Hello,
My static analysis tool reports a possible ABBA deadlock in the dlm
filesystem in Linux 5.10:
dlm_recover_waiters_pre()
mutex_lock(&ls->ls_waiters_mutex); --> line 5130
recover_convert_waiter()
_receive_convert_reply()
lock_rsb()
mutex_lock(&r->res_mutex); --> line 69
dlm_recover_waiters_post()
lock_rsb()
mutex_lock(&r->res_mutex); --> line 69
mutex_lock(&ls->ls_waiters_mutex); --> line 5307
When dlm_recover_waiters_pre() and dlm_recover_waiters_post() are
concurrently executed, the deadlock can occur.
I am not quite sure whether this possible deadlock is real and how to
fix it if it is real.
Any feedback would be appreciated, thanks :)
Reported-by: TOTE Robot <[email protected]>
Best wishes,
Jia-Ju Bai
On Thu, Aug 19, 2021 at 04:54:57PM +0800, Jia-Ju Bai wrote:
> Hello,
>
> My static analysis tool reports a possible ABBA deadlock in the dlm
> filesystem in Linux 5.10:
>
> dlm_recover_waiters_pre()
> ? mutex_lock(&ls->ls_waiters_mutex); --> line 5130
> ? recover_convert_waiter()
> ??? _receive_convert_reply()
> ????? lock_rsb()
> ??????? mutex_lock(&r->res_mutex); --> line 69
>
> dlm_recover_waiters_post()
> ? lock_rsb()
> ??? mutex_lock(&r->res_mutex); --> line 69
> ? mutex_lock(&ls->ls_waiters_mutex); --> line 5307
>
> When dlm_recover_waiters_pre() and dlm_recover_waiters_post() are
> concurrently executed, the deadlock can occur.
>
> I am not quite sure whether this possible deadlock is real and how to fix it
> if it is real.
> Any feedback would be appreciated, thanks :)
They won't be concurrent, "pre" runs before recovery, and "post" is after.
Dave
On 2021/8/19 23:55, David Teigland wrote:
> On Thu, Aug 19, 2021 at 04:54:57PM +0800, Jia-Ju Bai wrote:
>> Hello,
>>
>> My static analysis tool reports a possible ABBA deadlock in the dlm
>> filesystem in Linux 5.10:
>>
>> dlm_recover_waiters_pre()
>> mutex_lock(&ls->ls_waiters_mutex); --> line 5130
>> recover_convert_waiter()
>> _receive_convert_reply()
>> lock_rsb()
>> mutex_lock(&r->res_mutex); --> line 69
>>
>> dlm_recover_waiters_post()
>> lock_rsb()
>> mutex_lock(&r->res_mutex); --> line 69
>> mutex_lock(&ls->ls_waiters_mutex); --> line 5307
>>
>> When dlm_recover_waiters_pre() and dlm_recover_waiters_post() are
>> concurrently executed, the deadlock can occur.
>>
>> I am not quite sure whether this possible deadlock is real and how to fix it
>> if it is real.
>> Any feedback would be appreciated, thanks :)
> They won't be concurrent, "pre" runs before recovery, and "post" is after.
Okay, thanks for your reply :)
Best wishes,
Jia-Ju Bai