[clang] c955228 - [clang][x86] Add constexpr support for MMX _mm_set*_pi* intrinsics
Simon Pilgrim via cfe-commits
cfe-commits at lists.llvm.org
Fri Nov 8 07:30:24 PST 2024
Author: Simon Pilgrim
Date: 2024-11-08T15:30:02Z
New Revision: c9552283c0bf277eba490cde9fd913510f4111c0
URL: https://github.com/llvm/llvm-project/commit/c9552283c0bf277eba490cde9fd913510f4111c0
DIFF: https://github.com/llvm/llvm-project/commit/c9552283c0bf277eba490cde9fd913510f4111c0.diff
LOG: [clang][x86] Add constexpr support for MMX _mm_set*_pi* intrinsics
Added:
Modified:
clang/lib/Headers/mmintrin.h
clang/test/CodeGen/X86/builtin_test_helpers.h
clang/test/CodeGen/X86/mmx-builtins.c
Removed:
################################################################################
diff --git a/clang/lib/Headers/mmintrin.h b/clang/lib/Headers/mmintrin.h
index 0347c5ccf8254a..dc0fa5c523eeb6 100644
--- a/clang/lib/Headers/mmintrin.h
+++ b/clang/lib/Headers/mmintrin.h
@@ -49,6 +49,12 @@ typedef char __v16qi __attribute__((__vector_size__(16)));
__min_vector_width__(128)))
#endif
+#if defined(__cplusplus) && (__cplusplus >= 201103L)
+#define __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR __DEFAULT_FN_ATTRS_SSE2 constexpr
+#else
+#define __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR __DEFAULT_FN_ATTRS_SSE2
+#endif
+
#define __trunc64(x) \
(__m64) __builtin_shufflevector((__v2di)(x), __extension__(__v2di){}, 0)
#define __anyext128(x) \
@@ -1332,10 +1338,9 @@ _mm_cmpgt_pi32(__m64 __m1, __m64 __m2)
/// This intrinsic corresponds to the <c> PXOR </c> instruction.
///
/// \returns An initialized 64-bit integer vector with all elements set to zero.
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
-_mm_setzero_si64(void)
-{
- return __extension__ (__m64){ 0LL };
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
+_mm_setzero_si64(void) {
+ return __extension__(__m64){0LL};
}
/// Constructs a 64-bit integer vector initialized with the specified
@@ -1353,10 +1358,9 @@ _mm_setzero_si64(void)
/// A 32-bit integer value used to initialize the lower 32 bits of the
/// result.
/// \returns An initialized 64-bit integer vector.
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
-_mm_set_pi32(int __i1, int __i0)
-{
- return __extension__ (__m64)(__v2si){__i0, __i1};
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
+_mm_set_pi32(int __i1, int __i0) {
+ return __extension__(__m64)(__v2si){__i0, __i1};
}
/// Constructs a 64-bit integer vector initialized with the specified
@@ -1376,10 +1380,9 @@ _mm_set_pi32(int __i1, int __i0)
/// \param __s0
/// A 16-bit integer value used to initialize bits [15:0] of the result.
/// \returns An initialized 64-bit integer vector.
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
-_mm_set_pi16(short __s3, short __s2, short __s1, short __s0)
-{
- return __extension__ (__m64)(__v4hi){__s0, __s1, __s2, __s3};
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
+_mm_set_pi16(short __s3, short __s2, short __s1, short __s0) {
+ return __extension__(__m64)(__v4hi){__s0, __s1, __s2, __s3};
}
/// Constructs a 64-bit integer vector initialized with the specified
@@ -1407,12 +1410,11 @@ _mm_set_pi16(short __s3, short __s2, short __s1, short __s0)
/// \param __b0
/// An 8-bit integer value used to initialize bits [7:0] of the result.
/// \returns An initialized 64-bit integer vector.
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
_mm_set_pi8(char __b7, char __b6, char __b5, char __b4, char __b3, char __b2,
- char __b1, char __b0)
-{
- return __extension__ (__m64)(__v8qi){__b0, __b1, __b2, __b3,
- __b4, __b5, __b6, __b7};
+ char __b1, char __b0) {
+ return __extension__(__m64)(__v8qi){__b0, __b1, __b2, __b3,
+ __b4, __b5, __b6, __b7};
}
/// Constructs a 64-bit integer vector of [2 x i32], with each of the
@@ -1428,10 +1430,9 @@ _mm_set_pi8(char __b7, char __b6, char __b5, char __b4, char __b3, char __b2,
/// A 32-bit integer value used to initialize each vector element of the
/// result.
/// \returns An initialized 64-bit integer vector of [2 x i32].
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
-_mm_set1_pi32(int __i)
-{
- return _mm_set_pi32(__i, __i);
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
+_mm_set1_pi32(int __i) {
+ return _mm_set_pi32(__i, __i);
}
/// Constructs a 64-bit integer vector of [4 x i16], with each of the
@@ -1447,10 +1448,9 @@ _mm_set1_pi32(int __i)
/// A 16-bit integer value used to initialize each vector element of the
/// result.
/// \returns An initialized 64-bit integer vector of [4 x i16].
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
-_mm_set1_pi16(short __w)
-{
- return _mm_set_pi16(__w, __w, __w, __w);
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
+_mm_set1_pi16(short __w) {
+ return _mm_set_pi16(__w, __w, __w, __w);
}
/// Constructs a 64-bit integer vector of [8 x i8], with each of the
@@ -1465,10 +1465,9 @@ _mm_set1_pi16(short __w)
/// An 8-bit integer value used to initialize each vector element of the
/// result.
/// \returns An initialized 64-bit integer vector of [8 x i8].
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
-_mm_set1_pi8(char __b)
-{
- return _mm_set_pi8(__b, __b, __b, __b, __b, __b, __b, __b);
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
+_mm_set1_pi8(char __b) {
+ return _mm_set_pi8(__b, __b, __b, __b, __b, __b, __b, __b);
}
/// Constructs a 64-bit integer vector, initialized in reverse order with
@@ -1486,10 +1485,9 @@ _mm_set1_pi8(char __b)
/// A 32-bit integer value used to initialize the upper 32 bits of the
/// result.
/// \returns An initialized 64-bit integer vector.
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
-_mm_setr_pi32(int __i0, int __i1)
-{
- return _mm_set_pi32(__i1, __i0);
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
+_mm_setr_pi32(int __i0, int __i1) {
+ return _mm_set_pi32(__i1, __i0);
}
/// Constructs a 64-bit integer vector, initialized in reverse order with
@@ -1509,10 +1507,9 @@ _mm_setr_pi32(int __i0, int __i1)
/// \param __w3
/// A 16-bit integer value used to initialize bits [63:48] of the result.
/// \returns An initialized 64-bit integer vector.
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
-_mm_setr_pi16(short __w0, short __w1, short __w2, short __w3)
-{
- return _mm_set_pi16(__w3, __w2, __w1, __w0);
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
+_mm_setr_pi16(short __w0, short __w1, short __w2, short __w3) {
+ return _mm_set_pi16(__w3, __w2, __w1, __w0);
}
/// Constructs a 64-bit integer vector, initialized in reverse order with
@@ -1540,11 +1537,10 @@ _mm_setr_pi16(short __w0, short __w1, short __w2, short __w3)
/// \param __b7
/// An 8-bit integer value used to initialize bits [63:56] of the result.
/// \returns An initialized 64-bit integer vector.
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
_mm_setr_pi8(char __b0, char __b1, char __b2, char __b3, char __b4, char __b5,
- char __b6, char __b7)
-{
- return _mm_set_pi8(__b7, __b6, __b5, __b4, __b3, __b2, __b1, __b0);
+ char __b6, char __b7) {
+ return _mm_set_pi8(__b7, __b6, __b5, __b4, __b3, __b2, __b1, __b0);
}
#undef __anyext128
diff --git a/clang/test/CodeGen/X86/builtin_test_helpers.h b/clang/test/CodeGen/X86/builtin_test_helpers.h
index 0edec1441b33bd..c3307f749942b1 100644
--- a/clang/test/CodeGen/X86/builtin_test_helpers.h
+++ b/clang/test/CodeGen/X86/builtin_test_helpers.h
@@ -9,6 +9,25 @@ constexpr bool match_m64(__m64 _v, unsigned long long a) {
return v[0] == a;
}
+constexpr bool match_v1di(__m64 v, long long a) {
+ return v[0] == a;
+}
+
+constexpr bool match_v2si(__m64 _v, int a, int b) {
+ __v2si v = (__v2si)_v;
+ return v[0] == a && v[1] == b;
+}
+
+constexpr bool match_v4hi(__m64 _v, short a, short b, short c, short d) {
+ __v4hi v = (__v4hi)_v;
+ return v[0] == a && v[1] == b && v[2] == c && v[3] == d;
+}
+
+constexpr bool match_v8qi(__m64 _v, char a, char b, char c, char d, char e, char f, char g, char h) {
+ __v8qi v = (__v8qi)_v;
+ return v[0] == a && v[1] == b && v[2] == c && v[3] == d && v[4] == e && v[5] == f && v[6] == g && v[7] == h;
+}
+
constexpr bool match_m128(__m128 v, float a, float b, float c, float d) {
return v[0] == a && v[1] == b && v[2] == c && v[3] == d;
}
diff --git a/clang/test/CodeGen/X86/mmx-builtins.c b/clang/test/CodeGen/X86/mmx-builtins.c
index 50f47e1113e4f8..52cbe45ca238b9 100644
--- a/clang/test/CodeGen/X86/mmx-builtins.c
+++ b/clang/test/CodeGen/X86/mmx-builtins.c
@@ -9,6 +9,7 @@
#include <immintrin.h>
+#include "builtin_test_helpers.h"
__m64 test_mm_abs_pi8(__m64 a) {
// CHECK-LABEL: test_mm_abs_pi8
@@ -409,6 +410,7 @@ __m64 test_mm_set_pi8(char a, char b, char c, char d, char e, char f, char g, ch
// CHECK: insertelement <8 x i8>
return _mm_set_pi8(a, b, c, d, e, f, g, h);
}
+TEST_CONSTEXPR(match_v8qi(_mm_set_pi8(0, -1, 2, -3, 4, -5, 6, -7), -7, 6, -5, 4, -3, 2, -1, 0));
__m64 test_mm_set_pi16(short a, short b, short c, short d) {
// CHECK-LABEL: test_mm_set_pi16
@@ -418,6 +420,7 @@ __m64 test_mm_set_pi16(short a, short b, short c, short d) {
// CHECK: insertelement <4 x i16>
return _mm_set_pi16(a, b, c, d);
}
+TEST_CONSTEXPR(match_v4hi(_mm_set_pi16(101, 102, -103, -104), -104, -103, 102, 101));
__m64 test_mm_set_pi32(int a, int b) {
// CHECK-LABEL: test_mm_set_pi32
@@ -425,6 +428,7 @@ __m64 test_mm_set_pi32(int a, int b) {
// CHECK: insertelement <2 x i32>
return _mm_set_pi32(a, b);
}
+TEST_CONSTEXPR(match_v2si(_mm_set_pi32(5000, -1500), -1500, 5000));
__m64 test_mm_setr_pi8(char a, char b, char c, char d, char e, char f, char g, char h) {
// CHECK-LABEL: test_mm_setr_pi8
@@ -438,6 +442,7 @@ __m64 test_mm_setr_pi8(char a, char b, char c, char d, char e, char f, char g, c
// CHECK: insertelement <8 x i8>
return _mm_setr_pi8(a, b, c, d, e, f, g, h);
}
+TEST_CONSTEXPR(match_v8qi(_mm_setr_pi8(0, -1, 2, -3, 4, -5, 6, -7), 0, -1, 2, -3, 4, -5, 6, -7));
__m64 test_mm_setr_pi16(short a, short b, short c, short d) {
// CHECK-LABEL: test_mm_setr_pi16
@@ -447,6 +452,7 @@ __m64 test_mm_setr_pi16(short a, short b, short c, short d) {
// CHECK: insertelement <4 x i16>
return _mm_setr_pi16(a, b, c, d);
}
+TEST_CONSTEXPR(match_v4hi(_mm_setr_pi16(101, 102, -103, -104), 101, 102, -103, -104));
__m64 test_mm_setr_pi32(int a, int b) {
// CHECK-LABEL: test_mm_setr_pi32
@@ -454,6 +460,14 @@ __m64 test_mm_setr_pi32(int a, int b) {
// CHECK: insertelement <2 x i32>
return _mm_setr_pi32(a, b);
}
+TEST_CONSTEXPR(match_v2si(_mm_setr_pi32(5000, -1500), 5000, -1500));
+
+__m64 test_mm_setzero_si64() {
+ // CHECK-LABEL: test_mm_setzero_si64
+ // CHECK: zeroinitializer
+ return _mm_setzero_si64();
+}
+TEST_CONSTEXPR(match_m64(_mm_setzero_si64(), 0ULL));
__m64 test_mm_set1_pi8(char a) {
// CHECK-LABEL: test_mm_set1_pi8
@@ -467,6 +481,7 @@ __m64 test_mm_set1_pi8(char a) {
// CHECK: insertelement <8 x i8>
return _mm_set1_pi8(a);
}
+TEST_CONSTEXPR(match_v8qi(_mm_set1_pi8(99), 99, 99, 99, 99, 99, 99, 99, 99));
__m64 test_mm_set1_pi16(short a) {
// CHECK-LABEL: test_mm_set1_pi16
@@ -476,6 +491,7 @@ __m64 test_mm_set1_pi16(short a) {
// CHECK: insertelement <4 x i16>
return _mm_set1_pi16(a);
}
+TEST_CONSTEXPR(match_v4hi(_mm_set1_pi16(-128), -128, -128, -128, -128));
__m64 test_mm_set1_pi32(int a) {
// CHECK-LABEL: test_mm_set1_pi32
@@ -483,6 +499,7 @@ __m64 test_mm_set1_pi32(int a) {
// CHECK: insertelement <2 x i32>
return _mm_set1_pi32(a);
}
+TEST_CONSTEXPR(match_v2si(_mm_set1_pi32(55), 55, 55));
__m64 test_mm_shuffle_pi8(__m64 a, __m64 b) {
// CHECK-LABEL: test_mm_shuffle_pi8
More information about the cfe-commits
mailing list