Re: [RESEND PATCH v5 00/12] Allow preemption during IPI completion waiting to improve real-time performance

[Steven Rostedt]

On Fri, 22 May 2026 12:57:04 +0200
Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx> wrote:

> On 2026-05-13 20:45:12 [+0800], Chuyi Zhou wrote:
> > Introduction
> > ============
> > 
> > The vast majority of smp_call_function*() callers block until remote CPUs
> > complete the IPI function execution. As smp_call_function*() runs with
> > preemption disabled throughout, scheduling latency increases dramatically
> > with the number of remote CPUs and other factors (such as interrupts being
> > disabled).  
> …
> 
> Thank you for working on this.
> I went through the series and I don't see anything wrong with it
> conceptionally. I made a few comments on how to get
> smp_call_function_many_cond() a bit improved/ readable but this is about
> it.
> 
> For 1-3, 6-12:
>  Reviewed-by: Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx>
> 
> I am not too keen about that smp_processor_id() change, not sure if
> flush_tlb_kernel_range() could get its own structure. Don't want to make
> this series too complicated, that is the call of the x86 folks here.
> 

Sashiko may have found some minor issues too:

  https://sashiko.dev/#/patchset/20260513124524.2569867-1-zhouchuyi%40bytedance.com

Thanks,

-- Steve