CodeGen peephole: fold redundant phys reg copies

[oota-llvm.git] / test / CodeGen / X86 / vec_extract.ll

diff --git a/test/CodeGen/X86/vec_extract.ll b/test/CodeGen/X86/vec_extract.ll
index 2e369828c5b4fd3473cd8f3b587019631dea1acc..3b478880590d21db913da58b8f1eaf5d9e50e100 100644 (file)
--- a/test/CodeGen/X86/vec_extract.ll
+++ b/test/CodeGen/X86/vec_extract.ll
@@ -1,12 +1,18 @@
 ; RUN: llc < %s -mcpu=corei7 -march=x86 -mattr=+sse2,-sse4.1 | FileCheck %s
 
+target triple = "x86_64-unknown-linux-gnu"
+
 define void @test1(<4 x float>* %F, float* %f) nounwind {
 ; CHECK-LABEL: test1:
-; CHECK:         addps %[[X:xmm[0-9]+]], %[[X]]
-; CHECK-NEXT:    movss %[[X]], {{.*}}(%{{.*}})
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %ecx
+; CHECK-NEXT:    movaps (%ecx), %xmm0
+; CHECK-NEXT:    addps %xmm0, %xmm0
+; CHECK-NEXT:    movss %xmm0, (%eax)
 ; CHECK-NEXT:    retl
 entry:
-       %tmp = load <4 x float>* %F             ; <<4 x float>> [#uses=2]
+       %tmp = load <4 x float>, <4 x float>* %F                ; <<4 x float>> [#uses=2]
        %tmp7 = fadd <4 x float> %tmp, %tmp             ; <<4 x float>> [#uses=1]
        %tmp2 = extractelement <4 x float> %tmp7, i32 0         ; <float> [#uses=1]
        store float %tmp2, float* %f
@@ -15,12 +21,18 @@ entry:
 
 define float @test2(<4 x float>* %F, float* %f) nounwind {
 ; CHECK-LABEL: test2:
-; CHECK:         addps %[[X:xmm[0-9]+]], %[[X]]
-; CHECK-NEXT:    movhlps %[[X]], %[[X2:xmm[0-9]+]]
-; CHECK-NEXT:    movss %[[X2]], [[mem:.*\(%.*\)]]
-; CHECK-NEXT:    flds [[mem]]
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    pushl %eax
+; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; CHECK-NEXT:    movaps (%eax), %xmm0
+; CHECK-NEXT:    addps %xmm0, %xmm0
+; CHECK-NEXT:    shufpd {{.*#+}} xmm0 = xmm0[1,0]
+; CHECK-NEXT:    movss %xmm0, (%esp)
+; CHECK-NEXT:    flds (%esp)
+; CHECK-NEXT:    popl %eax
+; CHECK-NEXT:    retl
 entry:
-       %tmp = load <4 x float>* %F             ; <<4 x float>> [#uses=2]
+       %tmp = load <4 x float>, <4 x float>* %F                ; <<4 x float>> [#uses=2]
        %tmp7 = fadd <4 x float> %tmp, %tmp             ; <<4 x float>> [#uses=1]
        %tmp2 = extractelement <4 x float> %tmp7, i32 2         ; <float> [#uses=1]
        ret float %tmp2
@@ -28,11 +40,14 @@ entry:
 
 define void @test3(float* %R, <4 x float>* %P1) nounwind {
 ; CHECK-LABEL: test3:
-; CHECK:         movss {{.*}}(%{{.*}}), %[[X:xmm[0-9]+]]
-; CHECK-NEXT:    movss %[[X]], {{.*}}(%{{.*}})
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %ecx
+; CHECK-NEXT:    movss 12(%ecx), %xmm0
+; CHECK-NEXT:    movss %xmm0, (%eax)
 ; CHECK-NEXT:    retl
 entry:
-       %X = load <4 x float>* %P1              ; <<4 x float>> [#uses=1]
+       %X = load <4 x float>, <4 x float>* %P1         ; <<4 x float>> [#uses=1]
        %tmp = extractelement <4 x float> %X, i32 3             ; <float> [#uses=1]
        store float %tmp, float* %R
        ret void
@@ -40,11 +55,15 @@ entry:
 
 define double @test4(double %A) nounwind {
 ; CHECK-LABEL: test4:
-; CHECK:         calll foo
-; CHECK-NEXT:    unpckhpd %[[X:xmm[0-9]+]], %[[X]]
-; CHECK-NEXT:    addsd {{.*}}(%{{.*}}), %[[X2]]
-; CHECK-NEXT:    movsd %[[X2]], [[mem:.*\(%.*\)]]
-; CHECK-NEXT:    fldl [[mem]]
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    subl $12, %esp
+; CHECK-NEXT:    calll foo
+; CHECK-NEXT:    shufpd {{.*#+}} xmm0 = xmm0[1,0]
+; CHECK-NEXT:    addsd {{[0-9]+}}(%esp), %xmm0
+; CHECK-NEXT:    movsd %xmm0, (%esp)
+; CHECK-NEXT:    fldl (%esp)
+; CHECK-NEXT:    addl $12, %esp
+; CHECK-NEXT:    retl
 entry:
        %tmp1 = call <2 x double> @foo( )               ; <<2 x double>> [#uses=1]
        %tmp2 = extractelement <2 x double> %tmp1, i32 1                ; <double> [#uses=1]