mirror of
https://github.com/RPCSX/llvm.git
synced 2026-01-31 01:05:23 +01:00
I did this a long time ago with a janky python script, but now clang-format has built-in support for this. I fed clang-format every line with a #include and let it re-sort things according to the precise LLVM rules for include ordering baked into clang-format these days. I've reverted a number of files where the results of sorting includes isn't healthy. Either places where we have legacy code relying on particular include ordering (where possible, I'll fix these separately) or where we have particular formatting around #include lines that I didn't want to disturb in this patch. This patch is *entirely* mechanical. If you get merge conflicts or anything, just ignore the changes in this patch and run clang-format over your #include lines in the files. Sorry for any noise here, but it is important to keep these things stable. I was seeing an increasing number of patches with irrelevant re-ordering of #include lines because clang-format was used. This patch at least isolates that churn, makes it easy to skip when resolving conflicts, and gets us to a clean baseline (again). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@304787 91177308-0d34-0410-b5e6-96231b3b80d8
176 lines
5.8 KiB
C++
176 lines
5.8 KiB
C++
//===---------- PPCTLSDynamicCall.cpp - TLS Dynamic Call Fixup ------------===//
|
|
//
|
|
// The LLVM Compiler Infrastructure
|
|
//
|
|
// This file is distributed under the University of Illinois Open Source
|
|
// License. See LICENSE.TXT for details.
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
//
|
|
// This pass expands ADDItls{ld,gd}LADDR[32] machine instructions into
|
|
// separate ADDItls[gd]L[32] and GETtlsADDR[32] instructions, both of
|
|
// which define GPR3. A copy is added from GPR3 to the target virtual
|
|
// register of the original instruction. The GETtlsADDR[32] is really
|
|
// a call instruction, so its target register is constrained to be GPR3.
|
|
// This is not true of ADDItls[gd]L[32], but there is a legacy linker
|
|
// optimization bug that requires the target register of the addi of
|
|
// a local- or general-dynamic TLS access sequence to be GPR3.
|
|
//
|
|
// This is done in a late pass so that TLS variable accesses can be
|
|
// fully commoned by MachineCSE.
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
#include "PPC.h"
|
|
#include "PPCInstrBuilder.h"
|
|
#include "PPCInstrInfo.h"
|
|
#include "PPCTargetMachine.h"
|
|
#include "llvm/CodeGen/LiveIntervalAnalysis.h"
|
|
#include "llvm/CodeGen/MachineFunctionPass.h"
|
|
#include "llvm/CodeGen/MachineInstrBuilder.h"
|
|
#include "llvm/Support/Debug.h"
|
|
#include "llvm/Support/raw_ostream.h"
|
|
|
|
using namespace llvm;
|
|
|
|
#define DEBUG_TYPE "ppc-tls-dynamic-call"
|
|
|
|
namespace llvm {
|
|
void initializePPCTLSDynamicCallPass(PassRegistry&);
|
|
}
|
|
|
|
namespace {
|
|
struct PPCTLSDynamicCall : public MachineFunctionPass {
|
|
static char ID;
|
|
PPCTLSDynamicCall() : MachineFunctionPass(ID) {
|
|
initializePPCTLSDynamicCallPass(*PassRegistry::getPassRegistry());
|
|
}
|
|
|
|
const PPCInstrInfo *TII;
|
|
LiveIntervals *LIS;
|
|
|
|
protected:
|
|
bool processBlock(MachineBasicBlock &MBB) {
|
|
bool Changed = false;
|
|
bool Is64Bit = MBB.getParent()->getSubtarget<PPCSubtarget>().isPPC64();
|
|
|
|
for (MachineBasicBlock::iterator I = MBB.begin(), IE = MBB.end();
|
|
I != IE;) {
|
|
MachineInstr &MI = *I;
|
|
|
|
if (MI.getOpcode() != PPC::ADDItlsgdLADDR &&
|
|
MI.getOpcode() != PPC::ADDItlsldLADDR &&
|
|
MI.getOpcode() != PPC::ADDItlsgdLADDR32 &&
|
|
MI.getOpcode() != PPC::ADDItlsldLADDR32) {
|
|
++I;
|
|
continue;
|
|
}
|
|
|
|
DEBUG(dbgs() << "TLS Dynamic Call Fixup:\n " << MI);
|
|
|
|
unsigned OutReg = MI.getOperand(0).getReg();
|
|
unsigned InReg = MI.getOperand(1).getReg();
|
|
DebugLoc DL = MI.getDebugLoc();
|
|
unsigned GPR3 = Is64Bit ? PPC::X3 : PPC::R3;
|
|
unsigned Opc1, Opc2;
|
|
const unsigned OrigRegs[] = {OutReg, InReg, GPR3};
|
|
|
|
switch (MI.getOpcode()) {
|
|
default:
|
|
llvm_unreachable("Opcode inconsistency error");
|
|
case PPC::ADDItlsgdLADDR:
|
|
Opc1 = PPC::ADDItlsgdL;
|
|
Opc2 = PPC::GETtlsADDR;
|
|
break;
|
|
case PPC::ADDItlsldLADDR:
|
|
Opc1 = PPC::ADDItlsldL;
|
|
Opc2 = PPC::GETtlsldADDR;
|
|
break;
|
|
case PPC::ADDItlsgdLADDR32:
|
|
Opc1 = PPC::ADDItlsgdL32;
|
|
Opc2 = PPC::GETtlsADDR32;
|
|
break;
|
|
case PPC::ADDItlsldLADDR32:
|
|
Opc1 = PPC::ADDItlsldL32;
|
|
Opc2 = PPC::GETtlsldADDR32;
|
|
break;
|
|
}
|
|
|
|
// Don't really need to save data to the stack - the clobbered
|
|
// registers are already saved when the SDNode (e.g. PPCaddiTlsgdLAddr)
|
|
// gets translated to the pseudo instruction (e.g. ADDItlsgdLADDR).
|
|
BuildMI(MBB, I, DL, TII->get(PPC::ADJCALLSTACKDOWN)).addImm(0)
|
|
.addImm(0);
|
|
|
|
// Expand into two ops built prior to the existing instruction.
|
|
MachineInstr *Addi = BuildMI(MBB, I, DL, TII->get(Opc1), GPR3)
|
|
.addReg(InReg);
|
|
Addi->addOperand(MI.getOperand(2));
|
|
|
|
// The ADDItls* instruction is the first instruction in the
|
|
// repair range.
|
|
MachineBasicBlock::iterator First = I;
|
|
--First;
|
|
|
|
MachineInstr *Call = (BuildMI(MBB, I, DL, TII->get(Opc2), GPR3)
|
|
.addReg(GPR3));
|
|
Call->addOperand(MI.getOperand(3));
|
|
|
|
BuildMI(MBB, I, DL, TII->get(PPC::ADJCALLSTACKUP)).addImm(0).addImm(0);
|
|
|
|
BuildMI(MBB, I, DL, TII->get(TargetOpcode::COPY), OutReg)
|
|
.addReg(GPR3);
|
|
|
|
// The COPY is the last instruction in the repair range.
|
|
MachineBasicBlock::iterator Last = I;
|
|
--Last;
|
|
|
|
// Move past the original instruction and remove it.
|
|
++I;
|
|
MI.removeFromParent();
|
|
|
|
// Repair the live intervals.
|
|
LIS->repairIntervalsInRange(&MBB, First, Last, OrigRegs);
|
|
Changed = true;
|
|
}
|
|
|
|
return Changed;
|
|
}
|
|
|
|
public:
|
|
bool runOnMachineFunction(MachineFunction &MF) override {
|
|
TII = MF.getSubtarget<PPCSubtarget>().getInstrInfo();
|
|
LIS = &getAnalysis<LiveIntervals>();
|
|
|
|
bool Changed = false;
|
|
|
|
for (MachineFunction::iterator I = MF.begin(); I != MF.end();) {
|
|
MachineBasicBlock &B = *I++;
|
|
if (processBlock(B))
|
|
Changed = true;
|
|
}
|
|
|
|
return Changed;
|
|
}
|
|
|
|
void getAnalysisUsage(AnalysisUsage &AU) const override {
|
|
AU.addRequired<LiveIntervals>();
|
|
AU.addPreserved<LiveIntervals>();
|
|
AU.addRequired<SlotIndexes>();
|
|
AU.addPreserved<SlotIndexes>();
|
|
MachineFunctionPass::getAnalysisUsage(AU);
|
|
}
|
|
};
|
|
}
|
|
|
|
INITIALIZE_PASS_BEGIN(PPCTLSDynamicCall, DEBUG_TYPE,
|
|
"PowerPC TLS Dynamic Call Fixup", false, false)
|
|
INITIALIZE_PASS_DEPENDENCY(LiveIntervals)
|
|
INITIALIZE_PASS_DEPENDENCY(SlotIndexes)
|
|
INITIALIZE_PASS_END(PPCTLSDynamicCall, DEBUG_TYPE,
|
|
"PowerPC TLS Dynamic Call Fixup", false, false)
|
|
|
|
char PPCTLSDynamicCall::ID = 0;
|
|
FunctionPass*
|
|
llvm::createPPCTLSDynamicCallPass() { return new PPCTLSDynamicCall(); }
|