[clang] 8a8e7f3 - [clang][x86] Add constexpr support for all remaining BMI1 intrinsics (#110581)
via cfe-commits
cfe-commits at lists.llvm.org
Tue Oct 1 03:10:11 PDT 2024
Author: Simon Pilgrim
Date: 2024-10-01T11:10:05+01:00
New Revision: 8a8e7f3e43861e9a36c8f39fc21a2f50a6ccc9b3
URL: https://github.com/llvm/llvm-project/commit/8a8e7f3e43861e9a36c8f39fc21a2f50a6ccc9b3
DIFF: https://github.com/llvm/llvm-project/commit/8a8e7f3e43861e9a36c8f39fc21a2f50a6ccc9b3.diff
LOG: [clang][x86] Add constexpr support for all remaining BMI1 intrinsics (#110581)
BEXTR/TZCNT are already handled, so we just need to tag the rest of the intrinsics, which are all expanded to generic patterns.
Added:
Modified:
clang/docs/ReleaseNotes.rst
clang/lib/Headers/bmiintrin.h
clang/test/CodeGen/X86/bmi-builtins.c
Removed:
################################################################################
diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index d7b4a119b77706..a55a3b8687e46e 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -518,14 +518,13 @@ X86 Support
* Supported MINMAX intrinsics of ``*_(mask(z)))_minmax(ne)_p[s|d|h|bh]`` and
``*_(mask(z)))_minmax_s[s|d|h]``.
-- The following bit manipulation intrinsics can now be used in constant expressions:
- all lzcnt intrinsics in lzcntintrin.h
- all bextr intrinsics in bmiintrin.h
- all tzcnt intrinsics in bmiintrin.h
- all bzhi intrinsics in bmi2intrin.h
- all pdep intrinsics in bmi2intrin.h
- all pext intrinsics in bmi2intrin.h
- all intrinsics in tbmintrin.h
+- All intrinsics in lzcntintrin.h can now be used in constant expressions.
+
+- All intrinsics in bmiintrin.h can now be used in constant expressions.
+
+- All bzhi/pdep/pext intrinsics in bmi2intrin.h can now be used in constant expressions.
+
+- All intrinsics in tbmintrin.h can now be used in constant expressions.
Arm and AArch64 Support
^^^^^^^^^^^^^^^^^^^^^^^
diff --git a/clang/lib/Headers/bmiintrin.h b/clang/lib/Headers/bmiintrin.h
index bf12a22d0f9904..c7ce96096d9714 100644
--- a/clang/lib/Headers/bmiintrin.h
+++ b/clang/lib/Headers/bmiintrin.h
@@ -169,12 +169,13 @@ _mm_tzcnt_64(unsigned long long __X)
#if !defined(__SCE__) || __has_feature(modules) || defined(__BMI__)
/* Define the default attributes for the functions in this file. */
-#define __DEFAULT_FN_ATTRS __attribute__((__always_inline__, __nodebug__, __target__("bmi")))
-
#if defined(__cplusplus) && (__cplusplus >= 201103L)
-#define __DEFAULT_FN_ATTRS_CONSTEXPR __DEFAULT_FN_ATTRS constexpr
+#define __DEFAULT_FN_ATTRS \
+ __attribute__((__always_inline__, __nodebug__, __target__("bmi"))) constexpr
+constexpr
#else
-#define __DEFAULT_FN_ATTRS_CONSTEXPR __DEFAULT_FN_ATTRS
+#define __DEFAULT_FN_ATTRS \
+ __attribute__((__always_inline__, __nodebug__, __target__("bmi")))
#endif
/// Performs a bitwise AND of the second operand with the one's
@@ -234,7 +235,7 @@ __andn_u32(unsigned int __X, unsigned int __Y)
/// \returns An unsigned integer whose least significant bits contain the
/// extracted bits.
/// \see _bextr_u32
-static __inline__ unsigned int __DEFAULT_FN_ATTRS_CONSTEXPR
+static __inline__ unsigned int __DEFAULT_FN_ATTRS
__bextr_u32(unsigned int __X, unsigned int __Y) {
return __builtin_ia32_bextr_u32(__X, __Y);
}
@@ -258,7 +259,7 @@ __bextr_u32(unsigned int __X, unsigned int __Y) {
/// \returns An unsigned integer whose least significant bits contain the
/// extracted bits.
/// \see __bextr_u32
-static __inline__ unsigned int __DEFAULT_FN_ATTRS_CONSTEXPR
+static __inline__ unsigned int __DEFAULT_FN_ATTRS
_bextr_u32(unsigned int __X, unsigned int __Y, unsigned int __Z) {
return __builtin_ia32_bextr_u32(__X, ((__Y & 0xff) | ((__Z & 0xff) << 8)));
}
@@ -280,7 +281,7 @@ _bextr_u32(unsigned int __X, unsigned int __Y, unsigned int __Z) {
/// \returns An unsigned integer whose least significant bits contain the
/// extracted bits.
/// \see __bextr_u32
-static __inline__ unsigned int __DEFAULT_FN_ATTRS_CONSTEXPR
+static __inline__ unsigned int __DEFAULT_FN_ATTRS
_bextr2_u32(unsigned int __X, unsigned int __Y) {
return __builtin_ia32_bextr_u32(__X, __Y);
}
@@ -453,7 +454,7 @@ __andn_u64 (unsigned long long __X, unsigned long long __Y)
/// \returns An unsigned 64-bit integer whose least significant bits contain the
/// extracted bits.
/// \see _bextr_u64
-static __inline__ unsigned long long __DEFAULT_FN_ATTRS_CONSTEXPR
+static __inline__ unsigned long long __DEFAULT_FN_ATTRS
__bextr_u64(unsigned long long __X, unsigned long long __Y) {
return __builtin_ia32_bextr_u64(__X, __Y);
}
@@ -477,7 +478,7 @@ __bextr_u64(unsigned long long __X, unsigned long long __Y) {
/// \returns An unsigned 64-bit integer whose least significant bits contain the
/// extracted bits.
/// \see __bextr_u64
-static __inline__ unsigned long long __DEFAULT_FN_ATTRS_CONSTEXPR
+static __inline__ unsigned long long __DEFAULT_FN_ATTRS
_bextr_u64(unsigned long long __X, unsigned int __Y, unsigned int __Z) {
return __builtin_ia32_bextr_u64(__X, ((__Y & 0xff) | ((__Z & 0xff) << 8)));
}
@@ -499,7 +500,7 @@ _bextr_u64(unsigned long long __X, unsigned int __Y, unsigned int __Z) {
/// \returns An unsigned 64-bit integer whose least significant bits contain the
/// extracted bits.
/// \see __bextr_u64
-static __inline__ unsigned long long __DEFAULT_FN_ATTRS_CONSTEXPR
+static __inline__ unsigned long long __DEFAULT_FN_ATTRS
_bextr2_u64(unsigned long long __X, unsigned long long __Y) {
return __builtin_ia32_bextr_u64(__X, __Y);
}
@@ -615,7 +616,6 @@ __blsr_u64(unsigned long long __X)
#endif /* __x86_64__ */
#undef __DEFAULT_FN_ATTRS
-#undef __DEFAULT_FN_ATTRS_CONSTEXPR
#endif /* !defined(__SCE__) || __has_feature(modules) || defined(__BMI__) */
diff --git a/clang/test/CodeGen/X86/bmi-builtins.c b/clang/test/CodeGen/X86/bmi-builtins.c
index 6c0b2c440ea081..ded40ca59781e2 100644
--- a/clang/test/CodeGen/X86/bmi-builtins.c
+++ b/clang/test/CodeGen/X86/bmi-builtins.c
@@ -237,6 +237,9 @@ unsigned long long test_blsr_u64(unsigned long long __X) {
// Test constexpr handling.
#if defined(__cplusplus) && (__cplusplus >= 201103L)
+char andnu32[__andn_u32(0x01234567, 0xFECDBA98) == (~0x01234567 & 0xFECDBA98) ? 1 : -1];
+char andn2u32[_andn_u32(0x01234567, 0xFECDBA98) == (~0x01234567 & 0xFECDBA98) ? 1 : -1];
+
char bextr32_0[__bextr_u32(0x00000000, 0x00000000) == 0x00000000 ? 1 : -1];
char bextr32_1[__bextr_u32(0x000003F0, 0xFFFF1004) == 0x0000003F ? 1 : -1];
char bextr32_2[__bextr_u32(0x000003F0, 0xFFFF3008) == 0x00000003 ? 1 : -1];
@@ -249,6 +252,15 @@ char bextr32_6[_bextr_u32(0x00000000, 0x00000000, 0x00000000) == 0x00000000 ? 1
char bextr32_7[_bextr_u32(0x000003F0, 0xFFFFFF04, 0xFFFFFF10) == 0x0000003F ? 1 : -1];
char bextr32_8[_bextr_u32(0x000003F0, 0xFFFFFF08, 0xFFFFFF30) == 0x00000003 ? 1 : -1];
+char blsiu32[__blsi_u32(0x89ABCDEF) == (0x89ABCDEF & -0x89ABCDEF) ? 1 : -1];
+char blsi2u32[_blsi_u32(0x89ABCDEF) == (0x89ABCDEF & -0x89ABCDEF) ? 1 : -1];
+
+char blsmasku32[__blsmsk_u32(0x89ABCDEF) == (0x89ABCDEF ^ (0x89ABCDEF - 1)) ? 1 : -1];
+char blsmask2u32[_blsmsk_u32(0x89ABCDEF) == (0x89ABCDEF ^ (0x89ABCDEF - 1)) ? 1 : -1];
+
+char blsru32[__blsr_u32(0x89ABCDEF) == (0x89ABCDEF & (0x89ABCDEF - 1)) ? 1 : -1];
+char blsr2u32[_blsr_u32(0x89ABCDEF) == (0x89ABCDEF & (0x89ABCDEF - 1)) ? 1 : -1];
+
char tzcntu16_0[__tzcnt_u16(0x0000) == 16 ? 1 : -1];
char tzcntu16_1[__tzcnt_u16(0x0001) == 0 ? 1 : -1];
char tzcntu16_2[__tzcnt_u16(0x0010) == 4 ? 1 : -1];
@@ -270,6 +282,9 @@ char tzcnt3u32_1[_mm_tzcnt_32(0x00000001) == 0 ? 1 : -1];
char tzcnt3u32_2[_mm_tzcnt_32(0x00000080) == 7 ? 1 : -1];
#ifdef __x86_64__
+char andnu64[__andn_u64(0x0123456789ABCDEFULL, 0xFECDBA9876543210ULL) == (~0x0123456789ABCDEFULL & 0xFECDBA9876543210ULL) ? 1 : -1];
+char andn2u64[_andn_u64(0x0123456789ABCDEFULL, 0xFECDBA9876543210ULL) == (~0x0123456789ABCDEFULL & 0xFECDBA9876543210ULL) ? 1 : -1];
+
char bextr64_0[__bextr_u64(0x0000000000000000ULL, 0x0000000000000000ULL) == 0x0000000000000000ULL ? 1 : -1];
char bextr64_1[__bextr_u64(0xF000000000000001ULL, 0x0000000000004001ULL) == 0x7800000000000000ULL ? 1 : -1];
char bextr64_2[__bextr_u64(0xF000000000000001ULL, 0xFFFFFFFFFFFF1001ULL) == 0x0000000000000000ULL ? 1 : -1];
@@ -282,6 +297,15 @@ char bextr64_6[_bextr_u64(0x0000000000000000ULL, 0x0000000000000000ULL, 0x000000
char bextr64_7[_bextr_u64(0xF000000000000001ULL, 0x0000000000000001ULL, 0x0000000000000040ULL) == 0x7800000000000000ULL ? 1 : -1];
char bextr64_8[_bextr_u64(0xF000000000000001ULL, 0xFFFFFFFFFFFFFF01ULL, 0xFFFFFFFFFFFFFF10ULL) == 0x0000000000000000ULL ? 1 : -1];
+char blsiu64[__blsi_u64(0x0123456789ABCDEFULL) == (0x0123456789ABCDEFULL & -0x0123456789ABCDEFULL) ? 1 : -1];
+char blsi2u64[_blsi_u64(0x0123456789ABCDEFULL) == (0x0123456789ABCDEFULL & -0x0123456789ABCDEFULL) ? 1 : -1];
+
+char blsmasku64[__blsmsk_u64(0x0123456789ABCDEFULL) == (0x0123456789ABCDEFULL ^ (0x0123456789ABCDEFULL - 1)) ? 1 : -1];
+char blsmask2u64[_blsmsk_u64(0x0123456789ABCDEFULL) == (0x0123456789ABCDEFULL ^ (0x0123456789ABCDEFULL - 1)) ? 1 : -1];
+
+char blsru64[__blsr_u64(0x0123456789ABCDEFULL) == (0x0123456789ABCDEFULL & (0x0123456789ABCDEFULL - 1)) ? 1 : -1];
+char blsr2u64[_blsr_u64(0x0123456789ABCDEFULL) == (0x0123456789ABCDEFULL & (0x0123456789ABCDEFULL - 1)) ? 1 : -1];
+
char tzcntu64_0[__tzcnt_u64(0x0000000000000000ULL) == 64 ? 1 : -1];
char tzcntu64_1[__tzcnt_u64(0x0000000000000001ULL) == 0 ? 1 : -1];
char tzcntu64_2[__tzcnt_u64(0x0000000800000000ULL) == 35 ? 1 : -1];
More information about the cfe-commits
mailing list