This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [POWERPC] Don't override gcc -mcpu with wrong asm .machine
- From: Joseph Myers <joseph at codesourcery dot com>
- To: Tulio Magno Quites Machado Filho <tuliom at linux dot vnet dot ibm dot com>
- Cc: Alan Modra <amodra at gmail dot com>, <libc-alpha at sourceware dot org>
- Date: Fri, 20 Oct 2017 20:02:36 +0000
- Subject: Re: [POWERPC] Don't override gcc -mcpu with wrong asm .machine
- Authentication-results: sourceware.org; auth=none
- References: <20171020130241.GA27132@bubble.grove.modra.org> <87shedaerp.fsf@linux.vnet.ibm.com>
On Fri, 20 Oct 2017, Tulio Magno Quites Machado Filho wrote:
> I understand that ppc64le was never tested on POWER4, but I'm afraid this
> failure is actually showing there is another problem.
There is a configure test that for powerpc64le the (compiler + options
used) is building for POWER8 or later. *But* this does not ensure that
the power8 sysdeps directories are used.
The preferred approach for deciding which sub-architecture sysdeps
directories to use is that used in sysdeps/arm/preconfigure.ac - examine
what machine the compiler builds for. However, powerpc is still using the
older approach of requiring --with-cpu= options to select such sysdeps
directories, and there is no requirement to use --with-cpu=power8 for
powerpc64le. (Some architectures use a third approach, of basing things
on the host triplet.)
It would be a good idea to move powerpc to testing what the compiler does,
and generally to obsolete --with-cpu= for all architectures.
--
Joseph S. Myers
joseph@codesourcery.com