From 64c23e86873ee410554d6d1c76b60da47025e96f Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Fri, 14 Jan 2011 13:07:30 +0100 Subject: make the feature checks in ->fallocate future proof Instead of various home grown checks that might need updates for new flags just check for any bit outside the mask of the features supported by the filesystem. This makes the check future proof for any newly added flag. Signed-off-by: Christoph Hellwig Signed-off-by: Al Viro --- fs/ext4/extents.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/ext4/extents.c') diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c index c4068f6abf03..4bdd160854eb 100644 --- a/fs/ext4/extents.c +++ b/fs/ext4/extents.c @@ -3645,7 +3645,7 @@ long ext4_fallocate(struct inode *inode, int mode, loff_t offset, loff_t len) unsigned int credits, blkbits = inode->i_blkbits; /* We only support the FALLOC_FL_KEEP_SIZE mode */ - if (mode && (mode != FALLOC_FL_KEEP_SIZE)) + if (mode & ~FALLOC_FL_KEEP_SIZE) return -EOPNOTSUPP; /* -- cgit v1.2.3 From 2fe17c1075836b66678ed2a305fd09b6773883aa Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Fri, 14 Jan 2011 13:07:43 +0100 Subject: fallocate should be a file operation Currently all filesystems except XFS implement fallocate asynchronously, while XFS forced a commit. Both of these are suboptimal - in case of O_SYNC I/O we really want our allocation on disk, especially for the !KEEP_SIZE case where we actually grow the file with user-visible zeroes. On the other hand always commiting the transaction is a bad idea for fast-path uses of fallocate like for example in recent Samba versions. Given that block allocation is a data plane operation anyway change it from an inode operation to a file operation so that we have the file structure available that lets us check for O_SYNC. This also includes moving the code around for a few of the filesystems, and remove the already unnedded S_ISDIR checks given that we only wire up fallocate for regular files. Signed-off-by: Christoph Hellwig Signed-off-by: Al Viro --- Documentation/filesystems/Locking | 3 +- fs/btrfs/file.c | 113 +++++++++++++++++ fs/btrfs/inode.c | 111 ---------------- fs/ext4/ext4.h | 2 +- fs/ext4/extents.c | 9 +- fs/ext4/file.c | 2 +- fs/gfs2/file.c | 258 ++++++++++++++++++++++++++++++++++++++ fs/gfs2/ops_inode.c | 258 -------------------------------------- fs/ocfs2/file.c | 8 +- fs/open.c | 4 +- fs/xfs/linux-2.6/xfs_file.c | 56 +++++++++ fs/xfs/linux-2.6/xfs_iops.c | 60 --------- include/linux/fs.h | 4 +- 13 files changed, 440 insertions(+), 448 deletions(-) (limited to 'fs/ext4/extents.c') diff --git a/Documentation/filesystems/Locking b/Documentation/filesystems/Locking index 651d5237c155..4471a416c274 100644 --- a/Documentation/filesystems/Locking +++ b/Documentation/filesystems/Locking @@ -60,7 +60,6 @@ ata *); ssize_t (*listxattr) (struct dentry *, char *, size_t); int (*removexattr) (struct dentry *, const char *); void (*truncate_range)(struct inode *, loff_t, loff_t); - long (*fallocate)(struct inode *inode, int mode, loff_t offset, loff_t len); int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, u64 len); locking rules: @@ -88,7 +87,6 @@ getxattr: no listxattr: no removexattr: yes truncate_range: yes -fallocate: no fiemap: no Additionally, ->rmdir(), ->unlink() and ->rename() have ->i_mutex on victim. @@ -437,6 +435,7 @@ prototypes: ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int); int (*setlease)(struct file *, long, struct file_lock **); + long (*fallocate)(struct file *, int, loff_t, loff_t); }; locking rules: diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index 66836d85763b..a9e0a4eaf3d9 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -24,6 +24,7 @@ #include #include #include +#include #include #include #include @@ -1237,6 +1238,117 @@ static int btrfs_file_mmap(struct file *filp, struct vm_area_struct *vma) return 0; } +static long btrfs_fallocate(struct file *file, int mode, + loff_t offset, loff_t len) +{ + struct inode *inode = file->f_path.dentry->d_inode; + struct extent_state *cached_state = NULL; + u64 cur_offset; + u64 last_byte; + u64 alloc_start; + u64 alloc_end; + u64 alloc_hint = 0; + u64 locked_end; + u64 mask = BTRFS_I(inode)->root->sectorsize - 1; + struct extent_map *em; + int ret; + + alloc_start = offset & ~mask; + alloc_end = (offset + len + mask) & ~mask; + + /* We only support the FALLOC_FL_KEEP_SIZE mode */ + if (mode & ~FALLOC_FL_KEEP_SIZE) + return -EOPNOTSUPP; + + /* + * wait for ordered IO before we have any locks. We'll loop again + * below with the locks held. + */ + btrfs_wait_ordered_range(inode, alloc_start, alloc_end - alloc_start); + + mutex_lock(&inode->i_mutex); + ret = inode_newsize_ok(inode, alloc_end); + if (ret) + goto out; + + if (alloc_start > inode->i_size) { + ret = btrfs_cont_expand(inode, alloc_start); + if (ret) + goto out; + } + + ret = btrfs_check_data_free_space(inode, alloc_end - alloc_start); + if (ret) + goto out; + + locked_end = alloc_end - 1; + while (1) { + struct btrfs_ordered_extent *ordered; + + /* the extent lock is ordered inside the running + * transaction + */ + lock_extent_bits(&BTRFS_I(inode)->io_tree, alloc_start, + locked_end, 0, &cached_state, GFP_NOFS); + ordered = btrfs_lookup_first_ordered_extent(inode, + alloc_end - 1); + if (ordered && + ordered->file_offset + ordered->len > alloc_start && + ordered->file_offset < alloc_end) { + btrfs_put_ordered_extent(ordered); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, + alloc_start, locked_end, + &cached_state, GFP_NOFS); + /* + * we can't wait on the range with the transaction + * running or with the extent lock held + */ + btrfs_wait_ordered_range(inode, alloc_start, + alloc_end - alloc_start); + } else { + if (ordered) + btrfs_put_ordered_extent(ordered); + break; + } + } + + cur_offset = alloc_start; + while (1) { + em = btrfs_get_extent(inode, NULL, 0, cur_offset, + alloc_end - cur_offset, 0); + BUG_ON(IS_ERR(em) || !em); + last_byte = min(extent_map_end(em), alloc_end); + last_byte = (last_byte + mask) & ~mask; + if (em->block_start == EXTENT_MAP_HOLE || + (cur_offset >= inode->i_size && + !test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) { + ret = btrfs_prealloc_file_range(inode, mode, cur_offset, + last_byte - cur_offset, + 1 << inode->i_blkbits, + offset + len, + &alloc_hint); + if (ret < 0) { + free_extent_map(em); + break; + } + } + free_extent_map(em); + + cur_offset = last_byte; + if (cur_offset >= alloc_end) { + ret = 0; + break; + } + } + unlock_extent_cached(&BTRFS_I(inode)->io_tree, alloc_start, locked_end, + &cached_state, GFP_NOFS); + + btrfs_free_reserved_data_space(inode, alloc_end - alloc_start); +out: + mutex_unlock(&inode->i_mutex); + return ret; +} + const struct file_operations btrfs_file_operations = { .llseek = generic_file_llseek, .read = do_sync_read, @@ -1248,6 +1360,7 @@ const struct file_operations btrfs_file_operations = { .open = generic_file_open, .release = btrfs_release_file, .fsync = btrfs_sync_file, + .fallocate = btrfs_fallocate, .unlocked_ioctl = btrfs_ioctl, #ifdef CONFIG_COMPAT .compat_ioctl = btrfs_ioctl, diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 64daf2acd0d5..902afbf50811 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -7098,116 +7098,6 @@ int btrfs_prealloc_file_range_trans(struct inode *inode, min_size, actual_len, alloc_hint, trans); } -static long btrfs_fallocate(struct inode *inode, int mode, - loff_t offset, loff_t len) -{ - struct extent_state *cached_state = NULL; - u64 cur_offset; - u64 last_byte; - u64 alloc_start; - u64 alloc_end; - u64 alloc_hint = 0; - u64 locked_end; - u64 mask = BTRFS_I(inode)->root->sectorsize - 1; - struct extent_map *em; - int ret; - - alloc_start = offset & ~mask; - alloc_end = (offset + len + mask) & ~mask; - - /* We only support the FALLOC_FL_KEEP_SIZE mode */ - if (mode & ~FALLOC_FL_KEEP_SIZE) - return -EOPNOTSUPP; - - /* - * wait for ordered IO before we have any locks. We'll loop again - * below with the locks held. - */ - btrfs_wait_ordered_range(inode, alloc_start, alloc_end - alloc_start); - - mutex_lock(&inode->i_mutex); - ret = inode_newsize_ok(inode, alloc_end); - if (ret) - goto out; - - if (alloc_start > inode->i_size) { - ret = btrfs_cont_expand(inode, alloc_start); - if (ret) - goto out; - } - - ret = btrfs_check_data_free_space(inode, alloc_end - alloc_start); - if (ret) - goto out; - - locked_end = alloc_end - 1; - while (1) { - struct btrfs_ordered_extent *ordered; - - /* the extent lock is ordered inside the running - * transaction - */ - lock_extent_bits(&BTRFS_I(inode)->io_tree, alloc_start, - locked_end, 0, &cached_state, GFP_NOFS); - ordered = btrfs_lookup_first_ordered_extent(inode, - alloc_end - 1); - if (ordered && - ordered->file_offset + ordered->len > alloc_start && - ordered->file_offset < alloc_end) { - btrfs_put_ordered_extent(ordered); - unlock_extent_cached(&BTRFS_I(inode)->io_tree, - alloc_start, locked_end, - &cached_state, GFP_NOFS); - /* - * we can't wait on the range with the transaction - * running or with the extent lock held - */ - btrfs_wait_ordered_range(inode, alloc_start, - alloc_end - alloc_start); - } else { - if (ordered) - btrfs_put_ordered_extent(ordered); - break; - } - } - - cur_offset = alloc_start; - while (1) { - em = btrfs_get_extent(inode, NULL, 0, cur_offset, - alloc_end - cur_offset, 0); - BUG_ON(IS_ERR(em) || !em); - last_byte = min(extent_map_end(em), alloc_end); - last_byte = (last_byte + mask) & ~mask; - if (em->block_start == EXTENT_MAP_HOLE || - (cur_offset >= inode->i_size && - !test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) { - ret = btrfs_prealloc_file_range(inode, mode, cur_offset, - last_byte - cur_offset, - 1 << inode->i_blkbits, - offset + len, - &alloc_hint); - if (ret < 0) { - free_extent_map(em); - break; - } - } - free_extent_map(em); - - cur_offset = last_byte; - if (cur_offset >= alloc_end) { - ret = 0; - break; - } - } - unlock_extent_cached(&BTRFS_I(inode)->io_tree, alloc_start, locked_end, - &cached_state, GFP_NOFS); - - btrfs_free_reserved_data_space(inode, alloc_end - alloc_start); -out: - mutex_unlock(&inode->i_mutex); - return ret; -} - static int btrfs_set_page_dirty(struct page *page) { return __set_page_dirty_nobuffers(page); @@ -7310,7 +7200,6 @@ static const struct inode_operations btrfs_file_inode_operations = { .listxattr = btrfs_listxattr, .removexattr = btrfs_removexattr, .permission = btrfs_permission, - .fallocate = btrfs_fallocate, .fiemap = btrfs_fiemap, }; static const struct inode_operations btrfs_special_inode_operations = { diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index 1de65f572033..0c8d97b56f34 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -2065,7 +2065,7 @@ extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, extern void ext4_ext_truncate(struct inode *); extern void ext4_ext_init(struct super_block *); extern void ext4_ext_release(struct super_block *); -extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset, +extern long ext4_fallocate(struct file *file, int mode, loff_t offset, loff_t len); extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset, ssize_t len); diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c index 4bdd160854eb..63a75810b7c3 100644 --- a/fs/ext4/extents.c +++ b/fs/ext4/extents.c @@ -3627,14 +3627,15 @@ static void ext4_falloc_update_inode(struct inode *inode, } /* - * preallocate space for a file. This implements ext4's fallocate inode + * preallocate space for a file. This implements ext4's fallocate file * operation, which gets called from sys_fallocate system call. * For block-mapped files, posix_fallocate should fall back to the method * of writing zeroes to the required new blocks (the same behavior which is * expected for file systems which do not support fallocate() system call). */ -long ext4_fallocate(struct inode *inode, int mode, loff_t offset, loff_t len) +long ext4_fallocate(struct file *file, int mode, loff_t offset, loff_t len) { + struct inode *inode = file->f_path.dentry->d_inode; handle_t *handle; loff_t new_size; unsigned int max_blocks; @@ -3655,10 +3656,6 @@ long ext4_fallocate(struct inode *inode, int mode, loff_t offset, loff_t len) if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))) return -EOPNOTSUPP; - /* preallocation to directories is currently not supported */ - if (S_ISDIR(inode->i_mode)) - return -ENODEV; - map.m_lblk = offset >> blkbits; /* * We can't just convert len to max_blocks because diff --git a/fs/ext4/file.c b/fs/ext4/file.c index bb003dc9ffff..2e8322c8aa88 100644 --- a/fs/ext4/file.c +++ b/fs/ext4/file.c @@ -210,6 +210,7 @@ const struct file_operations ext4_file_operations = { .fsync = ext4_sync_file, .splice_read = generic_file_splice_read, .splice_write = generic_file_splice_write, + .fallocate = ext4_fallocate, }; const struct inode_operations ext4_file_inode_operations = { @@ -223,7 +224,6 @@ const struct inode_operations ext4_file_inode_operations = { .removexattr = generic_removexattr, #endif .check_acl = ext4_check_acl, - .fallocate = ext4_fallocate, .fiemap = ext4_fiemap, }; diff --git a/fs/gfs2/file.c b/fs/gfs2/file.c index fca6689e12e6..7cfdcb913363 100644 --- a/fs/gfs2/file.c +++ b/fs/gfs2/file.c @@ -19,6 +19,8 @@ #include #include #include +#include +#include #include #include #include @@ -610,6 +612,260 @@ static ssize_t gfs2_file_aio_write(struct kiocb *iocb, const struct iovec *iov, return generic_file_aio_write(iocb, iov, nr_segs, pos); } +static void empty_write_end(struct page *page, unsigned from, + unsigned to) +{ + struct gfs2_inode *ip = GFS2_I(page->mapping->host); + + page_zero_new_buffers(page, from, to); + flush_dcache_page(page); + mark_page_accessed(page); + + if (!gfs2_is_writeback(ip)) + gfs2_page_add_databufs(ip, page, from, to); + + block_commit_write(page, from, to); +} + +static int write_empty_blocks(struct page *page, unsigned from, unsigned to) +{ + unsigned start, end, next; + struct buffer_head *bh, *head; + int error; + + if (!page_has_buffers(page)) { + error = __block_write_begin(page, from, to - from, gfs2_block_map); + if (unlikely(error)) + return error; + + empty_write_end(page, from, to); + return 0; + } + + bh = head = page_buffers(page); + next = end = 0; + while (next < from) { + next += bh->b_size; + bh = bh->b_this_page; + } + start = next; + do { + next += bh->b_size; + if (buffer_mapped(bh)) { + if (end) { + error = __block_write_begin(page, start, end - start, + gfs2_block_map); + if (unlikely(error)) + return error; + empty_write_end(page, start, end); + end = 0; + } + start = next; + } + else + end = next; + bh = bh->b_this_page; + } while (next < to); + + if (end) { + error = __block_write_begin(page, start, end - start, gfs2_block_map); + if (unlikely(error)) + return error; + empty_write_end(page, start, end); + } + + return 0; +} + +static int fallocate_chunk(struct inode *inode, loff_t offset, loff_t len, + int mode) +{ + struct gfs2_inode *ip = GFS2_I(inode); + struct buffer_head *dibh; + int error; + u64 start = offset >> PAGE_CACHE_SHIFT; + unsigned int start_offset = offset & ~PAGE_CACHE_MASK; + u64 end = (offset + len - 1) >> PAGE_CACHE_SHIFT; + pgoff_t curr; + struct page *page; + unsigned int end_offset = (offset + len) & ~PAGE_CACHE_MASK; + unsigned int from, to; + + if (!end_offset) + end_offset = PAGE_CACHE_SIZE; + + error = gfs2_meta_inode_buffer(ip, &dibh); + if (unlikely(error)) + goto out; + + gfs2_trans_add_bh(ip->i_gl, dibh, 1); + + if (gfs2_is_stuffed(ip)) { + error = gfs2_unstuff_dinode(ip, NULL); + if (unlikely(error)) + goto out; + } + + curr = start; + offset = start << PAGE_CACHE_SHIFT; + from = start_offset; + to = PAGE_CACHE_SIZE; + while (curr <= end) { + page = grab_cache_page_write_begin(inode->i_mapping, curr, + AOP_FLAG_NOFS); + if (unlikely(!page)) { + error = -ENOMEM; + goto out; + } + + if (curr == end) + to = end_offset; + error = write_empty_blocks(page, from, to); + if (!error && offset + to > inode->i_size && + !(mode & FALLOC_FL_KEEP_SIZE)) { + i_size_write(inode, offset + to); + } + unlock_page(page); + page_cache_release(page); + if (error) + goto out; + curr++; + offset += PAGE_CACHE_SIZE; + from = 0; + } + + gfs2_dinode_out(ip, dibh->b_data); + mark_inode_dirty(inode); + + brelse(dibh); + +out: + return error; +} + +static void calc_max_reserv(struct gfs2_inode *ip, loff_t max, loff_t *len, + unsigned int *data_blocks, unsigned int *ind_blocks) +{ + const struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); + unsigned int max_blocks = ip->i_alloc->al_rgd->rd_free_clone; + unsigned int tmp, max_data = max_blocks - 3 * (sdp->sd_max_height - 1); + + for (tmp = max_data; tmp > sdp->sd_diptrs;) { + tmp = DIV_ROUND_UP(tmp, sdp->sd_inptrs); + max_data -= tmp; + } + /* This calculation isn't the exact reverse of gfs2_write_calc_reserve, + so it might end up with fewer data blocks */ + if (max_data <= *data_blocks) + return; + *data_blocks = max_data; + *ind_blocks = max_blocks - max_data; + *len = ((loff_t)max_data - 3) << sdp->sd_sb.sb_bsize_shift; + if (*len > max) { + *len = max; + gfs2_write_calc_reserv(ip, max, data_blocks, ind_blocks); + } +} + +static long gfs2_fallocate(struct file *file, int mode, loff_t offset, + loff_t len) +{ + struct inode *inode = file->f_path.dentry->d_inode; + struct gfs2_sbd *sdp = GFS2_SB(inode); + struct gfs2_inode *ip = GFS2_I(inode); + unsigned int data_blocks = 0, ind_blocks = 0, rblocks; + loff_t bytes, max_bytes; + struct gfs2_alloc *al; + int error; + loff_t next = (offset + len - 1) >> sdp->sd_sb.sb_bsize_shift; + next = (next + 1) << sdp->sd_sb.sb_bsize_shift; + + /* We only support the FALLOC_FL_KEEP_SIZE mode */ + if (mode & ~FALLOC_FL_KEEP_SIZE) + return -EOPNOTSUPP; + + offset = (offset >> sdp->sd_sb.sb_bsize_shift) << + sdp->sd_sb.sb_bsize_shift; + + len = next - offset; + bytes = sdp->sd_max_rg_data * sdp->sd_sb.sb_bsize / 2; + if (!bytes) + bytes = UINT_MAX; + + gfs2_holder_init(ip->i_gl, LM_ST_EXCLUSIVE, 0, &ip->i_gh); + error = gfs2_glock_nq(&ip->i_gh); + if (unlikely(error)) + goto out_uninit; + + if (!gfs2_write_alloc_required(ip, offset, len)) + goto out_unlock; + + while (len > 0) { + if (len < bytes) + bytes = len; + al = gfs2_alloc_get(ip); + if (!al) { + error = -ENOMEM; + goto out_unlock; + } + + error = gfs2_quota_lock_check(ip); + if (error) + goto out_alloc_put; + +retry: + gfs2_write_calc_reserv(ip, bytes, &data_blocks, &ind_blocks); + + al->al_requested = data_blocks + ind_blocks; + error = gfs2_inplace_reserve(ip); + if (error) { + if (error == -ENOSPC && bytes > sdp->sd_sb.sb_bsize) { + bytes >>= 1; + goto retry; + } + goto out_qunlock; + } + max_bytes = bytes; + calc_max_reserv(ip, len, &max_bytes, &data_blocks, &ind_blocks); + al->al_requested = data_blocks + ind_blocks; + + rblocks = RES_DINODE + ind_blocks + RES_STATFS + RES_QUOTA + + RES_RG_HDR + gfs2_rg_blocks(al); + if (gfs2_is_jdata(ip)) + rblocks += data_blocks ? data_blocks : 1; + + error = gfs2_trans_begin(sdp, rblocks, + PAGE_CACHE_SIZE/sdp->sd_sb.sb_bsize); + if (error) + goto out_trans_fail; + + error = fallocate_chunk(inode, offset, max_bytes, mode); + gfs2_trans_end(sdp); + + if (error) + goto out_trans_fail; + + len -= max_bytes; + offset += max_bytes; + gfs2_inplace_release(ip); + gfs2_quota_unlock(ip); + gfs2_alloc_put(ip); + } + goto out_unlock; + +out_trans_fail: + gfs2_inplace_release(ip); +out_qunlock: + gfs2_quota_unlock(ip); +out_alloc_put: + gfs2_alloc_put(ip); +out_unlock: + gfs2_glock_dq(&ip->i_gh); +out_uninit: + gfs2_holder_uninit(&ip->i_gh); + return error; +} + #ifdef CONFIG_GFS2_FS_LOCKING_DLM /** @@ -765,6 +1021,7 @@ const struct file_operations gfs2_file_fops = { .splice_read = generic_file_splice_read, .splice_write = generic_file_splice_write, .setlease = gfs2_setlease, + .fallocate = gfs2_fallocate, }; const struct file_operations gfs2_dir_fops = { @@ -794,6 +1051,7 @@ const struct file_operations gfs2_file_fops_nolock = { .splice_read = generic_file_splice_read, .splice_write = generic_file_splice_write, .setlease = generic_setlease, + .fallocate = gfs2_fallocate, }; const struct file_operations gfs2_dir_fops_nolock = { diff --git a/fs/gfs2/ops_inode.c b/fs/gfs2/ops_inode.c index c09528c07f3d..d8b26ac2e20b 100644 --- a/fs/gfs2/ops_inode.c +++ b/fs/gfs2/ops_inode.c @@ -18,8 +18,6 @@ #include #include #include -#include -#include #include #include "gfs2.h" @@ -1257,261 +1255,6 @@ static int gfs2_removexattr(struct dentry *dentry, const char *name) return ret; } -static void empty_write_end(struct page *page, unsigned from, - unsigned to) -{ - struct gfs2_inode *ip = GFS2_I(page->mapping->host); - - page_zero_new_buffers(page, from, to); - flush_dcache_page(page); - mark_page_accessed(page); - - if (!gfs2_is_writeback(ip)) - gfs2_page_add_databufs(ip, page, from, to); - - block_commit_write(page, from, to); -} - - -static int write_empty_blocks(struct page *page, unsigned from, unsigned to) -{ - unsigned start, end, next; - struct buffer_head *bh, *head; - int error; - - if (!page_has_buffers(page)) { - error = __block_write_begin(page, from, to - from, gfs2_block_map); - if (unlikely(error)) - return error; - - empty_write_end(page, from, to); - return 0; - } - - bh = head = page_buffers(page); - next = end = 0; - while (next < from) { - next += bh->b_size; - bh = bh->b_this_page; - } - start = next; - do { - next += bh->b_size; - if (buffer_mapped(bh)) { - if (end) { - error = __block_write_begin(page, start, end - start, - gfs2_block_map); - if (unlikely(error)) - return error; - empty_write_end(page, start, end); - end = 0; - } - start = next; - } - else - end = next; - bh = bh->b_this_page; - } while (next < to); - - if (end) { - error = __block_write_begin(page, start, end - start, gfs2_block_map); - if (unlikely(error)) - return error; - empty_write_end(page, start, end); - } - - return 0; -} - -static int fallocate_chunk(struct inode *inode, loff_t offset, loff_t len, - int mode) -{ - struct gfs2_inode *ip = GFS2_I(inode); - struct buffer_head *dibh; - int error; - u64 start = offset >> PAGE_CACHE_SHIFT; - unsigned int start_offset = offset & ~PAGE_CACHE_MASK; - u64 end = (offset + len - 1) >> PAGE_CACHE_SHIFT; - pgoff_t curr; - struct page *page; - unsigned int end_offset = (offset + len) & ~PAGE_CACHE_MASK; - unsigned int from, to; - - if (!end_offset) - end_offset = PAGE_CACHE_SIZE; - - error = gfs2_meta_inode_buffer(ip, &dibh); - if (unlikely(error)) - goto out; - - gfs2_trans_add_bh(ip->i_gl, dibh, 1); - - if (gfs2_is_stuffed(ip)) { - error = gfs2_unstuff_dinode(ip, NULL); - if (unlikely(error)) - goto out; - } - - curr = start; - offset = start << PAGE_CACHE_SHIFT; - from = start_offset; - to = PAGE_CACHE_SIZE; - while (curr <= end) { - page = grab_cache_page_write_begin(inode->i_mapping, curr, - AOP_FLAG_NOFS); - if (unlikely(!page)) { - error = -ENOMEM; - goto out; - } - - if (curr == end) - to = end_offset; - error = write_empty_blocks(page, from, to); - if (!error && offset + to > inode->i_size && - !(mode & FALLOC_FL_KEEP_SIZE)) { - i_size_write(inode, offset + to); - } - unlock_page(page); - page_cache_release(page); - if (error) - goto out; - curr++; - offset += PAGE_CACHE_SIZE; - from = 0; - } - - gfs2_dinode_out(ip, dibh->b_data); - mark_inode_dirty(inode); - - brelse(dibh); - -out: - return error; -} - -static void calc_max_reserv(struct gfs2_inode *ip, loff_t max, loff_t *len, - unsigned int *data_blocks, unsigned int *ind_blocks) -{ - const struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); - unsigned int max_blocks = ip->i_alloc->al_rgd->rd_free_clone; - unsigned int tmp, max_data = max_blocks - 3 * (sdp->sd_max_height - 1); - - for (tmp = max_data; tmp > sdp->sd_diptrs;) { - tmp = DIV_ROUND_UP(tmp, sdp->sd_inptrs); - max_data -= tmp; - } - /* This calculation isn't the exact reverse of gfs2_write_calc_reserve, - so it might end up with fewer data blocks */ - if (max_data <= *data_blocks) - return; - *data_blocks = max_data; - *ind_blocks = max_blocks - max_data; - *len = ((loff_t)max_data - 3) << sdp->sd_sb.sb_bsize_shift; - if (*len > max) { - *len = max; - gfs2_write_calc_reserv(ip, max, data_blocks, ind_blocks); - } -} - -static long gfs2_fallocate(struct inode *inode, int mode, loff_t offset, - loff_t len) -{ - struct gfs2_sbd *sdp = GFS2_SB(inode); - struct gfs2_inode *ip = GFS2_I(inode); - unsigned int data_blocks = 0, ind_blocks = 0, rblocks; - loff_t bytes, max_bytes; - struct gfs2_alloc *al; - int error; - loff_t next = (offset + len - 1) >> sdp->sd_sb.sb_bsize_shift; - next = (next + 1) << sdp->sd_sb.sb_bsize_shift; - - /* We only support the FALLOC_FL_KEEP_SIZE mode */ - if (mode & ~FALLOC_FL_KEEP_SIZE) - return -EOPNOTSUPP; - - offset = (offset >> sdp->sd_sb.sb_bsize_shift) << - sdp->sd_sb.sb_bsize_shift; - - len = next - offset; - bytes = sdp->sd_max_rg_data * sdp->sd_sb.sb_bsize / 2; - if (!bytes) - bytes = UINT_MAX; - - gfs2_holder_init(ip->i_gl, LM_ST_EXCLUSIVE, 0, &ip->i_gh); - error = gfs2_glock_nq(&ip->i_gh); - if (unlikely(error)) - goto out_uninit; - - if (!gfs2_write_alloc_required(ip, offset, len)) - goto out_unlock; - - while (len > 0) { - if (len < bytes) - bytes = len; - al = gfs2_alloc_get(ip); - if (!al) { - error = -ENOMEM; - goto out_unlock; - } - - error = gfs2_quota_lock_check(ip); - if (error) - goto out_alloc_put; - -retry: - gfs2_write_calc_reserv(ip, bytes, &data_blocks, &ind_blocks); - - al->al_requested = data_blocks + ind_blocks; - error = gfs2_inplace_reserve(ip); - if (error) { - if (error == -ENOSPC && bytes > sdp->sd_sb.sb_bsize) { - bytes >>= 1; - goto retry; - } - goto out_qunlock; - } - max_bytes = bytes; - calc_max_reserv(ip, len, &max_bytes, &data_blocks, &ind_blocks); - al->al_requested = data_blocks + ind_blocks; - - rblocks = RES_DINODE + ind_blocks + RES_STATFS + RES_QUOTA + - RES_RG_HDR + gfs2_rg_blocks(al); - if (gfs2_is_jdata(ip)) - rblocks += data_blocks ? data_blocks : 1; - - error = gfs2_trans_begin(sdp, rblocks, - PAGE_CACHE_SIZE/sdp->sd_sb.sb_bsize); - if (error) - goto out_trans_fail; - - error = fallocate_chunk(inode, offset, max_bytes, mode); - gfs2_trans_end(sdp); - - if (error) - goto out_trans_fail; - - len -= max_bytes; - offset += max_bytes; - gfs2_inplace_release(ip); - gfs2_quota_unlock(ip); - gfs2_alloc_put(ip); - } - goto out_unlock; - -out_trans_fail: - gfs2_inplace_release(ip); -out_qunlock: - gfs2_quota_unlock(ip); -out_alloc_put: - gfs2_alloc_put(ip); -out_unlock: - gfs2_glock_dq(&ip->i_gh); -out_uninit: - gfs2_holder_uninit(&ip->i_gh); - return error; -} - - static int gfs2_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, u64 start, u64 len) { @@ -1562,7 +1305,6 @@ const struct inode_operations gfs2_file_iops = { .getxattr = gfs2_getxattr, .listxattr = gfs2_listxattr, .removexattr = gfs2_removexattr, - .fallocate = gfs2_fallocate, .fiemap = gfs2_fiemap, }; diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c index cf254ce8c941..a6651956482e 100644 --- a/fs/ocfs2/file.c +++ b/fs/ocfs2/file.c @@ -1989,9 +1989,10 @@ int ocfs2_change_file_space(struct file *file, unsigned int cmd, return __ocfs2_change_file_space(file, inode, file->f_pos, cmd, sr, 0); } -static long ocfs2_fallocate(struct inode *inode, int mode, loff_t offset, +static long ocfs2_fallocate(struct file *file, int mode, loff_t offset, loff_t len) { + struct inode *inode = file->f_path.dentry->d_inode; struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); struct ocfs2_space_resv sr; int change_size = 1; @@ -2002,9 +2003,6 @@ static long ocfs2_fallocate(struct inode *inode, int mode, loff_t offset, if (!ocfs2_writes_unwritten_extents(osb)) return -EOPNOTSUPP; - if (S_ISDIR(inode->i_mode)) - return -ENODEV; - if (mode & FALLOC_FL_KEEP_SIZE) change_size = 0; @@ -2612,7 +2610,6 @@ const struct inode_operations ocfs2_file_iops = { .getxattr = generic_getxattr, .listxattr = ocfs2_listxattr, .removexattr = generic_removexattr, - .fallocate = ocfs2_fallocate, .fiemap = ocfs2_fiemap, }; @@ -2644,6 +2641,7 @@ const struct file_operations ocfs2_fops = { .flock = ocfs2_flock, .splice_read = ocfs2_file_splice_read, .splice_write = ocfs2_file_splice_write, + .fallocate = ocfs2_fallocate, }; const struct file_operations ocfs2_dops = { diff --git a/fs/open.c b/fs/open.c index 5b6ef7e2859e..e52389e1f05b 100644 --- a/fs/open.c +++ b/fs/open.c @@ -255,10 +255,10 @@ int do_fallocate(struct file *file, int mode, loff_t offset, loff_t len) if (((offset + len) > inode->i_sb->s_maxbytes) || ((offset + len) < 0)) return -EFBIG; - if (!inode->i_op->fallocate) + if (!file->f_op->fallocate) return -EOPNOTSUPP; - return inode->i_op->fallocate(inode, mode, offset, len); + return file->f_op->fallocate(file, mode, offset, len); } SYSCALL_DEFINE(fallocate)(int fd, int mode, loff_t offset, loff_t len) diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c index ef51eb43e137..a55c1b46b219 100644 --- a/fs/xfs/linux-2.6/xfs_file.c +++ b/fs/xfs/linux-2.6/xfs_file.c @@ -37,6 +37,7 @@ #include "xfs_trace.h" #include +#include static const struct vm_operations_struct xfs_file_vm_ops; @@ -882,6 +883,60 @@ out_unlock: return ret; } +STATIC long +xfs_file_fallocate( + struct file *file, + int mode, + loff_t offset, + loff_t len) +{ + struct inode *inode = file->f_path.dentry->d_inode; + long error; + loff_t new_size = 0; + xfs_flock64_t bf; + xfs_inode_t *ip = XFS_I(inode); + int cmd = XFS_IOC_RESVSP; + + if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE)) + return -EOPNOTSUPP; + + bf.l_whence = 0; + bf.l_start = offset; + bf.l_len = len; + + xfs_ilock(ip, XFS_IOLOCK_EXCL); + + if (mode & FALLOC_FL_PUNCH_HOLE) + cmd = XFS_IOC_UNRESVSP; + + /* check the new inode size is valid before allocating */ + if (!(mode & FALLOC_FL_KEEP_SIZE) && + offset + len > i_size_read(inode)) { + new_size = offset + len; + error = inode_newsize_ok(inode, new_size); + if (error) + goto out_unlock; + } + + error = -xfs_change_file_space(ip, cmd, &bf, 0, XFS_ATTR_NOLOCK); + if (error) + goto out_unlock; + + /* Change file size if needed */ + if (new_size) { + struct iattr iattr; + + iattr.ia_valid = ATTR_SIZE; + iattr.ia_size = new_size; + error = -xfs_setattr(ip, &iattr, XFS_ATTR_NOLOCK); + } + +out_unlock: + xfs_iunlock(ip, XFS_IOLOCK_EXCL); + return error; +} + + STATIC int xfs_file_open( struct inode *inode, @@ -1000,6 +1055,7 @@ const struct file_operations xfs_file_operations = { .open = xfs_file_open, .release = xfs_file_release, .fsync = xfs_file_fsync, + .fallocate = xfs_file_fallocate, }; const struct file_operations xfs_dir_file_operations = { diff --git a/fs/xfs/linux-2.6/xfs_iops.c b/fs/xfs/linux-2.6/xfs_iops.c index a4ecc2188a09..bd5727852fd6 100644 --- a/fs/xfs/linux-2.6/xfs_iops.c +++ b/fs/xfs/linux-2.6/xfs_iops.c @@ -46,7 +46,6 @@ #include #include #include -#include #include #include @@ -505,64 +504,6 @@ xfs_vn_setattr( return -xfs_setattr(XFS_I(dentry->d_inode), iattr, 0); } -STATIC long -xfs_vn_fallocate( - struct inode *inode, - int mode, - loff_t offset, - loff_t len) -{ - long error; - loff_t new_size = 0; - xfs_flock64_t bf; - xfs_inode_t *ip = XFS_I(inode); - int cmd = XFS_IOC_RESVSP; - - if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE)) - return -EOPNOTSUPP; - - /* preallocation on directories not yet supported */ - error = -ENODEV; - if (S_ISDIR(inode->i_mode)) - goto out_error; - - bf.l_whence = 0; - bf.l_start = offset; - bf.l_len = len; - - xfs_ilock(ip, XFS_IOLOCK_EXCL); - - if (mode & FALLOC_FL_PUNCH_HOLE) - cmd = XFS_IOC_UNRESVSP; - - /* check the new inode size is valid before allocating */ - if (!(mode & FALLOC_FL_KEEP_SIZE) && - offset + len > i_size_read(inode)) { - new_size = offset + len; - error = inode_newsize_ok(inode, new_size); - if (error) - goto out_unlock; - } - - error = -xfs_change_file_space(ip, cmd, &bf, 0, XFS_ATTR_NOLOCK); - if (error) - goto out_unlock; - - /* Change file size if needed */ - if (new_size) { - struct iattr iattr; - - iattr.ia_valid = ATTR_SIZE; - iattr.ia_size = new_size; - error = -xfs_setattr(ip, &iattr, XFS_ATTR_NOLOCK); - } - -out_unlock: - xfs_iunlock(ip, XFS_IOLOCK_EXCL); -out_error: - return error; -} - #define XFS_FIEMAP_FLAGS (FIEMAP_FLAG_SYNC|FIEMAP_FLAG_XATTR) /* @@ -656,7 +597,6 @@ static const struct inode_operations xfs_inode_operations = { .getxattr = generic_getxattr, .removexattr = generic_removexattr, .listxattr = xfs_vn_listxattr, - .fallocate = xfs_vn_fallocate, .fiemap = xfs_vn_fiemap, }; diff --git a/include/linux/fs.h b/include/linux/fs.h index 177b4ddea418..09b5bd6a7c6b 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -1552,6 +1552,8 @@ struct file_operations { ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int); ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int); int (*setlease)(struct file *, long, struct file_lock **); + long (*fallocate)(struct file *file, int mode, loff_t offset, + loff_t len); }; #define IPERM_FLAG_RCU 0x0001 @@ -1582,8 +1584,6 @@ struct inode_operations { ssize_t (*listxattr) (struct dentry *, char *, size_t); int (*removexattr) (struct dentry *, const char *); void (*truncate_range)(struct inode *, loff_t, loff_t); - long (*fallocate)(struct inode *inode, int mode, loff_t offset, - loff_t len); int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, u64 len); } ____cacheline_aligned; -- cgit v1.2.3 From e9e3bcecf44c04b9e6b505fd8e2eb9cea58fb94d Mon Sep 17 00:00:00 2001 From: Eric Sandeen Date: Sat, 12 Feb 2011 08:17:34 -0500 Subject: ext4: serialize unaligned asynchronous DIO ext4 has a data corruption case when doing non-block-aligned asynchronous direct IO into a sparse file, as demonstrated by xfstest 240. The root cause is that while ext4 preallocates space in the hole, mappings of that space still look "new" and dio_zero_block() will zero out the unwritten portions. When more than one AIO thread is going, they both find this "new" block and race to zero out their portion; this is uncoordinated and causes data corruption. Dave Chinner fixed this for xfs by simply serializing all unaligned asynchronous direct IO. I've done the same here. The difference is that we only wait on conversions, not all IO. This is a very big hammer, and I'm not very pleased with stuffing this into ext4_file_write(). But since ext4 is DIO_LOCKING, we need to serialize it at this high level. I tried to move this into ext4_ext_direct_IO, but by then we have the i_mutex already, and we will wait on the work queue to do conversions - which must also take the i_mutex. So that won't work. This was originally exposed by qemu-kvm installing to a raw disk image with a normal sector-63 alignment. I've tested a backport of this patch with qemu, and it does avoid the corruption. It is also quite a lot slower (14 min for package installs, vs. 8 min for well-aligned) but I'll take slow correctness over fast corruption any day. Mingming suggested that we can track outstanding conversions, and wait on those so that non-sparse files won't be affected, and I've implemented that here; unaligned AIO to nonsparse files won't take a perf hit. [tytso@mit.edu: Keep the mutex as a hashed array instead of bloating the ext4 inode] [tytso@mit.edu: Fix up namespace issues so that global variables are protected with an "ext4_" prefix.] Signed-off-by: Eric Sandeen Signed-off-by: "Theodore Ts'o" --- fs/ext4/ext4.h | 10 ++++++++++ fs/ext4/extents.c | 10 ++++++---- fs/ext4/file.c | 60 ++++++++++++++++++++++++++++++++++++++++++++++++++++++- fs/ext4/page-io.c | 25 ++++++++++++----------- fs/ext4/super.c | 13 +++++++++++- 5 files changed, 100 insertions(+), 18 deletions(-) (limited to 'fs/ext4/extents.c') diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index 0c8d97b56f34..3aa0b72b3b94 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -848,6 +848,7 @@ struct ext4_inode_info { atomic_t i_ioend_count; /* Number of outstanding io_end structs */ /* current io_end structure for async DIO write*/ ext4_io_end_t *cur_aio_dio; + atomic_t i_aiodio_unwritten; /* Nr. of inflight conversions pending */ spinlock_t i_block_reservation_lock; @@ -2119,6 +2120,15 @@ static inline void set_bitmap_uptodate(struct buffer_head *bh) #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) +/* For ioend & aio unwritten conversion wait queues */ +#define EXT4_WQ_HASH_SZ 37 +#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ + EXT4_WQ_HASH_SZ]) +#define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\ + EXT4_WQ_HASH_SZ]) +extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; +extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ]; + #endif /* __KERNEL__ */ #endif /* _EXT4_H */ diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c index 63a75810b7c3..ccce8a7e94ed 100644 --- a/fs/ext4/extents.c +++ b/fs/ext4/extents.c @@ -3174,9 +3174,10 @@ ext4_ext_handle_uninitialized_extents(handle_t *handle, struct inode *inode, * that this IO needs to convertion to written when IO is * completed */ - if (io) + if (io && !(io->flag & EXT4_IO_END_UNWRITTEN)) { io->flag = EXT4_IO_END_UNWRITTEN; - else + atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten); + } else ext4_set_inode_state(inode, EXT4_STATE_DIO_UNWRITTEN); if (ext4_should_dioread_nolock(inode)) map->m_flags |= EXT4_MAP_UNINIT; @@ -3463,9 +3464,10 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, * that we need to perform convertion when IO is done. */ if ((flags & EXT4_GET_BLOCKS_PRE_IO)) { - if (io) + if (io && !(io->flag & EXT4_IO_END_UNWRITTEN)) { io->flag = EXT4_IO_END_UNWRITTEN; - else + atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten); + } else ext4_set_inode_state(inode, EXT4_STATE_DIO_UNWRITTEN); } diff --git a/fs/ext4/file.c b/fs/ext4/file.c index 2e8322c8aa88..7b80d543b89e 100644 --- a/fs/ext4/file.c +++ b/fs/ext4/file.c @@ -55,11 +55,47 @@ static int ext4_release_file(struct inode *inode, struct file *filp) return 0; } +static void ext4_aiodio_wait(struct inode *inode) +{ + wait_queue_head_t *wq = ext4_ioend_wq(inode); + + wait_event(*wq, (atomic_read(&EXT4_I(inode)->i_aiodio_unwritten) == 0)); +} + +/* + * This tests whether the IO in question is block-aligned or not. + * Ext4 utilizes unwritten extents when hole-filling during direct IO, and they + * are converted to written only after the IO is complete. Until they are + * mapped, these blocks appear as holes, so dio_zero_block() will assume that + * it needs to zero out portions of the start and/or end block. If 2 AIO + * threads are at work on the same unwritten block, they must be synchronized + * or one thread will zero the other's data, causing corruption. + */ +static int +ext4_unaligned_aio(struct inode *inode, const struct iovec *iov, + unsigned long nr_segs, loff_t pos) +{ + struct super_block *sb = inode->i_sb; + int blockmask = sb->s_blocksize - 1; + size_t count = iov_length(iov, nr_segs); + loff_t final_size = pos + count; + + if (pos >= inode->i_size) + return 0; + + if ((pos & blockmask) || (final_size & blockmask)) + return 1; + + return 0; +} + static ssize_t ext4_file_write(struct kiocb *iocb, const struct iovec *iov, unsigned long nr_segs, loff_t pos) { struct inode *inode = iocb->ki_filp->f_path.dentry->d_inode; + int unaligned_aio = 0; + int ret; /* * If we have encountered a bitmap-format file, the size limit @@ -78,9 +114,31 @@ ext4_file_write(struct kiocb *iocb, const struct iovec *iov, nr_segs = iov_shorten((struct iovec *)iov, nr_segs, sbi->s_bitmap_maxbytes - pos); } + } else if (unlikely((iocb->ki_filp->f_flags & O_DIRECT) && + !is_sync_kiocb(iocb))) { + unaligned_aio = ext4_unaligned_aio(inode, iov, nr_segs, pos); } - return generic_file_aio_write(iocb, iov, nr_segs, pos); + /* Unaligned direct AIO must be serialized; see comment above */ + if (unaligned_aio) { + static unsigned long unaligned_warn_time; + + /* Warn about this once per day */ + if (printk_timed_ratelimit(&unaligned_warn_time, 60*60*24*HZ)) + ext4_msg(inode->i_sb, KERN_WARNING, + "Unaligned AIO/DIO on inode %ld by %s; " + "performance will be poor.", + inode->i_ino, current->comm); + mutex_lock(ext4_aio_mutex(inode)); + ext4_aiodio_wait(inode); + } + + ret = generic_file_aio_write(iocb, iov, nr_segs, pos); + + if (unaligned_aio) + mutex_unlock(ext4_aio_mutex(inode)); + + return ret; } static const struct vm_operations_struct ext4_file_vm_ops = { diff --git a/fs/ext4/page-io.c b/fs/ext4/page-io.c index 4e9b0a242f4c..955cc309142f 100644 --- a/fs/ext4/page-io.c +++ b/fs/ext4/page-io.c @@ -32,14 +32,8 @@ static struct kmem_cache *io_page_cachep, *io_end_cachep; -#define WQ_HASH_SZ 37 -#define to_ioend_wq(v) (&ioend_wq[((unsigned long)v) % WQ_HASH_SZ]) -static wait_queue_head_t ioend_wq[WQ_HASH_SZ]; - int __init ext4_init_pageio(void) { - int i; - io_page_cachep = KMEM_CACHE(ext4_io_page, SLAB_RECLAIM_ACCOUNT); if (io_page_cachep == NULL) return -ENOMEM; @@ -48,9 +42,6 @@ int __init ext4_init_pageio(void) kmem_cache_destroy(io_page_cachep); return -ENOMEM; } - for (i = 0; i < WQ_HASH_SZ; i++) - init_waitqueue_head(&ioend_wq[i]); - return 0; } @@ -62,7 +53,7 @@ void ext4_exit_pageio(void) void ext4_ioend_wait(struct inode *inode) { - wait_queue_head_t *wq = to_ioend_wq(inode); + wait_queue_head_t *wq = ext4_ioend_wq(inode); wait_event(*wq, (atomic_read(&EXT4_I(inode)->i_ioend_count) == 0)); } @@ -87,7 +78,7 @@ void ext4_free_io_end(ext4_io_end_t *io) for (i = 0; i < io->num_io_pages; i++) put_io_page(io->pages[i]); io->num_io_pages = 0; - wq = to_ioend_wq(io->inode); + wq = ext4_ioend_wq(io->inode); if (atomic_dec_and_test(&EXT4_I(io->inode)->i_ioend_count) && waitqueue_active(wq)) wake_up_all(wq); @@ -102,6 +93,7 @@ int ext4_end_io_nolock(ext4_io_end_t *io) struct inode *inode = io->inode; loff_t offset = io->offset; ssize_t size = io->size; + wait_queue_head_t *wq; int ret = 0; ext4_debug("ext4_end_io_nolock: io 0x%p from inode %lu,list->next 0x%p," @@ -126,7 +118,16 @@ int ext4_end_io_nolock(ext4_io_end_t *io) if (io->iocb) aio_complete(io->iocb, io->result, 0); /* clear the DIO AIO unwritten flag */ - io->flag &= ~EXT4_IO_END_UNWRITTEN; + if (io->flag & EXT4_IO_END_UNWRITTEN) { + io->flag &= ~EXT4_IO_END_UNWRITTEN; + /* Wake up anyone waiting on unwritten extent conversion */ + wq = ext4_ioend_wq(io->inode); + if (atomic_dec_and_test(&EXT4_I(inode)->i_aiodio_unwritten) && + waitqueue_active(wq)) { + wake_up_all(wq); + } + } + return ret; } diff --git a/fs/ext4/super.c b/fs/ext4/super.c index 86b05486dc63..f6a318f836b2 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -833,6 +833,7 @@ static struct inode *ext4_alloc_inode(struct super_block *sb) ei->i_sync_tid = 0; ei->i_datasync_tid = 0; atomic_set(&ei->i_ioend_count, 0); + atomic_set(&ei->i_aiodio_unwritten, 0); return &ei->vfs_inode; } @@ -4800,11 +4801,21 @@ static void ext4_exit_feat_adverts(void) kfree(ext4_feat); } +/* Shared across all ext4 file systems */ +wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; +struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ]; + static int __init ext4_init_fs(void) { - int err; + int i, err; ext4_check_flag_values(); + + for (i = 0; i < EXT4_WQ_HASH_SZ; i++) { + mutex_init(&ext4__aio_mutex[i]); + init_waitqueue_head(&ext4__ioend_wq[i]); + } + err = ext4_init_pageio(); if (err) return err; -- cgit v1.2.3