[X86] Remove mask arguments from permvar builtins/intrinsics. Use a select in IR instead.

Someday maybe we'll use selects for all the builtins.

llvm-svn: 332825

6 files changed, 92 insertions, 152 deletions

diff --git a/clang/lib/Headers/avx512bwintrin.h b/clang/lib/Headers/avx512bwintrin.h
index 499bb8a6906..fd1b871ae16 100644
--- a/clang/lib/Headers/avx512bwintrin.h
+++ b/clang/lib/Headers/avx512bwintrin.h
@@ -2034,30 +2034,25 @@ _mm512_maskz_broadcastw_epi16 (__mmask32 __M, __m128i __A)
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_permutexvar_epi16 (__m512i __A, __m512i __B)
 {
-  return (__m512i) __builtin_ia32_permvarhi512_mask ((__v32hi) __B,
-                 (__v32hi) __A,
-                 (__v32hi) _mm512_undefined_epi32 (),
-                 (__mmask32) -1);
+  return (__m512i)__builtin_ia32_permvarhi512((__v32hi)__B, (__v32hi)__A);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_maskz_permutexvar_epi16 (__mmask32 __M, __m512i __A,
         __m512i __B)
 {
-  return (__m512i) __builtin_ia32_permvarhi512_mask ((__v32hi) __B,
-                 (__v32hi) __A,
-                 (__v32hi) _mm512_setzero_hi(),
-                 (__mmask32) __M);
+  return (__m512i)__builtin_ia32_selectw_512((__mmask32)__M,
+                                    (__v32hi)_mm512_permutexvar_epi16(__A, __B),
+                                    (__v32hi)_mm512_setzero_hi());
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_permutexvar_epi16 (__m512i __W, __mmask32 __M, __m512i __A,
              __m512i __B)
 {
-  return (__m512i) __builtin_ia32_permvarhi512_mask ((__v32hi) __B,
-                 (__v32hi) __A,
-                 (__v32hi) __W,
-                 (__mmask32) __M);
+  return (__m512i)__builtin_ia32_selectw_512((__mmask32)__M,
+                                    (__v32hi)_mm512_permutexvar_epi16(__A, __B),
+                                    (__v32hi)__W);
 }
 
 #define _mm512_alignr_epi8(A, B, N) __extension__ ({\
diff --git a/clang/lib/Headers/avx512fintrin.h b/clang/lib/Headers/avx512fintrin.h
index ee5fb5ae0be..92d1ad26a94 100644
--- a/clang/lib/Headers/avx512fintrin.h
+++ b/clang/lib/Headers/avx512fintrin.h
@@ -8646,113 +8646,93 @@ _mm_mask3_fnmsub_sd (__m128d __W, __m128d __X, __m128d __Y, __mmask8 __U)
 static __inline__ __m512d __DEFAULT_FN_ATTRS
 _mm512_permutexvar_pd (__m512i __X, __m512d __Y)
 {
-  return (__m512d) __builtin_ia32_permvardf512_mask ((__v8df) __Y,
-                 (__v8di) __X,
-                 (__v8df) _mm512_undefined_pd (),
-                 (__mmask8) -1);
+  return (__m512d)__builtin_ia32_permvardf512((__v8df) __Y, (__v8di) __X);
 }
 
 static __inline__ __m512d __DEFAULT_FN_ATTRS
 _mm512_mask_permutexvar_pd (__m512d __W, __mmask8 __U, __m512i __X, __m512d __Y)
 {
-  return (__m512d) __builtin_ia32_permvardf512_mask ((__v8df) __Y,
-                 (__v8di) __X,
-                 (__v8df) __W,
-                 (__mmask8) __U);
+  return (__m512d)__builtin_ia32_selectpd_512((__mmask8)__U,
+                                        (__v8df)_mm512_permutexvar_pd(__X, __Y),
+                                        (__v8df)__W);
 }
 
 static __inline__ __m512d __DEFAULT_FN_ATTRS
 _mm512_maskz_permutexvar_pd (__mmask8 __U, __m512i __X, __m512d __Y)
 {
-  return (__m512d) __builtin_ia32_permvardf512_mask ((__v8df) __Y,
-                 (__v8di) __X,
-                 (__v8df) _mm512_setzero_pd (),
-                 (__mmask8) __U);
+  return (__m512d)__builtin_ia32_selectpd_512((__mmask8)__U,
+                                        (__v8df)_mm512_permutexvar_pd(__X, __Y),
+                                        (__v8df)_mm512_setzero_pd());
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
-_mm512_maskz_permutexvar_epi64 (__mmask8 __M, __m512i __X, __m512i __Y)
+_mm512_permutexvar_epi64 (__m512i __X, __m512i __Y)
 {
-  return (__m512i) __builtin_ia32_permvardi512_mask ((__v8di) __Y,
-                 (__v8di) __X,
-                 (__v8di) _mm512_setzero_si512 (),
-                 __M);
+  return (__m512i)__builtin_ia32_permvardi512((__v8di)__Y, (__v8di)__X);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
-_mm512_permutexvar_epi64 (__m512i __X, __m512i __Y)
+_mm512_maskz_permutexvar_epi64 (__mmask8 __M, __m512i __X, __m512i __Y)
 {
-  return (__m512i) __builtin_ia32_permvardi512_mask ((__v8di) __Y,
-                 (__v8di) __X,
-                 (__v8di) _mm512_undefined_epi32 (),
-                 (__mmask8) -1);
+  return (__m512i)__builtin_ia32_selectq_512((__mmask8)__M,
+                                     (__v8di)_mm512_permutexvar_epi64(__X, __Y),
+                                     (__v8di)_mm512_setzero_si512());
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_permutexvar_epi64 (__m512i __W, __mmask8 __M, __m512i __X,
              __m512i __Y)
 {
-  return (__m512i) __builtin_ia32_permvardi512_mask ((__v8di) __Y,
-                 (__v8di) __X,
-                 (__v8di) __W,
-                 __M);
+  return (__m512i)__builtin_ia32_selectq_512((__mmask8)__M,
+                                     (__v8di)_mm512_permutexvar_epi64(__X, __Y),
+                                     (__v8di)__W);
 }
 
 static __inline__ __m512 __DEFAULT_FN_ATTRS
 _mm512_permutexvar_ps (__m512i __X, __m512 __Y)
 {
-  return (__m512) __builtin_ia32_permvarsf512_mask ((__v16sf) __Y,
-                (__v16si) __X,
-                (__v16sf) _mm512_undefined_ps (),
-                (__mmask16) -1);
+  return (__m512)__builtin_ia32_permvarsf512((__v16sf)__Y, (__v16si)__X);
 }
 
 static __inline__ __m512 __DEFAULT_FN_ATTRS
 _mm512_mask_permutexvar_ps (__m512 __W, __mmask16 __U, __m512i __X, __m512 __Y)
 {
-  return (__m512) __builtin_ia32_permvarsf512_mask ((__v16sf) __Y,
-                (__v16si) __X,
-                (__v16sf) __W,
-                (__mmask16) __U);
+  return (__m512)__builtin_ia32_selectps_512((__mmask16)__U,
+                                       (__v16sf)_mm512_permutexvar_ps(__X, __Y),
+                                       (__v16sf)__W);
 }
 
 static __inline__ __m512 __DEFAULT_FN_ATTRS
 _mm512_maskz_permutexvar_ps (__mmask16 __U, __m512i __X, __m512 __Y)
 {
-  return (__m512) __builtin_ia32_permvarsf512_mask ((__v16sf) __Y,
-                (__v16si) __X,
-                (__v16sf) _mm512_setzero_ps (),
-                (__mmask16) __U);
+  return (__m512)__builtin_ia32_selectps_512((__mmask16)__U,
+                                       (__v16sf)_mm512_permutexvar_ps(__X, __Y),
+                                       (__v16sf)_mm512_setzero_ps());
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
-_mm512_maskz_permutexvar_epi32 (__mmask16 __M, __m512i __X, __m512i __Y)
+_mm512_permutexvar_epi32 (__m512i __X, __m512i __Y)
 {
-  return (__m512i) __builtin_ia32_permvarsi512_mask ((__v16si) __Y,
-                 (__v16si) __X,
-                 (__v16si) _mm512_setzero_si512 (),
-                 __M);
+  return (__m512i)__builtin_ia32_permvarsi512((__v16si)__Y, (__v16si)__X);
 }
 
+#define _mm512_permutevar_epi32 _mm512_permutexvar_epi32
+
 static __inline__ __m512i __DEFAULT_FN_ATTRS
-_mm512_permutexvar_epi32 (__m512i __X, __m512i __Y)
+_mm512_maskz_permutexvar_epi32 (__mmask16 __M, __m512i __X, __m512i __Y)
 {
-  return (__m512i) __builtin_ia32_permvarsi512_mask ((__v16si) __Y,
-                 (__v16si) __X,
-                 (__v16si) _mm512_undefined_epi32 (),
-                 (__mmask16) -1);
+  return (__m512i)__builtin_ia32_selectd_512((__mmask16)__M,
+                                    (__v16si)_mm512_permutexvar_epi32(__X, __Y),
+                                    (__v16si)_mm512_setzero_si512());
 }
 
-#define _mm512_permutevar_epi32 _mm512_permutexvar_epi32
-
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_permutexvar_epi32 (__m512i __W, __mmask16 __M, __m512i __X,
              __m512i __Y)
 {
-  return (__m512i) __builtin_ia32_permvarsi512_mask ((__v16si) __Y,
-                 (__v16si) __X,
-                 (__v16si) __W,
-                 __M);
+  return (__m512i)__builtin_ia32_selectd_512((__mmask16)__M,
+                                    (__v16si)_mm512_permutexvar_epi32(__X, __Y),
+                                    (__v16si)__W);
 }
 
 #define _mm512_mask_permutevar_epi32 _mm512_mask_permutexvar_epi32
diff --git a/clang/lib/Headers/avx512vbmiintrin.h b/clang/lib/Headers/avx512vbmiintrin.h
index 837238eda97..a8ae95c7f29 100644
--- a/clang/lib/Headers/avx512vbmiintrin.h
+++ b/clang/lib/Headers/avx512vbmiintrin.h
@@ -78,30 +78,25 @@ _mm512_maskz_permutex2var_epi8 (__mmask64 __U, __m512i __A,
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_permutexvar_epi8 (__m512i __A, __m512i __B)
 {
-  return (__m512i) __builtin_ia32_permvarqi512_mask ((__v64qi) __B,
-                 (__v64qi) __A,
-                 (__v64qi) _mm512_undefined_epi32 (),
-                 (__mmask64) -1);
+  return (__m512i)__builtin_ia32_permvarqi512((__v64qi) __B, (__v64qi) __A);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_maskz_permutexvar_epi8 (__mmask64 __M, __m512i __A,
         __m512i __B)
 {
-  return (__m512i) __builtin_ia32_permvarqi512_mask ((__v64qi) __B,
-                 (__v64qi) __A,
-                 (__v64qi) _mm512_setzero_si512(),
-                 (__mmask64) __M);
+  return (__m512i)__builtin_ia32_selectb_512((__mmask64)__M,
+                                     (__v64qi)_mm512_permutexvar_epi8(__A, __B),
+                                     (__v64qi)_mm512_setzero_si512());
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_permutexvar_epi8 (__m512i __W, __mmask64 __M, __m512i __A,
              __m512i __B)
 {
-  return (__m512i) __builtin_ia32_permvarqi512_mask ((__v64qi) __B,
-                 (__v64qi) __A,
-                 (__v64qi) __W,
-                 (__mmask64) __M);
+  return (__m512i)__builtin_ia32_selectb_512((__mmask64)__M,
+                                     (__v64qi)_mm512_permutexvar_epi8(__A, __B),
+                                     (__v64qi)__W);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
diff --git a/clang/lib/Headers/avx512vbmivlintrin.h b/clang/lib/Headers/avx512vbmivlintrin.h
index 105c6d142fa..b2336554eff 100644
--- a/clang/lib/Headers/avx512vbmivlintrin.h
+++ b/clang/lib/Headers/avx512vbmivlintrin.h
@@ -129,58 +129,48 @@ _mm256_maskz_permutex2var_epi8 (__mmask32 __U, __m256i __A,
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_permutexvar_epi8 (__m128i __A, __m128i __B)
 {
-  return (__m128i) __builtin_ia32_permvarqi128_mask ((__v16qi) __B,
-                 (__v16qi) __A,
-                 (__v16qi) _mm_undefined_si128 (),
-                 (__mmask16) -1);
+  return (__m128i)__builtin_ia32_permvarqi128((__v16qi)__B, (__v16qi)__A);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_maskz_permutexvar_epi8 (__mmask16 __M, __m128i __A, __m128i __B)
 {
-  return (__m128i) __builtin_ia32_permvarqi128_mask ((__v16qi) __B,
-                 (__v16qi) __A,
-                 (__v16qi) _mm_setzero_si128 (),
-                 (__mmask16) __M);
+  return (__m128i)__builtin_ia32_selectb_128((__mmask16)__M,
+                                        (__v16qi)_mm_permutexvar_epi8(__A, __B),
+                                        (__v16qi)_mm_setzero_si128());
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mask_permutexvar_epi8 (__m128i __W, __mmask16 __M, __m128i __A,
           __m128i __B)
 {
-  return (__m128i) __builtin_ia32_permvarqi128_mask ((__v16qi) __B,
-                 (__v16qi) __A,
-                 (__v16qi) __W,
-                 (__mmask16) __M);
+  return (__m128i)__builtin_ia32_selectb_128((__mmask16)__M,
+                                        (__v16qi)_mm_permutexvar_epi8(__A, __B),
+                                        (__v16qi)__W);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_permutexvar_epi8 (__m256i __A, __m256i __B)
 {
-  return (__m256i) __builtin_ia32_permvarqi256_mask ((__v32qi) __B,
-                 (__v32qi) __A,
-                 (__v32qi) _mm256_undefined_si256 (),
-                 (__mmask32) -1);
+  return (__m256i)__builtin_ia32_permvarqi256((__v32qi) __B, (__v32qi) __A);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_maskz_permutexvar_epi8 (__mmask32 __M, __m256i __A,
         __m256i __B)
 {
-  return (__m256i) __builtin_ia32_permvarqi256_mask ((__v32qi) __B,
-                 (__v32qi) __A,
-                 (__v32qi) _mm256_setzero_si256 (),
-                 (__mmask32) __M);
+  return (__m256i)__builtin_ia32_selectb_256((__mmask32)__M,
+                                     (__v32qi)_mm256_permutexvar_epi8(__A, __B),
+                                     (__v32qi)_mm256_setzero_si256());
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_mask_permutexvar_epi8 (__m256i __W, __mmask32 __M, __m256i __A,
              __m256i __B)
 {
-  return (__m256i) __builtin_ia32_permvarqi256_mask ((__v32qi) __B,
-                 (__v32qi) __A,
-                 (__v32qi) __W,
-                 (__mmask32) __M);
+  return (__m256i)__builtin_ia32_selectb_256((__mmask32)__M,
+                                     (__v32qi)_mm256_permutexvar_epi8(__A, __B),
+                                     (__v32qi)__W);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
diff --git a/clang/lib/Headers/avx512vlbwintrin.h b/clang/lib/Headers/avx512vlbwintrin.h
index 36b19b99aba..6c2d3eadc9c 100644
--- a/clang/lib/Headers/avx512vlbwintrin.h
+++ b/clang/lib/Headers/avx512vlbwintrin.h
@@ -2664,58 +2664,48 @@ _mm_maskz_set1_epi16 (__mmask8 __M, short __A)
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_permutexvar_epi16 (__m128i __A, __m128i __B)
 {
-  return (__m128i) __builtin_ia32_permvarhi128_mask ((__v8hi) __B,
-                 (__v8hi) __A,
-                 (__v8hi) _mm_undefined_si128 (),
-                 (__mmask8) -1);
+  return (__m128i)__builtin_ia32_permvarhi128((__v8hi) __B, (__v8hi) __A);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_maskz_permutexvar_epi16 (__mmask8 __M, __m128i __A, __m128i __B)
 {
-  return (__m128i) __builtin_ia32_permvarhi128_mask ((__v8hi) __B,
-                 (__v8hi) __A,
-                 (__v8hi) _mm_setzero_si128 (),
-                 (__mmask8) __M);
+  return (__m128i)__builtin_ia32_selectw_128((__mmask8)__M,
+                                        (__v8hi)_mm_permutexvar_epi16(__A, __B),
+                                        (__v8hi) _mm_setzero_si128());
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mask_permutexvar_epi16 (__m128i __W, __mmask8 __M, __m128i __A,
           __m128i __B)
 {
-  return (__m128i) __builtin_ia32_permvarhi128_mask ((__v8hi) __B,
-                 (__v8hi) __A,
-                 (__v8hi) __W,
-                 (__mmask8) __M);
+  return (__m128i)__builtin_ia32_selectw_128((__mmask8)__M,
+                                        (__v8hi)_mm_permutexvar_epi16(__A, __B),
+                                        (__v8hi)__W);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_permutexvar_epi16 (__m256i __A, __m256i __B)
 {
-  return (__m256i) __builtin_ia32_permvarhi256_mask ((__v16hi) __B,
-                 (__v16hi) __A,
-                 (__v16hi) _mm256_undefined_si256 (),
-                 (__mmask16) -1);
+  return (__m256i)__builtin_ia32_permvarhi256((__v16hi) __B, (__v16hi) __A);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_maskz_permutexvar_epi16 (__mmask16 __M, __m256i __A,
         __m256i __B)
 {
-  return (__m256i) __builtin_ia32_permvarhi256_mask ((__v16hi) __B,
-                 (__v16hi) __A,
-                 (__v16hi) _mm256_setzero_si256 (),
-                 (__mmask16) __M);
+  return (__m256i)__builtin_ia32_selectw_256((__mmask16)__M,
+                                    (__v16hi)_mm256_permutexvar_epi16(__A, __B),
+                                    (__v16hi)_mm256_setzero_si256());
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_mask_permutexvar_epi16 (__m256i __W, __mmask16 __M, __m256i __A,
              __m256i __B)
 {
-  return (__m256i) __builtin_ia32_permvarhi256_mask ((__v16hi) __B,
-                 (__v16hi) __A,
-                 (__v16hi) __W,
-                 (__mmask16) __M);
+  return (__m256i)__builtin_ia32_selectw_256((__mmask16)__M,
+                                    (__v16hi)_mm256_permutexvar_epi16(__A, __B),
+                                    (__v16hi)__W);
 }
 
 #define _mm_mask_alignr_epi8(W, U, A, B, N) __extension__ ({ \
diff --git a/clang/lib/Headers/avx512vlintrin.h b/clang/lib/Headers/avx512vlintrin.h
index de89cfdfacb..213d911fd5d 100644
--- a/clang/lib/Headers/avx512vlintrin.h
+++ b/clang/lib/Headers/avx512vlintrin.h
@@ -8108,57 +8108,47 @@ _mm256_mask_cvtepi64_storeu_epi16 (void * __P, __mmask8 __M, __m256i __A)
 static __inline__ __m256d __DEFAULT_FN_ATTRS
 _mm256_permutexvar_pd (__m256i __X, __m256d __Y)
 {
-  return (__m256d) __builtin_ia32_permvardf256_mask ((__v4df) __Y,
-                 (__v4di) __X,
-                 (__v4df) _mm256_undefined_si256 (),
-                 (__mmask8) -1);
+  return (__m256d)__builtin_ia32_permvardf256((__v4df)__Y, (__v4di)__X);
 }
 
 static __inline__ __m256d __DEFAULT_FN_ATTRS
 _mm256_mask_permutexvar_pd (__m256d __W, __mmask8 __U, __m256i __X,
           __m256d __Y)
 {
-  return (__m256d) __builtin_ia32_permvardf256_mask ((__v4df) __Y,
-                 (__v4di) __X,
-                 (__v4df) __W,
-                 (__mmask8) __U);
+  return (__m256d)__builtin_ia32_selectpd_256((__mmask8)__U,
+                                        (__v4df)_mm256_permutexvar_pd(__X, __Y),
+                                        (__v4df)__W);
 }
 
 static __inline__ __m256d __DEFAULT_FN_ATTRS
 _mm256_maskz_permutexvar_pd (__mmask8 __U, __m256i __X, __m256d __Y)
 {
-  return (__m256d) __builtin_ia32_permvardf256_mask ((__v4df) __Y,
-                 (__v4di) __X,
-                 (__v4df) _mm256_setzero_pd (),
-                 (__mmask8) __U);
+  return (__m256d)__builtin_ia32_selectpd_256((__mmask8)__U,
+                                        (__v4df)_mm256_permutexvar_pd(__X, __Y),
+                                        (__v4df)_mm256_setzero_pd());
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
-_mm256_maskz_permutexvar_epi64 (__mmask8 __M, __m256i __X, __m256i __Y)
+_mm256_permutexvar_epi64 ( __m256i __X, __m256i __Y)
 {
-  return (__m256i) __builtin_ia32_permvardi256_mask ((__v4di) __Y,
-                 (__v4di) __X,
-                 (__v4di) _mm256_setzero_si256 (),
-                 (__mmask8) __M);
+  return (__m256i)__builtin_ia32_permvardi256((__v4di) __Y, (__v4di) __X);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
-_mm256_permutexvar_epi64 ( __m256i __X, __m256i __Y)
+_mm256_maskz_permutexvar_epi64 (__mmask8 __M, __m256i __X, __m256i __Y)
 {
-  return (__m256i) __builtin_ia32_permvardi256_mask ((__v4di) __Y,
-                 (__v4di) __X,
-                 (__v4di) _mm256_undefined_si256 (),
-                 (__mmask8) -1);
+  return (__m256i)__builtin_ia32_selectq_256((__mmask8)__M,
+                                     (__v4di)_mm256_permutexvar_epi64(__X, __Y),
+                                     (__v4di)_mm256_setzero_si256());
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_mask_permutexvar_epi64 (__m256i __W, __mmask8 __M, __m256i __X,
              __m256i __Y)
 {
-  return (__m256i) __builtin_ia32_permvardi256_mask ((__v4di) __Y,
-                 (__v4di) __X,
-                 (__v4di) __W,
-                 __M);
+  return (__m256i)__builtin_ia32_selectq_256((__mmask8)__M,
+                                     (__v4di)_mm256_permutexvar_epi64(__X, __Y),
+                                     (__v4di)__W);
 }
 
 #define _mm256_permutexvar_ps(A, B) _mm256_permutevar8x32_ps((B), (A))