3 files changed, 21 insertions, 12 deletions
diff --git a/lib/CodeGen/CallingConvLower.cpp b/lib/CodeGen/CallingConvLower.cpp
index cdae33c2441..22b91409240 100644
--- a/lib/CodeGen/CallingConvLower.cpp
+++ b/lib/CodeGen/CallingConvLower.cpp
@@ -50,7 +50,7 @@ void CCState::HandleByVal(unsigned ValNo, MVT ValVT,
   if (MinAlign > (int)Align)
     Align = MinAlign;
   MF.getFrameInfo()->ensureMaxAlignment(Align);
-  TM.getTargetLowering()->HandleByVal(this, Size);
+  TM.getTargetLowering()->HandleByVal(this, Size, Align);
   unsigned Offset = AllocateStack(Size, Align);
   addLoc(CCValAssign::getMem(ValNo, ValVT, Offset, LocVT, LocInfo));
 }
diff --git a/lib/Target/ARM/ARMISelLowering.cpp b/lib/Target/ARM/ARMISelLowering.cpp
index 6ae44eb3537..b2eb5784879 100644
--- a/lib/Target/ARM/ARMISelLowering.cpp
+++ b/lib/Target/ARM/ARMISelLowering.cpp
@@ -1656,22 +1656,31 @@ ARMTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
 /// and then confiscate the rest of the parameter registers to insure
 /// this.
 void
-ARMTargetLowering::HandleByVal(CCState *State, unsigned &size) const {
+ARMTargetLowering::HandleByVal(
+    CCState *State, unsigned &size, unsigned Align) const {
   unsigned reg = State->AllocateReg(GPRArgRegs, 4);
   assert((State->getCallOrPrologue() == Prologue ||
           State->getCallOrPrologue() == Call) &&
          "unhandled ParmContext");
   if ((!State->isFirstByValRegValid()) &&
       (ARM::R0 <= reg) && (reg <= ARM::R3)) {
-    State->setFirstByValReg(reg);
-    // At a call site, a byval parameter that is split between
-    // registers and memory needs its size truncated here.  In a
-    // function prologue, such byval parameters are reassembled in
-    // memory, and are not truncated.
-    if (State->getCallOrPrologue() == Call) {
-      unsigned excess = 4 * (ARM::R4 - reg);
-      assert(size >= excess && "expected larger existing stack allocation");
-      size -= excess;
+    if (Subtarget->isAAPCS_ABI() && Align > 4) {
+      unsigned AlignInRegs = Align / 4;
+      unsigned Waste = (ARM::R4 - reg) % AlignInRegs;
+      for (unsigned i = 0; i < Waste; ++i)
+        reg = State->AllocateReg(GPRArgRegs, 4);
+    }
+    if (reg != 0) {
+      State->setFirstByValReg(reg);
+      // At a call site, a byval parameter that is split between
+      // registers and memory needs its size truncated here.  In a
+      // function prologue, such byval parameters are reassembled in
+      // memory, and are not truncated.
+      if (State->getCallOrPrologue() == Call) {
+        unsigned excess = 4 * (ARM::R4 - reg);
+        assert(size >= excess && "expected larger existing stack allocation");
+        size -= excess;
+      }
     }
   }
   // Confiscate any remaining parameter registers to preclude their
diff --git a/lib/Target/ARM/ARMISelLowering.h b/lib/Target/ARM/ARMISelLowering.h
index b5020c97108..9acab0b0834 100644
--- a/lib/Target/ARM/ARMISelLowering.h
+++ b/lib/Target/ARM/ARMISelLowering.h
@@ -480,7 +480,7 @@ namespace llvm {
                 SmallVectorImpl<SDValue> &InVals) const;
 
     /// HandleByVal - Target-specific cleanup for ByVal support.
-    virtual void HandleByVal(CCState *, unsigned &) const;
+    virtual void HandleByVal(CCState *, unsigned &, unsigned) const;
 
     /// IsEligibleForTailCallOptimization - Check whether the call is eligible
     /// for tail call optimization. Targets which want to do tail call