Hi,
PR63335 reports that the builtins vec_any_nge, vec_all_nge, vec_any_nle,
and vec_all_nle produce incorrect results for vector double. There is
some special handling for these intrinsics for the various vector
integral types. There is a test that excludes vector float from this
handling, but when vector double was added with VSX, the required
similar test was apparently not added. This patch fixes that. I've
added a test based on the attachment in the bugzilla.
Bootstrapped and tested on powerpc64le-unknown-linux-gnu. Is this ok
for trunk? I expect we should backport the fix to 4.8 and 4.9 as well.
Thanks,
Bill
[gcc]
2014-09-24 Bill Schmidt <wschm...@linux.vnet.ibm.com>
PR target/63335
* config/rs6000/rs6000-c.c (altivec_build_resolved_builtin):
Exclude VSX_BUILTIN_XVCMPGEDP_P from special handling.
[gcc/testsuite]
2014-09-24 Bill Schmidt <wschm...@linux.vnet.ibm.com>
PR target/63335
* gcc.target/powerpc/pr63335.c: New test.
Index: gcc/config/rs6000/rs6000-c.c
===================================================================
--- gcc/config/rs6000/rs6000-c.c (revision 215563)
+++ gcc/config/rs6000/rs6000-c.c (working copy)
@@ -4267,7 +4267,8 @@ altivec_build_resolved_builtin (tree *args, int n,
argument) is reversed. Patch the arguments here before building
the resolved CALL_EXPR. */
if (desc->code == ALTIVEC_BUILTIN_VEC_VCMPGE_P
- && desc->overloaded_code != ALTIVEC_BUILTIN_VCMPGEFP_P)
+ && desc->overloaded_code != ALTIVEC_BUILTIN_VCMPGEFP_P
+ && desc->overloaded_code != VSX_BUILTIN_XVCMPGEDP_P)
{
tree t;
t = args[2], args[2] = args[1], args[1] = t;
Index: gcc/testsuite/gcc.target/powerpc/pr63335.c
===================================================================
--- gcc/testsuite/gcc.target/powerpc/pr63335.c (revision 0)
+++ gcc/testsuite/gcc.target/powerpc/pr63335.c (working copy)
@@ -0,0 +1,30 @@
+/* { dg-do run { target { powerpc64*-*-* } } } */
+/* { dg-require-effective-target powerpc_vsx_ok } */
+/* { dg-options "-mvsx" } */
+
+#include <altivec.h>
+
+void abort (void);
+
+vector double vec = (vector double) {99.0, 99.0};
+
+int main() {
+
+ int actual = vec_all_nge(vec, vec);
+ if ( actual != 0)
+ abort();
+
+ actual = vec_all_nle(vec, vec);
+ if ( actual != 0)
+ abort();
+
+ actual = vec_any_nge(vec, vec);
+ if ( actual != 0)
+ abort();
+
+ actual = vec_any_nle(vec, vec);
+ if ( actual != 0)
+ abort();
+
+ return 0;
+}
