diff options
| author | Miao Xie <miaox@cn.fujitsu.com> | 2013-01-15 07:29:12 +0100 |
|---|---|---|
| committer | Josef Bacik <jbacik@fusionio.com> | 2013-01-24 18:51:25 +0100 |
| commit | 2cba30f172afdfa00f3e844f42f21eb3b972d01c (patch) | |
| tree | 8030e3d762a432f92a56151736b7f149c22454bc /fs | |
| parent | Btrfs: Add ACCESS_ONCE() to transaction->abort accesses (diff) | |
| download | linux-2cba30f172afdfa00f3e844f42f21eb3b972d01c.tar.xz linux-2cba30f172afdfa00f3e844f42f21eb3b972d01c.zip | |
Btrfs: fix missed transaction->aborted check
First, though the current transaction->aborted check can stop the commit early
and avoid unnecessary operations, it is too early, and some transaction handles
don't end, those handles may set transaction->aborted after the check.
Second, when we commit the transaction, we will wake up some worker threads to
flush the space cache and inode cache. Those threads also allocate some transaction
handles and may set transaction->aborted if some serious error happens.
So we need more check for ->aborted when committing the transaction. Fix it.
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Diffstat (limited to 'fs')
| -rw-r--r-- | fs/btrfs/transaction.c | 16 |
1 files changed, 16 insertions, 0 deletions
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 0ef29611fade..f15494699f3b 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -1575,6 +1575,11 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans, wait_event(cur_trans->writer_wait, atomic_read(&cur_trans->num_writers) == 1); + /* ->aborted might be set after the previous check, so check it */ + if (unlikely(ACCESS_ONCE(cur_trans->aborted))) { + ret = cur_trans->aborted; + goto cleanup_transaction; + } /* * the reloc mutex makes sure that we stop * the balancing code from coming in and moving @@ -1658,6 +1663,17 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans, goto cleanup_transaction; } + /* + * The tasks which save the space cache and inode cache may also + * update ->aborted, check it. + */ + if (unlikely(ACCESS_ONCE(cur_trans->aborted))) { + ret = cur_trans->aborted; + mutex_unlock(&root->fs_info->tree_log_mutex); + mutex_unlock(&root->fs_info->reloc_mutex); + goto cleanup_transaction; + } + btrfs_prepare_extent_commit(trans, root); cur_trans = root->fs_info->running_transaction; |