Re: [PATCH] rcu: Report a quiescent state when it's exactly in the state

From: Steven Rostedt
Date: Fri May 11 2018 - 13:29:17 EST


On Fri, 11 May 2018 10:27:35 -0700
"Paul E. McKenney" <paulmck@xxxxxxxxxxxxxxxxxx> wrote:

> On Fri, May 11, 2018 at 12:27:12PM -0400, Steven Rostedt wrote:
> > On Fri, 11 May 2018 12:25:28 -0400
> > Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:
> >
> > > I would also say that one should never call schedule() directly without
> > > changing its state to something other than TASK_RUNNING. Hence, calling
> > > schedule directly is saying you are ready to sleep. But that is not the
> > > case with cond_resched() which should always be called with the state
> > > as TASK_RUNNING.
> >
> > To continue this, with tracing, when a task is scheduled out in the
> > RUNNING state, it is considered preempted, otherwise it is not.
>
> I suppose another option would be for cond_resched_tasks_rcu_qs() to set
> (and later clear) a per-CPU variable that causes rcu_note_context_switch()
> to ignore its "preempt" parameter. Byungchul's approach seems more
> straightforward, though.

I agree that I prefer Byungchul's approach better ;-)

-- Steve