On Wed, 21 Mar 2018 11:50:01 -0700
Christoph Hellwig [off-list ref] wrote:

On Wed, Mar 21, 2018 at 12:38:54PM -0300, Arnaldo Carvalho de Melo wrote:

quoted

 	assert_spin_locked(&cmd->t_state_lock);
-	WARN_ON_ONCE(!irqs_disabled());
+	WARN_ON_ONCE_NONRT(!irqs_disabled());  

I can't find where WARN_ON_ONCE_NONRT is defined.

It's only in the RT patch set. But that may be changing soon.

That being said I think we can just kill these asserts.  If we have irqs
disabled spin_unlock_irq a few lines below should already warn.

I agree with Linus. This should just be some kind of
lockdep_assert_held_irqs_disabeld() or something like that.

-- Steve