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

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.

Reviewers: hfinkel

Subscribers: jholewinski, wdng, llvm-commits

Differential Revision: https://reviews.llvm.org/D34707

llvm-svn: 306541

2 files changed, 16 insertions, 5 deletions

diff --git a/llvm/lib/Transforms/Utils/LowerMemIntrinsics.cpp b/llvm/lib/Transforms/Utils/LowerMemIntrinsics.cpp
index 0a51f9a0e4a..ef829d43b7d 100644
--- a/llvm/lib/Transforms/Utils/LowerMemIntrinsics.cpp
+++ b/llvm/lib/Transforms/Utils/LowerMemIntrinsics.cpp
@@ -27,8 +27,11 @@ void llvm::createMemCpyLoop(Instruction *InsertBefore,
   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.
@@ -167,6 +170,7 @@ static void createMemMoveLoop(Instruction *InsertBefore,
 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 =
@@ -174,8 +178,11 @@ static void createMemSetLoop(Instruction *InsertBefore,
   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();
@@ -183,8 +190,8 @@ static void createMemSetLoop(Instruction *InsertBefore,
                                   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,
@@ -192,7 +199,7 @@ static void createMemSetLoop(Instruction *InsertBefore,
       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,
diff --git a/llvm/test/CodeGen/NVPTX/lower-aggr-copies.ll b/llvm/test/CodeGen/NVPTX/lower-aggr-copies.ll
index 192d4becb05..f522c6722ee 100644
--- a/llvm/test/CodeGen/NVPTX/lower-aggr-copies.ll
+++ b/llvm/test/CodeGen/NVPTX/lower-aggr-copies.ll
@@ -17,6 +17,8 @@ entry:
   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]]
@@ -73,6 +75,8 @@ entry:
 
 ; 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]]