Re: [PATCH v7 2/3] sched/task: Add the put_task_struct_atomic_safe() function

[Wander Lairson Costa]

On Fri, May 5, 2023 at 10:33 AM Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
>
> On Thu, May 04, 2023 at 03:21:11PM -0300, Wander Lairson Costa wrote:
> > On Thu, May 04, 2023 at 05:24:24PM +0200, Peter Zijlstra wrote:
> > > On Thu, May 04, 2023 at 11:55:15AM -0300, Wander Lairson Costa wrote:
> > >
> > > > > Then I'm thinking something trivial like so:
> > > > >
> > > > > static inline void put_task_struct(struct task_struct *t)
> > > > > {
> > > > >         if (!refcount_dec_and_test(&t->usage))
> > > > >                 return;
> > > > >
> > > > >         if (IS_ENABLED(CONFIG_PREEMPT_RT) && !preemptible())
> > > > >                 call_rcu(&t->rcu, __put_task_struct_rcu);
> > > > >
> > > > >         __put_task_struct(t);
> > > > > }
> > > > >
> > > >
> > > > That's what v5 [1] does. What would be the path in this case? Should I
> > > > resend it as v8?
> > >
> > > It's almost what v5 does. v5 also has a !in_task() thing. v5 also
> > > violates codingstyle :-)
> >
> > IIRC, the in_task() is there because preemptible() doesn't check if it
> > is running in interrupt context.
>
> #define preemptible()   (preempt_count() == 0 && !irqs_disabled())
>
> When in interrupt context preempt_count() will have a non-zero value in
> HARDIRQ_MASK and IRQs must be disabled, so preemptible() evaluates to
> (false && false), last time I checked that ends up being false.
>

When I first tested, I started with in_irq(), then I saw some warnings
and added preemptible().
Never tested with preemptible() alone. Thanks, I will change that and test it.