[PATCH v2] workqueue: Warn when work flush own workqueue

[PATCH v2] workqueue: Warn when work flush own workqueue

[Qianli Zhao]

From: Qianli Zhao <zhaoqianli@xiaomi.com>

Flushing own workqueue or work self in work context will lead to
a deadlock.
Catch this incorrect usage and issue a warning when issue happened

crash> ps 10856
PID    PPID  CPU       TASK       ST   COMM
10856    2   2  ffffffc873428080  UN  [kworker/u16:15]
crash> bt 10856
PID: 10856  TASK: ffffffc873428080  CPU: 2   COMMAND: "kworker/u16:15"
 #0 [ffffff80270cb9a0] __switch_to at ffffff99bba8533c
 #1 [ffffff80270cba30] __schedule at ffffff99bcda18dc
 #2 [ffffff80270cba50] schedule at ffffff99bcda1cdc
 #3 [ffffff80270cbaf0] schedule_timeout at ffffff99bcda6674
 #4 [ffffff80270cbb70] wait_for_common at ffffff99bcda2c68
 #5 [ffffff80270cbb80] wait_for_completion at ffffff99bcda2b60
 #6 [ffffff80270cbc30] flush_workqueue at ffffff99bbad7a60
 #7 [ffffff80270cbc90] drain_workqueue at ffffff99bbad80fc
 #8 [ffffff80270cbcb0] destroy_workqueue at ffffff99bbad92f8
 #9 [ffffff80270cbda0] dfc_svc_init at ffffff99bbfbfb6c
 #10 [ffffff80270cbdf0] process_one_work at ffffff99bbadc478
 #11 [ffffff80270cbe50] worker_thread at ffffff99bbadc9dc
 #12 [ffffff80270cbeb0] kthread at ffffff99bbae1f84

Signed-off-by: Qianli Zhao <zhaoqianli@xiaomi.com>
---
Changes in V2:
- update changelog
- update comment
---
 kernel/workqueue.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/kernel/workqueue.c b/kernel/workqueue.c
index 3f5d4bf..9798d77 100644
--- a/kernel/workqueue.c
+++ b/kernel/workqueue.c
@@ -2585,6 +2585,7 @@ static int rescuer_thread(void *__rescuer)
  * @target_work: work item being flushed (NULL for workqueue flushes)
  *
  * %current is trying to flush the whole @target_wq or @target_work on it.
+ * If a work flushing own workqueue or itself will lead to a deadlock.
  * If @target_wq doesn't have %WQ_MEM_RECLAIM, verify that %current is not
  * reclaiming memory or running on a workqueue which doesn't have
  * %WQ_MEM_RECLAIM as that can break forward-progress guarantee leading to
@@ -2594,13 +2595,16 @@ static void check_flush_dependency(struct workqueue_struct *target_wq,
 				   struct work_struct *target_work)
 {
 	work_func_t target_func = target_work ? target_work->func : NULL;
-	struct worker *worker;
+	struct worker *worker = current_wq_worker();
+
+	WARN_ONCE(worker && worker->current_pwq->wq == target_wq &&
+		  worker->task == current,
+		  "workqueue: current work function:%ps is flushing own workqueue:%s",
+		  worker->current_func, target_wq->name);
 
 	if (target_wq->flags & WQ_MEM_RECLAIM)
 		return;
 
-	worker = current_wq_worker();
-
 	WARN_ONCE(current->flags & PF_MEMALLOC,
 		  "workqueue: PF_MEMALLOC task %d(%s) is flushing !WQ_MEM_RECLAIM %s:%ps",
 		  current->pid, current->comm, target_wq->name, target_func);
-- 
2.7.4

Re: [PATCH v2] workqueue: Warn when work flush own workqueue

[qianli zhao]

Markus

Thanks for your suggestion,and sorry for my poor wording.

On Tue, Aug 25, 2020 at 4:00 PM Markus Elfring <Markus.Elfring@web.de> wrote:
>
> > Flushing own workqueue or work self in work context will lead to
> > a deadlock.
>
> I imagine that the wording “or work self” can become clearer another bit.
>
>
> > Catch this incorrect usage and issue a warning when issue happened
>
> * Would you like to mark the end of such a sentence with a dot?
>
> * How do you think about to adjust the repetition of the word “issue”?

How about below changelog?

workqueue: Warn when work flush own workqueue

Flushing itself or own workqueue in work context will
lead to a deadlock.
Catch this incorrect usage and warning when issue happened.

>
>
> …
> > - update comment
> > ---
> >  kernel/workqueue.c | 10 +++++++---
>
> I suggest to replace these triple dashes by a blank line.
Ok
>
>
> …
> > @@ -2585,6 +2585,7 @@ static int rescuer_thread(void *__rescuer)
> >   * @target_work: work item being flushed (NULL for workqueue flushes)
> >   *
> >   * %current is trying to flush the whole @target_wq or @target_work on it.
> > + * If a work flushing own workqueue or itself will lead to a deadlock.
>
> I stumble on understanding challenges for the wording “work flushing”.
> Can an adjustment help in comparison to the term “work item”?

How about below comment?

* If a work item flushing own workqueue or itself will lead to a deadlock.

>
> Regards,
> Markus

Re: [PATCH v2] workqueue: Warn when work flush own workqueue

[Dan Carpenter]

We have asked Markus to stop harrassing people about commit messages but
he refuses so I think he was banned from vger at the start of the month.

On Tue, Aug 25, 2020 at 04:55:52PM +0800, qianli zhao wrote:
> Markus
> 
> Thanks for your suggestion,and sorry for my poor wording.
> 
> On Tue, Aug 25, 2020 at 4:00 PM Markus Elfring <Markus.Elfring@web.de> wrote:
> >
> > > Flushing own workqueue or work self in work context will lead to
> > > a deadlock.
> >
> > I imagine that the wording “or work self” can become clearer another bit.
> >
> >
> > > Catch this incorrect usage and issue a warning when issue happened
> >
> > * Would you like to mark the end of such a sentence with a dot?
> >
> > * How do you think about to adjust the repetition of the word “issue”?
> 
> How about below changelog?
> 
> workqueue: Warn when work flush own workqueue
> 
> Flushing itself or own workqueue in work context will
> lead to a deadlock.
> Catch this incorrect usage and warning when issue happened.

"If a workqueue flushes itself then that will lead to a deadlock.  Print
a warning and a stack trace when this happens."

> 
> >
> >
> > …
> > > - update comment
> > > ---
> > >  kernel/workqueue.c | 10 +++++++---
> >
> > I suggest to replace these triple dashes by a blank line.
> Ok

This does not matter at all.  Keep the dashes or remove them.  It
doesn't matter at all.

> >
> >
> > …
> > > @@ -2585,6 +2585,7 @@ static int rescuer_thread(void *__rescuer)
> > >   * @target_work: work item being flushed (NULL for workqueue flushes)
> > >   *
> > >   * %current is trying to flush the whole @target_wq or @target_work on it.
> > > + * If a work flushing own workqueue or itself will lead to a deadlock.
> >
> > I stumble on understanding challenges for the wording “work flushing”.
> > Can an adjustment help in comparison to the term “work item”?
> 
> How about below comment?
> 
> * If a work item flushing own workqueue or itself will lead to a deadlock.

Write this:

	* If a work queue flushes itself, that will lead to a deadlock

regards,
dan carpenter