diff options
Diffstat (limited to 'fs/zonefs')
| -rw-r--r-- | fs/zonefs/Kconfig | 11 | ||||
| -rw-r--r-- | fs/zonefs/Makefile | 6 | ||||
| -rw-r--r-- | fs/zonefs/file.c | 847 | ||||
| -rw-r--r-- | fs/zonefs/super.c | 1451 | ||||
| -rw-r--r-- | fs/zonefs/sysfs.c | 134 | ||||
| -rw-r--r-- | fs/zonefs/trace.h | 106 | ||||
| -rw-r--r-- | fs/zonefs/zonefs.h | 289 |
7 files changed, 2844 insertions, 0 deletions
diff --git a/fs/zonefs/Kconfig b/fs/zonefs/Kconfig new file mode 100644 index 0000000000..827278f937 --- /dev/null +++ b/fs/zonefs/Kconfig @@ -0,0 +1,11 @@ +config ZONEFS_FS + tristate "zonefs filesystem support" + depends on BLOCK + depends on BLK_DEV_ZONED + select FS_IOMAP + select CRC32 + help + zonefs is a simple file system which exposes zones of a zoned block + device (e.g. host-managed or host-aware SMR disk drives) as files. + + If unsure, say N. diff --git a/fs/zonefs/Makefile b/fs/zonefs/Makefile new file mode 100644 index 0000000000..645f7229de --- /dev/null +++ b/fs/zonefs/Makefile @@ -0,0 +1,6 @@ +# SPDX-License-Identifier: GPL-2.0 +ccflags-y += -I$(src) + +obj-$(CONFIG_ZONEFS_FS) += zonefs.o + +zonefs-y := super.o file.o sysfs.o diff --git a/fs/zonefs/file.c b/fs/zonefs/file.c new file mode 100644 index 0000000000..b2c9b35df8 --- /dev/null +++ b/fs/zonefs/file.c @@ -0,0 +1,847 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Simple file system for zoned block devices exposing zones as files. + * + * Copyright (C) 2022 Western Digital Corporation or its affiliates. + */ +#include <linux/module.h> +#include <linux/pagemap.h> +#include <linux/iomap.h> +#include <linux/init.h> +#include <linux/slab.h> +#include <linux/blkdev.h> +#include <linux/statfs.h> +#include <linux/writeback.h> +#include <linux/quotaops.h> +#include <linux/seq_file.h> +#include <linux/parser.h> +#include <linux/uio.h> +#include <linux/mman.h> +#include <linux/sched/mm.h> +#include <linux/task_io_accounting_ops.h> + +#include "zonefs.h" + +#include "trace.h" + +static int zonefs_read_iomap_begin(struct inode *inode, loff_t offset, + loff_t length, unsigned int flags, + struct iomap *iomap, struct iomap *srcmap) +{ + struct zonefs_inode_info *zi = ZONEFS_I(inode); + struct zonefs_zone *z = zonefs_inode_zone(inode); + struct super_block *sb = inode->i_sb; + loff_t isize; + + /* + * All blocks are always mapped below EOF. If reading past EOF, + * act as if there is a hole up to the file maximum size. + */ + mutex_lock(&zi->i_truncate_mutex); + iomap->bdev = inode->i_sb->s_bdev; + iomap->offset = ALIGN_DOWN(offset, sb->s_blocksize); + isize = i_size_read(inode); + if (iomap->offset >= isize) { + iomap->type = IOMAP_HOLE; + iomap->addr = IOMAP_NULL_ADDR; + iomap->length = length; + } else { + iomap->type = IOMAP_MAPPED; + iomap->addr = (z->z_sector << SECTOR_SHIFT) + iomap->offset; + iomap->length = isize - iomap->offset; + } + mutex_unlock(&zi->i_truncate_mutex); + + trace_zonefs_iomap_begin(inode, iomap); + + return 0; +} + +static const struct iomap_ops zonefs_read_iomap_ops = { + .iomap_begin = zonefs_read_iomap_begin, +}; + +static int zonefs_write_iomap_begin(struct inode *inode, loff_t offset, + loff_t length, unsigned int flags, + struct iomap *iomap, struct iomap *srcmap) +{ + struct zonefs_inode_info *zi = ZONEFS_I(inode); + struct zonefs_zone *z = zonefs_inode_zone(inode); + struct super_block *sb = inode->i_sb; + loff_t isize; + + /* All write I/Os should always be within the file maximum size */ + if (WARN_ON_ONCE(offset + length > z->z_capacity)) + return -EIO; + + /* + * Sequential zones can only accept direct writes. This is already + * checked when writes are issued, so warn if we see a page writeback + * operation. + */ + if (WARN_ON_ONCE(zonefs_zone_is_seq(z) && !(flags & IOMAP_DIRECT))) + return -EIO; + + /* + * For conventional zones, all blocks are always mapped. For sequential + * zones, all blocks after always mapped below the inode size (zone + * write pointer) and unwriten beyond. + */ + mutex_lock(&zi->i_truncate_mutex); + iomap->bdev = inode->i_sb->s_bdev; + iomap->offset = ALIGN_DOWN(offset, sb->s_blocksize); + iomap->addr = (z->z_sector << SECTOR_SHIFT) + iomap->offset; + isize = i_size_read(inode); + if (iomap->offset >= isize) { + iomap->type = IOMAP_UNWRITTEN; + iomap->length = z->z_capacity - iomap->offset; + } else { + iomap->type = IOMAP_MAPPED; + iomap->length = isize - iomap->offset; + } + mutex_unlock(&zi->i_truncate_mutex); + + trace_zonefs_iomap_begin(inode, iomap); + + return 0; +} + +static const struct iomap_ops zonefs_write_iomap_ops = { + .iomap_begin = zonefs_write_iomap_begin, +}; + +static int zonefs_read_folio(struct file *unused, struct folio *folio) +{ + return iomap_read_folio(folio, &zonefs_read_iomap_ops); +} + +static void zonefs_readahead(struct readahead_control *rac) +{ + iomap_readahead(rac, &zonefs_read_iomap_ops); +} + +/* + * Map blocks for page writeback. This is used only on conventional zone files, + * which implies that the page range can only be within the fixed inode size. + */ +static int zonefs_write_map_blocks(struct iomap_writepage_ctx *wpc, + struct inode *inode, loff_t offset) +{ + struct zonefs_zone *z = zonefs_inode_zone(inode); + + if (WARN_ON_ONCE(zonefs_zone_is_seq(z))) + return -EIO; + if (WARN_ON_ONCE(offset >= i_size_read(inode))) + return -EIO; + + /* If the mapping is already OK, nothing needs to be done */ + if (offset >= wpc->iomap.offset && + offset < wpc->iomap.offset + wpc->iomap.length) + return 0; + + return zonefs_write_iomap_begin(inode, offset, + z->z_capacity - offset, + IOMAP_WRITE, &wpc->iomap, NULL); +} + +static const struct iomap_writeback_ops zonefs_writeback_ops = { + .map_blocks = zonefs_write_map_blocks, +}; + +static int zonefs_writepages(struct address_space *mapping, + struct writeback_control *wbc) +{ + struct iomap_writepage_ctx wpc = { }; + + return iomap_writepages(mapping, wbc, &wpc, &zonefs_writeback_ops); +} + +static int zonefs_swap_activate(struct swap_info_struct *sis, + struct file *swap_file, sector_t *span) +{ + struct inode *inode = file_inode(swap_file); + + if (zonefs_inode_is_seq(inode)) { + zonefs_err(inode->i_sb, + "swap file: not a conventional zone file\n"); + return -EINVAL; + } + + return iomap_swapfile_activate(sis, swap_file, span, + &zonefs_read_iomap_ops); +} + +const struct address_space_operations zonefs_file_aops = { + .read_folio = zonefs_read_folio, + .readahead = zonefs_readahead, + .writepages = zonefs_writepages, + .dirty_folio = iomap_dirty_folio, + .release_folio = iomap_release_folio, + .invalidate_folio = iomap_invalidate_folio, + .migrate_folio = filemap_migrate_folio, + .is_partially_uptodate = iomap_is_partially_uptodate, + .error_remove_page = generic_error_remove_page, + .swap_activate = zonefs_swap_activate, +}; + +int zonefs_file_truncate(struct inode *inode, loff_t isize) +{ + struct zonefs_inode_info *zi = ZONEFS_I(inode); + struct zonefs_zone *z = zonefs_inode_zone(inode); + loff_t old_isize; + enum req_op op; + int ret = 0; + + /* + * Only sequential zone files can be truncated and truncation is allowed + * only down to a 0 size, which is equivalent to a zone reset, and to + * the maximum file size, which is equivalent to a zone finish. + */ + if (!zonefs_zone_is_seq(z)) + return -EPERM; + + if (!isize) + op = REQ_OP_ZONE_RESET; + else if (isize == z->z_capacity) + op = REQ_OP_ZONE_FINISH; + else + return -EPERM; + + inode_dio_wait(inode); + + /* Serialize against page faults */ + filemap_invalidate_lock(inode->i_mapping); + + /* Serialize against zonefs_iomap_begin() */ + mutex_lock(&zi->i_truncate_mutex); + + old_isize = i_size_read(inode); + if (isize == old_isize) + goto unlock; + + ret = zonefs_inode_zone_mgmt(inode, op); + if (ret) + goto unlock; + + /* + * If the mount option ZONEFS_MNTOPT_EXPLICIT_OPEN is set, + * take care of open zones. + */ + if (z->z_flags & ZONEFS_ZONE_OPEN) { + /* + * Truncating a zone to EMPTY or FULL is the equivalent of + * closing the zone. For a truncation to 0, we need to + * re-open the zone to ensure new writes can be processed. + * For a truncation to the maximum file size, the zone is + * closed and writes cannot be accepted anymore, so clear + * the open flag. + */ + if (!isize) + ret = zonefs_inode_zone_mgmt(inode, REQ_OP_ZONE_OPEN); + else + z->z_flags &= ~ZONEFS_ZONE_OPEN; + } + + zonefs_update_stats(inode, isize); + truncate_setsize(inode, isize); + z->z_wpoffset = isize; + zonefs_inode_account_active(inode); + +unlock: + mutex_unlock(&zi->i_truncate_mutex); + filemap_invalidate_unlock(inode->i_mapping); + + return ret; +} + +static int zonefs_file_fsync(struct file *file, loff_t start, loff_t end, + int datasync) +{ + struct inode *inode = file_inode(file); + int ret = 0; + + if (unlikely(IS_IMMUTABLE(inode))) + return -EPERM; + + /* + * Since only direct writes are allowed in sequential files, page cache + * flush is needed only for conventional zone files. + */ + if (zonefs_inode_is_cnv(inode)) + ret = file_write_and_wait_range(file, start, end); + if (!ret) + ret = blkdev_issue_flush(inode->i_sb->s_bdev); + + if (ret) + zonefs_io_error(inode, true); + + return ret; +} + +static vm_fault_t zonefs_filemap_page_mkwrite(struct vm_fault *vmf) +{ + struct inode *inode = file_inode(vmf->vma->vm_file); + vm_fault_t ret; + + if (unlikely(IS_IMMUTABLE(inode))) + return VM_FAULT_SIGBUS; + + /* + * Sanity check: only conventional zone files can have shared + * writeable mappings. + */ + if (zonefs_inode_is_seq(inode)) + return VM_FAULT_NOPAGE; + + sb_start_pagefault(inode->i_sb); + file_update_time(vmf->vma->vm_file); + + /* Serialize against truncates */ + filemap_invalidate_lock_shared(inode->i_mapping); + ret = iomap_page_mkwrite(vmf, &zonefs_write_iomap_ops); + filemap_invalidate_unlock_shared(inode->i_mapping); + + sb_end_pagefault(inode->i_sb); + return ret; +} + +static const struct vm_operations_struct zonefs_file_vm_ops = { + .fault = filemap_fault, + .map_pages = filemap_map_pages, + .page_mkwrite = zonefs_filemap_page_mkwrite, +}; + +static int zonefs_file_mmap(struct file *file, struct vm_area_struct *vma) +{ + /* + * Conventional zones accept random writes, so their files can support + * shared writable mappings. For sequential zone files, only read + * mappings are possible since there are no guarantees for write + * ordering between msync() and page cache writeback. + */ + if (zonefs_inode_is_seq(file_inode(file)) && + (vma->vm_flags & VM_SHARED) && (vma->vm_flags & VM_MAYWRITE)) + return -EINVAL; + + file_accessed(file); + vma->vm_ops = &zonefs_file_vm_ops; + + return 0; +} + +static loff_t zonefs_file_llseek(struct file *file, loff_t offset, int whence) +{ + loff_t isize = i_size_read(file_inode(file)); + + /* + * Seeks are limited to below the zone size for conventional zones + * and below the zone write pointer for sequential zones. In both + * cases, this limit is the inode size. + */ + return generic_file_llseek_size(file, offset, whence, isize, isize); +} + +static int zonefs_file_write_dio_end_io(struct kiocb *iocb, ssize_t size, + int error, unsigned int flags) +{ + struct inode *inode = file_inode(iocb->ki_filp); + struct zonefs_inode_info *zi = ZONEFS_I(inode); + + if (error) { + zonefs_io_error(inode, true); + return error; + } + + if (size && zonefs_inode_is_seq(inode)) { + /* + * Note that we may be seeing completions out of order, + * but that is not a problem since a write completed + * successfully necessarily means that all preceding writes + * were also successful. So we can safely increase the inode + * size to the write end location. + */ + mutex_lock(&zi->i_truncate_mutex); + if (i_size_read(inode) < iocb->ki_pos + size) { + zonefs_update_stats(inode, iocb->ki_pos + size); + zonefs_i_size_write(inode, iocb->ki_pos + size); + } + mutex_unlock(&zi->i_truncate_mutex); + } + + return 0; +} + +static const struct iomap_dio_ops zonefs_write_dio_ops = { + .end_io = zonefs_file_write_dio_end_io, +}; + +/* + * Do not exceed the LFS limits nor the file zone size. If pos is under the + * limit it becomes a short access. If it exceeds the limit, return -EFBIG. + */ +static loff_t zonefs_write_check_limits(struct file *file, loff_t pos, + loff_t count) +{ + struct inode *inode = file_inode(file); + struct zonefs_zone *z = zonefs_inode_zone(inode); + loff_t limit = rlimit(RLIMIT_FSIZE); + loff_t max_size = z->z_capacity; + + if (limit != RLIM_INFINITY) { + if (pos >= limit) { + send_sig(SIGXFSZ, current, 0); + return -EFBIG; + } + count = min(count, limit - pos); + } + + if (!(file->f_flags & O_LARGEFILE)) + max_size = min_t(loff_t, MAX_NON_LFS, max_size); + + if (unlikely(pos >= max_size)) + return -EFBIG; + + return min(count, max_size - pos); +} + +static ssize_t zonefs_write_checks(struct kiocb *iocb, struct iov_iter *from) +{ + struct file *file = iocb->ki_filp; + struct inode *inode = file_inode(file); + struct zonefs_inode_info *zi = ZONEFS_I(inode); + struct zonefs_zone *z = zonefs_inode_zone(inode); + loff_t count; + + if (IS_SWAPFILE(inode)) + return -ETXTBSY; + + if (!iov_iter_count(from)) + return 0; + + if ((iocb->ki_flags & IOCB_NOWAIT) && !(iocb->ki_flags & IOCB_DIRECT)) + return -EINVAL; + + if (iocb->ki_flags & IOCB_APPEND) { + if (zonefs_zone_is_cnv(z)) + return -EINVAL; + mutex_lock(&zi->i_truncate_mutex); + iocb->ki_pos = z->z_wpoffset; + mutex_unlock(&zi->i_truncate_mutex); + } + + count = zonefs_write_check_limits(file, iocb->ki_pos, + iov_iter_count(from)); + if (count < 0) + return count; + + iov_iter_truncate(from, count); + return iov_iter_count(from); +} + +/* + * Handle direct writes. For sequential zone files, this is the only possible + * write path. For these files, check that the user is issuing writes + * sequentially from the end of the file. This code assumes that the block layer + * delivers write requests to the device in sequential order. This is always the + * case if a block IO scheduler implementing the ELEVATOR_F_ZBD_SEQ_WRITE + * elevator feature is being used (e.g. mq-deadline). The block layer always + * automatically select such an elevator for zoned block devices during the + * device initialization. + */ +static ssize_t zonefs_file_dio_write(struct kiocb *iocb, struct iov_iter *from) +{ + struct inode *inode = file_inode(iocb->ki_filp); + struct zonefs_inode_info *zi = ZONEFS_I(inode); + struct zonefs_zone *z = zonefs_inode_zone(inode); + struct super_block *sb = inode->i_sb; + ssize_t ret, count; + + /* + * For async direct IOs to sequential zone files, refuse IOCB_NOWAIT + * as this can cause write reordering (e.g. the first aio gets EAGAIN + * on the inode lock but the second goes through but is now unaligned). + */ + if (zonefs_zone_is_seq(z) && !is_sync_kiocb(iocb) && + (iocb->ki_flags & IOCB_NOWAIT)) + return -EOPNOTSUPP; + + if (iocb->ki_flags & IOCB_NOWAIT) { + if (!inode_trylock(inode)) + return -EAGAIN; + } else { + inode_lock(inode); + } + + count = zonefs_write_checks(iocb, from); + if (count <= 0) { + ret = count; + goto inode_unlock; + } + + if ((iocb->ki_pos | count) & (sb->s_blocksize - 1)) { + ret = -EINVAL; + goto inode_unlock; + } + + /* Enforce sequential writes (append only) in sequential zones */ + if (zonefs_zone_is_seq(z)) { + mutex_lock(&zi->i_truncate_mutex); + if (iocb->ki_pos != z->z_wpoffset) { + mutex_unlock(&zi->i_truncate_mutex); + ret = -EINVAL; + goto inode_unlock; + } + mutex_unlock(&zi->i_truncate_mutex); + } + + /* + * iomap_dio_rw() may return ENOTBLK if there was an issue with + * page invalidation. Overwrite that error code with EBUSY so that + * the user can make sense of the error. + */ + ret = iomap_dio_rw(iocb, from, &zonefs_write_iomap_ops, + &zonefs_write_dio_ops, 0, NULL, 0); + if (ret == -ENOTBLK) + ret = -EBUSY; + + if (zonefs_zone_is_seq(z) && + (ret > 0 || ret == -EIOCBQUEUED)) { + if (ret > 0) + count = ret; + + /* + * Update the zone write pointer offset assuming the write + * operation succeeded. If it did not, the error recovery path + * will correct it. Also do active seq file accounting. + */ + mutex_lock(&zi->i_truncate_mutex); + z->z_wpoffset += count; + zonefs_inode_account_active(inode); + mutex_unlock(&zi->i_truncate_mutex); + } + +inode_unlock: + inode_unlock(inode); + + return ret; +} + +static ssize_t zonefs_file_buffered_write(struct kiocb *iocb, + struct iov_iter *from) +{ + struct inode *inode = file_inode(iocb->ki_filp); + ssize_t ret; + + /* + * Direct IO writes are mandatory for sequential zone files so that the + * write IO issuing order is preserved. + */ + if (zonefs_inode_is_seq(inode)) + return -EIO; + + if (iocb->ki_flags & IOCB_NOWAIT) { + if (!inode_trylock(inode)) + return -EAGAIN; + } else { + inode_lock(inode); + } + + ret = zonefs_write_checks(iocb, from); + if (ret <= 0) + goto inode_unlock; + + ret = iomap_file_buffered_write(iocb, from, &zonefs_write_iomap_ops); + if (ret == -EIO) + zonefs_io_error(inode, true); + +inode_unlock: + inode_unlock(inode); + if (ret > 0) + ret = generic_write_sync(iocb, ret); + + return ret; +} + +static ssize_t zonefs_file_write_iter(struct kiocb *iocb, struct iov_iter *from) +{ + struct inode *inode = file_inode(iocb->ki_filp); + struct zonefs_zone *z = zonefs_inode_zone(inode); + + if (unlikely(IS_IMMUTABLE(inode))) + return -EPERM; + + if (sb_rdonly(inode->i_sb)) + return -EROFS; + + /* Write operations beyond the zone capacity are not allowed */ + if (iocb->ki_pos >= z->z_capacity) + return -EFBIG; + + if (iocb->ki_flags & IOCB_DIRECT) { + ssize_t ret = zonefs_file_dio_write(iocb, from); + + if (ret != -ENOTBLK) + return ret; + } + + return zonefs_file_buffered_write(iocb, from); +} + +static int zonefs_file_read_dio_end_io(struct kiocb *iocb, ssize_t size, + int error, unsigned int flags) +{ + if (error) { + zonefs_io_error(file_inode(iocb->ki_filp), false); + return error; + } + + return 0; +} + +static const struct iomap_dio_ops zonefs_read_dio_ops = { + .end_io = zonefs_file_read_dio_end_io, +}; + +static ssize_t zonefs_file_read_iter(struct kiocb *iocb, struct iov_iter *to) +{ + struct inode *inode = file_inode(iocb->ki_filp); + struct zonefs_inode_info *zi = ZONEFS_I(inode); + struct zonefs_zone *z = zonefs_inode_zone(inode); + struct super_block *sb = inode->i_sb; + loff_t isize; + ssize_t ret; + + /* Offline zones cannot be read */ + if (unlikely(IS_IMMUTABLE(inode) && !(inode->i_mode & 0777))) + return -EPERM; + + if (iocb->ki_pos >= z->z_capacity) + return 0; + + if (iocb->ki_flags & IOCB_NOWAIT) { + if (!inode_trylock_shared(inode)) + return -EAGAIN; + } else { + inode_lock_shared(inode); + } + + /* Limit read operations to written data */ + mutex_lock(&zi->i_truncate_mutex); + isize = i_size_read(inode); + if (iocb->ki_pos >= isize) { + mutex_unlock(&zi->i_truncate_mutex); + ret = 0; + goto inode_unlock; + } + iov_iter_truncate(to, isize - iocb->ki_pos); + mutex_unlock(&zi->i_truncate_mutex); + + if (iocb->ki_flags & IOCB_DIRECT) { + size_t count = iov_iter_count(to); + + if ((iocb->ki_pos | count) & (sb->s_blocksize - 1)) { + ret = -EINVAL; + goto inode_unlock; + } + file_accessed(iocb->ki_filp); + ret = iomap_dio_rw(iocb, to, &zonefs_read_iomap_ops, + &zonefs_read_dio_ops, 0, NULL, 0); + } else { + ret = generic_file_read_iter(iocb, to); + if (ret == -EIO) + zonefs_io_error(inode, false); + } + +inode_unlock: + inode_unlock_shared(inode); + + return ret; +} + +static ssize_t zonefs_file_splice_read(struct file *in, loff_t *ppos, + struct pipe_inode_info *pipe, + size_t len, unsigned int flags) +{ + struct inode *inode = file_inode(in); + struct zonefs_inode_info *zi = ZONEFS_I(inode); + struct zonefs_zone *z = zonefs_inode_zone(inode); + loff_t isize; + ssize_t ret = 0; + + /* Offline zones cannot be read */ + if (unlikely(IS_IMMUTABLE(inode) && !(inode->i_mode & 0777))) + return -EPERM; + + if (*ppos >= z->z_capacity) + return 0; + + inode_lock_shared(inode); + + /* Limit read operations to written data */ + mutex_lock(&zi->i_truncate_mutex); + isize = i_size_read(inode); + if (*ppos >= isize) + len = 0; + else + len = min_t(loff_t, len, isize - *ppos); + mutex_unlock(&zi->i_truncate_mutex); + + if (len > 0) { + ret = filemap_splice_read(in, ppos, pipe, len, flags); + if (ret == -EIO) + zonefs_io_error(inode, false); + } + + inode_unlock_shared(inode); + return ret; +} + +/* + * Write open accounting is done only for sequential files. + */ +static inline bool zonefs_seq_file_need_wro(struct inode *inode, + struct file *file) +{ + if (zonefs_inode_is_cnv(inode)) + return false; + + if (!(file->f_mode & FMODE_WRITE)) + return false; + + return true; +} + +static int zonefs_seq_file_write_open(struct inode *inode) +{ + struct zonefs_inode_info *zi = ZONEFS_I(inode); + struct zonefs_zone *z = zonefs_inode_zone(inode); + int ret = 0; + + mutex_lock(&zi->i_truncate_mutex); + + if (!zi->i_wr_refcnt) { + struct zonefs_sb_info *sbi = ZONEFS_SB(inode->i_sb); + unsigned int wro = atomic_inc_return(&sbi->s_wro_seq_files); + + if (sbi->s_mount_opts & ZONEFS_MNTOPT_EXPLICIT_OPEN) { + + if (sbi->s_max_wro_seq_files + && wro > sbi->s_max_wro_seq_files) { + atomic_dec(&sbi->s_wro_seq_files); + ret = -EBUSY; + goto unlock; + } + + if (i_size_read(inode) < z->z_capacity) { + ret = zonefs_inode_zone_mgmt(inode, + REQ_OP_ZONE_OPEN); + if (ret) { + atomic_dec(&sbi->s_wro_seq_files); + goto unlock; + } + z->z_flags |= ZONEFS_ZONE_OPEN; + zonefs_inode_account_active(inode); + } + } + } + + zi->i_wr_refcnt++; + +unlock: + mutex_unlock(&zi->i_truncate_mutex); + + return ret; +} + +static int zonefs_file_open(struct inode *inode, struct file *file) +{ + int ret; + + file->f_mode |= FMODE_CAN_ODIRECT; + ret = generic_file_open(inode, file); + if (ret) + return ret; + + if (zonefs_seq_file_need_wro(inode, file)) + return zonefs_seq_file_write_open(inode); + + return 0; +} + +static void zonefs_seq_file_write_close(struct inode *inode) +{ + struct zonefs_inode_info *zi = ZONEFS_I(inode); + struct zonefs_zone *z = zonefs_inode_zone(inode); + struct super_block *sb = inode->i_sb; + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + int ret = 0; + + mutex_lock(&zi->i_truncate_mutex); + + zi->i_wr_refcnt--; + if (zi->i_wr_refcnt) + goto unlock; + + /* + * The file zone may not be open anymore (e.g. the file was truncated to + * its maximum size or it was fully written). For this case, we only + * need to decrement the write open count. + */ + if (z->z_flags & ZONEFS_ZONE_OPEN) { + ret = zonefs_inode_zone_mgmt(inode, REQ_OP_ZONE_CLOSE); + if (ret) { + __zonefs_io_error(inode, false); + /* + * Leaving zones explicitly open may lead to a state + * where most zones cannot be written (zone resources + * exhausted). So take preventive action by remounting + * read-only. + */ + if (z->z_flags & ZONEFS_ZONE_OPEN && + !(sb->s_flags & SB_RDONLY)) { + zonefs_warn(sb, + "closing zone at %llu failed %d\n", + z->z_sector, ret); + zonefs_warn(sb, + "remounting filesystem read-only\n"); + sb->s_flags |= SB_RDONLY; + } + goto unlock; + } + + z->z_flags &= ~ZONEFS_ZONE_OPEN; + zonefs_inode_account_active(inode); + } + + atomic_dec(&sbi->s_wro_seq_files); + +unlock: + mutex_unlock(&zi->i_truncate_mutex); +} + +static int zonefs_file_release(struct inode *inode, struct file *file) +{ + /* + * If we explicitly open a zone we must close it again as well, but the + * zone management operation can fail (either due to an IO error or as + * the zone has gone offline or read-only). Make sure we don't fail the + * close(2) for user-space. + */ + if (zonefs_seq_file_need_wro(inode, file)) + zonefs_seq_file_write_close(inode); + + return 0; +} + +const struct file_operations zonefs_file_operations = { + .open = zonefs_file_open, + .release = zonefs_file_release, + .fsync = zonefs_file_fsync, + .mmap = zonefs_file_mmap, + .llseek = zonefs_file_llseek, + .read_iter = zonefs_file_read_iter, + .write_iter = zonefs_file_write_iter, + .splice_read = zonefs_file_splice_read, + .splice_write = iter_file_splice_write, + .iopoll = iocb_bio_iopoll, +}; diff --git a/fs/zonefs/super.c b/fs/zonefs/super.c new file mode 100644 index 0000000000..9d1a9808fb --- /dev/null +++ b/fs/zonefs/super.c @@ -0,0 +1,1451 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Simple file system for zoned block devices exposing zones as files. + * + * Copyright (C) 2019 Western Digital Corporation or its affiliates. + */ +#include <linux/module.h> +#include <linux/pagemap.h> +#include <linux/magic.h> +#include <linux/iomap.h> +#include <linux/init.h> +#include <linux/slab.h> +#include <linux/blkdev.h> +#include <linux/statfs.h> +#include <linux/writeback.h> +#include <linux/quotaops.h> +#include <linux/seq_file.h> +#include <linux/parser.h> +#include <linux/uio.h> +#include <linux/mman.h> +#include <linux/sched/mm.h> +#include <linux/crc32.h> +#include <linux/task_io_accounting_ops.h> + +#include "zonefs.h" + +#define CREATE_TRACE_POINTS +#include "trace.h" + +/* + * Get the name of a zone group directory. + */ +static const char *zonefs_zgroup_name(enum zonefs_ztype ztype) +{ + switch (ztype) { + case ZONEFS_ZTYPE_CNV: + return "cnv"; + case ZONEFS_ZTYPE_SEQ: + return "seq"; + default: + WARN_ON_ONCE(1); + return "???"; + } +} + +/* + * Manage the active zone count. + */ +static void zonefs_account_active(struct super_block *sb, + struct zonefs_zone *z) +{ + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + + if (zonefs_zone_is_cnv(z)) + return; + + /* + * For zones that transitioned to the offline or readonly condition, + * we only need to clear the active state. + */ + if (z->z_flags & (ZONEFS_ZONE_OFFLINE | ZONEFS_ZONE_READONLY)) + goto out; + + /* + * If the zone is active, that is, if it is explicitly open or + * partially written, check if it was already accounted as active. + */ + if ((z->z_flags & ZONEFS_ZONE_OPEN) || + (z->z_wpoffset > 0 && z->z_wpoffset < z->z_capacity)) { + if (!(z->z_flags & ZONEFS_ZONE_ACTIVE)) { + z->z_flags |= ZONEFS_ZONE_ACTIVE; + atomic_inc(&sbi->s_active_seq_files); + } + return; + } + +out: + /* The zone is not active. If it was, update the active count */ + if (z->z_flags & ZONEFS_ZONE_ACTIVE) { + z->z_flags &= ~ZONEFS_ZONE_ACTIVE; + atomic_dec(&sbi->s_active_seq_files); + } +} + +/* + * Manage the active zone count. Called with zi->i_truncate_mutex held. + */ +void zonefs_inode_account_active(struct inode *inode) +{ + lockdep_assert_held(&ZONEFS_I(inode)->i_truncate_mutex); + + return zonefs_account_active(inode->i_sb, zonefs_inode_zone(inode)); +} + +/* + * Execute a zone management operation. + */ +static int zonefs_zone_mgmt(struct super_block *sb, + struct zonefs_zone *z, enum req_op op) +{ + int ret; + + /* + * With ZNS drives, closing an explicitly open zone that has not been + * written will change the zone state to "closed", that is, the zone + * will remain active. Since this can then cause failure of explicit + * open operation on other zones if the drive active zone resources + * are exceeded, make sure that the zone does not remain active by + * resetting it. + */ + if (op == REQ_OP_ZONE_CLOSE && !z->z_wpoffset) + op = REQ_OP_ZONE_RESET; + + trace_zonefs_zone_mgmt(sb, z, op); + ret = blkdev_zone_mgmt(sb->s_bdev, op, z->z_sector, + z->z_size >> SECTOR_SHIFT, GFP_NOFS); + if (ret) { + zonefs_err(sb, + "Zone management operation %s at %llu failed %d\n", + blk_op_str(op), z->z_sector, ret); + return ret; + } + + return 0; +} + +int zonefs_inode_zone_mgmt(struct inode *inode, enum req_op op) +{ + lockdep_assert_held(&ZONEFS_I(inode)->i_truncate_mutex); + + return zonefs_zone_mgmt(inode->i_sb, zonefs_inode_zone(inode), op); +} + +void zonefs_i_size_write(struct inode *inode, loff_t isize) +{ + struct zonefs_zone *z = zonefs_inode_zone(inode); + + i_size_write(inode, isize); + + /* + * A full zone is no longer open/active and does not need + * explicit closing. + */ + if (isize >= z->z_capacity) { + struct zonefs_sb_info *sbi = ZONEFS_SB(inode->i_sb); + + if (z->z_flags & ZONEFS_ZONE_ACTIVE) + atomic_dec(&sbi->s_active_seq_files); + z->z_flags &= ~(ZONEFS_ZONE_OPEN | ZONEFS_ZONE_ACTIVE); + } +} + +void zonefs_update_stats(struct inode *inode, loff_t new_isize) +{ + struct super_block *sb = inode->i_sb; + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + loff_t old_isize = i_size_read(inode); + loff_t nr_blocks; + + if (new_isize == old_isize) + return; + + spin_lock(&sbi->s_lock); + + /* + * This may be called for an update after an IO error. + * So beware of the values seen. + */ + if (new_isize < old_isize) { + nr_blocks = (old_isize - new_isize) >> sb->s_blocksize_bits; + if (sbi->s_used_blocks > nr_blocks) + sbi->s_used_blocks -= nr_blocks; + else + sbi->s_used_blocks = 0; + } else { + sbi->s_used_blocks += + (new_isize - old_isize) >> sb->s_blocksize_bits; + if (sbi->s_used_blocks > sbi->s_blocks) + sbi->s_used_blocks = sbi->s_blocks; + } + + spin_unlock(&sbi->s_lock); +} + +/* + * Check a zone condition. Return the amount of written (and still readable) + * data in the zone. + */ +static loff_t zonefs_check_zone_condition(struct super_block *sb, + struct zonefs_zone *z, + struct blk_zone *zone) +{ + switch (zone->cond) { + case BLK_ZONE_COND_OFFLINE: + zonefs_warn(sb, "Zone %llu: offline zone\n", + z->z_sector); + z->z_flags |= ZONEFS_ZONE_OFFLINE; + return 0; + case BLK_ZONE_COND_READONLY: + /* + * The write pointer of read-only zones is invalid, so we cannot + * determine the zone wpoffset (inode size). We thus keep the + * zone wpoffset as is, which leads to an empty file + * (wpoffset == 0) on mount. For a runtime error, this keeps + * the inode size as it was when last updated so that the user + * can recover data. + */ + zonefs_warn(sb, "Zone %llu: read-only zone\n", + z->z_sector); + z->z_flags |= ZONEFS_ZONE_READONLY; + if (zonefs_zone_is_cnv(z)) + return z->z_capacity; + return z->z_wpoffset; + case BLK_ZONE_COND_FULL: + /* The write pointer of full zones is invalid. */ + return z->z_capacity; + default: + if (zonefs_zone_is_cnv(z)) + return z->z_capacity; + return (zone->wp - zone->start) << SECTOR_SHIFT; + } +} + +/* + * Check a zone condition and adjust its inode access permissions for + * offline and readonly zones. + */ +static void zonefs_inode_update_mode(struct inode *inode) +{ + struct zonefs_zone *z = zonefs_inode_zone(inode); + + if (z->z_flags & ZONEFS_ZONE_OFFLINE) { + /* Offline zones cannot be read nor written */ + inode->i_flags |= S_IMMUTABLE; + inode->i_mode &= ~0777; + } else if (z->z_flags & ZONEFS_ZONE_READONLY) { + /* Readonly zones cannot be written */ + inode->i_flags |= S_IMMUTABLE; + if (z->z_flags & ZONEFS_ZONE_INIT_MODE) + inode->i_mode &= ~0777; + else + inode->i_mode &= ~0222; + } + + z->z_flags &= ~ZONEFS_ZONE_INIT_MODE; + z->z_mode = inode->i_mode; +} + +struct zonefs_ioerr_data { + struct inode *inode; + bool write; +}; + +static int zonefs_io_error_cb(struct blk_zone *zone, unsigned int idx, + void *data) +{ + struct zonefs_ioerr_data *err = data; + struct inode *inode = err->inode; + struct zonefs_zone *z = zonefs_inode_zone(inode); + struct super_block *sb = inode->i_sb; + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + loff_t isize, data_size; + + /* + * Check the zone condition: if the zone is not "bad" (offline or + * read-only), read errors are simply signaled to the IO issuer as long + * as there is no inconsistency between the inode size and the amount of + * data writen in the zone (data_size). + */ + data_size = zonefs_check_zone_condition(sb, z, zone); + isize = i_size_read(inode); + if (!(z->z_flags & (ZONEFS_ZONE_READONLY | ZONEFS_ZONE_OFFLINE)) && + !err->write && isize == data_size) + return 0; + + /* + * At this point, we detected either a bad zone or an inconsistency + * between the inode size and the amount of data written in the zone. + * For the latter case, the cause may be a write IO error or an external + * action on the device. Two error patterns exist: + * 1) The inode size is lower than the amount of data in the zone: + * a write operation partially failed and data was writen at the end + * of the file. This can happen in the case of a large direct IO + * needing several BIOs and/or write requests to be processed. + * 2) The inode size is larger than the amount of data in the zone: + * this can happen with a deferred write error with the use of the + * device side write cache after getting successful write IO + * completions. Other possibilities are (a) an external corruption, + * e.g. an application reset the zone directly, or (b) the device + * has a serious problem (e.g. firmware bug). + * + * In all cases, warn about inode size inconsistency and handle the + * IO error according to the zone condition and to the mount options. + */ + if (zonefs_zone_is_seq(z) && isize != data_size) + zonefs_warn(sb, + "inode %lu: invalid size %lld (should be %lld)\n", + inode->i_ino, isize, data_size); + + /* + * First handle bad zones signaled by hardware. The mount options + * errors=zone-ro and errors=zone-offline result in changing the + * zone condition to read-only and offline respectively, as if the + * condition was signaled by the hardware. + */ + if ((z->z_flags & ZONEFS_ZONE_OFFLINE) || + (sbi->s_mount_opts & ZONEFS_MNTOPT_ERRORS_ZOL)) { + zonefs_warn(sb, "inode %lu: read/write access disabled\n", + inode->i_ino); + if (!(z->z_flags & ZONEFS_ZONE_OFFLINE)) + z->z_flags |= ZONEFS_ZONE_OFFLINE; + zonefs_inode_update_mode(inode); + data_size = 0; + } else if ((z->z_flags & ZONEFS_ZONE_READONLY) || + (sbi->s_mount_opts & ZONEFS_MNTOPT_ERRORS_ZRO)) { + zonefs_warn(sb, "inode %lu: write access disabled\n", + inode->i_ino); + if (!(z->z_flags & ZONEFS_ZONE_READONLY)) + z->z_flags |= ZONEFS_ZONE_READONLY; + zonefs_inode_update_mode(inode); + data_size = isize; + } else if (sbi->s_mount_opts & ZONEFS_MNTOPT_ERRORS_RO && + data_size > isize) { + /* Do not expose garbage data */ + data_size = isize; + } + + /* + * If the filesystem is mounted with the explicit-open mount option, we + * need to clear the ZONEFS_ZONE_OPEN flag if the zone transitioned to + * the read-only or offline condition, to avoid attempting an explicit + * close of the zone when the inode file is closed. + */ + if ((sbi->s_mount_opts & ZONEFS_MNTOPT_EXPLICIT_OPEN) && + (z->z_flags & (ZONEFS_ZONE_READONLY | ZONEFS_ZONE_OFFLINE))) + z->z_flags &= ~ZONEFS_ZONE_OPEN; + + /* + * If error=remount-ro was specified, any error result in remounting + * the volume as read-only. + */ + if ((sbi->s_mount_opts & ZONEFS_MNTOPT_ERRORS_RO) && !sb_rdonly(sb)) { + zonefs_warn(sb, "remounting filesystem read-only\n"); + sb->s_flags |= SB_RDONLY; + } + + /* + * Update block usage stats and the inode size to prevent access to + * invalid data. + */ + zonefs_update_stats(inode, data_size); + zonefs_i_size_write(inode, data_size); + z->z_wpoffset = data_size; + zonefs_inode_account_active(inode); + + return 0; +} + +/* + * When an file IO error occurs, check the file zone to see if there is a change + * in the zone condition (e.g. offline or read-only). For a failed write to a + * sequential zone, the zone write pointer position must also be checked to + * eventually correct the file size and zonefs inode write pointer offset + * (which can be out of sync with the drive due to partial write failures). + */ +void __zonefs_io_error(struct inode *inode, bool write) +{ + struct zonefs_zone *z = zonefs_inode_zone(inode); + struct super_block *sb = inode->i_sb; + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + unsigned int noio_flag; + unsigned int nr_zones = 1; + struct zonefs_ioerr_data err = { + .inode = inode, + .write = write, + }; + int ret; + + /* + * The only files that have more than one zone are conventional zone + * files with aggregated conventional zones, for which the inode zone + * size is always larger than the device zone size. + */ + if (z->z_size > bdev_zone_sectors(sb->s_bdev)) + nr_zones = z->z_size >> + (sbi->s_zone_sectors_shift + SECTOR_SHIFT); + + /* + * Memory allocations in blkdev_report_zones() can trigger a memory + * reclaim which may in turn cause a recursion into zonefs as well as + * struct request allocations for the same device. The former case may + * end up in a deadlock on the inode truncate mutex, while the latter + * may prevent IO forward progress. Executing the report zones under + * the GFP_NOIO context avoids both problems. + */ + noio_flag = memalloc_noio_save(); + ret = blkdev_report_zones(sb->s_bdev, z->z_sector, nr_zones, + zonefs_io_error_cb, &err); + if (ret != nr_zones) + zonefs_err(sb, "Get inode %lu zone information failed %d\n", + inode->i_ino, ret); + memalloc_noio_restore(noio_flag); +} + +static struct kmem_cache *zonefs_inode_cachep; + +static struct inode *zonefs_alloc_inode(struct super_block *sb) +{ + struct zonefs_inode_info *zi; + + zi = alloc_inode_sb(sb, zonefs_inode_cachep, GFP_KERNEL); + if (!zi) + return NULL; + + inode_init_once(&zi->i_vnode); + mutex_init(&zi->i_truncate_mutex); + zi->i_wr_refcnt = 0; + + return &zi->i_vnode; +} + +static void zonefs_free_inode(struct inode *inode) +{ + kmem_cache_free(zonefs_inode_cachep, ZONEFS_I(inode)); +} + +/* + * File system stat. + */ +static int zonefs_statfs(struct dentry *dentry, struct kstatfs *buf) +{ + struct super_block *sb = dentry->d_sb; + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + enum zonefs_ztype t; + + buf->f_type = ZONEFS_MAGIC; + buf->f_bsize = sb->s_blocksize; + buf->f_namelen = ZONEFS_NAME_MAX; + + spin_lock(&sbi->s_lock); + + buf->f_blocks = sbi->s_blocks; + if (WARN_ON(sbi->s_used_blocks > sbi->s_blocks)) + buf->f_bfree = 0; + else + buf->f_bfree = buf->f_blocks - sbi->s_used_blocks; + buf->f_bavail = buf->f_bfree; + + for (t = 0; t < ZONEFS_ZTYPE_MAX; t++) { + if (sbi->s_zgroup[t].g_nr_zones) + buf->f_files += sbi->s_zgroup[t].g_nr_zones + 1; + } + buf->f_ffree = 0; + + spin_unlock(&sbi->s_lock); + + buf->f_fsid = uuid_to_fsid(sbi->s_uuid.b); + + return 0; +} + +enum { + Opt_errors_ro, Opt_errors_zro, Opt_errors_zol, Opt_errors_repair, + Opt_explicit_open, Opt_err, +}; + +static const match_table_t tokens = { + { Opt_errors_ro, "errors=remount-ro"}, + { Opt_errors_zro, "errors=zone-ro"}, + { Opt_errors_zol, "errors=zone-offline"}, + { Opt_errors_repair, "errors=repair"}, + { Opt_explicit_open, "explicit-open" }, + { Opt_err, NULL} +}; + +static int zonefs_parse_options(struct super_block *sb, char *options) +{ + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + substring_t args[MAX_OPT_ARGS]; + char *p; + + if (!options) + return 0; + + while ((p = strsep(&options, ",")) != NULL) { + int token; + + if (!*p) + continue; + + token = match_token(p, tokens, args); + switch (token) { + case Opt_errors_ro: + sbi->s_mount_opts &= ~ZONEFS_MNTOPT_ERRORS_MASK; + sbi->s_mount_opts |= ZONEFS_MNTOPT_ERRORS_RO; + break; + case Opt_errors_zro: + sbi->s_mount_opts &= ~ZONEFS_MNTOPT_ERRORS_MASK; + sbi->s_mount_opts |= ZONEFS_MNTOPT_ERRORS_ZRO; + break; + case Opt_errors_zol: + sbi->s_mount_opts &= ~ZONEFS_MNTOPT_ERRORS_MASK; + sbi->s_mount_opts |= ZONEFS_MNTOPT_ERRORS_ZOL; + break; + case Opt_errors_repair: + sbi->s_mount_opts &= ~ZONEFS_MNTOPT_ERRORS_MASK; + sbi->s_mount_opts |= ZONEFS_MNTOPT_ERRORS_REPAIR; + break; + case Opt_explicit_open: + sbi->s_mount_opts |= ZONEFS_MNTOPT_EXPLICIT_OPEN; + break; + default: + return -EINVAL; + } + } + + return 0; +} + +static int zonefs_show_options(struct seq_file *seq, struct dentry *root) +{ + struct zonefs_sb_info *sbi = ZONEFS_SB(root->d_sb); + + if (sbi->s_mount_opts & ZONEFS_MNTOPT_ERRORS_RO) + seq_puts(seq, ",errors=remount-ro"); + if (sbi->s_mount_opts & ZONEFS_MNTOPT_ERRORS_ZRO) + seq_puts(seq, ",errors=zone-ro"); + if (sbi->s_mount_opts & ZONEFS_MNTOPT_ERRORS_ZOL) + seq_puts(seq, ",errors=zone-offline"); + if (sbi->s_mount_opts & ZONEFS_MNTOPT_ERRORS_REPAIR) + seq_puts(seq, ",errors=repair"); + + return 0; +} + +static int zonefs_remount(struct super_block *sb, int *flags, char *data) +{ + sync_filesystem(sb); + + return zonefs_parse_options(sb, data); +} + +static int zonefs_inode_setattr(struct mnt_idmap *idmap, + struct dentry *dentry, struct iattr *iattr) +{ + struct inode *inode = d_inode(dentry); + int ret; + + if (unlikely(IS_IMMUTABLE(inode))) + return -EPERM; + + ret = setattr_prepare(&nop_mnt_idmap, dentry, iattr); + if (ret) + return ret; + + /* + * Since files and directories cannot be created nor deleted, do not + * allow setting any write attributes on the sub-directories grouping + * files by zone type. + */ + if ((iattr->ia_valid & ATTR_MODE) && S_ISDIR(inode->i_mode) && + (iattr->ia_mode & 0222)) + return -EPERM; + + if (((iattr->ia_valid & ATTR_UID) && + !uid_eq(iattr->ia_uid, inode->i_uid)) || + ((iattr->ia_valid & ATTR_GID) && + !gid_eq(iattr->ia_gid, inode->i_gid))) { + ret = dquot_transfer(&nop_mnt_idmap, inode, iattr); + if (ret) + return ret; + } + + if (iattr->ia_valid & ATTR_SIZE) { + ret = zonefs_file_truncate(inode, iattr->ia_size); + if (ret) + return ret; + } + + setattr_copy(&nop_mnt_idmap, inode, iattr); + + if (S_ISREG(inode->i_mode)) { + struct zonefs_zone *z = zonefs_inode_zone(inode); + + z->z_mode = inode->i_mode; + z->z_uid = inode->i_uid; + z->z_gid = inode->i_gid; + } + + return 0; +} + +static const struct inode_operations zonefs_file_inode_operations = { + .setattr = zonefs_inode_setattr, +}; + +static long zonefs_fname_to_fno(const struct qstr *fname) +{ + const char *name = fname->name; + unsigned int len = fname->len; + long fno = 0, shift = 1; + const char *rname; + char c = *name; + unsigned int i; + + /* + * File names are always a base-10 number string without any + * leading 0s. + */ + if (!isdigit(c)) + return -ENOENT; + + if (len > 1 && c == '0') + return -ENOENT; + + if (len == 1) + return c - '0'; + + for (i = 0, rname = name + len - 1; i < len; i++, rname--) { + c = *rname; + if (!isdigit(c)) + return -ENOENT; + fno += (c - '0') * shift; + shift *= 10; + } + + return fno; +} + +static struct inode *zonefs_get_file_inode(struct inode *dir, + struct dentry *dentry) +{ + struct zonefs_zone_group *zgroup = dir->i_private; + struct super_block *sb = dir->i_sb; + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + struct zonefs_zone *z; + struct inode *inode; + ino_t ino; + long fno; + + /* Get the file number from the file name */ + fno = zonefs_fname_to_fno(&dentry->d_name); + if (fno < 0) + return ERR_PTR(fno); + + if (!zgroup->g_nr_zones || fno >= zgroup->g_nr_zones) + return ERR_PTR(-ENOENT); + + z = &zgroup->g_zones[fno]; + ino = z->z_sector >> sbi->s_zone_sectors_shift; + inode = iget_locked(sb, ino); + if (!inode) + return ERR_PTR(-ENOMEM); + if (!(inode->i_state & I_NEW)) { + WARN_ON_ONCE(inode->i_private != z); + return inode; + } + + inode->i_ino = ino; + inode->i_mode = z->z_mode; + inode->i_mtime = inode->i_atime = inode_set_ctime_to_ts(inode, + inode_get_ctime(dir)); + inode->i_uid = z->z_uid; + inode->i_gid = z->z_gid; + inode->i_size = z->z_wpoffset; + inode->i_blocks = z->z_capacity >> SECTOR_SHIFT; + inode->i_private = z; + + inode->i_op = &zonefs_file_inode_operations; + inode->i_fop = &zonefs_file_operations; + inode->i_mapping->a_ops = &zonefs_file_aops; + + /* Update the inode access rights depending on the zone condition */ + zonefs_inode_update_mode(inode); + + unlock_new_inode(inode); + + return inode; +} + +static struct inode *zonefs_get_zgroup_inode(struct super_block *sb, + enum zonefs_ztype ztype) +{ + struct inode *root = d_inode(sb->s_root); + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + struct inode *inode; + ino_t ino = bdev_nr_zones(sb->s_bdev) + ztype + 1; + + inode = iget_locked(sb, ino); + if (!inode) + return ERR_PTR(-ENOMEM); + if (!(inode->i_state & I_NEW)) + return inode; + + inode->i_ino = ino; + inode_init_owner(&nop_mnt_idmap, inode, root, S_IFDIR | 0555); + inode->i_size = sbi->s_zgroup[ztype].g_nr_zones; + inode->i_mtime = inode->i_atime = inode_set_ctime_to_ts(inode, + inode_get_ctime(root)); + inode->i_private = &sbi->s_zgroup[ztype]; + set_nlink(inode, 2); + + inode->i_op = &zonefs_dir_inode_operations; + inode->i_fop = &zonefs_dir_operations; + + unlock_new_inode(inode); + + return inode; +} + + +static struct inode *zonefs_get_dir_inode(struct inode *dir, + struct dentry *dentry) +{ + struct super_block *sb = dir->i_sb; + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + const char *name = dentry->d_name.name; + enum zonefs_ztype ztype; + + /* + * We only need to check for the "seq" directory and + * the "cnv" directory if we have conventional zones. + */ + if (dentry->d_name.len != 3) + return ERR_PTR(-ENOENT); + + for (ztype = 0; ztype < ZONEFS_ZTYPE_MAX; ztype++) { + if (sbi->s_zgroup[ztype].g_nr_zones && + memcmp(name, zonefs_zgroup_name(ztype), 3) == 0) + break; + } + if (ztype == ZONEFS_ZTYPE_MAX) + return ERR_PTR(-ENOENT); + + return zonefs_get_zgroup_inode(sb, ztype); +} + +static struct dentry *zonefs_lookup(struct inode *dir, struct dentry *dentry, + unsigned int flags) +{ + struct inode *inode; + + if (dentry->d_name.len > ZONEFS_NAME_MAX) + return ERR_PTR(-ENAMETOOLONG); + + if (dir == d_inode(dir->i_sb->s_root)) + inode = zonefs_get_dir_inode(dir, dentry); + else + inode = zonefs_get_file_inode(dir, dentry); + if (IS_ERR(inode)) + return ERR_CAST(inode); + + return d_splice_alias(inode, dentry); +} + +static int zonefs_readdir_root(struct file *file, struct dir_context *ctx) +{ + struct inode *inode = file_inode(file); + struct super_block *sb = inode->i_sb; + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + enum zonefs_ztype ztype = ZONEFS_ZTYPE_CNV; + ino_t base_ino = bdev_nr_zones(sb->s_bdev) + 1; + + if (ctx->pos >= inode->i_size) + return 0; + + if (!dir_emit_dots(file, ctx)) + return 0; + + if (ctx->pos == 2) { + if (!sbi->s_zgroup[ZONEFS_ZTYPE_CNV].g_nr_zones) + ztype = ZONEFS_ZTYPE_SEQ; + + if (!dir_emit(ctx, zonefs_zgroup_name(ztype), 3, + base_ino + ztype, DT_DIR)) + return 0; + ctx->pos++; + } + + if (ctx->pos == 3 && ztype != ZONEFS_ZTYPE_SEQ) { + ztype = ZONEFS_ZTYPE_SEQ; + if (!dir_emit(ctx, zonefs_zgroup_name(ztype), 3, + base_ino + ztype, DT_DIR)) + return 0; + ctx->pos++; + } + + return 0; +} + +static int zonefs_readdir_zgroup(struct file *file, + struct dir_context *ctx) +{ + struct inode *inode = file_inode(file); + struct zonefs_zone_group *zgroup = inode->i_private; + struct super_block *sb = inode->i_sb; + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + struct zonefs_zone *z; + int fname_len; + char *fname; + ino_t ino; + int f; + + /* + * The size of zone group directories is equal to the number + * of zone files in the group and does note include the "." and + * ".." entries. Hence the "+ 2" here. + */ + if (ctx->pos >= inode->i_size + 2) + return 0; + + if (!dir_emit_dots(file, ctx)) + return 0; + + fname = kmalloc(ZONEFS_NAME_MAX, GFP_KERNEL); + if (!fname) + return -ENOMEM; + + for (f = ctx->pos - 2; f < zgroup->g_nr_zones; f++) { + z = &zgroup->g_zones[f]; + ino = z->z_sector >> sbi->s_zone_sectors_shift; + fname_len = snprintf(fname, ZONEFS_NAME_MAX - 1, "%u", f); + if (!dir_emit(ctx, fname, fname_len, ino, DT_REG)) + break; + ctx->pos++; + } + + kfree(fname); + + return 0; +} + +static int zonefs_readdir(struct file *file, struct dir_context *ctx) +{ + struct inode *inode = file_inode(file); + + if (inode == d_inode(inode->i_sb->s_root)) + return zonefs_readdir_root(file, ctx); + + return zonefs_readdir_zgroup(file, ctx); +} + +const struct inode_operations zonefs_dir_inode_operations = { + .lookup = zonefs_lookup, + .setattr = zonefs_inode_setattr, +}; + +const struct file_operations zonefs_dir_operations = { + .llseek = generic_file_llseek, + .read = generic_read_dir, + .iterate_shared = zonefs_readdir, +}; + +struct zonefs_zone_data { + struct super_block *sb; + unsigned int nr_zones[ZONEFS_ZTYPE_MAX]; + sector_t cnv_zone_start; + struct blk_zone *zones; +}; + +static int zonefs_get_zone_info_cb(struct blk_zone *zone, unsigned int idx, + void *data) +{ + struct zonefs_zone_data *zd = data; + struct super_block *sb = zd->sb; + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + + /* + * We do not care about the first zone: it contains the super block + * and not exposed as a file. + */ + if (!idx) + return 0; + + /* + * Count the number of zones that will be exposed as files. + * For sequential zones, we always have as many files as zones. + * FOr conventional zones, the number of files depends on if we have + * conventional zones aggregation enabled. + */ + switch (zone->type) { + case BLK_ZONE_TYPE_CONVENTIONAL: + if (sbi->s_features & ZONEFS_F_AGGRCNV) { + /* One file per set of contiguous conventional zones */ + if (!(sbi->s_zgroup[ZONEFS_ZTYPE_CNV].g_nr_zones) || + zone->start != zd->cnv_zone_start) + sbi->s_zgroup[ZONEFS_ZTYPE_CNV].g_nr_zones++; + zd->cnv_zone_start = zone->start + zone->len; + } else { + /* One file per zone */ + sbi->s_zgroup[ZONEFS_ZTYPE_CNV].g_nr_zones++; + } + break; + case BLK_ZONE_TYPE_SEQWRITE_REQ: + case BLK_ZONE_TYPE_SEQWRITE_PREF: + sbi->s_zgroup[ZONEFS_ZTYPE_SEQ].g_nr_zones++; + break; + default: + zonefs_err(zd->sb, "Unsupported zone type 0x%x\n", + zone->type); + return -EIO; + } + + memcpy(&zd->zones[idx], zone, sizeof(struct blk_zone)); + + return 0; +} + +static int zonefs_get_zone_info(struct zonefs_zone_data *zd) +{ + struct block_device *bdev = zd->sb->s_bdev; + int ret; + + zd->zones = kvcalloc(bdev_nr_zones(bdev), sizeof(struct blk_zone), + GFP_KERNEL); + if (!zd->zones) + return -ENOMEM; + + /* Get zones information from the device */ + ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, + zonefs_get_zone_info_cb, zd); + if (ret < 0) { + zonefs_err(zd->sb, "Zone report failed %d\n", ret); + return ret; + } + + if (ret != bdev_nr_zones(bdev)) { + zonefs_err(zd->sb, "Invalid zone report (%d/%u zones)\n", + ret, bdev_nr_zones(bdev)); + return -EIO; + } + + return 0; +} + +static inline void zonefs_free_zone_info(struct zonefs_zone_data *zd) +{ + kvfree(zd->zones); +} + +/* + * Create a zone group and populate it with zone files. + */ +static int zonefs_init_zgroup(struct super_block *sb, + struct zonefs_zone_data *zd, + enum zonefs_ztype ztype) +{ + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + struct zonefs_zone_group *zgroup = &sbi->s_zgroup[ztype]; + struct blk_zone *zone, *next, *end; + struct zonefs_zone *z; + unsigned int n = 0; + int ret; + + /* Allocate the zone group. If it is empty, we have nothing to do. */ + if (!zgroup->g_nr_zones) + return 0; + + zgroup->g_zones = kvcalloc(zgroup->g_nr_zones, + sizeof(struct zonefs_zone), GFP_KERNEL); + if (!zgroup->g_zones) + return -ENOMEM; + + /* + * Initialize the zone groups using the device zone information. + * We always skip the first zone as it contains the super block + * and is not use to back a file. + */ + end = zd->zones + bdev_nr_zones(sb->s_bdev); + for (zone = &zd->zones[1]; zone < end; zone = next) { + + next = zone + 1; + if (zonefs_zone_type(zone) != ztype) + continue; + + if (WARN_ON_ONCE(n >= zgroup->g_nr_zones)) + return -EINVAL; + + /* + * For conventional zones, contiguous zones can be aggregated + * together to form larger files. Note that this overwrites the + * length of the first zone of the set of contiguous zones + * aggregated together. If one offline or read-only zone is + * found, assume that all zones aggregated have the same + * condition. + */ + if (ztype == ZONEFS_ZTYPE_CNV && + (sbi->s_features & ZONEFS_F_AGGRCNV)) { + for (; next < end; next++) { + if (zonefs_zone_type(next) != ztype) + break; + zone->len += next->len; + zone->capacity += next->capacity; + if (next->cond == BLK_ZONE_COND_READONLY && + zone->cond != BLK_ZONE_COND_OFFLINE) + zone->cond = BLK_ZONE_COND_READONLY; + else if (next->cond == BLK_ZONE_COND_OFFLINE) + zone->cond = BLK_ZONE_COND_OFFLINE; + } + } + + z = &zgroup->g_zones[n]; + if (ztype == ZONEFS_ZTYPE_CNV) + z->z_flags |= ZONEFS_ZONE_CNV; + z->z_sector = zone->start; + z->z_size = zone->len << SECTOR_SHIFT; + if (z->z_size > bdev_zone_sectors(sb->s_bdev) << SECTOR_SHIFT && + !(sbi->s_features & ZONEFS_F_AGGRCNV)) { + zonefs_err(sb, + "Invalid zone size %llu (device zone sectors %llu)\n", + z->z_size, + bdev_zone_sectors(sb->s_bdev) << SECTOR_SHIFT); + return -EINVAL; + } + + z->z_capacity = min_t(loff_t, MAX_LFS_FILESIZE, + zone->capacity << SECTOR_SHIFT); + z->z_wpoffset = zonefs_check_zone_condition(sb, z, zone); + + z->z_mode = S_IFREG | sbi->s_perm; + z->z_uid = sbi->s_uid; + z->z_gid = sbi->s_gid; + + /* + * Let zonefs_inode_update_mode() know that we will need + * special initialization of the inode mode the first time + * it is accessed. + */ + z->z_flags |= ZONEFS_ZONE_INIT_MODE; + + sb->s_maxbytes = max(z->z_capacity, sb->s_maxbytes); + sbi->s_blocks += z->z_capacity >> sb->s_blocksize_bits; + sbi->s_used_blocks += z->z_wpoffset >> sb->s_blocksize_bits; + + /* + * For sequential zones, make sure that any open zone is closed + * first to ensure that the initial number of open zones is 0, + * in sync with the open zone accounting done when the mount + * option ZONEFS_MNTOPT_EXPLICIT_OPEN is used. + */ + if (ztype == ZONEFS_ZTYPE_SEQ && + (zone->cond == BLK_ZONE_COND_IMP_OPEN || + zone->cond == BLK_ZONE_COND_EXP_OPEN)) { + ret = zonefs_zone_mgmt(sb, z, REQ_OP_ZONE_CLOSE); + if (ret) + return ret; + } + + zonefs_account_active(sb, z); + + n++; + } + + if (WARN_ON_ONCE(n != zgroup->g_nr_zones)) + return -EINVAL; + + zonefs_info(sb, "Zone group \"%s\" has %u file%s\n", + zonefs_zgroup_name(ztype), + zgroup->g_nr_zones, + zgroup->g_nr_zones > 1 ? "s" : ""); + + return 0; +} + +static void zonefs_free_zgroups(struct super_block *sb) +{ + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + enum zonefs_ztype ztype; + + if (!sbi) + return; + + for (ztype = 0; ztype < ZONEFS_ZTYPE_MAX; ztype++) { + kvfree(sbi->s_zgroup[ztype].g_zones); + sbi->s_zgroup[ztype].g_zones = NULL; + } +} + +/* + * Create a zone group and populate it with zone files. + */ +static int zonefs_init_zgroups(struct super_block *sb) +{ + struct zonefs_zone_data zd; + enum zonefs_ztype ztype; + int ret; + + /* First get the device zone information */ + memset(&zd, 0, sizeof(struct zonefs_zone_data)); + zd.sb = sb; + ret = zonefs_get_zone_info(&zd); + if (ret) + goto cleanup; + + /* Allocate and initialize the zone groups */ + for (ztype = 0; ztype < ZONEFS_ZTYPE_MAX; ztype++) { + ret = zonefs_init_zgroup(sb, &zd, ztype); + if (ret) { + zonefs_info(sb, + "Zone group \"%s\" initialization failed\n", + zonefs_zgroup_name(ztype)); + break; + } + } + +cleanup: + zonefs_free_zone_info(&zd); + if (ret) + zonefs_free_zgroups(sb); + + return ret; +} + +/* + * Read super block information from the device. + */ +static int zonefs_read_super(struct super_block *sb) +{ + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + struct zonefs_super *super; + u32 crc, stored_crc; + struct page *page; + struct bio_vec bio_vec; + struct bio bio; + int ret; + + page = alloc_page(GFP_KERNEL); + if (!page) + return -ENOMEM; + + bio_init(&bio, sb->s_bdev, &bio_vec, 1, REQ_OP_READ); + bio.bi_iter.bi_sector = 0; + __bio_add_page(&bio, page, PAGE_SIZE, 0); + + ret = submit_bio_wait(&bio); + if (ret) + goto free_page; + + super = page_address(page); + + ret = -EINVAL; + if (le32_to_cpu(super->s_magic) != ZONEFS_MAGIC) + goto free_page; + + stored_crc = le32_to_cpu(super->s_crc); + super->s_crc = 0; + crc = crc32(~0U, (unsigned char *)super, sizeof(struct zonefs_super)); + if (crc != stored_crc) { + zonefs_err(sb, "Invalid checksum (Expected 0x%08x, got 0x%08x)", + crc, stored_crc); + goto free_page; + } + + sbi->s_features = le64_to_cpu(super->s_features); + if (sbi->s_features & ~ZONEFS_F_DEFINED_FEATURES) { + zonefs_err(sb, "Unknown features set 0x%llx\n", + sbi->s_features); + goto free_page; + } + + if (sbi->s_features & ZONEFS_F_UID) { + sbi->s_uid = make_kuid(current_user_ns(), + le32_to_cpu(super->s_uid)); + if (!uid_valid(sbi->s_uid)) { + zonefs_err(sb, "Invalid UID feature\n"); + goto free_page; + } + } + + if (sbi->s_features & ZONEFS_F_GID) { + sbi->s_gid = make_kgid(current_user_ns(), + le32_to_cpu(super->s_gid)); + if (!gid_valid(sbi->s_gid)) { + zonefs_err(sb, "Invalid GID feature\n"); + goto free_page; + } + } + + if (sbi->s_features & ZONEFS_F_PERM) + sbi->s_perm = le32_to_cpu(super->s_perm); + + if (memchr_inv(super->s_reserved, 0, sizeof(super->s_reserved))) { + zonefs_err(sb, "Reserved area is being used\n"); + goto free_page; + } + + import_uuid(&sbi->s_uuid, super->s_uuid); + ret = 0; + +free_page: + __free_page(page); + + return ret; +} + +static const struct super_operations zonefs_sops = { + .alloc_inode = zonefs_alloc_inode, + .free_inode = zonefs_free_inode, + .statfs = zonefs_statfs, + .remount_fs = zonefs_remount, + .show_options = zonefs_show_options, +}; + +static int zonefs_get_zgroup_inodes(struct super_block *sb) +{ + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + struct inode *dir_inode; + enum zonefs_ztype ztype; + + for (ztype = 0; ztype < ZONEFS_ZTYPE_MAX; ztype++) { + if (!sbi->s_zgroup[ztype].g_nr_zones) + continue; + + dir_inode = zonefs_get_zgroup_inode(sb, ztype); + if (IS_ERR(dir_inode)) + return PTR_ERR(dir_inode); + + sbi->s_zgroup[ztype].g_inode = dir_inode; + } + + return 0; +} + +static void zonefs_release_zgroup_inodes(struct super_block *sb) +{ + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + enum zonefs_ztype ztype; + + if (!sbi) + return; + + for (ztype = 0; ztype < ZONEFS_ZTYPE_MAX; ztype++) { + if (sbi->s_zgroup[ztype].g_inode) { + iput(sbi->s_zgroup[ztype].g_inode); + sbi->s_zgroup[ztype].g_inode = NULL; + } + } +} + +/* + * Check that the device is zoned. If it is, get the list of zones and create + * sub-directories and files according to the device zone configuration and + * format options. + */ +static int zonefs_fill_super(struct super_block *sb, void *data, int silent) +{ + struct zonefs_sb_info *sbi; + struct inode *inode; + enum zonefs_ztype ztype; + int ret; + + if (!bdev_is_zoned(sb->s_bdev)) { + zonefs_err(sb, "Not a zoned block device\n"); + return -EINVAL; + } + + /* + * Initialize super block information: the maximum file size is updated + * when the zone files are created so that the format option + * ZONEFS_F_AGGRCNV which increases the maximum file size of a file + * beyond the zone size is taken into account. + */ + sbi = kzalloc(sizeof(*sbi), GFP_KERNEL); + if (!sbi) + return -ENOMEM; + + spin_lock_init(&sbi->s_lock); + sb->s_fs_info = sbi; + sb->s_magic = ZONEFS_MAGIC; + sb->s_maxbytes = 0; + sb->s_op = &zonefs_sops; + sb->s_time_gran = 1; + + /* + * The block size is set to the device zone write granularity to ensure + * that write operations are always aligned according to the device + * interface constraints. + */ + sb_set_blocksize(sb, bdev_zone_write_granularity(sb->s_bdev)); + sbi->s_zone_sectors_shift = ilog2(bdev_zone_sectors(sb->s_bdev)); + sbi->s_uid = GLOBAL_ROOT_UID; + sbi->s_gid = GLOBAL_ROOT_GID; + sbi->s_perm = 0640; + sbi->s_mount_opts = ZONEFS_MNTOPT_ERRORS_RO; + + atomic_set(&sbi->s_wro_seq_files, 0); + sbi->s_max_wro_seq_files = bdev_max_open_zones(sb->s_bdev); + atomic_set(&sbi->s_active_seq_files, 0); + sbi->s_max_active_seq_files = bdev_max_active_zones(sb->s_bdev); + + ret = zonefs_read_super(sb); + if (ret) + return ret; + + ret = zonefs_parse_options(sb, data); + if (ret) + return ret; + + zonefs_info(sb, "Mounting %u zones", bdev_nr_zones(sb->s_bdev)); + + if (!sbi->s_max_wro_seq_files && + !sbi->s_max_active_seq_files && + sbi->s_mount_opts & ZONEFS_MNTOPT_EXPLICIT_OPEN) { + zonefs_info(sb, + "No open and active zone limits. Ignoring explicit_open mount option\n"); + sbi->s_mount_opts &= ~ZONEFS_MNTOPT_EXPLICIT_OPEN; + } + + /* Initialize the zone groups */ + ret = zonefs_init_zgroups(sb); + if (ret) + goto cleanup; + + /* Create the root directory inode */ + ret = -ENOMEM; + inode = new_inode(sb); + if (!inode) + goto cleanup; + + inode->i_ino = bdev_nr_zones(sb->s_bdev); + inode->i_mode = S_IFDIR | 0555; + inode->i_mtime = inode->i_atime = inode_set_ctime_current(inode); + inode->i_op = &zonefs_dir_inode_operations; + inode->i_fop = &zonefs_dir_operations; + inode->i_size = 2; + set_nlink(inode, 2); + for (ztype = 0; ztype < ZONEFS_ZTYPE_MAX; ztype++) { + if (sbi->s_zgroup[ztype].g_nr_zones) { + inc_nlink(inode); + inode->i_size++; + } + } + + sb->s_root = d_make_root(inode); + if (!sb->s_root) + goto cleanup; + + /* + * Take a reference on the zone groups directory inodes + * to keep them in the inode cache. + */ + ret = zonefs_get_zgroup_inodes(sb); + if (ret) + goto cleanup; + + ret = zonefs_sysfs_register(sb); + if (ret) + goto cleanup; + + return 0; + +cleanup: + zonefs_release_zgroup_inodes(sb); + zonefs_free_zgroups(sb); + + return ret; +} + +static struct dentry *zonefs_mount(struct file_system_type *fs_type, + int flags, const char *dev_name, void *data) +{ + return mount_bdev(fs_type, flags, dev_name, data, zonefs_fill_super); +} + +static void zonefs_kill_super(struct super_block *sb) +{ + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + + /* Release the reference on the zone group directory inodes */ + zonefs_release_zgroup_inodes(sb); + + kill_block_super(sb); + + zonefs_sysfs_unregister(sb); + zonefs_free_zgroups(sb); + kfree(sbi); +} + +/* + * File system definition and registration. + */ +static struct file_system_type zonefs_type = { + .owner = THIS_MODULE, + .name = "zonefs", + .mount = zonefs_mount, + .kill_sb = zonefs_kill_super, + .fs_flags = FS_REQUIRES_DEV, +}; + +static int __init zonefs_init_inodecache(void) +{ + zonefs_inode_cachep = kmem_cache_create("zonefs_inode_cache", + sizeof(struct zonefs_inode_info), 0, + (SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD | SLAB_ACCOUNT), + NULL); + if (zonefs_inode_cachep == NULL) + return -ENOMEM; + return 0; +} + +static void zonefs_destroy_inodecache(void) +{ + /* + * Make sure all delayed rcu free inodes are flushed before we + * destroy the inode cache. + */ + rcu_barrier(); + kmem_cache_destroy(zonefs_inode_cachep); +} + +static int __init zonefs_init(void) +{ + int ret; + + BUILD_BUG_ON(sizeof(struct zonefs_super) != ZONEFS_SUPER_SIZE); + + ret = zonefs_init_inodecache(); + if (ret) + return ret; + + ret = zonefs_sysfs_init(); + if (ret) + goto destroy_inodecache; + + ret = register_filesystem(&zonefs_type); + if (ret) + goto sysfs_exit; + + return 0; + +sysfs_exit: + zonefs_sysfs_exit(); +destroy_inodecache: + zonefs_destroy_inodecache(); + + return ret; +} + +static void __exit zonefs_exit(void) +{ + unregister_filesystem(&zonefs_type); + zonefs_sysfs_exit(); + zonefs_destroy_inodecache(); +} + +MODULE_AUTHOR("Damien Le Moal"); +MODULE_DESCRIPTION("Zone file system for zoned block devices"); +MODULE_LICENSE("GPL"); +MODULE_ALIAS_FS("zonefs"); +module_init(zonefs_init); +module_exit(zonefs_exit); diff --git a/fs/zonefs/sysfs.c b/fs/zonefs/sysfs.c new file mode 100644 index 0000000000..8ccb65c2b4 --- /dev/null +++ b/fs/zonefs/sysfs.c @@ -0,0 +1,134 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Simple file system for zoned block devices exposing zones as files. + * + * Copyright (C) 2022 Western Digital Corporation or its affiliates. + */ +#include <linux/fs.h> +#include <linux/seq_file.h> +#include <linux/blkdev.h> + +#include "zonefs.h" + +struct zonefs_sysfs_attr { + struct attribute attr; + ssize_t (*show)(struct zonefs_sb_info *sbi, char *buf); +}; + +#define ZONEFS_SYSFS_ATTR_RO(name) \ +static struct zonefs_sysfs_attr zonefs_sysfs_attr_##name = __ATTR_RO(name) + +#define ATTR_LIST(name) &zonefs_sysfs_attr_##name.attr + +static ssize_t zonefs_sysfs_attr_show(struct kobject *kobj, + struct attribute *attr, char *buf) +{ + struct zonefs_sb_info *sbi = + container_of(kobj, struct zonefs_sb_info, s_kobj); + struct zonefs_sysfs_attr *zonefs_attr = + container_of(attr, struct zonefs_sysfs_attr, attr); + + if (!zonefs_attr->show) + return 0; + + return zonefs_attr->show(sbi, buf); +} + +static ssize_t max_wro_seq_files_show(struct zonefs_sb_info *sbi, char *buf) +{ + return sysfs_emit(buf, "%u\n", sbi->s_max_wro_seq_files); +} +ZONEFS_SYSFS_ATTR_RO(max_wro_seq_files); + +static ssize_t nr_wro_seq_files_show(struct zonefs_sb_info *sbi, char *buf) +{ + return sysfs_emit(buf, "%d\n", atomic_read(&sbi->s_wro_seq_files)); +} +ZONEFS_SYSFS_ATTR_RO(nr_wro_seq_files); + +static ssize_t max_active_seq_files_show(struct zonefs_sb_info *sbi, char *buf) +{ + return sysfs_emit(buf, "%u\n", sbi->s_max_active_seq_files); +} +ZONEFS_SYSFS_ATTR_RO(max_active_seq_files); + +static ssize_t nr_active_seq_files_show(struct zonefs_sb_info *sbi, char *buf) +{ + return sysfs_emit(buf, "%d\n", atomic_read(&sbi->s_active_seq_files)); +} +ZONEFS_SYSFS_ATTR_RO(nr_active_seq_files); + +static struct attribute *zonefs_sysfs_attrs[] = { + ATTR_LIST(max_wro_seq_files), + ATTR_LIST(nr_wro_seq_files), + ATTR_LIST(max_active_seq_files), + ATTR_LIST(nr_active_seq_files), + NULL, +}; +ATTRIBUTE_GROUPS(zonefs_sysfs); + +static void zonefs_sysfs_sb_release(struct kobject *kobj) +{ + struct zonefs_sb_info *sbi = + container_of(kobj, struct zonefs_sb_info, s_kobj); + + complete(&sbi->s_kobj_unregister); +} + +static const struct sysfs_ops zonefs_sysfs_attr_ops = { + .show = zonefs_sysfs_attr_show, +}; + +static const struct kobj_type zonefs_sb_ktype = { + .default_groups = zonefs_sysfs_groups, + .sysfs_ops = &zonefs_sysfs_attr_ops, + .release = zonefs_sysfs_sb_release, +}; + +static struct kobject *zonefs_sysfs_root; + +int zonefs_sysfs_register(struct super_block *sb) +{ + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + int ret; + + init_completion(&sbi->s_kobj_unregister); + ret = kobject_init_and_add(&sbi->s_kobj, &zonefs_sb_ktype, + zonefs_sysfs_root, "%s", sb->s_id); + if (ret) { + kobject_put(&sbi->s_kobj); + wait_for_completion(&sbi->s_kobj_unregister); + return ret; + } + + sbi->s_sysfs_registered = true; + + return 0; +} + +void zonefs_sysfs_unregister(struct super_block *sb) +{ + struct zonefs_sb_info *sbi = ZONEFS_SB(sb); + + if (!sbi || !sbi->s_sysfs_registered) + return; + + kobject_del(&sbi->s_kobj); + kobject_put(&sbi->s_kobj); + wait_for_completion(&sbi->s_kobj_unregister); +} + +int __init zonefs_sysfs_init(void) +{ + zonefs_sysfs_root = kobject_create_and_add("zonefs", fs_kobj); + if (!zonefs_sysfs_root) + return -ENOMEM; + + return 0; +} + +void zonefs_sysfs_exit(void) +{ + kobject_put(zonefs_sysfs_root); + zonefs_sysfs_root = NULL; +} diff --git a/fs/zonefs/trace.h b/fs/zonefs/trace.h new file mode 100644 index 0000000000..9969db3a9c --- /dev/null +++ b/fs/zonefs/trace.h @@ -0,0 +1,106 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* + * zonefs filesystem driver tracepoints. + * + * Copyright (C) 2021 Western Digital Corporation or its affiliates. + */ + +#undef TRACE_SYSTEM +#define TRACE_SYSTEM zonefs + +#if !defined(_TRACE_ZONEFS_H) || defined(TRACE_HEADER_MULTI_READ) +#define _TRACE_ZONEFS_H + +#include <linux/tracepoint.h> +#include <linux/trace_seq.h> +#include <linux/blkdev.h> + +#include "zonefs.h" + +#define show_dev(dev) MAJOR(dev), MINOR(dev) + +TRACE_EVENT(zonefs_zone_mgmt, + TP_PROTO(struct super_block *sb, struct zonefs_zone *z, + enum req_op op), + TP_ARGS(sb, z, op), + TP_STRUCT__entry( + __field(dev_t, dev) + __field(ino_t, ino) + __field(enum req_op, op) + __field(sector_t, sector) + __field(sector_t, nr_sectors) + ), + TP_fast_assign( + __entry->dev = sb->s_dev; + __entry->ino = + z->z_sector >> ZONEFS_SB(sb)->s_zone_sectors_shift; + __entry->op = op; + __entry->sector = z->z_sector; + __entry->nr_sectors = z->z_size >> SECTOR_SHIFT; + ), + TP_printk("bdev=(%d,%d), ino=%lu op=%s, sector=%llu, nr_sectors=%llu", + show_dev(__entry->dev), (unsigned long)__entry->ino, + blk_op_str(__entry->op), __entry->sector, + __entry->nr_sectors + ) +); + +TRACE_EVENT(zonefs_file_dio_append, + TP_PROTO(struct inode *inode, ssize_t size, ssize_t ret), + TP_ARGS(inode, size, ret), + TP_STRUCT__entry( + __field(dev_t, dev) + __field(ino_t, ino) + __field(sector_t, sector) + __field(ssize_t, size) + __field(loff_t, wpoffset) + __field(ssize_t, ret) + ), + TP_fast_assign( + __entry->dev = inode->i_sb->s_dev; + __entry->ino = inode->i_ino; + __entry->sector = zonefs_inode_zone(inode)->z_sector; + __entry->size = size; + __entry->wpoffset = + zonefs_inode_zone(inode)->z_wpoffset; + __entry->ret = ret; + ), + TP_printk("bdev=(%d, %d), ino=%lu, sector=%llu, size=%zu, wpoffset=%llu, ret=%zu", + show_dev(__entry->dev), (unsigned long)__entry->ino, + __entry->sector, __entry->size, __entry->wpoffset, + __entry->ret + ) +); + +TRACE_EVENT(zonefs_iomap_begin, + TP_PROTO(struct inode *inode, struct iomap *iomap), + TP_ARGS(inode, iomap), + TP_STRUCT__entry( + __field(dev_t, dev) + __field(ino_t, ino) + __field(u64, addr) + __field(loff_t, offset) + __field(u64, length) + ), + TP_fast_assign( + __entry->dev = inode->i_sb->s_dev; + __entry->ino = inode->i_ino; + __entry->addr = iomap->addr; + __entry->offset = iomap->offset; + __entry->length = iomap->length; + ), + TP_printk("bdev=(%d,%d), ino=%lu, addr=%llu, offset=%llu, length=%llu", + show_dev(__entry->dev), (unsigned long)__entry->ino, + __entry->addr, __entry->offset, __entry->length + ) +); + +#endif /* _TRACE_ZONEFS_H */ + +#undef TRACE_INCLUDE_PATH +#define TRACE_INCLUDE_PATH . +#undef TRACE_INCLUDE_FILE +#define TRACE_INCLUDE_FILE trace + +/* This part must be outside protection */ +#include <trace/define_trace.h> diff --git a/fs/zonefs/zonefs.h b/fs/zonefs/zonefs.h new file mode 100644 index 0000000000..8175652241 --- /dev/null +++ b/fs/zonefs/zonefs.h @@ -0,0 +1,289 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* + * Simple zone file system for zoned block devices. + * + * Copyright (C) 2019 Western Digital Corporation or its affiliates. + */ +#ifndef __ZONEFS_H__ +#define __ZONEFS_H__ + +#include <linux/fs.h> +#include <linux/magic.h> +#include <linux/uuid.h> +#include <linux/mutex.h> +#include <linux/rwsem.h> +#include <linux/kobject.h> + +/* + * Maximum length of file names: this only needs to be large enough to fit + * the zone group directory names and a decimal zone number for file names. + * 16 characters is plenty. + */ +#define ZONEFS_NAME_MAX 16 + +/* + * Zone types: ZONEFS_ZTYPE_SEQ is used for all sequential zone types + * defined in linux/blkzoned.h, that is, BLK_ZONE_TYPE_SEQWRITE_REQ and + * BLK_ZONE_TYPE_SEQWRITE_PREF. + */ +enum zonefs_ztype { + ZONEFS_ZTYPE_CNV, + ZONEFS_ZTYPE_SEQ, + ZONEFS_ZTYPE_MAX, +}; + +static inline enum zonefs_ztype zonefs_zone_type(struct blk_zone *zone) +{ + if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL) + return ZONEFS_ZTYPE_CNV; + return ZONEFS_ZTYPE_SEQ; +} + +#define ZONEFS_ZONE_INIT_MODE (1U << 0) +#define ZONEFS_ZONE_OPEN (1U << 1) +#define ZONEFS_ZONE_ACTIVE (1U << 2) +#define ZONEFS_ZONE_OFFLINE (1U << 3) +#define ZONEFS_ZONE_READONLY (1U << 4) +#define ZONEFS_ZONE_CNV (1U << 31) + +/* + * In-memory per-file inode zone data. + */ +struct zonefs_zone { + /* Zone state flags */ + unsigned int z_flags; + + /* Zone start sector (512B unit) */ + sector_t z_sector; + + /* Zone size (bytes) */ + loff_t z_size; + + /* Zone capacity (file maximum size, bytes) */ + loff_t z_capacity; + + /* Write pointer offset in the zone (sequential zones only, bytes) */ + loff_t z_wpoffset; + + /* Saved inode uid, gid and access rights */ + umode_t z_mode; + kuid_t z_uid; + kgid_t z_gid; +}; + +/* + * In memory zone group information: all zones of a group are exposed + * as files, one file per zone. + */ +struct zonefs_zone_group { + struct inode *g_inode; + unsigned int g_nr_zones; + struct zonefs_zone *g_zones; +}; + +/* + * In-memory inode data. + */ +struct zonefs_inode_info { + struct inode i_vnode; + + /* + * To serialise fully against both syscall and mmap based IO and + * sequential file truncation, two locks are used. For serializing + * zonefs_seq_file_truncate() against zonefs_iomap_begin(), that is, + * file truncate operations against block mapping, i_truncate_mutex is + * used. i_truncate_mutex also protects against concurrent accesses + * and changes to the inode private data, and in particular changes to + * a sequential file size on completion of direct IO writes. + * Serialization of mmap read IOs with truncate and syscall IO + * operations is done with invalidate_lock in addition to + * i_truncate_mutex. Only zonefs_seq_file_truncate() takes both lock + * (invalidate_lock first, i_truncate_mutex second). + */ + struct mutex i_truncate_mutex; + + /* guarded by i_truncate_mutex */ + unsigned int i_wr_refcnt; +}; + +static inline struct zonefs_inode_info *ZONEFS_I(struct inode *inode) +{ + return container_of(inode, struct zonefs_inode_info, i_vnode); +} + +static inline bool zonefs_zone_is_cnv(struct zonefs_zone *z) +{ + return z->z_flags & ZONEFS_ZONE_CNV; +} + +static inline bool zonefs_zone_is_seq(struct zonefs_zone *z) +{ + return !zonefs_zone_is_cnv(z); +} + +static inline struct zonefs_zone *zonefs_inode_zone(struct inode *inode) +{ + return inode->i_private; +} + +static inline bool zonefs_inode_is_cnv(struct inode *inode) +{ + return zonefs_zone_is_cnv(zonefs_inode_zone(inode)); +} + +static inline bool zonefs_inode_is_seq(struct inode *inode) +{ + return zonefs_zone_is_seq(zonefs_inode_zone(inode)); +} + +/* + * On-disk super block (block 0). + */ +#define ZONEFS_LABEL_LEN 64 +#define ZONEFS_UUID_SIZE 16 +#define ZONEFS_SUPER_SIZE 4096 + +struct zonefs_super { + + /* Magic number */ + __le32 s_magic; + + /* Checksum */ + __le32 s_crc; + + /* Volume label */ + char s_label[ZONEFS_LABEL_LEN]; + + /* 128-bit uuid */ + __u8 s_uuid[ZONEFS_UUID_SIZE]; + + /* Features */ + __le64 s_features; + + /* UID/GID to use for files */ + __le32 s_uid; + __le32 s_gid; + + /* File permissions */ + __le32 s_perm; + + /* Padding to ZONEFS_SUPER_SIZE bytes */ + __u8 s_reserved[3988]; + +} __packed; + +/* + * Feature flags: specified in the s_features field of the on-disk super + * block struct zonefs_super and in-memory in the s_feartures field of + * struct zonefs_sb_info. + */ +enum zonefs_features { + /* + * Aggregate contiguous conventional zones into a single file. + */ + ZONEFS_F_AGGRCNV = 1ULL << 0, + /* + * Use super block specified UID for files instead of default 0. + */ + ZONEFS_F_UID = 1ULL << 1, + /* + * Use super block specified GID for files instead of default 0. + */ + ZONEFS_F_GID = 1ULL << 2, + /* + * Use super block specified file permissions instead of default 640. + */ + ZONEFS_F_PERM = 1ULL << 3, +}; + +#define ZONEFS_F_DEFINED_FEATURES \ + (ZONEFS_F_AGGRCNV | ZONEFS_F_UID | ZONEFS_F_GID | ZONEFS_F_PERM) + +/* + * Mount options for zone write pointer error handling. + */ +#define ZONEFS_MNTOPT_ERRORS_RO (1 << 0) /* Make zone file readonly */ +#define ZONEFS_MNTOPT_ERRORS_ZRO (1 << 1) /* Make zone file offline */ +#define ZONEFS_MNTOPT_ERRORS_ZOL (1 << 2) /* Make zone file offline */ +#define ZONEFS_MNTOPT_ERRORS_REPAIR (1 << 3) /* Remount read-only */ +#define ZONEFS_MNTOPT_ERRORS_MASK \ + (ZONEFS_MNTOPT_ERRORS_RO | ZONEFS_MNTOPT_ERRORS_ZRO | \ + ZONEFS_MNTOPT_ERRORS_ZOL | ZONEFS_MNTOPT_ERRORS_REPAIR) +#define ZONEFS_MNTOPT_EXPLICIT_OPEN (1 << 4) /* Explicit open/close of zones on open/close */ + +/* + * In-memory Super block information. + */ +struct zonefs_sb_info { + + unsigned long s_mount_opts; + + spinlock_t s_lock; + + unsigned long long s_features; + kuid_t s_uid; + kgid_t s_gid; + umode_t s_perm; + uuid_t s_uuid; + unsigned int s_zone_sectors_shift; + + struct zonefs_zone_group s_zgroup[ZONEFS_ZTYPE_MAX]; + + loff_t s_blocks; + loff_t s_used_blocks; + + unsigned int s_max_wro_seq_files; + atomic_t s_wro_seq_files; + + unsigned int s_max_active_seq_files; + atomic_t s_active_seq_files; + + bool s_sysfs_registered; + struct kobject s_kobj; + struct completion s_kobj_unregister; +}; + +static inline struct zonefs_sb_info *ZONEFS_SB(struct super_block *sb) +{ + return sb->s_fs_info; +} + +#define zonefs_info(sb, format, args...) \ + pr_info("zonefs (%s): " format, sb->s_id, ## args) +#define zonefs_err(sb, format, args...) \ + pr_err("zonefs (%s) ERROR: " format, sb->s_id, ## args) +#define zonefs_warn(sb, format, args...) \ + pr_warn("zonefs (%s) WARNING: " format, sb->s_id, ## args) + +/* In super.c */ +void zonefs_inode_account_active(struct inode *inode); +int zonefs_inode_zone_mgmt(struct inode *inode, enum req_op op); +void zonefs_i_size_write(struct inode *inode, loff_t isize); +void zonefs_update_stats(struct inode *inode, loff_t new_isize); +void __zonefs_io_error(struct inode *inode, bool write); + +static inline void zonefs_io_error(struct inode *inode, bool write) +{ + struct zonefs_inode_info *zi = ZONEFS_I(inode); + + mutex_lock(&zi->i_truncate_mutex); + __zonefs_io_error(inode, write); + mutex_unlock(&zi->i_truncate_mutex); +} + +/* In super.c */ +extern const struct inode_operations zonefs_dir_inode_operations; +extern const struct file_operations zonefs_dir_operations; + +/* In file.c */ +extern const struct address_space_operations zonefs_file_aops; +extern const struct file_operations zonefs_file_operations; +int zonefs_file_truncate(struct inode *inode, loff_t isize); + +/* In sysfs.c */ +int zonefs_sysfs_register(struct super_block *sb); +void zonefs_sysfs_unregister(struct super_block *sb); +int zonefs_sysfs_init(void); +void zonefs_sysfs_exit(void); + +#endif |