diff options
| author | Qu Wenruo <quwenruo@cn.fujitsu.com> | 2015-01-19 02:42:41 -0500 |
|---|---|---|
| committer | Chris Mason <clm@fb.com> | 2015-01-20 20:20:21 -0500 |
| commit | a53f4f8e9c8ebe6c9ee3b34c368913aae9876e22 (patch) | |
| tree | dfe7b1c873db8cee0c10d3bbdd339b88f941ce88 /fs | |
| parent | 6c9fe14f9d64cc12401a825a60ec5c5723496ca4 (diff) | |
btrfs: Don't call btrfs_start_transaction() on frozen fs to avoid deadlock.
Commit 6b5fe46dfa52 (btrfs: do commit in sync_fs if there are pending
changes) will call btrfs_start_transaction() in sync_fs(), to handle
some operations needed to be done in next transaction.
However this can cause deadlock if the filesystem is frozen, with the
following sys_r+w output:
[ 143.255932] Call Trace:
[ 143.255936] [<ffffffff816c0e09>] schedule+0x29/0x70
[ 143.255939] [<ffffffff811cb7f3>] __sb_start_write+0xb3/0x100
[ 143.255971] [<ffffffffa040ec06>] start_transaction+0x2e6/0x5a0
[btrfs]
[ 143.255992] [<ffffffffa040f1eb>] btrfs_start_transaction+0x1b/0x20
[btrfs]
[ 143.256003] [<ffffffffa03dc0ba>] btrfs_sync_fs+0xca/0xd0 [btrfs]
[ 143.256007] [<ffffffff811f7be0>] sync_fs_one_sb+0x20/0x30
[ 143.256011] [<ffffffff811cbd01>] iterate_supers+0xe1/0xf0
[ 143.256014] [<ffffffff811f7d75>] sys_sync+0x55/0x90
[ 143.256017] [<ffffffff816c49d2>] system_call_fastpath+0x12/0x17
[ 143.256111] Call Trace:
[ 143.256114] [<ffffffff816c0e09>] schedule+0x29/0x70
[ 143.256119] [<ffffffff816c3405>] rwsem_down_write_failed+0x1c5/0x2d0
[ 143.256123] [<ffffffff8133f013>] call_rwsem_down_write_failed+0x13/0x20
[ 143.256131] [<ffffffff811caae8>] thaw_super+0x28/0xc0
[ 143.256135] [<ffffffff811db3e5>] do_vfs_ioctl+0x3f5/0x540
[ 143.256187] [<ffffffff811db5c1>] SyS_ioctl+0x91/0xb0
[ 143.256213] [<ffffffff816c49d2>] system_call_fastpath+0x12/0x17
The reason is like the following:
(Holding s_umount)
VFS sync_fs staff:
|- btrfs_sync_fs()
|- btrfs_start_transaction()
|- sb_start_intwrite()
(Waiting thaw_fs to unfreeze)
VFS thaw_fs staff:
thaw_fs()
(Waiting sync_fs to release
s_umount)
So deadlock happens.
This can be easily triggered by fstest/generic/068 with inode_cache
mount option.
The fix is to check if the fs is frozen, if the fs is frozen, just
return and waiting for the next transaction.
Cc: David Sterba <dsterba@suse.cz>
Reported-by: Gui Hecheng <guihc.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
[enhanced comment, changed to SB_FREEZE_WRITE]
Signed-off-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Chris Mason <clm@fb.com>
Diffstat (limited to 'fs')
| -rw-r--r-- | fs/btrfs/super.c | 10 |
1 files changed, 10 insertions, 0 deletions
diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index c86fb5438454..6f49b2872a64 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c | |||
| @@ -1000,6 +1000,16 @@ int btrfs_sync_fs(struct super_block *sb, int wait) | |||
| 1000 | */ | 1000 | */ |
| 1001 | if (fs_info->pending_changes == 0) | 1001 | if (fs_info->pending_changes == 0) |
| 1002 | return 0; | 1002 | return 0; |
| 1003 | /* | ||
| 1004 | * A non-blocking test if the fs is frozen. We must not | ||
| 1005 | * start a new transaction here otherwise a deadlock | ||
| 1006 | * happens. The pending operations are delayed to the | ||
| 1007 | * next commit after thawing. | ||
| 1008 | */ | ||
| 1009 | if (__sb_start_write(sb, SB_FREEZE_WRITE, false)) | ||
| 1010 | __sb_end_write(sb, SB_FREEZE_WRITE); | ||
| 1011 | else | ||
| 1012 | return 0; | ||
| 1003 | trans = btrfs_start_transaction(root, 0); | 1013 | trans = btrfs_start_transaction(root, 0); |
| 1004 | } | 1014 | } |
| 1005 | if (IS_ERR(trans)) | 1015 | if (IS_ERR(trans)) |