From 68aaa637162787dc3374080efe03366f70b344f1 Mon Sep 17 00:00:00 2001 From: Kent Overstreet Date: Thu, 20 Feb 2025 09:17:17 -0500 Subject: [PATCH 1/7] bcachefs: print op->nonce on data update inconsistency "nonce inconstancy" is popping up again, causing us to go emergency read-only. This one looks less serious, i.e. specific to the encryption path and not indicative of a data corruption bug. But we'll need more info to track it down. Signed-off-by: Kent Overstreet --- fs/bcachefs/data_update.c | 1 + 1 file changed, 1 insertion(+) diff --git a/fs/bcachefs/data_update.c b/fs/bcachefs/data_update.c index 337494facac6..642fbc60ecab 100644 --- a/fs/bcachefs/data_update.c +++ b/fs/bcachefs/data_update.c @@ -340,6 +340,7 @@ restart_drop_extra_replicas: struct printbuf buf = PRINTBUF; prt_str(&buf, "about to insert invalid key in data update path"); + prt_printf(&buf, "\nop.nonce: %u", m->op.nonce); prt_str(&buf, "\nold: "); bch2_bkey_val_to_text(&buf, c, old); prt_str(&buf, "\nk: "); From c522093b02835f2e897b83e9764e7919edac5d08 Mon Sep 17 00:00:00 2001 From: Alan Huang Date: Thu, 20 Feb 2025 18:56:08 +0800 Subject: [PATCH 2/7] bcachefs: Fix memmove when move keys down The fix alone doesn't fix [1], but should be applied before debugging that. [1] https://syzkaller.appspot.com/bug?extid=38a0cbd267eff2d286ff Signed-off-by: Alan Huang Signed-off-by: Kent Overstreet --- fs/bcachefs/btree_io.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/bcachefs/btree_io.c b/fs/bcachefs/btree_io.c index e371e60e3133..dece27d9db04 100644 --- a/fs/bcachefs/btree_io.c +++ b/fs/bcachefs/btree_io.c @@ -996,7 +996,7 @@ drop_this_key: } got_good_key: le16_add_cpu(&i->u64s, -next_good_key); - memmove_u64s_down(k, bkey_p_next(k), (u64 *) vstruct_end(i) - (u64 *) k); + memmove_u64s_down(k, (u64 *) k + next_good_key, (u64 *) vstruct_end(i) - (u64 *) k); set_btree_node_need_rewrite(b); } fsck_err: From cf3e6960263a2ecdf5528056b321e41557e9b03d Mon Sep 17 00:00:00 2001 From: Kent Overstreet Date: Sat, 22 Feb 2025 09:05:36 -0500 Subject: [PATCH 3/7] bcachefs: fix bch2_extent_ptr_eq() Reviewed-by: Thorsten Blum Signed-off-by: Kent Overstreet --- fs/bcachefs/extents.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/bcachefs/extents.h b/fs/bcachefs/extents.h index 620b284aa34f..204d765dd74c 100644 --- a/fs/bcachefs/extents.h +++ b/fs/bcachefs/extents.h @@ -704,7 +704,7 @@ static inline bool bch2_extent_ptr_eq(struct bch_extent_ptr ptr1, ptr1.unwritten == ptr2.unwritten && ptr1.offset == ptr2.offset && ptr1.dev == ptr2.dev && - ptr1.dev == ptr2.dev); + ptr1.gen == ptr2.gen); } void bch2_ptr_swab(struct bkey_s); From 4804f3ac2649475509b1836a4d252c04de143249 Mon Sep 17 00:00:00 2001 From: Kent Overstreet Date: Wed, 26 Feb 2025 19:30:38 -0500 Subject: [PATCH 4/7] bcachefs: Revert directory i_size This turned out to have several bugs, which were missed because the fsck code wasn't properly reporting errors - whoops. Kicking it out for now, hopefully it can make 6.15. Cc: Hongbo Li Signed-off-by: Kent Overstreet --- fs/bcachefs/dirent.h | 5 ----- fs/bcachefs/fs-common.c | 11 ----------- fs/bcachefs/fsck.c | 21 --------------------- fs/bcachefs/sb-downgrade.c | 5 +---- 4 files changed, 1 insertion(+), 41 deletions(-) diff --git a/fs/bcachefs/dirent.h b/fs/bcachefs/dirent.h index a633f83c1ac7..362b3b2f2f2e 100644 --- a/fs/bcachefs/dirent.h +++ b/fs/bcachefs/dirent.h @@ -31,11 +31,6 @@ static inline unsigned dirent_val_u64s(unsigned len) sizeof(u64)); } -static inline unsigned int dirent_occupied_size(const struct qstr *name) -{ - return (BKEY_U64s + dirent_val_u64s(name->len)) * sizeof(u64); -} - int bch2_dirent_read_target(struct btree_trans *, subvol_inum, struct bkey_s_c_dirent, subvol_inum *); diff --git a/fs/bcachefs/fs-common.c b/fs/bcachefs/fs-common.c index d70d9f634cea..2c3d46ac70c6 100644 --- a/fs/bcachefs/fs-common.c +++ b/fs/bcachefs/fs-common.c @@ -152,7 +152,6 @@ int bch2_create_trans(struct btree_trans *trans, if (is_subdir_for_nlink(new_inode)) dir_u->bi_nlink++; dir_u->bi_mtime = dir_u->bi_ctime = now; - dir_u->bi_size += dirent_occupied_size(name); ret = bch2_inode_write(trans, &dir_iter, dir_u); if (ret) @@ -221,7 +220,6 @@ int bch2_link_trans(struct btree_trans *trans, } dir_u->bi_mtime = dir_u->bi_ctime = now; - dir_u->bi_size += dirent_occupied_size(name); dir_hash = bch2_hash_info_init(c, dir_u); @@ -324,7 +322,6 @@ int bch2_unlink_trans(struct btree_trans *trans, dir_u->bi_mtime = dir_u->bi_ctime = inode_u->bi_ctime = now; dir_u->bi_nlink -= is_subdir_for_nlink(inode_u); - dir_u->bi_size -= dirent_occupied_size(name); ret = bch2_hash_delete_at(trans, bch2_dirent_hash_desc, &dir_hash, &dirent_iter, @@ -463,14 +460,6 @@ int bch2_rename_trans(struct btree_trans *trans, goto err; } - if (mode == BCH_RENAME) { - src_dir_u->bi_size -= dirent_occupied_size(src_name); - dst_dir_u->bi_size += dirent_occupied_size(dst_name); - } - - if (mode == BCH_RENAME_OVERWRITE) - src_dir_u->bi_size -= dirent_occupied_size(src_name); - if (src_inode_u->bi_parent_subvol) src_inode_u->bi_parent_subvol = dst_dir.subvol; diff --git a/fs/bcachefs/fsck.c b/fs/bcachefs/fsck.c index 9bf316e7b845..0e85131d0af8 100644 --- a/fs/bcachefs/fsck.c +++ b/fs/bcachefs/fsck.c @@ -1978,31 +1978,10 @@ fsck_err: return ret; } -static int check_dir_i_size_notnested(struct btree_trans *trans, struct inode_walker *w) -{ - struct bch_fs *c = trans->c; - int ret = 0; - - darray_for_each(w->inodes, i) - if (fsck_err_on(i->inode.bi_size != i->i_size, - trans, inode_dir_wrong_nlink, - "directory %llu:%u with wrong i_size: got %llu, should be %llu", - w->last_pos.inode, i->snapshot, i->inode.bi_size, i->i_size)) { - i->inode.bi_size = i->i_size; - ret = bch2_fsck_write_inode(trans, &i->inode); - if (ret) - break; - } -fsck_err: - bch_err_fn(c, ret); - return ret; -} - static int check_subdir_dirents_count(struct btree_trans *trans, struct inode_walker *w) { u32 restart_count = trans->restart_count; return check_subdir_count_notnested(trans, w) ?: - check_dir_i_size_notnested(trans, w) ?: trans_was_restarted(trans, restart_count); } diff --git a/fs/bcachefs/sb-downgrade.c b/fs/bcachefs/sb-downgrade.c index 35e07bc8fbd3..051214fdc735 100644 --- a/fs/bcachefs/sb-downgrade.c +++ b/fs/bcachefs/sb-downgrade.c @@ -90,10 +90,7 @@ BIT_ULL(BCH_RECOVERY_PASS_check_allocations), \ BCH_FSCK_ERR_accounting_mismatch, \ BCH_FSCK_ERR_accounting_key_replicas_nr_devs_0, \ - BCH_FSCK_ERR_accounting_key_junk_at_end) \ - x(directory_size, \ - BIT_ULL(BCH_RECOVERY_PASS_check_dirents), \ - BCH_FSCK_ERR_directory_size_mismatch) \ + BCH_FSCK_ERR_accounting_key_junk_at_end) #define DOWNGRADE_TABLE() \ x(bucket_stripe_sectors, \ From 7909d1fb90e290ffd7b8570f4e2f97fe2fb381d0 Mon Sep 17 00:00:00 2001 From: Kent Overstreet Date: Tue, 25 Feb 2025 22:35:28 -0500 Subject: [PATCH 5/7] bcachefs: Check for -BCH_ERR_open_buckets_empty in journal resize This fixes occasional failures from journal resize. Signed-off-by: Kent Overstreet --- fs/bcachefs/journal.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/fs/bcachefs/journal.c b/fs/bcachefs/journal.c index 24c294d4634e..5dabbf3c0965 100644 --- a/fs/bcachefs/journal.c +++ b/fs/bcachefs/journal.c @@ -1194,7 +1194,9 @@ int bch2_set_nr_journal_buckets(struct bch_fs *c, struct bch_dev *ca, closure_sync(&cl); - if (ret && ret != -BCH_ERR_bucket_alloc_blocked) + if (ret && + ret != -BCH_ERR_bucket_alloc_blocked && + ret != -BCH_ERR_open_buckets_empty) break; } From 677bdb7346b6fd806ea45b11cbfe36de0b0cd644 Mon Sep 17 00:00:00 2001 From: Alan Huang Date: Wed, 26 Feb 2025 17:33:22 +0800 Subject: [PATCH 6/7] bcachefs: Fix deadlock This fixes two deadlocks: 1.pcpu_alloc_mutex involved one as pointed by syzbot[1] 2.recursion deadlock. The root cause is that we hold the bc lock during alloc_percpu, fix it by following the pattern used by __btree_node_mem_alloc(). [1] https://lore.kernel.org/all/66f97d9a.050a0220.6bad9.001d.GAE@google.com/T/ Reported-by: syzbot+fe63f377148a6371a9db@syzkaller.appspotmail.com Tested-by: syzbot+fe63f377148a6371a9db@syzkaller.appspotmail.com Signed-off-by: Alan Huang Signed-off-by: Kent Overstreet --- fs/bcachefs/btree_cache.c | 9 +++++---- fs/bcachefs/btree_key_cache.c | 2 +- fs/bcachefs/btree_locking.c | 5 +++-- fs/bcachefs/btree_locking.h | 2 +- fs/bcachefs/six.c | 5 +++-- fs/bcachefs/six.h | 7 ++++--- 6 files changed, 17 insertions(+), 13 deletions(-) diff --git a/fs/bcachefs/btree_cache.c b/fs/bcachefs/btree_cache.c index ca755e8d1a37..1ec1f90e0eb3 100644 --- a/fs/bcachefs/btree_cache.c +++ b/fs/bcachefs/btree_cache.c @@ -203,7 +203,7 @@ struct btree *__bch2_btree_node_mem_alloc(struct bch_fs *c) return NULL; } - bch2_btree_lock_init(&b->c, 0); + bch2_btree_lock_init(&b->c, 0, GFP_KERNEL); __bch2_btree_node_to_freelist(bc, b); return b; @@ -795,17 +795,18 @@ struct btree *bch2_btree_node_mem_alloc(struct btree_trans *trans, bool pcpu_rea } b = __btree_node_mem_alloc(c, GFP_NOWAIT|__GFP_NOWARN); - if (!b) { + if (b) { + bch2_btree_lock_init(&b->c, pcpu_read_locks ? SIX_LOCK_INIT_PCPU : 0, GFP_NOWAIT); + } else { mutex_unlock(&bc->lock); bch2_trans_unlock(trans); b = __btree_node_mem_alloc(c, GFP_KERNEL); if (!b) goto err; + bch2_btree_lock_init(&b->c, pcpu_read_locks ? SIX_LOCK_INIT_PCPU : 0, GFP_KERNEL); mutex_lock(&bc->lock); } - bch2_btree_lock_init(&b->c, pcpu_read_locks ? SIX_LOCK_INIT_PCPU : 0); - BUG_ON(!six_trylock_intent(&b->c.lock)); BUG_ON(!six_trylock_write(&b->c.lock)); diff --git a/fs/bcachefs/btree_key_cache.c b/fs/bcachefs/btree_key_cache.c index 1821f40c161a..edce59433375 100644 --- a/fs/bcachefs/btree_key_cache.c +++ b/fs/bcachefs/btree_key_cache.c @@ -156,7 +156,7 @@ bkey_cached_alloc(struct btree_trans *trans, struct btree_path *path, unsigned k } if (ck) { - bch2_btree_lock_init(&ck->c, pcpu_readers ? SIX_LOCK_INIT_PCPU : 0); + bch2_btree_lock_init(&ck->c, pcpu_readers ? SIX_LOCK_INIT_PCPU : 0, GFP_KERNEL); ck->c.cached = true; goto lock; } diff --git a/fs/bcachefs/btree_locking.c b/fs/bcachefs/btree_locking.c index 10b805a60f52..caef65adeae4 100644 --- a/fs/bcachefs/btree_locking.c +++ b/fs/bcachefs/btree_locking.c @@ -7,9 +7,10 @@ static struct lock_class_key bch2_btree_node_lock_key; void bch2_btree_lock_init(struct btree_bkey_cached_common *b, - enum six_lock_init_flags flags) + enum six_lock_init_flags flags, + gfp_t gfp) { - __six_lock_init(&b->lock, "b->c.lock", &bch2_btree_node_lock_key, flags); + __six_lock_init(&b->lock, "b->c.lock", &bch2_btree_node_lock_key, flags, gfp); lockdep_set_notrack_class(&b->lock); } diff --git a/fs/bcachefs/btree_locking.h b/fs/bcachefs/btree_locking.h index b54ef48eb8cc..b33ab7af8440 100644 --- a/fs/bcachefs/btree_locking.h +++ b/fs/bcachefs/btree_locking.h @@ -13,7 +13,7 @@ #include "btree_iter.h" #include "six.h" -void bch2_btree_lock_init(struct btree_bkey_cached_common *, enum six_lock_init_flags); +void bch2_btree_lock_init(struct btree_bkey_cached_common *, enum six_lock_init_flags, gfp_t gfp); void bch2_trans_unlock_noassert(struct btree_trans *); void bch2_trans_unlock_write(struct btree_trans *); diff --git a/fs/bcachefs/six.c b/fs/bcachefs/six.c index 7e7c66a1e1a6..7c403427fbdb 100644 --- a/fs/bcachefs/six.c +++ b/fs/bcachefs/six.c @@ -850,7 +850,8 @@ void six_lock_exit(struct six_lock *lock) EXPORT_SYMBOL_GPL(six_lock_exit); void __six_lock_init(struct six_lock *lock, const char *name, - struct lock_class_key *key, enum six_lock_init_flags flags) + struct lock_class_key *key, enum six_lock_init_flags flags, + gfp_t gfp) { atomic_set(&lock->state, 0); raw_spin_lock_init(&lock->wait_lock); @@ -873,7 +874,7 @@ void __six_lock_init(struct six_lock *lock, const char *name, * failure if they wish by checking lock->readers, but generally * will not want to treat it as an error. */ - lock->readers = alloc_percpu(unsigned); + lock->readers = alloc_percpu_gfp(unsigned, gfp); } #endif } diff --git a/fs/bcachefs/six.h b/fs/bcachefs/six.h index c142e06b7a3a..59b851cf8bac 100644 --- a/fs/bcachefs/six.h +++ b/fs/bcachefs/six.h @@ -164,18 +164,19 @@ enum six_lock_init_flags { }; void __six_lock_init(struct six_lock *lock, const char *name, - struct lock_class_key *key, enum six_lock_init_flags flags); + struct lock_class_key *key, enum six_lock_init_flags flags, + gfp_t gfp); /** * six_lock_init - initialize a six lock * @lock: lock to initialize * @flags: optional flags, i.e. SIX_LOCK_INIT_PCPU */ -#define six_lock_init(lock, flags) \ +#define six_lock_init(lock, flags, gfp) \ do { \ static struct lock_class_key __key; \ \ - __six_lock_init((lock), #lock, &__key, flags); \ + __six_lock_init((lock), #lock, &__key, flags, gfp); \ } while (0) /** From eb54d2695b57426638fed0ec066ae17a18c4426c Mon Sep 17 00:00:00 2001 From: Kent Overstreet Date: Wed, 26 Feb 2025 10:57:26 -0500 Subject: [PATCH 7/7] bcachefs: Fix truncate sometimes failing and returning 1 __bch_truncate_folio() may return 1 to indicate dirtyness of the folio being truncated, needed for fpunch to get the i_size writes correct. But truncate was forgetting to clear ret, and sometimes returning it as an error. Signed-off-by: Kent Overstreet --- fs/bcachefs/fs-io.c | 1 + 1 file changed, 1 insertion(+) diff --git a/fs/bcachefs/fs-io.c b/fs/bcachefs/fs-io.c index 94bf34b9b65f..717e7b94c66f 100644 --- a/fs/bcachefs/fs-io.c +++ b/fs/bcachefs/fs-io.c @@ -466,6 +466,7 @@ int bchfs_truncate(struct mnt_idmap *idmap, ret = bch2_truncate_folio(inode, iattr->ia_size); if (unlikely(ret < 0)) goto err; + ret = 0; truncate_setsize(&inode->v, iattr->ia_size);