2 files changed, 56 insertions, 1 deletions

diff --git a/llvm/lib/Transforms/IPO/MergeFunctions.cpp b/llvm/lib/Transforms/IPO/MergeFunctions.cpp
index a1764ca33c1..6e907b0f120 100644
--- a/llvm/lib/Transforms/IPO/MergeFunctions.cpp
+++ b/llvm/lib/Transforms/IPO/MergeFunctions.cpp
@@ -784,8 +784,23 @@ void MergeFunctions::writeThunkOrAlias(Function *F, Function *G) {
 // Helper for writeThunk,
 // Selects proper bitcast operation,
 // but a bit simpler then CastInst::getCastOpcode.
-static Value* createCast(IRBuilder<false> &Builder, Value *V, Type *DestTy) {
+static Value *createCast(IRBuilder<false> &Builder, Value *V, Type *DestTy) {
   Type *SrcTy = V->getType();
+  if (SrcTy->isStructTy()) {
+    assert(DestTy->isStructTy());
+    assert(SrcTy->getStructNumElements() == DestTy->getStructNumElements());
+    Value *Result = UndefValue::get(DestTy);
+    for (unsigned int I = 0, E = SrcTy->getStructNumElements(); I < E; ++I) {
+      Value *Element = createCast(
+          Builder, Builder.CreateExtractValue(V, ArrayRef<unsigned int>(I)),
+          DestTy->getStructElementType(I));
+
+      Result =
+          Builder.CreateInsertValue(Result, Element, ArrayRef<unsigned int>(I));
+    }
+    return Result;
+  }
+  assert(!DestTy->isStructTy());
   if (SrcTy->isIntegerTy() && DestTy->isPointerTy())
     return Builder.CreateIntToPtr(V, DestTy);
   else if (SrcTy->isPointerTy() && DestTy->isIntegerTy())
diff --git a/llvm/test/Transforms/MergeFunc/mergefunc-struct-return.ll b/llvm/test/Transforms/MergeFunc/mergefunc-struct-return.ll
new file mode 100644
index 00000000000..d2cbe43da0c
--- /dev/null
+++ b/llvm/test/Transforms/MergeFunc/mergefunc-struct-return.ll
@@ -0,0 +1,40 @@
+; RUN: opt -mergefunc -S < %s | FileCheck %s
+
+; This test makes sure that the mergefunc pass, uses extract and insert value
+; to convert the struct result type; as struct types cannot be bitcast.
+
+target datalayout = "e-m:w-p:32:32-i64:64-f80:32-n8:16:32-S32"
+
+%kv1 = type { i32*, i32* }
+%kv2 = type { i8*, i8* }
+
+declare void @noop()
+
+define %kv1 @fn1() {
+; CHECK-LABEL: @fn1(
+  %tmp = alloca %kv1
+  %v1 = getelementptr %kv1* %tmp, i32 0, i32 0
+  store i32* null, i32** %v1
+  %v2 = getelementptr %kv1* %tmp, i32 0, i32 0
+  store i32* null, i32** %v2
+  call void @noop()
+  %v3 = load %kv1* %tmp
+  ret %kv1 %v3
+}
+
+define %kv2 @fn2() {
+; CHECK-LABEL: @fn2(
+; CHECK: %1 = tail call %kv1 @fn1()
+; CHECK: %2 = extractvalue %kv1 %1, 0
+; CHECK: %3 = bitcast i32* %2 to i8*
+; CHECK: %4 = insertvalue %kv2 undef, i8* %3, 0
+  %tmp = alloca %kv2
+  %v1 = getelementptr %kv2* %tmp, i32 0, i32 0
+  store i8* null, i8** %v1
+  %v2 = getelementptr %kv2* %tmp, i32 0, i32 0
+  store i8* null, i8** %v2
+  call void @noop()
+
+  %v3 = load %kv2* %tmp
+  ret %kv2 %v3
+}