Hi!
On Thu, Jan 04, 2018 at 08:16:06AM -0600, Bill Schmidt wrote:
> https://gcc.gnu.org/PR83677 reports that generation of xxpermr is always
> wrong. It effectively inverts the order of the two input registers from
> what they should be. This patch addresses that and provides a test case
> modified from the original report.
How confusing. Maybe it would be easier to read if the operands number
1 and 2 in the pattern had swapped numbers.
> Bootstrapped and tested on powerpc64le-linux-gnu with no regressions.
> Is this okay for trunk and shortly for backport to GCC 7? I will check
> on 6, but I'm pretty certain this was introduced in 7, as 6 has only
> minimal POWER9 support.
Okay for trunk and all branches where it is needed. Thanks!
One minor testcase thingie:
> --- gcc/testsuite/gcc.target/powerpc/pr83677.c (nonexistent)
> +++ gcc/testsuite/gcc.target/powerpc/pr83677.c (working copy)
> @@ -0,0 +1,166 @@
> +/* { dg-do run { target { powerpc64*-*-* && { lp64 && p9vector_hw } } } } */
powerpc*-*-* please; or why would that not work?
Segher
