[X86] add _mm_loadu_si64

[RKSimon]

tools/clang/test/CodeGen/sse2-builtins.c

1527 ↗

(On Diff #61199)

Replace the hardcoded %4 argument with a general pattern match

1530 ↗

(On Diff #61199)

Is the store/load necessary? This appears to be just the -O0 stack behaviour

[AsafBadouh]

small changes according to Simon review.

[RKSimon]

tools/clang/test/CodeGen/sse2-builtins.c

1526 ↗

(On Diff #61200)

Please can add the alignment operand to the CHECK (it should be align 1)?

[AsafBadouh]

add align to CHECK

[craig.topper]

tools/clang/test/CodeGen/sse2-builtins.c

1526 ↗

(On Diff #61661)

You need {{$}} after the 1 to indicate end of line or it can match 16 too.

[RKSimon]

LGTM

cfe/trunk/lib/Headers/emmintrin.h

 _mm_loadu_pd(double const *__dp)
 {
   struct __loadu_pd {
     __m128d __v;
   } __attribute__((__packed__, __may_alias__));
   return ((struct __loadu_pd*)__dp)->__v;
 }
 
+static __inline__ __m128i __DEFAULT_FN_ATTRS
+_mm_loadu_si64(void const *__a)
+{
+  struct __loadu_si64 {
+    long long __v;
+  } __attribute__((__packed__, __may_alias__));
+  long long __u = ((struct __loadu_si64*)__a)->__v;
+  return (__m128i){__u, 0L};
+}
+
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_load_sd(double const *__dp)
 {
   struct __mm_load_sd_struct {
     double __u;
   } __attribute__((__packed__, __may_alias__));
   double __u = ((struct __mm_load_sd_struct*)__dp)->__u;
   return (__m128d){ __u, 0 };

cfe/trunk/test/CodeGen/sse2-builtins.c

__m128d test_mm_xor_pd(__m128d A, __m128d B) {
   return _mm_xor_pd(A, B);
 }
 
 __m128i test_mm_xor_si128(__m128i A, __m128i B) {
   // CHECK-LABEL: test_mm_xor_si128
   // CHECK: xor <2 x i64> %{{.*}}, %{{.*}}
   return _mm_xor_si128(A, B);
 }
+
+__m128i test_mm_loadu_si64(void const* A) {
+  // CHECK-LABEL: test_mm_loadu_si64
+  // CHECK: load i64, i64* %{{.*}}, align 1{{$}}
+  // CHECK: insertelement <2 x i64> undef, i64 %{{.*}}, i32 0
+  // CHECK: insertelement <2 x i64> %{{.*}}, i64 0, i32 1
+  return _mm_loadu_si64(A);
+}
+