[PATCH v2 3/3] powerpc64le: Add glibc-hwcaps support

[Florian Weimer]

* Segher Boessenkool:

> Hi Florian,
>
> On Thu, Nov 05, 2020 at 12:40:29PM +0100, Florian Weimer wrote:
>> * Segher Boessenkool:
>> > Are you saying that it is *normal* for people to put very different code
>> > into libc like this?  Wow.
>> >
>> >> We have been through this with x86-64 already.  I don't want to produce
>> >> the same bug.
>> >
>> > No code in libc should ever use MMA, imnsho.
>> 
>> Oh, I see now.  I think we don't agree about the scope of the
>> glibc-hwcaps feature.
>
> I did not know about this new feature at all, I thought this was about
> the exiting power10/ etc. directories :-)
>
>> It's going to be used to ELF multilibs in general, not just glibc
>> components.  So a BLAS implementation could use it and drop its
>> implementation DSOs into the appropriate directories.
>> 
>> That's why vector features such as MMA matter in this context.
>
> Sure.
>
>> Given this additional context, I hope we can agree that a rule for
>> programmers like “build with -mcpu=power10 for the power10 glibc-hwcaps
>> subdirectory“ has a lot of value.
>
> But you cannot change what -mcpu=power10 means...  It should keep the
> same scheme as we have used for very long already.

I don't have a problem with selecting the power10 subdirectory for CPUs
which support ISA 3.1 *and* MMA (plus ISA 3.0 plus float128 in
hardware).  This would align glibc with the current GCC option.

The new glibc scheme allows us to add something between power10 and
power9 easily once the need arises (without backwards-incompatible
/etc/ld.so.cache format changes *cough* *cough*).

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