Handle target builtin options that are all required rather than

only one of a group of possibilities.

This changes the syntax in the builtin files to represent:

, as the and operator
| as the or operator

The former syntax matches how the backend tablegen files represent
multiple subtarget features being required.

Updated the builtin and intrinsic headers accordingly for the new
syntax.

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@251388 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/include/clang/Basic/BuiltinsX86.def b/include/clang/Basic/BuiltinsX86.def
index 42948469b7011070e5a3577f6b232408a74e1d05..c77b2ba4d750edafc53fc43fe53bd56ed56ba3d7 100644 (file)
--- a/include/clang/Basic/BuiltinsX86.def
+++ b/include/clang/Basic/BuiltinsX86.def
@@ -718,38 +718,38 @@ TARGET_BUILTIN(__builtin_ia32_sha256msg1, "V4iV4iV4i", "", "sha")
 TARGET_BUILTIN(__builtin_ia32_sha256msg2, "V4iV4iV4i", "", "sha")
 
 // FMA
-TARGET_BUILTIN(__builtin_ia32_vfmaddps, "V4fV4fV4fV4f", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmaddpd, "V2dV2dV2dV2d", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmaddss, "V4fV4fV4fV4f", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmaddsd, "V2dV2dV2dV2d", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubps, "V4fV4fV4fV4f", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubpd, "V2dV2dV2dV2d", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubss, "V4fV4fV4fV4f", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubsd, "V2dV2dV2dV2d", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmaddps, "V4fV4fV4fV4f", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmaddpd, "V2dV2dV2dV2d", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmaddss, "V4fV4fV4fV4f", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmaddsd, "V2dV2dV2dV2d", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmsubps, "V4fV4fV4fV4f", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmsubpd, "V2dV2dV2dV2d", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmsubss, "V4fV4fV4fV4f", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmsubsd, "V2dV2dV2dV2d", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmaddsubps, "V4fV4fV4fV4f", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmaddsubpd, "V2dV2dV2dV2d", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubaddps, "V4fV4fV4fV4f", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubaddpd, "V2dV2dV2dV2d", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmaddps256, "V8fV8fV8fV8f", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmaddpd256, "V4dV4dV4dV4d", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubps256, "V8fV8fV8fV8f", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubpd256, "V4dV4dV4dV4d", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmaddps256, "V8fV8fV8fV8f", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmaddpd256, "V4dV4dV4dV4d", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmsubps256, "V8fV8fV8fV8f", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfnmsubpd256, "V4dV4dV4dV4d", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmaddsubps256, "V8fV8fV8fV8f", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmaddsubpd256, "V4dV4dV4dV4d", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubaddps256, "V8fV8fV8fV8f", "", "fma,fma4")
-TARGET_BUILTIN(__builtin_ia32_vfmsubaddpd256, "V4dV4dV4dV4d", "", "fma,fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmaddps, "V4fV4fV4fV4f", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmaddpd, "V2dV2dV2dV2d", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmaddss, "V4fV4fV4fV4f", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmaddsd, "V2dV2dV2dV2d", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmsubps, "V4fV4fV4fV4f", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmsubpd, "V2dV2dV2dV2d", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmsubss, "V4fV4fV4fV4f", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmsubsd, "V2dV2dV2dV2d", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfnmaddps, "V4fV4fV4fV4f", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfnmaddpd, "V2dV2dV2dV2d", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfnmaddss, "V4fV4fV4fV4f", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfnmaddsd, "V2dV2dV2dV2d", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfnmsubps, "V4fV4fV4fV4f", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfnmsubpd, "V2dV2dV2dV2d", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfnmsubss, "V4fV4fV4fV4f", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfnmsubsd, "V2dV2dV2dV2d", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmaddsubps, "V4fV4fV4fV4f", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmaddsubpd, "V2dV2dV2dV2d", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmsubaddps, "V4fV4fV4fV4f", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmsubaddpd, "V2dV2dV2dV2d", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmaddps256, "V8fV8fV8fV8f", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmaddpd256, "V4dV4dV4dV4d", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmsubps256, "V8fV8fV8fV8f", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmsubpd256, "V4dV4dV4dV4d", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfnmaddps256, "V8fV8fV8fV8f", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfnmaddpd256, "V4dV4dV4dV4d", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfnmsubps256, "V8fV8fV8fV8f", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfnmsubpd256, "V4dV4dV4dV4d", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmaddsubps256, "V8fV8fV8fV8f", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmaddsubpd256, "V4dV4dV4dV4d", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmsubaddps256, "V8fV8fV8fV8f", "", "fma|fma4")
+TARGET_BUILTIN(__builtin_ia32_vfmsubaddpd256, "V4dV4dV4dV4d", "", "fma|fma4")
 
 TARGET_BUILTIN(__builtin_ia32_vfmaddpd128_mask, "V2dV2dV2dV2dUc", "", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_vfmaddpd128_mask3, "V2dV2dV2dV2dUc", "", "avx512vl")

diff --git a/lib/CodeGen/CGBuiltin.cpp b/lib/CodeGen/CGBuiltin.cpp
index c3fd54e7b225153e4826c4302b31112f3e6e74d4..5274b53ccc1c10c19a535a090f44c1cfb2135f12 100644 (file)
--- a/lib/CodeGen/CGBuiltin.cpp
+++ b/lib/CodeGen/CGBuiltin.cpp
@@ -341,10 +341,15 @@ bool CodeGenFunction::checkBuiltinTargetFeatures(
   // true, otherwise return false.
   SmallVector<StringRef, 1> AttrFeatures;
   StringRef(FeatureList).split(AttrFeatures, ",");
-  for (const auto &Feature : AttrFeatures)
-    if (FeatureMap[Feature])
-      return true;
-  return false;
+  return std::all_of(AttrFeatures.begin(), AttrFeatures.end(),
+                     [&](StringRef &Feature) {
+                       SmallVector<StringRef, 1> OrFeatures;
+                       Feature.split(OrFeatures, "|");
+                       return std::any_of(OrFeatures.begin(), OrFeatures.end(),
+                                          [&](StringRef &Feature) {
+                                            return FeatureMap[Feature];
+                                          });
+                     });
 }
 
 RValue CodeGenFunction::EmitBuiltinExpr(const FunctionDecl *FD,

diff --git a/lib/Headers/avx512vlintrin.h b/lib/Headers/avx512vlintrin.h
index 622385ae77f7c0e744bd04b9e658830c441d8986..8f13536fbb0e660b964fecfefb8a7525aa7aa805 100644 (file)
--- a/lib/Headers/avx512vlintrin.h
+++ b/lib/Headers/avx512vlintrin.h
@@ -28,18 +28,18 @@
 #ifndef __AVX512VLINTRIN_H
 #define __AVX512VLINTRIN_H
 
-/* Define the default attributes for the functions in this file. */
 #define __DEFAULT_FN_ATTRS __attribute__((__always_inline__, __nodebug__, __target__("avx512vl")))
+#define __DEFAULT_FN_ATTRS_BOTH __attribute__((__always_inline__, __nodebug__, __target__("avx512vl, avx512bw")))
 
 /* Integer compare */
 
-static __inline__ __mmask8 __DEFAULT_FN_ATTRS
+static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm_cmpeq_epi32_mask(__m128i __a, __m128i __b) {
   return (__mmask8)__builtin_ia32_pcmpeqd128_mask((__v4si)__a, (__v4si)__b,
                                                   (__mmask8)-1);
 }
 
-static __inline__ __mmask8 __DEFAULT_FN_ATTRS
+static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm_mask_cmpeq_epi32_mask(__mmask8 __u, __m128i __a, __m128i __b) {
   return (__mmask8)__builtin_ia32_pcmpeqd128_mask((__v4si)__a, (__v4si)__b,
                                                   __u);
@@ -57,13 +57,13 @@ _mm_mask_cmpeq_epu32_mask(__mmask8 __u, __m128i __a, __m128i __b) {
                                                 __u);
 }
 
-static __inline__ __mmask8 __DEFAULT_FN_ATTRS
+static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm256_cmpeq_epi32_mask(__m256i __a, __m256i __b) {
   return (__mmask8)__builtin_ia32_pcmpeqd256_mask((__v8si)__a, (__v8si)__b,
                                                   (__mmask8)-1);
 }
 
-static __inline__ __mmask8 __DEFAULT_FN_ATTRS
+static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm256_mask_cmpeq_epi32_mask(__mmask8 __u, __m256i __a, __m256i __b) {
   return (__mmask8)__builtin_ia32_pcmpeqd256_mask((__v8si)__a, (__v8si)__b,
                                                   __u);
@@ -81,13 +81,13 @@ _mm256_mask_cmpeq_epu32_mask(__mmask8 __u, __m256i __a, __m256i __b) {
                                                 __u);
 }
 
-static __inline__ __mmask8 __DEFAULT_FN_ATTRS
+static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm_cmpeq_epi64_mask(__m128i __a, __m128i __b) {
   return (__mmask8)__builtin_ia32_pcmpeqq128_mask((__v2di)__a, (__v2di)__b,
                                                   (__mmask8)-1);
 }
 
-static __inline__ __mmask8 __DEFAULT_FN_ATTRS
+static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm_mask_cmpeq_epi64_mask(__mmask8 __u, __m128i __a, __m128i __b) {
   return (__mmask8)__builtin_ia32_pcmpeqq128_mask((__v2di)__a, (__v2di)__b,
                                                   __u);
@@ -105,13 +105,13 @@ _mm_mask_cmpeq_epu64_mask(__mmask8 __u, __m128i __a, __m128i __b) {
                                                 __u);
 }
 
-static __inline__ __mmask8 __DEFAULT_FN_ATTRS
+static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm256_cmpeq_epi64_mask(__m256i __a, __m256i __b) {
   return (__mmask8)__builtin_ia32_pcmpeqq256_mask((__v4di)__a, (__v4di)__b,
                                                   (__mmask8)-1);
 }
 
-static __inline__ __mmask8 __DEFAULT_FN_ATTRS
+static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm256_mask_cmpeq_epi64_mask(__mmask8 __u, __m256i __a, __m256i __b) {
   return (__mmask8)__builtin_ia32_pcmpeqq256_mask((__v4di)__a, (__v4di)__b,
                                                   __u);
@@ -226,16 +226,13 @@ _mm256_mask_cmpge_epu64_mask(__mmask8 __u, __m256i __a, __m256i __b) {
                                                 __u);
 }
 
-
-
-
-static __inline__ __mmask8 __DEFAULT_FN_ATTRS
+static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm_cmpgt_epi32_mask(__m128i __a, __m128i __b) {
   return (__mmask8)__builtin_ia32_pcmpgtd128_mask((__v4si)__a, (__v4si)__b,
                                                   (__mmask8)-1);
 }
 
-static __inline__ __mmask8 __DEFAULT_FN_ATTRS
+static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm_mask_cmpgt_epi32_mask(__mmask8 __u, __m128i __a, __m128i __b) {
   return (__mmask8)__builtin_ia32_pcmpgtd128_mask((__v4si)__a, (__v4si)__b,
                                                   __u);
@@ -253,13 +250,13 @@ _mm_mask_cmpgt_epu32_mask(__mmask8 __u, __m128i __a, __m128i __b) {
                                                 __u);
 }
 
-static __inline__ __mmask8 __DEFAULT_FN_ATTRS
+static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm256_cmpgt_epi32_mask(__m256i __a, __m256i __b) {
   return (__mmask8)__builtin_ia32_pcmpgtd256_mask((__v8si)__a, (__v8si)__b,
                                                   (__mmask8)-1);
 }
 
-static __inline__ __mmask8 __DEFAULT_FN_ATTRS
+static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm256_mask_cmpgt_epi32_mask(__mmask8 __u, __m256i __a, __m256i __b) {
   return (__mmask8)__builtin_ia32_pcmpgtd256_mask((__v8si)__a, (__v8si)__b,
                                                   __u);
@@ -277,13 +274,13 @@ _mm256_mask_cmpgt_epu32_mask(__mmask8 __u, __m256i __a, __m256i __b) {
                                                 __u);
 }
 
-static __inline__ __mmask8 __DEFAULT_FN_ATTRS
+static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm_cmpgt_epi64_mask(__m128i __a, __m128i __b) {
   return (__mmask8)__builtin_ia32_pcmpgtq128_mask((__v2di)__a, (__v2di)__b,
                                                   (__mmask8)-1);
 }
 
-static __inline__ __mmask8 __DEFAULT_FN_ATTRS
+static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm_mask_cmpgt_epi64_mask(__mmask8 __u, __m128i __a, __m128i __b) {
   return (__mmask8)__builtin_ia32_pcmpgtq128_mask((__v2di)__a, (__v2di)__b,
                                                   __u);
@@ -301,13 +298,13 @@ _mm_mask_cmpgt_epu64_mask(__mmask8 __u, __m128i __a, __m128i __b) {
                                                 __u);
 }
 
-static __inline__ __mmask8 __DEFAULT_FN_ATTRS
+static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm256_cmpgt_epi64_mask(__m256i __a, __m256i __b) {
   return (__mmask8)__builtin_ia32_pcmpgtq256_mask((__v4di)__a, (__v4di)__b,
                                                   (__mmask8)-1);
 }
 
-static __inline__ __mmask8 __DEFAULT_FN_ATTRS
+static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm256_mask_cmpgt_epi64_mask(__mmask8 __u, __m256i __a, __m256i __b) {
   return (__mmask8)__builtin_ia32_pcmpgtq256_mask((__v4di)__a, (__v4di)__b,
                                                   __u);
@@ -4604,5 +4601,6 @@ _mm256_maskz_permutex2var_epi64 (__mmask8 __U, __m256i __A,
 }
 
 #undef __DEFAULT_FN_ATTRS
+#undef __DEFAULT_FN_ATTRS_BOTH
 
 #endif /* __AVX512VLINTRIN_H */

diff --git a/test/CodeGen/target-builtin-noerror.c b/test/CodeGen/target-builtin-noerror.c
index 569fb0a60ac0bcc83d3cb5d34892bfc413f53f27..7d86b9684624ec5818d6ac1cd77e74dd1ac36362 100644 (file)
--- a/test/CodeGen/target-builtin-noerror.c
+++ b/test/CodeGen/target-builtin-noerror.c
@@ -29,3 +29,16 @@ int qq() {
   else
     return qq_noavx();
 }
+
+// Test that fma and fma4 are both separately and combined valid for an fma intrinsic.
+__m128 __attribute__((target("fma"))) fma_1(__m128 a, __m128 b, __m128 c) {
+  return __builtin_ia32_vfmaddps(a, b, c);
+}
+
+__m128 __attribute__((target("fma4"))) fma_2(__m128 a, __m128 b, __m128 c) {
+  return __builtin_ia32_vfmaddps(a, b, c);
+}
+
+__m128 __attribute__((target("fma,fma4"))) fma_3(__m128 a, __m128 b, __m128 c) {
+  return __builtin_ia32_vfmaddps(a, b, c);
+}