Re: [RFC 1/1] sched/rt: Avoid RTO push to stopped CPUs during shutdown

[wangqk1 <wangqk1@xiaopeng.com>]

On Tue, Oct 13, 2026 at 02:03:02PM +0200, Peter Zijlstra wrote:
> This is horrific; lets not do this. The bug really is at the place where
> you randomly and without warning take CPUs down. We're not adding random
> hacks for things like that.

Understood, thanks for the quick feedback. I am dropping this patch.

One note on the trigger: this was seen on the normal poweroff path, not
on panic or crash. On arm64, machine_power_off() calls smp_send_stop()
before do_kernel_power_off(), so every orderly shutdown stops secondary
CPUs without hotplug teardown, and a CPU still running the rto push
irq_work can then pick a stale rto_mask bit.

I agree the fix belongs in the path that takes the CPUs down, and that
path will need to account for the rto overload state. If I find a viable
way to do that, I will bring it up again. Happy to test or review
anything in that area.

Thanks,
wangqk1