2020-06-12 01:05:28

by Zhiqiang Liu

[permalink] [raw]
Subject: [PATCH V3] bcache: fix potential deadlock problem in btree_gc_coalesce


coccicheck reports:
drivers/md//bcache/btree.c:1538:1-7: preceding lock on line 1417

In btree_gc_coalesce func, if the coalescing process fails, we will goto
to out_nocoalesce tag directly without releasing new_nodes[i]->write_lock.
Then, it will cause a deadlock when trying to acquire new_nodes[i]->write_lock
for freeing new_nodes[i] before return.

btree_gc_coalesce func details as follows:
if alloc new_nodes[i] fails:
goto out_nocoalesce;
mutex_lock(&new_nodes[i]->write_lock) // obtain new_nodes[i]->write_lock
for (i = nodes - 1; i > 0; --i) // main coalescing process
……
if coalescing process fails:
goto out_nocoalesce; // Here, directly goto out_nocoalesce
// tag will cause a deadlock
……
mutex_unlock(&new_nodes[i]->write_lock) // release new_nodes[i]->write_lock
return; // coalesing succ, return
out_nocoalesce:
btree_node_free(new_nodes[i]) // free new_nodes[i]
mutex_lock(&new_nodes[i]->write_lock); // obtain new_nodes[i]->write_lock
clear_bit(BTREE_NODE_dirty, &ew_nodes[i]->flags); // set flag for reuse
mutex_unlock(&new_nodes[i]->write_lock); // release new_nodes[i]->write_lock

To fix the problem, we add a new tag 'out_unlock_nocoalesce' for releasing
new_nodes[i]->write_lock before out_nocoalesce tag. If coalescing process fails,
we will go to out_unlock_nocoalesce tag for releasing new_nodes[i]->write_lock
before free new_nodes[i] in out_nocoalesce tag.

Fixes: 2a285686c109816 ("bcache: btree locking rework")
Signed-off-by: Zhiqiang Liu <[email protected]>
---
Changelog:
V3: improve commit log again (suggested by Markus and Coly)
V2: improve commit log
V1: initial version

drivers/md/bcache/btree.c | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/drivers/md/bcache/btree.c b/drivers/md/bcache/btree.c
index 72856e5f23a3..fd1f288fd801 100644
--- a/drivers/md/bcache/btree.c
+++ b/drivers/md/bcache/btree.c
@@ -1389,7 +1389,7 @@ static int btree_gc_coalesce(struct btree *b, struct btree_op *op,
if (__set_blocks(n1, n1->keys + n2->keys,
block_bytes(b->c)) >
btree_blocks(new_nodes[i]))
- goto out_nocoalesce;
+ goto out_unlock_nocoalesce;

keys = n2->keys;
/* Take the key of the node we're getting rid of */
@@ -1418,7 +1418,7 @@ static int btree_gc_coalesce(struct btree *b, struct btree_op *op,

if (__bch_keylist_realloc(&keylist,
bkey_u64s(&new_nodes[i]->key)))
- goto out_nocoalesce;
+ goto out_unlock_nocoalesce;

bch_btree_node_write(new_nodes[i], &cl);
bch_keylist_add(&keylist, &new_nodes[i]->key);
@@ -1464,6 +1464,10 @@ static int btree_gc_coalesce(struct btree *b, struct btree_op *op,
/* Invalidated our iterator */
return -EINTR;

+out_unlock_nocoalesce:
+ for (i = 0; i < nodes; i++)
+ mutex_unlock(&new_nodes[i]->write_lock);
+
out_nocoalesce:
closure_sync(&cl);

--
2.19.1