RPCSX/llvm
1
0
Fork 0
mirror of https://github.com/RPCSX/llvm.git synced 2024-12-11 05:35:11 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

NVPTX: Fix not preserving volatile when expanding memset

Browse Source 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@293851 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Matt Arsenault 2017-02-02 01:20:34 +00:00
parent 367ed08f22
commit 8c93ddba10
2 changed files with 17 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
lib/Target/NVPTX/NVPTXLowerAggrCopies.cpp
View File

@ -204,8 +204,8 @@ void convertMemMoveToLoop(Instruction *ConvertedInst, Value *SrcAddr,
// Lower memset to loop.
void convertMemSetToLoop(Instruction *ConvertedInst, Value *DstAddr,
Value *CopyLen, Value *SetValue, LLVMContext &Context,
Function &F) {
Value *CopyLen, Value *SetValue,
bool IsVolatile, LLVMContext &Context, Function &F) {
BasicBlock *OrigBB = ConvertedInst->getParent();
BasicBlock *NewBB =
ConvertedInst->getParent()->splitBasicBlock(ConvertedInst, "split");
@ -226,7 +226,7 @@ void convertMemSetToLoop(Instruction *ConvertedInst, Value *DstAddr,
LoopBuilder.CreateStore(
SetValue,
LoopBuilder.CreateInBoundsGEP(SetValue->getType(), DstAddr, LoopIndex),
false);
IsVolatile);
Value *NewIndex =
LoopBuilder.CreateAdd(LoopIndex, ConstantInt::get(CopyLen->getType(), 1));
@ -325,6 +325,7 @@ bool NVPTXLowerAggrCopies::runOnFunction(Function &F) {
/* DstAddr */ Memset->getRawDest(),
/* CopyLen */ Memset->getLength(),
/* SetValue */ Memset->getValue(),
/* IsVolatile */ Memset->isVolatile(),
/* Context */ Context,
/* Function F */ F);
}

13
test/CodeGen/NVPTX/lower-aggr-copies.ll
View File

@ -86,6 +86,19 @@ entry:
; PTX-NEXT: @%p[[PRED]] bra LBB[[LABEL]]
}
define i8* @volatile_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 true)
ret i8* %dst
; IR-LABEL: @volatile_memset_caller
; IR: [[VAL:%[0-9]+]] = trunc i32 %c to i8
; IR: loadstoreloop:
; IR: [[STOREPTR:%[0-9]+]] = getelementptr inbounds i8, i8* %dst, i64
; IR-NEXT: store volatile i8 [[VAL]], i8* [[STOREPTR]]
}
define i8* @memmove_caller(i8* %dst, i8* %src, i64 %n) #0 {
entry:
tail call void @llvm.memmove.p0i8.p0i8.i64(i8* %dst, i8* %src, i64 %n, i32 1, i1 false)