This revision was landed with ongoing or failed builds.
This revision was automatically updated to reflect the committed changes.
Closed by commit rGee815ff2ce8d: [PowerPC] Fix the implicit casting for the
emulated intrinsics (authored by maryammo).
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D144293/new/
https://reviews.llvm.org/D144293
Files:
clang/lib/Headers/ppc_wrappers/emmintrin.h
clang/lib/Headers/ppc_wrappers/smmintrin.h
clang/test/CodeGen/PowerPC/ppc-smmintrin.c
Index: clang/test/CodeGen/PowerPC/ppc-smmintrin.c
===================================================================
--- clang/test/CodeGen/PowerPC/ppc-smmintrin.c
+++ clang/test/CodeGen/PowerPC/ppc-smmintrin.c
@@ -73,7 +73,7 @@
// CHECK-LABEL: define available_externally <2 x i64> @_mm_blend_epi16(<2 x
i64> noundef %{{[0-9a-zA-Z_.]+}}, <2 x i64> noundef %{{[0-9a-zA-Z_.]+}}, i32
noundef signext %{{[0-9a-zA-Z_.]+}})
// CHECK: %[[TRUNC:[0-9a-zA-Z_.]+]] = trunc i32 %{{[0-9a-zA-Z_.]+}} to i8
-// CHECK: call <16 x i8> @vec_splats(signed char)(i8 noundef signext
%[[TRUNC]])
+// CHECK: call <16 x i8> @vec_splats(unsigned char)(i8 noundef zeroext
%[[TRUNC]])
// CHECK: call <16 x i8> @llvm.ppc.altivec.vgbbd(<16 x i8> %{{[0-9a-zA-Z_.]+}})
// CHECK: %[[PACK:[0-9a-zA-Z_.]+]] = call <8 x i16> @vec_unpackh(signed char
vector[16])
// CHECK: store <8 x i16> %[[PACK]], ptr %{{[0-9a-zA-Z_.]+}}, align 16
@@ -232,8 +232,8 @@
test_round() {
_mm_round_ps(mn1, 0);
_mm_round_ss(mn1, mn2, 0);
- _mm_round_pd(mn1, 0);
- _mm_round_sd(mn1, mn2, 0);
+ _mm_round_pd(md1, 0);
+ _mm_round_sd(md1, md2, 0);
}
// CHECK-LABEL: @test_round
Index: clang/lib/Headers/ppc_wrappers/smmintrin.h
===================================================================
--- clang/lib/Headers/ppc_wrappers/smmintrin.h
+++ clang/lib/Headers/ppc_wrappers/smmintrin.h
@@ -305,9 +305,9 @@
extern __inline __m128i
__attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_blend_epi16(__m128i __A, __m128i __B, const int __imm8) {
- __v16qi __charmask = vec_splats((signed char)__imm8);
+ __v16qu __charmask = vec_splats((unsigned char)__imm8);
__charmask = vec_gb(__charmask);
- __v8hu __shortmask = (__v8hu)vec_unpackh(__charmask);
+ __v8hu __shortmask = (__v8hu)vec_unpackh((__v16qi)__charmask);
#ifdef __BIG_ENDIAN__
__shortmask = vec_reve(__shortmask);
#endif
Index: clang/lib/Headers/ppc_wrappers/emmintrin.h
===================================================================
--- clang/lib/Headers/ppc_wrappers/emmintrin.h
+++ clang/lib/Headers/ppc_wrappers/emmintrin.h
@@ -46,6 +46,7 @@
/* SSE2 */
typedef __vector double __v2df;
+typedef __vector float __v4f;
typedef __vector long long __v2di;
typedef __vector unsigned long long __v2du;
typedef __vector int __v4si;
@@ -951,7 +952,7 @@
_mm_cvtpi32_pd(__m64 __A) {
__v4si __temp;
__v2di __tmp2;
- __v2df __result;
+ __v4f __result;
__temp = (__v4si)vec_splats(__A);
__tmp2 = (__v2di)vec_unpackl(__temp);
Index: clang/test/CodeGen/PowerPC/ppc-smmintrin.c
===================================================================
--- clang/test/CodeGen/PowerPC/ppc-smmintrin.c
+++ clang/test/CodeGen/PowerPC/ppc-smmintrin.c
@@ -73,7 +73,7 @@
// CHECK-LABEL: define available_externally <2 x i64> @_mm_blend_epi16(<2 x i64> noundef %{{[0-9a-zA-Z_.]+}}, <2 x i64> noundef %{{[0-9a-zA-Z_.]+}}, i32 noundef signext %{{[0-9a-zA-Z_.]+}})
// CHECK: %[[TRUNC:[0-9a-zA-Z_.]+]] = trunc i32 %{{[0-9a-zA-Z_.]+}} to i8
-// CHECK: call <16 x i8> @vec_splats(signed char)(i8 noundef signext %[[TRUNC]])
+// CHECK: call <16 x i8> @vec_splats(unsigned char)(i8 noundef zeroext %[[TRUNC]])
// CHECK: call <16 x i8> @llvm.ppc.altivec.vgbbd(<16 x i8> %{{[0-9a-zA-Z_.]+}})
// CHECK: %[[PACK:[0-9a-zA-Z_.]+]] = call <8 x i16> @vec_unpackh(signed char vector[16])
// CHECK: store <8 x i16> %[[PACK]], ptr %{{[0-9a-zA-Z_.]+}}, align 16
@@ -232,8 +232,8 @@
test_round() {
_mm_round_ps(mn1, 0);
_mm_round_ss(mn1, mn2, 0);
- _mm_round_pd(mn1, 0);
- _mm_round_sd(mn1, mn2, 0);
+ _mm_round_pd(md1, 0);
+ _mm_round_sd(md1, md2, 0);
}
// CHECK-LABEL: @test_round
Index: clang/lib/Headers/ppc_wrappers/smmintrin.h
===================================================================
--- clang/lib/Headers/ppc_wrappers/smmintrin.h
+++ clang/lib/Headers/ppc_wrappers/smmintrin.h
@@ -305,9 +305,9 @@
extern __inline __m128i
__attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_blend_epi16(__m128i __A, __m128i __B, const int __imm8) {
- __v16qi __charmask = vec_splats((signed char)__imm8);
+ __v16qu __charmask = vec_splats((unsigned char)__imm8);
__charmask = vec_gb(__charmask);
- __v8hu __shortmask = (__v8hu)vec_unpackh(__charmask);
+ __v8hu __shortmask = (__v8hu)vec_unpackh((__v16qi)__charmask);
#ifdef __BIG_ENDIAN__
__shortmask = vec_reve(__shortmask);
#endif
Index: clang/lib/Headers/ppc_wrappers/emmintrin.h
===================================================================
--- clang/lib/Headers/ppc_wrappers/emmintrin.h
+++ clang/lib/Headers/ppc_wrappers/emmintrin.h
@@ -46,6 +46,7 @@
/* SSE2 */
typedef __vector double __v2df;
+typedef __vector float __v4f;
typedef __vector long long __v2di;
typedef __vector unsigned long long __v2du;
typedef __vector int __v4si;
@@ -951,7 +952,7 @@
_mm_cvtpi32_pd(__m64 __A) {
__v4si __temp;
__v2di __tmp2;
- __v2df __result;
+ __v4f __result;
__temp = (__v4si)vec_splats(__A);
__tmp2 = (__v2di)vec_unpackl(__temp);
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
