[Headers] Add tests for _mm256_insert_epi64 and fix its definition

[filcab]

Summary

The definition for _mm256_insert_epi64 was taking an int, which would get
truncated before being inserted in the vector.

[filcab]

Ping + adding Craig Topper, since he seems active in this area.

[filcab]

Ping! + Updated with attribution I forgot.

[craig.topper]

LGTM. Sorry I didn't notice this review request earlier.

[filcab]

It happens. Thanks for the review!

cfe/trunk/lib/Headers/avxintrin.h

 {
   __v32qi __c = (__v32qi)__a;
   __c[__imm & 31] = __b;
   return (__m256i)__c;
 }
 
 #ifdef __x86_64__
 static __inline __m256i __attribute__((__always_inline__, __nodebug__))
-_mm256_insert_epi64(__m256i __a, int __b, int const __imm)
+_mm256_insert_epi64(__m256i __a, long __b, int const __imm)
 {
   __v4di __c = (__v4di)__a;
   __c[__imm & 3] = __b;
   return (__m256i)__c;
 }
 #endif
 
 /* Conversion */

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

__m256d test_256_blend_pd(__m256d __a, __m256d __b) {
   return _mm256_blend_pd(__a, __b, 0x35);
 }
 
 __m256 test_256_blend_ps(__m256 __a, __m256 __b) {
   // CHECK-LABEL: @test_256_blend_ps
   // CHECK: shufflevector <8 x float> %{{.*}}, <8 x float> %{{.*}}, <8 x i32> <i32 8, i32 1, i32 10, i32 3, i32 12, i32 13, i32 6, i32 7>
   return _mm256_blend_ps(__a, __b, 0x35);
 }
+
+__m256i test_256_insert_epi8(__m256i __a) {
+  // CHECK-LABEL: @test_256_insert_epi8
+  // CHECK: insertelement <32 x i8> {{.*}}, i8 {{.*}}, i32 {{.*}}
+  return _mm256_insert_epi8(__a, 42, 3);
+}
+
+__m256i test_256_insert_epi16(__m256i __a) {
+  // CHECK-LABEL: @test_256_insert_epi16
+  // CHECK: insertelement <16 x i16> {{.*}}, i16 {{.*}}, i32 {{.*}}
+  return _mm256_insert_epi16(__a, 42, 3);
+}
+
+__m256i test_256_insert_epi32(__m256i __a) {
+  // CHECK-LABEL: @test_256_insert_epi32
+  // CHECK: insertelement <8 x i32> {{.*}}, i32 {{.*}}, i32 {{.*}}
+  return _mm256_insert_epi32(__a, 42, 3);
+}
+
+__m256i test_256_insert_epi64(__m256i __a) {
+  // CHECK-LABEL: @test_256_insert_epi64
+  // CHECK: insertelement <4 x i64> {{.*}}, i64 {{.*}}, i32 {{.*}}
+  return _mm256_insert_epi64(__a, 42, 3);
+}