[PATCH v2 3/3] powerpc64le: Add glibc-hwcaps support
Florian Weimer
fweimer@redhat.com
Mon Nov 23 10:20:05 GMT 2020
* Segher Boessenkool:
> On Mon, Nov 16, 2020 at 11:51:27AM -0300, Tulio Magno Quites Machado Filho wrote:
>> But then, we're back to this point you had raised:
>>
>> > I think we need documentation what it means for a processor to implement
>> > ISA 3.0, and not altivec.
>>
>> Unfortunately, I think this documentation will only exist after a new -mcpu
>> value is created.
>
> Nothing that uses the vector registers can work, and disabling all
> instructions that touch a vector register gets you 99.9% there. This is
> completely analoguous with -msoft-float (which really means "no FPRs").
>
> With -mno-altivec you also have no VSCR and VRSAVE registers, just like
> you lose FPSCR with -msoft-float.
>
> But yes, this should be documented in the ABI, it matters for the
> calling sequences etc.
>
> A "normal" compilation may use the vector registers if the -mcpu= you
> used allows that, so it cannot run on hardware without vector regs. The
> compiler can also use vector registers if you did not explicitly ask for
> them (again analogous to the floating point registers), if you do not
> explicitly forbid it of course.
Fair enough. I take it that the subdirectory logic in the patch is
correct. If changes are needed, they will come in the form of new
subdirectories.
Thanks,
Florian
--
Red Hat GmbH, https://de.redhat.com/ , Registered seat: Grasbrunn,
Commercial register: Amtsgericht Muenchen, HRB 153243,
Managing Directors: Charles Cachera, Brian Klemm, Laurie Krebs, Michael O'Neill
More information about the Libc-alpha
mailing list