Implement "punpckldq %xmm0, $xmm0" as "pshufd $0x50, %xmm0, %xmm" unless optimizing for code size.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@56711 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/test/CodeGen/X86/vec_extract.ll b/test/CodeGen/X86/vec_extract.ll
index eddba87..b45f939 100644
--- a/test/CodeGen/X86/vec_extract.ll
+++ b/test/CodeGen/X86/vec_extract.ll
@@ -4,7 +4,7 @@
; RUN: grep pshufd %t | count 1
; RUN: grep unpckhpd %t | count 1
-define void @test1(<4 x float>* %F, float* %f) {
+define void @test1(<4 x float>* %F, float* %f) nounwind {
%tmp = load <4 x float>* %F ; <<4 x float>> [#uses=2]
%tmp7 = add <4 x float> %tmp, %tmp ; <<4 x float>> [#uses=1]
%tmp2 = extractelement <4 x float> %tmp7, i32 0 ; <float> [#uses=1]
@@ -12,21 +12,21 @@
ret void
}
-define float @test2(<4 x float>* %F, float* %f) {
+define float @test2(<4 x float>* %F, float* %f) nounwind {
%tmp = load <4 x float>* %F ; <<4 x float>> [#uses=2]
%tmp7 = add <4 x float> %tmp, %tmp ; <<4 x float>> [#uses=1]
%tmp2 = extractelement <4 x float> %tmp7, i32 2 ; <float> [#uses=1]
ret float %tmp2
}
-define void @test3(float* %R, <4 x float>* %P1) {
+define void @test3(float* %R, <4 x float>* %P1) nounwind {
%X = load <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
}
-define double @test4(double %A) {
+define double @test4(double %A) nounwind {
%tmp1 = call <2 x double> @foo( ) ; <<2 x double>> [#uses=1]
%tmp2 = extractelement <2 x double> %tmp1, i32 1 ; <double> [#uses=1]
%tmp3 = add double %tmp2, %A ; <double> [#uses=1]
diff --git a/test/CodeGen/X86/vec_shuffle-23.ll b/test/CodeGen/X86/vec_shuffle-23.ll
new file mode 100644
index 0000000..34d84ef
--- /dev/null
+++ b/test/CodeGen/X86/vec_shuffle-23.ll
@@ -0,0 +1,19 @@
+; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse2 | not grep punpck
+; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse2 | grep pshufd
+; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse2 -optimize-size | grep punpck
+
+define i32 @t() nounwind {
+entry:
+ %a = alloca <4 x i32> ; <<4 x i32>*> [#uses=2]
+ %b = alloca <4 x i32> ; <<4 x i32>*> [#uses=5]
+ volatile store <4 x i32> < i32 0, i32 1, i32 2, i32 3 >, <4 x i32>* %a
+ %tmp = load <4 x i32>* %a ; <<4 x i32>> [#uses=1]
+ store <4 x i32> %tmp, <4 x i32>* %b
+ %tmp1 = load <4 x i32>* %b ; <<4 x i32>> [#uses=1]
+ %tmp2 = load <4 x i32>* %b ; <<4 x i32>> [#uses=1]
+ %punpckldq = shufflevector <4 x i32> %tmp1, <4 x i32> %tmp2, <4 x i32> < i32 0, i32 4, i32 1, i32 5 > ; <<4 x i32>> [#uses=1]
+ store <4 x i32> %punpckldq, <4 x i32>* %b
+ %tmp3 = load <4 x i32>* %b ; <<4 x i32>> [#uses=1]
+ %result = extractelement <4 x i32> %tmp3, i32 0 ; <i32> [#uses=1]
+ ret i32 %result
+}