[Bug tree-optimization/65930] Reduction with sign-change not handled

[rguenther at suse dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65930

--- Comment #25 from rguenther at suse dot de <rguenther at suse dot de> ---
On Thu, 24 Oct 2019, tnfchris at gcc dot gnu.org wrote:

> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65930
> 
> --- Comment #22 from Tamar Christina <tnfchris at gcc dot gnu.org> ---
> It also causes failures on AArch64.
> 
> 
> FAIL: gcc.target/aarch64/sve/reduc_2.c
> FAIL: gcc.target/aarch64/sve/slp_7.c
> FAIL: gcc.target/aarch64/sve/slp_5.c

At least for this one there's the comment

/* ??? At present we don't treat the int8_t and int16_t loops as
   reductions.  */

but I guess it's the same for the rest.  You have to fixup the testcases
yourself - they are too complex and fragile.

> FAIL: gcc.target/aarch64/sve/reduc_1.c
> 
> I've also noticed a few ICEs on SPEC but haven't pinpointed it to this commit
> yet. Running a bisect.

Might be PR92205, I'm testing a patch for that.