btrfs: fix BUG_ON condition in btrfs_cancel_balance
PATCH 1/1 btrfs: fix BUG_ON condition in btrfs_cancel_balance
Pausing and canceling balance can race to intterupt balance lead to BUG_ON panic in btrfs_cancel_balance. The BUG_ON condition in btrfs_cancel_balance does not take this race scenario into account.
However, the race condition has no other side effects. We can fix that.
Reproducing it with panic trace like this:
kernel BUG at fs/btrfs/volumes.c:4618!
RIP: 0010:btrfs_cancel_balance+0x5cf/0x6a0 Call Trace: <TASK> ? do_nanosleep+0x60/0x120 ? hrtimer_nanosleep+0xb7/0x1a0 ? sched_core_clone_cookie+0x70/0x70 btrfs_ioctl_balance_ctl+0x55/0x70 btrfs_ioctl+0xa46/0xd20 __x64_sys_ioctl+0x7d/0xa0 do_syscall_64+0x38/0x80 entry_SYSCALL_64_after_hwframe+0x63/0xcd
Race scenario as follows:
Quoted 11 lines
mutex_unlock(&fs_info->balance_mutex);
--------------------
.......issue pause and cancel req in another thread
--------------------
ret =__btrfs_balance(fs_info);
mutex_lock(&fs_info->balance_mutex);if (ret == -ECANCELED && atomic_read(&fs_info->balance_pause_req)) { btrfs_info(fs_info, "balance: paused"); btrfs_exclop_balance(fs_info, BTRFS_EXCLOP_BALANCE_PAUSED); }
--- fs/btrfs/volumes.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
fs/btrfs/volumes.c +1 -2
@@ -4638,8 +4638,7 @@ int btrfs_cancel_balance(struct btrfs_fs_info *fs_info) } } -BUG_ON (fs_info->balance_ctl || -test_bit (BTRFS_FS_BALANCE_RUNNING , &fs_info->flags)); +BUG_ON (test_bit (BTRFS_FS_BALANCE_RUNNING , &fs_info->flags));atomic_dec (&fs_info->balance_cancel_req);mutex_unlock (&fs_info->balance_mutex);return 0 ;
Signature · 1 line
2.34.1
Quoted 3 lines · xiaoshoukui
Pausing and canceling balance can race to intterupt balance lead to
BUG_ONpanic inbtrfs_cancel_balance. TheBUG_ONcondition inbtrfs_cancel_balancedoes not take this race scenario into account.
Seems that it's from times the balance was not cancellable the same way as now. Also it's a good time to switch the BUG_ON to an assertion or handle it properly.
Quoted 40 lines
However, the race condition has no other side effects. We can fix that.
Reproducing it with panic trace like this:
kernel BUG atfs/btrfs/volumes.c:4618!RIP: 0010:btrfs_cancel_balance+0x5cf/0x6a0 Call Trace: <TASK> ? do_nanosleep+0x60/0x120 ? hrtimer_nanosleep+0xb7/0x1a0 ? sched_core_clone_cookie+0x70/0x70 btrfs_ioctl_balance_ctl+0x55/0x70 btrfs_ioctl+0xa46/0xd20 __x64_sys_ioctl+0x7d/0xa0 do_syscall_64+0x38/0x80 entry_SYSCALL_64_after_hwframe+0x63/0xcdRace scenario as follows:
Quoted 11 lines
mutex_unlock(&fs_info->balance_mutex);
--------------------
.......issue pause and cancel req in another thread
--------------------
ret =__btrfs_balance(fs_info);
mutex_lock(&fs_info->balance_mutex);if (ret == -ECANCELED && atomic_read(&fs_info->balance_pause_req)) { btrfs_info(fs_info, "balance: paused"); btrfs_exclop_balance(fs_info, BTRFS_EXCLOP_BALANCE_PAUSED); }Signed-off-by xiaoshoukui <[email protected]>--- fs/btrfs/volumes.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)fs/btrfs/volumes.c +0
-0
fs/btrfs/volumes.c int btrfs_cancel_balance(struct btrfs_fs_info *fs_info) · line 4641
Quoted 6 lines
fs/btrfs/volumes.c +1
-2@@ -4638,8 +4638,7 @@ int btrfs_cancel_balance(struct btrfs_fs_info *fs_info) } } -BUG_ON (fs_info->balance_ctl || -test_bit (BTRFS_FS_BALANCE_RUNNING , &fs_info->flags)); +BUG_ON (test_bit (BTRFS_FS_BALANCE_RUNNING , &fs_info->flags));
I'll change to to ASSERT, this is really to verify that the state tracking works properly.
Quoted 5 lines
fs/btrfs/volumes.c +0
-0atomic_dec (&fs_info->balance_cancel_req);mutex_unlock (&fs_info->balance_mutex);return 0 ;Signature · 1 line
2.34.1
Quoted 3 lines
Seems that it's from times the balance was not cancellable the same way as now. Also it's a good time to switch the
BUG_ONto an assertion or handle it properly.
That's the point. Canceling the balance only takes into account the normal scenarios. Replacing the BUG ON here with an assertion would make the code cleaner.
Quoted 2 lines
I'll change to to ASSERT, this is really to verify that the state tracking works properly.
The ASSERT and BUG ON macros have already helped us uncover many hidden issues.