mips64-linux-gnuabi64 testsuite breakage
YunQiang Su
wzssyqa@gmail.com
Tue May 9 03:15:28 GMT 2023
Maciej W. Rozycki <macro@orcam.me.uk> 于2023年5月8日周一 01:46写道:
>
> On Thu, 4 May 2023, YunQiang Su wrote:
>
> > > + # If Vendor is IMG, then MIPSr6 is used
> > > + case ${target} in
> > > + mips*64*-img-*)
> > > + mips_cpu=mips64r6
> > > + ;;
> > > + mips*-img-*)
> > > + mips_cpu=mips32r6
> > > + ;;
> > > + esac
> > >
> > > Not only this change is wrong (you can't just arbitrarily change a
> > > configuration that has been in the wild for ~8.5 years as it'll break
> > > things for people who rely on the established semantics), but it hasn't
> > > been properly verified either, as a change is supposed not to cause
> > > regressions in the testsuite.
> > >
> > > Please revert the part quoted and fix the regressions. Thank you.
> >
> > Reverted.
>
> Thank you.
>
> FAOD if you think a triplet that implicitly defaults to R6 is useful,
No, I don't think that it is useful at all. It is so uglyyyyyyy.
In fact, we should use the correct cpu name in triple, just like
mipsisa32r6 etc.
> then feel free to submit a change for say `mips*-cipunited-*' to do so (or
> maybe `mips*-cip-*' if you want to keep it short and the company policy
> allows using such a short name), as this would be a configuration that had
> no previous mention/semantics in our tree and therefore I would have no
> concerns about it.
>
No, I won't. I submit the patch about `img`, is due to that, this vendor value
did/does/being used.
IMG/MIPS/Wave/MIPS does provide the toolchain with img as its vendor name,
and the default arch is R6.
https://codescape.mips.com/components/toolchain/2021.09-01/downloads.html
> Overall I'd suggest the `--with-arch=' approach previously discussed and
> already implemented for MIPS targets in GCC and also some other targets in
> binutils. It is more generic, although it may complicate testing, so it
> may have to be deployed with care.
>
> > And I am working on fix the tesutsuite for `-gnuabi64` and `mipsisaXXr6`.
>
> I look forward to your fixes.
>
> Maciej
--
YunQiang Su
More information about the Binutils
mailing list