diff options
author | Nikolay Borisov <nborisov@suse.com> | 2018-04-18 09:41:54 +0300 |
---|---|---|
committer | David Sterba <dsterba@suse.com> | 2018-04-20 19:17:25 +0200 |
commit | 5e388e95815408c27f3612190d089afc0774b870 (patch) | |
tree | ad9882fb36439900d634df34ea76796957732c3d /fs/btrfs/delayed-ref.c | |
parent | 92d32170847bfff2dd08af2c016085779f2fd2a1 (diff) | |
download | linux-rpi-5e388e95815408c27f3612190d089afc0774b870.tar.gz linux-rpi-5e388e95815408c27f3612190d089afc0774b870.tar.bz2 linux-rpi-5e388e95815408c27f3612190d089afc0774b870.zip |
btrfs: Fix race condition between delayed refs and blockgroup removal
When the delayed refs for a head are all run, eventually
cleanup_ref_head is called which (in case of deletion) obtains a
reference for the relevant btrfs_space_info struct by querying the bg
for the range. This is problematic because when the last extent of a
bg is deleted a race window emerges between removal of that bg and the
subsequent invocation of cleanup_ref_head. This can result in cache being null
and either a null pointer dereference or assertion failure.
task: ffff8d04d31ed080 task.stack: ffff9e5dc10cc000
RIP: 0010:assfail.constprop.78+0x18/0x1a [btrfs]
RSP: 0018:ffff9e5dc10cfbe8 EFLAGS: 00010292
RAX: 0000000000000044 RBX: 0000000000000000 RCX: 0000000000000000
RDX: ffff8d04ffc1f868 RSI: ffff8d04ffc178c8 RDI: ffff8d04ffc178c8
RBP: ffff8d04d29e5ea0 R08: 00000000000001f0 R09: 0000000000000001
R10: ffff9e5dc0507d58 R11: 0000000000000001 R12: ffff8d04d29e5ea0
R13: ffff8d04d29e5f08 R14: ffff8d04efe29b40 R15: ffff8d04efe203e0
FS: 00007fbf58ead500(0000) GS:ffff8d04ffc00000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007fe6c6975648 CR3: 0000000013b2a000 CR4: 00000000000006f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
__btrfs_run_delayed_refs+0x10e7/0x12c0 [btrfs]
btrfs_run_delayed_refs+0x68/0x250 [btrfs]
btrfs_should_end_transaction+0x42/0x60 [btrfs]
btrfs_truncate_inode_items+0xaac/0xfc0 [btrfs]
btrfs_evict_inode+0x4c6/0x5c0 [btrfs]
evict+0xc6/0x190
do_unlinkat+0x19c/0x300
do_syscall_64+0x74/0x140
entry_SYSCALL_64_after_hwframe+0x3d/0xa2
RIP: 0033:0x7fbf589c57a7
To fix this, introduce a new flag "is_system" to head_ref structs,
which is populated at insertion time. This allows to decouple the
querying for the spaceinfo from querying the possibly deleted bg.
Fixes: d7eae3403f46 ("Btrfs: rework delayed ref total_bytes_pinned accounting")
CC: stable@vger.kernel.org # 4.14+
Suggested-by: Omar Sandoval <osandov@osandov.com>
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Diffstat (limited to 'fs/btrfs/delayed-ref.c')
-rw-r--r-- | fs/btrfs/delayed-ref.c | 19 |
1 files changed, 14 insertions, 5 deletions
diff --git a/fs/btrfs/delayed-ref.c b/fs/btrfs/delayed-ref.c index 9e98295de7ce..e1b0651686f7 100644 --- a/fs/btrfs/delayed-ref.c +++ b/fs/btrfs/delayed-ref.c @@ -540,8 +540,10 @@ add_delayed_ref_head(struct btrfs_fs_info *fs_info, struct btrfs_delayed_ref_head *head_ref, struct btrfs_qgroup_extent_record *qrecord, u64 bytenr, u64 num_bytes, u64 ref_root, u64 reserved, - int action, int is_data, int *qrecord_inserted_ret, + int action, int is_data, int is_system, + int *qrecord_inserted_ret, int *old_ref_mod, int *new_ref_mod) + { struct btrfs_delayed_ref_head *existing; struct btrfs_delayed_ref_root *delayed_refs; @@ -585,6 +587,7 @@ add_delayed_ref_head(struct btrfs_fs_info *fs_info, head_ref->ref_mod = count_mod; head_ref->must_insert_reserved = must_insert_reserved; head_ref->is_data = is_data; + head_ref->is_system = is_system; head_ref->ref_tree = RB_ROOT; INIT_LIST_HEAD(&head_ref->ref_add_list); RB_CLEAR_NODE(&head_ref->href_node); @@ -772,6 +775,7 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info, struct btrfs_delayed_ref_root *delayed_refs; struct btrfs_qgroup_extent_record *record = NULL; int qrecord_inserted; + int is_system = (ref_root == BTRFS_CHUNK_TREE_OBJECTID); BUG_ON(extent_op && extent_op->is_data); ref = kmem_cache_alloc(btrfs_delayed_tree_ref_cachep, GFP_NOFS); @@ -800,8 +804,8 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info, */ head_ref = add_delayed_ref_head(fs_info, trans, head_ref, record, bytenr, num_bytes, 0, 0, action, 0, - &qrecord_inserted, old_ref_mod, - new_ref_mod); + is_system, &qrecord_inserted, + old_ref_mod, new_ref_mod); add_delayed_tree_ref(fs_info, trans, head_ref, &ref->node, bytenr, num_bytes, parent, ref_root, level, action); @@ -868,7 +872,7 @@ int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info, */ head_ref = add_delayed_ref_head(fs_info, trans, head_ref, record, bytenr, num_bytes, ref_root, reserved, - action, 1, &qrecord_inserted, + action, 1, 0, &qrecord_inserted, old_ref_mod, new_ref_mod); add_delayed_data_ref(fs_info, trans, head_ref, &ref->node, bytenr, @@ -898,9 +902,14 @@ int btrfs_add_delayed_extent_op(struct btrfs_fs_info *fs_info, delayed_refs = &trans->transaction->delayed_refs; spin_lock(&delayed_refs->lock); + /* + * extent_ops just modify the flags of an extent and they don't result + * in ref count changes, hence it's safe to pass false/0 for is_system + * argument + */ add_delayed_ref_head(fs_info, trans, head_ref, NULL, bytenr, num_bytes, 0, 0, BTRFS_UPDATE_DELAYED_HEAD, - extent_op->is_data, NULL, NULL, NULL); + extent_op->is_data, 0, NULL, NULL, NULL); spin_unlock(&delayed_refs->lock); return 0; |