Fixed a bug in lowering shuffle_vectors to insertps

Summary:
We were being too strict and not accounting for undefs.
Added a test case and fixed another one where we improved codegen.

Reviewers: grosbach, nadav, delena

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D4039

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@210361 91177308-0d34-0410-b5e6-96231b3b80d8

2 files changed, 15 insertions, 2 deletions

diff --git a/test/CodeGen/X86/avx-shuffle.ll b/test/CodeGen/X86/avx-shuffle.ll
index f407ba4cc1..f3f7e554a3 100644
--- a/test/CodeGen/X86/avx-shuffle.ll
+++ b/test/CodeGen/X86/avx-shuffle.ll
@@ -5,8 +5,10 @@ define <4 x float> @test1(<4 x float> %a) nounwind {
   %b = shufflevector <4 x float> zeroinitializer, <4 x float> %a, <4 x i32> <i32 2, i32 5, i32 undef, i32 undef>
   ret <4 x float> %b
 ; CHECK-LABEL: test1:
-; CHECK: vshufps
-; CHECK: vpshufd
+;; TODO: This test could be improved by removing the xor instruction and
+;; having vinsertps zero out the needed elements.
+; CHECK: vxorps
+; CHECK: vinsertps
 }
 
 ; rdar://10538417
diff --git a/test/CodeGen/X86/sse41.ll b/test/CodeGen/X86/sse41.ll
index a3c62016c4..a77ede228d 100644
--- a/test/CodeGen/X86/sse41.ll
+++ b/test/CodeGen/X86/sse41.ll
@@ -692,3 +692,14 @@ define <4 x float> @insertps_from_broadcast_multiple_use(<4 x float> %a, <4 x fl
   %13 = fadd <4 x float> %11, %12
   ret <4 x float> %13
 }
+
+define <4 x float> @insertps_with_undefs(<4 x float> %a, float* %b) {
+; CHECK-LABEL: insertps_with_undefs:
+; CHECK-NOT: shufps
+; CHECK: insertps    $32, %xmm0
+; CHECK: ret
+  %1 = load float* %b, align 4
+  %2 = insertelement <4 x float> undef, float %1, i32 0
+  %result = shufflevector <4 x float> %a, <4 x float> %2, <4 x i32> <i32 4, i32 undef, i32 0, i32 7>
+  ret <4 x float> %result
+}