[x86] add test for SSE sqrtss register dep (PR22206)

1 files changed, 33 insertions, 28 deletions

diff --git a/llvm/test/CodeGen/X86/sse_partial_update.ll b/llvm/test/CodeGen/X86/sse_partial_update.ll
index aa4bc99b763..117f81c29f4 100644
--- a/llvm/test/CodeGen/X86/sse_partial_update.ll
+++ b/llvm/test/CodeGen/X86/sse_partial_update.ll
@@ -10,19 +10,17 @@
 
 define void @rsqrtss(<4 x float> %a) nounwind uwtable ssp {
 ; CHECK-LABEL: rsqrtss:
-; CHECK:       ## %bb.0: ## %entry
+; CHECK:       ## %bb.0:
 ; CHECK-NEXT:    rsqrtss %xmm0, %xmm0
 ; CHECK-NEXT:    cvtss2sd %xmm0, %xmm2
 ; CHECK-NEXT:    movshdup {{.*#+}} xmm0 = xmm0[1,1,3,3]
 ; CHECK-NEXT:    cvtss2sd %xmm0, %xmm1
 ; CHECK-NEXT:    movaps %xmm2, %xmm0
 ; CHECK-NEXT:    jmp _callee ## TAILCALL
-entry:
-
-  %0 = tail call <4 x float> @llvm.x86.sse.rsqrt.ss(<4 x float> %a) nounwind
-  %a.addr.0.extract = extractelement <4 x float> %0, i32 0
+  %t0 = tail call <4 x float> @llvm.x86.sse.rsqrt.ss(<4 x float> %a) nounwind
+  %a.addr.0.extract = extractelement <4 x float> %t0, i32 0
   %conv = fpext float %a.addr.0.extract to double
-  %a.addr.4.extract = extractelement <4 x float> %0, i32 1
+  %a.addr.4.extract = extractelement <4 x float> %t0, i32 1
   %conv3 = fpext float %a.addr.4.extract to double
   tail call void @callee(double %conv, double %conv3) nounwind
   ret void
@@ -32,19 +30,17 @@ declare <4 x float> @llvm.x86.sse.rsqrt.ss(<4 x float>) nounwind readnone
 
 define void @rcpss(<4 x float> %a) nounwind uwtable ssp {
 ; CHECK-LABEL: rcpss:
-; CHECK:       ## %bb.0: ## %entry
+; CHECK:       ## %bb.0:
 ; CHECK-NEXT:    rcpss %xmm0, %xmm0
 ; CHECK-NEXT:    cvtss2sd %xmm0, %xmm2
 ; CHECK-NEXT:    movshdup {{.*#+}} xmm0 = xmm0[1,1,3,3]
 ; CHECK-NEXT:    cvtss2sd %xmm0, %xmm1
 ; CHECK-NEXT:    movaps %xmm2, %xmm0
 ; CHECK-NEXT:    jmp _callee ## TAILCALL
-entry:
-
-  %0 = tail call <4 x float> @llvm.x86.sse.rcp.ss(<4 x float> %a) nounwind
-  %a.addr.0.extract = extractelement <4 x float> %0, i32 0
+  %t0 = tail call <4 x float> @llvm.x86.sse.rcp.ss(<4 x float> %a) nounwind
+  %a.addr.0.extract = extractelement <4 x float> %t0, i32 0
   %conv = fpext float %a.addr.0.extract to double
-  %a.addr.4.extract = extractelement <4 x float> %0, i32 1
+  %a.addr.4.extract = extractelement <4 x float> %t0, i32 1
   %conv3 = fpext float %a.addr.4.extract to double
   tail call void @callee(double %conv, double %conv3) nounwind
   ret void
@@ -53,7 +49,7 @@ declare <4 x float> @llvm.x86.sse.rcp.ss(<4 x float>) nounwind readnone
 
 define void @sqrtss(<4 x float> %a) nounwind uwtable ssp {
 ; CHECK-LABEL: sqrtss:
-; CHECK:       ## %bb.0: ## %entry
+; CHECK:       ## %bb.0:
 ; CHECK-NEXT:    sqrtss %xmm0, %xmm1
 ; CHECK-NEXT:    cvtss2sd %xmm1, %xmm2
 ; CHECK-NEXT:    movshdup {{.*#+}} xmm0 = xmm0[1,1,3,3]
@@ -61,12 +57,10 @@ define void @sqrtss(<4 x float> %a) nounwind uwtable ssp {
 ; CHECK-NEXT:    cvtss2sd %xmm0, %xmm1
 ; CHECK-NEXT:    movaps %xmm2, %xmm0
 ; CHECK-NEXT:    jmp _callee ## TAILCALL
-entry:
-
-  %0 = tail call <4 x float> @llvm.x86.sse.sqrt.ss(<4 x float> %a) nounwind
-  %a.addr.0.extract = extractelement <4 x float> %0, i32 0
+  %t0 = tail call <4 x float> @llvm.x86.sse.sqrt.ss(<4 x float> %a) nounwind
+  %a.addr.0.extract = extractelement <4 x float> %t0, i32 0
   %conv = fpext float %a.addr.0.extract to double
-  %a.addr.4.extract = extractelement <4 x float> %0, i32 1
+  %a.addr.4.extract = extractelement <4 x float> %t0, i32 1
   %conv3 = fpext float %a.addr.4.extract to double
   tail call void @callee(double %conv, double %conv3) nounwind
   ret void
@@ -75,7 +69,7 @@ declare <4 x float> @llvm.x86.sse.sqrt.ss(<4 x float>) nounwind readnone
 
 define void @sqrtsd(<2 x double> %a) nounwind uwtable ssp {
 ; CHECK-LABEL: sqrtsd:
-; CHECK:       ## %bb.0: ## %entry
+; CHECK:       ## %bb.0:
 ; CHECK-NEXT:    sqrtsd %xmm0, %xmm1
 ; CHECK-NEXT:    cvtsd2ss %xmm1, %xmm2
 ; CHECK-NEXT:    unpckhpd {{.*#+}} xmm0 = xmm0[1,1]
@@ -83,15 +77,13 @@ define void @sqrtsd(<2 x double> %a) nounwind uwtable ssp {
 ; CHECK-NEXT:    cvtsd2ss %xmm0, %xmm1
 ; CHECK-NEXT:    movaps %xmm2, %xmm0
 ; CHECK-NEXT:    jmp _callee2 ## TAILCALL
-entry:
-
- %0 = tail call <2 x double> @llvm.x86.sse2.sqrt.sd(<2 x double> %a) nounwind
- %a0 = extractelement <2 x double> %0, i32 0
- %conv = fptrunc double %a0 to float
- %a1 = extractelement <2 x double> %0, i32 1
- %conv3 = fptrunc double %a1 to float
- tail call void @callee2(float %conv, float %conv3) nounwind
- ret void
+  %t0 = tail call <2 x double> @llvm.x86.sse2.sqrt.sd(<2 x double> %a) nounwind
+  %a0 = extractelement <2 x double> %t0, i32 0
+  %conv = fptrunc double %a0 to float
+  %a1 = extractelement <2 x double> %t0, i32 1
+  %conv3 = fptrunc double %a1 to float
+  tail call void @callee2(float %conv, float %conv3) nounwind
+  ret void
 }
 
 declare void @callee2(float, float)
@@ -133,3 +125,16 @@ define <2 x double> @load_fold_cvtss2sd_int_minsize(<4 x float> *%a) minsize {
 
 declare <2 x double> @llvm.x86.sse2.cvtss2sd(<2 x double>, <4 x float>) nounwind readnone
 
+define float @PR22206(<4 x float> %a) {
+; CHECK-LABEL: PR22206:
+; CHECK:       ## %bb.0:
+; CHECK-NEXT:    sqrtss %xmm0, %xmm1
+; CHECK-NEXT:    addss %xmm1, %xmm0
+; CHECK-NEXT:    retq
+  %res = tail call <4 x float> @llvm.x86.sse.sqrt.ss(<4 x float> %a) nounwind
+  %new = extractelement <4 x float> %res, i32 0
+  %orig = extractelement <4 x float> %a, i32 0
+  %add = fadd float %new, %orig
+  ret float %add
+}
+