Re: [RFC PATCH 00/86] Make the kernel preemptible

[Steven Rostedt]

On Wed, 08 Nov 2023 16:38:11 +0100
Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote:

> On Wed, Nov 08 2023 at 11:13, Peter Zijlstra wrote:
> > On Wed, Nov 08, 2023 at 02:04:02AM -0800, Ankur Arora wrote:
> > I'm not understanding, those should stay obviously.
> >
> > The current preempt_dynamic stuff has 5 toggles:
> >
> > /*
> >  * SC:cond_resched
> >  * SC:might_resched
> >  * SC:preempt_schedule
> >  * SC:preempt_schedule_notrace
> >  * SC:irqentry_exit_cond_resched
> >  *
> >  *
> >  * NONE:
> >  *   cond_resched               <- __cond_resched
> >  *   might_resched              <- RET0
> >  *   preempt_schedule           <- NOP
> >  *   preempt_schedule_notrace   <- NOP
> >  *   irqentry_exit_cond_resched <- NOP
> >  *
> >  * VOLUNTARY:
> >  *   cond_resched               <- __cond_resched
> >  *   might_resched              <- __cond_resched
> >  *   preempt_schedule           <- NOP
> >  *   preempt_schedule_notrace   <- NOP
> >  *   irqentry_exit_cond_resched <- NOP
> >  *
> >  * FULL:
> >  *   cond_resched               <- RET0
> >  *   might_resched              <- RET0
> >  *   preempt_schedule           <- preempt_schedule
> >  *   preempt_schedule_notrace   <- preempt_schedule_notrace
> >  *   irqentry_exit_cond_resched <- irqentry_exit_cond_resched
> >  */
> >
> > If you kill voluntary as we know it today, you can remove cond_resched
> > and might_resched, but the remaining 3 are still needed to switch
> > between NONE and FULL.  
> 
> No. The whole point of LAZY is to keep preempt_schedule(),
> preempt_schedule_notrace(), irqentry_exit_cond_resched() always enabled.

Right.

 * NONE:
 *   cond_resched               <- __cond_resched
 *   might_resched              <- RET0
 *   preempt_schedule           <- NOP
 *   preempt_schedule_notrace   <- NOP
 *   irqentry_exit_cond_resched <- NOP

Peter, how can you say we can get rid of cond_resched() in NONE when you
show that NONE still uses it? I thought the entire point of this was to get
rid of all the cond_resched() and they are there for PREEMPT_NONE as well as
VOLUNTARY. As you showed above, the only difference between NONE and
VOLUNTARY was the might_sleep.

> 
> Look at my PoC: https://lore.kernel.org/lkml/87jzshhexi.ffs@tglx/

And I've been saying that many times already ;-)

Thanks Thomas for reiterating it.

-- Steve