Re: [PATCH v5 0/2] Start removing X86_X32_ABI

[Shuah Khan]

On 10/22/26 09:06, Sebastian Andrzej Siewior wrote:
> On 2026-10-21 12:31:46 [-0600], Shuah Khan wrote:
> > On 10/19/26 04:35, Sebastian Andrzej Siewior wrote:
> > > On 2026-10-15 13:47:37 [-0600], Shuah Khan wrote:
> > > > For selftests:
> > > >
> > > > Acked-by: Shuah Khan <skhan@xxxxxxxxxxxxxxxxxxx>
> > >
> > > Thank you.
> > >
> > > > Do you plan to cleanup x32 checks from selftests/x86 Makefiles?
> > >
> > > Could you please give me an example? I've been looking through
> > > 	tools/testing/selftests/x86
> > >
> > > and didn't find anything Makefile + x32 related. Just
> > > syscall_numbering.c which has x32 bits.
> >
> > Take a look at x86 selftest directory. You will find sigreturn.c
> > and syscall_numbering.c that have references to x32 syscalls.
> >
> > Makefile has handling for TARGETS_C_32BIT_ONLY.
>
> Okay. If x32 marked as broken it does not hurt to the selftests as-is,
> right?
> I think sigreturn.c can stay as-is except for the comment referring to
> x32.

Right the comment needs cleaning up.

> syscall_numbering.c has plenty of x32 bits and I would go and remove
> them once start removing the x32 bits. The reason for 1/2 in this series
> is that the nolibc selftests expect x32 to be working on x86-64.

Perfect.

> But this isn't the case here afaik. Now that I test it run into
>     https://lore.kernel.org/all/20261022142425.3Wx-RrU5@xxxxxxxxxxxxx/
>
> so the test is currently broken.

Probably. I haven't investigated.

> Did I understand you properly?

Yes. My objective is to get these items on your list to take a
look and address whichever you see fit and we accomplished it. :)

thanks,
-- Shuah