On Thu, Jul 01, 2021 at 04:47:21PM -0500, Segher Boessenkool wrote:
> Hi!
>
> On Thu, Jul 01, 2021 at 10:59:15PM +0930, Alan Modra wrote:
> > * lib/target-supports.exp (check_effective_target_has_arch_pwr10): New.
>
> Mike added this already, please make sure to not add it twice :-)
Yup, rebasing took it out of my patch and a little edit took it out of
my changelog.
> [...]
> > gcc.target/powerpc/pr86731-fwrapv-longlong.c: Match power10 insns.
>
> (It still allows older as well, so "Also match" maybe?)
OK.
> Did you make sure all of these are correct and expected?
Yes, they still are. I checked that there was a corresponding
testsuite regression fix for each change too.
> Are the
> testcases still strict enough.
I think so.
, or should you add -mno-pcrel to the
> options, instead? Or maybe test both -mpcrel and -mno-pcrel? Etc.
I think adding -mno-pcrel would be a bad idea, since it would reduce
power10 code coverage, and you'll get both by simply running the
testsuite on power10 and say, power9.
>
> > * gcc.target/powerpc/lvsl-lvsr.c: Avoid file name match.
>
> You also add a "p?", is that expected? Should be in the changelog
> then :-)
It was in the changelog.. I mentioned lvsl-lvsr.c twice (which I
suppose might fall foul of the changelog commit checking). Changing
to
* gcc.target/powerpc/lvsl-lvsr.c: Likewise. Avoid file name match.
> > -/* { dg-final { scan-assembler-times {\mlxvd2x\M|\mlxv\M} 2 } } */
> > +/* { dg-final { scan-assembler-times {\mlxvd2x\M|\mp?lxv\M} 2 } } */
>
>
> > @@ -1,12 +1,12 @@
> > /* Test expected code generation for lvsl and lvsr on little endian.
> > - Note that lvsl and lvsr are each produced once, but the filename
> > - causes them to appear twice in the file. */
> > + Note that \s is used in the lvsl/lvsr matches so we don't match
> > + on '.file "lvsl-lvsr.c"'. */
>
> Even better is to not put the instruction names in the filename, but
> heh, maybe that would be too simple ;-)
>
>
> Segher
--
Alan Modra
Australia Development Lab, IBM
