Add zero-length check to memcpy/memset load store loop expansion

[tejohnson]

Summary

I was testing using this expansion logic in other cases besides
NVPTX, and found some runtime failures due to the lack of a check
for a zero length memcpy/memset before the loop. There is already
such a check in the memmove expansion code though.

[hfinkel]

LGTM

lib/Transforms/Utils/LowerMemIntrinsics.cpp

void llvm::createMemCpyLoop(Instruction *InsertBefore,
 
   BasicBlock *OrigBB = InsertBefore->getParent();
   Function *F = OrigBB->getParent();
   BasicBlock *NewBB =
     InsertBefore->getParent()->splitBasicBlock(InsertBefore, "split");
   BasicBlock *LoopBB = BasicBlock::Create(F->getContext(), "loadstoreloop",
                                           F, NewBB);
 
-  OrigBB->getTerminator()->setSuccessor(0, LoopBB);
   IRBuilder<> Builder(OrigBB->getTerminator());
+  Builder.CreateCondBr(
+      Builder.CreateICmpEQ(ConstantInt::get(TypeOfCopyLen, 0), CopyLen), NewBB,
+      LoopBB);
+  OrigBB->getTerminator()->eraseFromParent();
 
   // SrcAddr and DstAddr are expected to be pointer types,
   // so no check is made here.
   unsigned SrcAS = cast<PointerType>(SrcAddr->getType())->getAddressSpace();
   unsigned DstAS = cast<PointerType>(DstAddr->getType())->getAddressSpace();
 
   // Cast pointers to (char *)
   SrcAddr = Builder.CreateBitCast(SrcAddr, Builder.getInt8PtrTy(SrcAS));
static void createMemMoveLoop(Instruction *InsertBefore,
 
   BranchInst::Create(ExitBB, FwdLoopBB, CompareN, ElseTerm);
   ElseTerm->eraseFromParent();
 }
 
 static void createMemSetLoop(Instruction *InsertBefore,
                              Value *DstAddr, Value *CopyLen, Value *SetValue,
                              unsigned Align, bool IsVolatile) {
+  Type *TypeOfCopyLen = CopyLen->getType();
   BasicBlock *OrigBB = InsertBefore->getParent();
   Function *F = OrigBB->getParent();
   BasicBlock *NewBB =
       OrigBB->splitBasicBlock(InsertBefore, "split");
   BasicBlock *LoopBB
     = BasicBlock::Create(F->getContext(), "loadstoreloop", F, NewBB);
 
-  OrigBB->getTerminator()->setSuccessor(0, LoopBB);
   IRBuilder<> Builder(OrigBB->getTerminator());
+  Builder.CreateCondBr(
+      Builder.CreateICmpEQ(ConstantInt::get(TypeOfCopyLen, 0), CopyLen), NewBB,
+      LoopBB);
+  OrigBB->getTerminator()->eraseFromParent();
 
   // Cast pointer to the type of value getting stored
   unsigned dstAS = cast<PointerType>(DstAddr->getType())->getAddressSpace();
   DstAddr = Builder.CreateBitCast(DstAddr,
                                   PointerType::get(SetValue->getType(), dstAS));
 
   IRBuilder<> LoopBuilder(LoopBB);
-  PHINode *LoopIndex = LoopBuilder.CreatePHI(CopyLen->getType(), 0);
-  LoopIndex->addIncoming(ConstantInt::get(CopyLen->getType(), 0), OrigBB);
+  PHINode *LoopIndex = LoopBuilder.CreatePHI(TypeOfCopyLen, 0);
+  LoopIndex->addIncoming(ConstantInt::get(TypeOfCopyLen, 0), OrigBB);
 
   LoopBuilder.CreateStore(
       SetValue,
       LoopBuilder.CreateInBoundsGEP(SetValue->getType(), DstAddr, LoopIndex),
       IsVolatile);
 
   Value *NewIndex =
-      LoopBuilder.CreateAdd(LoopIndex, ConstantInt::get(CopyLen->getType(), 1));
+      LoopBuilder.CreateAdd(LoopIndex, ConstantInt::get(TypeOfCopyLen, 1));
   LoopIndex->addIncoming(NewIndex, LoopBB);
 
   LoopBuilder.CreateCondBr(LoopBuilder.CreateICmpULT(NewIndex, CopyLen), LoopBB,
                            NewBB);
 }
 
 void llvm::expandMemCpyAsLoop(MemCpyInst *Memcpy) {
   createMemCpyLoop(/* InsertBefore */ Memcpy,

test/CodeGen/NVPTX/lower-aggr-copies.ll

 declare void @llvm.memset.p0i8.i64(i8* nocapture, i8, i64, i32, i1) #1
 
 define i8* @memcpy_caller(i8* %dst, i8* %src, i64 %n) #0 {
 entry:
   tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* %dst, i8* %src, i64 %n, i32 1, i1 false)
   ret i8* %dst
 
 ; IR-LABEL:   @memcpy_caller
+; IR:         [[CMPREG:%[0-9]+]] = icmp eq i64 0, %n
+; IR:         br i1 [[CMPREG]], label %split, label %loadstoreloop
 ; IR:         loadstoreloop:
 ; IR:         [[LOADPTR:%[0-9]+]] = getelementptr inbounds i8, i8* %src, i64
 ; IR-NEXT:    [[VAL:%[0-9]+]] = load i8, i8* [[LOADPTR]]
 ; IR-NEXT:    [[STOREPTR:%[0-9]+]] = getelementptr inbounds i8, i8* %dst, i64
 ; IR-NEXT:    store i8 [[VAL]], i8* [[STOREPTR]]
 
 ; PTX-LABEL:  .visible .func (.param .b64 func_retval0) memcpy_caller
 ; PTX:        LBB[[LABEL:[_0-9]+]]:

 define i8* @memset_caller(i8* %dst, i32 %c, i64 %n) #0 {
 entry:
   %0 = trunc i32 %c to i8
   tail call void @llvm.memset.p0i8.i64(i8* %dst, i8 %0, i64 %n, i32 1, i1 false)
   ret i8* %dst
 
 ; IR-LABEL:   @memset_caller
 ; IR:         [[VAL:%[0-9]+]] = trunc i32 %c to i8
+; IR:         [[CMPREG:%[0-9]+]] = icmp eq i64 0, %n
+; IR:         br i1 [[CMPREG]], label %split, label %loadstoreloop
 ; IR:         loadstoreloop:
 ; IR:         [[STOREPTR:%[0-9]+]] = getelementptr inbounds i8, i8* %dst, i64
 ; IR-NEXT:    store i8 [[VAL]], i8* [[STOREPTR]]
 
 ; PTX-LABEL:  .visible .func (.param .b64 func_retval0) memset_caller(
 ; PTX:        ld.param.u32 %r[[C:[0-9]+]]
 ; PTX:        cvt.u16.u32  %rs[[REG:[0-9]+]], %r[[C]];
 ; PTX:        LBB[[LABEL:[_0-9]+]]: