Re: net/bluetooth: workqueue destruction WARNING in hci_unregister_dev

On 09/13/2016, 05:35 PM, Tejun Heo wrote:

Hello,

On Sat, Sep 10, 2016 at 11:33:48AM +0200, Dmitry Vyukov wrote:

quoted

Hit the WARNING with the patch. It showed "Showing busy workqueues and
worker pools:" after the WARNING, but then no queue info. Was it
already destroyed and removed from the list?...

Hmm...  It either means that the work item which was in flight when
WARN_ON() ran finished by the time the debug printout got to it or
that it's something unrelated to busy work items.

quoted

[ 198.113838] WARNING: CPU: 2 PID: 26691 at kernel/workqueue.c:4042
destroy_workqueue+0x17b/0x630

I don't seem to have the same source code that you have.  Which exact
WARN_ON() is this?

I assume Dmitry sees the same what I am still seeing, so I reported this
some time ago:
https://lkml.org/lkml/2016/3/21/492

This warning is trigerred there and still occurs with "HEAD":
  (pwq != wq->dfl_pwq) && (pwq->refcnt > 1)
and the state dump is in the log empty too:
destroy_workqueue: name='hci0' pwq=ffff88006b5c8f00
wq->dfl_pwq=ffff88006b5c9b00 pwq->refcnt=2 pwq->nr_active=0 delayed_works:
  pwq 13:
 cpus=2-3 node=1 flags=0x4 nice=-20 active=0/1
    in-flight: 2669:wq_barrier_func

thanks,
-- 
js
suse labs