Re: [patch V6 07/11] rseq: Implement time slice extension enforcement timer

[Thomas Gleixner <tglx@linutronix.de>]

On Thu, Dec 18 2025 at 16:18, Peter Zijlstra wrote:
> On Mon, Dec 15, 2025 at 05:52:22PM +0100, Thomas Gleixner wrote:
>
>> +static void rseq_cancel_slice_extension_timer(void)
>> +{
>> +	struct slice_timer *st = this_cpu_ptr(&slice_timer);
>> +
>> +	/*
>> +	 * st->cookie can be safely read as preemption is disabled and the
>> +	 * timer is CPU local.
>> +	 *
>> +	 * As this is most probably the first expiring timer, the cancel is
>> +	 * expensive as it has to reprogram the hardware, but that's less
>> +	 * expensive than going through a full hrtimer_interrupt() cycle
>> +	 * for nothing.
>
> So I have these hrtick patches that skip some of that reprogramming --
> at the cost of causing those spurious interrupts. Overall that was a
> win.
>
> Should we look at the cost of a spurious hrtimer interrupt? IIRC each
> base will stop at the first iteration if the timer is 'early', which
> wasn't that bad.

Correct, but it's still going to reprogram the timer, so contrary to
cancel this takes the full overhead of the interrupt and in this case
because the expiry is short it will trigger most of the time.

Thanks,

        tglx