This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [RFC] Fix PR tree-optimization/34036


On Nov 19, 2007 11:07 AM, Eric Botcazou <ebotcazou@libertysurf.fr> wrote:
> > This caused the following regression for cris-elf:
> > Running /tmp/hpautotest-gcc1/gcc/gcc/testsuite/gcc.dg/tree-ssa/tree-ssa.exp
> > ... FAIL: gcc.dg/tree-ssa/pr23109.c scan-tree-dump-not recip "reciptmp"
>
> Thanks for the heads up.  This testcase apparently expects -ftrapping-math to
> pass on cris and now -funsafe-math-optimizations enables -fno-trapping-math:
>
> "/* LIM only performs the transformation in the no-trapping-math case.  In
>    the future we will do it for trapping-math as well in recip, check that
>    this is not wrongly optimized.  */
> /* { dg-final { scan-tree-dump-not "reciptmp" "lim" } } */
> /* { dg-final { scan-tree-dump-not "reciptmp" "recip" } } */
> /* { dg-final { cleanup-tree-dump "recip" } } */
> /* { dg-final { cleanup-tree-dump "lim" } } */"
>
>
> Richard, do you want me to go over all the testcases using -fu-m-o (23) and
> add -ftrapping-math to them (unless there is already -fno-trapping-math) or
> only to those which really need -ftrapping-math to pass?  This will trigger
> the warning in any cases.
>
> Similarly, do you want me to remove -fno-trapping-math from all of them?

Huhm.  So in fact we now changed the semantics of -funsafe-math-optimizations.
W/o -fassociative-math all its optimizations were in unsafe-math, so I think we
should retain them there.  So, apart from adjusting the testcases (I think it's
enough to fix those that we run into problems with - but if you have time to
audit the rest, that would be great), this change (enabling -fno-trapping-math
for -fu-m-o) deserves a changes.html entry.  Can you please add one?

CCing Mark, in case he also has an opinion on the changed semantics.

Thanks,
Richard.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]