MFC 261991:

Upgrade our copy of llvm/clang to 3.4 release.  This version supports
all of the features in the current working draft of the upcoming C++
standard, provisionally named C++1y.

The code generator's performance is greatly increased, and the loop
auto-vectorizer is now enabled at -Os and -O2 in addition to -O3.  The
PowerPC backend has made several major improvements to code generation
quality and compile time, and the X86, SPARC, ARM32, Aarch64 and SystemZ
backends have all seen major feature work.

Release notes for llvm and clang can be found here:
<http://llvm.org/releases/3.4/docs/ReleaseNotes.html>
<http://llvm.org/releases/3.4/tools/clang/docs/ReleaseNotes.html>

MFC 262121 (by emaste):

Update lldb for clang/llvm 3.4 import

This commit largely restores the lldb source to the upstream r196259
snapshot with the addition of threaded inferior support and a few bug
fixes.

Specific upstream lldb revisions restored include:
   SVN      git
  181387  779e6ac
  181703  7bef4e2
  182099  b31044e
  182650  f2dcf35
  182683  0d91b80
  183862  15c1774
  183929  99447a6
  184177  0b2934b
  184948  4dc3761
  184954  007e7bc
  186990  eebd175

Sponsored by:	DARPA, AFRL

MFC 262186 (by emaste):

Fix mismerge in r262121

A break statement was lost in the merge.  The error had no functional
impact, but restore it to reduce the diff against upstream.

MFC 262303:

Pull in r197521 from upstream clang trunk (by rdivacky):

  Use the integrated assembler by default on FreeBSD/ppc and ppc64.

Requested by:	jhibbits

MFC 262611:

Pull in r196874 from upstream llvm trunk:

  Fix a crash that occurs when PWD is invalid.

  MCJIT needs to be able to run in hostile environments, even when PWD
  is invalid. There's no need to crash MCJIT in this case.

  The obvious fix is to simply leave MCContext's CompilationDir empty
  when PWD can't be determined. This way, MCJIT clients,
  and other clients that link with LLVM don't need a valid working directory.

  If we do want to guarantee valid CompilationDir, that should be done
  only for clients of getCompilationDir(). This is as simple as checking
  for an empty string.

  The only current use of getCompilationDir is EmitGenDwarfInfo, which
  won't conceivably run with an invalid working dir. However, in the
  purely hypothetically and untestable case that this happens, the
  AT_comp_dir will be omitted from the compilation_unit DIE.

This should help fix assertions occurring with ports-mgmt/tinderbox,
when it is using jails, and sometimes invalidates clang's current
working directory.

Reported by:	decke

MFC 262809:

Pull in r203007 from upstream clang trunk:

  Don't produce an alias between destructors with different calling conventions.

  Fixes pr19007.

(Please note that is an LLVM PR identifier, not a FreeBSD one.)

This should fix Firefox and/or libxul crashes (due to problems with
regparm/stdcall calling conventions) on i386.

Reported by:	multiple users on freebsd-current
PR:		bin/187103

MFC 263048:

Repair recognition of "CC" as an alias for the C++ compiler, since it
was silently broken by upstream for a Windows-specific use-case.

Apparently some versions of CMake still rely on this archaic feature...

Reported by:	rakuco

MFC 263049:

Garbage collect the old way of adding the libstdc++ include directories
in clang's InitHeaderSearch.cpp.  This has been superseded by David
Chisnall's commit in r255321.

Moreover, if libc++ is used, the libstdc++ include directories should
not be in the search path at all.  These directories are now only used
if you pass -stdlib=libstdc++.

1 files changed, 149 insertions, 86 deletions

diff --git a/contrib/llvm/tools/clang/lib/CodeGen/CGCall.cpp b/contrib/llvm/tools/clang/lib/CodeGen/CGCall.cpp
index cdb2dd7..22f2467 100644
--- a/contrib/llvm/tools/clang/lib/CodeGen/CGCall.cpp
+++ b/contrib/llvm/tools/clang/lib/CodeGen/CGCall.cpp
@@ -1,4 +1,4 @@
-//===--- CGCall.cpp - Encapsulate calling convention details ----*- C++ -*-===//
+//===--- CGCall.cpp - Encapsulate calling convention details --------------===//
 //
 //                     The LLVM Compiler Infrastructure
 //
@@ -22,6 +22,7 @@
 #include "clang/AST/DeclCXX.h"
 #include "clang/AST/DeclObjC.h"
 #include "clang/Basic/TargetInfo.h"
+#include "clang/CodeGen/CGFunctionInfo.h"
 #include "clang/Frontend/CodeGenOptions.h"
 #include "llvm/ADT/StringExtras.h"
 #include "llvm/IR/Attributes.h"
@@ -105,24 +106,12 @@ static const CGFunctionInfo &arrangeFreeFunctionType(CodeGenTypes &CGT,
   return arrangeLLVMFunctionInfo(CGT, prefix, FTP, FTP->getExtInfo());
 }
 
-/// Given the formal ext-info of a C++ instance method, adjust it
-/// according to the C++ ABI in effect.
-static void adjustCXXMethodInfo(CodeGenTypes &CGT,
-                                FunctionType::ExtInfo &extInfo,
-                                bool isVariadic) {
-  if (extInfo.getCC() == CC_Default) {
-    CallingConv CC = CGT.getContext().getDefaultCXXMethodCallConv(isVariadic);
-    extInfo = extInfo.withCallingConv(CC);
-  }
-}
-
 /// Arrange the argument and result information for a free function (i.e.
 /// not a C++ or ObjC instance method) of the given type.
 static const CGFunctionInfo &arrangeCXXMethodType(CodeGenTypes &CGT,
                                       SmallVectorImpl<CanQualType> &prefix,
                                             CanQual<FunctionProtoType> FTP) {
   FunctionType::ExtInfo extInfo = FTP->getExtInfo();
-  adjustCXXMethodInfo(CGT, extInfo, FTP->isVariadic());
   return arrangeLLVMFunctionInfo(CGT, prefix, FTP, extInfo);
 }
 
@@ -162,6 +151,8 @@ static CallingConv getCallingConventionForDecl(const Decl *D) {
 
 /// Arrange the argument and result information for a call to an
 /// unknown C++ non-static member function of the given abstract type.
+/// (Zero value of RD means we don't have any meaningful "this" argument type,
+///  so fall back to a generic pointer type).
 /// The member function must be an ordinary function, i.e. not a
 /// constructor or destructor.
 const CGFunctionInfo &
@@ -170,7 +161,10 @@ CodeGenTypes::arrangeCXXMethodType(const CXXRecordDecl *RD,
   SmallVector<CanQualType, 16> argTypes;
 
   // Add the 'this' pointer.
-  argTypes.push_back(GetThisType(Context, RD));
+  if (RD)
+    argTypes.push_back(GetThisType(Context, RD));
+  else
+    argTypes.push_back(Context.VoidPtrTy);
 
   return ::arrangeCXXMethodType(*this, argTypes,
               FTP->getCanonicalTypeUnqualified().getAs<FunctionProtoType>());
@@ -182,14 +176,15 @@ CodeGenTypes::arrangeCXXMethodType(const CXXRecordDecl *RD,
 /// constructor or destructor.
 const CGFunctionInfo &
 CodeGenTypes::arrangeCXXMethodDeclaration(const CXXMethodDecl *MD) {
-  assert(!isa<CXXConstructorDecl>(MD) && "wrong method for contructors!");
+  assert(!isa<CXXConstructorDecl>(MD) && "wrong method for constructors!");
   assert(!isa<CXXDestructorDecl>(MD) && "wrong method for destructors!");
 
   CanQual<FunctionProtoType> prototype = GetFormalType(MD);
 
   if (MD->isInstance()) {
     // The abstract case is perfectly fine.
-    return arrangeCXXMethodType(MD->getParent(), prototype.getTypePtr());
+    const CXXRecordDecl *ThisType = TheCXXABI.getThisArgumentTypeForMethod(MD);
+    return arrangeCXXMethodType(ThisType, prototype.getTypePtr());
   }
 
   return arrangeFreeFunctionType(prototype);
@@ -202,7 +197,10 @@ CodeGenTypes::arrangeCXXConstructorDeclaration(const CXXConstructorDecl *D,
                                                CXXCtorType ctorKind) {
   SmallVector<CanQualType, 16> argTypes;
   argTypes.push_back(GetThisType(Context, D->getParent()));
-  CanQualType resultType = Context.VoidTy;
+
+  GlobalDecl GD(D, ctorKind);
+  CanQualType resultType =
+    TheCXXABI.HasThisReturn(GD) ? argTypes.front() : Context.VoidTy;
 
   TheCXXABI.BuildConstructorSignature(D, ctorKind, resultType, argTypes);
 
@@ -215,7 +213,6 @@ CodeGenTypes::arrangeCXXConstructorDeclaration(const CXXConstructorDecl *D,
     argTypes.push_back(FTP->getArgType(i));
 
   FunctionType::ExtInfo extInfo = FTP->getExtInfo();
-  adjustCXXMethodInfo(*this, extInfo, FTP->isVariadic());
   return arrangeLLVMFunctionInfo(resultType, argTypes, extInfo, required);
 }
 
@@ -227,7 +224,10 @@ CodeGenTypes::arrangeCXXDestructor(const CXXDestructorDecl *D,
                                    CXXDtorType dtorKind) {
   SmallVector<CanQualType, 2> argTypes;
   argTypes.push_back(GetThisType(Context, D->getParent()));
-  CanQualType resultType = Context.VoidTy;
+
+  GlobalDecl GD(D, dtorKind);
+  CanQualType resultType =
+    TheCXXABI.HasThisReturn(GD) ? argTypes.front() : Context.VoidTy;
 
   TheCXXABI.BuildDestructorSignature(D, dtorKind, resultType, argTypes);
 
@@ -236,7 +236,6 @@ CodeGenTypes::arrangeCXXDestructor(const CXXDestructorDecl *D,
   assert(FTP->isVariadic() == 0 && "dtor with formal parameters");
 
   FunctionType::ExtInfo extInfo = FTP->getExtInfo();
-  adjustCXXMethodInfo(*this, extInfo, false);
   return arrangeLLVMFunctionInfo(resultType, argTypes, extInfo,
                                  RequiredArgs::All);
 }
@@ -324,6 +323,7 @@ CodeGenTypes::arrangeGlobalDeclaration(GlobalDecl GD) {
 /// additional number of formal parameters considered required.
 static const CGFunctionInfo &
 arrangeFreeFunctionLikeCall(CodeGenTypes &CGT,
+                            CodeGenModule &CGM,
                             const CallArgList &args,
                             const FunctionType *fnType,
                             unsigned numExtraRequiredArgs) {
@@ -342,8 +342,9 @@ arrangeFreeFunctionLikeCall(CodeGenTypes &CGT,
   // explicitly use the variadic convention for unprototyped calls,
   // treat all of the arguments as required but preserve the nominal
   // possibility of variadics.
-  } else if (CGT.CGM.getTargetCodeGenInfo()
-               .isNoProtoCallVariadic(args, cast<FunctionNoProtoType>(fnType))) {
+  } else if (CGM.getTargetCodeGenInfo()
+                .isNoProtoCallVariadic(args,
+                                       cast<FunctionNoProtoType>(fnType))) {
     required = RequiredArgs(args.size());
   }
 
@@ -358,7 +359,7 @@ arrangeFreeFunctionLikeCall(CodeGenTypes &CGT,
 const CGFunctionInfo &
 CodeGenTypes::arrangeFreeFunctionCall(const CallArgList &args,
                                       const FunctionType *fnType) {
-  return arrangeFreeFunctionLikeCall(*this, args, fnType, 0);
+  return arrangeFreeFunctionLikeCall(*this, CGM, args, fnType, 0);
 }
 
 /// A block function call is essentially a free-function call with an
@@ -366,7 +367,7 @@ CodeGenTypes::arrangeFreeFunctionCall(const CallArgList &args,
 const CGFunctionInfo &
 CodeGenTypes::arrangeBlockFunctionCall(const CallArgList &args,
                                        const FunctionType *fnType) {
-  return arrangeFreeFunctionLikeCall(*this, args, fnType, 1);
+  return arrangeFreeFunctionLikeCall(*this, CGM, args, fnType, 1);
 }
 
 const CGFunctionInfo &
@@ -395,7 +396,6 @@ CodeGenTypes::arrangeCXXMethodCall(const CallArgList &args,
     argTypes.push_back(Context.getCanonicalParamType(i->Ty));
 
   FunctionType::ExtInfo info = FPT->getExtInfo();
-  adjustCXXMethodInfo(*this, info, FPT->isVariadic());
   return arrangeLLVMFunctionInfo(GetReturnType(FPT->getResultType()),
                                  argTypes, info, required);
 }
@@ -644,6 +644,10 @@ EnterStructPointerForCoercedAccess(llvm::Value *SrcPtr,
 /// CoerceIntOrPtrToIntOrPtr - Convert a value Val to the specific Ty where both
 /// are either integers or pointers.  This does a truncation of the value if it
 /// is too large or a zero extension if it is too small.
+///
+/// This behaves as if the value were coerced through memory, so on big-endian
+/// targets the high bits are preserved in a truncation, while little-endian
+/// targets preserve the low bits.
 static llvm::Value *CoerceIntOrPtrToIntOrPtr(llvm::Value *Val,
                                              llvm::Type *Ty,
                                              CodeGenFunction &CGF) {
@@ -663,8 +667,25 @@ static llvm::Value *CoerceIntOrPtrToIntOrPtr(llvm::Value *Val,
   if (isa<llvm::PointerType>(DestIntTy))
     DestIntTy = CGF.IntPtrTy;
 
-  if (Val->getType() != DestIntTy)
-    Val = CGF.Builder.CreateIntCast(Val, DestIntTy, false, "coerce.val.ii");
+  if (Val->getType() != DestIntTy) {
+    const llvm::DataLayout &DL = CGF.CGM.getDataLayout();
+    if (DL.isBigEndian()) {
+      // Preserve the high bits on big-endian targets.
+      // That is what memory coercion does.
+      uint64_t SrcSize = DL.getTypeAllocSizeInBits(Val->getType());
+      uint64_t DstSize = DL.getTypeAllocSizeInBits(DestIntTy);
+      if (SrcSize > DstSize) {
+        Val = CGF.Builder.CreateLShr(Val, SrcSize - DstSize, "coerce.highbits");
+        Val = CGF.Builder.CreateTrunc(Val, DestIntTy, "coerce.val.ii");
+      } else {
+        Val = CGF.Builder.CreateZExt(Val, DestIntTy, "coerce.val.ii");
+        Val = CGF.Builder.CreateShl(Val, DstSize - SrcSize, "coerce.highbits");
+      }
+    } else {
+      // Little-endian targets preserve the low bits. No shifts required.
+      Val = CGF.Builder.CreateIntCast(Val, DestIntTy, false, "coerce.val.ii");
+    }
+  }
 
   if (isa<llvm::PointerType>(Ty))
     Val = CGF.Builder.CreateIntToPtr(Val, Ty, "coerce.val.ip");
@@ -1026,25 +1047,29 @@ void CodeGenModule::ConstructAttributeList(const CGFunctionInfo &FI,
     // Attributes that should go on the function, but not the call site.
     if (!CodeGenOpts.DisableFPElim) {
       FuncAttrs.addAttribute("no-frame-pointer-elim", "false");
-      FuncAttrs.addAttribute("no-frame-pointer-elim-non-leaf", "false");
     } else if (CodeGenOpts.OmitLeafFramePointer) {
       FuncAttrs.addAttribute("no-frame-pointer-elim", "false");
-      FuncAttrs.addAttribute("no-frame-pointer-elim-non-leaf", "true");
+      FuncAttrs.addAttribute("no-frame-pointer-elim-non-leaf");
     } else {
       FuncAttrs.addAttribute("no-frame-pointer-elim", "true");
-      FuncAttrs.addAttribute("no-frame-pointer-elim-non-leaf", "true");
+      FuncAttrs.addAttribute("no-frame-pointer-elim-non-leaf");
     }
 
     FuncAttrs.addAttribute("less-precise-fpmad",
-                           CodeGenOpts.LessPreciseFPMAD ? "true" : "false");
+                           llvm::toStringRef(CodeGenOpts.LessPreciseFPMAD));
     FuncAttrs.addAttribute("no-infs-fp-math",
-                           CodeGenOpts.NoInfsFPMath ? "true" : "false");
+                           llvm::toStringRef(CodeGenOpts.NoInfsFPMath));
     FuncAttrs.addAttribute("no-nans-fp-math",
-                           CodeGenOpts.NoNaNsFPMath ? "true" : "false");
+                           llvm::toStringRef(CodeGenOpts.NoNaNsFPMath));
     FuncAttrs.addAttribute("unsafe-fp-math",
-                           CodeGenOpts.UnsafeFPMath ? "true" : "false");
+                           llvm::toStringRef(CodeGenOpts.UnsafeFPMath));
     FuncAttrs.addAttribute("use-soft-float",
-                           CodeGenOpts.SoftFloat ? "true" : "false");
+                           llvm::toStringRef(CodeGenOpts.SoftFloat));
+    FuncAttrs.addAttribute("stack-protector-buffer-size",
+                           llvm::utostr(CodeGenOpts.SSPBufferSize));
+
+    if (!CodeGenOpts.StackRealignment)
+      FuncAttrs.addAttribute("no-realign-stack");
   }
 
   QualType RetTy = FI.getReturnType();
@@ -1052,12 +1077,15 @@ void CodeGenModule::ConstructAttributeList(const CGFunctionInfo &FI,
   const ABIArgInfo &RetAI = FI.getReturnInfo();
   switch (RetAI.getKind()) {
   case ABIArgInfo::Extend:
-   if (RetTy->hasSignedIntegerRepresentation())
-     RetAttrs.addAttribute(llvm::Attribute::SExt);
-   else if (RetTy->hasUnsignedIntegerRepresentation())
-     RetAttrs.addAttribute(llvm::Attribute::ZExt);
-    break;
+    if (RetTy->hasSignedIntegerRepresentation())
+      RetAttrs.addAttribute(llvm::Attribute::SExt);
+    else if (RetTy->hasUnsignedIntegerRepresentation())
+      RetAttrs.addAttribute(llvm::Attribute::ZExt);
+    // FALL THROUGH
   case ABIArgInfo::Direct:
+    if (RetAI.getInReg())
+      RetAttrs.addAttribute(llvm::Attribute::InReg);
+    break;
   case ABIArgInfo::Ignore:
     break;
 
@@ -1265,7 +1293,8 @@ void CodeGenFunction::EmitFunctionProlog(const CGFunctionInfo &FI,
       } else {
         // Load scalar value from indirect argument.
         CharUnits Alignment = getContext().getTypeAlignInChars(Ty);
-        V = EmitLoadOfScalar(V, false, Alignment.getQuantity(), Ty);
+        V = EmitLoadOfScalar(V, false, Alignment.getQuantity(), Ty,
+                             Arg->getLocStart());
 
         if (isPromoted)
           V = emitArgumentDemotion(*this, Arg, V);
@@ -1296,6 +1325,13 @@ void CodeGenFunction::EmitFunctionProlog(const CGFunctionInfo &FI,
         if (isPromoted)
           V = emitArgumentDemotion(*this, Arg, V);
 
+        if (const CXXMethodDecl *MD =
+            dyn_cast_or_null<CXXMethodDecl>(CurCodeDecl)) {
+          if (MD->isVirtual() && Arg == CXXABIThisDecl)
+            V = CGM.getCXXABI().
+                adjustThisParameterInVirtualFunctionPrologue(*this, CurGD, V);
+        }
+
         // Because of merging of function types from multiple decls it is
         // possible for the type of an argument to not match the corresponding
         // type in the function type. Since we are codegening the callee
@@ -1373,7 +1409,7 @@ void CodeGenFunction::EmitFunctionProlog(const CGFunctionInfo &FI,
 
       // Match to what EmitParmDecl is expecting for this type.
       if (CodeGenFunction::hasScalarEvaluationKind(Ty)) {
-        V = EmitLoadOfScalar(V, false, AlignmentToUse, Ty);
+        V = EmitLoadOfScalar(V, false, AlignmentToUse, Ty, Arg->getLocStart());
         if (isPromoted)
           V = emitArgumentDemotion(*this, Arg, V);
       }
@@ -1611,20 +1647,9 @@ static llvm::StoreInst *findDominatingStoreToReturnValue(CodeGenFunction &CGF) {
   return store;
 }
 
-/// Check whether 'this' argument of a callsite matches 'this' of the caller.
-static bool checkThisPointer(llvm::Value *ThisArg, llvm::Value *This) {
-  if (ThisArg == This)
-    return true;
-  // Check whether ThisArg is a bitcast of This.
-  llvm::BitCastInst *Bitcast;
-  if ((Bitcast = dyn_cast<llvm::BitCastInst>(ThisArg)) &&
-      Bitcast->getOperand(0) == This)
-    return true;
-  return false;
-}
-
 void CodeGenFunction::EmitFunctionEpilog(const CGFunctionInfo &FI,
-                                         bool EmitRetDbgLoc) {
+                                         bool EmitRetDbgLoc,
+                                         SourceLocation EndLoc) {
   // Functions with no result always return void.
   if (ReturnValue == 0) {
     Builder.CreateRetVoid();
@@ -1641,7 +1666,8 @@ void CodeGenFunction::EmitFunctionEpilog(const CGFunctionInfo &FI,
     switch (getEvaluationKind(RetTy)) {
     case TEK_Complex: {
       ComplexPairTy RT =
-        EmitLoadOfComplex(MakeNaturalAlignAddrLValue(ReturnValue, RetTy));
+        EmitLoadOfComplex(MakeNaturalAlignAddrLValue(ReturnValue, RetTy),
+                          EndLoc);
       EmitStoreOfComplex(RT,
                        MakeNaturalAlignAddrLValue(CurFn->arg_begin(), RetTy),
                          /*isInit*/ true);
@@ -1669,8 +1695,10 @@ void CodeGenFunction::EmitFunctionEpilog(const CGFunctionInfo &FI,
       // If there is a dominating store to ReturnValue, we can elide
       // the load, zap the store, and usually zap the alloca.
       if (llvm::StoreInst *SI = findDominatingStoreToReturnValue(*this)) {
-        // Reuse the debug location from the store unless we're told not to.
-        if (EmitRetDbgLoc)
+        // Reuse the debug location from the store unless there is
+        // cleanup code to be emitted between the store and return
+        // instruction.
+        if (EmitRetDbgLoc && !AutoreleaseResult)
           RetDbgLoc = SI->getDebugLoc();
         // Get the stored value and nuke the now-dead store.
         RV = SI->getValueOperand();
@@ -1717,26 +1745,14 @@ void CodeGenFunction::EmitFunctionEpilog(const CGFunctionInfo &FI,
     llvm_unreachable("Invalid ABI kind for return argument");
   }
 
-  // If this function returns 'this', the last instruction is a CallInst
-  // that returns 'this', and 'this' argument of the CallInst points to
-  // the same object as CXXThisValue, use the return value from the CallInst.
-  // We will not need to keep 'this' alive through the callsite. It also enables
-  // optimizations in the backend, such as tail call optimization.
-  if (CalleeWithThisReturn && CGM.getCXXABI().HasThisReturn(CurGD)) {
-    llvm::BasicBlock *IP = Builder.GetInsertBlock();
-    llvm::CallInst *Callsite;
-    if (!IP->empty() && (Callsite = dyn_cast<llvm::CallInst>(&IP->back())) &&
-        Callsite->getCalledFunction() == CalleeWithThisReturn &&
-        checkThisPointer(Callsite->getOperand(0), CXXThisValue))
-      RV = Builder.CreateBitCast(Callsite, RetAI.getCoerceToType());
-  }
   llvm::Instruction *Ret = RV ? Builder.CreateRet(RV) : Builder.CreateRetVoid();
   if (!RetDbgLoc.isUnknown())
     Ret->setDebugLoc(RetDbgLoc);
 }
 
 void CodeGenFunction::EmitDelegateCallArg(CallArgList &args,
-                                          const VarDecl *param) {
+                                          const VarDecl *param,
+                                          SourceLocation loc) {
   // StartFunction converted the ABI-lowered parameter(s) into a
   // local alloca.  We need to turn that into an r-value suitable
   // for EmitCall.
@@ -1757,7 +1773,7 @@ void CodeGenFunction::EmitDelegateCallArg(CallArgList &args,
     return args.add(RValue::get(Builder.CreateLoad(local)), type);
   }
 
-  args.add(convertTempToRValue(local, type), type);
+  args.add(convertTempToRValue(local, type, loc), type);
 }
 
 static bool isProvablyNull(llvm::Value *addr) {
@@ -1816,7 +1832,7 @@ static void emitWriteback(CodeGenFunction &CGF,
     CGF.EmitARCIntrinsicUse(writeback.ToUse);
 
     // Load the old value (primitively).
-    llvm::Value *oldValue = CGF.EmitLoadOfScalar(srcLV);
+    llvm::Value *oldValue = CGF.EmitLoadOfScalar(srcLV, SourceLocation());
 
     // Put the new value in place (primitively).
     CGF.EmitStoreOfScalar(value, srcLV, /*init*/ false);
@@ -1841,6 +1857,19 @@ static void emitWritebacks(CodeGenFunction &CGF,
     emitWriteback(CGF, *i);
 }
 
+static void deactivateArgCleanupsBeforeCall(CodeGenFunction &CGF,
+                                            const CallArgList &CallArgs) {
+  assert(CGF.getTarget().getCXXABI().isArgumentDestroyedByCallee());
+  ArrayRef<CallArgList::CallArgCleanup> Cleanups =
+    CallArgs.getCleanupsToDeactivate();
+  // Iterate in reverse to increase the likelihood of popping the cleanup.
+  for (ArrayRef<CallArgList::CallArgCleanup>::reverse_iterator
+         I = Cleanups.rbegin(), E = Cleanups.rend(); I != E; ++I) {
+    CGF.DeactivateCleanupBlock(I->Cleanup, I->IsActiveIP);
+    I->IsActiveIP->eraseFromParent();
+  }
+}
+
 static const Expr *maybeGetUnaryAddrOfOperand(const Expr *E) {
   if (const UnaryOperator *uop = dyn_cast<UnaryOperator>(E->IgnoreParens()))
     if (uop->getOpcode() == UO_AddrOf)
@@ -1932,7 +1961,7 @@ static void emitWritebackArg(CodeGenFunction &CGF, CallArgList &args,
 
   // Perform a copy if necessary.
   if (shouldCopy) {
-    RValue srcRV = CGF.EmitLoadOfLValue(srcLV);
+    RValue srcRV = CGF.EmitLoadOfLValue(srcLV, SourceLocation());
     assert(srcRV.isScalar());
 
     llvm::Value *src = srcRV.getScalarVal();
@@ -1989,16 +2018,47 @@ void CodeGenFunction::EmitCallArg(CallArgList &args, const Expr *E,
 
   if (E->isGLValue()) {
     assert(E->getObjectKind() == OK_Ordinary);
-    return args.add(EmitReferenceBindingToExpr(E, /*InitializedDecl=*/0),
-                    type);
+    return args.add(EmitReferenceBindingToExpr(E), type);
+  }
+
+  bool HasAggregateEvalKind = hasAggregateEvaluationKind(type);
+
+  // In the Microsoft C++ ABI, aggregate arguments are destructed by the callee.
+  // However, we still have to push an EH-only cleanup in case we unwind before
+  // we make it to the call.
+  if (HasAggregateEvalKind &&
+      CGM.getTarget().getCXXABI().isArgumentDestroyedByCallee()) {
+    const CXXRecordDecl *RD = type->getAsCXXRecordDecl();
+    if (RD && RD->hasNonTrivialDestructor()) {
+      AggValueSlot Slot = CreateAggTemp(type, "agg.arg.tmp");
+      Slot.setExternallyDestructed();
+      EmitAggExpr(E, Slot);
+      RValue RV = Slot.asRValue();
+      args.add(RV, type);
+
+      pushDestroy(EHCleanup, RV.getAggregateAddr(), type, destroyCXXObject,
+                  /*useEHCleanupForArray*/ true);
+      // This unreachable is a temporary marker which will be removed later.
+      llvm::Instruction *IsActive = Builder.CreateUnreachable();
+      args.addArgCleanupDeactivation(EHStack.getInnermostEHScope(), IsActive);
+      return;
+    }
   }
 
-  if (hasAggregateEvaluationKind(type) &&
-      isa<ImplicitCastExpr>(E) &&
+  if (HasAggregateEvalKind && isa<ImplicitCastExpr>(E) &&
       cast<CastExpr>(E)->getCastKind() == CK_LValueToRValue) {
     LValue L = EmitLValue(cast<CastExpr>(E)->getSubExpr());
     assert(L.isSimple());
-    args.add(L.asAggregateRValue(), type, /*NeedsCopy*/true);
+    if (L.getAlignment() >= getContext().getTypeAlignInChars(type)) {
+      args.add(L.asAggregateRValue(), type, /*NeedsCopy*/true);
+    } else {
+      // We can't represent a misaligned lvalue in the CallArgList, so copy
+      // to an aligned temporary now.
+      llvm::Value *tmp = CreateMemTemp(type);
+      EmitAggregateCopy(tmp, L.getAddress(), type, L.isVolatile(),
+                        L.getAlignment());
+      args.add(RValue::getAggregate(tmp), type);
+    }
     return;
   }
 
@@ -2129,7 +2189,7 @@ static void checkArgMatches(llvm::Value *Elt, unsigned &ArgNo,
 }
 
 void CodeGenFunction::ExpandTypeToArgs(QualType Ty, RValue RV,
-                                       SmallVector<llvm::Value*,16> &Args,
+                                       SmallVectorImpl<llvm::Value *> &Args,
                                        llvm::FunctionType *IRFuncTy) {
   if (const ConstantArrayType *AT = getContext().getAsConstantArrayType(Ty)) {
     unsigned NumElts = AT->getSize().getZExtValue();
@@ -2137,7 +2197,7 @@ void CodeGenFunction::ExpandTypeToArgs(QualType Ty, RValue RV,
     llvm::Value *Addr = RV.getAggregateAddr();
     for (unsigned Elt = 0; Elt < NumElts; ++Elt) {
       llvm::Value *EltAddr = Builder.CreateConstGEP2_32(Addr, 0, Elt);
-      RValue EltRV = convertTempToRValue(EltAddr, EltTy);
+      RValue EltRV = convertTempToRValue(EltAddr, EltTy, SourceLocation());
       ExpandTypeToArgs(EltTy, EltRV, Args, IRFuncTy);
     }
   } else if (const RecordType *RT = Ty->getAs<RecordType>()) {
@@ -2161,7 +2221,7 @@ void CodeGenFunction::ExpandTypeToArgs(QualType Ty, RValue RV,
         }
       }
       if (LargestFD) {
-        RValue FldRV = EmitRValueForField(LV, LargestFD);
+        RValue FldRV = EmitRValueForField(LV, LargestFD, SourceLocation());
         ExpandTypeToArgs(LargestFD->getType(), FldRV, Args, IRFuncTy);
       }
     } else {
@@ -2169,7 +2229,7 @@ void CodeGenFunction::ExpandTypeToArgs(QualType Ty, RValue RV,
            i != e; ++i) {
         FieldDecl *FD = *i;
 
-        RValue FldRV = EmitRValueForField(LV, FD);
+        RValue FldRV = EmitRValueForField(LV, FD, SourceLocation());
         ExpandTypeToArgs(FD->getType(), FldRV, Args, IRFuncTy);
       }
     }
@@ -2396,6 +2456,9 @@ RValue CodeGenFunction::EmitCall(const CGFunctionInfo &CallInfo,
     }
   }
 
+  if (!CallArgs.getCleanupsToDeactivate().empty())
+    deactivateArgCleanupsBeforeCall(*this, CallArgs);
+
   // If the callee is a bitcast of a function to a varargs pointer to function
   // type, check to see if we can remove the bitcast.  This handles some cases
   // with unprototyped functions.
@@ -2484,7 +2547,7 @@ RValue CodeGenFunction::EmitCall(const CGFunctionInfo &CallInfo,
 
   switch (RetAI.getKind()) {
   case ABIArgInfo::Indirect:
-    return convertTempToRValue(Args[0], RetTy);
+    return convertTempToRValue(Args[0], RetTy, SourceLocation());
 
   case ABIArgInfo::Ignore:
     // If we are ignoring an argument that had a result, make sure to
@@ -2542,7 +2605,7 @@ RValue CodeGenFunction::EmitCall(const CGFunctionInfo &CallInfo,
     }
     CreateCoercedStore(CI, StorePtr, DestIsVolatile, *this);
 
-    return convertTempToRValue(DestPtr, RetTy);
+    return convertTempToRValue(DestPtr, RetTy, SourceLocation());
   }
 
   case ABIArgInfo::Expand: