goto out_unlock;
+	}
+
+	spin_unlock(&ci->i_ceph_lock);
+	up_read(&mdsc->snap_rwsem);
+	mutex_unlock(&session->s_mutex);
+
+	/* open target session */
+	tsession = ceph_mdsc_open_export_target_session(mdsc, target);
+	if (!IS_ERR(tsession)) {
+		if (mds > target) {
+			mutex_lock(&session->s_mutex);
+			mutex_lock_nested(&tsession->s_mutex,
+					  SINGLE_DEPTH_NESTING);
+		} else {
+			mutex_lock(&tsession->s_mutex);
+			mutex_lock_nested(&session->s_mutex,
+					  SINGLE_DEPTH_NESTING);
+		}
+		new_cap = ceph_get_cap(mdsc, NULL);
+	} else {
+		WARN_ON(1);
+		tsession = NULL;
+		target = -1;
+		mutex_lock(&session->s_mutex);
+	}
+	goto retry;
+
+out_unlock:
+	spin_unlock(&ci->i_ceph_lock);
+	up_read(&mdsc->snap_rwsem);
+	mutex_unlock(&session->s_mutex);
+	if (tsession) {
+		mutex_unlock(&tsession->s_mutex);
+		ceph_put_mds_session(tsession);
+	}
+	if (new_cap)
+		ceph_put_cap(mdsc, new_cap);
+}
+
+/*
+ * Handle cap IMPORT.
+ *
+ * caller holds s_mutex. acquires i_ceph_lock
+ */
+static void handle_cap_import(struct ceph_mds_client *mdsc,
+			      struct inode *inode, struct ceph_mds_caps *im,
+			      struct ceph_mds_cap_peer *ph,
+			      struct ceph_mds_session *session,
+			      struct ceph_cap **target_cap, int *old_issued)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_cap *cap, *ocap, *new_cap = NULL;
+	int mds = session->s_mds;
+	int issued;
+	unsigned caps = le32_to_cpu(im->caps);
+	unsigned wanted = le32_to_cpu(im->wanted);
+	unsigned seq = le32_to_cpu(im->seq);
+	unsigned mseq = le32_to_cpu(im->migrate_seq);
+	u64 realmino = le64_to_cpu(im->realm);
+	u64 cap_id = le64_to_cpu(im->cap_id);
+	u64 p_cap_id;
+	int peer;
+
+	if (ph) {
+		p_cap_id = le64_to_cpu(ph->cap_id);
+		peer = le32_to_cpu(ph->mds);
+	} else {
+		p_cap_id = 0;
+		peer = -1;
+	}
+
+	dout("handle_cap_import inode %p ci %p mds%d mseq %d peer %d\n",
+	     inode, ci, mds, mseq, peer);
+retry:
+	cap = __get_cap_for_mds(ci, mds);
+	if (!cap) {
+		if (!new_cap) {
+			spin_unlock(&ci->i_ceph_lock);
+			new_cap = ceph_get_cap(mdsc, NULL);
+			spin_lock(&ci->i_ceph_lock);
+			goto retry;
+		}
+		cap = new_cap;
+	} else {
+		if (new_cap) {
+			ceph_put_cap(mdsc, new_cap);
+			new_cap = NULL;
+		}
+	}
+
+	__ceph_caps_issued(ci, &issued);
+	issued |= __ceph_caps_dirty(ci);
+
+	ceph_add_cap(inode, session, cap_id, caps, wanted, seq, mseq,
+		     realmino, CEPH_CAP_FLAG_AUTH, &new_cap);
+
+	ocap = peer >= 0 ? __get_cap_for_mds(ci, peer) : NULL;
+	if (ocap && ocap->cap_id == p_cap_id) {
+		dout(" remove export cap %p mds%d flags %d\n",
+		     ocap, peer, ph->flags);
+		if ((ph->flags & CEPH_CAP_FLAG_AUTH) &&
+		    (ocap->seq != le32_to_cpu(ph->seq) ||
+		     ocap->mseq != le32_to_cpu(ph->mseq))) {
+			pr_err_ratelimited("handle_cap_import: "
+					"mismatched seq/mseq: ino (%llx.%llx) "
+					"mds%d seq %d mseq %d importer mds%d "
+					"has peer seq %d mseq %d\n",
+					ceph_vinop(inode), peer, ocap->seq,
+					ocap->mseq, mds, le32_to_cpu(ph->seq),
+					le32_to_cpu(ph->mseq));
+		}
+		ceph_remove_cap(ocap, (ph->flags & CEPH_CAP_FLAG_RELEASE));
+	}
+
+	*old_issued = issued;
+	*target_cap = cap;
+}
+
+/*
+ * Handle a caps message from the MDS.
+ *
+ * Identify the appropriate session, inode, and call the right handler
+ * based on the cap op.
+ */
+void ceph_handle_caps(struct ceph_mds_session *session,
+		      struct ceph_msg *msg)
+{
+	struct ceph_mds_client *mdsc = session->s_mdsc;
+	struct inode *inode;
+	struct ceph_inode_info *ci;
+	struct ceph_cap *cap;
+	struct ceph_mds_caps *h;
+	struct ceph_mds_cap_peer *peer = NULL;
+	struct ceph_snap_realm *realm = NULL;
+	int op;
+	int msg_version = le16_to_cpu(msg->hdr.version);
+	u32 seq, mseq;
+	struct ceph_vino vino;
+	void *snaptrace;
+	size_t snaptrace_len;
+	void *p, *end;
+	struct cap_extra_info extra_info = {};
+	bool queue_trunc;
+	bool close_sessions = false;
+
+	dout("handle_caps from mds%d\n", session->s_mds);
+
+	if (!ceph_inc_mds_stopping_blocker(mdsc, session))
+		return;
+
+	/* decode */
+	end = msg->front.iov_base + msg->front.iov_len;
+	if (msg->front.iov_len < sizeof(*h))
+		goto bad;
+	h = msg->front.iov_base;
+	op = le32_to_cpu(h->op);
+	vino.ino = le64_to_cpu(h->ino);
+	vino.snap = CEPH_NOSNAP;
+	seq = le32_to_cpu(h->seq);
+	mseq = le32_to_cpu(h->migrate_seq);
+
+	snaptrace = h + 1;
+	snaptrace_len = le32_to_cpu(h->snap_trace_len);
+	p = snaptrace + snaptrace_len;
+
+	if (msg_version >= 2) {
+		u32 flock_len;
+		ceph_decode_32_safe(&p, end, flock_len, bad);
+		if (p + flock_len > end)
+			goto bad;
+		p += flock_len;
+	}
+
+	if (msg_version >= 3) {
+		if (op == CEPH_CAP_OP_IMPORT) {
+			if (p + sizeof(*peer) > end)
+				goto bad;
+			peer = p;
+			p += sizeof(*peer);
+		} else if (op == CEPH_CAP_OP_EXPORT) {
+			/* recorded in unused fields */
+			peer = (void *)&h->size;
+		}
+	}
+
+	if (msg_version >= 4) {
+		ceph_decode_64_safe(&p, end, extra_info.inline_version, bad);
+		ceph_decode_32_safe(&p, end, extra_info.inline_len, bad);
+		if (p + extra_info.inline_len > end)
+			goto bad;
+		extra_info.inline_data = p;
+		p += extra_info.inline_len;
+	}
+
+	if (msg_version >= 5) {
+		struct ceph_osd_client	*osdc = &mdsc->fsc->client->osdc;
+		u32			epoch_barrier;
+
+		ceph_decode_32_safe(&p, end, epoch_barrier, bad);
+		ceph_osdc_update_epoch_barrier(osdc, epoch_barrier);
+	}
+
+	if (msg_version >= 8) {
+		u32 pool_ns_len;
+
+		/* version >= 6 */
+		ceph_decode_skip_64(&p, end, bad);	// flush_tid
+		/* version >= 7 */
+		ceph_decode_skip_32(&p, end, bad);	// caller_uid
+		ceph_decode_skip_32(&p, end, bad);	// caller_gid
+		/* version >= 8 */
+		ceph_decode_32_safe(&p, end, pool_ns_len, bad);
+		if (pool_ns_len > 0) {
+			ceph_decode_need(&p, end, pool_ns_len, bad);
+			extra_info.pool_ns =
+				ceph_find_or_create_string(p, pool_ns_len);
+			p += pool_ns_len;
+		}
+	}
+
+	if (msg_version >= 9) {
+		struct ceph_timespec *btime;
+
+		if (p + sizeof(*btime) > end)
+			goto bad;
+		btime = p;
+		ceph_decode_timespec64(&extra_info.btime, btime);
+		p += sizeof(*btime);
+		ceph_decode_64_safe(&p, end, extra_info.change_attr, bad);
+	}
+
+	if (msg_version >= 11) {
+		/* version >= 10 */
+		ceph_decode_skip_32(&p, end, bad); // flags
+		/* version >= 11 */
+		extra_info.dirstat_valid = true;
+		ceph_decode_64_safe(&p, end, extra_info.nfiles, bad);
+		ceph_decode_64_safe(&p, end, extra_info.nsubdirs, bad);
+	}
+
+	/* lookup ino */
+	inode = ceph_find_inode(mdsc->fsc->sb, vino);
+	dout(" op %s ino %llx.%llx inode %p\n", ceph_cap_op_name(op), vino.ino,
+	     vino.snap, inode);
+
+	mutex_lock(&session->s_mutex);
+	dout(" mds%d seq %lld cap seq %u\n", session->s_mds, session->s_seq,
+	     (unsigned)seq);
+
+	if (!inode) {
+		dout(" i don't have ino %llx\n", vino.ino);
+
+		if (op == CEPH_CAP_OP_IMPORT) {
+			cap = ceph_get_cap(mdsc, NULL);
+			cap->cap_ino = vino.ino;
+			cap->queue_release = 1;
+			cap->cap_id = le64_to_cpu(h->cap_id);
+			cap->mseq = mseq;
+			cap->seq = seq;
+			cap->issue_seq = seq;
+			spin_lock(&session->s_cap_lock);
+			__ceph_queue_cap_release(session, cap);
+			spin_unlock(&session->s_cap_lock);
+		}
+		goto flush_cap_releases;
+	}
+	ci = ceph_inode(inode);
+
+	/* these will work even if we don't have a cap yet */
+	switch (op) {
+	case CEPH_CAP_OP_FLUSHSNAP_ACK:
+		handle_cap_flushsnap_ack(inode, le64_to_cpu(msg->hdr.tid),
+					 h, session);
+		goto done;
+
+	case CEPH_CAP_OP_EXPORT:
+		handle_cap_export(inode, h, peer, session);
+		goto done_unlocked;
+
+	case CEPH_CAP_OP_IMPORT:
+		realm = NULL;
+		if (snaptrace_len) {
+			down_write(&mdsc->snap_rwsem);
+			if (ceph_update_snap_trace(mdsc, snaptrace,
+						   snaptrace + snaptrace_len,
+						   false, &realm)) {
+				up_write(&mdsc->snap_rwsem);
+				close_sessions = true;
+				goto done;
+			}
+			downgrade_write(&mdsc->snap_rwsem);
+		} else {
+			down_read(&mdsc->snap_rwsem);
+		}
+		spin_lock(&ci->i_ceph_lock);
+		handle_cap_import(mdsc, inode, h, peer, session,
+				  &cap, &extra_info.issued);
+		handle_cap_grant(inode, session, cap,
+				 h, msg->middle, &extra_info);
+		if (realm)
+			ceph_put_snap_realm(mdsc, realm);
+		goto done_unlocked;
+	}
+
+	/* the rest require a cap */
+	spin_lock(&ci->i_ceph_lock);
+	cap = __get_cap_for_mds(ceph_inode(inode), session->s_mds);
+	if (!cap) {
+		dout(" no cap on %p ino %llx.%llx from mds%d\n",
+		     inode, ceph_ino(inode), ceph_snap(inode),
+		     session->s_mds);
+		spin_unlock(&ci->i_ceph_lock);
+		goto flush_cap_releases;
+	}
+
+	/* note that each of these drops i_ceph_lock for us */
+	switch (op) {
+	case CEPH_CAP_OP_REVOKE:
+	case CEPH_CAP_OP_GRANT:
+		__ceph_caps_issued(ci, &extra_info.issued);
+		extra_info.issued |= __ceph_caps_dirty(ci);
+		handle_cap_grant(inode, session, cap,
+				 h, msg->middle, &extra_info);
+		goto done_unlocked;
+
+	case CEPH_CAP_OP_FLUSH_ACK:
+		handle_cap_flush_ack(inode, le64_to_cpu(msg->hdr.tid),
+				     h, session, cap);
+		break;
+
+	case CEPH_CAP_OP_TRUNC:
+		queue_trunc = handle_cap_trunc(inode, h, session);
+		spin_unlock(&ci->i_ceph_lock);
+		if (queue_trunc)
+			ceph_queue_vmtruncate(inode);
+		break;
+
+	default:
+		spin_unlock(&ci->i_ceph_lock);
+		pr_err("ceph_handle_caps: unknown cap op %d %s\n", op,
+		       ceph_cap_op_name(op));
+	}
+
+done:
+	mutex_unlock(&session->s_mutex);
+done_unlocked:
+	iput(inode);
+out:
+	ceph_dec_mds_stopping_blocker(mdsc);
+
+	ceph_put_string(extra_info.pool_ns);
+
+	/* Defer closing the sessions after s_mutex lock being released */
+	if (close_sessions)
+		ceph_mdsc_close_sessions(mdsc);
+
+	return;
+
+flush_cap_releases:
+	/*
+	 * send any cap release message to try to move things
+	 * along for the mds (who clearly thinks we still have this
+	 * cap).
+	 */
+	ceph_flush_cap_releases(mdsc, session);
+	goto done;
+
+bad:
+	pr_err("ceph_handle_caps: corrupt message\n");
+	ceph_msg_dump(msg);
+	goto out;
+}
+
+/*
+ * Delayed work handler to process end of delayed cap release LRU list.
+ *
+ * If new caps are added to the list while processing it, these won't get
+ * processed in this run.  In this case, the ci->i_hold_caps_max will be
+ * returned so that the work can be scheduled accordingly.
+ */
+unsigned long ceph_check_delayed_caps(struct ceph_mds_client *mdsc)
+{
+	struct inode *inode;
+	struct ceph_inode_info *ci;
+	struct ceph_mount_options *opt = mdsc->fsc->mount_options;
+	unsigned long delay_max = opt->caps_wanted_delay_max * HZ;
+	unsigned long loop_start = jiffies;
+	unsigned long delay = 0;
+
+	dout("check_delayed_caps\n");
+	spin_lock(&mdsc->cap_delay_lock);
+	while (!list_empty(&mdsc->cap_delay_list)) {
+		ci = list_first_entry(&mdsc->cap_delay_list,
+				      struct ceph_inode_info,
+				      i_cap_delay_list);
+		if (time_before(loop_start, ci->i_hold_caps_max - delay_max)) {
+			dout("%s caps added recently.  Exiting loop", __func__);
+			delay = ci->i_hold_caps_max;
+			break;
+		}
+		if ((ci->i_ceph_flags & CEPH_I_FLUSH) == 0 &&
+		    time_before(jiffies, ci->i_hold_caps_max))
+			break;
+		list_del_init(&ci->i_cap_delay_list);
+
+		inode = igrab(&ci->netfs.inode);
+		if (inode) {
+			spin_unlock(&mdsc->cap_delay_lock);
+			dout("check_delayed_caps on %p\n", inode);
+			ceph_check_caps(ci, 0, NULL);
+			iput(inode);
+			spin_lock(&mdsc->cap_delay_lock);
+		}
+	}
+	spin_unlock(&mdsc->cap_delay_lock);
+
+	return delay;
+}
+
+/*
+ * Flush all dirty caps to the mds
+ */
+static void flush_dirty_session_caps(struct ceph_mds_session *s)
+{
+	struct ceph_mds_client *mdsc = s->s_mdsc;
+	struct ceph_inode_info *ci;
+	struct inode *inode;
+
+	dout("flush_dirty_caps\n");
+	spin_lock(&mdsc->cap_dirty_lock);
+	while (!list_empty(&s->s_cap_dirty)) {
+		ci = list_first_entry(&s->s_cap_dirty, struct ceph_inode_info,
+				      i_dirty_item);
+		inode = &ci->netfs.inode;
+		ihold(inode);
+		dout("flush_dirty_caps %llx.%llx\n", ceph_vinop(inode));
+		spin_unlock(&mdsc->cap_dirty_lock);
+		ceph_wait_on_async_create(inode);
+		ceph_check_caps(ci, CHECK_CAPS_FLUSH, NULL);
+		iput(inode);
+		spin_lock(&mdsc->cap_dirty_lock);
+	}
+	spin_unlock(&mdsc->cap_dirty_lock);
+	dout("flush_dirty_caps done\n");
+}
+
+void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc)
+{
+	ceph_mdsc_iterate_sessions(mdsc, flush_dirty_session_caps, true);
+}
+
+void __ceph_touch_fmode(struct ceph_inode_info *ci,
+			struct ceph_mds_client *mdsc, int fmode)
+{
+	unsigned long now = jiffies;
+	if (fmode & CEPH_FILE_MODE_RD)
+		ci->i_last_rd = now;
+	if (fmode & CEPH_FILE_MODE_WR)
+		ci->i_last_wr = now;
+	/* queue periodic check */
+	if (fmode &&
+	    __ceph_is_any_real_caps(ci) &&
+	    list_empty(&ci->i_cap_delay_list))
+		__cap_delay_requeue(mdsc, ci);
+}
+
+void ceph_get_fmode(struct ceph_inode_info *ci, int fmode, int count)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(ci->netfs.inode.i_sb);
+	int bits = (fmode << 1) | 1;
+	bool already_opened = false;
+	int i;
+
+	if (count == 1)
+		atomic64_inc(&mdsc->metric.opened_files);
+
+	spin_lock(&ci->i_ceph_lock);
+	for (i = 0; i < CEPH_FILE_MODE_BITS; i++) {
+		/*
+		 * If any of the mode ref is larger than 0,
+		 * that means it has been already opened by
+		 * others. Just skip checking the PIN ref.
+		 */
+		if (i && ci->i_nr_by_mode[i])
+			already_opened = true;
+
+		if (bits & (1 << i))
+			ci->i_nr_by_mode[i] += count;
+	}
+
+	if (!already_opened)
+		percpu_counter_inc(&mdsc->metric.opened_inodes);
+	spin_unlock(&ci->i_ceph_lock);
+}
+
+/*
+ * Drop open file reference.  If we were the last open file,
+ * we may need to release capabilities to the MDS (or schedule
+ * their delayed release).
+ */
+void ceph_put_fmode(struct ceph_inode_info *ci, int fmode, int count)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(ci->netfs.inode.i_sb);
+	int bits = (fmode << 1) | 1;
+	bool is_closed = true;
+	int i;
+
+	if (count == 1)
+		atomic64_dec(&mdsc->metric.opened_files);
+
+	spin_lock(&ci->i_ceph_lock);
+	for (i = 0; i < CEPH_FILE_MODE_BITS; i++) {
+		if (bits & (1 << i)) {
+			BUG_ON(ci->i_nr_by_mode[i] < count);
+			ci->i_nr_by_mode[i] -= count;
+		}
+
+		/*
+		 * If any of the mode ref is not 0 after
+		 * decreased, that means it is still opened
+		 * by others. Just skip checking the PIN ref.
+		 */
+		if (i && ci->i_nr_by_mode[i])
+			is_closed = false;
+	}
+
+	if (is_closed)
+		percpu_counter_dec(&mdsc->metric.opened_inodes);
+	spin_unlock(&ci->i_ceph_lock);
+}
+
+/*
+ * For a soon-to-be unlinked file, drop the LINK caps. If it
+ * looks like the link count will hit 0, drop any other caps (other
+ * than PIN) we don't specifically want (due to the file still being
+ * open).
+ */
+int ceph_drop_caps_for_unlink(struct inode *inode)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	int drop = CEPH_CAP_LINK_SHARED | CEPH_CAP_LINK_EXCL;
+
+	spin_lock(&ci->i_ceph_lock);
+	if (inode->i_nlink == 1) {
+		drop |= ~(__ceph_caps_wanted(ci) | CEPH_CAP_PIN);
+
+		if (__ceph_caps_dirty(ci)) {
+			struct ceph_mds_client *mdsc =
+				ceph_inode_to_client(inode)->mdsc;
+			__cap_delay_requeue_front(mdsc, ci);
+		}
+	}
+	spin_unlock(&ci->i_ceph_lock);
+	return drop;
+}
+
+/*
+ * Helpers for embedding cap and dentry lease releases into mds
+ * requests.
+ *
+ * @force is used by dentry_release (below) to force inclusion of a
+ * record for the directory inode, even when there aren't any caps to
+ * drop.
+ */
+int ceph_encode_inode_release(void **p, struct inode *inode,
+			      int mds, int drop, int unless, int force)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_cap *cap;
+	struct ceph_mds_request_release *rel = *p;
+	int used, dirty;
+	int ret = 0;
+
+	spin_lock(&ci->i_ceph_lock);
+	used = __ceph_caps_used(ci);
+	dirty = __ceph_caps_dirty(ci);
+
+	dout("encode_inode_release %p mds%d used|dirty %s drop %s unless %s\n",
+	     inode, mds, ceph_cap_string(used|dirty), ceph_cap_string(drop),
+	     ceph_cap_string(unless));
+
+	/* only drop unused, clean caps */
+	drop &= ~(used | dirty);
+
+	cap = __get_cap_for_mds(ci, mds);
+	if (cap && __cap_is_valid(cap)) {
+		unless &= cap->issued;
+		if (unless) {
+			if (unless & CEPH_CAP_AUTH_EXCL)
+				drop &= ~CEPH_CAP_AUTH_SHARED;
+			if (unless & CEPH_CAP_LINK_EXCL)
+				drop &= ~CEPH_CAP_LINK_SHARED;
+			if (unless & CEPH_CAP_XATTR_EXCL)
+				drop &= ~CEPH_CAP_XATTR_SHARED;
+			if (unless & CEPH_CAP_FILE_EXCL)
+				drop &= ~CEPH_CAP_FILE_SHARED;
+		}
+
+		if (force || (cap->issued & drop)) {
+			if (cap->issued & drop) {
+				int wanted = __ceph_caps_wanted(ci);
+				dout("encode_inode_release %p cap %p "
+				     "%s -> %s, wanted %s -> %s\n", inode, cap,
+				     ceph_cap_string(cap->issued),
+				     ceph_cap_string(cap->issued & ~drop),
+				     ceph_cap_string(cap->mds_wanted),
+				     ceph_cap_string(wanted));
+
+				cap->issued &= ~drop;
+				cap->implemented &= ~drop;
+				cap->mds_wanted = wanted;
+				if (cap == ci->i_auth_cap &&
+				    !(wanted & CEPH_CAP_ANY_FILE_WR))
+					ci->i_requested_max_size = 0;
+			} else {
+				dout("encode_inode_release %p cap %p %s"
+				     " (force)\n", inode, cap,
+				     ceph_cap_string(cap->issued));
+			}
+
+			rel->ino = cpu_to_le64(ceph_ino(inode));
+			rel->cap_id = cpu_to_le64(cap->cap_id);
+			rel->seq = cpu_to_le32(cap->seq);
+			rel->issue_seq = cpu_to_le32(cap->issue_seq);
+			rel->mseq = cpu_to_le32(cap->mseq);
+			rel->caps = cpu_to_le32(cap->implemented);
+			rel->wanted = cpu_to_le32(cap->mds_wanted);
+			rel->dname_len = 0;
+			rel->dname_seq = 0;
+			*p += sizeof(*rel);
+			ret = 1;
+		} else {
+			dout("encode_inode_release %p cap %p %s (noop)\n",
+			     inode, cap, ceph_cap_string(cap->issued));
+		}
+	}
+	spin_unlock(&ci->i_ceph_lock);
+	return ret;
+}
+
+int ceph_encode_dentry_release(void **p, struct dentry *dentry,
+			       struct inode *dir,
+			       int mds, int drop, int unless)
+{
+	struct dentry *parent = NULL;
+	struct ceph_mds_request_release *rel = *p;
+	struct ceph_dentry_info *di = ceph_dentry(dentry);
+	int force = 0;
+	int ret;
+
+	/*
+	 * force an record for the directory caps if we have a dentry lease.
+	 * this is racy (can't take i_ceph_lock and d_lock together), but it
+	 * doesn't have to be perfect; the mds will revoke anything we don't
+	 * release.
+	 */
+	spin_lock(&dentry->d_lock);
+	if (di->lease_session && di->lease_session->s_mds == mds)
+		force = 1;
+	if (!dir) {
+		parent = dget(dentry->d_parent);
+		dir = d_inode(parent);
+	}
+	spin_unlock(&dentry->d_lock);
+
+	ret = ceph_encode_inode_release(p, dir, mds, drop, unless, force);
+	dput(parent);
+
+	spin_lock(&dentry->d_lock);
+	if (ret && di->lease_session && di->lease_session->s_mds == mds) {
+		dout("encode_dentry_release %p mds%d seq %d\n",
+		     dentry, mds, (int)di->lease_seq);
+		rel->dname_len = cpu_to_le32(dentry->d_name.len);
+		memcpy(*p, dentry->d_name.name, dentry->d_name.len);
+		*p += dentry->d_name.len;
+		rel->dname_seq = cpu_to_le32(di->lease_seq);
+		__ceph_mdsc_drop_dentry_lease(dentry);
+	}
+	spin_unlock(&dentry->d_lock);
+	return ret;
+}
+
+static int remove_capsnaps(struct ceph_mds_client *mdsc, struct inode *inode)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_cap_snap *capsnap;
+	int capsnap_release = 0;
+
+	lockdep_assert_held(&ci->i_ceph_lock);
+
+	dout("removing capsnaps, ci is %p, inode is %p\n", ci, inode);
+
+	while (!list_empty(&ci->i_cap_snaps)) {
+		capsnap = list_first_entry(&ci->i_cap_snaps,
+					   struct ceph_cap_snap, ci_item);
+		__ceph_remove_capsnap(inode, capsnap, NULL, NULL);
+		ceph_put_snap_context(capsnap->context);
+		ceph_put_cap_snap(capsnap);
+		capsnap_release++;
+	}
+	wake_up_all(&ci->i_cap_wq);
+	wake_up_all(&mdsc->cap_flushing_wq);
+	return capsnap_release;
+}
+
+int ceph_purge_inode_cap(struct inode *inode, struct ceph_cap *cap, bool *invalidate)
+{
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	bool is_auth;
+	bool dirty_dropped = false;
+	int iputs = 0;
+
+	lockdep_assert_held(&ci->i_ceph_lock);
+
+	dout("removing cap %p, ci is %p, inode is %p\n",
+	     cap, ci, &ci->netfs.inode);
+
+	is_auth = (cap == ci->i_auth_cap);
+	__ceph_remove_cap(cap, false);
+	if (is_auth) {
+		struct ceph_cap_flush *cf;
+
+		if (ceph_inode_is_shutdown(inode)) {
+			if (inode->i_data.nrpages > 0)
+				*invalidate = true;
+			if (ci->i_wrbuffer_ref > 0)
+				mapping_set_error(&inode->i_data, -EIO);
+		}
+
+		spin_lock(&mdsc->cap_dirty_lock);
+
+		/* trash all of the cap flushes for this inode */
+		while (!list_empty(&ci->i_cap_flush_list)) {
+			cf = list_first_entry(&ci->i_cap_flush_list,
+					      struct ceph_cap_flush, i_list);
+			list_del_init(&cf->g_list);
+			list_del_init(&cf->i_list);
+			if (!cf->is_capsnap)
+				ceph_free_cap_flush(cf);
+		}
+
+		if (!list_empty(&ci->i_dirty_item)) {
+			pr_warn_ratelimited(
+				" dropping dirty %s state for %p %lld\n",
+				ceph_cap_string(ci->i_dirty_caps),
+				inode, ceph_ino(inode));
+			ci->i_dirty_caps = 0;
+			list_del_init(&ci->i_dirty_item);
+			dirty_dropped = true;
+		}
+		if (!list_empty(&ci->i_flushing_item)) {
+			pr_warn_ratelimited(
+				" dropping dirty+flushing %s state for %p %lld\n",
+				ceph_cap_string(ci->i_flushing_caps),
+				inode, ceph_ino(inode));
+			ci->i_flushing_caps = 0;
+			list_del_init(&ci->i_flushing_item);
+			mdsc->num_cap_flushing--;
+			dirty_dropped = true;
+		}
+		spin_unlock(&mdsc->cap_dirty_lock);
+
+		if (dirty_dropped) {
+			mapping_set_error(inode->i_mapping, -EIO);
+
+			if (ci->i_wrbuffer_ref_head == 0 &&
+			    ci->i_wr_ref == 0 &&
+			    ci->i_dirty_caps == 0 &&
+			    ci->i_flushing_caps == 0) {
+				ceph_put_snap_context(ci->i_head_snapc);
+				ci->i_head_snapc = NULL;
+			}
+		}
+
+		if (atomic_read(&ci->i_filelock_ref) > 0) {
+			/* make further file lock syscall return -EIO */
+			ci->i_ceph_flags |= CEPH_I_ERROR_FILELOCK;
+			pr_warn_ratelimited(" dropping file locks for %p %lld\n",
+					    inode, ceph_ino(inode));
+		}
+
+		if (!ci->i_dirty_caps && ci->i_prealloc_cap_flush) {
+			cf = ci->i_prealloc_cap_flush;
+			ci->i_prealloc_cap_flush = NULL;
+			if (!cf->is_capsnap)
+				ceph_free_cap_flush(cf);
+		}
+
+		if (!list_empty(&ci->i_cap_snaps))
+			iputs = remove_capsnaps(mdsc, inode);
+	}
+	if (dirty_dropped)
+		++iputs;
+	return iputs;
+}
diff --git a/fs/ceph/ceph_frag.c b/fs/ceph/ceph_frag.c
new file mode 100644
index 000000000..6f67d5b88
--- /dev/null
+++ b/fs/ceph/ceph_frag.c
@@ -0,0 +1,23 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Ceph 'frag' type
+ */
+#include <linux/module.h>
+#include <linux/ceph/types.h>
+
+int ceph_frag_compare(__u32 a, __u32 b)
+{
+	unsigned va = ceph_frag_value(a);
+	unsigned vb = ceph_frag_value(b);
+	if (va < vb)
+		return -1;
+	if (va > vb)
+		return 1;
+	va = ceph_frag_bits(a);
+	vb = ceph_frag_bits(b);
+	if (va < vb)
+		return -1;
+	if (va > vb)
+		return 1;
+	return 0;
+}
diff --git a/fs/ceph/debugfs.c b/fs/ceph/debugfs.c
new file mode 100644
index 000000000..3904333fa
--- /dev/null
+++ b/fs/ceph/debugfs.c
@@ -0,0 +1,485 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/ceph/ceph_debug.h>
+
+#include <linux/device.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <linux/ctype.h>
+#include <linux/debugfs.h>
+#include <linux/seq_file.h>
+#include <linux/math64.h>
+#include <linux/ktime.h>
+
+#include <linux/ceph/libceph.h>
+#include <linux/ceph/mon_client.h>
+#include <linux/ceph/auth.h>
+#include <linux/ceph/debugfs.h>
+
+#include "super.h"
+
+#ifdef CONFIG_DEBUG_FS
+
+#include "mds_client.h"
+#include "metric.h"
+
+static int mdsmap_show(struct seq_file *s, void *p)
+{
+	int i;
+	struct ceph_fs_client *fsc = s->private;
+	struct ceph_mdsmap *mdsmap;
+
+	if (!fsc->mdsc || !fsc->mdsc->mdsmap)
+		return 0;
+	mdsmap = fsc->mdsc->mdsmap;
+	seq_printf(s, "epoch %d\n", mdsmap->m_epoch);
+	seq_printf(s, "root %d\n", mdsmap->m_root);
+	seq_printf(s, "max_mds %d\n", mdsmap->m_max_mds);
+	seq_printf(s, "session_timeout %d\n", mdsmap->m_session_timeout);
+	seq_printf(s, "session_autoclose %d\n", mdsmap->m_session_autoclose);
+	for (i = 0; i < mdsmap->possible_max_rank; i++) {
+		struct ceph_entity_addr *addr = &mdsmap->m_info[i].addr;
+		int state = mdsmap->m_info[i].state;
+		seq_printf(s, "\tmds%d\t%s\t(%s)\n", i,
+			       ceph_pr_addr(addr),
+			       ceph_mds_state_name(state));
+	}
+	return 0;
+}
+
+/*
+ * mdsc debugfs
+ */
+static int mdsc_show(struct seq_file *s, void *p)
+{
+	struct ceph_fs_client *fsc = s->private;
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	struct ceph_mds_request *req;
+	struct rb_node *rp;
+	int pathlen = 0;
+	u64 pathbase;
+	char *path;
+
+	mutex_lock(&mdsc->mutex);
+	for (rp = rb_first(&mdsc->request_tree); rp; rp = rb_next(rp)) {
+		req = rb_entry(rp, struct ceph_mds_request, r_node);
+
+		if (req->r_request && req->r_session)
+			seq_printf(s, "%lld\tmds%d\t", req->r_tid,
+				   req->r_session->s_mds);
+		else if (!req->r_request)
+			seq_printf(s, "%lld\t(no request)\t", req->r_tid);
+		else
+			seq_printf(s, "%lld\t(no session)\t", req->r_tid);
+
+		seq_printf(s, "%s", ceph_mds_op_name(req->r_op));
+
+		if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags))
+			seq_puts(s, "\t(unsafe)");
+		else
+			seq_puts(s, "\t");
+
+		if (req->r_inode) {
+			seq_printf(s, " #%llx", ceph_ino(req->r_inode));
+		} else if (req->r_dentry) {
+			path = ceph_mdsc_build_path(req->r_dentry, &pathlen,
+						    &pathbase, 0);
+			if (IS_ERR(path))
+				path = NULL;
+			spin_lock(&req->r_dentry->d_lock);
+			seq_printf(s, " #%llx/%pd (%s)",
+				   ceph_ino(d_inode(req->r_dentry->d_parent)),
+				   req->r_dentry,
+				   path ? path : "");
+			spin_unlock(&req->r_dentry->d_lock);
+			ceph_mdsc_free_path(path, pathlen);
+		} else if (req->r_path1) {
+			seq_printf(s, " #%llx/%s", req->r_ino1.ino,
+				   req->r_path1);
+		} else {
+			seq_printf(s, " #%llx", req->r_ino1.ino);
+		}
+
+		if (req->r_old_dentry) {
+			path = ceph_mdsc_build_path(req->r_old_dentry, &pathlen,
+						    &pathbase, 0);
+			if (IS_ERR(path))
+				path = NULL;
+			spin_lock(&req->r_old_dentry->d_lock);
+			seq_printf(s, " #%llx/%pd (%s)",
+				   req->r_old_dentry_dir ?
+				   ceph_ino(req->r_old_dentry_dir) : 0,
+				   req->r_old_dentry,
+				   path ? path : "");
+			spin_unlock(&req->r_old_dentry->d_lock);
+			ceph_mdsc_free_path(path, pathlen);
+		} else if (req->r_path2 && req->r_op != CEPH_MDS_OP_SYMLINK) {
+			if (req->r_ino2.ino)
+				seq_printf(s, " #%llx/%s", req->r_ino2.ino,
+					   req->r_path2);
+			else
+				seq_printf(s, " %s", req->r_path2);
+		}
+
+		seq_puts(s, "\n");
+	}
+	mutex_unlock(&mdsc->mutex);
+
+	return 0;
+}
+
+#define CEPH_LAT_METRIC_SHOW(name, total, avg, min, max, sq) {		\
+	s64 _total, _avg, _min, _max, _sq, _st;				\
+	_avg = ktime_to_us(avg);					\
+	_min = ktime_to_us(min == KTIME_MAX ? 0 : min);			\
+	_max = ktime_to_us(max);					\
+	_total = total - 1;						\
+	_sq = _total > 0 ? DIV64_U64_ROUND_CLOSEST(sq, _total) : 0;	\
+	_st = int_sqrt64(_sq);						\
+	_st = ktime_to_us(_st);						\
+	seq_printf(s, "%-14s%-12lld%-16lld%-16lld%-16lld%lld\n",	\
+		   name, total, _avg, _min, _max, _st);			\
+}
+
+#define CEPH_SZ_METRIC_SHOW(name, total, avg, min, max, sum) {		\
+	u64 _min = min == U64_MAX ? 0 : min;				\
+	seq_printf(s, "%-14s%-12lld%-16llu%-16llu%-16llu%llu\n",	\
+		   name, total, avg, _min, max, sum);			\
+}
+
+static int metrics_file_show(struct seq_file *s, void *p)
+{
+	struct ceph_fs_client *fsc = s->private;
+	struct ceph_client_metric *m = &fsc->mdsc->metric;
+
+	seq_printf(s, "item                               total\n");
+	seq_printf(s, "------------------------------------------\n");
+	seq_printf(s, "%-35s%lld\n", "total inodes",
+		   percpu_counter_sum(&m->total_inodes));
+	seq_printf(s, "%-35s%lld\n", "opened files",
+		   atomic64_read(&m->opened_files));
+	seq_printf(s, "%-35s%lld\n", "pinned i_caps",
+		   atomic64_read(&m->total_caps));
+	seq_printf(s, "%-35s%lld\n", "opened inodes",
+		   percpu_counter_sum(&m->opened_inodes));
+	return 0;
+}
+
+static const char * const metric_str[] = {
+	"read",
+	"write",
+	"metadata",
+	"copyfrom"
+};
+static int metrics_latency_show(struct seq_file *s, void *p)
+{
+	struct ceph_fs_client *fsc = s->private;
+	struct ceph_client_metric *cm = &fsc->mdsc->metric;
+	struct ceph_metric *m;
+	s64 total, avg, min, max, sq;
+	int i;
+
+	seq_printf(s, "item          total       avg_lat(us)     min_lat(us)     max_lat(us)     stdev(us)\n");
+	seq_printf(s, "-----------------------------------------------------------------------------------\n");
+
+	for (i = 0; i < METRIC_MAX; i++) {
+		m = &cm->metric[i];
+		spin_lock(&m->lock);
+		total = m->total;
+		avg = m->latency_avg;
+		min = m->latency_min;
+		max = m->latency_max;
+		sq = m->latency_sq_sum;
+		spin_unlock(&m->lock);
+		CEPH_LAT_METRIC_SHOW(metric_str[i], total, avg, min, max, sq);
+	}
+
+	return 0;
+}
+
+static int metrics_size_show(struct seq_file *s, void *p)
+{
+	struct ceph_fs_client *fsc = s->private;
+	struct ceph_client_metric *cm = &fsc->mdsc->metric;
+	struct ceph_metric *m;
+	s64 total;
+	u64 sum, avg, min, max;
+	int i;
+
+	seq_printf(s, "item          total       avg_sz(bytes)   min_sz(bytes)   max_sz(bytes)  total_sz(bytes)\n");
+	seq_printf(s, "----------------------------------------------------------------------------------------\n");
+
+	for (i = 0; i < METRIC_MAX; i++) {
+		/* skip 'metadata' as it doesn't use the size metric */
+		if (i == METRIC_METADATA)
+			continue;
+		m = &cm->metric[i];
+		spin_lock(&m->lock);
+		total = m->total;
+		sum = m->size_sum;
+		avg = total > 0 ? DIV64_U64_ROUND_CLOSEST(sum, total) : 0;
+		min = m->size_min;
+		max = m->size_max;
+		spin_unlock(&m->lock);
+		CEPH_SZ_METRIC_SHOW(metric_str[i], total, avg, min, max, sum);
+	}
+
+	return 0;
+}
+
+static int metrics_caps_show(struct seq_file *s, void *p)
+{
+	struct ceph_fs_client *fsc = s->private;
+	struct ceph_client_metric *m = &fsc->mdsc->metric;
+	int nr_caps = 0;
+
+	seq_printf(s, "item          total           miss            hit\n");
+	seq_printf(s, "-------------------------------------------------\n");
+
+	seq_printf(s, "%-14s%-16lld%-16lld%lld\n", "d_lease",
+		   atomic64_read(&m->total_dentries),
+		   percpu_counter_sum(&m->d_lease_mis),
+		   percpu_counter_sum(&m->d_lease_hit));
+
+	nr_caps = atomic64_read(&m->total_caps);
+	seq_printf(s, "%-14s%-16d%-16lld%lld\n", "caps", nr_caps,
+		   percpu_counter_sum(&m->i_caps_mis),
+		   percpu_counter_sum(&m->i_caps_hit));
+
+	return 0;
+}
+
+static int caps_show_cb(struct inode *inode, int mds, void *p)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct seq_file *s = p;
+	struct ceph_cap *cap;
+
+	spin_lock(&ci->i_ceph_lock);
+	cap = __get_cap_for_mds(ci, mds);
+	if (cap)
+		seq_printf(s, "0x%-17llx%-3d%-17s%-17s\n", ceph_ino(inode),
+			   cap->session->s_mds,
+			   ceph_cap_string(cap->issued),
+			   ceph_cap_string(cap->implemented));
+	spin_unlock(&ci->i_ceph_lock);
+	return 0;
+}
+
+static int caps_show(struct seq_file *s, void *p)
+{
+	struct ceph_fs_client *fsc = s->private;
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	int total, avail, used, reserved, min, i;
+	struct cap_wait	*cw;
+
+	ceph_reservation_status(fsc, &total, &avail, &used, &reserved, &min);
+	seq_printf(s, "total\t\t%d\n"
+		   "avail\t\t%d\n"
+		   "used\t\t%d\n"
+		   "reserved\t%d\n"
+		   "min\t\t%d\n\n",
+		   total, avail, used, reserved, min);
+	seq_printf(s, "ino              mds  issued           implemented\n");
+	seq_printf(s, "--------------------------------------------------\n");
+
+	mutex_lock(&mdsc->mutex);
+	for (i = 0; i < mdsc->max_sessions; i++) {
+		struct ceph_mds_session *session;
+
+		session = __ceph_lookup_mds_session(mdsc, i);
+		if (!session)
+			continue;
+		mutex_unlock(&mdsc->mutex);
+		mutex_lock(&session->s_mutex);
+		ceph_iterate_session_caps(session, caps_show_cb, s);
+		mutex_unlock(&session->s_mutex);
+		ceph_put_mds_session(session);
+		mutex_lock(&mdsc->mutex);
+	}
+	mutex_unlock(&mdsc->mutex);
+
+	seq_printf(s, "\n\nWaiters:\n--------\n");
+	seq_printf(s, "tgid         ino                need             want\n");
+	seq_printf(s, "-----------------------------------------------------\n");
+
+	spin_lock(&mdsc->caps_list_lock);
+	list_for_each_entry(cw, &mdsc->cap_wait_list, list) {
+		seq_printf(s, "%-13d0x%-17llx%-17s%-17s\n", cw->tgid, cw->ino,
+				ceph_cap_string(cw->need),
+				ceph_cap_string(cw->want));
+	}
+	spin_unlock(&mdsc->caps_list_lock);
+
+	return 0;
+}
+
+static int mds_sessions_show(struct seq_file *s, void *ptr)
+{
+	struct ceph_fs_client *fsc = s->private;
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	struct ceph_auth_client *ac = fsc->client->monc.auth;
+	struct ceph_options *opt = fsc->client->options;
+	int mds;
+
+	mutex_lock(&mdsc->mutex);
+
+	/* The 'num' portion of an 'entity name' */
+	seq_printf(s, "global_id %llu\n", ac->global_id);
+
+	/* The -o name mount argument */
+	seq_printf(s, "name \"%s\"\n", opt->name ? opt->name : "");
+
+	/* The list of MDS session rank+state */
+	for (mds = 0; mds < mdsc->max_sessions; mds++) {
+		struct ceph_mds_session *session =
+			__ceph_lookup_mds_session(mdsc, mds);
+		if (!session) {
+			continue;
+		}
+		mutex_unlock(&mdsc->mutex);
+		seq_printf(s, "mds.%d %s\n",
+				session->s_mds,
+				ceph_session_state_name(session->s_state));
+
+		ceph_put_mds_session(session);
+		mutex_lock(&mdsc->mutex);
+	}
+	mutex_unlock(&mdsc->mutex);
+
+	return 0;
+}
+
+static int status_show(struct seq_file *s, void *p)
+{
+	struct ceph_fs_client *fsc = s->private;
+	struct ceph_entity_inst *inst = &fsc->client->msgr.inst;
+	struct ceph_entity_addr *client_addr = ceph_client_addr(fsc->client);
+
+	seq_printf(s, "instance: %s.%lld %s/%u\n", ENTITY_NAME(inst->name),
+		   ceph_pr_addr(client_addr), le32_to_cpu(client_addr->nonce));
+	seq_printf(s, "blocklisted: %s\n", fsc->blocklisted ? "true" : "false");
+
+	return 0;
+}
+
+DEFINE_SHOW_ATTRIBUTE(mdsmap);
+DEFINE_SHOW_ATTRIBUTE(mdsc);
+DEFINE_SHOW_ATTRIBUTE(caps);
+DEFINE_SHOW_ATTRIBUTE(mds_sessions);
+DEFINE_SHOW_ATTRIBUTE(status);
+DEFINE_SHOW_ATTRIBUTE(metrics_file);
+DEFINE_SHOW_ATTRIBUTE(metrics_latency);
+DEFINE_SHOW_ATTRIBUTE(metrics_size);
+DEFINE_SHOW_ATTRIBUTE(metrics_caps);
+
+
+/*
+ * debugfs
+ */
+static int congestion_kb_set(void *data, u64 val)
+{
+	struct ceph_fs_client *fsc = (struct ceph_fs_client *)data;
+
+	fsc->mount_options->congestion_kb = (int)val;
+	return 0;
+}
+
+static int congestion_kb_get(void *data, u64 *val)
+{
+	struct ceph_fs_client *fsc = (struct ceph_fs_client *)data;
+
+	*val = (u64)fsc->mount_options->congestion_kb;
+	return 0;
+}
+
+DEFINE_SIMPLE_ATTRIBUTE(congestion_kb_fops, congestion_kb_get,
+			congestion_kb_set, "%llu\n");
+
+
+void ceph_fs_debugfs_cleanup(struct ceph_fs_client *fsc)
+{
+	dout("ceph_fs_debugfs_cleanup\n");
+	debugfs_remove(fsc->debugfs_bdi);
+	debugfs_remove(fsc->debugfs_congestion_kb);
+	debugfs_remove(fsc->debugfs_mdsmap);
+	debugfs_remove(fsc->debugfs_mds_sessions);
+	debugfs_remove(fsc->debugfs_caps);
+	debugfs_remove(fsc->debugfs_status);
+	debugfs_remove(fsc->debugfs_mdsc);
+	debugfs_remove_recursive(fsc->debugfs_metrics_dir);
+}
+
+void ceph_fs_debugfs_init(struct ceph_fs_client *fsc)
+{
+	char name[100];
+
+	dout("ceph_fs_debugfs_init\n");
+	fsc->debugfs_congestion_kb =
+		debugfs_create_file("writeback_congestion_kb",
+				    0600,
+				    fsc->client->debugfs_dir,
+				    fsc,
+				    &congestion_kb_fops);
+
+	snprintf(name, sizeof(name), "../../bdi/%s",
+		 bdi_dev_name(fsc->sb->s_bdi));
+	fsc->debugfs_bdi =
+		debugfs_create_symlink("bdi",
+				       fsc->client->debugfs_dir,
+				       name);
+
+	fsc->debugfs_mdsmap = debugfs_create_file("mdsmap",
+					0400,
+					fsc->client->debugfs_dir,
+					fsc,
+					&mdsmap_fops);
+
+	fsc->debugfs_mds_sessions = debugfs_create_file("mds_sessions",
+					0400,
+					fsc->client->debugfs_dir,
+					fsc,
+					&mds_sessions_fops);
+
+	fsc->debugfs_mdsc = debugfs_create_file("mdsc",
+						0400,
+						fsc->client->debugfs_dir,
+						fsc,
+						&mdsc_fops);
+
+	fsc->debugfs_caps = debugfs_create_file("caps",
+						0400,
+						fsc->client->debugfs_dir,
+						fsc,
+						&caps_fops);
+
+	fsc->debugfs_status = debugfs_create_file("status",
+						  0400,
+						  fsc->client->debugfs_dir,
+						  fsc,
+						  &status_fops);
+
+	fsc->debugfs_metrics_dir = debugfs_create_dir("metrics",
+						      fsc->client->debugfs_dir);
+
+	debugfs_create_file("file", 0400, fsc->debugfs_metrics_dir, fsc,
+			    &metrics_file_fops);
+	debugfs_create_file("latency", 0400, fsc->debugfs_metrics_dir, fsc,
+			    &metrics_latency_fops);
+	debugfs_create_file("size", 0400, fsc->debugfs_metrics_dir, fsc,
+			    &metrics_size_fops);
+	debugfs_create_file("caps", 0400, fsc->debugfs_metrics_dir, fsc,
+			    &metrics_caps_fops);
+}
+
+
+#else  /* CONFIG_DEBUG_FS */
+
+void ceph_fs_debugfs_init(struct ceph_fs_client *fsc)
+{
+}
+
+void ceph_fs_debugfs_cleanup(struct ceph_fs_client *fsc)
+{
+}
+
+#endif  /* CONFIG_DEBUG_FS */
diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
new file mode 100644
index 000000000..e7e2ebac3
--- /dev/null
+++ b/fs/ceph/dir.c
@@ -0,0 +1,2064 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/ceph/ceph_debug.h>
+
+#include <linux/spinlock.h>
+#include <linux/namei.h>
+#include <linux/slab.h>
+#include <linux/sched.h>
+#include <linux/xattr.h>
+
+#include "super.h"
+#include "mds_client.h"
+
+/*
+ * Directory operations: readdir, lookup, create, link, unlink,
+ * rename, etc.
+ */
+
+/*
+ * Ceph MDS operations are specified in terms of a base ino and
+ * relative path.  Thus, the client can specify an operation on a
+ * specific inode (e.g., a getattr due to fstat(2)), or as a path
+ * relative to, say, the root directory.
+ *
+ * Normally, we limit ourselves to strict inode ops (no path component)
+ * or dentry operations (a single path component relative to an ino).  The
+ * exception to this is open_root_dentry(), which will open the mount
+ * point by name.
+ */
+
+const struct dentry_operations ceph_dentry_ops;
+
+static bool __dentry_lease_is_valid(struct ceph_dentry_info *di);
+static int __dir_lease_try_check(const struct dentry *dentry);
+
+/*
+ * Initialize ceph dentry state.
+ */
+static int ceph_d_init(struct dentry *dentry)
+{
+	struct ceph_dentry_info *di;
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(dentry->d_sb);
+
+	di = kmem_cache_zalloc(ceph_dentry_cachep, GFP_KERNEL);
+	if (!di)
+		return -ENOMEM;          /* oh well */
+
+	di->dentry = dentry;
+	di->lease_session = NULL;
+	di->time = jiffies;
+	dentry->d_fsdata = di;
+	INIT_LIST_HEAD(&di->lease_list);
+
+	atomic64_inc(&mdsc->metric.total_dentries);
+
+	return 0;
+}
+
+/*
+ * for f_pos for readdir:
+ * - hash order:
+ *	(0xff << 52) | ((24 bits hash) << 28) |
+ *	(the nth entry has hash collision);
+ * - frag+name order;
+ *	((frag value) << 28) | (the nth entry in frag);
+ */
+#define OFFSET_BITS	28
+#define OFFSET_MASK	((1 << OFFSET_BITS) - 1)
+#define HASH_ORDER	(0xffull << (OFFSET_BITS + 24))
+loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order)
+{
+	loff_t fpos = ((loff_t)high << 28) | (loff_t)off;
+	if (hash_order)
+		fpos |= HASH_ORDER;
+	return fpos;
+}
+
+static bool is_hash_order(loff_t p)
+{
+	return (p & HASH_ORDER) == HASH_ORDER;
+}
+
+static unsigned fpos_frag(loff_t p)
+{
+	return p >> OFFSET_BITS;
+}
+
+static unsigned fpos_hash(loff_t p)
+{
+	return ceph_frag_value(fpos_frag(p));
+}
+
+static unsigned fpos_off(loff_t p)
+{
+	return p & OFFSET_MASK;
+}
+
+static int fpos_cmp(loff_t l, loff_t r)
+{
+	int v = ceph_frag_compare(fpos_frag(l), fpos_frag(r));
+	if (v)
+		return v;
+	return (int)(fpos_off(l) - fpos_off(r));
+}
+
+/*
+ * make note of the last dentry we read, so we can
+ * continue at the same lexicographical point,
+ * regardless of what dir changes take place on the
+ * server.
+ */
+static int note_last_dentry(struct ceph_dir_file_info *dfi, const char *name,
+		            int len, unsigned next_offset)
+{
+	char *buf = kmalloc(len+1, GFP_KERNEL);
+	if (!buf)
+		return -ENOMEM;
+	kfree(dfi->last_name);
+	dfi->last_name = buf;
+	memcpy(dfi->last_name, name, len);
+	dfi->last_name[len] = 0;
+	dfi->next_offset = next_offset;
+	dout("note_last_dentry '%s'\n", dfi->last_name);
+	return 0;
+}
+
+
+static struct dentry *
+__dcache_find_get_entry(struct dentry *parent, u64 idx,
+			struct ceph_readdir_cache_control *cache_ctl)
+{
+	struct inode *dir = d_inode(parent);
+	struct dentry *dentry;
+	unsigned idx_mask = (PAGE_SIZE / sizeof(struct dentry *)) - 1;
+	loff_t ptr_pos = idx * sizeof(struct dentry *);
+	pgoff_t ptr_pgoff = ptr_pos >> PAGE_SHIFT;
+
+	if (ptr_pos >= i_size_read(dir))
+		return NULL;
+
+	if (!cache_ctl->page || ptr_pgoff != page_index(cache_ctl->page)) {
+		ceph_readdir_cache_release(cache_ctl);
+		cache_ctl->page = find_lock_page(&dir->i_data, ptr_pgoff);
+		if (!cache_ctl->page) {
+			dout(" page %lu not found\n", ptr_pgoff);
+			return ERR_PTR(-EAGAIN);
+		}
+		/* reading/filling the cache are serialized by
+		   i_rwsem, no need to use page lock */
+		unlock_page(cache_ctl->page);
+		cache_ctl->dentries = kmap(cache_ctl->page);
+	}
+
+	cache_ctl->index = idx & idx_mask;
+
+	rcu_read_lock();
+	spin_lock(&parent->d_lock);
+	/* check i_size again here, because empty directory can be
+	 * marked as complete while not holding the i_rwsem. */
+	if (ceph_dir_is_complete_ordered(dir) && ptr_pos < i_size_read(dir))
+		dentry = cache_ctl->dentries[cache_ctl->index];
+	else
+		dentry = NULL;
+	spin_unlock(&parent->d_lock);
+	if (dentry && !lockref_get_not_dead(&dentry->d_lockref))
+		dentry = NULL;
+	rcu_read_unlock();
+	return dentry ? : ERR_PTR(-EAGAIN);
+}
+
+/*
+ * When possible, we try to satisfy a readdir by peeking at the
+ * dcache.  We make this work by carefully ordering dentries on
+ * d_child when we initially get results back from the MDS, and
+ * falling back to a "normal" sync readdir if any dentries in the dir
+ * are dropped.
+ *
+ * Complete dir indicates that we have all dentries in the dir.  It is
+ * defined IFF we hold CEPH_CAP_FILE_SHARED (which will be revoked by
+ * the MDS if/when the directory is modified).
+ */
+static int __dcache_readdir(struct file *file,  struct dir_context *ctx,
+			    int shared_gen)
+{
+	struct ceph_dir_file_info *dfi = file->private_data;
+	struct dentry *parent = file->f_path.dentry;
+	struct inode *dir = d_inode(parent);
+	struct dentry *dentry, *last = NULL;
+	struct ceph_dentry_info *di;
+	struct ceph_readdir_cache_control cache_ctl = {};
+	u64 idx = 0;
+	int err = 0;
+
+	dout("__dcache_readdir %p v%u at %llx\n", dir, (unsigned)shared_gen, ctx->pos);
+
+	/* search start position */
+	if (ctx->pos > 2) {
+		u64 count = div_u64(i_size_read(dir), sizeof(struct dentry *));
+		while (count > 0) {
+			u64 step = count >> 1;
+			dentry = __dcache_find_get_entry(parent, idx + step,
+							 &cache_ctl);
+			if (!dentry) {
+				/* use linar search */
+				idx = 0;
+				break;
+			}
+			if (IS_ERR(dentry)) {
+				err = PTR_ERR(dentry);
+				goto out;
+			}
+			di = ceph_dentry(dentry);
+			spin_lock(&dentry->d_lock);
+			if (fpos_cmp(di->offset, ctx->pos) < 0) {
+				idx += step + 1;
+				count -= step + 1;
+			} else {
+				count = step;
+			}
+			spin_unlock(&dentry->d_lock);
+			dput(dentry);
+		}
+
+		dout("__dcache_readdir %p cache idx %llu\n", dir, idx);
+	}
+
+
+	for (;;) {
+		bool emit_dentry = false;
+		dentry = __dcache_find_get_entry(parent, idx++, &cache_ctl);
+		if (!dentry) {
+			dfi->file_info.flags |= CEPH_F_ATEND;
+			err = 0;
+			break;
+		}
+		if (IS_ERR(dentry)) {
+			err = PTR_ERR(dentry);
+			goto out;
+		}
+
+		spin_lock(&dentry->d_lock);
+		di = ceph_dentry(dentry);
+		if (d_unhashed(dentry) ||
+		    d_really_is_negative(dentry) ||
+		    di->lease_shared_gen != shared_gen) {
+			spin_unlock(&dentry->d_lock);
+			dput(dentry);
+			err = -EAGAIN;
+			goto out;
+		}
+		if (fpos_cmp(ctx->pos, di->offset) <= 0) {
+			__ceph_dentry_dir_lease_touch(di);
+			emit_dentry = true;
+		}
+		spin_unlock(&dentry->d_lock);
+
+		if (emit_dentry) {
+			dout(" %llx dentry %p %pd %p\n", di->offset,
+			     dentry, dentry, d_inode(dentry));
+			ctx->pos = di->offset;
+			if (!dir_emit(ctx, dentry->d_name.name,
+				      dentry->d_name.len, ceph_present_inode(d_inode(dentry)),
+				      d_inode(dentry)->i_mode >> 12)) {
+				dput(dentry);
+				err = 0;
+				break;
+			}
+			ctx->pos++;
+
+			if (last)
+				dput(last);
+			last = dentry;
+		} else {
+			dput(dentry);
+		}
+	}
+out:
+	ceph_readdir_cache_release(&cache_ctl);
+	if (last) {
+		int ret;
+		di = ceph_dentry(last);
+		ret = note_last_dentry(dfi, last->d_name.name, last->d_name.len,
+				       fpos_off(di->offset) + 1);
+		if (ret < 0)
+			err = ret;
+		dput(last);
+		/* last_name no longer match cache index */
+		if (dfi->readdir_cache_idx >= 0) {
+			dfi->readdir_cache_idx = -1;
+			dfi->dir_release_count = 0;
+		}
+	}
+	return err;
+}
+
+static bool need_send_readdir(struct ceph_dir_file_info *dfi, loff_t pos)
+{
+	if (!dfi->last_readdir)
+		return true;
+	if (is_hash_order(pos))
+		return !ceph_frag_contains_value(dfi->frag, fpos_hash(pos));
+	else
+		return dfi->frag != fpos_frag(pos);
+}
+
+static int ceph_readdir(struct file *file, struct dir_context *ctx)
+{
+	struct ceph_dir_file_info *dfi = file->private_data;
+	struct inode *inode = file_inode(file);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	int i;
+	int err;
+	unsigned frag = -1;
+	struct ceph_mds_reply_info_parsed *rinfo;
+
+	dout("readdir %p file %p pos %llx\n", inode, file, ctx->pos);
+	if (dfi->file_info.flags & CEPH_F_ATEND)
+		return 0;
+
+	/* always start with . and .. */
+	if (ctx->pos == 0) {
+		dout("readdir off 0 -> '.'\n");
+		if (!dir_emit(ctx, ".", 1, ceph_present_inode(inode),
+			    inode->i_mode >> 12))
+			return 0;
+		ctx->pos = 1;
+	}
+	if (ctx->pos == 1) {
+		u64 ino;
+		struct dentry *dentry = file->f_path.dentry;
+
+		spin_lock(&dentry->d_lock);
+		ino = ceph_present_inode(dentry->d_parent->d_inode);
+		spin_unlock(&dentry->d_lock);
+
+		dout("readdir off 1 -> '..'\n");
+		if (!dir_emit(ctx, "..", 2, ino, inode->i_mode >> 12))
+			return 0;
+		ctx->pos = 2;
+	}
+
+	spin_lock(&ci->i_ceph_lock);
+	/* request Fx cap. if have Fx, we don't need to release Fs cap
+	 * for later create/unlink. */
+	__ceph_touch_fmode(ci, mdsc, CEPH_FILE_MODE_WR);
+	/* can we use the dcache? */
+	if (ceph_test_mount_opt(fsc, DCACHE) &&
+	    !ceph_test_mount_opt(fsc, NOASYNCREADDIR) &&
+	    ceph_snap(inode) != CEPH_SNAPDIR &&
+	    __ceph_dir_is_complete_ordered(ci) &&
+	    __ceph_caps_issued_mask_metric(ci, CEPH_CAP_FILE_SHARED, 1)) {
+		int shared_gen = atomic_read(&ci->i_shared_gen);
+
+		spin_unlock(&ci->i_ceph_lock);
+		err = __dcache_readdir(file, ctx, shared_gen);
+		if (err != -EAGAIN)
+			return err;
+	} else {
+		spin_unlock(&ci->i_ceph_lock);
+	}
+
+	/* proceed with a normal readdir */
+more:
+	/* do we have the correct frag content buffered? */
+	if (need_send_readdir(dfi, ctx->pos)) {
+		struct ceph_mds_request *req;
+		int op = ceph_snap(inode) == CEPH_SNAPDIR ?
+			CEPH_MDS_OP_LSSNAP : CEPH_MDS_OP_READDIR;
+
+		/* discard old result, if any */
+		if (dfi->last_readdir) {
+			ceph_mdsc_put_request(dfi->last_readdir);
+			dfi->last_readdir = NULL;
+		}
+
+		if (is_hash_order(ctx->pos)) {
+			/* fragtree isn't always accurate. choose frag
+			 * based on previous reply when possible. */
+			if (frag == (unsigned)-1)
+				frag = ceph_choose_frag(ci, fpos_hash(ctx->pos),
+							NULL, NULL);
+		} else {
+			frag = fpos_frag(ctx->pos);
+		}
+
+		dout("readdir fetching %llx.%llx frag %x offset '%s'\n",
+		     ceph_vinop(inode), frag, dfi->last_name);
+		req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
+		if (IS_ERR(req))
+			return PTR_ERR(req);
+		err = ceph_alloc_readdir_reply_buffer(req, inode);
+		if (err) {
+			ceph_mdsc_put_request(req);
+			return err;
+		}
+		/* hints to request -> mds selection code */
+		req->r_direct_mode = USE_AUTH_MDS;
+		if (op == CEPH_MDS_OP_READDIR) {
+			req->r_direct_hash = ceph_frag_value(frag);
+			__set_bit(CEPH_MDS_R_DIRECT_IS_HASH, &req->r_req_flags);
+			req->r_inode_drop = CEPH_CAP_FILE_EXCL;
+		}
+		if (dfi->last_name) {
+			req->r_path2 = kstrdup(dfi->last_name, GFP_KERNEL);
+			if (!req->r_path2) {
+				ceph_mdsc_put_request(req);
+				return -ENOMEM;
+			}
+		} else if (is_hash_order(ctx->pos)) {
+			req->r_args.readdir.offset_hash =
+				cpu_to_le32(fpos_hash(ctx->pos));
+		}
+
+		req->r_dir_release_cnt = dfi->dir_release_count;
+		req->r_dir_ordered_cnt = dfi->dir_ordered_count;
+		req->r_readdir_cache_idx = dfi->readdir_cache_idx;
+		req->r_readdir_offset = dfi->next_offset;
+		req->r_args.readdir.frag = cpu_to_le32(frag);
+		req->r_args.readdir.flags =
+				cpu_to_le16(CEPH_READDIR_REPLY_BITFLAGS);
+
+		req->r_inode = inode;
+		ihold(inode);
+		req->r_dentry = dget(file->f_path.dentry);
+		err = ceph_mdsc_do_request(mdsc, NULL, req);
+		if (err < 0) {
+			ceph_mdsc_put_request(req);
+			return err;
+		}
+		dout("readdir got and parsed readdir result=%d on "
+		     "frag %x, end=%d, complete=%d, hash_order=%d\n",
+		     err, frag,
+		     (int)req->r_reply_info.dir_end,
+		     (int)req->r_reply_info.dir_complete,
+		     (int)req->r_reply_info.hash_order);
+
+		rinfo = &req->r_reply_info;
+		if (le32_to_cpu(rinfo->dir_dir->frag) != frag) {
+			frag = le32_to_cpu(rinfo->dir_dir->frag);
+			if (!rinfo->hash_order) {
+				dfi->next_offset = req->r_readdir_offset;
+				/* adjust ctx->pos to beginning of frag */
+				ctx->pos = ceph_make_fpos(frag,
+							  dfi->next_offset,
+							  false);
+			}
+		}
+
+		dfi->frag = frag;
+		dfi->last_readdir = req;
+
+		if (test_bit(CEPH_MDS_R_DID_PREPOPULATE, &req->r_req_flags)) {
+			dfi->readdir_cache_idx = req->r_readdir_cache_idx;
+			if (dfi->readdir_cache_idx < 0) {
+				/* preclude from marking dir ordered */
+				dfi->dir_ordered_count = 0;
+			} else if (ceph_frag_is_leftmost(frag) &&
+				   dfi->next_offset == 2) {
+				/* note dir version at start of readdir so
+				 * we can tell if any dentries get dropped */
+				dfi->dir_release_count = req->r_dir_release_cnt;
+				dfi->dir_ordered_count = req->r_dir_ordered_cnt;
+			}
+		} else {
+			dout("readdir !did_prepopulate\n");
+			/* disable readdir cache */
+			dfi->readdir_cache_idx = -1;
+			/* preclude from marking dir complete */
+			dfi->dir_release_count = 0;
+		}
+
+		/* note next offset and last dentry name */
+		if (rinfo->dir_nr > 0) {
+			struct ceph_mds_reply_dir_entry *rde =
+					rinfo->dir_entries + (rinfo->dir_nr-1);
+			unsigned next_offset = req->r_reply_info.dir_end ?
+					2 : (fpos_off(rde->offset) + 1);
+			err = note_last_dentry(dfi, rde->name, rde->name_len,
+					       next_offset);
+			if (err) {
+				ceph_mdsc_put_request(dfi->last_readdir);
+				dfi->last_readdir = NULL;
+				return err;
+			}
+		} else if (req->r_reply_info.dir_end) {
+			dfi->next_offset = 2;
+			/* keep last name */
+		}
+	}
+
+	rinfo = &dfi->last_readdir->r_reply_info;
+	dout("readdir frag %x num %d pos %llx chunk first %llx\n",
+	     dfi->frag, rinfo->dir_nr, ctx->pos,
+	     rinfo->dir_nr ? rinfo->dir_entries[0].offset : 0LL);
+
+	i = 0;
+	/* search start position */
+	if (rinfo->dir_nr > 0) {
+		int step, nr = rinfo->dir_nr;
+		while (nr > 0) {
+			step = nr >> 1;
+			if (rinfo->dir_entries[i + step].offset < ctx->pos) {
+				i +=  step + 1;
+				nr -= step + 1;
+			} else {
+				nr = step;
+			}
+		}
+	}
+	for (; i < rinfo->dir_nr; i++) {
+		struct ceph_mds_reply_dir_entry *rde = rinfo->dir_entries + i;
+
+		BUG_ON(rde->offset < ctx->pos);
+
+		ctx->pos = rde->offset;
+		dout("readdir (%d/%d) -> %llx '%.*s' %p\n",
+		     i, rinfo->dir_nr, ctx->pos,
+		     rde->name_len, rde->name, &rde->inode.in);
+
+		BUG_ON(!rde->inode.in);
+
+		if (!dir_emit(ctx, rde->name, rde->name_len,
+			      ceph_present_ino(inode->i_sb, le64_to_cpu(rde->inode.in->ino)),
+			      le32_to_cpu(rde->inode.in->mode) >> 12)) {
+			/*
+			 * NOTE: Here no need to put the 'dfi->last_readdir',
+			 * because when dir_emit stops us it's most likely
+			 * doesn't have enough memory, etc. So for next readdir
+			 * it will continue.
+			 */
+			dout("filldir stopping us...\n");
+			return 0;
+		}
+		ctx->pos++;
+	}
+
+	ceph_mdsc_put_request(dfi->last_readdir);
+	dfi->last_readdir = NULL;
+
+	if (dfi->next_offset > 2) {
+		frag = dfi->frag;
+		goto more;
+	}
+
+	/* more frags? */
+	if (!ceph_frag_is_rightmost(dfi->frag)) {
+		frag = ceph_frag_next(dfi->frag);
+		if (is_hash_order(ctx->pos)) {
+			loff_t new_pos = ceph_make_fpos(ceph_frag_value(frag),
+							dfi->next_offset, true);
+			if (new_pos > ctx->pos)
+				ctx->pos = new_pos;
+			/* keep last_name */
+		} else {
+			ctx->pos = ceph_make_fpos(frag, dfi->next_offset,
+							false);
+			kfree(dfi->last_name);
+			dfi->last_name = NULL;
+		}
+		dout("readdir next frag is %x\n", frag);
+		goto more;
+	}
+	dfi->file_info.flags |= CEPH_F_ATEND;
+
+	/*
+	 * if dir_release_count still matches the dir, no dentries
+	 * were released during the whole readdir, and we should have
+	 * the complete dir contents in our cache.
+	 */
+	if (atomic64_read(&ci->i_release_count) ==
+					dfi->dir_release_count) {
+		spin_lock(&ci->i_ceph_lock);
+		if (dfi->dir_ordered_count ==
+				atomic64_read(&ci->i_ordered_count)) {
+			dout(" marking %p complete and ordered\n", inode);
+			/* use i_size to track number of entries in
+			 * readdir cache */
+			BUG_ON(dfi->readdir_cache_idx < 0);
+			i_size_write(inode, dfi->readdir_cache_idx *
+				     sizeof(struct dentry*));
+		} else {
+			dout(" marking %p complete\n", inode);
+		}
+		__ceph_dir_set_complete(ci, dfi->dir_release_count,
+					dfi->dir_ordered_count);
+		spin_unlock(&ci->i_ceph_lock);
+	}
+
+	dout("readdir %p file %p done.\n", inode, file);
+	return 0;
+}
+
+static void reset_readdir(struct ceph_dir_file_info *dfi)
+{
+	if (dfi->last_readdir) {
+		ceph_mdsc_put_request(dfi->last_readdir);
+		dfi->last_readdir = NULL;
+	}
+	kfree(dfi->last_name);
+	dfi->last_name = NULL;
+	dfi->dir_release_count = 0;
+	dfi->readdir_cache_idx = -1;
+	dfi->next_offset = 2;  /* compensate for . and .. */
+	dfi->file_info.flags &= ~CEPH_F_ATEND;
+}
+
+/*
+ * discard buffered readdir content on seekdir(0), or seek to new frag,
+ * or seek prior to current chunk
+ */
+static bool need_reset_readdir(struct ceph_dir_file_info *dfi, loff_t new_pos)
+{
+	struct ceph_mds_reply_info_parsed *rinfo;
+	loff_t chunk_offset;
+	if (new_pos == 0)
+		return true;
+	if (is_hash_order(new_pos)) {
+		/* no need to reset last_name for a forward seek when
+		 * dentries are sotred in hash order */
+	} else if (dfi->frag != fpos_frag(new_pos)) {
+		return true;
+	}
+	rinfo = dfi->last_readdir ? &dfi->last_readdir->r_reply_info : NULL;
+	if (!rinfo || !rinfo->dir_nr)
+		return true;
+	chunk_offset = rinfo->dir_entries[0].offset;
+	return new_pos < chunk_offset ||
+	       is_hash_order(new_pos) != is_hash_order(chunk_offset);
+}
+
+static loff_t ceph_dir_llseek(struct file *file, loff_t offset, int whence)
+{
+	struct ceph_dir_file_info *dfi = file->private_data;
+	struct inode *inode = file->f_mapping->host;
+	loff_t retval;
+
+	inode_lock(inode);
+	retval = -EINVAL;
+	switch (whence) {
+	case SEEK_CUR:
+		offset += file->f_pos;
+		break;
+	case SEEK_SET:
+		break;
+	case SEEK_END:
+		retval = -EOPNOTSUPP;
+		goto out;
+	default:
+		goto out;
+	}
+
+	if (offset >= 0) {
+		if (need_reset_readdir(dfi, offset)) {
+			dout("dir_llseek dropping %p content\n", file);
+			reset_readdir(dfi);
+		} else if (is_hash_order(offset) && offset > file->f_pos) {
+			/* for hash offset, we don't know if a forward seek
+			 * is within same frag */
+			dfi->dir_release_count = 0;
+			dfi->readdir_cache_idx = -1;
+		}
+
+		if (offset != file->f_pos) {
+			file->f_pos = offset;
+			file->f_version = 0;
+			dfi->file_info.flags &= ~CEPH_F_ATEND;
+		}
+		retval = offset;
+	}
+out:
+	inode_unlock(inode);
+	return retval;
+}
+
+/*
+ * Handle lookups for the hidden .snap directory.
+ */
+struct dentry *ceph_handle_snapdir(struct ceph_mds_request *req,
+				   struct dentry *dentry)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(dentry->d_sb);
+	struct inode *parent = d_inode(dentry->d_parent); /* we hold i_rwsem */
+
+	/* .snap dir? */
+	if (ceph_snap(parent) == CEPH_NOSNAP &&
+	    strcmp(dentry->d_name.name, fsc->mount_options->snapdir_name) == 0) {
+		struct dentry *res;
+		struct inode *inode = ceph_get_snapdir(parent);
+
+		res = d_splice_alias(inode, dentry);
+		dout("ENOENT on snapdir %p '%pd', linking to snapdir %p. Spliced dentry %p\n",
+		     dentry, dentry, inode, res);
+		if (res)
+			dentry = res;
+	}
+	return dentry;
+}
+
+/*
+ * Figure out final result of a lookup/open request.
+ *
+ * Mainly, make sure we return the final req->r_dentry (if it already
+ * existed) in place of the original VFS-provided dentry when they
+ * differ.
+ *
+ * Gracefully handle the case where the MDS replies with -ENOENT and
+ * no trace (which it may do, at its discretion, e.g., if it doesn't
+ * care to issue a lease on the negative dentry).
+ */
+struct dentry *ceph_finish_lookup(struct ceph_mds_request *req,
+				  struct dentry *dentry, int err)
+{
+	if (err == -ENOENT) {
+		/* no trace? */
+		err = 0;
+		if (!req->r_reply_info.head->is_dentry) {
+			dout("ENOENT and no trace, dentry %p inode %p\n",
+			     dentry, d_inode(dentry));
+			if (d_really_is_positive(dentry)) {
+				d_drop(dentry);
+				err = -ENOENT;
+			} else {
+				d_add(dentry, NULL);
+			}
+		}
+	}
+	if (err)
+		dentry = ERR_PTR(err);
+	else if (dentry != req->r_dentry)
+		dentry = dget(req->r_dentry);   /* we got spliced */
+	else
+		dentry = NULL;
+	return dentry;
+}
+
+static bool is_root_ceph_dentry(struct inode *inode, struct dentry *dentry)
+{
+	return ceph_ino(inode) == CEPH_INO_ROOT &&
+		strncmp(dentry->d_name.name, ".ceph", 5) == 0;
+}
+
+/*
+ * Look up a single dir entry.  If there is a lookup intent, inform
+ * the MDS so that it gets our 'caps wanted' value in a single op.
+ */
+static struct dentry *ceph_lookup(struct inode *dir, struct dentry *dentry,
+				  unsigned int flags)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(dir->i_sb);
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(dir->i_sb);
+	struct ceph_mds_request *req;
+	int op;
+	int mask;
+	int err;
+
+	dout("lookup %p dentry %p '%pd'\n",
+	     dir, dentry, dentry);
+
+	if (dentry->d_name.len > NAME_MAX)
+		return ERR_PTR(-ENAMETOOLONG);
+
+	/* can we conclude ENOENT locally? */
+	if (d_really_is_negative(dentry)) {
+		struct ceph_inode_info *ci = ceph_inode(dir);
+		struct ceph_dentry_info *di = ceph_dentry(dentry);
+
+		spin_lock(&ci->i_ceph_lock);
+		dout(" dir %p flags are 0x%lx\n", dir, ci->i_ceph_flags);
+		if (strncmp(dentry->d_name.name,
+			    fsc->mount_options->snapdir_name,
+			    dentry->d_name.len) &&
+		    !is_root_ceph_dentry(dir, dentry) &&
+		    ceph_test_mount_opt(fsc, DCACHE) &&
+		    __ceph_dir_is_complete(ci) &&
+		    __ceph_caps_issued_mask_metric(ci, CEPH_CAP_FILE_SHARED, 1)) {
+			__ceph_touch_fmode(ci, mdsc, CEPH_FILE_MODE_RD);
+			spin_unlock(&ci->i_ceph_lock);
+			dout(" dir %p complete, -ENOENT\n", dir);
+			d_add(dentry, NULL);
+			di->lease_shared_gen = atomic_read(&ci->i_shared_gen);
+			return NULL;
+		}
+		spin_unlock(&ci->i_ceph_lock);
+	}
+
+	op = ceph_snap(dir) == CEPH_SNAPDIR ?
+		CEPH_MDS_OP_LOOKUPSNAP : CEPH_MDS_OP_LOOKUP;
+	req = ceph_mdsc_create_request(mdsc, op, USE_ANY_MDS);
+	if (IS_ERR(req))
+		return ERR_CAST(req);
+	req->r_dentry = dget(dentry);
+	req->r_num_caps = 2;
+
+	mask = CEPH_STAT_CAP_INODE | CEPH_CAP_AUTH_SHARED;
+	if (ceph_security_xattr_wanted(dir))
+		mask |= CEPH_CAP_XATTR_SHARED;
+	req->r_args.getattr.mask = cpu_to_le32(mask);
+
+	ihold(dir);
+	req->r_parent = dir;
+	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
+	err = ceph_mdsc_do_request(mdsc, NULL, req);
+	if (err == -ENOENT) {
+		struct dentry *res;
+
+		res = ceph_handle_snapdir(req, dentry);
+		if (IS_ERR(res)) {
+			err = PTR_ERR(res);
+		} else {
+			dentry = res;
+			err = 0;
+		}
+	}
+	dentry = ceph_finish_lookup(req, dentry, err);
+	ceph_mdsc_put_request(req);  /* will dput(dentry) */
+	dout("lookup result=%p\n", dentry);
+	return dentry;
+}
+
+/*
+ * If we do a create but get no trace back from the MDS, follow up with
+ * a lookup (the VFS expects us to link up the provided dentry).
+ */
+int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry)
+{
+	struct dentry *result = ceph_lookup(dir, dentry, 0);
+
+	if (result && !IS_ERR(result)) {
+		/*
+		 * We created the item, then did a lookup, and found
+		 * it was already linked to another inode we already
+		 * had in our cache (and thus got spliced). To not
+		 * confuse VFS (especially when inode is a directory),
+		 * we don't link our dentry to that inode, return an
+		 * error instead.
+		 *
+		 * This event should be rare and it happens only when
+		 * we talk to old MDS. Recent MDS does not send traceless
+		 * reply for request that creates new inode.
+		 */
+		d_drop(result);
+		return -ESTALE;
+	}
+	return PTR_ERR(result);
+}
+
+static int ceph_mknod(struct user_namespace *mnt_userns, struct inode *dir,
+		      struct dentry *dentry, umode_t mode, dev_t rdev)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(dir->i_sb);
+	struct ceph_mds_request *req;
+	struct ceph_acl_sec_ctx as_ctx = {};
+	int err;
+
+	if (ceph_snap(dir) != CEPH_NOSNAP)
+		return -EROFS;
+
+	err = ceph_wait_on_conflict_unlink(dentry);
+	if (err)
+		return err;
+
+	if (ceph_quota_is_max_files_exceeded(dir)) {
+		err = -EDQUOT;
+		goto out;
+	}
+
+	err = ceph_pre_init_acls(dir, &mode, &as_ctx);
+	if (err < 0)
+		goto out;
+	err = ceph_security_init_secctx(dentry, mode, &as_ctx);
+	if (err < 0)
+		goto out;
+
+	dout("mknod in dir %p dentry %p mode 0%ho rdev %d\n",
+	     dir, dentry, mode, rdev);
+	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_MKNOD, USE_AUTH_MDS);
+	if (IS_ERR(req)) {
+		err = PTR_ERR(req);
+		goto out;
+	}
+	req->r_dentry = dget(dentry);
+	req->r_num_caps = 2;
+	req->r_parent = dir;
+	ihold(dir);
+	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
+	req->r_args.mknod.mode = cpu_to_le32(mode);
+	req->r_args.mknod.rdev = cpu_to_le32(rdev);
+	req->r_dentry_drop = CEPH_CAP_FILE_SHARED | CEPH_CAP_AUTH_EXCL;
+	req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
+	if (as_ctx.pagelist) {
+		req->r_pagelist = as_ctx.pagelist;
+		as_ctx.pagelist = NULL;
+	}
+	err = ceph_mdsc_do_request(mdsc, dir, req);
+	if (!err && !req->r_reply_info.head->is_dentry)
+		err = ceph_handle_notrace_create(dir, dentry);
+	ceph_mdsc_put_request(req);
+out:
+	if (!err)
+		ceph_init_inode_acls(d_inode(dentry), &as_ctx);
+	else
+		d_drop(dentry);
+	ceph_release_acl_sec_ctx(&as_ctx);
+	return err;
+}
+
+static int ceph_create(struct user_namespace *mnt_userns, struct inode *dir,
+		       struct dentry *dentry, umode_t mode, bool excl)
+{
+	return ceph_mknod(mnt_userns, dir, dentry, mode, 0);
+}
+
+static int ceph_symlink(struct user_namespace *mnt_userns, struct inode *dir,
+			struct dentry *dentry, const char *dest)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(dir->i_sb);
+	struct ceph_mds_request *req;
+	struct ceph_acl_sec_ctx as_ctx = {};
+	int err;
+
+	if (ceph_snap(dir) != CEPH_NOSNAP)
+		return -EROFS;
+
+	err = ceph_wait_on_conflict_unlink(dentry);
+	if (err)
+		return err;
+
+	if (ceph_quota_is_max_files_exceeded(dir)) {
+		err = -EDQUOT;
+		goto out;
+	}
+
+	err = ceph_security_init_secctx(dentry, S_IFLNK | 0777, &as_ctx);
+	if (err < 0)
+		goto out;
+
+	dout("symlink in dir %p dentry %p to '%s'\n", dir, dentry, dest);
+	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SYMLINK, USE_AUTH_MDS);
+	if (IS_ERR(req)) {
+		err = PTR_ERR(req);
+		goto out;
+	}
+	req->r_path2 = kstrdup(dest, GFP_KERNEL);
+	if (!req->r_path2) {
+		err = -ENOMEM;
+		ceph_mdsc_put_request(req);
+		goto out;
+	}
+	req->r_parent = dir;
+	ihold(dir);
+
+	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
+	req->r_dentry = dget(dentry);
+	req->r_num_caps = 2;
+	req->r_dentry_drop = CEPH_CAP_FILE_SHARED | CEPH_CAP_AUTH_EXCL;
+	req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
+	if (as_ctx.pagelist) {
+		req->r_pagelist = as_ctx.pagelist;
+		as_ctx.pagelist = NULL;
+	}
+	err = ceph_mdsc_do_request(mdsc, dir, req);
+	if (!err && !req->r_reply_info.head->is_dentry)
+		err = ceph_handle_notrace_create(dir, dentry);
+	ceph_mdsc_put_request(req);
+out:
+	if (err)
+		d_drop(dentry);
+	ceph_release_acl_sec_ctx(&as_ctx);
+	return err;
+}
+
+static int ceph_mkdir(struct user_namespace *mnt_userns, struct inode *dir,
+		      struct dentry *dentry, umode_t mode)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(dir->i_sb);
+	struct ceph_mds_request *req;
+	struct ceph_acl_sec_ctx as_ctx = {};
+	int err;
+	int op;
+
+	err = ceph_wait_on_conflict_unlink(dentry);
+	if (err)
+		return err;
+
+	if (ceph_snap(dir) == CEPH_SNAPDIR) {
+		/* mkdir .snap/foo is a MKSNAP */
+		op = CEPH_MDS_OP_MKSNAP;
+		dout("mksnap dir %p snap '%pd' dn %p\n", dir,
+		     dentry, dentry);
+	} else if (ceph_snap(dir) == CEPH_NOSNAP) {
+		dout("mkdir dir %p dn %p mode 0%ho\n", dir, dentry, mode);
+		op = CEPH_MDS_OP_MKDIR;
+	} else {
+		err = -EROFS;
+		goto out;
+	}
+
+	if (op == CEPH_MDS_OP_MKDIR &&
+	    ceph_quota_is_max_files_exceeded(dir)) {
+		err = -EDQUOT;
+		goto out;
+	}
+
+	mode |= S_IFDIR;
+	err = ceph_pre_init_acls(dir, &mode, &as_ctx);
+	if (err < 0)
+		goto out;
+	err = ceph_security_init_secctx(dentry, mode, &as_ctx);
+	if (err < 0)
+		goto out;
+
+	req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
+	if (IS_ERR(req)) {
+		err = PTR_ERR(req);
+		goto out;
+	}
+
+	req->r_dentry = dget(dentry);
+	req->r_num_caps = 2;
+	req->r_parent = dir;
+	ihold(dir);
+	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
+	req->r_args.mkdir.mode = cpu_to_le32(mode);
+	req->r_dentry_drop = CEPH_CAP_FILE_SHARED | CEPH_CAP_AUTH_EXCL;
+	req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
+	if (as_ctx.pagelist) {
+		req->r_pagelist = as_ctx.pagelist;
+		as_ctx.pagelist = NULL;
+	}
+	err = ceph_mdsc_do_request(mdsc, dir, req);
+	if (!err &&
+	    !req->r_reply_info.head->is_target &&
+	    !req->r_reply_info.head->is_dentry)
+		err = ceph_handle_notrace_create(dir, dentry);
+	ceph_mdsc_put_request(req);
+out:
+	if (!err)
+		ceph_init_inode_acls(d_inode(dentry), &as_ctx);
+	else
+		d_drop(dentry);
+	ceph_release_acl_sec_ctx(&as_ctx);
+	return err;
+}
+
+static int ceph_link(struct dentry *old_dentry, struct inode *dir,
+		     struct dentry *dentry)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(dir->i_sb);
+	struct ceph_mds_request *req;
+	int err;
+
+	err = ceph_wait_on_conflict_unlink(dentry);
+	if (err)
+		return err;
+
+	if (ceph_snap(dir) != CEPH_NOSNAP)
+		return -EROFS;
+
+	dout("link in dir %p old_dentry %p dentry %p\n", dir,
+	     old_dentry, dentry);
+	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_LINK, USE_AUTH_MDS);
+	if (IS_ERR(req)) {
+		d_drop(dentry);
+		return PTR_ERR(req);
+	}
+	req->r_dentry = dget(dentry);
+	req->r_num_caps = 2;
+	req->r_old_dentry = dget(old_dentry);
+	req->r_parent = dir;
+	ihold(dir);
+	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
+	req->r_dentry_drop = CEPH_CAP_FILE_SHARED;
+	req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
+	/* release LINK_SHARED on source inode (mds will lock it) */
+	req->r_old_inode_drop = CEPH_CAP_LINK_SHARED | CEPH_CAP_LINK_EXCL;
+	err = ceph_mdsc_do_request(mdsc, dir, req);
+	if (err) {
+		d_drop(dentry);
+	} else if (!req->r_reply_info.head->is_dentry) {
+		ihold(d_inode(old_dentry));
+		d_instantiate(dentry, d_inode(old_dentry));
+	}
+	ceph_mdsc_put_request(req);
+	return err;
+}
+
+static void ceph_async_unlink_cb(struct ceph_mds_client *mdsc,
+				 struct ceph_mds_request *req)
+{
+	struct dentry *dentry = req->r_dentry;
+	struct ceph_fs_client *fsc = ceph_sb_to_client(dentry->d_sb);
+	struct ceph_dentry_info *di = ceph_dentry(dentry);
+	int result = req->r_err ? req->r_err :
+			le32_to_cpu(req->r_reply_info.head->result);
+
+	if (!test_bit(CEPH_DENTRY_ASYNC_UNLINK_BIT, &di->flags))
+		pr_warn("%s dentry %p:%pd async unlink bit is not set\n",
+			__func__, dentry, dentry);
+
+	spin_lock(&fsc->async_unlink_conflict_lock);
+	hash_del_rcu(&di->hnode);
+	spin_unlock(&fsc->async_unlink_conflict_lock);
+
+	spin_lock(&dentry->d_lock);
+	di->flags &= ~CEPH_DENTRY_ASYNC_UNLINK;
+	wake_up_bit(&di->flags, CEPH_DENTRY_ASYNC_UNLINK_BIT);
+	spin_unlock(&dentry->d_lock);
+
+	synchronize_rcu();
+
+	if (result == -EJUKEBOX)
+		goto out;
+
+	/* If op failed, mark everyone involved for errors */
+	if (result) {
+		int pathlen = 0;
+		u64 base = 0;
+		char *path = ceph_mdsc_build_path(dentry, &pathlen,
+						  &base, 0);
+
+		/* mark error on parent + clear complete */
+		mapping_set_error(req->r_parent->i_mapping, result);
+		ceph_dir_clear_complete(req->r_parent);
+
+		/* drop the dentry -- we don't know its status */
+		if (!d_unhashed(dentry))
+			d_drop(dentry);
+
+		/* mark inode itself for an error (since metadata is bogus) */
+		mapping_set_error(req->r_old_inode->i_mapping, result);
+
+		pr_warn("async unlink failure path=(%llx)%s result=%d!\n",
+			base, IS_ERR(path) ? "<<bad>>" : path, result);
+		ceph_mdsc_free_path(path, pathlen);
+	}
+out:
+	iput(req->r_old_inode);
+	ceph_mdsc_release_dir_caps(req);
+}
+
+static int get_caps_for_async_unlink(struct inode *dir, struct dentry *dentry)
+{
+	struct ceph_inode_info *ci = ceph_inode(dir);
+	struct ceph_dentry_info *di;
+	int got = 0, want = CEPH_CAP_FILE_EXCL | CEPH_CAP_DIR_UNLINK;
+
+	spin_lock(&ci->i_ceph_lock);
+	if ((__ceph_caps_issued(ci, NULL) & want) == want) {
+		ceph_take_cap_refs(ci, want, false);
+		got = want;
+	}
+	spin_unlock(&ci->i_ceph_lock);
+
+	/* If we didn't get anything, return 0 */
+	if (!got)
+		return 0;
+
+        spin_lock(&dentry->d_lock);
+        di = ceph_dentry(dentry);
+	/*
+	 * - We are holding Fx, which implies Fs caps.
+	 * - Only support async unlink for primary linkage
+	 */
+	if (atomic_read(&ci->i_shared_gen) != di->lease_shared_gen ||
+	    !(di->flags & CEPH_DENTRY_PRIMARY_LINK))
+		want = 0;
+        spin_unlock(&dentry->d_lock);
+
+	/* Do we still want what we've got? */
+	if (want == got)
+		return got;
+
+	ceph_put_cap_refs(ci, got);
+	return 0;
+}
+
+/*
+ * rmdir and unlink are differ only by the metadata op code
+ */
+static int ceph_unlink(struct inode *dir, struct dentry *dentry)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(dir->i_sb);
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	struct inode *inode = d_inode(dentry);
+	struct ceph_mds_request *req;
+	bool try_async = ceph_test_mount_opt(fsc, ASYNC_DIROPS);
+	int err = -EROFS;
+	int op;
+
+	if (ceph_snap(dir) == CEPH_SNAPDIR) {
+		/* rmdir .snap/foo is RMSNAP */
+		dout("rmsnap dir %p '%pd' dn %p\n", dir, dentry, dentry);
+		op = CEPH_MDS_OP_RMSNAP;
+	} else if (ceph_snap(dir) == CEPH_NOSNAP) {
+		dout("unlink/rmdir dir %p dn %p inode %p\n",
+		     dir, dentry, inode);
+		op = d_is_dir(dentry) ?
+			CEPH_MDS_OP_RMDIR : CEPH_MDS_OP_UNLINK;
+	} else
+		goto out;
+retry:
+	req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
+	if (IS_ERR(req)) {
+		err = PTR_ERR(req);
+		goto out;
+	}
+	req->r_dentry = dget(dentry);
+	req->r_num_caps = 2;
+	req->r_parent = dir;
+	ihold(dir);
+	req->r_dentry_drop = CEPH_CAP_FILE_SHARED;
+	req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
+	req->r_inode_drop = ceph_drop_caps_for_unlink(inode);
+
+	if (try_async && op == CEPH_MDS_OP_UNLINK &&
+	    (req->r_dir_caps = get_caps_for_async_unlink(dir, dentry))) {
+		struct ceph_dentry_info *di = ceph_dentry(dentry);
+
+		dout("async unlink on %llu/%.*s caps=%s", ceph_ino(dir),
+		     dentry->d_name.len, dentry->d_name.name,
+		     ceph_cap_string(req->r_dir_caps));
+		set_bit(CEPH_MDS_R_ASYNC, &req->r_req_flags);
+		req->r_callback = ceph_async_unlink_cb;
+		req->r_old_inode = d_inode(dentry);
+		ihold(req->r_old_inode);
+
+		spin_lock(&dentry->d_lock);
+		di->flags |= CEPH_DENTRY_ASYNC_UNLINK;
+		spin_unlock(&dentry->d_lock);
+
+		spin_lock(&fsc->async_unlink_conflict_lock);
+		hash_add_rcu(fsc->async_unlink_conflict, &di->hnode,
+			     dentry->d_name.hash);
+		spin_unlock(&fsc->async_unlink_conflict_lock);
+
+		err = ceph_mdsc_submit_request(mdsc, dir, req);
+		if (!err) {
+			/*
+			 * We have enough caps, so we assume that the unlink
+			 * will succeed. Fix up the target inode and dcache.
+			 */
+			drop_nlink(inode);
+			d_delete(dentry);
+		} else {
+			spin_lock(&fsc->async_unlink_conflict_lock);
+			hash_del_rcu(&di->hnode);
+			spin_unlock(&fsc->async_unlink_conflict_lock);
+
+			spin_lock(&dentry->d_lock);
+			di->flags &= ~CEPH_DENTRY_ASYNC_UNLINK;
+			spin_unlock(&dentry->d_lock);
+
+			if (err == -EJUKEBOX) {
+				try_async = false;
+				ceph_mdsc_put_request(req);
+				goto retry;
+			}
+		}
+	} else {
+		set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
+		err = ceph_mdsc_do_request(mdsc, dir, req);
+		if (!err && !req->r_reply_info.head->is_dentry)
+			d_delete(dentry);
+	}
+
+	ceph_mdsc_put_request(req);
+out:
+	return err;
+}
+
+static int ceph_rename(struct user_namespace *mnt_userns, struct inode *old_dir,
+		       struct dentry *old_dentry, struct inode *new_dir,
+		       struct dentry *new_dentry, unsigned int flags)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(old_dir->i_sb);
+	struct ceph_mds_request *req;
+	int op = CEPH_MDS_OP_RENAME;
+	int err;
+
+	if (flags)
+		return -EINVAL;
+
+	if (ceph_snap(old_dir) != ceph_snap(new_dir))
+		return -EXDEV;
+	if (ceph_snap(old_dir) != CEPH_NOSNAP) {
+		if (old_dir == new_dir && ceph_snap(old_dir) == CEPH_SNAPDIR)
+			op = CEPH_MDS_OP_RENAMESNAP;
+		else
+			return -EROFS;
+	}
+	/* don't allow cross-quota renames */
+	if ((old_dir != new_dir) &&
+	    (!ceph_quota_is_same_realm(old_dir, new_dir)))
+		return -EXDEV;
+
+	err = ceph_wait_on_conflict_unlink(new_dentry);
+	if (err)
+		return err;
+
+	dout("rename dir %p dentry %p to dir %p dentry %p\n",
+	     old_dir, old_dentry, new_dir, new_dentry);
+	req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
+	if (IS_ERR(req))
+		return PTR_ERR(req);
+	ihold(old_dir);
+	req->r_dentry = dget(new_dentry);
+	req->r_num_caps = 2;
+	req->r_old_dentry = dget(old_dentry);
+	req->r_old_dentry_dir = old_dir;
+	req->r_parent = new_dir;
+	ihold(new_dir);
+	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
+	req->r_old_dentry_drop = CEPH_CAP_FILE_SHARED;
+	req->r_old_dentry_unless = CEPH_CAP_FILE_EXCL;
+	req->r_dentry_drop = CEPH_CAP_FILE_SHARED;
+	req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
+	/* release LINK_RDCACHE on source inode (mds will lock it) */
+	req->r_old_inode_drop = CEPH_CAP_LINK_SHARED | CEPH_CAP_LINK_EXCL;
+	if (d_really_is_positive(new_dentry)) {
+		req->r_inode_drop =
+			ceph_drop_caps_for_unlink(d_inode(new_dentry));
+	}
+	err = ceph_mdsc_do_request(mdsc, old_dir, req);
+	if (!err && !req->r_reply_info.head->is_dentry) {
+		/*
+		 * Normally d_move() is done by fill_trace (called by
+		 * do_request, above).  If there is no trace, we need
+		 * to do it here.
+		 */
+		d_move(old_dentry, new_dentry);
+	}
+	ceph_mdsc_put_request(req);
+	return err;
+}
+
+/*
+ * Move dentry to tail of mdsc->dentry_leases list when lease is updated.
+ * Leases at front of the list will expire first. (Assume all leases have
+ * similar duration)
+ *
+ * Called under dentry->d_lock.
+ */
+void __ceph_dentry_lease_touch(struct ceph_dentry_info *di)
+{
+	struct dentry *dn = di->dentry;
+	struct ceph_mds_client *mdsc;
+
+	dout("dentry_lease_touch %p %p '%pd'\n", di, dn, dn);
+
+	di->flags |= CEPH_DENTRY_LEASE_LIST;
+	if (di->flags & CEPH_DENTRY_SHRINK_LIST) {
+		di->flags |= CEPH_DENTRY_REFERENCED;
+		return;
+	}
+
+	mdsc = ceph_sb_to_client(dn->d_sb)->mdsc;
+	spin_lock(&mdsc->dentry_list_lock);
+	list_move_tail(&di->lease_list, &mdsc->dentry_leases);
+	spin_unlock(&mdsc->dentry_list_lock);
+}
+
+static void __dentry_dir_lease_touch(struct ceph_mds_client* mdsc,
+				     struct ceph_dentry_info *di)
+{
+	di->flags &= ~(CEPH_DENTRY_LEASE_LIST | CEPH_DENTRY_REFERENCED);
+	di->lease_gen = 0;
+	di->time = jiffies;
+	list_move_tail(&di->lease_list, &mdsc->dentry_dir_leases);
+}
+
+/*
+ * When dir lease is used, add dentry to tail of mdsc->dentry_dir_leases
+ * list if it's not in the list, otherwise set 'referenced' flag.
+ *
+ * Called under dentry->d_lock.
+ */
+void __ceph_dentry_dir_lease_touch(struct ceph_dentry_info *di)
+{
+	struct dentry *dn = di->dentry;
+	struct ceph_mds_client *mdsc;
+
+	dout("dentry_dir_lease_touch %p %p '%pd' (offset 0x%llx)\n",
+	     di, dn, dn, di->offset);
+
+	if (!list_empty(&di->lease_list)) {
+		if (di->flags & CEPH_DENTRY_LEASE_LIST) {
+			/* don't remove dentry from dentry lease list
+			 * if its lease is valid */
+			if (__dentry_lease_is_valid(di))
+				return;
+		} else {
+			di->flags |= CEPH_DENTRY_REFERENCED;
+			return;
+		}
+	}
+
+	if (di->flags & CEPH_DENTRY_SHRINK_LIST) {
+		di->flags |= CEPH_DENTRY_REFERENCED;
+		di->flags &= ~CEPH_DENTRY_LEASE_LIST;
+		return;
+	}
+
+	mdsc = ceph_sb_to_client(dn->d_sb)->mdsc;
+	spin_lock(&mdsc->dentry_list_lock);
+	__dentry_dir_lease_touch(mdsc, di),
+	spin_unlock(&mdsc->dentry_list_lock);
+}
+
+static void __dentry_lease_unlist(struct ceph_dentry_info *di)
+{
+	struct ceph_mds_client *mdsc;
+	if (di->flags & CEPH_DENTRY_SHRINK_LIST)
+		return;
+	if (list_empty(&di->lease_list))
+		return;
+
+	mdsc = ceph_sb_to_client(di->dentry->d_sb)->mdsc;
+	spin_lock(&mdsc->dentry_list_lock);
+	list_del_init(&di->lease_list);
+	spin_unlock(&mdsc->dentry_list_lock);
+}
+
+enum {
+	KEEP	= 0,
+	DELETE	= 1,
+	TOUCH	= 2,
+	STOP	= 4,
+};
+
+struct ceph_lease_walk_control {
+	bool dir_lease;
+	bool expire_dir_lease;
+	unsigned long nr_to_scan;
+	unsigned long dir_lease_ttl;
+};
+
+static unsigned long
+__dentry_leases_walk(struct ceph_mds_client *mdsc,
+		     struct ceph_lease_walk_control *lwc,
+		     int (*check)(struct dentry*, void*))
+{
+	struct ceph_dentry_info *di, *tmp;
+	struct dentry *dentry, *last = NULL;
+	struct list_head* list;
+        LIST_HEAD(dispose);
+	unsigned long freed = 0;
+	int ret = 0;
+
+	list = lwc->dir_lease ? &mdsc->dentry_dir_leases : &mdsc->dentry_leases;
+	spin_lock(&mdsc->dentry_list_lock);
+	list_for_each_entry_safe(di, tmp, list, lease_list) {
+		if (!lwc->nr_to_scan)
+			break;
+		--lwc->nr_to_scan;
+
+		dentry = di->dentry;
+		if (last == dentry)
+			break;
+
+		if (!spin_trylock(&dentry->d_lock))
+			continue;
+
+		if (__lockref_is_dead(&dentry->d_lockref)) {
+			list_del_init(&di->lease_list);
+			goto next;
+		}
+
+		ret = check(dentry, lwc);
+		if (ret & TOUCH) {
+			/* move it into tail of dir lease list */
+			__dentry_dir_lease_touch(mdsc, di);
+			if (!last)
+				last = dentry;
+		}
+		if (ret & DELETE) {
+			/* stale lease */
+			di->flags &= ~CEPH_DENTRY_REFERENCED;
+			if (dentry->d_lockref.count > 0) {
+				/* update_dentry_lease() will re-add
+				 * it to lease list, or
+				 * ceph_d_delete() will return 1 when
+				 * last reference is dropped */
+				list_del_init(&di->lease_list);
+			} else {
+				di->flags |= CEPH_DENTRY_SHRINK_LIST;
+				list_move_tail(&di->lease_list, &dispose);
+				dget_dlock(dentry);
+			}
+		}
+next:
+		spin_unlock(&dentry->d_lock);
+		if (ret & STOP)
+			break;
+	}
+	spin_unlock(&mdsc->dentry_list_lock);
+
+	while (!list_empty(&dispose)) {
+		di = list_first_entry(&dispose, struct ceph_dentry_info,
+				      lease_list);
+		dentry = di->dentry;
+		spin_lock(&dentry->d_lock);
+
+		list_del_init(&di->lease_list);
+		di->flags &= ~CEPH_DENTRY_SHRINK_LIST;
+		if (di->flags & CEPH_DENTRY_REFERENCED) {
+			spin_lock(&mdsc->dentry_list_lock);
+			if (di->flags & CEPH_DENTRY_LEASE_LIST) {
+				list_add_tail(&di->lease_list,
+					      &mdsc->dentry_leases);
+			} else {
+				__dentry_dir_lease_touch(mdsc, di);
+			}
+			spin_unlock(&mdsc->dentry_list_lock);
+		} else {
+			freed++;
+		}
+
+		spin_unlock(&dentry->d_lock);
+		/* ceph_d_delete() does the trick */
+		dput(dentry);
+	}
+	return freed;
+}
+
+static int __dentry_lease_check(struct dentry *dentry, void *arg)
+{
+	struct ceph_dentry_info *di = ceph_dentry(dentry);
+	int ret;
+
+	if (__dentry_lease_is_valid(di))
+		return STOP;
+	ret = __dir_lease_try_check(dentry);
+	if (ret == -EBUSY)
+		return KEEP;
+	if (ret > 0)
+		return TOUCH;
+	return DELETE;
+}
+
+static int __dir_lease_check(struct dentry *dentry, void *arg)
+{
+	struct ceph_lease_walk_control *lwc = arg;
+	struct ceph_dentry_info *di = ceph_dentry(dentry);
+
+	int ret = __dir_lease_try_check(dentry);
+	if (ret == -EBUSY)
+		return KEEP;
+	if (ret > 0) {
+		if (time_before(jiffies, di->time + lwc->dir_lease_ttl))
+			return STOP;
+		/* Move dentry to tail of dir lease list if we don't want
+		 * to delete it. So dentries in the list are checked in a
+		 * round robin manner */
+		if (!lwc->expire_dir_lease)
+			return TOUCH;
+		if (dentry->d_lockref.count > 0 ||
+		    (di->flags & CEPH_DENTRY_REFERENCED))
+			return TOUCH;
+		/* invalidate dir lease */
+		di->lease_shared_gen = 0;
+	}
+	return DELETE;
+}
+
+int ceph_trim_dentries(struct ceph_mds_client *mdsc)
+{
+	struct ceph_lease_walk_control lwc;
+	unsigned long count;
+	unsigned long freed;
+
+	spin_lock(&mdsc->caps_list_lock);
+        if (mdsc->caps_use_max > 0 &&
+            mdsc->caps_use_count > mdsc->caps_use_max)
+		count = mdsc->caps_use_count - mdsc->caps_use_max;
+	else
+		count = 0;
+        spin_unlock(&mdsc->caps_list_lock);
+
+	lwc.dir_lease = false;
+	lwc.nr_to_scan  = CEPH_CAPS_PER_RELEASE * 2;
+	freed = __dentry_leases_walk(mdsc, &lwc, __dentry_lease_check);
+	if (!lwc.nr_to_scan) /* more invalid leases */
+		return -EAGAIN;
+
+	if (lwc.nr_to_scan < CEPH_CAPS_PER_RELEASE)
+		lwc.nr_to_scan = CEPH_CAPS_PER_RELEASE;
+
+	lwc.dir_lease = true;
+	lwc.expire_dir_lease = freed < count;
+	lwc.dir_lease_ttl = mdsc->fsc->mount_options->caps_wanted_delay_max * HZ;
+	freed +=__dentry_leases_walk(mdsc, &lwc, __dir_lease_check);
+	if (!lwc.nr_to_scan) /* more to check */
+		return -EAGAIN;
+
+	return freed > 0 ? 1 : 0;
+}
+
+/*
+ * Ensure a dentry lease will no longer revalidate.
+ */
+void ceph_invalidate_dentry_lease(struct dentry *dentry)
+{
+	struct ceph_dentry_info *di = ceph_dentry(dentry);
+	spin_lock(&dentry->d_lock);
+	di->time = jiffies;
+	di->lease_shared_gen = 0;
+	di->flags &= ~CEPH_DENTRY_PRIMARY_LINK;
+	__dentry_lease_unlist(di);
+	spin_unlock(&dentry->d_lock);
+}
+
+/*
+ * Check if dentry lease is valid.  If not, delete the lease.  Try to
+ * renew if the least is more than half up.
+ */
+static bool __dentry_lease_is_valid(struct ceph_dentry_info *di)
+{
+	struct ceph_mds_session *session;
+
+	if (!di->lease_gen)
+		return false;
+
+	session = di->lease_session;
+	if (session) {
+		u32 gen;
+		unsigned long ttl;
+
+		gen = atomic_read(&session->s_cap_gen);
+		ttl = session->s_cap_ttl;
+
+		if (di->lease_gen == gen &&
+		    time_before(jiffies, ttl) &&
+		    time_before(jiffies, di->time))
+			return true;
+	}
+	di->lease_gen = 0;
+	return false;
+}
+
+static int dentry_lease_is_valid(struct dentry *dentry, unsigned int flags)
+{
+	struct ceph_dentry_info *di;
+	struct ceph_mds_session *session = NULL;
+	u32 seq = 0;
+	int valid = 0;
+
+	spin_lock(&dentry->d_lock);
+	di = ceph_dentry(dentry);
+	if (di && __dentry_lease_is_valid(di)) {
+		valid = 1;
+
+		if (di->lease_renew_after &&
+		    time_after(jiffies, di->lease_renew_after)) {
+			/*
+			 * We should renew. If we're in RCU walk mode
+			 * though, we can't do that so just return
+			 * -ECHILD.
+			 */
+			if (flags & LOOKUP_RCU) {
+				valid = -ECHILD;
+			} else {
+				session = ceph_get_mds_session(di->lease_session);
+				seq = di->lease_seq;
+				di->lease_renew_after = 0;
+				di->lease_renew_from = jiffies;
+			}
+		}
+	}
+	spin_unlock(&dentry->d_lock);
+
+	if (session) {
+		ceph_mdsc_lease_send_msg(session, dentry,
+					 CEPH_MDS_LEASE_RENEW, seq);
+		ceph_put_mds_session(session);
+	}
+	dout("dentry_lease_is_valid - dentry %p = %d\n", dentry, valid);
+	return valid;
+}
+
+/*
+ * Called under dentry->d_lock.
+ */
+static int __dir_lease_try_check(const struct dentry *dentry)
+{
+	struct ceph_dentry_info *di = ceph_dentry(dentry);
+	struct inode *dir;
+	struct ceph_inode_info *ci;
+	int valid = 0;
+
+	if (!di->lease_shared_gen)
+		return 0;
+	if (IS_ROOT(dentry))
+		return 0;
+
+	dir = d_inode(dentry->d_parent);
+	ci = ceph_inode(dir);
+
+	if (spin_trylock(&ci->i_ceph_lock)) {
+		if (atomic_read(&ci->i_shared_gen) == di->lease_shared_gen &&
+		    __ceph_caps_issued_mask(ci, CEPH_CAP_FILE_SHARED, 0))
+			valid = 1;
+		spin_unlock(&ci->i_ceph_lock);
+	} else {
+		valid = -EBUSY;
+	}
+
+	if (!valid)
+		di->lease_shared_gen = 0;
+	return valid;
+}
+
+/*
+ * Check if directory-wide content lease/cap is valid.
+ */
+static int dir_lease_is_valid(struct inode *dir, struct dentry *dentry,
+			      struct ceph_mds_client *mdsc)
+{
+	struct ceph_inode_info *ci = ceph_inode(dir);
+	int valid;
+	int shared_gen;
+
+	spin_lock(&ci->i_ceph_lock);
+	valid = __ceph_caps_issued_mask(ci, CEPH_CAP_FILE_SHARED, 1);
+	if (valid) {
+		__ceph_touch_fmode(ci, mdsc, CEPH_FILE_MODE_RD);
+		shared_gen = atomic_read(&ci->i_shared_gen);
+	}
+	spin_unlock(&ci->i_ceph_lock);
+	if (valid) {
+		struct ceph_dentry_info *di;
+		spin_lock(&dentry->d_lock);
+		di = ceph_dentry(dentry);
+		if (dir == d_inode(dentry->d_parent) &&
+		    di && di->lease_shared_gen == shared_gen)
+			__ceph_dentry_dir_lease_touch(di);
+		else
+			valid = 0;
+		spin_unlock(&dentry->d_lock);
+	}
+	dout("dir_lease_is_valid dir %p v%u dentry %p = %d\n",
+	     dir, (unsigned)atomic_read(&ci->i_shared_gen), dentry, valid);
+	return valid;
+}
+
+/*
+ * Check if cached dentry can be trusted.
+ */
+static int ceph_d_revalidate(struct dentry *dentry, unsigned int flags)
+{
+	int valid = 0;
+	struct dentry *parent;
+	struct inode *dir, *inode;
+	struct ceph_mds_client *mdsc;
+
+	if (flags & LOOKUP_RCU) {
+		parent = READ_ONCE(dentry->d_parent);
+		dir = d_inode_rcu(parent);
+		if (!dir)
+			return -ECHILD;
+		inode = d_inode_rcu(dentry);
+	} else {
+		parent = dget_parent(dentry);
+		dir = d_inode(parent);
+		inode = d_inode(dentry);
+	}
+
+	dout("d_revalidate %p '%pd' inode %p offset 0x%llx\n", dentry,
+	     dentry, inode, ceph_dentry(dentry)->offset);
+
+	mdsc = ceph_sb_to_client(dir->i_sb)->mdsc;
+
+	/* always trust cached snapped dentries, snapdir dentry */
+	if (ceph_snap(dir) != CEPH_NOSNAP) {
+		dout("d_revalidate %p '%pd' inode %p is SNAPPED\n", dentry,
+		     dentry, inode);
+		valid = 1;
+	} else if (inode && ceph_snap(inode) == CEPH_SNAPDIR) {
+		valid = 1;
+	} else {
+		valid = dentry_lease_is_valid(dentry, flags);
+		if (valid == -ECHILD)
+			return valid;
+		if (valid || dir_lease_is_valid(dir, dentry, mdsc)) {
+			if (inode)
+				valid = ceph_is_any_caps(inode);
+			else
+				valid = 1;
+		}
+	}
+
+	if (!valid) {
+		struct ceph_mds_request *req;
+		int op, err;
+		u32 mask;
+
+		if (flags & LOOKUP_RCU)
+			return -ECHILD;
+
+		percpu_counter_inc(&mdsc->metric.d_lease_mis);
+
+		op = ceph_snap(dir) == CEPH_SNAPDIR ?
+			CEPH_MDS_OP_LOOKUPSNAP : CEPH_MDS_OP_LOOKUP;
+		req = ceph_mdsc_create_request(mdsc, op, USE_ANY_MDS);
+		if (!IS_ERR(req)) {
+			req->r_dentry = dget(dentry);
+			req->r_num_caps = 2;
+			req->r_parent = dir;
+			ihold(dir);
+
+			mask = CEPH_STAT_CAP_INODE | CEPH_CAP_AUTH_SHARED;
+			if (ceph_security_xattr_wanted(dir))
+				mask |= CEPH_CAP_XATTR_SHARED;
+			req->r_args.getattr.mask = cpu_to_le32(mask);
+
+			err = ceph_mdsc_do_request(mdsc, NULL, req);
+			switch (err) {
+			case 0:
+				if (d_really_is_positive(dentry) &&
+				    d_inode(dentry) == req->r_target_inode)
+					valid = 1;
+				break;
+			case -ENOENT:
+				if (d_really_is_negative(dentry))
+					valid = 1;
+				fallthrough;
+			default:
+				break;
+			}
+			ceph_mdsc_put_request(req);
+			dout("d_revalidate %p lookup result=%d\n",
+			     dentry, err);
+		}
+	} else {
+		percpu_counter_inc(&mdsc->metric.d_lease_hit);
+	}
+
+	dout("d_revalidate %p %s\n", dentry, valid ? "valid" : "invalid");
+	if (!valid)
+		ceph_dir_clear_complete(dir);
+
+	if (!(flags & LOOKUP_RCU))
+		dput(parent);
+	return valid;
+}
+
+/*
+ * Delete unused dentry that doesn't have valid lease
+ *
+ * Called under dentry->d_lock.
+ */
+static int ceph_d_delete(const struct dentry *dentry)
+{
+	struct ceph_dentry_info *di;
+
+	/* won't release caps */
+	if (d_really_is_negative(dentry))
+		return 0;
+	if (ceph_snap(d_inode(dentry)) != CEPH_NOSNAP)
+		return 0;
+	/* vaild lease? */
+	di = ceph_dentry(dentry);
+	if (di) {
+		if (__dentry_lease_is_valid(di))
+			return 0;
+		if (__dir_lease_try_check(dentry))
+			return 0;
+	}
+	return 1;
+}
+
+/*
+ * Release our ceph_dentry_info.
+ */
+static void ceph_d_release(struct dentry *dentry)
+{
+	struct ceph_dentry_info *di = ceph_dentry(dentry);
+	struct ceph_fs_client *fsc = ceph_sb_to_client(dentry->d_sb);
+
+	dout("d_release %p\n", dentry);
+
+	atomic64_dec(&fsc->mdsc->metric.total_dentries);
+
+	spin_lock(&dentry->d_lock);
+	__dentry_lease_unlist(di);
+	dentry->d_fsdata = NULL;
+	spin_unlock(&dentry->d_lock);
+
+	ceph_put_mds_session(di->lease_session);
+	kmem_cache_free(ceph_dentry_cachep, di);
+}
+
+/*
+ * When the VFS prunes a dentry from the cache, we need to clear the
+ * complete flag on the parent directory.
+ *
+ * Called under dentry->d_lock.
+ */
+static void ceph_d_prune(struct dentry *dentry)
+{
+	struct ceph_inode_info *dir_ci;
+	struct ceph_dentry_info *di;
+
+	dout("ceph_d_prune %pd %p\n", dentry, dentry);
+
+	/* do we have a valid parent? */
+	if (IS_ROOT(dentry))
+		return;
+
+	/* we hold d_lock, so d_parent is stable */
+	dir_ci = ceph_inode(d_inode(dentry->d_parent));
+	if (dir_ci->i_vino.snap == CEPH_SNAPDIR)
+		return;
+
+	/* who calls d_delete() should also disable dcache readdir */
+	if (d_really_is_negative(dentry))
+		return;
+
+	/* d_fsdata does not get cleared until d_release */
+	if (!d_unhashed(dentry)) {
+		__ceph_dir_clear_complete(dir_ci);
+		return;
+	}
+
+	/* Disable dcache readdir just in case that someone called d_drop()
+	 * or d_invalidate(), but MDS didn't revoke CEPH_CAP_FILE_SHARED
+	 * properly (dcache readdir is still enabled) */
+	di = ceph_dentry(dentry);
+	if (di->offset > 0 &&
+	    di->lease_shared_gen == atomic_read(&dir_ci->i_shared_gen))
+		__ceph_dir_clear_ordered(dir_ci);
+}
+
+/*
+ * read() on a dir.  This weird interface hack only works if mounted
+ * with '-o dirstat'.
+ */
+static ssize_t ceph_read_dir(struct file *file, char __user *buf, size_t size,
+			     loff_t *ppos)
+{
+	struct ceph_dir_file_info *dfi = file->private_data;
+	struct inode *inode = file_inode(file);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	int left;
+	const int bufsize = 1024;
+
+	if (!ceph_test_mount_opt(ceph_sb_to_client(inode->i_sb), DIRSTAT))
+		return -EISDIR;
+
+	if (!dfi->dir_info) {
+		dfi->dir_info = kmalloc(bufsize, GFP_KERNEL);
+		if (!dfi->dir_info)
+			return -ENOMEM;
+		dfi->dir_info_len =
+			snprintf(dfi->dir_info, bufsize,
+				"entries:   %20lld\n"
+				" files:    %20lld\n"
+				" subdirs:  %20lld\n"
+				"rentries:  %20lld\n"
+				" rfiles:   %20lld\n"
+				" rsubdirs: %20lld\n"
+				"rbytes:    %20lld\n"
+				"rctime:    %10lld.%09ld\n",
+				ci->i_files + ci->i_subdirs,
+				ci->i_files,
+				ci->i_subdirs,
+				ci->i_rfiles + ci->i_rsubdirs,
+				ci->i_rfiles,
+				ci->i_rsubdirs,
+				ci->i_rbytes,
+				ci->i_rctime.tv_sec,
+				ci->i_rctime.tv_nsec);
+	}
+
+	if (*ppos >= dfi->dir_info_len)
+		return 0;
+	size = min_t(unsigned, size, dfi->dir_info_len-*ppos);
+	left = copy_to_user(buf, dfi->dir_info + *ppos, size);
+	if (left == size)
+		return -EFAULT;
+	*ppos += (size - left);
+	return size - left;
+}
+
+
+
+/*
+ * Return name hash for a given dentry.  This is dependent on
+ * the parent directory's hash function.
+ */
+unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn)
+{
+	struct ceph_inode_info *dci = ceph_inode(dir);
+	unsigned hash;
+
+	switch (dci->i_dir_layout.dl_dir_hash) {
+	case 0:	/* for backward compat */
+	case CEPH_STR_HASH_LINUX:
+		return dn->d_name.hash;
+
+	default:
+		spin_lock(&dn->d_lock);
+		hash = ceph_str_hash(dci->i_dir_layout.dl_dir_hash,
+				     dn->d_name.name, dn->d_name.len);
+		spin_unlock(&dn->d_lock);
+		return hash;
+	}
+}
+
+const struct file_operations ceph_dir_fops = {
+	.read = ceph_read_dir,
+	.iterate = ceph_readdir,
+	.llseek = ceph_dir_llseek,
+	.open = ceph_open,
+	.release = ceph_release,
+	.unlocked_ioctl = ceph_ioctl,
+	.compat_ioctl = compat_ptr_ioctl,
+	.fsync = ceph_fsync,
+	.lock = ceph_lock,
+	.flock = ceph_flock,
+};
+
+const struct file_operations ceph_snapdir_fops = {
+	.iterate = ceph_readdir,
+	.llseek = ceph_dir_llseek,
+	.open = ceph_open,
+	.release = ceph_release,
+};
+
+const struct inode_operations ceph_dir_iops = {
+	.lookup = ceph_lookup,
+	.permission = ceph_permission,
+	.getattr = ceph_getattr,
+	.setattr = ceph_setattr,
+	.listxattr = ceph_listxattr,
+	.get_acl = ceph_get_acl,
+	.set_acl = ceph_set_acl,
+	.mknod = ceph_mknod,
+	.symlink = ceph_symlink,
+	.mkdir = ceph_mkdir,
+	.link = ceph_link,
+	.unlink = ceph_unlink,
+	.rmdir = ceph_unlink,
+	.rename = ceph_rename,
+	.create = ceph_create,
+	.atomic_open = ceph_atomic_open,
+};
+
+const struct inode_operations ceph_snapdir_iops = {
+	.lookup = ceph_lookup,
+	.permission = ceph_permission,
+	.getattr = ceph_getattr,
+	.mkdir = ceph_mkdir,
+	.rmdir = ceph_unlink,
+	.rename = ceph_rename,
+};
+
+const struct dentry_operations ceph_dentry_ops = {
+	.d_revalidate = ceph_d_revalidate,
+	.d_delete = ceph_d_delete,
+	.d_release = ceph_d_release,
+	.d_prune = ceph_d_prune,
+	.d_init = ceph_d_init,
+};
diff --git a/fs/ceph/export.c b/fs/ceph/export.c
new file mode 100644
index 000000000..f780e4e0d
--- /dev/null
+++ b/fs/ceph/export.c
@@ -0,0 +1,584 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/ceph/ceph_debug.h>
+
+#include <linux/exportfs.h>
+#include <linux/slab.h>
+#include <asm/unaligned.h>
+
+#include "super.h"
+#include "mds_client.h"
+
+/*
+ * Basic fh
+ */
+struct ceph_nfs_fh {
+	u64 ino;
+} __attribute__ ((packed));
+
+/*
+ * Larger fh that includes parent ino.
+ */
+struct ceph_nfs_confh {
+	u64 ino, parent_ino;
+} __attribute__ ((packed));
+
+/*
+ * fh for snapped inode
+ */
+struct ceph_nfs_snapfh {
+	u64 ino;
+	u64 snapid;
+	u64 parent_ino;
+	u32 hash;
+} __attribute__ ((packed));
+
+static int ceph_encode_snapfh(struct inode *inode, u32 *rawfh, int *max_len,
+			      struct inode *parent_inode)
+{
+	static const int snap_handle_length =
+		sizeof(struct ceph_nfs_snapfh) >> 2;
+	struct ceph_nfs_snapfh *sfh = (void *)rawfh;
+	u64 snapid = ceph_snap(inode);
+	int ret;
+	bool no_parent = true;
+
+	if (*max_len < snap_handle_length) {
+		*max_len = snap_handle_length;
+		ret = FILEID_INVALID;
+		goto out;
+	}
+
+	ret =  -EINVAL;
+	if (snapid != CEPH_SNAPDIR) {
+		struct inode *dir;
+		struct dentry *dentry = d_find_alias(inode);
+		if (!dentry)
+			goto out;
+
+		rcu_read_lock();
+		dir = d_inode_rcu(dentry->d_parent);
+		if (ceph_snap(dir) != CEPH_SNAPDIR) {
+			sfh->parent_ino = ceph_ino(dir);
+			sfh->hash = ceph_dentry_hash(dir, dentry);
+			no_parent = false;
+		}
+		rcu_read_unlock();
+		dput(dentry);
+	}
+
+	if (no_parent) {
+		if (!S_ISDIR(inode->i_mode))
+			goto out;
+		sfh->parent_ino = sfh->ino;
+		sfh->hash = 0;
+	}
+	sfh->ino = ceph_ino(inode);
+	sfh->snapid = snapid;
+
+	*max_len = snap_handle_length;
+	ret = FILEID_BTRFS_WITH_PARENT;
+out:
+	dout("encode_snapfh %llx.%llx ret=%d\n", ceph_vinop(inode), ret);
+	return ret;
+}
+
+static int ceph_encode_fh(struct inode *inode, u32 *rawfh, int *max_len,
+			  struct inode *parent_inode)
+{
+	static const int handle_length =
+		sizeof(struct ceph_nfs_fh) >> 2;
+	static const int connected_handle_length =
+		sizeof(struct ceph_nfs_confh) >> 2;
+	int type;
+
+	if (ceph_snap(inode) != CEPH_NOSNAP)
+		return ceph_encode_snapfh(inode, rawfh, max_len, parent_inode);
+
+	if (parent_inode && (*max_len < connected_handle_length)) {
+		*max_len = connected_handle_length;
+		return FILEID_INVALID;
+	} else if (*max_len < handle_length) {
+		*max_len = handle_length;
+		return FILEID_INVALID;
+	}
+
+	if (parent_inode) {
+		struct ceph_nfs_confh *cfh = (void *)rawfh;
+		dout("encode_fh %llx with parent %llx\n",
+		     ceph_ino(inode), ceph_ino(parent_inode));
+		cfh->ino = ceph_ino(inode);
+		cfh->parent_ino = ceph_ino(parent_inode);
+		*max_len = connected_handle_length;
+		type = FILEID_INO32_GEN_PARENT;
+	} else {
+		struct ceph_nfs_fh *fh = (void *)rawfh;
+		dout("encode_fh %llx\n", ceph_ino(inode));
+		fh->ino = ceph_ino(inode);
+		*max_len = handle_length;
+		type = FILEID_INO32_GEN;
+	}
+	return type;
+}
+
+static struct inode *__lookup_inode(struct super_block *sb, u64 ino)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_client(sb)->mdsc;
+	struct inode *inode;
+	struct ceph_vino vino;
+	int err;
+
+	vino.ino = ino;
+	vino.snap = CEPH_NOSNAP;
+
+	if (ceph_vino_is_reserved(vino))
+		return ERR_PTR(-ESTALE);
+
+	inode = ceph_find_inode(sb, vino);
+	if (!inode) {
+		struct ceph_mds_request *req;
+		int mask;
+
+		req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_LOOKUPINO,
+					       USE_ANY_MDS);
+		if (IS_ERR(req))
+			return ERR_CAST(req);
+
+		mask = CEPH_STAT_CAP_INODE;
+		if (ceph_security_xattr_wanted(d_inode(sb->s_root)))
+			mask |= CEPH_CAP_XATTR_SHARED;
+		req->r_args.lookupino.mask = cpu_to_le32(mask);
+
+		req->r_ino1 = vino;
+		req->r_num_caps = 1;
+		err = ceph_mdsc_do_request(mdsc, NULL, req);
+		inode = req->r_target_inode;
+		if (inode)
+			ihold(inode);
+		ceph_mdsc_put_request(req);
+		if (!inode)
+			return err < 0 ? ERR_PTR(err) : ERR_PTR(-ESTALE);
+	} else {
+		if (ceph_inode_is_shutdown(inode)) {
+			iput(inode);
+			return ERR_PTR(-ESTALE);
+		}
+	}
+	return inode;
+}
+
+struct inode *ceph_lookup_inode(struct super_block *sb, u64 ino)
+{
+	struct inode *inode = __lookup_inode(sb, ino);
+	if (IS_ERR(inode))
+		return inode;
+	if (inode->i_nlink == 0) {
+		iput(inode);
+		return ERR_PTR(-ESTALE);
+	}
+	return inode;
+}
+
+static struct dentry *__fh_to_dentry(struct super_block *sb, u64 ino)
+{
+	struct inode *inode = __lookup_inode(sb, ino);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	int err;
+
+	if (IS_ERR(inode))
+		return ERR_CAST(inode);
+	/* We need LINK caps to reliably check i_nlink */
+	err = ceph_do_getattr(inode, CEPH_CAP_LINK_SHARED, false);
+	if (err) {
+		iput(inode);
+		return ERR_PTR(err);
+	}
+	/* -ESTALE if inode as been unlinked and no file is open */
+	if ((inode->i_nlink == 0) && !__ceph_is_file_opened(ci)) {
+		iput(inode);
+		return ERR_PTR(-ESTALE);
+	}
+	return d_obtain_alias(inode);
+}
+
+static struct dentry *__snapfh_to_dentry(struct super_block *sb,
+					  struct ceph_nfs_snapfh *sfh,
+					  bool want_parent)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_client(sb)->mdsc;
+	struct ceph_mds_request *req;
+	struct inode *inode;
+	struct ceph_vino vino;
+	int mask;
+	int err;
+	bool unlinked = false;
+
+	if (want_parent) {
+		vino.ino = sfh->parent_ino;
+		if (sfh->snapid == CEPH_SNAPDIR)
+			vino.snap = CEPH_NOSNAP;
+		else if (sfh->ino == sfh->parent_ino)
+			vino.snap = CEPH_SNAPDIR;
+		else
+			vino.snap = sfh->snapid;
+	} else {
+		vino.ino = sfh->ino;
+		vino.snap = sfh->snapid;
+	}
+
+	if (ceph_vino_is_reserved(vino))
+		return ERR_PTR(-ESTALE);
+
+	inode = ceph_find_inode(sb, vino);
+	if (inode) {
+		if (ceph_inode_is_shutdown(inode)) {
+			iput(inode);
+			return ERR_PTR(-ESTALE);
+		}
+		return d_obtain_alias(inode);
+	}
+
+	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_LOOKUPINO,
+				       USE_ANY_MDS);
+	if (IS_ERR(req))
+		return ERR_CAST(req);
+
+	mask = CEPH_STAT_CAP_INODE;
+	if (ceph_security_xattr_wanted(d_inode(sb->s_root)))
+		mask |= CEPH_CAP_XATTR_SHARED;
+	req->r_args.lookupino.mask = cpu_to_le32(mask);
+	if (vino.snap < CEPH_NOSNAP) {
+		req->r_args.lookupino.snapid = cpu_to_le64(vino.snap);
+		if (!want_parent && sfh->ino != sfh->parent_ino) {
+			req->r_args.lookupino.parent =
+					cpu_to_le64(sfh->parent_ino);
+			req->r_args.lookupino.hash =
+					cpu_to_le32(sfh->hash);
+		}
+	}
+
+	req->r_ino1 = vino;
+	req->r_num_caps = 1;
+	err = ceph_mdsc_do_request(mdsc, NULL, req);
+	inode = req->r_target_inode;
+	if (inode) {
+		if (vino.snap == CEPH_SNAPDIR) {
+			if (inode->i_nlink == 0)
+				unlinked = true;
+			inode = ceph_get_snapdir(inode);
+		} else if (ceph_snap(inode) == vino.snap) {
+			ihold(inode);
+		} else {
+			/* mds does not support lookup snapped inode */
+			inode = ERR_PTR(-EOPNOTSUPP);
+		}
+	} else {
+		inode = ERR_PTR(-ESTALE);
+	}
+	ceph_mdsc_put_request(req);
+
+	if (want_parent) {
+		dout("snapfh_to_parent %llx.%llx\n err=%d\n",
+		     vino.ino, vino.snap, err);
+	} else {
+		dout("snapfh_to_dentry %llx.%llx parent %llx hash %x err=%d",
+		      vino.ino, vino.snap, sfh->parent_ino, sfh->hash, err);
+	}
+	if (IS_ERR(inode))
+		return ERR_CAST(inode);
+	/* see comments in ceph_get_parent() */
+	return unlinked ? d_obtain_root(inode) : d_obtain_alias(inode);
+}
+
+/*
+ * convert regular fh to dentry
+ */
+static struct dentry *ceph_fh_to_dentry(struct super_block *sb,
+					struct fid *fid,
+					int fh_len, int fh_type)
+{
+	struct ceph_nfs_fh *fh = (void *)fid->raw;
+
+	if (fh_type == FILEID_BTRFS_WITH_PARENT) {
+		struct ceph_nfs_snapfh *sfh = (void *)fid->raw;
+		return __snapfh_to_dentry(sb, sfh, false);
+	}
+
+	if (fh_type != FILEID_INO32_GEN  &&
+	    fh_type != FILEID_INO32_GEN_PARENT)
+		return NULL;
+	if (fh_len < sizeof(*fh) / 4)
+		return NULL;
+
+	dout("fh_to_dentry %llx\n", fh->ino);
+	return __fh_to_dentry(sb, fh->ino);
+}
+
+static struct dentry *__get_parent(struct super_block *sb,
+				   struct dentry *child, u64 ino)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_client(sb)->mdsc;
+	struct ceph_mds_request *req;
+	struct inode *inode;
+	int mask;
+	int err;
+
+	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_LOOKUPPARENT,
+				       USE_ANY_MDS);
+	if (IS_ERR(req))
+		return ERR_CAST(req);
+
+	if (child) {
+		req->r_inode = d_inode(child);
+		ihold(d_inode(child));
+	} else {
+		req->r_ino1 = (struct ceph_vino) {
+			.ino = ino,
+			.snap = CEPH_NOSNAP,
+		};
+	}
+
+	mask = CEPH_STAT_CAP_INODE;
+	if (ceph_security_xattr_wanted(d_inode(sb->s_root)))
+		mask |= CEPH_CAP_XATTR_SHARED;
+	req->r_args.getattr.mask = cpu_to_le32(mask);
+
+	req->r_num_caps = 1;
+	err = ceph_mdsc_do_request(mdsc, NULL, req);
+	if (err) {
+		ceph_mdsc_put_request(req);
+		return ERR_PTR(err);
+	}
+
+	inode = req->r_target_inode;
+	if (inode)
+		ihold(inode);
+	ceph_mdsc_put_request(req);
+	if (!inode)
+		return ERR_PTR(-ENOENT);
+
+	return d_obtain_alias(inode);
+}
+
+static struct dentry *ceph_get_parent(struct dentry *child)
+{
+	struct inode *inode = d_inode(child);
+	struct dentry *dn;
+
+	if (ceph_snap(inode) != CEPH_NOSNAP) {
+		struct inode* dir;
+		bool unlinked = false;
+		/* do not support non-directory */
+		if (!d_is_dir(child)) {
+			dn = ERR_PTR(-EINVAL);
+			goto out;
+		}
+		dir = __lookup_inode(inode->i_sb, ceph_ino(inode));
+		if (IS_ERR(dir)) {
+			dn = ERR_CAST(dir);
+			goto out;
+		}
+		/* There can be multiple paths to access snapped inode.
+		 * For simplicity, treat snapdir of head inode as parent */
+		if (ceph_snap(inode) != CEPH_SNAPDIR) {
+			struct inode *snapdir = ceph_get_snapdir(dir);
+			if (dir->i_nlink == 0)
+				unlinked = true;
+			iput(dir);
+			if (IS_ERR(snapdir)) {
+				dn = ERR_CAST(snapdir);
+				goto out;
+			}
+			dir = snapdir;
+		}
+		/* If directory has already been deleted, futher get_parent
+		 * will fail. Do not mark snapdir dentry as disconnected,
+		 * this prevent exportfs from doing futher get_parent. */
+		if (unlinked)
+			dn = d_obtain_root(dir);
+		else
+			dn = d_obtain_alias(dir);
+	} else {
+		dn = __get_parent(child->d_sb, child, 0);
+	}
+out:
+	dout("get_parent %p ino %llx.%llx err=%ld\n",
+	     child, ceph_vinop(inode), (long)PTR_ERR_OR_ZERO(dn));
+	return dn;
+}
+
+/*
+ * convert regular fh to parent
+ */
+static struct dentry *ceph_fh_to_parent(struct super_block *sb,
+					struct fid *fid,
+					int fh_len, int fh_type)
+{
+	struct ceph_nfs_confh *cfh = (void *)fid->raw;
+	struct dentry *dentry;
+
+	if (fh_type == FILEID_BTRFS_WITH_PARENT) {
+		struct ceph_nfs_snapfh *sfh = (void *)fid->raw;
+		return __snapfh_to_dentry(sb, sfh, true);
+	}
+
+	if (fh_type != FILEID_INO32_GEN_PARENT)
+		return NULL;
+	if (fh_len < sizeof(*cfh) / 4)
+		return NULL;
+
+	dout("fh_to_parent %llx\n", cfh->parent_ino);
+	dentry = __get_parent(sb, NULL, cfh->ino);
+	if (unlikely(dentry == ERR_PTR(-ENOENT)))
+		dentry = __fh_to_dentry(sb, cfh->parent_ino);
+	return dentry;
+}
+
+static int __get_snap_name(struct dentry *parent, char *name,
+			   struct dentry *child)
+{
+	struct inode *inode = d_inode(child);
+	struct inode *dir = d_inode(parent);
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	struct ceph_mds_request *req = NULL;
+	char *last_name = NULL;
+	unsigned next_offset = 2;
+	int err = -EINVAL;
+
+	if (ceph_ino(inode) != ceph_ino(dir))
+		goto out;
+	if (ceph_snap(inode) == CEPH_SNAPDIR) {
+		if (ceph_snap(dir) == CEPH_NOSNAP) {
+			strcpy(name, fsc->mount_options->snapdir_name);
+			err = 0;
+		}
+		goto out;
+	}
+	if (ceph_snap(dir) != CEPH_SNAPDIR)
+		goto out;
+
+	while (1) {
+		struct ceph_mds_reply_info_parsed *rinfo;
+		struct ceph_mds_reply_dir_entry *rde;
+		int i;
+
+		req = ceph_mdsc_create_request(fsc->mdsc, CEPH_MDS_OP_LSSNAP,
+					       USE_AUTH_MDS);
+		if (IS_ERR(req)) {
+			err = PTR_ERR(req);
+			req = NULL;
+			goto out;
+		}
+		err = ceph_alloc_readdir_reply_buffer(req, inode);
+		if (err)
+			goto out;
+
+		req->r_direct_mode = USE_AUTH_MDS;
+		req->r_readdir_offset = next_offset;
+		req->r_args.readdir.flags =
+				cpu_to_le16(CEPH_READDIR_REPLY_BITFLAGS);
+		if (last_name) {
+			req->r_path2 = last_name;
+			last_name = NULL;
+		}
+
+		req->r_inode = dir;
+		ihold(dir);
+		req->r_dentry = dget(parent);
+
+		inode_lock(dir);
+		err = ceph_mdsc_do_request(fsc->mdsc, NULL, req);
+		inode_unlock(dir);
+
+		if (err < 0)
+			goto out;
+
+		rinfo = &req->r_reply_info;
+		for (i = 0; i < rinfo->dir_nr; i++) {
+			rde = rinfo->dir_entries + i;
+			BUG_ON(!rde->inode.in);
+			if (ceph_snap(inode) ==
+			    le64_to_cpu(rde->inode.in->snapid)) {
+				memcpy(name, rde->name, rde->name_len);
+				name[rde->name_len] = '\0';
+				err = 0;
+				goto out;
+			}
+		}
+
+		if (rinfo->dir_end)
+			break;
+
+		BUG_ON(rinfo->dir_nr <= 0);
+		rde = rinfo->dir_entries + (rinfo->dir_nr - 1);
+		next_offset += rinfo->dir_nr;
+		last_name = kstrndup(rde->name, rde->name_len, GFP_KERNEL);
+		if (!last_name) {
+			err = -ENOMEM;
+			goto out;
+		}
+
+		ceph_mdsc_put_request(req);
+		req = NULL;
+	}
+	err = -ENOENT;
+out:
+	if (req)
+		ceph_mdsc_put_request(req);
+	kfree(last_name);
+	dout("get_snap_name %p ino %llx.%llx err=%d\n",
+	     child, ceph_vinop(inode), err);
+	return err;
+}
+
+static int ceph_get_name(struct dentry *parent, char *name,
+			 struct dentry *child)
+{
+	struct ceph_mds_client *mdsc;
+	struct ceph_mds_request *req;
+	struct inode *inode = d_inode(child);
+	int err;
+
+	if (ceph_snap(inode) != CEPH_NOSNAP)
+		return __get_snap_name(parent, name, child);
+
+	mdsc = ceph_inode_to_client(inode)->mdsc;
+	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_LOOKUPNAME,
+				       USE_ANY_MDS);
+	if (IS_ERR(req))
+		return PTR_ERR(req);
+
+	inode_lock(d_inode(parent));
+
+	req->r_inode = inode;
+	ihold(inode);
+	req->r_ino2 = ceph_vino(d_inode(parent));
+	req->r_parent = d_inode(parent);
+	ihold(req->r_parent);
+	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
+	req->r_num_caps = 2;
+	err = ceph_mdsc_do_request(mdsc, NULL, req);
+
+	inode_unlock(d_inode(parent));
+
+	if (!err) {
+		struct ceph_mds_reply_info_parsed *rinfo = &req->r_reply_info;
+		memcpy(name, rinfo->dname, rinfo->dname_len);
+		name[rinfo->dname_len] = 0;
+		dout("get_name %p ino %llx.%llx name %s\n",
+		     child, ceph_vinop(inode), name);
+	} else {
+		dout("get_name %p ino %llx.%llx err %d\n",
+		     child, ceph_vinop(inode), err);
+	}
+
+	ceph_mdsc_put_request(req);
+	return err;
+}
+
+const struct export_operations ceph_export_ops = {
+	.encode_fh = ceph_encode_fh,
+	.fh_to_dentry = ceph_fh_to_dentry,
+	.fh_to_parent = ceph_fh_to_parent,
+	.get_parent = ceph_get_parent,
+	.get_name = ceph_get_name,
+};
diff --git a/fs/ceph/file.c b/fs/ceph/file.c
new file mode 100644
index 000000000..882eccfd6
--- /dev/null
+++ b/fs/ceph/file.c
@@ -0,0 +1,2600 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/ceph/ceph_debug.h>
+#include <linux/ceph/striper.h>
+
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/file.h>
+#include <linux/mount.h>
+#include <linux/namei.h>
+#include <linux/writeback.h>
+#include <linux/falloc.h>
+#include <linux/iversion.h>
+#include <linux/ktime.h>
+
+#include "super.h"
+#include "mds_client.h"
+#include "cache.h"
+#include "io.h"
+#include "metric.h"
+
+static __le32 ceph_flags_sys2wire(u32 flags)
+{
+	u32 wire_flags = 0;
+
+	switch (flags & O_ACCMODE) {
+	case O_RDONLY:
+		wire_flags |= CEPH_O_RDONLY;
+		break;
+	case O_WRONLY:
+		wire_flags |= CEPH_O_WRONLY;
+		break;
+	case O_RDWR:
+		wire_flags |= CEPH_O_RDWR;
+		break;
+	}
+
+	flags &= ~O_ACCMODE;
+
+#define ceph_sys2wire(a) if (flags & a) { wire_flags |= CEPH_##a; flags &= ~a; }
+
+	ceph_sys2wire(O_CREAT);
+	ceph_sys2wire(O_EXCL);
+	ceph_sys2wire(O_TRUNC);
+	ceph_sys2wire(O_DIRECTORY);
+	ceph_sys2wire(O_NOFOLLOW);
+
+#undef ceph_sys2wire
+
+	if (flags)
+		dout("unused open flags: %x\n", flags);
+
+	return cpu_to_le32(wire_flags);
+}
+
+/*
+ * Ceph file operations
+ *
+ * Implement basic open/close functionality, and implement
+ * read/write.
+ *
+ * We implement three modes of file I/O:
+ *  - buffered uses the generic_file_aio_{read,write} helpers
+ *
+ *  - synchronous is used when there is multi-client read/write
+ *    sharing, avoids the page cache, and synchronously waits for an
+ *    ack from the OSD.
+ *
+ *  - direct io takes the variant of the sync path that references
+ *    user pages directly.
+ *
+ * fsync() flushes and waits on dirty pages, but just queues metadata
+ * for writeback: since the MDS can recover size and mtime there is no
+ * need to wait for MDS acknowledgement.
+ */
+
+/*
+ * How many pages to get in one call to iov_iter_get_pages().  This
+ * determines the size of the on-stack array used as a buffer.
+ */
+#define ITER_GET_BVECS_PAGES	64
+
+static ssize_t __iter_get_bvecs(struct iov_iter *iter, size_t maxsize,
+				struct bio_vec *bvecs)
+{
+	size_t size = 0;
+	int bvec_idx = 0;
+
+	if (maxsize > iov_iter_count(iter))
+		maxsize = iov_iter_count(iter);
+
+	while (size < maxsize) {
+		struct page *pages[ITER_GET_BVECS_PAGES];
+		ssize_t bytes;
+		size_t start;
+		int idx = 0;
+
+		bytes = iov_iter_get_pages2(iter, pages, maxsize - size,
+					   ITER_GET_BVECS_PAGES, &start);
+		if (bytes < 0)
+			return size ?: bytes;
+
+		size += bytes;
+
+		for ( ; bytes; idx++, bvec_idx++) {
+			struct bio_vec bv = {
+				.bv_page = pages[idx],
+				.bv_len = min_t(int, bytes, PAGE_SIZE - start),
+				.bv_offset = start,
+			};
+
+			bvecs[bvec_idx] = bv;
+			bytes -= bv.bv_len;
+			start = 0;
+		}
+	}
+
+	return size;
+}
+
+/*
+ * iov_iter_get_pages() only considers one iov_iter segment, no matter
+ * what maxsize or maxpages are given.  For ITER_BVEC that is a single
+ * page.
+ *
+ * Attempt to get up to @maxsize bytes worth of pages from @iter.
+ * Return the number of bytes in the created bio_vec array, or an error.
+ */
+static ssize_t iter_get_bvecs_alloc(struct iov_iter *iter, size_t maxsize,
+				    struct bio_vec **bvecs, int *num_bvecs)
+{
+	struct bio_vec *bv;
+	size_t orig_count = iov_iter_count(iter);
+	ssize_t bytes;
+	int npages;
+
+	iov_iter_truncate(iter, maxsize);
+	npages = iov_iter_npages(iter, INT_MAX);
+	iov_iter_reexpand(iter, orig_count);
+
+	/*
+	 * __iter_get_bvecs() may populate only part of the array -- zero it
+	 * out.
+	 */
+	bv = kvmalloc_array(npages, sizeof(*bv), GFP_KERNEL | __GFP_ZERO);
+	if (!bv)
+		return -ENOMEM;
+
+	bytes = __iter_get_bvecs(iter, maxsize, bv);
+	if (bytes < 0) {
+		/*
+		 * No pages were pinned -- just free the array.
+		 */
+		kvfree(bv);
+		return bytes;
+	}
+
+	*bvecs = bv;
+	*num_bvecs = npages;
+	return bytes;
+}
+
+static void put_bvecs(struct bio_vec *bvecs, int num_bvecs, bool should_dirty)
+{
+	int i;
+
+	for (i = 0; i < num_bvecs; i++) {
+		if (bvecs[i].bv_page) {
+			if (should_dirty)
+				set_page_dirty_lock(bvecs[i].bv_page);
+			put_page(bvecs[i].bv_page);
+		}
+	}
+	kvfree(bvecs);
+}
+
+/*
+ * Prepare an open request.  Preallocate ceph_cap to avoid an
+ * inopportune ENOMEM later.
+ */
+static struct ceph_mds_request *
+prepare_open_request(struct super_block *sb, int flags, int create_mode)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(sb);
+	struct ceph_mds_request *req;
+	int want_auth = USE_ANY_MDS;
+	int op = (flags & O_CREAT) ? CEPH_MDS_OP_CREATE : CEPH_MDS_OP_OPEN;
+
+	if (flags & (O_WRONLY|O_RDWR|O_CREAT|O_TRUNC))
+		want_auth = USE_AUTH_MDS;
+
+	req = ceph_mdsc_create_request(mdsc, op, want_auth);
+	if (IS_ERR(req))
+		goto out;
+	req->r_fmode = ceph_flags_to_mode(flags);
+	req->r_args.open.flags = ceph_flags_sys2wire(flags);
+	req->r_args.open.mode = cpu_to_le32(create_mode);
+out:
+	return req;
+}
+
+static int ceph_init_file_info(struct inode *inode, struct file *file,
+					int fmode, bool isdir)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_mount_options *opt =
+		ceph_inode_to_client(&ci->netfs.inode)->mount_options;
+	struct ceph_file_info *fi;
+	int ret;
+
+	dout("%s %p %p 0%o (%s)\n", __func__, inode, file,
+			inode->i_mode, isdir ? "dir" : "regular");
+	BUG_ON(inode->i_fop->release != ceph_release);
+
+	if (isdir) {
+		struct ceph_dir_file_info *dfi =
+			kmem_cache_zalloc(ceph_dir_file_cachep, GFP_KERNEL);
+		if (!dfi)
+			return -ENOMEM;
+
+		file->private_data = dfi;
+		fi = &dfi->file_info;
+		dfi->next_offset = 2;
+		dfi->readdir_cache_idx = -1;
+	} else {
+		fi = kmem_cache_zalloc(ceph_file_cachep, GFP_KERNEL);
+		if (!fi)
+			return -ENOMEM;
+
+		if (opt->flags & CEPH_MOUNT_OPT_NOPAGECACHE)
+			fi->flags |= CEPH_F_SYNC;
+
+		file->private_data = fi;
+	}
+
+	ceph_get_fmode(ci, fmode, 1);
+	fi->fmode = fmode;
+
+	spin_lock_init(&fi->rw_contexts_lock);
+	INIT_LIST_HEAD(&fi->rw_contexts);
+	fi->filp_gen = READ_ONCE(ceph_inode_to_client(inode)->filp_gen);
+
+	if ((file->f_mode & FMODE_WRITE) && ceph_has_inline_data(ci)) {
+		ret = ceph_uninline_data(file);
+		if (ret < 0)
+			goto error;
+	}
+
+	return 0;
+
+error:
+	ceph_fscache_unuse_cookie(inode, file->f_mode & FMODE_WRITE);
+	ceph_put_fmode(ci, fi->fmode, 1);
+	kmem_cache_free(ceph_file_cachep, fi);
+	/* wake up anyone waiting for caps on this inode */
+	wake_up_all(&ci->i_cap_wq);
+	return ret;
+}
+
+/*
+ * initialize private struct file data.
+ * if we fail, clean up by dropping fmode reference on the ceph_inode
+ */
+static int ceph_init_file(struct inode *inode, struct file *file, int fmode)
+{
+	int ret = 0;
+
+	switch (inode->i_mode & S_IFMT) {
+	case S_IFREG:
+		ceph_fscache_use_cookie(inode, file->f_mode & FMODE_WRITE);
+		fallthrough;
+	case S_IFDIR:
+		ret = ceph_init_file_info(inode, file, fmode,
+						S_ISDIR(inode->i_mode));
+		break;
+
+	case S_IFLNK:
+		dout("init_file %p %p 0%o (symlink)\n", inode, file,
+		     inode->i_mode);
+		break;
+
+	default:
+		dout("init_file %p %p 0%o (special)\n", inode, file,
+		     inode->i_mode);
+		/*
+		 * we need to drop the open ref now, since we don't
+		 * have .release set to ceph_release.
+		 */
+		BUG_ON(inode->i_fop->release == ceph_release);
+
+		/* call the proper open fop */
+		ret = inode->i_fop->open(inode, file);
+	}
+	return ret;
+}
+
+/*
+ * try renew caps after session gets killed.
+ */
+int ceph_renew_caps(struct inode *inode, int fmode)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_mds_request *req;
+	int err, flags, wanted;
+
+	spin_lock(&ci->i_ceph_lock);
+	__ceph_touch_fmode(ci, mdsc, fmode);
+	wanted = __ceph_caps_file_wanted(ci);
+	if (__ceph_is_any_real_caps(ci) &&
+	    (!(wanted & CEPH_CAP_ANY_WR) || ci->i_auth_cap)) {
+		int issued = __ceph_caps_issued(ci, NULL);
+		spin_unlock(&ci->i_ceph_lock);
+		dout("renew caps %p want %s issued %s updating mds_wanted\n",
+		     inode, ceph_cap_string(wanted), ceph_cap_string(issued));
+		ceph_check_caps(ci, 0, NULL);
+		return 0;
+	}
+	spin_unlock(&ci->i_ceph_lock);
+
+	flags = 0;
+	if ((wanted & CEPH_CAP_FILE_RD) && (wanted & CEPH_CAP_FILE_WR))
+		flags = O_RDWR;
+	else if (wanted & CEPH_CAP_FILE_RD)
+		flags = O_RDONLY;
+	else if (wanted & CEPH_CAP_FILE_WR)
+		flags = O_WRONLY;
+#ifdef O_LAZY
+	if (wanted & CEPH_CAP_FILE_LAZYIO)
+		flags |= O_LAZY;
+#endif
+
+	req = prepare_open_request(inode->i_sb, flags, 0);
+	if (IS_ERR(req)) {
+		err = PTR_ERR(req);
+		goto out;
+	}
+
+	req->r_inode = inode;
+	ihold(inode);
+	req->r_num_caps = 1;
+
+	err = ceph_mdsc_do_request(mdsc, NULL, req);
+	ceph_mdsc_put_request(req);
+out:
+	dout("renew caps %p open result=%d\n", inode, err);
+	return err < 0 ? err : 0;
+}
+
+/*
+ * If we already have the requisite capabilities, we can satisfy
+ * the open request locally (no need to request new caps from the
+ * MDS).  We do, however, need to inform the MDS (asynchronously)
+ * if our wanted caps set expands.
+ */
+int ceph_open(struct inode *inode, struct file *file)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_fs_client *fsc = ceph_sb_to_client(inode->i_sb);
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	struct ceph_mds_request *req;
+	struct ceph_file_info *fi = file->private_data;
+	int err;
+	int flags, fmode, wanted;
+
+	if (fi) {
+		dout("open file %p is already opened\n", file);
+		return 0;
+	}
+
+	/* filter out O_CREAT|O_EXCL; vfs did that already.  yuck. */
+	flags = file->f_flags & ~(O_CREAT|O_EXCL);
+	if (S_ISDIR(inode->i_mode))
+		flags = O_DIRECTORY;  /* mds likes to know */
+
+	dout("open inode %p ino %llx.%llx file %p flags %d (%d)\n", inode,
+	     ceph_vinop(inode), file, flags, file->f_flags);
+	fmode = ceph_flags_to_mode(flags);
+	wanted = ceph_caps_for_mode(fmode);
+
+	/* snapped files are read-only */
+	if (ceph_snap(inode) != CEPH_NOSNAP && (file->f_mode & FMODE_WRITE))
+		return -EROFS;
+
+	/* trivially open snapdir */
+	if (ceph_snap(inode) == CEPH_SNAPDIR) {
+		return ceph_init_file(inode, file, fmode);
+	}
+
+	/*
+	 * No need to block if we have caps on the auth MDS (for
+	 * write) or any MDS (for read).  Update wanted set
+	 * asynchronously.
+	 */
+	spin_lock(&ci->i_ceph_lock);
+	if (__ceph_is_any_real_caps(ci) &&
+	    (((fmode & CEPH_FILE_MODE_WR) == 0) || ci->i_auth_cap)) {
+		int mds_wanted = __ceph_caps_mds_wanted(ci, true);
+		int issued = __ceph_caps_issued(ci, NULL);
+
+		dout("open %p fmode %d want %s issued %s using existing\n",
+		     inode, fmode, ceph_cap_string(wanted),
+		     ceph_cap_string(issued));
+		__ceph_touch_fmode(ci, mdsc, fmode);
+		spin_unlock(&ci->i_ceph_lock);
+
+		/* adjust wanted? */
+		if ((issued & wanted) != wanted &&
+		    (mds_wanted & wanted) != wanted &&
+		    ceph_snap(inode) != CEPH_SNAPDIR)
+			ceph_check_caps(ci, 0, NULL);
+
+		return ceph_init_file(inode, file, fmode);
+	} else if (ceph_snap(inode) != CEPH_NOSNAP &&
+		   (ci->i_snap_caps & wanted) == wanted) {
+		__ceph_touch_fmode(ci, mdsc, fmode);
+		spin_unlock(&ci->i_ceph_lock);
+		return ceph_init_file(inode, file, fmode);
+	}
+
+	spin_unlock(&ci->i_ceph_lock);
+
+	dout("open fmode %d wants %s\n", fmode, ceph_cap_string(wanted));
+	req = prepare_open_request(inode->i_sb, flags, 0);
+	if (IS_ERR(req)) {
+		err = PTR_ERR(req);
+		goto out;
+	}
+	req->r_inode = inode;
+	ihold(inode);
+
+	req->r_num_caps = 1;
+	err = ceph_mdsc_do_request(mdsc, NULL, req);
+	if (!err)
+		err = ceph_init_file(inode, file, req->r_fmode);
+	ceph_mdsc_put_request(req);
+	dout("open result=%d on %llx.%llx\n", err, ceph_vinop(inode));
+out:
+	return err;
+}
+
+/* Clone the layout from a synchronous create, if the dir now has Dc caps */
+static void
+cache_file_layout(struct inode *dst, struct inode *src)
+{
+	struct ceph_inode_info *cdst = ceph_inode(dst);
+	struct ceph_inode_info *csrc = ceph_inode(src);
+
+	spin_lock(&cdst->i_ceph_lock);
+	if ((__ceph_caps_issued(cdst, NULL) & CEPH_CAP_DIR_CREATE) &&
+	    !ceph_file_layout_is_valid(&cdst->i_cached_layout)) {
+		memcpy(&cdst->i_cached_layout, &csrc->i_layout,
+			sizeof(cdst->i_cached_layout));
+		rcu_assign_pointer(cdst->i_cached_layout.pool_ns,
+				   ceph_try_get_string(csrc->i_layout.pool_ns));
+	}
+	spin_unlock(&cdst->i_ceph_lock);
+}
+
+/*
+ * Try to set up an async create. We need caps, a file layout, and inode number,
+ * and either a lease on the dentry or complete dir info. If any of those
+ * criteria are not satisfied, then return false and the caller can go
+ * synchronous.
+ */
+static int try_prep_async_create(struct inode *dir, struct dentry *dentry,
+				 struct ceph_file_layout *lo, u64 *pino)
+{
+	struct ceph_inode_info *ci = ceph_inode(dir);
+	struct ceph_dentry_info *di = ceph_dentry(dentry);
+	int got = 0, want = CEPH_CAP_FILE_EXCL | CEPH_CAP_DIR_CREATE;
+	u64 ino;
+
+	spin_lock(&ci->i_ceph_lock);
+	/* No auth cap means no chance for Dc caps */
+	if (!ci->i_auth_cap)
+		goto no_async;
+
+	/* Any delegated inos? */
+	if (xa_empty(&ci->i_auth_cap->session->s_delegated_inos))
+		goto no_async;
+
+	if (!ceph_file_layout_is_valid(&ci->i_cached_layout))
+		goto no_async;
+
+	if ((__ceph_caps_issued(ci, NULL) & want) != want)
+		goto no_async;
+
+	if (d_in_lookup(dentry)) {
+		if (!__ceph_dir_is_complete(ci))
+			goto no_async;
+		spin_lock(&dentry->d_lock);
+		di->lease_shared_gen = atomic_read(&ci->i_shared_gen);
+		spin_unlock(&dentry->d_lock);
+	} else if (atomic_read(&ci->i_shared_gen) !=
+		   READ_ONCE(di->lease_shared_gen)) {
+		goto no_async;
+	}
+
+	ino = ceph_get_deleg_ino(ci->i_auth_cap->session);
+	if (!ino)
+		goto no_async;
+
+	*pino = ino;
+	ceph_take_cap_refs(ci, want, false);
+	memcpy(lo, &ci->i_cached_layout, sizeof(*lo));
+	rcu_assign_pointer(lo->pool_ns,
+			   ceph_try_get_string(ci->i_cached_layout.pool_ns));
+	got = want;
+no_async:
+	spin_unlock(&ci->i_ceph_lock);
+	return got;
+}
+
+static void restore_deleg_ino(struct inode *dir, u64 ino)
+{
+	struct ceph_inode_info *ci = ceph_inode(dir);
+	struct ceph_mds_session *s = NULL;
+
+	spin_lock(&ci->i_ceph_lock);
+	if (ci->i_auth_cap)
+		s = ceph_get_mds_session(ci->i_auth_cap->session);
+	spin_unlock(&ci->i_ceph_lock);
+	if (s) {
+		int err = ceph_restore_deleg_ino(s, ino);
+		if (err)
+			pr_warn("ceph: unable to restore delegated ino 0x%llx to session: %d\n",
+				ino, err);
+		ceph_put_mds_session(s);
+	}
+}
+
+static void wake_async_create_waiters(struct inode *inode,
+				      struct ceph_mds_session *session)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+
+	spin_lock(&ci->i_ceph_lock);
+	if (ci->i_ceph_flags & CEPH_I_ASYNC_CREATE) {
+		ci->i_ceph_flags &= ~CEPH_I_ASYNC_CREATE;
+		wake_up_bit(&ci->i_ceph_flags, CEPH_ASYNC_CREATE_BIT);
+	}
+	ceph_kick_flushing_inode_caps(session, ci);
+	spin_unlock(&ci->i_ceph_lock);
+}
+
+static void ceph_async_create_cb(struct ceph_mds_client *mdsc,
+                                 struct ceph_mds_request *req)
+{
+	struct dentry *dentry = req->r_dentry;
+	struct inode *dinode = d_inode(dentry);
+	struct inode *tinode = req->r_target_inode;
+	int result = req->r_err ? req->r_err :
+			le32_to_cpu(req->r_reply_info.head->result);
+
+	WARN_ON_ONCE(dinode && tinode && dinode != tinode);
+
+	/* MDS changed -- caller must resubmit */
+	if (result == -EJUKEBOX)
+		goto out;
+
+	mapping_set_error(req->r_parent->i_mapping, result);
+
+	if (result) {
+		int pathlen = 0;
+		u64 base = 0;
+		char *path = ceph_mdsc_build_path(req->r_dentry, &pathlen,
+						  &base, 0);
+
+		pr_warn("async create failure path=(%llx)%s result=%d!\n",
+			base, IS_ERR(path) ? "<<bad>>" : path, result);
+		ceph_mdsc_free_path(path, pathlen);
+
+		ceph_dir_clear_complete(req->r_parent);
+		if (!d_unhashed(dentry))
+			d_drop(dentry);
+
+		if (dinode) {
+			mapping_set_error(dinode->i_mapping, result);
+			ceph_inode_shutdown(dinode);
+			wake_async_create_waiters(dinode, req->r_session);
+		}
+	}
+
+	if (tinode) {
+		u64 ino = ceph_vino(tinode).ino;
+
+		if (req->r_deleg_ino != ino)
+			pr_warn("%s: inode number mismatch! err=%d deleg_ino=0x%llx target=0x%llx\n",
+				__func__, req->r_err, req->r_deleg_ino, ino);
+
+		mapping_set_error(tinode->i_mapping, result);
+		wake_async_create_waiters(tinode, req->r_session);
+	} else if (!result) {
+		pr_warn("%s: no req->r_target_inode for 0x%llx\n", __func__,
+			req->r_deleg_ino);
+	}
+out:
+	ceph_mdsc_release_dir_caps(req);
+}
+
+static int ceph_finish_async_create(struct inode *dir, struct dentry *dentry,
+				    struct file *file, umode_t mode,
+				    struct ceph_mds_request *req,
+				    struct ceph_acl_sec_ctx *as_ctx,
+				    struct ceph_file_layout *lo)
+{
+	int ret;
+	char xattr_buf[4];
+	struct ceph_mds_reply_inode in = { };
+	struct ceph_mds_reply_info_in iinfo = { .in = &in };
+	struct ceph_inode_info *ci = ceph_inode(dir);
+	struct ceph_dentry_info *di = ceph_dentry(dentry);
+	struct inode *inode;
+	struct timespec64 now;
+	struct ceph_string *pool_ns;
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(dir->i_sb);
+	struct ceph_vino vino = { .ino = req->r_deleg_ino,
+				  .snap = CEPH_NOSNAP };
+
+	ktime_get_real_ts64(&now);
+
+	inode = ceph_get_inode(dentry->d_sb, vino);
+	if (IS_ERR(inode))
+		return PTR_ERR(inode);
+
+	iinfo.inline_version = CEPH_INLINE_NONE;
+	iinfo.change_attr = 1;
+	ceph_encode_timespec64(&iinfo.btime, &now);
+
+	if (req->r_pagelist) {
+		iinfo.xattr_len = req->r_pagelist->length;
+		iinfo.xattr_data = req->r_pagelist->mapped_tail;
+	} else {
+		/* fake it */
+		iinfo.xattr_len = ARRAY_SIZE(xattr_buf);
+		iinfo.xattr_data = xattr_buf;
+		memset(iinfo.xattr_data, 0, iinfo.xattr_len);
+	}
+
+	in.ino = cpu_to_le64(vino.ino);
+	in.snapid = cpu_to_le64(CEPH_NOSNAP);
+	in.version = cpu_to_le64(1);	// ???
+	in.cap.caps = in.cap.wanted = cpu_to_le32(CEPH_CAP_ALL_FILE);
+	in.cap.cap_id = cpu_to_le64(1);
+	in.cap.realm = cpu_to_le64(ci->i_snap_realm->ino);
+	in.cap.flags = CEPH_CAP_FLAG_AUTH;
+	in.ctime = in.mtime = in.atime = iinfo.btime;
+	in.truncate_seq = cpu_to_le32(1);
+	in.truncate_size = cpu_to_le64(-1ULL);
+	in.xattr_version = cpu_to_le64(1);
+	in.uid = cpu_to_le32(from_kuid(&init_user_ns, current_fsuid()));
+	if (dir->i_mode & S_ISGID) {
+		in.gid = cpu_to_le32(from_kgid(&init_user_ns, dir->i_gid));
+
+		/* Directories always inherit the setgid bit. */
+		if (S_ISDIR(mode))
+			mode |= S_ISGID;
+	} else {
+		in.gid = cpu_to_le32(from_kgid(&init_user_ns, current_fsgid()));
+	}
+	in.mode = cpu_to_le32((u32)mode);
+
+	in.nlink = cpu_to_le32(1);
+	in.max_size = cpu_to_le64(lo->stripe_unit);
+
+	ceph_file_layout_to_legacy(lo, &in.layout);
+	/* lo is private, so pool_ns can't change */
+	pool_ns = rcu_dereference_raw(lo->pool_ns);
+	if (pool_ns) {
+		iinfo.pool_ns_len = pool_ns->len;
+		iinfo.pool_ns_data = pool_ns->str;
+	}
+
+	down_read(&mdsc->snap_rwsem);
+	ret = ceph_fill_inode(inode, NULL, &iinfo, NULL, req->r_session,
+			      req->r_fmode, NULL);
+	up_read(&mdsc->snap_rwsem);
+	if (ret) {
+		dout("%s failed to fill inode: %d\n", __func__, ret);
+		ceph_dir_clear_complete(dir);
+		if (!d_unhashed(dentry))
+			d_drop(dentry);
+		if (inode->i_state & I_NEW)
+			discard_new_inode(inode);
+	} else {
+		struct dentry *dn;
+
+		dout("%s d_adding new inode 0x%llx to 0x%llx/%s\n", __func__,
+			vino.ino, ceph_ino(dir), dentry->d_name.name);
+		ceph_dir_clear_ordered(dir);
+		ceph_init_inode_acls(inode, as_ctx);
+		if (inode->i_state & I_NEW) {
+			/*
+			 * If it's not I_NEW, then someone created this before
+			 * we got here. Assume the server is aware of it at
+			 * that point and don't worry about setting
+			 * CEPH_I_ASYNC_CREATE.
+			 */
+			ceph_inode(inode)->i_ceph_flags = CEPH_I_ASYNC_CREATE;
+			unlock_new_inode(inode);
+		}
+		if (d_in_lookup(dentry) || d_really_is_negative(dentry)) {
+			if (!d_unhashed(dentry))
+				d_drop(dentry);
+			dn = d_splice_alias(inode, dentry);
+			WARN_ON_ONCE(dn && dn != dentry);
+		}
+		file->f_mode |= FMODE_CREATED;
+		ret = finish_open(file, dentry, ceph_open);
+	}
+
+	spin_lock(&dentry->d_lock);
+	di->flags &= ~CEPH_DENTRY_ASYNC_CREATE;
+	wake_up_bit(&di->flags, CEPH_DENTRY_ASYNC_CREATE_BIT);
+	spin_unlock(&dentry->d_lock);
+
+	return ret;
+}
+
+/*
+ * Do a lookup + open with a single request.  If we get a non-existent
+ * file or symlink, return 1 so the VFS can retry.
+ */
+int ceph_atomic_open(struct inode *dir, struct dentry *dentry,
+		     struct file *file, unsigned flags, umode_t mode)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(dir->i_sb);
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	struct ceph_mds_request *req;
+	struct dentry *dn;
+	struct ceph_acl_sec_ctx as_ctx = {};
+	bool try_async = ceph_test_mount_opt(fsc, ASYNC_DIROPS);
+	int mask;
+	int err;
+
+	dout("atomic_open %p dentry %p '%pd' %s flags %d mode 0%o\n",
+	     dir, dentry, dentry,
+	     d_unhashed(dentry) ? "unhashed" : "hashed", flags, mode);
+
+	if (dentry->d_name.len > NAME_MAX)
+		return -ENAMETOOLONG;
+
+	err = ceph_wait_on_conflict_unlink(dentry);
+	if (err)
+		return err;
+	/*
+	 * Do not truncate the file, since atomic_open is called before the
+	 * permission check. The caller will do the truncation afterward.
+	 */
+	flags &= ~O_TRUNC;
+
+	if (flags & O_CREAT) {
+		if (ceph_quota_is_max_files_exceeded(dir))
+			return -EDQUOT;
+		err = ceph_pre_init_acls(dir, &mode, &as_ctx);
+		if (err < 0)
+			return err;
+		err = ceph_security_init_secctx(dentry, mode, &as_ctx);
+		if (err < 0)
+			goto out_ctx;
+		/* Async create can't handle more than a page of xattrs */
+		if (as_ctx.pagelist &&
+		    !list_is_singular(&as_ctx.pagelist->head))
+			try_async = false;
+	} else if (!d_in_lookup(dentry)) {
+		/* If it's not being looked up, it's negative */
+		return -ENOENT;
+	}
+retry:
+	/* do the open */
+	req = prepare_open_request(dir->i_sb, flags, mode);
+	if (IS_ERR(req)) {
+		err = PTR_ERR(req);
+		goto out_ctx;
+	}
+	req->r_dentry = dget(dentry);
+	req->r_num_caps = 2;
+	mask = CEPH_STAT_CAP_INODE | CEPH_CAP_AUTH_SHARED;
+	if (ceph_security_xattr_wanted(dir))
+		mask |= CEPH_CAP_XATTR_SHARED;
+	req->r_args.open.mask = cpu_to_le32(mask);
+	req->r_parent = dir;
+	ihold(dir);
+
+	if (flags & O_CREAT) {
+		struct ceph_file_layout lo;
+
+		req->r_dentry_drop = CEPH_CAP_FILE_SHARED | CEPH_CAP_AUTH_EXCL;
+		req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
+		if (as_ctx.pagelist) {
+			req->r_pagelist = as_ctx.pagelist;
+			as_ctx.pagelist = NULL;
+		}
+		if (try_async &&
+		    (req->r_dir_caps =
+		      try_prep_async_create(dir, dentry, &lo,
+					    &req->r_deleg_ino))) {
+			struct ceph_dentry_info *di = ceph_dentry(dentry);
+
+			set_bit(CEPH_MDS_R_ASYNC, &req->r_req_flags);
+			req->r_args.open.flags |= cpu_to_le32(CEPH_O_EXCL);
+			req->r_callback = ceph_async_create_cb;
+
+			spin_lock(&dentry->d_lock);
+			di->flags |= CEPH_DENTRY_ASYNC_CREATE;
+			spin_unlock(&dentry->d_lock);
+
+			err = ceph_mdsc_submit_request(mdsc, dir, req);
+			if (!err) {
+				err = ceph_finish_async_create(dir, dentry,
+							file, mode, req,
+							&as_ctx, &lo);
+			} else if (err == -EJUKEBOX) {
+				restore_deleg_ino(dir, req->r_deleg_ino);
+				ceph_mdsc_put_request(req);
+				try_async = false;
+				ceph_put_string(rcu_dereference_raw(lo.pool_ns));
+				goto retry;
+			}
+			ceph_put_string(rcu_dereference_raw(lo.pool_ns));
+			goto out_req;
+		}
+	}
+
+	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
+	err = ceph_mdsc_do_request(mdsc, (flags & O_CREAT) ? dir : NULL, req);
+	if (err == -ENOENT) {
+		dentry = ceph_handle_snapdir(req, dentry);
+		if (IS_ERR(dentry)) {
+			err = PTR_ERR(dentry);
+			goto out_req;
+		}
+		err = 0;
+	}
+
+	if (!err && (flags & O_CREAT) && !req->r_reply_info.head->is_dentry)
+		err = ceph_handle_notrace_create(dir, dentry);
+
+	if (d_in_lookup(dentry)) {
+		dn = ceph_finish_lookup(req, dentry, err);
+		if (IS_ERR(dn))
+			err = PTR_ERR(dn);
+	} else {
+		/* we were given a hashed negative dentry */
+		dn = NULL;
+	}
+	if (err)
+		goto out_req;
+	if (dn || d_really_is_negative(dentry) || d_is_symlink(dentry)) {
+		/* make vfs retry on splice, ENOENT, or symlink */
+		dout("atomic_open finish_no_open on dn %p\n", dn);
+		err = finish_no_open(file, dn);
+	} else {
+		dout("atomic_open finish_open on dn %p\n", dn);
+		if (req->r_op == CEPH_MDS_OP_CREATE && req->r_reply_info.has_create_ino) {
+			struct inode *newino = d_inode(dentry);
+
+			cache_file_layout(dir, newino);
+			ceph_init_inode_acls(newino, &as_ctx);
+			file->f_mode |= FMODE_CREATED;
+		}
+		err = finish_open(file, dentry, ceph_open);
+	}
+out_req:
+	ceph_mdsc_put_request(req);
+out_ctx:
+	ceph_release_acl_sec_ctx(&as_ctx);
+	dout("atomic_open result=%d\n", err);
+	return err;
+}
+
+int ceph_release(struct inode *inode, struct file *file)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+
+	if (S_ISDIR(inode->i_mode)) {
+		struct ceph_dir_file_info *dfi = file->private_data;
+		dout("release inode %p dir file %p\n", inode, file);
+		WARN_ON(!list_empty(&dfi->file_info.rw_contexts));
+
+		ceph_put_fmode(ci, dfi->file_info.fmode, 1);
+
+		if (dfi->last_readdir)
+			ceph_mdsc_put_request(dfi->last_readdir);
+		kfree(dfi->last_name);
+		kfree(dfi->dir_info);
+		kmem_cache_free(ceph_dir_file_cachep, dfi);
+	} else {
+		struct ceph_file_info *fi = file->private_data;
+		dout("release inode %p regular file %p\n", inode, file);
+		WARN_ON(!list_empty(&fi->rw_contexts));
+
+		ceph_fscache_unuse_cookie(inode, file->f_mode & FMODE_WRITE);
+		ceph_put_fmode(ci, fi->fmode, 1);
+
+		kmem_cache_free(ceph_file_cachep, fi);
+	}
+
+	/* wake up anyone waiting for caps on this inode */
+	wake_up_all(&ci->i_cap_wq);
+	return 0;
+}
+
+enum {
+	HAVE_RETRIED = 1,
+	CHECK_EOF =    2,
+	READ_INLINE =  3,
+};
+
+/*
+ * Completely synchronous read and write methods.  Direct from __user
+ * buffer to osd, or directly to user pages (if O_DIRECT).
+ *
+ * If the read spans object boundary, just do multiple reads.  (That's not
+ * atomic, but good enough for now.)
+ *
+ * If we get a short result from the OSD, check against i_size; we need to
+ * only return a short read to the caller if we hit EOF.
+ */
+static ssize_t ceph_sync_read(struct kiocb *iocb, struct iov_iter *to,
+			      int *retry_op)
+{
+	struct file *file = iocb->ki_filp;
+	struct inode *inode = file_inode(file);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	struct ceph_osd_client *osdc = &fsc->client->osdc;
+	ssize_t ret;
+	u64 off = iocb->ki_pos;
+	u64 len = iov_iter_count(to);
+	u64 i_size = i_size_read(inode);
+
+	dout("sync_read on file %p %llu~%u %s\n", file, off, (unsigned)len,
+	     (file->f_flags & O_DIRECT) ? "O_DIRECT" : "");
+
+	if (!len)
+		return 0;
+	/*
+	 * flush any page cache pages in this range.  this
+	 * will make concurrent normal and sync io slow,
+	 * but it will at least behave sensibly when they are
+	 * in sequence.
+	 */
+	ret = filemap_write_and_wait_range(inode->i_mapping,
+					   off, off + len - 1);
+	if (ret < 0)
+		return ret;
+
+	ret = 0;
+	while ((len = iov_iter_count(to)) > 0) {
+		struct ceph_osd_request *req;
+		struct page **pages;
+		int num_pages;
+		size_t page_off;
+		bool more;
+		int idx;
+		size_t left;
+
+		req = ceph_osdc_new_request(osdc, &ci->i_layout,
+					ci->i_vino, off, &len, 0, 1,
+					CEPH_OSD_OP_READ, CEPH_OSD_FLAG_READ,
+					NULL, ci->i_truncate_seq,
+					ci->i_truncate_size, false);
+		if (IS_ERR(req)) {
+			ret = PTR_ERR(req);
+			break;
+		}
+
+		more = len < iov_iter_count(to);
+
+		num_pages = calc_pages_for(off, len);
+		page_off = off & ~PAGE_MASK;
+		pages = ceph_alloc_page_vector(num_pages, GFP_KERNEL);
+		if (IS_ERR(pages)) {
+			ceph_osdc_put_request(req);
+			ret = PTR_ERR(pages);
+			break;
+		}
+
+		osd_req_op_extent_osd_data_pages(req, 0, pages, len, page_off,
+						 false, false);
+		ceph_osdc_start_request(osdc, req);
+		ret = ceph_osdc_wait_request(osdc, req);
+
+		ceph_update_read_metrics(&fsc->mdsc->metric,
+					 req->r_start_latency,
+					 req->r_end_latency,
+					 len, ret);
+
+		ceph_osdc_put_request(req);
+
+		i_size = i_size_read(inode);
+		dout("sync_read %llu~%llu got %zd i_size %llu%s\n",
+		     off, len, ret, i_size, (more ? " MORE" : ""));
+
+		if (ret == -ENOENT)
+			ret = 0;
+		if (ret >= 0 && ret < len && (off + ret < i_size)) {
+			int zlen = min(len - ret, i_size - off - ret);
+			int zoff = page_off + ret;
+			dout("sync_read zero gap %llu~%llu\n",
+                             off + ret, off + ret + zlen);
+			ceph_zero_page_vector_range(zoff, zlen, pages);
+			ret += zlen;
+		}
+
+		idx = 0;
+		left = ret > 0 ? ret : 0;
+		while (left > 0) {
+			size_t len, copied;
+			page_off = off & ~PAGE_MASK;
+			len = min_t(size_t, left, PAGE_SIZE - page_off);
+			SetPageUptodate(pages[idx]);
+			copied = copy_page_to_iter(pages[idx++],
+						   page_off, len, to);
+			off += copied;
+			left -= copied;
+			if (copied < len) {
+				ret = -EFAULT;
+				break;
+			}
+		}
+		ceph_release_page_vector(pages, num_pages);
+
+		if (ret < 0) {
+			if (ret == -EBLOCKLISTED)
+				fsc->blocklisted = true;
+			break;
+		}
+
+		if (off >= i_size || !more)
+			break;
+	}
+
+	if (off > iocb->ki_pos) {
+		if (off >= i_size) {
+			*retry_op = CHECK_EOF;
+			ret = i_size - iocb->ki_pos;
+			iocb->ki_pos = i_size;
+		} else {
+			ret = off - iocb->ki_pos;
+			iocb->ki_pos = off;
+		}
+	}
+
+	dout("sync_read result %zd retry_op %d\n", ret, *retry_op);
+	return ret;
+}
+
+struct ceph_aio_request {
+	struct kiocb *iocb;
+	size_t total_len;
+	bool write;
+	bool should_dirty;
+	int error;
+	struct list_head osd_reqs;
+	unsigned num_reqs;
+	atomic_t pending_reqs;
+	struct timespec64 mtime;
+	struct ceph_cap_flush *prealloc_cf;
+};
+
+struct ceph_aio_work {
+	struct work_struct work;
+	struct ceph_osd_request *req;
+};
+
+static void ceph_aio_retry_work(struct work_struct *work);
+
+static void ceph_aio_complete(struct inode *inode,
+			      struct ceph_aio_request *aio_req)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	int ret;
+
+	if (!atomic_dec_and_test(&aio_req->pending_reqs))
+		return;
+
+	if (aio_req->iocb->ki_flags & IOCB_DIRECT)
+		inode_dio_end(inode);
+
+	ret = aio_req->error;
+	if (!ret)
+		ret = aio_req->total_len;
+
+	dout("ceph_aio_complete %p rc %d\n", inode, ret);
+
+	if (ret >= 0 && aio_req->write) {
+		int dirty;
+
+		loff_t endoff = aio_req->iocb->ki_pos + aio_req->total_len;
+		if (endoff > i_size_read(inode)) {
+			if (ceph_inode_set_size(inode, endoff))
+				ceph_check_caps(ci, CHECK_CAPS_AUTHONLY, NULL);
+		}
+
+		spin_lock(&ci->i_ceph_lock);
+		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
+					       &aio_req->prealloc_cf);
+		spin_unlock(&ci->i_ceph_lock);
+		if (dirty)
+			__mark_inode_dirty(inode, dirty);
+
+	}
+
+	ceph_put_cap_refs(ci, (aio_req->write ? CEPH_CAP_FILE_WR :
+						CEPH_CAP_FILE_RD));
+
+	aio_req->iocb->ki_complete(aio_req->iocb, ret);
+
+	ceph_free_cap_flush(aio_req->prealloc_cf);
+	kfree(aio_req);
+}
+
+static void ceph_aio_complete_req(struct ceph_osd_request *req)
+{
+	int rc = req->r_result;
+	struct inode *inode = req->r_inode;
+	struct ceph_aio_request *aio_req = req->r_priv;
+	struct ceph_osd_data *osd_data = osd_req_op_extent_osd_data(req, 0);
+	struct ceph_client_metric *metric = &ceph_sb_to_mdsc(inode->i_sb)->metric;
+	unsigned int len = osd_data->bvec_pos.iter.bi_size;
+
+	BUG_ON(osd_data->type != CEPH_OSD_DATA_TYPE_BVECS);
+	BUG_ON(!osd_data->num_bvecs);
+
+	dout("ceph_aio_complete_req %p rc %d bytes %u\n", inode, rc, len);
+
+	if (rc == -EOLDSNAPC) {
+		struct ceph_aio_work *aio_work;
+		BUG_ON(!aio_req->write);
+
+		aio_work = kmalloc(sizeof(*aio_work), GFP_NOFS);
+		if (aio_work) {
+			INIT_WORK(&aio_work->work, ceph_aio_retry_work);
+			aio_work->req = req;
+			queue_work(ceph_inode_to_client(inode)->inode_wq,
+				   &aio_work->work);
+			return;
+		}
+		rc = -ENOMEM;
+	} else if (!aio_req->write) {
+		if (rc == -ENOENT)
+			rc = 0;
+		if (rc >= 0 && len > rc) {
+			struct iov_iter i;
+			int zlen = len - rc;
+
+			/*
+			 * If read is satisfied by single OSD request,
+			 * it can pass EOF. Otherwise read is within
+			 * i_size.
+			 */
+			if (aio_req->num_reqs == 1) {
+				loff_t i_size = i_size_read(inode);
+				loff_t endoff = aio_req->iocb->ki_pos + rc;
+				if (endoff < i_size)
+					zlen = min_t(size_t, zlen,
+						     i_size - endoff);
+				aio_req->total_len = rc + zlen;
+			}
+
+			iov_iter_bvec(&i, ITER_DEST, osd_data->bvec_pos.bvecs,
+				      osd_data->num_bvecs, len);
+			iov_iter_advance(&i, rc);
+			iov_iter_zero(zlen, &i);
+		}
+	}
+
+	/* r_start_latency == 0 means the request was not submitted */
+	if (req->r_start_latency) {
+		if (aio_req->write)
+			ceph_update_write_metrics(metric, req->r_start_latency,
+						  req->r_end_latency, len, rc);
+		else
+			ceph_update_read_metrics(metric, req->r_start_latency,
+						 req->r_end_latency, len, rc);
+	}
+
+	put_bvecs(osd_data->bvec_pos.bvecs, osd_data->num_bvecs,
+		  aio_req->should_dirty);
+	ceph_osdc_put_request(req);
+
+	if (rc < 0)
+		cmpxchg(&aio_req->error, 0, rc);
+
+	ceph_aio_complete(inode, aio_req);
+	return;
+}
+
+static void ceph_aio_retry_work(struct work_struct *work)
+{
+	struct ceph_aio_work *aio_work =
+		container_of(work, struct ceph_aio_work, work);
+	struct ceph_osd_request *orig_req = aio_work->req;
+	struct ceph_aio_request *aio_req = orig_req->r_priv;
+	struct inode *inode = orig_req->r_inode;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_snap_context *snapc;
+	struct ceph_osd_request *req;
+	int ret;
+
+	spin_lock(&ci->i_ceph_lock);
+	if (__ceph_have_pending_cap_snap(ci)) {
+		struct ceph_cap_snap *capsnap =
+			list_last_entry(&ci->i_cap_snaps,
+					struct ceph_cap_snap,
+					ci_item);
+		snapc = ceph_get_snap_context(capsnap->context);
+	} else {
+		BUG_ON(!ci->i_head_snapc);
+		snapc = ceph_get_snap_context(ci->i_head_snapc);
+	}
+	spin_unlock(&ci->i_ceph_lock);
+
+	req = ceph_osdc_alloc_request(orig_req->r_osdc, snapc, 1,
+			false, GFP_NOFS);
+	if (!req) {
+		ret = -ENOMEM;
+		req = orig_req;
+		goto out;
+	}
+
+	req->r_flags = /* CEPH_OSD_FLAG_ORDERSNAP | */ CEPH_OSD_FLAG_WRITE;
+	ceph_oloc_copy(&req->r_base_oloc, &orig_req->r_base_oloc);
+	ceph_oid_copy(&req->r_base_oid, &orig_req->r_base_oid);
+
+	req->r_ops[0] = orig_req->r_ops[0];
+
+	req->r_mtime = aio_req->mtime;
+	req->r_data_offset = req->r_ops[0].extent.offset;
+
+	ret = ceph_osdc_alloc_messages(req, GFP_NOFS);
+	if (ret) {
+		ceph_osdc_put_request(req);
+		req = orig_req;
+		goto out;
+	}
+
+	ceph_osdc_put_request(orig_req);
+
+	req->r_callback = ceph_aio_complete_req;
+	req->r_inode = inode;
+	req->r_priv = aio_req;
+
+	ceph_osdc_start_request(req->r_osdc, req);
+out:
+	if (ret < 0) {
+		req->r_result = ret;
+		ceph_aio_complete_req(req);
+	}
+
+	ceph_put_snap_context(snapc);
+	kfree(aio_work);
+}
+
+static ssize_t
+ceph_direct_read_write(struct kiocb *iocb, struct iov_iter *iter,
+		       struct ceph_snap_context *snapc,
+		       struct ceph_cap_flush **pcf)
+{
+	struct file *file = iocb->ki_filp;
+	struct inode *inode = file_inode(file);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	struct ceph_client_metric *metric = &fsc->mdsc->metric;
+	struct ceph_vino vino;
+	struct ceph_osd_request *req;
+	struct bio_vec *bvecs;
+	struct ceph_aio_request *aio_req = NULL;
+	int num_pages = 0;
+	int flags;
+	int ret = 0;
+	struct timespec64 mtime = current_time(inode);
+	size_t count = iov_iter_count(iter);
+	loff_t pos = iocb->ki_pos;
+	bool write = iov_iter_rw(iter) == WRITE;
+	bool should_dirty = !write && user_backed_iter(iter);
+
+	if (write && ceph_snap(file_inode(file)) != CEPH_NOSNAP)
+		return -EROFS;
+
+	dout("sync_direct_%s on file %p %lld~%u snapc %p seq %lld\n",
+	     (write ? "write" : "read"), file, pos, (unsigned)count,
+	     snapc, snapc ? snapc->seq : 0);
+
+	if (write) {
+		int ret2;
+
+		ceph_fscache_invalidate(inode, true);
+
+		ret2 = invalidate_inode_pages2_range(inode->i_mapping,
+					pos >> PAGE_SHIFT,
+					(pos + count - 1) >> PAGE_SHIFT);
+		if (ret2 < 0)
+			dout("invalidate_inode_pages2_range returned %d\n", ret2);
+
+		flags = /* CEPH_OSD_FLAG_ORDERSNAP | */ CEPH_OSD_FLAG_WRITE;
+	} else {
+		flags = CEPH_OSD_FLAG_READ;
+	}
+
+	while (iov_iter_count(iter) > 0) {
+		u64 size = iov_iter_count(iter);
+		ssize_t len;
+
+		if (write)
+			size = min_t(u64, size, fsc->mount_options->wsize);
+		else
+			size = min_t(u64, size, fsc->mount_options->rsize);
+
+		vino = ceph_vino(inode);
+		req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
+					    vino, pos, &size, 0,
+					    1,
+					    write ? CEPH_OSD_OP_WRITE :
+						    CEPH_OSD_OP_READ,
+					    flags, snapc,
+					    ci->i_truncate_seq,
+					    ci->i_truncate_size,
+					    false);
+		if (IS_ERR(req)) {
+			ret = PTR_ERR(req);
+			break;
+		}
+
+		len = iter_get_bvecs_alloc(iter, size, &bvecs, &num_pages);
+		if (len < 0) {
+			ceph_osdc_put_request(req);
+			ret = len;
+			break;
+		}
+		if (len != size)
+			osd_req_op_extent_update(req, 0, len);
+
+		/*
+		 * To simplify error handling, allow AIO when IO within i_size
+		 * or IO can be satisfied by single OSD request.
+		 */
+		if (pos == iocb->ki_pos && !is_sync_kiocb(iocb) &&
+		    (len == count || pos + count <= i_size_read(inode))) {
+			aio_req = kzalloc(sizeof(*aio_req), GFP_KERNEL);
+			if (aio_req) {
+				aio_req->iocb = iocb;
+				aio_req->write = write;
+				aio_req->should_dirty = should_dirty;
+				INIT_LIST_HEAD(&aio_req->osd_reqs);
+				if (write) {
+					aio_req->mtime = mtime;
+					swap(aio_req->prealloc_cf, *pcf);
+				}
+			}
+			/* ignore error */
+		}
+
+		if (write) {
+			/*
+			 * throw out any page cache pages in this range. this
+			 * may block.
+			 */
+			truncate_inode_pages_range(inode->i_mapping, pos,
+						   PAGE_ALIGN(pos + len) - 1);
+
+			req->r_mtime = mtime;
+		}
+
+		osd_req_op_extent_osd_data_bvecs(req, 0, bvecs, num_pages, len);
+
+		if (aio_req) {
+			aio_req->total_len += len;
+			aio_req->num_reqs++;
+			atomic_inc(&aio_req->pending_reqs);
+
+			req->r_callback = ceph_aio_complete_req;
+			req->r_inode = inode;
+			req->r_priv = aio_req;
+			list_add_tail(&req->r_private_item, &aio_req->osd_reqs);
+
+			pos += len;
+			continue;
+		}
+
+		ceph_osdc_start_request(req->r_osdc, req);
+		ret = ceph_osdc_wait_request(&fsc->client->osdc, req);
+
+		if (write)
+			ceph_update_write_metrics(metric, req->r_start_latency,
+						  req->r_end_latency, len, ret);
+		else
+			ceph_update_read_metrics(metric, req->r_start_latency,
+						 req->r_end_latency, len, ret);
+
+		size = i_size_read(inode);
+		if (!write) {
+			if (ret == -ENOENT)
+				ret = 0;
+			if (ret >= 0 && ret < len && pos + ret < size) {
+				struct iov_iter i;
+				int zlen = min_t(size_t, len - ret,
+						 size - pos - ret);
+
+				iov_iter_bvec(&i, ITER_DEST, bvecs, num_pages, len);
+				iov_iter_advance(&i, ret);
+				iov_iter_zero(zlen, &i);
+				ret += zlen;
+			}
+			if (ret >= 0)
+				len = ret;
+		}
+
+		put_bvecs(bvecs, num_pages, should_dirty);
+		ceph_osdc_put_request(req);
+		if (ret < 0)
+			break;
+
+		pos += len;
+		if (!write && pos >= size)
+			break;
+
+		if (write && pos > size) {
+			if (ceph_inode_set_size(inode, pos))
+				ceph_check_caps(ceph_inode(inode),
+						CHECK_CAPS_AUTHONLY,
+						NULL);
+		}
+	}
+
+	if (aio_req) {
+		LIST_HEAD(osd_reqs);
+
+		if (aio_req->num_reqs == 0) {
+			kfree(aio_req);
+			return ret;
+		}
+
+		ceph_get_cap_refs(ci, write ? CEPH_CAP_FILE_WR :
+					      CEPH_CAP_FILE_RD);
+
+		list_splice(&aio_req->osd_reqs, &osd_reqs);
+		inode_dio_begin(inode);
+		while (!list_empty(&osd_reqs)) {
+			req = list_first_entry(&osd_reqs,
+					       struct ceph_osd_request,
+					       r_private_item);
+			list_del_init(&req->r_private_item);
+			if (ret >= 0)
+				ceph_osdc_start_request(req->r_osdc, req);
+			if (ret < 0) {
+				req->r_result = ret;
+				ceph_aio_complete_req(req);
+			}
+		}
+		return -EIOCBQUEUED;
+	}
+
+	if (ret != -EOLDSNAPC && pos > iocb->ki_pos) {
+		ret = pos - iocb->ki_pos;
+		iocb->ki_pos = pos;
+	}
+	return ret;
+}
+
+/*
+ * Synchronous write, straight from __user pointer or user pages.
+ *
+ * If write spans object boundary, just do multiple writes.  (For a
+ * correct atomic write, we should e.g. take write locks on all
+ * objects, rollback on failure, etc.)
+ */
+static ssize_t
+ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos,
+		struct ceph_snap_context *snapc)
+{
+	struct file *file = iocb->ki_filp;
+	struct inode *inode = file_inode(file);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	struct ceph_vino vino;
+	struct ceph_osd_request *req;
+	struct page **pages;
+	u64 len;
+	int num_pages;
+	int written = 0;
+	int flags;
+	int ret;
+	bool check_caps = false;
+	struct timespec64 mtime = current_time(inode);
+	size_t count = iov_iter_count(from);
+
+	if (ceph_snap(file_inode(file)) != CEPH_NOSNAP)
+		return -EROFS;
+
+	dout("sync_write on file %p %lld~%u snapc %p seq %lld\n",
+	     file, pos, (unsigned)count, snapc, snapc->seq);
+
+	ret = filemap_write_and_wait_range(inode->i_mapping,
+					   pos, pos + count - 1);
+	if (ret < 0)
+		return ret;
+
+	ceph_fscache_invalidate(inode, false);
+	ret = invalidate_inode_pages2_range(inode->i_mapping,
+					    pos >> PAGE_SHIFT,
+					    (pos + count - 1) >> PAGE_SHIFT);
+	if (ret < 0)
+		dout("invalidate_inode_pages2_range returned %d\n", ret);
+
+	flags = /* CEPH_OSD_FLAG_ORDERSNAP | */ CEPH_OSD_FLAG_WRITE;
+
+	while ((len = iov_iter_count(from)) > 0) {
+		size_t left;
+		int n;
+
+		vino = ceph_vino(inode);
+		req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
+					    vino, pos, &len, 0, 1,
+					    CEPH_OSD_OP_WRITE, flags, snapc,
+					    ci->i_truncate_seq,
+					    ci->i_truncate_size,
+					    false);
+		if (IS_ERR(req)) {
+			ret = PTR_ERR(req);
+			break;
+		}
+
+		/*
+		 * write from beginning of first page,
+		 * regardless of io alignment
+		 */
+		num_pages = (len + PAGE_SIZE - 1) >> PAGE_SHIFT;
+
+		pages = ceph_alloc_page_vector(num_pages, GFP_KERNEL);
+		if (IS_ERR(pages)) {
+			ret = PTR_ERR(pages);
+			goto out;
+		}
+
+		left = len;
+		for (n = 0; n < num_pages; n++) {
+			size_t plen = min_t(size_t, left, PAGE_SIZE);
+			ret = copy_page_from_iter(pages[n], 0, plen, from);
+			if (ret != plen) {
+				ret = -EFAULT;
+				break;
+			}
+			left -= ret;
+		}
+
+		if (ret < 0) {
+			ceph_release_page_vector(pages, num_pages);
+			goto out;
+		}
+
+		req->r_inode = inode;
+
+		osd_req_op_extent_osd_data_pages(req, 0, pages, len, 0,
+						false, true);
+
+		req->r_mtime = mtime;
+		ceph_osdc_start_request(&fsc->client->osdc, req);
+		ret = ceph_osdc_wait_request(&fsc->client->osdc, req);
+
+		ceph_update_write_metrics(&fsc->mdsc->metric, req->r_start_latency,
+					  req->r_end_latency, len, ret);
+out:
+		ceph_osdc_put_request(req);
+		if (ret != 0) {
+			ceph_set_error_write(ci);
+			break;
+		}
+
+		ceph_clear_error_write(ci);
+		pos += len;
+		written += len;
+		if (pos > i_size_read(inode)) {
+			check_caps = ceph_inode_set_size(inode, pos);
+			if (check_caps)
+				ceph_check_caps(ceph_inode(inode),
+						CHECK_CAPS_AUTHONLY,
+						NULL);
+		}
+
+	}
+
+	if (ret != -EOLDSNAPC && written > 0) {
+		ret = written;
+		iocb->ki_pos = pos;
+	}
+	return ret;
+}
+
+/*
+ * Wrap generic_file_aio_read with checks for cap bits on the inode.
+ * Atomically grab references, so that those bits are not released
+ * back to the MDS mid-read.
+ *
+ * Hmm, the sync read case isn't actually async... should it be?
+ */
+static ssize_t ceph_read_iter(struct kiocb *iocb, struct iov_iter *to)
+{
+	struct file *filp = iocb->ki_filp;
+	struct ceph_file_info *fi = filp->private_data;
+	size_t len = iov_iter_count(to);
+	struct inode *inode = file_inode(filp);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	bool direct_lock = iocb->ki_flags & IOCB_DIRECT;
+	ssize_t ret;
+	int want = 0, got = 0;
+	int retry_op = 0, read = 0;
+
+again:
+	dout("aio_read %p %llx.%llx %llu~%u trying to get caps on %p\n",
+	     inode, ceph_vinop(inode), iocb->ki_pos, (unsigned)len, inode);
+
+	if (ceph_inode_is_shutdown(inode))
+		return -ESTALE;
+
+	if (direct_lock)
+		ceph_start_io_direct(inode);
+	else
+		ceph_start_io_read(inode);
+
+	if (!(fi->flags & CEPH_F_SYNC) && !direct_lock)
+		want |= CEPH_CAP_FILE_CACHE;
+	if (fi->fmode & CEPH_FILE_MODE_LAZY)
+		want |= CEPH_CAP_FILE_LAZYIO;
+
+	ret = ceph_get_caps(filp, CEPH_CAP_FILE_RD, want, -1, &got);
+	if (ret < 0) {
+		if (direct_lock)
+			ceph_end_io_direct(inode);
+		else
+			ceph_end_io_read(inode);
+		return ret;
+	}
+
+	if ((got & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) == 0 ||
+	    (iocb->ki_flags & IOCB_DIRECT) ||
+	    (fi->flags & CEPH_F_SYNC)) {
+
+		dout("aio_sync_read %p %llx.%llx %llu~%u got cap refs on %s\n",
+		     inode, ceph_vinop(inode), iocb->ki_pos, (unsigned)len,
+		     ceph_cap_string(got));
+
+		if (!ceph_has_inline_data(ci)) {
+			if (!retry_op && (iocb->ki_flags & IOCB_DIRECT)) {
+				ret = ceph_direct_read_write(iocb, to,
+							     NULL, NULL);
+				if (ret >= 0 && ret < len)
+					retry_op = CHECK_EOF;
+			} else {
+				ret = ceph_sync_read(iocb, to, &retry_op);
+			}
+		} else {
+			retry_op = READ_INLINE;
+		}
+	} else {
+		CEPH_DEFINE_RW_CONTEXT(rw_ctx, got);
+		dout("aio_read %p %llx.%llx %llu~%u got cap refs on %s\n",
+		     inode, ceph_vinop(inode), iocb->ki_pos, (unsigned)len,
+		     ceph_cap_string(got));
+		ceph_add_rw_context(fi, &rw_ctx);
+		ret = generic_file_read_iter(iocb, to);
+		ceph_del_rw_context(fi, &rw_ctx);
+	}
+
+	dout("aio_read %p %llx.%llx dropping cap refs on %s = %d\n",
+	     inode, ceph_vinop(inode), ceph_cap_string(got), (int)ret);
+	ceph_put_cap_refs(ci, got);
+
+	if (direct_lock)
+		ceph_end_io_direct(inode);
+	else
+		ceph_end_io_read(inode);
+
+	if (retry_op > HAVE_RETRIED && ret >= 0) {
+		int statret;
+		struct page *page = NULL;
+		loff_t i_size;
+		if (retry_op == READ_INLINE) {
+			page = __page_cache_alloc(GFP_KERNEL);
+			if (!page)
+				return -ENOMEM;
+		}
+
+		statret = __ceph_do_getattr(inode, page,
+					    CEPH_STAT_CAP_INLINE_DATA, !!page);
+		if (statret < 0) {
+			if (page)
+				__free_page(page);
+			if (statret == -ENODATA) {
+				BUG_ON(retry_op != READ_INLINE);
+				goto again;
+			}
+			return statret;
+		}
+
+		i_size = i_size_read(inode);
+		if (retry_op == READ_INLINE) {
+			BUG_ON(ret > 0 || read > 0);
+			if (iocb->ki_pos < i_size &&
+			    iocb->ki_pos < PAGE_SIZE) {
+				loff_t end = min_t(loff_t, i_size,
+						   iocb->ki_pos + len);
+				end = min_t(loff_t, end, PAGE_SIZE);
+				if (statret < end)
+					zero_user_segment(page, statret, end);
+				ret = copy_page_to_iter(page,
+						iocb->ki_pos & ~PAGE_MASK,
+						end - iocb->ki_pos, to);
+				iocb->ki_pos += ret;
+				read += ret;
+			}
+			if (iocb->ki_pos < i_size && read < len) {
+				size_t zlen = min_t(size_t, len - read,
+						    i_size - iocb->ki_pos);
+				ret = iov_iter_zero(zlen, to);
+				iocb->ki_pos += ret;
+				read += ret;
+			}
+			__free_pages(page, 0);
+			return read;
+		}
+
+		/* hit EOF or hole? */
+		if (retry_op == CHECK_EOF && iocb->ki_pos < i_size &&
+		    ret < len) {
+			dout("sync_read hit hole, ppos %lld < size %lld"
+			     ", reading more\n", iocb->ki_pos, i_size);
+
+			read += ret;
+			len -= ret;
+			retry_op = HAVE_RETRIED;
+			goto again;
+		}
+	}
+
+	if (ret >= 0)
+		ret += read;
+
+	return ret;
+}
+
+/*
+ * Take cap references to avoid releasing caps to MDS mid-write.
+ *
+ * If we are synchronous, and write with an old snap context, the OSD
+ * may return EOLDSNAPC.  In that case, retry the write.. _after_
+ * dropping our cap refs and allowing the pending snap to logically
+ * complete _before_ this write occurs.
+ *
+ * If we are near ENOSPC, write synchronously.
+ */
+static ssize_t ceph_write_iter(struct kiocb *iocb, struct iov_iter *from)
+{
+	struct file *file = iocb->ki_filp;
+	struct ceph_file_info *fi = file->private_data;
+	struct inode *inode = file_inode(file);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	struct ceph_osd_client *osdc = &fsc->client->osdc;
+	struct ceph_cap_flush *prealloc_cf;
+	ssize_t count, written = 0;
+	int err, want = 0, got;
+	bool direct_lock = false;
+	u32 map_flags;
+	u64 pool_flags;
+	loff_t pos;
+	loff_t limit = max(i_size_read(inode), fsc->max_file_size);
+
+	if (ceph_inode_is_shutdown(inode))
+		return -ESTALE;
+
+	if (ceph_snap(inode) != CEPH_NOSNAP)
+		return -EROFS;
+
+	prealloc_cf = ceph_alloc_cap_flush();
+	if (!prealloc_cf)
+		return -ENOMEM;
+
+	if ((iocb->ki_flags & (IOCB_DIRECT | IOCB_APPEND)) == IOCB_DIRECT)
+		direct_lock = true;
+
+retry_snap:
+	if (direct_lock)
+		ceph_start_io_direct(inode);
+	else
+		ceph_start_io_write(inode);
+
+	/* We can write back this queue in page reclaim */
+	current->backing_dev_info = inode_to_bdi(inode);
+
+	if (iocb->ki_flags & IOCB_APPEND) {
+		err = ceph_do_getattr(inode, CEPH_STAT_CAP_SIZE, false);
+		if (err < 0)
+			goto out;
+	}
+
+	err = generic_write_checks(iocb, from);
+	if (err <= 0)
+		goto out;
+
+	pos = iocb->ki_pos;
+	if (unlikely(pos >= limit)) {
+		err = -EFBIG;
+		goto out;
+	} else {
+		iov_iter_truncate(from, limit - pos);
+	}
+
+	count = iov_iter_count(from);
+	if (ceph_quota_is_max_bytes_exceeded(inode, pos + count)) {
+		err = -EDQUOT;
+		goto out;
+	}
+
+	down_read(&osdc->lock);
+	map_flags = osdc->osdmap->flags;
+	pool_flags = ceph_pg_pool_flags(osdc->osdmap, ci->i_layout.pool_id);
+	up_read(&osdc->lock);
+	if ((map_flags & CEPH_OSDMAP_FULL) ||
+	    (pool_flags & CEPH_POOL_FLAG_FULL)) {
+		err = -ENOSPC;
+		goto out;
+	}
+
+	err = file_remove_privs(file);
+	if (err)
+		goto out;
+
+	dout("aio_write %p %llx.%llx %llu~%zd getting caps. i_size %llu\n",
+	     inode, ceph_vinop(inode), pos, count, i_size_read(inode));
+	if (!(fi->flags & CEPH_F_SYNC) && !direct_lock)
+		want |= CEPH_CAP_FILE_BUFFER;
+	if (fi->fmode & CEPH_FILE_MODE_LAZY)
+		want |= CEPH_CAP_FILE_LAZYIO;
+	got = 0;
+	err = ceph_get_caps(file, CEPH_CAP_FILE_WR, want, pos + count, &got);
+	if (err < 0)
+		goto out;
+
+	err = file_update_time(file);
+	if (err)
+		goto out_caps;
+
+	inode_inc_iversion_raw(inode);
+
+	dout("aio_write %p %llx.%llx %llu~%zd got cap refs on %s\n",
+	     inode, ceph_vinop(inode), pos, count, ceph_cap_string(got));
+
+	if ((got & (CEPH_CAP_FILE_BUFFER|CEPH_CAP_FILE_LAZYIO)) == 0 ||
+	    (iocb->ki_flags & IOCB_DIRECT) || (fi->flags & CEPH_F_SYNC) ||
+	    (ci->i_ceph_flags & CEPH_I_ERROR_WRITE)) {
+		struct ceph_snap_context *snapc;
+		struct iov_iter data;
+
+		spin_lock(&ci->i_ceph_lock);
+		if (__ceph_have_pending_cap_snap(ci)) {
+			struct ceph_cap_snap *capsnap =
+					list_last_entry(&ci->i_cap_snaps,
+							struct ceph_cap_snap,
+							ci_item);
+			snapc = ceph_get_snap_context(capsnap->context);
+		} else {
+			BUG_ON(!ci->i_head_snapc);
+			snapc = ceph_get_snap_context(ci->i_head_snapc);
+		}
+		spin_unlock(&ci->i_ceph_lock);
+
+		/* we might need to revert back to that point */
+		data = *from;
+		if (iocb->ki_flags & IOCB_DIRECT)
+			written = ceph_direct_read_write(iocb, &data, snapc,
+							 &prealloc_cf);
+		else
+			written = ceph_sync_write(iocb, &data, pos, snapc);
+		if (direct_lock)
+			ceph_end_io_direct(inode);
+		else
+			ceph_end_io_write(inode);
+		if (written > 0)
+			iov_iter_advance(from, written);
+		ceph_put_snap_context(snapc);
+	} else {
+		/*
+		 * No need to acquire the i_truncate_mutex. Because
+		 * the MDS revokes Fwb caps before sending truncate
+		 * message to us. We can't get Fwb cap while there
+		 * are pending vmtruncate. So write and vmtruncate
+		 * can not run at the same time
+		 */
+		written = generic_perform_write(iocb, from);
+		if (likely(written >= 0))
+			iocb->ki_pos = pos + written;
+		ceph_end_io_write(inode);
+	}
+
+	if (written >= 0) {
+		int dirty;
+
+		spin_lock(&ci->i_ceph_lock);
+		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
+					       &prealloc_cf);
+		spin_unlock(&ci->i_ceph_lock);
+		if (dirty)
+			__mark_inode_dirty(inode, dirty);
+		if (ceph_quota_is_max_bytes_approaching(inode, iocb->ki_pos))
+			ceph_check_caps(ci, CHECK_CAPS_FLUSH, NULL);
+	}
+
+	dout("aio_write %p %llx.%llx %llu~%u  dropping cap refs on %s\n",
+	     inode, ceph_vinop(inode), pos, (unsigned)count,
+	     ceph_cap_string(got));
+	ceph_put_cap_refs(ci, got);
+
+	if (written == -EOLDSNAPC) {
+		dout("aio_write %p %llx.%llx %llu~%u" "got EOLDSNAPC, retrying\n",
+		     inode, ceph_vinop(inode), pos, (unsigned)count);
+		goto retry_snap;
+	}
+
+	if (written >= 0) {
+		if ((map_flags & CEPH_OSDMAP_NEARFULL) ||
+		    (pool_flags & CEPH_POOL_FLAG_NEARFULL))
+			iocb->ki_flags |= IOCB_DSYNC;
+		written = generic_write_sync(iocb, written);
+	}
+
+	goto out_unlocked;
+out_caps:
+	ceph_put_cap_refs(ci, got);
+out:
+	if (direct_lock)
+		ceph_end_io_direct(inode);
+	else
+		ceph_end_io_write(inode);
+out_unlocked:
+	ceph_free_cap_flush(prealloc_cf);
+	current->backing_dev_info = NULL;
+	return written ? written : err;
+}
+
+/*
+ * llseek.  be sure to verify file size on SEEK_END.
+ */
+static loff_t ceph_llseek(struct file *file, loff_t offset, int whence)
+{
+	if (whence == SEEK_END || whence == SEEK_DATA || whence == SEEK_HOLE) {
+		struct inode *inode = file_inode(file);
+		int ret;
+
+		ret = ceph_do_getattr(inode, CEPH_STAT_CAP_SIZE, false);
+		if (ret < 0)
+			return ret;
+	}
+	return generic_file_llseek(file, offset, whence);
+}
+
+static inline void ceph_zero_partial_page(
+	struct inode *inode, loff_t offset, unsigned size)
+{
+	struct page *page;
+	pgoff_t index = offset >> PAGE_SHIFT;
+
+	page = find_lock_page(inode->i_mapping, index);
+	if (page) {
+		wait_on_page_writeback(page);
+		zero_user(page, offset & (PAGE_SIZE - 1), size);
+		unlock_page(page);
+		put_page(page);
+	}
+}
+
+static void ceph_zero_pagecache_range(struct inode *inode, loff_t offset,
+				      loff_t length)
+{
+	loff_t nearly = round_up(offset, PAGE_SIZE);
+	if (offset < nearly) {
+		loff_t size = nearly - offset;
+		if (length < size)
+			size = length;
+		ceph_zero_partial_page(inode, offset, size);
+		offset += size;
+		length -= size;
+	}
+	if (length >= PAGE_SIZE) {
+		loff_t size = round_down(length, PAGE_SIZE);
+		truncate_pagecache_range(inode, offset, offset + size - 1);
+		offset += size;
+		length -= size;
+	}
+	if (length)
+		ceph_zero_partial_page(inode, offset, length);
+}
+
+static int ceph_zero_partial_object(struct inode *inode,
+				    loff_t offset, loff_t *length)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	struct ceph_osd_request *req;
+	int ret = 0;
+	loff_t zero = 0;
+	int op;
+
+	if (ceph_inode_is_shutdown(inode))
+		return -EIO;
+
+	if (!length) {
+		op = offset ? CEPH_OSD_OP_DELETE : CEPH_OSD_OP_TRUNCATE;
+		length = &zero;
+	} else {
+		op = CEPH_OSD_OP_ZERO;
+	}
+
+	req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
+					ceph_vino(inode),
+					offset, length,
+					0, 1, op,
+					CEPH_OSD_FLAG_WRITE,
+					NULL, 0, 0, false);
+	if (IS_ERR(req)) {
+		ret = PTR_ERR(req);
+		goto out;
+	}
+
+	req->r_mtime = inode->i_mtime;
+	ceph_osdc_start_request(&fsc->client->osdc, req);
+	ret = ceph_osdc_wait_request(&fsc->client->osdc, req);
+	if (ret == -ENOENT)
+		ret = 0;
+	ceph_osdc_put_request(req);
+
+out:
+	return ret;
+}
+
+static int ceph_zero_objects(struct inode *inode, loff_t offset, loff_t length)
+{
+	int ret = 0;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	s32 stripe_unit = ci->i_layout.stripe_unit;
+	s32 stripe_count = ci->i_layout.stripe_count;
+	s32 object_size = ci->i_layout.object_size;
+	u64 object_set_size = object_size * stripe_count;
+	u64 nearly, t;
+
+	/* round offset up to next period boundary */
+	nearly = offset + object_set_size - 1;
+	t = nearly;
+	nearly -= do_div(t, object_set_size);
+
+	while (length && offset < nearly) {
+		loff_t size = length;
+		ret = ceph_zero_partial_object(inode, offset, &size);
+		if (ret < 0)
+			return ret;
+		offset += size;
+		length -= size;
+	}
+	while (length >= object_set_size) {
+		int i;
+		loff_t pos = offset;
+		for (i = 0; i < stripe_count; ++i) {
+			ret = ceph_zero_partial_object(inode, pos, NULL);
+			if (ret < 0)
+				return ret;
+			pos += stripe_unit;
+		}
+		offset += object_set_size;
+		length -= object_set_size;
+	}
+	while (length) {
+		loff_t size = length;
+		ret = ceph_zero_partial_object(inode, offset, &size);
+		if (ret < 0)
+			return ret;
+		offset += size;
+		length -= size;
+	}
+	return ret;
+}
+
+static long ceph_fallocate(struct file *file, int mode,
+				loff_t offset, loff_t length)
+{
+	struct ceph_file_info *fi = file->private_data;
+	struct inode *inode = file_inode(file);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_cap_flush *prealloc_cf;
+	int want, got = 0;
+	int dirty;
+	int ret = 0;
+	loff_t endoff = 0;
+	loff_t size;
+
+	dout("%s %p %llx.%llx mode %x, offset %llu length %llu\n", __func__,
+	     inode, ceph_vinop(inode), mode, offset, length);
+
+	if (mode != (FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE))
+		return -EOPNOTSUPP;
+
+	if (!S_ISREG(inode->i_mode))
+		return -EOPNOTSUPP;
+
+	prealloc_cf = ceph_alloc_cap_flush();
+	if (!prealloc_cf)
+		return -ENOMEM;
+
+	inode_lock(inode);
+
+	if (ceph_snap(inode) != CEPH_NOSNAP) {
+		ret = -EROFS;
+		goto unlock;
+	}
+
+	size = i_size_read(inode);
+
+	/* Are we punching a hole beyond EOF? */
+	if (offset >= size)
+		goto unlock;
+	if ((offset + length) > size)
+		length = size - offset;
+
+	if (fi->fmode & CEPH_FILE_MODE_LAZY)
+		want = CEPH_CAP_FILE_BUFFER | CEPH_CAP_FILE_LAZYIO;
+	else
+		want = CEPH_CAP_FILE_BUFFER;
+
+	ret = ceph_get_caps(file, CEPH_CAP_FILE_WR, want, endoff, &got);
+	if (ret < 0)
+		goto unlock;
+
+	ret = file_modified(file);
+	if (ret)
+		goto put_caps;
+
+	filemap_invalidate_lock(inode->i_mapping);
+	ceph_fscache_invalidate(inode, false);
+	ceph_zero_pagecache_range(inode, offset, length);
+	ret = ceph_zero_objects(inode, offset, length);
+
+	if (!ret) {
+		spin_lock(&ci->i_ceph_lock);
+		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
+					       &prealloc_cf);
+		spin_unlock(&ci->i_ceph_lock);
+		if (dirty)
+			__mark_inode_dirty(inode, dirty);
+	}
+	filemap_invalidate_unlock(inode->i_mapping);
+
+put_caps:
+	ceph_put_cap_refs(ci, got);
+unlock:
+	inode_unlock(inode);
+	ceph_free_cap_flush(prealloc_cf);
+	return ret;
+}
+
+/*
+ * This function tries to get FILE_WR capabilities for dst_ci and FILE_RD for
+ * src_ci.  Two attempts are made to obtain both caps, and an error is return if
+ * this fails; zero is returned on success.
+ */
+static int get_rd_wr_caps(struct file *src_filp, int *src_got,
+			  struct file *dst_filp,
+			  loff_t dst_endoff, int *dst_got)
+{
+	int ret = 0;
+	bool retrying = false;
+
+retry_caps:
+	ret = ceph_get_caps(dst_filp, CEPH_CAP_FILE_WR, CEPH_CAP_FILE_BUFFER,
+			    dst_endoff, dst_got);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * Since we're already holding the FILE_WR capability for the dst file,
+	 * we would risk a deadlock by using ceph_get_caps.  Thus, we'll do some
+	 * retry dance instead to try to get both capabilities.
+	 */
+	ret = ceph_try_get_caps(file_inode(src_filp),
+				CEPH_CAP_FILE_RD, CEPH_CAP_FILE_SHARED,
+				false, src_got);
+	if (ret <= 0) {
+		/* Start by dropping dst_ci caps and getting src_ci caps */
+		ceph_put_cap_refs(ceph_inode(file_inode(dst_filp)), *dst_got);
+		if (retrying) {
+			if (!ret)
+				/* ceph_try_get_caps masks EAGAIN */
+				ret = -EAGAIN;
+			return ret;
+		}
+		ret = ceph_get_caps(src_filp, CEPH_CAP_FILE_RD,
+				    CEPH_CAP_FILE_SHARED, -1, src_got);
+		if (ret < 0)
+			return ret;
+		/*... drop src_ci caps too, and retry */
+		ceph_put_cap_refs(ceph_inode(file_inode(src_filp)), *src_got);
+		retrying = true;
+		goto retry_caps;
+	}
+	return ret;
+}
+
+static void put_rd_wr_caps(struct ceph_inode_info *src_ci, int src_got,
+			   struct ceph_inode_info *dst_ci, int dst_got)
+{
+	ceph_put_cap_refs(src_ci, src_got);
+	ceph_put_cap_refs(dst_ci, dst_got);
+}
+
+/*
+ * This function does several size-related checks, returning an error if:
+ *  - source file is smaller than off+len
+ *  - destination file size is not OK (inode_newsize_ok())
+ *  - max bytes quotas is exceeded
+ */
+static int is_file_size_ok(struct inode *src_inode, struct inode *dst_inode,
+			   loff_t src_off, loff_t dst_off, size_t len)
+{
+	loff_t size, endoff;
+
+	size = i_size_read(src_inode);
+	/*
+	 * Don't copy beyond source file EOF.  Instead of simply setting length
+	 * to (size - src_off), just drop to VFS default implementation, as the
+	 * local i_size may be stale due to other clients writing to the source
+	 * inode.
+	 */
+	if (src_off + len > size) {
+		dout("Copy beyond EOF (%llu + %zu > %llu)\n",
+		     src_off, len, size);
+		return -EOPNOTSUPP;
+	}
+	size = i_size_read(dst_inode);
+
+	endoff = dst_off + len;
+	if (inode_newsize_ok(dst_inode, endoff))
+		return -EOPNOTSUPP;
+
+	if (ceph_quota_is_max_bytes_exceeded(dst_inode, endoff))
+		return -EDQUOT;
+
+	return 0;
+}
+
+static struct ceph_osd_request *
+ceph_alloc_copyfrom_request(struct ceph_osd_client *osdc,
+			    u64 src_snapid,
+			    struct ceph_object_id *src_oid,
+			    struct ceph_object_locator *src_oloc,
+			    struct ceph_object_id *dst_oid,
+			    struct ceph_object_locator *dst_oloc,
+			    u32 truncate_seq, u64 truncate_size)
+{
+	struct ceph_osd_request *req;
+	int ret;
+	u32 src_fadvise_flags =
+		CEPH_OSD_OP_FLAG_FADVISE_SEQUENTIAL |
+		CEPH_OSD_OP_FLAG_FADVISE_NOCACHE;
+	u32 dst_fadvise_flags =
+		CEPH_OSD_OP_FLAG_FADVISE_SEQUENTIAL |
+		CEPH_OSD_OP_FLAG_FADVISE_DONTNEED;
+
+	req = ceph_osdc_alloc_request(osdc, NULL, 1, false, GFP_KERNEL);
+	if (!req)
+		return ERR_PTR(-ENOMEM);
+
+	req->r_flags = CEPH_OSD_FLAG_WRITE;
+
+	ceph_oloc_copy(&req->r_t.base_oloc, dst_oloc);
+	ceph_oid_copy(&req->r_t.base_oid, dst_oid);
+
+	ret = osd_req_op_copy_from_init(req, src_snapid, 0,
+					src_oid, src_oloc,
+					src_fadvise_flags,
+					dst_fadvise_flags,
+					truncate_seq,
+					truncate_size,
+					CEPH_OSD_COPY_FROM_FLAG_TRUNCATE_SEQ);
+	if (ret)
+		goto out;
+
+	ret = ceph_osdc_alloc_messages(req, GFP_KERNEL);
+	if (ret)
+		goto out;
+
+	return req;
+
+out:
+	ceph_osdc_put_request(req);
+	return ERR_PTR(ret);
+}
+
+static ssize_t ceph_do_objects_copy(struct ceph_inode_info *src_ci, u64 *src_off,
+				    struct ceph_inode_info *dst_ci, u64 *dst_off,
+				    struct ceph_fs_client *fsc,
+				    size_t len, unsigned int flags)
+{
+	struct ceph_object_locator src_oloc, dst_oloc;
+	struct ceph_object_id src_oid, dst_oid;
+	struct ceph_osd_client *osdc;
+	struct ceph_osd_request *req;
+	size_t bytes = 0;
+	u64 src_objnum, src_objoff, dst_objnum, dst_objoff;
+	u32 src_objlen, dst_objlen;
+	u32 object_size = src_ci->i_layout.object_size;
+	int ret;
+
+	src_oloc.pool = src_ci->i_layout.pool_id;
+	src_oloc.pool_ns = ceph_try_get_string(src_ci->i_layout.pool_ns);
+	dst_oloc.pool = dst_ci->i_layout.pool_id;
+	dst_oloc.pool_ns = ceph_try_get_string(dst_ci->i_layout.pool_ns);
+	osdc = &fsc->client->osdc;
+
+	while (len >= object_size) {
+		ceph_calc_file_object_mapping(&src_ci->i_layout, *src_off,
+					      object_size, &src_objnum,
+					      &src_objoff, &src_objlen);
+		ceph_calc_file_object_mapping(&dst_ci->i_layout, *dst_off,
+					      object_size, &dst_objnum,
+					      &dst_objoff, &dst_objlen);
+		ceph_oid_init(&src_oid);
+		ceph_oid_printf(&src_oid, "%llx.%08llx",
+				src_ci->i_vino.ino, src_objnum);
+		ceph_oid_init(&dst_oid);
+		ceph_oid_printf(&dst_oid, "%llx.%08llx",
+				dst_ci->i_vino.ino, dst_objnum);
+		/* Do an object remote copy */
+		req = ceph_alloc_copyfrom_request(osdc, src_ci->i_vino.snap,
+						  &src_oid, &src_oloc,
+						  &dst_oid, &dst_oloc,
+						  dst_ci->i_truncate_seq,
+						  dst_ci->i_truncate_size);
+		if (IS_ERR(req))
+			ret = PTR_ERR(req);
+		else {
+			ceph_osdc_start_request(osdc, req);
+			ret = ceph_osdc_wait_request(osdc, req);
+			ceph_update_copyfrom_metrics(&fsc->mdsc->metric,
+						     req->r_start_latency,
+						     req->r_end_latency,
+						     object_size, ret);
+			ceph_osdc_put_request(req);
+		}
+		if (ret) {
+			if (ret == -EOPNOTSUPP) {
+				fsc->have_copy_from2 = false;
+				pr_notice("OSDs don't support copy-from2; disabling copy offload\n");
+			}
+			dout("ceph_osdc_copy_from returned %d\n", ret);
+			if (!bytes)
+				bytes = ret;
+			goto out;
+		}
+		len -= object_size;
+		bytes += object_size;
+		*src_off += object_size;
+		*dst_off += object_size;
+	}
+
+out:
+	ceph_oloc_destroy(&src_oloc);
+	ceph_oloc_destroy(&dst_oloc);
+	return bytes;
+}
+
+static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
+				      struct file *dst_file, loff_t dst_off,
+				      size_t len, unsigned int flags)
+{
+	struct inode *src_inode = file_inode(src_file);
+	struct inode *dst_inode = file_inode(dst_file);
+	struct ceph_inode_info *src_ci = ceph_inode(src_inode);
+	struct ceph_inode_info *dst_ci = ceph_inode(dst_inode);
+	struct ceph_cap_flush *prealloc_cf;
+	struct ceph_fs_client *src_fsc = ceph_inode_to_client(src_inode);
+	loff_t size;
+	ssize_t ret = -EIO, bytes;
+	u64 src_objnum, dst_objnum, src_objoff, dst_objoff;
+	u32 src_objlen, dst_objlen;
+	int src_got = 0, dst_got = 0, err, dirty;
+
+	if (src_inode->i_sb != dst_inode->i_sb) {
+		struct ceph_fs_client *dst_fsc = ceph_inode_to_client(dst_inode);
+
+		if (ceph_fsid_compare(&src_fsc->client->fsid,
+				      &dst_fsc->client->fsid)) {
+			dout("Copying files across clusters: src: %pU dst: %pU\n",
+			     &src_fsc->client->fsid, &dst_fsc->client->fsid);
+			return -EXDEV;
+		}
+	}
+	if (ceph_snap(dst_inode) != CEPH_NOSNAP)
+		return -EROFS;
+
+	/*
+	 * Some of the checks below will return -EOPNOTSUPP, which will force a
+	 * fallback to the default VFS copy_file_range implementation.  This is
+	 * desirable in several cases (for ex, the 'len' is smaller than the
+	 * size of the objects, or in cases where that would be more
+	 * efficient).
+	 */
+
+	if (ceph_test_mount_opt(src_fsc, NOCOPYFROM))
+		return -EOPNOTSUPP;
+
+	if (!src_fsc->have_copy_from2)
+		return -EOPNOTSUPP;
+
+	/*
+	 * Striped file layouts require that we copy partial objects, but the
+	 * OSD copy-from operation only supports full-object copies.  Limit
+	 * this to non-striped file layouts for now.
+	 */
+	if ((src_ci->i_layout.stripe_unit != dst_ci->i_layout.stripe_unit) ||
+	    (src_ci->i_layout.stripe_count != 1) ||
+	    (dst_ci->i_layout.stripe_count != 1) ||
+	    (src_ci->i_layout.object_size != dst_ci->i_layout.object_size)) {
+		dout("Invalid src/dst files layout\n");
+		return -EOPNOTSUPP;
+	}
+
+	if (len < src_ci->i_layout.object_size)
+		return -EOPNOTSUPP; /* no remote copy will be done */
+
+	prealloc_cf = ceph_alloc_cap_flush();
+	if (!prealloc_cf)
+		return -ENOMEM;
+
+	/* Start by sync'ing the source and destination files */
+	ret = file_write_and_wait_range(src_file, src_off, (src_off + len));
+	if (ret < 0) {
+		dout("failed to write src file (%zd)\n", ret);
+		goto out;
+	}
+	ret = file_write_and_wait_range(dst_file, dst_off, (dst_off + len));
+	if (ret < 0) {
+		dout("failed to write dst file (%zd)\n", ret);
+		goto out;
+	}
+
+	/*
+	 * We need FILE_WR caps for dst_ci and FILE_RD for src_ci as other
+	 * clients may have dirty data in their caches.  And OSDs know nothing
+	 * about caps, so they can't safely do the remote object copies.
+	 */
+	err = get_rd_wr_caps(src_file, &src_got,
+			     dst_file, (dst_off + len), &dst_got);
+	if (err < 0) {
+		dout("get_rd_wr_caps returned %d\n", err);
+		ret = -EOPNOTSUPP;
+		goto out;
+	}
+
+	ret = is_file_size_ok(src_inode, dst_inode, src_off, dst_off, len);
+	if (ret < 0)
+		goto out_caps;
+
+	/* Drop dst file cached pages */
+	ceph_fscache_invalidate(dst_inode, false);
+	ret = invalidate_inode_pages2_range(dst_inode->i_mapping,
+					    dst_off >> PAGE_SHIFT,
+					    (dst_off + len) >> PAGE_SHIFT);
+	if (ret < 0) {
+		dout("Failed to invalidate inode pages (%zd)\n", ret);
+		ret = 0; /* XXX */
+	}
+	ceph_calc_file_object_mapping(&src_ci->i_layout, src_off,
+				      src_ci->i_layout.object_size,
+				      &src_objnum, &src_objoff, &src_objlen);
+	ceph_calc_file_object_mapping(&dst_ci->i_layout, dst_off,
+				      dst_ci->i_layout.object_size,
+				      &dst_objnum, &dst_objoff, &dst_objlen);
+	/* object-level offsets need to the same */
+	if (src_objoff != dst_objoff) {
+		ret = -EOPNOTSUPP;
+		goto out_caps;
+	}
+
+	/*
+	 * Do a manual copy if the object offset isn't object aligned.
+	 * 'src_objlen' contains the bytes left until the end of the object,
+	 * starting at the src_off
+	 */
+	if (src_objoff) {
+		dout("Initial partial copy of %u bytes\n", src_objlen);
+
+		/*
+		 * we need to temporarily drop all caps as we'll be calling
+		 * {read,write}_iter, which will get caps again.
+		 */
+		put_rd_wr_caps(src_ci, src_got, dst_ci, dst_got);
+		ret = do_splice_direct(src_file, &src_off, dst_file,
+				       &dst_off, src_objlen, flags);
+		/* Abort on short copies or on error */
+		if (ret < (long)src_objlen) {
+			dout("Failed partial copy (%zd)\n", ret);
+			goto out;
+		}
+		len -= ret;
+		err = get_rd_wr_caps(src_file, &src_got,
+				     dst_file, (dst_off + len), &dst_got);
+		if (err < 0)
+			goto out;
+		err = is_file_size_ok(src_inode, dst_inode,
+				      src_off, dst_off, len);
+		if (err < 0)
+			goto out_caps;
+	}
+
+	size = i_size_read(dst_inode);
+	bytes = ceph_do_objects_copy(src_ci, &src_off, dst_ci, &dst_off,
+				     src_fsc, len, flags);
+	if (bytes <= 0) {
+		if (!ret)
+			ret = bytes;
+		goto out_caps;
+	}
+	dout("Copied %zu bytes out of %zu\n", bytes, len);
+	len -= bytes;
+	ret += bytes;
+
+	file_update_time(dst_file);
+	inode_inc_iversion_raw(dst_inode);
+
+	if (dst_off > size) {
+		/* Let the MDS know about dst file size change */
+		if (ceph_inode_set_size(dst_inode, dst_off) ||
+		    ceph_quota_is_max_bytes_approaching(dst_inode, dst_off))
+			ceph_check_caps(dst_ci, CHECK_CAPS_AUTHONLY | CHECK_CAPS_FLUSH,
+					NULL);
+	}
+	/* Mark Fw dirty */
+	spin_lock(&dst_ci->i_ceph_lock);
+	dirty = __ceph_mark_dirty_caps(dst_ci, CEPH_CAP_FILE_WR, &prealloc_cf);
+	spin_unlock(&dst_ci->i_ceph_lock);
+	if (dirty)
+		__mark_inode_dirty(dst_inode, dirty);
+
+out_caps:
+	put_rd_wr_caps(src_ci, src_got, dst_ci, dst_got);
+
+	/*
+	 * Do the final manual copy if we still have some bytes left, unless
+	 * there were errors in remote object copies (len >= object_size).
+	 */
+	if (len && (len < src_ci->i_layout.object_size)) {
+		dout("Final partial copy of %zu bytes\n", len);
+		bytes = do_splice_direct(src_file, &src_off, dst_file,
+					 &dst_off, len, flags);
+		if (bytes > 0)
+			ret += bytes;
+		else
+			dout("Failed partial copy (%zd)\n", bytes);
+	}
+
+out:
+	ceph_free_cap_flush(prealloc_cf);
+
+	return ret;
+}
+
+static ssize_t ceph_copy_file_range(struct file *src_file, loff_t src_off,
+				    struct file *dst_file, loff_t dst_off,
+				    size_t len, unsigned int flags)
+{
+	ssize_t ret;
+
+	ret = __ceph_copy_file_range(src_file, src_off, dst_file, dst_off,
+				     len, flags);
+
+	if (ret == -EOPNOTSUPP || ret == -EXDEV)
+		ret = generic_copy_file_range(src_file, src_off, dst_file,
+					      dst_off, len, flags);
+	return ret;
+}
+
+const struct file_operations ceph_file_fops = {
+	.open = ceph_open,
+	.release = ceph_release,
+	.llseek = ceph_llseek,
+	.read_iter = ceph_read_iter,
+	.write_iter = ceph_write_iter,
+	.mmap = ceph_mmap,
+	.fsync = ceph_fsync,
+	.lock = ceph_lock,
+	.setlease = simple_nosetlease,
+	.flock = ceph_flock,
+	.splice_read = generic_file_splice_read,
+	.splice_write = iter_file_splice_write,
+	.unlocked_ioctl = ceph_ioctl,
+	.compat_ioctl = compat_ptr_ioctl,
+	.fallocate	= ceph_fallocate,
+	.copy_file_range = ceph_copy_file_range,
+};
diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
new file mode 100644
index 000000000..29384ec1a
--- /dev/null
+++ b/fs/ceph/inode.c
@@ -0,0 +1,2546 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/ceph/ceph_debug.h>
+
+#include <linux/module.h>
+#include <linux/fs.h>
+#include <linux/slab.h>
+#include <linux/string.h>
+#include <linux/uaccess.h>
+#include <linux/kernel.h>
+#include <linux/writeback.h>
+#include <linux/vmalloc.h>
+#include <linux/xattr.h>
+#include <linux/posix_acl.h>
+#include <linux/random.h>
+#include <linux/sort.h>
+#include <linux/iversion.h>
+
+#include "super.h"
+#include "mds_client.h"
+#include "cache.h"
+#include <linux/ceph/decode.h>
+
+/*
+ * Ceph inode operations
+ *
+ * Implement basic inode helpers (get, alloc) and inode ops (getattr,
+ * setattr, etc.), xattr helpers, and helpers for assimilating
+ * metadata returned by the MDS into our cache.
+ *
+ * Also define helpers for doing asynchronous writeback, invalidation,
+ * and truncation for the benefit of those who can't afford to block
+ * (typically because they are in the message handler path).
+ */
+
+static const struct inode_operations ceph_symlink_iops;
+
+static void ceph_inode_work(struct work_struct *work);
+
+/*
+ * find or create an inode, given the ceph ino number
+ */
+static int ceph_set_ino_cb(struct inode *inode, void *data)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
+
+	ci->i_vino = *(struct ceph_vino *)data;
+	inode->i_ino = ceph_vino_to_ino_t(ci->i_vino);
+	inode_set_iversion_raw(inode, 0);
+	percpu_counter_inc(&mdsc->metric.total_inodes);
+
+	return 0;
+}
+
+struct inode *ceph_get_inode(struct super_block *sb, struct ceph_vino vino)
+{
+	struct inode *inode;
+
+	if (ceph_vino_is_reserved(vino))
+		return ERR_PTR(-EREMOTEIO);
+
+	inode = iget5_locked(sb, (unsigned long)vino.ino, ceph_ino_compare,
+			     ceph_set_ino_cb, &vino);
+	if (!inode)
+		return ERR_PTR(-ENOMEM);
+
+	dout("get_inode on %llu=%llx.%llx got %p new %d\n", ceph_present_inode(inode),
+	     ceph_vinop(inode), inode, !!(inode->i_state & I_NEW));
+	return inode;
+}
+
+/*
+ * get/constuct snapdir inode for a given directory
+ */
+struct inode *ceph_get_snapdir(struct inode *parent)
+{
+	struct ceph_vino vino = {
+		.ino = ceph_ino(parent),
+		.snap = CEPH_SNAPDIR,
+	};
+	struct inode *inode = ceph_get_inode(parent->i_sb, vino);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+
+	if (IS_ERR(inode))
+		return inode;
+
+	if (!S_ISDIR(parent->i_mode)) {
+		pr_warn_once("bad snapdir parent type (mode=0%o)\n",
+			     parent->i_mode);
+		goto err;
+	}
+
+	if (!(inode->i_state & I_NEW) && !S_ISDIR(inode->i_mode)) {
+		pr_warn_once("bad snapdir inode type (mode=0%o)\n",
+			     inode->i_mode);
+		goto err;
+	}
+
+	inode->i_mode = parent->i_mode;
+	inode->i_uid = parent->i_uid;
+	inode->i_gid = parent->i_gid;
+	inode->i_mtime = parent->i_mtime;
+	inode->i_ctime = parent->i_ctime;
+	inode->i_atime = parent->i_atime;
+	ci->i_rbytes = 0;
+	ci->i_btime = ceph_inode(parent)->i_btime;
+
+	if (inode->i_state & I_NEW) {
+		inode->i_op = &ceph_snapdir_iops;
+		inode->i_fop = &ceph_snapdir_fops;
+		ci->i_snap_caps = CEPH_CAP_PIN; /* so we can open */
+		unlock_new_inode(inode);
+	}
+
+	return inode;
+err:
+	if ((inode->i_state & I_NEW))
+		discard_new_inode(inode);
+	else
+		iput(inode);
+	return ERR_PTR(-ENOTDIR);
+}
+
+const struct inode_operations ceph_file_iops = {
+	.permission = ceph_permission,
+	.setattr = ceph_setattr,
+	.getattr = ceph_getattr,
+	.listxattr = ceph_listxattr,
+	.get_acl = ceph_get_acl,
+	.set_acl = ceph_set_acl,
+};
+
+
+/*
+ * We use a 'frag tree' to keep track of the MDS's directory fragments
+ * for a given inode (usually there is just a single fragment).  We
+ * need to know when a child frag is delegated to a new MDS, or when
+ * it is flagged as replicated, so we can direct our requests
+ * accordingly.
+ */
+
+/*
+ * find/create a frag in the tree
+ */
+static struct ceph_inode_frag *__get_or_create_frag(struct ceph_inode_info *ci,
+						    u32 f)
+{
+	struct rb_node **p;
+	struct rb_node *parent = NULL;
+	struct ceph_inode_frag *frag;
+	int c;
+
+	p = &ci->i_fragtree.rb_node;
+	while (*p) {
+		parent = *p;
+		frag = rb_entry(parent, struct ceph_inode_frag, node);
+		c = ceph_frag_compare(f, frag->frag);
+		if (c < 0)
+			p = &(*p)->rb_left;
+		else if (c > 0)
+			p = &(*p)->rb_right;
+		else
+			return frag;
+	}
+
+	frag = kmalloc(sizeof(*frag), GFP_NOFS);
+	if (!frag)
+		return ERR_PTR(-ENOMEM);
+
+	frag->frag = f;
+	frag->split_by = 0;
+	frag->mds = -1;
+	frag->ndist = 0;
+
+	rb_link_node(&frag->node, parent, p);
+	rb_insert_color(&frag->node, &ci->i_fragtree);
+
+	dout("get_or_create_frag added %llx.%llx frag %x\n",
+	     ceph_vinop(&ci->netfs.inode), f);
+	return frag;
+}
+
+/*
+ * find a specific frag @f
+ */
+struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, u32 f)
+{
+	struct rb_node *n = ci->i_fragtree.rb_node;
+
+	while (n) {
+		struct ceph_inode_frag *frag =
+			rb_entry(n, struct ceph_inode_frag, node);
+		int c = ceph_frag_compare(f, frag->frag);
+		if (c < 0)
+			n = n->rb_left;
+		else if (c > 0)
+			n = n->rb_right;
+		else
+			return frag;
+	}
+	return NULL;
+}
+
+/*
+ * Choose frag containing the given value @v.  If @pfrag is
+ * specified, copy the frag delegation info to the caller if
+ * it is present.
+ */
+static u32 __ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
+			      struct ceph_inode_frag *pfrag, int *found)
+{
+	u32 t = ceph_frag_make(0, 0);
+	struct ceph_inode_frag *frag;
+	unsigned nway, i;
+	u32 n;
+
+	if (found)
+		*found = 0;
+
+	while (1) {
+		WARN_ON(!ceph_frag_contains_value(t, v));
+		frag = __ceph_find_frag(ci, t);
+		if (!frag)
+			break; /* t is a leaf */
+		if (frag->split_by == 0) {
+			if (pfrag)
+				memcpy(pfrag, frag, sizeof(*pfrag));
+			if (found)
+				*found = 1;
+			break;
+		}
+
+		/* choose child */
+		nway = 1 << frag->split_by;
+		dout("choose_frag(%x) %x splits by %d (%d ways)\n", v, t,
+		     frag->split_by, nway);
+		for (i = 0; i < nway; i++) {
+			n = ceph_frag_make_child(t, frag->split_by, i);
+			if (ceph_frag_contains_value(n, v)) {
+				t = n;
+				break;
+			}
+		}
+		BUG_ON(i == nway);
+	}
+	dout("choose_frag(%x) = %x\n", v, t);
+
+	return t;
+}
+
+u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
+		     struct ceph_inode_frag *pfrag, int *found)
+{
+	u32 ret;
+	mutex_lock(&ci->i_fragtree_mutex);
+	ret = __ceph_choose_frag(ci, v, pfrag, found);
+	mutex_unlock(&ci->i_fragtree_mutex);
+	return ret;
+}
+
+/*
+ * Process dirfrag (delegation) info from the mds.  Include leaf
+ * fragment in tree ONLY if ndist > 0.  Otherwise, only
+ * branches/splits are included in i_fragtree)
+ */
+static int ceph_fill_dirfrag(struct inode *inode,
+			     struct ceph_mds_reply_dirfrag *dirinfo)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_inode_frag *frag;
+	u32 id = le32_to_cpu(dirinfo->frag);
+	int mds = le32_to_cpu(dirinfo->auth);
+	int ndist = le32_to_cpu(dirinfo->ndist);
+	int diri_auth = -1;
+	int i;
+	int err = 0;
+
+	spin_lock(&ci->i_ceph_lock);
+	if (ci->i_auth_cap)
+		diri_auth = ci->i_auth_cap->mds;
+	spin_unlock(&ci->i_ceph_lock);
+
+	if (mds == -1) /* CDIR_AUTH_PARENT */
+		mds = diri_auth;
+
+	mutex_lock(&ci->i_fragtree_mutex);
+	if (ndist == 0 && mds == diri_auth) {
+		/* no delegation info needed. */
+		frag = __ceph_find_frag(ci, id);
+		if (!frag)
+			goto out;
+		if (frag->split_by == 0) {
+			/* tree leaf, remove */
+			dout("fill_dirfrag removed %llx.%llx frag %x"
+			     " (no ref)\n", ceph_vinop(inode), id);
+			rb_erase(&frag->node, &ci->i_fragtree);
+			kfree(frag);
+		} else {
+			/* tree branch, keep and clear */
+			dout("fill_dirfrag cleared %llx.%llx frag %x"
+			     " referral\n", ceph_vinop(inode), id);
+			frag->mds = -1;
+			frag->ndist = 0;
+		}
+		goto out;
+	}
+
+
+	/* find/add this frag to store mds delegation info */
+	frag = __get_or_create_frag(ci, id);
+	if (IS_ERR(frag)) {
+		/* this is not the end of the world; we can continue
+		   with bad/inaccurate delegation info */
+		pr_err("fill_dirfrag ENOMEM on mds ref %llx.%llx fg %x\n",
+		       ceph_vinop(inode), le32_to_cpu(dirinfo->frag));
+		err = -ENOMEM;
+		goto out;
+	}
+
+	frag->mds = mds;
+	frag->ndist = min_t(u32, ndist, CEPH_MAX_DIRFRAG_REP);
+	for (i = 0; i < frag->ndist; i++)
+		frag->dist[i] = le32_to_cpu(dirinfo->dist[i]);
+	dout("fill_dirfrag %llx.%llx frag %x ndist=%d\n",
+	     ceph_vinop(inode), frag->frag, frag->ndist);
+
+out:
+	mutex_unlock(&ci->i_fragtree_mutex);
+	return err;
+}
+
+static int frag_tree_split_cmp(const void *l, const void *r)
+{
+	struct ceph_frag_tree_split *ls = (struct ceph_frag_tree_split*)l;
+	struct ceph_frag_tree_split *rs = (struct ceph_frag_tree_split*)r;
+	return ceph_frag_compare(le32_to_cpu(ls->frag),
+				 le32_to_cpu(rs->frag));
+}
+
+static bool is_frag_child(u32 f, struct ceph_inode_frag *frag)
+{
+	if (!frag)
+		return f == ceph_frag_make(0, 0);
+	if (ceph_frag_bits(f) != ceph_frag_bits(frag->frag) + frag->split_by)
+		return false;
+	return ceph_frag_contains_value(frag->frag, ceph_frag_value(f));
+}
+
+static int ceph_fill_fragtree(struct inode *inode,
+			      struct ceph_frag_tree_head *fragtree,
+			      struct ceph_mds_reply_dirfrag *dirinfo)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_inode_frag *frag, *prev_frag = NULL;
+	struct rb_node *rb_node;
+	unsigned i, split_by, nsplits;
+	u32 id;
+	bool update = false;
+
+	mutex_lock(&ci->i_fragtree_mutex);
+	nsplits = le32_to_cpu(fragtree->nsplits);
+	if (nsplits != ci->i_fragtree_nsplits) {
+		update = true;
+	} else if (nsplits) {
+		i = prandom_u32_max(nsplits);
+		id = le32_to_cpu(fragtree->splits[i].frag);
+		if (!__ceph_find_frag(ci, id))
+			update = true;
+	} else if (!RB_EMPTY_ROOT(&ci->i_fragtree)) {
+		rb_node = rb_first(&ci->i_fragtree);
+		frag = rb_entry(rb_node, struct ceph_inode_frag, node);
+		if (frag->frag != ceph_frag_make(0, 0) || rb_next(rb_node))
+			update = true;
+	}
+	if (!update && dirinfo) {
+		id = le32_to_cpu(dirinfo->frag);
+		if (id != __ceph_choose_frag(ci, id, NULL, NULL))
+			update = true;
+	}
+	if (!update)
+		goto out_unlock;
+
+	if (nsplits > 1) {
+		sort(fragtree->splits, nsplits, sizeof(fragtree->splits[0]),
+		     frag_tree_split_cmp, NULL);
+	}
+
+	dout("fill_fragtree %llx.%llx\n", ceph_vinop(inode));
+	rb_node = rb_first(&ci->i_fragtree);
+	for (i = 0; i < nsplits; i++) {
+		id = le32_to_cpu(fragtree->splits[i].frag);
+		split_by = le32_to_cpu(fragtree->splits[i].by);
+		if (split_by == 0 || ceph_frag_bits(id) + split_by > 24) {
+			pr_err("fill_fragtree %llx.%llx invalid split %d/%u, "
+			       "frag %x split by %d\n", ceph_vinop(inode),
+			       i, nsplits, id, split_by);
+			continue;
+		}
+		frag = NULL;
+		while (rb_node) {
+			frag = rb_entry(rb_node, struct ceph_inode_frag, node);
+			if (ceph_frag_compare(frag->frag, id) >= 0) {
+				if (frag->frag != id)
+					frag = NULL;
+				else
+					rb_node = rb_next(rb_node);
+				break;
+			}
+			rb_node = rb_next(rb_node);
+			/* delete stale split/leaf node */
+			if (frag->split_by > 0 ||
+			    !is_frag_child(frag->frag, prev_frag)) {
+				rb_erase(&frag->node, &ci->i_fragtree);
+				if (frag->split_by > 0)
+					ci->i_fragtree_nsplits--;
+				kfree(frag);
+			}
+			frag = NULL;
+		}
+		if (!frag) {
+			frag = __get_or_create_frag(ci, id);
+			if (IS_ERR(frag))
+				continue;
+		}
+		if (frag->split_by == 0)
+			ci->i_fragtree_nsplits++;
+		frag->split_by = split_by;
+		dout(" frag %x split by %d\n", frag->frag, frag->split_by);
+		prev_frag = frag;
+	}
+	while (rb_node) {
+		frag = rb_entry(rb_node, struct ceph_inode_frag, node);
+		rb_node = rb_next(rb_node);
+		/* delete stale split/leaf node */
+		if (frag->split_by > 0 ||
+		    !is_frag_child(frag->frag, prev_frag)) {
+			rb_erase(&frag->node, &ci->i_fragtree);
+			if (frag->split_by > 0)
+				ci->i_fragtree_nsplits--;
+			kfree(frag);
+		}
+	}
+out_unlock:
+	mutex_unlock(&ci->i_fragtree_mutex);
+	return 0;
+}
+
+/*
+ * initialize a newly allocated inode.
+ */
+struct inode *ceph_alloc_inode(struct super_block *sb)
+{
+	struct ceph_inode_info *ci;
+	int i;
+
+	ci = alloc_inode_sb(sb, ceph_inode_cachep, GFP_NOFS);
+	if (!ci)
+		return NULL;
+
+	dout("alloc_inode %p\n", &ci->netfs.inode);
+
+	/* Set parameters for the netfs library */
+	netfs_inode_init(&ci->netfs, &ceph_netfs_ops);
+
+	spin_lock_init(&ci->i_ceph_lock);
+
+	ci->i_version = 0;
+	ci->i_inline_version = 0;
+	ci->i_time_warp_seq = 0;
+	ci->i_ceph_flags = 0;
+	atomic64_set(&ci->i_ordered_count, 1);
+	atomic64_set(&ci->i_release_count, 1);
+	atomic64_set(&ci->i_complete_seq[0], 0);
+	atomic64_set(&ci->i_complete_seq[1], 0);
+	ci->i_symlink = NULL;
+
+	ci->i_max_bytes = 0;
+	ci->i_max_files = 0;
+
+	memset(&ci->i_dir_layout, 0, sizeof(ci->i_dir_layout));
+	memset(&ci->i_cached_layout, 0, sizeof(ci->i_cached_layout));
+	RCU_INIT_POINTER(ci->i_layout.pool_ns, NULL);
+
+	ci->i_fragtree = RB_ROOT;
+	mutex_init(&ci->i_fragtree_mutex);
+
+	ci->i_xattrs.blob = NULL;
+	ci->i_xattrs.prealloc_blob = NULL;
+	ci->i_xattrs.dirty = false;
+	ci->i_xattrs.index = RB_ROOT;
+	ci->i_xattrs.count = 0;
+	ci->i_xattrs.names_size = 0;
+	ci->i_xattrs.vals_size = 0;
+	ci->i_xattrs.version = 0;
+	ci->i_xattrs.index_version = 0;
+
+	ci->i_caps = RB_ROOT;
+	ci->i_auth_cap = NULL;
+	ci->i_dirty_caps = 0;
+	ci->i_flushing_caps = 0;
+	INIT_LIST_HEAD(&ci->i_dirty_item);
+	INIT_LIST_HEAD(&ci->i_flushing_item);
+	ci->i_prealloc_cap_flush = NULL;
+	INIT_LIST_HEAD(&ci->i_cap_flush_list);
+	init_waitqueue_head(&ci->i_cap_wq);
+	ci->i_hold_caps_max = 0;
+	INIT_LIST_HEAD(&ci->i_cap_delay_list);
+	INIT_LIST_HEAD(&ci->i_cap_snaps);
+	ci->i_head_snapc = NULL;
+	ci->i_snap_caps = 0;
+
+	ci->i_last_rd = ci->i_last_wr = jiffies - 3600 * HZ;
+	for (i = 0; i < CEPH_FILE_MODE_BITS; i++)
+		ci->i_nr_by_mode[i] = 0;
+
+	mutex_init(&ci->i_truncate_mutex);
+	ci->i_truncate_seq = 0;
+	ci->i_truncate_size = 0;
+	ci->i_truncate_pending = 0;
+
+	ci->i_max_size = 0;
+	ci->i_reported_size = 0;
+	ci->i_wanted_max_size = 0;
+	ci->i_requested_max_size = 0;
+
+	ci->i_pin_ref = 0;
+	ci->i_rd_ref = 0;
+	ci->i_rdcache_ref = 0;
+	ci->i_wr_ref = 0;
+	ci->i_wb_ref = 0;
+	ci->i_fx_ref = 0;
+	ci->i_wrbuffer_ref = 0;
+	ci->i_wrbuffer_ref_head = 0;
+	atomic_set(&ci->i_filelock_ref, 0);
+	atomic_set(&ci->i_shared_gen, 1);
+	ci->i_rdcache_gen = 0;
+	ci->i_rdcache_revoking = 0;
+
+	INIT_LIST_HEAD(&ci->i_unsafe_dirops);
+	INIT_LIST_HEAD(&ci->i_unsafe_iops);
+	spin_lock_init(&ci->i_unsafe_lock);
+
+	ci->i_snap_realm = NULL;
+	INIT_LIST_HEAD(&ci->i_snap_realm_item);
+	INIT_LIST_HEAD(&ci->i_snap_flush_item);
+
+	INIT_WORK(&ci->i_work, ceph_inode_work);
+	ci->i_work_mask = 0;
+	memset(&ci->i_btime, '\0', sizeof(ci->i_btime));
+	return &ci->netfs.inode;
+}
+
+void ceph_free_inode(struct inode *inode)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+
+	kfree(ci->i_symlink);
+	kmem_cache_free(ceph_inode_cachep, ci);
+}
+
+void ceph_evict_inode(struct inode *inode)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
+	struct ceph_inode_frag *frag;
+	struct rb_node *n;
+
+	dout("evict_inode %p ino %llx.%llx\n", inode, ceph_vinop(inode));
+
+	percpu_counter_dec(&mdsc->metric.total_inodes);
+
+	truncate_inode_pages_final(&inode->i_data);
+	if (inode->i_state & I_PINNING_FSCACHE_WB)
+		ceph_fscache_unuse_cookie(inode, true);
+	clear_inode(inode);
+
+	ceph_fscache_unregister_inode_cookie(ci);
+
+	__ceph_remove_caps(ci);
+
+	if (__ceph_has_quota(ci, QUOTA_GET_ANY))
+		ceph_adjust_quota_realms_count(inode, false);
+
+	/*
+	 * we may still have a snap_realm reference if there are stray
+	 * caps in i_snap_caps.
+	 */
+	if (ci->i_snap_realm) {
+		if (ceph_snap(inode) == CEPH_NOSNAP) {
+			dout(" dropping residual ref to snap realm %p\n",
+			     ci->i_snap_realm);
+			ceph_change_snap_realm(inode, NULL);
+		} else {
+			ceph_put_snapid_map(mdsc, ci->i_snapid_map);
+			ci->i_snap_realm = NULL;
+		}
+	}
+
+	while ((n = rb_first(&ci->i_fragtree)) != NULL) {
+		frag = rb_entry(n, struct ceph_inode_frag, node);
+		rb_erase(n, &ci->i_fragtree);
+		kfree(frag);
+	}
+	ci->i_fragtree_nsplits = 0;
+
+	__ceph_destroy_xattrs(ci);
+	if (ci->i_xattrs.blob)
+		ceph_buffer_put(ci->i_xattrs.blob);
+	if (ci->i_xattrs.prealloc_blob)
+		ceph_buffer_put(ci->i_xattrs.prealloc_blob);
+
+	ceph_put_string(rcu_dereference_raw(ci->i_layout.pool_ns));
+	ceph_put_string(rcu_dereference_raw(ci->i_cached_layout.pool_ns));
+}
+
+static inline blkcnt_t calc_inode_blocks(u64 size)
+{
+	return (size + (1<<9) - 1) >> 9;
+}
+
+/*
+ * Helpers to fill in size, ctime, mtime, and atime.  We have to be
+ * careful because either the client or MDS may have more up to date
+ * info, depending on which capabilities are held, and whether
+ * time_warp_seq or truncate_seq have increased.  (Ordinarily, mtime
+ * and size are monotonically increasing, except when utimes() or
+ * truncate() increments the corresponding _seq values.)
+ */
+int ceph_fill_file_size(struct inode *inode, int issued,
+			u32 truncate_seq, u64 truncate_size, u64 size)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	int queue_trunc = 0;
+	loff_t isize = i_size_read(inode);
+
+	if (ceph_seq_cmp(truncate_seq, ci->i_truncate_seq) > 0 ||
+	    (truncate_seq == ci->i_truncate_seq && size > isize)) {
+		dout("size %lld -> %llu\n", isize, size);
+		if (size > 0 && S_ISDIR(inode->i_mode)) {
+			pr_err("fill_file_size non-zero size for directory\n");
+			size = 0;
+		}
+		i_size_write(inode, size);
+		inode->i_blocks = calc_inode_blocks(size);
+		/*
+		 * If we're expanding, then we should be able to just update
+		 * the existing cookie.
+		 */
+		if (size > isize)
+			ceph_fscache_update(inode);
+		ci->i_reported_size = size;
+		if (truncate_seq != ci->i_truncate_seq) {
+			dout("truncate_seq %u -> %u\n",
+			     ci->i_truncate_seq, truncate_seq);
+			ci->i_truncate_seq = truncate_seq;
+
+			/* the MDS should have revoked these caps */
+			WARN_ON_ONCE(issued & (CEPH_CAP_FILE_RD |
+					       CEPH_CAP_FILE_LAZYIO));
+			/*
+			 * If we hold relevant caps, or in the case where we're
+			 * not the only client referencing this file and we
+			 * don't hold those caps, then we need to check whether
+			 * the file is either opened or mmaped
+			 */
+			if ((issued & (CEPH_CAP_FILE_CACHE|
+				       CEPH_CAP_FILE_BUFFER)) ||
+			    mapping_mapped(inode->i_mapping) ||
+			    __ceph_is_file_opened(ci)) {
+				ci->i_truncate_pending++;
+				queue_trunc = 1;
+			}
+		}
+	}
+	if (ceph_seq_cmp(truncate_seq, ci->i_truncate_seq) >= 0 &&
+	    ci->i_truncate_size != truncate_size) {
+		dout("truncate_size %lld -> %llu\n", ci->i_truncate_size,
+		     truncate_size);
+		ci->i_truncate_size = truncate_size;
+	}
+	return queue_trunc;
+}
+
+void ceph_fill_file_time(struct inode *inode, int issued,
+			 u64 time_warp_seq, struct timespec64 *ctime,
+			 struct timespec64 *mtime, struct timespec64 *atime)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	int warn = 0;
+
+	if (issued & (CEPH_CAP_FILE_EXCL|
+		      CEPH_CAP_FILE_WR|
+		      CEPH_CAP_FILE_BUFFER|
+		      CEPH_CAP_AUTH_EXCL|
+		      CEPH_CAP_XATTR_EXCL)) {
+		if (ci->i_version == 0 ||
+		    timespec64_compare(ctime, &inode->i_ctime) > 0) {
+			dout("ctime %lld.%09ld -> %lld.%09ld inc w/ cap\n",
+			     inode->i_ctime.tv_sec, inode->i_ctime.tv_nsec,
+			     ctime->tv_sec, ctime->tv_nsec);
+			inode->i_ctime = *ctime;
+		}
+		if (ci->i_version == 0 ||
+		    ceph_seq_cmp(time_warp_seq, ci->i_time_warp_seq) > 0) {
+			/* the MDS did a utimes() */
+			dout("mtime %lld.%09ld -> %lld.%09ld "
+			     "tw %d -> %d\n",
+			     inode->i_mtime.tv_sec, inode->i_mtime.tv_nsec,
+			     mtime->tv_sec, mtime->tv_nsec,
+			     ci->i_time_warp_seq, (int)time_warp_seq);
+
+			inode->i_mtime = *mtime;
+			inode->i_atime = *atime;
+			ci->i_time_warp_seq = time_warp_seq;
+		} else if (time_warp_seq == ci->i_time_warp_seq) {
+			/* nobody did utimes(); take the max */
+			if (timespec64_compare(mtime, &inode->i_mtime) > 0) {
+				dout("mtime %lld.%09ld -> %lld.%09ld inc\n",
+				     inode->i_mtime.tv_sec,
+				     inode->i_mtime.tv_nsec,
+				     mtime->tv_sec, mtime->tv_nsec);
+				inode->i_mtime = *mtime;
+			}
+			if (timespec64_compare(atime, &inode->i_atime) > 0) {
+				dout("atime %lld.%09ld -> %lld.%09ld inc\n",
+				     inode->i_atime.tv_sec,
+				     inode->i_atime.tv_nsec,
+				     atime->tv_sec, atime->tv_nsec);
+				inode->i_atime = *atime;
+			}
+		} else if (issued & CEPH_CAP_FILE_EXCL) {
+			/* we did a utimes(); ignore mds values */
+		} else {
+			warn = 1;
+		}
+	} else {
+		/* we have no write|excl caps; whatever the MDS says is true */
+		if (ceph_seq_cmp(time_warp_seq, ci->i_time_warp_seq) >= 0) {
+			inode->i_ctime = *ctime;
+			inode->i_mtime = *mtime;
+			inode->i_atime = *atime;
+			ci->i_time_warp_seq = time_warp_seq;
+		} else {
+			warn = 1;
+		}
+	}
+	if (warn) /* time_warp_seq shouldn't go backwards */
+		dout("%p mds time_warp_seq %llu < %u\n",
+		     inode, time_warp_seq, ci->i_time_warp_seq);
+}
+
+/*
+ * Populate an inode based on info from mds.  May be called on new or
+ * existing inodes.
+ */
+int ceph_fill_inode(struct inode *inode, struct page *locked_page,
+		    struct ceph_mds_reply_info_in *iinfo,
+		    struct ceph_mds_reply_dirfrag *dirinfo,
+		    struct ceph_mds_session *session, int cap_fmode,
+		    struct ceph_cap_reservation *caps_reservation)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
+	struct ceph_mds_reply_inode *info = iinfo->in;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	int issued, new_issued, info_caps;
+	struct timespec64 mtime, atime, ctime;
+	struct ceph_buffer *xattr_blob = NULL;
+	struct ceph_buffer *old_blob = NULL;
+	struct ceph_string *pool_ns = NULL;
+	struct ceph_cap *new_cap = NULL;
+	int err = 0;
+	bool wake = false;
+	bool queue_trunc = false;
+	bool new_version = false;
+	bool fill_inline = false;
+	umode_t mode = le32_to_cpu(info->mode);
+	dev_t rdev = le32_to_cpu(info->rdev);
+
+	lockdep_assert_held(&mdsc->snap_rwsem);
+
+	dout("%s %p ino %llx.%llx v %llu had %llu\n", __func__,
+	     inode, ceph_vinop(inode), le64_to_cpu(info->version),
+	     ci->i_version);
+
+	/* Once I_NEW is cleared, we can't change type or dev numbers */
+	if (inode->i_state & I_NEW) {
+		inode->i_mode = mode;
+	} else {
+		if (inode_wrong_type(inode, mode)) {
+			pr_warn_once("inode type changed! (ino %llx.%llx is 0%o, mds says 0%o)\n",
+				     ceph_vinop(inode), inode->i_mode, mode);
+			return -ESTALE;
+		}
+
+		if ((S_ISCHR(mode) || S_ISBLK(mode)) && inode->i_rdev != rdev) {
+			pr_warn_once("dev inode rdev changed! (ino %llx.%llx is %u:%u, mds says %u:%u)\n",
+				     ceph_vinop(inode), MAJOR(inode->i_rdev),
+				     MINOR(inode->i_rdev), MAJOR(rdev),
+				     MINOR(rdev));
+			return -ESTALE;
+		}
+	}
+
+	info_caps = le32_to_cpu(info->cap.caps);
+
+	/* prealloc new cap struct */
+	if (info_caps && ceph_snap(inode) == CEPH_NOSNAP) {
+		new_cap = ceph_get_cap(mdsc, caps_reservation);
+		if (!new_cap)
+			return -ENOMEM;
+	}
+
+	/*
+	 * prealloc xattr data, if it looks like we'll need it.  only
+	 * if len > 4 (meaning there are actually xattrs; the first 4
+	 * bytes are the xattr count).
+	 */
+	if (iinfo->xattr_len > 4) {
+		xattr_blob = ceph_buffer_new(iinfo->xattr_len, GFP_NOFS);
+		if (!xattr_blob)
+			pr_err("%s ENOMEM xattr blob %d bytes\n", __func__,
+			       iinfo->xattr_len);
+	}
+
+	if (iinfo->pool_ns_len > 0)
+		pool_ns = ceph_find_or_create_string(iinfo->pool_ns_data,
+						     iinfo->pool_ns_len);
+
+	if (ceph_snap(inode) != CEPH_NOSNAP && !ci->i_snapid_map)
+		ci->i_snapid_map = ceph_get_snapid_map(mdsc, ceph_snap(inode));
+
+	spin_lock(&ci->i_ceph_lock);
+
+	/*
+	 * provided version will be odd if inode value is projected,
+	 * even if stable.  skip the update if we have newer stable
+	 * info (ours>=theirs, e.g. due to racing mds replies), unless
+	 * we are getting projected (unstable) info (in which case the
+	 * version is odd, and we want ours>theirs).
+	 *   us   them
+	 *   2    2     skip
+	 *   3    2     skip
+	 *   3    3     update
+	 */
+	if (ci->i_version == 0 ||
+	    ((info->cap.flags & CEPH_CAP_FLAG_AUTH) &&
+	     le64_to_cpu(info->version) > (ci->i_version & ~1)))
+		new_version = true;
+
+	/* Update change_attribute */
+	inode_set_max_iversion_raw(inode, iinfo->change_attr);
+
+	__ceph_caps_issued(ci, &issued);
+	issued |= __ceph_caps_dirty(ci);
+	new_issued = ~issued & info_caps;
+
+	/* directories have fl_stripe_unit set to zero */
+	if (le32_to_cpu(info->layout.fl_stripe_unit))
+		inode->i_blkbits =
+			fls(le32_to_cpu(info->layout.fl_stripe_unit)) - 1;
+	else
+		inode->i_blkbits = CEPH_BLOCK_SHIFT;
+
+	__ceph_update_quota(ci, iinfo->max_bytes, iinfo->max_files);
+
+	if ((new_version || (new_issued & CEPH_CAP_AUTH_SHARED)) &&
+	    (issued & CEPH_CAP_AUTH_EXCL) == 0) {
+		inode->i_mode = mode;
+		inode->i_uid = make_kuid(&init_user_ns, le32_to_cpu(info->uid));
+		inode->i_gid = make_kgid(&init_user_ns, le32_to_cpu(info->gid));
+		dout("%p mode 0%o uid.gid %d.%d\n", inode, inode->i_mode,
+		     from_kuid(&init_user_ns, inode->i_uid),
+		     from_kgid(&init_user_ns, inode->i_gid));
+		ceph_decode_timespec64(&ci->i_btime, &iinfo->btime);
+		ceph_decode_timespec64(&ci->i_snap_btime, &iinfo->snap_btime);
+	}
+
+	if ((new_version || (new_issued & CEPH_CAP_LINK_SHARED)) &&
+	    (issued & CEPH_CAP_LINK_EXCL) == 0)
+		set_nlink(inode, le32_to_cpu(info->nlink));
+
+	if (new_version || (new_issued & CEPH_CAP_ANY_RD)) {
+		/* be careful with mtime, atime, size */
+		ceph_decode_timespec64(&atime, &info->atime);
+		ceph_decode_timespec64(&mtime, &info->mtime);
+		ceph_decode_timespec64(&ctime, &info->ctime);
+		ceph_fill_file_time(inode, issued,
+				le32_to_cpu(info->time_warp_seq),
+				&ctime, &mtime, &atime);
+	}
+
+	if (new_version || (info_caps & CEPH_CAP_FILE_SHARED)) {
+		ci->i_files = le64_to_cpu(info->files);
+		ci->i_subdirs = le64_to_cpu(info->subdirs);
+	}
+
+	if (new_version ||
+	    (new_issued & (CEPH_CAP_ANY_FILE_RD | CEPH_CAP_ANY_FILE_WR))) {
+		s64 old_pool = ci->i_layout.pool_id;
+		struct ceph_string *old_ns;
+
+		ceph_file_layout_from_legacy(&ci->i_layout, &info->layout);
+		old_ns = rcu_dereference_protected(ci->i_layout.pool_ns,
+					lockdep_is_held(&ci->i_ceph_lock));
+		rcu_assign_pointer(ci->i_layout.pool_ns, pool_ns);
+
+		if (ci->i_layout.pool_id != old_pool || pool_ns != old_ns)
+			ci->i_ceph_flags &= ~CEPH_I_POOL_PERM;
+
+		pool_ns = old_ns;
+
+		queue_trunc = ceph_fill_file_size(inode, issued,
+					le32_to_cpu(info->truncate_seq),
+					le64_to_cpu(info->truncate_size),
+					le64_to_cpu(info->size));
+		/* only update max_size on auth cap */
+		if ((info->cap.flags & CEPH_CAP_FLAG_AUTH) &&
+		    ci->i_max_size != le64_to_cpu(info->max_size)) {
+			dout("max_size %lld -> %llu\n", ci->i_max_size,
+					le64_to_cpu(info->max_size));
+			ci->i_max_size = le64_to_cpu(info->max_size);
+		}
+	}
+
+	/* layout and rstat are not tracked by capability, update them if
+	 * the inode info is from auth mds */
+	if (new_version || (info->cap.flags & CEPH_CAP_FLAG_AUTH)) {
+		if (S_ISDIR(inode->i_mode)) {
+			ci->i_dir_layout = iinfo->dir_layout;
+			ci->i_rbytes = le64_to_cpu(info->rbytes);
+			ci->i_rfiles = le64_to_cpu(info->rfiles);
+			ci->i_rsubdirs = le64_to_cpu(info->rsubdirs);
+			ci->i_dir_pin = iinfo->dir_pin;
+			ci->i_rsnaps = iinfo->rsnaps;
+			ceph_decode_timespec64(&ci->i_rctime, &info->rctime);
+		}
+	}
+
+	/* xattrs */
+	/* note that if i_xattrs.len <= 4, i_xattrs.data will still be NULL. */
+	if ((ci->i_xattrs.version == 0 || !(issued & CEPH_CAP_XATTR_EXCL))  &&
+	    le64_to_cpu(info->xattr_version) > ci->i_xattrs.version) {
+		if (ci->i_xattrs.blob)
+			old_blob = ci->i_xattrs.blob;
+		ci->i_xattrs.blob = xattr_blob;
+		if (xattr_blob)
+			memcpy(ci->i_xattrs.blob->vec.iov_base,
+			       iinfo->xattr_data, iinfo->xattr_len);
+		ci->i_xattrs.version = le64_to_cpu(info->xattr_version);
+		ceph_forget_all_cached_acls(inode);
+		ceph_security_invalidate_secctx(inode);
+		xattr_blob = NULL;
+	}
+
+	/* finally update i_version */
+	if (le64_to_cpu(info->version) > ci->i_version)
+		ci->i_version = le64_to_cpu(info->version);
+
+	inode->i_mapping->a_ops = &ceph_aops;
+
+	switch (inode->i_mode & S_IFMT) {
+	case S_IFIFO:
+	case S_IFBLK:
+	case S_IFCHR:
+	case S_IFSOCK:
+		inode->i_blkbits = PAGE_SHIFT;
+		init_special_inode(inode, inode->i_mode, rdev);
+		inode->i_op = &ceph_file_iops;
+		break;
+	case S_IFREG:
+		inode->i_op = &ceph_file_iops;
+		inode->i_fop = &ceph_file_fops;
+		break;
+	case S_IFLNK:
+		inode->i_op = &ceph_symlink_iops;
+		if (!ci->i_symlink) {
+			u32 symlen = iinfo->symlink_len;
+			char *sym;
+
+			spin_unlock(&ci->i_ceph_lock);
+
+			if (symlen != i_size_read(inode)) {
+				pr_err("%s %llx.%llx BAD symlink "
+					"size %lld\n", __func__,
+					ceph_vinop(inode),
+					i_size_read(inode));
+				i_size_write(inode, symlen);
+				inode->i_blocks = calc_inode_blocks(symlen);
+			}
+
+			err = -ENOMEM;
+			sym = kstrndup(iinfo->symlink, symlen, GFP_NOFS);
+			if (!sym)
+				goto out;
+
+			spin_lock(&ci->i_ceph_lock);
+			if (!ci->i_symlink)
+				ci->i_symlink = sym;
+			else
+				kfree(sym); /* lost a race */
+		}
+		inode->i_link = ci->i_symlink;
+		break;
+	case S_IFDIR:
+		inode->i_op = &ceph_dir_iops;
+		inode->i_fop = &ceph_dir_fops;
+		break;
+	default:
+		pr_err("%s %llx.%llx BAD mode 0%o\n", __func__,
+		       ceph_vinop(inode), inode->i_mode);
+	}
+
+	/* were we issued a capability? */
+	if (info_caps) {
+		if (ceph_snap(inode) == CEPH_NOSNAP) {
+			ceph_add_cap(inode, session,
+				     le64_to_cpu(info->cap.cap_id),
+				     info_caps,
+				     le32_to_cpu(info->cap.wanted),
+				     le32_to_cpu(info->cap.seq),
+				     le32_to_cpu(info->cap.mseq),
+				     le64_to_cpu(info->cap.realm),
+				     info->cap.flags, &new_cap);
+
+			/* set dir completion flag? */
+			if (S_ISDIR(inode->i_mode) &&
+			    ci->i_files == 0 && ci->i_subdirs == 0 &&
+			    (info_caps & CEPH_CAP_FILE_SHARED) &&
+			    (issued & CEPH_CAP_FILE_EXCL) == 0 &&
+			    !__ceph_dir_is_complete(ci)) {
+				dout(" marking %p complete (empty)\n", inode);
+				i_size_write(inode, 0);
+				__ceph_dir_set_complete(ci,
+					atomic64_read(&ci->i_release_count),
+					atomic64_read(&ci->i_ordered_count));
+			}
+
+			wake = true;
+		} else {
+			dout(" %p got snap_caps %s\n", inode,
+			     ceph_cap_string(info_caps));
+			ci->i_snap_caps |= info_caps;
+		}
+	}
+
+	if (iinfo->inline_version > 0 &&
+	    iinfo->inline_version >= ci->i_inline_version) {
+		int cache_caps = CEPH_CAP_FILE_CACHE | CEPH_CAP_FILE_LAZYIO;
+		ci->i_inline_version = iinfo->inline_version;
+		if (ceph_has_inline_data(ci) &&
+		    (locked_page || (info_caps & cache_caps)))
+			fill_inline = true;
+	}
+
+	if (cap_fmode >= 0) {
+		if (!info_caps)
+			pr_warn("mds issued no caps on %llx.%llx\n",
+				ceph_vinop(inode));
+		__ceph_touch_fmode(ci, mdsc, cap_fmode);
+	}
+
+	spin_unlock(&ci->i_ceph_lock);
+
+	ceph_fscache_register_inode_cookie(inode);
+
+	if (fill_inline)
+		ceph_fill_inline_data(inode, locked_page,
+				      iinfo->inline_data, iinfo->inline_len);
+
+	if (wake)
+		wake_up_all(&ci->i_cap_wq);
+
+	/* queue truncate if we saw i_size decrease */
+	if (queue_trunc)
+		ceph_queue_vmtruncate(inode);
+
+	/* populate frag tree */
+	if (S_ISDIR(inode->i_mode))
+		ceph_fill_fragtree(inode, &info->fragtree, dirinfo);
+
+	/* update delegation info? */
+	if (dirinfo)
+		ceph_fill_dirfrag(inode, dirinfo);
+
+	err = 0;
+out:
+	if (new_cap)
+		ceph_put_cap(mdsc, new_cap);
+	ceph_buffer_put(old_blob);
+	ceph_buffer_put(xattr_blob);
+	ceph_put_string(pool_ns);
+	return err;
+}
+
+/*
+ * caller should hold session s_mutex and dentry->d_lock.
+ */
+static void __update_dentry_lease(struct inode *dir, struct dentry *dentry,
+				  struct ceph_mds_reply_lease *lease,
+				  struct ceph_mds_session *session,
+				  unsigned long from_time,
+				  struct ceph_mds_session **old_lease_session)
+{
+	struct ceph_dentry_info *di = ceph_dentry(dentry);
+	unsigned mask = le16_to_cpu(lease->mask);
+	long unsigned duration = le32_to_cpu(lease->duration_ms);
+	long unsigned ttl = from_time + (duration * HZ) / 1000;
+	long unsigned half_ttl = from_time + (duration * HZ / 2) / 1000;
+
+	dout("update_dentry_lease %p duration %lu ms ttl %lu\n",
+	     dentry, duration, ttl);
+
+	/* only track leases on regular dentries */
+	if (ceph_snap(dir) != CEPH_NOSNAP)
+		return;
+
+	if (mask & CEPH_LEASE_PRIMARY_LINK)
+		di->flags |= CEPH_DENTRY_PRIMARY_LINK;
+	else
+		di->flags &= ~CEPH_DENTRY_PRIMARY_LINK;
+
+	di->lease_shared_gen = atomic_read(&ceph_inode(dir)->i_shared_gen);
+	if (!(mask & CEPH_LEASE_VALID)) {
+		__ceph_dentry_dir_lease_touch(di);
+		return;
+	}
+
+	if (di->lease_gen == atomic_read(&session->s_cap_gen) &&
+	    time_before(ttl, di->time))
+		return;  /* we already have a newer lease. */
+
+	if (di->lease_session && di->lease_session != session) {
+		*old_lease_session = di->lease_session;
+		di->lease_session = NULL;
+	}
+
+	if (!di->lease_session)
+		di->lease_session = ceph_get_mds_session(session);
+	di->lease_gen = atomic_read(&session->s_cap_gen);
+	di->lease_seq = le32_to_cpu(lease->seq);
+	di->lease_renew_after = half_ttl;
+	di->lease_renew_from = 0;
+	di->time = ttl;
+
+	__ceph_dentry_lease_touch(di);
+}
+
+static inline void update_dentry_lease(struct inode *dir, struct dentry *dentry,
+					struct ceph_mds_reply_lease *lease,
+					struct ceph_mds_session *session,
+					unsigned long from_time)
+{
+	struct ceph_mds_session *old_lease_session = NULL;
+	spin_lock(&dentry->d_lock);
+	__update_dentry_lease(dir, dentry, lease, session, from_time,
+			      &old_lease_session);
+	spin_unlock(&dentry->d_lock);
+	ceph_put_mds_session(old_lease_session);
+}
+
+/*
+ * update dentry lease without having parent inode locked
+ */
+static void update_dentry_lease_careful(struct dentry *dentry,
+					struct ceph_mds_reply_lease *lease,
+					struct ceph_mds_session *session,
+					unsigned long from_time,
+					char *dname, u32 dname_len,
+					struct ceph_vino *pdvino,
+					struct ceph_vino *ptvino)
+
+{
+	struct inode *dir;
+	struct ceph_mds_session *old_lease_session = NULL;
+
+	spin_lock(&dentry->d_lock);
+	/* make sure dentry's name matches target */
+	if (dentry->d_name.len != dname_len ||
+	    memcmp(dentry->d_name.name, dname, dname_len))
+		goto out_unlock;
+
+	dir = d_inode(dentry->d_parent);
+	/* make sure parent matches dvino */
+	if (!ceph_ino_compare(dir, pdvino))
+		goto out_unlock;
+
+	/* make sure dentry's inode matches target. NULL ptvino means that
+	 * we expect a negative dentry */
+	if (ptvino) {
+		if (d_really_is_negative(dentry))
+			goto out_unlock;
+		if (!ceph_ino_compare(d_inode(dentry), ptvino))
+			goto out_unlock;
+	} else {
+		if (d_really_is_positive(dentry))
+			goto out_unlock;
+	}
+
+	__update_dentry_lease(dir, dentry, lease, session,
+			      from_time, &old_lease_session);
+out_unlock:
+	spin_unlock(&dentry->d_lock);
+	ceph_put_mds_session(old_lease_session);
+}
+
+/*
+ * splice a dentry to an inode.
+ * caller must hold directory i_rwsem for this to be safe.
+ */
+static int splice_dentry(struct dentry **pdn, struct inode *in)
+{
+	struct dentry *dn = *pdn;
+	struct dentry *realdn;
+
+	BUG_ON(d_inode(dn));
+
+	if (S_ISDIR(in->i_mode)) {
+		/* If inode is directory, d_splice_alias() below will remove
+		 * 'realdn' from its origin parent. We need to ensure that
+		 * origin parent's readdir cache will not reference 'realdn'
+		 */
+		realdn = d_find_any_alias(in);
+		if (realdn) {
+			struct ceph_dentry_info *di = ceph_dentry(realdn);
+			spin_lock(&realdn->d_lock);
+
+			realdn->d_op->d_prune(realdn);
+
+			di->time = jiffies;
+			di->lease_shared_gen = 0;
+			di->offset = 0;
+
+			spin_unlock(&realdn->d_lock);
+			dput(realdn);
+		}
+	}
+
+	/* dn must be unhashed */
+	if (!d_unhashed(dn))
+		d_drop(dn);
+	realdn = d_splice_alias(in, dn);
+	if (IS_ERR(realdn)) {
+		pr_err("splice_dentry error %ld %p inode %p ino %llx.%llx\n",
+		       PTR_ERR(realdn), dn, in, ceph_vinop(in));
+		return PTR_ERR(realdn);
+	}
+
+	if (realdn) {
+		dout("dn %p (%d) spliced with %p (%d) "
+		     "inode %p ino %llx.%llx\n",
+		     dn, d_count(dn),
+		     realdn, d_count(realdn),
+		     d_inode(realdn), ceph_vinop(d_inode(realdn)));
+		dput(dn);
+		*pdn = realdn;
+	} else {
+		BUG_ON(!ceph_dentry(dn));
+		dout("dn %p attached to %p ino %llx.%llx\n",
+		     dn, d_inode(dn), ceph_vinop(d_inode(dn)));
+	}
+	return 0;
+}
+
+/*
+ * Incorporate results into the local cache.  This is either just
+ * one inode, or a directory, dentry, and possibly linked-to inode (e.g.,
+ * after a lookup).
+ *
+ * A reply may contain
+ *         a directory inode along with a dentry.
+ *  and/or a target inode
+ *
+ * Called with snap_rwsem (read).
+ */
+int ceph_fill_trace(struct super_block *sb, struct ceph_mds_request *req)
+{
+	struct ceph_mds_session *session = req->r_session;
+	struct ceph_mds_reply_info_parsed *rinfo = &req->r_reply_info;
+	struct inode *in = NULL;
+	struct ceph_vino tvino, dvino;
+	struct ceph_fs_client *fsc = ceph_sb_to_client(sb);
+	int err = 0;
+
+	dout("fill_trace %p is_dentry %d is_target %d\n", req,
+	     rinfo->head->is_dentry, rinfo->head->is_target);
+
+	if (!rinfo->head->is_target && !rinfo->head->is_dentry) {
+		dout("fill_trace reply is empty!\n");
+		if (rinfo->head->result == 0 && req->r_parent)
+			ceph_invalidate_dir_request(req);
+		return 0;
+	}
+
+	if (rinfo->head->is_dentry) {
+		struct inode *dir = req->r_parent;
+
+		if (dir) {
+			err = ceph_fill_inode(dir, NULL, &rinfo->diri,
+					      rinfo->dirfrag, session, -1,
+					      &req->r_caps_reservation);
+			if (err < 0)
+				goto done;
+		} else {
+			WARN_ON_ONCE(1);
+		}
+
+		if (dir && req->r_op == CEPH_MDS_OP_LOOKUPNAME &&
+		    test_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags) &&
+		    !test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags)) {
+			struct qstr dname;
+			struct dentry *dn, *parent;
+
+			BUG_ON(!rinfo->head->is_target);
+			BUG_ON(req->r_dentry);
+
+			parent = d_find_any_alias(dir);
+			BUG_ON(!parent);
+
+			dname.name = rinfo->dname;
+			dname.len = rinfo->dname_len;
+			dname.hash = full_name_hash(parent, dname.name, dname.len);
+			tvino.ino = le64_to_cpu(rinfo->targeti.in->ino);
+			tvino.snap = le64_to_cpu(rinfo->targeti.in->snapid);
+retry_lookup:
+			dn = d_lookup(parent, &dname);
+			dout("d_lookup on parent=%p name=%.*s got %p\n",
+			     parent, dname.len, dname.name, dn);
+
+			if (!dn) {
+				dn = d_alloc(parent, &dname);
+				dout("d_alloc %p '%.*s' = %p\n", parent,
+				     dname.len, dname.name, dn);
+				if (!dn) {
+					dput(parent);
+					err = -ENOMEM;
+					goto done;
+				}
+				err = 0;
+			} else if (d_really_is_positive(dn) &&
+				   (ceph_ino(d_inode(dn)) != tvino.ino ||
+				    ceph_snap(d_inode(dn)) != tvino.snap)) {
+				dout(" dn %p points to wrong inode %p\n",
+				     dn, d_inode(dn));
+				ceph_dir_clear_ordered(dir);
+				d_delete(dn);
+				dput(dn);
+				goto retry_lookup;
+			}
+
+			req->r_dentry = dn;
+			dput(parent);
+		}
+	}
+
+	if (rinfo->head->is_target) {
+		/* Should be filled in by handle_reply */
+		BUG_ON(!req->r_target_inode);
+
+		in = req->r_target_inode;
+		err = ceph_fill_inode(in, req->r_locked_page, &rinfo->targeti,
+				NULL, session,
+				(!test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags) &&
+				 !test_bit(CEPH_MDS_R_ASYNC, &req->r_req_flags) &&
+				 rinfo->head->result == 0) ?  req->r_fmode : -1,
+				&req->r_caps_reservation);
+		if (err < 0) {
+			pr_err("ceph_fill_inode badness %p %llx.%llx\n",
+				in, ceph_vinop(in));
+			req->r_target_inode = NULL;
+			if (in->i_state & I_NEW)
+				discard_new_inode(in);
+			else
+				iput(in);
+			goto done;
+		}
+		if (in->i_state & I_NEW)
+			unlock_new_inode(in);
+	}
+
+	/*
+	 * ignore null lease/binding on snapdir ENOENT, or else we
+	 * will have trouble splicing in the virtual snapdir later
+	 */
+	if (rinfo->head->is_dentry &&
+            !test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags) &&
+	    test_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags) &&
+	    (rinfo->head->is_target || strncmp(req->r_dentry->d_name.name,
+					       fsc->mount_options->snapdir_name,
+					       req->r_dentry->d_name.len))) {
+		/*
+		 * lookup link rename   : null -> possibly existing inode
+		 * mknod symlink mkdir  : null -> new inode
+		 * unlink               : linked -> null
+		 */
+		struct inode *dir = req->r_parent;
+		struct dentry *dn = req->r_dentry;
+		bool have_dir_cap, have_lease;
+
+		BUG_ON(!dn);
+		BUG_ON(!dir);
+		BUG_ON(d_inode(dn->d_parent) != dir);
+
+		dvino.ino = le64_to_cpu(rinfo->diri.in->ino);
+		dvino.snap = le64_to_cpu(rinfo->diri.in->snapid);
+
+		BUG_ON(ceph_ino(dir) != dvino.ino);
+		BUG_ON(ceph_snap(dir) != dvino.snap);
+
+		/* do we have a lease on the whole dir? */
+		have_dir_cap =
+			(le32_to_cpu(rinfo->diri.in->cap.caps) &
+			 CEPH_CAP_FILE_SHARED);
+
+		/* do we have a dn lease? */
+		have_lease = have_dir_cap ||
+			le32_to_cpu(rinfo->dlease->duration_ms);
+		if (!have_lease)
+			dout("fill_trace  no dentry lease or dir cap\n");
+
+		/* rename? */
+		if (req->r_old_dentry && req->r_op == CEPH_MDS_OP_RENAME) {
+			struct inode *olddir = req->r_old_dentry_dir;
+			BUG_ON(!olddir);
+
+			dout(" src %p '%pd' dst %p '%pd'\n",
+			     req->r_old_dentry,
+			     req->r_old_dentry,
+			     dn, dn);
+			dout("fill_trace doing d_move %p -> %p\n",
+			     req->r_old_dentry, dn);
+
+			/* d_move screws up sibling dentries' offsets */
+			ceph_dir_clear_ordered(dir);
+			ceph_dir_clear_ordered(olddir);
+
+			d_move(req->r_old_dentry, dn);
+			dout(" src %p '%pd' dst %p '%pd'\n",
+			     req->r_old_dentry,
+			     req->r_old_dentry,
+			     dn, dn);
+
+			/* ensure target dentry is invalidated, despite
+			   rehashing bug in vfs_rename_dir */
+			ceph_invalidate_dentry_lease(dn);
+
+			dout("dn %p gets new offset %lld\n", req->r_old_dentry,
+			     ceph_dentry(req->r_old_dentry)->offset);
+
+			/* swap r_dentry and r_old_dentry in case that
+			 * splice_dentry() gets called later. This is safe
+			 * because no other place will use them */
+			req->r_dentry = req->r_old_dentry;
+			req->r_old_dentry = dn;
+			dn = req->r_dentry;
+		}
+
+		/* null dentry? */
+		if (!rinfo->head->is_target) {
+			dout("fill_trace null dentry\n");
+			if (d_really_is_positive(dn)) {
+				dout("d_delete %p\n", dn);
+				ceph_dir_clear_ordered(dir);
+				d_delete(dn);
+			} else if (have_lease) {
+				if (d_unhashed(dn))
+					d_add(dn, NULL);
+			}
+
+			if (!d_unhashed(dn) && have_lease)
+				update_dentry_lease(dir, dn,
+						    rinfo->dlease, session,
+						    req->r_request_started);
+			goto done;
+		}
+
+		/* attach proper inode */
+		if (d_really_is_negative(dn)) {
+			ceph_dir_clear_ordered(dir);
+			ihold(in);
+			err = splice_dentry(&req->r_dentry, in);
+			if (err < 0)
+				goto done;
+			dn = req->r_dentry;  /* may have spliced */
+		} else if (d_really_is_positive(dn) && d_inode(dn) != in) {
+			dout(" %p links to %p %llx.%llx, not %llx.%llx\n",
+			     dn, d_inode(dn), ceph_vinop(d_inode(dn)),
+			     ceph_vinop(in));
+			d_invalidate(dn);
+			have_lease = false;
+		}
+
+		if (have_lease) {
+			update_dentry_lease(dir, dn,
+					    rinfo->dlease, session,
+					    req->r_request_started);
+		}
+		dout(" final dn %p\n", dn);
+	} else if ((req->r_op == CEPH_MDS_OP_LOOKUPSNAP ||
+		    req->r_op == CEPH_MDS_OP_MKSNAP) &&
+	           test_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags) &&
+		   !test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags)) {
+		struct inode *dir = req->r_parent;
+
+		/* fill out a snapdir LOOKUPSNAP dentry */
+		BUG_ON(!dir);
+		BUG_ON(ceph_snap(dir) != CEPH_SNAPDIR);
+		BUG_ON(!req->r_dentry);
+		dout(" linking snapped dir %p to dn %p\n", in, req->r_dentry);
+		ceph_dir_clear_ordered(dir);
+		ihold(in);
+		err = splice_dentry(&req->r_dentry, in);
+		if (err < 0)
+			goto done;
+	} else if (rinfo->head->is_dentry && req->r_dentry) {
+		/* parent inode is not locked, be carefull */
+		struct ceph_vino *ptvino = NULL;
+		dvino.ino = le64_to_cpu(rinfo->diri.in->ino);
+		dvino.snap = le64_to_cpu(rinfo->diri.in->snapid);
+		if (rinfo->head->is_target) {
+			tvino.ino = le64_to_cpu(rinfo->targeti.in->ino);
+			tvino.snap = le64_to_cpu(rinfo->targeti.in->snapid);
+			ptvino = &tvino;
+		}
+		update_dentry_lease_careful(req->r_dentry, rinfo->dlease,
+					    session, req->r_request_started,
+					    rinfo->dname, rinfo->dname_len,
+					    &dvino, ptvino);
+	}
+done:
+	dout("fill_trace done err=%d\n", err);
+	return err;
+}
+
+/*
+ * Prepopulate our cache with readdir results, leases, etc.
+ */
+static int readdir_prepopulate_inodes_only(struct ceph_mds_request *req,
+					   struct ceph_mds_session *session)
+{
+	struct ceph_mds_reply_info_parsed *rinfo = &req->r_reply_info;
+	int i, err = 0;
+
+	for (i = 0; i < rinfo->dir_nr; i++) {
+		struct ceph_mds_reply_dir_entry *rde = rinfo->dir_entries + i;
+		struct ceph_vino vino;
+		struct inode *in;
+		int rc;
+
+		vino.ino = le64_to_cpu(rde->inode.in->ino);
+		vino.snap = le64_to_cpu(rde->inode.in->snapid);
+
+		in = ceph_get_inode(req->r_dentry->d_sb, vino);
+		if (IS_ERR(in)) {
+			err = PTR_ERR(in);
+			dout("new_inode badness got %d\n", err);
+			continue;
+		}
+		rc = ceph_fill_inode(in, NULL, &rde->inode, NULL, session,
+				     -1, &req->r_caps_reservation);
+		if (rc < 0) {
+			pr_err("ceph_fill_inode badness on %p got %d\n",
+			       in, rc);
+			err = rc;
+			if (in->i_state & I_NEW) {
+				ihold(in);
+				discard_new_inode(in);
+			}
+		} else if (in->i_state & I_NEW) {
+			unlock_new_inode(in);
+		}
+
+		iput(in);
+	}
+
+	return err;
+}
+
+void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl)
+{
+	if (ctl->page) {
+		kunmap(ctl->page);
+		put_page(ctl->page);
+		ctl->page = NULL;
+	}
+}
+
+static int fill_readdir_cache(struct inode *dir, struct dentry *dn,
+			      struct ceph_readdir_cache_control *ctl,
+			      struct ceph_mds_request *req)
+{
+	struct ceph_inode_info *ci = ceph_inode(dir);
+	unsigned nsize = PAGE_SIZE / sizeof(struct dentry*);
+	unsigned idx = ctl->index % nsize;
+	pgoff_t pgoff = ctl->index / nsize;
+
+	if (!ctl->page || pgoff != page_index(ctl->page)) {
+		ceph_readdir_cache_release(ctl);
+		if (idx == 0)
+			ctl->page = grab_cache_page(&dir->i_data, pgoff);
+		else
+			ctl->page = find_lock_page(&dir->i_data, pgoff);
+		if (!ctl->page) {
+			ctl->index = -1;
+			return idx == 0 ? -ENOMEM : 0;
+		}
+		/* reading/filling the cache are serialized by
+		 * i_rwsem, no need to use page lock */
+		unlock_page(ctl->page);
+		ctl->dentries = kmap(ctl->page);
+		if (idx == 0)
+			memset(ctl->dentries, 0, PAGE_SIZE);
+	}
+
+	if (req->r_dir_release_cnt == atomic64_read(&ci->i_release_count) &&
+	    req->r_dir_ordered_cnt == atomic64_read(&ci->i_ordered_count)) {
+		dout("readdir cache dn %p idx %d\n", dn, ctl->index);
+		ctl->dentries[idx] = dn;
+		ctl->index++;
+	} else {
+		dout("disable readdir cache\n");
+		ctl->index = -1;
+	}
+	return 0;
+}
+
+int ceph_readdir_prepopulate(struct ceph_mds_request *req,
+			     struct ceph_mds_session *session)
+{
+	struct dentry *parent = req->r_dentry;
+	struct ceph_inode_info *ci = ceph_inode(d_inode(parent));
+	struct ceph_mds_reply_info_parsed *rinfo = &req->r_reply_info;
+	struct qstr dname;
+	struct dentry *dn;
+	struct inode *in;
+	int err = 0, skipped = 0, ret, i;
+	u32 frag = le32_to_cpu(req->r_args.readdir.frag);
+	u32 last_hash = 0;
+	u32 fpos_offset;
+	struct ceph_readdir_cache_control cache_ctl = {};
+
+	if (test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags))
+		return readdir_prepopulate_inodes_only(req, session);
+
+	if (rinfo->hash_order) {
+		if (req->r_path2) {
+			last_hash = ceph_str_hash(ci->i_dir_layout.dl_dir_hash,
+						  req->r_path2,
+						  strlen(req->r_path2));
+			last_hash = ceph_frag_value(last_hash);
+		} else if (rinfo->offset_hash) {
+			/* mds understands offset_hash */
+			WARN_ON_ONCE(req->r_readdir_offset != 2);
+			last_hash = le32_to_cpu(req->r_args.readdir.offset_hash);
+		}
+	}
+
+	if (rinfo->dir_dir &&
+	    le32_to_cpu(rinfo->dir_dir->frag) != frag) {
+		dout("readdir_prepopulate got new frag %x -> %x\n",
+		     frag, le32_to_cpu(rinfo->dir_dir->frag));
+		frag = le32_to_cpu(rinfo->dir_dir->frag);
+		if (!rinfo->hash_order)
+			req->r_readdir_offset = 2;
+	}
+
+	if (le32_to_cpu(rinfo->head->op) == CEPH_MDS_OP_LSSNAP) {
+		dout("readdir_prepopulate %d items under SNAPDIR dn %p\n",
+		     rinfo->dir_nr, parent);
+	} else {
+		dout("readdir_prepopulate %d items under dn %p\n",
+		     rinfo->dir_nr, parent);
+		if (rinfo->dir_dir)
+			ceph_fill_dirfrag(d_inode(parent), rinfo->dir_dir);
+
+		if (ceph_frag_is_leftmost(frag) &&
+		    req->r_readdir_offset == 2 &&
+		    !(rinfo->hash_order && last_hash)) {
+			/* note dir version at start of readdir so we can
+			 * tell if any dentries get dropped */
+			req->r_dir_release_cnt =
+				atomic64_read(&ci->i_release_count);
+			req->r_dir_ordered_cnt =
+				atomic64_read(&ci->i_ordered_count);
+			req->r_readdir_cache_idx = 0;
+		}
+	}
+
+	cache_ctl.index = req->r_readdir_cache_idx;
+	fpos_offset = req->r_readdir_offset;
+
+	/* FIXME: release caps/leases if error occurs */
+	for (i = 0; i < rinfo->dir_nr; i++) {
+		struct ceph_mds_reply_dir_entry *rde = rinfo->dir_entries + i;
+		struct ceph_vino tvino;
+
+		dname.name = rde->name;
+		dname.len = rde->name_len;
+		dname.hash = full_name_hash(parent, dname.name, dname.len);
+
+		tvino.ino = le64_to_cpu(rde->inode.in->ino);
+		tvino.snap = le64_to_cpu(rde->inode.in->snapid);
+
+		if (rinfo->hash_order) {
+			u32 hash = ceph_str_hash(ci->i_dir_layout.dl_dir_hash,
+						 rde->name, rde->name_len);
+			hash = ceph_frag_value(hash);
+			if (hash != last_hash)
+				fpos_offset = 2;
+			last_hash = hash;
+			rde->offset = ceph_make_fpos(hash, fpos_offset++, true);
+		} else {
+			rde->offset = ceph_make_fpos(frag, fpos_offset++, false);
+		}
+
+retry_lookup:
+		dn = d_lookup(parent, &dname);
+		dout("d_lookup on parent=%p name=%.*s got %p\n",
+		     parent, dname.len, dname.name, dn);
+
+		if (!dn) {
+			dn = d_alloc(parent, &dname);
+			dout("d_alloc %p '%.*s' = %p\n", parent,
+			     dname.len, dname.name, dn);
+			if (!dn) {
+				dout("d_alloc badness\n");
+				err = -ENOMEM;
+				goto out;
+			}
+		} else if (d_really_is_positive(dn) &&
+			   (ceph_ino(d_inode(dn)) != tvino.ino ||
+			    ceph_snap(d_inode(dn)) != tvino.snap)) {
+			struct ceph_dentry_info *di = ceph_dentry(dn);
+			dout(" dn %p points to wrong inode %p\n",
+			     dn, d_inode(dn));
+
+			spin_lock(&dn->d_lock);
+			if (di->offset > 0 &&
+			    di->lease_shared_gen ==
+			    atomic_read(&ci->i_shared_gen)) {
+				__ceph_dir_clear_ordered(ci);
+				di->offset = 0;
+			}
+			spin_unlock(&dn->d_lock);
+
+			d_delete(dn);
+			dput(dn);
+			goto retry_lookup;
+		}
+
+		/* inode */
+		if (d_really_is_positive(dn)) {
+			in = d_inode(dn);
+		} else {
+			in = ceph_get_inode(parent->d_sb, tvino);
+			if (IS_ERR(in)) {
+				dout("new_inode badness\n");
+				d_drop(dn);
+				dput(dn);
+				err = PTR_ERR(in);
+				goto out;
+			}
+		}
+
+		ret = ceph_fill_inode(in, NULL, &rde->inode, NULL, session,
+				      -1, &req->r_caps_reservation);
+		if (ret < 0) {
+			pr_err("ceph_fill_inode badness on %p\n", in);
+			if (d_really_is_negative(dn)) {
+				if (in->i_state & I_NEW) {
+					ihold(in);
+					discard_new_inode(in);
+				}
+				iput(in);
+			}
+			d_drop(dn);
+			err = ret;
+			goto next_item;
+		}
+		if (in->i_state & I_NEW)
+			unlock_new_inode(in);
+
+		if (d_really_is_negative(dn)) {
+			if (ceph_security_xattr_deadlock(in)) {
+				dout(" skip splicing dn %p to inode %p"
+				     " (security xattr deadlock)\n", dn, in);
+				iput(in);
+				skipped++;
+				goto next_item;
+			}
+
+			err = splice_dentry(&dn, in);
+			if (err < 0)
+				goto next_item;
+		}
+
+		ceph_dentry(dn)->offset = rde->offset;
+
+		update_dentry_lease(d_inode(parent), dn,
+				    rde->lease, req->r_session,
+				    req->r_request_started);
+
+		if (err == 0 && skipped == 0 && cache_ctl.index >= 0) {
+			ret = fill_readdir_cache(d_inode(parent), dn,
+						 &cache_ctl, req);
+			if (ret < 0)
+				err = ret;
+		}
+next_item:
+		dput(dn);
+	}
+out:
+	if (err == 0 && skipped == 0) {
+		set_bit(CEPH_MDS_R_DID_PREPOPULATE, &req->r_req_flags);
+		req->r_readdir_cache_idx = cache_ctl.index;
+	}
+	ceph_readdir_cache_release(&cache_ctl);
+	dout("readdir_prepopulate done\n");
+	return err;
+}
+
+bool ceph_inode_set_size(struct inode *inode, loff_t size)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	bool ret;
+
+	spin_lock(&ci->i_ceph_lock);
+	dout("set_size %p %llu -> %llu\n", inode, i_size_read(inode), size);
+	i_size_write(inode, size);
+	ceph_fscache_update(inode);
+	inode->i_blocks = calc_inode_blocks(size);
+
+	ret = __ceph_should_report_size(ci);
+
+	spin_unlock(&ci->i_ceph_lock);
+
+	return ret;
+}
+
+void ceph_queue_inode_work(struct inode *inode, int work_bit)
+{
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	set_bit(work_bit, &ci->i_work_mask);
+
+	ihold(inode);
+	if (queue_work(fsc->inode_wq, &ci->i_work)) {
+		dout("queue_inode_work %p, mask=%lx\n", inode, ci->i_work_mask);
+	} else {
+		dout("queue_inode_work %p already queued, mask=%lx\n",
+		     inode, ci->i_work_mask);
+		iput(inode);
+	}
+}
+
+static void ceph_do_invalidate_pages(struct inode *inode)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	u32 orig_gen;
+	int check = 0;
+
+	ceph_fscache_invalidate(inode, false);
+
+	mutex_lock(&ci->i_truncate_mutex);
+
+	if (ceph_inode_is_shutdown(inode)) {
+		pr_warn_ratelimited("%s: inode %llx.%llx is shut down\n",
+				    __func__, ceph_vinop(inode));
+		mapping_set_error(inode->i_mapping, -EIO);
+		truncate_pagecache(inode, 0);
+		mutex_unlock(&ci->i_truncate_mutex);
+		goto out;
+	}
+
+	spin_lock(&ci->i_ceph_lock);
+	dout("invalidate_pages %p gen %d revoking %d\n", inode,
+	     ci->i_rdcache_gen, ci->i_rdcache_revoking);
+	if (ci->i_rdcache_revoking != ci->i_rdcache_gen) {
+		if (__ceph_caps_revoking_other(ci, NULL, CEPH_CAP_FILE_CACHE))
+			check = 1;
+		spin_unlock(&ci->i_ceph_lock);
+		mutex_unlock(&ci->i_truncate_mutex);
+		goto out;
+	}
+	orig_gen = ci->i_rdcache_gen;
+	spin_unlock(&ci->i_ceph_lock);
+
+	if (invalidate_inode_pages2(inode->i_mapping) < 0) {
+		pr_err("invalidate_inode_pages2 %llx.%llx failed\n",
+		       ceph_vinop(inode));
+	}
+
+	spin_lock(&ci->i_ceph_lock);
+	if (orig_gen == ci->i_rdcache_gen &&
+	    orig_gen == ci->i_rdcache_revoking) {
+		dout("invalidate_pages %p gen %d successful\n", inode,
+		     ci->i_rdcache_gen);
+		ci->i_rdcache_revoking--;
+		check = 1;
+	} else {
+		dout("invalidate_pages %p gen %d raced, now %d revoking %d\n",
+		     inode, orig_gen, ci->i_rdcache_gen,
+		     ci->i_rdcache_revoking);
+		if (__ceph_caps_revoking_other(ci, NULL, CEPH_CAP_FILE_CACHE))
+			check = 1;
+	}
+	spin_unlock(&ci->i_ceph_lock);
+	mutex_unlock(&ci->i_truncate_mutex);
+out:
+	if (check)
+		ceph_check_caps(ci, 0, NULL);
+}
+
+/*
+ * Make sure any pending truncation is applied before doing anything
+ * that may depend on it.
+ */
+void __ceph_do_pending_vmtruncate(struct inode *inode)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	u64 to;
+	int wrbuffer_refs, finish = 0;
+
+	mutex_lock(&ci->i_truncate_mutex);
+retry:
+	spin_lock(&ci->i_ceph_lock);
+	if (ci->i_truncate_pending == 0) {
+		dout("__do_pending_vmtruncate %p none pending\n", inode);
+		spin_unlock(&ci->i_ceph_lock);
+		mutex_unlock(&ci->i_truncate_mutex);
+		return;
+	}
+
+	/*
+	 * make sure any dirty snapped pages are flushed before we
+	 * possibly truncate them.. so write AND block!
+	 */
+	if (ci->i_wrbuffer_ref_head < ci->i_wrbuffer_ref) {
+		spin_unlock(&ci->i_ceph_lock);
+		dout("__do_pending_vmtruncate %p flushing snaps first\n",
+		     inode);
+		filemap_write_and_wait_range(&inode->i_data, 0,
+					     inode->i_sb->s_maxbytes);
+		goto retry;
+	}
+
+	/* there should be no reader or writer */
+	WARN_ON_ONCE(ci->i_rd_ref || ci->i_wr_ref);
+
+	to = ci->i_truncate_size;
+	wrbuffer_refs = ci->i_wrbuffer_ref;
+	dout("__do_pending_vmtruncate %p (%d) to %lld\n", inode,
+	     ci->i_truncate_pending, to);
+	spin_unlock(&ci->i_ceph_lock);
+
+	ceph_fscache_resize(inode, to);
+	truncate_pagecache(inode, to);
+
+	spin_lock(&ci->i_ceph_lock);
+	if (to == ci->i_truncate_size) {
+		ci->i_truncate_pending = 0;
+		finish = 1;
+	}
+	spin_unlock(&ci->i_ceph_lock);
+	if (!finish)
+		goto retry;
+
+	mutex_unlock(&ci->i_truncate_mutex);
+
+	if (wrbuffer_refs == 0)
+		ceph_check_caps(ci, 0, NULL);
+
+	wake_up_all(&ci->i_cap_wq);
+}
+
+static void ceph_inode_work(struct work_struct *work)
+{
+	struct ceph_inode_info *ci = container_of(work, struct ceph_inode_info,
+						 i_work);
+	struct inode *inode = &ci->netfs.inode;
+
+	if (test_and_clear_bit(CEPH_I_WORK_WRITEBACK, &ci->i_work_mask)) {
+		dout("writeback %p\n", inode);
+		filemap_fdatawrite(&inode->i_data);
+	}
+	if (test_and_clear_bit(CEPH_I_WORK_INVALIDATE_PAGES, &ci->i_work_mask))
+		ceph_do_invalidate_pages(inode);
+
+	if (test_and_clear_bit(CEPH_I_WORK_VMTRUNCATE, &ci->i_work_mask))
+		__ceph_do_pending_vmtruncate(inode);
+
+	if (test_and_clear_bit(CEPH_I_WORK_CHECK_CAPS, &ci->i_work_mask))
+		ceph_check_caps(ci, 0, NULL);
+
+	if (test_and_clear_bit(CEPH_I_WORK_FLUSH_SNAPS, &ci->i_work_mask))
+		ceph_flush_snaps(ci, NULL);
+
+	iput(inode);
+}
+
+/*
+ * symlinks
+ */
+static const struct inode_operations ceph_symlink_iops = {
+	.get_link = simple_get_link,
+	.setattr = ceph_setattr,
+	.getattr = ceph_getattr,
+	.listxattr = ceph_listxattr,
+};
+
+int __ceph_setattr(struct inode *inode, struct iattr *attr)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	unsigned int ia_valid = attr->ia_valid;
+	struct ceph_mds_request *req;
+	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
+	struct ceph_cap_flush *prealloc_cf;
+	int issued;
+	int release = 0, dirtied = 0;
+	int mask = 0;
+	int err = 0;
+	int inode_dirty_flags = 0;
+	bool lock_snap_rwsem = false;
+
+	prealloc_cf = ceph_alloc_cap_flush();
+	if (!prealloc_cf)
+		return -ENOMEM;
+
+	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETATTR,
+				       USE_AUTH_MDS);
+	if (IS_ERR(req)) {
+		ceph_free_cap_flush(prealloc_cf);
+		return PTR_ERR(req);
+	}
+
+	spin_lock(&ci->i_ceph_lock);
+	issued = __ceph_caps_issued(ci, NULL);
+
+	if (!ci->i_head_snapc &&
+	    (issued & (CEPH_CAP_ANY_EXCL | CEPH_CAP_FILE_WR))) {
+		lock_snap_rwsem = true;
+		if (!down_read_trylock(&mdsc->snap_rwsem)) {
+			spin_unlock(&ci->i_ceph_lock);
+			down_read(&mdsc->snap_rwsem);
+			spin_lock(&ci->i_ceph_lock);
+			issued = __ceph_caps_issued(ci, NULL);
+		}
+	}
+
+	dout("setattr %p issued %s\n", inode, ceph_cap_string(issued));
+
+	if (ia_valid & ATTR_UID) {
+		dout("setattr %p uid %d -> %d\n", inode,
+		     from_kuid(&init_user_ns, inode->i_uid),
+		     from_kuid(&init_user_ns, attr->ia_uid));
+		if (issued & CEPH_CAP_AUTH_EXCL) {
+			inode->i_uid = attr->ia_uid;
+			dirtied |= CEPH_CAP_AUTH_EXCL;
+		} else if ((issued & CEPH_CAP_AUTH_SHARED) == 0 ||
+			   !uid_eq(attr->ia_uid, inode->i_uid)) {
+			req->r_args.setattr.uid = cpu_to_le32(
+				from_kuid(&init_user_ns, attr->ia_uid));
+			mask |= CEPH_SETATTR_UID;
+			release |= CEPH_CAP_AUTH_SHARED;
+		}
+	}
+	if (ia_valid & ATTR_GID) {
+		dout("setattr %p gid %d -> %d\n", inode,
+		     from_kgid(&init_user_ns, inode->i_gid),
+		     from_kgid(&init_user_ns, attr->ia_gid));
+		if (issued & CEPH_CAP_AUTH_EXCL) {
+			inode->i_gid = attr->ia_gid;
+			dirtied |= CEPH_CAP_AUTH_EXCL;
+		} else if ((issued & CEPH_CAP_AUTH_SHARED) == 0 ||
+			   !gid_eq(attr->ia_gid, inode->i_gid)) {
+			req->r_args.setattr.gid = cpu_to_le32(
+				from_kgid(&init_user_ns, attr->ia_gid));
+			mask |= CEPH_SETATTR_GID;
+			release |= CEPH_CAP_AUTH_SHARED;
+		}
+	}
+	if (ia_valid & ATTR_MODE) {
+		dout("setattr %p mode 0%o -> 0%o\n", inode, inode->i_mode,
+		     attr->ia_mode);
+		if (issued & CEPH_CAP_AUTH_EXCL) {
+			inode->i_mode = attr->ia_mode;
+			dirtied |= CEPH_CAP_AUTH_EXCL;
+		} else if ((issued & CEPH_CAP_AUTH_SHARED) == 0 ||
+			   attr->ia_mode != inode->i_mode) {
+			inode->i_mode = attr->ia_mode;
+			req->r_args.setattr.mode = cpu_to_le32(attr->ia_mode);
+			mask |= CEPH_SETATTR_MODE;
+			release |= CEPH_CAP_AUTH_SHARED;
+		}
+	}
+
+	if (ia_valid & ATTR_ATIME) {
+		dout("setattr %p atime %lld.%ld -> %lld.%ld\n", inode,
+		     inode->i_atime.tv_sec, inode->i_atime.tv_nsec,
+		     attr->ia_atime.tv_sec, attr->ia_atime.tv_nsec);
+		if (issued & CEPH_CAP_FILE_EXCL) {
+			ci->i_time_warp_seq++;
+			inode->i_atime = attr->ia_atime;
+			dirtied |= CEPH_CAP_FILE_EXCL;
+		} else if ((issued & CEPH_CAP_FILE_WR) &&
+			   timespec64_compare(&inode->i_atime,
+					    &attr->ia_atime) < 0) {
+			inode->i_atime = attr->ia_atime;
+			dirtied |= CEPH_CAP_FILE_WR;
+		} else if ((issued & CEPH_CAP_FILE_SHARED) == 0 ||
+			   !timespec64_equal(&inode->i_atime, &attr->ia_atime)) {
+			ceph_encode_timespec64(&req->r_args.setattr.atime,
+					       &attr->ia_atime);
+			mask |= CEPH_SETATTR_ATIME;
+			release |= CEPH_CAP_FILE_SHARED |
+				   CEPH_CAP_FILE_RD | CEPH_CAP_FILE_WR;
+		}
+	}
+	if (ia_valid & ATTR_SIZE) {
+		loff_t isize = i_size_read(inode);
+
+		dout("setattr %p size %lld -> %lld\n", inode, isize, attr->ia_size);
+		if ((issued & CEPH_CAP_FILE_EXCL) && attr->ia_size >= isize) {
+			if (attr->ia_size > isize) {
+				i_size_write(inode, attr->ia_size);
+				inode->i_blocks = calc_inode_blocks(attr->ia_size);
+				ci->i_reported_size = attr->ia_size;
+				dirtied |= CEPH_CAP_FILE_EXCL;
+				ia_valid |= ATTR_MTIME;
+			}
+		} else if ((issued & CEPH_CAP_FILE_SHARED) == 0 ||
+			   attr->ia_size != isize) {
+			req->r_args.setattr.size = cpu_to_le64(attr->ia_size);
+			req->r_args.setattr.old_size = cpu_to_le64(isize);
+			mask |= CEPH_SETATTR_SIZE;
+			release |= CEPH_CAP_FILE_SHARED | CEPH_CAP_FILE_EXCL |
+				   CEPH_CAP_FILE_RD | CEPH_CAP_FILE_WR;
+		}
+	}
+	if (ia_valid & ATTR_MTIME) {
+		dout("setattr %p mtime %lld.%ld -> %lld.%ld\n", inode,
+		     inode->i_mtime.tv_sec, inode->i_mtime.tv_nsec,
+		     attr->ia_mtime.tv_sec, attr->ia_mtime.tv_nsec);
+		if (issued & CEPH_CAP_FILE_EXCL) {
+			ci->i_time_warp_seq++;
+			inode->i_mtime = attr->ia_mtime;
+			dirtied |= CEPH_CAP_FILE_EXCL;
+		} else if ((issued & CEPH_CAP_FILE_WR) &&
+			   timespec64_compare(&inode->i_mtime,
+					    &attr->ia_mtime) < 0) {
+			inode->i_mtime = attr->ia_mtime;
+			dirtied |= CEPH_CAP_FILE_WR;
+		} else if ((issued & CEPH_CAP_FILE_SHARED) == 0 ||
+			   !timespec64_equal(&inode->i_mtime, &attr->ia_mtime)) {
+			ceph_encode_timespec64(&req->r_args.setattr.mtime,
+					       &attr->ia_mtime);
+			mask |= CEPH_SETATTR_MTIME;
+			release |= CEPH_CAP_FILE_SHARED |
+				   CEPH_CAP_FILE_RD | CEPH_CAP_FILE_WR;
+		}
+	}
+
+	/* these do nothing */
+	if (ia_valid & ATTR_CTIME) {
+		bool only = (ia_valid & (ATTR_SIZE|ATTR_MTIME|ATTR_ATIME|
+					 ATTR_MODE|ATTR_UID|ATTR_GID)) == 0;
+		dout("setattr %p ctime %lld.%ld -> %lld.%ld (%s)\n", inode,
+		     inode->i_ctime.tv_sec, inode->i_ctime.tv_nsec,
+		     attr->ia_ctime.tv_sec, attr->ia_ctime.tv_nsec,
+		     only ? "ctime only" : "ignored");
+		if (only) {
+			/*
+			 * if kernel wants to dirty ctime but nothing else,
+			 * we need to choose a cap to dirty under, or do
+			 * a almost-no-op setattr
+			 */
+			if (issued & CEPH_CAP_AUTH_EXCL)
+				dirtied |= CEPH_CAP_AUTH_EXCL;
+			else if (issued & CEPH_CAP_FILE_EXCL)
+				dirtied |= CEPH_CAP_FILE_EXCL;
+			else if (issued & CEPH_CAP_XATTR_EXCL)
+				dirtied |= CEPH_CAP_XATTR_EXCL;
+			else
+				mask |= CEPH_SETATTR_CTIME;
+		}
+	}
+	if (ia_valid & ATTR_FILE)
+		dout("setattr %p ATTR_FILE ... hrm!\n", inode);
+
+	if (dirtied) {
+		inode_dirty_flags = __ceph_mark_dirty_caps(ci, dirtied,
+							   &prealloc_cf);
+		inode->i_ctime = attr->ia_ctime;
+		inode_inc_iversion_raw(inode);
+	}
+
+	release &= issued;
+	spin_unlock(&ci->i_ceph_lock);
+	if (lock_snap_rwsem)
+		up_read(&mdsc->snap_rwsem);
+
+	if (inode_dirty_flags)
+		__mark_inode_dirty(inode, inode_dirty_flags);
+
+	if (mask) {
+		req->r_inode = inode;
+		ihold(inode);
+		req->r_inode_drop = release;
+		req->r_args.setattr.mask = cpu_to_le32(mask);
+		req->r_num_caps = 1;
+		req->r_stamp = attr->ia_ctime;
+		err = ceph_mdsc_do_request(mdsc, NULL, req);
+	}
+	dout("setattr %p result=%d (%s locally, %d remote)\n", inode, err,
+	     ceph_cap_string(dirtied), mask);
+
+	ceph_mdsc_put_request(req);
+	ceph_free_cap_flush(prealloc_cf);
+
+	if (err >= 0 && (mask & CEPH_SETATTR_SIZE))
+		__ceph_do_pending_vmtruncate(inode);
+
+	return err;
+}
+
+/*
+ * setattr
+ */
+int ceph_setattr(struct user_namespace *mnt_userns, struct dentry *dentry,
+		 struct iattr *attr)
+{
+	struct inode *inode = d_inode(dentry);
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	int err;
+
+	if (ceph_snap(inode) != CEPH_NOSNAP)
+		return -EROFS;
+
+	if (ceph_inode_is_shutdown(inode))
+		return -ESTALE;
+
+	err = setattr_prepare(&init_user_ns, dentry, attr);
+	if (err != 0)
+		return err;
+
+	if ((attr->ia_valid & ATTR_SIZE) &&
+	    attr->ia_size > max(i_size_read(inode), fsc->max_file_size))
+		return -EFBIG;
+
+	if ((attr->ia_valid & ATTR_SIZE) &&
+	    ceph_quota_is_max_bytes_exceeded(inode, attr->ia_size))
+		return -EDQUOT;
+
+	err = __ceph_setattr(inode, attr);
+
+	if (err >= 0 && (attr->ia_valid & ATTR_MODE))
+		err = posix_acl_chmod(&init_user_ns, inode, attr->ia_mode);
+
+	return err;
+}
+
+int ceph_try_to_choose_auth_mds(struct inode *inode, int mask)
+{
+	int issued = ceph_caps_issued(ceph_inode(inode));
+
+	/*
+	 * If any 'x' caps is issued we can just choose the auth MDS
+	 * instead of the random replica MDSes. Because only when the
+	 * Locker is in LOCK_EXEC state will the loner client could
+	 * get the 'x' caps. And if we send the getattr requests to
+	 * any replica MDS it must auth pin and tries to rdlock from
+	 * the auth MDS, and then the auth MDS need to do the Locker
+	 * state transition to LOCK_SYNC. And after that the lock state
+	 * will change back.
+	 *
+	 * This cost much when doing the Locker state transition and
+	 * usually will need to revoke caps from clients.
+	 *
+	 * And for the 'Xs' caps for getxattr we will also choose the
+	 * auth MDS, because the MDS side code is buggy due to setxattr
+	 * won't notify the replica MDSes when the values changed and
+	 * the replica MDS will return the old values. Though we will
+	 * fix it in MDS code, but this still makes sense for old ceph.
+	 */
+	if (((mask & CEPH_CAP_ANY_SHARED) && (issued & CEPH_CAP_ANY_EXCL))
+	    || (mask & (CEPH_STAT_RSTAT | CEPH_STAT_CAP_XATTR)))
+		return USE_AUTH_MDS;
+	else
+		return USE_ANY_MDS;
+}
+
+/*
+ * Verify that we have a lease on the given mask.  If not,
+ * do a getattr against an mds.
+ */
+int __ceph_do_getattr(struct inode *inode, struct page *locked_page,
+		      int mask, bool force)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(inode->i_sb);
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	struct ceph_mds_request *req;
+	int mode;
+	int err;
+
+	if (ceph_snap(inode) == CEPH_SNAPDIR) {
+		dout("do_getattr inode %p SNAPDIR\n", inode);
+		return 0;
+	}
+
+	dout("do_getattr inode %p mask %s mode 0%o\n",
+	     inode, ceph_cap_string(mask), inode->i_mode);
+	if (!force && ceph_caps_issued_mask_metric(ceph_inode(inode), mask, 1))
+			return 0;
+
+	mode = ceph_try_to_choose_auth_mds(inode, mask);
+	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_GETATTR, mode);
+	if (IS_ERR(req))
+		return PTR_ERR(req);
+	req->r_inode = inode;
+	ihold(inode);
+	req->r_num_caps = 1;
+	req->r_args.getattr.mask = cpu_to_le32(mask);
+	req->r_locked_page = locked_page;
+	err = ceph_mdsc_do_request(mdsc, NULL, req);
+	if (locked_page && err == 0) {
+		u64 inline_version = req->r_reply_info.targeti.inline_version;
+		if (inline_version == 0) {
+			/* the reply is supposed to contain inline data */
+			err = -EINVAL;
+		} else if (inline_version == CEPH_INLINE_NONE ||
+			   inline_version == 1) {
+			err = -ENODATA;
+		} else {
+			err = req->r_reply_info.targeti.inline_len;
+		}
+	}
+	ceph_mdsc_put_request(req);
+	dout("do_getattr result=%d\n", err);
+	return err;
+}
+
+int ceph_do_getvxattr(struct inode *inode, const char *name, void *value,
+		      size_t size)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(inode->i_sb);
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	struct ceph_mds_request *req;
+	int mode = USE_AUTH_MDS;
+	int err;
+	char *xattr_value;
+	size_t xattr_value_len;
+
+	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_GETVXATTR, mode);
+	if (IS_ERR(req)) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	req->r_feature_needed = CEPHFS_FEATURE_OP_GETVXATTR;
+	req->r_path2 = kstrdup(name, GFP_NOFS);
+	if (!req->r_path2) {
+		err = -ENOMEM;
+		goto put;
+	}
+
+	ihold(inode);
+	req->r_inode = inode;
+	err = ceph_mdsc_do_request(mdsc, NULL, req);
+	if (err < 0)
+		goto put;
+
+	xattr_value = req->r_reply_info.xattr_info.xattr_value;
+	xattr_value_len = req->r_reply_info.xattr_info.xattr_value_len;
+
+	dout("do_getvxattr xattr_value_len:%zu, size:%zu\n", xattr_value_len, size);
+
+	err = (int)xattr_value_len;
+	if (size == 0)
+		goto put;
+
+	if (xattr_value_len > size) {
+		err = -ERANGE;
+		goto put;
+	}
+
+	memcpy(value, xattr_value, xattr_value_len);
+put:
+	ceph_mdsc_put_request(req);
+out:
+	dout("do_getvxattr result=%d\n", err);
+	return err;
+}
+
+
+/*
+ * Check inode permissions.  We verify we have a valid value for
+ * the AUTH cap, then call the generic handler.
+ */
+int ceph_permission(struct user_namespace *mnt_userns, struct inode *inode,
+		    int mask)
+{
+	int err;
+
+	if (mask & MAY_NOT_BLOCK)
+		return -ECHILD;
+
+	err = ceph_do_getattr(inode, CEPH_CAP_AUTH_SHARED, false);
+
+	if (!err)
+		err = generic_permission(&init_user_ns, inode, mask);
+	return err;
+}
+
+/* Craft a mask of needed caps given a set of requested statx attrs. */
+static int statx_to_caps(u32 want, umode_t mode)
+{
+	int mask = 0;
+
+	if (want & (STATX_MODE|STATX_UID|STATX_GID|STATX_CTIME|STATX_BTIME))
+		mask |= CEPH_CAP_AUTH_SHARED;
+
+	if (want & (STATX_NLINK|STATX_CTIME)) {
+		/*
+		 * The link count for directories depends on inode->i_subdirs,
+		 * and that is only updated when Fs caps are held.
+		 */
+		if (S_ISDIR(mode))
+			mask |= CEPH_CAP_FILE_SHARED;
+		else
+			mask |= CEPH_CAP_LINK_SHARED;
+	}
+
+	if (want & (STATX_ATIME|STATX_MTIME|STATX_CTIME|STATX_SIZE|
+		    STATX_BLOCKS))
+		mask |= CEPH_CAP_FILE_SHARED;
+
+	if (want & (STATX_CTIME))
+		mask |= CEPH_CAP_XATTR_SHARED;
+
+	return mask;
+}
+
+/*
+ * Get all the attributes. If we have sufficient caps for the requested attrs,
+ * then we can avoid talking to the MDS at all.
+ */
+int ceph_getattr(struct user_namespace *mnt_userns, const struct path *path,
+		 struct kstat *stat, u32 request_mask, unsigned int flags)
+{
+	struct inode *inode = d_inode(path->dentry);
+	struct super_block *sb = inode->i_sb;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	u32 valid_mask = STATX_BASIC_STATS;
+	int err = 0;
+
+	if (ceph_inode_is_shutdown(inode))
+		return -ESTALE;
+
+	/* Skip the getattr altogether if we're asked not to sync */
+	if ((flags & AT_STATX_SYNC_TYPE) != AT_STATX_DONT_SYNC) {
+		err = ceph_do_getattr(inode,
+				statx_to_caps(request_mask, inode->i_mode),
+				flags & AT_STATX_FORCE_SYNC);
+		if (err)
+			return err;
+	}
+
+	generic_fillattr(&init_user_ns, inode, stat);
+	stat->ino = ceph_present_inode(inode);
+
+	/*
+	 * btime on newly-allocated inodes is 0, so if this is still set to
+	 * that, then assume that it's not valid.
+	 */
+	if (ci->i_btime.tv_sec || ci->i_btime.tv_nsec) {
+		stat->btime = ci->i_btime;
+		valid_mask |= STATX_BTIME;
+	}
+
+	if (ceph_snap(inode) == CEPH_NOSNAP)
+		stat->dev = sb->s_dev;
+	else
+		stat->dev = ci->i_snapid_map ? ci->i_snapid_map->dev : 0;
+
+	if (S_ISDIR(inode->i_mode)) {
+		if (ceph_test_mount_opt(ceph_sb_to_client(sb), RBYTES)) {
+			stat->size = ci->i_rbytes;
+		} else if (ceph_snap(inode) == CEPH_SNAPDIR) {
+			struct ceph_inode_info *pci;
+			struct ceph_snap_realm *realm;
+			struct inode *parent;
+
+			parent = ceph_lookup_inode(sb, ceph_ino(inode));
+			if (IS_ERR(parent))
+				return PTR_ERR(parent);
+
+			pci = ceph_inode(parent);
+			spin_lock(&pci->i_ceph_lock);
+			realm = pci->i_snap_realm;
+			if (realm)
+				stat->size = realm->num_snaps;
+			else
+				stat->size = 0;
+			spin_unlock(&pci->i_ceph_lock);
+			iput(parent);
+		} else {
+			stat->size = ci->i_files + ci->i_subdirs;
+		}
+		stat->blocks = 0;
+		stat->blksize = 65536;
+		/*
+		 * Some applications rely on the number of st_nlink
+		 * value on directories to be either 0 (if unlinked)
+		 * or 2 + number of subdirectories.
+		 */
+		if (stat->nlink == 1)
+			/* '.' + '..' + subdirs */
+			stat->nlink = 1 + 1 + ci->i_subdirs;
+	}
+
+	stat->result_mask = request_mask & valid_mask;
+	return err;
+}
+
+void ceph_inode_shutdown(struct inode *inode)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct rb_node *p;
+	int iputs = 0;
+	bool invalidate = false;
+
+	spin_lock(&ci->i_ceph_lock);
+	ci->i_ceph_flags |= CEPH_I_SHUTDOWN;
+	p = rb_first(&ci->i_caps);
+	while (p) {
+		struct ceph_cap *cap = rb_entry(p, struct ceph_cap, ci_node);
+
+		p = rb_next(p);
+		iputs += ceph_purge_inode_cap(inode, cap, &invalidate);
+	}
+	spin_unlock(&ci->i_ceph_lock);
+
+	if (invalidate)
+		ceph_queue_invalidate(inode);
+	while (iputs--)
+		iput(inode);
+}
diff --git a/fs/ceph/io.c b/fs/ceph/io.c
new file mode 100644
index 000000000..c456509b3
--- /dev/null
+++ b/fs/ceph/io.c
@@ -0,0 +1,163 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2016 Trond Myklebust
+ * Copyright (c) 2019 Jeff Layton
+ *
+ * I/O and data path helper functionality.
+ *
+ * Heavily borrowed from equivalent code in fs/nfs/io.c
+ */
+
+#include <linux/ceph/ceph_debug.h>
+
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/rwsem.h>
+#include <linux/fs.h>
+
+#include "super.h"
+#include "io.h"
+
+/* Call with exclusively locked inode->i_rwsem */
+static void ceph_block_o_direct(struct ceph_inode_info *ci, struct inode *inode)
+{
+	lockdep_assert_held_write(&inode->i_rwsem);
+
+	if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ODIRECT) {
+		spin_lock(&ci->i_ceph_lock);
+		ci->i_ceph_flags &= ~CEPH_I_ODIRECT;
+		spin_unlock(&ci->i_ceph_lock);
+		inode_dio_wait(inode);
+	}
+}
+
+/**
+ * ceph_start_io_read - declare the file is being used for buffered reads
+ * @inode: file inode
+ *
+ * Declare that a buffered read operation is about to start, and ensure
+ * that we block all direct I/O.
+ * On exit, the function ensures that the CEPH_I_ODIRECT flag is unset,
+ * and holds a shared lock on inode->i_rwsem to ensure that the flag
+ * cannot be changed.
+ * In practice, this means that buffered read operations are allowed to
+ * execute in parallel, thanks to the shared lock, whereas direct I/O
+ * operations need to wait to grab an exclusive lock in order to set
+ * CEPH_I_ODIRECT.
+ * Note that buffered writes and truncates both take a write lock on
+ * inode->i_rwsem, meaning that those are serialised w.r.t. the reads.
+ */
+void
+ceph_start_io_read(struct inode *inode)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+
+	/* Be an optimist! */
+	down_read(&inode->i_rwsem);
+	if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ODIRECT))
+		return;
+	up_read(&inode->i_rwsem);
+	/* Slow path.... */
+	down_write(&inode->i_rwsem);
+	ceph_block_o_direct(ci, inode);
+	downgrade_write(&inode->i_rwsem);
+}
+
+/**
+ * ceph_end_io_read - declare that the buffered read operation is done
+ * @inode: file inode
+ *
+ * Declare that a buffered read operation is done, and release the shared
+ * lock on inode->i_rwsem.
+ */
+void
+ceph_end_io_read(struct inode *inode)
+{
+	up_read(&inode->i_rwsem);
+}
+
+/**
+ * ceph_start_io_write - declare the file is being used for buffered writes
+ * @inode: file inode
+ *
+ * Declare that a buffered write operation is about to start, and ensure
+ * that we block all direct I/O.
+ */
+void
+ceph_start_io_write(struct inode *inode)
+{
+	down_write(&inode->i_rwsem);
+	ceph_block_o_direct(ceph_inode(inode), inode);
+}
+
+/**
+ * ceph_end_io_write - declare that the buffered write operation is done
+ * @inode: file inode
+ *
+ * Declare that a buffered write operation is done, and release the
+ * lock on inode->i_rwsem.
+ */
+void
+ceph_end_io_write(struct inode *inode)
+{
+	up_write(&inode->i_rwsem);
+}
+
+/* Call with exclusively locked inode->i_rwsem */
+static void ceph_block_buffered(struct ceph_inode_info *ci, struct inode *inode)
+{
+	lockdep_assert_held_write(&inode->i_rwsem);
+
+	if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ODIRECT)) {
+		spin_lock(&ci->i_ceph_lock);
+		ci->i_ceph_flags |= CEPH_I_ODIRECT;
+		spin_unlock(&ci->i_ceph_lock);
+		/* FIXME: unmap_mapping_range? */
+		filemap_write_and_wait(inode->i_mapping);
+	}
+}
+
+/**
+ * ceph_start_io_direct - declare the file is being used for direct i/o
+ * @inode: file inode
+ *
+ * Declare that a direct I/O operation is about to start, and ensure
+ * that we block all buffered I/O.
+ * On exit, the function ensures that the CEPH_I_ODIRECT flag is set,
+ * and holds a shared lock on inode->i_rwsem to ensure that the flag
+ * cannot be changed.
+ * In practice, this means that direct I/O operations are allowed to
+ * execute in parallel, thanks to the shared lock, whereas buffered I/O
+ * operations need to wait to grab an exclusive lock in order to clear
+ * CEPH_I_ODIRECT.
+ * Note that buffered writes and truncates both take a write lock on
+ * inode->i_rwsem, meaning that those are serialised w.r.t. O_DIRECT.
+ */
+void
+ceph_start_io_direct(struct inode *inode)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+
+	/* Be an optimist! */
+	down_read(&inode->i_rwsem);
+	if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ODIRECT)
+		return;
+	up_read(&inode->i_rwsem);
+	/* Slow path.... */
+	down_write(&inode->i_rwsem);
+	ceph_block_buffered(ci, inode);
+	downgrade_write(&inode->i_rwsem);
+}
+
+/**
+ * ceph_end_io_direct - declare that the direct i/o operation is done
+ * @inode: file inode
+ *
+ * Declare that a direct I/O operation is done, and release the shared
+ * lock on inode->i_rwsem.
+ */
+void
+ceph_end_io_direct(struct inode *inode)
+{
+	up_read(&inode->i_rwsem);
+}
diff --git a/fs/ceph/io.h b/fs/ceph/io.h
new file mode 100644
index 000000000..fa594cd77
--- /dev/null
+++ b/fs/ceph/io.h
@@ -0,0 +1,12 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _FS_CEPH_IO_H
+#define _FS_CEPH_IO_H
+
+void ceph_start_io_read(struct inode *inode);
+void ceph_end_io_read(struct inode *inode);
+void ceph_start_io_write(struct inode *inode);
+void ceph_end_io_write(struct inode *inode);
+void ceph_start_io_direct(struct inode *inode);
+void ceph_end_io_direct(struct inode *inode);
+
+#endif /* FS_CEPH_IO_H */
diff --git a/fs/ceph/ioctl.c b/fs/ceph/ioctl.c
new file mode 100644
index 000000000..6e061bf62
--- /dev/null
+++ b/fs/ceph/ioctl.c
@@ -0,0 +1,295 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/ceph/ceph_debug.h>
+#include <linux/in.h>
+
+#include "super.h"
+#include "mds_client.h"
+#include "ioctl.h"
+#include <linux/ceph/striper.h>
+
+/*
+ * ioctls
+ */
+
+/*
+ * get and set the file layout
+ */
+static long ceph_ioctl_get_layout(struct file *file, void __user *arg)
+{
+	struct ceph_inode_info *ci = ceph_inode(file_inode(file));
+	struct ceph_ioctl_layout l;
+	int err;
+
+	err = ceph_do_getattr(file_inode(file), CEPH_STAT_CAP_LAYOUT, false);
+	if (!err) {
+		l.stripe_unit = ci->i_layout.stripe_unit;
+		l.stripe_count = ci->i_layout.stripe_count;
+		l.object_size = ci->i_layout.object_size;
+		l.data_pool = ci->i_layout.pool_id;
+		l.preferred_osd = -1;
+		if (copy_to_user(arg, &l, sizeof(l)))
+			return -EFAULT;
+	}
+
+	return err;
+}
+
+static long __validate_layout(struct ceph_mds_client *mdsc,
+			      struct ceph_ioctl_layout *l)
+{
+	int i, err;
+
+	/* validate striping parameters */
+	if ((l->object_size & ~PAGE_MASK) ||
+	    (l->stripe_unit & ~PAGE_MASK) ||
+	    ((unsigned)l->stripe_unit != 0 &&
+	     ((unsigned)l->object_size % (unsigned)l->stripe_unit)))
+		return -EINVAL;
+
+	/* make sure it's a valid data pool */
+	mutex_lock(&mdsc->mutex);
+	err = -EINVAL;
+	for (i = 0; i < mdsc->mdsmap->m_num_data_pg_pools; i++)
+		if (mdsc->mdsmap->m_data_pg_pools[i] == l->data_pool) {
+			err = 0;
+			break;
+		}
+	mutex_unlock(&mdsc->mutex);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+static long ceph_ioctl_set_layout(struct file *file, void __user *arg)
+{
+	struct inode *inode = file_inode(file);
+	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
+	struct ceph_mds_request *req;
+	struct ceph_ioctl_layout l;
+	struct ceph_inode_info *ci = ceph_inode(file_inode(file));
+	struct ceph_ioctl_layout nl;
+	int err;
+
+	if (copy_from_user(&l, arg, sizeof(l)))
+		return -EFAULT;
+
+	/* validate changed params against current layout */
+	err = ceph_do_getattr(file_inode(file), CEPH_STAT_CAP_LAYOUT, false);
+	if (err)
+		return err;
+
+	memset(&nl, 0, sizeof(nl));
+	if (l.stripe_count)
+		nl.stripe_count = l.stripe_count;
+	else
+		nl.stripe_count = ci->i_layout.stripe_count;
+	if (l.stripe_unit)
+		nl.stripe_unit = l.stripe_unit;
+	else
+		nl.stripe_unit = ci->i_layout.stripe_unit;
+	if (l.object_size)
+		nl.object_size = l.object_size;
+	else
+		nl.object_size = ci->i_layout.object_size;
+	if (l.data_pool)
+		nl.data_pool = l.data_pool;
+	else
+		nl.data_pool = ci->i_layout.pool_id;
+
+	/* this is obsolete, and always -1 */
+	nl.preferred_osd = -1;
+
+	err = __validate_layout(mdsc, &nl);
+	if (err)
+		return err;
+
+	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETLAYOUT,
+				       USE_AUTH_MDS);
+	if (IS_ERR(req))
+		return PTR_ERR(req);
+	req->r_inode = inode;
+	ihold(inode);
+	req->r_num_caps = 1;
+
+	req->r_inode_drop = CEPH_CAP_FILE_SHARED | CEPH_CAP_FILE_EXCL;
+
+	req->r_args.setlayout.layout.fl_stripe_unit =
+		cpu_to_le32(l.stripe_unit);
+	req->r_args.setlayout.layout.fl_stripe_count =
+		cpu_to_le32(l.stripe_count);
+	req->r_args.setlayout.layout.fl_object_size =
+		cpu_to_le32(l.object_size);
+	req->r_args.setlayout.layout.fl_pg_pool = cpu_to_le32(l.data_pool);
+
+	err = ceph_mdsc_do_request(mdsc, NULL, req);
+	ceph_mdsc_put_request(req);
+	return err;
+}
+
+/*
+ * Set a layout policy on a directory inode. All items in the tree
+ * rooted at this inode will inherit this layout on creation,
+ * (It doesn't apply retroactively )
+ * unless a subdirectory has its own layout policy.
+ */
+static long ceph_ioctl_set_layout_policy (struct file *file, void __user *arg)
+{
+	struct inode *inode = file_inode(file);
+	struct ceph_mds_request *req;
+	struct ceph_ioctl_layout l;
+	int err;
+	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
+
+	/* copy and validate */
+	if (copy_from_user(&l, arg, sizeof(l)))
+		return -EFAULT;
+
+	err = __validate_layout(mdsc, &l);
+	if (err)
+		return err;
+
+	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETDIRLAYOUT,
+				       USE_AUTH_MDS);
+
+	if (IS_ERR(req))
+		return PTR_ERR(req);
+	req->r_inode = inode;
+	ihold(inode);
+	req->r_num_caps = 1;
+
+	req->r_args.setlayout.layout.fl_stripe_unit =
+			cpu_to_le32(l.stripe_unit);
+	req->r_args.setlayout.layout.fl_stripe_count =
+			cpu_to_le32(l.stripe_count);
+	req->r_args.setlayout.layout.fl_object_size =
+			cpu_to_le32(l.object_size);
+	req->r_args.setlayout.layout.fl_pg_pool =
+			cpu_to_le32(l.data_pool);
+
+	err = ceph_mdsc_do_request(mdsc, inode, req);
+	ceph_mdsc_put_request(req);
+	return err;
+}
+
+/*
+ * Return object name, size/offset information, and location (OSD
+ * number, network address) for a given file offset.
+ */
+static long ceph_ioctl_get_dataloc(struct file *file, void __user *arg)
+{
+	struct ceph_ioctl_dataloc dl;
+	struct inode *inode = file_inode(file);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_osd_client *osdc =
+		&ceph_sb_to_client(inode->i_sb)->client->osdc;
+	struct ceph_object_locator oloc;
+	CEPH_DEFINE_OID_ONSTACK(oid);
+	u32 xlen;
+	u64 tmp;
+	struct ceph_pg pgid;
+	int r;
+
+	/* copy and validate */
+	if (copy_from_user(&dl, arg, sizeof(dl)))
+		return -EFAULT;
+
+	down_read(&osdc->lock);
+	ceph_calc_file_object_mapping(&ci->i_layout, dl.file_offset, 1,
+				      &dl.object_no, &dl.object_offset, &xlen);
+	dl.file_offset -= dl.object_offset;
+	dl.object_size = ci->i_layout.object_size;
+	dl.block_size = ci->i_layout.stripe_unit;
+
+	/* block_offset = object_offset % block_size */
+	tmp = dl.object_offset;
+	dl.block_offset = do_div(tmp, dl.block_size);
+
+	snprintf(dl.object_name, sizeof(dl.object_name), "%llx.%08llx",
+		 ceph_ino(inode), dl.object_no);
+
+	oloc.pool = ci->i_layout.pool_id;
+	oloc.pool_ns = ceph_try_get_string(ci->i_layout.pool_ns);
+	ceph_oid_printf(&oid, "%s", dl.object_name);
+
+	r = ceph_object_locator_to_pg(osdc->osdmap, &oid, &oloc, &pgid);
+
+	ceph_oloc_destroy(&oloc);
+	if (r < 0) {
+		up_read(&osdc->lock);
+		return r;
+	}
+
+	dl.osd = ceph_pg_to_acting_primary(osdc->osdmap, &pgid);
+	if (dl.osd >= 0) {
+		struct ceph_entity_addr *a =
+			ceph_osd_addr(osdc->osdmap, dl.osd);
+		if (a)
+			memcpy(&dl.osd_addr, &a->in_addr, sizeof(dl.osd_addr));
+	} else {
+		memset(&dl.osd_addr, 0, sizeof(dl.osd_addr));
+	}
+	up_read(&osdc->lock);
+
+	/* send result back to user */
+	if (copy_to_user(arg, &dl, sizeof(dl)))
+		return -EFAULT;
+
+	return 0;
+}
+
+static long ceph_ioctl_lazyio(struct file *file)
+{
+	struct ceph_file_info *fi = file->private_data;
+	struct inode *inode = file_inode(file);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;
+
+	if ((fi->fmode & CEPH_FILE_MODE_LAZY) == 0) {
+		spin_lock(&ci->i_ceph_lock);
+		fi->fmode |= CEPH_FILE_MODE_LAZY;
+		ci->i_nr_by_mode[ffs(CEPH_FILE_MODE_LAZY)]++;
+		__ceph_touch_fmode(ci, mdsc, fi->fmode);
+		spin_unlock(&ci->i_ceph_lock);
+		dout("ioctl_layzio: file %p marked lazy\n", file);
+
+		ceph_check_caps(ci, 0, NULL);
+	} else {
+		dout("ioctl_layzio: file %p already lazy\n", file);
+	}
+	return 0;
+}
+
+static long ceph_ioctl_syncio(struct file *file)
+{
+	struct ceph_file_info *fi = file->private_data;
+
+	fi->flags |= CEPH_F_SYNC;
+	return 0;
+}
+
+long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
+{
+	dout("ioctl file %p cmd %u arg %lu\n", file, cmd, arg);
+	switch (cmd) {
+	case CEPH_IOC_GET_LAYOUT:
+		return ceph_ioctl_get_layout(file, (void __user *)arg);
+
+	case CEPH_IOC_SET_LAYOUT:
+		return ceph_ioctl_set_layout(file, (void __user *)arg);
+
+	case CEPH_IOC_SET_LAYOUT_POLICY:
+		return ceph_ioctl_set_layout_policy(file, (void __user *)arg);
+
+	case CEPH_IOC_GET_DATALOC:
+		return ceph_ioctl_get_dataloc(file, (void __user *)arg);
+
+	case CEPH_IOC_LAZYIO:
+		return ceph_ioctl_lazyio(file);
+
+	case CEPH_IOC_SYNCIO:
+		return ceph_ioctl_syncio(file);
+	}
+
+	return -ENOTTY;
+}
diff --git a/fs/ceph/ioctl.h b/fs/ceph/ioctl.h
new file mode 100644
index 000000000..51f7f1d39
--- /dev/null
+++ b/fs/ceph/ioctl.h
@@ -0,0 +1,101 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef FS_CEPH_IOCTL_H
+#define FS_CEPH_IOCTL_H
+
+#include <linux/ioctl.h>
+#include <linux/types.h>
+
+#define CEPH_IOCTL_MAGIC 0x97
+
+/*
+ * CEPH_IOC_GET_LAYOUT - get file layout or dir layout policy
+ * CEPH_IOC_SET_LAYOUT - set file layout
+ * CEPH_IOC_SET_LAYOUT_POLICY - set dir layout policy
+ *
+ * The file layout specifies how file data is striped over objects in
+ * the distributed object store, which object pool they belong to (if
+ * it differs from the default), and an optional 'preferred osd' to
+ * store them on.
+ *
+ * Files get a new layout based on the policy set on the containing
+ * directory or one of its ancestors.  The GET_LAYOUT ioctl will let
+ * you examine the layout for a file or the policy on a directory.
+ *
+ * SET_LAYOUT will let you set a layout on a newly created file.  This
+ * only works immediately after the file is created and before any
+ * data is written to it.
+ *
+ * SET_LAYOUT_POLICY will let you set a layout policy (default layout)
+ * on a directory that will apply to any new files created in that
+ * directory (or any child directory that doesn't specify a layout of
+ * its own).
+ */
+
+/* use u64 to align sanely on all archs */
+struct ceph_ioctl_layout {
+	__u64 stripe_unit, stripe_count, object_size;
+	__u64 data_pool;
+
+	/* obsolete.  new values ignored, always return -1 */
+	__s64 preferred_osd;
+};
+
+#define CEPH_IOC_GET_LAYOUT _IOR(CEPH_IOCTL_MAGIC, 1,		\
+				   struct ceph_ioctl_layout)
+#define CEPH_IOC_SET_LAYOUT _IOW(CEPH_IOCTL_MAGIC, 2,		\
+				   struct ceph_ioctl_layout)
+#define CEPH_IOC_SET_LAYOUT_POLICY _IOW(CEPH_IOCTL_MAGIC, 5,	\
+				   struct ceph_ioctl_layout)
+
+/*
+ * CEPH_IOC_GET_DATALOC - get location of file data in the cluster
+ *
+ * Extract identity, address of the OSD and object storing a given
+ * file offset.
+ */
+struct ceph_ioctl_dataloc {
+	__u64 file_offset;           /* in+out: file offset */
+	__u64 object_offset;         /* out: offset in object */
+	__u64 object_no;             /* out: object # */
+	__u64 object_size;           /* out: object size */
+	char object_name[64];        /* out: object name */
+	__u64 block_offset;          /* out: offset in block */
+	__u64 block_size;            /* out: block length */
+	__s64 osd;                   /* out: osd # */
+	struct sockaddr_storage osd_addr; /* out: osd address */
+};
+
+#define CEPH_IOC_GET_DATALOC _IOWR(CEPH_IOCTL_MAGIC, 3,	\
+				   struct ceph_ioctl_dataloc)
+
+/*
+ * CEPH_IOC_LAZYIO - relax consistency
+ *
+ * Normally Ceph switches to synchronous IO when multiple clients have
+ * the file open (and or more for write).  Reads and writes bypass the
+ * page cache and go directly to the OSD.  Setting this flag on a file
+ * descriptor will allow buffered IO for this file in cases where the
+ * application knows it won't interfere with other nodes (or doesn't
+ * care).
+ */
+#define CEPH_IOC_LAZYIO _IO(CEPH_IOCTL_MAGIC, 4)
+
+/*
+ * CEPH_IOC_SYNCIO - force synchronous IO
+ *
+ * This ioctl sets a file flag that forces the synchronous IO that
+ * bypasses the page cache, even if it is not necessary.  This is
+ * essentially the opposite behavior of IOC_LAZYIO.  This forces the
+ * same read/write path as a file opened by multiple clients when one
+ * or more of those clients is opened for write.
+ *
+ * Note that this type of sync IO takes a different path than a file
+ * opened with O_SYNC/D_SYNC (writes hit the page cache and are
+ * immediately flushed on page boundaries).  It is very similar to
+ * O_DIRECT (writes bypass the page cache) excep that O_DIRECT writes
+ * are not copied (user page must remain stable) and O_DIRECT writes
+ * have alignment restrictions (on the buffer and file offset).
+ */
+#define CEPH_IOC_SYNCIO _IO(CEPH_IOCTL_MAGIC, 5)
+
+#endif
diff --git a/fs/ceph/locks.c b/fs/ceph/locks.c
new file mode 100644
index 000000000..b191426bf
--- /dev/null
+++ b/fs/ceph/locks.c
@@ -0,0 +1,493 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/ceph/ceph_debug.h>
+
+#include <linux/file.h>
+#include <linux/namei.h>
+#include <linux/random.h>
+
+#include "super.h"
+#include "mds_client.h"
+#include <linux/ceph/pagelist.h>
+
+static u64 lock_secret;
+static int ceph_lock_wait_for_completion(struct ceph_mds_client *mdsc,
+                                         struct ceph_mds_request *req);
+
+static inline u64 secure_addr(void *addr)
+{
+	u64 v = lock_secret ^ (u64)(unsigned long)addr;
+	/*
+	 * Set the most significant bit, so that MDS knows the 'owner'
+	 * is sufficient to identify the owner of lock. (old code uses
+	 * both 'owner' and 'pid')
+	 */
+	v |= (1ULL << 63);
+	return v;
+}
+
+void __init ceph_flock_init(void)
+{
+	get_random_bytes(&lock_secret, sizeof(lock_secret));
+}
+
+static void ceph_fl_copy_lock(struct file_lock *dst, struct file_lock *src)
+{
+	struct inode *inode = file_inode(dst->fl_file);
+	atomic_inc(&ceph_inode(inode)->i_filelock_ref);
+}
+
+static void ceph_fl_release_lock(struct file_lock *fl)
+{
+	struct inode *inode = file_inode(fl->fl_file);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	if (atomic_dec_and_test(&ci->i_filelock_ref)) {
+		/* clear error when all locks are released */
+		spin_lock(&ci->i_ceph_lock);
+		ci->i_ceph_flags &= ~CEPH_I_ERROR_FILELOCK;
+		spin_unlock(&ci->i_ceph_lock);
+	}
+}
+
+static const struct file_lock_operations ceph_fl_lock_ops = {
+	.fl_copy_lock = ceph_fl_copy_lock,
+	.fl_release_private = ceph_fl_release_lock,
+};
+
+/*
+ * Implement fcntl and flock locking functions.
+ */
+static int ceph_lock_message(u8 lock_type, u16 operation, struct inode *inode,
+			     int cmd, u8 wait, struct file_lock *fl)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
+	struct ceph_mds_request *req;
+	int err;
+	u64 length = 0;
+	u64 owner;
+
+	if (operation == CEPH_MDS_OP_SETFILELOCK) {
+		/*
+		 * increasing i_filelock_ref closes race window between
+		 * handling request reply and adding file_lock struct to
+		 * inode. Otherwise, auth caps may get trimmed in the
+		 * window. Caller function will decrease the counter.
+		 */
+		fl->fl_ops = &ceph_fl_lock_ops;
+		fl->fl_ops->fl_copy_lock(fl, NULL);
+	}
+
+	if (operation != CEPH_MDS_OP_SETFILELOCK || cmd == CEPH_LOCK_UNLOCK)
+		wait = 0;
+
+	req = ceph_mdsc_create_request(mdsc, operation, USE_AUTH_MDS);
+	if (IS_ERR(req))
+		return PTR_ERR(req);
+	req->r_inode = inode;
+	ihold(inode);
+	req->r_num_caps = 1;
+
+	/* mds requires start and length rather than start and end */
+	if (LLONG_MAX == fl->fl_end)
+		length = 0;
+	else
+		length = fl->fl_end - fl->fl_start + 1;
+
+	owner = secure_addr(fl->fl_owner);
+
+	dout("ceph_lock_message: rule: %d, op: %d, owner: %llx, pid: %llu, "
+	     "start: %llu, length: %llu, wait: %d, type: %d\n", (int)lock_type,
+	     (int)operation, owner, (u64)fl->fl_pid, fl->fl_start, length,
+	     wait, fl->fl_type);
+
+	req->r_args.filelock_change.rule = lock_type;
+	req->r_args.filelock_change.type = cmd;
+	req->r_args.filelock_change.owner = cpu_to_le64(owner);
+	req->r_args.filelock_change.pid = cpu_to_le64((u64)fl->fl_pid);
+	req->r_args.filelock_change.start = cpu_to_le64(fl->fl_start);
+	req->r_args.filelock_change.length = cpu_to_le64(length);
+	req->r_args.filelock_change.wait = wait;
+
+	err = ceph_mdsc_submit_request(mdsc, inode, req);
+	if (!err)
+		err = ceph_mdsc_wait_request(mdsc, req, wait ?
+					ceph_lock_wait_for_completion : NULL);
+	if (!err && operation == CEPH_MDS_OP_GETFILELOCK) {
+		fl->fl_pid = -le64_to_cpu(req->r_reply_info.filelock_reply->pid);
+		if (CEPH_LOCK_SHARED == req->r_reply_info.filelock_reply->type)
+			fl->fl_type = F_RDLCK;
+		else if (CEPH_LOCK_EXCL == req->r_reply_info.filelock_reply->type)
+			fl->fl_type = F_WRLCK;
+		else
+			fl->fl_type = F_UNLCK;
+
+		fl->fl_start = le64_to_cpu(req->r_reply_info.filelock_reply->start);
+		length = le64_to_cpu(req->r_reply_info.filelock_reply->start) +
+						 le64_to_cpu(req->r_reply_info.filelock_reply->length);
+		if (length >= 1)
+			fl->fl_end = length -1;
+		else
+			fl->fl_end = 0;
+
+	}
+	ceph_mdsc_put_request(req);
+	dout("ceph_lock_message: rule: %d, op: %d, pid: %llu, start: %llu, "
+	     "length: %llu, wait: %d, type: %d, err code %d\n", (int)lock_type,
+	     (int)operation, (u64)fl->fl_pid, fl->fl_start,
+	     length, wait, fl->fl_type, err);
+	return err;
+}
+
+static int ceph_lock_wait_for_completion(struct ceph_mds_client *mdsc,
+                                         struct ceph_mds_request *req)
+{
+	struct ceph_mds_request *intr_req;
+	struct inode *inode = req->r_inode;
+	int err, lock_type;
+
+	BUG_ON(req->r_op != CEPH_MDS_OP_SETFILELOCK);
+	if (req->r_args.filelock_change.rule == CEPH_LOCK_FCNTL)
+		lock_type = CEPH_LOCK_FCNTL_INTR;
+	else if (req->r_args.filelock_change.rule == CEPH_LOCK_FLOCK)
+		lock_type = CEPH_LOCK_FLOCK_INTR;
+	else
+		BUG_ON(1);
+	BUG_ON(req->r_args.filelock_change.type == CEPH_LOCK_UNLOCK);
+
+	err = wait_for_completion_interruptible(&req->r_completion);
+	if (!err)
+		return 0;
+
+	dout("ceph_lock_wait_for_completion: request %llu was interrupted\n",
+	     req->r_tid);
+
+	mutex_lock(&mdsc->mutex);
+	if (test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags)) {
+		err = 0;
+	} else {
+		/*
+		 * ensure we aren't running concurrently with
+		 * ceph_fill_trace or ceph_readdir_prepopulate, which
+		 * rely on locks (dir mutex) held by our caller.
+		 */
+		mutex_lock(&req->r_fill_mutex);
+		req->r_err = err;
+		set_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags);
+		mutex_unlock(&req->r_fill_mutex);
+
+		if (!req->r_session) {
+			// haven't sent the request
+			err = 0;
+		}
+	}
+	mutex_unlock(&mdsc->mutex);
+	if (!err)
+		return 0;
+
+	intr_req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETFILELOCK,
+					    USE_AUTH_MDS);
+	if (IS_ERR(intr_req))
+		return PTR_ERR(intr_req);
+
+	intr_req->r_inode = inode;
+	ihold(inode);
+	intr_req->r_num_caps = 1;
+
+	intr_req->r_args.filelock_change = req->r_args.filelock_change;
+	intr_req->r_args.filelock_change.rule = lock_type;
+	intr_req->r_args.filelock_change.type = CEPH_LOCK_UNLOCK;
+
+	err = ceph_mdsc_do_request(mdsc, inode, intr_req);
+	ceph_mdsc_put_request(intr_req);
+
+	if (err && err != -ERESTARTSYS)
+		return err;
+
+	wait_for_completion_killable(&req->r_safe_completion);
+	return 0;
+}
+
+static int try_unlock_file(struct file *file, struct file_lock *fl)
+{
+	int err;
+	unsigned int orig_flags = fl->fl_flags;
+	fl->fl_flags |= FL_EXISTS;
+	err = locks_lock_file_wait(file, fl);
+	fl->fl_flags = orig_flags;
+	if (err == -ENOENT) {
+		if (!(orig_flags & FL_EXISTS))
+			err = 0;
+		return err;
+	}
+	return 1;
+}
+
+/*
+ * Attempt to set an fcntl lock.
+ * For now, this just goes away to the server. Later it may be more awesome.
+ */
+int ceph_lock(struct file *file, int cmd, struct file_lock *fl)
+{
+	struct inode *inode = file_inode(file);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	int err = 0;
+	u16 op = CEPH_MDS_OP_SETFILELOCK;
+	u8 wait = 0;
+	u8 lock_cmd;
+
+	if (!(fl->fl_flags & FL_POSIX))
+		return -ENOLCK;
+
+	if (ceph_inode_is_shutdown(inode))
+		return -ESTALE;
+
+	dout("ceph_lock, fl_owner: %p\n", fl->fl_owner);
+
+	/* set wait bit as appropriate, then make command as Ceph expects it*/
+	if (IS_GETLK(cmd))
+		op = CEPH_MDS_OP_GETFILELOCK;
+	else if (IS_SETLKW(cmd))
+		wait = 1;
+
+	spin_lock(&ci->i_ceph_lock);
+	if (ci->i_ceph_flags & CEPH_I_ERROR_FILELOCK) {
+		err = -EIO;
+	}
+	spin_unlock(&ci->i_ceph_lock);
+	if (err < 0) {
+		if (op == CEPH_MDS_OP_SETFILELOCK && F_UNLCK == fl->fl_type)
+			posix_lock_file(file, fl, NULL);
+		return err;
+	}
+
+	if (F_RDLCK == fl->fl_type)
+		lock_cmd = CEPH_LOCK_SHARED;
+	else if (F_WRLCK == fl->fl_type)
+		lock_cmd = CEPH_LOCK_EXCL;
+	else
+		lock_cmd = CEPH_LOCK_UNLOCK;
+
+	if (op == CEPH_MDS_OP_SETFILELOCK && F_UNLCK == fl->fl_type) {
+		err = try_unlock_file(file, fl);
+		if (err <= 0)
+			return err;
+	}
+
+	err = ceph_lock_message(CEPH_LOCK_FCNTL, op, inode, lock_cmd, wait, fl);
+	if (!err) {
+		if (op == CEPH_MDS_OP_SETFILELOCK && F_UNLCK != fl->fl_type) {
+			dout("mds locked, locking locally\n");
+			err = posix_lock_file(file, fl, NULL);
+			if (err) {
+				/* undo! This should only happen if
+				 * the kernel detects local
+				 * deadlock. */
+				ceph_lock_message(CEPH_LOCK_FCNTL, op, inode,
+						  CEPH_LOCK_UNLOCK, 0, fl);
+				dout("got %d on posix_lock_file, undid lock\n",
+				     err);
+			}
+		}
+	}
+	return err;
+}
+
+int ceph_flock(struct file *file, int cmd, struct file_lock *fl)
+{
+	struct inode *inode = file_inode(file);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	int err = 0;
+	u8 wait = 0;
+	u8 lock_cmd;
+
+	if (!(fl->fl_flags & FL_FLOCK))
+		return -ENOLCK;
+
+	if (ceph_inode_is_shutdown(inode))
+		return -ESTALE;
+
+	dout("ceph_flock, fl_file: %p\n", fl->fl_file);
+
+	spin_lock(&ci->i_ceph_lock);
+	if (ci->i_ceph_flags & CEPH_I_ERROR_FILELOCK) {
+		err = -EIO;
+	}
+	spin_unlock(&ci->i_ceph_lock);
+	if (err < 0) {
+		if (F_UNLCK == fl->fl_type)
+			locks_lock_file_wait(file, fl);
+		return err;
+	}
+
+	if (IS_SETLKW(cmd))
+		wait = 1;
+
+	if (F_RDLCK == fl->fl_type)
+		lock_cmd = CEPH_LOCK_SHARED;
+	else if (F_WRLCK == fl->fl_type)
+		lock_cmd = CEPH_LOCK_EXCL;
+	else
+		lock_cmd = CEPH_LOCK_UNLOCK;
+
+	if (F_UNLCK == fl->fl_type) {
+		err = try_unlock_file(file, fl);
+		if (err <= 0)
+			return err;
+	}
+
+	err = ceph_lock_message(CEPH_LOCK_FLOCK, CEPH_MDS_OP_SETFILELOCK,
+				inode, lock_cmd, wait, fl);
+	if (!err && F_UNLCK != fl->fl_type) {
+		err = locks_lock_file_wait(file, fl);
+		if (err) {
+			ceph_lock_message(CEPH_LOCK_FLOCK,
+					  CEPH_MDS_OP_SETFILELOCK,
+					  inode, CEPH_LOCK_UNLOCK, 0, fl);
+			dout("got %d on locks_lock_file_wait, undid lock\n", err);
+		}
+	}
+	return err;
+}
+
+/*
+ * Fills in the passed counter variables, so you can prepare pagelist metadata
+ * before calling ceph_encode_locks.
+ */
+void ceph_count_locks(struct inode *inode, int *fcntl_count, int *flock_count)
+{
+	struct file_lock *lock;
+	struct file_lock_context *ctx;
+
+	*fcntl_count = 0;
+	*flock_count = 0;
+
+	ctx = inode->i_flctx;
+	if (ctx) {
+		spin_lock(&ctx->flc_lock);
+		list_for_each_entry(lock, &ctx->flc_posix, fl_list)
+			++(*fcntl_count);
+		list_for_each_entry(lock, &ctx->flc_flock, fl_list)
+			++(*flock_count);
+		spin_unlock(&ctx->flc_lock);
+	}
+	dout("counted %d flock locks and %d fcntl locks\n",
+	     *flock_count, *fcntl_count);
+}
+
+/*
+ * Given a pointer to a lock, convert it to a ceph filelock
+ */
+static int lock_to_ceph_filelock(struct file_lock *lock,
+				 struct ceph_filelock *cephlock)
+{
+	int err = 0;
+	cephlock->start = cpu_to_le64(lock->fl_start);
+	cephlock->length = cpu_to_le64(lock->fl_end - lock->fl_start + 1);
+	cephlock->client = cpu_to_le64(0);
+	cephlock->pid = cpu_to_le64((u64)lock->fl_pid);
+	cephlock->owner = cpu_to_le64(secure_addr(lock->fl_owner));
+
+	switch (lock->fl_type) {
+	case F_RDLCK:
+		cephlock->type = CEPH_LOCK_SHARED;
+		break;
+	case F_WRLCK:
+		cephlock->type = CEPH_LOCK_EXCL;
+		break;
+	case F_UNLCK:
+		cephlock->type = CEPH_LOCK_UNLOCK;
+		break;
+	default:
+		dout("Have unknown lock type %d\n", lock->fl_type);
+		err = -EINVAL;
+	}
+
+	return err;
+}
+
+/*
+ * Encode the flock and fcntl locks for the given inode into the ceph_filelock
+ * array. Must be called with inode->i_lock already held.
+ * If we encounter more of a specific lock type than expected, return -ENOSPC.
+ */
+int ceph_encode_locks_to_buffer(struct inode *inode,
+				struct ceph_filelock *flocks,
+				int num_fcntl_locks, int num_flock_locks)
+{
+	struct file_lock *lock;
+	struct file_lock_context *ctx = inode->i_flctx;
+	int err = 0;
+	int seen_fcntl = 0;
+	int seen_flock = 0;
+	int l = 0;
+
+	dout("encoding %d flock and %d fcntl locks\n", num_flock_locks,
+	     num_fcntl_locks);
+
+	if (!ctx)
+		return 0;
+
+	spin_lock(&ctx->flc_lock);
+	list_for_each_entry(lock, &ctx->flc_posix, fl_list) {
+		++seen_fcntl;
+		if (seen_fcntl > num_fcntl_locks) {
+			err = -ENOSPC;
+			goto fail;
+		}
+		err = lock_to_ceph_filelock(lock, &flocks[l]);
+		if (err)
+			goto fail;
+		++l;
+	}
+	list_for_each_entry(lock, &ctx->flc_flock, fl_list) {
+		++seen_flock;
+		if (seen_flock > num_flock_locks) {
+			err = -ENOSPC;
+			goto fail;
+		}
+		err = lock_to_ceph_filelock(lock, &flocks[l]);
+		if (err)
+			goto fail;
+		++l;
+	}
+fail:
+	spin_unlock(&ctx->flc_lock);
+	return err;
+}
+
+/*
+ * Copy the encoded flock and fcntl locks into the pagelist.
+ * Format is: #fcntl locks, sequential fcntl locks, #flock locks,
+ * sequential flock locks.
+ * Returns zero on success.
+ */
+int ceph_locks_to_pagelist(struct ceph_filelock *flocks,
+			   struct ceph_pagelist *pagelist,
+			   int num_fcntl_locks, int num_flock_locks)
+{
+	int err = 0;
+	__le32 nlocks;
+
+	nlocks = cpu_to_le32(num_fcntl_locks);
+	err = ceph_pagelist_append(pagelist, &nlocks, sizeof(nlocks));
+	if (err)
+		goto out_fail;
+
+	if (num_fcntl_locks > 0) {
+		err = ceph_pagelist_append(pagelist, flocks,
+					   num_fcntl_locks * sizeof(*flocks));
+		if (err)
+			goto out_fail;
+	}
+
+	nlocks = cpu_to_le32(num_flock_locks);
+	err = ceph_pagelist_append(pagelist, &nlocks, sizeof(nlocks));
+	if (err)
+		goto out_fail;
+
+	if (num_flock_locks > 0) {
+		err = ceph_pagelist_append(pagelist, &flocks[num_fcntl_locks],
+					   num_flock_locks * sizeof(*flocks));
+	}
+out_fail:
+	return err;
+}
diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
new file mode 100644
index 000000000..82874be94
--- /dev/null
+++ b/fs/ceph/mds_client.c
@@ -0,0 +1,5608 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/ceph/ceph_debug.h>
+
+#include <linux/fs.h>
+#include <linux/wait.h>
+#include <linux/slab.h>
+#include <linux/gfp.h>
+#include <linux/sched.h>
+#include <linux/debugfs.h>
+#include <linux/seq_file.h>
+#include <linux/ratelimit.h>
+#include <linux/bits.h>
+#include <linux/ktime.h>
+#include <linux/bitmap.h>
+
+#include "super.h"
+#include "mds_client.h"
+
+#include <linux/ceph/ceph_features.h>
+#include <linux/ceph/messenger.h>
+#include <linux/ceph/decode.h>
+#include <linux/ceph/pagelist.h>
+#include <linux/ceph/auth.h>
+#include <linux/ceph/debugfs.h>
+
+#define RECONNECT_MAX_SIZE (INT_MAX - PAGE_SIZE)
+
+/*
+ * A cluster of MDS (metadata server) daemons is responsible for
+ * managing the file system namespace (the directory hierarchy and
+ * inodes) and for coordinating shared access to storage.  Metadata is
+ * partitioning hierarchically across a number of servers, and that
+ * partition varies over time as the cluster adjusts the distribution
+ * in order to balance load.
+ *
+ * The MDS client is primarily responsible to managing synchronous
+ * metadata requests for operations like open, unlink, and so forth.
+ * If there is a MDS failure, we find out about it when we (possibly
+ * request and) receive a new MDS map, and can resubmit affected
+ * requests.
+ *
+ * For the most part, though, we take advantage of a lossless
+ * communications channel to the MDS, and do not need to worry about
+ * timing out or resubmitting requests.
+ *
+ * We maintain a stateful "session" with each MDS we interact with.
+ * Within each session, we sent periodic heartbeat messages to ensure
+ * any capabilities or leases we have been issues remain valid.  If
+ * the session times out and goes stale, our leases and capabilities
+ * are no longer valid.
+ */
+
+struct ceph_reconnect_state {
+	struct ceph_mds_session *session;
+	int nr_caps, nr_realms;
+	struct ceph_pagelist *pagelist;
+	unsigned msg_version;
+	bool allow_multi;
+};
+
+static void __wake_requests(struct ceph_mds_client *mdsc,
+			    struct list_head *head);
+static void ceph_cap_release_work(struct work_struct *work);
+static void ceph_cap_reclaim_work(struct work_struct *work);
+
+static const struct ceph_connection_operations mds_con_ops;
+
+
+/*
+ * mds reply parsing
+ */
+
+static int parse_reply_info_quota(void **p, void *end,
+				  struct ceph_mds_reply_info_in *info)
+{
+	u8 struct_v, struct_compat;
+	u32 struct_len;
+
+	ceph_decode_8_safe(p, end, struct_v, bad);
+	ceph_decode_8_safe(p, end, struct_compat, bad);
+	/* struct_v is expected to be >= 1. we only
+	 * understand encoding with struct_compat == 1. */
+	if (!struct_v || struct_compat != 1)
+		goto bad;
+	ceph_decode_32_safe(p, end, struct_len, bad);
+	ceph_decode_need(p, end, struct_len, bad);
+	end = *p + struct_len;
+	ceph_decode_64_safe(p, end, info->max_bytes, bad);
+	ceph_decode_64_safe(p, end, info->max_files, bad);
+	*p = end;
+	return 0;
+bad:
+	return -EIO;
+}
+
+/*
+ * parse individual inode info
+ */
+static int parse_reply_info_in(void **p, void *end,
+			       struct ceph_mds_reply_info_in *info,
+			       u64 features)
+{
+	int err = 0;
+	u8 struct_v = 0;
+
+	if (features == (u64)-1) {
+		u32 struct_len;
+		u8 struct_compat;
+		ceph_decode_8_safe(p, end, struct_v, bad);
+		ceph_decode_8_safe(p, end, struct_compat, bad);
+		/* struct_v is expected to be >= 1. we only understand
+		 * encoding with struct_compat == 1. */
+		if (!struct_v || struct_compat != 1)
+			goto bad;
+		ceph_decode_32_safe(p, end, struct_len, bad);
+		ceph_decode_need(p, end, struct_len, bad);
+		end = *p + struct_len;
+	}
+
+	ceph_decode_need(p, end, sizeof(struct ceph_mds_reply_inode), bad);
+	info->in = *p;
+	*p += sizeof(struct ceph_mds_reply_inode) +
+		sizeof(*info->in->fragtree.splits) *
+		le32_to_cpu(info->in->fragtree.nsplits);
+
+	ceph_decode_32_safe(p, end, info->symlink_len, bad);
+	ceph_decode_need(p, end, info->symlink_len, bad);
+	info->symlink = *p;
+	*p += info->symlink_len;
+
+	ceph_decode_copy_safe(p, end, &info->dir_layout,
+			      sizeof(info->dir_layout), bad);
+	ceph_decode_32_safe(p, end, info->xattr_len, bad);
+	ceph_decode_need(p, end, info->xattr_len, bad);
+	info->xattr_data = *p;
+	*p += info->xattr_len;
+
+	if (features == (u64)-1) {
+		/* inline data */
+		ceph_decode_64_safe(p, end, info->inline_version, bad);
+		ceph_decode_32_safe(p, end, info->inline_len, bad);
+		ceph_decode_need(p, end, info->inline_len, bad);
+		info->inline_data = *p;
+		*p += info->inline_len;
+		/* quota */
+		err = parse_reply_info_quota(p, end, info);
+		if (err < 0)
+			goto out_bad;
+		/* pool namespace */
+		ceph_decode_32_safe(p, end, info->pool_ns_len, bad);
+		if (info->pool_ns_len > 0) {
+			ceph_decode_need(p, end, info->pool_ns_len, bad);
+			info->pool_ns_data = *p;
+			*p += info->pool_ns_len;
+		}
+
+		/* btime */
+		ceph_decode_need(p, end, sizeof(info->btime), bad);
+		ceph_decode_copy(p, &info->btime, sizeof(info->btime));
+
+		/* change attribute */
+		ceph_decode_64_safe(p, end, info->change_attr, bad);
+
+		/* dir pin */
+		if (struct_v >= 2) {
+			ceph_decode_32_safe(p, end, info->dir_pin, bad);
+		} else {
+			info->dir_pin = -ENODATA;
+		}
+
+		/* snapshot birth time, remains zero for v<=2 */
+		if (struct_v >= 3) {
+			ceph_decode_need(p, end, sizeof(info->snap_btime), bad);
+			ceph_decode_copy(p, &info->snap_btime,
+					 sizeof(info->snap_btime));
+		} else {
+			memset(&info->snap_btime, 0, sizeof(info->snap_btime));
+		}
+
+		/* snapshot count, remains zero for v<=3 */
+		if (struct_v >= 4) {
+			ceph_decode_64_safe(p, end, info->rsnaps, bad);
+		} else {
+			info->rsnaps = 0;
+		}
+
+		*p = end;
+	} else {
+		if (features & CEPH_FEATURE_MDS_INLINE_DATA) {
+			ceph_decode_64_safe(p, end, info->inline_version, bad);
+			ceph_decode_32_safe(p, end, info->inline_len, bad);
+			ceph_decode_need(p, end, info->inline_len, bad);
+			info->inline_data = *p;
+			*p += info->inline_len;
+		} else
+			info->inline_version = CEPH_INLINE_NONE;
+
+		if (features & CEPH_FEATURE_MDS_QUOTA) {
+			err = parse_reply_info_quota(p, end, info);
+			if (err < 0)
+				goto out_bad;
+		} else {
+			info->max_bytes = 0;
+			info->max_files = 0;
+		}
+
+		info->pool_ns_len = 0;
+		info->pool_ns_data = NULL;
+		if (features & CEPH_FEATURE_FS_FILE_LAYOUT_V2) {
+			ceph_decode_32_safe(p, end, info->pool_ns_len, bad);
+			if (info->pool_ns_len > 0) {
+				ceph_decode_need(p, end, info->pool_ns_len, bad);
+				info->pool_ns_data = *p;
+				*p += info->pool_ns_len;
+			}
+		}
+
+		if (features & CEPH_FEATURE_FS_BTIME) {
+			ceph_decode_need(p, end, sizeof(info->btime), bad);
+			ceph_decode_copy(p, &info->btime, sizeof(info->btime));
+			ceph_decode_64_safe(p, end, info->change_attr, bad);
+		}
+
+		info->dir_pin = -ENODATA;
+		/* info->snap_btime and info->rsnaps remain zero */
+	}
+	return 0;
+bad:
+	err = -EIO;
+out_bad:
+	return err;
+}
+
+static int parse_reply_info_dir(void **p, void *end,
+				struct ceph_mds_reply_dirfrag **dirfrag,
+				u64 features)
+{
+	if (features == (u64)-1) {
+		u8 struct_v, struct_compat;
+		u32 struct_len;
+		ceph_decode_8_safe(p, end, struct_v, bad);
+		ceph_decode_8_safe(p, end, struct_compat, bad);
+		/* struct_v is expected to be >= 1. we only understand
+		 * encoding whose struct_compat == 1. */
+		if (!struct_v || struct_compat != 1)
+			goto bad;
+		ceph_decode_32_safe(p, end, struct_len, bad);
+		ceph_decode_need(p, end, struct_len, bad);
+		end = *p + struct_len;
+	}
+
+	ceph_decode_need(p, end, sizeof(**dirfrag), bad);
+	*dirfrag = *p;
+	*p += sizeof(**dirfrag) + sizeof(u32) * le32_to_cpu((*dirfrag)->ndist);
+	if (unlikely(*p > end))
+		goto bad;
+	if (features == (u64)-1)
+		*p = end;
+	return 0;
+bad:
+	return -EIO;
+}
+
+static int parse_reply_info_lease(void **p, void *end,
+				  struct ceph_mds_reply_lease **lease,
+				  u64 features)
+{
+	if (features == (u64)-1) {
+		u8 struct_v, struct_compat;
+		u32 struct_len;
+		ceph_decode_8_safe(p, end, struct_v, bad);
+		ceph_decode_8_safe(p, end, struct_compat, bad);
+		/* struct_v is expected to be >= 1. we only understand
+		 * encoding whose struct_compat == 1. */
+		if (!struct_v || struct_compat != 1)
+			goto bad;
+		ceph_decode_32_safe(p, end, struct_len, bad);
+		ceph_decode_need(p, end, struct_len, bad);
+		end = *p + struct_len;
+	}
+
+	ceph_decode_need(p, end, sizeof(**lease), bad);
+	*lease = *p;
+	*p += sizeof(**lease);
+	if (features == (u64)-1)
+		*p = end;
+	return 0;
+bad:
+	return -EIO;
+}
+
+/*
+ * parse a normal reply, which may contain a (dir+)dentry and/or a
+ * target inode.
+ */
+static int parse_reply_info_trace(void **p, void *end,
+				  struct ceph_mds_reply_info_parsed *info,
+				  u64 features)
+{
+	int err;
+
+	if (info->head->is_dentry) {
+		err = parse_reply_info_in(p, end, &info->diri, features);
+		if (err < 0)
+			goto out_bad;
+
+		err = parse_reply_info_dir(p, end, &info->dirfrag, features);
+		if (err < 0)
+			goto out_bad;
+
+		ceph_decode_32_safe(p, end, info->dname_len, bad);
+		ceph_decode_need(p, end, info->dname_len, bad);
+		info->dname = *p;
+		*p += info->dname_len;
+
+		err = parse_reply_info_lease(p, end, &info->dlease, features);
+		if (err < 0)
+			goto out_bad;
+	}
+
+	if (info->head->is_target) {
+		err = parse_reply_info_in(p, end, &info->targeti, features);
+		if (err < 0)
+			goto out_bad;
+	}
+
+	if (unlikely(*p != end))
+		goto bad;
+	return 0;
+
+bad:
+	err = -EIO;
+out_bad:
+	pr_err("problem parsing mds trace %d\n", err);
+	return err;
+}
+
+/*
+ * parse readdir results
+ */
+static int parse_reply_info_readdir(void **p, void *end,
+				struct ceph_mds_reply_info_parsed *info,
+				u64 features)
+{
+	u32 num, i = 0;
+	int err;
+
+	err = parse_reply_info_dir(p, end, &info->dir_dir, features);
+	if (err < 0)
+		goto out_bad;
+
+	ceph_decode_need(p, end, sizeof(num) + 2, bad);
+	num = ceph_decode_32(p);
+	{
+		u16 flags = ceph_decode_16(p);
+		info->dir_end = !!(flags & CEPH_READDIR_FRAG_END);
+		info->dir_complete = !!(flags & CEPH_READDIR_FRAG_COMPLETE);
+		info->hash_order = !!(flags & CEPH_READDIR_HASH_ORDER);
+		info->offset_hash = !!(flags & CEPH_READDIR_OFFSET_HASH);
+	}
+	if (num == 0)
+		goto done;
+
+	BUG_ON(!info->dir_entries);
+	if ((unsigned long)(info->dir_entries + num) >
+	    (unsigned long)info->dir_entries + info->dir_buf_size) {
+		pr_err("dir contents are larger than expected\n");
+		WARN_ON(1);
+		goto bad;
+	}
+
+	info->dir_nr = num;
+	while (num) {
+		struct ceph_mds_reply_dir_entry *rde = info->dir_entries + i;
+		/* dentry */
+		ceph_decode_32_safe(p, end, rde->name_len, bad);
+		ceph_decode_need(p, end, rde->name_len, bad);
+		rde->name = *p;
+		*p += rde->name_len;
+		dout("parsed dir dname '%.*s'\n", rde->name_len, rde->name);
+
+		/* dentry lease */
+		err = parse_reply_info_lease(p, end, &rde->lease, features);
+		if (err)
+			goto out_bad;
+		/* inode */
+		err = parse_reply_info_in(p, end, &rde->inode, features);
+		if (err < 0)
+			goto out_bad;
+		/* ceph_readdir_prepopulate() will update it */
+		rde->offset = 0;
+		i++;
+		num--;
+	}
+
+done:
+	/* Skip over any unrecognized fields */
+	*p = end;
+	return 0;
+
+bad:
+	err = -EIO;
+out_bad:
+	pr_err("problem parsing dir contents %d\n", err);
+	return err;
+}
+
+/*
+ * parse fcntl F_GETLK results
+ */
+static int parse_reply_info_filelock(void **p, void *end,
+				     struct ceph_mds_reply_info_parsed *info,
+				     u64 features)
+{
+	if (*p + sizeof(*info->filelock_reply) > end)
+		goto bad;
+
+	info->filelock_reply = *p;
+
+	/* Skip over any unrecognized fields */
+	*p = end;
+	return 0;
+bad:
+	return -EIO;
+}
+
+
+#if BITS_PER_LONG == 64
+
+#define DELEGATED_INO_AVAILABLE		xa_mk_value(1)
+
+static int ceph_parse_deleg_inos(void **p, void *end,
+				 struct ceph_mds_session *s)
+{
+	u32 sets;
+
+	ceph_decode_32_safe(p, end, sets, bad);
+	dout("got %u sets of delegated inodes\n", sets);
+	while (sets--) {
+		u64 start, len;
+
+		ceph_decode_64_safe(p, end, start, bad);
+		ceph_decode_64_safe(p, end, len, bad);
+
+		/* Don't accept a delegation of system inodes */
+		if (start < CEPH_INO_SYSTEM_BASE) {
+			pr_warn_ratelimited("ceph: ignoring reserved inode range delegation (start=0x%llx len=0x%llx)\n",
+					start, len);
+			continue;
+		}
+		while (len--) {
+			int err = xa_insert(&s->s_delegated_inos, start++,
+					    DELEGATED_INO_AVAILABLE,
+					    GFP_KERNEL);
+			if (!err) {
+				dout("added delegated inode 0x%llx\n",
+				     start - 1);
+			} else if (err == -EBUSY) {
+				pr_warn("MDS delegated inode 0x%llx more than once.\n",
+					start - 1);
+			} else {
+				return err;
+			}
+		}
+	}
+	return 0;
+bad:
+	return -EIO;
+}
+
+u64 ceph_get_deleg_ino(struct ceph_mds_session *s)
+{
+	unsigned long ino;
+	void *val;
+
+	xa_for_each(&s->s_delegated_inos, ino, val) {
+		val = xa_erase(&s->s_delegated_inos, ino);
+		if (val == DELEGATED_INO_AVAILABLE)
+			return ino;
+	}
+	return 0;
+}
+
+int ceph_restore_deleg_ino(struct ceph_mds_session *s, u64 ino)
+{
+	return xa_insert(&s->s_delegated_inos, ino, DELEGATED_INO_AVAILABLE,
+			 GFP_KERNEL);
+}
+#else /* BITS_PER_LONG == 64 */
+/*
+ * FIXME: xarrays can't handle 64-bit indexes on a 32-bit arch. For now, just
+ * ignore delegated_inos on 32 bit arch. Maybe eventually add xarrays for top
+ * and bottom words?
+ */
+static int ceph_parse_deleg_inos(void **p, void *end,
+				 struct ceph_mds_session *s)
+{
+	u32 sets;
+
+	ceph_decode_32_safe(p, end, sets, bad);
+	if (sets)
+		ceph_decode_skip_n(p, end, sets * 2 * sizeof(__le64), bad);
+	return 0;
+bad:
+	return -EIO;
+}
+
+u64 ceph_get_deleg_ino(struct ceph_mds_session *s)
+{
+	return 0;
+}
+
+int ceph_restore_deleg_ino(struct ceph_mds_session *s, u64 ino)
+{
+	return 0;
+}
+#endif /* BITS_PER_LONG == 64 */
+
+/*
+ * parse create results
+ */
+static int parse_reply_info_create(void **p, void *end,
+				  struct ceph_mds_reply_info_parsed *info,
+				  u64 features, struct ceph_mds_session *s)
+{
+	int ret;
+
+	if (features == (u64)-1 ||
+	    (features & CEPH_FEATURE_REPLY_CREATE_INODE)) {
+		if (*p == end) {
+			/* Malformed reply? */
+			info->has_create_ino = false;
+		} else if (test_bit(CEPHFS_FEATURE_DELEG_INO, &s->s_features)) {
+			info->has_create_ino = true;
+			/* struct_v, struct_compat, and len */
+			ceph_decode_skip_n(p, end, 2 + sizeof(u32), bad);
+			ceph_decode_64_safe(p, end, info->ino, bad);
+			ret = ceph_parse_deleg_inos(p, end, s);
+			if (ret)
+				return ret;
+		} else {
+			/* legacy */
+			ceph_decode_64_safe(p, end, info->ino, bad);
+			info->has_create_ino = true;
+		}
+	} else {
+		if (*p != end)
+			goto bad;
+	}
+
+	/* Skip over any unrecognized fields */
+	*p = end;
+	return 0;
+bad:
+	return -EIO;
+}
+
+static int parse_reply_info_getvxattr(void **p, void *end,
+				      struct ceph_mds_reply_info_parsed *info,
+				      u64 features)
+{
+	u32 value_len;
+
+	ceph_decode_skip_8(p, end, bad); /* skip current version: 1 */
+	ceph_decode_skip_8(p, end, bad); /* skip first version: 1 */
+	ceph_decode_skip_32(p, end, bad); /* skip payload length */
+
+	ceph_decode_32_safe(p, end, value_len, bad);
+
+	if (value_len == end - *p) {
+	  info->xattr_info.xattr_value = *p;
+	  info->xattr_info.xattr_value_len = value_len;
+	  *p = end;
+	  return value_len;
+	}
+bad:
+	return -EIO;
+}
+
+/*
+ * parse extra results
+ */
+static int parse_reply_info_extra(void **p, void *end,
+				  struct ceph_mds_reply_info_parsed *info,
+				  u64 features, struct ceph_mds_session *s)
+{
+	u32 op = le32_to_cpu(info->head->op);
+
+	if (op == CEPH_MDS_OP_GETFILELOCK)
+		return parse_reply_info_filelock(p, end, info, features);
+	else if (op == CEPH_MDS_OP_READDIR || op == CEPH_MDS_OP_LSSNAP)
+		return parse_reply_info_readdir(p, end, info, features);
+	else if (op == CEPH_MDS_OP_CREATE)
+		return parse_reply_info_create(p, end, info, features, s);
+	else if (op == CEPH_MDS_OP_GETVXATTR)
+		return parse_reply_info_getvxattr(p, end, info, features);
+	else
+		return -EIO;
+}
+
+/*
+ * parse entire mds reply
+ */
+static int parse_reply_info(struct ceph_mds_session *s, struct ceph_msg *msg,
+			    struct ceph_mds_reply_info_parsed *info,
+			    u64 features)
+{
+	void *p, *end;
+	u32 len;
+	int err;
+
+	info->head = msg->front.iov_base;
+	p = msg->front.iov_base + sizeof(struct ceph_mds_reply_head);
+	end = p + msg->front.iov_len - sizeof(struct ceph_mds_reply_head);
+
+	/* trace */
+	ceph_decode_32_safe(&p, end, len, bad);
+	if (len > 0) {
+		ceph_decode_need(&p, end, len, bad);
+		err = parse_reply_info_trace(&p, p+len, info, features);
+		if (err < 0)
+			goto out_bad;
+	}
+
+	/* extra */
+	ceph_decode_32_safe(&p, end, len, bad);
+	if (len > 0) {
+		ceph_decode_need(&p, end, len, bad);
+		err = parse_reply_info_extra(&p, p+len, info, features, s);
+		if (err < 0)
+			goto out_bad;
+	}
+
+	/* snap blob */
+	ceph_decode_32_safe(&p, end, len, bad);
+	info->snapblob_len = len;
+	info->snapblob = p;
+	p += len;
+
+	if (p != end)
+		goto bad;
+	return 0;
+
+bad:
+	err = -EIO;
+out_bad:
+	pr_err("mds parse_reply err %d\n", err);
+	ceph_msg_dump(msg);
+	return err;
+}
+
+static void destroy_reply_info(struct ceph_mds_reply_info_parsed *info)
+{
+	if (!info->dir_entries)
+		return;
+	free_pages((unsigned long)info->dir_entries, get_order(info->dir_buf_size));
+}
+
+/*
+ * In async unlink case the kclient won't wait for the first reply
+ * from MDS and just drop all the links and unhash the dentry and then
+ * succeeds immediately.
+ *
+ * For any new create/link/rename,etc requests followed by using the
+ * same file names we must wait for the first reply of the inflight
+ * unlink request, or the MDS possibly will fail these following
+ * requests with -EEXIST if the inflight async unlink request was
+ * delayed for some reasons.
+ *
+ * And the worst case is that for the none async openc request it will
+ * successfully open the file if the CDentry hasn't been unlinked yet,
+ * but later the previous delayed async unlink request will remove the
+ * CDenty. That means the just created file is possiblly deleted later
+ * by accident.
+ *
+ * We need to wait for the inflight async unlink requests to finish
+ * when creating new files/directories by using the same file names.
+ */
+int ceph_wait_on_conflict_unlink(struct dentry *dentry)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(dentry->d_sb);
+	struct dentry *pdentry = dentry->d_parent;
+	struct dentry *udentry, *found = NULL;
+	struct ceph_dentry_info *di;
+	struct qstr dname;
+	u32 hash = dentry->d_name.hash;
+	int err;
+
+	dname.name = dentry->d_name.name;
+	dname.len = dentry->d_name.len;
+
+	rcu_read_lock();
+	hash_for_each_possible_rcu(fsc->async_unlink_conflict, di,
+				   hnode, hash) {
+		udentry = di->dentry;
+
+		spin_lock(&udentry->d_lock);
+		if (udentry->d_name.hash != hash)
+			goto next;
+		if (unlikely(udentry->d_parent != pdentry))
+			goto next;
+		if (!hash_hashed(&di->hnode))
+			goto next;
+
+		if (!test_bit(CEPH_DENTRY_ASYNC_UNLINK_BIT, &di->flags))
+			pr_warn("%s dentry %p:%pd async unlink bit is not set\n",
+				__func__, dentry, dentry);
+
+		if (!d_same_name(udentry, pdentry, &dname))
+			goto next;
+
+		found = dget_dlock(udentry);
+		spin_unlock(&udentry->d_lock);
+		break;
+next:
+		spin_unlock(&udentry->d_lock);
+	}
+	rcu_read_unlock();
+
+	if (likely(!found))
+		return 0;
+
+	dout("%s dentry %p:%pd conflict with old %p:%pd\n", __func__,
+	     dentry, dentry, found, found);
+
+	err = wait_on_bit(&di->flags, CEPH_DENTRY_ASYNC_UNLINK_BIT,
+			  TASK_KILLABLE);
+	dput(found);
+	return err;
+}
+
+
+/*
+ * sessions
+ */
+const char *ceph_session_state_name(int s)
+{
+	switch (s) {
+	case CEPH_MDS_SESSION_NEW: return "new";
+	case CEPH_MDS_SESSION_OPENING: return "opening";
+	case CEPH_MDS_SESSION_OPEN: return "open";
+	case CEPH_MDS_SESSION_HUNG: return "hung";
+	case CEPH_MDS_SESSION_CLOSING: return "closing";
+	case CEPH_MDS_SESSION_CLOSED: return "closed";
+	case CEPH_MDS_SESSION_RESTARTING: return "restarting";
+	case CEPH_MDS_SESSION_RECONNECTING: return "reconnecting";
+	case CEPH_MDS_SESSION_REJECTED: return "rejected";
+	default: return "???";
+	}
+}
+
+struct ceph_mds_session *ceph_get_mds_session(struct ceph_mds_session *s)
+{
+	if (refcount_inc_not_zero(&s->s_ref))
+		return s;
+	return NULL;
+}
+
+void ceph_put_mds_session(struct ceph_mds_session *s)
+{
+	if (IS_ERR_OR_NULL(s))
+		return;
+
+	if (refcount_dec_and_test(&s->s_ref)) {
+		if (s->s_auth.authorizer)
+			ceph_auth_destroy_authorizer(s->s_auth.authorizer);
+		WARN_ON(mutex_is_locked(&s->s_mutex));
+		xa_destroy(&s->s_delegated_inos);
+		kfree(s);
+	}
+}
+
+/*
+ * called under mdsc->mutex
+ */
+struct ceph_mds_session *__ceph_lookup_mds_session(struct ceph_mds_client *mdsc,
+						   int mds)
+{
+	if (mds >= mdsc->max_sessions || !mdsc->sessions[mds])
+		return NULL;
+	return ceph_get_mds_session(mdsc->sessions[mds]);
+}
+
+static bool __have_session(struct ceph_mds_client *mdsc, int mds)
+{
+	if (mds >= mdsc->max_sessions || !mdsc->sessions[mds])
+		return false;
+	else
+		return true;
+}
+
+static int __verify_registered_session(struct ceph_mds_client *mdsc,
+				       struct ceph_mds_session *s)
+{
+	if (s->s_mds >= mdsc->max_sessions ||
+	    mdsc->sessions[s->s_mds] != s)
+		return -ENOENT;
+	return 0;
+}
+
+/*
+ * create+register a new session for given mds.
+ * called under mdsc->mutex.
+ */
+static struct ceph_mds_session *register_session(struct ceph_mds_client *mdsc,
+						 int mds)
+{
+	struct ceph_mds_session *s;
+
+	if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_FENCE_IO)
+		return ERR_PTR(-EIO);
+
+	if (mds >= mdsc->mdsmap->possible_max_rank)
+		return ERR_PTR(-EINVAL);
+
+	s = kzalloc(sizeof(*s), GFP_NOFS);
+	if (!s)
+		return ERR_PTR(-ENOMEM);
+
+	if (mds >= mdsc->max_sessions) {
+		int newmax = 1 << get_count_order(mds + 1);
+		struct ceph_mds_session **sa;
+
+		dout("%s: realloc to %d\n", __func__, newmax);
+		sa = kcalloc(newmax, sizeof(void *), GFP_NOFS);
+		if (!sa)
+			goto fail_realloc;
+		if (mdsc->sessions) {
+			memcpy(sa, mdsc->sessions,
+			       mdsc->max_sessions * sizeof(void *));
+			kfree(mdsc->sessions);
+		}
+		mdsc->sessions = sa;
+		mdsc->max_sessions = newmax;
+	}
+
+	dout("%s: mds%d\n", __func__, mds);
+	s->s_mdsc = mdsc;
+	s->s_mds = mds;
+	s->s_state = CEPH_MDS_SESSION_NEW;
+	mutex_init(&s->s_mutex);
+
+	ceph_con_init(&s->s_con, s, &mds_con_ops, &mdsc->fsc->client->msgr);
+
+	atomic_set(&s->s_cap_gen, 1);
+	s->s_cap_ttl = jiffies - 1;
+
+	spin_lock_init(&s->s_cap_lock);
+	INIT_LIST_HEAD(&s->s_caps);
+	refcount_set(&s->s_ref, 1);
+	INIT_LIST_HEAD(&s->s_waiting);
+	INIT_LIST_HEAD(&s->s_unsafe);
+	xa_init(&s->s_delegated_inos);
+	INIT_LIST_HEAD(&s->s_cap_releases);
+	INIT_WORK(&s->s_cap_release_work, ceph_cap_release_work);
+
+	INIT_LIST_HEAD(&s->s_cap_dirty);
+	INIT_LIST_HEAD(&s->s_cap_flushing);
+
+	mdsc->sessions[mds] = s;
+	atomic_inc(&mdsc->num_sessions);
+	refcount_inc(&s->s_ref);  /* one ref to sessions[], one to caller */
+
+	ceph_con_open(&s->s_con, CEPH_ENTITY_TYPE_MDS, mds,
+		      ceph_mdsmap_get_addr(mdsc->mdsmap, mds));
+
+	return s;
+
+fail_realloc:
+	kfree(s);
+	return ERR_PTR(-ENOMEM);
+}
+
+/*
+ * called under mdsc->mutex
+ */
+static void __unregister_session(struct ceph_mds_client *mdsc,
+			       struct ceph_mds_session *s)
+{
+	dout("__unregister_session mds%d %p\n", s->s_mds, s);
+	BUG_ON(mdsc->sessions[s->s_mds] != s);
+	mdsc->sessions[s->s_mds] = NULL;
+	ceph_con_close(&s->s_con);
+	ceph_put_mds_session(s);
+	atomic_dec(&mdsc->num_sessions);
+}
+
+/*
+ * drop session refs in request.
+ *
+ * should be last request ref, or hold mdsc->mutex
+ */
+static void put_request_session(struct ceph_mds_request *req)
+{
+	if (req->r_session) {
+		ceph_put_mds_session(req->r_session);
+		req->r_session = NULL;
+	}
+}
+
+void ceph_mdsc_iterate_sessions(struct ceph_mds_client *mdsc,
+				void (*cb)(struct ceph_mds_session *),
+				bool check_state)
+{
+	int mds;
+
+	mutex_lock(&mdsc->mutex);
+	for (mds = 0; mds < mdsc->max_sessions; ++mds) {
+		struct ceph_mds_session *s;
+
+		s = __ceph_lookup_mds_session(mdsc, mds);
+		if (!s)
+			continue;
+
+		if (check_state && !check_session_state(s)) {
+			ceph_put_mds_session(s);
+			continue;
+		}
+
+		mutex_unlock(&mdsc->mutex);
+		cb(s);
+		ceph_put_mds_session(s);
+		mutex_lock(&mdsc->mutex);
+	}
+	mutex_unlock(&mdsc->mutex);
+}
+
+void ceph_mdsc_release_request(struct kref *kref)
+{
+	struct ceph_mds_request *req = container_of(kref,
+						    struct ceph_mds_request,
+						    r_kref);
+	ceph_mdsc_release_dir_caps_no_check(req);
+	destroy_reply_info(&req->r_reply_info);
+	if (req->r_request)
+		ceph_msg_put(req->r_request);
+	if (req->r_reply)
+		ceph_msg_put(req->r_reply);
+	if (req->r_inode) {
+		ceph_put_cap_refs(ceph_inode(req->r_inode), CEPH_CAP_PIN);
+		iput(req->r_inode);
+	}
+	if (req->r_parent) {
+		ceph_put_cap_refs(ceph_inode(req->r_parent), CEPH_CAP_PIN);
+		iput(req->r_parent);
+	}
+	iput(req->r_target_inode);
+	if (req->r_dentry)
+		dput(req->r_dentry);
+	if (req->r_old_dentry)
+		dput(req->r_old_dentry);
+	if (req->r_old_dentry_dir) {
+		/*
+		 * track (and drop pins for) r_old_dentry_dir
+		 * separately, since r_old_dentry's d_parent may have
+		 * changed between the dir mutex being dropped and
+		 * this request being freed.
+		 */
+		ceph_put_cap_refs(ceph_inode(req->r_old_dentry_dir),
+				  CEPH_CAP_PIN);
+		iput(req->r_old_dentry_dir);
+	}
+	kfree(req->r_path1);
+	kfree(req->r_path2);
+	put_cred(req->r_cred);
+	if (req->r_pagelist)
+		ceph_pagelist_release(req->r_pagelist);
+	put_request_session(req);
+	ceph_unreserve_caps(req->r_mdsc, &req->r_caps_reservation);
+	WARN_ON_ONCE(!list_empty(&req->r_wait));
+	kmem_cache_free(ceph_mds_request_cachep, req);
+}
+
+DEFINE_RB_FUNCS(request, struct ceph_mds_request, r_tid, r_node)
+
+/*
+ * lookup session, bump ref if found.
+ *
+ * called under mdsc->mutex.
+ */
+static struct ceph_mds_request *
+lookup_get_request(struct ceph_mds_client *mdsc, u64 tid)
+{
+	struct ceph_mds_request *req;
+
+	req = lookup_request(&mdsc->request_tree, tid);
+	if (req)
+		ceph_mdsc_get_request(req);
+
+	return req;
+}
+
+/*
+ * Register an in-flight request, and assign a tid.  Link to directory
+ * are modifying (if any).
+ *
+ * Called under mdsc->mutex.
+ */
+static void __register_request(struct ceph_mds_client *mdsc,
+			       struct ceph_mds_request *req,
+			       struct inode *dir)
+{
+	int ret = 0;
+
+	req->r_tid = ++mdsc->last_tid;
+	if (req->r_num_caps) {
+		ret = ceph_reserve_caps(mdsc, &req->r_caps_reservation,
+					req->r_num_caps);
+		if (ret < 0) {
+			pr_err("__register_request %p "
+			       "failed to reserve caps: %d\n", req, ret);
+			/* set req->r_err to fail early from __do_request */
+			req->r_err = ret;
+			return;
+		}
+	}
+	dout("__register_request %p tid %lld\n", req, req->r_tid);
+	ceph_mdsc_get_request(req);
+	insert_request(&mdsc->request_tree, req);
+
+	req->r_cred = get_current_cred();
+
+	if (mdsc->oldest_tid == 0 && req->r_op != CEPH_MDS_OP_SETFILELOCK)
+		mdsc->oldest_tid = req->r_tid;
+
+	if (dir) {
+		struct ceph_inode_info *ci = ceph_inode(dir);
+
+		ihold(dir);
+		req->r_unsafe_dir = dir;
+		spin_lock(&ci->i_unsafe_lock);
+		list_add_tail(&req->r_unsafe_dir_item, &ci->i_unsafe_dirops);
+		spin_unlock(&ci->i_unsafe_lock);
+	}
+}
+
+static void __unregister_request(struct ceph_mds_client *mdsc,
+				 struct ceph_mds_request *req)
+{
+	dout("__unregister_request %p tid %lld\n", req, req->r_tid);
+
+	/* Never leave an unregistered request on an unsafe list! */
+	list_del_init(&req->r_unsafe_item);
+
+	if (req->r_tid == mdsc->oldest_tid) {
+		struct rb_node *p = rb_next(&req->r_node);
+		mdsc->oldest_tid = 0;
+		while (p) {
+			struct ceph_mds_request *next_req =
+				rb_entry(p, struct ceph_mds_request, r_node);
+			if (next_req->r_op != CEPH_MDS_OP_SETFILELOCK) {
+				mdsc->oldest_tid = next_req->r_tid;
+				break;
+			}
+			p = rb_next(p);
+		}
+	}
+
+	erase_request(&mdsc->request_tree, req);
+
+	if (req->r_unsafe_dir) {
+		struct ceph_inode_info *ci = ceph_inode(req->r_unsafe_dir);
+		spin_lock(&ci->i_unsafe_lock);
+		list_del_init(&req->r_unsafe_dir_item);
+		spin_unlock(&ci->i_unsafe_lock);
+	}
+	if (req->r_target_inode &&
+	    test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
+		struct ceph_inode_info *ci = ceph_inode(req->r_target_inode);
+		spin_lock(&ci->i_unsafe_lock);
+		list_del_init(&req->r_unsafe_target_item);
+		spin_unlock(&ci->i_unsafe_lock);
+	}
+
+	if (req->r_unsafe_dir) {
+		iput(req->r_unsafe_dir);
+		req->r_unsafe_dir = NULL;
+	}
+
+	complete_all(&req->r_safe_completion);
+
+	ceph_mdsc_put_request(req);
+}
+
+/*
+ * Walk back up the dentry tree until we hit a dentry representing a
+ * non-snapshot inode. We do this using the rcu_read_lock (which must be held
+ * when calling this) to ensure that the objects won't disappear while we're
+ * working with them. Once we hit a candidate dentry, we attempt to take a
+ * reference to it, and return that as the result.
+ */
+static struct inode *get_nonsnap_parent(struct dentry *dentry)
+{
+	struct inode *inode = NULL;
+
+	while (dentry && !IS_ROOT(dentry)) {
+		inode = d_inode_rcu(dentry);
+		if (!inode || ceph_snap(inode) == CEPH_NOSNAP)
+			break;
+		dentry = dentry->d_parent;
+	}
+	if (inode)
+		inode = igrab(inode);
+	return inode;
+}
+
+/*
+ * Choose mds to send request to next.  If there is a hint set in the
+ * request (e.g., due to a prior forward hint from the mds), use that.
+ * Otherwise, consult frag tree and/or caps to identify the
+ * appropriate mds.  If all else fails, choose randomly.
+ *
+ * Called under mdsc->mutex.
+ */
+static int __choose_mds(struct ceph_mds_client *mdsc,
+			struct ceph_mds_request *req,
+			bool *random)
+{
+	struct inode *inode;
+	struct ceph_inode_info *ci;
+	struct ceph_cap *cap;
+	int mode = req->r_direct_mode;
+	int mds = -1;
+	u32 hash = req->r_direct_hash;
+	bool is_hash = test_bit(CEPH_MDS_R_DIRECT_IS_HASH, &req->r_req_flags);
+
+	if (random)
+		*random = false;
+
+	/*
+	 * is there a specific mds we should try?  ignore hint if we have
+	 * no session and the mds is not up (active or recovering).
+	 */
+	if (req->r_resend_mds >= 0 &&
+	    (__have_session(mdsc, req->r_resend_mds) ||
+	     ceph_mdsmap_get_state(mdsc->mdsmap, req->r_resend_mds) > 0)) {
+		dout("%s using resend_mds mds%d\n", __func__,
+		     req->r_resend_mds);
+		return req->r_resend_mds;
+	}
+
+	if (mode == USE_RANDOM_MDS)
+		goto random;
+
+	inode = NULL;
+	if (req->r_inode) {
+		if (ceph_snap(req->r_inode) != CEPH_SNAPDIR) {
+			inode = req->r_inode;
+			ihold(inode);
+		} else {
+			/* req->r_dentry is non-null for LSSNAP request */
+			rcu_read_lock();
+			inode = get_nonsnap_parent(req->r_dentry);
+			rcu_read_unlock();
+			dout("%s using snapdir's parent %p\n", __func__, inode);
+		}
+	} else if (req->r_dentry) {
+		/* ignore race with rename; old or new d_parent is okay */
+		struct dentry *parent;
+		struct inode *dir;
+
+		rcu_read_lock();
+		parent = READ_ONCE(req->r_dentry->d_parent);
+		dir = req->r_parent ? : d_inode_rcu(parent);
+
+		if (!dir || dir->i_sb != mdsc->fsc->sb) {
+			/*  not this fs or parent went negative */
+			inode = d_inode(req->r_dentry);
+			if (inode)
+				ihold(inode);
+		} else if (ceph_snap(dir) != CEPH_NOSNAP) {
+			/* direct snapped/virtual snapdir requests
+			 * based on parent dir inode */
+			inode = get_nonsnap_parent(parent);
+			dout("%s using nonsnap parent %p\n", __func__, inode);
+		} else {
+			/* dentry target */
+			inode = d_inode(req->r_dentry);
+			if (!inode || mode == USE_AUTH_MDS) {
+				/* dir + name */
+				inode = igrab(dir);
+				hash = ceph_dentry_hash(dir, req->r_dentry);
+				is_hash = true;
+			} else {
+				ihold(inode);
+			}
+		}
+		rcu_read_unlock();
+	}
+
+	dout("%s %p is_hash=%d (0x%x) mode %d\n", __func__, inode, (int)is_hash,
+	     hash, mode);
+	if (!inode)
+		goto random;
+	ci = ceph_inode(inode);
+
+	if (is_hash && S_ISDIR(inode->i_mode)) {
+		struct ceph_inode_frag frag;
+		int found;
+
+		ceph_choose_frag(ci, hash, &frag, &found);
+		if (found) {
+			if (mode == USE_ANY_MDS && frag.ndist > 0) {
+				u8 r;
+
+				/* choose a random replica */
+				get_random_bytes(&r, 1);
+				r %= frag.ndist;
+				mds = frag.dist[r];
+				dout("%s %p %llx.%llx frag %u mds%d (%d/%d)\n",
+				     __func__, inode, ceph_vinop(inode),
+				     frag.frag, mds, (int)r, frag.ndist);
+				if (ceph_mdsmap_get_state(mdsc->mdsmap, mds) >=
+				    CEPH_MDS_STATE_ACTIVE &&
+				    !ceph_mdsmap_is_laggy(mdsc->mdsmap, mds))
+					goto out;
+			}
+
+			/* since this file/dir wasn't known to be
+			 * replicated, then we want to look for the
+			 * authoritative mds. */
+			if (frag.mds >= 0) {
+				/* choose auth mds */
+				mds = frag.mds;
+				dout("%s %p %llx.%llx frag %u mds%d (auth)\n",
+				     __func__, inode, ceph_vinop(inode),
+				     frag.frag, mds);
+				if (ceph_mdsmap_get_state(mdsc->mdsmap, mds) >=
+				    CEPH_MDS_STATE_ACTIVE) {
+					if (!ceph_mdsmap_is_laggy(mdsc->mdsmap,
+								  mds))
+						goto out;
+				}
+			}
+			mode = USE_AUTH_MDS;
+		}
+	}
+
+	spin_lock(&ci->i_ceph_lock);
+	cap = NULL;
+	if (mode == USE_AUTH_MDS)
+		cap = ci->i_auth_cap;
+	if (!cap && !RB_EMPTY_ROOT(&ci->i_caps))
+		cap = rb_entry(rb_first(&ci->i_caps), struct ceph_cap, ci_node);
+	if (!cap) {
+		spin_unlock(&ci->i_ceph_lock);
+		iput(inode);
+		goto random;
+	}
+	mds = cap->session->s_mds;
+	dout("%s %p %llx.%llx mds%d (%scap %p)\n", __func__,
+	     inode, ceph_vinop(inode), mds,
+	     cap == ci->i_auth_cap ? "auth " : "", cap);
+	spin_unlock(&ci->i_ceph_lock);
+out:
+	iput(inode);
+	return mds;
+
+random:
+	if (random)
+		*random = true;
+
+	mds = ceph_mdsmap_get_random_mds(mdsc->mdsmap);
+	dout("%s chose random mds%d\n", __func__, mds);
+	return mds;
+}
+
+
+/*
+ * session messages
+ */
+struct ceph_msg *ceph_create_session_msg(u32 op, u64 seq)
+{
+	struct ceph_msg *msg;
+	struct ceph_mds_session_head *h;
+
+	msg = ceph_msg_new(CEPH_MSG_CLIENT_SESSION, sizeof(*h), GFP_NOFS,
+			   false);
+	if (!msg) {
+		pr_err("ENOMEM creating session %s msg\n",
+		       ceph_session_op_name(op));
+		return NULL;
+	}
+	h = msg->front.iov_base;
+	h->op = cpu_to_le32(op);
+	h->seq = cpu_to_le64(seq);
+
+	return msg;
+}
+
+static const unsigned char feature_bits[] = CEPHFS_FEATURES_CLIENT_SUPPORTED;
+#define FEATURE_BYTES(c) (DIV_ROUND_UP((size_t)feature_bits[c - 1] + 1, 64) * 8)
+static int encode_supported_features(void **p, void *end)
+{
+	static const size_t count = ARRAY_SIZE(feature_bits);
+
+	if (count > 0) {
+		size_t i;
+		size_t size = FEATURE_BYTES(count);
+		unsigned long bit;
+
+		if (WARN_ON_ONCE(*p + 4 + size > end))
+			return -ERANGE;
+
+		ceph_encode_32(p, size);
+		memset(*p, 0, size);
+		for (i = 0; i < count; i++) {
+			bit = feature_bits[i];
+			((unsigned char *)(*p))[bit / 8] |= BIT(bit % 8);
+		}
+		*p += size;
+	} else {
+		if (WARN_ON_ONCE(*p + 4 > end))
+			return -ERANGE;
+
+		ceph_encode_32(p, 0);
+	}
+
+	return 0;
+}
+
+static const unsigned char metric_bits[] = CEPHFS_METRIC_SPEC_CLIENT_SUPPORTED;
+#define METRIC_BYTES(cnt) (DIV_ROUND_UP((size_t)metric_bits[cnt - 1] + 1, 64) * 8)
+static int encode_metric_spec(void **p, void *end)
+{
+	static const size_t count = ARRAY_SIZE(metric_bits);
+
+	/* header */
+	if (WARN_ON_ONCE(*p + 2 > end))
+		return -ERANGE;
+
+	ceph_encode_8(p, 1); /* version */
+	ceph_encode_8(p, 1); /* compat */
+
+	if (count > 0) {
+		size_t i;
+		size_t size = METRIC_BYTES(count);
+
+		if (WARN_ON_ONCE(*p + 4 + 4 + size > end))
+			return -ERANGE;
+
+		/* metric spec info length */
+		ceph_encode_32(p, 4 + size);
+
+		/* metric spec */
+		ceph_encode_32(p, size);
+		memset(*p, 0, size);
+		for (i = 0; i < count; i++)
+			((unsigned char *)(*p))[i / 8] |= BIT(metric_bits[i] % 8);
+		*p += size;
+	} else {
+		if (WARN_ON_ONCE(*p + 4 + 4 > end))
+			return -ERANGE;
+
+		/* metric spec info length */
+		ceph_encode_32(p, 4);
+		/* metric spec */
+		ceph_encode_32(p, 0);
+	}
+
+	return 0;
+}
+
+/*
+ * session message, specialization for CEPH_SESSION_REQUEST_OPEN
+ * to include additional client metadata fields.
+ */
+static struct ceph_msg *create_session_open_msg(struct ceph_mds_client *mdsc, u64 seq)
+{
+	struct ceph_msg *msg;
+	struct ceph_mds_session_head *h;
+	int i;
+	int extra_bytes = 0;
+	int metadata_key_count = 0;
+	struct ceph_options *opt = mdsc->fsc->client->options;
+	struct ceph_mount_options *fsopt = mdsc->fsc->mount_options;
+	size_t size, count;
+	void *p, *end;
+	int ret;
+
+	const char* metadata[][2] = {
+		{"hostname", mdsc->nodename},
+		{"kernel_version", init_utsname()->release},
+		{"entity_id", opt->name ? : ""},
+		{"root", fsopt->server_path ? : "/"},
+		{NULL, NULL}
+	};
+
+	/* Calculate serialized length of metadata */
+	extra_bytes = 4;  /* map length */
+	for (i = 0; metadata[i][0]; ++i) {
+		extra_bytes += 8 + strlen(metadata[i][0]) +
+			strlen(metadata[i][1]);
+		metadata_key_count++;
+	}
+
+	/* supported feature */
+	size = 0;
+	count = ARRAY_SIZE(feature_bits);
+	if (count > 0)
+		size = FEATURE_BYTES(count);
+	extra_bytes += 4 + size;
+
+	/* metric spec */
+	size = 0;
+	count = ARRAY_SIZE(metric_bits);
+	if (count > 0)
+		size = METRIC_BYTES(count);
+	extra_bytes += 2 + 4 + 4 + size;
+
+	/* Allocate the message */
+	msg = ceph_msg_new(CEPH_MSG_CLIENT_SESSION, sizeof(*h) + extra_bytes,
+			   GFP_NOFS, false);
+	if (!msg) {
+		pr_err("ENOMEM creating session open msg\n");
+		return ERR_PTR(-ENOMEM);
+	}
+	p = msg->front.iov_base;
+	end = p + msg->front.iov_len;
+
+	h = p;
+	h->op = cpu_to_le32(CEPH_SESSION_REQUEST_OPEN);
+	h->seq = cpu_to_le64(seq);
+
+	/*
+	 * Serialize client metadata into waiting buffer space, using
+	 * the format that userspace expects for map<string, string>
+	 *
+	 * ClientSession messages with metadata are v4
+	 */
+	msg->hdr.version = cpu_to_le16(4);
+	msg->hdr.compat_version = cpu_to_le16(1);
+
+	/* The write pointer, following the session_head structure */
+	p += sizeof(*h);
+
+	/* Number of entries in the map */
+	ceph_encode_32(&p, metadata_key_count);
+
+	/* Two length-prefixed strings for each entry in the map */
+	for (i = 0; metadata[i][0]; ++i) {
+		size_t const key_len = strlen(metadata[i][0]);
+		size_t const val_len = strlen(metadata[i][1]);
+
+		ceph_encode_32(&p, key_len);
+		memcpy(p, metadata[i][0], key_len);
+		p += key_len;
+		ceph_encode_32(&p, val_len);
+		memcpy(p, metadata[i][1], val_len);
+		p += val_len;
+	}
+
+	ret = encode_supported_features(&p, end);
+	if (ret) {
+		pr_err("encode_supported_features failed!\n");
+		ceph_msg_put(msg);
+		return ERR_PTR(ret);
+	}
+
+	ret = encode_metric_spec(&p, end);
+	if (ret) {
+		pr_err("encode_metric_spec failed!\n");
+		ceph_msg_put(msg);
+		return ERR_PTR(ret);
+	}
+
+	msg->front.iov_len = p - msg->front.iov_base;
+	msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
+
+	return msg;
+}
+
+/*
+ * send session open request.
+ *
+ * called under mdsc->mutex
+ */
+static int __open_session(struct ceph_mds_client *mdsc,
+			  struct ceph_mds_session *session)
+{
+	struct ceph_msg *msg;
+	int mstate;
+	int mds = session->s_mds;
+
+	if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_FENCE_IO)
+		return -EIO;
+
+	/* wait for mds to go active? */
+	mstate = ceph_mdsmap_get_state(mdsc->mdsmap, mds);
+	dout("open_session to mds%d (%s)\n", mds,
+	     ceph_mds_state_name(mstate));
+	session->s_state = CEPH_MDS_SESSION_OPENING;
+	session->s_renew_requested = jiffies;
+
+	/* send connect message */
+	msg = create_session_open_msg(mdsc, session->s_seq);
+	if (IS_ERR(msg))
+		return PTR_ERR(msg);
+	ceph_con_send(&session->s_con, msg);
+	return 0;
+}
+
+/*
+ * open sessions for any export targets for the given mds
+ *
+ * called under mdsc->mutex
+ */
+static struct ceph_mds_session *
+__open_export_target_session(struct ceph_mds_client *mdsc, int target)
+{
+	struct ceph_mds_session *session;
+	int ret;
+
+	session = __ceph_lookup_mds_session(mdsc, target);
+	if (!session) {
+		session = register_session(mdsc, target);
+		if (IS_ERR(session))
+			return session;
+	}
+	if (session->s_state == CEPH_MDS_SESSION_NEW ||
+	    session->s_state == CEPH_MDS_SESSION_CLOSING) {
+		ret = __open_session(mdsc, session);
+		if (ret)
+			return ERR_PTR(ret);
+	}
+
+	return session;
+}
+
+struct ceph_mds_session *
+ceph_mdsc_open_export_target_session(struct ceph_mds_client *mdsc, int target)
+{
+	struct ceph_mds_session *session;
+
+	dout("open_export_target_session to mds%d\n", target);
+
+	mutex_lock(&mdsc->mutex);
+	session = __open_export_target_session(mdsc, target);
+	mutex_unlock(&mdsc->mutex);
+
+	return session;
+}
+
+static void __open_export_target_sessions(struct ceph_mds_client *mdsc,
+					  struct ceph_mds_session *session)
+{
+	struct ceph_mds_info *mi;
+	struct ceph_mds_session *ts;
+	int i, mds = session->s_mds;
+
+	if (mds >= mdsc->mdsmap->possible_max_rank)
+		return;
+
+	mi = &mdsc->mdsmap->m_info[mds];
+	dout("open_export_target_sessions for mds%d (%d targets)\n",
+	     session->s_mds, mi->num_export_targets);
+
+	for (i = 0; i < mi->num_export_targets; i++) {
+		ts = __open_export_target_session(mdsc, mi->export_targets[i]);
+		ceph_put_mds_session(ts);
+	}
+}
+
+void ceph_mdsc_open_export_target_sessions(struct ceph_mds_client *mdsc,
+					   struct ceph_mds_session *session)
+{
+	mutex_lock(&mdsc->mutex);
+	__open_export_target_sessions(mdsc, session);
+	mutex_unlock(&mdsc->mutex);
+}
+
+/*
+ * session caps
+ */
+
+static void detach_cap_releases(struct ceph_mds_session *session,
+				struct list_head *target)
+{
+	lockdep_assert_held(&session->s_cap_lock);
+
+	list_splice_init(&session->s_cap_releases, target);
+	session->s_num_cap_releases = 0;
+	dout("dispose_cap_releases mds%d\n", session->s_mds);
+}
+
+static void dispose_cap_releases(struct ceph_mds_client *mdsc,
+				 struct list_head *dispose)
+{
+	while (!list_empty(dispose)) {
+		struct ceph_cap *cap;
+		/* zero out the in-progress message */
+		cap = list_first_entry(dispose, struct ceph_cap, session_caps);
+		list_del(&cap->session_caps);
+		ceph_put_cap(mdsc, cap);
+	}
+}
+
+static void cleanup_session_requests(struct ceph_mds_client *mdsc,
+				     struct ceph_mds_session *session)
+{
+	struct ceph_mds_request *req;
+	struct rb_node *p;
+
+	dout("cleanup_session_requests mds%d\n", session->s_mds);
+	mutex_lock(&mdsc->mutex);
+	while (!list_empty(&session->s_unsafe)) {
+		req = list_first_entry(&session->s_unsafe,
+				       struct ceph_mds_request, r_unsafe_item);
+		pr_warn_ratelimited(" dropping unsafe request %llu\n",
+				    req->r_tid);
+		if (req->r_target_inode)
+			mapping_set_error(req->r_target_inode->i_mapping, -EIO);
+		if (req->r_unsafe_dir)
+			mapping_set_error(req->r_unsafe_dir->i_mapping, -EIO);
+		__unregister_request(mdsc, req);
+	}
+	/* zero r_attempts, so kick_requests() will re-send requests */
+	p = rb_first(&mdsc->request_tree);
+	while (p) {
+		req = rb_entry(p, struct ceph_mds_request, r_node);
+		p = rb_next(p);
+		if (req->r_session &&
+		    req->r_session->s_mds == session->s_mds)
+			req->r_attempts = 0;
+	}
+	mutex_unlock(&mdsc->mutex);
+}
+
+/*
+ * Helper to safely iterate over all caps associated with a session, with
+ * special care taken to handle a racing __ceph_remove_cap().
+ *
+ * Caller must hold session s_mutex.
+ */
+int ceph_iterate_session_caps(struct ceph_mds_session *session,
+			      int (*cb)(struct inode *, int mds, void *),
+			      void *arg)
+{
+	struct list_head *p;
+	struct ceph_cap *cap;
+	struct inode *inode, *last_inode = NULL;
+	struct ceph_cap *old_cap = NULL;
+	int ret;
+
+	dout("iterate_session_caps %p mds%d\n", session, session->s_mds);
+	spin_lock(&session->s_cap_lock);
+	p = session->s_caps.next;
+	while (p != &session->s_caps) {
+		int mds;
+
+		cap = list_entry(p, struct ceph_cap, session_caps);
+		inode = igrab(&cap->ci->netfs.inode);
+		if (!inode) {
+			p = p->next;
+			continue;
+		}
+		session->s_cap_iterator = cap;
+		mds = cap->mds;
+		spin_unlock(&session->s_cap_lock);
+
+		if (last_inode) {
+			iput(last_inode);
+			last_inode = NULL;
+		}
+		if (old_cap) {
+			ceph_put_cap(session->s_mdsc, old_cap);
+			old_cap = NULL;
+		}
+
+		ret = cb(inode, mds, arg);
+		last_inode = inode;
+
+		spin_lock(&session->s_cap_lock);
+		p = p->next;
+		if (!cap->ci) {
+			dout("iterate_session_caps  finishing cap %p removal\n",
+			     cap);
+			BUG_ON(cap->session != session);
+			cap->session = NULL;
+			list_del_init(&cap->session_caps);
+			session->s_nr_caps--;
+			atomic64_dec(&session->s_mdsc->metric.total_caps);
+			if (cap->queue_release)
+				__ceph_queue_cap_release(session, cap);
+			else
+				old_cap = cap;  /* put_cap it w/o locks held */
+		}
+		if (ret < 0)
+			goto out;
+	}
+	ret = 0;
+out:
+	session->s_cap_iterator = NULL;
+	spin_unlock(&session->s_cap_lock);
+
+	iput(last_inode);
+	if (old_cap)
+		ceph_put_cap(session->s_mdsc, old_cap);
+
+	return ret;
+}
+
+static int remove_session_caps_cb(struct inode *inode, int mds, void *arg)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	bool invalidate = false;
+	struct ceph_cap *cap;
+	int iputs = 0;
+
+	spin_lock(&ci->i_ceph_lock);
+	cap = __get_cap_for_mds(ci, mds);
+	if (cap) {
+		dout(" removing cap %p, ci is %p, inode is %p\n",
+		     cap, ci, &ci->netfs.inode);
+
+		iputs = ceph_purge_inode_cap(inode, cap, &invalidate);
+	}
+	spin_unlock(&ci->i_ceph_lock);
+
+	if (cap)
+		wake_up_all(&ci->i_cap_wq);
+	if (invalidate)
+		ceph_queue_invalidate(inode);
+	while (iputs--)
+		iput(inode);
+	return 0;
+}
+
+/*
+ * caller must hold session s_mutex
+ */
+static void remove_session_caps(struct ceph_mds_session *session)
+{
+	struct ceph_fs_client *fsc = session->s_mdsc->fsc;
+	struct super_block *sb = fsc->sb;
+	LIST_HEAD(dispose);
+
+	dout("remove_session_caps on %p\n", session);
+	ceph_iterate_session_caps(session, remove_session_caps_cb, fsc);
+
+	wake_up_all(&fsc->mdsc->cap_flushing_wq);
+
+	spin_lock(&session->s_cap_lock);
+	if (session->s_nr_caps > 0) {
+		struct inode *inode;
+		struct ceph_cap *cap, *prev = NULL;
+		struct ceph_vino vino;
+		/*
+		 * iterate_session_caps() skips inodes that are being
+		 * deleted, we need to wait until deletions are complete.
+		 * __wait_on_freeing_inode() is designed for the job,
+		 * but it is not exported, so use lookup inode function
+		 * to access it.
+		 */
+		while (!list_empty(&session->s_caps)) {
+			cap = list_entry(session->s_caps.next,
+					 struct ceph_cap, session_caps);
+			if (cap == prev)
+				break;
+			prev = cap;
+			vino = cap->ci->i_vino;
+			spin_unlock(&session->s_cap_lock);
+
+			inode = ceph_find_inode(sb, vino);
+			iput(inode);
+
+			spin_lock(&session->s_cap_lock);
+		}
+	}
+
+	// drop cap expires and unlock s_cap_lock
+	detach_cap_releases(session, &dispose);
+
+	BUG_ON(session->s_nr_caps > 0);
+	BUG_ON(!list_empty(&session->s_cap_flushing));
+	spin_unlock(&session->s_cap_lock);
+	dispose_cap_releases(session->s_mdsc, &dispose);
+}
+
+enum {
+	RECONNECT,
+	RENEWCAPS,
+	FORCE_RO,
+};
+
+/*
+ * wake up any threads waiting on this session's caps.  if the cap is
+ * old (didn't get renewed on the client reconnect), remove it now.
+ *
+ * caller must hold s_mutex.
+ */
+static int wake_up_session_cb(struct inode *inode, int mds, void *arg)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	unsigned long ev = (unsigned long)arg;
+
+	if (ev == RECONNECT) {
+		spin_lock(&ci->i_ceph_lock);
+		ci->i_wanted_max_size = 0;
+		ci->i_requested_max_size = 0;
+		spin_unlock(&ci->i_ceph_lock);
+	} else if (ev == RENEWCAPS) {
+		struct ceph_cap *cap;
+
+		spin_lock(&ci->i_ceph_lock);
+		cap = __get_cap_for_mds(ci, mds);
+		/* mds did not re-issue stale cap */
+		if (cap && cap->cap_gen < atomic_read(&cap->session->s_cap_gen))
+			cap->issued = cap->implemented = CEPH_CAP_PIN;
+		spin_unlock(&ci->i_ceph_lock);
+	} else if (ev == FORCE_RO) {
+	}
+	wake_up_all(&ci->i_cap_wq);
+	return 0;
+}
+
+static void wake_up_session_caps(struct ceph_mds_session *session, int ev)
+{
+	dout("wake_up_session_caps %p mds%d\n", session, session->s_mds);
+	ceph_iterate_session_caps(session, wake_up_session_cb,
+				  (void *)(unsigned long)ev);
+}
+
+/*
+ * Send periodic message to MDS renewing all currently held caps.  The
+ * ack will reset the expiration for all caps from this session.
+ *
+ * caller holds s_mutex
+ */
+static int send_renew_caps(struct ceph_mds_client *mdsc,
+			   struct ceph_mds_session *session)
+{
+	struct ceph_msg *msg;
+	int state;
+
+	if (time_after_eq(jiffies, session->s_cap_ttl) &&
+	    time_after_eq(session->s_cap_ttl, session->s_renew_requested))
+		pr_info("mds%d caps stale\n", session->s_mds);
+	session->s_renew_requested = jiffies;
+
+	/* do not try to renew caps until a recovering mds has reconnected
+	 * with its clients. */
+	state = ceph_mdsmap_get_state(mdsc->mdsmap, session->s_mds);
+	if (state < CEPH_MDS_STATE_RECONNECT) {
+		dout("send_renew_caps ignoring mds%d (%s)\n",
+		     session->s_mds, ceph_mds_state_name(state));
+		return 0;
+	}
+
+	dout("send_renew_caps to mds%d (%s)\n", session->s_mds,
+		ceph_mds_state_name(state));
+	msg = ceph_create_session_msg(CEPH_SESSION_REQUEST_RENEWCAPS,
+				      ++session->s_renew_seq);
+	if (!msg)
+		return -ENOMEM;
+	ceph_con_send(&session->s_con, msg);
+	return 0;
+}
+
+static int send_flushmsg_ack(struct ceph_mds_client *mdsc,
+			     struct ceph_mds_session *session, u64 seq)
+{
+	struct ceph_msg *msg;
+
+	dout("send_flushmsg_ack to mds%d (%s)s seq %lld\n",
+	     session->s_mds, ceph_session_state_name(session->s_state), seq);
+	msg = ceph_create_session_msg(CEPH_SESSION_FLUSHMSG_ACK, seq);
+	if (!msg)
+		return -ENOMEM;
+	ceph_con_send(&session->s_con, msg);
+	return 0;
+}
+
+
+/*
+ * Note new cap ttl, and any transition from stale -> not stale (fresh?).
+ *
+ * Called under session->s_mutex
+ */
+static void renewed_caps(struct ceph_mds_client *mdsc,
+			 struct ceph_mds_session *session, int is_renew)
+{
+	int was_stale;
+	int wake = 0;
+
+	spin_lock(&session->s_cap_lock);
+	was_stale = is_renew && time_after_eq(jiffies, session->s_cap_ttl);
+
+	session->s_cap_ttl = session->s_renew_requested +
+		mdsc->mdsmap->m_session_timeout*HZ;
+
+	if (was_stale) {
+		if (time_before(jiffies, session->s_cap_ttl)) {
+			pr_info("mds%d caps renewed\n", session->s_mds);
+			wake = 1;
+		} else {
+			pr_info("mds%d caps still stale\n", session->s_mds);
+		}
+	}
+	dout("renewed_caps mds%d ttl now %lu, was %s, now %s\n",
+	     session->s_mds, session->s_cap_ttl, was_stale ? "stale" : "fresh",
+	     time_before(jiffies, session->s_cap_ttl) ? "stale" : "fresh");
+	spin_unlock(&session->s_cap_lock);
+
+	if (wake)
+		wake_up_session_caps(session, RENEWCAPS);
+}
+
+/*
+ * send a session close request
+ */
+static int request_close_session(struct ceph_mds_session *session)
+{
+	struct ceph_msg *msg;
+
+	dout("request_close_session mds%d state %s seq %lld\n",
+	     session->s_mds, ceph_session_state_name(session->s_state),
+	     session->s_seq);
+	msg = ceph_create_session_msg(CEPH_SESSION_REQUEST_CLOSE,
+				      session->s_seq);
+	if (!msg)
+		return -ENOMEM;
+	ceph_con_send(&session->s_con, msg);
+	return 1;
+}
+
+/*
+ * Called with s_mutex held.
+ */
+static int __close_session(struct ceph_mds_client *mdsc,
+			 struct ceph_mds_session *session)
+{
+	if (session->s_state >= CEPH_MDS_SESSION_CLOSING)
+		return 0;
+	session->s_state = CEPH_MDS_SESSION_CLOSING;
+	return request_close_session(session);
+}
+
+static bool drop_negative_children(struct dentry *dentry)
+{
+	struct dentry *child;
+	bool all_negative = true;
+
+	if (!d_is_dir(dentry))
+		goto out;
+
+	spin_lock(&dentry->d_lock);
+	list_for_each_entry(child, &dentry->d_subdirs, d_child) {
+		if (d_really_is_positive(child)) {
+			all_negative = false;
+			break;
+		}
+	}
+	spin_unlock(&dentry->d_lock);
+
+	if (all_negative)
+		shrink_dcache_parent(dentry);
+out:
+	return all_negative;
+}
+
+/*
+ * Trim old(er) caps.
+ *
+ * Because we can't cache an inode without one or more caps, we do
+ * this indirectly: if a cap is unused, we prune its aliases, at which
+ * point the inode will hopefully get dropped to.
+ *
+ * Yes, this is a bit sloppy.  Our only real goal here is to respond to
+ * memory pressure from the MDS, though, so it needn't be perfect.
+ */
+static int trim_caps_cb(struct inode *inode, int mds, void *arg)
+{
+	int *remaining = arg;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	int used, wanted, oissued, mine;
+	struct ceph_cap *cap;
+
+	if (*remaining <= 0)
+		return -1;
+
+	spin_lock(&ci->i_ceph_lock);
+	cap = __get_cap_for_mds(ci, mds);
+	if (!cap) {
+		spin_unlock(&ci->i_ceph_lock);
+		return 0;
+	}
+	mine = cap->issued | cap->implemented;
+	used = __ceph_caps_used(ci);
+	wanted = __ceph_caps_file_wanted(ci);
+	oissued = __ceph_caps_issued_other(ci, cap);
+
+	dout("trim_caps_cb %p cap %p mine %s oissued %s used %s wanted %s\n",
+	     inode, cap, ceph_cap_string(mine), ceph_cap_string(oissued),
+	     ceph_cap_string(used), ceph_cap_string(wanted));
+	if (cap == ci->i_auth_cap) {
+		if (ci->i_dirty_caps || ci->i_flushing_caps ||
+		    !list_empty(&ci->i_cap_snaps))
+			goto out;
+		if ((used | wanted) & CEPH_CAP_ANY_WR)
+			goto out;
+		/* Note: it's possible that i_filelock_ref becomes non-zero
+		 * after dropping auth caps. It doesn't hurt because reply
+		 * of lock mds request will re-add auth caps. */
+		if (atomic_read(&ci->i_filelock_ref) > 0)
+			goto out;
+	}
+	/* The inode has cached pages, but it's no longer used.
+	 * we can safely drop it */
+	if (S_ISREG(inode->i_mode) &&
+	    wanted == 0 && used == CEPH_CAP_FILE_CACHE &&
+	    !(oissued & CEPH_CAP_FILE_CACHE)) {
+	  used = 0;
+	  oissued = 0;
+	}
+	if ((used | wanted) & ~oissued & mine)
+		goto out;   /* we need these caps */
+
+	if (oissued) {
+		/* we aren't the only cap.. just remove us */
+		ceph_remove_cap(cap, true);
+		(*remaining)--;
+	} else {
+		struct dentry *dentry;
+		/* try dropping referring dentries */
+		spin_unlock(&ci->i_ceph_lock);
+		dentry = d_find_any_alias(inode);
+		if (dentry && drop_negative_children(dentry)) {
+			int count;
+			dput(dentry);
+			d_prune_aliases(inode);
+			count = atomic_read(&inode->i_count);
+			if (count == 1)
+				(*remaining)--;
+			dout("trim_caps_cb %p cap %p pruned, count now %d\n",
+			     inode, cap, count);
+		} else {
+			dput(dentry);
+		}
+		return 0;
+	}
+
+out:
+	spin_unlock(&ci->i_ceph_lock);
+	return 0;
+}
+
+/*
+ * Trim session cap count down to some max number.
+ */
+int ceph_trim_caps(struct ceph_mds_client *mdsc,
+		   struct ceph_mds_session *session,
+		   int max_caps)
+{
+	int trim_caps = session->s_nr_caps - max_caps;
+
+	dout("trim_caps mds%d start: %d / %d, trim %d\n",
+	     session->s_mds, session->s_nr_caps, max_caps, trim_caps);
+	if (trim_caps > 0) {
+		int remaining = trim_caps;
+
+		ceph_iterate_session_caps(session, trim_caps_cb, &remaining);
+		dout("trim_caps mds%d done: %d / %d, trimmed %d\n",
+		     session->s_mds, session->s_nr_caps, max_caps,
+			trim_caps - remaining);
+	}
+
+	ceph_flush_cap_releases(mdsc, session);
+	return 0;
+}
+
+static int check_caps_flush(struct ceph_mds_client *mdsc,
+			    u64 want_flush_tid)
+{
+	int ret = 1;
+
+	spin_lock(&mdsc->cap_dirty_lock);
+	if (!list_empty(&mdsc->cap_flush_list)) {
+		struct ceph_cap_flush *cf =
+			list_first_entry(&mdsc->cap_flush_list,
+					 struct ceph_cap_flush, g_list);
+		if (cf->tid <= want_flush_tid) {
+			dout("check_caps_flush still flushing tid "
+			     "%llu <= %llu\n", cf->tid, want_flush_tid);
+			ret = 0;
+		}
+	}
+	spin_unlock(&mdsc->cap_dirty_lock);
+	return ret;
+}
+
+/*
+ * flush all dirty inode data to disk.
+ *
+ * returns true if we've flushed through want_flush_tid
+ */
+static void wait_caps_flush(struct ceph_mds_client *mdsc,
+			    u64 want_flush_tid)
+{
+	dout("check_caps_flush want %llu\n", want_flush_tid);
+
+	wait_event(mdsc->cap_flushing_wq,
+		   check_caps_flush(mdsc, want_flush_tid));
+
+	dout("check_caps_flush ok, flushed thru %llu\n", want_flush_tid);
+}
+
+/*
+ * called under s_mutex
+ */
+static void ceph_send_cap_releases(struct ceph_mds_client *mdsc,
+				   struct ceph_mds_session *session)
+{
+	struct ceph_msg *msg = NULL;
+	struct ceph_mds_cap_release *head;
+	struct ceph_mds_cap_item *item;
+	struct ceph_osd_client *osdc = &mdsc->fsc->client->osdc;
+	struct ceph_cap *cap;
+	LIST_HEAD(tmp_list);
+	int num_cap_releases;
+	__le32	barrier, *cap_barrier;
+
+	down_read(&osdc->lock);
+	barrier = cpu_to_le32(osdc->epoch_barrier);
+	up_read(&osdc->lock);
+
+	spin_lock(&session->s_cap_lock);
+again:
+	list_splice_init(&session->s_cap_releases, &tmp_list);
+	num_cap_releases = session->s_num_cap_releases;
+	session->s_num_cap_releases = 0;
+	spin_unlock(&session->s_cap_lock);
+
+	while (!list_empty(&tmp_list)) {
+		if (!msg) {
+			msg = ceph_msg_new(CEPH_MSG_CLIENT_CAPRELEASE,
+					PAGE_SIZE, GFP_NOFS, false);
+			if (!msg)
+				goto out_err;
+			head = msg->front.iov_base;
+			head->num = cpu_to_le32(0);
+			msg->front.iov_len = sizeof(*head);
+
+			msg->hdr.version = cpu_to_le16(2);
+			msg->hdr.compat_version = cpu_to_le16(1);
+		}
+
+		cap = list_first_entry(&tmp_list, struct ceph_cap,
+					session_caps);
+		list_del(&cap->session_caps);
+		num_cap_releases--;
+
+		head = msg->front.iov_base;
+		put_unaligned_le32(get_unaligned_le32(&head->num) + 1,
+				   &head->num);
+		item = msg->front.iov_base + msg->front.iov_len;
+		item->ino = cpu_to_le64(cap->cap_ino);
+		item->cap_id = cpu_to_le64(cap->cap_id);
+		item->migrate_seq = cpu_to_le32(cap->mseq);
+		item->seq = cpu_to_le32(cap->issue_seq);
+		msg->front.iov_len += sizeof(*item);
+
+		ceph_put_cap(mdsc, cap);
+
+		if (le32_to_cpu(head->num) == CEPH_CAPS_PER_RELEASE) {
+			// Append cap_barrier field
+			cap_barrier = msg->front.iov_base + msg->front.iov_len;
+			*cap_barrier = barrier;
+			msg->front.iov_len += sizeof(*cap_barrier);
+
+			msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
+			dout("send_cap_releases mds%d %p\n", session->s_mds, msg);
+			ceph_con_send(&session->s_con, msg);
+			msg = NULL;
+		}
+	}
+
+	BUG_ON(num_cap_releases != 0);
+
+	spin_lock(&session->s_cap_lock);
+	if (!list_empty(&session->s_cap_releases))
+		goto again;
+	spin_unlock(&session->s_cap_lock);
+
+	if (msg) {
+		// Append cap_barrier field
+		cap_barrier = msg->front.iov_base + msg->front.iov_len;
+		*cap_barrier = barrier;
+		msg->front.iov_len += sizeof(*cap_barrier);
+
+		msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
+		dout("send_cap_releases mds%d %p\n", session->s_mds, msg);
+		ceph_con_send(&session->s_con, msg);
+	}
+	return;
+out_err:
+	pr_err("send_cap_releases mds%d, failed to allocate message\n",
+		session->s_mds);
+	spin_lock(&session->s_cap_lock);
+	list_splice(&tmp_list, &session->s_cap_releases);
+	session->s_num_cap_releases += num_cap_releases;
+	spin_unlock(&session->s_cap_lock);
+}
+
+static void ceph_cap_release_work(struct work_struct *work)
+{
+	struct ceph_mds_session *session =
+		container_of(work, struct ceph_mds_session, s_cap_release_work);
+
+	mutex_lock(&session->s_mutex);
+	if (session->s_state == CEPH_MDS_SESSION_OPEN ||
+	    session->s_state == CEPH_MDS_SESSION_HUNG)
+		ceph_send_cap_releases(session->s_mdsc, session);
+	mutex_unlock(&session->s_mutex);
+	ceph_put_mds_session(session);
+}
+
+void ceph_flush_cap_releases(struct ceph_mds_client *mdsc,
+		             struct ceph_mds_session *session)
+{
+	if (mdsc->stopping)
+		return;
+
+	ceph_get_mds_session(session);
+	if (queue_work(mdsc->fsc->cap_wq,
+		       &session->s_cap_release_work)) {
+		dout("cap release work queued\n");
+	} else {
+		ceph_put_mds_session(session);
+		dout("failed to queue cap release work\n");
+	}
+}
+
+/*
+ * caller holds session->s_cap_lock
+ */
+void __ceph_queue_cap_release(struct ceph_mds_session *session,
+			      struct ceph_cap *cap)
+{
+	list_add_tail(&cap->session_caps, &session->s_cap_releases);
+	session->s_num_cap_releases++;
+
+	if (!(session->s_num_cap_releases % CEPH_CAPS_PER_RELEASE))
+		ceph_flush_cap_releases(session->s_mdsc, session);
+}
+
+static void ceph_cap_reclaim_work(struct work_struct *work)
+{
+	struct ceph_mds_client *mdsc =
+		container_of(work, struct ceph_mds_client, cap_reclaim_work);
+	int ret = ceph_trim_dentries(mdsc);
+	if (ret == -EAGAIN)
+		ceph_queue_cap_reclaim_work(mdsc);
+}
+
+void ceph_queue_cap_reclaim_work(struct ceph_mds_client *mdsc)
+{
+	if (mdsc->stopping)
+		return;
+
+        if (queue_work(mdsc->fsc->cap_wq, &mdsc->cap_reclaim_work)) {
+                dout("caps reclaim work queued\n");
+        } else {
+                dout("failed to queue caps release work\n");
+        }
+}
+
+void ceph_reclaim_caps_nr(struct ceph_mds_client *mdsc, int nr)
+{
+	int val;
+	if (!nr)
+		return;
+	val = atomic_add_return(nr, &mdsc->cap_reclaim_pending);
+	if ((val % CEPH_CAPS_PER_RELEASE) < nr) {
+		atomic_set(&mdsc->cap_reclaim_pending, 0);
+		ceph_queue_cap_reclaim_work(mdsc);
+	}
+}
+
+/*
+ * requests
+ */
+
+int ceph_alloc_readdir_reply_buffer(struct ceph_mds_request *req,
+				    struct inode *dir)
+{
+	struct ceph_inode_info *ci = ceph_inode(dir);
+	struct ceph_mds_reply_info_parsed *rinfo = &req->r_reply_info;
+	struct ceph_mount_options *opt = req->r_mdsc->fsc->mount_options;
+	size_t size = sizeof(struct ceph_mds_reply_dir_entry);
+	unsigned int num_entries;
+	int order;
+
+	spin_lock(&ci->i_ceph_lock);
+	num_entries = ci->i_files + ci->i_subdirs;
+	spin_unlock(&ci->i_ceph_lock);
+	num_entries = max(num_entries, 1U);
+	num_entries = min(num_entries, opt->max_readdir);
+
+	order = get_order(size * num_entries);
+	while (order >= 0) {
+		rinfo->dir_entries = (void*)__get_free_pages(GFP_KERNEL |
+							     __GFP_NOWARN |
+							     __GFP_ZERO,
+							     order);
+		if (rinfo->dir_entries)
+			break;
+		order--;
+	}
+	if (!rinfo->dir_entries)
+		return -ENOMEM;
+
+	num_entries = (PAGE_SIZE << order) / size;
+	num_entries = min(num_entries, opt->max_readdir);
+
+	rinfo->dir_buf_size = PAGE_SIZE << order;
+	req->r_num_caps = num_entries + 1;
+	req->r_args.readdir.max_entries = cpu_to_le32(num_entries);
+	req->r_args.readdir.max_bytes = cpu_to_le32(opt->max_readdir_bytes);
+	return 0;
+}
+
+/*
+ * Create an mds request.
+ */
+struct ceph_mds_request *
+ceph_mdsc_create_request(struct ceph_mds_client *mdsc, int op, int mode)
+{
+	struct ceph_mds_request *req;
+
+	req = kmem_cache_zalloc(ceph_mds_request_cachep, GFP_NOFS);
+	if (!req)
+		return ERR_PTR(-ENOMEM);
+
+	mutex_init(&req->r_fill_mutex);
+	req->r_mdsc = mdsc;
+	req->r_started = jiffies;
+	req->r_start_latency = ktime_get();
+	req->r_resend_mds = -1;
+	INIT_LIST_HEAD(&req->r_unsafe_dir_item);
+	INIT_LIST_HEAD(&req->r_unsafe_target_item);
+	req->r_fmode = -1;
+	req->r_feature_needed = -1;
+	kref_init(&req->r_kref);
+	RB_CLEAR_NODE(&req->r_node);
+	INIT_LIST_HEAD(&req->r_wait);
+	init_completion(&req->r_completion);
+	init_completion(&req->r_safe_completion);
+	INIT_LIST_HEAD(&req->r_unsafe_item);
+
+	ktime_get_coarse_real_ts64(&req->r_stamp);
+
+	req->r_op = op;
+	req->r_direct_mode = mode;
+	return req;
+}
+
+/*
+ * return oldest (lowest) request, tid in request tree, 0 if none.
+ *
+ * called under mdsc->mutex.
+ */
+static struct ceph_mds_request *__get_oldest_req(struct ceph_mds_client *mdsc)
+{
+	if (RB_EMPTY_ROOT(&mdsc->request_tree))
+		return NULL;
+	return rb_entry(rb_first(&mdsc->request_tree),
+			struct ceph_mds_request, r_node);
+}
+
+static inline  u64 __get_oldest_tid(struct ceph_mds_client *mdsc)
+{
+	return mdsc->oldest_tid;
+}
+
+/*
+ * Build a dentry's path.  Allocate on heap; caller must kfree.  Based
+ * on build_path_from_dentry in fs/cifs/dir.c.
+ *
+ * If @stop_on_nosnap, generate path relative to the first non-snapped
+ * inode.
+ *
+ * Encode hidden .snap dirs as a double /, i.e.
+ *   foo/.snap/bar -> foo//bar
+ */
+char *ceph_mdsc_build_path(struct dentry *dentry, int *plen, u64 *pbase,
+			   int stop_on_nosnap)
+{
+	struct dentry *temp;
+	char *path;
+	int pos;
+	unsigned seq;
+	u64 base;
+
+	if (!dentry)
+		return ERR_PTR(-EINVAL);
+
+	path = __getname();
+	if (!path)
+		return ERR_PTR(-ENOMEM);
+retry:
+	pos = PATH_MAX - 1;
+	path[pos] = '\0';
+
+	seq = read_seqbegin(&rename_lock);
+	rcu_read_lock();
+	temp = dentry;
+	for (;;) {
+		struct inode *inode;
+
+		spin_lock(&temp->d_lock);
+		inode = d_inode(temp);
+		if (inode && ceph_snap(inode) == CEPH_SNAPDIR) {
+			dout("build_path path+%d: %p SNAPDIR\n",
+			     pos, temp);
+		} else if (stop_on_nosnap && inode && dentry != temp &&
+			   ceph_snap(inode) == CEPH_NOSNAP) {
+			spin_unlock(&temp->d_lock);
+			pos++; /* get rid of any prepended '/' */
+			break;
+		} else {
+			pos -= temp->d_name.len;
+			if (pos < 0) {
+				spin_unlock(&temp->d_lock);
+				break;
+			}
+			memcpy(path + pos, temp->d_name.name, temp->d_name.len);
+		}
+		spin_unlock(&temp->d_lock);
+		temp = READ_ONCE(temp->d_parent);
+
+		/* Are we at the root? */
+		if (IS_ROOT(temp))
+			break;
+
+		/* Are we out of buffer? */
+		if (--pos < 0)
+			break;
+
+		path[pos] = '/';
+	}
+	base = ceph_ino(d_inode(temp));
+	rcu_read_unlock();
+
+	if (read_seqretry(&rename_lock, seq))
+		goto retry;
+
+	if (pos < 0) {
+		/*
+		 * A rename didn't occur, but somehow we didn't end up where
+		 * we thought we would. Throw a warning and try again.
+		 */
+		pr_warn("build_path did not end path lookup where "
+			"expected, pos is %d\n", pos);
+		goto retry;
+	}
+
+	*pbase = base;
+	*plen = PATH_MAX - 1 - pos;
+	dout("build_path on %p %d built %llx '%.*s'\n",
+	     dentry, d_count(dentry), base, *plen, path + pos);
+	return path + pos;
+}
+
+static int build_dentry_path(struct dentry *dentry, struct inode *dir,
+			     const char **ppath, int *ppathlen, u64 *pino,
+			     bool *pfreepath, bool parent_locked)
+{
+	char *path;
+
+	rcu_read_lock();
+	if (!dir)
+		dir = d_inode_rcu(dentry->d_parent);
+	if (dir && parent_locked && ceph_snap(dir) == CEPH_NOSNAP) {
+		*pino = ceph_ino(dir);
+		rcu_read_unlock();
+		*ppath = dentry->d_name.name;
+		*ppathlen = dentry->d_name.len;
+		return 0;
+	}
+	rcu_read_unlock();
+	path = ceph_mdsc_build_path(dentry, ppathlen, pino, 1);
+	if (IS_ERR(path))
+		return PTR_ERR(path);
+	*ppath = path;
+	*pfreepath = true;
+	return 0;
+}
+
+static int build_inode_path(struct inode *inode,
+			    const char **ppath, int *ppathlen, u64 *pino,
+			    bool *pfreepath)
+{
+	struct dentry *dentry;
+	char *path;
+
+	if (ceph_snap(inode) == CEPH_NOSNAP) {
+		*pino = ceph_ino(inode);
+		*ppathlen = 0;
+		return 0;
+	}
+	dentry = d_find_alias(inode);
+	path = ceph_mdsc_build_path(dentry, ppathlen, pino, 1);
+	dput(dentry);
+	if (IS_ERR(path))
+		return PTR_ERR(path);
+	*ppath = path;
+	*pfreepath = true;
+	return 0;
+}
+
+/*
+ * request arguments may be specified via an inode *, a dentry *, or
+ * an explicit ino+path.
+ */
+static int set_request_path_attr(struct inode *rinode, struct dentry *rdentry,
+				  struct inode *rdiri, const char *rpath,
+				  u64 rino, const char **ppath, int *pathlen,
+				  u64 *ino, bool *freepath, bool parent_locked)
+{
+	int r = 0;
+
+	if (rinode) {
+		r = build_inode_path(rinode, ppath, pathlen, ino, freepath);
+		dout(" inode %p %llx.%llx\n", rinode, ceph_ino(rinode),
+		     ceph_snap(rinode));
+	} else if (rdentry) {
+		r = build_dentry_path(rdentry, rdiri, ppath, pathlen, ino,
+					freepath, parent_locked);
+		dout(" dentry %p %llx/%.*s\n", rdentry, *ino, *pathlen,
+		     *ppath);
+	} else if (rpath || rino) {
+		*ino = rino;
+		*ppath = rpath;
+		*pathlen = rpath ? strlen(rpath) : 0;
+		dout(" path %.*s\n", *pathlen, rpath);
+	}
+
+	return r;
+}
+
+static void encode_timestamp_and_gids(void **p,
+				      const struct ceph_mds_request *req)
+{
+	struct ceph_timespec ts;
+	int i;
+
+	ceph_encode_timespec64(&ts, &req->r_stamp);
+	ceph_encode_copy(p, &ts, sizeof(ts));
+
+	/* gid_list */
+	ceph_encode_32(p, req->r_cred->group_info->ngroups);
+	for (i = 0; i < req->r_cred->group_info->ngroups; i++)
+		ceph_encode_64(p, from_kgid(&init_user_ns,
+					    req->r_cred->group_info->gid[i]));
+}
+
+/*
+ * called under mdsc->mutex
+ */
+static struct ceph_msg *create_request_message(struct ceph_mds_session *session,
+					       struct ceph_mds_request *req,
+					       bool drop_cap_releases)
+{
+	int mds = session->s_mds;
+	struct ceph_mds_client *mdsc = session->s_mdsc;
+	struct ceph_msg *msg;
+	struct ceph_mds_request_head_old *head;
+	const char *path1 = NULL;
+	const char *path2 = NULL;
+	u64 ino1 = 0, ino2 = 0;
+	int pathlen1 = 0, pathlen2 = 0;
+	bool freepath1 = false, freepath2 = false;
+	int len;
+	u16 releases;
+	void *p, *end;
+	int ret;
+	bool legacy = !(session->s_con.peer_features & CEPH_FEATURE_FS_BTIME);
+
+	ret = set_request_path_attr(req->r_inode, req->r_dentry,
+			      req->r_parent, req->r_path1, req->r_ino1.ino,
+			      &path1, &pathlen1, &ino1, &freepath1,
+			      test_bit(CEPH_MDS_R_PARENT_LOCKED,
+					&req->r_req_flags));
+	if (ret < 0) {
+		msg = ERR_PTR(ret);
+		goto out;
+	}
+
+	/* If r_old_dentry is set, then assume that its parent is locked */
+	ret = set_request_path_attr(NULL, req->r_old_dentry,
+			      req->r_old_dentry_dir,
+			      req->r_path2, req->r_ino2.ino,
+			      &path2, &pathlen2, &ino2, &freepath2, true);
+	if (ret < 0) {
+		msg = ERR_PTR(ret);
+		goto out_free1;
+	}
+
+	len = legacy ? sizeof(*head) : sizeof(struct ceph_mds_request_head);
+	len += pathlen1 + pathlen2 + 2*(1 + sizeof(u32) + sizeof(u64)) +
+		sizeof(struct ceph_timespec);
+	len += sizeof(u32) + (sizeof(u64) * req->r_cred->group_info->ngroups);
+
+	/* calculate (max) length for cap releases */
+	len += sizeof(struct ceph_mds_request_release) *
+		(!!req->r_inode_drop + !!req->r_dentry_drop +
+		 !!req->r_old_inode_drop + !!req->r_old_dentry_drop);
+
+	if (req->r_dentry_drop)
+		len += pathlen1;
+	if (req->r_old_dentry_drop)
+		len += pathlen2;
+
+	msg = ceph_msg_new2(CEPH_MSG_CLIENT_REQUEST, len, 1, GFP_NOFS, false);
+	if (!msg) {
+		msg = ERR_PTR(-ENOMEM);
+		goto out_free2;
+	}
+
+	msg->hdr.tid = cpu_to_le64(req->r_tid);
+
+	/*
+	 * The old ceph_mds_request_head didn't contain a version field, and
+	 * one was added when we moved the message version from 3->4.
+	 */
+	if (legacy) {
+		msg->hdr.version = cpu_to_le16(3);
+		head = msg->front.iov_base;
+		p = msg->front.iov_base + sizeof(*head);
+	} else {
+		struct ceph_mds_request_head *new_head = msg->front.iov_base;
+
+		msg->hdr.version = cpu_to_le16(4);
+		new_head->version = cpu_to_le16(CEPH_MDS_REQUEST_HEAD_VERSION);
+		head = (struct ceph_mds_request_head_old *)&new_head->oldest_client_tid;
+		p = msg->front.iov_base + sizeof(*new_head);
+	}
+
+	end = msg->front.iov_base + msg->front.iov_len;
+
+	head->mdsmap_epoch = cpu_to_le32(mdsc->mdsmap->m_epoch);
+	head->op = cpu_to_le32(req->r_op);
+	head->caller_uid = cpu_to_le32(from_kuid(&init_user_ns,
+						 req->r_cred->fsuid));
+	head->caller_gid = cpu_to_le32(from_kgid(&init_user_ns,
+						 req->r_cred->fsgid));
+	head->ino = cpu_to_le64(req->r_deleg_ino);
+	head->args = req->r_args;
+
+	ceph_encode_filepath(&p, end, ino1, path1);
+	ceph_encode_filepath(&p, end, ino2, path2);
+
+	/* make note of release offset, in case we need to replay */
+	req->r_request_release_offset = p - msg->front.iov_base;
+
+	/* cap releases */
+	releases = 0;
+	if (req->r_inode_drop)
+		releases += ceph_encode_inode_release(&p,
+		      req->r_inode ? req->r_inode : d_inode(req->r_dentry),
+		      mds, req->r_inode_drop, req->r_inode_unless,
+		      req->r_op == CEPH_MDS_OP_READDIR);
+	if (req->r_dentry_drop)
+		releases += ceph_encode_dentry_release(&p, req->r_dentry,
+				req->r_parent, mds, req->r_dentry_drop,
+				req->r_dentry_unless);
+	if (req->r_old_dentry_drop)
+		releases += ceph_encode_dentry_release(&p, req->r_old_dentry,
+				req->r_old_dentry_dir, mds,
+				req->r_old_dentry_drop,
+				req->r_old_dentry_unless);
+	if (req->r_old_inode_drop)
+		releases += ceph_encode_inode_release(&p,
+		      d_inode(req->r_old_dentry),
+		      mds, req->r_old_inode_drop, req->r_old_inode_unless, 0);
+
+	if (drop_cap_releases) {
+		releases = 0;
+		p = msg->front.iov_base + req->r_request_release_offset;
+	}
+
+	head->num_releases = cpu_to_le16(releases);
+
+	encode_timestamp_and_gids(&p, req);
+
+	if (WARN_ON_ONCE(p > end)) {
+		ceph_msg_put(msg);
+		msg = ERR_PTR(-ERANGE);
+		goto out_free2;
+	}
+
+	msg->front.iov_len = p - msg->front.iov_base;
+	msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
+
+	if (req->r_pagelist) {
+		struct ceph_pagelist *pagelist = req->r_pagelist;
+		ceph_msg_data_add_pagelist(msg, pagelist);
+		msg->hdr.data_len = cpu_to_le32(pagelist->length);
+	} else {
+		msg->hdr.data_len = 0;
+	}
+
+	msg->hdr.data_off = cpu_to_le16(0);
+
+out_free2:
+	if (freepath2)
+		ceph_mdsc_free_path((char *)path2, pathlen2);
+out_free1:
+	if (freepath1)
+		ceph_mdsc_free_path((char *)path1, pathlen1);
+out:
+	return msg;
+}
+
+/*
+ * called under mdsc->mutex if error, under no mutex if
+ * success.
+ */
+static void complete_request(struct ceph_mds_client *mdsc,
+			     struct ceph_mds_request *req)
+{
+	req->r_end_latency = ktime_get();
+
+	if (req->r_callback)
+		req->r_callback(mdsc, req);
+	complete_all(&req->r_completion);
+}
+
+static struct ceph_mds_request_head_old *
+find_old_request_head(void *p, u64 features)
+{
+	bool legacy = !(features & CEPH_FEATURE_FS_BTIME);
+	struct ceph_mds_request_head *new_head;
+
+	if (legacy)
+		return (struct ceph_mds_request_head_old *)p;
+	new_head = (struct ceph_mds_request_head *)p;
+	return (struct ceph_mds_request_head_old *)&new_head->oldest_client_tid;
+}
+
+/*
+ * called under mdsc->mutex
+ */
+static int __prepare_send_request(struct ceph_mds_session *session,
+				  struct ceph_mds_request *req,
+				  bool drop_cap_releases)
+{
+	int mds = session->s_mds;
+	struct ceph_mds_client *mdsc = session->s_mdsc;
+	struct ceph_mds_request_head_old *rhead;
+	struct ceph_msg *msg;
+	int flags = 0, max_retry;
+
+	/*
+	 * The type of 'r_attempts' in kernel 'ceph_mds_request'
+	 * is 'int', while in 'ceph_mds_request_head' the type of
+	 * 'num_retry' is '__u8'. So in case the request retries
+	 *  exceeding 256 times, the MDS will receive a incorrect
+	 *  retry seq.
+	 *
+	 * In this case it's ususally a bug in MDS and continue
+	 * retrying the request makes no sense.
+	 *
+	 * In future this could be fixed in ceph code, so avoid
+	 * using the hardcode here.
+	 */
+	max_retry = sizeof_field(struct ceph_mds_request_head, num_retry);
+	max_retry = 1 << (max_retry * BITS_PER_BYTE);
+	if (req->r_attempts >= max_retry) {
+		pr_warn_ratelimited("%s request tid %llu seq overflow\n",
+				    __func__, req->r_tid);
+		return -EMULTIHOP;
+	}
+
+	req->r_attempts++;
+	if (req->r_inode) {
+		struct ceph_cap *cap =
+			ceph_get_cap_for_mds(ceph_inode(req->r_inode), mds);
+
+		if (cap)
+			req->r_sent_on_mseq = cap->mseq;
+		else
+			req->r_sent_on_mseq = -1;
+	}
+	dout("%s %p tid %lld %s (attempt %d)\n", __func__, req,
+	     req->r_tid, ceph_mds_op_name(req->r_op), req->r_attempts);
+
+	if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
+		void *p;
+
+		/*
+		 * Replay.  Do not regenerate message (and rebuild
+		 * paths, etc.); just use the original message.
+		 * Rebuilding paths will break for renames because
+		 * d_move mangles the src name.
+		 */
+		msg = req->r_request;
+		rhead = find_old_request_head(msg->front.iov_base,
+					      session->s_con.peer_features);
+
+		flags = le32_to_cpu(rhead->flags);
+		flags |= CEPH_MDS_FLAG_REPLAY;
+		rhead->flags = cpu_to_le32(flags);
+
+		if (req->r_target_inode)
+			rhead->ino = cpu_to_le64(ceph_ino(req->r_target_inode));
+
+		rhead->num_retry = req->r_attempts - 1;
+
+		/* remove cap/dentry releases from message */
+		rhead->num_releases = 0;
+
+		p = msg->front.iov_base + req->r_request_release_offset;
+		encode_timestamp_and_gids(&p, req);
+
+		msg->front.iov_len = p - msg->front.iov_base;
+		msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
+		return 0;
+	}
+
+	if (req->r_request) {
+		ceph_msg_put(req->r_request);
+		req->r_request = NULL;
+	}
+	msg = create_request_message(session, req, drop_cap_releases);
+	if (IS_ERR(msg)) {
+		req->r_err = PTR_ERR(msg);
+		return PTR_ERR(msg);
+	}
+	req->r_request = msg;
+
+	rhead = find_old_request_head(msg->front.iov_base,
+				      session->s_con.peer_features);
+	rhead->oldest_client_tid = cpu_to_le64(__get_oldest_tid(mdsc));
+	if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags))
+		flags |= CEPH_MDS_FLAG_REPLAY;
+	if (test_bit(CEPH_MDS_R_ASYNC, &req->r_req_flags))
+		flags |= CEPH_MDS_FLAG_ASYNC;
+	if (req->r_parent)
+		flags |= CEPH_MDS_FLAG_WANT_DENTRY;
+	rhead->flags = cpu_to_le32(flags);
+	rhead->num_fwd = req->r_num_fwd;
+	rhead->num_retry = req->r_attempts - 1;
+
+	dout(" r_parent = %p\n", req->r_parent);
+	return 0;
+}
+
+/*
+ * called under mdsc->mutex
+ */
+static int __send_request(struct ceph_mds_session *session,
+			  struct ceph_mds_request *req,
+			  bool drop_cap_releases)
+{
+	int err;
+
+	err = __prepare_send_request(session, req, drop_cap_releases);
+	if (!err) {
+		ceph_msg_get(req->r_request);
+		ceph_con_send(&session->s_con, req->r_request);
+	}
+
+	return err;
+}
+
+/*
+ * send request, or put it on the appropriate wait list.
+ */
+static void __do_request(struct ceph_mds_client *mdsc,
+			struct ceph_mds_request *req)
+{
+	struct ceph_mds_session *session = NULL;
+	int mds = -1;
+	int err = 0;
+	bool random;
+
+	if (req->r_err || test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags)) {
+		if (test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags))
+			__unregister_request(mdsc, req);
+		return;
+	}
+
+	if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_FENCE_IO) {
+		dout("do_request metadata corrupted\n");
+		err = -EIO;
+		goto finish;
+	}
+	if (req->r_timeout &&
+	    time_after_eq(jiffies, req->r_started + req->r_timeout)) {
+		dout("do_request timed out\n");
+		err = -ETIMEDOUT;
+		goto finish;
+	}
+	if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) {
+		dout("do_request forced umount\n");
+		err = -EIO;
+		goto finish;
+	}
+	if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_MOUNTING) {
+		if (mdsc->mdsmap_err) {
+			err = mdsc->mdsmap_err;
+			dout("do_request mdsmap err %d\n", err);
+			goto finish;
+		}
+		if (mdsc->mdsmap->m_epoch == 0) {
+			dout("do_request no mdsmap, waiting for map\n");
+			list_add(&req->r_wait, &mdsc->waiting_for_map);
+			return;
+		}
+		if (!(mdsc->fsc->mount_options->flags &
+		      CEPH_MOUNT_OPT_MOUNTWAIT) &&
+		    !ceph_mdsmap_is_cluster_available(mdsc->mdsmap)) {
+			err = -EHOSTUNREACH;
+			goto finish;
+		}
+	}
+
+	put_request_session(req);
+
+	mds = __choose_mds(mdsc, req, &random);
+	if (mds < 0 ||
+	    ceph_mdsmap_get_state(mdsc->mdsmap, mds) < CEPH_MDS_STATE_ACTIVE) {
+		if (test_bit(CEPH_MDS_R_ASYNC, &req->r_req_flags)) {
+			err = -EJUKEBOX;
+			goto finish;
+		}
+		dout("do_request no mds or not active, waiting for map\n");
+		list_add(&req->r_wait, &mdsc->waiting_for_map);
+		return;
+	}
+
+	/* get, open session */
+	session = __ceph_lookup_mds_session(mdsc, mds);
+	if (!session) {
+		session = register_session(mdsc, mds);
+		if (IS_ERR(session)) {
+			err = PTR_ERR(session);
+			goto finish;
+		}
+	}
+	req->r_session = ceph_get_mds_session(session);
+
+	dout("do_request mds%d session %p state %s\n", mds, session,
+	     ceph_session_state_name(session->s_state));
+
+	/*
+	 * The old ceph will crash the MDSs when see unknown OPs
+	 */
+	if (req->r_feature_needed > 0 &&
+	    !test_bit(req->r_feature_needed, &session->s_features)) {
+		err = -EOPNOTSUPP;
+		goto out_session;
+	}
+
+	if (session->s_state != CEPH_MDS_SESSION_OPEN &&
+	    session->s_state != CEPH_MDS_SESSION_HUNG) {
+		/*
+		 * We cannot queue async requests since the caps and delegated
+		 * inodes are bound to the session. Just return -EJUKEBOX and
+		 * let the caller retry a sync request in that case.
+		 */
+		if (test_bit(CEPH_MDS_R_ASYNC, &req->r_req_flags)) {
+			err = -EJUKEBOX;
+			goto out_session;
+		}
+
+		/*
+		 * If the session has been REJECTED, then return a hard error,
+		 * unless it's a CLEANRECOVER mount, in which case we'll queue
+		 * it to the mdsc queue.
+		 */
+		if (session->s_state == CEPH_MDS_SESSION_REJECTED) {
+			if (ceph_test_mount_opt(mdsc->fsc, CLEANRECOVER))
+				list_add(&req->r_wait, &mdsc->waiting_for_map);
+			else
+				err = -EACCES;
+			goto out_session;
+		}
+
+		if (session->s_state == CEPH_MDS_SESSION_NEW ||
+		    session->s_state == CEPH_MDS_SESSION_CLOSING) {
+			err = __open_session(mdsc, session);
+			if (err)
+				goto out_session;
+			/* retry the same mds later */
+			if (random)
+				req->r_resend_mds = mds;
+		}
+		list_add(&req->r_wait, &session->s_waiting);
+		goto out_session;
+	}
+
+	/* send request */
+	req->r_resend_mds = -1;   /* forget any previous mds hint */
+
+	if (req->r_request_started == 0)   /* note request start time */
+		req->r_request_started = jiffies;
+
+	/*
+	 * For async create we will choose the auth MDS of frag in parent
+	 * directory to send the request and ususally this works fine, but
+	 * if the migrated the dirtory to another MDS before it could handle
+	 * it the request will be forwarded.
+	 *
+	 * And then the auth cap will be changed.
+	 */
+	if (test_bit(CEPH_MDS_R_ASYNC, &req->r_req_flags) && req->r_num_fwd) {
+		struct ceph_dentry_info *di = ceph_dentry(req->r_dentry);
+		struct ceph_inode_info *ci;
+		struct ceph_cap *cap;
+
+		/*
+		 * The request maybe handled very fast and the new inode
+		 * hasn't been linked to the dentry yet. We need to wait
+		 * for the ceph_finish_async_create(), which shouldn't be
+		 * stuck too long or fail in thoery, to finish when forwarding
+		 * the request.
+		 */
+		if (!d_inode(req->r_dentry)) {
+			err = wait_on_bit(&di->flags, CEPH_DENTRY_ASYNC_CREATE_BIT,
+					  TASK_KILLABLE);
+			if (err) {
+				mutex_lock(&req->r_fill_mutex);
+				set_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags);
+				mutex_unlock(&req->r_fill_mutex);
+				goto out_session;
+			}
+		}
+
+		ci = ceph_inode(d_inode(req->r_dentry));
+
+		spin_lock(&ci->i_ceph_lock);
+		cap = ci->i_auth_cap;
+		if (ci->i_ceph_flags & CEPH_I_ASYNC_CREATE && mds != cap->mds) {
+			dout("do_request session changed for auth cap %d -> %d\n",
+			     cap->session->s_mds, session->s_mds);
+
+			/* Remove the auth cap from old session */
+			spin_lock(&cap->session->s_cap_lock);
+			cap->session->s_nr_caps--;
+			list_del_init(&cap->session_caps);
+			spin_unlock(&cap->session->s_cap_lock);
+
+			/* Add the auth cap to the new session */
+			cap->mds = mds;
+			cap->session = session;
+			spin_lock(&session->s_cap_lock);
+			session->s_nr_caps++;
+			list_add_tail(&cap->session_caps, &session->s_caps);
+			spin_unlock(&session->s_cap_lock);
+
+			change_auth_cap_ses(ci, session);
+		}
+		spin_unlock(&ci->i_ceph_lock);
+	}
+
+	err = __send_request(session, req, false);
+
+out_session:
+	ceph_put_mds_session(session);
+finish:
+	if (err) {
+		dout("__do_request early error %d\n", err);
+		req->r_err = err;
+		complete_request(mdsc, req);
+		__unregister_request(mdsc, req);
+	}
+	return;
+}
+
+/*
+ * called under mdsc->mutex
+ */
+static void __wake_requests(struct ceph_mds_client *mdsc,
+			    struct list_head *head)
+{
+	struct ceph_mds_request *req;
+	LIST_HEAD(tmp_list);
+
+	list_splice_init(head, &tmp_list);
+
+	while (!list_empty(&tmp_list)) {
+		req = list_entry(tmp_list.next,
+				 struct ceph_mds_request, r_wait);
+		list_del_init(&req->r_wait);
+		dout(" wake request %p tid %llu\n", req, req->r_tid);
+		__do_request(mdsc, req);
+	}
+}
+
+/*
+ * Wake up threads with requests pending for @mds, so that they can
+ * resubmit their requests to a possibly different mds.
+ */
+static void kick_requests(struct ceph_mds_client *mdsc, int mds)
+{
+	struct ceph_mds_request *req;
+	struct rb_node *p = rb_first(&mdsc->request_tree);
+
+	dout("kick_requests mds%d\n", mds);
+	while (p) {
+		req = rb_entry(p, struct ceph_mds_request, r_node);
+		p = rb_next(p);
+		if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags))
+			continue;
+		if (req->r_attempts > 0)
+			continue; /* only new requests */
+		if (req->r_session &&
+		    req->r_session->s_mds == mds) {
+			dout(" kicking tid %llu\n", req->r_tid);
+			list_del_init(&req->r_wait);
+			__do_request(mdsc, req);
+		}
+	}
+}
+
+int ceph_mdsc_submit_request(struct ceph_mds_client *mdsc, struct inode *dir,
+			      struct ceph_mds_request *req)
+{
+	int err = 0;
+
+	/* take CAP_PIN refs for r_inode, r_parent, r_old_dentry */
+	if (req->r_inode)
+		ceph_get_cap_refs(ceph_inode(req->r_inode), CEPH_CAP_PIN);
+	if (req->r_parent) {
+		struct ceph_inode_info *ci = ceph_inode(req->r_parent);
+		int fmode = (req->r_op & CEPH_MDS_OP_WRITE) ?
+			    CEPH_FILE_MODE_WR : CEPH_FILE_MODE_RD;
+		spin_lock(&ci->i_ceph_lock);
+		ceph_take_cap_refs(ci, CEPH_CAP_PIN, false);
+		__ceph_touch_fmode(ci, mdsc, fmode);
+		spin_unlock(&ci->i_ceph_lock);
+	}
+	if (req->r_old_dentry_dir)
+		ceph_get_cap_refs(ceph_inode(req->r_old_dentry_dir),
+				  CEPH_CAP_PIN);
+
+	if (req->r_inode) {
+		err = ceph_wait_on_async_create(req->r_inode);
+		if (err) {
+			dout("%s: wait for async create returned: %d\n",
+			     __func__, err);
+			return err;
+		}
+	}
+
+	if (!err && req->r_old_inode) {
+		err = ceph_wait_on_async_create(req->r_old_inode);
+		if (err) {
+			dout("%s: wait for async create returned: %d\n",
+			     __func__, err);
+			return err;
+		}
+	}
+
+	dout("submit_request on %p for inode %p\n", req, dir);
+	mutex_lock(&mdsc->mutex);
+	__register_request(mdsc, req, dir);
+	__do_request(mdsc, req);
+	err = req->r_err;
+	mutex_unlock(&mdsc->mutex);
+	return err;
+}
+
+int ceph_mdsc_wait_request(struct ceph_mds_client *mdsc,
+			   struct ceph_mds_request *req,
+			   ceph_mds_request_wait_callback_t wait_func)
+{
+	int err;
+
+	/* wait */
+	dout("do_request waiting\n");
+	if (wait_func) {
+		err = wait_func(mdsc, req);
+	} else {
+		long timeleft = wait_for_completion_killable_timeout(
+					&req->r_completion,
+					ceph_timeout_jiffies(req->r_timeout));
+		if (timeleft > 0)
+			err = 0;
+		else if (!timeleft)
+			err = -ETIMEDOUT;  /* timed out */
+		else
+			err = timeleft;  /* killed */
+	}
+	dout("do_request waited, got %d\n", err);
+	mutex_lock(&mdsc->mutex);
+
+	/* only abort if we didn't race with a real reply */
+	if (test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags)) {
+		err = le32_to_cpu(req->r_reply_info.head->result);
+	} else if (err < 0) {
+		dout("aborted request %lld with %d\n", req->r_tid, err);
+
+		/*
+		 * ensure we aren't running concurrently with
+		 * ceph_fill_trace or ceph_readdir_prepopulate, which
+		 * rely on locks (dir mutex) held by our caller.
+		 */
+		mutex_lock(&req->r_fill_mutex);
+		req->r_err = err;
+		set_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags);
+		mutex_unlock(&req->r_fill_mutex);
+
+		if (req->r_parent &&
+		    (req->r_op & CEPH_MDS_OP_WRITE))
+			ceph_invalidate_dir_request(req);
+	} else {
+		err = req->r_err;
+	}
+
+	mutex_unlock(&mdsc->mutex);
+	return err;
+}
+
+/*
+ * Synchrously perform an mds request.  Take care of all of the
+ * session setup, forwarding, retry details.
+ */
+int ceph_mdsc_do_request(struct ceph_mds_client *mdsc,
+			 struct inode *dir,
+			 struct ceph_mds_request *req)
+{
+	int err;
+
+	dout("do_request on %p\n", req);
+
+	/* issue */
+	err = ceph_mdsc_submit_request(mdsc, dir, req);
+	if (!err)
+		err = ceph_mdsc_wait_request(mdsc, req, NULL);
+	dout("do_request %p done, result %d\n", req, err);
+	return err;
+}
+
+/*
+ * Invalidate dir's completeness, dentry lease state on an aborted MDS
+ * namespace request.
+ */
+void ceph_invalidate_dir_request(struct ceph_mds_request *req)
+{
+	struct inode *dir = req->r_parent;
+	struct inode *old_dir = req->r_old_dentry_dir;
+
+	dout("invalidate_dir_request %p %p (complete, lease(s))\n", dir, old_dir);
+
+	ceph_dir_clear_complete(dir);
+	if (old_dir)
+		ceph_dir_clear_complete(old_dir);
+	if (req->r_dentry)
+		ceph_invalidate_dentry_lease(req->r_dentry);
+	if (req->r_old_dentry)
+		ceph_invalidate_dentry_lease(req->r_old_dentry);
+}
+
+/*
+ * Handle mds reply.
+ *
+ * We take the session mutex and parse and process the reply immediately.
+ * This preserves the logical ordering of replies, capabilities, etc., sent
+ * by the MDS as they are applied to our local cache.
+ */
+static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg)
+{
+	struct ceph_mds_client *mdsc = session->s_mdsc;
+	struct ceph_mds_request *req;
+	struct ceph_mds_reply_head *head = msg->front.iov_base;
+	struct ceph_mds_reply_info_parsed *rinfo;  /* parsed reply info */
+	struct ceph_snap_realm *realm;
+	u64 tid;
+	int err, result;
+	int mds = session->s_mds;
+	bool close_sessions = false;
+
+	if (msg->front.iov_len < sizeof(*head)) {
+		pr_err("mdsc_handle_reply got corrupt (short) reply\n");
+		ceph_msg_dump(msg);
+		return;
+	}
+
+	/* get request, session */
+	tid = le64_to_cpu(msg->hdr.tid);
+	mutex_lock(&mdsc->mutex);
+	req = lookup_get_request(mdsc, tid);
+	if (!req) {
+		dout("handle_reply on unknown tid %llu\n", tid);
+		mutex_unlock(&mdsc->mutex);
+		return;
+	}
+	dout("handle_reply %p\n", req);
+
+	/* correct session? */
+	if (req->r_session != session) {
+		pr_err("mdsc_handle_reply got %llu on session mds%d"
+		       " not mds%d\n", tid, session->s_mds,
+		       req->r_session ? req->r_session->s_mds : -1);
+		mutex_unlock(&mdsc->mutex);
+		goto out;
+	}
+
+	/* dup? */
+	if ((test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags) && !head->safe) ||
+	    (test_bit(CEPH_MDS_R_GOT_SAFE, &req->r_req_flags) && head->safe)) {
+		pr_warn("got a dup %s reply on %llu from mds%d\n",
+			   head->safe ? "safe" : "unsafe", tid, mds);
+		mutex_unlock(&mdsc->mutex);
+		goto out;
+	}
+	if (test_bit(CEPH_MDS_R_GOT_SAFE, &req->r_req_flags)) {
+		pr_warn("got unsafe after safe on %llu from mds%d\n",
+			   tid, mds);
+		mutex_unlock(&mdsc->mutex);
+		goto out;
+	}
+
+	result = le32_to_cpu(head->result);
+
+	if (head->safe) {
+		set_bit(CEPH_MDS_R_GOT_SAFE, &req->r_req_flags);
+		__unregister_request(mdsc, req);
+
+		/* last request during umount? */
+		if (mdsc->stopping && !__get_oldest_req(mdsc))
+			complete_all(&mdsc->safe_umount_waiters);
+
+		if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
+			/*
+			 * We already handled the unsafe response, now do the
+			 * cleanup.  No need to examine the response; the MDS
+			 * doesn't include any result info in the safe
+			 * response.  And even if it did, there is nothing
+			 * useful we could do with a revised return value.
+			 */
+			dout("got safe reply %llu, mds%d\n", tid, mds);
+
+			mutex_unlock(&mdsc->mutex);
+			goto out;
+		}
+	} else {
+		set_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags);
+		list_add_tail(&req->r_unsafe_item, &req->r_session->s_unsafe);
+	}
+
+	dout("handle_reply tid %lld result %d\n", tid, result);
+	rinfo = &req->r_reply_info;
+	if (test_bit(CEPHFS_FEATURE_REPLY_ENCODING, &session->s_features))
+		err = parse_reply_info(session, msg, rinfo, (u64)-1);
+	else
+		err = parse_reply_info(session, msg, rinfo, session->s_con.peer_features);
+	mutex_unlock(&mdsc->mutex);
+
+	/* Must find target inode outside of mutexes to avoid deadlocks */
+	if ((err >= 0) && rinfo->head->is_target) {
+		struct inode *in;
+		struct ceph_vino tvino = {
+			.ino  = le64_to_cpu(rinfo->targeti.in->ino),
+			.snap = le64_to_cpu(rinfo->targeti.in->snapid)
+		};
+
+		in = ceph_get_inode(mdsc->fsc->sb, tvino);
+		if (IS_ERR(in)) {
+			err = PTR_ERR(in);
+			mutex_lock(&session->s_mutex);
+			goto out_err;
+		}
+		req->r_target_inode = in;
+	}
+
+	mutex_lock(&session->s_mutex);
+	if (err < 0) {
+		pr_err("mdsc_handle_reply got corrupt reply mds%d(tid:%lld)\n", mds, tid);
+		ceph_msg_dump(msg);
+		goto out_err;
+	}
+
+	/* snap trace */
+	realm = NULL;
+	if (rinfo->snapblob_len) {
+		down_write(&mdsc->snap_rwsem);
+		err = ceph_update_snap_trace(mdsc, rinfo->snapblob,
+				rinfo->snapblob + rinfo->snapblob_len,
+				le32_to_cpu(head->op) == CEPH_MDS_OP_RMSNAP,
+				&realm);
+		if (err) {
+			up_write(&mdsc->snap_rwsem);
+			close_sessions = true;
+			if (err == -EIO)
+				ceph_msg_dump(msg);
+			goto out_err;
+		}
+		downgrade_write(&mdsc->snap_rwsem);
+	} else {
+		down_read(&mdsc->snap_rwsem);
+	}
+
+	/* insert trace into our cache */
+	mutex_lock(&req->r_fill_mutex);
+	current->journal_info = req;
+	err = ceph_fill_trace(mdsc->fsc->sb, req);
+	if (err == 0) {
+		if (result == 0 && (req->r_op == CEPH_MDS_OP_READDIR ||
+				    req->r_op == CEPH_MDS_OP_LSSNAP))
+			ceph_readdir_prepopulate(req, req->r_session);
+	}
+	current->journal_info = NULL;
+	mutex_unlock(&req->r_fill_mutex);
+
+	up_read(&mdsc->snap_rwsem);
+	if (realm)
+		ceph_put_snap_realm(mdsc, realm);
+
+	if (err == 0) {
+		if (req->r_target_inode &&
+		    test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
+			struct ceph_inode_info *ci =
+				ceph_inode(req->r_target_inode);
+			spin_lock(&ci->i_unsafe_lock);
+			list_add_tail(&req->r_unsafe_target_item,
+				      &ci->i_unsafe_iops);
+			spin_unlock(&ci->i_unsafe_lock);
+		}
+
+		ceph_unreserve_caps(mdsc, &req->r_caps_reservation);
+	}
+out_err:
+	mutex_lock(&mdsc->mutex);
+	if (!test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags)) {
+		if (err) {
+			req->r_err = err;
+		} else {
+			req->r_reply =  ceph_msg_get(msg);
+			set_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags);
+		}
+	} else {
+		dout("reply arrived after request %lld was aborted\n", tid);
+	}
+	mutex_unlock(&mdsc->mutex);
+
+	mutex_unlock(&session->s_mutex);
+
+	/* kick calling process */
+	complete_request(mdsc, req);
+
+	ceph_update_metadata_metrics(&mdsc->metric, req->r_start_latency,
+				     req->r_end_latency, err);
+out:
+	ceph_mdsc_put_request(req);
+
+	/* Defer closing the sessions after s_mutex lock being released */
+	if (close_sessions)
+		ceph_mdsc_close_sessions(mdsc);
+	return;
+}
+
+
+
+/*
+ * handle mds notification that our request has been forwarded.
+ */
+static void handle_forward(struct ceph_mds_client *mdsc,
+			   struct ceph_mds_session *session,
+			   struct ceph_msg *msg)
+{
+	struct ceph_mds_request *req;
+	u64 tid = le64_to_cpu(msg->hdr.tid);
+	u32 next_mds;
+	u32 fwd_seq;
+	int err = -EINVAL;
+	void *p = msg->front.iov_base;
+	void *end = p + msg->front.iov_len;
+	bool aborted = false;
+
+	ceph_decode_need(&p, end, 2*sizeof(u32), bad);
+	next_mds = ceph_decode_32(&p);
+	fwd_seq = ceph_decode_32(&p);
+
+	mutex_lock(&mdsc->mutex);
+	req = lookup_get_request(mdsc, tid);
+	if (!req) {
+		mutex_unlock(&mdsc->mutex);
+		dout("forward tid %llu to mds%d - req dne\n", tid, next_mds);
+		return;  /* dup reply? */
+	}
+
+	if (test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags)) {
+		dout("forward tid %llu aborted, unregistering\n", tid);
+		__unregister_request(mdsc, req);
+	} else if (fwd_seq <= req->r_num_fwd) {
+		/*
+		 * The type of 'num_fwd' in ceph 'MClientRequestForward'
+		 * is 'int32_t', while in 'ceph_mds_request_head' the
+		 * type is '__u8'. So in case the request bounces between
+		 * MDSes exceeding 256 times, the client will get stuck.
+		 *
+		 * In this case it's ususally a bug in MDS and continue
+		 * bouncing the request makes no sense.
+		 *
+		 * In future this could be fixed in ceph code, so avoid
+		 * using the hardcode here.
+		 */
+		int max = sizeof_field(struct ceph_mds_request_head, num_fwd);
+		max = 1 << (max * BITS_PER_BYTE);
+		if (req->r_num_fwd >= max) {
+			mutex_lock(&req->r_fill_mutex);
+			req->r_err = -EMULTIHOP;
+			set_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags);
+			mutex_unlock(&req->r_fill_mutex);
+			aborted = true;
+			pr_warn_ratelimited("forward tid %llu seq overflow\n",
+					    tid);
+		} else {
+			dout("forward tid %llu to mds%d - old seq %d <= %d\n",
+			     tid, next_mds, req->r_num_fwd, fwd_seq);
+		}
+	} else {
+		/* resend. forward race not possible; mds would drop */
+		dout("forward tid %llu to mds%d (we resend)\n", tid, next_mds);
+		BUG_ON(req->r_err);
+		BUG_ON(test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags));
+		req->r_attempts = 0;
+		req->r_num_fwd = fwd_seq;
+		req->r_resend_mds = next_mds;
+		put_request_session(req);
+		__do_request(mdsc, req);
+	}
+	mutex_unlock(&mdsc->mutex);
+
+	/* kick calling process */
+	if (aborted)
+		complete_request(mdsc, req);
+	ceph_mdsc_put_request(req);
+	return;
+
+bad:
+	pr_err("mdsc_handle_forward decode error err=%d\n", err);
+	ceph_msg_dump(msg);
+}
+
+static int __decode_session_metadata(void **p, void *end,
+				     bool *blocklisted)
+{
+	/* map<string,string> */
+	u32 n;
+	bool err_str;
+	ceph_decode_32_safe(p, end, n, bad);
+	while (n-- > 0) {
+		u32 len;
+		ceph_decode_32_safe(p, end, len, bad);
+		ceph_decode_need(p, end, len, bad);
+		err_str = !strncmp(*p, "error_string", len);
+		*p += len;
+		ceph_decode_32_safe(p, end, len, bad);
+		ceph_decode_need(p, end, len, bad);
+		/*
+		 * Match "blocklisted (blacklisted)" from newer MDSes,
+		 * or "blacklisted" from older MDSes.
+		 */
+		if (err_str && strnstr(*p, "blacklisted", len))
+			*blocklisted = true;
+		*p += len;
+	}
+	return 0;
+bad:
+	return -1;
+}
+
+/*
+ * handle a mds session control message
+ */
+static void handle_session(struct ceph_mds_session *session,
+			   struct ceph_msg *msg)
+{
+	struct ceph_mds_client *mdsc = session->s_mdsc;
+	int mds = session->s_mds;
+	int msg_version = le16_to_cpu(msg->hdr.version);
+	void *p = msg->front.iov_base;
+	void *end = p + msg->front.iov_len;
+	struct ceph_mds_session_head *h;
+	u32 op;
+	u64 seq, features = 0;
+	int wake = 0;
+	bool blocklisted = false;
+
+	/* decode */
+	ceph_decode_need(&p, end, sizeof(*h), bad);
+	h = p;
+	p += sizeof(*h);
+
+	op = le32_to_cpu(h->op);
+	seq = le64_to_cpu(h->seq);
+
+	if (msg_version >= 3) {
+		u32 len;
+		/* version >= 2 and < 5, decode metadata, skip otherwise
+		 * as it's handled via flags.
+		 */
+		if (msg_version >= 5)
+			ceph_decode_skip_map(&p, end, string, string, bad);
+		else if (__decode_session_metadata(&p, end, &blocklisted) < 0)
+			goto bad;
+
+		/* version >= 3, feature bits */
+		ceph_decode_32_safe(&p, end, len, bad);
+		if (len) {
+			ceph_decode_64_safe(&p, end, features, bad);
+			p += len - sizeof(features);
+		}
+	}
+
+	if (msg_version >= 5) {
+		u32 flags, len;
+
+		/* version >= 4 */
+		ceph_decode_skip_16(&p, end, bad); /* struct_v, struct_cv */
+		ceph_decode_32_safe(&p, end, len, bad); /* len */
+		ceph_decode_skip_n(&p, end, len, bad); /* metric_spec */
+
+		/* version >= 5, flags   */
+		ceph_decode_32_safe(&p, end, flags, bad);
+		if (flags & CEPH_SESSION_BLOCKLISTED) {
+			pr_warn("mds%d session blocklisted\n", session->s_mds);
+			blocklisted = true;
+		}
+	}
+
+	mutex_lock(&mdsc->mutex);
+	if (op == CEPH_SESSION_CLOSE) {
+		ceph_get_mds_session(session);
+		__unregister_session(mdsc, session);
+	}
+	/* FIXME: this ttl calculation is generous */
+	session->s_ttl = jiffies + HZ*mdsc->mdsmap->m_session_autoclose;
+	mutex_unlock(&mdsc->mutex);
+
+	mutex_lock(&session->s_mutex);
+
+	dout("handle_session mds%d %s %p state %s seq %llu\n",
+	     mds, ceph_session_op_name(op), session,
+	     ceph_session_state_name(session->s_state), seq);
+
+	if (session->s_state == CEPH_MDS_SESSION_HUNG) {
+		session->s_state = CEPH_MDS_SESSION_OPEN;
+		pr_info("mds%d came back\n", session->s_mds);
+	}
+
+	switch (op) {
+	case CEPH_SESSION_OPEN:
+		if (session->s_state == CEPH_MDS_SESSION_RECONNECTING)
+			pr_info("mds%d reconnect success\n", session->s_mds);
+
+		if (session->s_state == CEPH_MDS_SESSION_OPEN) {
+			pr_notice("mds%d is already opened\n", session->s_mds);
+		} else {
+			session->s_state = CEPH_MDS_SESSION_OPEN;
+			session->s_features = features;
+			renewed_caps(mdsc, session, 0);
+			if (test_bit(CEPHFS_FEATURE_METRIC_COLLECT,
+				     &session->s_features))
+				metric_schedule_delayed(&mdsc->metric);
+		}
+
+		/*
+		 * The connection maybe broken and the session in client
+		 * side has been reinitialized, need to update the seq
+		 * anyway.
+		 */
+		if (!session->s_seq && seq)
+			session->s_seq = seq;
+
+		wake = 1;
+		if (mdsc->stopping)
+			__close_session(mdsc, session);
+		break;
+
+	case CEPH_SESSION_RENEWCAPS:
+		if (session->s_renew_seq == seq)
+			renewed_caps(mdsc, session, 1);
+		break;
+
+	case CEPH_SESSION_CLOSE:
+		if (session->s_state == CEPH_MDS_SESSION_RECONNECTING)
+			pr_info("mds%d reconnect denied\n", session->s_mds);
+		session->s_state = CEPH_MDS_SESSION_CLOSED;
+		cleanup_session_requests(mdsc, session);
+		remove_session_caps(session);
+		wake = 2; /* for good measure */
+		wake_up_all(&mdsc->session_close_wq);
+		break;
+
+	case CEPH_SESSION_STALE:
+		pr_info("mds%d caps went stale, renewing\n",
+			session->s_mds);
+		atomic_inc(&session->s_cap_gen);
+		session->s_cap_ttl = jiffies - 1;
+		send_renew_caps(mdsc, session);
+		break;
+
+	case CEPH_SESSION_RECALL_STATE:
+		ceph_trim_caps(mdsc, session, le32_to_cpu(h->max_caps));
+		break;
+
+	case CEPH_SESSION_FLUSHMSG:
+		/* flush cap releases */
+		spin_lock(&session->s_cap_lock);
+		if (session->s_num_cap_releases)
+			ceph_flush_cap_releases(mdsc, session);
+		spin_unlock(&session->s_cap_lock);
+
+		send_flushmsg_ack(mdsc, session, seq);
+		break;
+
+	case CEPH_SESSION_FORCE_RO:
+		dout("force_session_readonly %p\n", session);
+		spin_lock(&session->s_cap_lock);
+		session->s_readonly = true;
+		spin_unlock(&session->s_cap_lock);
+		wake_up_session_caps(session, FORCE_RO);
+		break;
+
+	case CEPH_SESSION_REJECT:
+		WARN_ON(session->s_state != CEPH_MDS_SESSION_OPENING);
+		pr_info("mds%d rejected session\n", session->s_mds);
+		session->s_state = CEPH_MDS_SESSION_REJECTED;
+		cleanup_session_requests(mdsc, session);
+		remove_session_caps(session);
+		if (blocklisted)
+			mdsc->fsc->blocklisted = true;
+		wake = 2; /* for good measure */
+		break;
+
+	default:
+		pr_err("mdsc_handle_session bad op %d mds%d\n", op, mds);
+		WARN_ON(1);
+	}
+
+	mutex_unlock(&session->s_mutex);
+	if (wake) {
+		mutex_lock(&mdsc->mutex);
+		__wake_requests(mdsc, &session->s_waiting);
+		if (wake == 2)
+			kick_requests(mdsc, mds);
+		mutex_unlock(&mdsc->mutex);
+	}
+	if (op == CEPH_SESSION_CLOSE)
+		ceph_put_mds_session(session);
+	return;
+
+bad:
+	pr_err("mdsc_handle_session corrupt message mds%d len %d\n", mds,
+	       (int)msg->front.iov_len);
+	ceph_msg_dump(msg);
+	return;
+}
+
+void ceph_mdsc_release_dir_caps(struct ceph_mds_request *req)
+{
+	int dcaps;
+
+	dcaps = xchg(&req->r_dir_caps, 0);
+	if (dcaps) {
+		dout("releasing r_dir_caps=%s\n", ceph_cap_string(dcaps));
+		ceph_put_cap_refs(ceph_inode(req->r_parent), dcaps);
+	}
+}
+
+void ceph_mdsc_release_dir_caps_no_check(struct ceph_mds_request *req)
+{
+	int dcaps;
+
+	dcaps = xchg(&req->r_dir_caps, 0);
+	if (dcaps) {
+		dout("releasing r_dir_caps=%s\n", ceph_cap_string(dcaps));
+		ceph_put_cap_refs_no_check_caps(ceph_inode(req->r_parent),
+						dcaps);
+	}
+}
+
+/*
+ * called under session->mutex.
+ */
+static void replay_unsafe_requests(struct ceph_mds_client *mdsc,
+				   struct ceph_mds_session *session)
+{
+	struct ceph_mds_request *req, *nreq;
+	struct rb_node *p;
+
+	dout("replay_unsafe_requests mds%d\n", session->s_mds);
+
+	mutex_lock(&mdsc->mutex);
+	list_for_each_entry_safe(req, nreq, &session->s_unsafe, r_unsafe_item)
+		__send_request(session, req, true);
+
+	/*
+	 * also re-send old requests when MDS enters reconnect stage. So that MDS
+	 * can process completed request in clientreplay stage.
+	 */
+	p = rb_first(&mdsc->request_tree);
+	while (p) {
+		req = rb_entry(p, struct ceph_mds_request, r_node);
+		p = rb_next(p);
+		if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags))
+			continue;
+		if (req->r_attempts == 0)
+			continue; /* only old requests */
+		if (!req->r_session)
+			continue;
+		if (req->r_session->s_mds != session->s_mds)
+			continue;
+
+		ceph_mdsc_release_dir_caps_no_check(req);
+
+		__send_request(session, req, true);
+	}
+	mutex_unlock(&mdsc->mutex);
+}
+
+static int send_reconnect_partial(struct ceph_reconnect_state *recon_state)
+{
+	struct ceph_msg *reply;
+	struct ceph_pagelist *_pagelist;
+	struct page *page;
+	__le32 *addr;
+	int err = -ENOMEM;
+
+	if (!recon_state->allow_multi)
+		return -ENOSPC;
+
+	/* can't handle message that contains both caps and realm */
+	BUG_ON(!recon_state->nr_caps == !recon_state->nr_realms);
+
+	/* pre-allocate new pagelist */
+	_pagelist = ceph_pagelist_alloc(GFP_NOFS);
+	if (!_pagelist)
+		return -ENOMEM;
+
+	reply = ceph_msg_new2(CEPH_MSG_CLIENT_RECONNECT, 0, 1, GFP_NOFS, false);
+	if (!reply)
+		goto fail_msg;
+
+	/* placeholder for nr_caps */
+	err = ceph_pagelist_encode_32(_pagelist, 0);
+	if (err < 0)
+		goto fail;
+
+	if (recon_state->nr_caps) {
+		/* currently encoding caps */
+		err = ceph_pagelist_encode_32(recon_state->pagelist, 0);
+		if (err)
+			goto fail;
+	} else {
+		/* placeholder for nr_realms (currently encoding relams) */
+		err = ceph_pagelist_encode_32(_pagelist, 0);
+		if (err < 0)
+			goto fail;
+	}
+
+	err = ceph_pagelist_encode_8(recon_state->pagelist, 1);
+	if (err)
+		goto fail;
+
+	page = list_first_entry(&recon_state->pagelist->head, struct page, lru);
+	addr = kmap_atomic(page);
+	if (recon_state->nr_caps) {
+		/* currently encoding caps */
+		*addr = cpu_to_le32(recon_state->nr_caps);
+	} else {
+		/* currently encoding relams */
+		*(addr + 1) = cpu_to_le32(recon_state->nr_realms);
+	}
+	kunmap_atomic(addr);
+
+	reply->hdr.version = cpu_to_le16(5);
+	reply->hdr.compat_version = cpu_to_le16(4);
+
+	reply->hdr.data_len = cpu_to_le32(recon_state->pagelist->length);
+	ceph_msg_data_add_pagelist(reply, recon_state->pagelist);
+
+	ceph_con_send(&recon_state->session->s_con, reply);
+	ceph_pagelist_release(recon_state->pagelist);
+
+	recon_state->pagelist = _pagelist;
+	recon_state->nr_caps = 0;
+	recon_state->nr_realms = 0;
+	recon_state->msg_version = 5;
+	return 0;
+fail:
+	ceph_msg_put(reply);
+fail_msg:
+	ceph_pagelist_release(_pagelist);
+	return err;
+}
+
+static struct dentry* d_find_primary(struct inode *inode)
+{
+	struct dentry *alias, *dn = NULL;
+
+	if (hlist_empty(&inode->i_dentry))
+		return NULL;
+
+	spin_lock(&inode->i_lock);
+	if (hlist_empty(&inode->i_dentry))
+		goto out_unlock;
+
+	if (S_ISDIR(inode->i_mode)) {
+		alias = hlist_entry(inode->i_dentry.first, struct dentry, d_u.d_alias);
+		if (!IS_ROOT(alias))
+			dn = dget(alias);
+		goto out_unlock;
+	}
+
+	hlist_for_each_entry(alias, &inode->i_dentry, d_u.d_alias) {
+		spin_lock(&alias->d_lock);
+		if (!d_unhashed(alias) &&
+		    (ceph_dentry(alias)->flags & CEPH_DENTRY_PRIMARY_LINK)) {
+			dn = dget_dlock(alias);
+		}
+		spin_unlock(&alias->d_lock);
+		if (dn)
+			break;
+	}
+out_unlock:
+	spin_unlock(&inode->i_lock);
+	return dn;
+}
+
+/*
+ * Encode information about a cap for a reconnect with the MDS.
+ */
+static int reconnect_caps_cb(struct inode *inode, int mds, void *arg)
+{
+	union {
+		struct ceph_mds_cap_reconnect v2;
+		struct ceph_mds_cap_reconnect_v1 v1;
+	} rec;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_reconnect_state *recon_state = arg;
+	struct ceph_pagelist *pagelist = recon_state->pagelist;
+	struct dentry *dentry;
+	struct ceph_cap *cap;
+	char *path;
+	int pathlen = 0, err;
+	u64 pathbase;
+	u64 snap_follows;
+
+	dentry = d_find_primary(inode);
+	if (dentry) {
+		/* set pathbase to parent dir when msg_version >= 2 */
+		path = ceph_mdsc_build_path(dentry, &pathlen, &pathbase,
+					    recon_state->msg_version >= 2);
+		dput(dentry);
+		if (IS_ERR(path)) {
+			err = PTR_ERR(path);
+			goto out_err;
+		}
+	} else {
+		path = NULL;
+		pathbase = 0;
+	}
+
+	spin_lock(&ci->i_ceph_lock);
+	cap = __get_cap_for_mds(ci, mds);
+	if (!cap) {
+		spin_unlock(&ci->i_ceph_lock);
+		err = 0;
+		goto out_err;
+	}
+	dout(" adding %p ino %llx.%llx cap %p %lld %s\n",
+	     inode, ceph_vinop(inode), cap, cap->cap_id,
+	     ceph_cap_string(cap->issued));
+
+	cap->seq = 0;        /* reset cap seq */
+	cap->issue_seq = 0;  /* and issue_seq */
+	cap->mseq = 0;       /* and migrate_seq */
+	cap->cap_gen = atomic_read(&cap->session->s_cap_gen);
+
+	/* These are lost when the session goes away */
+	if (S_ISDIR(inode->i_mode)) {
+		if (cap->issued & CEPH_CAP_DIR_CREATE) {
+			ceph_put_string(rcu_dereference_raw(ci->i_cached_layout.pool_ns));
+			memset(&ci->i_cached_layout, 0, sizeof(ci->i_cached_layout));
+		}
+		cap->issued &= ~CEPH_CAP_ANY_DIR_OPS;
+	}
+
+	if (recon_state->msg_version >= 2) {
+		rec.v2.cap_id = cpu_to_le64(cap->cap_id);
+		rec.v2.wanted = cpu_to_le32(__ceph_caps_wanted(ci));
+		rec.v2.issued = cpu_to_le32(cap->issued);
+		rec.v2.snaprealm = cpu_to_le64(ci->i_snap_realm->ino);
+		rec.v2.pathbase = cpu_to_le64(pathbase);
+		rec.v2.flock_len = (__force __le32)
+			((ci->i_ceph_flags & CEPH_I_ERROR_FILELOCK) ? 0 : 1);
+	} else {
+		rec.v1.cap_id = cpu_to_le64(cap->cap_id);
+		rec.v1.wanted = cpu_to_le32(__ceph_caps_wanted(ci));
+		rec.v1.issued = cpu_to_le32(cap->issued);
+		rec.v1.size = cpu_to_le64(i_size_read(inode));
+		ceph_encode_timespec64(&rec.v1.mtime, &inode->i_mtime);
+		ceph_encode_timespec64(&rec.v1.atime, &inode->i_atime);
+		rec.v1.snaprealm = cpu_to_le64(ci->i_snap_realm->ino);
+		rec.v1.pathbase = cpu_to_le64(pathbase);
+	}
+
+	if (list_empty(&ci->i_cap_snaps)) {
+		snap_follows = ci->i_head_snapc ? ci->i_head_snapc->seq : 0;
+	} else {
+		struct ceph_cap_snap *capsnap =
+			list_first_entry(&ci->i_cap_snaps,
+					 struct ceph_cap_snap, ci_item);
+		snap_follows = capsnap->follows;
+	}
+	spin_unlock(&ci->i_ceph_lock);
+
+	if (recon_state->msg_version >= 2) {
+		int num_fcntl_locks, num_flock_locks;
+		struct ceph_filelock *flocks = NULL;
+		size_t struct_len, total_len = sizeof(u64);
+		u8 struct_v = 0;
+
+encode_again:
+		if (rec.v2.flock_len) {
+			ceph_count_locks(inode, &num_fcntl_locks, &num_flock_locks);
+		} else {
+			num_fcntl_locks = 0;
+			num_flock_locks = 0;
+		}
+		if (num_fcntl_locks + num_flock_locks > 0) {
+			flocks = kmalloc_array(num_fcntl_locks + num_flock_locks,
+					       sizeof(struct ceph_filelock),
+					       GFP_NOFS);
+			if (!flocks) {
+				err = -ENOMEM;
+				goto out_err;
+			}
+			err = ceph_encode_locks_to_buffer(inode, flocks,
+							  num_fcntl_locks,
+							  num_flock_locks);
+			if (err) {
+				kfree(flocks);
+				flocks = NULL;
+				if (err == -ENOSPC)
+					goto encode_again;
+				goto out_err;
+			}
+		} else {
+			kfree(flocks);
+			flocks = NULL;
+		}
+
+		if (recon_state->msg_version >= 3) {
+			/* version, compat_version and struct_len */
+			total_len += 2 * sizeof(u8) + sizeof(u32);
+			struct_v = 2;
+		}
+		/*
+		 * number of encoded locks is stable, so copy to pagelist
+		 */
+		struct_len = 2 * sizeof(u32) +
+			    (num_fcntl_locks + num_flock_locks) *
+			    sizeof(struct ceph_filelock);
+		rec.v2.flock_len = cpu_to_le32(struct_len);
+
+		struct_len += sizeof(u32) + pathlen + sizeof(rec.v2);
+
+		if (struct_v >= 2)
+			struct_len += sizeof(u64); /* snap_follows */
+
+		total_len += struct_len;
+
+		if (pagelist->length + total_len > RECONNECT_MAX_SIZE) {
+			err = send_reconnect_partial(recon_state);
+			if (err)
+				goto out_freeflocks;
+			pagelist = recon_state->pagelist;
+		}
+
+		err = ceph_pagelist_reserve(pagelist, total_len);
+		if (err)
+			goto out_freeflocks;
+
+		ceph_pagelist_encode_64(pagelist, ceph_ino(inode));
+		if (recon_state->msg_version >= 3) {
+			ceph_pagelist_encode_8(pagelist, struct_v);
+			ceph_pagelist_encode_8(pagelist, 1);
+			ceph_pagelist_encode_32(pagelist, struct_len);
+		}
+		ceph_pagelist_encode_string(pagelist, path, pathlen);
+		ceph_pagelist_append(pagelist, &rec, sizeof(rec.v2));
+		ceph_locks_to_pagelist(flocks, pagelist,
+				       num_fcntl_locks, num_flock_locks);
+		if (struct_v >= 2)
+			ceph_pagelist_encode_64(pagelist, snap_follows);
+out_freeflocks:
+		kfree(flocks);
+	} else {
+		err = ceph_pagelist_reserve(pagelist,
+					    sizeof(u64) + sizeof(u32) +
+					    pathlen + sizeof(rec.v1));
+		if (err)
+			goto out_err;
+
+		ceph_pagelist_encode_64(pagelist, ceph_ino(inode));
+		ceph_pagelist_encode_string(pagelist, path, pathlen);
+		ceph_pagelist_append(pagelist, &rec, sizeof(rec.v1));
+	}
+
+out_err:
+	ceph_mdsc_free_path(path, pathlen);
+	if (!err)
+		recon_state->nr_caps++;
+	return err;
+}
+
+static int encode_snap_realms(struct ceph_mds_client *mdsc,
+			      struct ceph_reconnect_state *recon_state)
+{
+	struct rb_node *p;
+	struct ceph_pagelist *pagelist = recon_state->pagelist;
+	int err = 0;
+
+	if (recon_state->msg_version >= 4) {
+		err = ceph_pagelist_encode_32(pagelist, mdsc->num_snap_realms);
+		if (err < 0)
+			goto fail;
+	}
+
+	/*
+	 * snaprealms.  we provide mds with the ino, seq (version), and
+	 * parent for all of our realms.  If the mds has any newer info,
+	 * it will tell us.
+	 */
+	for (p = rb_first(&mdsc->snap_realms); p; p = rb_next(p)) {
+		struct ceph_snap_realm *realm =
+		       rb_entry(p, struct ceph_snap_realm, node);
+		struct ceph_mds_snaprealm_reconnect sr_rec;
+
+		if (recon_state->msg_version >= 4) {
+			size_t need = sizeof(u8) * 2 + sizeof(u32) +
+				      sizeof(sr_rec);
+
+			if (pagelist->length + need > RECONNECT_MAX_SIZE) {
+				err = send_reconnect_partial(recon_state);
+				if (err)
+					goto fail;
+				pagelist = recon_state->pagelist;
+			}
+
+			err = ceph_pagelist_reserve(pagelist, need);
+			if (err)
+				goto fail;
+
+			ceph_pagelist_encode_8(pagelist, 1);
+			ceph_pagelist_encode_8(pagelist, 1);
+			ceph_pagelist_encode_32(pagelist, sizeof(sr_rec));
+		}
+
+		dout(" adding snap realm %llx seq %lld parent %llx\n",
+		     realm->ino, realm->seq, realm->parent_ino);
+		sr_rec.ino = cpu_to_le64(realm->ino);
+		sr_rec.seq = cpu_to_le64(realm->seq);
+		sr_rec.parent = cpu_to_le64(realm->parent_ino);
+
+		err = ceph_pagelist_append(pagelist, &sr_rec, sizeof(sr_rec));
+		if (err)
+			goto fail;
+
+		recon_state->nr_realms++;
+	}
+fail:
+	return err;
+}
+
+
+/*
+ * If an MDS fails and recovers, clients need to reconnect in order to
+ * reestablish shared state.  This includes all caps issued through
+ * this session _and_ the snap_realm hierarchy.  Because it's not
+ * clear which snap realms the mds cares about, we send everything we
+ * know about.. that ensures we'll then get any new info the
+ * recovering MDS might have.
+ *
+ * This is a relatively heavyweight operation, but it's rare.
+ */
+static void send_mds_reconnect(struct ceph_mds_client *mdsc,
+			       struct ceph_mds_session *session)
+{
+	struct ceph_msg *reply;
+	int mds = session->s_mds;
+	int err = -ENOMEM;
+	struct ceph_reconnect_state recon_state = {
+		.session = session,
+	};
+	LIST_HEAD(dispose);
+
+	pr_info("mds%d reconnect start\n", mds);
+
+	recon_state.pagelist = ceph_pagelist_alloc(GFP_NOFS);
+	if (!recon_state.pagelist)
+		goto fail_nopagelist;
+
+	reply = ceph_msg_new2(CEPH_MSG_CLIENT_RECONNECT, 0, 1, GFP_NOFS, false);
+	if (!reply)
+		goto fail_nomsg;
+
+	xa_destroy(&session->s_delegated_inos);
+
+	mutex_lock(&session->s_mutex);
+	session->s_state = CEPH_MDS_SESSION_RECONNECTING;
+	session->s_seq = 0;
+
+	dout("session %p state %s\n", session,
+	     ceph_session_state_name(session->s_state));
+
+	atomic_inc(&session->s_cap_gen);
+
+	spin_lock(&session->s_cap_lock);
+	/* don't know if session is readonly */
+	session->s_readonly = 0;
+	/*
+	 * notify __ceph_remove_cap() that we are composing cap reconnect.
+	 * If a cap get released before being added to the cap reconnect,
+	 * __ceph_remove_cap() should skip queuing cap release.
+	 */
+	session->s_cap_reconnect = 1;
+	/* drop old cap expires; we're about to reestablish that state */
+	detach_cap_releases(session, &dispose);
+	spin_unlock(&session->s_cap_lock);
+	dispose_cap_releases(mdsc, &dispose);
+
+	/* trim unused caps to reduce MDS's cache rejoin time */
+	if (mdsc->fsc->sb->s_root)
+		shrink_dcache_parent(mdsc->fsc->sb->s_root);
+
+	ceph_con_close(&session->s_con);
+	ceph_con_open(&session->s_con,
+		      CEPH_ENTITY_TYPE_MDS, mds,
+		      ceph_mdsmap_get_addr(mdsc->mdsmap, mds));
+
+	/* replay unsafe requests */
+	replay_unsafe_requests(mdsc, session);
+
+	ceph_early_kick_flushing_caps(mdsc, session);
+
+	down_read(&mdsc->snap_rwsem);
+
+	/* placeholder for nr_caps */
+	err = ceph_pagelist_encode_32(recon_state.pagelist, 0);
+	if (err)
+		goto fail;
+
+	if (test_bit(CEPHFS_FEATURE_MULTI_RECONNECT, &session->s_features)) {
+		recon_state.msg_version = 3;
+		recon_state.allow_multi = true;
+	} else if (session->s_con.peer_features & CEPH_FEATURE_MDSENC) {
+		recon_state.msg_version = 3;
+	} else {
+		recon_state.msg_version = 2;
+	}
+	/* trsaverse this session's caps */
+	err = ceph_iterate_session_caps(session, reconnect_caps_cb, &recon_state);
+
+	spin_lock(&session->s_cap_lock);
+	session->s_cap_reconnect = 0;
+	spin_unlock(&session->s_cap_lock);
+
+	if (err < 0)
+		goto fail;
+
+	/* check if all realms can be encoded into current message */
+	if (mdsc->num_snap_realms) {
+		size_t total_len =
+			recon_state.pagelist->length +
+			mdsc->num_snap_realms *
+			sizeof(struct ceph_mds_snaprealm_reconnect);
+		if (recon_state.msg_version >= 4) {
+			/* number of realms */
+			total_len += sizeof(u32);
+			/* version, compat_version and struct_len */
+			total_len += mdsc->num_snap_realms *
+				     (2 * sizeof(u8) + sizeof(u32));
+		}
+		if (total_len > RECONNECT_MAX_SIZE) {
+			if (!recon_state.allow_multi) {
+				err = -ENOSPC;
+				goto fail;
+			}
+			if (recon_state.nr_caps) {
+				err = send_reconnect_partial(&recon_state);
+				if (err)
+					goto fail;
+			}
+			recon_state.msg_version = 5;
+		}
+	}
+
+	err = encode_snap_realms(mdsc, &recon_state);
+	if (err < 0)
+		goto fail;
+
+	if (recon_state.msg_version >= 5) {
+		err = ceph_pagelist_encode_8(recon_state.pagelist, 0);
+		if (err < 0)
+			goto fail;
+	}
+
+	if (recon_state.nr_caps || recon_state.nr_realms) {
+		struct page *page =
+			list_first_entry(&recon_state.pagelist->head,
+					struct page, lru);
+		__le32 *addr = kmap_atomic(page);
+		if (recon_state.nr_caps) {
+			WARN_ON(recon_state.nr_realms != mdsc->num_snap_realms);
+			*addr = cpu_to_le32(recon_state.nr_caps);
+		} else if (recon_state.msg_version >= 4) {
+			*(addr + 1) = cpu_to_le32(recon_state.nr_realms);
+		}
+		kunmap_atomic(addr);
+	}
+
+	reply->hdr.version = cpu_to_le16(recon_state.msg_version);
+	if (recon_state.msg_version >= 4)
+		reply->hdr.compat_version = cpu_to_le16(4);
+
+	reply->hdr.data_len = cpu_to_le32(recon_state.pagelist->length);
+	ceph_msg_data_add_pagelist(reply, recon_state.pagelist);
+
+	ceph_con_send(&session->s_con, reply);
+
+	mutex_unlock(&session->s_mutex);
+
+	mutex_lock(&mdsc->mutex);
+	__wake_requests(mdsc, &session->s_waiting);
+	mutex_unlock(&mdsc->mutex);
+
+	up_read(&mdsc->snap_rwsem);
+	ceph_pagelist_release(recon_state.pagelist);
+	return;
+
+fail:
+	ceph_msg_put(reply);
+	up_read(&mdsc->snap_rwsem);
+	mutex_unlock(&session->s_mutex);
+fail_nomsg:
+	ceph_pagelist_release(recon_state.pagelist);
+fail_nopagelist:
+	pr_err("error %d preparing reconnect for mds%d\n", err, mds);
+	return;
+}
+
+
+/*
+ * compare old and new mdsmaps, kicking requests
+ * and closing out old connections as necessary
+ *
+ * called under mdsc->mutex.
+ */
+static void check_new_map(struct ceph_mds_client *mdsc,
+			  struct ceph_mdsmap *newmap,
+			  struct ceph_mdsmap *oldmap)
+{
+	int i, j, err;
+	int oldstate, newstate;
+	struct ceph_mds_session *s;
+	unsigned long targets[DIV_ROUND_UP(CEPH_MAX_MDS, sizeof(unsigned long))] = {0};
+
+	dout("check_new_map new %u old %u\n",
+	     newmap->m_epoch, oldmap->m_epoch);
+
+	if (newmap->m_info) {
+		for (i = 0; i < newmap->possible_max_rank; i++) {
+			for (j = 0; j < newmap->m_info[i].num_export_targets; j++)
+				set_bit(newmap->m_info[i].export_targets[j], targets);
+		}
+	}
+
+	for (i = 0; i < oldmap->possible_max_rank && i < mdsc->max_sessions; i++) {
+		if (!mdsc->sessions[i])
+			continue;
+		s = mdsc->sessions[i];
+		oldstate = ceph_mdsmap_get_state(oldmap, i);
+		newstate = ceph_mdsmap_get_state(newmap, i);
+
+		dout("check_new_map mds%d state %s%s -> %s%s (session %s)\n",
+		     i, ceph_mds_state_name(oldstate),
+		     ceph_mdsmap_is_laggy(oldmap, i) ? " (laggy)" : "",
+		     ceph_mds_state_name(newstate),
+		     ceph_mdsmap_is_laggy(newmap, i) ? " (laggy)" : "",
+		     ceph_session_state_name(s->s_state));
+
+		if (i >= newmap->possible_max_rank) {
+			/* force close session for stopped mds */
+			ceph_get_mds_session(s);
+			__unregister_session(mdsc, s);
+			__wake_requests(mdsc, &s->s_waiting);
+			mutex_unlock(&mdsc->mutex);
+
+			mutex_lock(&s->s_mutex);
+			cleanup_session_requests(mdsc, s);
+			remove_session_caps(s);
+			mutex_unlock(&s->s_mutex);
+
+			ceph_put_mds_session(s);
+
+			mutex_lock(&mdsc->mutex);
+			kick_requests(mdsc, i);
+			continue;
+		}
+
+		if (memcmp(ceph_mdsmap_get_addr(oldmap, i),
+			   ceph_mdsmap_get_addr(newmap, i),
+			   sizeof(struct ceph_entity_addr))) {
+			/* just close it */
+			mutex_unlock(&mdsc->mutex);
+			mutex_lock(&s->s_mutex);
+			mutex_lock(&mdsc->mutex);
+			ceph_con_close(&s->s_con);
+			mutex_unlock(&s->s_mutex);
+			s->s_state = CEPH_MDS_SESSION_RESTARTING;
+		} else if (oldstate == newstate) {
+			continue;  /* nothing new with this mds */
+		}
+
+		/*
+		 * send reconnect?
+		 */
+		if (s->s_state == CEPH_MDS_SESSION_RESTARTING &&
+		    newstate >= CEPH_MDS_STATE_RECONNECT) {
+			mutex_unlock(&mdsc->mutex);
+			clear_bit(i, targets);
+			send_mds_reconnect(mdsc, s);
+			mutex_lock(&mdsc->mutex);
+		}
+
+		/*
+		 * kick request on any mds that has gone active.
+		 */
+		if (oldstate < CEPH_MDS_STATE_ACTIVE &&
+		    newstate >= CEPH_MDS_STATE_ACTIVE) {
+			if (oldstate != CEPH_MDS_STATE_CREATING &&
+			    oldstate != CEPH_MDS_STATE_STARTING)
+				pr_info("mds%d recovery completed\n", s->s_mds);
+			kick_requests(mdsc, i);
+			mutex_unlock(&mdsc->mutex);
+			mutex_lock(&s->s_mutex);
+			mutex_lock(&mdsc->mutex);
+			ceph_kick_flushing_caps(mdsc, s);
+			mutex_unlock(&s->s_mutex);
+			wake_up_session_caps(s, RECONNECT);
+		}
+	}
+
+	/*
+	 * Only open and reconnect sessions that don't exist yet.
+	 */
+	for (i = 0; i < newmap->possible_max_rank; i++) {
+		/*
+		 * In case the import MDS is crashed just after
+		 * the EImportStart journal is flushed, so when
+		 * a standby MDS takes over it and is replaying
+		 * the EImportStart journal the new MDS daemon
+		 * will wait the client to reconnect it, but the
+		 * client may never register/open the session yet.
+		 *
+		 * Will try to reconnect that MDS daemon if the
+		 * rank number is in the export targets array and
+		 * is the up:reconnect state.
+		 */
+		newstate = ceph_mdsmap_get_state(newmap, i);
+		if (!test_bit(i, targets) || newstate != CEPH_MDS_STATE_RECONNECT)
+			continue;
+
+		/*
+		 * The session maybe registered and opened by some
+		 * requests which were choosing random MDSes during
+		 * the mdsc->mutex's unlock/lock gap below in rare
+		 * case. But the related MDS daemon will just queue
+		 * that requests and be still waiting for the client's
+		 * reconnection request in up:reconnect state.
+		 */
+		s = __ceph_lookup_mds_session(mdsc, i);
+		if (likely(!s)) {
+			s = __open_export_target_session(mdsc, i);
+			if (IS_ERR(s)) {
+				err = PTR_ERR(s);
+				pr_err("failed to open export target session, err %d\n",
+				       err);
+				continue;
+			}
+		}
+		dout("send reconnect to export target mds.%d\n", i);
+		mutex_unlock(&mdsc->mutex);
+		send_mds_reconnect(mdsc, s);
+		ceph_put_mds_session(s);
+		mutex_lock(&mdsc->mutex);
+	}
+
+	for (i = 0; i < newmap->possible_max_rank && i < mdsc->max_sessions; i++) {
+		s = mdsc->sessions[i];
+		if (!s)
+			continue;
+		if (!ceph_mdsmap_is_laggy(newmap, i))
+			continue;
+		if (s->s_state == CEPH_MDS_SESSION_OPEN ||
+		    s->s_state == CEPH_MDS_SESSION_HUNG ||
+		    s->s_state == CEPH_MDS_SESSION_CLOSING) {
+			dout(" connecting to export targets of laggy mds%d\n",
+			     i);
+			__open_export_target_sessions(mdsc, s);
+		}
+	}
+}
+
+
+
+/*
+ * leases
+ */
+
+/*
+ * caller must hold session s_mutex, dentry->d_lock
+ */
+void __ceph_mdsc_drop_dentry_lease(struct dentry *dentry)
+{
+	struct ceph_dentry_info *di = ceph_dentry(dentry);
+
+	ceph_put_mds_session(di->lease_session);
+	di->lease_session = NULL;
+}
+
+static void handle_lease(struct ceph_mds_client *mdsc,
+			 struct ceph_mds_session *session,
+			 struct ceph_msg *msg)
+{
+	struct super_block *sb = mdsc->fsc->sb;
+	struct inode *inode;
+	struct dentry *parent, *dentry;
+	struct ceph_dentry_info *di;
+	int mds = session->s_mds;
+	struct ceph_mds_lease *h = msg->front.iov_base;
+	u32 seq;
+	struct ceph_vino vino;
+	struct qstr dname;
+	int release = 0;
+
+	dout("handle_lease from mds%d\n", mds);
+
+	if (!ceph_inc_mds_stopping_blocker(mdsc, session))
+		return;
+
+	/* decode */
+	if (msg->front.iov_len < sizeof(*h) + sizeof(u32))
+		goto bad;
+	vino.ino = le64_to_cpu(h->ino);
+	vino.snap = CEPH_NOSNAP;
+	seq = le32_to_cpu(h->seq);
+	dname.len = get_unaligned_le32(h + 1);
+	if (msg->front.iov_len < sizeof(*h) + sizeof(u32) + dname.len)
+		goto bad;
+	dname.name = (void *)(h + 1) + sizeof(u32);
+
+	/* lookup inode */
+	inode = ceph_find_inode(sb, vino);
+	dout("handle_lease %s, ino %llx %p %.*s\n",
+	     ceph_lease_op_name(h->action), vino.ino, inode,
+	     dname.len, dname.name);
+
+	mutex_lock(&session->s_mutex);
+	if (!inode) {
+		dout("handle_lease no inode %llx\n", vino.ino);
+		goto release;
+	}
+
+	/* dentry */
+	parent = d_find_alias(inode);
+	if (!parent) {
+		dout("no parent dentry on inode %p\n", inode);
+		WARN_ON(1);
+		goto release;  /* hrm... */
+	}
+	dname.hash = full_name_hash(parent, dname.name, dname.len);
+	dentry = d_lookup(parent, &dname);
+	dput(parent);
+	if (!dentry)
+		goto release;
+
+	spin_lock(&dentry->d_lock);
+	di = ceph_dentry(dentry);
+	switch (h->action) {
+	case CEPH_MDS_LEASE_REVOKE:
+		if (di->lease_session == session) {
+			if (ceph_seq_cmp(di->lease_seq, seq) > 0)
+				h->seq = cpu_to_le32(di->lease_seq);
+			__ceph_mdsc_drop_dentry_lease(dentry);
+		}
+		release = 1;
+		break;
+
+	case CEPH_MDS_LEASE_RENEW:
+		if (di->lease_session == session &&
+		    di->lease_gen == atomic_read(&session->s_cap_gen) &&
+		    di->lease_renew_from &&
+		    di->lease_renew_after == 0) {
+			unsigned long duration =
+				msecs_to_jiffies(le32_to_cpu(h->duration_ms));
+
+			di->lease_seq = seq;
+			di->time = di->lease_renew_from + duration;
+			di->lease_renew_after = di->lease_renew_from +
+				(duration >> 1);
+			di->lease_renew_from = 0;
+		}
+		break;
+	}
+	spin_unlock(&dentry->d_lock);
+	dput(dentry);
+
+	if (!release)
+		goto out;
+
+release:
+	/* let's just reuse the same message */
+	h->action = CEPH_MDS_LEASE_REVOKE_ACK;
+	ceph_msg_get(msg);
+	ceph_con_send(&session->s_con, msg);
+
+out:
+	mutex_unlock(&session->s_mutex);
+	iput(inode);
+
+	ceph_dec_mds_stopping_blocker(mdsc);
+	return;
+
+bad:
+	ceph_dec_mds_stopping_blocker(mdsc);
+
+	pr_err("corrupt lease message\n");
+	ceph_msg_dump(msg);
+}
+
+void ceph_mdsc_lease_send_msg(struct ceph_mds_session *session,
+			      struct dentry *dentry, char action,
+			      u32 seq)
+{
+	struct ceph_msg *msg;
+	struct ceph_mds_lease *lease;
+	struct inode *dir;
+	int len = sizeof(*lease) + sizeof(u32) + NAME_MAX;
+
+	dout("lease_send_msg identry %p %s to mds%d\n",
+	     dentry, ceph_lease_op_name(action), session->s_mds);
+
+	msg = ceph_msg_new(CEPH_MSG_CLIENT_LEASE, len, GFP_NOFS, false);
+	if (!msg)
+		return;
+	lease = msg->front.iov_base;
+	lease->action = action;
+	lease->seq = cpu_to_le32(seq);
+
+	spin_lock(&dentry->d_lock);
+	dir = d_inode(dentry->d_parent);
+	lease->ino = cpu_to_le64(ceph_ino(dir));
+	lease->first = lease->last = cpu_to_le64(ceph_snap(dir));
+
+	put_unaligned_le32(dentry->d_name.len, lease + 1);
+	memcpy((void *)(lease + 1) + 4,
+	       dentry->d_name.name, dentry->d_name.len);
+	spin_unlock(&dentry->d_lock);
+
+	ceph_con_send(&session->s_con, msg);
+}
+
+/*
+ * lock unlock the session, to wait ongoing session activities
+ */
+static void lock_unlock_session(struct ceph_mds_session *s)
+{
+	mutex_lock(&s->s_mutex);
+	mutex_unlock(&s->s_mutex);
+}
+
+static void maybe_recover_session(struct ceph_mds_client *mdsc)
+{
+	struct ceph_fs_client *fsc = mdsc->fsc;
+
+	if (!ceph_test_mount_opt(fsc, CLEANRECOVER))
+		return;
+
+	if (READ_ONCE(fsc->mount_state) != CEPH_MOUNT_MOUNTED)
+		return;
+
+	if (!READ_ONCE(fsc->blocklisted))
+		return;
+
+	pr_info("auto reconnect after blocklisted\n");
+	ceph_force_reconnect(fsc->sb);
+}
+
+bool check_session_state(struct ceph_mds_session *s)
+{
+	switch (s->s_state) {
+	case CEPH_MDS_SESSION_OPEN:
+		if (s->s_ttl && time_after(jiffies, s->s_ttl)) {
+			s->s_state = CEPH_MDS_SESSION_HUNG;
+			pr_info("mds%d hung\n", s->s_mds);
+		}
+		break;
+	case CEPH_MDS_SESSION_CLOSING:
+	case CEPH_MDS_SESSION_NEW:
+	case CEPH_MDS_SESSION_RESTARTING:
+	case CEPH_MDS_SESSION_CLOSED:
+	case CEPH_MDS_SESSION_REJECTED:
+		return false;
+	}
+
+	return true;
+}
+
+/*
+ * If the sequence is incremented while we're waiting on a REQUEST_CLOSE reply,
+ * then we need to retransmit that request.
+ */
+void inc_session_sequence(struct ceph_mds_session *s)
+{
+	lockdep_assert_held(&s->s_mutex);
+
+	s->s_seq++;
+
+	if (s->s_state == CEPH_MDS_SESSION_CLOSING) {
+		int ret;
+
+		dout("resending session close request for mds%d\n", s->s_mds);
+		ret = request_close_session(s);
+		if (ret < 0)
+			pr_err("unable to close session to mds%d: %d\n",
+			       s->s_mds, ret);
+	}
+}
+
+/*
+ * delayed work -- periodically trim expired leases, renew caps with mds.  If
+ * the @delay parameter is set to 0 or if it's more than 5 secs, the default
+ * workqueue delay value of 5 secs will be used.
+ */
+static void schedule_delayed(struct ceph_mds_client *mdsc, unsigned long delay)
+{
+	unsigned long max_delay = HZ * 5;
+
+	/* 5 secs default delay */
+	if (!delay || (delay > max_delay))
+		delay = max_delay;
+	schedule_delayed_work(&mdsc->delayed_work,
+			      round_jiffies_relative(delay));
+}
+
+static void delayed_work(struct work_struct *work)
+{
+	struct ceph_mds_client *mdsc =
+		container_of(work, struct ceph_mds_client, delayed_work.work);
+	unsigned long delay;
+	int renew_interval;
+	int renew_caps;
+	int i;
+
+	dout("mdsc delayed_work\n");
+
+	if (mdsc->stopping >= CEPH_MDSC_STOPPING_FLUSHED)
+		return;
+
+	mutex_lock(&mdsc->mutex);
+	renew_interval = mdsc->mdsmap->m_session_timeout >> 2;
+	renew_caps = time_after_eq(jiffies, HZ*renew_interval +
+				   mdsc->last_renew_caps);
+	if (renew_caps)
+		mdsc->last_renew_caps = jiffies;
+
+	for (i = 0; i < mdsc->max_sessions; i++) {
+		struct ceph_mds_session *s = __ceph_lookup_mds_session(mdsc, i);
+		if (!s)
+			continue;
+
+		if (!check_session_state(s)) {
+			ceph_put_mds_session(s);
+			continue;
+		}
+		mutex_unlock(&mdsc->mutex);
+
+		mutex_lock(&s->s_mutex);
+		if (renew_caps)
+			send_renew_caps(mdsc, s);
+		else
+			ceph_con_keepalive(&s->s_con);
+		if (s->s_state == CEPH_MDS_SESSION_OPEN ||
+		    s->s_state == CEPH_MDS_SESSION_HUNG)
+			ceph_send_cap_releases(mdsc, s);
+		mutex_unlock(&s->s_mutex);
+		ceph_put_mds_session(s);
+
+		mutex_lock(&mdsc->mutex);
+	}
+	mutex_unlock(&mdsc->mutex);
+
+	delay = ceph_check_delayed_caps(mdsc);
+
+	ceph_queue_cap_reclaim_work(mdsc);
+
+	ceph_trim_snapid_map(mdsc);
+
+	maybe_recover_session(mdsc);
+
+	schedule_delayed(mdsc, delay);
+}
+
+int ceph_mdsc_init(struct ceph_fs_client *fsc)
+
+{
+	struct ceph_mds_client *mdsc;
+	int err;
+
+	mdsc = kzalloc(sizeof(struct ceph_mds_client), GFP_NOFS);
+	if (!mdsc)
+		return -ENOMEM;
+	mdsc->fsc = fsc;
+	mutex_init(&mdsc->mutex);
+	mdsc->mdsmap = kzalloc(sizeof(*mdsc->mdsmap), GFP_NOFS);
+	if (!mdsc->mdsmap) {
+		err = -ENOMEM;
+		goto err_mdsc;
+	}
+
+	init_completion(&mdsc->safe_umount_waiters);
+	spin_lock_init(&mdsc->stopping_lock);
+	atomic_set(&mdsc->stopping_blockers, 0);
+	init_completion(&mdsc->stopping_waiter);
+	init_waitqueue_head(&mdsc->session_close_wq);
+	INIT_LIST_HEAD(&mdsc->waiting_for_map);
+	mdsc->quotarealms_inodes = RB_ROOT;
+	mutex_init(&mdsc->quotarealms_inodes_mutex);
+	init_rwsem(&mdsc->snap_rwsem);
+	mdsc->snap_realms = RB_ROOT;
+	INIT_LIST_HEAD(&mdsc->snap_empty);
+	spin_lock_init(&mdsc->snap_empty_lock);
+	mdsc->request_tree = RB_ROOT;
+	INIT_DELAYED_WORK(&mdsc->delayed_work, delayed_work);
+	mdsc->last_renew_caps = jiffies;
+	INIT_LIST_HEAD(&mdsc->cap_delay_list);
+	INIT_LIST_HEAD(&mdsc->cap_wait_list);
+	spin_lock_init(&mdsc->cap_delay_lock);
+	INIT_LIST_HEAD(&mdsc->snap_flush_list);
+	spin_lock_init(&mdsc->snap_flush_lock);
+	mdsc->last_cap_flush_tid = 1;
+	INIT_LIST_HEAD(&mdsc->cap_flush_list);
+	INIT_LIST_HEAD(&mdsc->cap_dirty_migrating);
+	spin_lock_init(&mdsc->cap_dirty_lock);
+	init_waitqueue_head(&mdsc->cap_flushing_wq);
+	INIT_WORK(&mdsc->cap_reclaim_work, ceph_cap_reclaim_work);
+	err = ceph_metric_init(&mdsc->metric);
+	if (err)
+		goto err_mdsmap;
+
+	spin_lock_init(&mdsc->dentry_list_lock);
+	INIT_LIST_HEAD(&mdsc->dentry_leases);
+	INIT_LIST_HEAD(&mdsc->dentry_dir_leases);
+
+	ceph_caps_init(mdsc);
+	ceph_adjust_caps_max_min(mdsc, fsc->mount_options);
+
+	spin_lock_init(&mdsc->snapid_map_lock);
+	mdsc->snapid_map_tree = RB_ROOT;
+	INIT_LIST_HEAD(&mdsc->snapid_map_lru);
+
+	init_rwsem(&mdsc->pool_perm_rwsem);
+	mdsc->pool_perm_tree = RB_ROOT;
+
+	strscpy(mdsc->nodename, utsname()->nodename,
+		sizeof(mdsc->nodename));
+
+	fsc->mdsc = mdsc;
+	return 0;
+
+err_mdsmap:
+	kfree(mdsc->mdsmap);
+err_mdsc:
+	kfree(mdsc);
+	return err;
+}
+
+/*
+ * Wait for safe replies on open mds requests.  If we time out, drop
+ * all requests from the tree to avoid dangling dentry refs.
+ */
+static void wait_requests(struct ceph_mds_client *mdsc)
+{
+	struct ceph_options *opts = mdsc->fsc->client->options;
+	struct ceph_mds_request *req;
+
+	mutex_lock(&mdsc->mutex);
+	if (__get_oldest_req(mdsc)) {
+		mutex_unlock(&mdsc->mutex);
+
+		dout("wait_requests waiting for requests\n");
+		wait_for_completion_timeout(&mdsc->safe_umount_waiters,
+				    ceph_timeout_jiffies(opts->mount_timeout));
+
+		/* tear down remaining requests */
+		mutex_lock(&mdsc->mutex);
+		while ((req = __get_oldest_req(mdsc))) {
+			dout("wait_requests timed out on tid %llu\n",
+			     req->r_tid);
+			list_del_init(&req->r_wait);
+			__unregister_request(mdsc, req);
+		}
+	}
+	mutex_unlock(&mdsc->mutex);
+	dout("wait_requests done\n");
+}
+
+void send_flush_mdlog(struct ceph_mds_session *s)
+{
+	struct ceph_msg *msg;
+
+	/*
+	 * Pre-luminous MDS crashes when it sees an unknown session request
+	 */
+	if (!CEPH_HAVE_FEATURE(s->s_con.peer_features, SERVER_LUMINOUS))
+		return;
+
+	mutex_lock(&s->s_mutex);
+	dout("request mdlog flush to mds%d (%s)s seq %lld\n", s->s_mds,
+	     ceph_session_state_name(s->s_state), s->s_seq);
+	msg = ceph_create_session_msg(CEPH_SESSION_REQUEST_FLUSH_MDLOG,
+				      s->s_seq);
+	if (!msg) {
+		pr_err("failed to request mdlog flush to mds%d (%s) seq %lld\n",
+		       s->s_mds, ceph_session_state_name(s->s_state), s->s_seq);
+	} else {
+		ceph_con_send(&s->s_con, msg);
+	}
+	mutex_unlock(&s->s_mutex);
+}
+
+/*
+ * called before mount is ro, and before dentries are torn down.
+ * (hmm, does this still race with new lookups?)
+ */
+void ceph_mdsc_pre_umount(struct ceph_mds_client *mdsc)
+{
+	dout("pre_umount\n");
+	mdsc->stopping = CEPH_MDSC_STOPPING_BEGIN;
+
+	ceph_mdsc_iterate_sessions(mdsc, send_flush_mdlog, true);
+	ceph_mdsc_iterate_sessions(mdsc, lock_unlock_session, false);
+	ceph_flush_dirty_caps(mdsc);
+	wait_requests(mdsc);
+
+	/*
+	 * wait for reply handlers to drop their request refs and
+	 * their inode/dcache refs
+	 */
+	ceph_msgr_flush();
+
+	ceph_cleanup_quotarealms_inodes(mdsc);
+}
+
+/*
+ * flush the mdlog and wait for all write mds requests to flush.
+ */
+static void flush_mdlog_and_wait_mdsc_unsafe_requests(struct ceph_mds_client *mdsc,
+						 u64 want_tid)
+{
+	struct ceph_mds_request *req = NULL, *nextreq;
+	struct ceph_mds_session *last_session = NULL;
+	struct rb_node *n;
+
+	mutex_lock(&mdsc->mutex);
+	dout("%s want %lld\n", __func__, want_tid);
+restart:
+	req = __get_oldest_req(mdsc);
+	while (req && req->r_tid <= want_tid) {
+		/* find next request */
+		n = rb_next(&req->r_node);
+		if (n)
+			nextreq = rb_entry(n, struct ceph_mds_request, r_node);
+		else
+			nextreq = NULL;
+		if (req->r_op != CEPH_MDS_OP_SETFILELOCK &&
+		    (req->r_op & CEPH_MDS_OP_WRITE)) {
+			struct ceph_mds_session *s = req->r_session;
+
+			if (!s) {
+				req = nextreq;
+				continue;
+			}
+
+			/* write op */
+			ceph_mdsc_get_request(req);
+			if (nextreq)
+				ceph_mdsc_get_request(nextreq);
+			s = ceph_get_mds_session(s);
+			mutex_unlock(&mdsc->mutex);
+
+			/* send flush mdlog request to MDS */
+			if (last_session != s) {
+				send_flush_mdlog(s);
+				ceph_put_mds_session(last_session);
+				last_session = s;
+			} else {
+				ceph_put_mds_session(s);
+			}
+			dout("%s wait on %llu (want %llu)\n", __func__,
+			     req->r_tid, want_tid);
+			wait_for_completion(&req->r_safe_completion);
+
+			mutex_lock(&mdsc->mutex);
+			ceph_mdsc_put_request(req);
+			if (!nextreq)
+				break;  /* next dne before, so we're done! */
+			if (RB_EMPTY_NODE(&nextreq->r_node)) {
+				/* next request was removed from tree */
+				ceph_mdsc_put_request(nextreq);
+				goto restart;
+			}
+			ceph_mdsc_put_request(nextreq);  /* won't go away */
+		}
+		req = nextreq;
+	}
+	mutex_unlock(&mdsc->mutex);
+	ceph_put_mds_session(last_session);
+	dout("%s done\n", __func__);
+}
+
+void ceph_mdsc_sync(struct ceph_mds_client *mdsc)
+{
+	u64 want_tid, want_flush;
+
+	if (READ_ONCE(mdsc->fsc->mount_state) >= CEPH_MOUNT_SHUTDOWN)
+		return;
+
+	dout("sync\n");
+	mutex_lock(&mdsc->mutex);
+	want_tid = mdsc->last_tid;
+	mutex_unlock(&mdsc->mutex);
+
+	ceph_flush_dirty_caps(mdsc);
+	spin_lock(&mdsc->cap_dirty_lock);
+	want_flush = mdsc->last_cap_flush_tid;
+	if (!list_empty(&mdsc->cap_flush_list)) {
+		struct ceph_cap_flush *cf =
+			list_last_entry(&mdsc->cap_flush_list,
+					struct ceph_cap_flush, g_list);
+		cf->wake = true;
+	}
+	spin_unlock(&mdsc->cap_dirty_lock);
+
+	dout("sync want tid %lld flush_seq %lld\n",
+	     want_tid, want_flush);
+
+	flush_mdlog_and_wait_mdsc_unsafe_requests(mdsc, want_tid);
+	wait_caps_flush(mdsc, want_flush);
+}
+
+/*
+ * true if all sessions are closed, or we force unmount
+ */
+static bool done_closing_sessions(struct ceph_mds_client *mdsc, int skipped)
+{
+	if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN)
+		return true;
+	return atomic_read(&mdsc->num_sessions) <= skipped;
+}
+
+/*
+ * called after sb is ro or when metadata corrupted.
+ */
+void ceph_mdsc_close_sessions(struct ceph_mds_client *mdsc)
+{
+	struct ceph_options *opts = mdsc->fsc->client->options;
+	struct ceph_mds_session *session;
+	int i;
+	int skipped = 0;
+
+	dout("close_sessions\n");
+
+	/* close sessions */
+	mutex_lock(&mdsc->mutex);
+	for (i = 0; i < mdsc->max_sessions; i++) {
+		session = __ceph_lookup_mds_session(mdsc, i);
+		if (!session)
+			continue;
+		mutex_unlock(&mdsc->mutex);
+		mutex_lock(&session->s_mutex);
+		if (__close_session(mdsc, session) <= 0)
+			skipped++;
+		mutex_unlock(&session->s_mutex);
+		ceph_put_mds_session(session);
+		mutex_lock(&mdsc->mutex);
+	}
+	mutex_unlock(&mdsc->mutex);
+
+	dout("waiting for sessions to close\n");
+	wait_event_timeout(mdsc->session_close_wq,
+			   done_closing_sessions(mdsc, skipped),
+			   ceph_timeout_jiffies(opts->mount_timeout));
+
+	/* tear down remaining sessions */
+	mutex_lock(&mdsc->mutex);
+	for (i = 0; i < mdsc->max_sessions; i++) {
+		if (mdsc->sessions[i]) {
+			session = ceph_get_mds_session(mdsc->sessions[i]);
+			__unregister_session(mdsc, session);
+			mutex_unlock(&mdsc->mutex);
+			mutex_lock(&session->s_mutex);
+			remove_session_caps(session);
+			mutex_unlock(&session->s_mutex);
+			ceph_put_mds_session(session);
+			mutex_lock(&mdsc->mutex);
+		}
+	}
+	WARN_ON(!list_empty(&mdsc->cap_delay_list));
+	mutex_unlock(&mdsc->mutex);
+
+	ceph_cleanup_snapid_map(mdsc);
+	ceph_cleanup_global_and_empty_realms(mdsc);
+
+	cancel_work_sync(&mdsc->cap_reclaim_work);
+	cancel_delayed_work_sync(&mdsc->delayed_work); /* cancel timer */
+
+	dout("stopped\n");
+}
+
+void ceph_mdsc_force_umount(struct ceph_mds_client *mdsc)
+{
+	struct ceph_mds_session *session;
+	int mds;
+
+	dout("force umount\n");
+
+	mutex_lock(&mdsc->mutex);
+	for (mds = 0; mds < mdsc->max_sessions; mds++) {
+		session = __ceph_lookup_mds_session(mdsc, mds);
+		if (!session)
+			continue;
+
+		if (session->s_state == CEPH_MDS_SESSION_REJECTED)
+			__unregister_session(mdsc, session);
+		__wake_requests(mdsc, &session->s_waiting);
+		mutex_unlock(&mdsc->mutex);
+
+		mutex_lock(&session->s_mutex);
+		__close_session(mdsc, session);
+		if (session->s_state == CEPH_MDS_SESSION_CLOSING) {
+			cleanup_session_requests(mdsc, session);
+			remove_session_caps(session);
+		}
+		mutex_unlock(&session->s_mutex);
+		ceph_put_mds_session(session);
+
+		mutex_lock(&mdsc->mutex);
+		kick_requests(mdsc, mds);
+	}
+	__wake_requests(mdsc, &mdsc->waiting_for_map);
+	mutex_unlock(&mdsc->mutex);
+}
+
+static void ceph_mdsc_stop(struct ceph_mds_client *mdsc)
+{
+	dout("stop\n");
+	/*
+	 * Make sure the delayed work stopped before releasing
+	 * the resources.
+	 *
+	 * Because the cancel_delayed_work_sync() will only
+	 * guarantee that the work finishes executing. But the
+	 * delayed work will re-arm itself again after that.
+	 */
+	flush_delayed_work(&mdsc->delayed_work);
+
+	if (mdsc->mdsmap)
+		ceph_mdsmap_destroy(mdsc->mdsmap);
+	kfree(mdsc->sessions);
+	ceph_caps_finalize(mdsc);
+	ceph_pool_perm_destroy(mdsc);
+}
+
+void ceph_mdsc_destroy(struct ceph_fs_client *fsc)
+{
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	dout("mdsc_destroy %p\n", mdsc);
+
+	if (!mdsc)
+		return;
+
+	/* flush out any connection work with references to us */
+	ceph_msgr_flush();
+
+	ceph_mdsc_stop(mdsc);
+
+	ceph_metric_destroy(&mdsc->metric);
+
+	fsc->mdsc = NULL;
+	kfree(mdsc);
+	dout("mdsc_destroy %p done\n", mdsc);
+}
+
+void ceph_mdsc_handle_fsmap(struct ceph_mds_client *mdsc, struct ceph_msg *msg)
+{
+	struct ceph_fs_client *fsc = mdsc->fsc;
+	const char *mds_namespace = fsc->mount_options->mds_namespace;
+	void *p = msg->front.iov_base;
+	void *end = p + msg->front.iov_len;
+	u32 epoch;
+	u32 num_fs;
+	u32 mount_fscid = (u32)-1;
+	int err = -EINVAL;
+
+	ceph_decode_need(&p, end, sizeof(u32), bad);
+	epoch = ceph_decode_32(&p);
+
+	dout("handle_fsmap epoch %u\n", epoch);
+
+	/* struct_v, struct_cv, map_len, epoch, legacy_client_fscid */
+	ceph_decode_skip_n(&p, end, 2 + sizeof(u32) * 3, bad);
+
+	ceph_decode_32_safe(&p, end, num_fs, bad);
+	while (num_fs-- > 0) {
+		void *info_p, *info_end;
+		u32 info_len;
+		u32 fscid, namelen;
+
+		ceph_decode_need(&p, end, 2 + sizeof(u32), bad);
+		p += 2;		// info_v, info_cv
+		info_len = ceph_decode_32(&p);
+		ceph_decode_need(&p, end, info_len, bad);
+		info_p = p;
+		info_end = p + info_len;
+		p = info_end;
+
+		ceph_decode_need(&info_p, info_end, sizeof(u32) * 2, bad);
+		fscid = ceph_decode_32(&info_p);
+		namelen = ceph_decode_32(&info_p);
+		ceph_decode_need(&info_p, info_end, namelen, bad);
+
+		if (mds_namespace &&
+		    strlen(mds_namespace) == namelen &&
+		    !strncmp(mds_namespace, (char *)info_p, namelen)) {
+			mount_fscid = fscid;
+			break;
+		}
+	}
+
+	ceph_monc_got_map(&fsc->client->monc, CEPH_SUB_FSMAP, epoch);
+	if (mount_fscid != (u32)-1) {
+		fsc->client->monc.fs_cluster_id = mount_fscid;
+		ceph_monc_want_map(&fsc->client->monc, CEPH_SUB_MDSMAP,
+				   0, true);
+		ceph_monc_renew_subs(&fsc->client->monc);
+	} else {
+		err = -ENOENT;
+		goto err_out;
+	}
+	return;
+
+bad:
+	pr_err("error decoding fsmap %d. Shutting down mount.\n", err);
+	ceph_umount_begin(mdsc->fsc->sb);
+	ceph_msg_dump(msg);
+err_out:
+	mutex_lock(&mdsc->mutex);
+	mdsc->mdsmap_err = err;
+	__wake_requests(mdsc, &mdsc->waiting_for_map);
+	mutex_unlock(&mdsc->mutex);
+}
+
+/*
+ * handle mds map update.
+ */
+void ceph_mdsc_handle_mdsmap(struct ceph_mds_client *mdsc, struct ceph_msg *msg)
+{
+	u32 epoch;
+	u32 maplen;
+	void *p = msg->front.iov_base;
+	void *end = p + msg->front.iov_len;
+	struct ceph_mdsmap *newmap, *oldmap;
+	struct ceph_fsid fsid;
+	int err = -EINVAL;
+
+	ceph_decode_need(&p, end, sizeof(fsid)+2*sizeof(u32), bad);
+	ceph_decode_copy(&p, &fsid, sizeof(fsid));
+	if (ceph_check_fsid(mdsc->fsc->client, &fsid) < 0)
+		return;
+	epoch = ceph_decode_32(&p);
+	maplen = ceph_decode_32(&p);
+	dout("handle_map epoch %u len %d\n", epoch, (int)maplen);
+
+	/* do we need it? */
+	mutex_lock(&mdsc->mutex);
+	if (mdsc->mdsmap && epoch <= mdsc->mdsmap->m_epoch) {
+		dout("handle_map epoch %u <= our %u\n",
+		     epoch, mdsc->mdsmap->m_epoch);
+		mutex_unlock(&mdsc->mutex);
+		return;
+	}
+
+	newmap = ceph_mdsmap_decode(&p, end, ceph_msgr2(mdsc->fsc->client));
+	if (IS_ERR(newmap)) {
+		err = PTR_ERR(newmap);
+		goto bad_unlock;
+	}
+
+	/* swap into place */
+	if (mdsc->mdsmap) {
+		oldmap = mdsc->mdsmap;
+		mdsc->mdsmap = newmap;
+		check_new_map(mdsc, newmap, oldmap);
+		ceph_mdsmap_destroy(oldmap);
+	} else {
+		mdsc->mdsmap = newmap;  /* first mds map */
+	}
+	mdsc->fsc->max_file_size = min((loff_t)mdsc->mdsmap->m_max_file_size,
+					MAX_LFS_FILESIZE);
+
+	__wake_requests(mdsc, &mdsc->waiting_for_map);
+	ceph_monc_got_map(&mdsc->fsc->client->monc, CEPH_SUB_MDSMAP,
+			  mdsc->mdsmap->m_epoch);
+
+	mutex_unlock(&mdsc->mutex);
+	schedule_delayed(mdsc, 0);
+	return;
+
+bad_unlock:
+	mutex_unlock(&mdsc->mutex);
+bad:
+	pr_err("error decoding mdsmap %d. Shutting down mount.\n", err);
+	ceph_umount_begin(mdsc->fsc->sb);
+	ceph_msg_dump(msg);
+	return;
+}
+
+static struct ceph_connection *mds_get_con(struct ceph_connection *con)
+{
+	struct ceph_mds_session *s = con->private;
+
+	if (ceph_get_mds_session(s))
+		return con;
+	return NULL;
+}
+
+static void mds_put_con(struct ceph_connection *con)
+{
+	struct ceph_mds_session *s = con->private;
+
+	ceph_put_mds_session(s);
+}
+
+/*
+ * if the client is unresponsive for long enough, the mds will kill
+ * the session entirely.
+ */
+static void mds_peer_reset(struct ceph_connection *con)
+{
+	struct ceph_mds_session *s = con->private;
+	struct ceph_mds_client *mdsc = s->s_mdsc;
+
+	pr_warn("mds%d closed our session\n", s->s_mds);
+	if (READ_ONCE(mdsc->fsc->mount_state) != CEPH_MOUNT_FENCE_IO)
+		send_mds_reconnect(mdsc, s);
+}
+
+static void mds_dispatch(struct ceph_connection *con, struct ceph_msg *msg)
+{
+	struct ceph_mds_session *s = con->private;
+	struct ceph_mds_client *mdsc = s->s_mdsc;
+	int type = le16_to_cpu(msg->hdr.type);
+
+	mutex_lock(&mdsc->mutex);
+	if (__verify_registered_session(mdsc, s) < 0) {
+		mutex_unlock(&mdsc->mutex);
+		goto out;
+	}
+	mutex_unlock(&mdsc->mutex);
+
+	switch (type) {
+	case CEPH_MSG_MDS_MAP:
+		ceph_mdsc_handle_mdsmap(mdsc, msg);
+		break;
+	case CEPH_MSG_FS_MAP_USER:
+		ceph_mdsc_handle_fsmap(mdsc, msg);
+		break;
+	case CEPH_MSG_CLIENT_SESSION:
+		handle_session(s, msg);
+		break;
+	case CEPH_MSG_CLIENT_REPLY:
+		handle_reply(s, msg);
+		break;
+	case CEPH_MSG_CLIENT_REQUEST_FORWARD:
+		handle_forward(mdsc, s, msg);
+		break;
+	case CEPH_MSG_CLIENT_CAPS:
+		ceph_handle_caps(s, msg);
+		break;
+	case CEPH_MSG_CLIENT_SNAP:
+		ceph_handle_snap(mdsc, s, msg);
+		break;
+	case CEPH_MSG_CLIENT_LEASE:
+		handle_lease(mdsc, s, msg);
+		break;
+	case CEPH_MSG_CLIENT_QUOTA:
+		ceph_handle_quota(mdsc, s, msg);
+		break;
+
+	default:
+		pr_err("received unknown message type %d %s\n", type,
+		       ceph_msg_type_name(type));
+	}
+out:
+	ceph_msg_put(msg);
+}
+
+/*
+ * authentication
+ */
+
+/*
+ * Note: returned pointer is the address of a structure that's
+ * managed separately.  Caller must *not* attempt to free it.
+ */
+static struct ceph_auth_handshake *
+mds_get_authorizer(struct ceph_connection *con, int *proto, int force_new)
+{
+	struct ceph_mds_session *s = con->private;
+	struct ceph_mds_client *mdsc = s->s_mdsc;
+	struct ceph_auth_client *ac = mdsc->fsc->client->monc.auth;
+	struct ceph_auth_handshake *auth = &s->s_auth;
+	int ret;
+
+	ret = __ceph_auth_get_authorizer(ac, auth, CEPH_ENTITY_TYPE_MDS,
+					 force_new, proto, NULL, NULL);
+	if (ret)
+		return ERR_PTR(ret);
+
+	return auth;
+}
+
+static int mds_add_authorizer_challenge(struct ceph_connection *con,
+				    void *challenge_buf, int challenge_buf_len)
+{
+	struct ceph_mds_session *s = con->private;
+	struct ceph_mds_client *mdsc = s->s_mdsc;
+	struct ceph_auth_client *ac = mdsc->fsc->client->monc.auth;
+
+	return ceph_auth_add_authorizer_challenge(ac, s->s_auth.authorizer,
+					    challenge_buf, challenge_buf_len);
+}
+
+static int mds_verify_authorizer_reply(struct ceph_connection *con)
+{
+	struct ceph_mds_session *s = con->private;
+	struct ceph_mds_client *mdsc = s->s_mdsc;
+	struct ceph_auth_client *ac = mdsc->fsc->client->monc.auth;
+	struct ceph_auth_handshake *auth = &s->s_auth;
+
+	return ceph_auth_verify_authorizer_reply(ac, auth->authorizer,
+		auth->authorizer_reply_buf, auth->authorizer_reply_buf_len,
+		NULL, NULL, NULL, NULL);
+}
+
+static int mds_invalidate_authorizer(struct ceph_connection *con)
+{
+	struct ceph_mds_session *s = con->private;
+	struct ceph_mds_client *mdsc = s->s_mdsc;
+	struct ceph_auth_client *ac = mdsc->fsc->client->monc.auth;
+
+	ceph_auth_invalidate_authorizer(ac, CEPH_ENTITY_TYPE_MDS);
+
+	return ceph_monc_validate_auth(&mdsc->fsc->client->monc);
+}
+
+static int mds_get_auth_request(struct ceph_connection *con,
+				void *buf, int *buf_len,
+				void **authorizer, int *authorizer_len)
+{
+	struct ceph_mds_session *s = con->private;
+	struct ceph_auth_client *ac = s->s_mdsc->fsc->client->monc.auth;
+	struct ceph_auth_handshake *auth = &s->s_auth;
+	int ret;
+
+	ret = ceph_auth_get_authorizer(ac, auth, CEPH_ENTITY_TYPE_MDS,
+				       buf, buf_len);
+	if (ret)
+		return ret;
+
+	*authorizer = auth->authorizer_buf;
+	*authorizer_len = auth->authorizer_buf_len;
+	return 0;
+}
+
+static int mds_handle_auth_reply_more(struct ceph_connection *con,
+				      void *reply, int reply_len,
+				      void *buf, int *buf_len,
+				      void **authorizer, int *authorizer_len)
+{
+	struct ceph_mds_session *s = con->private;
+	struct ceph_auth_client *ac = s->s_mdsc->fsc->client->monc.auth;
+	struct ceph_auth_handshake *auth = &s->s_auth;
+	int ret;
+
+	ret = ceph_auth_handle_svc_reply_more(ac, auth, reply, reply_len,
+					      buf, buf_len);
+	if (ret)
+		return ret;
+
+	*authorizer = auth->authorizer_buf;
+	*authorizer_len = auth->authorizer_buf_len;
+	return 0;
+}
+
+static int mds_handle_auth_done(struct ceph_connection *con,
+				u64 global_id, void *reply, int reply_len,
+				u8 *session_key, int *session_key_len,
+				u8 *con_secret, int *con_secret_len)
+{
+	struct ceph_mds_session *s = con->private;
+	struct ceph_auth_client *ac = s->s_mdsc->fsc->client->monc.auth;
+	struct ceph_auth_handshake *auth = &s->s_auth;
+
+	return ceph_auth_handle_svc_reply_done(ac, auth, reply, reply_len,
+					       session_key, session_key_len,
+					       con_secret, con_secret_len);
+}
+
+static int mds_handle_auth_bad_method(struct ceph_connection *con,
+				      int used_proto, int result,
+				      const int *allowed_protos, int proto_cnt,
+				      const int *allowed_modes, int mode_cnt)
+{
+	struct ceph_mds_session *s = con->private;
+	struct ceph_mon_client *monc = &s->s_mdsc->fsc->client->monc;
+	int ret;
+
+	if (ceph_auth_handle_bad_authorizer(monc->auth, CEPH_ENTITY_TYPE_MDS,
+					    used_proto, result,
+					    allowed_protos, proto_cnt,
+					    allowed_modes, mode_cnt)) {
+		ret = ceph_monc_validate_auth(monc);
+		if (ret)
+			return ret;
+	}
+
+	return -EACCES;
+}
+
+static struct ceph_msg *mds_alloc_msg(struct ceph_connection *con,
+				struct ceph_msg_header *hdr, int *skip)
+{
+	struct ceph_msg *msg;
+	int type = (int) le16_to_cpu(hdr->type);
+	int front_len = (int) le32_to_cpu(hdr->front_len);
+
+	if (con->in_msg)
+		return con->in_msg;
+
+	*skip = 0;
+	msg = ceph_msg_new(type, front_len, GFP_NOFS, false);
+	if (!msg) {
+		pr_err("unable to allocate msg type %d len %d\n",
+		       type, front_len);
+		return NULL;
+	}
+
+	return msg;
+}
+
+static int mds_sign_message(struct ceph_msg *msg)
+{
+       struct ceph_mds_session *s = msg->con->private;
+       struct ceph_auth_handshake *auth = &s->s_auth;
+
+       return ceph_auth_sign_message(auth, msg);
+}
+
+static int mds_check_message_signature(struct ceph_msg *msg)
+{
+       struct ceph_mds_session *s = msg->con->private;
+       struct ceph_auth_handshake *auth = &s->s_auth;
+
+       return ceph_auth_check_message_signature(auth, msg);
+}
+
+static const struct ceph_connection_operations mds_con_ops = {
+	.get = mds_get_con,
+	.put = mds_put_con,
+	.alloc_msg = mds_alloc_msg,
+	.dispatch = mds_dispatch,
+	.peer_reset = mds_peer_reset,
+	.get_authorizer = mds_get_authorizer,
+	.add_authorizer_challenge = mds_add_authorizer_challenge,
+	.verify_authorizer_reply = mds_verify_authorizer_reply,
+	.invalidate_authorizer = mds_invalidate_authorizer,
+	.sign_message = mds_sign_message,
+	.check_message_signature = mds_check_message_signature,
+	.get_auth_request = mds_get_auth_request,
+	.handle_auth_reply_more = mds_handle_auth_reply_more,
+	.handle_auth_done = mds_handle_auth_done,
+	.handle_auth_bad_method = mds_handle_auth_bad_method,
+};
+
+/* eof */
diff --git a/fs/ceph/mds_client.h b/fs/ceph/mds_client.h
new file mode 100644
index 000000000..0913959cc
--- /dev/null
+++ b/fs/ceph/mds_client.h
@@ -0,0 +1,597 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _FS_CEPH_MDS_CLIENT_H
+#define _FS_CEPH_MDS_CLIENT_H
+
+#include <linux/completion.h>
+#include <linux/kref.h>
+#include <linux/list.h>
+#include <linux/mutex.h>
+#include <linux/rbtree.h>
+#include <linux/spinlock.h>
+#include <linux/refcount.h>
+#include <linux/utsname.h>
+#include <linux/ktime.h>
+
+#include <linux/ceph/types.h>
+#include <linux/ceph/messenger.h>
+#include <linux/ceph/mdsmap.h>
+#include <linux/ceph/auth.h>
+
+#include "metric.h"
+#include "super.h"
+
+/* The first 8 bits are reserved for old ceph releases */
+enum ceph_feature_type {
+	CEPHFS_FEATURE_MIMIC = 8,
+	CEPHFS_FEATURE_REPLY_ENCODING,
+	CEPHFS_FEATURE_RECLAIM_CLIENT,
+	CEPHFS_FEATURE_LAZY_CAP_WANTED,
+	CEPHFS_FEATURE_MULTI_RECONNECT,
+	CEPHFS_FEATURE_DELEG_INO,
+	CEPHFS_FEATURE_METRIC_COLLECT,
+	CEPHFS_FEATURE_ALTERNATE_NAME,
+	CEPHFS_FEATURE_NOTIFY_SESSION_STATE,
+	CEPHFS_FEATURE_OP_GETVXATTR,
+
+	CEPHFS_FEATURE_MAX = CEPHFS_FEATURE_OP_GETVXATTR,
+};
+
+#define CEPHFS_FEATURES_CLIENT_SUPPORTED {	\
+	0, 1, 2, 3, 4, 5, 6, 7,			\
+	CEPHFS_FEATURE_MIMIC,			\
+	CEPHFS_FEATURE_REPLY_ENCODING,		\
+	CEPHFS_FEATURE_LAZY_CAP_WANTED,		\
+	CEPHFS_FEATURE_MULTI_RECONNECT,		\
+	CEPHFS_FEATURE_DELEG_INO,		\
+	CEPHFS_FEATURE_METRIC_COLLECT,		\
+	CEPHFS_FEATURE_NOTIFY_SESSION_STATE,	\
+	CEPHFS_FEATURE_OP_GETVXATTR,		\
+}
+
+/*
+ * Some lock dependencies:
+ *
+ * session->s_mutex
+ *         mdsc->mutex
+ *
+ *         mdsc->snap_rwsem
+ *
+ *         ci->i_ceph_lock
+ *                 mdsc->snap_flush_lock
+ *                 mdsc->cap_delay_lock
+ *
+ */
+
+struct ceph_fs_client;
+struct ceph_cap;
+
+/*
+ * parsed info about a single inode.  pointers are into the encoded
+ * on-wire structures within the mds reply message payload.
+ */
+struct ceph_mds_reply_info_in {
+	struct ceph_mds_reply_inode *in;
+	struct ceph_dir_layout dir_layout;
+	u32 symlink_len;
+	char *symlink;
+	u32 xattr_len;
+	char *xattr_data;
+	u64 inline_version;
+	u32 inline_len;
+	char *inline_data;
+	u32 pool_ns_len;
+	char *pool_ns_data;
+	u64 max_bytes;
+	u64 max_files;
+	s32 dir_pin;
+	struct ceph_timespec btime;
+	struct ceph_timespec snap_btime;
+	u64 rsnaps;
+	u64 change_attr;
+};
+
+struct ceph_mds_reply_dir_entry {
+	char                          *name;
+	u32                           name_len;
+	struct ceph_mds_reply_lease   *lease;
+	struct ceph_mds_reply_info_in inode;
+	loff_t			      offset;
+};
+
+struct ceph_mds_reply_xattr {
+	char *xattr_value;
+	size_t xattr_value_len;
+};
+
+/*
+ * parsed info about an mds reply, including information about
+ * either: 1) the target inode and/or its parent directory and dentry,
+ * and directory contents (for readdir results), or
+ * 2) the file range lock info (for fcntl F_GETLK results).
+ */
+struct ceph_mds_reply_info_parsed {
+	struct ceph_mds_reply_head    *head;
+
+	/* trace */
+	struct ceph_mds_reply_info_in diri, targeti;
+	struct ceph_mds_reply_dirfrag *dirfrag;
+	char                          *dname;
+	u32                           dname_len;
+	struct ceph_mds_reply_lease   *dlease;
+	struct ceph_mds_reply_xattr   xattr_info;
+
+	/* extra */
+	union {
+		/* for fcntl F_GETLK results */
+		struct ceph_filelock *filelock_reply;
+
+		/* for readdir results */
+		struct {
+			struct ceph_mds_reply_dirfrag *dir_dir;
+			size_t			      dir_buf_size;
+			int                           dir_nr;
+			bool			      dir_end;
+			bool			      dir_complete;
+			bool			      hash_order;
+			bool			      offset_hash;
+			struct ceph_mds_reply_dir_entry  *dir_entries;
+		};
+
+		/* for create results */
+		struct {
+			bool has_create_ino;
+			u64 ino;
+		};
+	};
+
+	/* encoded blob describing snapshot contexts for certain
+	   operations (e.g., open) */
+	void *snapblob;
+	int snapblob_len;
+};
+
+
+/*
+ * cap releases are batched and sent to the MDS en masse.
+ *
+ * Account for per-message overhead of mds_cap_release header
+ * and __le32 for osd epoch barrier trailing field.
+ */
+#define CEPH_CAPS_PER_RELEASE ((PAGE_SIZE - sizeof(u32) -		\
+				sizeof(struct ceph_mds_cap_release)) /	\
+			        sizeof(struct ceph_mds_cap_item))
+
+
+/*
+ * state associated with each MDS<->client session
+ */
+enum {
+	CEPH_MDS_SESSION_NEW = 1,
+	CEPH_MDS_SESSION_OPENING = 2,
+	CEPH_MDS_SESSION_OPEN = 3,
+	CEPH_MDS_SESSION_HUNG = 4,
+	CEPH_MDS_SESSION_RESTARTING = 5,
+	CEPH_MDS_SESSION_RECONNECTING = 6,
+	CEPH_MDS_SESSION_CLOSING = 7,
+	CEPH_MDS_SESSION_CLOSED = 8,
+	CEPH_MDS_SESSION_REJECTED = 9,
+};
+
+struct ceph_mds_session {
+	struct ceph_mds_client *s_mdsc;
+	int               s_mds;
+	int               s_state;
+	unsigned long     s_ttl;      /* time until mds kills us */
+	unsigned long	  s_features;
+	u64               s_seq;      /* incoming msg seq # */
+	struct mutex      s_mutex;    /* serialize session messages */
+
+	struct ceph_connection s_con;
+
+	struct ceph_auth_handshake s_auth;
+
+	atomic_t          s_cap_gen;  /* inc each time we get mds stale msg */
+	unsigned long     s_cap_ttl;  /* when session caps expire. protected by s_mutex */
+
+	/* protected by s_cap_lock */
+	spinlock_t        s_cap_lock;
+	refcount_t        s_ref;
+	struct list_head  s_caps;     /* all caps issued by this session */
+	struct ceph_cap  *s_cap_iterator;
+	int               s_nr_caps;
+	int               s_num_cap_releases;
+	int		  s_cap_reconnect;
+	int		  s_readonly;
+	struct list_head  s_cap_releases; /* waiting cap_release messages */
+	struct work_struct s_cap_release_work;
+
+	/* See ceph_inode_info->i_dirty_item. */
+	struct list_head  s_cap_dirty;	      /* inodes w/ dirty caps */
+
+	/* See ceph_inode_info->i_flushing_item. */
+	struct list_head  s_cap_flushing;     /* inodes w/ flushing caps */
+
+	unsigned long     s_renew_requested; /* last time we sent a renew req */
+	u64               s_renew_seq;
+
+	struct list_head  s_waiting;  /* waiting requests */
+	struct list_head  s_unsafe;   /* unsafe requests */
+	struct xarray	  s_delegated_inos;
+};
+
+/*
+ * modes of choosing which MDS to send a request to
+ */
+enum {
+	USE_ANY_MDS,
+	USE_RANDOM_MDS,
+	USE_AUTH_MDS,   /* prefer authoritative mds for this metadata item */
+};
+
+struct ceph_mds_request;
+struct ceph_mds_client;
+
+/*
+ * request completion callback
+ */
+typedef void (*ceph_mds_request_callback_t) (struct ceph_mds_client *mdsc,
+					     struct ceph_mds_request *req);
+/*
+ * wait for request completion callback
+ */
+typedef int (*ceph_mds_request_wait_callback_t) (struct ceph_mds_client *mdsc,
+						 struct ceph_mds_request *req);
+
+/*
+ * an in-flight mds request
+ */
+struct ceph_mds_request {
+	u64 r_tid;                   /* transaction id */
+	struct rb_node r_node;
+	struct ceph_mds_client *r_mdsc;
+
+	struct kref       r_kref;
+	int r_op;                    /* mds op code */
+
+	/* operation on what? */
+	struct inode *r_inode;              /* arg1 */
+	struct dentry *r_dentry;            /* arg1 */
+	struct dentry *r_old_dentry;        /* arg2: rename from or link from */
+	struct inode *r_old_dentry_dir;     /* arg2: old dentry's parent dir */
+	char *r_path1, *r_path2;
+	struct ceph_vino r_ino1, r_ino2;
+
+	struct inode *r_parent;		    /* parent dir inode */
+	struct inode *r_target_inode;       /* resulting inode */
+
+#define CEPH_MDS_R_DIRECT_IS_HASH	(1) /* r_direct_hash is valid */
+#define CEPH_MDS_R_ABORTED		(2) /* call was aborted */
+#define CEPH_MDS_R_GOT_UNSAFE		(3) /* got an unsafe reply */
+#define CEPH_MDS_R_GOT_SAFE		(4) /* got a safe reply */
+#define CEPH_MDS_R_GOT_RESULT		(5) /* got a result */
+#define CEPH_MDS_R_DID_PREPOPULATE	(6) /* prepopulated readdir */
+#define CEPH_MDS_R_PARENT_LOCKED	(7) /* is r_parent->i_rwsem wlocked? */
+#define CEPH_MDS_R_ASYNC		(8) /* async request */
+	unsigned long	r_req_flags;
+
+	struct mutex r_fill_mutex;
+
+	union ceph_mds_request_args r_args;
+	int r_fmode;        /* file mode, if expecting cap */
+	int r_request_release_offset;
+	const struct cred *r_cred;
+	struct timespec64 r_stamp;
+
+	/* for choosing which mds to send this request to */
+	int r_direct_mode;
+	u32 r_direct_hash;      /* choose dir frag based on this dentry hash */
+
+	/* data payload is used for xattr ops */
+	struct ceph_pagelist *r_pagelist;
+
+	/* what caps shall we drop? */
+	int r_inode_drop, r_inode_unless;
+	int r_dentry_drop, r_dentry_unless;
+	int r_old_dentry_drop, r_old_dentry_unless;
+	struct inode *r_old_inode;
+	int r_old_inode_drop, r_old_inode_unless;
+
+	struct ceph_msg  *r_request;  /* original request */
+	struct ceph_msg  *r_reply;
+	struct ceph_mds_reply_info_parsed r_reply_info;
+	int r_err;
+	u32               r_readdir_offset;
+
+	struct page *r_locked_page;
+	int r_dir_caps;
+	int r_num_caps;
+
+	unsigned long r_timeout;  /* optional.  jiffies, 0 is "wait forever" */
+	unsigned long r_started;  /* start time to measure timeout against */
+	unsigned long r_start_latency;  /* start time to measure latency */
+	unsigned long r_end_latency;    /* finish time to measure latency */
+	unsigned long r_request_started; /* start time for mds request only,
+					    used to measure lease durations */
+
+	/* link unsafe requests to parent directory, for fsync */
+	struct inode	*r_unsafe_dir;
+	struct list_head r_unsafe_dir_item;
+
+	/* unsafe requests that modify the target inode */
+	struct list_head r_unsafe_target_item;
+
+	struct ceph_mds_session *r_session;
+
+	int               r_attempts;   /* resend attempts */
+	int               r_num_fwd;    /* number of forward attempts */
+	int               r_resend_mds; /* mds to resend to next, if any*/
+	u32               r_sent_on_mseq; /* cap mseq request was sent at*/
+	u64		  r_deleg_ino;
+
+	struct list_head  r_wait;
+	struct completion r_completion;
+	struct completion r_safe_completion;
+	ceph_mds_request_callback_t r_callback;
+	struct list_head  r_unsafe_item;  /* per-session unsafe list item */
+
+	long long	  r_dir_release_cnt;
+	long long	  r_dir_ordered_cnt;
+	int		  r_readdir_cache_idx;
+
+	int		  r_feature_needed;
+
+	struct ceph_cap_reservation r_caps_reservation;
+};
+
+struct ceph_pool_perm {
+	struct rb_node node;
+	int perm;
+	s64 pool;
+	size_t pool_ns_len;
+	char pool_ns[];
+};
+
+struct ceph_snapid_map {
+	struct rb_node node;
+	struct list_head lru;
+	atomic_t ref;
+	u64 snap;
+	dev_t dev;
+	unsigned long last_used;
+};
+
+/*
+ * node for list of quotarealm inodes that are not visible from the filesystem
+ * mountpoint, but required to handle, e.g. quotas.
+ */
+struct ceph_quotarealm_inode {
+	struct rb_node node;
+	u64 ino;
+	unsigned long timeout; /* last time a lookup failed for this inode */
+	struct mutex mutex;
+	struct inode *inode;
+};
+
+struct cap_wait {
+	struct list_head	list;
+	u64			ino;
+	pid_t			tgid;
+	int			need;
+	int			want;
+};
+
+enum {
+	CEPH_MDSC_STOPPING_BEGIN = 1,
+	CEPH_MDSC_STOPPING_FLUSHING = 2,
+	CEPH_MDSC_STOPPING_FLUSHED = 3,
+};
+
+/*
+ * mds client state
+ */
+struct ceph_mds_client {
+	struct ceph_fs_client  *fsc;
+	struct mutex            mutex;         /* all nested structures */
+
+	struct ceph_mdsmap      *mdsmap;
+	struct completion       safe_umount_waiters;
+	wait_queue_head_t       session_close_wq;
+	struct list_head        waiting_for_map;
+	int 			mdsmap_err;
+
+	struct ceph_mds_session **sessions;    /* NULL for mds if no session */
+	atomic_t		num_sessions;
+	int                     max_sessions;  /* len of sessions array */
+
+	spinlock_t              stopping_lock;  /* protect snap_empty */
+	int                     stopping;      /* the stage of shutting down */
+	atomic_t                stopping_blockers;
+	struct completion	stopping_waiter;
+
+	atomic64_t		quotarealms_count; /* # realms with quota */
+	/*
+	 * We keep a list of inodes we don't see in the mountpoint but that we
+	 * need to track quota realms.
+	 */
+	struct rb_root		quotarealms_inodes;
+	struct mutex		quotarealms_inodes_mutex;
+
+	/*
+	 * snap_rwsem will cover cap linkage into snaprealms, and
+	 * realm snap contexts.  (later, we can do per-realm snap
+	 * contexts locks..)  the empty list contains realms with no
+	 * references (implying they contain no inodes with caps) that
+	 * should be destroyed.
+	 */
+	u64			last_snap_seq;
+	struct rw_semaphore     snap_rwsem;
+	struct rb_root          snap_realms;
+	struct list_head        snap_empty;
+	int			num_snap_realms;
+	spinlock_t              snap_empty_lock;  /* protect snap_empty */
+
+	u64                    last_tid;      /* most recent mds request */
+	u64                    oldest_tid;    /* oldest incomplete mds request,
+						 excluding setfilelock requests */
+	struct rb_root         request_tree;  /* pending mds requests */
+	struct delayed_work    delayed_work;  /* delayed work */
+	unsigned long    last_renew_caps;  /* last time we renewed our caps */
+	struct list_head cap_delay_list;   /* caps with delayed release */
+	spinlock_t       cap_delay_lock;   /* protects cap_delay_list */
+	struct list_head snap_flush_list;  /* cap_snaps ready to flush */
+	spinlock_t       snap_flush_lock;
+
+	u64               last_cap_flush_tid;
+	struct list_head  cap_flush_list;
+	struct list_head  cap_dirty_migrating; /* ...that are migration... */
+	int               num_cap_flushing; /* # caps we are flushing */
+	spinlock_t        cap_dirty_lock;   /* protects above items */
+	wait_queue_head_t cap_flushing_wq;
+
+	struct work_struct cap_reclaim_work;
+	atomic_t	   cap_reclaim_pending;
+
+	/*
+	 * Cap reservations
+	 *
+	 * Maintain a global pool of preallocated struct ceph_caps, referenced
+	 * by struct ceph_caps_reservations.  This ensures that we preallocate
+	 * memory needed to successfully process an MDS response.  (If an MDS
+	 * sends us cap information and we fail to process it, we will have
+	 * problems due to the client and MDS being out of sync.)
+	 *
+	 * Reservations are 'owned' by a ceph_cap_reservation context.
+	 */
+	spinlock_t	caps_list_lock;
+	struct		list_head caps_list; /* unused (reserved or
+						unreserved) */
+	struct		list_head cap_wait_list;
+	int		caps_total_count;    /* total caps allocated */
+	int		caps_use_count;      /* in use */
+	int		caps_use_max;	     /* max used caps */
+	int		caps_reserve_count;  /* unused, reserved */
+	int		caps_avail_count;    /* unused, unreserved */
+	int		caps_min_count;      /* keep at least this many
+						(unreserved) */
+	spinlock_t	  dentry_list_lock;
+	struct list_head  dentry_leases;     /* fifo list */
+	struct list_head  dentry_dir_leases; /* lru list */
+
+	struct ceph_client_metric metric;
+
+	spinlock_t		snapid_map_lock;
+	struct rb_root		snapid_map_tree;
+	struct list_head	snapid_map_lru;
+
+	struct rw_semaphore     pool_perm_rwsem;
+	struct rb_root		pool_perm_tree;
+
+	char nodename[__NEW_UTS_LEN + 1];
+};
+
+extern const char *ceph_mds_op_name(int op);
+
+extern bool check_session_state(struct ceph_mds_session *s);
+void inc_session_sequence(struct ceph_mds_session *s);
+
+extern struct ceph_mds_session *
+__ceph_lookup_mds_session(struct ceph_mds_client *, int mds);
+
+extern const char *ceph_session_state_name(int s);
+
+extern struct ceph_mds_session *
+ceph_get_mds_session(struct ceph_mds_session *s);
+extern void ceph_put_mds_session(struct ceph_mds_session *s);
+
+extern int ceph_send_msg_mds(struct ceph_mds_client *mdsc,
+			     struct ceph_msg *msg, int mds);
+
+extern int ceph_mdsc_init(struct ceph_fs_client *fsc);
+extern void ceph_mdsc_close_sessions(struct ceph_mds_client *mdsc);
+extern void ceph_mdsc_force_umount(struct ceph_mds_client *mdsc);
+extern void ceph_mdsc_destroy(struct ceph_fs_client *fsc);
+
+extern void ceph_mdsc_sync(struct ceph_mds_client *mdsc);
+
+extern void ceph_invalidate_dir_request(struct ceph_mds_request *req);
+extern int ceph_alloc_readdir_reply_buffer(struct ceph_mds_request *req,
+					   struct inode *dir);
+extern struct ceph_mds_request *
+ceph_mdsc_create_request(struct ceph_mds_client *mdsc, int op, int mode);
+extern int ceph_mdsc_submit_request(struct ceph_mds_client *mdsc,
+				    struct inode *dir,
+				    struct ceph_mds_request *req);
+int ceph_mdsc_wait_request(struct ceph_mds_client *mdsc,
+			struct ceph_mds_request *req,
+			ceph_mds_request_wait_callback_t wait_func);
+extern int ceph_mdsc_do_request(struct ceph_mds_client *mdsc,
+				struct inode *dir,
+				struct ceph_mds_request *req);
+extern void ceph_mdsc_release_dir_caps(struct ceph_mds_request *req);
+extern void ceph_mdsc_release_dir_caps_no_check(struct ceph_mds_request *req);
+static inline void ceph_mdsc_get_request(struct ceph_mds_request *req)
+{
+	kref_get(&req->r_kref);
+}
+extern void ceph_mdsc_release_request(struct kref *kref);
+static inline void ceph_mdsc_put_request(struct ceph_mds_request *req)
+{
+	kref_put(&req->r_kref, ceph_mdsc_release_request);
+}
+
+extern void send_flush_mdlog(struct ceph_mds_session *s);
+extern void ceph_mdsc_iterate_sessions(struct ceph_mds_client *mdsc,
+				       void (*cb)(struct ceph_mds_session *),
+				       bool check_state);
+extern struct ceph_msg *ceph_create_session_msg(u32 op, u64 seq);
+extern void __ceph_queue_cap_release(struct ceph_mds_session *session,
+				    struct ceph_cap *cap);
+extern void ceph_flush_cap_releases(struct ceph_mds_client *mdsc,
+				    struct ceph_mds_session *session);
+extern void ceph_queue_cap_reclaim_work(struct ceph_mds_client *mdsc);
+extern void ceph_reclaim_caps_nr(struct ceph_mds_client *mdsc, int nr);
+extern int ceph_iterate_session_caps(struct ceph_mds_session *session,
+				     int (*cb)(struct inode *, int mds, void *),
+				     void *arg);
+extern void ceph_mdsc_pre_umount(struct ceph_mds_client *mdsc);
+
+static inline void ceph_mdsc_free_path(char *path, int len)
+{
+	if (!IS_ERR_OR_NULL(path))
+		__putname(path - (PATH_MAX - 1 - len));
+}
+
+extern char *ceph_mdsc_build_path(struct dentry *dentry, int *plen, u64 *base,
+				  int stop_on_nosnap);
+
+extern void __ceph_mdsc_drop_dentry_lease(struct dentry *dentry);
+extern void ceph_mdsc_lease_send_msg(struct ceph_mds_session *session,
+				     struct dentry *dentry, char action,
+				     u32 seq);
+
+extern void ceph_mdsc_handle_mdsmap(struct ceph_mds_client *mdsc,
+				    struct ceph_msg *msg);
+extern void ceph_mdsc_handle_fsmap(struct ceph_mds_client *mdsc,
+				   struct ceph_msg *msg);
+
+extern struct ceph_mds_session *
+ceph_mdsc_open_export_target_session(struct ceph_mds_client *mdsc, int target);
+extern void ceph_mdsc_open_export_target_sessions(struct ceph_mds_client *mdsc,
+					  struct ceph_mds_session *session);
+
+extern int ceph_trim_caps(struct ceph_mds_client *mdsc,
+			  struct ceph_mds_session *session,
+			  int max_caps);
+
+static inline int ceph_wait_on_async_create(struct inode *inode)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+
+	return wait_on_bit(&ci->i_ceph_flags, CEPH_ASYNC_CREATE_BIT,
+			   TASK_KILLABLE);
+}
+
+extern int ceph_wait_on_conflict_unlink(struct dentry *dentry);
+extern u64 ceph_get_deleg_ino(struct ceph_mds_session *session);
+extern int ceph_restore_deleg_ino(struct ceph_mds_session *session, u64 ino);
+#endif
diff --git a/fs/ceph/mdsmap.c b/fs/ceph/mdsmap.c
new file mode 100644
index 000000000..3fbabc98e
--- /dev/null
+++ b/fs/ceph/mdsmap.c
@@ -0,0 +1,436 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/ceph/ceph_debug.h>
+
+#include <linux/bug.h>
+#include <linux/err.h>
+#include <linux/random.h>
+#include <linux/slab.h>
+#include <linux/types.h>
+
+#include <linux/ceph/mdsmap.h>
+#include <linux/ceph/messenger.h>
+#include <linux/ceph/decode.h>
+
+#include "super.h"
+
+#define CEPH_MDS_IS_READY(i, ignore_laggy) \
+	(m->m_info[i].state > 0 && ignore_laggy ? true : !m->m_info[i].laggy)
+
+static int __mdsmap_get_random_mds(struct ceph_mdsmap *m, bool ignore_laggy)
+{
+	int n = 0;
+	int i, j;
+
+	/* count */
+	for (i = 0; i < m->possible_max_rank; i++)
+		if (CEPH_MDS_IS_READY(i, ignore_laggy))
+			n++;
+	if (n == 0)
+		return -1;
+
+	/* pick */
+	n = prandom_u32_max(n);
+	for (j = 0, i = 0; i < m->possible_max_rank; i++) {
+		if (CEPH_MDS_IS_READY(i, ignore_laggy))
+			j++;
+		if (j > n)
+			break;
+	}
+
+	return i;
+}
+
+/*
+ * choose a random mds that is "up" (i.e. has a state > 0), or -1.
+ */
+int ceph_mdsmap_get_random_mds(struct ceph_mdsmap *m)
+{
+	int mds;
+
+	mds = __mdsmap_get_random_mds(m, false);
+	if (mds == m->possible_max_rank || mds == -1)
+		mds = __mdsmap_get_random_mds(m, true);
+
+	return mds == m->possible_max_rank ? -1 : mds;
+}
+
+#define __decode_and_drop_type(p, end, type, bad)		\
+	do {							\
+		if (*p + sizeof(type) > end)			\
+			goto bad;				\
+		*p += sizeof(type);				\
+	} while (0)
+
+#define __decode_and_drop_set(p, end, type, bad)		\
+	do {							\
+		u32 n;						\
+		size_t need;					\
+		ceph_decode_32_safe(p, end, n, bad);		\
+		need = sizeof(type) * n;			\
+		ceph_decode_need(p, end, need, bad);		\
+		*p += need;					\
+	} while (0)
+
+#define __decode_and_drop_map(p, end, ktype, vtype, bad)	\
+	do {							\
+		u32 n;						\
+		size_t need;					\
+		ceph_decode_32_safe(p, end, n, bad);		\
+		need = (sizeof(ktype) + sizeof(vtype)) * n;	\
+		ceph_decode_need(p, end, need, bad);		\
+		*p += need;					\
+	} while (0)
+
+
+static int __decode_and_drop_compat_set(void **p, void* end)
+{
+	int i;
+	/* compat, ro_compat, incompat*/
+	for (i = 0; i < 3; i++) {
+		u32 n;
+		ceph_decode_need(p, end, sizeof(u64) + sizeof(u32), bad);
+		/* mask */
+		*p += sizeof(u64);
+		/* names (map<u64, string>) */
+		n = ceph_decode_32(p);
+		while (n-- > 0) {
+			u32 len;
+			ceph_decode_need(p, end, sizeof(u64) + sizeof(u32),
+					 bad);
+			*p += sizeof(u64);
+			len = ceph_decode_32(p);
+			ceph_decode_need(p, end, len, bad);
+			*p += len;
+		}
+	}
+	return 0;
+bad:
+	return -1;
+}
+
+/*
+ * Decode an MDS map
+ *
+ * Ignore any fields we don't care about (there are quite a few of
+ * them).
+ */
+struct ceph_mdsmap *ceph_mdsmap_decode(void **p, void *end, bool msgr2)
+{
+	struct ceph_mdsmap *m;
+	const void *start = *p;
+	int i, j, n;
+	int err;
+	u8 mdsmap_v;
+	u16 mdsmap_ev;
+	u32 target;
+
+	m = kzalloc(sizeof(*m), GFP_NOFS);
+	if (!m)
+		return ERR_PTR(-ENOMEM);
+
+	ceph_decode_need(p, end, 1 + 1, bad);
+	mdsmap_v = ceph_decode_8(p);
+	*p += sizeof(u8);			/* mdsmap_cv */
+	if (mdsmap_v >= 4) {
+	       u32 mdsmap_len;
+	       ceph_decode_32_safe(p, end, mdsmap_len, bad);
+	       if (end < *p + mdsmap_len)
+		       goto bad;
+	       end = *p + mdsmap_len;
+	}
+
+	ceph_decode_need(p, end, 8*sizeof(u32) + sizeof(u64), bad);
+	m->m_epoch = ceph_decode_32(p);
+	m->m_client_epoch = ceph_decode_32(p);
+	m->m_last_failure = ceph_decode_32(p);
+	m->m_root = ceph_decode_32(p);
+	m->m_session_timeout = ceph_decode_32(p);
+	m->m_session_autoclose = ceph_decode_32(p);
+	m->m_max_file_size = ceph_decode_64(p);
+	m->m_max_mds = ceph_decode_32(p);
+
+	/*
+	 * pick out the active nodes as the m_num_active_mds, the
+	 * m_num_active_mds maybe larger than m_max_mds when decreasing
+	 * the max_mds in cluster side, in other case it should less
+	 * than or equal to m_max_mds.
+	 */
+	m->m_num_active_mds = n = ceph_decode_32(p);
+
+	/*
+	 * the possible max rank, it maybe larger than the m_num_active_mds,
+	 * for example if the mds_max == 2 in the cluster, when the MDS(0)
+	 * was laggy and being replaced by a new MDS, we will temporarily
+	 * receive a new mds map with n_num_mds == 1 and the active MDS(1),
+	 * and the mds rank >= m_num_active_mds.
+	 */
+	m->possible_max_rank = max(m->m_num_active_mds, m->m_max_mds);
+
+	m->m_info = kcalloc(m->possible_max_rank, sizeof(*m->m_info), GFP_NOFS);
+	if (!m->m_info)
+		goto nomem;
+
+	/* pick out active nodes from mds_info (state > 0) */
+	for (i = 0; i < n; i++) {
+		u64 global_id;
+		u32 namelen;
+		s32 mds, inc, state;
+		u8 info_v;
+		void *info_end = NULL;
+		struct ceph_entity_addr addr;
+		u32 num_export_targets;
+		void *pexport_targets = NULL;
+		struct ceph_timespec laggy_since;
+		struct ceph_mds_info *info;
+		bool laggy;
+
+		ceph_decode_need(p, end, sizeof(u64) + 1, bad);
+		global_id = ceph_decode_64(p);
+		info_v= ceph_decode_8(p);
+		if (info_v >= 4) {
+			u32 info_len;
+			ceph_decode_need(p, end, 1 + sizeof(u32), bad);
+			*p += sizeof(u8);	/* info_cv */
+			info_len = ceph_decode_32(p);
+			info_end = *p + info_len;
+			if (info_end > end)
+				goto bad;
+		}
+
+		ceph_decode_need(p, end, sizeof(u64) + sizeof(u32), bad);
+		*p += sizeof(u64);
+		namelen = ceph_decode_32(p);  /* skip mds name */
+		*p += namelen;
+
+		ceph_decode_32_safe(p, end, mds, bad);
+		ceph_decode_32_safe(p, end, inc, bad);
+		ceph_decode_32_safe(p, end, state, bad);
+		*p += sizeof(u64);		/* state_seq */
+		if (info_v >= 8)
+			err = ceph_decode_entity_addrvec(p, end, msgr2, &addr);
+		else
+			err = ceph_decode_entity_addr(p, end, &addr);
+		if (err)
+			goto corrupt;
+
+		ceph_decode_copy_safe(p, end, &laggy_since, sizeof(laggy_since),
+				      bad);
+		laggy = laggy_since.tv_sec != 0 || laggy_since.tv_nsec != 0;
+		*p += sizeof(u32);
+		ceph_decode_32_safe(p, end, namelen, bad);
+		*p += namelen;
+		if (info_v >= 2) {
+			ceph_decode_32_safe(p, end, num_export_targets, bad);
+			pexport_targets = *p;
+			*p += num_export_targets * sizeof(u32);
+		} else {
+			num_export_targets = 0;
+		}
+
+		if (info_end && *p != info_end) {
+			if (*p > info_end)
+				goto bad;
+			*p = info_end;
+		}
+
+		dout("mdsmap_decode %d/%d %lld mds%d.%d %s %s%s\n",
+		     i+1, n, global_id, mds, inc,
+		     ceph_pr_addr(&addr),
+		     ceph_mds_state_name(state),
+		     laggy ? "(laggy)" : "");
+
+		if (mds < 0 || mds >= m->possible_max_rank) {
+			pr_warn("mdsmap_decode got incorrect mds(%d)\n", mds);
+			continue;
+		}
+
+		if (state <= 0) {
+			dout("mdsmap_decode got incorrect state(%s)\n",
+			     ceph_mds_state_name(state));
+			continue;
+		}
+
+		info = &m->m_info[mds];
+		info->global_id = global_id;
+		info->state = state;
+		info->addr = addr;
+		info->laggy = laggy;
+		info->num_export_targets = num_export_targets;
+		if (num_export_targets) {
+			info->export_targets = kcalloc(num_export_targets,
+						       sizeof(u32), GFP_NOFS);
+			if (!info->export_targets)
+				goto nomem;
+			for (j = 0; j < num_export_targets; j++) {
+				target = ceph_decode_32(&pexport_targets);
+				info->export_targets[j] = target;
+			}
+		} else {
+			info->export_targets = NULL;
+		}
+	}
+
+	/* pg_pools */
+	ceph_decode_32_safe(p, end, n, bad);
+	m->m_num_data_pg_pools = n;
+	m->m_data_pg_pools = kcalloc(n, sizeof(u64), GFP_NOFS);
+	if (!m->m_data_pg_pools)
+		goto nomem;
+	ceph_decode_need(p, end, sizeof(u64)*(n+1), bad);
+	for (i = 0; i < n; i++)
+		m->m_data_pg_pools[i] = ceph_decode_64(p);
+	m->m_cas_pg_pool = ceph_decode_64(p);
+	m->m_enabled = m->m_epoch > 1;
+
+	mdsmap_ev = 1;
+	if (mdsmap_v >= 2) {
+		ceph_decode_16_safe(p, end, mdsmap_ev, bad_ext);
+	}
+	if (mdsmap_ev >= 3) {
+		if (__decode_and_drop_compat_set(p, end) < 0)
+			goto bad_ext;
+	}
+	/* metadata_pool */
+	if (mdsmap_ev < 5) {
+		__decode_and_drop_type(p, end, u32, bad_ext);
+	} else {
+		__decode_and_drop_type(p, end, u64, bad_ext);
+	}
+
+	/* created + modified + tableserver */
+	__decode_and_drop_type(p, end, struct ceph_timespec, bad_ext);
+	__decode_and_drop_type(p, end, struct ceph_timespec, bad_ext);
+	__decode_and_drop_type(p, end, u32, bad_ext);
+
+	/* in */
+	{
+		int num_laggy = 0;
+		ceph_decode_32_safe(p, end, n, bad_ext);
+		ceph_decode_need(p, end, sizeof(u32) * n, bad_ext);
+
+		for (i = 0; i < n; i++) {
+			s32 mds = ceph_decode_32(p);
+			if (mds >= 0 && mds < m->possible_max_rank) {
+				if (m->m_info[mds].laggy)
+					num_laggy++;
+			}
+		}
+		m->m_num_laggy = num_laggy;
+
+		if (n > m->possible_max_rank) {
+			void *new_m_info = krealloc(m->m_info,
+						    n * sizeof(*m->m_info),
+						    GFP_NOFS | __GFP_ZERO);
+			if (!new_m_info)
+				goto nomem;
+			m->m_info = new_m_info;
+		}
+		m->possible_max_rank = n;
+	}
+
+	/* inc */
+	__decode_and_drop_map(p, end, u32, u32, bad_ext);
+	/* up */
+	__decode_and_drop_map(p, end, u32, u64, bad_ext);
+	/* failed */
+	__decode_and_drop_set(p, end, u32, bad_ext);
+	/* stopped */
+	__decode_and_drop_set(p, end, u32, bad_ext);
+
+	if (mdsmap_ev >= 4) {
+		/* last_failure_osd_epoch */
+		__decode_and_drop_type(p, end, u32, bad_ext);
+	}
+	if (mdsmap_ev >= 6) {
+		/* ever_allowed_snaps */
+		__decode_and_drop_type(p, end, u8, bad_ext);
+		/* explicitly_allowed_snaps */
+		__decode_and_drop_type(p, end, u8, bad_ext);
+	}
+	if (mdsmap_ev >= 7) {
+		/* inline_data_enabled */
+		__decode_and_drop_type(p, end, u8, bad_ext);
+	}
+	if (mdsmap_ev >= 8) {
+		/* enabled */
+		ceph_decode_8_safe(p, end, m->m_enabled, bad_ext);
+		/* fs_name */
+		ceph_decode_skip_string(p, end, bad_ext);
+	}
+	/* damaged */
+	if (mdsmap_ev >= 9) {
+		size_t need;
+		ceph_decode_32_safe(p, end, n, bad_ext);
+		need = sizeof(u32) * n;
+		ceph_decode_need(p, end, need, bad_ext);
+		*p += need;
+		m->m_damaged = n > 0;
+	} else {
+		m->m_damaged = false;
+	}
+	if (mdsmap_ev >= 17) {
+		/* balancer */
+		ceph_decode_skip_string(p, end, bad_ext);
+		/* standby_count_wanted */
+		ceph_decode_skip_32(p, end, bad_ext);
+		/* old_max_mds */
+		ceph_decode_skip_32(p, end, bad_ext);
+		/* min_compat_client */
+		ceph_decode_skip_8(p, end, bad_ext);
+		/* required_client_features */
+		ceph_decode_skip_set(p, end, 64, bad_ext);
+		ceph_decode_64_safe(p, end, m->m_max_xattr_size, bad_ext);
+	} else {
+		/* This forces the usage of the (sync) SETXATTR Op */
+		m->m_max_xattr_size = 0;
+	}
+bad_ext:
+	dout("mdsmap_decode m_enabled: %d, m_damaged: %d, m_num_laggy: %d\n",
+	     !!m->m_enabled, !!m->m_damaged, m->m_num_laggy);
+	*p = end;
+	dout("mdsmap_decode success epoch %u\n", m->m_epoch);
+	return m;
+nomem:
+	err = -ENOMEM;
+	goto out_err;
+corrupt:
+	pr_err("corrupt mdsmap\n");
+	print_hex_dump(KERN_DEBUG, "mdsmap: ",
+		       DUMP_PREFIX_OFFSET, 16, 1,
+		       start, end - start, true);
+out_err:
+	ceph_mdsmap_destroy(m);
+	return ERR_PTR(err);
+bad:
+	err = -EINVAL;
+	goto corrupt;
+}
+
+void ceph_mdsmap_destroy(struct ceph_mdsmap *m)
+{
+	int i;
+
+	if (m->m_info) {
+		for (i = 0; i < m->possible_max_rank; i++)
+			kfree(m->m_info[i].export_targets);
+		kfree(m->m_info);
+	}
+	kfree(m->m_data_pg_pools);
+	kfree(m);
+}
+
+bool ceph_mdsmap_is_cluster_available(struct ceph_mdsmap *m)
+{
+	int i, nr_active = 0;
+	if (!m->m_enabled)
+		return false;
+	if (m->m_damaged)
+		return false;
+	if (m->m_num_laggy == m->m_num_active_mds)
+		return false;
+	for (i = 0; i < m->possible_max_rank; i++) {
+		if (m->m_info[i].state == CEPH_MDS_STATE_ACTIVE)
+			nr_active++;
+	}
+	return nr_active > 0;
+}
diff --git a/fs/ceph/metric.c b/fs/ceph/metric.c
new file mode 100644
index 000000000..9560b7bc6
--- /dev/null
+++ b/fs/ceph/metric.c
@@ -0,0 +1,353 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#include <linux/ceph/ceph_debug.h>
+
+#include <linux/types.h>
+#include <linux/percpu_counter.h>
+#include <linux/math64.h>
+
+#include "metric.h"
+#include "mds_client.h"
+
+static void ktime_to_ceph_timespec(struct ceph_timespec *ts, ktime_t val)
+{
+	struct timespec64 t = ktime_to_timespec64(val);
+	ceph_encode_timespec64(ts, &t);
+}
+
+static bool ceph_mdsc_send_metrics(struct ceph_mds_client *mdsc,
+				   struct ceph_mds_session *s)
+{
+	struct ceph_metric_head *head;
+	struct ceph_metric_cap *cap;
+	struct ceph_metric_read_latency *read;
+	struct ceph_metric_write_latency *write;
+	struct ceph_metric_metadata_latency *meta;
+	struct ceph_metric_dlease *dlease;
+	struct ceph_opened_files *files;
+	struct ceph_pinned_icaps *icaps;
+	struct ceph_opened_inodes *inodes;
+	struct ceph_read_io_size *rsize;
+	struct ceph_write_io_size *wsize;
+	struct ceph_client_metric *m = &mdsc->metric;
+	u64 nr_caps = atomic64_read(&m->total_caps);
+	u32 header_len = sizeof(struct ceph_metric_header);
+	struct ceph_msg *msg;
+	s64 sum;
+	s32 items = 0;
+	s32 len;
+
+	len = sizeof(*head) + sizeof(*cap) + sizeof(*read) + sizeof(*write)
+	      + sizeof(*meta) + sizeof(*dlease) + sizeof(*files)
+	      + sizeof(*icaps) + sizeof(*inodes) + sizeof(*rsize)
+	      + sizeof(*wsize);
+
+	msg = ceph_msg_new(CEPH_MSG_CLIENT_METRICS, len, GFP_NOFS, true);
+	if (!msg) {
+		pr_err("send metrics to mds%d, failed to allocate message\n",
+		       s->s_mds);
+		return false;
+	}
+
+	head = msg->front.iov_base;
+
+	/* encode the cap metric */
+	cap = (struct ceph_metric_cap *)(head + 1);
+	cap->header.type = cpu_to_le32(CLIENT_METRIC_TYPE_CAP_INFO);
+	cap->header.ver = 1;
+	cap->header.compat = 1;
+	cap->header.data_len = cpu_to_le32(sizeof(*cap) - header_len);
+	cap->hit = cpu_to_le64(percpu_counter_sum(&m->i_caps_hit));
+	cap->mis = cpu_to_le64(percpu_counter_sum(&m->i_caps_mis));
+	cap->total = cpu_to_le64(nr_caps);
+	items++;
+
+	/* encode the read latency metric */
+	read = (struct ceph_metric_read_latency *)(cap + 1);
+	read->header.type = cpu_to_le32(CLIENT_METRIC_TYPE_READ_LATENCY);
+	read->header.ver = 2;
+	read->header.compat = 1;
+	read->header.data_len = cpu_to_le32(sizeof(*read) - header_len);
+	sum = m->metric[METRIC_READ].latency_sum;
+	ktime_to_ceph_timespec(&read->lat, sum);
+	ktime_to_ceph_timespec(&read->avg, m->metric[METRIC_READ].latency_avg);
+	read->sq_sum = cpu_to_le64(m->metric[METRIC_READ].latency_sq_sum);
+	read->count = cpu_to_le64(m->metric[METRIC_READ].total);
+	items++;
+
+	/* encode the write latency metric */
+	write = (struct ceph_metric_write_latency *)(read + 1);
+	write->header.type = cpu_to_le32(CLIENT_METRIC_TYPE_WRITE_LATENCY);
+	write->header.ver = 2;
+	write->header.compat = 1;
+	write->header.data_len = cpu_to_le32(sizeof(*write) - header_len);
+	sum = m->metric[METRIC_WRITE].latency_sum;
+	ktime_to_ceph_timespec(&write->lat, sum);
+	ktime_to_ceph_timespec(&write->avg, m->metric[METRIC_WRITE].latency_avg);
+	write->sq_sum = cpu_to_le64(m->metric[METRIC_WRITE].latency_sq_sum);
+	write->count = cpu_to_le64(m->metric[METRIC_WRITE].total);
+	items++;
+
+	/* encode the metadata latency metric */
+	meta = (struct ceph_metric_metadata_latency *)(write + 1);
+	meta->header.type = cpu_to_le32(CLIENT_METRIC_TYPE_METADATA_LATENCY);
+	meta->header.ver = 2;
+	meta->header.compat = 1;
+	meta->header.data_len = cpu_to_le32(sizeof(*meta) - header_len);
+	sum = m->metric[METRIC_METADATA].latency_sum;
+	ktime_to_ceph_timespec(&meta->lat, sum);
+	ktime_to_ceph_timespec(&meta->avg, m->metric[METRIC_METADATA].latency_avg);
+	meta->sq_sum = cpu_to_le64(m->metric[METRIC_METADATA].latency_sq_sum);
+	meta->count = cpu_to_le64(m->metric[METRIC_METADATA].total);
+	items++;
+
+	/* encode the dentry lease metric */
+	dlease = (struct ceph_metric_dlease *)(meta + 1);
+	dlease->header.type = cpu_to_le32(CLIENT_METRIC_TYPE_DENTRY_LEASE);
+	dlease->header.ver = 1;
+	dlease->header.compat = 1;
+	dlease->header.data_len = cpu_to_le32(sizeof(*dlease) - header_len);
+	dlease->hit = cpu_to_le64(percpu_counter_sum(&m->d_lease_hit));
+	dlease->mis = cpu_to_le64(percpu_counter_sum(&m->d_lease_mis));
+	dlease->total = cpu_to_le64(atomic64_read(&m->total_dentries));
+	items++;
+
+	sum = percpu_counter_sum(&m->total_inodes);
+
+	/* encode the opened files metric */
+	files = (struct ceph_opened_files *)(dlease + 1);
+	files->header.type = cpu_to_le32(CLIENT_METRIC_TYPE_OPENED_FILES);
+	files->header.ver = 1;
+	files->header.compat = 1;
+	files->header.data_len = cpu_to_le32(sizeof(*files) - header_len);
+	files->opened_files = cpu_to_le64(atomic64_read(&m->opened_files));
+	files->total = cpu_to_le64(sum);
+	items++;
+
+	/* encode the pinned icaps metric */
+	icaps = (struct ceph_pinned_icaps *)(files + 1);
+	icaps->header.type = cpu_to_le32(CLIENT_METRIC_TYPE_PINNED_ICAPS);
+	icaps->header.ver = 1;
+	icaps->header.compat = 1;
+	icaps->header.data_len = cpu_to_le32(sizeof(*icaps) - header_len);
+	icaps->pinned_icaps = cpu_to_le64(nr_caps);
+	icaps->total = cpu_to_le64(sum);
+	items++;
+
+	/* encode the opened inodes metric */
+	inodes = (struct ceph_opened_inodes *)(icaps + 1);
+	inodes->header.type = cpu_to_le32(CLIENT_METRIC_TYPE_OPENED_INODES);
+	inodes->header.ver = 1;
+	inodes->header.compat = 1;
+	inodes->header.data_len = cpu_to_le32(sizeof(*inodes) - header_len);
+	inodes->opened_inodes = cpu_to_le64(percpu_counter_sum(&m->opened_inodes));
+	inodes->total = cpu_to_le64(sum);
+	items++;
+
+	/* encode the read io size metric */
+	rsize = (struct ceph_read_io_size *)(inodes + 1);
+	rsize->header.type = cpu_to_le32(CLIENT_METRIC_TYPE_READ_IO_SIZES);
+	rsize->header.ver = 1;
+	rsize->header.compat = 1;
+	rsize->header.data_len = cpu_to_le32(sizeof(*rsize) - header_len);
+	rsize->total_ops = cpu_to_le64(m->metric[METRIC_READ].total);
+	rsize->total_size = cpu_to_le64(m->metric[METRIC_READ].size_sum);
+	items++;
+
+	/* encode the write io size metric */
+	wsize = (struct ceph_write_io_size *)(rsize + 1);
+	wsize->header.type = cpu_to_le32(CLIENT_METRIC_TYPE_WRITE_IO_SIZES);
+	wsize->header.ver = 1;
+	wsize->header.compat = 1;
+	wsize->header.data_len = cpu_to_le32(sizeof(*wsize) - header_len);
+	wsize->total_ops = cpu_to_le64(m->metric[METRIC_WRITE].total);
+	wsize->total_size = cpu_to_le64(m->metric[METRIC_WRITE].size_sum);
+	items++;
+
+	put_unaligned_le32(items, &head->num);
+	msg->front.iov_len = len;
+	msg->hdr.version = cpu_to_le16(1);
+	msg->hdr.compat_version = cpu_to_le16(1);
+	msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
+	ceph_con_send(&s->s_con, msg);
+
+	return true;
+}
+
+
+static void metric_get_session(struct ceph_mds_client *mdsc)
+{
+	struct ceph_mds_session *s;
+	int i;
+
+	mutex_lock(&mdsc->mutex);
+	for (i = 0; i < mdsc->max_sessions; i++) {
+		s = __ceph_lookup_mds_session(mdsc, i);
+		if (!s)
+			continue;
+
+		/*
+		 * Skip it if MDS doesn't support the metric collection,
+		 * or the MDS will close the session's socket connection
+		 * directly when it get this message.
+		 */
+		if (check_session_state(s) &&
+		    test_bit(CEPHFS_FEATURE_METRIC_COLLECT, &s->s_features)) {
+			mdsc->metric.session = s;
+			break;
+		}
+
+		ceph_put_mds_session(s);
+	}
+	mutex_unlock(&mdsc->mutex);
+}
+
+static void metric_delayed_work(struct work_struct *work)
+{
+	struct ceph_client_metric *m =
+		container_of(work, struct ceph_client_metric, delayed_work.work);
+	struct ceph_mds_client *mdsc =
+		container_of(m, struct ceph_mds_client, metric);
+
+	if (mdsc->stopping || disable_send_metrics)
+		return;
+
+	if (!m->session || !check_session_state(m->session)) {
+		if (m->session) {
+			ceph_put_mds_session(m->session);
+			m->session = NULL;
+		}
+		metric_get_session(mdsc);
+	}
+	if (m->session) {
+		ceph_mdsc_send_metrics(mdsc, m->session);
+		metric_schedule_delayed(m);
+	}
+}
+
+int ceph_metric_init(struct ceph_client_metric *m)
+{
+	struct ceph_metric *metric;
+	int ret, i;
+
+	if (!m)
+		return -EINVAL;
+
+	atomic64_set(&m->total_dentries, 0);
+	ret = percpu_counter_init(&m->d_lease_hit, 0, GFP_KERNEL);
+	if (ret)
+		return ret;
+
+	ret = percpu_counter_init(&m->d_lease_mis, 0, GFP_KERNEL);
+	if (ret)
+		goto err_d_lease_mis;
+
+	atomic64_set(&m->total_caps, 0);
+	ret = percpu_counter_init(&m->i_caps_hit, 0, GFP_KERNEL);
+	if (ret)
+		goto err_i_caps_hit;
+
+	ret = percpu_counter_init(&m->i_caps_mis, 0, GFP_KERNEL);
+	if (ret)
+		goto err_i_caps_mis;
+
+	for (i = 0; i < METRIC_MAX; i++) {
+		metric = &m->metric[i];
+		spin_lock_init(&metric->lock);
+		metric->size_sum = 0;
+		metric->size_min = U64_MAX;
+		metric->size_max = 0;
+		metric->total = 0;
+		metric->latency_sum = 0;
+		metric->latency_avg = 0;
+		metric->latency_sq_sum = 0;
+		metric->latency_min = KTIME_MAX;
+		metric->latency_max = 0;
+	}
+
+	atomic64_set(&m->opened_files, 0);
+	ret = percpu_counter_init(&m->opened_inodes, 0, GFP_KERNEL);
+	if (ret)
+		goto err_opened_inodes;
+	ret = percpu_counter_init(&m->total_inodes, 0, GFP_KERNEL);
+	if (ret)
+		goto err_total_inodes;
+
+	m->session = NULL;
+	INIT_DELAYED_WORK(&m->delayed_work, metric_delayed_work);
+
+	return 0;
+
+err_total_inodes:
+	percpu_counter_destroy(&m->opened_inodes);
+err_opened_inodes:
+	percpu_counter_destroy(&m->i_caps_mis);
+err_i_caps_mis:
+	percpu_counter_destroy(&m->i_caps_hit);
+err_i_caps_hit:
+	percpu_counter_destroy(&m->d_lease_mis);
+err_d_lease_mis:
+	percpu_counter_destroy(&m->d_lease_hit);
+
+	return ret;
+}
+
+void ceph_metric_destroy(struct ceph_client_metric *m)
+{
+	if (!m)
+		return;
+
+	cancel_delayed_work_sync(&m->delayed_work);
+
+	percpu_counter_destroy(&m->total_inodes);
+	percpu_counter_destroy(&m->opened_inodes);
+	percpu_counter_destroy(&m->i_caps_mis);
+	percpu_counter_destroy(&m->i_caps_hit);
+	percpu_counter_destroy(&m->d_lease_mis);
+	percpu_counter_destroy(&m->d_lease_hit);
+
+	ceph_put_mds_session(m->session);
+}
+
+#define METRIC_UPDATE_MIN_MAX(min, max, new)	\
+{						\
+	if (unlikely(new < min))		\
+		min = new;			\
+	if (unlikely(new > max))		\
+		max = new;			\
+}
+
+static inline void __update_mean_and_stdev(ktime_t total, ktime_t *lavg,
+					   ktime_t *sq_sump, ktime_t lat)
+{
+	ktime_t avg;
+
+	if (unlikely(total == 1)) {
+		*lavg = lat;
+	} else {
+		/* the sq is (lat - old_avg) * (lat - new_avg) */
+		avg = *lavg + div64_s64(lat - *lavg, total);
+		*sq_sump += (lat - *lavg)*(lat - avg);
+		*lavg = avg;
+	}
+}
+
+void ceph_update_metrics(struct ceph_metric *m,
+			 ktime_t r_start, ktime_t r_end,
+			 unsigned int size, int rc)
+{
+	ktime_t lat = ktime_sub(r_end, r_start);
+	ktime_t total;
+
+	if (unlikely(rc < 0 && rc != -ENOENT && rc != -ETIMEDOUT))
+		return;
+
+	spin_lock(&m->lock);
+	total = ++m->total;
+	m->size_sum += size;
+	METRIC_UPDATE_MIN_MAX(m->size_min, m->size_max, size);
+	m->latency_sum += lat;
+	METRIC_UPDATE_MIN_MAX(m->latency_min, m->latency_max, lat);
+	__update_mean_and_stdev(total, &m->latency_avg,	&m->latency_sq_sum,
+				lat);
+	spin_unlock(&m->lock);
+}
diff --git a/fs/ceph/metric.h b/fs/ceph/metric.h
new file mode 100644
index 000000000..0d0c44bd3
--- /dev/null
+++ b/fs/ceph/metric.h
@@ -0,0 +1,244 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _FS_CEPH_MDS_METRIC_H
+#define _FS_CEPH_MDS_METRIC_H
+
+#include <linux/ceph/types.h>
+#include <linux/percpu_counter.h>
+#include <linux/ktime.h>
+
+extern bool disable_send_metrics;
+
+enum ceph_metric_type {
+	CLIENT_METRIC_TYPE_CAP_INFO,
+	CLIENT_METRIC_TYPE_READ_LATENCY,
+	CLIENT_METRIC_TYPE_WRITE_LATENCY,
+	CLIENT_METRIC_TYPE_METADATA_LATENCY,
+	CLIENT_METRIC_TYPE_DENTRY_LEASE,
+	CLIENT_METRIC_TYPE_OPENED_FILES,
+	CLIENT_METRIC_TYPE_PINNED_ICAPS,
+	CLIENT_METRIC_TYPE_OPENED_INODES,
+	CLIENT_METRIC_TYPE_READ_IO_SIZES,
+	CLIENT_METRIC_TYPE_WRITE_IO_SIZES,
+	CLIENT_METRIC_TYPE_AVG_READ_LATENCY,
+	CLIENT_METRIC_TYPE_STDEV_READ_LATENCY,
+	CLIENT_METRIC_TYPE_AVG_WRITE_LATENCY,
+	CLIENT_METRIC_TYPE_STDEV_WRITE_LATENCY,
+	CLIENT_METRIC_TYPE_AVG_METADATA_LATENCY,
+	CLIENT_METRIC_TYPE_STDEV_METADATA_LATENCY,
+
+	CLIENT_METRIC_TYPE_MAX = CLIENT_METRIC_TYPE_STDEV_METADATA_LATENCY,
+};
+
+/*
+ * This will always have the highest metric bit value
+ * as the last element of the array.
+ */
+#define CEPHFS_METRIC_SPEC_CLIENT_SUPPORTED {	   \
+	CLIENT_METRIC_TYPE_CAP_INFO,		   \
+	CLIENT_METRIC_TYPE_READ_LATENCY,	   \
+	CLIENT_METRIC_TYPE_WRITE_LATENCY,	   \
+	CLIENT_METRIC_TYPE_METADATA_LATENCY,	   \
+	CLIENT_METRIC_TYPE_DENTRY_LEASE,	   \
+	CLIENT_METRIC_TYPE_OPENED_FILES,	   \
+	CLIENT_METRIC_TYPE_PINNED_ICAPS,	   \
+	CLIENT_METRIC_TYPE_OPENED_INODES,	   \
+	CLIENT_METRIC_TYPE_READ_IO_SIZES,	   \
+	CLIENT_METRIC_TYPE_WRITE_IO_SIZES,	   \
+	CLIENT_METRIC_TYPE_AVG_READ_LATENCY,	   \
+	CLIENT_METRIC_TYPE_STDEV_READ_LATENCY,	   \
+	CLIENT_METRIC_TYPE_AVG_WRITE_LATENCY,	   \
+	CLIENT_METRIC_TYPE_STDEV_WRITE_LATENCY,	   \
+	CLIENT_METRIC_TYPE_AVG_METADATA_LATENCY,   \
+	CLIENT_METRIC_TYPE_STDEV_METADATA_LATENCY, \
+						   \
+	CLIENT_METRIC_TYPE_MAX,			   \
+}
+
+struct ceph_metric_header {
+	__le32 type;     /* ceph metric type */
+	__u8  ver;
+	__u8  compat;
+	__le32 data_len; /* length of sizeof(hit + mis + total) */
+} __packed;
+
+/* metric caps header */
+struct ceph_metric_cap {
+	struct ceph_metric_header header;
+	__le64 hit;
+	__le64 mis;
+	__le64 total;
+} __packed;
+
+/* metric read latency header */
+struct ceph_metric_read_latency {
+	struct ceph_metric_header header;
+	struct ceph_timespec lat;
+	struct ceph_timespec avg;
+	__le64 sq_sum;
+	__le64 count;
+} __packed;
+
+/* metric write latency header */
+struct ceph_metric_write_latency {
+	struct ceph_metric_header header;
+	struct ceph_timespec lat;
+	struct ceph_timespec avg;
+	__le64 sq_sum;
+	__le64 count;
+} __packed;
+
+/* metric metadata latency header */
+struct ceph_metric_metadata_latency {
+	struct ceph_metric_header header;
+	struct ceph_timespec lat;
+	struct ceph_timespec avg;
+	__le64 sq_sum;
+	__le64 count;
+} __packed;
+
+/* metric dentry lease header */
+struct ceph_metric_dlease {
+	struct ceph_metric_header header;
+	__le64 hit;
+	__le64 mis;
+	__le64 total;
+} __packed;
+
+/* metric opened files header */
+struct ceph_opened_files {
+	struct ceph_metric_header header;
+	__le64 opened_files;
+	__le64 total;
+} __packed;
+
+/* metric pinned i_caps header */
+struct ceph_pinned_icaps {
+	struct ceph_metric_header header;
+	__le64 pinned_icaps;
+	__le64 total;
+} __packed;
+
+/* metric opened inodes header */
+struct ceph_opened_inodes {
+	struct ceph_metric_header header;
+	__le64 opened_inodes;
+	__le64 total;
+} __packed;
+
+/* metric read io size header */
+struct ceph_read_io_size {
+	struct ceph_metric_header header;
+	__le64 total_ops;
+	__le64 total_size;
+} __packed;
+
+/* metric write io size header */
+struct ceph_write_io_size {
+	struct ceph_metric_header header;
+	__le64 total_ops;
+	__le64 total_size;
+} __packed;
+
+struct ceph_metric_head {
+	__le32 num;	/* the number of metrics that will be sent */
+} __packed;
+
+enum metric_type {
+	METRIC_READ,
+	METRIC_WRITE,
+	METRIC_METADATA,
+	METRIC_COPYFROM,
+	METRIC_MAX
+};
+
+struct ceph_metric {
+	spinlock_t lock;
+	u64 total;
+	u64 size_sum;
+	u64 size_min;
+	u64 size_max;
+	ktime_t latency_sum;
+	ktime_t latency_avg;
+	ktime_t latency_sq_sum;
+	ktime_t latency_min;
+	ktime_t latency_max;
+};
+
+/* This is the global metrics */
+struct ceph_client_metric {
+	atomic64_t            total_dentries;
+	struct percpu_counter d_lease_hit;
+	struct percpu_counter d_lease_mis;
+
+	atomic64_t            total_caps;
+	struct percpu_counter i_caps_hit;
+	struct percpu_counter i_caps_mis;
+
+	struct ceph_metric metric[METRIC_MAX];
+
+	/* The total number of directories and files that are opened */
+	atomic64_t opened_files;
+
+	/* The total number of inodes that have opened files or directories */
+	struct percpu_counter opened_inodes;
+	struct percpu_counter total_inodes;
+
+	struct ceph_mds_session *session;
+	struct delayed_work delayed_work;  /* delayed work */
+};
+
+static inline void metric_schedule_delayed(struct ceph_client_metric *m)
+{
+	if (disable_send_metrics)
+		return;
+
+	/* per second */
+	schedule_delayed_work(&m->delayed_work, round_jiffies_relative(HZ));
+}
+
+extern int ceph_metric_init(struct ceph_client_metric *m);
+extern void ceph_metric_destroy(struct ceph_client_metric *m);
+
+static inline void ceph_update_cap_hit(struct ceph_client_metric *m)
+{
+	percpu_counter_inc(&m->i_caps_hit);
+}
+
+static inline void ceph_update_cap_mis(struct ceph_client_metric *m)
+{
+	percpu_counter_inc(&m->i_caps_mis);
+}
+
+extern void ceph_update_metrics(struct ceph_metric *m,
+				ktime_t r_start, ktime_t r_end,
+				unsigned int size, int rc);
+
+static inline void ceph_update_read_metrics(struct ceph_client_metric *m,
+					    ktime_t r_start, ktime_t r_end,
+					    unsigned int size, int rc)
+{
+	ceph_update_metrics(&m->metric[METRIC_READ],
+			    r_start, r_end, size, rc);
+}
+static inline void ceph_update_write_metrics(struct ceph_client_metric *m,
+					     ktime_t r_start, ktime_t r_end,
+					     unsigned int size, int rc)
+{
+	ceph_update_metrics(&m->metric[METRIC_WRITE],
+			    r_start, r_end, size, rc);
+}
+static inline void ceph_update_metadata_metrics(struct ceph_client_metric *m,
+						ktime_t r_start, ktime_t r_end,
+						int rc)
+{
+	ceph_update_metrics(&m->metric[METRIC_METADATA],
+			    r_start, r_end, 0, rc);
+}
+static inline void ceph_update_copyfrom_metrics(struct ceph_client_metric *m,
+						ktime_t r_start, ktime_t r_end,
+						unsigned int size, int rc)
+{
+	ceph_update_metrics(&m->metric[METRIC_COPYFROM],
+			    r_start, r_end, size, rc);
+}
+#endif /* _FS_CEPH_MDS_METRIC_H */
diff --git a/fs/ceph/quota.c b/fs/ceph/quota.c
new file mode 100644
index 000000000..f7fcf7f08
--- /dev/null
+++ b/fs/ceph/quota.c
@@ -0,0 +1,535 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * quota.c - CephFS quota
+ *
+ * Copyright (C) 2017-2018 SUSE
+ */
+
+#include <linux/statfs.h>
+
+#include "super.h"
+#include "mds_client.h"
+
+void ceph_adjust_quota_realms_count(struct inode *inode, bool inc)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
+	if (inc)
+		atomic64_inc(&mdsc->quotarealms_count);
+	else
+		atomic64_dec(&mdsc->quotarealms_count);
+}
+
+static inline bool ceph_has_realms_with_quotas(struct inode *inode)
+{
+	struct super_block *sb = inode->i_sb;
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(sb);
+	struct inode *root = d_inode(sb->s_root);
+
+	if (atomic64_read(&mdsc->quotarealms_count) > 0)
+		return true;
+	/* if root is the real CephFS root, we don't have quota realms */
+	if (root && ceph_ino(root) == CEPH_INO_ROOT)
+		return false;
+	/* MDS stray dirs have no quota realms */
+	if (ceph_vino_is_reserved(ceph_inode(inode)->i_vino))
+		return false;
+	/* otherwise, we can't know for sure */
+	return true;
+}
+
+void ceph_handle_quota(struct ceph_mds_client *mdsc,
+		       struct ceph_mds_session *session,
+		       struct ceph_msg *msg)
+{
+	struct super_block *sb = mdsc->fsc->sb;
+	struct ceph_mds_quota *h = msg->front.iov_base;
+	struct ceph_vino vino;
+	struct inode *inode;
+	struct ceph_inode_info *ci;
+
+	if (!ceph_inc_mds_stopping_blocker(mdsc, session))
+		return;
+
+	if (msg->front.iov_len < sizeof(*h)) {
+		pr_err("%s corrupt message mds%d len %d\n", __func__,
+		       session->s_mds, (int)msg->front.iov_len);
+		ceph_msg_dump(msg);
+		goto out;
+	}
+
+	/* lookup inode */
+	vino.ino = le64_to_cpu(h->ino);
+	vino.snap = CEPH_NOSNAP;
+	inode = ceph_find_inode(sb, vino);
+	if (!inode) {
+		pr_warn("Failed to find inode %llu\n", vino.ino);
+		goto out;
+	}
+	ci = ceph_inode(inode);
+
+	spin_lock(&ci->i_ceph_lock);
+	ci->i_rbytes = le64_to_cpu(h->rbytes);
+	ci->i_rfiles = le64_to_cpu(h->rfiles);
+	ci->i_rsubdirs = le64_to_cpu(h->rsubdirs);
+	__ceph_update_quota(ci, le64_to_cpu(h->max_bytes),
+		            le64_to_cpu(h->max_files));
+	spin_unlock(&ci->i_ceph_lock);
+
+	iput(inode);
+out:
+	ceph_dec_mds_stopping_blocker(mdsc);
+}
+
+static struct ceph_quotarealm_inode *
+find_quotarealm_inode(struct ceph_mds_client *mdsc, u64 ino)
+{
+	struct ceph_quotarealm_inode *qri = NULL;
+	struct rb_node **node, *parent = NULL;
+
+	mutex_lock(&mdsc->quotarealms_inodes_mutex);
+	node = &(mdsc->quotarealms_inodes.rb_node);
+	while (*node) {
+		parent = *node;
+		qri = container_of(*node, struct ceph_quotarealm_inode, node);
+
+		if (ino < qri->ino)
+			node = &((*node)->rb_left);
+		else if (ino > qri->ino)
+			node = &((*node)->rb_right);
+		else
+			break;
+	}
+	if (!qri || (qri->ino != ino)) {
+		/* Not found, create a new one and insert it */
+		qri = kmalloc(sizeof(*qri), GFP_KERNEL);
+		if (qri) {
+			qri->ino = ino;
+			qri->inode = NULL;
+			qri->timeout = 0;
+			mutex_init(&qri->mutex);
+			rb_link_node(&qri->node, parent, node);
+			rb_insert_color(&qri->node, &mdsc->quotarealms_inodes);
+		} else
+			pr_warn("Failed to alloc quotarealms_inode\n");
+	}
+	mutex_unlock(&mdsc->quotarealms_inodes_mutex);
+
+	return qri;
+}
+
+/*
+ * This function will try to lookup a realm inode which isn't visible in the
+ * filesystem mountpoint.  A list of these kind of inodes (not visible) is
+ * maintained in the mdsc and freed only when the filesystem is umounted.
+ *
+ * Note that these inodes are kept in this list even if the lookup fails, which
+ * allows to prevent useless lookup requests.
+ */
+static struct inode *lookup_quotarealm_inode(struct ceph_mds_client *mdsc,
+					     struct super_block *sb,
+					     struct ceph_snap_realm *realm)
+{
+	struct ceph_quotarealm_inode *qri;
+	struct inode *in;
+
+	qri = find_quotarealm_inode(mdsc, realm->ino);
+	if (!qri)
+		return NULL;
+
+	mutex_lock(&qri->mutex);
+	if (qri->inode && ceph_is_any_caps(qri->inode)) {
+		/* A request has already returned the inode */
+		mutex_unlock(&qri->mutex);
+		return qri->inode;
+	}
+	/* Check if this inode lookup has failed recently */
+	if (qri->timeout &&
+	    time_before_eq(jiffies, qri->timeout)) {
+		mutex_unlock(&qri->mutex);
+		return NULL;
+	}
+	if (qri->inode) {
+		/* get caps */
+		int ret = __ceph_do_getattr(qri->inode, NULL,
+					    CEPH_STAT_CAP_INODE, true);
+		if (ret >= 0)
+			in = qri->inode;
+		else
+			in = ERR_PTR(ret);
+	}  else {
+		in = ceph_lookup_inode(sb, realm->ino);
+	}
+
+	if (IS_ERR(in)) {
+		dout("Can't lookup inode %llx (err: %ld)\n",
+		     realm->ino, PTR_ERR(in));
+		qri->timeout = jiffies + msecs_to_jiffies(60 * 1000); /* XXX */
+	} else {
+		qri->timeout = 0;
+		qri->inode = in;
+	}
+	mutex_unlock(&qri->mutex);
+
+	return in;
+}
+
+void ceph_cleanup_quotarealms_inodes(struct ceph_mds_client *mdsc)
+{
+	struct ceph_quotarealm_inode *qri;
+	struct rb_node *node;
+
+	/*
+	 * It should now be safe to clean quotarealms_inode tree without holding
+	 * mdsc->quotarealms_inodes_mutex...
+	 */
+	mutex_lock(&mdsc->quotarealms_inodes_mutex);
+	while (!RB_EMPTY_ROOT(&mdsc->quotarealms_inodes)) {
+		node = rb_first(&mdsc->quotarealms_inodes);
+		qri = rb_entry(node, struct ceph_quotarealm_inode, node);
+		rb_erase(node, &mdsc->quotarealms_inodes);
+		iput(qri->inode);
+		kfree(qri);
+	}
+	mutex_unlock(&mdsc->quotarealms_inodes_mutex);
+}
+
+/*
+ * This function walks through the snaprealm for an inode and returns the
+ * ceph_snap_realm for the first snaprealm that has quotas set (max_files,
+ * max_bytes, or any, depending on the 'which_quota' argument).  If the root is
+ * reached, return the root ceph_snap_realm instead.
+ *
+ * Note that the caller is responsible for calling ceph_put_snap_realm() on the
+ * returned realm.
+ *
+ * Callers of this function need to hold mdsc->snap_rwsem.  However, if there's
+ * a need to do an inode lookup, this rwsem will be temporarily dropped.  Hence
+ * the 'retry' argument: if rwsem needs to be dropped and 'retry' is 'false'
+ * this function will return -EAGAIN; otherwise, the snaprealms walk-through
+ * will be restarted.
+ */
+static struct ceph_snap_realm *get_quota_realm(struct ceph_mds_client *mdsc,
+					       struct inode *inode,
+					       enum quota_get_realm which_quota,
+					       bool retry)
+{
+	struct ceph_inode_info *ci = NULL;
+	struct ceph_snap_realm *realm, *next;
+	struct inode *in;
+	bool has_quota;
+
+	if (ceph_snap(inode) != CEPH_NOSNAP)
+		return NULL;
+
+restart:
+	realm = ceph_inode(inode)->i_snap_realm;
+	if (realm)
+		ceph_get_snap_realm(mdsc, realm);
+	else
+		pr_err_ratelimited("get_quota_realm: ino (%llx.%llx) "
+				   "null i_snap_realm\n", ceph_vinop(inode));
+	while (realm) {
+		bool has_inode;
+
+		spin_lock(&realm->inodes_with_caps_lock);
+		has_inode = realm->inode;
+		in = has_inode ? igrab(realm->inode) : NULL;
+		spin_unlock(&realm->inodes_with_caps_lock);
+		if (has_inode && !in)
+			break;
+		if (!in) {
+			up_read(&mdsc->snap_rwsem);
+			in = lookup_quotarealm_inode(mdsc, inode->i_sb, realm);
+			down_read(&mdsc->snap_rwsem);
+			if (IS_ERR_OR_NULL(in))
+				break;
+			ceph_put_snap_realm(mdsc, realm);
+			if (!retry)
+				return ERR_PTR(-EAGAIN);
+			goto restart;
+		}
+
+		ci = ceph_inode(in);
+		has_quota = __ceph_has_quota(ci, which_quota);
+		iput(in);
+
+		next = realm->parent;
+		if (has_quota || !next)
+		       return realm;
+
+		ceph_get_snap_realm(mdsc, next);
+		ceph_put_snap_realm(mdsc, realm);
+		realm = next;
+	}
+	if (realm)
+		ceph_put_snap_realm(mdsc, realm);
+
+	return NULL;
+}
+
+bool ceph_quota_is_same_realm(struct inode *old, struct inode *new)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(old->i_sb);
+	struct ceph_snap_realm *old_realm, *new_realm;
+	bool is_same;
+
+restart:
+	/*
+	 * We need to lookup 2 quota realms atomically, i.e. with snap_rwsem.
+	 * However, get_quota_realm may drop it temporarily.  By setting the
+	 * 'retry' parameter to 'false', we'll get -EAGAIN if the rwsem was
+	 * dropped and we can then restart the whole operation.
+	 */
+	down_read(&mdsc->snap_rwsem);
+	old_realm = get_quota_realm(mdsc, old, QUOTA_GET_ANY, true);
+	new_realm = get_quota_realm(mdsc, new, QUOTA_GET_ANY, false);
+	if (PTR_ERR(new_realm) == -EAGAIN) {
+		up_read(&mdsc->snap_rwsem);
+		if (old_realm)
+			ceph_put_snap_realm(mdsc, old_realm);
+		goto restart;
+	}
+	is_same = (old_realm == new_realm);
+	up_read(&mdsc->snap_rwsem);
+
+	if (old_realm)
+		ceph_put_snap_realm(mdsc, old_realm);
+	if (new_realm)
+		ceph_put_snap_realm(mdsc, new_realm);
+
+	return is_same;
+}
+
+enum quota_check_op {
+	QUOTA_CHECK_MAX_FILES_OP,	/* check quota max_files limit */
+	QUOTA_CHECK_MAX_BYTES_OP,	/* check quota max_files limit */
+	QUOTA_CHECK_MAX_BYTES_APPROACHING_OP	/* check if quota max_files
+						   limit is approaching */
+};
+
+/*
+ * check_quota_exceeded() will walk up the snaprealm hierarchy and, for each
+ * realm, it will execute quota check operation defined by the 'op' parameter.
+ * The snaprealm walk is interrupted if the quota check detects that the quota
+ * is exceeded or if the root inode is reached.
+ */
+static bool check_quota_exceeded(struct inode *inode, enum quota_check_op op,
+				 loff_t delta)
+{
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
+	struct ceph_inode_info *ci;
+	struct ceph_snap_realm *realm, *next;
+	struct inode *in;
+	u64 max, rvalue;
+	bool exceeded = false;
+
+	if (ceph_snap(inode) != CEPH_NOSNAP)
+		return false;
+
+	down_read(&mdsc->snap_rwsem);
+restart:
+	realm = ceph_inode(inode)->i_snap_realm;
+	if (realm)
+		ceph_get_snap_realm(mdsc, realm);
+	else
+		pr_err_ratelimited("check_quota_exceeded: ino (%llx.%llx) "
+				   "null i_snap_realm\n", ceph_vinop(inode));
+	while (realm) {
+		bool has_inode;
+
+		spin_lock(&realm->inodes_with_caps_lock);
+		has_inode = realm->inode;
+		in = has_inode ? igrab(realm->inode) : NULL;
+		spin_unlock(&realm->inodes_with_caps_lock);
+		if (has_inode && !in)
+			break;
+		if (!in) {
+			up_read(&mdsc->snap_rwsem);
+			in = lookup_quotarealm_inode(mdsc, inode->i_sb, realm);
+			down_read(&mdsc->snap_rwsem);
+			if (IS_ERR_OR_NULL(in))
+				break;
+			ceph_put_snap_realm(mdsc, realm);
+			goto restart;
+		}
+		ci = ceph_inode(in);
+		spin_lock(&ci->i_ceph_lock);
+		if (op == QUOTA_CHECK_MAX_FILES_OP) {
+			max = ci->i_max_files;
+			rvalue = ci->i_rfiles + ci->i_rsubdirs;
+		} else {
+			max = ci->i_max_bytes;
+			rvalue = ci->i_rbytes;
+		}
+		spin_unlock(&ci->i_ceph_lock);
+		switch (op) {
+		case QUOTA_CHECK_MAX_FILES_OP:
+		case QUOTA_CHECK_MAX_BYTES_OP:
+			exceeded = (max && (rvalue + delta > max));
+			break;
+		case QUOTA_CHECK_MAX_BYTES_APPROACHING_OP:
+			if (max) {
+				if (rvalue >= max)
+					exceeded = true;
+				else {
+					/*
+					 * when we're writing more that 1/16th
+					 * of the available space
+					 */
+					exceeded =
+						(((max - rvalue) >> 4) < delta);
+				}
+			}
+			break;
+		default:
+			/* Shouldn't happen */
+			pr_warn("Invalid quota check op (%d)\n", op);
+			exceeded = true; /* Just break the loop */
+		}
+		iput(in);
+
+		next = realm->parent;
+		if (exceeded || !next)
+			break;
+		ceph_get_snap_realm(mdsc, next);
+		ceph_put_snap_realm(mdsc, realm);
+		realm = next;
+	}
+	if (realm)
+		ceph_put_snap_realm(mdsc, realm);
+	up_read(&mdsc->snap_rwsem);
+
+	return exceeded;
+}
+
+/*
+ * ceph_quota_is_max_files_exceeded - check if we can create a new file
+ * @inode:	directory where a new file is being created
+ *
+ * This functions returns true is max_files quota allows a new file to be
+ * created.  It is necessary to walk through the snaprealm hierarchy (until the
+ * FS root) to check all realms with quotas set.
+ */
+bool ceph_quota_is_max_files_exceeded(struct inode *inode)
+{
+	if (!ceph_has_realms_with_quotas(inode))
+		return false;
+
+	WARN_ON(!S_ISDIR(inode->i_mode));
+
+	return check_quota_exceeded(inode, QUOTA_CHECK_MAX_FILES_OP, 1);
+}
+
+/*
+ * ceph_quota_is_max_bytes_exceeded - check if we can write to a file
+ * @inode:	inode being written
+ * @newsize:	new size if write succeeds
+ *
+ * This functions returns true is max_bytes quota allows a file size to reach
+ * @newsize; it returns false otherwise.
+ */
+bool ceph_quota_is_max_bytes_exceeded(struct inode *inode, loff_t newsize)
+{
+	loff_t size = i_size_read(inode);
+
+	if (!ceph_has_realms_with_quotas(inode))
+		return false;
+
+	/* return immediately if we're decreasing file size */
+	if (newsize <= size)
+		return false;
+
+	return check_quota_exceeded(inode, QUOTA_CHECK_MAX_BYTES_OP, (newsize - size));
+}
+
+/*
+ * ceph_quota_is_max_bytes_approaching - check if we're reaching max_bytes
+ * @inode:	inode being written
+ * @newsize:	new size if write succeeds
+ *
+ * This function returns true if the new file size @newsize will be consuming
+ * more than 1/16th of the available quota space; it returns false otherwise.
+ */
+bool ceph_quota_is_max_bytes_approaching(struct inode *inode, loff_t newsize)
+{
+	loff_t size = ceph_inode(inode)->i_reported_size;
+
+	if (!ceph_has_realms_with_quotas(inode))
+		return false;
+
+	/* return immediately if we're decreasing file size */
+	if (newsize <= size)
+		return false;
+
+	return check_quota_exceeded(inode, QUOTA_CHECK_MAX_BYTES_APPROACHING_OP,
+				    (newsize - size));
+}
+
+/*
+ * ceph_quota_update_statfs - if root has quota update statfs with quota status
+ * @fsc:	filesystem client instance
+ * @buf:	statfs to update
+ *
+ * If the mounted filesystem root has max_bytes quota set, update the filesystem
+ * statistics with the quota status.
+ *
+ * This function returns true if the stats have been updated, false otherwise.
+ */
+bool ceph_quota_update_statfs(struct ceph_fs_client *fsc, struct kstatfs *buf)
+{
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	struct ceph_inode_info *ci;
+	struct ceph_snap_realm *realm;
+	struct inode *in;
+	u64 total = 0, used, free;
+	bool is_updated = false;
+
+	down_read(&mdsc->snap_rwsem);
+	realm = get_quota_realm(mdsc, d_inode(fsc->sb->s_root),
+				QUOTA_GET_MAX_BYTES, true);
+	up_read(&mdsc->snap_rwsem);
+	if (!realm)
+		return false;
+
+	spin_lock(&realm->inodes_with_caps_lock);
+	in = realm->inode ? igrab(realm->inode) : NULL;
+	spin_unlock(&realm->inodes_with_caps_lock);
+	if (in) {
+		ci = ceph_inode(in);
+		spin_lock(&ci->i_ceph_lock);
+		if (ci->i_max_bytes) {
+			total = ci->i_max_bytes >> CEPH_BLOCK_SHIFT;
+			used = ci->i_rbytes >> CEPH_BLOCK_SHIFT;
+			/* For quota size less than 4MB, use 4KB block size */
+			if (!total) {
+				total = ci->i_max_bytes >> CEPH_4K_BLOCK_SHIFT;
+				used = ci->i_rbytes >> CEPH_4K_BLOCK_SHIFT;
+	                        buf->f_frsize = 1 << CEPH_4K_BLOCK_SHIFT;
+			}
+			/* It is possible for a quota to be exceeded.
+			 * Report 'zero' in that case
+			 */
+			free = total > used ? total - used : 0;
+			/* For quota size less than 4KB, report the
+			 * total=used=4KB,free=0 when quota is full
+			 * and total=free=4KB, used=0 otherwise */
+			if (!total) {
+				total = 1;
+				free = ci->i_max_bytes > ci->i_rbytes ? 1 : 0;
+	                        buf->f_frsize = 1 << CEPH_4K_BLOCK_SHIFT;
+			}
+		}
+		spin_unlock(&ci->i_ceph_lock);
+		if (total) {
+			buf->f_blocks = total;
+			buf->f_bfree = free;
+			buf->f_bavail = free;
+			is_updated = true;
+		}
+		iput(in);
+	}
+	ceph_put_snap_realm(mdsc, realm);
+
+	return is_updated;
+}
+
diff --git a/fs/ceph/snap.c b/fs/ceph/snap.c
new file mode 100644
index 000000000..82f7592e1
--- /dev/null
+++ b/fs/ceph/snap.c
@@ -0,0 +1,1320 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/ceph/ceph_debug.h>
+
+#include <linux/fs.h>
+#include <linux/sort.h>
+#include <linux/slab.h>
+#include <linux/iversion.h>
+#include "super.h"
+#include "mds_client.h"
+#include <linux/ceph/decode.h>
+
+/* unused map expires after 5 minutes */
+#define CEPH_SNAPID_MAP_TIMEOUT	(5 * 60 * HZ)
+
+/*
+ * Snapshots in ceph are driven in large part by cooperation from the
+ * client.  In contrast to local file systems or file servers that
+ * implement snapshots at a single point in the system, ceph's
+ * distributed access to storage requires clients to help decide
+ * whether a write logically occurs before or after a recently created
+ * snapshot.
+ *
+ * This provides a perfect instantanous client-wide snapshot.  Between
+ * clients, however, snapshots may appear to be applied at slightly
+ * different points in time, depending on delays in delivering the
+ * snapshot notification.
+ *
+ * Snapshots are _not_ file system-wide.  Instead, each snapshot
+ * applies to the subdirectory nested beneath some directory.  This
+ * effectively divides the hierarchy into multiple "realms," where all
+ * of the files contained by each realm share the same set of
+ * snapshots.  An individual realm's snap set contains snapshots
+ * explicitly created on that realm, as well as any snaps in its
+ * parent's snap set _after_ the point at which the parent became it's
+ * parent (due to, say, a rename).  Similarly, snaps from prior parents
+ * during the time intervals during which they were the parent are included.
+ *
+ * The client is spared most of this detail, fortunately... it must only
+ * maintains a hierarchy of realms reflecting the current parent/child
+ * realm relationship, and for each realm has an explicit list of snaps
+ * inherited from prior parents.
+ *
+ * A snap_realm struct is maintained for realms containing every inode
+ * with an open cap in the system.  (The needed snap realm information is
+ * provided by the MDS whenever a cap is issued, i.e., on open.)  A 'seq'
+ * version number is used to ensure that as realm parameters change (new
+ * snapshot, new parent, etc.) the client's realm hierarchy is updated.
+ *
+ * The realm hierarchy drives the generation of a 'snap context' for each
+ * realm, which simply lists the resulting set of snaps for the realm.  This
+ * is attached to any writes sent to OSDs.
+ */
+/*
+ * Unfortunately error handling is a bit mixed here.  If we get a snap
+ * update, but don't have enough memory to update our realm hierarchy,
+ * it's not clear what we can do about it (besides complaining to the
+ * console).
+ */
+
+
+/*
+ * increase ref count for the realm
+ *
+ * caller must hold snap_rwsem.
+ */
+void ceph_get_snap_realm(struct ceph_mds_client *mdsc,
+			 struct ceph_snap_realm *realm)
+{
+	lockdep_assert_held(&mdsc->snap_rwsem);
+
+	/*
+	 * The 0->1 and 1->0 transitions must take the snap_empty_lock
+	 * atomically with the refcount change. Go ahead and bump the
+	 * nref here, unless it's 0, in which case we take the spinlock
+	 * and then do the increment and remove it from the list.
+	 */
+	if (atomic_inc_not_zero(&realm->nref))
+		return;
+
+	spin_lock(&mdsc->snap_empty_lock);
+	if (atomic_inc_return(&realm->nref) == 1)
+		list_del_init(&realm->empty_item);
+	spin_unlock(&mdsc->snap_empty_lock);
+}
+
+static void __insert_snap_realm(struct rb_root *root,
+				struct ceph_snap_realm *new)
+{
+	struct rb_node **p = &root->rb_node;
+	struct rb_node *parent = NULL;
+	struct ceph_snap_realm *r = NULL;
+
+	while (*p) {
+		parent = *p;
+		r = rb_entry(parent, struct ceph_snap_realm, node);
+		if (new->ino < r->ino)
+			p = &(*p)->rb_left;
+		else if (new->ino > r->ino)
+			p = &(*p)->rb_right;
+		else
+			BUG();
+	}
+
+	rb_link_node(&new->node, parent, p);
+	rb_insert_color(&new->node, root);
+}
+
+/*
+ * create and get the realm rooted at @ino and bump its ref count.
+ *
+ * caller must hold snap_rwsem for write.
+ */
+static struct ceph_snap_realm *ceph_create_snap_realm(
+	struct ceph_mds_client *mdsc,
+	u64 ino)
+{
+	struct ceph_snap_realm *realm;
+
+	lockdep_assert_held_write(&mdsc->snap_rwsem);
+
+	realm = kzalloc(sizeof(*realm), GFP_NOFS);
+	if (!realm)
+		return ERR_PTR(-ENOMEM);
+
+	/* Do not release the global dummy snaprealm until unmouting */
+	if (ino == CEPH_INO_GLOBAL_SNAPREALM)
+		atomic_set(&realm->nref, 2);
+	else
+		atomic_set(&realm->nref, 1);
+	realm->ino = ino;
+	INIT_LIST_HEAD(&realm->children);
+	INIT_LIST_HEAD(&realm->child_item);
+	INIT_LIST_HEAD(&realm->empty_item);
+	INIT_LIST_HEAD(&realm->dirty_item);
+	INIT_LIST_HEAD(&realm->rebuild_item);
+	INIT_LIST_HEAD(&realm->inodes_with_caps);
+	spin_lock_init(&realm->inodes_with_caps_lock);
+	__insert_snap_realm(&mdsc->snap_realms, realm);
+	mdsc->num_snap_realms++;
+
+	dout("%s %llx %p\n", __func__, realm->ino, realm);
+	return realm;
+}
+
+/*
+ * lookup the realm rooted at @ino.
+ *
+ * caller must hold snap_rwsem.
+ */
+static struct ceph_snap_realm *__lookup_snap_realm(struct ceph_mds_client *mdsc,
+						   u64 ino)
+{
+	struct rb_node *n = mdsc->snap_realms.rb_node;
+	struct ceph_snap_realm *r;
+
+	lockdep_assert_held(&mdsc->snap_rwsem);
+
+	while (n) {
+		r = rb_entry(n, struct ceph_snap_realm, node);
+		if (ino < r->ino)
+			n = n->rb_left;
+		else if (ino > r->ino)
+			n = n->rb_right;
+		else {
+			dout("%s %llx %p\n", __func__, r->ino, r);
+			return r;
+		}
+	}
+	return NULL;
+}
+
+struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
+					       u64 ino)
+{
+	struct ceph_snap_realm *r;
+	r = __lookup_snap_realm(mdsc, ino);
+	if (r)
+		ceph_get_snap_realm(mdsc, r);
+	return r;
+}
+
+static void __put_snap_realm(struct ceph_mds_client *mdsc,
+			     struct ceph_snap_realm *realm);
+
+/*
+ * called with snap_rwsem (write)
+ */
+static void __destroy_snap_realm(struct ceph_mds_client *mdsc,
+				 struct ceph_snap_realm *realm)
+{
+	lockdep_assert_held_write(&mdsc->snap_rwsem);
+
+	dout("%s %p %llx\n", __func__, realm, realm->ino);
+
+	rb_erase(&realm->node, &mdsc->snap_realms);
+	mdsc->num_snap_realms--;
+
+	if (realm->parent) {
+		list_del_init(&realm->child_item);
+		__put_snap_realm(mdsc, realm->parent);
+	}
+
+	kfree(realm->prior_parent_snaps);
+	kfree(realm->snaps);
+	ceph_put_snap_context(realm->cached_context);
+	kfree(realm);
+}
+
+/*
+ * caller holds snap_rwsem (write)
+ */
+static void __put_snap_realm(struct ceph_mds_client *mdsc,
+			     struct ceph_snap_realm *realm)
+{
+	lockdep_assert_held_write(&mdsc->snap_rwsem);
+
+	/*
+	 * We do not require the snap_empty_lock here, as any caller that
+	 * increments the value must hold the snap_rwsem.
+	 */
+	if (atomic_dec_and_test(&realm->nref))
+		__destroy_snap_realm(mdsc, realm);
+}
+
+/*
+ * See comments in ceph_get_snap_realm. Caller needn't hold any locks.
+ */
+void ceph_put_snap_realm(struct ceph_mds_client *mdsc,
+			 struct ceph_snap_realm *realm)
+{
+	if (!atomic_dec_and_lock(&realm->nref, &mdsc->snap_empty_lock))
+		return;
+
+	if (down_write_trylock(&mdsc->snap_rwsem)) {
+		spin_unlock(&mdsc->snap_empty_lock);
+		__destroy_snap_realm(mdsc, realm);
+		up_write(&mdsc->snap_rwsem);
+	} else {
+		list_add(&realm->empty_item, &mdsc->snap_empty);
+		spin_unlock(&mdsc->snap_empty_lock);
+	}
+}
+
+/*
+ * Clean up any realms whose ref counts have dropped to zero.  Note
+ * that this does not include realms who were created but not yet
+ * used.
+ *
+ * Called under snap_rwsem (write)
+ */
+static void __cleanup_empty_realms(struct ceph_mds_client *mdsc)
+{
+	struct ceph_snap_realm *realm;
+
+	lockdep_assert_held_write(&mdsc->snap_rwsem);
+
+	spin_lock(&mdsc->snap_empty_lock);
+	while (!list_empty(&mdsc->snap_empty)) {
+		realm = list_first_entry(&mdsc->snap_empty,
+				   struct ceph_snap_realm, empty_item);
+		list_del(&realm->empty_item);
+		spin_unlock(&mdsc->snap_empty_lock);
+		__destroy_snap_realm(mdsc, realm);
+		spin_lock(&mdsc->snap_empty_lock);
+	}
+	spin_unlock(&mdsc->snap_empty_lock);
+}
+
+void ceph_cleanup_global_and_empty_realms(struct ceph_mds_client *mdsc)
+{
+	struct ceph_snap_realm *global_realm;
+
+	down_write(&mdsc->snap_rwsem);
+	global_realm = __lookup_snap_realm(mdsc, CEPH_INO_GLOBAL_SNAPREALM);
+	if (global_realm)
+		ceph_put_snap_realm(mdsc, global_realm);
+	__cleanup_empty_realms(mdsc);
+	up_write(&mdsc->snap_rwsem);
+}
+
+/*
+ * adjust the parent realm of a given @realm.  adjust child list, and parent
+ * pointers, and ref counts appropriately.
+ *
+ * return true if parent was changed, 0 if unchanged, <0 on error.
+ *
+ * caller must hold snap_rwsem for write.
+ */
+static int adjust_snap_realm_parent(struct ceph_mds_client *mdsc,
+				    struct ceph_snap_realm *realm,
+				    u64 parentino)
+{
+	struct ceph_snap_realm *parent;
+
+	lockdep_assert_held_write(&mdsc->snap_rwsem);
+
+	if (realm->parent_ino == parentino)
+		return 0;
+
+	parent = ceph_lookup_snap_realm(mdsc, parentino);
+	if (!parent) {
+		parent = ceph_create_snap_realm(mdsc, parentino);
+		if (IS_ERR(parent))
+			return PTR_ERR(parent);
+	}
+	dout("%s %llx %p: %llx %p -> %llx %p\n", __func__, realm->ino,
+	     realm, realm->parent_ino, realm->parent, parentino, parent);
+	if (realm->parent) {
+		list_del_init(&realm->child_item);
+		ceph_put_snap_realm(mdsc, realm->parent);
+	}
+	realm->parent_ino = parentino;
+	realm->parent = parent;
+	list_add(&realm->child_item, &parent->children);
+	return 1;
+}
+
+
+static int cmpu64_rev(const void *a, const void *b)
+{
+	if (*(u64 *)a < *(u64 *)b)
+		return 1;
+	if (*(u64 *)a > *(u64 *)b)
+		return -1;
+	return 0;
+}
+
+
+/*
+ * build the snap context for a given realm.
+ */
+static int build_snap_context(struct ceph_snap_realm *realm,
+			      struct list_head *realm_queue,
+			      struct list_head *dirty_realms)
+{
+	struct ceph_snap_realm *parent = realm->parent;
+	struct ceph_snap_context *snapc;
+	int err = 0;
+	u32 num = realm->num_prior_parent_snaps + realm->num_snaps;
+
+	/*
+	 * build parent context, if it hasn't been built.
+	 * conservatively estimate that all parent snaps might be
+	 * included by us.
+	 */
+	if (parent) {
+		if (!parent->cached_context) {
+			/* add to the queue head */
+			list_add(&parent->rebuild_item, realm_queue);
+			return 1;
+		}
+		num += parent->cached_context->num_snaps;
+	}
+
+	/* do i actually need to update?  not if my context seq
+	   matches realm seq, and my parents' does to.  (this works
+	   because we rebuild_snap_realms() works _downward_ in
+	   hierarchy after each update.) */
+	if (realm->cached_context &&
+	    realm->cached_context->seq == realm->seq &&
+	    (!parent ||
+	     realm->cached_context->seq >= parent->cached_context->seq)) {
+		dout("%s %llx %p: %p seq %lld (%u snaps) (unchanged)\n",
+		     __func__, realm->ino, realm, realm->cached_context,
+		     realm->cached_context->seq,
+		     (unsigned int)realm->cached_context->num_snaps);
+		return 0;
+	}
+
+	/* alloc new snap context */
+	err = -ENOMEM;
+	if (num > (SIZE_MAX - sizeof(*snapc)) / sizeof(u64))
+		goto fail;
+	snapc = ceph_create_snap_context(num, GFP_NOFS);
+	if (!snapc)
+		goto fail;
+
+	/* build (reverse sorted) snap vector */
+	num = 0;
+	snapc->seq = realm->seq;
+	if (parent) {
+		u32 i;
+
+		/* include any of parent's snaps occurring _after_ my
+		   parent became my parent */
+		for (i = 0; i < parent->cached_context->num_snaps; i++)
+			if (parent->cached_context->snaps[i] >=
+			    realm->parent_since)
+				snapc->snaps[num++] =
+					parent->cached_context->snaps[i];
+		if (parent->cached_context->seq > snapc->seq)
+			snapc->seq = parent->cached_context->seq;
+	}
+	memcpy(snapc->snaps + num, realm->snaps,
+	       sizeof(u64)*realm->num_snaps);
+	num += realm->num_snaps;
+	memcpy(snapc->snaps + num, realm->prior_parent_snaps,
+	       sizeof(u64)*realm->num_prior_parent_snaps);
+	num += realm->num_prior_parent_snaps;
+
+	sort(snapc->snaps, num, sizeof(u64), cmpu64_rev, NULL);
+	snapc->num_snaps = num;
+	dout("%s %llx %p: %p seq %lld (%u snaps)\n", __func__, realm->ino,
+	     realm, snapc, snapc->seq, (unsigned int) snapc->num_snaps);
+
+	ceph_put_snap_context(realm->cached_context);
+	realm->cached_context = snapc;
+	/* queue realm for cap_snap creation */
+	list_add_tail(&realm->dirty_item, dirty_realms);
+	return 0;
+
+fail:
+	/*
+	 * if we fail, clear old (incorrect) cached_context... hopefully
+	 * we'll have better luck building it later
+	 */
+	if (realm->cached_context) {
+		ceph_put_snap_context(realm->cached_context);
+		realm->cached_context = NULL;
+	}
+	pr_err("%s %llx %p fail %d\n", __func__, realm->ino, realm, err);
+	return err;
+}
+
+/*
+ * rebuild snap context for the given realm and all of its children.
+ */
+static void rebuild_snap_realms(struct ceph_snap_realm *realm,
+				struct list_head *dirty_realms)
+{
+	LIST_HEAD(realm_queue);
+	int last = 0;
+	bool skip = false;
+
+	list_add_tail(&realm->rebuild_item, &realm_queue);
+
+	while (!list_empty(&realm_queue)) {
+		struct ceph_snap_realm *_realm, *child;
+
+		_realm = list_first_entry(&realm_queue,
+					  struct ceph_snap_realm,
+					  rebuild_item);
+
+		/*
+		 * If the last building failed dues to memory
+		 * issue, just empty the realm_queue and return
+		 * to avoid infinite loop.
+		 */
+		if (last < 0) {
+			list_del_init(&_realm->rebuild_item);
+			continue;
+		}
+
+		last = build_snap_context(_realm, &realm_queue, dirty_realms);
+		dout("%s %llx %p, %s\n", __func__, _realm->ino, _realm,
+		     last > 0 ? "is deferred" : !last ? "succeeded" : "failed");
+
+		/* is any child in the list ? */
+		list_for_each_entry(child, &_realm->children, child_item) {
+			if (!list_empty(&child->rebuild_item)) {
+				skip = true;
+				break;
+			}
+		}
+
+		if (!skip) {
+			list_for_each_entry(child, &_realm->children, child_item)
+				list_add_tail(&child->rebuild_item, &realm_queue);
+		}
+
+		/* last == 1 means need to build parent first */
+		if (last <= 0)
+			list_del_init(&_realm->rebuild_item);
+	}
+}
+
+
+/*
+ * helper to allocate and decode an array of snapids.  free prior
+ * instance, if any.
+ */
+static int dup_array(u64 **dst, __le64 *src, u32 num)
+{
+	u32 i;
+
+	kfree(*dst);
+	if (num) {
+		*dst = kcalloc(num, sizeof(u64), GFP_NOFS);
+		if (!*dst)
+			return -ENOMEM;
+		for (i = 0; i < num; i++)
+			(*dst)[i] = get_unaligned_le64(src + i);
+	} else {
+		*dst = NULL;
+	}
+	return 0;
+}
+
+static bool has_new_snaps(struct ceph_snap_context *o,
+			  struct ceph_snap_context *n)
+{
+	if (n->num_snaps == 0)
+		return false;
+	/* snaps are in descending order */
+	return n->snaps[0] > o->seq;
+}
+
+/*
+ * When a snapshot is applied, the size/mtime inode metadata is queued
+ * in a ceph_cap_snap (one for each snapshot) until writeback
+ * completes and the metadata can be flushed back to the MDS.
+ *
+ * However, if a (sync) write is currently in-progress when we apply
+ * the snapshot, we have to wait until the write succeeds or fails
+ * (and a final size/mtime is known).  In this case the
+ * cap_snap->writing = 1, and is said to be "pending."  When the write
+ * finishes, we __ceph_finish_cap_snap().
+ *
+ * Caller must hold snap_rwsem for read (i.e., the realm topology won't
+ * change).
+ */
+static void ceph_queue_cap_snap(struct ceph_inode_info *ci,
+				struct ceph_cap_snap **pcapsnap)
+{
+	struct inode *inode = &ci->netfs.inode;
+	struct ceph_snap_context *old_snapc, *new_snapc;
+	struct ceph_cap_snap *capsnap = *pcapsnap;
+	struct ceph_buffer *old_blob = NULL;
+	int used, dirty;
+
+	spin_lock(&ci->i_ceph_lock);
+	used = __ceph_caps_used(ci);
+	dirty = __ceph_caps_dirty(ci);
+
+	old_snapc = ci->i_head_snapc;
+	new_snapc = ci->i_snap_realm->cached_context;
+
+	/*
+	 * If there is a write in progress, treat that as a dirty Fw,
+	 * even though it hasn't completed yet; by the time we finish
+	 * up this capsnap it will be.
+	 */
+	if (used & CEPH_CAP_FILE_WR)
+		dirty |= CEPH_CAP_FILE_WR;
+
+	if (__ceph_have_pending_cap_snap(ci)) {
+		/* there is no point in queuing multiple "pending" cap_snaps,
+		   as no new writes are allowed to start when pending, so any
+		   writes in progress now were started before the previous
+		   cap_snap.  lucky us. */
+		dout("%s %p %llx.%llx already pending\n",
+		     __func__, inode, ceph_vinop(inode));
+		goto update_snapc;
+	}
+	if (ci->i_wrbuffer_ref_head == 0 &&
+	    !(dirty & (CEPH_CAP_ANY_EXCL|CEPH_CAP_FILE_WR))) {
+		dout("%s %p %llx.%llx nothing dirty|writing\n",
+		     __func__, inode, ceph_vinop(inode));
+		goto update_snapc;
+	}
+
+	BUG_ON(!old_snapc);
+
+	/*
+	 * There is no need to send FLUSHSNAP message to MDS if there is
+	 * no new snapshot. But when there is dirty pages or on-going
+	 * writes, we still need to create cap_snap. cap_snap is needed
+	 * by the write path and page writeback path.
+	 *
+	 * also see ceph_try_drop_cap_snap()
+	 */
+	if (has_new_snaps(old_snapc, new_snapc)) {
+		if (dirty & (CEPH_CAP_ANY_EXCL|CEPH_CAP_FILE_WR))
+			capsnap->need_flush = true;
+	} else {
+		if (!(used & CEPH_CAP_FILE_WR) &&
+		    ci->i_wrbuffer_ref_head == 0) {
+			dout("%s %p %llx.%llx no new_snap|dirty_page|writing\n",
+			     __func__, inode, ceph_vinop(inode));
+			goto update_snapc;
+		}
+	}
+
+	dout("%s %p %llx.%llx cap_snap %p queuing under %p %s %s\n",
+	     __func__, inode, ceph_vinop(inode), capsnap, old_snapc,
+	     ceph_cap_string(dirty), capsnap->need_flush ? "" : "no_flush");
+	ihold(inode);
+
+	capsnap->follows = old_snapc->seq;
+	capsnap->issued = __ceph_caps_issued(ci, NULL);
+	capsnap->dirty = dirty;
+
+	capsnap->mode = inode->i_mode;
+	capsnap->uid = inode->i_uid;
+	capsnap->gid = inode->i_gid;
+
+	if (dirty & CEPH_CAP_XATTR_EXCL) {
+		old_blob = __ceph_build_xattrs_blob(ci);
+		capsnap->xattr_blob =
+			ceph_buffer_get(ci->i_xattrs.blob);
+		capsnap->xattr_version = ci->i_xattrs.version;
+	} else {
+		capsnap->xattr_blob = NULL;
+		capsnap->xattr_version = 0;
+	}
+
+	capsnap->inline_data = ci->i_inline_version != CEPH_INLINE_NONE;
+
+	/* dirty page count moved from _head to this cap_snap;
+	   all subsequent writes page dirties occur _after_ this
+	   snapshot. */
+	capsnap->dirty_pages = ci->i_wrbuffer_ref_head;
+	ci->i_wrbuffer_ref_head = 0;
+	capsnap->context = old_snapc;
+	list_add_tail(&capsnap->ci_item, &ci->i_cap_snaps);
+
+	if (used & CEPH_CAP_FILE_WR) {
+		dout("%s %p %llx.%llx cap_snap %p snapc %p seq %llu used WR,"
+		     " now pending\n", __func__, inode, ceph_vinop(inode),
+		     capsnap, old_snapc, old_snapc->seq);
+		capsnap->writing = 1;
+	} else {
+		/* note mtime, size NOW. */
+		__ceph_finish_cap_snap(ci, capsnap);
+	}
+	*pcapsnap = NULL;
+	old_snapc = NULL;
+
+update_snapc:
+	if (ci->i_wrbuffer_ref_head == 0 &&
+	    ci->i_wr_ref == 0 &&
+	    ci->i_dirty_caps == 0 &&
+	    ci->i_flushing_caps == 0) {
+		ci->i_head_snapc = NULL;
+	} else {
+		ci->i_head_snapc = ceph_get_snap_context(new_snapc);
+		dout(" new snapc is %p\n", new_snapc);
+	}
+	spin_unlock(&ci->i_ceph_lock);
+
+	ceph_buffer_put(old_blob);
+	ceph_put_snap_context(old_snapc);
+}
+
+/*
+ * Finalize the size, mtime for a cap_snap.. that is, settle on final values
+ * to be used for the snapshot, to be flushed back to the mds.
+ *
+ * If capsnap can now be flushed, add to snap_flush list, and return 1.
+ *
+ * Caller must hold i_ceph_lock.
+ */
+int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
+			    struct ceph_cap_snap *capsnap)
+{
+	struct inode *inode = &ci->netfs.inode;
+	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
+
+	BUG_ON(capsnap->writing);
+	capsnap->size = i_size_read(inode);
+	capsnap->mtime = inode->i_mtime;
+	capsnap->atime = inode->i_atime;
+	capsnap->ctime = inode->i_ctime;
+	capsnap->btime = ci->i_btime;
+	capsnap->change_attr = inode_peek_iversion_raw(inode);
+	capsnap->time_warp_seq = ci->i_time_warp_seq;
+	capsnap->truncate_size = ci->i_truncate_size;
+	capsnap->truncate_seq = ci->i_truncate_seq;
+	if (capsnap->dirty_pages) {
+		dout("%s %p %llx.%llx cap_snap %p snapc %p %llu %s s=%llu "
+		     "still has %d dirty pages\n", __func__, inode,
+		     ceph_vinop(inode), capsnap, capsnap->context,
+		     capsnap->context->seq, ceph_cap_string(capsnap->dirty),
+		     capsnap->size, capsnap->dirty_pages);
+		return 0;
+	}
+
+	/* Fb cap still in use, delay it */
+	if (ci->i_wb_ref) {
+		dout("%s %p %llx.%llx cap_snap %p snapc %p %llu %s s=%llu "
+		     "used WRBUFFER, delaying\n", __func__, inode,
+		     ceph_vinop(inode), capsnap, capsnap->context,
+		     capsnap->context->seq, ceph_cap_string(capsnap->dirty),
+		     capsnap->size);
+		capsnap->writing = 1;
+		return 0;
+	}
+
+	ci->i_ceph_flags |= CEPH_I_FLUSH_SNAPS;
+	dout("%s %p %llx.%llx cap_snap %p snapc %p %llu %s s=%llu\n",
+	     __func__, inode, ceph_vinop(inode), capsnap, capsnap->context,
+	     capsnap->context->seq, ceph_cap_string(capsnap->dirty),
+	     capsnap->size);
+
+	spin_lock(&mdsc->snap_flush_lock);
+	if (list_empty(&ci->i_snap_flush_item)) {
+		ihold(inode);
+		list_add_tail(&ci->i_snap_flush_item, &mdsc->snap_flush_list);
+	}
+	spin_unlock(&mdsc->snap_flush_lock);
+	return 1;  /* caller may want to ceph_flush_snaps */
+}
+
+/*
+ * Queue cap_snaps for snap writeback for this realm and its children.
+ * Called under snap_rwsem, so realm topology won't change.
+ */
+static void queue_realm_cap_snaps(struct ceph_snap_realm *realm)
+{
+	struct ceph_inode_info *ci;
+	struct inode *lastinode = NULL;
+	struct ceph_cap_snap *capsnap = NULL;
+
+	dout("%s %p %llx inode\n", __func__, realm, realm->ino);
+
+	spin_lock(&realm->inodes_with_caps_lock);
+	list_for_each_entry(ci, &realm->inodes_with_caps, i_snap_realm_item) {
+		struct inode *inode = igrab(&ci->netfs.inode);
+		if (!inode)
+			continue;
+		spin_unlock(&realm->inodes_with_caps_lock);
+		iput(lastinode);
+		lastinode = inode;
+
+		/*
+		 * Allocate the capsnap memory outside of ceph_queue_cap_snap()
+		 * to reduce very possible but unnecessary frequently memory
+		 * allocate/free in this loop.
+		 */
+		if (!capsnap) {
+			capsnap = kmem_cache_zalloc(ceph_cap_snap_cachep, GFP_NOFS);
+			if (!capsnap) {
+				pr_err("ENOMEM allocating ceph_cap_snap on %p\n",
+				       inode);
+				return;
+			}
+		}
+		capsnap->cap_flush.is_capsnap = true;
+		refcount_set(&capsnap->nref, 1);
+		INIT_LIST_HEAD(&capsnap->cap_flush.i_list);
+		INIT_LIST_HEAD(&capsnap->cap_flush.g_list);
+		INIT_LIST_HEAD(&capsnap->ci_item);
+
+		ceph_queue_cap_snap(ci, &capsnap);
+		spin_lock(&realm->inodes_with_caps_lock);
+	}
+	spin_unlock(&realm->inodes_with_caps_lock);
+	iput(lastinode);
+
+	if (capsnap)
+		kmem_cache_free(ceph_cap_snap_cachep, capsnap);
+	dout("%s %p %llx done\n", __func__, realm, realm->ino);
+}
+
+/*
+ * Parse and apply a snapblob "snap trace" from the MDS.  This specifies
+ * the snap realm parameters from a given realm and all of its ancestors,
+ * up to the root.
+ *
+ * Caller must hold snap_rwsem for write.
+ */
+int ceph_update_snap_trace(struct ceph_mds_client *mdsc,
+			   void *p, void *e, bool deletion,
+			   struct ceph_snap_realm **realm_ret)
+{
+	struct ceph_mds_snap_realm *ri;    /* encoded */
+	__le64 *snaps;                     /* encoded */
+	__le64 *prior_parent_snaps;        /* encoded */
+	struct ceph_snap_realm *realm;
+	struct ceph_snap_realm *first_realm = NULL;
+	struct ceph_snap_realm *realm_to_rebuild = NULL;
+	struct ceph_client *client = mdsc->fsc->client;
+	int rebuild_snapcs;
+	int err = -ENOMEM;
+	int ret;
+	LIST_HEAD(dirty_realms);
+
+	lockdep_assert_held_write(&mdsc->snap_rwsem);
+
+	dout("%s deletion=%d\n", __func__, deletion);
+more:
+	realm = NULL;
+	rebuild_snapcs = 0;
+	ceph_decode_need(&p, e, sizeof(*ri), bad);
+	ri = p;
+	p += sizeof(*ri);
+	ceph_decode_need(&p, e, sizeof(u64)*(le32_to_cpu(ri->num_snaps) +
+			    le32_to_cpu(ri->num_prior_parent_snaps)), bad);
+	snaps = p;
+	p += sizeof(u64) * le32_to_cpu(ri->num_snaps);
+	prior_parent_snaps = p;
+	p += sizeof(u64) * le32_to_cpu(ri->num_prior_parent_snaps);
+
+	realm = ceph_lookup_snap_realm(mdsc, le64_to_cpu(ri->ino));
+	if (!realm) {
+		realm = ceph_create_snap_realm(mdsc, le64_to_cpu(ri->ino));
+		if (IS_ERR(realm)) {
+			err = PTR_ERR(realm);
+			goto fail;
+		}
+	}
+
+	/* ensure the parent is correct */
+	err = adjust_snap_realm_parent(mdsc, realm, le64_to_cpu(ri->parent));
+	if (err < 0)
+		goto fail;
+	rebuild_snapcs += err;
+
+	if (le64_to_cpu(ri->seq) > realm->seq) {
+		dout("%s updating %llx %p %lld -> %lld\n", __func__,
+		     realm->ino, realm, realm->seq, le64_to_cpu(ri->seq));
+		/* update realm parameters, snap lists */
+		realm->seq = le64_to_cpu(ri->seq);
+		realm->created = le64_to_cpu(ri->created);
+		realm->parent_since = le64_to_cpu(ri->parent_since);
+
+		realm->num_snaps = le32_to_cpu(ri->num_snaps);
+		err = dup_array(&realm->snaps, snaps, realm->num_snaps);
+		if (err < 0)
+			goto fail;
+
+		realm->num_prior_parent_snaps =
+			le32_to_cpu(ri->num_prior_parent_snaps);
+		err = dup_array(&realm->prior_parent_snaps, prior_parent_snaps,
+				realm->num_prior_parent_snaps);
+		if (err < 0)
+			goto fail;
+
+		if (realm->seq > mdsc->last_snap_seq)
+			mdsc->last_snap_seq = realm->seq;
+
+		rebuild_snapcs = 1;
+	} else if (!realm->cached_context) {
+		dout("%s %llx %p seq %lld new\n", __func__,
+		     realm->ino, realm, realm->seq);
+		rebuild_snapcs = 1;
+	} else {
+		dout("%s %llx %p seq %lld unchanged\n", __func__,
+		     realm->ino, realm, realm->seq);
+	}
+
+	dout("done with %llx %p, rebuild_snapcs=%d, %p %p\n", realm->ino,
+	     realm, rebuild_snapcs, p, e);
+
+	/*
+	 * this will always track the uppest parent realm from which
+	 * we need to rebuild the snapshot contexts _downward_ in
+	 * hierarchy.
+	 */
+	if (rebuild_snapcs)
+		realm_to_rebuild = realm;
+
+	/* rebuild_snapcs when we reach the _end_ (root) of the trace */
+	if (realm_to_rebuild && p >= e)
+		rebuild_snap_realms(realm_to_rebuild, &dirty_realms);
+
+	if (!first_realm)
+		first_realm = realm;
+	else
+		ceph_put_snap_realm(mdsc, realm);
+
+	if (p < e)
+		goto more;
+
+	/*
+	 * queue cap snaps _after_ we've built the new snap contexts,
+	 * so that i_head_snapc can be set appropriately.
+	 */
+	while (!list_empty(&dirty_realms)) {
+		realm = list_first_entry(&dirty_realms, struct ceph_snap_realm,
+					 dirty_item);
+		list_del_init(&realm->dirty_item);
+		queue_realm_cap_snaps(realm);
+	}
+
+	if (realm_ret)
+		*realm_ret = first_realm;
+	else
+		ceph_put_snap_realm(mdsc, first_realm);
+
+	__cleanup_empty_realms(mdsc);
+	return 0;
+
+bad:
+	err = -EIO;
+fail:
+	if (realm && !IS_ERR(realm))
+		ceph_put_snap_realm(mdsc, realm);
+	if (first_realm)
+		ceph_put_snap_realm(mdsc, first_realm);
+	pr_err("%s error %d\n", __func__, err);
+
+	/*
+	 * When receiving a corrupted snap trace we don't know what
+	 * exactly has happened in MDS side. And we shouldn't continue
+	 * writing to OSD, which may corrupt the snapshot contents.
+	 *
+	 * Just try to blocklist this kclient and then this kclient
+	 * must be remounted to continue after the corrupted metadata
+	 * fixed in the MDS side.
+	 */
+	WRITE_ONCE(mdsc->fsc->mount_state, CEPH_MOUNT_FENCE_IO);
+	ret = ceph_monc_blocklist_add(&client->monc, &client->msgr.inst.addr);
+	if (ret)
+		pr_err("%s failed to blocklist %s: %d\n", __func__,
+		       ceph_pr_addr(&client->msgr.inst.addr), ret);
+
+	WARN(1, "%s: %s%sdo remount to continue%s",
+	     __func__, ret ? "" : ceph_pr_addr(&client->msgr.inst.addr),
+	     ret ? "" : " was blocklisted, ",
+	     err == -EIO ? " after corrupted snaptrace is fixed" : "");
+
+	return err;
+}
+
+
+/*
+ * Send any cap_snaps that are queued for flush.  Try to carry
+ * s_mutex across multiple snap flushes to avoid locking overhead.
+ *
+ * Caller holds no locks.
+ */
+static void flush_snaps(struct ceph_mds_client *mdsc)
+{
+	struct ceph_inode_info *ci;
+	struct inode *inode;
+	struct ceph_mds_session *session = NULL;
+
+	dout("%s\n", __func__);
+	spin_lock(&mdsc->snap_flush_lock);
+	while (!list_empty(&mdsc->snap_flush_list)) {
+		ci = list_first_entry(&mdsc->snap_flush_list,
+				struct ceph_inode_info, i_snap_flush_item);
+		inode = &ci->netfs.inode;
+		ihold(inode);
+		spin_unlock(&mdsc->snap_flush_lock);
+		ceph_flush_snaps(ci, &session);
+		iput(inode);
+		spin_lock(&mdsc->snap_flush_lock);
+	}
+	spin_unlock(&mdsc->snap_flush_lock);
+
+	ceph_put_mds_session(session);
+	dout("%s done\n", __func__);
+}
+
+/**
+ * ceph_change_snap_realm - change the snap_realm for an inode
+ * @inode: inode to move to new snap realm
+ * @realm: new realm to move inode into (may be NULL)
+ *
+ * Detach an inode from its old snaprealm (if any) and attach it to
+ * the new snaprealm (if any). The old snap realm reference held by
+ * the inode is put. If realm is non-NULL, then the caller's reference
+ * to it is taken over by the inode.
+ */
+void ceph_change_snap_realm(struct inode *inode, struct ceph_snap_realm *realm)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;
+	struct ceph_snap_realm *oldrealm = ci->i_snap_realm;
+
+	lockdep_assert_held(&ci->i_ceph_lock);
+
+	if (oldrealm) {
+		spin_lock(&oldrealm->inodes_with_caps_lock);
+		list_del_init(&ci->i_snap_realm_item);
+		if (oldrealm->ino == ci->i_vino.ino)
+			oldrealm->inode = NULL;
+		spin_unlock(&oldrealm->inodes_with_caps_lock);
+		ceph_put_snap_realm(mdsc, oldrealm);
+	}
+
+	ci->i_snap_realm = realm;
+
+	if (realm) {
+		spin_lock(&realm->inodes_with_caps_lock);
+		list_add(&ci->i_snap_realm_item, &realm->inodes_with_caps);
+		if (realm->ino == ci->i_vino.ino)
+			realm->inode = inode;
+		spin_unlock(&realm->inodes_with_caps_lock);
+	}
+}
+
+/*
+ * Handle a snap notification from the MDS.
+ *
+ * This can take two basic forms: the simplest is just a snap creation
+ * or deletion notification on an existing realm.  This should update the
+ * realm and its children.
+ *
+ * The more difficult case is realm creation, due to snap creation at a
+ * new point in the file hierarchy, or due to a rename that moves a file or
+ * directory into another realm.
+ */
+void ceph_handle_snap(struct ceph_mds_client *mdsc,
+		      struct ceph_mds_session *session,
+		      struct ceph_msg *msg)
+{
+	struct super_block *sb = mdsc->fsc->sb;
+	int mds = session->s_mds;
+	u64 split;
+	int op;
+	int trace_len;
+	struct ceph_snap_realm *realm = NULL;
+	void *p = msg->front.iov_base;
+	void *e = p + msg->front.iov_len;
+	struct ceph_mds_snap_head *h;
+	int num_split_inos, num_split_realms;
+	__le64 *split_inos = NULL, *split_realms = NULL;
+	int i;
+	int locked_rwsem = 0;
+	bool close_sessions = false;
+
+	if (!ceph_inc_mds_stopping_blocker(mdsc, session))
+		return;
+
+	/* decode */
+	if (msg->front.iov_len < sizeof(*h))
+		goto bad;
+	h = p;
+	op = le32_to_cpu(h->op);
+	split = le64_to_cpu(h->split);   /* non-zero if we are splitting an
+					  * existing realm */
+	num_split_inos = le32_to_cpu(h->num_split_inos);
+	num_split_realms = le32_to_cpu(h->num_split_realms);
+	trace_len = le32_to_cpu(h->trace_len);
+	p += sizeof(*h);
+
+	dout("%s from mds%d op %s split %llx tracelen %d\n", __func__,
+	     mds, ceph_snap_op_name(op), split, trace_len);
+
+	down_write(&mdsc->snap_rwsem);
+	locked_rwsem = 1;
+
+	if (op == CEPH_SNAP_OP_SPLIT) {
+		struct ceph_mds_snap_realm *ri;
+
+		/*
+		 * A "split" breaks part of an existing realm off into
+		 * a new realm.  The MDS provides a list of inodes
+		 * (with caps) and child realms that belong to the new
+		 * child.
+		 */
+		split_inos = p;
+		p += sizeof(u64) * num_split_inos;
+		split_realms = p;
+		p += sizeof(u64) * num_split_realms;
+		ceph_decode_need(&p, e, sizeof(*ri), bad);
+		/* we will peek at realm info here, but will _not_
+		 * advance p, as the realm update will occur below in
+		 * ceph_update_snap_trace. */
+		ri = p;
+
+		realm = ceph_lookup_snap_realm(mdsc, split);
+		if (!realm) {
+			realm = ceph_create_snap_realm(mdsc, split);
+			if (IS_ERR(realm))
+				goto out;
+		}
+
+		dout("splitting snap_realm %llx %p\n", realm->ino, realm);
+		for (i = 0; i < num_split_inos; i++) {
+			struct ceph_vino vino = {
+				.ino = le64_to_cpu(split_inos[i]),
+				.snap = CEPH_NOSNAP,
+			};
+			struct inode *inode = ceph_find_inode(sb, vino);
+			struct ceph_inode_info *ci;
+
+			if (!inode)
+				continue;
+			ci = ceph_inode(inode);
+
+			spin_lock(&ci->i_ceph_lock);
+			if (!ci->i_snap_realm)
+				goto skip_inode;
+			/*
+			 * If this inode belongs to a realm that was
+			 * created after our new realm, we experienced
+			 * a race (due to another split notifications
+			 * arriving from a different MDS).  So skip
+			 * this inode.
+			 */
+			if (ci->i_snap_realm->created >
+			    le64_to_cpu(ri->created)) {
+				dout(" leaving %p %llx.%llx in newer realm %llx %p\n",
+				     inode, ceph_vinop(inode), ci->i_snap_realm->ino,
+				     ci->i_snap_realm);
+				goto skip_inode;
+			}
+			dout(" will move %p %llx.%llx to split realm %llx %p\n",
+			     inode, ceph_vinop(inode), realm->ino, realm);
+
+			ceph_get_snap_realm(mdsc, realm);
+			ceph_change_snap_realm(inode, realm);
+			spin_unlock(&ci->i_ceph_lock);
+			iput(inode);
+			continue;
+
+skip_inode:
+			spin_unlock(&ci->i_ceph_lock);
+			iput(inode);
+		}
+
+		/* we may have taken some of the old realm's children. */
+		for (i = 0; i < num_split_realms; i++) {
+			struct ceph_snap_realm *child =
+				__lookup_snap_realm(mdsc,
+					   le64_to_cpu(split_realms[i]));
+			if (!child)
+				continue;
+			adjust_snap_realm_parent(mdsc, child, realm->ino);
+		}
+	} else {
+		/*
+		 * In the non-split case both 'num_split_inos' and
+		 * 'num_split_realms' should be 0, making this a no-op.
+		 * However the MDS happens to populate 'split_realms' list
+		 * in one of the UPDATE op cases by mistake.
+		 *
+		 * Skip both lists just in case to ensure that 'p' is
+		 * positioned at the start of realm info, as expected by
+		 * ceph_update_snap_trace().
+		 */
+		p += sizeof(u64) * num_split_inos;
+		p += sizeof(u64) * num_split_realms;
+	}
+
+	/*
+	 * update using the provided snap trace. if we are deleting a
+	 * snap, we can avoid queueing cap_snaps.
+	 */
+	if (ceph_update_snap_trace(mdsc, p, e,
+				   op == CEPH_SNAP_OP_DESTROY,
+				   NULL)) {
+		close_sessions = true;
+		goto bad;
+	}
+
+	if (op == CEPH_SNAP_OP_SPLIT)
+		/* we took a reference when we created the realm, above */
+		ceph_put_snap_realm(mdsc, realm);
+
+	__cleanup_empty_realms(mdsc);
+
+	up_write(&mdsc->snap_rwsem);
+
+	flush_snaps(mdsc);
+	ceph_dec_mds_stopping_blocker(mdsc);
+	return;
+
+bad:
+	pr_err("%s corrupt snap message from mds%d\n", __func__, mds);
+	ceph_msg_dump(msg);
+out:
+	if (locked_rwsem)
+		up_write(&mdsc->snap_rwsem);
+
+	ceph_dec_mds_stopping_blocker(mdsc);
+
+	if (close_sessions)
+		ceph_mdsc_close_sessions(mdsc);
+	return;
+}
+
+struct ceph_snapid_map* ceph_get_snapid_map(struct ceph_mds_client *mdsc,
+					    u64 snap)
+{
+	struct ceph_snapid_map *sm, *exist;
+	struct rb_node **p, *parent;
+	int ret;
+
+	exist = NULL;
+	spin_lock(&mdsc->snapid_map_lock);
+	p = &mdsc->snapid_map_tree.rb_node;
+	while (*p) {
+		exist = rb_entry(*p, struct ceph_snapid_map, node);
+		if (snap > exist->snap) {
+			p = &(*p)->rb_left;
+		} else if (snap < exist->snap) {
+			p = &(*p)->rb_right;
+		} else {
+			if (atomic_inc_return(&exist->ref) == 1)
+				list_del_init(&exist->lru);
+			break;
+		}
+		exist = NULL;
+	}
+	spin_unlock(&mdsc->snapid_map_lock);
+	if (exist) {
+		dout("%s found snapid map %llx -> %x\n", __func__,
+		     exist->snap, exist->dev);
+		return exist;
+	}
+
+	sm = kmalloc(sizeof(*sm), GFP_NOFS);
+	if (!sm)
+		return NULL;
+
+	ret = get_anon_bdev(&sm->dev);
+	if (ret < 0) {
+		kfree(sm);
+		return NULL;
+	}
+
+	INIT_LIST_HEAD(&sm->lru);
+	atomic_set(&sm->ref, 1);
+	sm->snap = snap;
+
+	exist = NULL;
+	parent = NULL;
+	p = &mdsc->snapid_map_tree.rb_node;
+	spin_lock(&mdsc->snapid_map_lock);
+	while (*p) {
+		parent = *p;
+		exist = rb_entry(*p, struct ceph_snapid_map, node);
+		if (snap > exist->snap)
+			p = &(*p)->rb_left;
+		else if (snap < exist->snap)
+			p = &(*p)->rb_right;
+		else
+			break;
+		exist = NULL;
+	}
+	if (exist) {
+		if (atomic_inc_return(&exist->ref) == 1)
+			list_del_init(&exist->lru);
+	} else {
+		rb_link_node(&sm->node, parent, p);
+		rb_insert_color(&sm->node, &mdsc->snapid_map_tree);
+	}
+	spin_unlock(&mdsc->snapid_map_lock);
+	if (exist) {
+		free_anon_bdev(sm->dev);
+		kfree(sm);
+		dout("%s found snapid map %llx -> %x\n", __func__,
+		     exist->snap, exist->dev);
+		return exist;
+	}
+
+	dout("%s create snapid map %llx -> %x\n", __func__,
+	     sm->snap, sm->dev);
+	return sm;
+}
+
+void ceph_put_snapid_map(struct ceph_mds_client* mdsc,
+			 struct ceph_snapid_map *sm)
+{
+	if (!sm)
+		return;
+	if (atomic_dec_and_lock(&sm->ref, &mdsc->snapid_map_lock)) {
+		if (!RB_EMPTY_NODE(&sm->node)) {
+			sm->last_used = jiffies;
+			list_add_tail(&sm->lru, &mdsc->snapid_map_lru);
+			spin_unlock(&mdsc->snapid_map_lock);
+		} else {
+			/* already cleaned up by
+			 * ceph_cleanup_snapid_map() */
+			spin_unlock(&mdsc->snapid_map_lock);
+			kfree(sm);
+		}
+	}
+}
+
+void ceph_trim_snapid_map(struct ceph_mds_client *mdsc)
+{
+	struct ceph_snapid_map *sm;
+	unsigned long now;
+	LIST_HEAD(to_free);
+
+	spin_lock(&mdsc->snapid_map_lock);
+	now = jiffies;
+
+	while (!list_empty(&mdsc->snapid_map_lru)) {
+		sm = list_first_entry(&mdsc->snapid_map_lru,
+				      struct ceph_snapid_map, lru);
+		if (time_after(sm->last_used + CEPH_SNAPID_MAP_TIMEOUT, now))
+			break;
+
+		rb_erase(&sm->node, &mdsc->snapid_map_tree);
+		list_move(&sm->lru, &to_free);
+	}
+	spin_unlock(&mdsc->snapid_map_lock);
+
+	while (!list_empty(&to_free)) {
+		sm = list_first_entry(&to_free, struct ceph_snapid_map, lru);
+		list_del(&sm->lru);
+		dout("trim snapid map %llx -> %x\n", sm->snap, sm->dev);
+		free_anon_bdev(sm->dev);
+		kfree(sm);
+	}
+}
+
+void ceph_cleanup_snapid_map(struct ceph_mds_client *mdsc)
+{
+	struct ceph_snapid_map *sm;
+	struct rb_node *p;
+	LIST_HEAD(to_free);
+
+	spin_lock(&mdsc->snapid_map_lock);
+	while ((p = rb_first(&mdsc->snapid_map_tree))) {
+		sm = rb_entry(p, struct ceph_snapid_map, node);
+		rb_erase(p, &mdsc->snapid_map_tree);
+		RB_CLEAR_NODE(p);
+		list_move(&sm->lru, &to_free);
+	}
+	spin_unlock(&mdsc->snapid_map_lock);
+
+	while (!list_empty(&to_free)) {
+		sm = list_first_entry(&to_free, struct ceph_snapid_map, lru);
+		list_del(&sm->lru);
+		free_anon_bdev(sm->dev);
+		if (WARN_ON_ONCE(atomic_read(&sm->ref))) {
+			pr_err("snapid map %llx -> %x still in use\n",
+			       sm->snap, sm->dev);
+		}
+		kfree(sm);
+	}
+}
diff --git a/fs/ceph/strings.c b/fs/ceph/strings.c
new file mode 100644
index 000000000..e36e8948e
--- /dev/null
+++ b/fs/ceph/strings.c
@@ -0,0 +1,130 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Ceph fs string constants
+ */
+#include <linux/module.h>
+#include <linux/ceph/types.h>
+
+
+const char *ceph_mds_state_name(int s)
+{
+	switch (s) {
+		/* down and out */
+	case CEPH_MDS_STATE_DNE:        return "down:dne";
+	case CEPH_MDS_STATE_STOPPED:    return "down:stopped";
+		/* up and out */
+	case CEPH_MDS_STATE_BOOT:       return "up:boot";
+	case CEPH_MDS_STATE_STANDBY:    return "up:standby";
+	case CEPH_MDS_STATE_STANDBY_REPLAY:    return "up:standby-replay";
+	case CEPH_MDS_STATE_REPLAYONCE: return "up:oneshot-replay";
+	case CEPH_MDS_STATE_CREATING:   return "up:creating";
+	case CEPH_MDS_STATE_STARTING:   return "up:starting";
+		/* up and in */
+	case CEPH_MDS_STATE_REPLAY:     return "up:replay";
+	case CEPH_MDS_STATE_RESOLVE:    return "up:resolve";
+	case CEPH_MDS_STATE_RECONNECT:  return "up:reconnect";
+	case CEPH_MDS_STATE_REJOIN:     return "up:rejoin";
+	case CEPH_MDS_STATE_CLIENTREPLAY: return "up:clientreplay";
+	case CEPH_MDS_STATE_ACTIVE:     return "up:active";
+	case CEPH_MDS_STATE_STOPPING:   return "up:stopping";
+	}
+	return "???";
+}
+
+const char *ceph_session_op_name(int op)
+{
+	switch (op) {
+	case CEPH_SESSION_REQUEST_OPEN: return "request_open";
+	case CEPH_SESSION_OPEN: return "open";
+	case CEPH_SESSION_REQUEST_CLOSE: return "request_close";
+	case CEPH_SESSION_CLOSE: return "close";
+	case CEPH_SESSION_REQUEST_RENEWCAPS: return "request_renewcaps";
+	case CEPH_SESSION_RENEWCAPS: return "renewcaps";
+	case CEPH_SESSION_STALE: return "stale";
+	case CEPH_SESSION_RECALL_STATE: return "recall_state";
+	case CEPH_SESSION_FLUSHMSG: return "flushmsg";
+	case CEPH_SESSION_FLUSHMSG_ACK: return "flushmsg_ack";
+	case CEPH_SESSION_FORCE_RO: return "force_ro";
+	case CEPH_SESSION_REJECT: return "reject";
+	case CEPH_SESSION_REQUEST_FLUSH_MDLOG: return "flush_mdlog";
+	}
+	return "???";
+}
+
+const char *ceph_mds_op_name(int op)
+{
+	switch (op) {
+	case CEPH_MDS_OP_LOOKUP:  return "lookup";
+	case CEPH_MDS_OP_LOOKUPHASH:  return "lookuphash";
+	case CEPH_MDS_OP_LOOKUPPARENT:  return "lookupparent";
+	case CEPH_MDS_OP_LOOKUPINO:  return "lookupino";
+	case CEPH_MDS_OP_LOOKUPNAME:  return "lookupname";
+	case CEPH_MDS_OP_GETATTR:  return "getattr";
+	case CEPH_MDS_OP_GETVXATTR:  return "getvxattr";
+	case CEPH_MDS_OP_SETXATTR: return "setxattr";
+	case CEPH_MDS_OP_SETATTR: return "setattr";
+	case CEPH_MDS_OP_RMXATTR: return "rmxattr";
+	case CEPH_MDS_OP_SETLAYOUT: return "setlayou";
+	case CEPH_MDS_OP_SETDIRLAYOUT: return "setdirlayout";
+	case CEPH_MDS_OP_READDIR: return "readdir";
+	case CEPH_MDS_OP_MKNOD: return "mknod";
+	case CEPH_MDS_OP_LINK: return "link";
+	case CEPH_MDS_OP_UNLINK: return "unlink";
+	case CEPH_MDS_OP_RENAME: return "rename";
+	case CEPH_MDS_OP_MKDIR: return "mkdir";
+	case CEPH_MDS_OP_RMDIR: return "rmdir";
+	case CEPH_MDS_OP_SYMLINK: return "symlink";
+	case CEPH_MDS_OP_CREATE: return "create";
+	case CEPH_MDS_OP_OPEN: return "open";
+	case CEPH_MDS_OP_LOOKUPSNAP: return "lookupsnap";
+	case CEPH_MDS_OP_LSSNAP: return "lssnap";
+	case CEPH_MDS_OP_MKSNAP: return "mksnap";
+	case CEPH_MDS_OP_RMSNAP: return "rmsnap";
+	case CEPH_MDS_OP_RENAMESNAP: return "renamesnap";
+	case CEPH_MDS_OP_SETFILELOCK: return "setfilelock";
+	case CEPH_MDS_OP_GETFILELOCK: return "getfilelock";
+	}
+	return "???";
+}
+
+const char *ceph_cap_op_name(int op)
+{
+	switch (op) {
+	case CEPH_CAP_OP_GRANT: return "grant";
+	case CEPH_CAP_OP_REVOKE: return "revoke";
+	case CEPH_CAP_OP_TRUNC: return "trunc";
+	case CEPH_CAP_OP_EXPORT: return "export";
+	case CEPH_CAP_OP_IMPORT: return "import";
+	case CEPH_CAP_OP_UPDATE: return "update";
+	case CEPH_CAP_OP_DROP: return "drop";
+	case CEPH_CAP_OP_FLUSH: return "flush";
+	case CEPH_CAP_OP_FLUSH_ACK: return "flush_ack";
+	case CEPH_CAP_OP_FLUSHSNAP: return "flushsnap";
+	case CEPH_CAP_OP_FLUSHSNAP_ACK: return "flushsnap_ack";
+	case CEPH_CAP_OP_RELEASE: return "release";
+	case CEPH_CAP_OP_RENEW: return "renew";
+	}
+	return "???";
+}
+
+const char *ceph_lease_op_name(int o)
+{
+	switch (o) {
+	case CEPH_MDS_LEASE_REVOKE: return "revoke";
+	case CEPH_MDS_LEASE_RELEASE: return "release";
+	case CEPH_MDS_LEASE_RENEW: return "renew";
+	case CEPH_MDS_LEASE_REVOKE_ACK: return "revoke_ack";
+	}
+	return "???";
+}
+
+const char *ceph_snap_op_name(int o)
+{
+	switch (o) {
+	case CEPH_SNAP_OP_UPDATE: return "update";
+	case CEPH_SNAP_OP_CREATE: return "create";
+	case CEPH_SNAP_OP_DESTROY: return "destroy";
+	case CEPH_SNAP_OP_SPLIT: return "split";
+	}
+	return "???";
+}
diff --git a/fs/ceph/super.c b/fs/ceph/super.c
new file mode 100644
index 000000000..281b493fd
--- /dev/null
+++ b/fs/ceph/super.c
@@ -0,0 +1,1573 @@
+// SPDX-License-Identifier: GPL-2.0-only
+
+#include <linux/ceph/ceph_debug.h>
+
+#include <linux/backing-dev.h>
+#include <linux/ctype.h>
+#include <linux/fs.h>
+#include <linux/inet.h>
+#include <linux/in6.h>
+#include <linux/module.h>
+#include <linux/mount.h>
+#include <linux/fs_context.h>
+#include <linux/fs_parser.h>
+#include <linux/sched.h>
+#include <linux/seq_file.h>
+#include <linux/slab.h>
+#include <linux/statfs.h>
+#include <linux/string.h>
+
+#include "super.h"
+#include "mds_client.h"
+#include "cache.h"
+
+#include <linux/ceph/ceph_features.h>
+#include <linux/ceph/decode.h>
+#include <linux/ceph/mon_client.h>
+#include <linux/ceph/auth.h>
+#include <linux/ceph/debugfs.h>
+
+#include <uapi/linux/magic.h>
+
+static DEFINE_SPINLOCK(ceph_fsc_lock);
+static LIST_HEAD(ceph_fsc_list);
+
+/*
+ * Ceph superblock operations
+ *
+ * Handle the basics of mounting, unmounting.
+ */
+
+/*
+ * super ops
+ */
+static void ceph_put_super(struct super_block *s)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(s);
+
+	dout("put_super\n");
+	ceph_mdsc_close_sessions(fsc->mdsc);
+}
+
+static int ceph_statfs(struct dentry *dentry, struct kstatfs *buf)
+{
+	struct ceph_fs_client *fsc = ceph_inode_to_client(d_inode(dentry));
+	struct ceph_mon_client *monc = &fsc->client->monc;
+	struct ceph_statfs st;
+	int i, err;
+	u64 data_pool;
+
+	if (fsc->mdsc->mdsmap->m_num_data_pg_pools == 1) {
+		data_pool = fsc->mdsc->mdsmap->m_data_pg_pools[0];
+	} else {
+		data_pool = CEPH_NOPOOL;
+	}
+
+	dout("statfs\n");
+	err = ceph_monc_do_statfs(monc, data_pool, &st);
+	if (err < 0)
+		return err;
+
+	/* fill in kstatfs */
+	buf->f_type = CEPH_SUPER_MAGIC;  /* ?? */
+
+	/*
+	 * Express utilization in terms of large blocks to avoid
+	 * overflow on 32-bit machines.
+	 */
+	buf->f_frsize = 1 << CEPH_BLOCK_SHIFT;
+
+	/*
+	 * By default use root quota for stats; fallback to overall filesystem
+	 * usage if using 'noquotadf' mount option or if the root dir doesn't
+	 * have max_bytes quota set.
+	 */
+	if (ceph_test_mount_opt(fsc, NOQUOTADF) ||
+	    !ceph_quota_update_statfs(fsc, buf)) {
+		buf->f_blocks = le64_to_cpu(st.kb) >> (CEPH_BLOCK_SHIFT-10);
+		buf->f_bfree = le64_to_cpu(st.kb_avail) >> (CEPH_BLOCK_SHIFT-10);
+		buf->f_bavail = le64_to_cpu(st.kb_avail) >> (CEPH_BLOCK_SHIFT-10);
+	}
+
+	/*
+	 * NOTE: for the time being, we make bsize == frsize to humor
+	 * not-yet-ancient versions of glibc that are broken.
+	 * Someday, we will probably want to report a real block
+	 * size...  whatever that may mean for a network file system!
+	 */
+	buf->f_bsize = buf->f_frsize;
+
+	buf->f_files = le64_to_cpu(st.num_objects);
+	buf->f_ffree = -1;
+	buf->f_namelen = NAME_MAX;
+
+	/* Must convert the fsid, for consistent values across arches */
+	buf->f_fsid.val[0] = 0;
+	mutex_lock(&monc->mutex);
+	for (i = 0 ; i < sizeof(monc->monmap->fsid) / sizeof(__le32) ; ++i)
+		buf->f_fsid.val[0] ^= le32_to_cpu(((__le32 *)&monc->monmap->fsid)[i]);
+	mutex_unlock(&monc->mutex);
+
+	/* fold the fs_cluster_id into the upper bits */
+	buf->f_fsid.val[1] = monc->fs_cluster_id;
+
+	return 0;
+}
+
+static int ceph_sync_fs(struct super_block *sb, int wait)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(sb);
+
+	if (!wait) {
+		dout("sync_fs (non-blocking)\n");
+		ceph_flush_dirty_caps(fsc->mdsc);
+		dout("sync_fs (non-blocking) done\n");
+		return 0;
+	}
+
+	dout("sync_fs (blocking)\n");
+	ceph_osdc_sync(&fsc->client->osdc);
+	ceph_mdsc_sync(fsc->mdsc);
+	dout("sync_fs (blocking) done\n");
+	return 0;
+}
+
+/*
+ * mount options
+ */
+enum {
+	Opt_wsize,
+	Opt_rsize,
+	Opt_rasize,
+	Opt_caps_wanted_delay_min,
+	Opt_caps_wanted_delay_max,
+	Opt_caps_max,
+	Opt_readdir_max_entries,
+	Opt_readdir_max_bytes,
+	Opt_congestion_kb,
+	/* int args above */
+	Opt_snapdirname,
+	Opt_mds_namespace,
+	Opt_recover_session,
+	Opt_source,
+	Opt_mon_addr,
+	/* string args above */
+	Opt_dirstat,
+	Opt_rbytes,
+	Opt_asyncreaddir,
+	Opt_dcache,
+	Opt_ino32,
+	Opt_fscache,
+	Opt_poolperm,
+	Opt_require_active_mds,
+	Opt_acl,
+	Opt_quotadf,
+	Opt_copyfrom,
+	Opt_wsync,
+	Opt_pagecache,
+};
+
+enum ceph_recover_session_mode {
+	ceph_recover_session_no,
+	ceph_recover_session_clean
+};
+
+static const struct constant_table ceph_param_recover[] = {
+	{ "no",		ceph_recover_session_no },
+	{ "clean",	ceph_recover_session_clean },
+	{}
+};
+
+static const struct fs_parameter_spec ceph_mount_parameters[] = {
+	fsparam_flag_no ("acl",				Opt_acl),
+	fsparam_flag_no ("asyncreaddir",		Opt_asyncreaddir),
+	fsparam_s32	("caps_max",			Opt_caps_max),
+	fsparam_u32	("caps_wanted_delay_max",	Opt_caps_wanted_delay_max),
+	fsparam_u32	("caps_wanted_delay_min",	Opt_caps_wanted_delay_min),
+	fsparam_u32	("write_congestion_kb",		Opt_congestion_kb),
+	fsparam_flag_no ("copyfrom",			Opt_copyfrom),
+	fsparam_flag_no ("dcache",			Opt_dcache),
+	fsparam_flag_no ("dirstat",			Opt_dirstat),
+	fsparam_flag_no	("fsc",				Opt_fscache), // fsc|nofsc
+	fsparam_string	("fsc",				Opt_fscache), // fsc=...
+	fsparam_flag_no ("ino32",			Opt_ino32),
+	fsparam_string	("mds_namespace",		Opt_mds_namespace),
+	fsparam_flag_no ("poolperm",			Opt_poolperm),
+	fsparam_flag_no ("quotadf",			Opt_quotadf),
+	fsparam_u32	("rasize",			Opt_rasize),
+	fsparam_flag_no ("rbytes",			Opt_rbytes),
+	fsparam_u32	("readdir_max_bytes",		Opt_readdir_max_bytes),
+	fsparam_u32	("readdir_max_entries",		Opt_readdir_max_entries),
+	fsparam_enum	("recover_session",		Opt_recover_session, ceph_param_recover),
+	fsparam_flag_no ("require_active_mds",		Opt_require_active_mds),
+	fsparam_u32	("rsize",			Opt_rsize),
+	fsparam_string	("snapdirname",			Opt_snapdirname),
+	fsparam_string	("source",			Opt_source),
+	fsparam_string	("mon_addr",			Opt_mon_addr),
+	fsparam_u32	("wsize",			Opt_wsize),
+	fsparam_flag_no	("wsync",			Opt_wsync),
+	fsparam_flag_no	("pagecache",			Opt_pagecache),
+	{}
+};
+
+struct ceph_parse_opts_ctx {
+	struct ceph_options		*copts;
+	struct ceph_mount_options	*opts;
+};
+
+/*
+ * Remove adjacent slashes and then the trailing slash, unless it is
+ * the only remaining character.
+ *
+ * E.g. "//dir1////dir2///" --> "/dir1/dir2", "///" --> "/".
+ */
+static void canonicalize_path(char *path)
+{
+	int i, j = 0;
+
+	for (i = 0; path[i] != '\0'; i++) {
+		if (path[i] != '/' || j < 1 || path[j - 1] != '/')
+			path[j++] = path[i];
+	}
+
+	if (j > 1 && path[j - 1] == '/')
+		j--;
+	path[j] = '\0';
+}
+
+/*
+ * Check if the mds namespace in ceph_mount_options matches
+ * the passed in namespace string. First time match (when
+ * ->mds_namespace is NULL) is treated specially, since
+ * ->mds_namespace needs to be initialized by the caller.
+ */
+static int namespace_equals(struct ceph_mount_options *fsopt,
+			    const char *namespace, size_t len)
+{
+	return !(fsopt->mds_namespace &&
+		 (strlen(fsopt->mds_namespace) != len ||
+		  strncmp(fsopt->mds_namespace, namespace, len)));
+}
+
+static int ceph_parse_old_source(const char *dev_name, const char *dev_name_end,
+				 struct fs_context *fc)
+{
+	int r;
+	struct ceph_parse_opts_ctx *pctx = fc->fs_private;
+	struct ceph_mount_options *fsopt = pctx->opts;
+
+	if (*dev_name_end != ':')
+		return invalfc(fc, "separator ':' missing in source");
+
+	r = ceph_parse_mon_ips(dev_name, dev_name_end - dev_name,
+			       pctx->copts, fc->log.log, ',');
+	if (r)
+		return r;
+
+	fsopt->new_dev_syntax = false;
+	return 0;
+}
+
+static int ceph_parse_new_source(const char *dev_name, const char *dev_name_end,
+				 struct fs_context *fc)
+{
+	size_t len;
+	struct ceph_fsid fsid;
+	struct ceph_parse_opts_ctx *pctx = fc->fs_private;
+	struct ceph_mount_options *fsopt = pctx->opts;
+	char *fsid_start, *fs_name_start;
+
+	if (*dev_name_end != '=') {
+		dout("separator '=' missing in source");
+		return -EINVAL;
+	}
+
+	fsid_start = strchr(dev_name, '@');
+	if (!fsid_start)
+		return invalfc(fc, "missing cluster fsid");
+	++fsid_start; /* start of cluster fsid */
+
+	fs_name_start = strchr(fsid_start, '.');
+	if (!fs_name_start)
+		return invalfc(fc, "missing file system name");
+
+	if (ceph_parse_fsid(fsid_start, &fsid))
+		return invalfc(fc, "Invalid FSID");
+
+	++fs_name_start; /* start of file system name */
+	len = dev_name_end - fs_name_start;
+
+	if (!namespace_equals(fsopt, fs_name_start, len))
+		return invalfc(fc, "Mismatching mds_namespace");
+	kfree(fsopt->mds_namespace);
+	fsopt->mds_namespace = kstrndup(fs_name_start, len, GFP_KERNEL);
+	if (!fsopt->mds_namespace)
+		return -ENOMEM;
+	dout("file system (mds namespace) '%s'\n", fsopt->mds_namespace);
+
+	fsopt->new_dev_syntax = true;
+	return 0;
+}
+
+/*
+ * Parse the source parameter for new device format. Distinguish the device
+ * spec from the path. Try parsing new device format and fallback to old
+ * format if needed.
+ *
+ * New device syntax will looks like:
+ *     <device_spec>=/<path>
+ * where
+ *     <device_spec> is name@fsid.fsname
+ *     <path> is optional, but if present must begin with '/'
+ * (monitor addresses are passed via mount option)
+ *
+ * Old device syntax is:
+ *     <server_spec>[,<server_spec>...]:[<path>]
+ * where
+ *     <server_spec> is <ip>[:<port>]
+ *     <path> is optional, but if present must begin with '/'
+ */
+static int ceph_parse_source(struct fs_parameter *param, struct fs_context *fc)
+{
+	struct ceph_parse_opts_ctx *pctx = fc->fs_private;
+	struct ceph_mount_options *fsopt = pctx->opts;
+	char *dev_name = param->string, *dev_name_end;
+	int ret;
+
+	dout("%s '%s'\n", __func__, dev_name);
+	if (!dev_name || !*dev_name)
+		return invalfc(fc, "Empty source");
+
+	dev_name_end = strchr(dev_name, '/');
+	if (dev_name_end) {
+		/*
+		 * The server_path will include the whole chars from userland
+		 * including the leading '/'.
+		 */
+		kfree(fsopt->server_path);
+		fsopt->server_path = kstrdup(dev_name_end, GFP_KERNEL);
+		if (!fsopt->server_path)
+			return -ENOMEM;
+
+		canonicalize_path(fsopt->server_path);
+	} else {
+		dev_name_end = dev_name + strlen(dev_name);
+	}
+
+	dev_name_end--;		/* back up to separator */
+	if (dev_name_end < dev_name)
+		return invalfc(fc, "Path missing in source");
+
+	dout("device name '%.*s'\n", (int)(dev_name_end - dev_name), dev_name);
+	if (fsopt->server_path)
+		dout("server path '%s'\n", fsopt->server_path);
+
+	dout("trying new device syntax");
+	ret = ceph_parse_new_source(dev_name, dev_name_end, fc);
+	if (ret) {
+		if (ret != -EINVAL)
+			return ret;
+		dout("trying old device syntax");
+		ret = ceph_parse_old_source(dev_name, dev_name_end, fc);
+		if (ret)
+			return ret;
+	}
+
+	fc->source = param->string;
+	param->string = NULL;
+	return 0;
+}
+
+static int ceph_parse_mon_addr(struct fs_parameter *param,
+			       struct fs_context *fc)
+{
+	struct ceph_parse_opts_ctx *pctx = fc->fs_private;
+	struct ceph_mount_options *fsopt = pctx->opts;
+
+	kfree(fsopt->mon_addr);
+	fsopt->mon_addr = param->string;
+	param->string = NULL;
+
+	return ceph_parse_mon_ips(fsopt->mon_addr, strlen(fsopt->mon_addr),
+				  pctx->copts, fc->log.log, '/');
+}
+
+static int ceph_parse_mount_param(struct fs_context *fc,
+				  struct fs_parameter *param)
+{
+	struct ceph_parse_opts_ctx *pctx = fc->fs_private;
+	struct ceph_mount_options *fsopt = pctx->opts;
+	struct fs_parse_result result;
+	unsigned int mode;
+	int token, ret;
+
+	ret = ceph_parse_param(param, pctx->copts, fc->log.log);
+	if (ret != -ENOPARAM)
+		return ret;
+
+	token = fs_parse(fc, ceph_mount_parameters, param, &result);
+	dout("%s fs_parse '%s' token %d\n", __func__, param->key, token);
+	if (token < 0)
+		return token;
+
+	switch (token) {
+	case Opt_snapdirname:
+		kfree(fsopt->snapdir_name);
+		fsopt->snapdir_name = param->string;
+		param->string = NULL;
+		break;
+	case Opt_mds_namespace:
+		if (!namespace_equals(fsopt, param->string, strlen(param->string)))
+			return invalfc(fc, "Mismatching mds_namespace");
+		kfree(fsopt->mds_namespace);
+		fsopt->mds_namespace = param->string;
+		param->string = NULL;
+		break;
+	case Opt_recover_session:
+		mode = result.uint_32;
+		if (mode == ceph_recover_session_no)
+			fsopt->flags &= ~CEPH_MOUNT_OPT_CLEANRECOVER;
+		else if (mode == ceph_recover_session_clean)
+			fsopt->flags |= CEPH_MOUNT_OPT_CLEANRECOVER;
+		else
+			BUG();
+		break;
+	case Opt_source:
+		if (fc->source)
+			return invalfc(fc, "Multiple sources specified");
+		return ceph_parse_source(param, fc);
+	case Opt_mon_addr:
+		return ceph_parse_mon_addr(param, fc);
+	case Opt_wsize:
+		if (result.uint_32 < PAGE_SIZE ||
+		    result.uint_32 > CEPH_MAX_WRITE_SIZE)
+			goto out_of_range;
+		fsopt->wsize = ALIGN(result.uint_32, PAGE_SIZE);
+		break;
+	case Opt_rsize:
+		if (result.uint_32 < PAGE_SIZE ||
+		    result.uint_32 > CEPH_MAX_READ_SIZE)
+			goto out_of_range;
+		fsopt->rsize = ALIGN(result.uint_32, PAGE_SIZE);
+		break;
+	case Opt_rasize:
+		fsopt->rasize = ALIGN(result.uint_32, PAGE_SIZE);
+		break;
+	case Opt_caps_wanted_delay_min:
+		if (result.uint_32 < 1)
+			goto out_of_range;
+		fsopt->caps_wanted_delay_min = result.uint_32;
+		break;
+	case Opt_caps_wanted_delay_max:
+		if (result.uint_32 < 1)
+			goto out_of_range;
+		fsopt->caps_wanted_delay_max = result.uint_32;
+		break;
+	case Opt_caps_max:
+		if (result.int_32 < 0)
+			goto out_of_range;
+		fsopt->caps_max = result.int_32;
+		break;
+	case Opt_readdir_max_entries:
+		if (result.uint_32 < 1)
+			goto out_of_range;
+		fsopt->max_readdir = result.uint_32;
+		break;
+	case Opt_readdir_max_bytes:
+		if (result.uint_32 < PAGE_SIZE && result.uint_32 != 0)
+			goto out_of_range;
+		fsopt->max_readdir_bytes = result.uint_32;
+		break;
+	case Opt_congestion_kb:
+		if (result.uint_32 < 1024) /* at least 1M */
+			goto out_of_range;
+		fsopt->congestion_kb = result.uint_32;
+		break;
+	case Opt_dirstat:
+		if (!result.negated)
+			fsopt->flags |= CEPH_MOUNT_OPT_DIRSTAT;
+		else
+			fsopt->flags &= ~CEPH_MOUNT_OPT_DIRSTAT;
+		break;
+	case Opt_rbytes:
+		if (!result.negated)
+			fsopt->flags |= CEPH_MOUNT_OPT_RBYTES;
+		else
+			fsopt->flags &= ~CEPH_MOUNT_OPT_RBYTES;
+		break;
+	case Opt_asyncreaddir:
+		if (!result.negated)
+			fsopt->flags &= ~CEPH_MOUNT_OPT_NOASYNCREADDIR;
+		else
+			fsopt->flags |= CEPH_MOUNT_OPT_NOASYNCREADDIR;
+		break;
+	case Opt_dcache:
+		if (!result.negated)
+			fsopt->flags |= CEPH_MOUNT_OPT_DCACHE;
+		else
+			fsopt->flags &= ~CEPH_MOUNT_OPT_DCACHE;
+		break;
+	case Opt_ino32:
+		if (!result.negated)
+			fsopt->flags |= CEPH_MOUNT_OPT_INO32;
+		else
+			fsopt->flags &= ~CEPH_MOUNT_OPT_INO32;
+		break;
+
+	case Opt_fscache:
+#ifdef CONFIG_CEPH_FSCACHE
+		kfree(fsopt->fscache_uniq);
+		fsopt->fscache_uniq = NULL;
+		if (result.negated) {
+			fsopt->flags &= ~CEPH_MOUNT_OPT_FSCACHE;
+		} else {
+			fsopt->flags |= CEPH_MOUNT_OPT_FSCACHE;
+			fsopt->fscache_uniq = param->string;
+			param->string = NULL;
+		}
+		break;
+#else
+		return invalfc(fc, "fscache support is disabled");
+#endif
+	case Opt_poolperm:
+		if (!result.negated)
+			fsopt->flags &= ~CEPH_MOUNT_OPT_NOPOOLPERM;
+		else
+			fsopt->flags |= CEPH_MOUNT_OPT_NOPOOLPERM;
+		break;
+	case Opt_require_active_mds:
+		if (!result.negated)
+			fsopt->flags &= ~CEPH_MOUNT_OPT_MOUNTWAIT;
+		else
+			fsopt->flags |= CEPH_MOUNT_OPT_MOUNTWAIT;
+		break;
+	case Opt_quotadf:
+		if (!result.negated)
+			fsopt->flags &= ~CEPH_MOUNT_OPT_NOQUOTADF;
+		else
+			fsopt->flags |= CEPH_MOUNT_OPT_NOQUOTADF;
+		break;
+	case Opt_copyfrom:
+		if (!result.negated)
+			fsopt->flags &= ~CEPH_MOUNT_OPT_NOCOPYFROM;
+		else
+			fsopt->flags |= CEPH_MOUNT_OPT_NOCOPYFROM;
+		break;
+	case Opt_acl:
+		if (!result.negated) {
+#ifdef CONFIG_CEPH_FS_POSIX_ACL
+			fc->sb_flags |= SB_POSIXACL;
+#else
+			return invalfc(fc, "POSIX ACL support is disabled");
+#endif
+		} else {
+			fc->sb_flags &= ~SB_POSIXACL;
+		}
+		break;
+	case Opt_wsync:
+		if (!result.negated)
+			fsopt->flags &= ~CEPH_MOUNT_OPT_ASYNC_DIROPS;
+		else
+			fsopt->flags |= CEPH_MOUNT_OPT_ASYNC_DIROPS;
+		break;
+	case Opt_pagecache:
+		if (result.negated)
+			fsopt->flags |= CEPH_MOUNT_OPT_NOPAGECACHE;
+		else
+			fsopt->flags &= ~CEPH_MOUNT_OPT_NOPAGECACHE;
+		break;
+	default:
+		BUG();
+	}
+	return 0;
+
+out_of_range:
+	return invalfc(fc, "%s out of range", param->key);
+}
+
+static void destroy_mount_options(struct ceph_mount_options *args)
+{
+	dout("destroy_mount_options %p\n", args);
+	if (!args)
+		return;
+
+	kfree(args->snapdir_name);
+	kfree(args->mds_namespace);
+	kfree(args->server_path);
+	kfree(args->fscache_uniq);
+	kfree(args->mon_addr);
+	kfree(args);
+}
+
+static int strcmp_null(const char *s1, const char *s2)
+{
+	if (!s1 && !s2)
+		return 0;
+	if (s1 && !s2)
+		return -1;
+	if (!s1 && s2)
+		return 1;
+	return strcmp(s1, s2);
+}
+
+static int compare_mount_options(struct ceph_mount_options *new_fsopt,
+				 struct ceph_options *new_opt,
+				 struct ceph_fs_client *fsc)
+{
+	struct ceph_mount_options *fsopt1 = new_fsopt;
+	struct ceph_mount_options *fsopt2 = fsc->mount_options;
+	int ofs = offsetof(struct ceph_mount_options, snapdir_name);
+	int ret;
+
+	ret = memcmp(fsopt1, fsopt2, ofs);
+	if (ret)
+		return ret;
+
+	ret = strcmp_null(fsopt1->snapdir_name, fsopt2->snapdir_name);
+	if (ret)
+		return ret;
+
+	ret = strcmp_null(fsopt1->mds_namespace, fsopt2->mds_namespace);
+	if (ret)
+		return ret;
+
+	ret = strcmp_null(fsopt1->server_path, fsopt2->server_path);
+	if (ret)
+		return ret;
+
+	ret = strcmp_null(fsopt1->fscache_uniq, fsopt2->fscache_uniq);
+	if (ret)
+		return ret;
+
+	ret = strcmp_null(fsopt1->mon_addr, fsopt2->mon_addr);
+	if (ret)
+		return ret;
+
+	return ceph_compare_options(new_opt, fsc->client);
+}
+
+/**
+ * ceph_show_options - Show mount options in /proc/mounts
+ * @m: seq_file to write to
+ * @root: root of that (sub)tree
+ */
+static int ceph_show_options(struct seq_file *m, struct dentry *root)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(root->d_sb);
+	struct ceph_mount_options *fsopt = fsc->mount_options;
+	size_t pos;
+	int ret;
+
+	/* a comma between MNT/MS and client options */
+	seq_putc(m, ',');
+	pos = m->count;
+
+	ret = ceph_print_client_options(m, fsc->client, false);
+	if (ret)
+		return ret;
+
+	/* retract our comma if no client options */
+	if (m->count == pos)
+		m->count--;
+
+	if (fsopt->flags & CEPH_MOUNT_OPT_DIRSTAT)
+		seq_puts(m, ",dirstat");
+	if ((fsopt->flags & CEPH_MOUNT_OPT_RBYTES))
+		seq_puts(m, ",rbytes");
+	if (fsopt->flags & CEPH_MOUNT_OPT_NOASYNCREADDIR)
+		seq_puts(m, ",noasyncreaddir");
+	if ((fsopt->flags & CEPH_MOUNT_OPT_DCACHE) == 0)
+		seq_puts(m, ",nodcache");
+	if (fsopt->flags & CEPH_MOUNT_OPT_INO32)
+		seq_puts(m, ",ino32");
+	if (fsopt->flags & CEPH_MOUNT_OPT_FSCACHE) {
+		seq_show_option(m, "fsc", fsopt->fscache_uniq);
+	}
+	if (fsopt->flags & CEPH_MOUNT_OPT_NOPOOLPERM)
+		seq_puts(m, ",nopoolperm");
+	if (fsopt->flags & CEPH_MOUNT_OPT_NOQUOTADF)
+		seq_puts(m, ",noquotadf");
+
+#ifdef CONFIG_CEPH_FS_POSIX_ACL
+	if (root->d_sb->s_flags & SB_POSIXACL)
+		seq_puts(m, ",acl");
+	else
+		seq_puts(m, ",noacl");
+#endif
+
+	if ((fsopt->flags & CEPH_MOUNT_OPT_NOCOPYFROM) == 0)
+		seq_puts(m, ",copyfrom");
+
+	/* dump mds_namespace when old device syntax is in use */
+	if (fsopt->mds_namespace && !fsopt->new_dev_syntax)
+		seq_show_option(m, "mds_namespace", fsopt->mds_namespace);
+
+	if (fsopt->mon_addr)
+		seq_printf(m, ",mon_addr=%s", fsopt->mon_addr);
+
+	if (fsopt->flags & CEPH_MOUNT_OPT_CLEANRECOVER)
+		seq_show_option(m, "recover_session", "clean");
+
+	if (!(fsopt->flags & CEPH_MOUNT_OPT_ASYNC_DIROPS))
+		seq_puts(m, ",wsync");
+
+	if (fsopt->flags & CEPH_MOUNT_OPT_NOPAGECACHE)
+		seq_puts(m, ",nopagecache");
+
+	if (fsopt->wsize != CEPH_MAX_WRITE_SIZE)
+		seq_printf(m, ",wsize=%u", fsopt->wsize);
+	if (fsopt->rsize != CEPH_MAX_READ_SIZE)
+		seq_printf(m, ",rsize=%u", fsopt->rsize);
+	if (fsopt->rasize != CEPH_RASIZE_DEFAULT)
+		seq_printf(m, ",rasize=%u", fsopt->rasize);
+	if (fsopt->congestion_kb != default_congestion_kb())
+		seq_printf(m, ",write_congestion_kb=%u", fsopt->congestion_kb);
+	if (fsopt->caps_max)
+		seq_printf(m, ",caps_max=%d", fsopt->caps_max);
+	if (fsopt->caps_wanted_delay_min != CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT)
+		seq_printf(m, ",caps_wanted_delay_min=%u",
+			 fsopt->caps_wanted_delay_min);
+	if (fsopt->caps_wanted_delay_max != CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT)
+		seq_printf(m, ",caps_wanted_delay_max=%u",
+			   fsopt->caps_wanted_delay_max);
+	if (fsopt->max_readdir != CEPH_MAX_READDIR_DEFAULT)
+		seq_printf(m, ",readdir_max_entries=%u", fsopt->max_readdir);
+	if (fsopt->max_readdir_bytes != CEPH_MAX_READDIR_BYTES_DEFAULT)
+		seq_printf(m, ",readdir_max_bytes=%u", fsopt->max_readdir_bytes);
+	if (strcmp(fsopt->snapdir_name, CEPH_SNAPDIRNAME_DEFAULT))
+		seq_show_option(m, "snapdirname", fsopt->snapdir_name);
+
+	return 0;
+}
+
+/*
+ * handle any mon messages the standard library doesn't understand.
+ * return error if we don't either.
+ */
+static int extra_mon_dispatch(struct ceph_client *client, struct ceph_msg *msg)
+{
+	struct ceph_fs_client *fsc = client->private;
+	int type = le16_to_cpu(msg->hdr.type);
+
+	switch (type) {
+	case CEPH_MSG_MDS_MAP:
+		ceph_mdsc_handle_mdsmap(fsc->mdsc, msg);
+		return 0;
+	case CEPH_MSG_FS_MAP_USER:
+		ceph_mdsc_handle_fsmap(fsc->mdsc, msg);
+		return 0;
+	default:
+		return -1;
+	}
+}
+
+/*
+ * create a new fs client
+ *
+ * Success or not, this function consumes @fsopt and @opt.
+ */
+static struct ceph_fs_client *create_fs_client(struct ceph_mount_options *fsopt,
+					struct ceph_options *opt)
+{
+	struct ceph_fs_client *fsc;
+	int err;
+
+	fsc = kzalloc(sizeof(*fsc), GFP_KERNEL);
+	if (!fsc) {
+		err = -ENOMEM;
+		goto fail;
+	}
+
+	fsc->client = ceph_create_client(opt, fsc);
+	if (IS_ERR(fsc->client)) {
+		err = PTR_ERR(fsc->client);
+		goto fail;
+	}
+	opt = NULL; /* fsc->client now owns this */
+
+	fsc->client->extra_mon_dispatch = extra_mon_dispatch;
+	ceph_set_opt(fsc->client, ABORT_ON_FULL);
+
+	if (!fsopt->mds_namespace) {
+		ceph_monc_want_map(&fsc->client->monc, CEPH_SUB_MDSMAP,
+				   0, true);
+	} else {
+		ceph_monc_want_map(&fsc->client->monc, CEPH_SUB_FSMAP,
+				   0, false);
+	}
+
+	fsc->mount_options = fsopt;
+
+	fsc->sb = NULL;
+	fsc->mount_state = CEPH_MOUNT_MOUNTING;
+	fsc->filp_gen = 1;
+	fsc->have_copy_from2 = true;
+
+	atomic_long_set(&fsc->writeback_count, 0);
+	fsc->write_congested = false;
+
+	err = -ENOMEM;
+	/*
+	 * The number of concurrent works can be high but they don't need
+	 * to be processed in parallel, limit concurrency.
+	 */
+	fsc->inode_wq = alloc_workqueue("ceph-inode", WQ_UNBOUND, 0);
+	if (!fsc->inode_wq)
+		goto fail_client;
+	fsc->cap_wq = alloc_workqueue("ceph-cap", 0, 1);
+	if (!fsc->cap_wq)
+		goto fail_inode_wq;
+
+	hash_init(fsc->async_unlink_conflict);
+	spin_lock_init(&fsc->async_unlink_conflict_lock);
+
+	spin_lock(&ceph_fsc_lock);
+	list_add_tail(&fsc->metric_wakeup, &ceph_fsc_list);
+	spin_unlock(&ceph_fsc_lock);
+
+	return fsc;
+
+fail_inode_wq:
+	destroy_workqueue(fsc->inode_wq);
+fail_client:
+	ceph_destroy_client(fsc->client);
+fail:
+	kfree(fsc);
+	if (opt)
+		ceph_destroy_options(opt);
+	destroy_mount_options(fsopt);
+	return ERR_PTR(err);
+}
+
+static void flush_fs_workqueues(struct ceph_fs_client *fsc)
+{
+	flush_workqueue(fsc->inode_wq);
+	flush_workqueue(fsc->cap_wq);
+}
+
+static void destroy_fs_client(struct ceph_fs_client *fsc)
+{
+	dout("destroy_fs_client %p\n", fsc);
+
+	spin_lock(&ceph_fsc_lock);
+	list_del(&fsc->metric_wakeup);
+	spin_unlock(&ceph_fsc_lock);
+
+	ceph_mdsc_destroy(fsc);
+	destroy_workqueue(fsc->inode_wq);
+	destroy_workqueue(fsc->cap_wq);
+
+	destroy_mount_options(fsc->mount_options);
+
+	ceph_destroy_client(fsc->client);
+
+	kfree(fsc);
+	dout("destroy_fs_client %p done\n", fsc);
+}
+
+/*
+ * caches
+ */
+struct kmem_cache *ceph_inode_cachep;
+struct kmem_cache *ceph_cap_cachep;
+struct kmem_cache *ceph_cap_snap_cachep;
+struct kmem_cache *ceph_cap_flush_cachep;
+struct kmem_cache *ceph_dentry_cachep;
+struct kmem_cache *ceph_file_cachep;
+struct kmem_cache *ceph_dir_file_cachep;
+struct kmem_cache *ceph_mds_request_cachep;
+mempool_t *ceph_wb_pagevec_pool;
+
+static void ceph_inode_init_once(void *foo)
+{
+	struct ceph_inode_info *ci = foo;
+	inode_init_once(&ci->netfs.inode);
+}
+
+static int __init init_caches(void)
+{
+	int error = -ENOMEM;
+
+	ceph_inode_cachep = kmem_cache_create("ceph_inode_info",
+				      sizeof(struct ceph_inode_info),
+				      __alignof__(struct ceph_inode_info),
+				      SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD|
+				      SLAB_ACCOUNT, ceph_inode_init_once);
+	if (!ceph_inode_cachep)
+		return -ENOMEM;
+
+	ceph_cap_cachep = KMEM_CACHE(ceph_cap, SLAB_MEM_SPREAD);
+	if (!ceph_cap_cachep)
+		goto bad_cap;
+	ceph_cap_snap_cachep = KMEM_CACHE(ceph_cap_snap, SLAB_MEM_SPREAD);
+	if (!ceph_cap_snap_cachep)
+		goto bad_cap_snap;
+	ceph_cap_flush_cachep = KMEM_CACHE(ceph_cap_flush,
+					   SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD);
+	if (!ceph_cap_flush_cachep)
+		goto bad_cap_flush;
+
+	ceph_dentry_cachep = KMEM_CACHE(ceph_dentry_info,
+					SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD);
+	if (!ceph_dentry_cachep)
+		goto bad_dentry;
+
+	ceph_file_cachep = KMEM_CACHE(ceph_file_info, SLAB_MEM_SPREAD);
+	if (!ceph_file_cachep)
+		goto bad_file;
+
+	ceph_dir_file_cachep = KMEM_CACHE(ceph_dir_file_info, SLAB_MEM_SPREAD);
+	if (!ceph_dir_file_cachep)
+		goto bad_dir_file;
+
+	ceph_mds_request_cachep = KMEM_CACHE(ceph_mds_request, SLAB_MEM_SPREAD);
+	if (!ceph_mds_request_cachep)
+		goto bad_mds_req;
+
+	ceph_wb_pagevec_pool = mempool_create_kmalloc_pool(10, CEPH_MAX_WRITE_SIZE >> PAGE_SHIFT);
+	if (!ceph_wb_pagevec_pool)
+		goto bad_pagevec_pool;
+
+	return 0;
+
+bad_pagevec_pool:
+	kmem_cache_destroy(ceph_mds_request_cachep);
+bad_mds_req:
+	kmem_cache_destroy(ceph_dir_file_cachep);
+bad_dir_file:
+	kmem_cache_destroy(ceph_file_cachep);
+bad_file:
+	kmem_cache_destroy(ceph_dentry_cachep);
+bad_dentry:
+	kmem_cache_destroy(ceph_cap_flush_cachep);
+bad_cap_flush:
+	kmem_cache_destroy(ceph_cap_snap_cachep);
+bad_cap_snap:
+	kmem_cache_destroy(ceph_cap_cachep);
+bad_cap:
+	kmem_cache_destroy(ceph_inode_cachep);
+	return error;
+}
+
+static void destroy_caches(void)
+{
+	/*
+	 * Make sure all delayed rcu free inodes are flushed before we
+	 * destroy cache.
+	 */
+	rcu_barrier();
+
+	kmem_cache_destroy(ceph_inode_cachep);
+	kmem_cache_destroy(ceph_cap_cachep);
+	kmem_cache_destroy(ceph_cap_snap_cachep);
+	kmem_cache_destroy(ceph_cap_flush_cachep);
+	kmem_cache_destroy(ceph_dentry_cachep);
+	kmem_cache_destroy(ceph_file_cachep);
+	kmem_cache_destroy(ceph_dir_file_cachep);
+	kmem_cache_destroy(ceph_mds_request_cachep);
+	mempool_destroy(ceph_wb_pagevec_pool);
+}
+
+static void __ceph_umount_begin(struct ceph_fs_client *fsc)
+{
+	ceph_osdc_abort_requests(&fsc->client->osdc, -EIO);
+	ceph_mdsc_force_umount(fsc->mdsc);
+	fsc->filp_gen++; // invalidate open files
+}
+
+/*
+ * ceph_umount_begin - initiate forced umount.  Tear down the
+ * mount, skipping steps that may hang while waiting for server(s).
+ */
+void ceph_umount_begin(struct super_block *sb)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(sb);
+
+	dout("ceph_umount_begin - starting forced umount\n");
+	if (!fsc)
+		return;
+	fsc->mount_state = CEPH_MOUNT_SHUTDOWN;
+	__ceph_umount_begin(fsc);
+}
+
+static const struct super_operations ceph_super_ops = {
+	.alloc_inode	= ceph_alloc_inode,
+	.free_inode	= ceph_free_inode,
+	.write_inode    = ceph_write_inode,
+	.drop_inode	= generic_delete_inode,
+	.evict_inode	= ceph_evict_inode,
+	.sync_fs        = ceph_sync_fs,
+	.put_super	= ceph_put_super,
+	.show_options   = ceph_show_options,
+	.statfs		= ceph_statfs,
+	.umount_begin   = ceph_umount_begin,
+};
+
+/*
+ * Bootstrap mount by opening the root directory.  Note the mount
+ * @started time from caller, and time out if this takes too long.
+ */
+static struct dentry *open_root_dentry(struct ceph_fs_client *fsc,
+				       const char *path,
+				       unsigned long started)
+{
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	struct ceph_mds_request *req = NULL;
+	int err;
+	struct dentry *root;
+
+	/* open dir */
+	dout("open_root_inode opening '%s'\n", path);
+	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_GETATTR, USE_ANY_MDS);
+	if (IS_ERR(req))
+		return ERR_CAST(req);
+	req->r_path1 = kstrdup(path, GFP_NOFS);
+	if (!req->r_path1) {
+		root = ERR_PTR(-ENOMEM);
+		goto out;
+	}
+
+	req->r_ino1.ino = CEPH_INO_ROOT;
+	req->r_ino1.snap = CEPH_NOSNAP;
+	req->r_started = started;
+	req->r_timeout = fsc->client->options->mount_timeout;
+	req->r_args.getattr.mask = cpu_to_le32(CEPH_STAT_CAP_INODE);
+	req->r_num_caps = 2;
+	err = ceph_mdsc_do_request(mdsc, NULL, req);
+	if (err == 0) {
+		struct inode *inode = req->r_target_inode;
+		req->r_target_inode = NULL;
+		dout("open_root_inode success\n");
+		root = d_make_root(inode);
+		if (!root) {
+			root = ERR_PTR(-ENOMEM);
+			goto out;
+		}
+		dout("open_root_inode success, root dentry is %p\n", root);
+	} else {
+		root = ERR_PTR(err);
+	}
+out:
+	ceph_mdsc_put_request(req);
+	return root;
+}
+
+/*
+ * mount: join the ceph cluster, and open root directory.
+ */
+static struct dentry *ceph_real_mount(struct ceph_fs_client *fsc,
+				      struct fs_context *fc)
+{
+	int err;
+	unsigned long started = jiffies;  /* note the start time */
+	struct dentry *root;
+
+	dout("mount start %p\n", fsc);
+	mutex_lock(&fsc->client->mount_mutex);
+
+	if (!fsc->sb->s_root) {
+		const char *path = fsc->mount_options->server_path ?
+				     fsc->mount_options->server_path + 1 : "";
+
+		err = __ceph_open_session(fsc->client, started);
+		if (err < 0)
+			goto out;
+
+		/* setup fscache */
+		if (fsc->mount_options->flags & CEPH_MOUNT_OPT_FSCACHE) {
+			err = ceph_fscache_register_fs(fsc, fc);
+			if (err < 0)
+				goto out;
+		}
+
+		dout("mount opening path '%s'\n", path);
+
+		ceph_fs_debugfs_init(fsc);
+
+		root = open_root_dentry(fsc, path, started);
+		if (IS_ERR(root)) {
+			err = PTR_ERR(root);
+			goto out;
+		}
+		fsc->sb->s_root = dget(root);
+	} else {
+		root = dget(fsc->sb->s_root);
+	}
+
+	fsc->mount_state = CEPH_MOUNT_MOUNTED;
+	dout("mount success\n");
+	mutex_unlock(&fsc->client->mount_mutex);
+	return root;
+
+out:
+	mutex_unlock(&fsc->client->mount_mutex);
+	return ERR_PTR(err);
+}
+
+static int ceph_set_super(struct super_block *s, struct fs_context *fc)
+{
+	struct ceph_fs_client *fsc = s->s_fs_info;
+	int ret;
+
+	dout("set_super %p\n", s);
+
+	s->s_maxbytes = MAX_LFS_FILESIZE;
+
+	s->s_xattr = ceph_xattr_handlers;
+	fsc->sb = s;
+	fsc->max_file_size = 1ULL << 40; /* temp value until we get mdsmap */
+
+	s->s_op = &ceph_super_ops;
+	s->s_d_op = &ceph_dentry_ops;
+	s->s_export_op = &ceph_export_ops;
+
+	s->s_time_gran = 1;
+	s->s_time_min = 0;
+	s->s_time_max = U32_MAX;
+	s->s_flags |= SB_NODIRATIME | SB_NOATIME;
+
+	ret = set_anon_super_fc(s, fc);
+	if (ret != 0)
+		fsc->sb = NULL;
+	return ret;
+}
+
+/*
+ * share superblock if same fs AND options
+ */
+static int ceph_compare_super(struct super_block *sb, struct fs_context *fc)
+{
+	struct ceph_fs_client *new = fc->s_fs_info;
+	struct ceph_mount_options *fsopt = new->mount_options;
+	struct ceph_options *opt = new->client->options;
+	struct ceph_fs_client *fsc = ceph_sb_to_client(sb);
+
+	dout("ceph_compare_super %p\n", sb);
+
+	if (compare_mount_options(fsopt, opt, fsc)) {
+		dout("monitor(s)/mount options don't match\n");
+		return 0;
+	}
+	if ((opt->flags & CEPH_OPT_FSID) &&
+	    ceph_fsid_compare(&opt->fsid, &fsc->client->fsid)) {
+		dout("fsid doesn't match\n");
+		return 0;
+	}
+	if (fc->sb_flags != (sb->s_flags & ~SB_BORN)) {
+		dout("flags differ\n");
+		return 0;
+	}
+
+	if (fsc->blocklisted && !ceph_test_mount_opt(fsc, CLEANRECOVER)) {
+		dout("client is blocklisted (and CLEANRECOVER is not set)\n");
+		return 0;
+	}
+
+	if (fsc->mount_state == CEPH_MOUNT_SHUTDOWN) {
+		dout("client has been forcibly unmounted\n");
+		return 0;
+	}
+
+	return 1;
+}
+
+/*
+ * construct our own bdi so we can control readahead, etc.
+ */
+static atomic_long_t bdi_seq = ATOMIC_LONG_INIT(0);
+
+static int ceph_setup_bdi(struct super_block *sb, struct ceph_fs_client *fsc)
+{
+	int err;
+
+	err = super_setup_bdi_name(sb, "ceph-%ld",
+				   atomic_long_inc_return(&bdi_seq));
+	if (err)
+		return err;
+
+	/* set ra_pages based on rasize mount option? */
+	sb->s_bdi->ra_pages = fsc->mount_options->rasize >> PAGE_SHIFT;
+
+	/* set io_pages based on max osd read size */
+	sb->s_bdi->io_pages = fsc->mount_options->rsize >> PAGE_SHIFT;
+
+	return 0;
+}
+
+static int ceph_get_tree(struct fs_context *fc)
+{
+	struct ceph_parse_opts_ctx *pctx = fc->fs_private;
+	struct ceph_mount_options *fsopt = pctx->opts;
+	struct super_block *sb;
+	struct ceph_fs_client *fsc;
+	struct dentry *res;
+	int (*compare_super)(struct super_block *, struct fs_context *) =
+		ceph_compare_super;
+	int err;
+
+	dout("ceph_get_tree\n");
+
+	if (!fc->source)
+		return invalfc(fc, "No source");
+	if (fsopt->new_dev_syntax && !fsopt->mon_addr)
+		return invalfc(fc, "No monitor address");
+
+	/* create client (which we may/may not use) */
+	fsc = create_fs_client(pctx->opts, pctx->copts);
+	pctx->opts = NULL;
+	pctx->copts = NULL;
+	if (IS_ERR(fsc)) {
+		err = PTR_ERR(fsc);
+		goto out_final;
+	}
+
+	err = ceph_mdsc_init(fsc);
+	if (err < 0)
+		goto out;
+
+	if (ceph_test_opt(fsc->client, NOSHARE))
+		compare_super = NULL;
+
+	fc->s_fs_info = fsc;
+	sb = sget_fc(fc, compare_super, ceph_set_super);
+	fc->s_fs_info = NULL;
+	if (IS_ERR(sb)) {
+		err = PTR_ERR(sb);
+		goto out;
+	}
+
+	if (ceph_sb_to_client(sb) != fsc) {
+		destroy_fs_client(fsc);
+		fsc = ceph_sb_to_client(sb);
+		dout("get_sb got existing client %p\n", fsc);
+	} else {
+		dout("get_sb using new client %p\n", fsc);
+		err = ceph_setup_bdi(sb, fsc);
+		if (err < 0)
+			goto out_splat;
+	}
+
+	res = ceph_real_mount(fsc, fc);
+	if (IS_ERR(res)) {
+		err = PTR_ERR(res);
+		goto out_splat;
+	}
+	dout("root %p inode %p ino %llx.%llx\n", res,
+	     d_inode(res), ceph_vinop(d_inode(res)));
+	fc->root = fsc->sb->s_root;
+	return 0;
+
+out_splat:
+	if (!ceph_mdsmap_is_cluster_available(fsc->mdsc->mdsmap)) {
+		pr_info("No mds server is up or the cluster is laggy\n");
+		err = -EHOSTUNREACH;
+	}
+
+	ceph_mdsc_close_sessions(fsc->mdsc);
+	deactivate_locked_super(sb);
+	goto out_final;
+
+out:
+	destroy_fs_client(fsc);
+out_final:
+	dout("ceph_get_tree fail %d\n", err);
+	return err;
+}
+
+static void ceph_free_fc(struct fs_context *fc)
+{
+	struct ceph_parse_opts_ctx *pctx = fc->fs_private;
+
+	if (pctx) {
+		destroy_mount_options(pctx->opts);
+		ceph_destroy_options(pctx->copts);
+		kfree(pctx);
+	}
+}
+
+static int ceph_reconfigure_fc(struct fs_context *fc)
+{
+	struct ceph_parse_opts_ctx *pctx = fc->fs_private;
+	struct ceph_mount_options *fsopt = pctx->opts;
+	struct ceph_fs_client *fsc = ceph_sb_to_client(fc->root->d_sb);
+
+	if (fsopt->flags & CEPH_MOUNT_OPT_ASYNC_DIROPS)
+		ceph_set_mount_opt(fsc, ASYNC_DIROPS);
+	else
+		ceph_clear_mount_opt(fsc, ASYNC_DIROPS);
+
+	if (strcmp_null(fsc->mount_options->mon_addr, fsopt->mon_addr)) {
+		kfree(fsc->mount_options->mon_addr);
+		fsc->mount_options->mon_addr = fsopt->mon_addr;
+		fsopt->mon_addr = NULL;
+		pr_notice("ceph: monitor addresses recorded, but not used for reconnection");
+	}
+
+	sync_filesystem(fc->root->d_sb);
+	return 0;
+}
+
+static const struct fs_context_operations ceph_context_ops = {
+	.free		= ceph_free_fc,
+	.parse_param	= ceph_parse_mount_param,
+	.get_tree	= ceph_get_tree,
+	.reconfigure	= ceph_reconfigure_fc,
+};
+
+/*
+ * Set up the filesystem mount context.
+ */
+static int ceph_init_fs_context(struct fs_context *fc)
+{
+	struct ceph_parse_opts_ctx *pctx;
+	struct ceph_mount_options *fsopt;
+
+	pctx = kzalloc(sizeof(*pctx), GFP_KERNEL);
+	if (!pctx)
+		return -ENOMEM;
+
+	pctx->copts = ceph_alloc_options();
+	if (!pctx->copts)
+		goto nomem;
+
+	pctx->opts = kzalloc(sizeof(*pctx->opts), GFP_KERNEL);
+	if (!pctx->opts)
+		goto nomem;
+
+	fsopt = pctx->opts;
+	fsopt->flags = CEPH_MOUNT_OPT_DEFAULT;
+
+	fsopt->wsize = CEPH_MAX_WRITE_SIZE;
+	fsopt->rsize = CEPH_MAX_READ_SIZE;
+	fsopt->rasize = CEPH_RASIZE_DEFAULT;
+	fsopt->snapdir_name = kstrdup(CEPH_SNAPDIRNAME_DEFAULT, GFP_KERNEL);
+	if (!fsopt->snapdir_name)
+		goto nomem;
+
+	fsopt->caps_wanted_delay_min = CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT;
+	fsopt->caps_wanted_delay_max = CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT;
+	fsopt->max_readdir = CEPH_MAX_READDIR_DEFAULT;
+	fsopt->max_readdir_bytes = CEPH_MAX_READDIR_BYTES_DEFAULT;
+	fsopt->congestion_kb = default_congestion_kb();
+
+#ifdef CONFIG_CEPH_FS_POSIX_ACL
+	fc->sb_flags |= SB_POSIXACL;
+#endif
+
+	fc->fs_private = pctx;
+	fc->ops = &ceph_context_ops;
+	return 0;
+
+nomem:
+	destroy_mount_options(pctx->opts);
+	ceph_destroy_options(pctx->copts);
+	kfree(pctx);
+	return -ENOMEM;
+}
+
+/*
+ * Return true if it successfully increases the blocker counter,
+ * or false if the mdsc is in stopping and flushed state.
+ */
+static bool __inc_stopping_blocker(struct ceph_mds_client *mdsc)
+{
+	spin_lock(&mdsc->stopping_lock);
+	if (mdsc->stopping >= CEPH_MDSC_STOPPING_FLUSHING) {
+		spin_unlock(&mdsc->stopping_lock);
+		return false;
+	}
+	atomic_inc(&mdsc->stopping_blockers);
+	spin_unlock(&mdsc->stopping_lock);
+	return true;
+}
+
+static void __dec_stopping_blocker(struct ceph_mds_client *mdsc)
+{
+	spin_lock(&mdsc->stopping_lock);
+	if (!atomic_dec_return(&mdsc->stopping_blockers) &&
+	    mdsc->stopping >= CEPH_MDSC_STOPPING_FLUSHING)
+		complete_all(&mdsc->stopping_waiter);
+	spin_unlock(&mdsc->stopping_lock);
+}
+
+/* For metadata IO requests */
+bool ceph_inc_mds_stopping_blocker(struct ceph_mds_client *mdsc,
+				   struct ceph_mds_session *session)
+{
+	mutex_lock(&session->s_mutex);
+	inc_session_sequence(session);
+	mutex_unlock(&session->s_mutex);
+
+	return __inc_stopping_blocker(mdsc);
+}
+
+void ceph_dec_mds_stopping_blocker(struct ceph_mds_client *mdsc)
+{
+	__dec_stopping_blocker(mdsc);
+}
+
+static void ceph_kill_sb(struct super_block *s)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(s);
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	bool wait;
+
+	dout("kill_sb %p\n", s);
+
+	ceph_mdsc_pre_umount(mdsc);
+	flush_fs_workqueues(fsc);
+
+	/*
+	 * Though the kill_anon_super() will finally trigger the
+	 * sync_filesystem() anyway, we still need to do it here and
+	 * then bump the stage of shutdown. This will allow us to
+	 * drop any further message, which will increase the inodes'
+	 * i_count reference counters but makes no sense any more,
+	 * from MDSs.
+	 *
+	 * Without this when evicting the inodes it may fail in the
+	 * kill_anon_super(), which will trigger a warning when
+	 * destroying the fscrypt keyring and then possibly trigger
+	 * a further crash in ceph module when the iput() tries to
+	 * evict the inodes later.
+	 */
+	sync_filesystem(s);
+
+	spin_lock(&mdsc->stopping_lock);
+	mdsc->stopping = CEPH_MDSC_STOPPING_FLUSHING;
+	wait = !!atomic_read(&mdsc->stopping_blockers);
+	spin_unlock(&mdsc->stopping_lock);
+
+	if (wait && atomic_read(&mdsc->stopping_blockers)) {
+		long timeleft = wait_for_completion_killable_timeout(
+					&mdsc->stopping_waiter,
+					fsc->client->options->mount_timeout);
+		if (!timeleft) /* timed out */
+			pr_warn("umount timed out, %ld\n", timeleft);
+		else if (timeleft < 0) /* killed */
+			pr_warn("umount was killed, %ld\n", timeleft);
+	}
+
+	mdsc->stopping = CEPH_MDSC_STOPPING_FLUSHED;
+	kill_anon_super(s);
+
+	fsc->client->extra_mon_dispatch = NULL;
+	ceph_fs_debugfs_cleanup(fsc);
+
+	ceph_fscache_unregister_fs(fsc);
+
+	destroy_fs_client(fsc);
+}
+
+static struct file_system_type ceph_fs_type = {
+	.owner		= THIS_MODULE,
+	.name		= "ceph",
+	.init_fs_context = ceph_init_fs_context,
+	.kill_sb	= ceph_kill_sb,
+	.fs_flags	= FS_RENAME_DOES_D_MOVE,
+};
+MODULE_ALIAS_FS("ceph");
+
+int ceph_force_reconnect(struct super_block *sb)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(sb);
+	int err = 0;
+
+	fsc->mount_state = CEPH_MOUNT_RECOVER;
+	__ceph_umount_begin(fsc);
+
+	/* Make sure all page caches get invalidated.
+	 * see remove_session_caps_cb() */
+	flush_workqueue(fsc->inode_wq);
+
+	/* In case that we were blocklisted. This also reset
+	 * all mon/osd connections */
+	ceph_reset_client_addr(fsc->client);
+
+	ceph_osdc_clear_abort_err(&fsc->client->osdc);
+
+	fsc->blocklisted = false;
+	fsc->mount_state = CEPH_MOUNT_MOUNTED;
+
+	if (sb->s_root) {
+		err = __ceph_do_getattr(d_inode(sb->s_root), NULL,
+					CEPH_STAT_CAP_INODE, true);
+	}
+	return err;
+}
+
+static int __init init_ceph(void)
+{
+	int ret = init_caches();
+	if (ret)
+		goto out;
+
+	ceph_flock_init();
+	ret = register_filesystem(&ceph_fs_type);
+	if (ret)
+		goto out_caches;
+
+	pr_info("loaded (mds proto %d)\n", CEPH_MDSC_PROTOCOL);
+
+	return 0;
+
+out_caches:
+	destroy_caches();
+out:
+	return ret;
+}
+
+static void __exit exit_ceph(void)
+{
+	dout("exit_ceph\n");
+	unregister_filesystem(&ceph_fs_type);
+	destroy_caches();
+}
+
+static int param_set_metrics(const char *val, const struct kernel_param *kp)
+{
+	struct ceph_fs_client *fsc;
+	int ret;
+
+	ret = param_set_bool(val, kp);
+	if (ret) {
+		pr_err("Failed to parse sending metrics switch value '%s'\n",
+		       val);
+		return ret;
+	} else if (!disable_send_metrics) {
+		// wake up all the mds clients
+		spin_lock(&ceph_fsc_lock);
+		list_for_each_entry(fsc, &ceph_fsc_list, metric_wakeup) {
+			metric_schedule_delayed(&fsc->mdsc->metric);
+		}
+		spin_unlock(&ceph_fsc_lock);
+	}
+
+	return 0;
+}
+
+static const struct kernel_param_ops param_ops_metrics = {
+	.set = param_set_metrics,
+	.get = param_get_bool,
+};
+
+bool disable_send_metrics = false;
+module_param_cb(disable_send_metrics, &param_ops_metrics, &disable_send_metrics, 0644);
+MODULE_PARM_DESC(disable_send_metrics, "Enable sending perf metrics to ceph cluster (default: on)");
+
+/* for both v1 and v2 syntax */
+static bool mount_support = true;
+static const struct kernel_param_ops param_ops_mount_syntax = {
+	.get = param_get_bool,
+};
+module_param_cb(mount_syntax_v1, &param_ops_mount_syntax, &mount_support, 0444);
+module_param_cb(mount_syntax_v2, &param_ops_mount_syntax, &mount_support, 0444);
+
+module_init(init_ceph);
+module_exit(exit_ceph);
+
+MODULE_AUTHOR("Sage Weil <sage@newdream.net>");
+MODULE_AUTHOR("Yehuda Sadeh <yehuda@hq.newdream.net>");
+MODULE_AUTHOR("Patience Warnick <patience@newdream.net>");
+MODULE_DESCRIPTION("Ceph filesystem for Linux");
+MODULE_LICENSE("GPL");
diff --git a/fs/ceph/super.h b/fs/ceph/super.h
new file mode 100644
index 000000000..7ca74f5f7
--- /dev/null
+++ b/fs/ceph/super.h
@@ -0,0 +1,1380 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _FS_CEPH_SUPER_H
+#define _FS_CEPH_SUPER_H
+
+#include <linux/ceph/ceph_debug.h>
+
+#include <asm/unaligned.h>
+#include <linux/backing-dev.h>
+#include <linux/completion.h>
+#include <linux/exportfs.h>
+#include <linux/fs.h>
+#include <linux/mempool.h>
+#include <linux/pagemap.h>
+#include <linux/wait.h>
+#include <linux/writeback.h>
+#include <linux/slab.h>
+#include <linux/posix_acl.h>
+#include <linux/refcount.h>
+#include <linux/security.h>
+#include <linux/netfs.h>
+#include <linux/fscache.h>
+#include <linux/hashtable.h>
+
+#include <linux/ceph/libceph.h>
+
+/* large granularity for statfs utilization stats to facilitate
+ * large volume sizes on 32-bit machines. */
+#define CEPH_BLOCK_SHIFT   22  /* 4 MB */
+#define CEPH_BLOCK         (1 << CEPH_BLOCK_SHIFT)
+#define CEPH_4K_BLOCK_SHIFT 12  /* 4 KB */
+
+#define CEPH_MOUNT_OPT_CLEANRECOVER    (1<<1) /* auto reonnect (clean mode) after blocklisted */
+#define CEPH_MOUNT_OPT_DIRSTAT         (1<<4) /* `cat dirname` for stats */
+#define CEPH_MOUNT_OPT_RBYTES          (1<<5) /* dir st_bytes = rbytes */
+#define CEPH_MOUNT_OPT_NOASYNCREADDIR  (1<<7) /* no dcache readdir */
+#define CEPH_MOUNT_OPT_INO32           (1<<8) /* 32 bit inos */
+#define CEPH_MOUNT_OPT_DCACHE          (1<<9) /* use dcache for readdir etc */
+#define CEPH_MOUNT_OPT_FSCACHE         (1<<10) /* use fscache */
+#define CEPH_MOUNT_OPT_NOPOOLPERM      (1<<11) /* no pool permission check */
+#define CEPH_MOUNT_OPT_MOUNTWAIT       (1<<12) /* mount waits if no mds is up */
+#define CEPH_MOUNT_OPT_NOQUOTADF       (1<<13) /* no root dir quota in statfs */
+#define CEPH_MOUNT_OPT_NOCOPYFROM      (1<<14) /* don't use RADOS 'copy-from' op */
+#define CEPH_MOUNT_OPT_ASYNC_DIROPS    (1<<15) /* allow async directory ops */
+#define CEPH_MOUNT_OPT_NOPAGECACHE     (1<<16) /* bypass pagecache altogether */
+
+#define CEPH_MOUNT_OPT_DEFAULT			\
+	(CEPH_MOUNT_OPT_DCACHE |		\
+	 CEPH_MOUNT_OPT_NOCOPYFROM |		\
+	 CEPH_MOUNT_OPT_ASYNC_DIROPS)
+
+#define ceph_set_mount_opt(fsc, opt) \
+	(fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt
+#define ceph_clear_mount_opt(fsc, opt) \
+	(fsc)->mount_options->flags &= ~CEPH_MOUNT_OPT_##opt
+#define ceph_test_mount_opt(fsc, opt) \
+	(!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt))
+
+/* max size of osd read request, limited by libceph */
+#define CEPH_MAX_READ_SIZE              CEPH_MSG_MAX_DATA_LEN
+/* osd has a configurable limitaion of max write size.
+ * CEPH_MSG_MAX_DATA_LEN should be small enough. */
+#define CEPH_MAX_WRITE_SIZE		CEPH_MSG_MAX_DATA_LEN
+#define CEPH_RASIZE_DEFAULT             (8192*1024)    /* max readahead */
+#define CEPH_MAX_READDIR_DEFAULT        1024
+#define CEPH_MAX_READDIR_BYTES_DEFAULT  (512*1024)
+#define CEPH_SNAPDIRNAME_DEFAULT        ".snap"
+
+/*
+ * Delay telling the MDS we no longer want caps, in case we reopen
+ * the file.  Delay a minimum amount of time, even if we send a cap
+ * message for some other reason.  Otherwise, take the oppotunity to
+ * update the mds to avoid sending another message later.
+ */
+#define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT      5  /* cap release delay */
+#define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT     60  /* cap release delay */
+
+struct ceph_mount_options {
+	unsigned int flags;
+
+	unsigned int wsize;            /* max write size */
+	unsigned int rsize;            /* max read size */
+	unsigned int rasize;           /* max readahead */
+	unsigned int congestion_kb;    /* max writeback in flight */
+	unsigned int caps_wanted_delay_min, caps_wanted_delay_max;
+	int caps_max;
+	unsigned int max_readdir;       /* max readdir result (entries) */
+	unsigned int max_readdir_bytes; /* max readdir result (bytes) */
+
+	bool new_dev_syntax;
+
+	/*
+	 * everything above this point can be memcmp'd; everything below
+	 * is handled in compare_mount_options()
+	 */
+
+	char *snapdir_name;   /* default ".snap" */
+	char *mds_namespace;  /* default NULL */
+	char *server_path;    /* default NULL (means "/") */
+	char *fscache_uniq;   /* default NULL */
+	char *mon_addr;
+};
+
+/* mount state */
+enum {
+	CEPH_MOUNT_MOUNTING,
+	CEPH_MOUNT_MOUNTED,
+	CEPH_MOUNT_UNMOUNTING,
+	CEPH_MOUNT_UNMOUNTED,
+	CEPH_MOUNT_SHUTDOWN,
+	CEPH_MOUNT_RECOVER,
+	CEPH_MOUNT_FENCE_IO,
+};
+
+#define CEPH_ASYNC_CREATE_CONFLICT_BITS 8
+
+struct ceph_fs_client {
+	struct super_block *sb;
+
+	struct list_head metric_wakeup;
+
+	struct ceph_mount_options *mount_options;
+	struct ceph_client *client;
+
+	int mount_state;
+
+	bool blocklisted;
+
+	bool have_copy_from2;
+
+	u32 filp_gen;
+	loff_t max_file_size;
+
+	struct ceph_mds_client *mdsc;
+
+	atomic_long_t writeback_count;
+	bool write_congested;
+
+	struct workqueue_struct *inode_wq;
+	struct workqueue_struct *cap_wq;
+
+	DECLARE_HASHTABLE(async_unlink_conflict, CEPH_ASYNC_CREATE_CONFLICT_BITS);
+	spinlock_t async_unlink_conflict_lock;
+
+#ifdef CONFIG_DEBUG_FS
+	struct dentry *debugfs_dentry_lru, *debugfs_caps;
+	struct dentry *debugfs_congestion_kb;
+	struct dentry *debugfs_bdi;
+	struct dentry *debugfs_mdsc, *debugfs_mdsmap;
+	struct dentry *debugfs_status;
+	struct dentry *debugfs_mds_sessions;
+	struct dentry *debugfs_metrics_dir;
+#endif
+
+#ifdef CONFIG_CEPH_FSCACHE
+	struct fscache_volume *fscache;
+#endif
+};
+
+
+/*
+ * File i/o capability.  This tracks shared state with the metadata
+ * server that allows us to cache or writeback attributes or to read
+ * and write data.  For any given inode, we should have one or more
+ * capabilities, one issued by each metadata server, and our
+ * cumulative access is the OR of all issued capabilities.
+ *
+ * Each cap is referenced by the inode's i_caps rbtree and by per-mds
+ * session capability lists.
+ */
+struct ceph_cap {
+	struct ceph_inode_info *ci;
+	struct rb_node ci_node;          /* per-ci cap tree */
+	struct ceph_mds_session *session;
+	struct list_head session_caps;   /* per-session caplist */
+	u64 cap_id;       /* unique cap id (mds provided) */
+	union {
+		/* in-use caps */
+		struct {
+			int issued;       /* latest, from the mds */
+			int implemented;  /* implemented superset of
+					     issued (for revocation) */
+			int mds;	  /* mds index for this cap */
+			int mds_wanted;   /* caps wanted from this mds */
+		};
+		/* caps to release */
+		struct {
+			u64 cap_ino;
+			int queue_release;
+		};
+	};
+	u32 seq, issue_seq, mseq;
+	u32 cap_gen;      /* active/stale cycle */
+	unsigned long last_used;
+	struct list_head caps_item;
+};
+
+#define CHECK_CAPS_AUTHONLY   1  /* only check auth cap */
+#define CHECK_CAPS_FLUSH      2  /* flush any dirty caps */
+#define CHECK_CAPS_NOINVAL    4  /* don't invalidate pagecache */
+
+struct ceph_cap_flush {
+	u64 tid;
+	int caps;
+	bool wake; /* wake up flush waiters when finish ? */
+	bool is_capsnap; /* true means capsnap */
+	struct list_head g_list; // global
+	struct list_head i_list; // per inode
+};
+
+/*
+ * Snapped cap state that is pending flush to mds.  When a snapshot occurs,
+ * we first complete any in-process sync writes and writeback any dirty
+ * data before flushing the snapped state (tracked here) back to the MDS.
+ */
+struct ceph_cap_snap {
+	refcount_t nref;
+	struct list_head ci_item;
+
+	struct ceph_cap_flush cap_flush;
+
+	u64 follows;
+	int issued, dirty;
+	struct ceph_snap_context *context;
+
+	umode_t mode;
+	kuid_t uid;
+	kgid_t gid;
+
+	struct ceph_buffer *xattr_blob;
+	u64 xattr_version;
+
+	u64 size;
+	u64 change_attr;
+	struct timespec64 mtime, atime, ctime, btime;
+	u64 time_warp_seq;
+	u64 truncate_size;
+	u32 truncate_seq;
+	int writing;   /* a sync write is still in progress */
+	int dirty_pages;     /* dirty pages awaiting writeback */
+	bool inline_data;
+	bool need_flush;
+};
+
+static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap)
+{
+	if (refcount_dec_and_test(&capsnap->nref)) {
+		if (capsnap->xattr_blob)
+			ceph_buffer_put(capsnap->xattr_blob);
+		kmem_cache_free(ceph_cap_snap_cachep, capsnap);
+	}
+}
+
+/*
+ * The frag tree describes how a directory is fragmented, potentially across
+ * multiple metadata servers.  It is also used to indicate points where
+ * metadata authority is delegated, and whether/where metadata is replicated.
+ *
+ * A _leaf_ frag will be present in the i_fragtree IFF there is
+ * delegation info.  That is, if mds >= 0 || ndist > 0.
+ */
+#define CEPH_MAX_DIRFRAG_REP 4
+
+struct ceph_inode_frag {
+	struct rb_node node;
+
+	/* fragtree state */
+	u32 frag;
+	int split_by;         /* i.e. 2^(split_by) children */
+
+	/* delegation and replication info */
+	int mds;              /* -1 if same authority as parent */
+	int ndist;            /* >0 if replicated */
+	int dist[CEPH_MAX_DIRFRAG_REP];
+};
+
+/*
+ * We cache inode xattrs as an encoded blob until they are first used,
+ * at which point we parse them into an rbtree.
+ */
+struct ceph_inode_xattr {
+	struct rb_node node;
+
+	const char *name;
+	int name_len;
+	const char *val;
+	int val_len;
+	int dirty;
+
+	int should_free_name;
+	int should_free_val;
+};
+
+/*
+ * Ceph dentry state
+ */
+struct ceph_dentry_info {
+	struct dentry *dentry;
+	struct ceph_mds_session *lease_session;
+	struct list_head lease_list;
+	struct hlist_node hnode;
+	unsigned long flags;
+	int lease_shared_gen;
+	u32 lease_gen;
+	u32 lease_seq;
+	unsigned long lease_renew_after, lease_renew_from;
+	unsigned long time;
+	u64 offset;
+};
+
+#define CEPH_DENTRY_REFERENCED		(1 << 0)
+#define CEPH_DENTRY_LEASE_LIST		(1 << 1)
+#define CEPH_DENTRY_SHRINK_LIST		(1 << 2)
+#define CEPH_DENTRY_PRIMARY_LINK	(1 << 3)
+#define CEPH_DENTRY_ASYNC_UNLINK_BIT	(4)
+#define CEPH_DENTRY_ASYNC_UNLINK	(1 << CEPH_DENTRY_ASYNC_UNLINK_BIT)
+#define CEPH_DENTRY_ASYNC_CREATE_BIT	(5)
+#define CEPH_DENTRY_ASYNC_CREATE	(1 << CEPH_DENTRY_ASYNC_CREATE_BIT)
+
+struct ceph_inode_xattrs_info {
+	/*
+	 * (still encoded) xattr blob. we avoid the overhead of parsing
+	 * this until someone actually calls getxattr, etc.
+	 *
+	 * blob->vec.iov_len == 4 implies there are no xattrs; blob ==
+	 * NULL means we don't know.
+	*/
+	struct ceph_buffer *blob, *prealloc_blob;
+
+	struct rb_root index;
+	bool dirty;
+	int count;
+	int names_size;
+	int vals_size;
+	u64 version, index_version;
+};
+
+/*
+ * Ceph inode.
+ */
+struct ceph_inode_info {
+	struct netfs_inode netfs; /* Netfslib context and vfs inode */
+	struct ceph_vino i_vino;   /* ceph ino + snap */
+
+	spinlock_t i_ceph_lock;
+
+	u64 i_version;
+	u64 i_inline_version;
+	u32 i_time_warp_seq;
+
+	unsigned long i_ceph_flags;
+	atomic64_t i_release_count;
+	atomic64_t i_ordered_count;
+	atomic64_t i_complete_seq[2];
+
+	struct ceph_dir_layout i_dir_layout;
+	struct ceph_file_layout i_layout;
+	struct ceph_file_layout i_cached_layout;	// for async creates
+	char *i_symlink;
+
+	/* for dirs */
+	struct timespec64 i_rctime;
+	u64 i_rbytes, i_rfiles, i_rsubdirs, i_rsnaps;
+	u64 i_files, i_subdirs;
+
+	/* quotas */
+	u64 i_max_bytes, i_max_files;
+
+	s32 i_dir_pin;
+
+	struct rb_root i_fragtree;
+	int i_fragtree_nsplits;
+	struct mutex i_fragtree_mutex;
+
+	struct ceph_inode_xattrs_info i_xattrs;
+
+	/* capabilities.  protected _both_ by i_ceph_lock and cap->session's
+	 * s_mutex. */
+	struct rb_root i_caps;           /* cap list */
+	struct ceph_cap *i_auth_cap;     /* authoritative cap, if any */
+	unsigned i_dirty_caps, i_flushing_caps;     /* mask of dirtied fields */
+
+	/*
+	 * Link to the auth cap's session's s_cap_dirty list. s_cap_dirty
+	 * is protected by the mdsc->cap_dirty_lock, but each individual item
+	 * is also protected by the inode's i_ceph_lock. Walking s_cap_dirty
+	 * requires the mdsc->cap_dirty_lock. List presence for an item can
+	 * be tested under the i_ceph_lock. Changing anything requires both.
+	 */
+	struct list_head i_dirty_item;
+
+	/*
+	 * Link to session's s_cap_flushing list. Protected in a similar
+	 * fashion to i_dirty_item, but also by the s_mutex for changes. The
+	 * s_cap_flushing list can be walked while holding either the s_mutex
+	 * or msdc->cap_dirty_lock. List presence can also be checked while
+	 * holding the i_ceph_lock for this inode.
+	 */
+	struct list_head i_flushing_item;
+
+	/* we need to track cap writeback on a per-cap-bit basis, to allow
+	 * overlapping, pipelined cap flushes to the mds.  we can probably
+	 * reduce the tid to 8 bits if we're concerned about inode size. */
+	struct ceph_cap_flush *i_prealloc_cap_flush;
+	struct list_head i_cap_flush_list;
+	wait_queue_head_t i_cap_wq;      /* threads waiting on a capability */
+	unsigned long i_hold_caps_max; /* jiffies */
+	struct list_head i_cap_delay_list;  /* for delayed cap release to mds */
+	struct ceph_cap_reservation i_cap_migration_resv;
+	struct list_head i_cap_snaps;   /* snapped state pending flush to mds */
+	struct ceph_snap_context *i_head_snapc;  /* set if wr_buffer_head > 0 or
+						    dirty|flushing caps */
+	unsigned i_snap_caps;           /* cap bits for snapped files */
+
+	unsigned long i_last_rd;
+	unsigned long i_last_wr;
+	int i_nr_by_mode[CEPH_FILE_MODE_BITS];  /* open file counts */
+
+	struct mutex i_truncate_mutex;
+	u32 i_truncate_seq;        /* last truncate to smaller size */
+	u64 i_truncate_size;       /*  and the size we last truncated down to */
+	int i_truncate_pending;    /*  still need to call vmtruncate */
+
+	u64 i_max_size;            /* max file size authorized by mds */
+	u64 i_reported_size; /* (max_)size reported to or requested of mds */
+	u64 i_wanted_max_size;     /* offset we'd like to write too */
+	u64 i_requested_max_size;  /* max_size we've requested */
+
+	/* held references to caps */
+	int i_pin_ref;
+	int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref, i_fx_ref;
+	int i_wrbuffer_ref, i_wrbuffer_ref_head;
+	atomic_t i_filelock_ref;
+	atomic_t i_shared_gen;       /* increment each time we get FILE_SHARED */
+	u32 i_rdcache_gen;      /* incremented each time we get FILE_CACHE. */
+	u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */
+
+	struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */
+	struct list_head i_unsafe_iops;   /* uncommitted mds inode ops */
+	spinlock_t i_unsafe_lock;
+
+	union {
+		struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */
+		struct ceph_snapid_map *i_snapid_map; /* snapid -> dev_t */
+	};
+	struct list_head i_snap_realm_item;
+	struct list_head i_snap_flush_item;
+	struct timespec64 i_btime;
+	struct timespec64 i_snap_btime;
+
+	struct work_struct i_work;
+	unsigned long  i_work_mask;
+};
+
+struct ceph_netfs_request_data {
+	int caps;
+
+	/*
+	 * Maximum size of a file readahead request.
+	 * The fadvise could update the bdi's default ra_pages.
+	 */
+	unsigned int file_ra_pages;
+
+	/* Set it if fadvise disables file readahead entirely */
+	bool file_ra_disabled;
+};
+
+static inline struct ceph_inode_info *
+ceph_inode(const struct inode *inode)
+{
+	return container_of(inode, struct ceph_inode_info, netfs.inode);
+}
+
+static inline struct ceph_fs_client *
+ceph_inode_to_client(const struct inode *inode)
+{
+	return (struct ceph_fs_client *)inode->i_sb->s_fs_info;
+}
+
+static inline struct ceph_fs_client *
+ceph_sb_to_client(const struct super_block *sb)
+{
+	return (struct ceph_fs_client *)sb->s_fs_info;
+}
+
+static inline struct ceph_mds_client *
+ceph_sb_to_mdsc(const struct super_block *sb)
+{
+	return (struct ceph_mds_client *)ceph_sb_to_client(sb)->mdsc;
+}
+
+static inline struct ceph_vino
+ceph_vino(const struct inode *inode)
+{
+	return ceph_inode(inode)->i_vino;
+}
+
+static inline u32 ceph_ino_to_ino32(u64 vino)
+{
+	u32 ino = vino & 0xffffffff;
+	ino ^= vino >> 32;
+	if (!ino)
+		ino = 2;
+	return ino;
+}
+
+/*
+ * Inode numbers in cephfs are 64 bits, but inode->i_ino is 32-bits on
+ * some arches. We generally do not use this value inside the ceph driver, but
+ * we do want to set it to something, so that generic vfs code has an
+ * appropriate value for tracepoints and the like.
+ */
+static inline ino_t ceph_vino_to_ino_t(struct ceph_vino vino)
+{
+	if (sizeof(ino_t) == sizeof(u32))
+		return ceph_ino_to_ino32(vino.ino);
+	return (ino_t)vino.ino;
+}
+
+/* for printf-style formatting */
+#define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap
+
+static inline u64 ceph_ino(struct inode *inode)
+{
+	return ceph_inode(inode)->i_vino.ino;
+}
+
+static inline u64 ceph_snap(struct inode *inode)
+{
+	return ceph_inode(inode)->i_vino.snap;
+}
+
+/**
+ * ceph_present_ino - format an inode number for presentation to userland
+ * @sb: superblock where the inode lives
+ * @ino: inode number to (possibly) convert
+ *
+ * If the user mounted with the ino32 option, then the 64-bit value needs
+ * to be converted to something that can fit inside 32 bits. Note that
+ * internal kernel code never uses this value, so this is entirely for
+ * userland consumption.
+ */
+static inline u64 ceph_present_ino(struct super_block *sb, u64 ino)
+{
+	if (unlikely(ceph_test_mount_opt(ceph_sb_to_client(sb), INO32)))
+		return ceph_ino_to_ino32(ino);
+	return ino;
+}
+
+static inline u64 ceph_present_inode(struct inode *inode)
+{
+	return ceph_present_ino(inode->i_sb, ceph_ino(inode));
+}
+
+static inline int ceph_ino_compare(struct inode *inode, void *data)
+{
+	struct ceph_vino *pvino = (struct ceph_vino *)data;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	return ci->i_vino.ino == pvino->ino &&
+		ci->i_vino.snap == pvino->snap;
+}
+
+/*
+ * The MDS reserves a set of inodes for its own usage. These should never
+ * be accessible by clients, and so the MDS has no reason to ever hand these
+ * out. The range is CEPH_MDS_INO_MDSDIR_OFFSET..CEPH_INO_SYSTEM_BASE.
+ *
+ * These come from src/mds/mdstypes.h in the ceph sources.
+ */
+#define CEPH_MAX_MDS			0x100
+#define CEPH_NUM_STRAY			10
+#define CEPH_MDS_INO_MDSDIR_OFFSET	(1 * CEPH_MAX_MDS)
+#define CEPH_MDS_INO_LOG_OFFSET		(2 * CEPH_MAX_MDS)
+#define CEPH_INO_SYSTEM_BASE		((6*CEPH_MAX_MDS) + (CEPH_MAX_MDS * CEPH_NUM_STRAY))
+
+static inline bool ceph_vino_is_reserved(const struct ceph_vino vino)
+{
+	if (vino.ino >= CEPH_INO_SYSTEM_BASE ||
+	    vino.ino < CEPH_MDS_INO_MDSDIR_OFFSET)
+		return false;
+
+	/* Don't warn on mdsdirs */
+	WARN_RATELIMIT(vino.ino >= CEPH_MDS_INO_LOG_OFFSET,
+			"Attempt to access reserved inode number 0x%llx",
+			vino.ino);
+	return true;
+}
+
+static inline struct inode *ceph_find_inode(struct super_block *sb,
+					    struct ceph_vino vino)
+{
+	if (ceph_vino_is_reserved(vino))
+		return NULL;
+
+	/*
+	 * NB: The hashval will be run through the fs/inode.c hash function
+	 * anyway, so there is no need to squash the inode number down to
+	 * 32-bits first. Just use low-order bits on arches with 32-bit long.
+	 */
+	return ilookup5(sb, (unsigned long)vino.ino, ceph_ino_compare, &vino);
+}
+
+
+/*
+ * Ceph inode.
+ */
+#define CEPH_I_DIR_ORDERED	(1 << 0)  /* dentries in dir are ordered */
+#define CEPH_I_FLUSH		(1 << 2)  /* do not delay flush of dirty metadata */
+#define CEPH_I_POOL_PERM	(1 << 3)  /* pool rd/wr bits are valid */
+#define CEPH_I_POOL_RD		(1 << 4)  /* can read from pool */
+#define CEPH_I_POOL_WR		(1 << 5)  /* can write to pool */
+#define CEPH_I_SEC_INITED	(1 << 6)  /* security initialized */
+#define CEPH_I_KICK_FLUSH	(1 << 7)  /* kick flushing caps */
+#define CEPH_I_FLUSH_SNAPS	(1 << 8)  /* need flush snapss */
+#define CEPH_I_ERROR_WRITE	(1 << 9) /* have seen write errors */
+#define CEPH_I_ERROR_FILELOCK	(1 << 10) /* have seen file lock errors */
+#define CEPH_I_ODIRECT		(1 << 11) /* inode in direct I/O mode */
+#define CEPH_ASYNC_CREATE_BIT	(12)	  /* async create in flight for this */
+#define CEPH_I_ASYNC_CREATE	(1 << CEPH_ASYNC_CREATE_BIT)
+#define CEPH_I_SHUTDOWN		(1 << 13) /* inode is no longer usable */
+
+/*
+ * Masks of ceph inode work.
+ */
+#define CEPH_I_WORK_WRITEBACK		0
+#define CEPH_I_WORK_INVALIDATE_PAGES	1
+#define CEPH_I_WORK_VMTRUNCATE		2
+#define CEPH_I_WORK_CHECK_CAPS		3
+#define CEPH_I_WORK_FLUSH_SNAPS		4
+
+/*
+ * We set the ERROR_WRITE bit when we start seeing write errors on an inode
+ * and then clear it when they start succeeding. Note that we do a lockless
+ * check first, and only take the lock if it looks like it needs to be changed.
+ * The write submission code just takes this as a hint, so we're not too
+ * worried if a few slip through in either direction.
+ */
+static inline void ceph_set_error_write(struct ceph_inode_info *ci)
+{
+	if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) {
+		spin_lock(&ci->i_ceph_lock);
+		ci->i_ceph_flags |= CEPH_I_ERROR_WRITE;
+		spin_unlock(&ci->i_ceph_lock);
+	}
+}
+
+static inline void ceph_clear_error_write(struct ceph_inode_info *ci)
+{
+	if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) {
+		spin_lock(&ci->i_ceph_lock);
+		ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE;
+		spin_unlock(&ci->i_ceph_lock);
+	}
+}
+
+static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci,
+					   long long release_count,
+					   long long ordered_count)
+{
+	/*
+	 * Makes sure operations that setup readdir cache (update page
+	 * cache and i_size) are strongly ordered w.r.t. the following
+	 * atomic64_set() operations.
+	 */
+	smp_mb();
+	atomic64_set(&ci->i_complete_seq[0], release_count);
+	atomic64_set(&ci->i_complete_seq[1], ordered_count);
+}
+
+static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci)
+{
+	atomic64_inc(&ci->i_release_count);
+}
+
+static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci)
+{
+	atomic64_inc(&ci->i_ordered_count);
+}
+
+static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci)
+{
+	return atomic64_read(&ci->i_complete_seq[0]) ==
+		atomic64_read(&ci->i_release_count);
+}
+
+static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci)
+{
+	return  atomic64_read(&ci->i_complete_seq[0]) ==
+		atomic64_read(&ci->i_release_count) &&
+		atomic64_read(&ci->i_complete_seq[1]) ==
+		atomic64_read(&ci->i_ordered_count);
+}
+
+static inline void ceph_dir_clear_complete(struct inode *inode)
+{
+	__ceph_dir_clear_complete(ceph_inode(inode));
+}
+
+static inline void ceph_dir_clear_ordered(struct inode *inode)
+{
+	__ceph_dir_clear_ordered(ceph_inode(inode));
+}
+
+static inline bool ceph_dir_is_complete_ordered(struct inode *inode)
+{
+	bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode));
+	smp_rmb();
+	return ret;
+}
+
+/* find a specific frag @f */
+extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci,
+						u32 f);
+
+/*
+ * choose fragment for value @v.  copy frag content to pfrag, if leaf
+ * exists
+ */
+extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
+			    struct ceph_inode_frag *pfrag,
+			    int *found);
+
+static inline struct ceph_dentry_info *ceph_dentry(const struct dentry *dentry)
+{
+	return (struct ceph_dentry_info *)dentry->d_fsdata;
+}
+
+/*
+ * caps helpers
+ */
+static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci)
+{
+	return !RB_EMPTY_ROOT(&ci->i_caps);
+}
+
+extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented);
+extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t);
+extern int __ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, int mask,
+					  int t);
+extern int __ceph_caps_issued_other(struct ceph_inode_info *ci,
+				    struct ceph_cap *cap);
+
+static inline int ceph_caps_issued(struct ceph_inode_info *ci)
+{
+	int issued;
+	spin_lock(&ci->i_ceph_lock);
+	issued = __ceph_caps_issued(ci, NULL);
+	spin_unlock(&ci->i_ceph_lock);
+	return issued;
+}
+
+static inline int ceph_caps_issued_mask_metric(struct ceph_inode_info *ci,
+					       int mask, int touch)
+{
+	int r;
+	spin_lock(&ci->i_ceph_lock);
+	r = __ceph_caps_issued_mask_metric(ci, mask, touch);
+	spin_unlock(&ci->i_ceph_lock);
+	return r;
+}
+
+static inline int __ceph_caps_dirty(struct ceph_inode_info *ci)
+{
+	return ci->i_dirty_caps | ci->i_flushing_caps;
+}
+extern struct ceph_cap_flush *ceph_alloc_cap_flush(void);
+extern void ceph_free_cap_flush(struct ceph_cap_flush *cf);
+extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask,
+				  struct ceph_cap_flush **pcf);
+
+extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci,
+				      struct ceph_cap *ocap, int mask);
+extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask);
+extern int __ceph_caps_used(struct ceph_inode_info *ci);
+
+static inline bool __ceph_is_file_opened(struct ceph_inode_info *ci)
+{
+	return ci->i_nr_by_mode[0];
+}
+extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci);
+extern int __ceph_caps_wanted(struct ceph_inode_info *ci);
+
+/* what the mds thinks we want */
+extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check);
+
+extern void ceph_caps_init(struct ceph_mds_client *mdsc);
+extern void ceph_caps_finalize(struct ceph_mds_client *mdsc);
+extern void ceph_adjust_caps_max_min(struct ceph_mds_client *mdsc,
+				     struct ceph_mount_options *fsopt);
+extern int ceph_reserve_caps(struct ceph_mds_client *mdsc,
+			     struct ceph_cap_reservation *ctx, int need);
+extern void ceph_unreserve_caps(struct ceph_mds_client *mdsc,
+			       struct ceph_cap_reservation *ctx);
+extern void ceph_reservation_status(struct ceph_fs_client *client,
+				    int *total, int *avail, int *used,
+				    int *reserved, int *min);
+extern void change_auth_cap_ses(struct ceph_inode_info *ci,
+				struct ceph_mds_session *session);
+
+
+
+/*
+ * we keep buffered readdir results attached to file->private_data
+ */
+#define CEPH_F_SYNC     1
+#define CEPH_F_ATEND    2
+
+struct ceph_file_info {
+	short fmode;     /* initialized on open */
+	short flags;     /* CEPH_F_* */
+
+	spinlock_t rw_contexts_lock;
+	struct list_head rw_contexts;
+
+	u32 filp_gen;
+};
+
+struct ceph_dir_file_info {
+	struct ceph_file_info file_info;
+
+	/* readdir: position within the dir */
+	u32 frag;
+	struct ceph_mds_request *last_readdir;
+
+	/* readdir: position within a frag */
+	unsigned next_offset;  /* offset of next chunk (last_name's + 1) */
+	char *last_name;       /* last entry in previous chunk */
+	long long dir_release_count;
+	long long dir_ordered_count;
+	int readdir_cache_idx;
+
+	/* used for -o dirstat read() on directory thing */
+	char *dir_info;
+	int dir_info_len;
+};
+
+struct ceph_rw_context {
+	struct list_head list;
+	struct task_struct *thread;
+	int caps;
+};
+
+#define CEPH_DEFINE_RW_CONTEXT(_name, _caps)	\
+	struct ceph_rw_context _name = {	\
+		.thread = current,		\
+		.caps = _caps,			\
+	}
+
+static inline void ceph_add_rw_context(struct ceph_file_info *cf,
+				       struct ceph_rw_context *ctx)
+{
+	spin_lock(&cf->rw_contexts_lock);
+	list_add(&ctx->list, &cf->rw_contexts);
+	spin_unlock(&cf->rw_contexts_lock);
+}
+
+static inline void ceph_del_rw_context(struct ceph_file_info *cf,
+				       struct ceph_rw_context *ctx)
+{
+	spin_lock(&cf->rw_contexts_lock);
+	list_del(&ctx->list);
+	spin_unlock(&cf->rw_contexts_lock);
+}
+
+static inline struct ceph_rw_context*
+ceph_find_rw_context(struct ceph_file_info *cf)
+{
+	struct ceph_rw_context *ctx, *found = NULL;
+	spin_lock(&cf->rw_contexts_lock);
+	list_for_each_entry(ctx, &cf->rw_contexts, list) {
+		if (ctx->thread == current) {
+			found = ctx;
+			break;
+		}
+	}
+	spin_unlock(&cf->rw_contexts_lock);
+	return found;
+}
+
+struct ceph_readdir_cache_control {
+	struct page  *page;
+	struct dentry **dentries;
+	int index;
+};
+
+/*
+ * A "snap realm" describes a subset of the file hierarchy sharing
+ * the same set of snapshots that apply to it.  The realms themselves
+ * are organized into a hierarchy, such that children inherit (some of)
+ * the snapshots of their parents.
+ *
+ * All inodes within the realm that have capabilities are linked into a
+ * per-realm list.
+ */
+struct ceph_snap_realm {
+	u64 ino;
+	struct inode *inode;
+	atomic_t nref;
+	struct rb_node node;
+
+	u64 created, seq;
+	u64 parent_ino;
+	u64 parent_since;   /* snapid when our current parent became so */
+
+	u64 *prior_parent_snaps;      /* snaps inherited from any parents we */
+	u32 num_prior_parent_snaps;   /*  had prior to parent_since */
+	u64 *snaps;                   /* snaps specific to this realm */
+	u32 num_snaps;
+
+	struct ceph_snap_realm *parent;
+	struct list_head children;       /* list of child realms */
+	struct list_head child_item;
+
+	struct list_head empty_item;     /* if i have ref==0 */
+
+	struct list_head dirty_item;     /* if realm needs new context */
+
+	struct list_head rebuild_item;   /* rebuild snap realms _downward_ in hierarchy */
+
+	/* the current set of snaps for this realm */
+	struct ceph_snap_context *cached_context;
+
+	struct list_head inodes_with_caps;
+	spinlock_t inodes_with_caps_lock;
+};
+
+static inline int default_congestion_kb(void)
+{
+	int congestion_kb;
+
+	/*
+	 * Copied from NFS
+	 *
+	 * congestion size, scale with available memory.
+	 *
+	 *  64MB:    8192k
+	 * 128MB:   11585k
+	 * 256MB:   16384k
+	 * 512MB:   23170k
+	 *   1GB:   32768k
+	 *   2GB:   46340k
+	 *   4GB:   65536k
+	 *   8GB:   92681k
+	 *  16GB:  131072k
+	 *
+	 * This allows larger machines to have larger/more transfers.
+	 * Limit the default to 256M
+	 */
+	congestion_kb = (16*int_sqrt(totalram_pages())) << (PAGE_SHIFT-10);
+	if (congestion_kb > 256*1024)
+		congestion_kb = 256*1024;
+
+	return congestion_kb;
+}
+
+
+/* super.c */
+extern int ceph_force_reconnect(struct super_block *sb);
+/* snap.c */
+struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
+					       u64 ino);
+extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc,
+				struct ceph_snap_realm *realm);
+extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc,
+				struct ceph_snap_realm *realm);
+extern int ceph_update_snap_trace(struct ceph_mds_client *m,
+				  void *p, void *e, bool deletion,
+				  struct ceph_snap_realm **realm_ret);
+void ceph_change_snap_realm(struct inode *inode, struct ceph_snap_realm *realm);
+extern void ceph_handle_snap(struct ceph_mds_client *mdsc,
+			     struct ceph_mds_session *session,
+			     struct ceph_msg *msg);
+extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
+				  struct ceph_cap_snap *capsnap);
+extern void ceph_cleanup_global_and_empty_realms(struct ceph_mds_client *mdsc);
+
+extern struct ceph_snapid_map *ceph_get_snapid_map(struct ceph_mds_client *mdsc,
+						   u64 snap);
+extern void ceph_put_snapid_map(struct ceph_mds_client* mdsc,
+				struct ceph_snapid_map *sm);
+extern void ceph_trim_snapid_map(struct ceph_mds_client *mdsc);
+extern void ceph_cleanup_snapid_map(struct ceph_mds_client *mdsc);
+void ceph_umount_begin(struct super_block *sb);
+
+
+/*
+ * a cap_snap is "pending" if it is still awaiting an in-progress
+ * sync write (that may/may not still update size, mtime, etc.).
+ */
+static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci)
+{
+	return !list_empty(&ci->i_cap_snaps) &&
+	       list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap,
+			       ci_item)->writing;
+}
+
+/* inode.c */
+struct ceph_mds_reply_info_in;
+struct ceph_mds_reply_dirfrag;
+
+extern const struct inode_operations ceph_file_iops;
+
+extern struct inode *ceph_alloc_inode(struct super_block *sb);
+extern void ceph_evict_inode(struct inode *inode);
+extern void ceph_free_inode(struct inode *inode);
+
+extern struct inode *ceph_get_inode(struct super_block *sb,
+				    struct ceph_vino vino);
+extern struct inode *ceph_get_snapdir(struct inode *parent);
+extern int ceph_fill_file_size(struct inode *inode, int issued,
+			       u32 truncate_seq, u64 truncate_size, u64 size);
+extern void ceph_fill_file_time(struct inode *inode, int issued,
+				u64 time_warp_seq, struct timespec64 *ctime,
+				struct timespec64 *mtime,
+				struct timespec64 *atime);
+extern int ceph_fill_inode(struct inode *inode, struct page *locked_page,
+		    struct ceph_mds_reply_info_in *iinfo,
+		    struct ceph_mds_reply_dirfrag *dirinfo,
+		    struct ceph_mds_session *session, int cap_fmode,
+		    struct ceph_cap_reservation *caps_reservation);
+extern int ceph_fill_trace(struct super_block *sb,
+			   struct ceph_mds_request *req);
+extern int ceph_readdir_prepopulate(struct ceph_mds_request *req,
+				    struct ceph_mds_session *session);
+
+extern int ceph_inode_holds_cap(struct inode *inode, int mask);
+
+extern bool ceph_inode_set_size(struct inode *inode, loff_t size);
+extern void __ceph_do_pending_vmtruncate(struct inode *inode);
+
+void ceph_queue_inode_work(struct inode *inode, int work_bit);
+
+static inline void ceph_queue_vmtruncate(struct inode *inode)
+{
+	ceph_queue_inode_work(inode, CEPH_I_WORK_VMTRUNCATE);
+}
+
+static inline void ceph_queue_invalidate(struct inode *inode)
+{
+	ceph_queue_inode_work(inode, CEPH_I_WORK_INVALIDATE_PAGES);
+}
+
+static inline void ceph_queue_writeback(struct inode *inode)
+{
+	ceph_queue_inode_work(inode, CEPH_I_WORK_WRITEBACK);
+}
+
+static inline void ceph_queue_check_caps(struct inode *inode)
+{
+	ceph_queue_inode_work(inode, CEPH_I_WORK_CHECK_CAPS);
+}
+
+static inline void ceph_queue_flush_snaps(struct inode *inode)
+{
+	ceph_queue_inode_work(inode, CEPH_I_WORK_FLUSH_SNAPS);
+}
+
+extern int ceph_try_to_choose_auth_mds(struct inode *inode, int mask);
+extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page,
+			     int mask, bool force);
+static inline int ceph_do_getattr(struct inode *inode, int mask, bool force)
+{
+	return __ceph_do_getattr(inode, NULL, mask, force);
+}
+extern int ceph_permission(struct user_namespace *mnt_userns,
+			   struct inode *inode, int mask);
+extern int __ceph_setattr(struct inode *inode, struct iattr *attr);
+extern int ceph_setattr(struct user_namespace *mnt_userns,
+			struct dentry *dentry, struct iattr *attr);
+extern int ceph_getattr(struct user_namespace *mnt_userns,
+			const struct path *path, struct kstat *stat,
+			u32 request_mask, unsigned int flags);
+void ceph_inode_shutdown(struct inode *inode);
+
+static inline bool ceph_inode_is_shutdown(struct inode *inode)
+{
+	unsigned long flags = READ_ONCE(ceph_inode(inode)->i_ceph_flags);
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	int state = READ_ONCE(fsc->mount_state);
+
+	return (flags & CEPH_I_SHUTDOWN) || state >= CEPH_MOUNT_SHUTDOWN;
+}
+
+/* xattr.c */
+int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int);
+int ceph_do_getvxattr(struct inode *inode, const char *name, void *value, size_t size);
+ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t);
+extern ssize_t ceph_listxattr(struct dentry *, char *, size_t);
+extern struct ceph_buffer *__ceph_build_xattrs_blob(struct ceph_inode_info *ci);
+extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci);
+extern const struct xattr_handler *ceph_xattr_handlers[];
+
+struct ceph_acl_sec_ctx {
+#ifdef CONFIG_CEPH_FS_POSIX_ACL
+	void *default_acl;
+	void *acl;
+#endif
+#ifdef CONFIG_CEPH_FS_SECURITY_LABEL
+	void *sec_ctx;
+	u32 sec_ctxlen;
+#endif
+	struct ceph_pagelist *pagelist;
+};
+
+#ifdef CONFIG_SECURITY
+extern bool ceph_security_xattr_deadlock(struct inode *in);
+extern bool ceph_security_xattr_wanted(struct inode *in);
+#else
+static inline bool ceph_security_xattr_deadlock(struct inode *in)
+{
+	return false;
+}
+static inline bool ceph_security_xattr_wanted(struct inode *in)
+{
+	return false;
+}
+#endif
+
+#ifdef CONFIG_CEPH_FS_SECURITY_LABEL
+extern int ceph_security_init_secctx(struct dentry *dentry, umode_t mode,
+				     struct ceph_acl_sec_ctx *ctx);
+static inline void ceph_security_invalidate_secctx(struct inode *inode)
+{
+	security_inode_invalidate_secctx(inode);
+}
+#else
+static inline int ceph_security_init_secctx(struct dentry *dentry, umode_t mode,
+					    struct ceph_acl_sec_ctx *ctx)
+{
+	return 0;
+}
+static inline void ceph_security_invalidate_secctx(struct inode *inode)
+{
+}
+#endif
+
+void ceph_release_acl_sec_ctx(struct ceph_acl_sec_ctx *as_ctx);
+
+/* acl.c */
+#ifdef CONFIG_CEPH_FS_POSIX_ACL
+
+struct posix_acl *ceph_get_acl(struct inode *, int, bool);
+int ceph_set_acl(struct user_namespace *mnt_userns,
+		 struct inode *inode, struct posix_acl *acl, int type);
+int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
+		       struct ceph_acl_sec_ctx *as_ctx);
+void ceph_init_inode_acls(struct inode *inode,
+			  struct ceph_acl_sec_ctx *as_ctx);
+
+static inline void ceph_forget_all_cached_acls(struct inode *inode)
+{
+       forget_all_cached_acls(inode);
+}
+
+#else
+
+#define ceph_get_acl NULL
+#define ceph_set_acl NULL
+
+static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
+				     struct ceph_acl_sec_ctx *as_ctx)
+{
+	return 0;
+}
+static inline void ceph_init_inode_acls(struct inode *inode,
+					struct ceph_acl_sec_ctx *as_ctx)
+{
+}
+static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode)
+{
+	return 0;
+}
+
+static inline void ceph_forget_all_cached_acls(struct inode *inode)
+{
+}
+
+#endif
+
+/* caps.c */
+extern const char *ceph_cap_string(int c);
+extern void ceph_handle_caps(struct ceph_mds_session *session,
+			     struct ceph_msg *msg);
+extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc,
+				     struct ceph_cap_reservation *ctx);
+extern void ceph_add_cap(struct inode *inode,
+			 struct ceph_mds_session *session, u64 cap_id,
+			 unsigned issued, unsigned wanted,
+			 unsigned cap, unsigned seq, u64 realmino, int flags,
+			 struct ceph_cap **new_cap);
+extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release);
+extern void ceph_remove_cap(struct ceph_cap *cap, bool queue_release);
+extern void __ceph_remove_caps(struct ceph_inode_info *ci);
+extern void ceph_put_cap(struct ceph_mds_client *mdsc,
+			 struct ceph_cap *cap);
+extern int ceph_is_any_caps(struct inode *inode);
+
+extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc);
+extern int ceph_fsync(struct file *file, loff_t start, loff_t end,
+		      int datasync);
+extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc,
+					  struct ceph_mds_session *session);
+extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
+				    struct ceph_mds_session *session);
+void ceph_kick_flushing_inode_caps(struct ceph_mds_session *session,
+				   struct ceph_inode_info *ci);
+extern struct ceph_cap *__get_cap_for_mds(struct ceph_inode_info *ci,
+					  int mds);
+extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci,
+					     int mds);
+extern void ceph_take_cap_refs(struct ceph_inode_info *ci, int caps,
+				bool snap_rwsem_locked);
+extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps);
+extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had);
+extern void ceph_put_cap_refs_async(struct ceph_inode_info *ci, int had);
+extern void ceph_put_cap_refs_no_check_caps(struct ceph_inode_info *ci,
+					    int had);
+extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
+				       struct ceph_snap_context *snapc);
+extern void __ceph_remove_capsnap(struct inode *inode,
+				  struct ceph_cap_snap *capsnap,
+				  bool *wake_ci, bool *wake_mdsc);
+extern void ceph_remove_capsnap(struct inode *inode,
+				struct ceph_cap_snap *capsnap,
+				bool *wake_ci, bool *wake_mdsc);
+extern void ceph_flush_snaps(struct ceph_inode_info *ci,
+			     struct ceph_mds_session **psession);
+extern bool __ceph_should_report_size(struct ceph_inode_info *ci);
+extern void ceph_check_caps(struct ceph_inode_info *ci, int flags,
+			    struct ceph_mds_session *session);
+extern unsigned long ceph_check_delayed_caps(struct ceph_mds_client *mdsc);
+extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc);
+extern int  ceph_drop_caps_for_unlink(struct inode *inode);
+extern int ceph_encode_inode_release(void **p, struct inode *inode,
+				     int mds, int drop, int unless, int force);
+extern int ceph_encode_dentry_release(void **p, struct dentry *dn,
+				      struct inode *dir,
+				      int mds, int drop, int unless);
+
+extern int ceph_get_caps(struct file *filp, int need, int want,
+			 loff_t endoff, int *got);
+extern int ceph_try_get_caps(struct inode *inode,
+			     int need, int want, bool nonblock, int *got);
+
+/* for counting open files by mode */
+extern void ceph_get_fmode(struct ceph_inode_info *ci, int mode, int count);
+extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode, int count);
+extern void __ceph_touch_fmode(struct ceph_inode_info *ci,
+			       struct ceph_mds_client *mdsc, int fmode);
+
+/* addr.c */
+extern const struct address_space_operations ceph_aops;
+extern const struct netfs_request_ops ceph_netfs_ops;
+extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
+extern int ceph_uninline_data(struct file *file);
+extern int ceph_pool_perm_check(struct inode *inode, int need);
+extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc);
+int ceph_purge_inode_cap(struct inode *inode, struct ceph_cap *cap, bool *invalidate);
+
+static inline bool ceph_has_inline_data(struct ceph_inode_info *ci)
+{
+	if (ci->i_inline_version == CEPH_INLINE_NONE ||
+	    ci->i_inline_version == 1) /* initial version, no data */
+		return false;
+	return true;
+}
+
+/* file.c */
+extern const struct file_operations ceph_file_fops;
+
+extern int ceph_renew_caps(struct inode *inode, int fmode);
+extern int ceph_open(struct inode *inode, struct file *file);
+extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry,
+			    struct file *file, unsigned flags, umode_t mode);
+extern int ceph_release(struct inode *inode, struct file *filp);
+extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page,
+				  char *data, size_t len);
+
+/* dir.c */
+extern const struct file_operations ceph_dir_fops;
+extern const struct file_operations ceph_snapdir_fops;
+extern const struct inode_operations ceph_dir_iops;
+extern const struct inode_operations ceph_snapdir_iops;
+extern const struct dentry_operations ceph_dentry_ops;
+
+extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order);
+extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry);
+extern struct dentry *ceph_handle_snapdir(struct ceph_mds_request *req,
+			       struct dentry *dentry);
+extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req,
+					 struct dentry *dentry, int err);
+
+extern void __ceph_dentry_lease_touch(struct ceph_dentry_info *di);
+extern void __ceph_dentry_dir_lease_touch(struct ceph_dentry_info *di);
+extern void ceph_invalidate_dentry_lease(struct dentry *dentry);
+extern int ceph_trim_dentries(struct ceph_mds_client *mdsc);
+extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn);
+extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl);
+
+/* ioctl.c */
+extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
+
+/* export.c */
+extern const struct export_operations ceph_export_ops;
+struct inode *ceph_lookup_inode(struct super_block *sb, u64 ino);
+
+/* locks.c */
+extern __init void ceph_flock_init(void);
+extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl);
+extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl);
+extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num);
+extern int ceph_encode_locks_to_buffer(struct inode *inode,
+				       struct ceph_filelock *flocks,
+				       int num_fcntl_locks,
+				       int num_flock_locks);
+extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks,
+				  struct ceph_pagelist *pagelist,
+				  int num_fcntl_locks, int num_flock_locks);
+
+/* debugfs.c */
+extern void ceph_fs_debugfs_init(struct ceph_fs_client *client);
+extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client);
+
+/* quota.c */
+
+enum quota_get_realm {
+	QUOTA_GET_MAX_FILES,
+	QUOTA_GET_MAX_BYTES,
+	QUOTA_GET_ANY
+};
+
+static inline bool __ceph_has_quota(struct ceph_inode_info *ci,
+				    enum quota_get_realm which)
+{
+	bool has_quota = false;
+
+	switch (which) {
+	case QUOTA_GET_MAX_BYTES:
+		has_quota = !!ci->i_max_bytes;
+		break;
+	case QUOTA_GET_MAX_FILES:
+		has_quota = !!ci->i_max_files;
+		break;
+	default:
+		has_quota = !!(ci->i_max_files || ci->i_max_bytes);
+	}
+	return has_quota;
+}
+
+extern void ceph_adjust_quota_realms_count(struct inode *inode, bool inc);
+
+static inline void __ceph_update_quota(struct ceph_inode_info *ci,
+				       u64 max_bytes, u64 max_files)
+{
+	bool had_quota, has_quota;
+	had_quota = __ceph_has_quota(ci, QUOTA_GET_ANY);
+	ci->i_max_bytes = max_bytes;
+	ci->i_max_files = max_files;
+	has_quota = __ceph_has_quota(ci, QUOTA_GET_ANY);
+
+	if (had_quota != has_quota)
+		ceph_adjust_quota_realms_count(&ci->netfs.inode, has_quota);
+}
+
+extern void ceph_handle_quota(struct ceph_mds_client *mdsc,
+			      struct ceph_mds_session *session,
+			      struct ceph_msg *msg);
+extern bool ceph_quota_is_max_files_exceeded(struct inode *inode);
+extern bool ceph_quota_is_same_realm(struct inode *old, struct inode *new);
+extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode,
+					     loff_t newlen);
+extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode,
+						loff_t newlen);
+extern bool ceph_quota_update_statfs(struct ceph_fs_client *fsc,
+				     struct kstatfs *buf);
+extern void ceph_cleanup_quotarealms_inodes(struct ceph_mds_client *mdsc);
+
+bool ceph_inc_mds_stopping_blocker(struct ceph_mds_client *mdsc,
+			       struct ceph_mds_session *session);
+void ceph_dec_mds_stopping_blocker(struct ceph_mds_client *mdsc);
+#endif /* _FS_CEPH_SUPER_H */
diff --git a/fs/ceph/util.c b/fs/ceph/util.c
new file mode 100644
index 000000000..2c3487567
--- /dev/null
+++ b/fs/ceph/util.c
@@ -0,0 +1,100 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Some non-inline ceph helpers
+ */
+#include <linux/module.h>
+#include <linux/ceph/types.h>
+
+/*
+ * return true if @layout appears to be valid
+ */
+int ceph_file_layout_is_valid(const struct ceph_file_layout *layout)
+{
+	__u32 su = layout->stripe_unit;
+	__u32 sc = layout->stripe_count;
+	__u32 os = layout->object_size;
+
+	/* stripe unit, object size must be non-zero, 64k increment */
+	if (!su || (su & (CEPH_MIN_STRIPE_UNIT-1)))
+		return 0;
+	if (!os || (os & (CEPH_MIN_STRIPE_UNIT-1)))
+		return 0;
+	/* object size must be a multiple of stripe unit */
+	if (os < su || os % su)
+		return 0;
+	/* stripe count must be non-zero */
+	if (!sc)
+		return 0;
+	return 1;
+}
+
+void ceph_file_layout_from_legacy(struct ceph_file_layout *fl,
+				  struct ceph_file_layout_legacy *legacy)
+{
+	fl->stripe_unit = le32_to_cpu(legacy->fl_stripe_unit);
+	fl->stripe_count = le32_to_cpu(legacy->fl_stripe_count);
+	fl->object_size = le32_to_cpu(legacy->fl_object_size);
+	fl->pool_id = le32_to_cpu(legacy->fl_pg_pool);
+	if (fl->pool_id == 0 && fl->stripe_unit == 0 &&
+	    fl->stripe_count == 0 && fl->object_size == 0)
+		fl->pool_id = -1;
+}
+
+void ceph_file_layout_to_legacy(struct ceph_file_layout *fl,
+				struct ceph_file_layout_legacy *legacy)
+{
+	legacy->fl_stripe_unit = cpu_to_le32(fl->stripe_unit);
+	legacy->fl_stripe_count = cpu_to_le32(fl->stripe_count);
+	legacy->fl_object_size = cpu_to_le32(fl->object_size);
+	if (fl->pool_id >= 0)
+		legacy->fl_pg_pool = cpu_to_le32(fl->pool_id);
+	else
+		legacy->fl_pg_pool = 0;
+}
+
+int ceph_flags_to_mode(int flags)
+{
+	int mode;
+
+#ifdef O_DIRECTORY  /* fixme */
+	if ((flags & O_DIRECTORY) == O_DIRECTORY)
+		return CEPH_FILE_MODE_PIN;
+#endif
+
+	switch (flags & O_ACCMODE) {
+	case O_WRONLY:
+		mode = CEPH_FILE_MODE_WR;
+		break;
+	case O_RDONLY:
+		mode = CEPH_FILE_MODE_RD;
+		break;
+	case O_RDWR:
+	case O_ACCMODE: /* this is what the VFS does */
+		mode = CEPH_FILE_MODE_RDWR;
+		break;
+	}
+#ifdef O_LAZY
+	if (flags & O_LAZY)
+		mode |= CEPH_FILE_MODE_LAZY;
+#endif
+
+	return mode;
+}
+
+int ceph_caps_for_mode(int mode)
+{
+	int caps = CEPH_CAP_PIN;
+
+	if (mode & CEPH_FILE_MODE_RD)
+		caps |= CEPH_CAP_FILE_SHARED |
+			CEPH_CAP_FILE_RD | CEPH_CAP_FILE_CACHE;
+	if (mode & CEPH_FILE_MODE_WR)
+		caps |= CEPH_CAP_FILE_EXCL |
+			CEPH_CAP_FILE_WR | CEPH_CAP_FILE_BUFFER |
+			CEPH_CAP_AUTH_SHARED | CEPH_CAP_AUTH_EXCL |
+			CEPH_CAP_XATTR_SHARED | CEPH_CAP_XATTR_EXCL;
+	if (mode & CEPH_FILE_MODE_LAZY)
+		caps |= CEPH_CAP_FILE_LAZYIO;
+
+	return caps;
+}
diff --git a/fs/ceph/xattr.c b/fs/ceph/xattr.c
new file mode 100644
index 000000000..f31350cda
--- /dev/null
+++ b/fs/ceph/xattr.c
@@ -0,0 +1,1420 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/ceph/ceph_debug.h>
+#include <linux/ceph/pagelist.h>
+
+#include "super.h"
+#include "mds_client.h"
+
+#include <linux/ceph/decode.h>
+
+#include <linux/xattr.h>
+#include <linux/security.h>
+#include <linux/posix_acl_xattr.h>
+#include <linux/slab.h>
+
+#define XATTR_CEPH_PREFIX "ceph."
+#define XATTR_CEPH_PREFIX_LEN (sizeof (XATTR_CEPH_PREFIX) - 1)
+
+static int __remove_xattr(struct ceph_inode_info *ci,
+			  struct ceph_inode_xattr *xattr);
+
+static bool ceph_is_valid_xattr(const char *name)
+{
+	return !strncmp(name, XATTR_SECURITY_PREFIX, XATTR_SECURITY_PREFIX_LEN) ||
+	       !strncmp(name, XATTR_CEPH_PREFIX, XATTR_CEPH_PREFIX_LEN) ||
+	       !strncmp(name, XATTR_TRUSTED_PREFIX, XATTR_TRUSTED_PREFIX_LEN) ||
+	       !strncmp(name, XATTR_USER_PREFIX, XATTR_USER_PREFIX_LEN);
+}
+
+/*
+ * These define virtual xattrs exposing the recursive directory
+ * statistics and layout metadata.
+ */
+struct ceph_vxattr {
+	char *name;
+	size_t name_size;	/* strlen(name) + 1 (for '\0') */
+	ssize_t (*getxattr_cb)(struct ceph_inode_info *ci, char *val,
+			       size_t size);
+	bool (*exists_cb)(struct ceph_inode_info *ci);
+	unsigned int flags;
+};
+
+#define VXATTR_FLAG_READONLY		(1<<0)
+#define VXATTR_FLAG_HIDDEN		(1<<1)
+#define VXATTR_FLAG_RSTAT		(1<<2)
+#define VXATTR_FLAG_DIRSTAT		(1<<3)
+
+/* layouts */
+
+static bool ceph_vxattrcb_layout_exists(struct ceph_inode_info *ci)
+{
+	struct ceph_file_layout *fl = &ci->i_layout;
+	return (fl->stripe_unit > 0 || fl->stripe_count > 0 ||
+		fl->object_size > 0 || fl->pool_id >= 0 ||
+		rcu_dereference_raw(fl->pool_ns) != NULL);
+}
+
+static ssize_t ceph_vxattrcb_layout(struct ceph_inode_info *ci, char *val,
+				    size_t size)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(ci->netfs.inode.i_sb);
+	struct ceph_osd_client *osdc = &fsc->client->osdc;
+	struct ceph_string *pool_ns;
+	s64 pool = ci->i_layout.pool_id;
+	const char *pool_name;
+	const char *ns_field = " pool_namespace=";
+	char buf[128];
+	size_t len, total_len = 0;
+	ssize_t ret;
+
+	pool_ns = ceph_try_get_string(ci->i_layout.pool_ns);
+
+	dout("ceph_vxattrcb_layout %p\n", &ci->netfs.inode);
+	down_read(&osdc->lock);
+	pool_name = ceph_pg_pool_name_by_id(osdc->osdmap, pool);
+	if (pool_name) {
+		len = snprintf(buf, sizeof(buf),
+		"stripe_unit=%u stripe_count=%u object_size=%u pool=",
+		ci->i_layout.stripe_unit, ci->i_layout.stripe_count,
+	        ci->i_layout.object_size);
+		total_len = len + strlen(pool_name);
+	} else {
+		len = snprintf(buf, sizeof(buf),
+		"stripe_unit=%u stripe_count=%u object_size=%u pool=%lld",
+		ci->i_layout.stripe_unit, ci->i_layout.stripe_count,
+		ci->i_layout.object_size, pool);
+		total_len = len;
+	}
+
+	if (pool_ns)
+		total_len += strlen(ns_field) + pool_ns->len;
+
+	ret = total_len;
+	if (size >= total_len) {
+		memcpy(val, buf, len);
+		ret = len;
+		if (pool_name) {
+			len = strlen(pool_name);
+			memcpy(val + ret, pool_name, len);
+			ret += len;
+		}
+		if (pool_ns) {
+			len = strlen(ns_field);
+			memcpy(val + ret, ns_field, len);
+			ret += len;
+			memcpy(val + ret, pool_ns->str, pool_ns->len);
+			ret += pool_ns->len;
+		}
+	}
+	up_read(&osdc->lock);
+	ceph_put_string(pool_ns);
+	return ret;
+}
+
+/*
+ * The convention with strings in xattrs is that they should not be NULL
+ * terminated, since we're returning the length with them. snprintf always
+ * NULL terminates however, so call it on a temporary buffer and then memcpy
+ * the result into place.
+ */
+static __printf(3, 4)
+int ceph_fmt_xattr(char *val, size_t size, const char *fmt, ...)
+{
+	int ret;
+	va_list args;
+	char buf[96]; /* NB: reevaluate size if new vxattrs are added */
+
+	va_start(args, fmt);
+	ret = vsnprintf(buf, size ? sizeof(buf) : 0, fmt, args);
+	va_end(args);
+
+	/* Sanity check */
+	if (size && ret + 1 > sizeof(buf)) {
+		WARN_ONCE(true, "Returned length too big (%d)", ret);
+		return -E2BIG;
+	}
+
+	if (ret <= size)
+		memcpy(val, buf, ret);
+	return ret;
+}
+
+static ssize_t ceph_vxattrcb_layout_stripe_unit(struct ceph_inode_info *ci,
+						char *val, size_t size)
+{
+	return ceph_fmt_xattr(val, size, "%u", ci->i_layout.stripe_unit);
+}
+
+static ssize_t ceph_vxattrcb_layout_stripe_count(struct ceph_inode_info *ci,
+						 char *val, size_t size)
+{
+	return ceph_fmt_xattr(val, size, "%u", ci->i_layout.stripe_count);
+}
+
+static ssize_t ceph_vxattrcb_layout_object_size(struct ceph_inode_info *ci,
+						char *val, size_t size)
+{
+	return ceph_fmt_xattr(val, size, "%u", ci->i_layout.object_size);
+}
+
+static ssize_t ceph_vxattrcb_layout_pool(struct ceph_inode_info *ci,
+					 char *val, size_t size)
+{
+	ssize_t ret;
+	struct ceph_fs_client *fsc = ceph_sb_to_client(ci->netfs.inode.i_sb);
+	struct ceph_osd_client *osdc = &fsc->client->osdc;
+	s64 pool = ci->i_layout.pool_id;
+	const char *pool_name;
+
+	down_read(&osdc->lock);
+	pool_name = ceph_pg_pool_name_by_id(osdc->osdmap, pool);
+	if (pool_name) {
+		ret = strlen(pool_name);
+		if (ret <= size)
+			memcpy(val, pool_name, ret);
+	} else {
+		ret = ceph_fmt_xattr(val, size, "%lld", pool);
+	}
+	up_read(&osdc->lock);
+	return ret;
+}
+
+static ssize_t ceph_vxattrcb_layout_pool_namespace(struct ceph_inode_info *ci,
+						   char *val, size_t size)
+{
+	ssize_t ret = 0;
+	struct ceph_string *ns = ceph_try_get_string(ci->i_layout.pool_ns);
+
+	if (ns) {
+		ret = ns->len;
+		if (ret <= size)
+			memcpy(val, ns->str, ret);
+		ceph_put_string(ns);
+	}
+	return ret;
+}
+
+/* directories */
+
+static ssize_t ceph_vxattrcb_dir_entries(struct ceph_inode_info *ci, char *val,
+					 size_t size)
+{
+	return ceph_fmt_xattr(val, size, "%lld", ci->i_files + ci->i_subdirs);
+}
+
+static ssize_t ceph_vxattrcb_dir_files(struct ceph_inode_info *ci, char *val,
+				       size_t size)
+{
+	return ceph_fmt_xattr(val, size, "%lld", ci->i_files);
+}
+
+static ssize_t ceph_vxattrcb_dir_subdirs(struct ceph_inode_info *ci, char *val,
+					 size_t size)
+{
+	return ceph_fmt_xattr(val, size, "%lld", ci->i_subdirs);
+}
+
+static ssize_t ceph_vxattrcb_dir_rentries(struct ceph_inode_info *ci, char *val,
+					  size_t size)
+{
+	return ceph_fmt_xattr(val, size, "%lld",
+				ci->i_rfiles + ci->i_rsubdirs);
+}
+
+static ssize_t ceph_vxattrcb_dir_rfiles(struct ceph_inode_info *ci, char *val,
+					size_t size)
+{
+	return ceph_fmt_xattr(val, size, "%lld", ci->i_rfiles);
+}
+
+static ssize_t ceph_vxattrcb_dir_rsubdirs(struct ceph_inode_info *ci, char *val,
+					  size_t size)
+{
+	return ceph_fmt_xattr(val, size, "%lld", ci->i_rsubdirs);
+}
+
+static ssize_t ceph_vxattrcb_dir_rsnaps(struct ceph_inode_info *ci, char *val,
+					  size_t size)
+{
+	return ceph_fmt_xattr(val, size, "%lld", ci->i_rsnaps);
+}
+
+static ssize_t ceph_vxattrcb_dir_rbytes(struct ceph_inode_info *ci, char *val,
+					size_t size)
+{
+	return ceph_fmt_xattr(val, size, "%lld", ci->i_rbytes);
+}
+
+static ssize_t ceph_vxattrcb_dir_rctime(struct ceph_inode_info *ci, char *val,
+					size_t size)
+{
+	return ceph_fmt_xattr(val, size, "%lld.%09ld", ci->i_rctime.tv_sec,
+				ci->i_rctime.tv_nsec);
+}
+
+/* dir pin */
+static bool ceph_vxattrcb_dir_pin_exists(struct ceph_inode_info *ci)
+{
+	return ci->i_dir_pin != -ENODATA;
+}
+
+static ssize_t ceph_vxattrcb_dir_pin(struct ceph_inode_info *ci, char *val,
+				     size_t size)
+{
+	return ceph_fmt_xattr(val, size, "%d", (int)ci->i_dir_pin);
+}
+
+/* quotas */
+static bool ceph_vxattrcb_quota_exists(struct ceph_inode_info *ci)
+{
+	bool ret = false;
+	spin_lock(&ci->i_ceph_lock);
+	if ((ci->i_max_files || ci->i_max_bytes) &&
+	    ci->i_vino.snap == CEPH_NOSNAP &&
+	    ci->i_snap_realm &&
+	    ci->i_snap_realm->ino == ci->i_vino.ino)
+		ret = true;
+	spin_unlock(&ci->i_ceph_lock);
+	return ret;
+}
+
+static ssize_t ceph_vxattrcb_quota(struct ceph_inode_info *ci, char *val,
+				   size_t size)
+{
+	return ceph_fmt_xattr(val, size, "max_bytes=%llu max_files=%llu",
+				ci->i_max_bytes, ci->i_max_files);
+}
+
+static ssize_t ceph_vxattrcb_quota_max_bytes(struct ceph_inode_info *ci,
+					     char *val, size_t size)
+{
+	return ceph_fmt_xattr(val, size, "%llu", ci->i_max_bytes);
+}
+
+static ssize_t ceph_vxattrcb_quota_max_files(struct ceph_inode_info *ci,
+					     char *val, size_t size)
+{
+	return ceph_fmt_xattr(val, size, "%llu", ci->i_max_files);
+}
+
+/* snapshots */
+static bool ceph_vxattrcb_snap_btime_exists(struct ceph_inode_info *ci)
+{
+	return (ci->i_snap_btime.tv_sec != 0 || ci->i_snap_btime.tv_nsec != 0);
+}
+
+static ssize_t ceph_vxattrcb_snap_btime(struct ceph_inode_info *ci, char *val,
+					size_t size)
+{
+	return ceph_fmt_xattr(val, size, "%lld.%09ld", ci->i_snap_btime.tv_sec,
+				ci->i_snap_btime.tv_nsec);
+}
+
+static ssize_t ceph_vxattrcb_cluster_fsid(struct ceph_inode_info *ci,
+					  char *val, size_t size)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(ci->netfs.inode.i_sb);
+
+	return ceph_fmt_xattr(val, size, "%pU", &fsc->client->fsid);
+}
+
+static ssize_t ceph_vxattrcb_client_id(struct ceph_inode_info *ci,
+				       char *val, size_t size)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(ci->netfs.inode.i_sb);
+
+	return ceph_fmt_xattr(val, size, "client%lld",
+			      ceph_client_gid(fsc->client));
+}
+
+static ssize_t ceph_vxattrcb_caps(struct ceph_inode_info *ci, char *val,
+					size_t size)
+{
+	int issued;
+
+	spin_lock(&ci->i_ceph_lock);
+	issued = __ceph_caps_issued(ci, NULL);
+	spin_unlock(&ci->i_ceph_lock);
+
+	return ceph_fmt_xattr(val, size, "%s/0x%x",
+			      ceph_cap_string(issued), issued);
+}
+
+static ssize_t ceph_vxattrcb_auth_mds(struct ceph_inode_info *ci,
+				       char *val, size_t size)
+{
+	int ret;
+
+	spin_lock(&ci->i_ceph_lock);
+	ret = ceph_fmt_xattr(val, size, "%d",
+			     ci->i_auth_cap ? ci->i_auth_cap->session->s_mds : -1);
+	spin_unlock(&ci->i_ceph_lock);
+	return ret;
+}
+
+#define CEPH_XATTR_NAME(_type, _name)	XATTR_CEPH_PREFIX #_type "." #_name
+#define CEPH_XATTR_NAME2(_type, _name, _name2)	\
+	XATTR_CEPH_PREFIX #_type "." #_name "." #_name2
+
+#define XATTR_NAME_CEPH(_type, _name, _flags)				\
+	{								\
+		.name = CEPH_XATTR_NAME(_type, _name),			\
+		.name_size = sizeof (CEPH_XATTR_NAME(_type, _name)), \
+		.getxattr_cb = ceph_vxattrcb_ ## _type ## _ ## _name, \
+		.exists_cb = NULL,					\
+		.flags = (VXATTR_FLAG_READONLY | _flags),		\
+	}
+#define XATTR_RSTAT_FIELD(_type, _name)			\
+	XATTR_NAME_CEPH(_type, _name, VXATTR_FLAG_RSTAT)
+#define XATTR_RSTAT_FIELD_UPDATABLE(_type, _name)			\
+	{								\
+		.name = CEPH_XATTR_NAME(_type, _name),			\
+		.name_size = sizeof (CEPH_XATTR_NAME(_type, _name)),	\
+		.getxattr_cb = ceph_vxattrcb_ ## _type ## _ ## _name,	\
+		.exists_cb = NULL,					\
+		.flags = VXATTR_FLAG_RSTAT,				\
+	}
+#define XATTR_LAYOUT_FIELD(_type, _name, _field)			\
+	{								\
+		.name = CEPH_XATTR_NAME2(_type, _name, _field),	\
+		.name_size = sizeof (CEPH_XATTR_NAME2(_type, _name, _field)), \
+		.getxattr_cb = ceph_vxattrcb_ ## _name ## _ ## _field, \
+		.exists_cb = ceph_vxattrcb_layout_exists,	\
+		.flags = VXATTR_FLAG_HIDDEN,			\
+	}
+#define XATTR_QUOTA_FIELD(_type, _name)					\
+	{								\
+		.name = CEPH_XATTR_NAME(_type, _name),			\
+		.name_size = sizeof(CEPH_XATTR_NAME(_type, _name)),	\
+		.getxattr_cb = ceph_vxattrcb_ ## _type ## _ ## _name,	\
+		.exists_cb = ceph_vxattrcb_quota_exists,		\
+		.flags = VXATTR_FLAG_HIDDEN,				\
+	}
+
+static struct ceph_vxattr ceph_dir_vxattrs[] = {
+	{
+		.name = "ceph.dir.layout",
+		.name_size = sizeof("ceph.dir.layout"),
+		.getxattr_cb = ceph_vxattrcb_layout,
+		.exists_cb = ceph_vxattrcb_layout_exists,
+		.flags = VXATTR_FLAG_HIDDEN,
+	},
+	XATTR_LAYOUT_FIELD(dir, layout, stripe_unit),
+	XATTR_LAYOUT_FIELD(dir, layout, stripe_count),
+	XATTR_LAYOUT_FIELD(dir, layout, object_size),
+	XATTR_LAYOUT_FIELD(dir, layout, pool),
+	XATTR_LAYOUT_FIELD(dir, layout, pool_namespace),
+	XATTR_NAME_CEPH(dir, entries, VXATTR_FLAG_DIRSTAT),
+	XATTR_NAME_CEPH(dir, files, VXATTR_FLAG_DIRSTAT),
+	XATTR_NAME_CEPH(dir, subdirs, VXATTR_FLAG_DIRSTAT),
+	XATTR_RSTAT_FIELD(dir, rentries),
+	XATTR_RSTAT_FIELD(dir, rfiles),
+	XATTR_RSTAT_FIELD(dir, rsubdirs),
+	XATTR_RSTAT_FIELD(dir, rsnaps),
+	XATTR_RSTAT_FIELD(dir, rbytes),
+	XATTR_RSTAT_FIELD_UPDATABLE(dir, rctime),
+	{
+		.name = "ceph.dir.pin",
+		.name_size = sizeof("ceph.dir.pin"),
+		.getxattr_cb = ceph_vxattrcb_dir_pin,
+		.exists_cb = ceph_vxattrcb_dir_pin_exists,
+		.flags = VXATTR_FLAG_HIDDEN,
+	},
+	{
+		.name = "ceph.quota",
+		.name_size = sizeof("ceph.quota"),
+		.getxattr_cb = ceph_vxattrcb_quota,
+		.exists_cb = ceph_vxattrcb_quota_exists,
+		.flags = VXATTR_FLAG_HIDDEN,
+	},
+	XATTR_QUOTA_FIELD(quota, max_bytes),
+	XATTR_QUOTA_FIELD(quota, max_files),
+	{
+		.name = "ceph.snap.btime",
+		.name_size = sizeof("ceph.snap.btime"),
+		.getxattr_cb = ceph_vxattrcb_snap_btime,
+		.exists_cb = ceph_vxattrcb_snap_btime_exists,
+		.flags = VXATTR_FLAG_READONLY,
+	},
+	{
+		.name = "ceph.caps",
+		.name_size = sizeof("ceph.caps"),
+		.getxattr_cb = ceph_vxattrcb_caps,
+		.exists_cb = NULL,
+		.flags = VXATTR_FLAG_HIDDEN,
+	},
+	{ .name = NULL, 0 }	/* Required table terminator */
+};
+
+/* files */
+
+static struct ceph_vxattr ceph_file_vxattrs[] = {
+	{
+		.name = "ceph.file.layout",
+		.name_size = sizeof("ceph.file.layout"),
+		.getxattr_cb = ceph_vxattrcb_layout,
+		.exists_cb = ceph_vxattrcb_layout_exists,
+		.flags = VXATTR_FLAG_HIDDEN,
+	},
+	XATTR_LAYOUT_FIELD(file, layout, stripe_unit),
+	XATTR_LAYOUT_FIELD(file, layout, stripe_count),
+	XATTR_LAYOUT_FIELD(file, layout, object_size),
+	XATTR_LAYOUT_FIELD(file, layout, pool),
+	XATTR_LAYOUT_FIELD(file, layout, pool_namespace),
+	{
+		.name = "ceph.snap.btime",
+		.name_size = sizeof("ceph.snap.btime"),
+		.getxattr_cb = ceph_vxattrcb_snap_btime,
+		.exists_cb = ceph_vxattrcb_snap_btime_exists,
+		.flags = VXATTR_FLAG_READONLY,
+	},
+	{
+		.name = "ceph.caps",
+		.name_size = sizeof("ceph.caps"),
+		.getxattr_cb = ceph_vxattrcb_caps,
+		.exists_cb = NULL,
+		.flags = VXATTR_FLAG_HIDDEN,
+	},
+	{ .name = NULL, 0 }	/* Required table terminator */
+};
+
+static struct ceph_vxattr ceph_common_vxattrs[] = {
+	{
+		.name = "ceph.cluster_fsid",
+		.name_size = sizeof("ceph.cluster_fsid"),
+		.getxattr_cb = ceph_vxattrcb_cluster_fsid,
+		.exists_cb = NULL,
+		.flags = VXATTR_FLAG_READONLY,
+	},
+	{
+		.name = "ceph.client_id",
+		.name_size = sizeof("ceph.client_id"),
+		.getxattr_cb = ceph_vxattrcb_client_id,
+		.exists_cb = NULL,
+		.flags = VXATTR_FLAG_READONLY,
+	},
+	{
+		.name = "ceph.auth_mds",
+		.name_size = sizeof("ceph.auth_mds"),
+		.getxattr_cb = ceph_vxattrcb_auth_mds,
+		.exists_cb = NULL,
+		.flags = VXATTR_FLAG_READONLY,
+	},
+	{ .name = NULL, 0 }	/* Required table terminator */
+};
+
+static struct ceph_vxattr *ceph_inode_vxattrs(struct inode *inode)
+{
+	if (S_ISDIR(inode->i_mode))
+		return ceph_dir_vxattrs;
+	else if (S_ISREG(inode->i_mode))
+		return ceph_file_vxattrs;
+	return NULL;
+}
+
+static struct ceph_vxattr *ceph_match_vxattr(struct inode *inode,
+						const char *name)
+{
+	struct ceph_vxattr *vxattr = ceph_inode_vxattrs(inode);
+
+	if (vxattr) {
+		while (vxattr->name) {
+			if (!strcmp(vxattr->name, name))
+				return vxattr;
+			vxattr++;
+		}
+	}
+
+	vxattr = ceph_common_vxattrs;
+	while (vxattr->name) {
+		if (!strcmp(vxattr->name, name))
+			return vxattr;
+		vxattr++;
+	}
+
+	return NULL;
+}
+
+static int __set_xattr(struct ceph_inode_info *ci,
+			   const char *name, int name_len,
+			   const char *val, int val_len,
+			   int flags, int update_xattr,
+			   struct ceph_inode_xattr **newxattr)
+{
+	struct rb_node **p;
+	struct rb_node *parent = NULL;
+	struct ceph_inode_xattr *xattr = NULL;
+	int c;
+	int new = 0;
+
+	p = &ci->i_xattrs.index.rb_node;
+	while (*p) {
+		parent = *p;
+		xattr = rb_entry(parent, struct ceph_inode_xattr, node);
+		c = strncmp(name, xattr->name, min(name_len, xattr->name_len));
+		if (c < 0)
+			p = &(*p)->rb_left;
+		else if (c > 0)
+			p = &(*p)->rb_right;
+		else {
+			if (name_len == xattr->name_len)
+				break;
+			else if (name_len < xattr->name_len)
+				p = &(*p)->rb_left;
+			else
+				p = &(*p)->rb_right;
+		}
+		xattr = NULL;
+	}
+
+	if (update_xattr) {
+		int err = 0;
+
+		if (xattr && (flags & XATTR_CREATE))
+			err = -EEXIST;
+		else if (!xattr && (flags & XATTR_REPLACE))
+			err = -ENODATA;
+		if (err) {
+			kfree(name);
+			kfree(val);
+			kfree(*newxattr);
+			return err;
+		}
+		if (update_xattr < 0) {
+			if (xattr)
+				__remove_xattr(ci, xattr);
+			kfree(name);
+			kfree(*newxattr);
+			return 0;
+		}
+	}
+
+	if (!xattr) {
+		new = 1;
+		xattr = *newxattr;
+		xattr->name = name;
+		xattr->name_len = name_len;
+		xattr->should_free_name = update_xattr;
+
+		ci->i_xattrs.count++;
+		dout("__set_xattr count=%d\n", ci->i_xattrs.count);
+	} else {
+		kfree(*newxattr);
+		*newxattr = NULL;
+		if (xattr->should_free_val)
+			kfree(xattr->val);
+
+		if (update_xattr) {
+			kfree(name);
+			name = xattr->name;
+		}
+		ci->i_xattrs.names_size -= xattr->name_len;
+		ci->i_xattrs.vals_size -= xattr->val_len;
+	}
+	ci->i_xattrs.names_size += name_len;
+	ci->i_xattrs.vals_size += val_len;
+	if (val)
+		xattr->val = val;
+	else
+		xattr->val = "";
+
+	xattr->val_len = val_len;
+	xattr->dirty = update_xattr;
+	xattr->should_free_val = (val && update_xattr);
+
+	if (new) {
+		rb_link_node(&xattr->node, parent, p);
+		rb_insert_color(&xattr->node, &ci->i_xattrs.index);
+		dout("__set_xattr_val p=%p\n", p);
+	}
+
+	dout("__set_xattr_val added %llx.%llx xattr %p %.*s=%.*s\n",
+	     ceph_vinop(&ci->netfs.inode), xattr, name_len, name, val_len, val);
+
+	return 0;
+}
+
+static struct ceph_inode_xattr *__get_xattr(struct ceph_inode_info *ci,
+			   const char *name)
+{
+	struct rb_node **p;
+	struct rb_node *parent = NULL;
+	struct ceph_inode_xattr *xattr = NULL;
+	int name_len = strlen(name);
+	int c;
+
+	p = &ci->i_xattrs.index.rb_node;
+	while (*p) {
+		parent = *p;
+		xattr = rb_entry(parent, struct ceph_inode_xattr, node);
+		c = strncmp(name, xattr->name, xattr->name_len);
+		if (c == 0 && name_len > xattr->name_len)
+			c = 1;
+		if (c < 0)
+			p = &(*p)->rb_left;
+		else if (c > 0)
+			p = &(*p)->rb_right;
+		else {
+			dout("__get_xattr %s: found %.*s\n", name,
+			     xattr->val_len, xattr->val);
+			return xattr;
+		}
+	}
+
+	dout("__get_xattr %s: not found\n", name);
+
+	return NULL;
+}
+
+static void __free_xattr(struct ceph_inode_xattr *xattr)
+{
+	BUG_ON(!xattr);
+
+	if (xattr->should_free_name)
+		kfree(xattr->name);
+	if (xattr->should_free_val)
+		kfree(xattr->val);
+
+	kfree(xattr);
+}
+
+static int __remove_xattr(struct ceph_inode_info *ci,
+			  struct ceph_inode_xattr *xattr)
+{
+	if (!xattr)
+		return -ENODATA;
+
+	rb_erase(&xattr->node, &ci->i_xattrs.index);
+
+	if (xattr->should_free_name)
+		kfree(xattr->name);
+	if (xattr->should_free_val)
+		kfree(xattr->val);
+
+	ci->i_xattrs.names_size -= xattr->name_len;
+	ci->i_xattrs.vals_size -= xattr->val_len;
+	ci->i_xattrs.count--;
+	kfree(xattr);
+
+	return 0;
+}
+
+static char *__copy_xattr_names(struct ceph_inode_info *ci,
+				char *dest)
+{
+	struct rb_node *p;
+	struct ceph_inode_xattr *xattr = NULL;
+
+	p = rb_first(&ci->i_xattrs.index);
+	dout("__copy_xattr_names count=%d\n", ci->i_xattrs.count);
+
+	while (p) {
+		xattr = rb_entry(p, struct ceph_inode_xattr, node);
+		memcpy(dest, xattr->name, xattr->name_len);
+		dest[xattr->name_len] = '\0';
+
+		dout("dest=%s %p (%s) (%d/%d)\n", dest, xattr, xattr->name,
+		     xattr->name_len, ci->i_xattrs.names_size);
+
+		dest += xattr->name_len + 1;
+		p = rb_next(p);
+	}
+
+	return dest;
+}
+
+void __ceph_destroy_xattrs(struct ceph_inode_info *ci)
+{
+	struct rb_node *p, *tmp;
+	struct ceph_inode_xattr *xattr = NULL;
+
+	p = rb_first(&ci->i_xattrs.index);
+
+	dout("__ceph_destroy_xattrs p=%p\n", p);
+
+	while (p) {
+		xattr = rb_entry(p, struct ceph_inode_xattr, node);
+		tmp = p;
+		p = rb_next(tmp);
+		dout("__ceph_destroy_xattrs next p=%p (%.*s)\n", p,
+		     xattr->name_len, xattr->name);
+		rb_erase(tmp, &ci->i_xattrs.index);
+
+		__free_xattr(xattr);
+	}
+
+	ci->i_xattrs.names_size = 0;
+	ci->i_xattrs.vals_size = 0;
+	ci->i_xattrs.index_version = 0;
+	ci->i_xattrs.count = 0;
+	ci->i_xattrs.index = RB_ROOT;
+}
+
+static int __build_xattrs(struct inode *inode)
+	__releases(ci->i_ceph_lock)
+	__acquires(ci->i_ceph_lock)
+{
+	u32 namelen;
+	u32 numattr = 0;
+	void *p, *end;
+	u32 len;
+	const char *name, *val;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	u64 xattr_version;
+	struct ceph_inode_xattr **xattrs = NULL;
+	int err = 0;
+	int i;
+
+	dout("__build_xattrs() len=%d\n",
+	     ci->i_xattrs.blob ? (int)ci->i_xattrs.blob->vec.iov_len : 0);
+
+	if (ci->i_xattrs.index_version >= ci->i_xattrs.version)
+		return 0; /* already built */
+
+	__ceph_destroy_xattrs(ci);
+
+start:
+	/* updated internal xattr rb tree */
+	if (ci->i_xattrs.blob && ci->i_xattrs.blob->vec.iov_len > 4) {
+		p = ci->i_xattrs.blob->vec.iov_base;
+		end = p + ci->i_xattrs.blob->vec.iov_len;
+		ceph_decode_32_safe(&p, end, numattr, bad);
+		xattr_version = ci->i_xattrs.version;
+		spin_unlock(&ci->i_ceph_lock);
+
+		xattrs = kcalloc(numattr, sizeof(struct ceph_inode_xattr *),
+				 GFP_NOFS);
+		err = -ENOMEM;
+		if (!xattrs)
+			goto bad_lock;
+
+		for (i = 0; i < numattr; i++) {
+			xattrs[i] = kmalloc(sizeof(struct ceph_inode_xattr),
+					    GFP_NOFS);
+			if (!xattrs[i])
+				goto bad_lock;
+		}
+
+		spin_lock(&ci->i_ceph_lock);
+		if (ci->i_xattrs.version != xattr_version) {
+			/* lost a race, retry */
+			for (i = 0; i < numattr; i++)
+				kfree(xattrs[i]);
+			kfree(xattrs);
+			xattrs = NULL;
+			goto start;
+		}
+		err = -EIO;
+		while (numattr--) {
+			ceph_decode_32_safe(&p, end, len, bad);
+			namelen = len;
+			name = p;
+			p += len;
+			ceph_decode_32_safe(&p, end, len, bad);
+			val = p;
+			p += len;
+
+			err = __set_xattr(ci, name, namelen, val, len,
+					  0, 0, &xattrs[numattr]);
+
+			if (err < 0)
+				goto bad;
+		}
+		kfree(xattrs);
+	}
+	ci->i_xattrs.index_version = ci->i_xattrs.version;
+	ci->i_xattrs.dirty = false;
+
+	return err;
+bad_lock:
+	spin_lock(&ci->i_ceph_lock);
+bad:
+	if (xattrs) {
+		for (i = 0; i < numattr; i++)
+			kfree(xattrs[i]);
+		kfree(xattrs);
+	}
+	ci->i_xattrs.names_size = 0;
+	return err;
+}
+
+static int __get_required_blob_size(struct ceph_inode_info *ci, int name_size,
+				    int val_size)
+{
+	/*
+	 * 4 bytes for the length, and additional 4 bytes per each xattr name,
+	 * 4 bytes per each value
+	 */
+	int size = 4 + ci->i_xattrs.count*(4 + 4) +
+			     ci->i_xattrs.names_size +
+			     ci->i_xattrs.vals_size;
+	dout("__get_required_blob_size c=%d names.size=%d vals.size=%d\n",
+	     ci->i_xattrs.count, ci->i_xattrs.names_size,
+	     ci->i_xattrs.vals_size);
+
+	if (name_size)
+		size += 4 + 4 + name_size + val_size;
+
+	return size;
+}
+
+/*
+ * If there are dirty xattrs, reencode xattrs into the prealloc_blob
+ * and swap into place.  It returns the old i_xattrs.blob (or NULL) so
+ * that it can be freed by the caller as the i_ceph_lock is likely to be
+ * held.
+ */
+struct ceph_buffer *__ceph_build_xattrs_blob(struct ceph_inode_info *ci)
+{
+	struct rb_node *p;
+	struct ceph_inode_xattr *xattr = NULL;
+	struct ceph_buffer *old_blob = NULL;
+	void *dest;
+
+	dout("__build_xattrs_blob %p\n", &ci->netfs.inode);
+	if (ci->i_xattrs.dirty) {
+		int need = __get_required_blob_size(ci, 0, 0);
+
+		BUG_ON(need > ci->i_xattrs.prealloc_blob->alloc_len);
+
+		p = rb_first(&ci->i_xattrs.index);
+		dest = ci->i_xattrs.prealloc_blob->vec.iov_base;
+
+		ceph_encode_32(&dest, ci->i_xattrs.count);
+		while (p) {
+			xattr = rb_entry(p, struct ceph_inode_xattr, node);
+
+			ceph_encode_32(&dest, xattr->name_len);
+			memcpy(dest, xattr->name, xattr->name_len);
+			dest += xattr->name_len;
+			ceph_encode_32(&dest, xattr->val_len);
+			memcpy(dest, xattr->val, xattr->val_len);
+			dest += xattr->val_len;
+
+			p = rb_next(p);
+		}
+
+		/* adjust buffer len; it may be larger than we need */
+		ci->i_xattrs.prealloc_blob->vec.iov_len =
+			dest - ci->i_xattrs.prealloc_blob->vec.iov_base;
+
+		if (ci->i_xattrs.blob)
+			old_blob = ci->i_xattrs.blob;
+		ci->i_xattrs.blob = ci->i_xattrs.prealloc_blob;
+		ci->i_xattrs.prealloc_blob = NULL;
+		ci->i_xattrs.dirty = false;
+		ci->i_xattrs.version++;
+	}
+
+	return old_blob;
+}
+
+static inline int __get_request_mask(struct inode *in) {
+	struct ceph_mds_request *req = current->journal_info;
+	int mask = 0;
+	if (req && req->r_target_inode == in) {
+		if (req->r_op == CEPH_MDS_OP_LOOKUP ||
+		    req->r_op == CEPH_MDS_OP_LOOKUPINO ||
+		    req->r_op == CEPH_MDS_OP_LOOKUPPARENT ||
+		    req->r_op == CEPH_MDS_OP_GETATTR) {
+			mask = le32_to_cpu(req->r_args.getattr.mask);
+		} else if (req->r_op == CEPH_MDS_OP_OPEN ||
+			   req->r_op == CEPH_MDS_OP_CREATE) {
+			mask = le32_to_cpu(req->r_args.open.mask);
+		}
+	}
+	return mask;
+}
+
+ssize_t __ceph_getxattr(struct inode *inode, const char *name, void *value,
+		      size_t size)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_inode_xattr *xattr;
+	struct ceph_vxattr *vxattr;
+	int req_mask;
+	ssize_t err;
+
+	if (strncmp(name, XATTR_CEPH_PREFIX, XATTR_CEPH_PREFIX_LEN))
+		goto handle_non_vxattrs;
+
+	/* let's see if a virtual xattr was requested */
+	vxattr = ceph_match_vxattr(inode, name);
+	if (vxattr) {
+		int mask = 0;
+		if (vxattr->flags & VXATTR_FLAG_RSTAT)
+			mask |= CEPH_STAT_RSTAT;
+		if (vxattr->flags & VXATTR_FLAG_DIRSTAT)
+			mask |= CEPH_CAP_FILE_SHARED;
+		err = ceph_do_getattr(inode, mask, true);
+		if (err)
+			return err;
+		err = -ENODATA;
+		if (!(vxattr->exists_cb && !vxattr->exists_cb(ci))) {
+			err = vxattr->getxattr_cb(ci, value, size);
+			if (size && size < err)
+				err = -ERANGE;
+		}
+		return err;
+	} else {
+		err = ceph_do_getvxattr(inode, name, value, size);
+		/* this would happen with a new client and old server combo */
+		if (err == -EOPNOTSUPP)
+			err = -ENODATA;
+		return err;
+	}
+handle_non_vxattrs:
+	req_mask = __get_request_mask(inode);
+
+	spin_lock(&ci->i_ceph_lock);
+	dout("getxattr %p name '%s' ver=%lld index_ver=%lld\n", inode, name,
+	     ci->i_xattrs.version, ci->i_xattrs.index_version);
+
+	if (ci->i_xattrs.version == 0 ||
+	    !((req_mask & CEPH_CAP_XATTR_SHARED) ||
+	      __ceph_caps_issued_mask_metric(ci, CEPH_CAP_XATTR_SHARED, 1))) {
+		spin_unlock(&ci->i_ceph_lock);
+
+		/* security module gets xattr while filling trace */
+		if (current->journal_info) {
+			pr_warn_ratelimited("sync getxattr %p "
+					    "during filling trace\n", inode);
+			return -EBUSY;
+		}
+
+		/* get xattrs from mds (if we don't already have them) */
+		err = ceph_do_getattr(inode, CEPH_STAT_CAP_XATTR, true);
+		if (err)
+			return err;
+		spin_lock(&ci->i_ceph_lock);
+	}
+
+	err = __build_xattrs(inode);
+	if (err < 0)
+		goto out;
+
+	err = -ENODATA;  /* == ENOATTR */
+	xattr = __get_xattr(ci, name);
+	if (!xattr)
+		goto out;
+
+	err = -ERANGE;
+	if (size && size < xattr->val_len)
+		goto out;
+
+	err = xattr->val_len;
+	if (size == 0)
+		goto out;
+
+	memcpy(value, xattr->val, xattr->val_len);
+
+	if (current->journal_info &&
+	    !strncmp(name, XATTR_SECURITY_PREFIX, XATTR_SECURITY_PREFIX_LEN) &&
+	    security_ismaclabel(name + XATTR_SECURITY_PREFIX_LEN))
+		ci->i_ceph_flags |= CEPH_I_SEC_INITED;
+out:
+	spin_unlock(&ci->i_ceph_lock);
+	return err;
+}
+
+ssize_t ceph_listxattr(struct dentry *dentry, char *names, size_t size)
+{
+	struct inode *inode = d_inode(dentry);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	bool len_only = (size == 0);
+	u32 namelen;
+	int err;
+
+	spin_lock(&ci->i_ceph_lock);
+	dout("listxattr %p ver=%lld index_ver=%lld\n", inode,
+	     ci->i_xattrs.version, ci->i_xattrs.index_version);
+
+	if (ci->i_xattrs.version == 0 ||
+	    !__ceph_caps_issued_mask_metric(ci, CEPH_CAP_XATTR_SHARED, 1)) {
+		spin_unlock(&ci->i_ceph_lock);
+		err = ceph_do_getattr(inode, CEPH_STAT_CAP_XATTR, true);
+		if (err)
+			return err;
+		spin_lock(&ci->i_ceph_lock);
+	}
+
+	err = __build_xattrs(inode);
+	if (err < 0)
+		goto out;
+
+	/* add 1 byte for each xattr due to the null termination */
+	namelen = ci->i_xattrs.names_size + ci->i_xattrs.count;
+	if (!len_only) {
+		if (namelen > size) {
+			err = -ERANGE;
+			goto out;
+		}
+		names = __copy_xattr_names(ci, names);
+		size -= namelen;
+	}
+	err = namelen;
+out:
+	spin_unlock(&ci->i_ceph_lock);
+	return err;
+}
+
+static int ceph_sync_setxattr(struct inode *inode, const char *name,
+			      const char *value, size_t size, int flags)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(inode->i_sb);
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_mds_request *req;
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	struct ceph_osd_client *osdc = &fsc->client->osdc;
+	struct ceph_pagelist *pagelist = NULL;
+	int op = CEPH_MDS_OP_SETXATTR;
+	int err;
+
+	if (size > 0) {
+		/* copy value into pagelist */
+		pagelist = ceph_pagelist_alloc(GFP_NOFS);
+		if (!pagelist)
+			return -ENOMEM;
+
+		err = ceph_pagelist_append(pagelist, value, size);
+		if (err)
+			goto out;
+	} else if (!value) {
+		if (flags & CEPH_XATTR_REPLACE)
+			op = CEPH_MDS_OP_RMXATTR;
+		else
+			flags |= CEPH_XATTR_REMOVE;
+	}
+
+	dout("setxattr value size: %zu\n", size);
+
+	/* do request */
+	req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
+	if (IS_ERR(req)) {
+		err = PTR_ERR(req);
+		goto out;
+	}
+
+	req->r_path2 = kstrdup(name, GFP_NOFS);
+	if (!req->r_path2) {
+		ceph_mdsc_put_request(req);
+		err = -ENOMEM;
+		goto out;
+	}
+
+	if (op == CEPH_MDS_OP_SETXATTR) {
+		req->r_args.setxattr.flags = cpu_to_le32(flags);
+		req->r_args.setxattr.osdmap_epoch =
+			cpu_to_le32(osdc->osdmap->epoch);
+		req->r_pagelist = pagelist;
+		pagelist = NULL;
+	}
+
+	req->r_inode = inode;
+	ihold(inode);
+	req->r_num_caps = 1;
+	req->r_inode_drop = CEPH_CAP_XATTR_SHARED;
+
+	dout("xattr.ver (before): %lld\n", ci->i_xattrs.version);
+	err = ceph_mdsc_do_request(mdsc, NULL, req);
+	ceph_mdsc_put_request(req);
+	dout("xattr.ver (after): %lld\n", ci->i_xattrs.version);
+
+out:
+	if (pagelist)
+		ceph_pagelist_release(pagelist);
+	return err;
+}
+
+int __ceph_setxattr(struct inode *inode, const char *name,
+			const void *value, size_t size, int flags)
+{
+	struct ceph_vxattr *vxattr;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
+	struct ceph_cap_flush *prealloc_cf = NULL;
+	struct ceph_buffer *old_blob = NULL;
+	int issued;
+	int err;
+	int dirty = 0;
+	int name_len = strlen(name);
+	int val_len = size;
+	char *newname = NULL;
+	char *newval = NULL;
+	struct ceph_inode_xattr *xattr = NULL;
+	int required_blob_size;
+	bool check_realm = false;
+	bool lock_snap_rwsem = false;
+
+	if (ceph_snap(inode) != CEPH_NOSNAP)
+		return -EROFS;
+
+	vxattr = ceph_match_vxattr(inode, name);
+	if (vxattr) {
+		if (vxattr->flags & VXATTR_FLAG_READONLY)
+			return -EOPNOTSUPP;
+		if (value && !strncmp(vxattr->name, "ceph.quota", 10))
+			check_realm = true;
+	}
+
+	/* pass any unhandled ceph.* xattrs through to the MDS */
+	if (!strncmp(name, XATTR_CEPH_PREFIX, XATTR_CEPH_PREFIX_LEN))
+		goto do_sync_unlocked;
+
+	/* preallocate memory for xattr name, value, index node */
+	err = -ENOMEM;
+	newname = kmemdup(name, name_len + 1, GFP_NOFS);
+	if (!newname)
+		goto out;
+
+	if (val_len) {
+		newval = kmemdup(value, val_len, GFP_NOFS);
+		if (!newval)
+			goto out;
+	}
+
+	xattr = kmalloc(sizeof(struct ceph_inode_xattr), GFP_NOFS);
+	if (!xattr)
+		goto out;
+
+	prealloc_cf = ceph_alloc_cap_flush();
+	if (!prealloc_cf)
+		goto out;
+
+	spin_lock(&ci->i_ceph_lock);
+retry:
+	issued = __ceph_caps_issued(ci, NULL);
+	required_blob_size = __get_required_blob_size(ci, name_len, val_len);
+	if ((ci->i_xattrs.version == 0) || !(issued & CEPH_CAP_XATTR_EXCL) ||
+	    (required_blob_size > mdsc->mdsmap->m_max_xattr_size)) {
+		dout("%s do sync setxattr: version: %llu size: %d max: %llu\n",
+		     __func__, ci->i_xattrs.version, required_blob_size,
+		     mdsc->mdsmap->m_max_xattr_size);
+		goto do_sync;
+	}
+
+	if (!lock_snap_rwsem && !ci->i_head_snapc) {
+		lock_snap_rwsem = true;
+		if (!down_read_trylock(&mdsc->snap_rwsem)) {
+			spin_unlock(&ci->i_ceph_lock);
+			down_read(&mdsc->snap_rwsem);
+			spin_lock(&ci->i_ceph_lock);
+			goto retry;
+		}
+	}
+
+	dout("setxattr %p name '%s' issued %s\n", inode, name,
+	     ceph_cap_string(issued));
+	__build_xattrs(inode);
+
+	if (!ci->i_xattrs.prealloc_blob ||
+	    required_blob_size > ci->i_xattrs.prealloc_blob->alloc_len) {
+		struct ceph_buffer *blob;
+
+		spin_unlock(&ci->i_ceph_lock);
+		ceph_buffer_put(old_blob); /* Shouldn't be required */
+		dout(" pre-allocating new blob size=%d\n", required_blob_size);
+		blob = ceph_buffer_new(required_blob_size, GFP_NOFS);
+		if (!blob)
+			goto do_sync_unlocked;
+		spin_lock(&ci->i_ceph_lock);
+		/* prealloc_blob can't be released while holding i_ceph_lock */
+		if (ci->i_xattrs.prealloc_blob)
+			old_blob = ci->i_xattrs.prealloc_blob;
+		ci->i_xattrs.prealloc_blob = blob;
+		goto retry;
+	}
+
+	err = __set_xattr(ci, newname, name_len, newval, val_len,
+			  flags, value ? 1 : -1, &xattr);
+
+	if (!err) {
+		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_XATTR_EXCL,
+					       &prealloc_cf);
+		ci->i_xattrs.dirty = true;
+		inode->i_ctime = current_time(inode);
+	}
+
+	spin_unlock(&ci->i_ceph_lock);
+	ceph_buffer_put(old_blob);
+	if (lock_snap_rwsem)
+		up_read(&mdsc->snap_rwsem);
+	if (dirty)
+		__mark_inode_dirty(inode, dirty);
+	ceph_free_cap_flush(prealloc_cf);
+	return err;
+
+do_sync:
+	spin_unlock(&ci->i_ceph_lock);
+do_sync_unlocked:
+	if (lock_snap_rwsem)
+		up_read(&mdsc->snap_rwsem);
+
+	/* security module set xattr while filling trace */
+	if (current->journal_info) {
+		pr_warn_ratelimited("sync setxattr %p "
+				    "during filling trace\n", inode);
+		err = -EBUSY;
+	} else {
+		err = ceph_sync_setxattr(inode, name, value, size, flags);
+		if (err >= 0 && check_realm) {
+			/* check if snaprealm was created for quota inode */
+			spin_lock(&ci->i_ceph_lock);
+			if ((ci->i_max_files || ci->i_max_bytes) &&
+			    !(ci->i_snap_realm &&
+			      ci->i_snap_realm->ino == ci->i_vino.ino))
+				err = -EOPNOTSUPP;
+			spin_unlock(&ci->i_ceph_lock);
+		}
+	}
+out:
+	ceph_free_cap_flush(prealloc_cf);
+	kfree(newname);
+	kfree(newval);
+	kfree(xattr);
+	return err;
+}
+
+static int ceph_get_xattr_handler(const struct xattr_handler *handler,
+				  struct dentry *dentry, struct inode *inode,
+				  const char *name, void *value, size_t size)
+{
+	if (!ceph_is_valid_xattr(name))
+		return -EOPNOTSUPP;
+	return __ceph_getxattr(inode, name, value, size);
+}
+
+static int ceph_set_xattr_handler(const struct xattr_handler *handler,
+				  struct user_namespace *mnt_userns,
+				  struct dentry *unused, struct inode *inode,
+				  const char *name, const void *value,
+				  size_t size, int flags)
+{
+	if (!ceph_is_valid_xattr(name))
+		return -EOPNOTSUPP;
+	return __ceph_setxattr(inode, name, value, size, flags);
+}
+
+static const struct xattr_handler ceph_other_xattr_handler = {
+	.prefix = "",  /* match any name => handlers called with full name */
+	.get = ceph_get_xattr_handler,
+	.set = ceph_set_xattr_handler,
+};
+
+#ifdef CONFIG_SECURITY
+bool ceph_security_xattr_wanted(struct inode *in)
+{
+	return in->i_security != NULL;
+}
+
+bool ceph_security_xattr_deadlock(struct inode *in)
+{
+	struct ceph_inode_info *ci;
+	bool ret;
+	if (!in->i_security)
+		return false;
+	ci = ceph_inode(in);
+	spin_lock(&ci->i_ceph_lock);
+	ret = !(ci->i_ceph_flags & CEPH_I_SEC_INITED) &&
+	      !(ci->i_xattrs.version > 0 &&
+		__ceph_caps_issued_mask(ci, CEPH_CAP_XATTR_SHARED, 0));
+	spin_unlock(&ci->i_ceph_lock);
+	return ret;
+}
+
+#ifdef CONFIG_CEPH_FS_SECURITY_LABEL
+int ceph_security_init_secctx(struct dentry *dentry, umode_t mode,
+			   struct ceph_acl_sec_ctx *as_ctx)
+{
+	struct ceph_pagelist *pagelist = as_ctx->pagelist;
+	const char *name;
+	size_t name_len;
+	int err;
+
+	err = security_dentry_init_security(dentry, mode, &dentry->d_name,
+					    &name, &as_ctx->sec_ctx,
+					    &as_ctx->sec_ctxlen);
+	if (err < 0) {
+		WARN_ON_ONCE(err != -EOPNOTSUPP);
+		err = 0; /* do nothing */
+		goto out;
+	}
+
+	err = -ENOMEM;
+	if (!pagelist) {
+		pagelist = ceph_pagelist_alloc(GFP_KERNEL);
+		if (!pagelist)
+			goto out;
+		err = ceph_pagelist_reserve(pagelist, PAGE_SIZE);
+		if (err)
+			goto out;
+		ceph_pagelist_encode_32(pagelist, 1);
+	}
+
+	/*
+	 * FIXME: Make security_dentry_init_security() generic. Currently
+	 * It only supports single security module and only selinux has
+	 * dentry_init_security hook.
+	 */
+	name_len = strlen(name);
+	err = ceph_pagelist_reserve(pagelist,
+				    4 * 2 + name_len + as_ctx->sec_ctxlen);
+	if (err)
+		goto out;
+
+	if (as_ctx->pagelist) {
+		/* update count of KV pairs */
+		BUG_ON(pagelist->length <= sizeof(__le32));
+		if (list_is_singular(&pagelist->head)) {
+			le32_add_cpu((__le32*)pagelist->mapped_tail, 1);
+		} else {
+			struct page *page = list_first_entry(&pagelist->head,
+							     struct page, lru);
+			void *addr = kmap_atomic(page);
+			le32_add_cpu((__le32*)addr, 1);
+			kunmap_atomic(addr);
+		}
+	} else {
+		as_ctx->pagelist = pagelist;
+	}
+
+	ceph_pagelist_encode_32(pagelist, name_len);
+	ceph_pagelist_append(pagelist, name, name_len);
+
+	ceph_pagelist_encode_32(pagelist, as_ctx->sec_ctxlen);
+	ceph_pagelist_append(pagelist, as_ctx->sec_ctx, as_ctx->sec_ctxlen);
+
+	err = 0;
+out:
+	if (pagelist && !as_ctx->pagelist)
+		ceph_pagelist_release(pagelist);
+	return err;
+}
+#endif /* CONFIG_CEPH_FS_SECURITY_LABEL */
+#endif /* CONFIG_SECURITY */
+
+void ceph_release_acl_sec_ctx(struct ceph_acl_sec_ctx *as_ctx)
+{
+#ifdef CONFIG_CEPH_FS_POSIX_ACL
+	posix_acl_release(as_ctx->acl);
+	posix_acl_release(as_ctx->default_acl);
+#endif
+#ifdef CONFIG_CEPH_FS_SECURITY_LABEL
+	security_release_secctx(as_ctx->sec_ctx, as_ctx->sec_ctxlen);
+#endif
+	if (as_ctx->pagelist)
+		ceph_pagelist_release(as_ctx->pagelist);
+}
+
+/*
+ * List of handlers for synthetic system.* attributes. Other
+ * attributes are handled directly.
+ */
+const struct xattr_handler *ceph_xattr_handlers[] = {
+#ifdef CONFIG_CEPH_FS_POSIX_ACL
+	&posix_acl_access_xattr_handler,
+	&posix_acl_default_xattr_handler,
+#endif
+	&ceph_other_xattr_handler,
+	NULL,
+};