This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PR19525: Build testing of non-default multilibs
Eric Botcazou <ebotcazou@libertysurf.fr> writes:
>> ... I'll be damned, I really thought we didn't have a 64-bit-capable
>> SPARC (or, more accurately, a 64-bit-capable Solaris install).
>
> Do you mean "we, CodeSourcery"?
Yes.
>> > Besides, I'd like to have both sets of tools covered, so I'm going to
>> > test it against the other one. Which set of tools did you use (GNU
>> > binutils or Sun)?
>>
>> GNU binutils.
>
> Thanks. The patch works fine with the Sun tools, both with a 32-bit multilib
> compiler and a 64-bit multilib compiler. So if you can test the non-default
> multilibs of your 32-bit multilib compiler, SPARC/Solaris will be ok.
I've now done this, and also completed testing of the non-default
multilibs on the platforms where I had forgotten to do so before
(x86_64-linux, powerpc64-linux). In all cases, I see either no
change, or a huge improvement (thousands of FAILs go away) in the test
results for the non-default multilibs.
I thus believe that Richard's patch is correct for all the targets I
can test, except ia64-hpux which needs additional adjustment. I have
a modification in my local tree which fixes ia64-hpux. I would
suggest that Richard go ahead and check in his patch, and then I will
send in my modifications as an incremental fix.
zw