RTABI chapter 4.3.4 specifies __eabi_mem* calls. Specifically, __eabi_memset accepts parameters (ptr, size, value) in a different order than GNU's memset (ptr, value, size), therefore the special lowering in AAPCS mode. Implementation by Evzen Muller.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@131868 91177308-0d34-0410-b5e6-96231b3b80d8

4 files changed, 99 insertions, 4 deletions

diff --git a/lib/Target/ARM/ARMISelLowering.cpp b/lib/Target/ARM/ARMISelLowering.cpp
index e9a9963..5c1cded 100644
--- a/lib/Target/ARM/ARMISelLowering.cpp
+++ b/lib/Target/ARM/ARMISelLowering.cpp
@@ -396,6 +396,12 @@ ARMTargetLowering::ARMTargetLowering(TargetMachine &TM)
     setLibcallCallingConv(RTLIB::UDIV_I8, CallingConv::ARM_AAPCS);
     setLibcallCallingConv(RTLIB::UDIV_I16, CallingConv::ARM_AAPCS);
     setLibcallCallingConv(RTLIB::UDIV_I32, CallingConv::ARM_AAPCS);
+
+    // Memory operations
+    // RTABI chapter 4.3.4
+    setLibcallName(RTLIB::MEMCPY,  "__aeabi_memcpy");
+    setLibcallName(RTLIB::MEMMOVE, "__aeabi_memmove");
+    setLibcallName(RTLIB::MEMSET,  "__aeabi_memset");
   }
 
   if (Subtarget->isThumb1Only())
diff --git a/lib/Target/ARM/ARMSelectionDAGInfo.cpp b/lib/Target/ARM/ARMSelectionDAGInfo.cpp
index aa1e398..ef0aaf2 100644
--- a/lib/Target/ARM/ARMSelectionDAGInfo.cpp
+++ b/lib/Target/ARM/ARMSelectionDAGInfo.cpp
@@ -13,6 +13,8 @@
 
 #define DEBUG_TYPE "arm-selectiondag-info"
 #include "ARMTargetMachine.h"
+#include "llvm/DerivedTypes.h"
+#include "llvm/CodeGen/SelectionDAG.h"
 using namespace llvm;
 
 ARMSelectionDAGInfo::ARMSelectionDAGInfo(const TargetMachine &TM)
@@ -132,3 +134,65 @@ ARMSelectionDAGInfo::EmitTargetCodeForMemcpy(SelectionDAG &DAG, DebugLoc dl,
   }
   return DAG.getNode(ISD::TokenFactor, dl, MVT::Other, &TFOps[0], i);
 }
+
+// Adjust parameters for memset, EABI uses format (ptr, size, value),
+// GNU library uses (ptr, value, size)
+// See RTABI section 4.3.4
+SDValue
+ARMSelectionDAGInfo::EmitTargetCodeForMemset(SelectionDAG &DAG, DebugLoc dl,
+                                             SDValue Chain, SDValue Dst,
+                                             SDValue Src, SDValue Size,
+                                             unsigned Align, bool isVolatile,
+                                             MachinePointerInfo DstPtrInfo) const
+{
+  // Use default for non AAPCS subtargets
+  if (!Subtarget->isAAPCS_ABI())
+    return SDValue();
+
+  const ARMTargetLowering &TLI =
+    *static_cast<const ARMTargetLowering*>(DAG.getTarget().getTargetLowering());
+  TargetLowering::ArgListTy Args;
+  TargetLowering::ArgListEntry Entry;
+
+  // First argument: data pointer
+  const Type *IntPtrTy = TLI.getTargetData()->getIntPtrType(*DAG.getContext());
+  Entry.Node = Dst;
+  Entry.Ty = IntPtrTy;
+  Args.push_back(Entry);
+
+  // Second argument: buffer size
+  Entry.Node = Size;
+  Entry.Ty = IntPtrTy;
+  Entry.isSExt = false;
+  Args.push_back(Entry);
+
+  // Extend or truncate the argument to be an i32 value for the call.
+  if (Src.getValueType().bitsGT(MVT::i32))
+    Src = DAG.getNode(ISD::TRUNCATE, dl, MVT::i32, Src);
+  else
+    Src = DAG.getNode(ISD::ZERO_EXTEND, dl, MVT::i32, Src);
+
+  // Third argument: value to fill
+  Entry.Node = Src;
+  Entry.Ty = Type::getInt32Ty(*DAG.getContext());
+  Entry.isSExt = true;
+  Args.push_back(Entry);
+
+  // Emit __eabi_memset call
+  std::pair<SDValue,SDValue> CallResult =
+    TLI.LowerCallTo(Chain,
+                    Type::getVoidTy(*DAG.getContext()), // return type
+                    false, // return sign ext
+                    false, // return zero ext
+                    false, // is var arg
+                    false, // is in regs
+                    0,     // number of fixed arguments
+                    TLI.getLibcallCallingConv(RTLIB::MEMSET), // call conv
+                    false, // is tail call
+                    false, // is return val used
+                    DAG.getExternalSymbol(TLI.getLibcallName(RTLIB::MEMSET),
+                                          TLI.getPointerTy()), // callee
+                    Args, DAG, dl); // arg list, DAG and debug
+
+  return CallResult.second;
+}
diff --git a/lib/Target/ARM/ARMSelectionDAGInfo.h b/lib/Target/ARM/ARMSelectionDAGInfo.h
index 7533690..ec1bf5c 100644
--- a/lib/Target/ARM/ARMSelectionDAGInfo.h
+++ b/lib/Target/ARM/ARMSelectionDAGInfo.h
@@ -35,6 +35,15 @@ public:
                                   bool isVolatile, bool AlwaysInline,
                                   MachinePointerInfo DstPtrInfo,
                                   MachinePointerInfo SrcPtrInfo) const;
+
+  // Adjust parameters for memset, see RTABI section 4.3.4
+  virtual
+  SDValue EmitTargetCodeForMemset(SelectionDAG &DAG, DebugLoc dl,
+                                  SDValue Chain,
+                                  SDValue Op1, SDValue Op2,
+                                  SDValue Op3, unsigned Align,
+                                  bool isVolatile,
+                                  MachinePointerInfo DstPtrInfo) const;
 };
 
 }
diff --git a/test/CodeGen/ARM/memfunc.ll b/test/CodeGen/ARM/memfunc.ll
index 41d5944..032129d 100644
--- a/test/CodeGen/ARM/memfunc.ll
+++ b/test/CodeGen/ARM/memfunc.ll
@@ -1,10 +1,26 @@
-; RUN: llc < %s -march=arm
+; RUN: llc < %s -march=arm -o - | FileCheck %s
+; RUN: llc < %s -mtriple=arm-none-eabi -o - | FileCheck --check-prefix=EABI %s
+
+@from = common global [500 x i32] zeroinitializer, align 4
+@to = common global [500 x i32] zeroinitializer, align 4
 
 define void @f() {
 entry:
-        call void @llvm.memmove.i32( i8* null, i8* null, i32 64, i32 0 )
-        call void @llvm.memcpy.i32( i8* null, i8* null, i32 64, i32 0 )
-        call void @llvm.memset.i32( i8* null, i8 64, i32 0, i32 0 )
+
+        ; CHECK: memmove
+        ; EABI: __aeabi_memmove
+        call void @llvm.memmove.i32( i8* bitcast ([500 x i32]* @from to i8*), i8* bitcast ([500 x i32]* @to to i8*), i32 500, i32 0 )
+
+        ; CHECK: memcpy
+        ; EABI: __aeabi_memcpy
+        call void @llvm.memcpy.i32( i8* bitcast ([500 x i32]* @from to i8*), i8* bitcast ([500 x i32]* @to to i8*), i32 500, i32 0 )
+
+        ; EABI memset swaps arguments
+        ; CHECK: mov r1, #0
+        ; CHECK: memset
+        ; EABI: mov r2, #0
+        ; EABI: __aeabi_memset
+        call void @llvm.memset.i32( i8* bitcast ([500 x i32]* @from to i8*), i8 0, i32 500, i32 0 )
         unreachable
 }