Aaron Sawdey was running tests on the Power9 simulator, and he noticed that the
XXSPLTW to splat the SFmode value into the vector accessed the wrong element.
When I added the more general code for vector splat for power9, I called an
insn that reversed the element numbers on little endian, when I should have
called the direct function, since the XSCVDPSPN instruction leaves the
converted value in real element 0 in the hardware.
I did a bootstrap and make check with no regressions, can I check this into the
trunk? And can I back port the patch to GCC 6.2 after a burn-in period?
[gcc]
2016-06-30 Michael Meissner <[email protected]>
PR target/71720
* config/rs6000/vsx.md (vsx_splat_v4sf_internal): When splitting
the insns, use vsx_xxspltw_v4sf_direct which does not check for
little endian.
[gcc/testsuite]
2016-06-30 Michael Meissner <[email protected]>
PR target/71720
* gcc.target/powerpc/pr71720.c: New test.
--
Michael Meissner, IBM
IBM, M/S 2506R, 550 King Street, Littleton, MA 01460-6245, USA
email: [email protected], phone: +1 (978) 899-4797
Index: gcc/config/rs6000/vsx.md
===================================================================
--- gcc/config/rs6000/vsx.md (revision 237895)
+++ gcc/config/rs6000/vsx.md (working copy)
@@ -2512,9 +2512,8 @@ (define_insn_and_split "*vsx_splat_v4sf_
[(set (match_dup 0)
(unspec:V4SF [(match_dup 1)] UNSPEC_VSX_CVDPSPN))
(set (match_dup 0)
- (vec_duplicate:V4SF
- (vec_select:SF (match_dup 0)
- (parallel [(const_int 0)]))))]
+ (unspec:V4SF [(match_dup 0)
+ (const_int 0)] UNSPEC_VSX_XXSPLTW))]
""
[(set_attr "type" "vecload,vecperm,mftgpr")
(set_attr "length" "4,8,4")])
Index: gcc/testsuite/gcc.target/powerpc/pr71720.c
===================================================================
--- gcc/testsuite/gcc.target/powerpc/pr71720.c (revision 0)
+++ gcc/testsuite/gcc.target/powerpc/pr71720.c (working copy)
@@ -0,0 +1,15 @@
+/* { dg-do compile { target { powerpc64*-*-* && lp64 } } } */
+/* { dg-skip-if "do not override -mcpu" { powerpc*-*-* } { "-mcpu=*" } {
"-mcpu=power9" } } */
+/* { dg-require-effective-target powerpc_p9vector_ok } */
+/* { dg-options "-mcpu=power9 -O2 -mupper-regs-di" } */
+
+/* Verify that we generate xxspltw <reg>,<reg>,0 for V4SFmode splat. */
+
+vector float
+splat_v4sf (float f)
+{
+ return (vector float) { f, f, f, f };
+}
+
+/* { dg-final { scan-assembler "xscvdpspn " } } */
+/* { dg-final { scan-assembler "xxspltw .*,.*,0" } } */
