diff options
Diffstat (limited to 'fs/btrfs')
| -rw-r--r-- | fs/btrfs/btrfs_inode.h | 9 | ||||
| -rw-r--r-- | fs/btrfs/ctree.h | 3 | ||||
| -rw-r--r-- | fs/btrfs/dev-replace.c | 2 | ||||
| -rw-r--r-- | fs/btrfs/inode.c | 15 | ||||
| -rw-r--r-- | fs/btrfs/ioctl.c | 2 | ||||
| -rw-r--r-- | fs/btrfs/reflink.c | 15 | ||||
| -rw-r--r-- | fs/btrfs/space-info.c | 2 |
7 files changed, 40 insertions, 8 deletions
diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h index c974f7fe614a..8de4bf8edb9c 100644 --- a/fs/btrfs/btrfs_inode.h +++ b/fs/btrfs/btrfs_inode.h @@ -42,6 +42,15 @@ enum { * to an inode. */ BTRFS_INODE_NO_XATTRS, + /* + * Set when we are in a context where we need to start a transaction and + * have dirty pages with the respective file range locked. This is to + * ensure that when reserving space for the transaction, if we are low + * on available space and need to flush delalloc, we will not flush + * delalloc for this inode, because that could result in a deadlock (on + * the file range, inode's io_tree). + */ + BTRFS_INODE_NO_DELALLOC_FLUSH, }; /* in memory btrfs inode */ diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 62461239600f..e01545538e07 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -3001,7 +3001,8 @@ int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans, u32 min_type); int btrfs_start_delalloc_snapshot(struct btrfs_root *root); -int btrfs_start_delalloc_roots(struct btrfs_fs_info *fs_info, u64 nr); +int btrfs_start_delalloc_roots(struct btrfs_fs_info *fs_info, u64 nr, + bool in_reclaim_context); int btrfs_set_extent_delalloc(struct btrfs_inode *inode, u64 start, u64 end, unsigned int extra_bits, struct extent_state **cached_state); diff --git a/fs/btrfs/dev-replace.c b/fs/btrfs/dev-replace.c index 10638537b9ef..d29780463182 100644 --- a/fs/btrfs/dev-replace.c +++ b/fs/btrfs/dev-replace.c @@ -703,7 +703,7 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info, * flush all outstanding I/O and inode extent mappings before the * copy operation is declared as being finished */ - ret = btrfs_start_delalloc_roots(fs_info, U64_MAX); + ret = btrfs_start_delalloc_roots(fs_info, U64_MAX, false); if (ret) { mutex_unlock(&dev_replace->lock_finishing_cancel_unmount); return ret; diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 7e8d8169779d..07479250221d 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -9389,7 +9389,8 @@ static struct btrfs_delalloc_work *btrfs_alloc_delalloc_work(struct inode *inode * some fairly slow code that needs optimization. This walks the list * of all the inodes with pending delalloc and forces them to disk. */ -static int start_delalloc_inodes(struct btrfs_root *root, u64 *nr, bool snapshot) +static int start_delalloc_inodes(struct btrfs_root *root, u64 *nr, bool snapshot, + bool in_reclaim_context) { struct btrfs_inode *binode; struct inode *inode; @@ -9410,6 +9411,11 @@ static int start_delalloc_inodes(struct btrfs_root *root, u64 *nr, bool snapshot list_move_tail(&binode->delalloc_inodes, &root->delalloc_inodes); + + if (in_reclaim_context && + test_bit(BTRFS_INODE_NO_DELALLOC_FLUSH, &binode->runtime_flags)) + continue; + inode = igrab(&binode->vfs_inode); if (!inode) { cond_resched_lock(&root->delalloc_lock); @@ -9463,10 +9469,11 @@ int btrfs_start_delalloc_snapshot(struct btrfs_root *root) if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) return -EROFS; - return start_delalloc_inodes(root, &nr, true); + return start_delalloc_inodes(root, &nr, true, false); } -int btrfs_start_delalloc_roots(struct btrfs_fs_info *fs_info, u64 nr) +int btrfs_start_delalloc_roots(struct btrfs_fs_info *fs_info, u64 nr, + bool in_reclaim_context) { struct btrfs_root *root; struct list_head splice; @@ -9489,7 +9496,7 @@ int btrfs_start_delalloc_roots(struct btrfs_fs_info *fs_info, u64 nr) &fs_info->delalloc_roots); spin_unlock(&fs_info->delalloc_root_lock); - ret = start_delalloc_inodes(root, &nr, false); + ret = start_delalloc_inodes(root, &nr, false, in_reclaim_context); btrfs_put_root(root); if (ret < 0) goto out; diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index e8ca229a216b..bd46e107f955 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -4940,7 +4940,7 @@ long btrfs_ioctl(struct file *file, unsigned int case BTRFS_IOC_SYNC: { int ret; - ret = btrfs_start_delalloc_roots(fs_info, U64_MAX); + ret = btrfs_start_delalloc_roots(fs_info, U64_MAX, false); if (ret) return ret; ret = btrfs_sync_fs(inode->i_sb, 1); diff --git a/fs/btrfs/reflink.c b/fs/btrfs/reflink.c index 99aa87c08912..a646af95dd10 100644 --- a/fs/btrfs/reflink.c +++ b/fs/btrfs/reflink.c @@ -89,6 +89,19 @@ static int copy_inline_to_page(struct btrfs_inode *inode, if (ret) goto out_unlock; + /* + * After dirtying the page our caller will need to start a transaction, + * and if we are low on metadata free space, that can cause flushing of + * delalloc for all inodes in order to get metadata space released. + * However we are holding the range locked for the whole duration of + * the clone/dedupe operation, so we may deadlock if that happens and no + * other task releases enough space. So mark this inode as not being + * possible to flush to avoid such deadlock. We will clear that flag + * when we finish cloning all extents, since a transaction is started + * after finding each extent to clone. + */ + set_bit(BTRFS_INODE_NO_DELALLOC_FLUSH, &inode->runtime_flags); + if (comp_type == BTRFS_COMPRESS_NONE) { char *map; @@ -547,6 +560,8 @@ process_slot: out: btrfs_free_path(path); kvfree(buf); + clear_bit(BTRFS_INODE_NO_DELALLOC_FLUSH, &BTRFS_I(inode)->runtime_flags); + return ret; } diff --git a/fs/btrfs/space-info.c b/fs/btrfs/space-info.c index 64099565ab8f..67e55c5479b8 100644 --- a/fs/btrfs/space-info.c +++ b/fs/btrfs/space-info.c @@ -532,7 +532,7 @@ static void shrink_delalloc(struct btrfs_fs_info *fs_info, loops = 0; while ((delalloc_bytes || dio_bytes) && loops < 3) { - btrfs_start_delalloc_roots(fs_info, items); + btrfs_start_delalloc_roots(fs_info, items, true); loops++; if (wait_ordered && !trans) { |