Re: [PATCH v3 2/9] stop_machine: Add NMI-based execution path

From: Borislav Petkov

Date: Mon Nov 09 2026 - 21:31:46 EST


On Mon, Nov 09, 2026 at 06:06:47PM -0800, Chang S. Bae wrote:
> On 11/6/26 09:44, Nikolay Borisov wrote:
> >
> > I think this function should be moved over to patch 5 and by having
> > moved/subsumed patch 4, Patch 5 should become patch 3 really so related
> > functionality is introduced in subsequent patches.
>
> Yeah, the suggestion makes sense to me.
>
> On top of this fix,
> https://lore.kernel.org/lkml/d81a45f9-1ba2-4eed-92f2-c5152c5caa7b@xxxxxxxxx/
>
> my local pile now has:
>
> * x86/microcode: Remove microcode_nmi_handler_enable
> * x86/nmi: Simplify offline microcode handler invocation
> * x86/microcode: Use stop-machine NMI facility
> * x86/microcode: Distinguish NMI control path on stop-machine callback
> * stop_machine: Introduce stop_machine_nmi_cpuslocked()
> * stop_machine: Add NMI-based execution path
> * x86/nmi: Support NMI stop-machine handler
> * x86/apic: Implement self-NMI support
> * stop_machine: Define build option for NMI stop-machine
> * stop_machine: Refactor multi-CPU stop glue code
>
> I'll post a revision once sorting out the bugfix part and other feedback.

I think that's enough stuff to warrant a new submission. If the bugfix needs
to go before them, place it first in the series.

You can wait a bit until I finish looking at patch 2 and send you some more
feedback.

Thx.

--
Regards/Gruss,
Boris.

https://people.kernel.org/tglx/notes-about-netiquette