Re: BUG: Invalid wait context with 5.7-rc2?
From: Sebastian Andrzej Siewior
Date: Thu Apr 23 2020 - 14:50:00 EST
On 2020-04-23 10:10:29 [-0700], Paul E. McKenney wrote:
> > Curious if this was already on anyone's radar?
>
> Looks like __queue_work() is attempting to acquire a normal spinlock_t
> with interrupts disabled, which -rt doesn't like much. And I believe
> that lockdep has been upgraded to check for this.
>
> Adding Sebastian for his thoughts.
This is known. There is more than just workqueue in this regard. Please
disable CONFIG_PROVE_RAW_LOCK_NESTING until this is fixed (see Kconfig
help). Except for workqueue, there is also printk which triggers the
same warning and isn't as easy to fix as workqueue.
> Thanx, Paul
Sebastian