[X86] Rename __builtin_ia32_pslldqi128 to __builtin_ia32_pslldqi128_byteshift and...

The previous names took the shift amount in bits to match gcc and required a multiply by 8 in the header. This creates a misleading error message when we check the range of the immediate to the builtin since the allowed range also got multiplied by 8.

This commit changes the builtins to use a byte shift amount to match the underlying instruction and the Intel intrinsic.

Fixes the remaining issue from PR37795.

llvm-svn: 334773

diff --git a/clang/include/clang/Basic/BuiltinsX86.def b/clang/include/clang/Basic/BuiltinsX86.def
index 06f56c7..7cb7c46 100644 (file)
--- a/clang/include/clang/Basic/BuiltinsX86.def
+++ b/clang/include/clang/Basic/BuiltinsX86.def
@@ -363,8 +363,8 @@ TARGET_BUILTIN(__builtin_ia32_psrlqi128, "V2LLiV2LLii", "nc", "sse2")
 TARGET_BUILTIN(__builtin_ia32_psrawi128, "V8sV8si", "nc", "sse2")
 TARGET_BUILTIN(__builtin_ia32_psradi128, "V4iV4ii", "nc", "sse2")
 TARGET_BUILTIN(__builtin_ia32_pmaddwd128, "V4iV8sV8s", "nc", "sse2")
-TARGET_BUILTIN(__builtin_ia32_pslldqi128, "V2LLiV2LLiIi", "nc", "sse2")
-TARGET_BUILTIN(__builtin_ia32_psrldqi128, "V2LLiV2LLiIi", "nc", "sse2")
+TARGET_BUILTIN(__builtin_ia32_pslldqi128_byteshift, "V2LLiV2LLiIi", "nc", "sse2")
+TARGET_BUILTIN(__builtin_ia32_psrldqi128_byteshift, "V2LLiV2LLiIi", "nc", "sse2")
 
 TARGET_BUILTIN(__builtin_ia32_monitor, "vv*UiUi", "n", "sse3")
 TARGET_BUILTIN(__builtin_ia32_mwait, "vUiUi", "n", "sse3")
@@ -613,7 +613,7 @@ TARGET_BUILTIN(__builtin_ia32_psignw256, "V16sV16sV16s", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_psignd256, "V8iV8iV8i", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_psllwi256, "V16sV16si", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_psllw256, "V16sV16sV8s", "nc", "avx2")
-TARGET_BUILTIN(__builtin_ia32_pslldqi256, "V4LLiV4LLiIi", "nc", "avx2")
+TARGET_BUILTIN(__builtin_ia32_pslldqi256_byteshift, "V4LLiV4LLiIi", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_pslldi256, "V8iV8ii", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_pslld256, "V8iV8iV4i", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_psllqi256, "V4LLiV4LLii", "nc", "avx2")
@@ -622,7 +622,7 @@ TARGET_BUILTIN(__builtin_ia32_psrawi256, "V16sV16si", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_psraw256, "V16sV16sV8s", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_psradi256, "V8iV8ii", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_psrad256, "V8iV8iV4i", "nc", "avx2")
-TARGET_BUILTIN(__builtin_ia32_psrldqi256, "V4LLiV4LLiIi", "nc", "avx2")
+TARGET_BUILTIN(__builtin_ia32_psrldqi256_byteshift, "V4LLiV4LLiIi", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_psrlwi256, "V16sV16si", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_psrlw256, "V16sV16sV8s", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_psrldi256, "V8iV8ii", "nc", "avx2")
@@ -1399,8 +1399,8 @@ TARGET_BUILTIN(__builtin_ia32_psraw512, "V32sV32sV8s", "nc", "avx512bw")
 TARGET_BUILTIN(__builtin_ia32_psrawi512, "V32sV32si", "nc", "avx512bw")
 TARGET_BUILTIN(__builtin_ia32_psrlw512, "V32sV32sV8s", "nc", "avx512bw")
 TARGET_BUILTIN(__builtin_ia32_psrlwi512, "V32sV32si", "nc", "avx512bw")
-TARGET_BUILTIN(__builtin_ia32_pslldqi512, "V8LLiV8LLiIi", "nc", "avx512bw")
-TARGET_BUILTIN(__builtin_ia32_psrldqi512, "V8LLiV8LLiIi", "nc", "avx512bw")
+TARGET_BUILTIN(__builtin_ia32_pslldqi512_byteshift, "V8LLiV8LLiIi", "nc", "avx512bw")
+TARGET_BUILTIN(__builtin_ia32_psrldqi512_byteshift, "V8LLiV8LLiIi", "nc", "avx512bw")
 TARGET_BUILTIN(__builtin_ia32_movdqa32load128_mask, "V4iV4i*V4iUc", "n", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_movdqa32load256_mask, "V8iV8i*V8iUc", "n", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_movdqa32load512_mask, "V16iV16iC*V16iUs", "n", "avx512f")

diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index 08175b3..2393957 100644 (file)
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -9690,11 +9690,10 @@ Value *CodeGenFunction::EmitX86BuiltinExpr(unsigned BuiltinID,
                                        "vperm");
   }
 
-  case X86::BI__builtin_ia32_pslldqi128:
-  case X86::BI__builtin_ia32_pslldqi256:
-  case X86::BI__builtin_ia32_pslldqi512: {
-    // Shift value is in bits so divide by 8.
-    unsigned ShiftVal = cast<llvm::ConstantInt>(Ops[1])->getZExtValue() >> 3;
+  case X86::BI__builtin_ia32_pslldqi128_byteshift:
+  case X86::BI__builtin_ia32_pslldqi256_byteshift:
+  case X86::BI__builtin_ia32_pslldqi512_byteshift: {
+    unsigned ShiftVal = cast<llvm::ConstantInt>(Ops[1])->getZExtValue();
     llvm::Type *ResultType = Ops[0]->getType();
     // Builtin type is vXi64 so multiply by 8 to get bytes.
     unsigned NumElts = ResultType->getVectorNumElements() * 8;
@@ -9721,11 +9720,10 @@ Value *CodeGenFunction::EmitX86BuiltinExpr(unsigned BuiltinID,
                                             "pslldq");
     return Builder.CreateBitCast(SV, Ops[0]->getType(), "cast");
   }
-  case X86::BI__builtin_ia32_psrldqi128:
-  case X86::BI__builtin_ia32_psrldqi256:
-  case X86::BI__builtin_ia32_psrldqi512: {
-    // Shift value is in bits so divide by 8.
-    unsigned ShiftVal = cast<llvm::ConstantInt>(Ops[1])->getZExtValue() >> 3;
+  case X86::BI__builtin_ia32_psrldqi128_byteshift:
+  case X86::BI__builtin_ia32_psrldqi256_byteshift:
+  case X86::BI__builtin_ia32_psrldqi512_byteshift: {
+    unsigned ShiftVal = cast<llvm::ConstantInt>(Ops[1])->getZExtValue();
     llvm::Type *ResultType = Ops[0]->getType();
     // Builtin type is vXi64 so multiply by 8 to get bytes.
     unsigned NumElts = ResultType->getVectorNumElements() * 8;

diff --git a/clang/lib/Headers/avx2intrin.h b/clang/lib/Headers/avx2intrin.h
index 7af494a..22c00f9 100644 (file)
--- a/clang/lib/Headers/avx2intrin.h
+++ b/clang/lib/Headers/avx2intrin.h
@@ -515,10 +515,10 @@ _mm256_sign_epi32(__m256i __a, __m256i __b)
 }
 
 #define _mm256_slli_si256(a, imm) \
-  (__m256i)__builtin_ia32_pslldqi256((__v4di)(__m256i)(a), (int)(imm) * 8)
+  (__m256i)__builtin_ia32_pslldqi256_byteshift((__v4di)(__m256i)(a), (int)(imm))
 
 #define _mm256_bslli_epi128(a, imm) \
-  (__m256i)__builtin_ia32_pslldqi256((__v4di)(__m256i)(a), (int)(imm) * 8)
+  (__m256i)__builtin_ia32_pslldqi256_byteshift((__v4di)(__m256i)(a), (int)(imm))
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_slli_epi16(__m256i __a, int __count)
@@ -581,10 +581,10 @@ _mm256_sra_epi32(__m256i __a, __m128i __count)
 }
 
 #define _mm256_srli_si256(a, imm) \
-  (__m256i)__builtin_ia32_psrldqi256((__m256i)(a), (int)(imm) * 8)
+  (__m256i)__builtin_ia32_psrldqi256_byteshift((__m256i)(a), (int)(imm))
 
 #define _mm256_bsrli_epi128(a, imm) \
-  (__m256i)__builtin_ia32_psrldqi256((__m256i)(a), (int)(imm) * 8)
+  (__m256i)__builtin_ia32_psrldqi256_byteshift((__m256i)(a), (int)(imm))
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_srli_epi16(__m256i __a, int __count)

diff --git a/clang/lib/Headers/avx512bwintrin.h b/clang/lib/Headers/avx512bwintrin.h
index 471ec38..444df35 100644 (file)
--- a/clang/lib/Headers/avx512bwintrin.h
+++ b/clang/lib/Headers/avx512bwintrin.h
@@ -1387,7 +1387,7 @@ _mm512_maskz_slli_epi16(__mmask32 __U, __m512i __A, int __B)
 }
 
 #define _mm512_bslli_epi128(a, imm) \
-  (__m512i)__builtin_ia32_pslldqi512((__v8di)(__m512i)(a), (int)(imm) * 8)
+  (__m512i)__builtin_ia32_pslldqi512_byteshift((__v8di)(__m512i)(a), (int)(imm))
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_srlv_epi16(__m512i __A, __m512i __B)
@@ -1522,7 +1522,7 @@ _mm512_maskz_srli_epi16(__mmask32 __U, __m512i __A, int __B)
 }
 
 #define _mm512_bsrli_epi128(a, imm) \
-  (__m512i)__builtin_ia32_psrldqi512((__v8di)(__m512i)(a), (int)(imm) * 8)
+  (__m512i)__builtin_ia32_psrldqi512_byteshift((__v8di)(__m512i)(a), (int)(imm))
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_mov_epi16 (__m512i __W, __mmask32 __U, __m512i __A)

diff --git a/clang/lib/Headers/emmintrin.h b/clang/lib/Headers/emmintrin.h
index 6c64d41..df98a81 100644 (file)
--- a/clang/lib/Headers/emmintrin.h
+++ b/clang/lib/Headers/emmintrin.h
@@ -2794,10 +2794,10 @@ _mm_xor_si128(__m128i __a, __m128i __b)
 ///    \a a.
 /// \returns A 128-bit integer vector containing the left-shifted value.
 #define _mm_slli_si128(a, imm) \
-  (__m128i)__builtin_ia32_pslldqi128((__v2di)(__m128i)(a), (int)(imm) * 8)
+  (__m128i)__builtin_ia32_pslldqi128_byteshift((__v2di)(__m128i)(a), (int)(imm))
 
 #define _mm_bslli_si128(a, imm) \
-  (__m128i)__builtin_ia32_pslldqi128((__v2di)(__m128i)(a), (int)(imm) * 8)
+  (__m128i)__builtin_ia32_pslldqi128_byteshift((__v2di)(__m128i)(a), (int)(imm))
 
 /// Left-shifts each 16-bit value in the 128-bit integer vector operand
 ///    by the specified number of bits. Low-order bits are cleared.
@@ -3011,10 +3011,10 @@ _mm_sra_epi32(__m128i __a, __m128i __count)
 ///    \a a.
 /// \returns A 128-bit integer vector containing the right-shifted value.
 #define _mm_srli_si128(a, imm) \
-  (__m128i)__builtin_ia32_psrldqi128((__v2di)(__m128i)(a), (int)(imm) * 8)
+  (__m128i)__builtin_ia32_psrldqi128_byteshift((__v2di)(__m128i)(a), (int)(imm))
 
 #define _mm_bsrli_si128(a, imm) \
-  (__m128i)__builtin_ia32_psrldqi128((__v2di)(__m128i)(a), (int)(imm) * 8)
+  (__m128i)__builtin_ia32_psrldqi128_byteshift((__v2di)(__m128i)(a), (int)(imm))
 
 /// Right-shifts each of 16-bit values in the 128-bit integer vector
 ///    operand by the specified number of bits. High-order bits are cleared.

diff --git a/clang/lib/Sema/SemaChecking.cpp b/clang/lib/Sema/SemaChecking.cpp
index dfe4373..9de512c 100644 (file)
--- a/clang/lib/Sema/SemaChecking.cpp
+++ b/clang/lib/Sema/SemaChecking.cpp
@@ -2821,6 +2821,12 @@ bool Sema::CheckX86BuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) {
   case X86::BI__builtin_ia32_fpclasspd512_mask:
   case X86::BI__builtin_ia32_fpclasssd_mask:
   case X86::BI__builtin_ia32_fpclassss_mask:
+  case X86::BI__builtin_ia32_pslldqi128_byteshift:
+  case X86::BI__builtin_ia32_pslldqi256_byteshift:
+  case X86::BI__builtin_ia32_pslldqi512_byteshift:
+  case X86::BI__builtin_ia32_psrldqi128_byteshift:
+  case X86::BI__builtin_ia32_psrldqi256_byteshift:
+  case X86::BI__builtin_ia32_psrldqi512_byteshift:
     i = 1; l = 0; u = 255;
     break;
   case X86::BI__builtin_ia32_vperm2f128_pd256:
@@ -2917,14 +2923,6 @@ bool Sema::CheckX86BuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) {
   case X86::BI__builtin_ia32_rndscaless_round_mask:
     i = 4; l = 0; u = 255;
     break;
-  case X86::BI__builtin_ia32_pslldqi128:
-  case X86::BI__builtin_ia32_pslldqi256:
-  case X86::BI__builtin_ia32_pslldqi512:
-  case X86::BI__builtin_ia32_psrldqi128:
-  case X86::BI__builtin_ia32_psrldqi256:
-  case X86::BI__builtin_ia32_psrldqi512:
-    i = 1; l = 0; u = 2047;
-    break;
   }
   return SemaBuiltinConstantArgRange(TheCall, i, l, u);
 }