4 files changed, 45 insertions, 8 deletions

diff --git a/lib/CodeGen/AsmPrinter/DwarfDebug.cpp b/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
index 73bba6989f..1e706ccf1e 100644
--- a/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
+++ b/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
@@ -1624,9 +1624,37 @@ void DwarfDebug::beginFunction(const MachineFunction *MF) {
 }
 
 void DwarfDebug::addScopeVariable(LexicalScope *LS, DbgVariable *Var) {
-//  SmallVector<DbgVariable *, 8> &Vars = ScopeVariables.lookup(LS);
-  ScopeVariables[LS].push_back(Var);
-//  Vars.push_back(Var);
+  SmallVectorImpl<DbgVariable *> &Vars = ScopeVariables[LS];
+  DIVariable DV = Var->getVariable();
+  if (DV.getTag() == dwarf::DW_TAG_arg_variable) {
+    DISubprogram Ctxt(DV.getContext());
+    DIArray Variables = Ctxt.getVariables();
+    // If the variable is a parameter (arg_variable) and this is an optimized
+    // build (the subprogram has a 'variables' list) make sure we keep the
+    // parameters in order. Otherwise we would produce an incorrect function
+    // type with parameters out of order if function parameters were used out of
+    // order or unused (see the call to addScopeVariable in endFunction where
+    // the remaining unused variables (including parameters) are added).
+    if (unsigned NumVariables = Variables.getNumElements()) {
+      // Keep the parameters at the start of the variables list. Search through
+      // current variable list (Vars) and the full function variable list in
+      // lock-step looking for this parameter in the full list to find the
+      // insertion point.
+      SmallVectorImpl<DbgVariable *>::iterator I = Vars.begin();
+      unsigned j = 0;
+      while (I != Vars.end() && j != NumVariables &&
+             Variables.getElement(j) != DV &&
+             (*I)->getVariable().getTag() == dwarf::DW_TAG_arg_variable) {
+        if (Variables.getElement(j) == (*I)->getVariable())
+          ++I;
+        ++j;
+      }
+      Vars.insert(I, Var);
+      return;
+    }
+  }
+
+  Vars.push_back(Var);
 }
 
 // Gather and emit post-function debug information.
diff --git a/lib/Target/AArch64/Utils/CMakeLists.txt b/lib/Target/AArch64/Utils/CMakeLists.txt
index 2c28348d7d..2348e44f85 100644
--- a/lib/Target/AArch64/Utils/CMakeLists.txt
+++ b/lib/Target/AArch64/Utils/CMakeLists.txt
@@ -3,3 +3,5 @@ include_directories( ${CMAKE_CURRENT_BINARY_DIR}/.. ${CMAKE_CURRENT_SOURCE_DIR}/
 add_llvm_library(LLVMAArch64Utils
   AArch64BaseInfo.cpp
   )
+
+add_dependencies(LLVMAArch64Utils AArch64CommonTableGen)
diff --git a/lib/Target/R600/R600ControlFlowFinalizer.cpp b/lib/Target/R600/R600ControlFlowFinalizer.cpp
index ffe3414413..7d3a0f52de 100644
--- a/lib/Target/R600/R600ControlFlowFinalizer.cpp
+++ b/lib/Target/R600/R600ControlFlowFinalizer.cpp
@@ -116,8 +116,15 @@ private:
       const MachineOperand &MO = *I;
       if (!MO.isReg())
         continue;
-      if (MO.isDef())
-        DstMI = MO.getReg();
+      if (MO.isDef()) {
+        unsigned Reg = MO.getReg();
+        if (AMDGPU::R600_Reg128RegClass.contains(Reg))
+          DstMI = Reg;
+        else
+          DstMI = TRI.getMatchingSuperReg(Reg,
+              TRI.getSubRegFromChannel(TRI.getHWRegChan(Reg)),
+              &AMDGPU::R600_Reg128RegClass);
+      }
       if (MO.isUse()) {
         unsigned Reg = MO.getReg();
         if (AMDGPU::R600_Reg128RegClass.contains(Reg))
diff --git a/lib/Transforms/Vectorize/LoopVectorize.cpp b/lib/Transforms/Vectorize/LoopVectorize.cpp
index 08d372512d..11ee99ddf1 100644
--- a/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -964,7 +964,7 @@ void InnerLoopVectorizer::vectorizeMemoryInstruction(Instruction *Instr,
   Type *DataTy = VectorType::get(ScalarDataTy, VF);
   Value *Ptr = LI ? LI->getPointerOperand() : SI->getPointerOperand();
   unsigned Alignment = LI ? LI->getAlignment() : SI->getAlignment();
-
+  unsigned AddressSpace = Ptr->getType()->getPointerAddressSpace();
   unsigned ScalarAllocatedSize = DL->getTypeAllocSize(ScalarDataTy);
   unsigned VectorElementSize = DL->getTypeStoreSize(DataTy)/VF;
 
@@ -1039,7 +1039,7 @@ void InnerLoopVectorizer::vectorizeMemoryInstruction(Instruction *Instr,
         PartPtr = Builder.CreateGEP(PartPtr, Builder.getInt32(1 - VF));
       }
 
-      Value *VecPtr = Builder.CreateBitCast(PartPtr, DataTy->getPointerTo());
+      Value *VecPtr = Builder.CreateBitCast(PartPtr, DataTy->getPointerTo(AddressSpace));
       Builder.CreateStore(StoredVal[Part], VecPtr)->setAlignment(Alignment);
     }
   }
@@ -1055,7 +1055,7 @@ void InnerLoopVectorizer::vectorizeMemoryInstruction(Instruction *Instr,
       PartPtr = Builder.CreateGEP(PartPtr, Builder.getInt32(1 - VF));
     }
 
-    Value *VecPtr = Builder.CreateBitCast(PartPtr, DataTy->getPointerTo());
+    Value *VecPtr = Builder.CreateBitCast(PartPtr, DataTy->getPointerTo(AddressSpace));
     Value *LI = Builder.CreateLoad(VecPtr, "wide.load");
     cast<LoadInst>(LI)->setAlignment(Alignment);
     Entry[Part] = Reverse ? reverseVector(LI) :  LI;