From f46e01bb236330f9be8ae473ab4930045bba7f3b Mon Sep 17 00:00:00 2001
From: Craig Topper <craig.topper@gmail.com>
Date: Sun, 26 Feb 2017 06:45:51 +0000
Subject: [PATCH] [AVX-512] Fix the execution domain for AVX-512 integer
 broadcasts.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@296290 91177308-0d34-0410-b5e6-96231b3b80d8
---
 lib/Target/X86/X86InstrAVX512.td            | 1 +
 test/CodeGen/X86/avx2-vbroadcast.ll         | 8 ++++----
 test/Transforms/LoopVectorize/X86/avx512.ll | 2 +-
 3 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/lib/Target/X86/X86InstrAVX512.td b/lib/Target/X86/X86InstrAVX512.td
index 3f5433de423..a66908a41ac 100644
--- a/lib/Target/X86/X86InstrAVX512.td
+++ b/lib/Target/X86/X86InstrAVX512.td
@@ -965,6 +965,7 @@ def : Pat<(int_x86_avx512_vbroadcast_sd_512 addr:$src),
 multiclass avx512_int_broadcast_reg<bits<8> opc, X86VectorVTInfo _,
                                     SDPatternOperator OpNode,
                                     RegisterClass SrcRC> {
+  let ExeDomain = _.ExeDomain in
   defm r : AVX512_maskable<opc, MRMSrcReg, _, (outs _.RC:$dst),
                          (ins SrcRC:$src),
                          "vpbroadcast"##_.Suffix, "$src", "$src",
diff --git a/test/CodeGen/X86/avx2-vbroadcast.ll b/test/CodeGen/X86/avx2-vbroadcast.ll
index f65f485cc62..46e4357cd77 100644
--- a/test/CodeGen/X86/avx2-vbroadcast.ll
+++ b/test/CodeGen/X86/avx2-vbroadcast.ll
@@ -1419,7 +1419,7 @@ define void @isel_crash_4d(i32* %cV_R.addr) {
 ; X64-AVX512VL-NEXT:    movl (%rdi), %eax
 ; X64-AVX512VL-NEXT:    vpbroadcastd %eax, %xmm1
 ; X64-AVX512VL-NEXT:    vmovaps %xmm0, -{{[0-9]+}}(%rsp)
-; X64-AVX512VL-NEXT:    vmovaps %xmm1, -{{[0-9]+}}(%rsp)
+; X64-AVX512VL-NEXT:    vmovdqa %xmm1, -{{[0-9]+}}(%rsp)
 ; X64-AVX512VL-NEXT:    retq
 entry:
   %__a.addr.i = alloca <2 x i64>, align 16
@@ -1523,7 +1523,7 @@ define void @isel_crash_8d(i32* %cV_R.addr) {
 ; X64-AVX512VL-NEXT:    movl (%rdi), %eax
 ; X64-AVX512VL-NEXT:    vpbroadcastd %eax, %ymm1
 ; X64-AVX512VL-NEXT:    vmovaps %ymm0, {{[0-9]+}}(%rsp)
-; X64-AVX512VL-NEXT:    vmovaps %ymm1, {{[0-9]+}}(%rsp)
+; X64-AVX512VL-NEXT:    vmovdqa %ymm1, {{[0-9]+}}(%rsp)
 ; X64-AVX512VL-NEXT:    movq %rbp, %rsp
 ; X64-AVX512VL-NEXT:    popq %rbp
 ; X64-AVX512VL-NEXT:    retq
@@ -1580,7 +1580,7 @@ define void @isel_crash_2q(i64* %cV_R.addr) {
 ; X64-AVX512VL-NEXT:    movq (%rdi), %rax
 ; X64-AVX512VL-NEXT:    vpbroadcastq %rax, %xmm1
 ; X64-AVX512VL-NEXT:    vmovaps %xmm0, -{{[0-9]+}}(%rsp)
-; X64-AVX512VL-NEXT:    vmovaps %xmm1, -{{[0-9]+}}(%rsp)
+; X64-AVX512VL-NEXT:    vmovdqa %xmm1, -{{[0-9]+}}(%rsp)
 ; X64-AVX512VL-NEXT:    retq
 entry:
   %__a.addr.i = alloca <2 x i64>, align 16
@@ -1695,7 +1695,7 @@ define void @isel_crash_4q(i64* %cV_R.addr) {
 ; X64-AVX512VL-NEXT:    movq (%rdi), %rax
 ; X64-AVX512VL-NEXT:    vpbroadcastq %rax, %ymm1
 ; X64-AVX512VL-NEXT:    vmovaps %ymm0, {{[0-9]+}}(%rsp)
-; X64-AVX512VL-NEXT:    vmovaps %ymm1, {{[0-9]+}}(%rsp)
+; X64-AVX512VL-NEXT:    vmovdqa %ymm1, {{[0-9]+}}(%rsp)
 ; X64-AVX512VL-NEXT:    movq %rbp, %rsp
 ; X64-AVX512VL-NEXT:    popq %rbp
 ; X64-AVX512VL-NEXT:    retq
diff --git a/test/Transforms/LoopVectorize/X86/avx512.ll b/test/Transforms/LoopVectorize/X86/avx512.ll
index fb01454c253..1eb1cd3f5d7 100644
--- a/test/Transforms/LoopVectorize/X86/avx512.ll
+++ b/test/Transforms/LoopVectorize/X86/avx512.ll
@@ -7,7 +7,7 @@ target triple = "x86_64-apple-macosx10.9.0"
 ; loop.
 
 ; CHECK-LABEL: f:
-; CHECK: vmovups %zmm{{.}},
+; CHECK: vmovdqu32 %zmm{{.}},
 ; CHECK-NOT: %ymm
 
 define void @f(i32* %a, i32 %n) {
-- 
2.11.0