4 files changed, 38 insertions, 18 deletions
diff --git a/llvm/lib/Target/X86/X86InstrSSE.td b/llvm/lib/Target/X86/X86InstrSSE.td
index 451303054f5..d4676b57455 100644
--- a/llvm/lib/Target/X86/X86InstrSSE.td
+++ b/llvm/lib/Target/X86/X86InstrSSE.td
@@ -3186,7 +3186,7 @@ let Predicates = prds in {
 /// sse2_fp_unop_p - SSE2 unops in vector forms.
 multiclass sse2_fp_unop_p<bits<8> opc, string OpcodeStr,
                           SDNode OpNode, OpndItins itins> {
-let Predicates = [HasAVX] in {
+let Predicates = [HasAVX, NoVLX] in {
   def V#NAME#PDr : PDI<opc, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
                        !strconcat("v", OpcodeStr,
                                   "pd\t{$src, $dst|$dst, $src}"),
@@ -3245,7 +3245,7 @@ multiclass sse2_fp_unop_s<bits<8> opc, string OpcodeStr, SDNode OpNode,
 
 // Square root.
 defm SQRT  : sse1_fp_unop_s<0x51, "sqrt", fsqrt, SSE_SQRTSS>,
-             sse1_fp_unop_p<0x51, "sqrt", fsqrt, SSE_SQRTPS, [HasAVX]>,
+             sse1_fp_unop_p<0x51, "sqrt", fsqrt, SSE_SQRTPS, [HasAVX, NoVLX]>,
              sse2_fp_unop_s<0x51, "sqrt", fsqrt, SSE_SQRTSD>,
              sse2_fp_unop_p<0x51, "sqrt", fsqrt, SSE_SQRTPD>;
 
diff --git a/llvm/test/CodeGen/X86/avx-intrinsics-x86.ll b/llvm/test/CodeGen/X86/avx-intrinsics-x86.ll
index 44eb14160ee..b0cf4e3b29f 100644
--- a/llvm/test/CodeGen/X86/avx-intrinsics-x86.ll
+++ b/llvm/test/CodeGen/X86/avx-intrinsics-x86.ll
@@ -635,10 +635,15 @@ declare <8 x float> @llvm.x86.avx.rsqrt.ps.256(<8 x float>) nounwind readnone
 
 
 define <4 x double> @test_x86_avx_sqrt_pd_256(<4 x double> %a0) {
-; CHECK-LABEL: test_x86_avx_sqrt_pd_256:
-; CHECK:       # BB#0:
-; CHECK-NEXT:    vsqrtpd %ymm0, %ymm0 # encoding: [0xc5,0xfd,0x51,0xc0]
-; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
+; AVX-LABEL: test_x86_avx_sqrt_pd_256:
+; AVX:       # BB#0:
+; AVX-NEXT:    vsqrtpd %ymm0, %ymm0 # encoding: [0xc5,0xfd,0x51,0xc0]
+; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
+;
+; AVX512VL-LABEL: test_x86_avx_sqrt_pd_256:
+; AVX512VL:       # BB#0:
+; AVX512VL-NEXT:    vsqrtpd %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x51,0xc0]
+; AVX512VL-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
   %res = call <4 x double> @llvm.x86.avx.sqrt.pd.256(<4 x double> %a0) ; <<4 x double>> [#uses=1]
   ret <4 x double> %res
 }
@@ -646,10 +651,15 @@ declare <4 x double> @llvm.x86.avx.sqrt.pd.256(<4 x double>) nounwind readnone
 
 
 define <8 x float> @test_x86_avx_sqrt_ps_256(<8 x float> %a0) {
-; CHECK-LABEL: test_x86_avx_sqrt_ps_256:
-; CHECK:       # BB#0:
-; CHECK-NEXT:    vsqrtps %ymm0, %ymm0 # encoding: [0xc5,0xfc,0x51,0xc0]
-; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
+; AVX-LABEL: test_x86_avx_sqrt_ps_256:
+; AVX:       # BB#0:
+; AVX-NEXT:    vsqrtps %ymm0, %ymm0 # encoding: [0xc5,0xfc,0x51,0xc0]
+; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
+;
+; AVX512VL-LABEL: test_x86_avx_sqrt_ps_256:
+; AVX512VL:       # BB#0:
+; AVX512VL-NEXT:    vsqrtps %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x51,0xc0]
+; AVX512VL-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
   %res = call <8 x float> @llvm.x86.avx.sqrt.ps.256(<8 x float> %a0) ; <<8 x float>> [#uses=1]
   ret <8 x float> %res
 }
diff --git a/llvm/test/CodeGen/X86/sse-intrinsics-x86.ll b/llvm/test/CodeGen/X86/sse-intrinsics-x86.ll
index f178e18a259..5ba9f9a2645 100644
--- a/llvm/test/CodeGen/X86/sse-intrinsics-x86.ll
+++ b/llvm/test/CodeGen/X86/sse-intrinsics-x86.ll
@@ -475,10 +475,15 @@ define <4 x float> @test_x86_sse_sqrt_ps(<4 x float> %a0) {
 ; SSE-NEXT:    sqrtps %xmm0, %xmm0 ## encoding: [0x0f,0x51,0xc0]
 ; SSE-NEXT:    retl ## encoding: [0xc3]
 ;
-; VCHECK-LABEL: test_x86_sse_sqrt_ps:
-; VCHECK:       ## BB#0:
-; VCHECK-NEXT:    vsqrtps %xmm0, %xmm0 ## encoding: [0xc5,0xf8,0x51,0xc0]
-; VCHECK-NEXT:    retl ## encoding: [0xc3]
+; AVX2-LABEL: test_x86_sse_sqrt_ps:
+; AVX2:       ## BB#0:
+; AVX2-NEXT:    vsqrtps %xmm0, %xmm0 ## encoding: [0xc5,0xf8,0x51,0xc0]
+; AVX2-NEXT:    retl ## encoding: [0xc3]
+;
+; SKX-LABEL: test_x86_sse_sqrt_ps:
+; SKX:       ## BB#0:
+; SKX-NEXT:    vsqrtps %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf8,0x51,0xc0]
+; SKX-NEXT:    retl ## encoding: [0xc3]
   %res = call <4 x float> @llvm.x86.sse.sqrt.ps(<4 x float> %a0) ; <<4 x float>> [#uses=1]
   ret <4 x float> %res
 }
diff --git a/llvm/test/CodeGen/X86/sse2-intrinsics-x86.ll b/llvm/test/CodeGen/X86/sse2-intrinsics-x86.ll
index d4047faad9b..dcccdbfc2e6 100644
--- a/llvm/test/CodeGen/X86/sse2-intrinsics-x86.ll
+++ b/llvm/test/CodeGen/X86/sse2-intrinsics-x86.ll
@@ -1592,10 +1592,15 @@ define <2 x double> @test_x86_sse2_sqrt_pd(<2 x double> %a0) {
 ; SSE-NEXT:    sqrtpd %xmm0, %xmm0 ## encoding: [0x66,0x0f,0x51,0xc0]
 ; SSE-NEXT:    retl ## encoding: [0xc3]
 ;
-; VCHECK-LABEL: test_x86_sse2_sqrt_pd:
-; VCHECK:       ## BB#0:
-; VCHECK-NEXT:    vsqrtpd %xmm0, %xmm0 ## encoding: [0xc5,0xf9,0x51,0xc0]
-; VCHECK-NEXT:    retl ## encoding: [0xc3]
+; AVX2-LABEL: test_x86_sse2_sqrt_pd:
+; AVX2:       ## BB#0:
+; AVX2-NEXT:    vsqrtpd %xmm0, %xmm0 ## encoding: [0xc5,0xf9,0x51,0xc0]
+; AVX2-NEXT:    retl ## encoding: [0xc3]
+;
+; SKX-LABEL: test_x86_sse2_sqrt_pd:
+; SKX:       ## BB#0:
+; SKX-NEXT:    vsqrtpd %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf9,0x51,0xc0]
+; SKX-NEXT:    retl ## encoding: [0xc3]
   %res = call <2 x double> @llvm.x86.sse2.sqrt.pd(<2 x double> %a0) ; <<2 x double>> [#uses=1]
   ret <2 x double> %res
 }