[X86] Use unsigned types for vector arithmetic in intrinsics to avoid undefined behav...

This is really only needed for addition, subtraction, and multiplication, but I did the bitwise ops too for overall consistency. Clang currently doesn't set NSW for signed vector operations so the undefined behavior shouldn't happen today.

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@271778 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Headers/avx2intrin.h b/lib/Headers/avx2intrin.h
index 029199e4cd8e911c918e26bc62a0ef8acf0c3835..9df915a10b6028e8fd59fad11807e2bac4aa6c93 100644 (file)
--- a/lib/Headers/avx2intrin.h
+++ b/lib/Headers/avx2intrin.h
@@ -81,25 +81,25 @@ _mm256_packus_epi32(__m256i __V1, __m256i __V2)
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_add_epi8(__m256i __a, __m256i __b)
 {
-  return (__m256i)((__v32qi)__a + (__v32qi)__b);
+  return (__m256i)((__v32qu)__a + (__v32qu)__b);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_add_epi16(__m256i __a, __m256i __b)
 {
-  return (__m256i)((__v16hi)__a + (__v16hi)__b);
+  return (__m256i)((__v16hu)__a + (__v16hu)__b);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_add_epi32(__m256i __a, __m256i __b)
 {
-  return (__m256i)((__v8si)__a + (__v8si)__b);
+  return (__m256i)((__v8su)__a + (__v8su)__b);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_add_epi64(__m256i __a, __m256i __b)
 {
-  return (__m256i)((__v4di)__a + (__v4di)__b);
+  return (__m256i)((__v4du)__a + (__v4du)__b);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
@@ -133,13 +133,13 @@ _mm256_adds_epu16(__m256i __a, __m256i __b)
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_and_si256(__m256i __a, __m256i __b)
 {
-  return (__m256i)((__v4di)__a & (__v4di)__b);
+  return (__m256i)((__v4du)__a & (__v4du)__b);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_andnot_si256(__m256i __a, __m256i __b)
 {
-  return (__m256i)(~(__v4di)__a & (__v4di)__b);
+  return (__m256i)(~(__v4du)__a & (__v4du)__b);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
@@ -462,13 +462,13 @@ _mm256_mulhi_epi16(__m256i __a, __m256i __b)
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_mullo_epi16(__m256i __a, __m256i __b)
 {
-  return (__m256i)((__v16hi)__a * (__v16hi)__b);
+  return (__m256i)((__v16hu)__a * (__v16hu)__b);
 }
 
 static __inline__  __m256i __DEFAULT_FN_ATTRS
 _mm256_mullo_epi32 (__m256i __a, __m256i __b)
 {
-  return (__m256i)((__v8si)__a * (__v8si)__b);
+  return (__m256i)((__v8su)__a * (__v8su)__b);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
@@ -480,7 +480,7 @@ _mm256_mul_epu32(__m256i __a, __m256i __b)
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_or_si256(__m256i __a, __m256i __b)
 {
-  return (__m256i)((__v4di)__a | (__v4di)__b);
+  return (__m256i)((__v4du)__a | (__v4du)__b);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
@@ -658,25 +658,25 @@ _mm256_srl_epi64(__m256i __a, __m128i __count)
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_sub_epi8(__m256i __a, __m256i __b)
 {
-  return (__m256i)((__v32qi)__a - (__v32qi)__b);
+  return (__m256i)((__v32qu)__a - (__v32qu)__b);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_sub_epi16(__m256i __a, __m256i __b)
 {
-  return (__m256i)((__v16hi)__a - (__v16hi)__b);
+  return (__m256i)((__v16hu)__a - (__v16hu)__b);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_sub_epi32(__m256i __a, __m256i __b)
 {
-  return (__m256i)((__v8si)__a - (__v8si)__b);
+  return (__m256i)((__v8su)__a - (__v8su)__b);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_sub_epi64(__m256i __a, __m256i __b)
 {
-  return (__m256i)((__v4di)__a - (__v4di)__b);
+  return (__m256i)((__v4du)__a - (__v4du)__b);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
@@ -754,7 +754,7 @@ _mm256_unpacklo_epi64(__m256i __a, __m256i __b)
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_xor_si256(__m256i __a, __m256i __b)
 {
-  return (__m256i)((__v4di)__a ^ (__v4di)__b);
+  return (__m256i)((__v4du)__a ^ (__v4du)__b);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS

diff --git a/lib/Headers/avx512bwintrin.h b/lib/Headers/avx512bwintrin.h
index 01bea5bc838be044473f0df4912e5eb7281bc109..c7c8f8c3607848ec5332221d68bdd11f35abbb80 100644 (file)
--- a/lib/Headers/avx512bwintrin.h
+++ b/lib/Headers/avx512bwintrin.h
@@ -30,8 +30,6 @@
 
 typedef unsigned int __mmask32;
 typedef unsigned long long __mmask64;
-typedef char __v64qi __attribute__ ((__vector_size__ (64)));
-typedef short __v32hi __attribute__ ((__vector_size__ (64)));
 
 /* Define the default attributes for the functions in this file. */
 #define __DEFAULT_FN_ATTRS __attribute__((__always_inline__, __nodebug__, __target__("avx512bw")))
@@ -348,7 +346,7 @@ _mm512_mask_cmpneq_epu16_mask(__mmask32 __u, __m512i __a, __m512i __b) {
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_add_epi8 (__m512i __A, __m512i __B) {
-  return (__m512i) ((__v64qi) __A + (__v64qi) __B);
+  return (__m512i) ((__v64qu) __A + (__v64qu) __B);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
@@ -369,7 +367,7 @@ _mm512_maskz_add_epi8 (__mmask64 __U, __m512i __A, __m512i __B) {
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_sub_epi8 (__m512i __A, __m512i __B) {
-  return (__m512i) ((__v64qi) __A - (__v64qi) __B);
+  return (__m512i) ((__v64qu) __A - (__v64qu) __B);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
@@ -390,7 +388,7 @@ _mm512_maskz_sub_epi8 (__mmask64 __U, __m512i __A, __m512i __B) {
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_add_epi16 (__m512i __A, __m512i __B) {
-  return (__m512i) ((__v32hi) __A + (__v32hi) __B);
+  return (__m512i) ((__v32hu) __A + (__v32hu) __B);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
@@ -411,7 +409,7 @@ _mm512_maskz_add_epi16 (__mmask32 __U, __m512i __A, __m512i __B) {
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_sub_epi16 (__m512i __A, __m512i __B) {
-  return (__m512i) ((__v32hi) __A - (__v32hi) __B);
+  return (__m512i) ((__v32hu) __A - (__v32hu) __B);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
@@ -432,7 +430,7 @@ _mm512_maskz_sub_epi16 (__mmask32 __U, __m512i __A, __m512i __B) {
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mullo_epi16 (__m512i __A, __m512i __B) {
-  return (__m512i) ((__v32hi) __A * (__v32hi) __B);
+  return (__m512i) ((__v32hu) __A * (__v32hu) __B);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS

diff --git a/lib/Headers/avx512dqintrin.h b/lib/Headers/avx512dqintrin.h
index 72ce4153570ca3a34e1a60916abe92e9130f48c1..e3ceffabb89f4a9a78891207e886ac7bfdd2917d 100644 (file)
--- a/lib/Headers/avx512dqintrin.h
+++ b/lib/Headers/avx512dqintrin.h
@@ -33,7 +33,7 @@
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mullo_epi64 (__m512i __A, __m512i __B) {
-  return (__m512i) ((__v8di) __A * (__v8di) __B);
+  return (__m512i) ((__v8du) __A * (__v8du) __B);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
@@ -55,7 +55,7 @@ _mm512_maskz_mullo_epi64 (__mmask8 __U, __m512i __A, __m512i __B) {
 
 static __inline__ __m512d __DEFAULT_FN_ATTRS
 _mm512_xor_pd (__m512d __A, __m512d __B) {
-  return (__m512d) ((__v8di) __A ^ (__v8di) __B);
+  return (__m512d) ((__v8du) __A ^ (__v8du) __B);
 }
 
 static __inline__ __m512d __DEFAULT_FN_ATTRS
@@ -77,7 +77,7 @@ _mm512_maskz_xor_pd (__mmask8 __U, __m512d __A, __m512d __B) {
 
 static __inline__ __m512 __DEFAULT_FN_ATTRS
 _mm512_xor_ps (__m512 __A, __m512 __B) {
-  return (__m512) ((__v16si) __A ^ (__v16si) __B);
+  return (__m512) ((__v16su) __A ^ (__v16su) __B);
 }
 
 static __inline__ __m512 __DEFAULT_FN_ATTRS
@@ -99,7 +99,7 @@ _mm512_maskz_xor_ps (__mmask16 __U, __m512 __A, __m512 __B) {
 
 static __inline__ __m512d __DEFAULT_FN_ATTRS
 _mm512_or_pd (__m512d __A, __m512d __B) {
-  return (__m512d) ((__v8di) __A | (__v8di) __B);
+  return (__m512d) ((__v8du) __A | (__v8du) __B);
 }
 
 static __inline__ __m512d __DEFAULT_FN_ATTRS
@@ -121,7 +121,7 @@ _mm512_maskz_or_pd (__mmask8 __U, __m512d __A, __m512d __B) {
 
 static __inline__ __m512 __DEFAULT_FN_ATTRS
 _mm512_or_ps (__m512 __A, __m512 __B) {
-  return (__m512) ((__v16si) __A | (__v16si) __B);
+  return (__m512) ((__v16su) __A | (__v16su) __B);
 }
 
 static __inline__ __m512 __DEFAULT_FN_ATTRS
@@ -143,7 +143,7 @@ _mm512_maskz_or_ps (__mmask16 __U, __m512 __A, __m512 __B) {
 
 static __inline__ __m512d __DEFAULT_FN_ATTRS
 _mm512_and_pd (__m512d __A, __m512d __B) {
-  return (__m512d) ((__v8di) __A & (__v8di) __B);
+  return (__m512d) ((__v8du) __A & (__v8du) __B);
 }
 
 static __inline__ __m512d __DEFAULT_FN_ATTRS
@@ -165,7 +165,7 @@ _mm512_maskz_and_pd (__mmask8 __U, __m512d __A, __m512d __B) {
 
 static __inline__ __m512 __DEFAULT_FN_ATTRS
 _mm512_and_ps (__m512 __A, __m512 __B) {
-  return (__m512) ((__v16si) __A & (__v16si) __B);
+  return (__m512) ((__v16su) __A & (__v16su) __B);
 }
 
 static __inline__ __m512 __DEFAULT_FN_ATTRS

diff --git a/lib/Headers/avx512fintrin.h b/lib/Headers/avx512fintrin.h
index 0aed8243b0462bf263657ecb03046c19f1dac33c..701d5cf17bbd84e88e71791a7a2a30a2548c8747 100644 (file)
--- a/lib/Headers/avx512fintrin.h
+++ b/lib/Headers/avx512fintrin.h
@@ -34,6 +34,12 @@ typedef float __v16sf __attribute__((__vector_size__(64)));
 typedef long long __v8di __attribute__((__vector_size__(64)));
 typedef int __v16si __attribute__((__vector_size__(64)));
 
+/* Unsigned types */
+typedef unsigned char __v64qu __attribute__((__vector_size__(64)));
+typedef unsigned short __v32hu __attribute__((__vector_size__(64)));
+typedef unsigned long long __v8du __attribute__((__vector_size__(64)));
+typedef unsigned int __v16su __attribute__((__vector_size__(64)));
+
 typedef float __m512 __attribute__((__vector_size__(64)));
 typedef double __m512d __attribute__((__vector_size__(64)));
 typedef long long __m512i __attribute__((__vector_size__(64)));
@@ -455,7 +461,7 @@ _mm512_castsi512_si256 (__m512i __A)
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_and_epi32(__m512i __a, __m512i __b)
 {
-  return (__m512i)((__v16si)__a & (__v16si)__b);
+  return (__m512i)((__v16su)__a & (__v16su)__b);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
@@ -479,7 +485,7 @@ _mm512_maskz_and_epi32(__mmask16 __k, __m512i __a, __m512i __b)
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_and_epi64(__m512i __a, __m512i __b)
 {
-  return (__m512i)((__v8di)__a & (__v8di)__b);
+  return (__m512i)((__v8du)__a & (__v8du)__b);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
@@ -503,7 +509,7 @@ _mm512_maskz_and_epi64(__mmask8 __k, __m512i __a, __m512i __b)
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_andnot_si512 (__m512i __A, __m512i __B)
 {
-  return (__m512i)(~(__A) & __B);
+  return (__m512i)(~(__v8du)(__A) & (__v8du)__B);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
@@ -565,7 +571,7 @@ _mm512_maskz_andnot_epi64 (__mmask8 __U, __m512i __A, __m512i __B)
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_or_epi32(__m512i __a, __m512i __b)
 {
-  return (__m512i)((__v16si)__a | (__v16si)__b);
+  return (__m512i)((__v16su)__a | (__v16su)__b);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
@@ -589,7 +595,7 @@ _mm512_maskz_or_epi32(__mmask16 __k, __m512i __a, __m512i __b)
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_or_epi64(__m512i __a, __m512i __b)
 {
-  return (__m512i)((__v8di)__a | (__v8di)__b);
+  return (__m512i)((__v8du)__a | (__v8du)__b);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
@@ -613,7 +619,7 @@ _mm512_maskz_or_epi64(__mmask8 __k, __m512i __a, __m512i __b)
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_xor_epi32(__m512i __a, __m512i __b)
 {
-  return (__m512i)((__v16si)__a ^ (__v16si)__b);
+  return (__m512i)((__v16su)__a ^ (__v16su)__b);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
@@ -637,7 +643,7 @@ _mm512_maskz_xor_epi32(__mmask16 __k, __m512i __a, __m512i __b)
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_xor_epi64(__m512i __a, __m512i __b)
 {
-  return (__m512i)((__v8di)__a ^ (__v8di)__b);
+  return (__m512i)((__v8du)__a ^ (__v8du)__b);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
@@ -661,19 +667,19 @@ _mm512_maskz_xor_epi64(__mmask8 __k, __m512i __a, __m512i __b)
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_and_si512(__m512i __a, __m512i __b)
 {
-  return (__m512i)((__v8di)__a & (__v8di)__b);
+  return (__m512i)((__v8du)__a & (__v8du)__b);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_or_si512(__m512i __a, __m512i __b)
 {
-  return (__m512i)((__v8di)__a | (__v8di)__b);
+  return (__m512i)((__v8du)__a | (__v8du)__b);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_xor_si512(__m512i __a, __m512i __b)
 {
-  return (__m512i)((__v8di)__a ^ (__v8di)__b);
+  return (__m512i)((__v8du)__a ^ (__v8du)__b);
 }
 /* Arithmetic */
 
@@ -716,7 +722,7 @@ _mm512_sub_ps(__m512 __a, __m512 __b)
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_add_epi64 (__m512i __A, __m512i __B)
 {
-  return (__m512i) ((__v8di) __A + (__v8di) __B);
+  return (__m512i) ((__v8du) __A + (__v8du) __B);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
@@ -741,7 +747,7 @@ _mm512_maskz_add_epi64 (__mmask8 __U, __m512i __A, __m512i __B)
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_sub_epi64 (__m512i __A, __m512i __B)
 {
-  return (__m512i) ((__v8di) __A - (__v8di) __B);
+  return (__m512i) ((__v8du) __A - (__v8du) __B);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
@@ -766,7 +772,7 @@ _mm512_maskz_sub_epi64 (__mmask8 __U, __m512i __A, __m512i __B)
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_add_epi32 (__m512i __A, __m512i __B)
 {
-  return (__m512i) ((__v16si) __A + (__v16si) __B);
+  return (__m512i) ((__v16su) __A + (__v16su) __B);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
@@ -791,7 +797,7 @@ _mm512_maskz_add_epi32 (__mmask16 __U, __m512i __A, __m512i __B)
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_sub_epi32 (__m512i __A, __m512i __B)
 {
-  return (__m512i) ((__v16si) __A - (__v16si) __B);
+  return (__m512i) ((__v16su) __A - (__v16su) __B);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
@@ -1462,7 +1468,7 @@ _mm512_maskz_mul_epu32 (__mmask8 __M, __m512i __X, __m512i __Y)
 static __inline __m512i __DEFAULT_FN_ATTRS
 _mm512_mullo_epi32 (__m512i __A, __m512i __B)
 {
-  return (__m512i) ((__v16si) __A * (__v16si) __B);
+  return (__m512i) ((__v16su) __A * (__v16su) __B);
 }
 
 static __inline __m512i __DEFAULT_FN_ATTRS

diff --git a/lib/Headers/avx512vldqintrin.h b/lib/Headers/avx512vldqintrin.h
index e4929ca4301d46f3007209997ed530bc46bf42f8..8187bcd6b28e33590c3ed1475d74dcca5a7e4f76 100644 (file)
--- a/lib/Headers/avx512vldqintrin.h
+++ b/lib/Headers/avx512vldqintrin.h
@@ -33,7 +33,7 @@
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_mullo_epi64 (__m256i __A, __m256i __B) {
-  return (__m256i) ((__v4di) __A * (__v4di) __B);
+  return (__m256i) ((__v4du) __A * (__v4du) __B);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
@@ -55,7 +55,7 @@ _mm256_maskz_mullo_epi64 (__mmask8 __U, __m256i __A, __m256i __B) {
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mullo_epi64 (__m128i __A, __m128i __B) {
-  return (__m128i) ((__v2di) __A * (__v2di) __B);
+  return (__m128i) ((__v2du) __A * (__v2du) __B);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS

diff --git a/lib/Headers/avxintrin.h b/lib/Headers/avxintrin.h
index e46f8f4ac41d6e100ffb1c65f2a2f3bba7ff985e..540a35b0e624959dca9183d7e2088a415ca69fe1 100644 (file)
--- a/lib/Headers/avxintrin.h
+++ b/lib/Headers/avxintrin.h
@@ -35,6 +35,12 @@ typedef int __v8si __attribute__ ((__vector_size__ (32)));
 typedef short __v16hi __attribute__ ((__vector_size__ (32)));
 typedef char __v32qi __attribute__ ((__vector_size__ (32)));
 
+/* Unsigned types */
+typedef unsigned long long __v4du __attribute__ ((__vector_size__ (32)));
+typedef unsigned int __v8su __attribute__ ((__vector_size__ (32)));
+typedef unsigned short __v16hu __attribute__ ((__vector_size__ (32)));
+typedef unsigned char __v32qu __attribute__ ((__vector_size__ (32)));
+
 /* We need an explicitly signed variant for char. Note that this shouldn't
  * appear in the interface though. */
 typedef signed char __v32qs __attribute__((__vector_size__(32)));
@@ -522,7 +528,7 @@ _mm256_rcp_ps(__m256 __a)
 static __inline __m256d __DEFAULT_FN_ATTRS
 _mm256_and_pd(__m256d __a, __m256d __b)
 {
-  return (__m256d)((__v4di)__a & (__v4di)__b);
+  return (__m256d)((__v4du)__a & (__v4du)__b);
 }
 
 /// \brief Performs a bitwise AND of two 256-bit vectors of [8 x float].
@@ -540,7 +546,7 @@ _mm256_and_pd(__m256d __a, __m256d __b)
 static __inline __m256 __DEFAULT_FN_ATTRS
 _mm256_and_ps(__m256 __a, __m256 __b)
 {
-  return (__m256)((__v8si)__a & (__v8si)__b);
+  return (__m256)((__v8su)__a & (__v8su)__b);
 }
 
 /// \brief Performs a bitwise AND of two 256-bit vectors of [4 x double], using
@@ -561,7 +567,7 @@ _mm256_and_ps(__m256 __a, __m256 __b)
 static __inline __m256d __DEFAULT_FN_ATTRS
 _mm256_andnot_pd(__m256d __a, __m256d __b)
 {
-  return (__m256d)(~(__v4di)__a & (__v4di)__b);
+  return (__m256d)(~(__v4du)__a & (__v4du)__b);
 }
 
 /// \brief Performs a bitwise AND of two 256-bit vectors of [8 x float], using
@@ -582,7 +588,7 @@ _mm256_andnot_pd(__m256d __a, __m256d __b)
 static __inline __m256 __DEFAULT_FN_ATTRS
 _mm256_andnot_ps(__m256 __a, __m256 __b)
 {
-  return (__m256)(~(__v8si)__a & (__v8si)__b);
+  return (__m256)(~(__v8su)__a & (__v8su)__b);
 }
 
 /// \brief Performs a bitwise OR of two 256-bit vectors of [4 x double].
@@ -600,7 +606,7 @@ _mm256_andnot_ps(__m256 __a, __m256 __b)
 static __inline __m256d __DEFAULT_FN_ATTRS
 _mm256_or_pd(__m256d __a, __m256d __b)
 {
-  return (__m256d)((__v4di)__a | (__v4di)__b);
+  return (__m256d)((__v4du)__a | (__v4du)__b);
 }
 
 /// \brief Performs a bitwise OR of two 256-bit vectors of [8 x float].
@@ -618,7 +624,7 @@ _mm256_or_pd(__m256d __a, __m256d __b)
 static __inline __m256 __DEFAULT_FN_ATTRS
 _mm256_or_ps(__m256 __a, __m256 __b)
 {
-  return (__m256)((__v8si)__a | (__v8si)__b);
+  return (__m256)((__v8su)__a | (__v8su)__b);
 }
 
 /// \brief Performs a bitwise XOR of two 256-bit vectors of [4 x double].
@@ -636,7 +642,7 @@ _mm256_or_ps(__m256 __a, __m256 __b)
 static __inline __m256d __DEFAULT_FN_ATTRS
 _mm256_xor_pd(__m256d __a, __m256d __b)
 {
-  return (__m256d)((__v4di)__a ^ (__v4di)__b);
+  return (__m256d)((__v4du)__a ^ (__v4du)__b);
 }
 
 /// \brief Performs a bitwise XOR of two 256-bit vectors of [8 x float].
@@ -654,7 +660,7 @@ _mm256_xor_pd(__m256d __a, __m256d __b)
 static __inline __m256 __DEFAULT_FN_ATTRS
 _mm256_xor_ps(__m256 __a, __m256 __b)
 {
-  return (__m256)((__v8si)__a ^ (__v8si)__b);
+  return (__m256)((__v8su)__a ^ (__v8su)__b);
 }
 
 /* Horizontal arithmetic */

diff --git a/lib/Headers/emmintrin.h b/lib/Headers/emmintrin.h
index 04cca74a425c254cdb75cb5159f5629c37f91816..c9fdda8191e8c0790a7999c20defe042349586f5 100644 (file)
--- a/lib/Headers/emmintrin.h
+++ b/lib/Headers/emmintrin.h
@@ -37,7 +37,6 @@ typedef char __v16qi __attribute__((__vector_size__(16)));
 
 /* Unsigned types */
 typedef unsigned long long __v2du __attribute__ ((__vector_size__ (16)));
-typedef unsigned int __v4su __attribute__((__vector_size__(16)));
 typedef unsigned short __v8hu __attribute__((__vector_size__(16)));
 typedef unsigned char __v16qu __attribute__((__vector_size__(16)));
 
@@ -142,25 +141,25 @@ _mm_max_pd(__m128d __a, __m128d __b)
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_and_pd(__m128d __a, __m128d __b)
 {
-  return (__m128d)((__v4si)__a & (__v4si)__b);
+  return (__m128d)((__v4su)__a & (__v4su)__b);
 }
 
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_andnot_pd(__m128d __a, __m128d __b)
 {
-  return (__m128d)(~(__v4si)__a & (__v4si)__b);
+  return (__m128d)(~(__v4su)__a & (__v4su)__b);
 }
 
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_or_pd(__m128d __a, __m128d __b)
 {
-  return (__m128d)((__v4si)__a | (__v4si)__b);
+  return (__m128d)((__v4su)__a | (__v4su)__b);
 }
 
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_xor_pd(__m128d __a, __m128d __b)
 {
-  return (__m128d)((__v4si)__a ^ (__v4si)__b);
+  return (__m128d)((__v4su)__a ^ (__v4su)__b);
 }
 
 static __inline__ __m128d __DEFAULT_FN_ATTRS
@@ -643,19 +642,19 @@ _mm_storel_pd(double *__dp, __m128d __a)
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_add_epi8(__m128i __a, __m128i __b)
 {
-  return (__m128i)((__v16qi)__a + (__v16qi)__b);
+  return (__m128i)((__v16qu)__a + (__v16qu)__b);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_add_epi16(__m128i __a, __m128i __b)
 {
-  return (__m128i)((__v8hi)__a + (__v8hi)__b);
+  return (__m128i)((__v8hu)__a + (__v8hu)__b);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_add_epi32(__m128i __a, __m128i __b)
 {
-  return (__m128i)((__v4si)__a + (__v4si)__b);
+  return (__m128i)((__v4su)__a + (__v4su)__b);
 }
 
 static __inline__ __m64 __DEFAULT_FN_ATTRS
@@ -667,7 +666,7 @@ _mm_add_si64(__m64 __a, __m64 __b)
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_add_epi64(__m128i __a, __m128i __b)
 {
-  return (__m128i)((__v2di)__a + (__v2di)__b);
+  return (__m128i)((__v2du)__a + (__v2du)__b);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
@@ -764,7 +763,7 @@ _mm_mulhi_epu16(__m128i __a, __m128i __b)
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mullo_epi16(__m128i __a, __m128i __b)
 {
-  return (__m128i)((__v8hi)__a * (__v8hi)__b);
+  return (__m128i)((__v8hu)__a * (__v8hu)__b);
 }
 
 /// \brief Multiplies 32-bit unsigned integer values contained in the lower bits
@@ -842,7 +841,7 @@ _mm_sad_epu8(__m128i __a, __m128i __b)
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_sub_epi8(__m128i __a, __m128i __b)
 {
-  return (__m128i)((__v16qi)__a - (__v16qi)__b);
+  return (__m128i)((__v16qu)__a - (__v16qu)__b);
 }
 
 /// \brief Subtracts the corresponding 16-bit integer values in the operands.
@@ -860,7 +859,7 @@ _mm_sub_epi8(__m128i __a, __m128i __b)
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_sub_epi16(__m128i __a, __m128i __b)
 {
-  return (__m128i)((__v8hi)__a - (__v8hi)__b);
+  return (__m128i)((__v8hu)__a - (__v8hu)__b);
 }
 
 /// \brief Subtracts the corresponding 32-bit integer values in the operands.
@@ -878,7 +877,7 @@ _mm_sub_epi16(__m128i __a, __m128i __b)
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_sub_epi32(__m128i __a, __m128i __b)
 {
-  return (__m128i)((__v4si)__a - (__v4si)__b);
+  return (__m128i)((__v4su)__a - (__v4su)__b);
 }
 
 /// \brief Subtracts signed or unsigned 64-bit integer values and writes the
@@ -915,7 +914,7 @@ _mm_sub_si64(__m64 __a, __m64 __b)
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_sub_epi64(__m128i __a, __m128i __b)
 {
-  return (__m128i)((__v2di)__a - (__v2di)__b);
+  return (__m128i)((__v2du)__a - (__v2du)__b);
 }
 
 /// \brief Subtracts corresponding 8-bit signed integer values in the input and
@@ -1015,7 +1014,7 @@ _mm_subs_epu16(__m128i __a, __m128i __b)
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_and_si128(__m128i __a, __m128i __b)
 {
-  return (__m128i)((__v2di)__a & (__v2di)__b);
+  return (__m128i)((__v2du)__a & (__v2du)__b);
 }
 
 /// \brief Performs a bitwise AND of two 128-bit integer vectors, using the
@@ -1035,7 +1034,7 @@ _mm_and_si128(__m128i __a, __m128i __b)
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_andnot_si128(__m128i __a, __m128i __b)
 {
-  return (__m128i)(~(__v2di)__a & (__v2di)__b);
+  return (__m128i)(~(__v2du)__a & (__v2du)__b);
 }
 /// \brief Performs a bitwise OR of two 128-bit integer vectors.
 ///
@@ -1052,7 +1051,7 @@ _mm_andnot_si128(__m128i __a, __m128i __b)
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_or_si128(__m128i __a, __m128i __b)
 {
-  return (__m128i)((__v2di)__a | (__v2di)__b);
+  return (__m128i)((__v2du)__a | (__v2du)__b);
 }
 
 /// \brief Performs a bitwise exclusive OR of two 128-bit integer vectors.
@@ -1070,7 +1069,7 @@ _mm_or_si128(__m128i __a, __m128i __b)
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_xor_si128(__m128i __a, __m128i __b)
 {
-  return (__m128i)((__v2di)__a ^ (__v2di)__b);
+  return (__m128i)((__v2du)__a ^ (__v2du)__b);
 }
 
 /// \brief Left-shifts the 128-bit integer vector operand by the specified

diff --git a/lib/Headers/smmintrin.h b/lib/Headers/smmintrin.h
index 37abb7001d52355dc5988bde3d5c32463d3b5f38..e48ab034f46f25492081799a0b6f975455991745 100644 (file)
--- a/lib/Headers/smmintrin.h
+++ b/lib/Headers/smmintrin.h
@@ -121,7 +121,7 @@ _mm_blendv_epi8 (__m128i __V1, __m128i __V2, __m128i __M)
 static __inline__  __m128i __DEFAULT_FN_ATTRS
 _mm_mullo_epi32 (__m128i __V1, __m128i __V2)
 {
-  return (__m128i) ((__v4si)__V1 * (__v4si)__V2);
+  return (__m128i) ((__v4su)__V1 * (__v4su)__V2);
 }
 
 static __inline__  __m128i __DEFAULT_FN_ATTRS

diff --git a/lib/Headers/xmmintrin.h b/lib/Headers/xmmintrin.h
index 2a048c880f888444144ad43a23ce613feb1fe13a..937713cad5aa7e2398d6cdbecd96c85ef52e7dac 100644 (file)
--- a/lib/Headers/xmmintrin.h
+++ b/lib/Headers/xmmintrin.h
@@ -30,6 +30,9 @@ typedef int __v4si __attribute__((__vector_size__(16)));
 typedef float __v4sf __attribute__((__vector_size__(16)));
 typedef float __m128 __attribute__((__vector_size__(16)));
 
+/* Unsigned types */
+typedef unsigned int __v4su __attribute__((__vector_size__(16)));
+
 /* This header should only be included in a hosted environment as it depends on
  * a standard library to provide allocation routines. */
 #if __STDC_HOSTED__
@@ -414,7 +417,7 @@ _mm_max_ps(__m128 __a, __m128 __b)
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_and_ps(__m128 __a, __m128 __b)
 {
-  return (__m128)((__v4si)__a & (__v4si)__b);
+  return (__m128)((__v4su)__a & (__v4su)__b);
 }
 
 /// \brief Performs a bitwise AND of two 128-bit vectors of [4 x float], using
@@ -436,7 +439,7 @@ _mm_and_ps(__m128 __a, __m128 __b)
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_andnot_ps(__m128 __a, __m128 __b)
 {
-  return (__m128)(~(__v4si)__a & (__v4si)__b);
+  return (__m128)(~(__v4su)__a & (__v4su)__b);
 }
 
 /// \brief Performs a bitwise OR of two 128-bit vectors of [4 x float].
@@ -454,7 +457,7 @@ _mm_andnot_ps(__m128 __a, __m128 __b)
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_or_ps(__m128 __a, __m128 __b)
 {
-  return (__m128)((__v4si)__a | (__v4si)__b);
+  return (__m128)((__v4su)__a | (__v4su)__b);
 }
 
 /// \brief Performs a bitwise exclusive OR of two 128-bit vectors of
@@ -473,7 +476,7 @@ _mm_or_ps(__m128 __a, __m128 __b)
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_xor_ps(__m128 __a, __m128 __b)
 {
-  return (__m128)((__v4si)__a ^ (__v4si)__b);
+  return (__m128)((__v4su)__a ^ (__v4su)__b);
 }
 
 /// \brief Compares two 32-bit float values in the low-order bits of both