[ms-cxxabi] Use inalloca on win32 when passing non-trivial C++ objects

When a non-trivial parameter is present, clang now gathers up all the
parameters that lack inreg and puts them into a packed struct.  MSVC
always aligns each parameter to 4 bytes and no more, so this is a pretty
simple struct to lay out.

On win64, non-trivial records are passed indirectly.  Prior to this
change, clang was incorrectly using byval on win64.

I'm able to self-host a working clang with this change and additional
LLVM patches.

Reviewers: rsmith

Differential Revision: http://llvm-reviews.chandlerc.com/D2636

llvm-svn: 200597

clang/lib/CodeGen/CodeGenFunction.cpp

@@ -276,6 +276,14 @@ void CodeGenFunction::FinishFunction(SourceLocation EndLoc) {
 
   if (CGM.getCodeGenOpts().EmitDeclMetadata)
     EmitDeclMetadata();
+
+  for (SmallVectorImpl<std::pair<llvm::Instruction *, llvm::Value *> >::iterator
+           I = DeferredReplacements.begin(),
+           E = DeferredReplacements.end();
+       I != E; ++I) {
+    I->first->replaceAllUsesWith(I->second);
+    I->first->eraseFromParent();
+  }
 }
 
 /// ShouldInstrumentFunction - Return true if the current function should be
@@ -592,6 +600,14 @@ void CodeGenFunction::StartFunction(GlobalDecl GD,
     // Indirect aggregate return; emit returned value directly into sret slot.
     // This reduces code size, and affects correctness in C++.
     ReturnValue = CurFn->arg_begin();
+  } else if (CurFnInfo->getReturnInfo().getKind() == ABIArgInfo::InAlloca &&
+             !hasScalarEvaluationKind(CurFnInfo->getReturnType())) {
+    // Load the sret pointer from the argument struct and return into that.
+    unsigned Idx = CurFnInfo->getReturnInfo().getInAllocaFieldIndex();
+    llvm::Function::arg_iterator EI = CurFn->arg_end();
+    --EI;
+    llvm::Value *Addr = Builder.CreateStructGEP(EI, Idx);
+    ReturnValue = Builder.CreateLoad(Addr, "agg.result");
   } else {
     ReturnValue = CreateIRTemp(RetTy, "retval");