Skip to content

[Clang] Add constexpr support for AVX512 permutexvar intrinsics #167802

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 4 commits into from
Nov 13, 2025
Merged

[Clang] Add constexpr support for AVX512 permutexvar intrinsics #167802

merged 4 commits into from
Nov 13, 2025
+305 −137

Conversation

@chaitanyav
Copy link
Contributor

@chaitanyav chaitanyav commented Nov 13, 2025

Resolves: #167476

@chaitanyav chaitanyav marked this pull request as ready for review November 13, 2025 01:57
@llvmbot llvmbot added clang Clang issues not falling into any other category backend:X86 clang:frontend Language frontend issues, e.g. anything involving "Sema" clang:headers Headers provided by Clang, e.g. for intrinsics clang:bytecode Issues for the clang bytecode constexpr interpreter labels Nov 13, 2025
@chaitanyav chaitanyav self-assigned this Nov 13, 2025
@llvmbot
Copy link
Member

llvmbot commented Nov 13, 2025

@llvm/pr-subscribers-clang

Author: NagaChaitanya Vellanki (chaitanyav)

Changes

Resolves: #167476

Patch is 55.89 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/167802.diff

14 Files Affected:

  • (modified) clang/include/clang/Basic/BuiltinsX86.td (+13-10)
  • (modified) clang/lib/AST/ByteCode/InterpBuiltin.cpp (+44)
  • (modified) clang/lib/AST/ExprConstant.cpp (+64)
  • (modified) clang/lib/Headers/avx10_2_512bf16intrin.h (+1-1)
  • (modified) clang/lib/Headers/avx10_2bf16intrin.h (+2-2)
  • (modified) clang/lib/Headers/avx2intrin.h (+4-6)
  • (modified) clang/lib/Headers/avx512bwintrin.h (+7-11)
  • (modified) clang/lib/Headers/avx512fintrin.h (+27-38)
  • (modified) clang/lib/Headers/avx512vlbwintrin.h (+14-21)
  • (modified) clang/lib/Headers/avx512vlintrin.h (+22-32)
  • (modified) clang/test/CodeGen/X86/avx512bw-builtins.c (+11-2)
  • (modified) clang/test/CodeGen/X86/avx512f-builtins.c (+46-10)
  • (modified) clang/test/CodeGen/X86/avx512vl-builtins.c (+37)
  • (modified) clang/test/CodeGen/X86/avx512vlbw-builtins.c (+21-4)
diff --git a/clang/include/clang/Basic/BuiltinsX86.td b/clang/include/clang/Basic/BuiltinsX86.td
index cb08e2107f072..b261b681990e0 100644
--- a/clang/include/clang/Basic/BuiltinsX86.td
+++ b/clang/include/clang/Basic/BuiltinsX86.td
@@ -603,6 +603,11 @@ let Features = "avx", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWid
   def vec_set_v8si : X86Builtin<"_Vector<8, int>(_Vector<8, int>, int, _Constant int)">;
 }
 
+let Features = "avx2", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<256>] in {
+ def permvarsi256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<8, int>)">;
+ def permvarsf256 : X86Builtin<"_Vector<8, float>(_Vector<8, float>, _Vector<8, int>)">;
+}
+
 let Features = "avx2", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in {
   def mpsadbw256 : X86Builtin<"_Vector<32, char>(_Vector<32, char>, _Vector<32, char>, _Constant char)">;
 
@@ -617,9 +622,7 @@ let Features = "avx2", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] i
   def psrlw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<8, short>)">;
   def psrld256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<4, int>)">;
   def psrlq256 : X86Builtin<"_Vector<4, long long int>(_Vector<4, long long int>, _Vector<2, long long int>)">;
-  def permvarsi256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<8, int>)">;
   def permdf256 : X86Builtin<"_Vector<4, double>(_Vector<4, double>, _Constant int)">;
-  def permvarsf256 : X86Builtin<"_Vector<8, float>(_Vector<8, float>, _Vector<8, int>)">;
   def permti256 : X86Builtin<"_Vector<4, long long int>(_Vector<4, long long int>, _Vector<4, long long int>, _Constant int)">;
   def permdi256 : X86Builtin<"_Vector<4, long long int>(_Vector<4, long long int>, _Constant int)">;
 }
@@ -3052,38 +3055,38 @@ let Features = "avx512f", Attributes = [NoThrow, Const, RequiredVectorWidth<512>
   def permdi512 : X86Builtin<"_Vector<8, long long int>(_Vector<8, long long int>, _Constant int)">;
 }
 
-let Features = "avx512bw", Attributes = [NoThrow, Const, RequiredVectorWidth<512>] in {
+let Features = "avx512bw", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<512>] in {
   def permvarhi512 : X86Builtin<"_Vector<32, short>(_Vector<32, short>, _Vector<32, short>)">;
 }
 
-let Features = "avx512f", Attributes = [NoThrow, Const, RequiredVectorWidth<512>] in {
+let Features = "avx512f", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<512>] in {
   def permvardf512 : X86Builtin<"_Vector<8, double>(_Vector<8, double>, _Vector<8, long long int>)">;
   def permvardi512 : X86Builtin<"_Vector<8, long long int>(_Vector<8, long long int>, _Vector<8, long long int>)">;
   def permvarsf512 : X86Builtin<"_Vector<16, float>(_Vector<16, float>, _Vector<16, int>)">;
   def permvarsi512 : X86Builtin<"_Vector<16, int>(_Vector<16, int>, _Vector<16, int>)">;
 }
 
-let Features = "avx512vbmi", Attributes = [NoThrow, Const, RequiredVectorWidth<512>] in {
+let Features = "avx512vbmi", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<512>] in {
   def permvarqi512 : X86Builtin<"_Vector<64, char>(_Vector<64, char>, _Vector<64, char>)">;
 }
 
-let Features = "avx512vbmi,avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<128>] in {
+let Features = "avx512vbmi,avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in {
   def permvarqi128 : X86Builtin<"_Vector<16, char>(_Vector<16, char>, _Vector<16, char>)">;
 }
 
-let Features = "avx512vbmi,avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in {
+let Features = "avx512vbmi,avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<256>] in {
   def permvarqi256 : X86Builtin<"_Vector<32, char>(_Vector<32, char>, _Vector<32, char>)">;
 }
 
-let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<128>] in {
+let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in {
   def permvarhi128 : X86Builtin<"_Vector<8, short>(_Vector<8, short>, _Vector<8, short>)">;
 }
 
-let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in {
+let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<256>] in {
   def permvarhi256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">;
 }
 
-let Features = "avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in {
+let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<256>] in {
   def permvardf256 : X86Builtin<"_Vector<4, double>(_Vector<4, double>, _Vector<4, long long int>)">;
   def permvardi256 : X86Builtin<"_Vector<4, long long int>(_Vector<4, long long int>, _Vector<4, long long int>)">;
 }
diff --git a/clang/lib/AST/ByteCode/InterpBuiltin.cpp b/clang/lib/AST/ByteCode/InterpBuiltin.cpp
index 6c7b2f502cc51..c72a3566681b1 100644
--- a/clang/lib/AST/ByteCode/InterpBuiltin.cpp
+++ b/clang/lib/AST/ByteCode/InterpBuiltin.cpp
@@ -4414,6 +4414,50 @@ bool InterpretBuiltin(InterpState &S, CodePtr OpPC, const CallExpr *Call,
             return std::pair<unsigned, int>{0, static_cast<int>(DstIdx)};
           }
         });
+  case X86::BI__builtin_ia32_permvarsi256:
+  case X86::BI__builtin_ia32_permvarsf256:
+  case X86::BI__builtin_ia32_permvardf512:
+  case X86::BI__builtin_ia32_permvardi512:
+  case X86::BI__builtin_ia32_permvarhi128:
+    return interp__builtin_ia32_shuffle_generic(
+        S, OpPC, Call, [](unsigned DstIdx, unsigned ShuffleMask) {
+          int Offset = ShuffleMask & 0x7;
+          unsigned SrcIdx = 0;
+          return std::pair<unsigned, int>{SrcIdx, Offset};
+        });
+  case X86::BI__builtin_ia32_permvarqi128:
+  case X86::BI__builtin_ia32_permvarhi256:
+  case X86::BI__builtin_ia32_permvarsi512:
+  case X86::BI__builtin_ia32_permvarsf512:
+    return interp__builtin_ia32_shuffle_generic(
+        S, OpPC, Call, [](unsigned DstIdx, unsigned ShuffleMask) {
+          int Offset = ShuffleMask & 0xF;
+          unsigned SrcIdx = 0;
+          return std::pair<unsigned, int>{SrcIdx, Offset};
+        });
+  case X86::BI__builtin_ia32_permvardi256:
+  case X86::BI__builtin_ia32_permvardf256:
+    return interp__builtin_ia32_shuffle_generic(
+        S, OpPC, Call, [](unsigned DstIdx, unsigned ShuffleMask) {
+          int Offset = ShuffleMask & 0x3;
+          unsigned SrcIdx = 0;
+          return std::pair<unsigned, int>{SrcIdx, Offset};
+        });
+  case X86::BI__builtin_ia32_permvarqi256:
+  case X86::BI__builtin_ia32_permvarhi512:
+    return interp__builtin_ia32_shuffle_generic(
+        S, OpPC, Call, [](unsigned DstIdx, unsigned ShuffleMask) {
+          int Offset = ShuffleMask & 0x1F;
+          unsigned SrcIdx = 0;
+          return std::pair<unsigned, int>{SrcIdx, Offset};
+        });
+  case X86::BI__builtin_ia32_permvarqi512:
+    return interp__builtin_ia32_shuffle_generic(
+        S, OpPC, Call, [](unsigned DstIdx, unsigned ShuffleMask) {
+          int Offset = ShuffleMask & 0x3F;
+          unsigned SrcIdx = 0;
+          return std::pair<unsigned, int>{SrcIdx, Offset};
+        });
   case X86::BI__builtin_ia32_vpermi2varq128:
   case X86::BI__builtin_ia32_vpermi2varpd128:
     return interp__builtin_ia32_shuffle_generic(
diff --git a/clang/lib/AST/ExprConstant.cpp b/clang/lib/AST/ExprConstant.cpp
index 1bfea24b228e8..e9e448143477e 100644
--- a/clang/lib/AST/ExprConstant.cpp
+++ b/clang/lib/AST/ExprConstant.cpp
@@ -13551,6 +13551,70 @@ bool VectorExprEvaluator::VisitCallExpr(const CallExpr *E) {
       return false;
     return Success(R, E);
   }
+  case X86::BI__builtin_ia32_permvarsi256:
+  case X86::BI__builtin_ia32_permvarsf256:
+  case X86::BI__builtin_ia32_permvardf512:
+  case X86::BI__builtin_ia32_permvardi512:
+  case X86::BI__builtin_ia32_permvarhi128: {
+    APValue R;
+    if (!evalShuffleGeneric(Info, E, R,
+                            [](unsigned DstIdx, unsigned ShuffleMask) {
+                              int Offset = ShuffleMask & 0x7;
+                              unsigned SrcIdx = 0;
+                              return std::pair<unsigned, int>{SrcIdx, Offset};
+                            }))
+      return false;
+    return Success(R, E);
+  }
+  case X86::BI__builtin_ia32_permvarqi128:
+  case X86::BI__builtin_ia32_permvarhi256:
+  case X86::BI__builtin_ia32_permvarsi512:
+  case X86::BI__builtin_ia32_permvarsf512: {
+    APValue R;
+    if (!evalShuffleGeneric(Info, E, R,
+                            [](unsigned DstIdx, unsigned ShuffleMask) {
+                              int Offset = ShuffleMask & 0xF;
+                              unsigned SrcIdx = 0;
+                              return std::pair<unsigned, int>{SrcIdx, Offset};
+                            }))
+      return false;
+    return Success(R, E);
+  }
+  case X86::BI__builtin_ia32_permvardi256:
+  case X86::BI__builtin_ia32_permvardf256: {
+    APValue R;
+    if (!evalShuffleGeneric(Info, E, R,
+                            [](unsigned DstIdx, unsigned ShuffleMask) {
+                              int Offset = ShuffleMask & 0x3;
+                              unsigned SrcIdx = 0;
+                              return std::pair<unsigned, int>{SrcIdx, Offset};
+                            }))
+      return false;
+    return Success(R, E);
+  }
+  case X86::BI__builtin_ia32_permvarqi256:
+  case X86::BI__builtin_ia32_permvarhi512: {
+    APValue R;
+    if (!evalShuffleGeneric(Info, E, R,
+                            [](unsigned DstIdx, unsigned ShuffleMask) {
+                              int Offset = ShuffleMask & 0x1F;
+                              unsigned SrcIdx = 0;
+                              return std::pair<unsigned, int>{SrcIdx, Offset};
+                            }))
+      return false;
+    return Success(R, E);
+  }
+  case X86::BI__builtin_ia32_permvarqi512: {
+    APValue R;
+    if (!evalShuffleGeneric(Info, E, R,
+                            [](unsigned DstIdx, unsigned ShuffleMask) {
+                              int Offset = ShuffleMask & 0x3F;
+                              unsigned SrcIdx = 0;
+                              return std::pair<unsigned, int>{SrcIdx, Offset};
+                            }))
+      return false;
+    return Success(R, E);
+  }
   case X86::BI__builtin_ia32_vpermi2varq128:
   case X86::BI__builtin_ia32_vpermi2varpd128: {
     APValue R;
diff --git a/clang/lib/Headers/avx10_2_512bf16intrin.h b/clang/lib/Headers/avx10_2_512bf16intrin.h
index 46ec12a63ef9c..3201307af4731 100644
--- a/clang/lib/Headers/avx10_2_512bf16intrin.h
+++ b/clang/lib/Headers/avx10_2_512bf16intrin.h
@@ -179,7 +179,7 @@ _mm512_permutex2var_pbh(__m512bh __A, __m512i __I, __m512bh __B) {
                                                   (__v32hi)__B);
 }
 
-static __inline__ __m512bh __DEFAULT_FN_ATTRS512
+static __inline__ __m512bh __DEFAULT_FN_ATTRS512_CONSTEXPR
 _mm512_permutexvar_pbh(__m512i __A, __m512bh __B) {
   return (__m512bh)__builtin_ia32_permvarhi512((__v32hi)__B, (__v32hi)__A);
 }
diff --git a/clang/lib/Headers/avx10_2bf16intrin.h b/clang/lib/Headers/avx10_2bf16intrin.h
index 8fb8cd7cd0865..9f5b726d7b789 100644
--- a/clang/lib/Headers/avx10_2bf16intrin.h
+++ b/clang/lib/Headers/avx10_2bf16intrin.h
@@ -307,12 +307,12 @@ _mm256_permutex2var_pbh(__m256bh __A, __m256i __I, __m256bh __B) {
                                                   (__v16hi)__B);
 }
 
-static __inline__ __m128bh __DEFAULT_FN_ATTRS128
+static __inline__ __m128bh __DEFAULT_FN_ATTRS128_CONSTEXPR
 _mm_permutexvar_pbh(__m128i __A, __m128bh __B) {
   return (__m128bh)__builtin_ia32_permvarhi128((__v8hi)__B, (__v8hi)__A);
 }
 
-static __inline__ __m256bh __DEFAULT_FN_ATTRS256
+static __inline__ __m256bh __DEFAULT_FN_ATTRS256_CONSTEXPR
 _mm256_permutexvar_pbh(__m256i __A, __m256bh __B) {
   return (__m256bh)__builtin_ia32_permvarhi256((__v16hi)__B, (__v16hi)__A);
 }
diff --git a/clang/lib/Headers/avx2intrin.h b/clang/lib/Headers/avx2intrin.h
index 3cbaaece7b38e..3e3c13d8bd662 100644
--- a/clang/lib/Headers/avx2intrin.h
+++ b/clang/lib/Headers/avx2intrin.h
@@ -3214,9 +3214,8 @@ _mm_broadcastq_epi64(__m128i __X) {
 ///    A 256-bit vector of [8 x i32] containing indexes of values to use from
 ///    \a __a.
 /// \returns A 256-bit vector of [8 x i32] containing the result.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_permutevar8x32_epi32(__m256i __a, __m256i __b)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_permutevar8x32_epi32(__m256i __a, __m256i __b) {
   return (__m256i)__builtin_ia32_permvarsi256((__v8si)__a, (__v8si)__b);
 }
 
@@ -3272,9 +3271,8 @@ _mm256_permutevar8x32_epi32(__m256i __a, __m256i __b)
 ///    A 256-bit vector of [8 x i32] containing indexes of values to use from
 ///    \a __a.
 /// \returns A 256-bit vector of [8 x float] containing the result.
-static __inline__ __m256 __DEFAULT_FN_ATTRS256
-_mm256_permutevar8x32_ps(__m256 __a, __m256i __b)
-{
+static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_permutevar8x32_ps(__m256 __a, __m256i __b) {
   return (__m256)__builtin_ia32_permvarsf256((__v8sf)__a, (__v8si)__b);
 }
 
diff --git a/clang/lib/Headers/avx512bwintrin.h b/clang/lib/Headers/avx512bwintrin.h
index 4a02c96620335..3cfa32eb9e727 100644
--- a/clang/lib/Headers/avx512bwintrin.h
+++ b/clang/lib/Headers/avx512bwintrin.h
@@ -1846,25 +1846,21 @@ _mm512_maskz_broadcastw_epi16 (__mmask32 __M, __m128i __A)
                                              (__v32hi) _mm512_setzero_si512());
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_permutexvar_epi16 (__m512i __A, __m512i __B)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_permutexvar_epi16(__m512i __A, __m512i __B) {
   return (__m512i)__builtin_ia32_permvarhi512((__v32hi)__B, (__v32hi)__A);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_maskz_permutexvar_epi16 (__mmask32 __M, __m512i __A,
-        __m512i __B)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_permutexvar_epi16(__mmask32 __M, __m512i __A, __m512i __B) {
   return (__m512i)__builtin_ia32_selectw_512((__mmask32)__M,
                                     (__v32hi)_mm512_permutexvar_epi16(__A, __B),
                                     (__v32hi)_mm512_setzero_si512());
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_mask_permutexvar_epi16 (__m512i __W, __mmask32 __M, __m512i __A,
-             __m512i __B)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_permutexvar_epi16(__m512i __W, __mmask32 __M, __m512i __A,
+                              __m512i __B) {
   return (__m512i)__builtin_ia32_selectw_512((__mmask32)__M,
                                     (__v32hi)_mm512_permutexvar_epi16(__A, __B),
                                     (__v32hi)__W);
diff --git a/clang/lib/Headers/avx512fintrin.h b/clang/lib/Headers/avx512fintrin.h
index 997e9608e112f..79c37173ac838 100644
--- a/clang/lib/Headers/avx512fintrin.h
+++ b/clang/lib/Headers/avx512fintrin.h
@@ -7959,93 +7959,82 @@ _mm_mask3_fnmsub_sd (__m128d __W, __m128d __X, __m128d __Y, __mmask8 __U)
                                        (__v8di)_mm512_permutex_epi64((X), (C)), \
                                        (__v8di)_mm512_setzero_si512()))
 
-static __inline__ __m512d __DEFAULT_FN_ATTRS512
-_mm512_permutexvar_pd (__m512i __X, __m512d __Y)
-{
+static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_permutexvar_pd(__m512i __X, __m512d __Y) {
   return (__m512d)__builtin_ia32_permvardf512((__v8df) __Y, (__v8di) __X);
 }
 
-static __inline__ __m512d __DEFAULT_FN_ATTRS512
-_mm512_mask_permutexvar_pd (__m512d __W, __mmask8 __U, __m512i __X, __m512d __Y)
-{
+static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_permutexvar_pd(__m512d __W, __mmask8 __U, __m512i __X,
+                           __m512d __Y) {
   return (__m512d)__builtin_ia32_selectpd_512((__mmask8)__U,
                                         (__v8df)_mm512_permutexvar_pd(__X, __Y),
                                         (__v8df)__W);
 }
 
-static __inline__ __m512d __DEFAULT_FN_ATTRS512
-_mm512_maskz_permutexvar_pd (__mmask8 __U, __m512i __X, __m512d __Y)
-{
+static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_permutexvar_pd(__mmask8 __U, __m512i __X, __m512d __Y) {
   return (__m512d)__builtin_ia32_selectpd_512((__mmask8)__U,
                                         (__v8df)_mm512_permutexvar_pd(__X, __Y),
                                         (__v8df)_mm512_setzero_pd());
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_permutexvar_epi64 (__m512i __X, __m512i __Y)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_permutexvar_epi64(__m512i __X, __m512i __Y) {
   return (__m512i)__builtin_ia32_permvardi512((__v8di)__Y, (__v8di)__X);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_maskz_permutexvar_epi64 (__mmask8 __M, __m512i __X, __m512i __Y)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_permutexvar_epi64(__mmask8 __M, __m512i __X, __m512i __Y) {
   return (__m512i)__builtin_ia32_selectq_512((__mmask8)__M,
                                      (__v8di)_mm512_permutexvar_epi64(__X, __Y),
                                      (__v8di)_mm512_setzero_si512());
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_mask_permutexvar_epi64 (__m512i __W, __mmask8 __M, __m512i __X,
-             __m512i __Y)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_permutexvar_epi64(__m512i __W, __mmask8 __M, __m512i __X,
+                              __m512i __Y) {
   return (__m512i)__builtin_ia32_selectq_512((__mmask8)__M,
                                      (__v8di)_mm512_permutexvar_epi64(__X, __Y),
                                      (__v8di)__W);
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_permutexvar_ps (__m512i __X, __m512 __Y)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_permutexvar_ps(__m512i __X, __m512 __Y) {
   return (__m512)__builtin_ia32_permvarsf512((__v16sf)__Y, (__v16si)__X);
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_mask_permutexvar_ps (__m512 __W, __mmask16 __U, __m512i __X, __m512 __Y)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_permutexvar_ps(__m512 __W, __mmask16 __U, __m512i __X, __m512 __Y) {
   return (__m512)__builtin_ia32_selectps_512((__mmask16)__U,
                                        (__v16sf)_mm512_permutexvar_ps(__X, __Y),
                                        (__v16sf)__W);
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_maskz_permutexvar_ps (__mmask16 __U, __m512i __X, __m512 __Y)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_permutexvar_ps(__mmask16 __U, __m512i __X, __m512 __Y) {
   return (__m512)__builtin_ia32_selectps_512((__mmask16)__U,
                                        (__v16sf)_mm512_permutexvar_ps(__X, __Y),
                                        (__v16sf)_mm512_setzero_ps());
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_permutexvar_epi32 (__m512i __X, __m512i __Y)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_permutexvar_epi32(__m512i __X, __m512i __Y) {
   return (__m512i)__builtin_ia32_permvarsi512((__v16si)__Y, (__v16si)__X);
 }
 
 #define _mm512_permutevar_epi32 _mm512_permutexvar_epi32
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_maskz_permutexvar_epi32 (__mmask16 __M, __m512i __X, __m512i __Y)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_permutexvar_epi32(__mmask16 __M, __m512i __X, __m512i __Y) {
   return (__m512i)__builtin_ia32_selectd_512((__mmask16)__M,
                                     (__v16si)_mm512_permutexvar_epi32(__X, __Y),
                                     (__v16si)_mm512_setzero_si512());
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_mask_permutexvar_epi32 (__m512i __W, __mmask16 __M, __m512i __X,
-             __m512i __Y)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_permutexvar_epi32(__m512i __W, __mmask16 __M, __m512i __X,
+                              __m512i __Y) {
   return (__m512i)__builtin_ia32_selectd_512((__mmask16)__M,
                                     (__v16si)_mm512_permutexvar_epi32(__...
[truncated]

@llvmbot
Copy link
Member

llvmbot commented Nov 13, 2025

@llvm/pr-subscribers-backend-x86

Author: NagaChaitanya Vellanki (chaitanyav)

Changes

Resolves: #167476

Patch is 55.89 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/167802.diff

14 Files Affected:

  • (modified) clang/include/clang/Basic/BuiltinsX86.td (+13-10)
  • (modified) clang/lib/AST/ByteCode/InterpBuiltin.cpp (+44)
  • (modified) clang/lib/AST/ExprConstant.cpp (+64)
  • (modified) clang/lib/Headers/avx10_2_512bf16intrin.h (+1-1)
  • (modified) clang/lib/Headers/avx10_2bf16intrin.h (+2-2)
  • (modified) clang/lib/Headers/avx2intrin.h (+4-6)
  • (modified) clang/lib/Headers/avx512bwintrin.h (+7-11)
  • (modified) clang/lib/Headers/avx512fintrin.h (+27-38)
  • (modified) clang/lib/Headers/avx512vlbwintrin.h (+14-21)
  • (modified) clang/lib/Headers/avx512vlintrin.h (+22-32)
  • (modified) clang/test/CodeGen/X86/avx512bw-builtins.c (+11-2)
  • (modified) clang/test/CodeGen/X86/avx512f-builtins.c (+46-10)
  • (modified) clang/test/CodeGen/X86/avx512vl-builtins.c (+37)
  • (modified) clang/test/CodeGen/X86/avx512vlbw-builtins.c (+21-4)
diff --git a/clang/include/clang/Basic/BuiltinsX86.td b/clang/include/clang/Basic/BuiltinsX86.td
index cb08e2107f072..b261b681990e0 100644
--- a/clang/include/clang/Basic/BuiltinsX86.td
+++ b/clang/include/clang/Basic/BuiltinsX86.td
@@ -603,6 +603,11 @@ let Features = "avx", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWid
   def vec_set_v8si : X86Builtin<"_Vector<8, int>(_Vector<8, int>, int, _Constant int)">;
 }
 
+let Features = "avx2", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<256>] in {
+ def permvarsi256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<8, int>)">;
+ def permvarsf256 : X86Builtin<"_Vector<8, float>(_Vector<8, float>, _Vector<8, int>)">;
+}
+
 let Features = "avx2", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in {
   def mpsadbw256 : X86Builtin<"_Vector<32, char>(_Vector<32, char>, _Vector<32, char>, _Constant char)">;
 
@@ -617,9 +622,7 @@ let Features = "avx2", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] i
   def psrlw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<8, short>)">;
   def psrld256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<4, int>)">;
   def psrlq256 : X86Builtin<"_Vector<4, long long int>(_Vector<4, long long int>, _Vector<2, long long int>)">;
-  def permvarsi256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<8, int>)">;
   def permdf256 : X86Builtin<"_Vector<4, double>(_Vector<4, double>, _Constant int)">;
-  def permvarsf256 : X86Builtin<"_Vector<8, float>(_Vector<8, float>, _Vector<8, int>)">;
   def permti256 : X86Builtin<"_Vector<4, long long int>(_Vector<4, long long int>, _Vector<4, long long int>, _Constant int)">;
   def permdi256 : X86Builtin<"_Vector<4, long long int>(_Vector<4, long long int>, _Constant int)">;
 }
@@ -3052,38 +3055,38 @@ let Features = "avx512f", Attributes = [NoThrow, Const, RequiredVectorWidth<512>
   def permdi512 : X86Builtin<"_Vector<8, long long int>(_Vector<8, long long int>, _Constant int)">;
 }
 
-let Features = "avx512bw", Attributes = [NoThrow, Const, RequiredVectorWidth<512>] in {
+let Features = "avx512bw", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<512>] in {
   def permvarhi512 : X86Builtin<"_Vector<32, short>(_Vector<32, short>, _Vector<32, short>)">;
 }
 
-let Features = "avx512f", Attributes = [NoThrow, Const, RequiredVectorWidth<512>] in {
+let Features = "avx512f", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<512>] in {
   def permvardf512 : X86Builtin<"_Vector<8, double>(_Vector<8, double>, _Vector<8, long long int>)">;
   def permvardi512 : X86Builtin<"_Vector<8, long long int>(_Vector<8, long long int>, _Vector<8, long long int>)">;
   def permvarsf512 : X86Builtin<"_Vector<16, float>(_Vector<16, float>, _Vector<16, int>)">;
   def permvarsi512 : X86Builtin<"_Vector<16, int>(_Vector<16, int>, _Vector<16, int>)">;
 }
 
-let Features = "avx512vbmi", Attributes = [NoThrow, Const, RequiredVectorWidth<512>] in {
+let Features = "avx512vbmi", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<512>] in {
   def permvarqi512 : X86Builtin<"_Vector<64, char>(_Vector<64, char>, _Vector<64, char>)">;
 }
 
-let Features = "avx512vbmi,avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<128>] in {
+let Features = "avx512vbmi,avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in {
   def permvarqi128 : X86Builtin<"_Vector<16, char>(_Vector<16, char>, _Vector<16, char>)">;
 }
 
-let Features = "avx512vbmi,avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in {
+let Features = "avx512vbmi,avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<256>] in {
   def permvarqi256 : X86Builtin<"_Vector<32, char>(_Vector<32, char>, _Vector<32, char>)">;
 }
 
-let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<128>] in {
+let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in {
   def permvarhi128 : X86Builtin<"_Vector<8, short>(_Vector<8, short>, _Vector<8, short>)">;
 }
 
-let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in {
+let Features = "avx512bw,avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<256>] in {
   def permvarhi256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">;
 }
 
-let Features = "avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in {
+let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<256>] in {
   def permvardf256 : X86Builtin<"_Vector<4, double>(_Vector<4, double>, _Vector<4, long long int>)">;
   def permvardi256 : X86Builtin<"_Vector<4, long long int>(_Vector<4, long long int>, _Vector<4, long long int>)">;
 }
diff --git a/clang/lib/AST/ByteCode/InterpBuiltin.cpp b/clang/lib/AST/ByteCode/InterpBuiltin.cpp
index 6c7b2f502cc51..c72a3566681b1 100644
--- a/clang/lib/AST/ByteCode/InterpBuiltin.cpp
+++ b/clang/lib/AST/ByteCode/InterpBuiltin.cpp
@@ -4414,6 +4414,50 @@ bool InterpretBuiltin(InterpState &S, CodePtr OpPC, const CallExpr *Call,
             return std::pair<unsigned, int>{0, static_cast<int>(DstIdx)};
           }
         });
+  case X86::BI__builtin_ia32_permvarsi256:
+  case X86::BI__builtin_ia32_permvarsf256:
+  case X86::BI__builtin_ia32_permvardf512:
+  case X86::BI__builtin_ia32_permvardi512:
+  case X86::BI__builtin_ia32_permvarhi128:
+    return interp__builtin_ia32_shuffle_generic(
+        S, OpPC, Call, [](unsigned DstIdx, unsigned ShuffleMask) {
+          int Offset = ShuffleMask & 0x7;
+          unsigned SrcIdx = 0;
+          return std::pair<unsigned, int>{SrcIdx, Offset};
+        });
+  case X86::BI__builtin_ia32_permvarqi128:
+  case X86::BI__builtin_ia32_permvarhi256:
+  case X86::BI__builtin_ia32_permvarsi512:
+  case X86::BI__builtin_ia32_permvarsf512:
+    return interp__builtin_ia32_shuffle_generic(
+        S, OpPC, Call, [](unsigned DstIdx, unsigned ShuffleMask) {
+          int Offset = ShuffleMask & 0xF;
+          unsigned SrcIdx = 0;
+          return std::pair<unsigned, int>{SrcIdx, Offset};
+        });
+  case X86::BI__builtin_ia32_permvardi256:
+  case X86::BI__builtin_ia32_permvardf256:
+    return interp__builtin_ia32_shuffle_generic(
+        S, OpPC, Call, [](unsigned DstIdx, unsigned ShuffleMask) {
+          int Offset = ShuffleMask & 0x3;
+          unsigned SrcIdx = 0;
+          return std::pair<unsigned, int>{SrcIdx, Offset};
+        });
+  case X86::BI__builtin_ia32_permvarqi256:
+  case X86::BI__builtin_ia32_permvarhi512:
+    return interp__builtin_ia32_shuffle_generic(
+        S, OpPC, Call, [](unsigned DstIdx, unsigned ShuffleMask) {
+          int Offset = ShuffleMask & 0x1F;
+          unsigned SrcIdx = 0;
+          return std::pair<unsigned, int>{SrcIdx, Offset};
+        });
+  case X86::BI__builtin_ia32_permvarqi512:
+    return interp__builtin_ia32_shuffle_generic(
+        S, OpPC, Call, [](unsigned DstIdx, unsigned ShuffleMask) {
+          int Offset = ShuffleMask & 0x3F;
+          unsigned SrcIdx = 0;
+          return std::pair<unsigned, int>{SrcIdx, Offset};
+        });
   case X86::BI__builtin_ia32_vpermi2varq128:
   case X86::BI__builtin_ia32_vpermi2varpd128:
     return interp__builtin_ia32_shuffle_generic(
diff --git a/clang/lib/AST/ExprConstant.cpp b/clang/lib/AST/ExprConstant.cpp
index 1bfea24b228e8..e9e448143477e 100644
--- a/clang/lib/AST/ExprConstant.cpp
+++ b/clang/lib/AST/ExprConstant.cpp
@@ -13551,6 +13551,70 @@ bool VectorExprEvaluator::VisitCallExpr(const CallExpr *E) {
       return false;
     return Success(R, E);
   }
+  case X86::BI__builtin_ia32_permvarsi256:
+  case X86::BI__builtin_ia32_permvarsf256:
+  case X86::BI__builtin_ia32_permvardf512:
+  case X86::BI__builtin_ia32_permvardi512:
+  case X86::BI__builtin_ia32_permvarhi128: {
+    APValue R;
+    if (!evalShuffleGeneric(Info, E, R,
+                            [](unsigned DstIdx, unsigned ShuffleMask) {
+                              int Offset = ShuffleMask & 0x7;
+                              unsigned SrcIdx = 0;
+                              return std::pair<unsigned, int>{SrcIdx, Offset};
+                            }))
+      return false;
+    return Success(R, E);
+  }
+  case X86::BI__builtin_ia32_permvarqi128:
+  case X86::BI__builtin_ia32_permvarhi256:
+  case X86::BI__builtin_ia32_permvarsi512:
+  case X86::BI__builtin_ia32_permvarsf512: {
+    APValue R;
+    if (!evalShuffleGeneric(Info, E, R,
+                            [](unsigned DstIdx, unsigned ShuffleMask) {
+                              int Offset = ShuffleMask & 0xF;
+                              unsigned SrcIdx = 0;
+                              return std::pair<unsigned, int>{SrcIdx, Offset};
+                            }))
+      return false;
+    return Success(R, E);
+  }
+  case X86::BI__builtin_ia32_permvardi256:
+  case X86::BI__builtin_ia32_permvardf256: {
+    APValue R;
+    if (!evalShuffleGeneric(Info, E, R,
+                            [](unsigned DstIdx, unsigned ShuffleMask) {
+                              int Offset = ShuffleMask & 0x3;
+                              unsigned SrcIdx = 0;
+                              return std::pair<unsigned, int>{SrcIdx, Offset};
+                            }))
+      return false;
+    return Success(R, E);
+  }
+  case X86::BI__builtin_ia32_permvarqi256:
+  case X86::BI__builtin_ia32_permvarhi512: {
+    APValue R;
+    if (!evalShuffleGeneric(Info, E, R,
+                            [](unsigned DstIdx, unsigned ShuffleMask) {
+                              int Offset = ShuffleMask & 0x1F;
+                              unsigned SrcIdx = 0;
+                              return std::pair<unsigned, int>{SrcIdx, Offset};
+                            }))
+      return false;
+    return Success(R, E);
+  }
+  case X86::BI__builtin_ia32_permvarqi512: {
+    APValue R;
+    if (!evalShuffleGeneric(Info, E, R,
+                            [](unsigned DstIdx, unsigned ShuffleMask) {
+                              int Offset = ShuffleMask & 0x3F;
+                              unsigned SrcIdx = 0;
+                              return std::pair<unsigned, int>{SrcIdx, Offset};
+                            }))
+      return false;
+    return Success(R, E);
+  }
   case X86::BI__builtin_ia32_vpermi2varq128:
   case X86::BI__builtin_ia32_vpermi2varpd128: {
     APValue R;
diff --git a/clang/lib/Headers/avx10_2_512bf16intrin.h b/clang/lib/Headers/avx10_2_512bf16intrin.h
index 46ec12a63ef9c..3201307af4731 100644
--- a/clang/lib/Headers/avx10_2_512bf16intrin.h
+++ b/clang/lib/Headers/avx10_2_512bf16intrin.h
@@ -179,7 +179,7 @@ _mm512_permutex2var_pbh(__m512bh __A, __m512i __I, __m512bh __B) {
                                                   (__v32hi)__B);
 }
 
-static __inline__ __m512bh __DEFAULT_FN_ATTRS512
+static __inline__ __m512bh __DEFAULT_FN_ATTRS512_CONSTEXPR
 _mm512_permutexvar_pbh(__m512i __A, __m512bh __B) {
   return (__m512bh)__builtin_ia32_permvarhi512((__v32hi)__B, (__v32hi)__A);
 }
diff --git a/clang/lib/Headers/avx10_2bf16intrin.h b/clang/lib/Headers/avx10_2bf16intrin.h
index 8fb8cd7cd0865..9f5b726d7b789 100644
--- a/clang/lib/Headers/avx10_2bf16intrin.h
+++ b/clang/lib/Headers/avx10_2bf16intrin.h
@@ -307,12 +307,12 @@ _mm256_permutex2var_pbh(__m256bh __A, __m256i __I, __m256bh __B) {
                                                   (__v16hi)__B);
 }
 
-static __inline__ __m128bh __DEFAULT_FN_ATTRS128
+static __inline__ __m128bh __DEFAULT_FN_ATTRS128_CONSTEXPR
 _mm_permutexvar_pbh(__m128i __A, __m128bh __B) {
   return (__m128bh)__builtin_ia32_permvarhi128((__v8hi)__B, (__v8hi)__A);
 }
 
-static __inline__ __m256bh __DEFAULT_FN_ATTRS256
+static __inline__ __m256bh __DEFAULT_FN_ATTRS256_CONSTEXPR
 _mm256_permutexvar_pbh(__m256i __A, __m256bh __B) {
   return (__m256bh)__builtin_ia32_permvarhi256((__v16hi)__B, (__v16hi)__A);
 }
diff --git a/clang/lib/Headers/avx2intrin.h b/clang/lib/Headers/avx2intrin.h
index 3cbaaece7b38e..3e3c13d8bd662 100644
--- a/clang/lib/Headers/avx2intrin.h
+++ b/clang/lib/Headers/avx2intrin.h
@@ -3214,9 +3214,8 @@ _mm_broadcastq_epi64(__m128i __X) {
 ///    A 256-bit vector of [8 x i32] containing indexes of values to use from
 ///    \a __a.
 /// \returns A 256-bit vector of [8 x i32] containing the result.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_permutevar8x32_epi32(__m256i __a, __m256i __b)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_permutevar8x32_epi32(__m256i __a, __m256i __b) {
   return (__m256i)__builtin_ia32_permvarsi256((__v8si)__a, (__v8si)__b);
 }
 
@@ -3272,9 +3271,8 @@ _mm256_permutevar8x32_epi32(__m256i __a, __m256i __b)
 ///    A 256-bit vector of [8 x i32] containing indexes of values to use from
 ///    \a __a.
 /// \returns A 256-bit vector of [8 x float] containing the result.
-static __inline__ __m256 __DEFAULT_FN_ATTRS256
-_mm256_permutevar8x32_ps(__m256 __a, __m256i __b)
-{
+static __inline__ __m256 __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_permutevar8x32_ps(__m256 __a, __m256i __b) {
   return (__m256)__builtin_ia32_permvarsf256((__v8sf)__a, (__v8si)__b);
 }
 
diff --git a/clang/lib/Headers/avx512bwintrin.h b/clang/lib/Headers/avx512bwintrin.h
index 4a02c96620335..3cfa32eb9e727 100644
--- a/clang/lib/Headers/avx512bwintrin.h
+++ b/clang/lib/Headers/avx512bwintrin.h
@@ -1846,25 +1846,21 @@ _mm512_maskz_broadcastw_epi16 (__mmask32 __M, __m128i __A)
                                              (__v32hi) _mm512_setzero_si512());
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_permutexvar_epi16 (__m512i __A, __m512i __B)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_permutexvar_epi16(__m512i __A, __m512i __B) {
   return (__m512i)__builtin_ia32_permvarhi512((__v32hi)__B, (__v32hi)__A);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_maskz_permutexvar_epi16 (__mmask32 __M, __m512i __A,
-        __m512i __B)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_permutexvar_epi16(__mmask32 __M, __m512i __A, __m512i __B) {
   return (__m512i)__builtin_ia32_selectw_512((__mmask32)__M,
                                     (__v32hi)_mm512_permutexvar_epi16(__A, __B),
                                     (__v32hi)_mm512_setzero_si512());
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_mask_permutexvar_epi16 (__m512i __W, __mmask32 __M, __m512i __A,
-             __m512i __B)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_permutexvar_epi16(__m512i __W, __mmask32 __M, __m512i __A,
+                              __m512i __B) {
   return (__m512i)__builtin_ia32_selectw_512((__mmask32)__M,
                                     (__v32hi)_mm512_permutexvar_epi16(__A, __B),
                                     (__v32hi)__W);
diff --git a/clang/lib/Headers/avx512fintrin.h b/clang/lib/Headers/avx512fintrin.h
index 997e9608e112f..79c37173ac838 100644
--- a/clang/lib/Headers/avx512fintrin.h
+++ b/clang/lib/Headers/avx512fintrin.h
@@ -7959,93 +7959,82 @@ _mm_mask3_fnmsub_sd (__m128d __W, __m128d __X, __m128d __Y, __mmask8 __U)
                                        (__v8di)_mm512_permutex_epi64((X), (C)), \
                                        (__v8di)_mm512_setzero_si512()))
 
-static __inline__ __m512d __DEFAULT_FN_ATTRS512
-_mm512_permutexvar_pd (__m512i __X, __m512d __Y)
-{
+static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_permutexvar_pd(__m512i __X, __m512d __Y) {
   return (__m512d)__builtin_ia32_permvardf512((__v8df) __Y, (__v8di) __X);
 }
 
-static __inline__ __m512d __DEFAULT_FN_ATTRS512
-_mm512_mask_permutexvar_pd (__m512d __W, __mmask8 __U, __m512i __X, __m512d __Y)
-{
+static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_permutexvar_pd(__m512d __W, __mmask8 __U, __m512i __X,
+                           __m512d __Y) {
   return (__m512d)__builtin_ia32_selectpd_512((__mmask8)__U,
                                         (__v8df)_mm512_permutexvar_pd(__X, __Y),
                                         (__v8df)__W);
 }
 
-static __inline__ __m512d __DEFAULT_FN_ATTRS512
-_mm512_maskz_permutexvar_pd (__mmask8 __U, __m512i __X, __m512d __Y)
-{
+static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_permutexvar_pd(__mmask8 __U, __m512i __X, __m512d __Y) {
   return (__m512d)__builtin_ia32_selectpd_512((__mmask8)__U,
                                         (__v8df)_mm512_permutexvar_pd(__X, __Y),
                                         (__v8df)_mm512_setzero_pd());
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_permutexvar_epi64 (__m512i __X, __m512i __Y)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_permutexvar_epi64(__m512i __X, __m512i __Y) {
   return (__m512i)__builtin_ia32_permvardi512((__v8di)__Y, (__v8di)__X);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_maskz_permutexvar_epi64 (__mmask8 __M, __m512i __X, __m512i __Y)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_permutexvar_epi64(__mmask8 __M, __m512i __X, __m512i __Y) {
   return (__m512i)__builtin_ia32_selectq_512((__mmask8)__M,
                                      (__v8di)_mm512_permutexvar_epi64(__X, __Y),
                                      (__v8di)_mm512_setzero_si512());
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_mask_permutexvar_epi64 (__m512i __W, __mmask8 __M, __m512i __X,
-             __m512i __Y)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_permutexvar_epi64(__m512i __W, __mmask8 __M, __m512i __X,
+                              __m512i __Y) {
   return (__m512i)__builtin_ia32_selectq_512((__mmask8)__M,
                                      (__v8di)_mm512_permutexvar_epi64(__X, __Y),
                                      (__v8di)__W);
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_permutexvar_ps (__m512i __X, __m512 __Y)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_permutexvar_ps(__m512i __X, __m512 __Y) {
   return (__m512)__builtin_ia32_permvarsf512((__v16sf)__Y, (__v16si)__X);
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_mask_permutexvar_ps (__m512 __W, __mmask16 __U, __m512i __X, __m512 __Y)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_permutexvar_ps(__m512 __W, __mmask16 __U, __m512i __X, __m512 __Y) {
   return (__m512)__builtin_ia32_selectps_512((__mmask16)__U,
                                        (__v16sf)_mm512_permutexvar_ps(__X, __Y),
                                        (__v16sf)__W);
 }
 
-static __inline__ __m512 __DEFAULT_FN_ATTRS512
-_mm512_maskz_permutexvar_ps (__mmask16 __U, __m512i __X, __m512 __Y)
-{
+static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_permutexvar_ps(__mmask16 __U, __m512i __X, __m512 __Y) {
   return (__m512)__builtin_ia32_selectps_512((__mmask16)__U,
                                        (__v16sf)_mm512_permutexvar_ps(__X, __Y),
                                        (__v16sf)_mm512_setzero_ps());
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_permutexvar_epi32 (__m512i __X, __m512i __Y)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_permutexvar_epi32(__m512i __X, __m512i __Y) {
   return (__m512i)__builtin_ia32_permvarsi512((__v16si)__Y, (__v16si)__X);
 }
 
 #define _mm512_permutevar_epi32 _mm512_permutexvar_epi32
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_maskz_permutexvar_epi32 (__mmask16 __M, __m512i __X, __m512i __Y)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_permutexvar_epi32(__mmask16 __M, __m512i __X, __m512i __Y) {
   return (__m512i)__builtin_ia32_selectd_512((__mmask16)__M,
                                     (__v16si)_mm512_permutexvar_epi32(__X, __Y),
                                     (__v16si)_mm512_setzero_si512());
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_mask_permutexvar_epi32 (__m512i __W, __mmask16 __M, __m512i __X,
-             __m512i __Y)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_permutexvar_epi32(__m512i __W, __mmask16 __M, __m512i __X,
+                              __m512i __Y) {
   return (__m512i)__builtin_ia32_selectd_512((__mmask16)__M,
                                     (__v16si)_mm512_permutexvar_epi32(__...
[truncated]

RKSimon
RKSimon requested changes Nov 13, 2025
View reviewed changes
RKSimon
RKSimon approved these changes Nov 13, 2025
View reviewed changes
Copy link
Collaborator

@RKSimon RKSimon left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM - cheers

RKSimon
RKSimon requested changes Nov 13, 2025
View reviewed changes
Copy link
Collaborator

@RKSimon RKSimon left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

sorry - you missed the avx2 tests

clang/lib/Headers/avx2intrin.h
_mm256_permutevar8x32_epi32(__m256i __a, __m256i __b)
{
static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_permutevar8x32_epi32(__m256i __a, __m256i __b) {
Copy link
Collaborator

@RKSimon RKSimon Nov 13, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

test coverage?

@chaitanyav
[Clang] VectorExprEvaluator::VisitCallExpr / InterpretBuiltin - Allow…
a561197 
… AVX512 permutexvar intrinsics to be used in constexpr

Resolves: llvm#167476
@chaitanyav
Address code review comments
6752579 
- Group permvarsi256/permvarsf256 with other AVX2 constexpr builtins
- Remove unnecessary SrcIdx variable and use zero directly in pair
  construction
@chaitanyav
Add tests for permutevar8x32, permutevar8x32_epi32
3502a04
RKSimon
RKSimon approved these changes Nov 13, 2025
View reviewed changes
Copy link
Collaborator

@RKSimon RKSimon left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM - cheers

@RKSimon RKSimon enabled auto-merge (squash) November 13, 2025 13:38
@RKSimon RKSimon mentioned this pull request Nov 13, 2025
Open
@RKSimon RKSimon merged commit 91a1bde into llvm:main Nov 13, 2025
9 of 10 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@RKSimon RKSimon RKSimon approved these changes

@FreddyLeaf FreddyLeaf Awaiting requested review from FreddyLeaf

@tbaederr tbaederr Awaiting requested review from tbaederr

Assignees

@chaitanyav chaitanyav

Labels

backend:X86 clang:bytecode Issues for the clang bytecode constexpr interpreter clang:frontend Language frontend issues, e.g. anything involving "Sema" clang:headers Headers provided by Clang, e.g. for intrinsics clang Clang issues not falling into any other category

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

[Clang] VectorExprEvaluator::VisitCallExpr / InterpretBuiltin - allow AVX512 permutexvar intrinsics to be used in constexpr

3 participants

@chaitanyav @llvmbot @RKSimon