bcachefs: skip invisible entries in empty subvolume checking
PATCH 1/1 bcachefs: skip invisible entries in empty subvolume checking
When we are checking whether a subvolume is empty in the specified snapshot, entries that do not belong to this subvolume should be skipped.
This fixes the following case:
$ bcachefs subvolume create ./sub
$ cd sub
$ bcachefs subvolume create ./sub2
$ bcachefs subvolume snapshot . ./snap
$ ls -a snap
. ..
$ rmdir snap
rmdir: failed to remove 'snap': Directory not empty--- fs/bcachefs/dirent.c | 7 +++++-- fs/bcachefs/dirent.h | 2 +- fs/bcachefs/inode.c | 2 +- 3 files changed, 7 insertions(+), 4 deletions(-)
fs/bcachefs/dirent.c +5 -2
@@ -508,7 +508,7 @@ u64 bch2_dirent_lookup(struct bch_fs *c, subvol_inum dir,return ret; } -int bch2_empty_dir_snapshot (struct btree_trans *trans,u64 dir,u32 snapshot) +int bch2_empty_dir_snapshot (struct btree_trans *trans,u64 dir,u32 subvol,u32 snapshot) {struct btree_iter iter;struct bkey_s_c k;
@@ -518,6 +518,9 @@ int bch2_empty_dir_snapshot(struct btree_trans *trans, u64 dir, u32 snapshot) SPOS(dir, 0, snapshot), POS(dir, U64_MAX), 0, k, ret)if (k.k->type == KEY_TYPE_dirent) { +struct bkey_s_c_dirent d =bkey_s_c_to_dirent (k); +if (d.v->d_type ==DT_SUBVOL &&le32_to_cpu (d.v->d_parent_subvol) != subvol) +continue ; ret = -ENOTEMPTY ;break ; }
@@ -531,7 +534,7 @@ int bch2_empty_dir_trans(struct btree_trans *trans, subvol_inum dir)u32 snapshot;return bch2_subvolume_get_snapshot (trans, dir.subvol, &snapshot) ?: -bch2_empty_dir_snapshot (trans, dir.inum, snapshot); +bch2_empty_dir_snapshot (trans, dir.inum, dir.subvol, snapshot); }int bch2_readdir (struct bch_fs *c,subvol_inum inum,struct dir_context *ctx)
fs/bcachefs/dirent.h +1 -1
@@ -69,7 +69,7 @@ u64 bch2_dirent_lookup(struct bch_fs *, subvol_inum, const struct bch_hash_info *, const struct qstr *, subvol_inum *); -int bch2_empty_dir_snapshot (struct btree_trans *,u64 ,u32 ); +int bch2_empty_dir_snapshot (struct btree_trans *,u64 ,u32 ,u32 );int bch2_empty_dir_trans (struct btree_trans *,subvol_inum );int bch2_readdir (struct bch_fs *,subvol_inum ,struct dir_context *);
fs/bcachefs/inode.c +1 -1
@@ -1088,7 +1088,7 @@ static int may_delete_deleted_inode(struct btree_trans *trans, goto out;if (S_ISDIR (inode.bi_mode)) { - ret =bch2_empty_dir_snapshot (trans, pos.offset, pos.snapshot); + ret =bch2_empty_dir_snapshot (trans, pos.offset, inode.bi_subvol, pos.snapshot);if (fsck_err_on (ret == -ENOTEMPTY , c, deleted_inode_is_dir,"non empty directory %llu:%u in deleted_inodes btree" , pos.offset, pos.snapshot))
Signature · 1 line
2.43.0
Quoted 14 lines · Guoyu Ou
When we are checking whether a subvolume is empty in the specified snapshot, entries that do not belong to this subvolume should be skipped.
This fixes the following case:
$ bcachefs subvolume create ./sub $ cd sub $ bcachefs subvolume create ./sub2 $ bcachefs subvolume snapshot . ./snap $ ls -a snap . .. $ rmdir snap rmdir: failed to remove 'snap': Directory not empty
Quoted 1 line
Signed-off-by Guoyu Ou <[email protected]>
So that's an interesting corner case - can you add a ktest test for it?
https://evilpiepirate.org/git/ktest.git/
If you're not using it yet, please jump on IRC and ask if you run into any issues getting it going;
There's an issue with your fix: in may_delete_deleted_inode(), you're getting the subvol from inode.bi_subvol.
That's incorrect, because bi_subvol is only set on subvolume roots, because we can't go through every inode in the subvolume and change bi_subvol when taking a snapshot.
You're doing it right in bch2_empty_dir_trans(), which is the path that matters more; may_delete_deleted_inode() is only checking "did this filesystem get borked".
I would just pass in 0 from may_delete_deleted_inode(), and then have empty_dir_snapshot() ignore DT_SUBVOL dirents when subvol == 0. It makes the check less strict, but that's ok, the rest of fsck will still catch it.
Quoted 67 lines
--- fs/bcachefs/dirent.c | 7 +++++-- fs/bcachefs/dirent.h | 2 +- fs/bcachefs/inode.c | 2 +- 3 files changed, 7 insertions(+), 4 deletions(-)fs/bcachefs/dirent.c +5
-2@@ -508,7 +508,7 @@ u64 bch2_dirent_lookup(struct bch_fs *c, subvol_inum dir,return ret; } -int bch2_empty_dir_snapshot (struct btree_trans *trans,u64 dir,u32 snapshot) +int bch2_empty_dir_snapshot (struct btree_trans *trans,u64 dir,u32 subvol,u32 snapshot) {struct btree_iter iter;struct bkey_s_c k;@@ -518,6 +518,9 @@ int bch2_empty_dir_snapshot(struct btree_trans *trans, u64 dir, u32 snapshot) SPOS(dir, 0, snapshot), POS(dir, U64_MAX), 0, k, ret)if (k.k->type == KEY_TYPE_dirent) { +struct bkey_s_c_dirent d =bkey_s_c_to_dirent (k); +if (d.v->d_type ==DT_SUBVOL &&le32_to_cpu (d.v->d_parent_subvol) != subvol) +continue ; ret = -ENOTEMPTY ;break ; }@@ -531,7 +534,7 @@ int bch2_empty_dir_trans(struct btree_trans *trans, subvol_inum dir)u32 snapshot;return bch2_subvolume_get_snapshot (trans, dir.subvol, &snapshot) ?: -bch2_empty_dir_snapshot (trans, dir.inum, snapshot); +bch2_empty_dir_snapshot (trans, dir.inum, dir.subvol, snapshot); }int bch2_readdir (struct bch_fs *c,subvol_inum inum,struct dir_context *ctx)fs/bcachefs/dirent.h +1
-1@@ -69,7 +69,7 @@ u64 bch2_dirent_lookup(struct bch_fs *, subvol_inum, const struct bch_hash_info *, const struct qstr *, subvol_inum *); -int bch2_empty_dir_snapshot (struct btree_trans *,u64 ,u32 ); +int bch2_empty_dir_snapshot (struct btree_trans *,u64 ,u32 ,u32 );int bch2_empty_dir_trans (struct btree_trans *,subvol_inum );int bch2_readdir (struct bch_fs *,subvol_inum ,struct dir_context *);fs/bcachefs/inode.c +1
-1@@ -1088,7 +1088,7 @@ static int may_delete_deleted_inode(struct btree_trans *trans, goto out;if (S_ISDIR (inode.bi_mode)) { - ret =bch2_empty_dir_snapshot (trans, pos.offset, pos.snapshot); + ret =bch2_empty_dir_snapshot (trans, pos.offset, inode.bi_subvol, pos.snapshot);if (fsck_err_on (ret == -ENOTEMPTY , c, deleted_inode_is_dir,"non empty directory %llu:%u in deleted_inodes btree" , pos.offset, pos.snapshot))Signature · 1 line
2.43.0
Quoted 30 lines · Kent Overstreet
Quoted 14 lines · Guoyu Ou
When we are checking whether a subvolume is empty in the specified snapshot, entries that do not belong to this subvolume should be skipped.
This fixes the following case:
$ bcachefs subvolume create ./sub $ cd sub $ bcachefs subvolume create ./sub2 $ bcachefs subvolume snapshot . ./snap $ ls -a snap . .. $ rmdir snap rmdir: failed to remove 'snap': Directory not emptyQuoted 1 line
Signed-off-by Guoyu Ou <[email protected]>So that's an interesting corner case - can you add a ktest test for it?
https://evilpiepirate.org/git/ktest.git/If you're not using it yet, please jump on IRC and ask if you run into any issues getting it going;
There's an issue with your fix: in
may_delete_deleted_inode(), you're getting the subvol from inode.bi_subvol.That's incorrect, because
bi_subvolis only set on subvolume roots, because we can't go through every inode in the subvolume and changebi_subvolwhen taking a snapshot.
Quoted 8 lines
You're doing it right in
bch2_empty_dir_trans(), which is the path that matters more;may_delete_deleted_inode()is only checking "did this filesystem get borked".I would just pass in 0 from
may_delete_deleted_inode(), and then haveempty_dir_snapshot()ignoreDT_SUBVOLdirents when subvol == 0. It makes the check less strict, but that's ok, the rest of fsck will still catch it.
Got it. I will add a new test case and update this patch later.
Quoted 68 lines
Quoted 68 lines
--- fs/bcachefs/dirent.c | 7 +++++-- fs/bcachefs/dirent.h | 2 +- fs/bcachefs/inode.c | 2 +- 3 files changed, 7 insertions(+), 4 deletions(-)fs/bcachefs/dirent.c +5
-2@@ -508,7 +508,7 @@ u64 bch2_dirent_lookup(struct bch_fs *c, subvol_inum dir,return ret; } -int bch2_empty_dir_snapshot (struct btree_trans *trans,u64 dir,u32 snapshot) +int bch2_empty_dir_snapshot (struct btree_trans *trans,u64 dir,u32 subvol,u32 snapshot) {struct btree_iter iter;struct bkey_s_c k;@@ -518,6 +518,9 @@ int bch2_empty_dir_snapshot(struct btree_trans *trans, u64 dir, u32 snapshot) SPOS(dir, 0, snapshot), POS(dir, U64_MAX), 0, k, ret)if (k.k->type == KEY_TYPE_dirent) { +struct bkey_s_c_dirent d =bkey_s_c_to_dirent (k); +if (d.v->d_type ==DT_SUBVOL &&le32_to_cpu (d.v->d_parent_subvol) != subvol) +continue ; ret = -ENOTEMPTY ;break ; }@@ -531,7 +534,7 @@ int bch2_empty_dir_trans(struct btree_trans *trans, subvol_inum dir)u32 snapshot;return bch2_subvolume_get_snapshot (trans, dir.subvol, &snapshot) ?: -bch2_empty_dir_snapshot (trans, dir.inum, snapshot); +bch2_empty_dir_snapshot (trans, dir.inum, dir.subvol, snapshot); }int bch2_readdir (struct bch_fs *c,subvol_inum inum,struct dir_context *ctx)fs/bcachefs/dirent.h +1
-1@@ -69,7 +69,7 @@ u64 bch2_dirent_lookup(struct bch_fs *, subvol_inum, const struct bch_hash_info *, const struct qstr *, subvol_inum *); -int bch2_empty_dir_snapshot (struct btree_trans *,u64 ,u32 ); +int bch2_empty_dir_snapshot (struct btree_trans *,u64 ,u32 ,u32 );int bch2_empty_dir_trans (struct btree_trans *,subvol_inum );int bch2_readdir (struct bch_fs *,subvol_inum ,struct dir_context *);fs/bcachefs/inode.c +1
-2@@ -1088,7 +1088,7 @@ static int may_delete_deleted_inode(struct btree_trans *trans, goto out;if (S_ISDIR (inode.bi_mode)) { - ret =bch2_empty_dir_snapshot (trans, pos.offset, pos.snapshot); + ret =bch2_empty_dir_snapshot (trans, pos.offset, inode.bi_subvol, pos.snapshot);if (fsck_err_on (ret == -ENOTEMPTY , c, deleted_inode_is_dir,"non empty directory %llu:%u in deleted_inodes btree" , pos.offset, pos.snapshot)) --2.43.0