RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2025-01-19 10:13:13 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

[GlobalISel] Translate memset/memmove/memcpy from undef ptrs into nops

Browse Source 
If the source is undef, then just don't do anything.

This matches SelectionDAG's behaviour in SelectionDAG.cpp.

Also add a test showing that we do the right thing here.
(irtranslator-memfunc-undef.ll)

Differential Revision: https://reviews.llvm.org/D63095

llvm-svn: 362989
This commit is contained in:
Jessica Paquette 2019-06-10 21:53:56 +00:00
parent 5b49a8e579
commit b79dfdcfff
2 changed files with 54 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
lib/CodeGen/GlobalISel/IRTranslator.cpp
View File

@ -728,6 +728,19 @@ bool IRTranslator::translateGetElementPtr(const User &U,
bool IRTranslator::translateMemfunc(const CallInst &CI,
MachineIRBuilder &MIRBuilder,
unsigned ID) {
// If the source is undef, then just emit a nop.
if (isa<UndefValue>(CI.getArgOperand(1))) {
switch (ID) {
case Intrinsic::memmove:
case Intrinsic::memcpy:
case Intrinsic::memset:
return true;
default:
break;
}
}
LLT SizeTy = getLLTForType(*CI.getArgOperand(2)->getType(), *DL);
Type *DstTy = CI.getArgOperand(0)->getType();
if (cast<PointerType>(DstTy)->getAddressSpace() != 0 ||

41
test/CodeGen/AArch64/GlobalISel/irtranslator-memfunc-undef.ll Normal file
View File

@ -0,0 +1,41 @@
; RUN: llc -mtriple=aarch64-unknown-unknown -global-isel -verify-machineinstrs -stop-after=irtranslator %s -o - | FileCheck %s
define void @memset() {
; CHECK-LABEL: name: memset
; CHECK: bb.1.entry:
; CHECK-NEXT: [[FRAME_INDEX:%[0-9]+]]:_(p0) = G_FRAME_INDEX %stack.0.buf
; CHECK-NEXT: RET_ReallyLR
entry:
%buf = alloca [512 x i8], align 1
%ptr = getelementptr inbounds [512 x i8], [512 x i8]* %buf, i32 0, i32 0
call void @llvm.memset.p0i8.i32(i8* %ptr, i8 undef, i32 512, i1 false)
ret void
}
define void @memcpy() {
; CHECK-LABEL: name: memcpy
; CHECK: bb.1.entry:
; CHECK-NEXT: [[FRAME_INDEX:%[0-9]+]]:_(p0) = G_FRAME_INDEX %stack.0.buf
; CHECK-NEXT: RET_ReallyLR
entry:
%buf = alloca [512 x i8], align 1
%ptr = getelementptr inbounds [512 x i8], [512 x i8]* %buf, i32 0, i32 0
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %ptr, i8* undef, i32 512, i1 false)
ret void
}
define void @memmove() {
; CHECK-LABEL: name: memmove
; CHECK: bb.1.entry:
; CHECK-NEXT: [[FRAME_INDEX:%[0-9]+]]:_(p0) = G_FRAME_INDEX %stack.0.buf
; CHECK-NEXT: RET_ReallyLR
entry:
%buf = alloca [512 x i8], align 1
%ptr = getelementptr inbounds [512 x i8], [512 x i8]* %buf, i32 0, i32 0
call void @llvm.memmove.p0i8.p0i8.i32(i8* %ptr, i8* undef, i32 512, i1 false)
ret void
}
declare void @llvm.memset.p0i8.i32(i8* nocapture, i8, i32, i1) nounwind
declare void @llvm.memcpy.p0i8.p0i8.i32(i8* nocapture writeonly, i8* nocapture readonly, i32, i1) nounwind
declare void @llvm.memmove.p0i8.p0i8.i32(i8* nocapture, i8* nocapture, i32, i1) nounwind