[AVX512] Fix return types on int_x86_avx512_gatherXXX_di intrinsics

[cameron.mcinally]

Summary

The int_x86_avx512_gatherXXX_di intrinsics have incorrect return types. The return type should be the same as the passthrough vector type for each.

[delena]

Clang should be fixed as well.
Thanks.

[cameron.mcinally]

Thanks for the fast review.

I'm pretty new to Phab. Should I attach the Clang diff to this Differential or create an additional Differential against Clang?

include/llvm/IR/IntrinsicsX86.td

let TargetPrefix = "x86" in {
   def int_x86_avx512_gather3div2_df :
         GCCBuiltin<"__builtin_ia32_gather3div2df">,
           Intrinsic<[llvm_v2f64_ty],
           [llvm_v2f64_ty, llvm_ptr_ty, llvm_v2i64_ty, llvm_i8_ty, llvm_i32_ty],
           [IntrReadMem, IntrArgMemOnly]>;
 
   def int_x86_avx512_gather3div2_di :
         GCCBuiltin<"__builtin_ia32_gather3div2di">,
-          Intrinsic<[llvm_v4i32_ty],
+          Intrinsic<[llvm_v2i64_ty],
           [llvm_v2i64_ty, llvm_ptr_ty, llvm_v2i64_ty, llvm_i8_ty, llvm_i32_ty],
           [IntrReadMem, IntrArgMemOnly]>;
 
   def int_x86_avx512_gather3div4_df :
         GCCBuiltin<"__builtin_ia32_gather3div4df">,
           Intrinsic<[llvm_v4f64_ty],
           [llvm_v4f64_ty, llvm_ptr_ty, llvm_v4i64_ty, llvm_i8_ty, llvm_i32_ty],
           [IntrReadMem, IntrArgMemOnly]>;
 
   def int_x86_avx512_gather3div4_di :
         GCCBuiltin<"__builtin_ia32_gather3div4di">,
-          Intrinsic<[llvm_v8i32_ty],
+          Intrinsic<[llvm_v4i64_ty],
           [llvm_v4i64_ty, llvm_ptr_ty, llvm_v4i64_ty, llvm_i8_ty, llvm_i32_ty],
           [IntrReadMem, IntrArgMemOnly]>;
 
   def int_x86_avx512_gather3div4_sf :
         GCCBuiltin<"__builtin_ia32_gather3div4sf">,
           Intrinsic<[llvm_v4f32_ty],
           [llvm_v4f32_ty, llvm_ptr_ty, llvm_v2i64_ty, llvm_i8_ty, llvm_i32_ty],
           [IntrReadMem, IntrArgMemOnly]>;
let TargetPrefix = "x86" in {
   def int_x86_avx512_gather3siv2_df :
         GCCBuiltin<"__builtin_ia32_gather3siv2df">,
           Intrinsic<[llvm_v2f64_ty],
           [llvm_v2f64_ty, llvm_ptr_ty, llvm_v4i32_ty, llvm_i8_ty, llvm_i32_ty],
           [IntrReadMem, IntrArgMemOnly]>;
 
   def int_x86_avx512_gather3siv2_di :
         GCCBuiltin<"__builtin_ia32_gather3siv2di">,
-          Intrinsic<[llvm_v4i32_ty],
+          Intrinsic<[llvm_v2i64_ty],
           [llvm_v2i64_ty, llvm_ptr_ty, llvm_v4i32_ty, llvm_i8_ty, llvm_i32_ty],
           [IntrReadMem, IntrArgMemOnly]>;
 
   def int_x86_avx512_gather3siv4_df :
         GCCBuiltin<"__builtin_ia32_gather3siv4df">,
           Intrinsic<[llvm_v4f64_ty],
           [llvm_v4f64_ty, llvm_ptr_ty, llvm_v4i32_ty, llvm_i8_ty, llvm_i32_ty],
           [IntrReadMem, IntrArgMemOnly]>;
 
   def int_x86_avx512_gather3siv4_di :
         GCCBuiltin<"__builtin_ia32_gather3siv4di">,
-          Intrinsic<[llvm_v8i32_ty],
+          Intrinsic<[llvm_v4i64_ty],
           [llvm_v4i64_ty, llvm_ptr_ty, llvm_v4i32_ty, llvm_i8_ty, llvm_i32_ty],
           [IntrReadMem, IntrArgMemOnly]>;
 
   def int_x86_avx512_gather3siv4_sf :
         GCCBuiltin<"__builtin_ia32_gather3siv4sf">,
           Intrinsic<[llvm_v4f32_ty],
           [llvm_v4f32_ty, llvm_ptr_ty, llvm_v4i32_ty, llvm_i8_ty, llvm_i32_ty],
           [IntrReadMem, IntrArgMemOnly]>;

test/CodeGen/X86/avx512-gather-scatter-intrin.ll

 ; CHECK-NEXT:    vaddpd %xmm0, %xmm2, %xmm0
 ; CHECK-NEXT:    retq
   %res = call <2 x double> @llvm.x86.avx512.gather3div2.df(<2 x double> %x0, i8* %x1, <2 x i64> %x2, i8 %x3, i32 4)
   %res1 = call <2 x double> @llvm.x86.avx512.gather3div2.df(<2 x double> %x0, i8* %x1, <2 x i64> %x2, i8 -1, i32 2)
   %res2 = fadd <2 x double> %res, %res1
   ret <2 x double> %res2
 }
 
-declare <4 x i32> @llvm.x86.avx512.gather3div2.di(<2 x i64>, i8*, <2 x i64>, i8, i32)
+declare <2 x i64> @llvm.x86.avx512.gather3div2.di(<2 x i64>, i8*, <2 x i64>, i8, i32)
 
-define <4 x i32>@test_int_x86_avx512_gather3div2_di(<2 x i64> %x0, i8* %x1, <2 x i64> %x2, i8 %x3) {
+define <2 x i64>@test_int_x86_avx512_gather3div2_di(<2 x i64> %x0, i8* %x1, <2 x i64> %x2, i8 %x3) {
 ; CHECK-LABEL: test_int_x86_avx512_gather3div2_di:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    kmovb %esi, %k1
 ; CHECK-NEXT:    vpgatherqq (%rdi,%xmm1,8), %xmm0 {%k1}
-; CHECK-NEXT:    vpaddd %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpaddq  %xmm0, %xmm0, %xmm0
 ; CHECK-NEXT:    retq
-  %res = call <4 x i32> @llvm.x86.avx512.gather3div2.di(<2 x i64> %x0, i8* %x1, <2 x i64> %x2, i8 %x3, i32 8)
-  %res1 = call <4 x i32> @llvm.x86.avx512.gather3div2.di(<2 x i64> %x0, i8* %x1, <2 x i64> %x2, i8 %x3, i32 8)
-  %res2 = add <4 x i32> %res, %res1
-  ret <4 x i32> %res2
+  %res = call <2 x i64> @llvm.x86.avx512.gather3div2.di(<2 x i64> %x0, i8* %x1, <2 x i64> %x2, i8 %x3, i32 8)
+  %res1 = call <2 x i64> @llvm.x86.avx512.gather3div2.di(<2 x i64> %x0, i8* %x1, <2 x i64> %x2, i8 %x3, i32 8)
+  %res2 = add <2 x i64> %res, %res1
+  ret <2 x i64> %res2
 }
 
 declare <4 x double> @llvm.x86.avx512.gather3div4.df(<4 x double>, i8*, <4 x i64>, i8, i32)
 
 define <4 x double>@test_int_x86_avx512_gather3div4_df(<4 x double> %x0, i8* %x1, <4 x i64> %x2, i8 %x3) {
 ; CHECK-LABEL: test_int_x86_avx512_gather3div4_df:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    kmovb %esi, %k1
 ; CHECK-NEXT:    vmovapd %ymm0, %ymm2
 ; CHECK-NEXT:    vgatherqpd (%rdi,%ymm1,4), %ymm2 {%k1}
 ; CHECK-NEXT:    kxnorw %k0, %k0, %k1
 ; CHECK-NEXT:    vgatherqpd (%rdi,%ymm1,2), %ymm0 {%k1}
 ; CHECK-NEXT:    vaddpd %ymm0, %ymm2, %ymm0
 ; CHECK-NEXT:    retq
   %res = call <4 x double> @llvm.x86.avx512.gather3div4.df(<4 x double> %x0, i8* %x1, <4 x i64> %x2, i8 %x3, i32 4)
   %res1 = call <4 x double> @llvm.x86.avx512.gather3div4.df(<4 x double> %x0, i8* %x1, <4 x i64> %x2, i8 -1, i32 2)
   %res2 = fadd <4 x double> %res, %res1
   ret <4 x double> %res2
 }
 
-declare <8 x i32> @llvm.x86.avx512.gather3div4.di(<4 x i64>, i8*, <4 x i64>, i8, i32)
+declare <4 x i64> @llvm.x86.avx512.gather3div4.di(<4 x i64>, i8*, <4 x i64>, i8, i32)
 
-define <8 x i32>@test_int_x86_avx512_gather3div4_di(<4 x i64> %x0, i8* %x1, <4 x i64> %x2, i8 %x3) {
+define <4 x i64>@test_int_x86_avx512_gather3div4_di(<4 x i64> %x0, i8* %x1, <4 x i64> %x2, i8 %x3) {
 ; CHECK-LABEL: test_int_x86_avx512_gather3div4_di:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    kmovb %esi, %k1
 ; CHECK-NEXT:    vmovdqa64 %ymm0, %ymm2
 ; CHECK-NEXT:    vpgatherqq (%rdi,%ymm1,8), %ymm2 {%k1}
 ; CHECK-NEXT:    kxnorw %k0, %k0, %k1
 ; CHECK-NEXT:    vpgatherqq (%rdi,%ymm1,8), %ymm0 {%k1}
-; CHECK-NEXT:    vpaddd %ymm0, %ymm2, %ymm0
+; CHECK-NEXT:    vpaddq  %ymm0, %ymm2, %ymm0 
 ; CHECK-NEXT:    retq
-  %res = call <8 x i32> @llvm.x86.avx512.gather3div4.di(<4 x i64> %x0, i8* %x1, <4 x i64> %x2, i8 %x3, i32 8)
-  %res1 = call <8 x i32> @llvm.x86.avx512.gather3div4.di(<4 x i64> %x0, i8* %x1, <4 x i64> %x2, i8 -1, i32 8)
-  %res2 = add <8 x i32> %res, %res1
-  ret <8 x i32> %res2
+  %res = call <4 x i64> @llvm.x86.avx512.gather3div4.di(<4 x i64> %x0, i8* %x1, <4 x i64> %x2, i8 %x3, i32 8)
+  %res1 = call <4 x i64> @llvm.x86.avx512.gather3div4.di(<4 x i64> %x0, i8* %x1, <4 x i64> %x2, i8 -1, i32 8)
+  %res2 = add <4 x i64> %res, %res1
+  ret <4 x i64> %res2
 }
 
 declare <4 x float> @llvm.x86.avx512.gather3div4.sf(<4 x float>, i8*, <2 x i64>, i8, i32)
 
 define <4 x float>@test_int_x86_avx512_gather3div4_sf(<4 x float> %x0, i8* %x1, <2 x i64> %x2, i8 %x3) {
 ; CHECK-LABEL: test_int_x86_avx512_gather3div4_sf:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    kmovb %esi, %k1

 ; CHECK-NEXT:    vaddpd %xmm0, %xmm2, %xmm0
 ; CHECK-NEXT:    retq
   %res = call <2 x double> @llvm.x86.avx512.gather3siv2.df(<2 x double> %x0, i8* %x1, <4 x i32> %x2, i8 %x3, i32 4)
   %res1 = call <2 x double> @llvm.x86.avx512.gather3siv2.df(<2 x double> %x0, i8* %x1, <4 x i32> %x2, i8 -1, i32 2)
   %res2 = fadd <2 x double> %res, %res1
   ret <2 x double> %res2
 }
 
-declare <4 x i32> @llvm.x86.avx512.gather3siv2.di(<2 x i64>, i8*, <4 x i32>, i8, i32)
+declare <2 x i64> @llvm.x86.avx512.gather3siv2.di(<2 x i64>, i8*, <4 x i32>, i8, i32)
 
-define <4 x i32>@test_int_x86_avx512_gather3siv2_di(<2 x i64> %x0, i8* %x1, <4 x i32> %x2, i8 %x3) {
+define <2 x i64>@test_int_x86_avx512_gather3siv2_di(<2 x i64> %x0, i8* %x1, <4 x i32> %x2, i8 %x3) {
 ; CHECK-LABEL: test_int_x86_avx512_gather3siv2_di:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    kmovb %esi, %k1
 ; CHECK-NEXT:    vpgatherdq (%rdi,%xmm1,8), %xmm0 {%k1}
-; CHECK-NEXT:    vpaddd %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpaddq %xmm0, %xmm0, %xmm0
 ; CHECK-NEXT:    retq
-  %res = call <4 x i32> @llvm.x86.avx512.gather3siv2.di(<2 x i64> %x0, i8* %x1, <4 x i32> %x2, i8 %x3, i32 8)
-  %res1 = call <4 x i32> @llvm.x86.avx512.gather3siv2.di(<2 x i64> %x0, i8* %x1, <4 x i32> %x2, i8 %x3, i32 8)
-  %res2 = add <4 x i32> %res, %res1
-  ret <4 x i32> %res2
+  %res = call <2 x i64> @llvm.x86.avx512.gather3siv2.di(<2 x i64> %x0, i8* %x1, <4 x i32> %x2, i8 %x3, i32 8)
+  %res1 = call <2 x i64> @llvm.x86.avx512.gather3siv2.di(<2 x i64> %x0, i8* %x1, <4 x i32> %x2, i8 %x3, i32 8)
+  %res2 = add <2 x i64> %res, %res1
+  ret <2 x i64> %res2
 }
 
 declare <4 x double> @llvm.x86.avx512.gather3siv4.df(<4 x double>, i8*, <4 x i32>, i8, i32)
 
 define <4 x double>@test_int_x86_avx512_gather3siv4_df(<4 x double> %x0, i8* %x1, <4 x i32> %x2, i8 %x3) {
 ; CHECK-LABEL: test_int_x86_avx512_gather3siv4_df:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    kmovb %esi, %k1
 ; CHECK-NEXT:    vmovapd %ymm0, %ymm2
 ; CHECK-NEXT:    vgatherdpd (%rdi,%xmm1,4), %ymm2 {%k1}
 ; CHECK-NEXT:    kxnorw %k0, %k0, %k1
 ; CHECK-NEXT:    vgatherdpd (%rdi,%xmm1,2), %ymm0 {%k1}
 ; CHECK-NEXT:    vaddpd %ymm0, %ymm2, %ymm0
 ; CHECK-NEXT:    retq
   %res = call <4 x double> @llvm.x86.avx512.gather3siv4.df(<4 x double> %x0, i8* %x1, <4 x i32> %x2, i8 %x3, i32 4)
   %res1 = call <4 x double> @llvm.x86.avx512.gather3siv4.df(<4 x double> %x0, i8* %x1, <4 x i32> %x2, i8 -1, i32 2)
   %res2 = fadd <4 x double> %res, %res1
   ret <4 x double> %res2
 }
 
-declare <8 x i32> @llvm.x86.avx512.gather3siv4.di(<4 x i64>, i8*, <4 x i32>, i8, i32)
+declare <4 x i64> @llvm.x86.avx512.gather3siv4.di(<4 x i64>, i8*, <4 x i32>, i8, i32)
 
-define <8 x i32>@test_int_x86_avx512_gather3siv4_di(<4 x i64> %x0, i8* %x1, <4 x i32> %x2, i8 %x3) {
+define <4 x i64>@test_int_x86_avx512_gather3siv4_di(<4 x i64> %x0, i8* %x1, <4 x i32> %x2, i8 %x3) {
 ; CHECK-LABEL: test_int_x86_avx512_gather3siv4_di:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    kmovb %esi, %k1
 ; CHECK-NEXT:    vpgatherdq (%rdi,%xmm1,8), %ymm0 {%k1}
-; CHECK-NEXT:    vpaddd %ymm0, %ymm0, %ymm0
+; CHECK-NEXT:    vpaddq  %ymm0, %ymm0, %ymm0 
 ; CHECK-NEXT:    retq
-  %res = call <8 x i32> @llvm.x86.avx512.gather3siv4.di(<4 x i64> %x0, i8* %x1, <4 x i32> %x2, i8 %x3, i32 8)
-  %res1 = call <8 x i32> @llvm.x86.avx512.gather3siv4.di(<4 x i64> %x0, i8* %x1, <4 x i32> %x2, i8 %x3, i32 8)
-  %res2 = add <8 x i32> %res, %res1
-  ret <8 x i32> %res2
+  %res = call <4 x i64> @llvm.x86.avx512.gather3siv4.di(<4 x i64> %x0, i8* %x1, <4 x i32> %x2, i8 %x3, i32 8)
+  %res1 = call <4 x i64> @llvm.x86.avx512.gather3siv4.di(<4 x i64> %x0, i8* %x1, <4 x i32> %x2, i8 %x3, i32 8)
+  %res2 = add <4 x i64> %res, %res1
+  ret <4 x i64> %res2
 }
 
 declare <4 x float> @llvm.x86.avx512.gather3siv4.sf(<4 x float>, i8*, <4 x i32>, i8, i32)
 
 define <4 x float>@test_int_x86_avx512_gather3siv4_sf(<4 x float> %x0, i8* %x1, <4 x i32> %x2, i8 %x3) {
 ; CHECK-LABEL: test_int_x86_avx512_gather3siv4_sf:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    kmovb %esi, %k1