Re: [PATCH] [RFC] x86/cpu: rework instruction set selection

[Ingo Molnar <mingo@xxxxxxxxxx>]

* Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:

> On Sat, 26 Apr 2025 at 11:59, Arnd Bergmann <arnd@xxxxxxxx> wrote:
> >
> > Right. With the current set of features, CMOV is almost the
> > same as 686. My reasoning was that support for CMOV has a
> > very clear definition, with the instruction either being
> > available or not.
> 
> Yeah, I don't think there's any reason to make CMOV a reason to drop support.
> 
> It has questionable performance impact - I doubt anybody can measure 
> it - and the "maintenance burden" is basically a single compiler 
> flag.
> 
> (And yes, one use in a x86 header file that is pretty questionable 
> too: I think the reason for the cmov is actually i486-only behavior 
> and we could probably unify the 32-bit and 64-bit implementation)
> 
> Let's not drop Pentium support due to something as insignificant as 
> that.

Agreed on that. Idea to require CMOV dropped.

Note that the outcome of 486 removal will likely be that the few 
remaining community distros that still offer x86-32 builds are either 
already 686-CMOV-only (Debian), or are going to drop their 486 builds 
and keep their 686-CMOV-only builds (Gentoo and Archlinux32) by way of 
simple inertia. (There's an off chance that they'll change their 486 
builds to 586, but I think dropping the extra complication and 
standardizing on 686 will be the most likely outcome.)

No commercial distro builds x86-32 with a modern v6.x series kernel 
AFAICS.

Anyway, I agree that the maintenance cost on the kernel side to build 
non-CMOV kernels is very low.

Thanks,

        Ingo