https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89736
--- Comment #1 from kelvin at gcc dot gnu.org ---
Hi Segher,
I agree with your analysis. I'm not sure we have easy access to a platform
where I can demonstrate/reproduce the problem. Do you know where I can test
this?
I believe the proper fix is the following, which is not yet tested. Comments?
Index: gcc/testsuite/gcc.target/powerpc/pr87532-mc.c
===================================================================
--- gcc/testsuite/gcc.target/powerpc/pr87532-mc.c (revision 269764)
+++ gcc/testsuite/gcc.target/powerpc/pr87532-mc.c (working copy)
@@ -1,8 +1,8 @@
/* { dg-do run { target int128 } } */
-/* { dg-require-effective-target vmx_hw } */
-/* { dg-options "-maltivec -O2" } */
+/* { dg-require-effective-target vsx_hw } */
+/* { dg-options "-mvsx -O2" } */
-/* This test should run the same on any target that supports altivec/dfp
+/* This test should run the same on any target that supports vsx
instructions. Intentionally not specifying cpu in order to test
all code generation paths. */
