Improve the loading of load-anyext vectors by allowing the codegen to load
multiple scalars and insert them into a vector. Next, we shuffle the elements
into the correct places, as before.
Also fix a small dagcombine bug in SimplifyBinOpWithSameOpcodeHands, when the
migration of bitcasts happened too late in the SelectionDAG process.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@159991 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/test/CodeGen/X86/2012-07-10-extload64.ll b/test/CodeGen/X86/2012-07-10-extload64.ll
new file mode 100644
index 0000000..3284f5e
--- /dev/null
+++ b/test/CodeGen/X86/2012-07-10-extload64.ll
@@ -0,0 +1,13 @@
+; RUN: llc < %s -march=x86 -mcpu=corei7 -mtriple=i686-pc-win32 | FileCheck %s
+
+; CHECK: load_store
+define void @load_store(<4 x i16>* %in) {
+entry:
+ %A27 = load <4 x i16>* %in, align 4
+ %A28 = add <4 x i16> %A27, %A27
+ store <4 x i16> %A28, <4 x i16>* %in, align 4
+ ret void
+; CHECK: movd
+; CHECK: pinsrd
+; CHECK: ret
+}