[X86] Add all intrinsics for scalar rsqrt28/rcp28 to avx512erintrin.h. Add parenthese...

llvm-svn: 227722

diff --git a/clang/lib/Headers/avx512erintrin.h b/clang/lib/Headers/avx512erintrin.h
index 54db997..323f4f0 100644 (file)
--- a/clang/lib/Headers/avx512erintrin.h
+++ b/clang/lib/Headers/avx512erintrin.h
@@ -45,13 +45,13 @@
                                          (__mmask8)(M), (R)); })
 
 #define _mm512_rsqrt28_pd(A) \
-  _mm512_rsqrt28_round_pd(A, _MM_FROUND_CUR_DIRECTION)
+  _mm512_rsqrt28_round_pd((A), _MM_FROUND_CUR_DIRECTION)
 
 #define _mm512_mask_rsqrt28_pd(S, M, A) \
-  _mm512_mask_rsqrt28_round_pd(S, M, A, _MM_FROUND_CUR_DIRECTION)
+  _mm512_mask_rsqrt28_round_pd((S), (M), (A), _MM_FROUND_CUR_DIRECTION)
 
 #define _mm512_maskz_rsqrt28_pd(M, A) \
-  _mm512_maskz_rsqrt28_round_pd(M, A, _MM_FROUND_CUR_DIRECTION)
+  _mm512_maskz_rsqrt28_round_pd((M), (A), _MM_FROUND_CUR_DIRECTION)
 
 #define _mm512_rsqrt28_round_ps(A, R) __extension__ ({ \
   (__m512)__builtin_ia32_rsqrt28ps_mask((__v16sf)(__m512)(A), \
@@ -69,13 +69,13 @@
                                         (__mmask16)(M), (R)); })
 
 #define _mm512_rsqrt28_ps(A) \
-  _mm512_rsqrt28_round_ps(A, _MM_FROUND_CUR_DIRECTION)
+  _mm512_rsqrt28_round_ps((A), _MM_FROUND_CUR_DIRECTION)
 
 #define _mm512_mask_rsqrt28_ps(S, M, A) \
-  _mm512_mask_rsqrt28_round_ps(S, M, A, _MM_FROUND_CUR_DIRECTION)
+  _mm512_mask_rsqrt28_round_ps((S), (M), A, _MM_FROUND_CUR_DIRECTION)
 
 #define _mm512_maskz_rsqrt28_ps(M, A) \
-  _mm512_maskz_rsqrt28_round_ps(M, A, _MM_FROUND_CUR_DIRECTION)
+  _mm512_maskz_rsqrt28_round_ps((M), (A), _MM_FROUND_CUR_DIRECTION)
 
 #define _mm_rsqrt28_round_ss(A, B, R) __extension__ ({ \
   (__m128)__builtin_ia32_rsqrt28ss_mask((__v4sf)(__m128)(A), \
@@ -83,12 +83,53 @@
                                         (__v4sf)_mm_setzero_ps(), \
                                         (__mmask8)-1, (R)); })
 
+#define _mm_mask_rsqrt28_round_ss(S, M, A, B, R) __extension__ ({ \
+  (__m128)__builtin_ia32_rsqrt28ss_mask((__v4sf)(__m128)(A), \
+                                        (__v4sf)(__m128)(B), \
+                                        (__v4sf)(__m128)(S), \
+                                        (__mmask8)(M), (R)); })
+
+#define _mm_maskz_rsqrt28_round_ss(M, A, B, R) __extension__ ({ \
+  (__m128)__builtin_ia32_rsqrt28ss_mask((__v4sf)(__m128)(A), \
+                                        (__v4sf)(__m128)(B), \
+                                        (__v4sf)_mm_setzero_ps(), \
+                                        (__mmask8)(M), (R)); })
+
+#define _mm_rsqrt28_ss(A, B) \
+  _mm_rsqrt28_round_ss((A), (B), _MM_FROUND_CUR_DIRECTION)
+
+#define _mm_mask_rsqrt28_ss(S, M, A, B) \
+  _mm_mask_rsqrt28_round_ss((S), (M), (A), (B), _MM_FROUND_CUR_DIRECTION)
+
+#define _mm_maskz_rsqrt28_ss(M, A, B) \
+  _mm_maskz_rsqrt28_round_ss((M), (A), (B), _MM_FROUND_CUR_DIRECTION)
+
 #define _mm_rsqrt28_round_sd(A, B, R) __extension__ ({ \
   (__m128d)__builtin_ia32_rsqrt28sd_mask((__v2df)(__m128d)(A), \
                                          (__v2df)(__m128d)(B), \
                                          (__v2df)_mm_setzero_pd(), \
                                          (__mmask8)-1, (R)); })
 
+#define _mm_mask_rsqrt28_round_sd(S, M, A, B, R) __extension__ ({ \
+  (__m128d)__builtin_ia32_rsqrt28sd_mask((__v2df)(__m128d)(A), \
+                                         (__v2df)(__m128d)(B), \
+                                         (__v2df)(__m128d)(S), \
+                                         (__mmask8)(M), (R)); })
+
+#define _mm_maskz_rsqrt28_round_sd(M, A, B, R) __extension__ ({ \
+  (__m128d)__builtin_ia32_rsqrt28sd_mask((__v2df)(__m128d)(A), \
+                                         (__v2df)(__m128d)(B), \
+                                         (__v2df)_mm_setzero_pd(), \
+                                         (__mmask8)(M), (R)); })
+
+#define _mm_rsqrt28_sd(A, B) \
+  _mm_rsqrt28_round_sd((A), (B), _MM_FROUND_CUR_DIRECTION)
+
+#define _mm_mask_rsqrt28_sd(S, M, A, B) \
+  _mm_mask_rsqrt28_round_sd((S), (M), (A), (B), _MM_FROUND_CUR_DIRECTION)
+
+#define _mm_maskz_rsqrt28_sd(M, A, B) \
+  _mm_mask_rsqrt28_round_sd((M), (A), (B), _MM_FROUND_CUR_DIRECTION)
 
 // rcp28
 #define _mm512_rcp28_round_pd(A, R) __extension__ ({ \
@@ -107,13 +148,13 @@
                                        (__mmask8)(M), (R)); })
 
 #define _mm512_rcp28_pd(A) \
-  _mm512_rcp28_round_pd(A, _MM_FROUND_CUR_DIRECTION)
+  _mm512_rcp28_round_pd((A), _MM_FROUND_CUR_DIRECTION)
 
 #define _mm512_mask_rcp28_pd(S, M, A) \
-  _mm512_mask_rcp28_round_pd(S, M, A, _MM_FROUND_CUR_DIRECTION)
+  _mm512_mask_rcp28_round_pd((S), (M), (A), _MM_FROUND_CUR_DIRECTION)
 
 #define _mm512_maskz_rcp28_pd(M, A) \
-  _mm512_maskz_rcp28_round_pd(M, A, _MM_FROUND_CUR_DIRECTION)
+  _mm512_maskz_rcp28_round_pd((M), (A), _MM_FROUND_CUR_DIRECTION)
 
 #define _mm512_rcp28_round_ps(A, R) __extension__ ({ \
   (__m512)__builtin_ia32_rcp28ps_mask((__v16sf)(__m512)(A), \
@@ -131,19 +172,40 @@
                                       (__mmask16)(M), (R)); })
 
 #define _mm512_rcp28_ps(A) \
-  _mm512_rcp28_round_ps(A, _MM_FROUND_CUR_DIRECTION)
+  _mm512_rcp28_round_ps((A), _MM_FROUND_CUR_DIRECTION)
 
 #define _mm512_mask_rcp28_ps(S, M, A) \
-  _mm512_mask_rcp28_round_ps(S, M, A, _MM_FROUND_CUR_DIRECTION)
+  _mm512_mask_rcp28_round_ps((S), (M), (A), _MM_FROUND_CUR_DIRECTION)
 
 #define _mm512_maskz_rcp28_ps(M, A) \
-  _mm512_maskz_rcp28_round_ps(M, A, _MM_FROUND_CUR_DIRECTION)
+  _mm512_maskz_rcp28_round_ps((M), (A), _MM_FROUND_CUR_DIRECTION)
 
 #define _mm_rcp28_round_ss(A, B, R) __extension__ ({ \
-  (__m128) __builtin_ia32_rcp28ss_mask ((__v4sf)(__m128)(A), \
-                                        (__v4sf)(__m128)(B), \
-                                        (__v4sf)_mm_setzero_ps(), \
-                                        (__mmask8)-1, (R)); })
+  (__m128)__builtin_ia32_rcp28ss_mask((__v4sf)(__m128)(A), \
+                                      (__v4sf)(__m128)(B), \
+                                      (__v4sf)_mm_setzero_ps(), \
+                                      (__mmask8)-1, (R)); })
+
+#define _mm_mask_rcp28_round_ss(S, M, A, B, R) __extension__ ({ \
+  (__m128)__builtin_ia32_rcp28ss_mask((__v4sf)(__m128)(A), \
+                                      (__v4sf)(__m128)(B), \
+                                      (__v4sf)(__m128)(S), \
+                                      (__mmask8)(M), (R)); })
+
+#define _mm_maskz_rcp28_round_ss(M, A, B, R) __extension__ ({ \
+  (__m128)__builtin_ia32_rcp28ss_mask((__v4sf)(__m128)(A), \
+                                      (__v4sf)(__m128)(B), \
+                                      (__v4sf)_mm_setzero_ps(), \
+                                      (__mmask8)(M), (R)); })
+
+#define _mm_rcp28_ss(A, B) \
+  _mm_rcp28_round_ss((A), (B), _MM_FROUND_CUR_DIRECTION)
+
+#define _mm_mask_rcp28_ss(S, M, A, B) \
+  _mm_mask_rcp28_round_ss((S), (M), (A), (B), _MM_FROUND_CUR_DIRECTION)
+
+#define _mm_maskz_rcp28_ss(M, A, B) \
+  _mm_maskz_rcp28_round_ss((M), (A), (B), _MM_FROUND_CUR_DIRECTION)
 
 #define _mm_rcp28_round_sd(A, B, R) __extension__ ({ \
   (__m128d)__builtin_ia32_rcp28sd_mask((__v2df)(__m128d)(A), \
@@ -151,4 +213,25 @@
                                        (__v2df)_mm_setzero_pd(), \
                                        (__mmask8)-1, (R)); })
 
+#define _mm_mask_rcp28_round_sd(S, M, A, B, R) __extension__ ({ \
+  (__m128d)__builtin_ia32_rcp28sd_mask((__v2df)(__m128d)(A), \
+                                       (__v2df)(__m128d)(B), \
+                                       (__v2df)(__m128d)(S), \
+                                       (__mmask8)(M), (R)); })
+
+#define _mm_maskz_rcp28_round_sd(M, A, B, R) __extension__ ({ \
+  (__m128d)__builtin_ia32_rcp28sd_mask((__v2df)(__m128d)(A), \
+                                       (__v2df)(__m128d)(B), \
+                                       (__v2df)_mm_setzero_pd(), \
+                                       (__mmask8)(M), (R)); })
+
+#define _mm_rcp28_sd(A, B) \
+  _mm_rcp28_round_sd((A), (B), _MM_FROUND_CUR_DIRECTION)
+
+#define _mm_mask_rcp28_sd(S, M, A, B) \
+  _mm_mask_rcp28_round_sd((S), (M), (A), (B), _MM_FROUND_CUR_DIRECTION)
+
+#define _mm_maskz_rcp28_sd(M, A, B) \
+  _mm_maskz_rcp28_round_sd((M), (A), (B), _MM_FROUND_CUR_DIRECTION)
+
 #endif // __AVX512ERINTRIN_H