[PATCH] [PATCH, rs6000] Fix pr79941 (v2)

[Will Schmidt <will_schmidt at vnet dot ibm dot com>]

Hi,
Per PR79941, we are folding the vec_mul{e,o}* operations improperly. Those
entries were added to the intrinsics-to-be-folded list where the generic
multiplies should have been instead.  Test coverage in place was for the
generic multiplies, and this was missed by my testing.

The mul[eo]* unsigned operations were missing entries in builtin_function_type()
to indicate the overloaded arguments were unsigned.  This is corrected here,
and those operations now fold accurately to the desired instruction.

Testcases have been added for the mule/mulo operations, as well as a dg-do
run test based on the testcase in the PR.  I'll note that the variables in that
test case are now marked as volatile so they are not entirely optimized out during
the compile.

OK for trunk?
regtest is currently running on ppc64*.


gcc:
2017-03-09  Will Schmidt <will_schmidt@vnet.ibm.com>

     PR target/79941
     * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*UH
     entries to the case statement that marks unsigned arguments to
     overloaded functions.

testsuite:
2017-03-09  Will Schmidt <will_schmidt@vnet.ibm.com>

     PR target/79941
     * gcc.target/powerpc/fold-vec-mult-even_odd_misc.c: New test.
     * gcc.target/powerpc/fold-vec-mult-even_odd_char.c: New test.
     * gcc.target/powerpc/fold-vec-mult-even_odd_short.c: New test.
---
 gcc/config/rs6000/rs6000.c                         |    4 +
 .../gcc.target/powerpc/fold-vec-mule-char.c        |   38 ++++++++++++
 .../gcc.target/powerpc/fold-vec-mule-misc.c        |   61 ++++++++++++++++++++
 .../gcc.target/powerpc/fold-vec-mule-short.c       |   37 ++++++++++++
 4 files changed, 140 insertions(+)
 create mode 100644 gcc/testsuite/gcc.target/powerpc/fold-vec-mule-char.c
 create mode 100644 gcc/testsuite/gcc.target/powerpc/fold-vec-mule-misc.c
 create mode 100644 gcc/testsuite/gcc.target/powerpc/fold-vec-mule-short.c

diff --git a/gcc/config/rs6000/rs6000.c b/gcc/config/rs6000/rs6000.c
index 25b10f1..42109f5 100644
--- a/gcc/config/rs6000/rs6000.c
+++ b/gcc/config/rs6000/rs6000.c
@@ -18530,6 +18530,10 @@ builtin_function_type (machine_mode mode_ret, machine_mode mode_arg0,
     case ALTIVEC_BUILTIN_VMULEUH_UNS:
     case ALTIVEC_BUILTIN_VMULOUB_UNS:
     case ALTIVEC_BUILTIN_VMULOUH_UNS:
+    case ALTIVEC_BUILTIN_VMULEUB:
+    case ALTIVEC_BUILTIN_VMULEUH:
+    case ALTIVEC_BUILTIN_VMULOUB:
+    case ALTIVEC_BUILTIN_VMULOUH:
     case CRYPTO_BUILTIN_VCIPHER:
     case CRYPTO_BUILTIN_VCIPHERLAST:
     case CRYPTO_BUILTIN_VNCIPHER:
diff --git a/gcc/testsuite/gcc.target/powerpc/fold-vec-mule-char.c b/gcc/testsuite/gcc.target/powerpc/fold-vec-mule-char.c
new file mode 100644
index 0000000..88cd6cf
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/fold-vec-mule-char.c
@@ -0,0 +1,38 @@
+/* Verify that overloaded built-ins for vec_mule,vec_mulo with char
+   inputs produce the right results.  */
+
+/* { dg-do compile } */
+/* { dg-require-effective-target powerpc_altivec_ok } */
+/* { dg-options "-maltivec -O2 " } */
+
+#include <altivec.h>
+
+vector signed short
+test_even (vector signed char x, vector signed char y)
+{
+  return vec_mule (x, y);
+}
+
+vector unsigned short
+test_uns_even (vector unsigned char x, vector unsigned char y)
+{
+  return vec_mule (x, y);
+}
+
+vector signed short
+test_odd (vector signed char x, vector signed char y)
+{
+  return vec_mulo (x, y);
+}
+
+vector unsigned short
+test_uns_odd (vector unsigned char x, vector unsigned char y)
+{
+  return vec_mulo (x, y);
+}
+
+/* { dg-final { scan-assembler-times "vmuleub" 1 } } */
+/* { dg-final { scan-assembler-times "vmulesb" 1 } } */
+/* { dg-final { scan-assembler-times "vmuloub" 1 } } */
+/* { dg-final { scan-assembler-times "vmulosb" 1 } } */
+
diff --git a/gcc/testsuite/gcc.target/powerpc/fold-vec-mule-misc.c b/gcc/testsuite/gcc.target/powerpc/fold-vec-mule-misc.c
new file mode 100644
index 0000000..4bb6185
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/fold-vec-mule-misc.c
@@ -0,0 +1,61 @@
+/* PR target/79941 */
+
+/* { dg-do run } */
+/* { dg-require-effective-target powerpc_vsx_ok } */
+/* { dg-options "-mvsx -O2 -save-temps" } */
+
+#include <altivec.h>
+
+__attribute__((noinline)) void 
+test_eub_char ()
+{
+  volatile vector unsigned char v0 = {1, 0, 0, 0, 0, 0, 0, 0};
+  volatile vector unsigned char v1 = {0xff, 0, 0, 0, 0, 0, 0, 0};
+  vector unsigned short res = vec_vmuleub (v0, v1);
+  if (res[0] != (unsigned short)v0[0] * (unsigned short)v1[0])
+    __builtin_abort ();
+}
+
+__attribute__((noinline)) void 
+test_oub_char ()
+{
+  volatile vector unsigned char v0 = {0, 1, 0, 0, 0, 0, 0, 0};
+  volatile vector unsigned char v1 = {0, 0xff, 0, 0, 0, 0, 0, 0};
+  vector unsigned short res = vec_vmuloub (v0, v1);
+  if (res[0] != (unsigned short)v0[1] * (unsigned short)v1[1])
+    __builtin_abort ();
+}
+
+__attribute__((noinline)) void 
+test_euh_short ()
+{
+  volatile vector unsigned short v0 = {1, 0, 0, 0};
+  volatile vector unsigned short v1 = {0xff, 0, 0, 0};
+  vector unsigned int res = vec_vmuleuh (v0, v1);
+  if (res[0] != (unsigned int)v0[0] * (unsigned int)v1[0])
+    __builtin_abort ();
+}
+
+__attribute__((noinline)) void 
+test_ouh_short ()
+{
+  volatile vector unsigned short v0 = {0, 1, 0, 0};
+  volatile vector unsigned short v1 = {0, 0xff, 0, 0};
+  vector unsigned int res = vec_vmulouh (v0, v1);
+  if (res[0] != (unsigned int)v0[1] * (unsigned int)v1[1])
+    __builtin_abort ();
+}
+
+int main ()
+{
+  test_eub_char();
+  test_oub_char();
+  test_euh_short();
+  test_ouh_short();
+}
+
+/* { dg-final { scan-assembler-times "vmuleub" 1 } } */
+/* { dg-final { scan-assembler-times "vmuloub" 1 } } */
+/* { dg-final { scan-assembler-times "vmuleuh" 1 } } */
+/* { dg-final { scan-assembler-times "vmulouh" 1 } } */
+
diff --git a/gcc/testsuite/gcc.target/powerpc/fold-vec-mule-short.c b/gcc/testsuite/gcc.target/powerpc/fold-vec-mule-short.c
new file mode 100644
index 0000000..8c856aa
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/fold-vec-mule-short.c
@@ -0,0 +1,37 @@
+/* Verify that overloaded built-ins for vec_mule,vec_mulo with short
+   inputs produce the right results.  */
+
+/* { dg-do compile } */
+/* { dg-require-effective-target powerpc_altivec_ok } */
+/* { dg-options "-maltivec -O2 " } */
+
+#include <altivec.h>
+
+vector signed int
+test_even (vector signed short x, vector signed short y)
+{
+  return vec_mule (x, y);
+}
+
+vector unsigned int
+test_uns_even (vector unsigned short x, vector unsigned short y)
+{
+  return vec_mule (x, y);
+}
+
+vector signed int
+test_odd (vector signed short x, vector signed short y)
+{
+  return vec_mulo (x, y);
+}
+
+vector unsigned int
+test_uns_odd (vector unsigned short x, vector unsigned short y)
+{
+  return vec_mulo (x, y);
+}
+
+/* { dg-final { scan-assembler-times "vmuleuh" 1 } } */
+/* { dg-final { scan-assembler-times "vmulesh" 1 } } */
+/* { dg-final { scan-assembler-times "vmulouh" 1 } } */
+/* { dg-final { scan-assembler-times "vmulosh" 1 } } */