[X86] Fix the execution domain for scalar SQRT intrinsic instruction.

llvm-svn: 296284

3 files changed, 7 insertions, 7 deletions

diff --git a/llvm/lib/Target/X86/X86InstrSSE.td b/llvm/lib/Target/X86/X86InstrSSE.td
index 9764b05a859..0c169798289 100644
--- a/llvm/lib/Target/X86/X86InstrSSE.td
+++ b/llvm/lib/Target/X86/X86InstrSSE.td
@@ -3333,7 +3333,7 @@ multiclass sse_fp_unop_s<bits<8> opc, string OpcodeStr, RegisterClass RC,
             Sched<[itins.Sched.Folded, ReadAfterLd]>,
             Requires<[target, OptForSize]>;
 
-  let isCodeGenOnly = 1, Constraints = "$src1 = $dst" in {
+  let isCodeGenOnly = 1, Constraints = "$src1 = $dst", ExeDomain = d in {
   def r_Int : I<opc, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src1, VR128:$src2),
               !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
             []>, Sched<[itins.Sched.Folded, ReadAfterLd]>;
@@ -3377,7 +3377,7 @@ multiclass avx_fp_unop_s<bits<8> opc, string OpcodeStr, RegisterClass RC,
   def m : I<opc, MRMSrcMem, (outs RC:$dst), (ins RC:$src1, x86memop:$src2),
              !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
             [], itins.rm, d>, Sched<[itins.Sched.Folded, ReadAfterLd]>;
-  let isCodeGenOnly = 1 in {
+  let isCodeGenOnly = 1, ExeDomain = d in {
   def r_Int : I<opc, MRMSrcReg, (outs VR128:$dst),
                 (ins VR128:$src1, VR128:$src2),
              !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
diff --git a/llvm/test/CodeGen/X86/sse2-intrinsics-fast-isel.ll b/llvm/test/CodeGen/X86/sse2-intrinsics-fast-isel.ll
index 972a33f13cd..3071155172e 100644
--- a/llvm/test/CodeGen/X86/sse2-intrinsics-fast-isel.ll
+++ b/llvm/test/CodeGen/X86/sse2-intrinsics-fast-isel.ll
@@ -2934,13 +2934,13 @@ define <2 x double> @test_mm_sqrt_sd(<2 x double> %a0, <2 x double> %a1) nounwin
 ; X32-LABEL: test_mm_sqrt_sd:
 ; X32:       # BB#0:
 ; X32-NEXT:    sqrtsd %xmm0, %xmm1
-; X32-NEXT:    movaps %xmm1, %xmm0
+; X32-NEXT:    movapd %xmm1, %xmm0
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: test_mm_sqrt_sd:
 ; X64:       # BB#0:
 ; X64-NEXT:    sqrtsd %xmm0, %xmm1
-; X64-NEXT:    movaps %xmm1, %xmm0
+; X64-NEXT:    movapd %xmm1, %xmm0
 ; X64-NEXT:    retq
   %call = call <2 x double> @llvm.x86.sse2.sqrt.sd(<2 x double> %a0)
   %ext0 = extractelement <2 x double> %call, i32 0
diff --git a/llvm/test/CodeGen/X86/sse2-intrinsics-x86.ll b/llvm/test/CodeGen/X86/sse2-intrinsics-x86.ll
index 026f8e91969..b0a8744f5d8 100644
--- a/llvm/test/CodeGen/X86/sse2-intrinsics-x86.ll
+++ b/llvm/test/CodeGen/X86/sse2-intrinsics-x86.ll
@@ -1512,21 +1512,21 @@ define <2 x double> @test_x86_sse2_sqrt_sd_vec_load(<2 x double>* %a0) {
 ; SSE-LABEL: test_x86_sse2_sqrt_sd_vec_load:
 ; SSE:       ## BB#0:
 ; SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax ## encoding: [0x8b,0x44,0x24,0x04]
-; SSE-NEXT:    movaps (%eax), %xmm0 ## encoding: [0x0f,0x28,0x00]
+; SSE-NEXT:    movapd (%eax), %xmm0 ## encoding: [0x66,0x0f,0x28,0x00]
 ; SSE-NEXT:    sqrtsd %xmm0, %xmm0 ## encoding: [0xf2,0x0f,0x51,0xc0]
 ; SSE-NEXT:    retl ## encoding: [0xc3]
 ;
 ; AVX2-LABEL: test_x86_sse2_sqrt_sd_vec_load:
 ; AVX2:       ## BB#0:
 ; AVX2-NEXT:    movl {{[0-9]+}}(%esp), %eax ## encoding: [0x8b,0x44,0x24,0x04]
-; AVX2-NEXT:    vmovaps (%eax), %xmm0 ## encoding: [0xc5,0xf8,0x28,0x00]
+; AVX2-NEXT:    vmovapd (%eax), %xmm0 ## encoding: [0xc5,0xf9,0x28,0x00]
 ; AVX2-NEXT:    vsqrtsd %xmm0, %xmm0, %xmm0 ## encoding: [0xc5,0xfb,0x51,0xc0]
 ; AVX2-NEXT:    retl ## encoding: [0xc3]
 ;
 ; SKX-LABEL: test_x86_sse2_sqrt_sd_vec_load:
 ; SKX:       ## BB#0:
 ; SKX-NEXT:    movl {{[0-9]+}}(%esp), %eax ## encoding: [0x8b,0x44,0x24,0x04]
-; SKX-NEXT:    vmovaps (%eax), %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0x00]
+; SKX-NEXT:    vmovapd (%eax), %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf9,0x28,0x00]
 ; SKX-NEXT:    vsqrtsd %xmm0, %xmm0, %xmm0 ## encoding: [0xc5,0xfb,0x51,0xc0]
 ; SKX-NEXT:    retl ## encoding: [0xc3]
   %a1 = load <2 x double>, <2 x double>* %a0, align 16