RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-25 12:12:47 +01:00
Code Issues Projects Releases Wiki Activity

ARM: tweak WoA frame lowering

Browse Source 
Accept r11 when targeting Windows on ARM rather than just low registers.
Because we are in a thumb-2 only mode, this may be slightly more expensive in
code size, but results in better code for the environment since it spills the
frame register, which is generally desired for fast stack walking as per the
ABI.

llvm-svn: 249804
This commit is contained in:
Saleem Abdulrasool 2015-10-09 03:19:03 +00:00
parent b0ed8472d2
commit 3accf5501c
2 changed files with 30 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
lib/Target/ARM/ARMFrameLowering.cpp
View File

@ -1605,13 +1605,11 @@ void ARMFrameLowering::determineCalleeSaves(MachineFunction &MF,
// FIXME: We could add logic to be more precise about negative offsets // FIXME: We could add logic to be more precise about negative offsets
// and which instructions will need a scratch register for them. Is it // and which instructions will need a scratch register for them. Is it
// worth the effort and added fragility? // worth the effort and added fragility?
bool BigStack = bool BigStack = (RS && (MFI->estimateStackSize(MF) +
(RS && ((hasFP(MF) && AFI->hasStackFrame()) ? 4 : 0) >=
(MFI->estimateStackSize(MF) + estimateRSStackSizeLimit(MF, this))) ||
((hasFP(MF) && AFI->hasStackFrame()) ? 4:0) >= MFI->hasVarSizedObjects() ||
estimateRSStackSizeLimit(MF, this))) (MFI->adjustsStack() && !canSimplifyCallFramePseudos(MF));
|| MFI->hasVarSizedObjects()
|| (MFI->adjustsStack() && !canSimplifyCallFramePseudos(MF));
bool ExtraCSSpill = false; bool ExtraCSSpill = false;
if (BigStack || !CanEliminateFrame || RegInfo->cannotEliminateFrame(MF)) { if (BigStack || !CanEliminateFrame || RegInfo->cannotEliminateFrame(MF)) {
@ -1649,8 +1647,10 @@ void ARMFrameLowering::determineCalleeSaves(MachineFunction &MF,
if (CS1Spilled && !UnspilledCS1GPRs.empty()) { if (CS1Spilled && !UnspilledCS1GPRs.empty()) {
for (unsigned i = 0, e = UnspilledCS1GPRs.size(); i != e; ++i) { for (unsigned i = 0, e = UnspilledCS1GPRs.size(); i != e; ++i) {
unsigned Reg = UnspilledCS1GPRs[i]; unsigned Reg = UnspilledCS1GPRs[i];
// Don't spill high register if the function is thumb // Don't spill high register if the function is thumb. In the case of
// Windows on ARM, accept R11 (frame pointer)
if (!AFI->isThumbFunction() || if (!AFI->isThumbFunction() ||
(STI.isTargetWindows() && Reg == ARM::R11) ||
isARMLowRegister(Reg) || Reg == ARM::LR) { isARMLowRegister(Reg) || Reg == ARM::LR) {
SavedRegs.set(Reg); SavedRegs.set(Reg);
if (!MRI.isReserved(Reg)) if (!MRI.isReserved(Reg))

22
test/CodeGen/ARM/Windows/no-frame-register.ll Normal file
View File

@ -0,0 +1,22 @@
; RUN: llc -mtriple thumbv7-windows -filetype asm -o - %s | FileCheck %s
declare void @callee(i32)
define i32 @calleer(i32 %i) {
entry:
%i.addr = alloca i32, align 4
%j = alloca i32, align 4
store i32 %i, i32* %i.addr, align 4
%0 = load i32, i32* %i.addr, align 4
%add = add nsw i32 %0, 1
store i32 %add, i32* %j, align 4
%1 = load i32, i32* %j, align 4
call void @callee(i32 %1)
%2 = load i32, i32* %j, align 4
%add1 = add nsw i32 %2, 1
ret i32 %add1
}
; CHECK-NOT: push.w {r7, lr}
; CHECK: push.w {r11, lr}