Re: [PATCH, rs6000] Fix PR65914

[David Edelsohn <dje dot gcc at gmail dot com>]

On Fri, Jun 19, 2015 at 6:35 PM, Bill Schmidt
<wschmidt@linux.vnet.ibm.com> wrote:
> Hi,
>
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65914 demonstrates that we
> fail to match vector predicates with pseudos that are in the virtual
> stack register range.  The reduced test case provided with the bug
> report, when compiled for the C++14 standard, demonstrates that we need
> to be able to do this.  This patch loosens the restriction for the
> vector predicates so that all pseudos are accepted.  Thanks to Uli
> Weigand for his insights on this bug.
>
> As a side benefit, applying this patch fixes a number of libgomp tests
> that have been failing recently.
>
> Bootstrapped and tested on powerpc64le-unknown-linux-gnu with no
> regressions.  Is this ok for trunk?  After it burns in, I would propose
> to backport it to GCC 5.1 and GCC 4.9 (when the tree is open).
>
> Thanks,
> Bill
>
>
> [gcc]
>
> 2015-05-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
>
>         * config/rs6000/predicates.md (altivec_register_operand): Permit
>         virtual stack registers.
>         (vsx_register_operand): Likewise.
>         (vfloat_operand): Likewise.
>         (vint_operand): Likewise.
>         (vlogical_operand): Likewise.

Okay.

> [gcc/testsuite]
>
> 2015-05-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
>
>         * g++.dg/torture/pr65419.C:  New.

Maybe you should ask Richi or Jakub about the testcase because you are
placing it in a non-target-specific location.  It should succeed on
all targets, but it may expose latent bugs on other targets.

Thanks, David