Re: [PATCH resend] rs6000, pr 94833: fix vec_first_match_index for nulls

2020-06-10 Thread Segher Boessenkool
Hi Carl, On Wed, Jun 10, 2020 at 09:05:36AM -0700, Carl Love wrote: > I committed this patch to mainline and backported to GCC 9. > > I have looked at GCC 8. The functional issue is there, i.e. the > vcmpnez is used instead of vcmpne. However the test case > builtins-8-p9-runnable.c does not

RE: [PATCH resend] rs6000, pr 94833: fix vec_first_match_index for nulls

2020-06-10 Thread Carl Love via Gcc-patches
Segher, Bill: I committed this patch to mainline and backported to GCC 9. I have looked at GCC 8. The functional issue is there, i.e. the vcmpnez is used instead of vcmpne. However the test case builtins-8-p9-runnable.c does not exist in GCC 8. The patch consists of the functional fix: ---

Re: [PATCH resend] rs6000, pr 94833: fix vec_first_match_index for nulls

2020-05-14 Thread Segher Boessenkool
Hi! On Wed, May 13, 2020 at 10:14:24AM -0700, Carl Love wrote: > The following patch fixes PR94833, vec_first_match_index does not > function as described in its description. > > The builtin does not handle vector elements which are zero correctly. > The following patch fixes the issue and adds