Fixed vec_pack_to_short_fp32 in Clang altivec.h

[johnplatts]

Fixes bug in vec_pack_to_short_fp32 in clang/lib/Headers/altivec.h.

Resolves issue #60822

Reviewer: @lei137

[github-actions]

Thank you for submitting a Pull Request (PR) to the LLVM Project!

This PR will be automatically labeled and the relevant teams will be notified.

If you wish to, you can add reviewers by using the "Reviewers" section on this page.

If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using @ followed by their GitHub username.

If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers.

If you have further questions, they may be answered by the LLVM GitHub User Guide.

You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums.

[llvmbot]

@llvm/pr-subscribers-backend-x86

@llvm/pr-subscribers-clang

Author: John Platts (johnplatts)

Changes

Fixes bug in vec_pack_to_short_fp32 in clang/lib/Headers/altivec.h.

Resolves issue #60822

Reviewer: @lei137

---

Full diff: https://github.com/llvm/llvm-project/pull/129923.diff

1 Files Affected:

• (modified) clang/lib/Headers/altivec.h (+3-7)

diff --git a/clang/lib/Headers/altivec.h b/clang/lib/Headers/altivec.h
index 8da65055012f1..45d557238e3d9 100644
--- a/clang/lib/Headers/altivec.h
+++ b/clang/lib/Headers/altivec.h
@@ -7529,13 +7529,9 @@ vec_pack(vector double __a, vector double __b) {
 #ifdef __POWER9_VECTOR__
 static __inline__ vector unsigned short __ATTRS_o_ai
 vec_pack_to_short_fp32(vector float __a, vector float __b) {
-  vector float __resa = __builtin_vsx_xvcvsphp(__a);
-  vector float __resb = __builtin_vsx_xvcvsphp(__b);
-#ifdef __LITTLE_ENDIAN__
-  return (vector unsigned short)vec_mergee(__resa, __resb);
-#else
-  return (vector unsigned short)vec_mergeo(__resa, __resb);
-#endif
+  vector unsigned int __resa = (vector unsigned int)__builtin_vsx_xvcvsphp(__a);
+  vector unsigned int __resb = (vector unsigned int)__builtin_vsx_xvcvsphp(__b);
+  return vec_pack(__resa, __resb);
 }
 
 #endif

[llvmbot]

@llvm/pr-subscribers-backend-x86

Author: John Platts (johnplatts)

Changes

Fixes bug in vec_pack_to_short_fp32 in clang/lib/Headers/altivec.h.

Resolves issue #60822

Reviewer: @lei137

---

Full diff: https://github.com/llvm/llvm-project/pull/129923.diff

1 Files Affected:

• (modified) clang/lib/Headers/altivec.h (+3-7)

diff --git a/clang/lib/Headers/altivec.h b/clang/lib/Headers/altivec.h
index 8da65055012f1..45d557238e3d9 100644
--- a/clang/lib/Headers/altivec.h
+++ b/clang/lib/Headers/altivec.h
@@ -7529,13 +7529,9 @@ vec_pack(vector double __a, vector double __b) {
 #ifdef __POWER9_VECTOR__
 static __inline__ vector unsigned short __ATTRS_o_ai
 vec_pack_to_short_fp32(vector float __a, vector float __b) {
-  vector float __resa = __builtin_vsx_xvcvsphp(__a);
-  vector float __resb = __builtin_vsx_xvcvsphp(__b);
-#ifdef __LITTLE_ENDIAN__
-  return (vector unsigned short)vec_mergee(__resa, __resb);
-#else
-  return (vector unsigned short)vec_mergeo(__resa, __resb);
-#endif
+  vector unsigned int __resa = (vector unsigned int)__builtin_vsx_xvcvsphp(__a);
+  vector unsigned int __resb = (vector unsigned int)__builtin_vsx_xvcvsphp(__b);
+  return vec_pack(__resa, __resb);
 }
 
 #endif

[johnplatts]

I have discovered that there is a bug in the VSX vec_pack_to_short_fp32(a, b) intrinsic with GCC on big-endian POWER9 (but not little-endian POWER9), where vec_pack_to_short_fp32(a, b) returns {b[0], b[1], b[2], b[3], a[0], a[1], a[2], a[3]} instead of the expected {a[0], a[1], a[2], a[3], b[0], b[1], b[2], b[3]}.

PVIPR expects the vec_pack_to_short_fp32(a, b) intrinsic to return the same results on both big-endian POWER9 and little-endian POWER9.

[lei137]

Thank-you for the fix John!

It would be good if your patch can show the changes that is being introduced via this patch. In prep for that I have put up a PR to add the existing checks for this test. Can you add the changes that will be introduced as part of this PR? Here is the diff once my patch lands.

diff --git a/clang/test/CodeGen/PowerPC/builtins-ppc-p9vector.c b/clang/test/CodeGen/PowerPC/builtins-ppc-p9vector.c
index 68d32ee14c8f..824267b98564 100644
--- a/clang/test/CodeGen/PowerPC/builtins-ppc-p9vector.c
+++ b/clang/test/CodeGen/PowerPC/builtins-ppc-p9vector.c
@@ -854,15 +854,13 @@ vector unsigned short test74(void) {
 // CHECK-BE: @llvm.ppc.vsx.xvcvsphp(<4 x float>
 // CHECK-BE: @llvm.ppc.vsx.xvcvsphp(<4 x float>
 // CHECK-BE: [[REG0:%[0-9]+]] = call <4 x i32> @llvm.ppc.altivec.vperm
-// CHECK-BE-NEXT: [[REG1:%[0-9]+]] = bitcast <4 x i32> [[REG0]] to <4 x float>
-// CHECK-BE-NEXT: [[REG2:%[0-9]+]] = bitcast <4 x float> [[REG1]] to <8 x i16>
-// CHECK-BE-NEXT: ret <8 x i16> [[REG2]]
+// CHECK-BE-NEXT: [[REG1:%[0-9]+]] = bitcast <4 x i32> [[REG0]] to <8 x i16>
+// CHECK-BE-NEXT: ret <8 x i16> [[REG1]]
 // CHECK: @llvm.ppc.vsx.xvcvsphp(<4 x float>
 // CHECK: @llvm.ppc.vsx.xvcvsphp(<4 x float>
 // CHECK: [[REG0:%[0-9]+]] = call <4 x i32> @llvm.ppc.altivec.vperm
-// CHECK-NEXT: [[REG1:%[0-9]+]] = bitcast <4 x i32> [[REG0]] to <4 x float>
-// CHECK-NEXT: [[REG2:%[0-9]+]] = bitcast <4 x float> [[REG1]] to <8 x i16>
-// CHECK-NEXT: ret <8 x i16> [[REG2]]
+// CHECK-NEXT: [[REG1:%[0-9]+]] = bitcast <4 x i32> [[REG0]] to <8 x i16>
+// CHECK-NEXT: ret <8 x i16> [[REG1]]
   return vec_pack_to_short_fp32(vfa, vfb);
 }
 vector unsigned int test75(void) {

Thank-you!

[lei137]

@johnplatts Please resolve the conflicts for this PR.