[XCore] Target optimized library function __memcpy_4()

Summary:
If the src, dst and size of a memcpy are known to be 4 byte aligned we
can call __memcpy_4() instead of memcpy().

Reviewers: robertlytton

Reviewed By: robertlytton

CC: llvm-commits

Differential Revision: http://llvm-reviews.chandlerc.com/D2871

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@202395 91177308-0d34-0410-b5e6-96231b3b80d8

2 files changed, 42 insertions, 0 deletions

diff --git a/lib/Target/XCore/XCoreSelectionDAGInfo.cpp b/lib/Target/XCore/XCoreSelectionDAGInfo.cpp
index 44aeb6057c..68ede6ae6d 100644
--- a/lib/Target/XCore/XCoreSelectionDAGInfo.cpp
+++ b/lib/Target/XCore/XCoreSelectionDAGInfo.cpp
@@ -21,3 +21,36 @@ XCoreSelectionDAGInfo::XCoreSelectionDAGInfo(const XCoreTargetMachine &TM)
 
 XCoreSelectionDAGInfo::~XCoreSelectionDAGInfo() {
 }
+
+SDValue XCoreSelectionDAGInfo::
+EmitTargetCodeForMemcpy(SelectionDAG &DAG, SDLoc dl, SDValue Chain,
+                        SDValue Dst, SDValue Src, SDValue Size, unsigned Align,
+                        bool isVolatile, bool AlwaysInline,
+                        MachinePointerInfo DstPtrInfo,
+                        MachinePointerInfo SrcPtrInfo) const
+{
+  unsigned SizeBitWidth = Size.getValueType().getSizeInBits();
+  // Call __memcpy_4 if the src, dst and size are all 4 byte aligned.
+  if (!AlwaysInline && (Align & 3) == 0 &&
+      DAG.MaskedValueIsZero(Size, APInt(SizeBitWidth, 3))) {
+    const TargetLowering &TLI = *DAG.getTarget().getTargetLowering();
+    TargetLowering::ArgListTy Args;
+    TargetLowering::ArgListEntry Entry;
+    Entry.Ty = TLI.getDataLayout()->getIntPtrType(*DAG.getContext());
+    Entry.Node = Dst; Args.push_back(Entry);
+    Entry.Node = Src; Args.push_back(Entry);
+    Entry.Node = Size; Args.push_back(Entry);
+
+    TargetLowering::CallLoweringInfo
+    CLI(Chain, Type::getVoidTy(*DAG.getContext()), false, false, false, false,
+        0, TLI.getLibcallCallingConv(RTLIB::MEMCPY), /*isTailCall=*/false,
+        /*doesNotRet=*/false, /*isReturnValueUsed=*/false,
+        DAG.getExternalSymbol("__memcpy_4", TLI.getPointerTy()), Args, DAG, dl);
+    std::pair<SDValue,SDValue> CallResult =
+      TLI.LowerCallTo(CLI);
+    return CallResult.second;
+  }
+
+  // Otherwise have the target-independent code call memcpy.
+  return SDValue();
+}
diff --git a/lib/Target/XCore/XCoreSelectionDAGInfo.h b/lib/Target/XCore/XCoreSelectionDAGInfo.h
index 0386968638..31704f388a 100644
--- a/lib/Target/XCore/XCoreSelectionDAGInfo.h
+++ b/lib/Target/XCore/XCoreSelectionDAGInfo.h
@@ -24,6 +24,15 @@ class XCoreSelectionDAGInfo : public TargetSelectionDAGInfo {
 public:
   explicit XCoreSelectionDAGInfo(const XCoreTargetMachine &TM);
   ~XCoreSelectionDAGInfo();
+
+  virtual SDValue
+  EmitTargetCodeForMemcpy(SelectionDAG &DAG, SDLoc dl,
+                          SDValue Chain,
+                          SDValue Op1, SDValue Op2,
+                          SDValue Op3, unsigned Align, bool isVolatile,
+                          bool AlwaysInline,
+                          MachinePointerInfo DstPtrInfo,
+                          MachinePointerInfo SrcPtrInfo) const;
 };
 
 }