On Wed, Aug 28, 2019 at 03:14:33PM +0200, Sebastian Andrzej Siewior wrote:
> On 2019-08-28 05:54:26 [-0700], Paul E. McKenney wrote:
> > On Wed, Aug 28, 2019 at 11:27:39AM +0200, Sebastian Andrzej Siewior wrote:
> > > On 2019-08-27 08:53:06 [-0700], Paul E. McKenney wrote:
> > > > Am I understanding this correctly?
> > >
> > > Everything perfect except that it is not lockdep complaining but the
> > > WARN_ON_ONCE() in rcu_note_context_switch().
> >
> > This one, right?
> >
> > WARN_ON_ONCE(!preempt && t->rcu_read_lock_nesting > 0);
> >
> > Another approach would be to change that WARN_ON_ONCE(). This fix might
> > be too extreme, as it would suppress other issues:
> >
> > WARN_ON_ONCE(IS_ENABLED(CONFIG_PREEMPT_RT_BASE) && !preempt &&
> > t->rcu_read_lock_nesting > 0);
> >
> > But maybe what is happening under the covers is that preempt is being
> > set when sleeping on a spinlock. Is that the case?
>
> I would like to keep that check and that is why we have:
>
> | #if defined(CONFIG_PREEMPT_RT_FULL)
> | sleeping_l = t->sleeping_lock;
> | #endif
> | WARN_ON_ONCE(!preempt && t->rcu_read_lock_nesting > 0 &&
> !sleeping_l);
>
> in -RT and ->sleeping_lock is that counter that is incremented in
> spin_lock(). And the only reason why sleeping_lock_inc() was used in the
> patch was to disable _this_ warning.
Very good! I would prefer an (one-line) access function to keep the
number of #if uses down to dull roar, but other than that, looks good!
(Yeah, I know, this is tree_preempt.h, but still...)
Thanx, Paul
