2022-02-09 06:37:13

by Jia-Ju Bai

Subject: [BUG] scsi: qedi: possible ABBA deadlock in qedi_process_tmf_resp() and qedi_process_cmd_cleanup_resp()

Hello,

My static analysis tool reports a possible ABBA deadlock in the qedi
driver in Linux 5.16:

qedi_process_tmf_resp()
  spin_lock(&session->back_lock); --> Line 201 (Lock A)
  spin_lock(&qedi_conn->tmf_work_lock); --> Line 230 (Lock B)

qedi_process_cmd_cleanup_resp()
  spin_lock_bh(&qedi_conn->tmf_work_lock); --> Line 752 (Lock B)
  spin_lock_bh(&conn->session->back_lock); --> Line 784 (Lock A)

When qedi_process_tmf_resp() and qedi_process_cmd_cleanup_resp() 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

2022-02-09 08:02:34

by Mike Christie

Subject: Re: [BUG] scsi: qedi: possible ABBA deadlock in qedi_process_tmf_resp() and qedi_process_cmd_cleanup_resp()

On 2/8/22 5:23 AM, Jia-Ju Bai wrote:
> Hello,
>
> My static analysis tool reports a possible ABBA deadlock in the qedi driver in Linux 5.16:
>
> qedi_process_tmf_resp()
>   spin_lock(&session->back_lock); --> Line 201 (Lock A)
>   spin_lock(&qedi_conn->tmf_work_lock); --> Line 230 (Lock B)
>
> qedi_process_cmd_cleanup_resp()
>   spin_lock_bh(&qedi_conn->tmf_work_lock); --> Line 752 (Lock B)
>   spin_lock_bh(&conn->session->back_lock); --> Line 784 (Lock A)
>
> When qedi_process_tmf_resp() and qedi_process_cmd_cleanup_resp() 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 :)
>

That looks like a valid deadlock. I'll send a patch. Thanks.

> Reported-by: TOTE Robot <[email protected]>
>
>
> Best wishes,
> Jia-Ju Bai