Google Git Google Git
Sign in
swiftshader/SwiftShader/b8f1a3a/./third_party/llvm-16.0/llvm/lib/CodeGen/RegAllocGreedy.cpp
blob: b43a4d2a4b8550703972ff81e998fe2df14172be [file]
//===- RegAllocGreedy.cpp - greedy register allocator ---------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
//
// This file defines the RAGreedy function pass for register allocation in
// optimized builds.
//
//===----------------------------------------------------------------------===//
#include "RegAllocGreedy.h"
#include "AllocationOrder.h"
#include "InterferenceCache.h"
#include "LiveDebugVariables.h"
#include "RegAllocBase.h"
#include "RegAllocEvictionAdvisor.h"
#include "RegAllocPriorityAdvisor.h"
#include "SpillPlacement.h"
#include "SplitKit.h"
#include "llvm/ADT/ArrayRef.h"
#include "llvm/ADT/BitVector.h"
#include "llvm/ADT/IndexedMap.h"
#include "llvm/ADT/SmallPtrSet.h"
#include "llvm/ADT/SmallSet.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/Statistic.h"
#include "llvm/ADT/StringRef.h"
#include "llvm/Analysis/AliasAnalysis.h"
#include "llvm/Analysis/OptimizationRemarkEmitter.h"
#include "llvm/CodeGen/CalcSpillWeights.h"
#include "llvm/CodeGen/EdgeBundles.h"
#include "llvm/CodeGen/LiveInterval.h"
#include "llvm/CodeGen/LiveIntervalUnion.h"
#include "llvm/CodeGen/LiveIntervals.h"
#include "llvm/CodeGen/LiveRangeEdit.h"
#include "llvm/CodeGen/LiveRegMatrix.h"
#include "llvm/CodeGen/LiveStacks.h"
#include "llvm/CodeGen/MachineBasicBlock.h"
#include "llvm/CodeGen/MachineBlockFrequencyInfo.h"
#include "llvm/CodeGen/MachineDominators.h"
#include "llvm/CodeGen/MachineFrameInfo.h"
#include "llvm/CodeGen/MachineFunction.h"
#include "llvm/CodeGen/MachineFunctionPass.h"
#include "llvm/CodeGen/MachineInstr.h"
#include "llvm/CodeGen/MachineLoopInfo.h"
#include "llvm/CodeGen/MachineOperand.h"
#include "llvm/CodeGen/MachineOptimizationRemarkEmitter.h"
#include "llvm/CodeGen/MachineRegisterInfo.h"
#include "llvm/CodeGen/RegAllocRegistry.h"
#include "llvm/CodeGen/RegisterClassInfo.h"
#include "llvm/CodeGen/SlotIndexes.h"
#include "llvm/CodeGen/Spiller.h"
#include "llvm/CodeGen/TargetInstrInfo.h"
#include "llvm/CodeGen/TargetRegisterInfo.h"
#include "llvm/CodeGen/TargetSubtargetInfo.h"
#include "llvm/CodeGen/VirtRegMap.h"
#include "llvm/IR/DebugInfoMetadata.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/LLVMContext.h"
#include "llvm/InitializePasses.h"
#include "llvm/MC/MCRegisterInfo.h"
#include "llvm/Pass.h"
#include "llvm/Support/BlockFrequency.h"
#include "llvm/Support/BranchProbability.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/MathExtras.h"
#include "llvm/Support/Timer.h"
#include "llvm/Support/raw_ostream.h"
#include <algorithm>
#include <cassert>
#include <cstdint>
#include <utility>
using namespace llvm;
#define DEBUG_TYPE "regalloc"
STATISTIC(NumGlobalSplits, "Number of split global live ranges");
STATISTIC(NumLocalSplits, "Number of split local live ranges");
STATISTIC(NumEvicted, "Number of interferences evicted");
static cl::opt<SplitEditor::ComplementSpillMode> SplitSpillMode(
"split-spill-mode", cl::Hidden,
cl::desc("Spill mode for splitting live ranges"),
cl::values(clEnumValN(SplitEditor::SM_Partition, "default", "Default"),
clEnumValN(SplitEditor::SM_Size, "size", "Optimize for size"),
clEnumValN(SplitEditor::SM_Speed, "speed", "Optimize for speed")),
cl::init(SplitEditor::SM_Speed));
static cl::opt<unsigned>
LastChanceRecoloringMaxDepth("lcr-max-depth", cl::Hidden,
cl::desc("Last chance recoloring max depth"),
cl::init(5));
static cl::opt<unsigned> LastChanceRecoloringMaxInterference(
"lcr-max-interf", cl::Hidden,
cl::desc("Last chance recoloring maximum number of considered"
" interference at a time"),
cl::init(8));
static cl::opt<bool> ExhaustiveSearch(
"exhaustive-register-search", cl::NotHidden,
cl::desc("Exhaustive Search for registers bypassing the depth "
"and interference cutoffs of last chance recoloring"),
cl::Hidden);
static cl::opt<bool> EnableDeferredSpilling(
"enable-deferred-spilling", cl::Hidden,
cl::desc("Instead of spilling a variable right away, defer the actual "
"code insertion to the end of the allocation. That way the "
"allocator might still find a suitable coloring for this "
"variable because of other evicted variables."),
cl::init(false));
// FIXME: Find a good default for this flag and remove the flag.
static cl::opt<unsigned>
CSRFirstTimeCost("regalloc-csr-first-time-cost",
cl::desc("Cost for first time use of callee-saved register."),
cl::init(0), cl::Hidden);
static cl::opt<unsigned long> GrowRegionComplexityBudget(
"grow-region-complexity-budget",
cl::desc("growRegion() does not scale with the number of BB edges, so "
"limit its budget and bail out once we reach the limit."),
cl::init(10000), cl::Hidden);
static cl::opt<bool> GreedyRegClassPriorityTrumpsGlobalness(
"greedy-regclass-priority-trumps-globalness",
cl::desc("Change the greedy register allocator's live range priority "
"calculation to make the AllocationPriority of the register class "
"more important then whether the range is global"),
cl::Hidden);
static cl::opt<bool> GreedyReverseLocalAssignment(
"greedy-reverse-local-assignment",
cl::desc("Reverse allocation order of local live ranges, such that "
"shorter local live ranges will tend to be allocated first"),
cl::Hidden);
static RegisterRegAlloc greedyRegAlloc("greedy", "greedy register allocator",
createGreedyRegisterAllocator);
char RAGreedy::ID = 0;
char &llvm::RAGreedyID = RAGreedy::ID;
INITIALIZE_PASS_BEGIN(RAGreedy, "greedy",
"Greedy Register Allocator", false, false)
INITIALIZE_PASS_DEPENDENCY(LiveDebugVariables)
INITIALIZE_PASS_DEPENDENCY(SlotIndexes)
INITIALIZE_PASS_DEPENDENCY(LiveIntervals)
INITIALIZE_PASS_DEPENDENCY(RegisterCoalescer)
INITIALIZE_PASS_DEPENDENCY(MachineScheduler)
INITIALIZE_PASS_DEPENDENCY(LiveStacks)
INITIALIZE_PASS_DEPENDENCY(MachineDominatorTree)
INITIALIZE_PASS_DEPENDENCY(MachineLoopInfo)
INITIALIZE_PASS_DEPENDENCY(VirtRegMap)
INITIALIZE_PASS_DEPENDENCY(LiveRegMatrix)
INITIALIZE_PASS_DEPENDENCY(EdgeBundles)
INITIALIZE_PASS_DEPENDENCY(SpillPlacement)
INITIALIZE_PASS_DEPENDENCY(MachineOptimizationRemarkEmitterPass)
INITIALIZE_PASS_DEPENDENCY(RegAllocEvictionAdvisorAnalysis)
INITIALIZE_PASS_DEPENDENCY(RegAllocPriorityAdvisorAnalysis)
INITIALIZE_PASS_END(RAGreedy, "greedy",
"Greedy Register Allocator", false, false)
#ifndef NDEBUG
const char *const RAGreedy::StageName[] = {
"RS_New",
"RS_Assign",
"RS_Split",
"RS_Split2",
"RS_Spill",
"RS_Memory",
"RS_Done"
};
#endif
// Hysteresis to use when comparing floats.
// This helps stabilize decisions based on float comparisons.
const float Hysteresis = (2007 / 2048.0f); // 0.97998046875
FunctionPass* llvm::createGreedyRegisterAllocator() {
return new RAGreedy();
}
FunctionPass *llvm::createGreedyRegisterAllocator(RegClassFilterFunc Ftor) {
return new RAGreedy(Ftor);
}
RAGreedy::RAGreedy(RegClassFilterFunc F):
MachineFunctionPass(ID),
RegAllocBase(F) {
}
void RAGreedy::getAnalysisUsage(AnalysisUsage &AU) const {
AU.setPreservesCFG();
AU.addRequired<MachineBlockFrequencyInfo>();
AU.addPreserved<MachineBlockFrequencyInfo>();
AU.addRequired<LiveIntervals>();
AU.addPreserved<LiveIntervals>();
AU.addRequired<SlotIndexes>();
AU.addPreserved<SlotIndexes>();
AU.addRequired<LiveDebugVariables>();
AU.addPreserved<LiveDebugVariables>();
AU.addRequired<LiveStacks>();
AU.addPreserved<LiveStacks>();
AU.addRequired<MachineDominatorTree>();
AU.addPreserved<MachineDominatorTree>();
AU.addRequired<MachineLoopInfo>();
AU.addPreserved<MachineLoopInfo>();
AU.addRequired<VirtRegMap>();
AU.addPreserved<VirtRegMap>();
AU.addRequired<LiveRegMatrix>();
AU.addPreserved<LiveRegMatrix>();
AU.addRequired<EdgeBundles>();
AU.addRequired<SpillPlacement>();
AU.addRequired<MachineOptimizationRemarkEmitterPass>();
AU.addRequired<RegAllocEvictionAdvisorAnalysis>();
AU.addRequired<RegAllocPriorityAdvisorAnalysis>();
MachineFunctionPass::getAnalysisUsage(AU);
}
//===----------------------------------------------------------------------===//
// LiveRangeEdit delegate methods
//===----------------------------------------------------------------------===//
bool RAGreedy::LRE_CanEraseVirtReg(Register VirtReg) {
LiveInterval &LI = LIS->getInterval(VirtReg);
if (VRM->hasPhys(VirtReg)) {
Matrix->unassign(LI);
aboutToRemoveInterval(LI);
return true;
}
// Unassigned virtreg is probably in the priority queue.
// RegAllocBase will erase it after dequeueing.
// Nonetheless, clear the live-range so that the debug
// dump will show the right state for that VirtReg.
LI.clear();
return false;
}
void RAGreedy::LRE_WillShrinkVirtReg(Register VirtReg) {
if (!VRM->hasPhys(VirtReg))
return;
// Register is assigned, put it back on the queue for reassignment.
LiveInterval &LI = LIS->getInterval(VirtReg);
Matrix->unassign(LI);
RegAllocBase::enqueue(&LI);
}
void RAGreedy::LRE_DidCloneVirtReg(Register New, Register Old) {
ExtraInfo->LRE_DidCloneVirtReg(New, Old);
}
void RAGreedy::ExtraRegInfo::LRE_DidCloneVirtReg(Register New, Register Old) {
// Cloning a register we haven't even heard about yet? Just ignore it.
if (!Info.inBounds(Old))
return;
// LRE may clone a virtual register because dead code elimination causes it to
// be split into connected components. The new components are much smaller
// than the original, so they should get a new chance at being assigned.
// same stage as the parent.
Info[Old].Stage = RS_Assign;
Info.grow(New.id());
Info[New] = Info[Old];
}
void RAGreedy::releaseMemory() {
SpillerInstance.reset();
GlobalCand.clear();
}
void RAGreedy::enqueueImpl(const LiveInterval *LI) { enqueue(Queue, LI); }
void RAGreedy::enqueue(PQueue &CurQueue, const LiveInterval *LI) {
// Prioritize live ranges by size, assigning larger ranges first.
// The queue holds (size, reg) pairs.
const Register Reg = LI->reg();
assert(Reg.isVirtual() && "Can only enqueue virtual registers");
auto Stage = ExtraInfo->getOrInitStage(Reg);
if (Stage == RS_New) {
Stage = RS_Assign;
ExtraInfo->setStage(Reg, Stage);
}
unsigned Ret = PriorityAdvisor->getPriority(*LI);
// The virtual register number is a tie breaker for same-sized ranges.
// Give lower vreg numbers higher priority to assign them first.
CurQueue.push(std::make_pair(Ret, ~Reg));
}
unsigned DefaultPriorityAdvisor::getPriority(const LiveInterval &LI) const {
const unsigned Size = LI.getSize();
const Register Reg = LI.reg();
unsigned Prio;
LiveRangeStage Stage = RA.getExtraInfo().getStage(LI);
if (Stage == RS_Split) {
// Unsplit ranges that couldn't be allocated immediately are deferred until
// everything else has been allocated.
Prio = Size;
} else if (Stage == RS_Memory) {
// Memory operand should be considered last.
// Change the priority such that Memory operand are assigned in
// the reverse order that they came in.
// TODO: Make this a member variable and probably do something about hints.
static unsigned MemOp = 0;
Prio = MemOp++;
} else {
// Giant live ranges fall back to the global assignment heuristic, which
// prevents excessive spilling in pathological cases.
const TargetRegisterClass &RC = *MRI->getRegClass(Reg);
bool ForceGlobal = RC.GlobalPriority ||
(!ReverseLocalAssignment &&
(Size / SlotIndex::InstrDist) >
(2 * RegClassInfo.getNumAllocatableRegs(&RC)));
unsigned GlobalBit = 0;
if (Stage == RS_Assign && !ForceGlobal && !LI.empty() &&
LIS->intervalIsInOneMBB(LI)) {
// Allocate original local ranges in linear instruction order. Since they
// are singly defined, this produces optimal coloring in the absence of
// global interference and other constraints.
if (!ReverseLocalAssignment)
Prio = LI.beginIndex().getApproxInstrDistance(Indexes->getLastIndex());
else {
// Allocating bottom up may allow many short LRGs to be assigned first
// to one of the cheap registers. This could be much faster for very
// large blocks on targets with many physical registers.
Prio = Indexes->getZeroIndex().getApproxInstrDistance(LI.endIndex());
}
} else {
// Allocate global and split ranges in long->short order. Long ranges that
// don't fit should be spilled (or split) ASAP so they don't create
// interference. Mark a bit to prioritize global above local ranges.
Prio = Size;
GlobalBit = 1;
}
// Priority bit layout:
// 31 RS_Assign priority
// 30 Preference priority
// if (RegClassPriorityTrumpsGlobalness)
// 29-25 AllocPriority
// 24 GlobalBit
// else
// 29 Global bit
// 28-24 AllocPriority
// 0-23 Size/Instr distance
// Clamp the size to fit with the priority masking scheme
Prio = std::min(Prio, (unsigned)maxUIntN(24));
assert(isUInt<5>(RC.AllocationPriority) && "allocation priority overflow");
if (RegClassPriorityTrumpsGlobalness)
Prio |= RC.AllocationPriority << 25 | GlobalBit << 24;
else
Prio |= GlobalBit << 29 | RC.AllocationPriority << 24;
// Mark a higher bit to prioritize global and local above RS_Split.
Prio |= (1u << 31);
// Boost ranges that have a physical register hint.
if (VRM->hasKnownPreference(Reg))
Prio |= (1u << 30);
}
return Prio;
}
const LiveInterval *RAGreedy::dequeue() { return dequeue(Queue); }
const LiveInterval *RAGreedy::dequeue(PQueue &CurQueue) {
if (CurQueue.empty())
return nullptr;
LiveInterval *LI = &LIS->getInterval(~CurQueue.top().second);
CurQueue.pop();
return LI;
}
//===----------------------------------------------------------------------===//
// Direct Assignment
//===----------------------------------------------------------------------===//
/// tryAssign - Try to assign VirtReg to an available register.
MCRegister RAGreedy::tryAssign(const LiveInterval &VirtReg,
AllocationOrder &Order,
SmallVectorImpl<Register> &NewVRegs,
const SmallVirtRegSet &FixedRegisters) {
MCRegister PhysReg;
for (auto I = Order.begin(), E = Order.end(); I != E && !PhysReg; ++I) {
assert(*I);
if (!Matrix->checkInterference(VirtReg, *I)) {
if (I.isHint())
return *I;
else
PhysReg = *I;
}
}
if (!PhysReg.isValid())
return PhysReg;
// PhysReg is available, but there may be a better choice.
// If we missed a simple hint, try to cheaply evict interference from the
// preferred register.
if (Register Hint = MRI->getSimpleHint(VirtReg.reg()))
if (Order.isHint(Hint)) {
MCRegister PhysHint = Hint.asMCReg();
LLVM_DEBUG(dbgs() << "missed hint " << printReg(PhysHint, TRI) << '\n');
if (EvictAdvisor->canEvictHintInterference(VirtReg, PhysHint,
FixedRegisters)) {
evictInterference(VirtReg, PhysHint, NewVRegs);
return PhysHint;
}
// Record the missed hint, we may be able to recover
// at the end if the surrounding allocation changed.
SetOfBrokenHints.insert(&VirtReg);
}
// Try to evict interference from a cheaper alternative.
uint8_t Cost = RegCosts[PhysReg];
// Most registers have 0 additional cost.
if (!Cost)
return PhysReg;
LLVM_DEBUG(dbgs() << printReg(PhysReg, TRI) << " is available at cost "
<< (unsigned)Cost << '\n');
MCRegister CheapReg = tryEvict(VirtReg, Order, NewVRegs, Cost, FixedRegisters);
return CheapReg ? CheapReg : PhysReg;
}
//===----------------------------------------------------------------------===//
// Interference eviction
//===----------------------------------------------------------------------===//
Register RegAllocEvictionAdvisor::canReassign(const LiveInterval &VirtReg,
Register PrevReg) const {
auto Order =
AllocationOrder::create(VirtReg.reg(), *VRM, RegClassInfo, Matrix);
MCRegister PhysReg;
for (auto I = Order.begin(), E = Order.end(); I != E && !PhysReg; ++I) {
if ((*I).id() == PrevReg.id())
continue;
MCRegUnitIterator Units(*I, TRI);
for (; Units.isValid(); ++Units) {
// Instantiate a "subquery", not to be confused with the Queries array.
LiveIntervalUnion::Query subQ(VirtReg, Matrix->getLiveUnions()[*Units]);
if (subQ.checkInterference())
break;
}
// If no units have interference, break out with the current PhysReg.
if (!Units.isValid())
PhysReg = *I;
}
if (PhysReg)
LLVM_DEBUG(dbgs() << "can reassign: " << VirtReg << " from "
<< printReg(PrevReg, TRI) << " to "
<< printReg(PhysReg, TRI) << '\n');
return PhysReg;
}
/// evictInterference - Evict any interferring registers that prevent VirtReg
/// from being assigned to Physreg. This assumes that canEvictInterference
/// returned true.
void RAGreedy::evictInterference(const LiveInterval &VirtReg,
MCRegister PhysReg,
SmallVectorImpl<Register> &NewVRegs) {
// Make sure that VirtReg has a cascade number, and assign that cascade
// number to every evicted register. These live ranges than then only be
// evicted by a newer cascade, preventing infinite loops.
unsigned Cascade = ExtraInfo->getOrAssignNewCascade(VirtReg.reg());
LLVM_DEBUG(dbgs() << "evicting " << printReg(PhysReg, TRI)
<< " interference: Cascade " << Cascade << '\n');
// Collect all interfering virtregs first.
SmallVector<const LiveInterval *, 8> Intfs;
for (MCRegUnitIterator Units(PhysReg, TRI); Units.isValid(); ++Units) {
LiveIntervalUnion::Query &Q = Matrix->query(VirtReg, *Units);
// We usually have the interfering VRegs cached so collectInterferingVRegs()
// should be fast, we may need to recalculate if when different physregs
// overlap the same register unit so we had different SubRanges queried
// against it.
ArrayRef<const LiveInterval *> IVR = Q.interferingVRegs();
Intfs.append(IVR.begin(), IVR.end());
}
// Evict them second. This will invalidate the queries.
for (const LiveInterval *Intf : Intfs) {
// The same VirtReg may be present in multiple RegUnits. Skip duplicates.
if (!VRM->hasPhys(Intf->reg()))
continue;
Matrix->unassign(*Intf);
assert((ExtraInfo->getCascade(Intf->reg()) < Cascade ||
VirtReg.isSpillable() < Intf->isSpillable()) &&
"Cannot decrease cascade number, illegal eviction");
ExtraInfo->setCascade(Intf->reg(), Cascade);
++NumEvicted;
NewVRegs.push_back(Intf->reg());
}
}
/// Returns true if the given \p PhysReg is a callee saved register and has not
/// been used for allocation yet.
bool RegAllocEvictionAdvisor::isUnusedCalleeSavedReg(MCRegister PhysReg) const {
MCRegister CSR = RegClassInfo.getLastCalleeSavedAlias(PhysReg);
if (!CSR)
return false;
return !Matrix->isPhysRegUsed(PhysReg);
}
std::optional<unsigned>
RegAllocEvictionAdvisor::getOrderLimit(const LiveInterval &VirtReg,
const AllocationOrder &Order,
unsigned CostPerUseLimit) const {
unsigned OrderLimit = Order.getOrder().size();
if (CostPerUseLimit < uint8_t(~0u)) {
// Check of any registers in RC are below CostPerUseLimit.
const TargetRegisterClass *RC = MRI->getRegClass(VirtReg.reg());
uint8_t MinCost = RegClassInfo.getMinCost(RC);
if (MinCost >= CostPerUseLimit) {
LLVM_DEBUG(dbgs() << TRI->getRegClassName(RC) << " minimum cost = "
<< MinCost << ", no cheaper registers to be found.\n");
return std::nullopt;
}
// It is normal for register classes to have a long tail of registers with
// the same cost. We don't need to look at them if they're too expensive.
if (RegCosts[Order.getOrder().back()] >= CostPerUseLimit) {
OrderLimit = RegClassInfo.getLastCostChange(RC);
LLVM_DEBUG(dbgs() << "Only trying the first " << OrderLimit
<< " regs.\n");
}
}
return OrderLimit;
}
bool RegAllocEvictionAdvisor::canAllocatePhysReg(unsigned CostPerUseLimit,
MCRegister PhysReg) const {
if (RegCosts[PhysReg] >= CostPerUseLimit)
return false;
// The first use of a callee-saved register in a function has cost 1.
// Don't start using a CSR when the CostPerUseLimit is low.
if (CostPerUseLimit == 1 && isUnusedCalleeSavedReg(PhysReg)) {
LLVM_DEBUG(
dbgs() << printReg(PhysReg, TRI) << " would clobber CSR "
<< printReg(RegClassInfo.getLastCalleeSavedAlias(PhysReg), TRI)
<< '\n');
return false;
}
return true;
}
/// tryEvict - Try to evict all interferences for a physreg.
/// @param VirtReg Currently unassigned virtual register.
/// @param Order Physregs to try.
/// @return Physreg to assign VirtReg, or 0.
MCRegister RAGreedy::tryEvict(const LiveInterval &VirtReg,
AllocationOrder &Order,
SmallVectorImpl<Register> &NewVRegs,
uint8_t CostPerUseLimit,
const SmallVirtRegSet &FixedRegisters) {
NamedRegionTimer T("evict", "Evict", TimerGroupName, TimerGroupDescription,
TimePassesIsEnabled);
MCRegister BestPhys = EvictAdvisor->tryFindEvictionCandidate(
VirtReg, Order, CostPerUseLimit, FixedRegisters);
if (BestPhys.isValid())
evictInterference(VirtReg, BestPhys, NewVRegs);
return BestPhys;
}
//===----------------------------------------------------------------------===//
// Region Splitting
//===----------------------------------------------------------------------===//
/// addSplitConstraints - Fill out the SplitConstraints vector based on the
/// interference pattern in Physreg and its aliases. Add the constraints to
/// SpillPlacement and return the static cost of this split in Cost, assuming
/// that all preferences in SplitConstraints are met.
/// Return false if there are no bundles with positive bias.
bool RAGreedy::addSplitConstraints(InterferenceCache::Cursor Intf,
BlockFrequency &Cost) {
ArrayRef<SplitAnalysis::BlockInfo> UseBlocks = SA->getUseBlocks();
// Reset interference dependent info.
SplitConstraints.resize(UseBlocks.size());
BlockFrequency StaticCost = 0;
for (unsigned I = 0; I != UseBlocks.size(); ++I) {
const SplitAnalysis::BlockInfo &BI = UseBlocks[I];
SpillPlacement::BlockConstraint &BC = SplitConstraints[I];
BC.Number = BI.MBB->getNumber();
Intf.moveToBlock(BC.Number);
BC.Entry = BI.LiveIn ? SpillPlacement::PrefReg : SpillPlacement::DontCare;
BC.Exit = (BI.LiveOut &&
!LIS->getInstructionFromIndex(BI.LastInstr)->isImplicitDef())
? SpillPlacement::PrefReg
: SpillPlacement::DontCare;
BC.ChangesValue = BI.FirstDef.isValid();
if (!Intf.hasInterference())
continue;
// Number of spill code instructions to insert.
unsigned Ins = 0;
// Interference for the live-in value.
if (BI.LiveIn) {
if (Intf.first() <= Indexes->getMBBStartIdx(BC.Number)) {
BC.Entry = SpillPlacement::MustSpill;
++Ins;
} else if (Intf.first() < BI.FirstInstr) {
BC.Entry = SpillPlacement::PrefSpill;
++Ins;
} else if (Intf.first() < BI.LastInstr) {
++Ins;
}
// Abort if the spill cannot be inserted at the MBB' start
if (((BC.Entry == SpillPlacement::MustSpill) ||
(BC.Entry == SpillPlacement::PrefSpill)) &&
SlotIndex::isEarlierInstr(BI.FirstInstr,
SA->getFirstSplitPoint(BC.Number)))
return false;
}
// Interference for the live-out value.
if (BI.LiveOut) {
if (Intf.last() >= SA->getLastSplitPoint(BC.Number)) {
BC.Exit = SpillPlacement::MustSpill;
++Ins;
} else if (Intf.last() > BI.LastInstr) {
BC.Exit = SpillPlacement::PrefSpill;
++Ins;
} else if (Intf.last() > BI.FirstInstr) {
++Ins;
}
}
// Accumulate the total frequency of inserted spill code.
while (Ins--)
StaticCost += SpillPlacer->getBlockFrequency(BC.Number);
}
Cost = StaticCost;
// Add constraints for use-blocks. Note that these are the only constraints
// that may add a positive bias, it is downhill from here.
SpillPlacer->addConstraints(SplitConstraints);
return SpillPlacer->scanActiveBundles();
}
/// addThroughConstraints - Add constraints and links to SpillPlacer from the
/// live-through blocks in Blocks.
bool RAGreedy::addThroughConstraints(InterferenceCache::Cursor Intf,
ArrayRef<unsigned> Blocks) {
const unsigned GroupSize = 8;
SpillPlacement::BlockConstraint BCS[GroupSize];
unsigned TBS[GroupSize];
unsigned B = 0, T = 0;
for (unsigned Number : Blocks) {
Intf.moveToBlock(Number);
if (!Intf.hasInterference()) {
assert(T < GroupSize && "Array overflow");
TBS[T] = Number;
if (++T == GroupSize) {
SpillPlacer->addLinks(ArrayRef(TBS, T));
T = 0;
}
continue;
}
assert(B < GroupSize && "Array overflow");
BCS[B].Number = Number;
// Abort if the spill cannot be inserted at the MBB' start
MachineBasicBlock *MBB = MF->getBlockNumbered(Number);
auto FirstNonDebugInstr = MBB->getFirstNonDebugInstr();
if (FirstNonDebugInstr != MBB->end() &&
SlotIndex::isEarlierInstr(LIS->getInstructionIndex(*FirstNonDebugInstr),
SA->getFirstSplitPoint(Number)))
return false;
// Interference for the live-in value.
if (Intf.first() <= Indexes->getMBBStartIdx(Number))
BCS[B].Entry = SpillPlacement::MustSpill;
else
BCS[B].Entry = SpillPlacement::PrefSpill;
// Interference for the live-out value.
if (Intf.last() >= SA->getLastSplitPoint(Number))
BCS[B].Exit = SpillPlacement::MustSpill;
else
BCS[B].Exit = SpillPlacement::PrefSpill;
if (++B == GroupSize) {
SpillPlacer->addConstraints(ArrayRef(BCS, B));
B = 0;
}
}
SpillPlacer->addConstraints(ArrayRef(BCS, B));
SpillPlacer->addLinks(ArrayRef(TBS, T));
return true;
}
bool RAGreedy::growRegion(GlobalSplitCandidate &Cand) {
// Keep track of through blocks that have not been added to SpillPlacer.
BitVector Todo = SA->getThroughBlocks();
SmallVectorImpl<unsigned> &ActiveBlocks = Cand.ActiveBlocks;
unsigned AddedTo = 0;
#ifndef NDEBUG
unsigned Visited = 0;
#endif
unsigned long Budget = GrowRegionComplexityBudget;
while (true) {
ArrayRef<unsigned> NewBundles = SpillPlacer->getRecentPositive();
// Find new through blocks in the periphery of PrefRegBundles.
for (unsigned Bundle : NewBundles) {
// Look at all blocks connected to Bundle in the full graph.
ArrayRef<unsigned> Blocks = Bundles->getBlocks(Bundle);
// Limit compilation time by bailing out after we use all our budget.
if (Blocks.size() >= Budget)
return false;
Budget -= Blocks.size();
for (unsigned Block : Blocks) {
if (!Todo.test(Block))
continue;
Todo.reset(Block);
// This is a new through block. Add it to SpillPlacer later.
ActiveBlocks.push_back(Block);
#ifndef NDEBUG
++Visited;
#endif
}
}
// Any new blocks to add?
if (ActiveBlocks.size() == AddedTo)
break;
// Compute through constraints from the interference, or assume that all
// through blocks prefer spilling when forming compact regions.
auto NewBlocks = ArrayRef(ActiveBlocks).slice(AddedTo);
if (Cand.PhysReg) {
if (!addThroughConstraints(Cand.Intf, NewBlocks))
return false;
} else
// Provide a strong negative bias on through blocks to prevent unwanted
// liveness on loop backedges.
SpillPlacer->addPrefSpill(NewBlocks, /* Strong= */ true);
AddedTo = ActiveBlocks.size();
// Perhaps iterating can enable more bundles?
SpillPlacer->iterate();
}
LLVM_DEBUG(dbgs() << ", v=" << Visited);
return true;
}
/// calcCompactRegion - Compute the set of edge bundles that should be live
/// when splitting the current live range into compact regions. Compact
/// regions can be computed without looking at interference. They are the
/// regions formed by removing all the live-through blocks from the live range.
///
/// Returns false if the current live range is already compact, or if the
/// compact regions would form single block regions anyway.
bool RAGreedy::calcCompactRegion(GlobalSplitCandidate &Cand) {
// Without any through blocks, the live range is already compact.
if (!SA->getNumThroughBlocks())
return false;
// Compact regions don't correspond to any physreg.
Cand.reset(IntfCache, MCRegister::NoRegister);
LLVM_DEBUG(dbgs() << "Compact region bundles");
// Use the spill placer to determine the live bundles. GrowRegion pretends
// that all the through blocks have interference when PhysReg is unset.
SpillPlacer->prepare(Cand.LiveBundles);
// The static split cost will be zero since Cand.Intf reports no interference.
BlockFrequency Cost;
if (!addSplitConstraints(Cand.Intf, Cost)) {
LLVM_DEBUG(dbgs() << ", none.\n");
return false;
}
if (!growRegion(Cand)) {
LLVM_DEBUG(dbgs() << ", cannot spill all interferences.\n");
return false;
}
SpillPlacer->finish();
if (!Cand.LiveBundles.any()) {
LLVM_DEBUG(dbgs() << ", none.\n");
return false;
}
LLVM_DEBUG({
for (int I : Cand.LiveBundles.set_bits())
dbgs() << " EB#" << I;
dbgs() << ".\n";
});
return true;
}
/// calcSpillCost - Compute how expensive it would be to split the live range in
/// SA around all use blocks instead of forming bundle regions.
BlockFrequency RAGreedy::calcSpillCost() {
BlockFrequency Cost = 0;
ArrayRef<SplitAnalysis::BlockInfo> UseBlocks = SA->getUseBlocks();
for (const SplitAnalysis::BlockInfo &BI : UseBlocks) {
unsigned Number = BI.MBB->getNumber();
// We normally only need one spill instruction - a load or a store.
Cost += SpillPlacer->getBlockFrequency(Number);
// Unless the value is redefined in the block.
if (BI.LiveIn && BI.LiveOut && BI.FirstDef)
Cost += SpillPlacer->getBlockFrequency(Number);
}
return Cost;
}
/// calcGlobalSplitCost - Return the global split cost of following the split
/// pattern in LiveBundles. This cost should be added to the local cost of the
/// interference pattern in SplitConstraints.
///
BlockFrequency RAGreedy::calcGlobalSplitCost(GlobalSplitCandidate &Cand,
const AllocationOrder &Order) {
BlockFrequency GlobalCost = 0;
const BitVector &LiveBundles = Cand.LiveBundles;
ArrayRef<SplitAnalysis::BlockInfo> UseBlocks = SA->getUseBlocks();
for (unsigned I = 0; I != UseBlocks.size(); ++I) {
const SplitAnalysis::BlockInfo &BI = UseBlocks[I];
SpillPlacement::BlockConstraint &BC = SplitConstraints[I];
bool RegIn = LiveBundles[Bundles->getBundle(BC.Number, false)];
bool RegOut = LiveBundles[Bundles->getBundle(BC.Number, true)];
unsigned Ins = 0;
Cand.Intf.moveToBlock(BC.Number);
if (BI.LiveIn)
Ins += RegIn != (BC.Entry == SpillPlacement::PrefReg);
if (BI.LiveOut)
Ins += RegOut != (BC.Exit == SpillPlacement::PrefReg);
while (Ins--)
GlobalCost += SpillPlacer->getBlockFrequency(BC.Number);
}
for (unsigned Number : Cand.ActiveBlocks) {
bool RegIn = LiveBundles[Bundles->getBundle(Number, false)];
bool RegOut = LiveBundles[Bundles->getBundle(Number, true)];
if (!RegIn && !RegOut)
continue;
if (RegIn && RegOut) {
// We need double spill code if this block has interference.
Cand.Intf.moveToBlock(Number);
if (Cand.Intf.hasInterference()) {
GlobalCost += SpillPlacer->getBlockFrequency(Number);
GlobalCost += SpillPlacer->getBlockFrequency(Number);
}
continue;
}
// live-in / stack-out or stack-in live-out.
GlobalCost += SpillPlacer->getBlockFrequency(Number);
}
return GlobalCost;
}
/// splitAroundRegion - Split the current live range around the regions
/// determined by BundleCand and GlobalCand.
///
/// Before calling this function, GlobalCand and BundleCand must be initialized
/// so each bundle is assigned to a valid candidate, or NoCand for the
/// stack-bound bundles. The shared SA/SE SplitAnalysis and SplitEditor
/// objects must be initialized for the current live range, and intervals
/// created for the used candidates.
///
/// @param LREdit The LiveRangeEdit object handling the current split.
/// @param UsedCands List of used GlobalCand entries. Every BundleCand value
/// must appear in this list.
void RAGreedy::splitAroundRegion(LiveRangeEdit &LREdit,
ArrayRef<unsigned> UsedCands) {
// These are the intervals created for new global ranges. We may create more
// intervals for local ranges.
const unsigned NumGlobalIntvs = LREdit.size();
LLVM_DEBUG(dbgs() << "splitAroundRegion with " << NumGlobalIntvs
<< " globals.\n");
assert(NumGlobalIntvs && "No global intervals configured");
// Isolate even single instructions when dealing with a proper sub-class.
// That guarantees register class inflation for the stack interval because it
// is all copies.
Register Reg = SA->getParent().reg();
bool SingleInstrs = RegClassInfo.isProperSubClass(MRI->getRegClass(Reg));
// First handle all the blocks with uses.
ArrayRef<SplitAnalysis::BlockInfo> UseBlocks = SA->getUseBlocks();
for (const SplitAnalysis::BlockInfo &BI : UseBlocks) {
unsigned Number = BI.MBB->getNumber();
unsigned IntvIn = 0, IntvOut = 0;
SlotIndex IntfIn, IntfOut;
if (BI.LiveIn) {
unsigned CandIn = BundleCand[Bundles->getBundle(Number, false)];
if (CandIn != NoCand) {
GlobalSplitCandidate &Cand = GlobalCand[CandIn];
IntvIn = Cand.IntvIdx;
Cand.Intf.moveToBlock(Number);
IntfIn = Cand.Intf.first();
}
}
if (BI.LiveOut) {
unsigned CandOut = BundleCand[Bundles->getBundle(Number, true)];
if (CandOut != NoCand) {
GlobalSplitCandidate &Cand = GlobalCand[CandOut];
IntvOut = Cand.IntvIdx;
Cand.Intf.moveToBlock(Number);
IntfOut = Cand.Intf.last();
}
}
// Create separate intervals for isolated blocks with multiple uses.
if (!IntvIn && !IntvOut) {
LLVM_DEBUG(dbgs() << printMBBReference(*BI.MBB) << " isolated.\n");
if (SA->shouldSplitSingleBlock(BI, SingleInstrs))
SE->splitSingleBlock(BI);
continue;
}
if (IntvIn && IntvOut)
SE->splitLiveThroughBlock(Number, IntvIn, IntfIn, IntvOut, IntfOut);
else if (IntvIn)
SE->splitRegInBlock(BI, IntvIn, IntfIn);
else
SE->splitRegOutBlock(BI, IntvOut, IntfOut);
}
// Handle live-through blocks. The relevant live-through blocks are stored in
// the ActiveBlocks list with each candidate. We need to filter out
// duplicates.
BitVector Todo = SA->getThroughBlocks();
for (unsigned UsedCand : UsedCands) {
ArrayRef<unsigned> Blocks = GlobalCand[UsedCand].ActiveBlocks;
for (unsigned Number : Blocks) {
if (!Todo.test(Number))
continue;
Todo.reset(Number);
unsigned IntvIn = 0, IntvOut = 0;
SlotIndex IntfIn, IntfOut;
unsigned CandIn = BundleCand[Bundles->getBundle(Number, false)];
if (CandIn != NoCand) {
GlobalSplitCandidate &Cand = GlobalCand[CandIn];
IntvIn = Cand.IntvIdx;
Cand.Intf.moveToBlock(Number);
IntfIn = Cand.Intf.first();
}
unsigned CandOut = BundleCand[Bundles->getBundle(Number, true)];
if (CandOut != NoCand) {
GlobalSplitCandidate &Cand = GlobalCand[CandOut];
IntvOut = Cand.IntvIdx;
Cand.Intf.moveToBlock(Number);
IntfOut = Cand.Intf.last();
}
if (!IntvIn && !IntvOut)
continue;
SE->splitLiveThroughBlock(Number, IntvIn, IntfIn, IntvOut, IntfOut);
}
}
++NumGlobalSplits;
SmallVector<unsigned, 8> IntvMap;
SE->finish(&IntvMap);
DebugVars->splitRegister(Reg, LREdit.regs(), *LIS);
unsigned OrigBlocks = SA->getNumLiveBlocks();
// Sort out the new intervals created by splitting. We get four kinds:
// - Remainder intervals should not be split again.
// - Candidate intervals can be assigned to Cand.PhysReg.
// - Block-local splits are candidates for local splitting.
// - DCE leftovers should go back on the queue.
for (unsigned I = 0, E = LREdit.size(); I != E; ++I) {
const LiveInterval &Reg = LIS->getInterval(LREdit.get(I));
// Ignore old intervals from DCE.
if (ExtraInfo->getOrInitStage(Reg.reg()) != RS_New)
continue;
// Remainder interval. Don't try splitting again, spill if it doesn't
// allocate.
if (IntvMap[I] == 0) {
ExtraInfo->setStage(Reg, RS_Spill);
continue;
}
// Global intervals. Allow repeated splitting as long as the number of live
// blocks is strictly decreasing.
if (IntvMap[I] < NumGlobalIntvs) {
if (SA->countLiveBlocks(&Reg) >= OrigBlocks) {
LLVM_DEBUG(dbgs() << "Main interval covers the same " << OrigBlocks
<< " blocks as original.\n");
// Don't allow repeated splitting as a safe guard against looping.
ExtraInfo->setStage(Reg, RS_Split2);
}
continue;
}
// Other intervals are treated as new. This includes local intervals created
// for blocks with multiple uses, and anything created by DCE.
}
if (VerifyEnabled)
MF->verify(this, "After splitting live range around region");
}
MCRegister RAGreedy::tryRegionSplit(const LiveInterval &VirtReg,
AllocationOrder &Order,
SmallVectorImpl<Register> &NewVRegs) {
if (!TRI->shouldRegionSplitForVirtReg(*MF, VirtReg))
return MCRegister::NoRegister;
unsigned NumCands = 0;
BlockFrequency SpillCost = calcSpillCost();
BlockFrequency BestCost;
// Check if we can split this live range around a compact region.
bool HasCompact = calcCompactRegion(GlobalCand.front());
if (HasCompact) {
// Yes, keep GlobalCand[0] as the compact region candidate.
NumCands = 1;
BestCost = BlockFrequency::getMaxFrequency();
} else {
// No benefit from the compact region, our fallback will be per-block
// splitting. Make sure we find a solution that is cheaper than spilling.
BestCost = SpillCost;
LLVM_DEBUG(dbgs() << "Cost of isolating all blocks = ";
MBFI->printBlockFreq(dbgs(), BestCost) << '\n');
}
unsigned BestCand = calculateRegionSplitCost(VirtReg, Order, BestCost,
NumCands, false /*IgnoreCSR*/);
// No solutions found, fall back to single block splitting.
if (!HasCompact && BestCand == NoCand)
return MCRegister::NoRegister;
return doRegionSplit(VirtReg, BestCand, HasCompact, NewVRegs);
}
unsigned RAGreedy::calculateRegionSplitCost(const LiveInterval &VirtReg,
AllocationOrder &Order,
BlockFrequency &BestCost,
unsigned &NumCands,
bool IgnoreCSR) {
unsigned BestCand = NoCand;
for (MCPhysReg PhysReg : Order) {
assert(PhysReg);
if (IgnoreCSR && EvictAdvisor->isUnusedCalleeSavedReg(PhysReg))
continue;
// Discard bad candidates before we run out of interference cache cursors.
// This will only affect register classes with a lot of registers (>32).
if (NumCands == IntfCache.getMaxCursors()) {
unsigned WorstCount = ~0u;
unsigned Worst = 0;
for (unsigned CandIndex = 0; CandIndex != NumCands; ++CandIndex) {
if (CandIndex == BestCand || !GlobalCand[CandIndex].PhysReg)
continue;
unsigned Count = GlobalCand[CandIndex].LiveBundles.count();
if (Count < WorstCount) {
Worst = CandIndex;
WorstCount = Count;
}
}
--NumCands;
GlobalCand[Worst] = GlobalCand[NumCands];
if (BestCand == NumCands)
BestCand = Worst;
}
if (GlobalCand.size() <= NumCands)
GlobalCand.resize(NumCands+1);
GlobalSplitCandidate &Cand = GlobalCand[NumCands];
Cand.reset(IntfCache, PhysReg);
SpillPlacer->prepare(Cand.LiveBundles);
BlockFrequency Cost;
if (!addSplitConstraints(Cand.Intf, Cost)) {
LLVM_DEBUG(dbgs() << printReg(PhysReg, TRI) << "\tno positive bundles\n");
continue;
}
LLVM_DEBUG(dbgs() << printReg(PhysReg, TRI) << "\tstatic = ";
MBFI->printBlockFreq(dbgs(), Cost));
if (Cost >= BestCost) {
LLVM_DEBUG({
if (BestCand == NoCand)
dbgs() << " worse than no bundles\n";
else
dbgs() << " worse than "
<< printReg(GlobalCand[BestCand].PhysReg, TRI) << '\n';
});
continue;
}
if (!growRegion(Cand)) {
LLVM_DEBUG(dbgs() << ", cannot spill all interferences.\n");
continue;
}
SpillPlacer->finish();
// No live bundles, defer to splitSingleBlocks().
if (!Cand.LiveBundles.any()) {
LLVM_DEBUG(dbgs() << " no bundles.\n");
continue;
}
Cost += calcGlobalSplitCost(Cand, Order);
LLVM_DEBUG({
dbgs() << ", total = ";
MBFI->printBlockFreq(dbgs(), Cost) << " with bundles";
for (int I : Cand.LiveBundles.set_bits())
dbgs() << " EB#" << I;
dbgs() << ".\n";
});
if (Cost < BestCost) {
BestCand = NumCands;
BestCost = Cost;
}
++NumCands;
}
return BestCand;
}
unsigned RAGreedy::doRegionSplit(const LiveInterval &VirtReg, unsigned BestCand,
bool HasCompact,
SmallVectorImpl<Register> &NewVRegs) {
SmallVector<unsigned, 8> UsedCands;
// Prepare split editor.
LiveRangeEdit LREdit(&VirtReg, NewVRegs, *MF, *LIS, VRM, this, &DeadRemats);
SE->reset(LREdit, SplitSpillMode);
// Assign all edge bundles to the preferred candidate, or NoCand.
BundleCand.assign(Bundles->getNumBundles(), NoCand);
// Assign bundles for the best candidate region.
if (BestCand != NoCand) {
GlobalSplitCandidate &Cand = GlobalCand[BestCand];
if (unsigned B = Cand.getBundles(BundleCand, BestCand)) {
UsedCands.push_back(BestCand);
Cand.IntvIdx = SE->openIntv();
LLVM_DEBUG(dbgs() << "Split for " << printReg(Cand.PhysReg, TRI) << " in "
<< B << " bundles, intv " << Cand.IntvIdx << ".\n");
(void)B;
}
}
// Assign bundles for the compact region.
if (HasCompact) {
GlobalSplitCandidate &Cand = GlobalCand.front();
assert(!Cand.PhysReg && "Compact region has no physreg");
if (unsigned B = Cand.getBundles(BundleCand, 0)) {
UsedCands.push_back(0);
Cand.IntvIdx = SE->openIntv();
LLVM_DEBUG(dbgs() << "Split for compact region in " << B
<< " bundles, intv " << Cand.IntvIdx << ".\n");
(void)B;
}
}
splitAroundRegion(LREdit, UsedCands);
return 0;
}
//===----------------------------------------------------------------------===//
// Per-Block Splitting
//===----------------------------------------------------------------------===//
/// tryBlockSplit - Split a global live range around every block with uses. This
/// creates a lot of local live ranges, that will be split by tryLocalSplit if
/// they don't allocate.
unsigned RAGreedy::tryBlockSplit(const LiveInterval &VirtReg,
AllocationOrder &Order,
SmallVectorImpl<Register> &NewVRegs) {
assert(&SA->getParent() == &VirtReg && "Live range wasn't analyzed");
Register Reg = VirtReg.reg();
bool SingleInstrs = RegClassInfo.isProperSubClass(MRI->getRegClass(Reg));
LiveRangeEdit LREdit(&VirtReg, NewVRegs, *MF, *LIS, VRM, this, &DeadRemats);
SE->reset(LREdit, SplitSpillMode);
ArrayRef<SplitAnalysis::BlockInfo> UseBlocks = SA->getUseBlocks();
for (const SplitAnalysis::BlockInfo &BI : UseBlocks) {
if (SA->shouldSplitSingleBlock(BI, SingleInstrs))
SE->splitSingleBlock(BI);
}
// No blocks were split.
if (LREdit.empty())
return 0;
// We did split for some blocks.
SmallVector<unsigned, 8> IntvMap;
SE->finish(&IntvMap);
// Tell LiveDebugVariables about the new ranges.
DebugVars->splitRegister(Reg, LREdit.regs(), *LIS);
// Sort out the new intervals created by splitting. The remainder interval
// goes straight to spilling, the new local ranges get to stay RS_New.
for (unsigned I = 0, E = LREdit.size(); I != E; ++I) {
const LiveInterval &LI = LIS->getInterval(LREdit.get(I));
if (ExtraInfo->getOrInitStage(LI.reg()) == RS_New && IntvMap[I] == 0)
ExtraInfo->setStage(LI, RS_Spill);
}
if (VerifyEnabled)
MF->verify(this, "After splitting live range around basic blocks");
return 0;
}
//===----------------------------------------------------------------------===//
// Per-Instruction Splitting
//===----------------------------------------------------------------------===//
/// Get the number of allocatable registers that match the constraints of \p Reg
/// on \p MI and that are also in \p SuperRC.
static unsigned getNumAllocatableRegsForConstraints(
const MachineInstr *MI, Register Reg, const TargetRegisterClass *SuperRC,
const TargetInstrInfo *TII, const TargetRegisterInfo *TRI,
const RegisterClassInfo &RCI) {
assert(SuperRC && "Invalid register class");
const TargetRegisterClass *ConstrainedRC =
MI->getRegClassConstraintEffectForVReg(Reg, SuperRC, TII, TRI,
/* ExploreBundle */ true);
if (!ConstrainedRC)
return 0;
return RCI.getNumAllocatableRegs(ConstrainedRC);
}
static LaneBitmask getInstReadLaneMask(const MachineRegisterInfo &MRI,
const TargetRegisterInfo &TRI,
const MachineInstr &MI, Register Reg) {
LaneBitmask Mask;
for (const MachineOperand &MO : MI.operands()) {
if (!MO.isReg() || MO.getReg() != Reg)
continue;
unsigned SubReg = MO.getSubReg();
if (SubReg == 0 && MO.isUse()) {
Mask |= MRI.getMaxLaneMaskForVReg(Reg);
continue;
}
LaneBitmask SubRegMask = TRI.getSubRegIndexLaneMask(SubReg);
if (MO.isDef()) {
if (!MO.isUndef())
Mask |= ~SubRegMask;
} else
Mask |= SubRegMask;
}
return Mask;
}
/// Return true if \p MI at \P Use reads a subset of the lanes live in \p
/// VirtReg.
static bool readsLaneSubset(const MachineRegisterInfo &MRI,
const MachineInstr *MI, const LiveInterval &VirtReg,
const TargetRegisterInfo *TRI, SlotIndex Use) {
// Early check the common case.
if (MI->isCopy() &&
MI->getOperand(0).getSubReg() == MI->getOperand(1).getSubReg())
return false;
// FIXME: We're only considering uses, but should be consider defs too?
LaneBitmask ReadMask = getInstReadLaneMask(MRI, *TRI, *MI, VirtReg.reg());
LaneBitmask LiveAtMask;
for (const LiveInterval::SubRange &S : VirtReg.subranges()) {
if (S.liveAt(Use))
LiveAtMask |= S.LaneMask;
}
// If the live lanes aren't different from the lanes used by the instruction,
// this doesn't help.
return (ReadMask & ~(LiveAtMask & TRI->getCoveringLanes())).any();
}
/// tryInstructionSplit - Split a live range around individual instructions.
/// This is normally not worthwhile since the spiller is doing essentially the
/// same thing. However, when the live range is in a constrained register
/// class, it may help to insert copies such that parts of the live range can
/// be moved to a larger register class.
///
/// This is similar to spilling to a larger register class.
unsigned RAGreedy::tryInstructionSplit(const LiveInterval &VirtReg,
AllocationOrder &Order,
SmallVectorImpl<Register> &NewVRegs) {
const TargetRegisterClass *CurRC = MRI->getRegClass(VirtReg.reg());
// There is no point to this if there are no larger sub-classes.
bool SplitSubClass = true;
if (!RegClassInfo.isProperSubClass(CurRC)) {
if (!VirtReg.hasSubRanges())
return 0;
SplitSubClass = false;
}
// Always enable split spill mode, since we're effectively spilling to a
// register.
LiveRangeEdit LREdit(&VirtReg, NewVRegs, *MF, *LIS, VRM, this, &DeadRemats);
SE->reset(LREdit, SplitEditor::SM_Size);
ArrayRef<SlotIndex> Uses = SA->getUseSlots();
if (Uses.size() <= 1)
return 0;
LLVM_DEBUG(dbgs() << "Split around " << Uses.size()
<< " individual instrs.\n");
const TargetRegisterClass *SuperRC =
TRI->getLargestLegalSuperClass(CurRC, *MF);
unsigned SuperRCNumAllocatableRegs =
RegClassInfo.getNumAllocatableRegs(SuperRC);
// Split around every non-copy instruction if this split will relax
// the constraints on the virtual register.
// Otherwise, splitting just inserts uncoalescable copies that do not help
// the allocation.
for (const SlotIndex Use : Uses) {
if (const MachineInstr *MI = Indexes->getInstructionFromIndex(Use)) {
if (MI->isFullCopy() ||
(SplitSubClass &&
SuperRCNumAllocatableRegs ==
getNumAllocatableRegsForConstraints(MI, VirtReg.reg(), SuperRC,
TII, TRI, RegClassInfo)) ||
// TODO: Handle split for subranges with subclass constraints?
(!SplitSubClass && VirtReg.hasSubRanges() &&
!readsLaneSubset(*MRI, MI, VirtReg, TRI, Use))) {
LLVM_DEBUG(dbgs() << " skip:\t" << Use << '\t' << *MI);
continue;
}
}
SE->openIntv();
SlotIndex SegStart = SE->enterIntvBefore(Use);
SlotIndex SegStop = SE->leaveIntvAfter(Use);
SE->useIntv(SegStart, SegStop);
}
if (LREdit.empty()) {
LLVM_DEBUG(dbgs() << "All uses were copies.\n");
return 0;
}
SmallVector<unsigned, 8> IntvMap;
SE->finish(&IntvMap);
DebugVars->splitRegister(VirtReg.reg(), LREdit.regs(), *LIS);
// Assign all new registers to RS_Spill. This was the last chance.
ExtraInfo->setStage(LREdit.begin(), LREdit.end(), RS_Spill);
return 0;
}
//===----------------------------------------------------------------------===//
// Local Splitting
//===----------------------------------------------------------------------===//
/// calcGapWeights - Compute the maximum spill weight that needs to be evicted
/// in order to use PhysReg between two entries in SA->UseSlots.
///
/// GapWeight[I] represents the gap between UseSlots[I] and UseSlots[I + 1].
///
void RAGreedy::calcGapWeights(MCRegister PhysReg,
SmallVectorImpl<float> &GapWeight) {
assert(SA->getUseBlocks().size() == 1 && "Not a local interval");
const SplitAnalysis::BlockInfo &BI = SA->getUseBlocks().front();
ArrayRef<SlotIndex> Uses = SA->getUseSlots();
const unsigned NumGaps = Uses.size()-1;
// Start and end points for the interference check.
SlotIndex StartIdx =
BI.LiveIn ? BI.FirstInstr.getBaseIndex() : BI.FirstInstr;
SlotIndex StopIdx =
BI.LiveOut ? BI.LastInstr.getBoundaryIndex() : BI.LastInstr;
GapWeight.assign(NumGaps, 0.0f);
// Add interference from each overlapping register.
for (MCRegUnitIterator Units(PhysReg, TRI); Units.isValid(); ++Units) {
if (!Matrix->query(const_cast<LiveInterval&>(SA->getParent()), *Units)
.checkInterference())
continue;
// We know that VirtReg is a continuous interval from FirstInstr to
// LastInstr, so we don't need InterferenceQuery.
//
// Interference that overlaps an instruction is counted in both gaps
// surrounding the instruction. The exception is interference before
// StartIdx and after StopIdx.
//
LiveIntervalUnion::SegmentIter IntI =
Matrix->getLiveUnions()[*Units] .find(StartIdx);
for (unsigned Gap = 0; IntI.valid() && IntI.start() < StopIdx; ++IntI) {
// Skip the gaps before IntI.
while (Uses[Gap+1].getBoundaryIndex() < IntI.start())
if (++Gap == NumGaps)
break;
if (Gap == NumGaps)
break;
// Update the gaps covered by IntI.
const float weight = IntI.value()->weight();
for (; Gap != NumGaps; ++Gap) {
GapWeight[Gap] = std::max(GapWeight[Gap], weight);
if (Uses[Gap+1].getBaseIndex() >= IntI.stop())
break;
}
if (Gap == NumGaps)
break;
}
}
// Add fixed interference.
for (MCRegUnitIterator Units(PhysReg, TRI); Units.isValid(); ++Units) {
const LiveRange &LR = LIS->getRegUnit(*Units);
LiveRange::const_iterator I = LR.find(StartIdx);
LiveRange::const_iterator E = LR.end();
// Same loop as above. Mark any overlapped gaps as HUGE_VALF.
for (unsigned Gap = 0; I != E && I->start < StopIdx; ++I) {
while (Uses[Gap+1].getBoundaryIndex() < I->start)
if (++Gap == NumGaps)
break;
if (Gap == NumGaps)
break;
for (; Gap != NumGaps; ++Gap) {
GapWeight[Gap] = huge_valf;
if (Uses[Gap+1].getBaseIndex() >= I->end)
break;
}
if (Gap == NumGaps)
break;
}
}
}
/// tryLocalSplit - Try to split VirtReg into smaller intervals inside its only
/// basic block.
///
unsigned RAGreedy::tryLocalSplit(const LiveInterval &VirtReg,
AllocationOrder &Order,
SmallVectorImpl<Register> &NewVRegs) {
// TODO: the function currently only handles a single UseBlock; it should be
// possible to generalize.
if (SA->getUseBlocks().size() != 1)
return 0;
const SplitAnalysis::BlockInfo &BI = SA->getUseBlocks().front();
// Note that it is possible to have an interval that is live-in or live-out
// while only covering a single block - A phi-def can use undef values from
// predecessors, and the block could be a single-block loop.
// We don't bother doing anything clever about such a case, we simply assume
// that the interval is continuous from FirstInstr to LastInstr. We should
// make sure that we don't do anything illegal to such an interval, though.
ArrayRef<SlotIndex> Uses = SA->getUseSlots();
if (Uses.size() <= 2)
return 0;
const unsigned NumGaps = Uses.size()-1;
LLVM_DEBUG({
dbgs() << "tryLocalSplit: ";
for (const auto &Use : Uses)
dbgs() << ' ' << Use;
dbgs() << '\n';
});
// If VirtReg is live across any register mask operands, compute a list of
// gaps with register masks.
SmallVector<unsigned, 8> RegMaskGaps;
if (Matrix->checkRegMaskInterference(VirtReg)) {
// Get regmask slots for the whole block.
ArrayRef<SlotIndex> RMS = LIS->getRegMaskSlotsInBlock(BI.MBB->getNumber());
LLVM_DEBUG(dbgs() << RMS.size() << " regmasks in block:");
// Constrain to VirtReg's live range.
unsigned RI =
llvm::lower_bound(RMS, Uses.front().getRegSlot()) - RMS.begin();
unsigned RE = RMS.size();
for (unsigned I = 0; I != NumGaps && RI != RE; ++I) {
// Look for Uses[I] <= RMS <= Uses[I + 1].
assert(!SlotIndex::isEarlierInstr(RMS[RI], Uses[I]));
if (SlotIndex::isEarlierInstr(Uses[I + 1], RMS[RI]))
continue;
// Skip a regmask on the same instruction as the last use. It doesn't
// overlap the live range.
if (SlotIndex::isSameInstr(Uses[I + 1], RMS[RI]) && I + 1 == NumGaps)
break;
LLVM_DEBUG(dbgs() << ' ' << RMS[RI] << ':' << Uses[I] << '-'
<< Uses[I + 1]);
RegMaskGaps.push_back(I);
// Advance ri to the next gap. A regmask on one of the uses counts in
// both gaps.
while (RI != RE && SlotIndex::isEarlierInstr(RMS[RI], Uses[I + 1]))
++RI;
}
LLVM_DEBUG(dbgs() << '\n');
}
// Since we allow local split results to be split again, there is a risk of
// creating infinite loops. It is tempting to require that the new live
// ranges have less instructions than the original. That would guarantee
// convergence, but it is too strict. A live range with 3 instructions can be
// split 2+3 (including the COPY), and we want to allow that.
//
// Instead we use these rules:
//
// 1. Allow any split for ranges with getStage() < RS_Split2. (Except for the
// noop split, of course).
// 2. Require progress be made for ranges with getStage() == RS_Split2. All
// the new ranges must have fewer instructions than before the split.
// 3. New ranges with the same number of instructions are marked RS_Split2,
// smaller ranges are marked RS_New.
//
// These rules allow a 3 -> 2+3 split once, which we need. They also prevent
// excessive splitting and infinite loops.
//
bool ProgressRequired = ExtraInfo->getStage(VirtReg) >= RS_Split2;
// Best split candidate.
unsigned BestBefore = NumGaps;
unsigned BestAfter = 0;
float BestDiff = 0;
const float blockFreq =
SpillPlacer->getBlockFrequency(BI.MBB->getNumber()).getFrequency() *
(1.0f / MBFI->getEntryFreq());
SmallVector<float, 8> GapWeight;
for (MCPhysReg PhysReg : Order) {
assert(PhysReg);
// Keep track of the largest spill weight that would need to be evicted in
// order to make use of PhysReg between UseSlots[I] and UseSlots[I + 1].
calcGapWeights(PhysReg, GapWeight);
// Remove any gaps with regmask clobbers.
if (Matrix->checkRegMaskInterference(VirtReg, PhysReg))
for (unsigned I = 0, E = RegMaskGaps.size(); I != E; ++I)
GapWeight[RegMaskGaps[I]] = huge_valf;
// Try to find the best sequence of gaps to close.
// The new spill weight must be larger than any gap interference.
// We will split before Uses[SplitBefore] and after Uses[SplitAfter].
unsigned SplitBefore = 0, SplitAfter = 1;
// MaxGap should always be max(GapWeight[SplitBefore..SplitAfter-1]).
// It is the spill weight that needs to be evicted.
float MaxGap = GapWeight[0];
while (true) {
// Live before/after split?
const bool LiveBefore = SplitBefore != 0 || BI.LiveIn;
const bool LiveAfter = SplitAfter != NumGaps || BI.LiveOut;
LLVM_DEBUG(dbgs() << printReg(PhysReg, TRI) << ' ' << Uses[SplitBefore]
<< '-' << Uses[SplitAfter] << " I=" << MaxGap);
// Stop before the interval gets so big we wouldn't be making progress.
if (!LiveBefore && !LiveAfter) {
LLVM_DEBUG(dbgs() << " all\n");
break;
}
// Should the interval be extended or shrunk?
bool Shrink = true;
// How many gaps would the new range have?
unsigned NewGaps = LiveBefore + SplitAfter - SplitBefore + LiveAfter;
// Legally, without causing looping?
bool Legal = !ProgressRequired || NewGaps < NumGaps;
if (Legal && MaxGap < huge_valf) {
// Estimate the new spill weight. Each instruction reads or writes the
// register. Conservatively assume there are no read-modify-write
// instructions.
//
// Try to guess the size of the new interval.
const float EstWeight = normalizeSpillWeight(
blockFreq * (NewGaps + 1),
Uses[SplitBefore].distance(Uses[SplitAfter]) +
(LiveBefore + LiveAfter) * SlotIndex::InstrDist,
1);
// Would this split be possible to allocate?
// Never allocate all gaps, we wouldn't be making progress.
LLVM_DEBUG(dbgs() << " w=" << EstWeight);
if (EstWeight * Hysteresis >= MaxGap) {
Shrink = false;
float Diff = EstWeight - MaxGap;
if (Diff > BestDiff) {
LLVM_DEBUG(dbgs() << " (best)");
BestDiff = Hysteresis * Diff;
BestBefore = SplitBefore;
BestAfter = SplitAfter;
}
}
}
// Try to shrink.
if (Shrink) {
if (++SplitBefore < SplitAfter) {
LLVM_DEBUG(dbgs() << " shrink\n");
// Recompute the max when necessary.
if (GapWeight[SplitBefore - 1] >= MaxGap) {
MaxGap = GapWeight[SplitBefore];
for (unsigned I = SplitBefore + 1; I != SplitAfter; ++I)
MaxGap = std::max(MaxGap, GapWeight[I]);
}
continue;
}
MaxGap = 0;
}
// Try to extend the interval.
if (SplitAfter >= NumGaps) {
LLVM_DEBUG(dbgs() << " end\n");
break;
}
LLVM_DEBUG(dbgs() << " extend\n");
MaxGap = std::max(MaxGap, GapWeight[SplitAfter++]);
}
}
// Didn't find any candidates?
if (BestBefore == NumGaps)
return 0;
LLVM_DEBUG(dbgs() << "Best local split range: " << Uses[BestBefore] << '-'
<< Uses[BestAfter] << ", " << BestDiff << ", "
<< (BestAfter - BestBefore + 1) << " instrs\n");
LiveRangeEdit LREdit(&VirtReg, NewVRegs, *MF, *LIS, VRM, this, &DeadRemats);
SE->reset(LREdit);
SE->openIntv();
SlotIndex SegStart = SE->enterIntvBefore(Uses[BestBefore]);
SlotIndex SegStop = SE->leaveIntvAfter(Uses[BestAfter]);
SE->useIntv(SegStart, SegStop);
SmallVector<unsigned, 8> IntvMap;
SE->finish(&IntvMap);
DebugVars->splitRegister(VirtReg.reg(), LREdit.regs(), *LIS);
// If the new range has the same number of instructions as before, mark it as
// RS_Split2 so the next split will be forced to make progress. Otherwise,
// leave the new intervals as RS_New so they can compete.
bool LiveBefore = BestBefore != 0 || BI.LiveIn;
bool LiveAfter = BestAfter != NumGaps || BI.LiveOut;
unsigned NewGaps = LiveBefore + BestAfter - BestBefore + LiveAfter;
if (NewGaps >= NumGaps) {
LLVM_DEBUG(dbgs() << "Tagging non-progress ranges:");
assert(!ProgressRequired && "Didn't make progress when it was required.");
for (unsigned I = 0, E = IntvMap.size(); I != E; ++I)
if (IntvMap[I] == 1) {
ExtraInfo->setStage(LIS->getInterval(LREdit.get(I)), RS_Split2);
LLVM_DEBUG(dbgs() << ' ' << printReg(LREdit.get(I)));
}
LLVM_DEBUG(dbgs() << '\n');
}
++NumLocalSplits;
return 0;
}
//===----------------------------------------------------------------------===//
// Live Range Splitting
//===----------------------------------------------------------------------===//
/// trySplit - Try to split VirtReg or one of its interferences, making it
/// assignable.
/// @return Physreg when VirtReg may be assigned and/or new NewVRegs.
unsigned RAGreedy::trySplit(const LiveInterval &VirtReg, AllocationOrder &Order,
SmallVectorImpl<Register> &NewVRegs,
const SmallVirtRegSet &FixedRegisters) {
// Ranges must be Split2 or less.
if (ExtraInfo->getStage(VirtReg) >= RS_Spill)
return 0;
// Local intervals are handled separately.
if (LIS->intervalIsInOneMBB(VirtReg)) {
NamedRegionTimer T("local_split", "Local Splitting", TimerGroupName,
TimerGroupDescription, TimePassesIsEnabled);
SA->analyze(&VirtReg);
Register PhysReg = tryLocalSplit(VirtReg, Order, NewVRegs);
if (PhysReg || !NewVRegs.empty())
return PhysReg;
return tryInstructionSplit(VirtReg, Order, NewVRegs);
}
NamedRegionTimer T("global_split", "Global Splitting", TimerGroupName,
TimerGroupDescription, TimePassesIsEnabled);
SA->analyze(&VirtReg);
// First try to split around a region spanning multiple blocks. RS_Split2
// ranges already made dubious progress with region splitting, so they go
// straight to single block splitting.
if (ExtraInfo->getStage(VirtReg) < RS_Split2) {
MCRegister PhysReg = tryRegionSplit(VirtReg, Order, NewVRegs);
if (PhysReg || !NewVRegs.empty())
return PhysReg;
}
// Then isolate blocks.
return tryBlockSplit(VirtReg, Order, NewVRegs);
}
//===----------------------------------------------------------------------===//
// Last Chance Recoloring
//===----------------------------------------------------------------------===//
/// Return true if \p reg has any tied def operand.
static bool hasTiedDef(MachineRegisterInfo *MRI, unsigned reg) {
for (const MachineOperand &MO : MRI->def_operands(reg))
if (MO.isTied())
return true;
return false;
}
/// Return true if the existing assignment of \p Intf overlaps, but is not the
/// same, as \p PhysReg.
static bool assignedRegPartiallyOverlaps(const TargetRegisterInfo &TRI,
const VirtRegMap &VRM,
MCRegister PhysReg,
const LiveInterval &Intf) {
MCRegister AssignedReg = VRM.getPhys(Intf.reg());
if (PhysReg == AssignedReg)
return false;
return TRI.regsOverlap(PhysReg, AssignedReg);
}
/// mayRecolorAllInterferences - Check if the virtual registers that
/// interfere with \p VirtReg on \p PhysReg (or one of its aliases) may be
/// recolored to free \p PhysReg.
/// When true is returned, \p RecoloringCandidates has been augmented with all
/// the live intervals that need to be recolored in order to free \p PhysReg
/// for \p VirtReg.
/// \p FixedRegisters contains all the virtual registers that cannot be
/// recolored.
bool RAGreedy::mayRecolorAllInterferences(
MCRegister PhysReg, const LiveInterval &VirtReg,
SmallLISet &RecoloringCandidates, const SmallVirtRegSet &FixedRegisters) {
const TargetRegisterClass *CurRC = MRI->getRegClass(VirtReg.reg());
for (MCRegUnitIterator Units(PhysReg, TRI); Units.isValid(); ++Units) {
LiveIntervalUnion::Query &Q = Matrix->query(VirtReg, *Units);
// If there is LastChanceRecoloringMaxInterference or more interferences,
// chances are one would not be recolorable.
if (Q.interferingVRegs(LastChanceRecoloringMaxInterference).size() >=
LastChanceRecoloringMaxInterference &&
!ExhaustiveSearch) {
LLVM_DEBUG(dbgs() << "Early abort: too many interferences.\n");
CutOffInfo |= CO_Interf;
return false;
}
for (const LiveInterval *Intf : reverse(Q.interferingVRegs())) {
// If Intf is done and sits on the same register class as VirtReg, it
// would not be recolorable as it is in the same state as
// VirtReg. However there are at least two exceptions.
//
// If VirtReg has tied defs and Intf doesn't, then
// there is still a point in examining if it can be recolorable.
//
// Additionally, if the register class has overlapping tuple members, it
// may still be recolorable using a different tuple. This is more likely
// if the existing assignment aliases with the candidate.
//
if (((ExtraInfo->getStage(*Intf) == RS_Done &&
MRI->getRegClass(Intf->reg()) == CurRC &&
!assignedRegPartiallyOverlaps(*TRI, *VRM, PhysReg, *Intf)) &&
!(hasTiedDef(MRI, VirtReg.reg()) &&
!hasTiedDef(MRI, Intf->reg()))) ||
FixedRegisters.count(Intf->reg())) {
LLVM_DEBUG(
dbgs() << "Early abort: the interference is not recolorable.\n");
return false;
}
RecoloringCandidates.insert(Intf);
}
}
return true;
}
/// tryLastChanceRecoloring - Try to assign a color to \p VirtReg by recoloring
/// its interferences.
/// Last chance recoloring chooses a color for \p VirtReg and recolors every
/// virtual register that was using it. The recoloring process may recursively
/// use the last chance recoloring. Therefore, when a virtual register has been
/// assigned a color by this mechanism, it is marked as Fixed, i.e., it cannot
/// be last-chance-recolored again during this recoloring "session".
/// E.g.,
/// Let
/// vA can use {R1, R2 }
/// vB can use { R2, R3}
/// vC can use {R1 }
/// Where vA, vB, and vC cannot be split anymore (they are reloads for
/// instance) and they all interfere.
///
/// vA is assigned R1
/// vB is assigned R2
/// vC tries to evict vA but vA is already done.
/// Regular register allocation fails.
///
/// Last chance recoloring kicks in:
/// vC does as if vA was evicted => vC uses R1.
/// vC is marked as fixed.
/// vA needs to find a color.
/// None are available.
/// vA cannot evict vC: vC is a fixed virtual register now.
/// vA does as if vB was evicted => vA uses R2.
/// vB needs to find a color.
/// R3 is available.
/// Recoloring => vC = R1, vA = R2, vB = R3
///
/// \p Order defines the preferred allocation order for \p VirtReg.
/// \p NewRegs will contain any new virtual register that have been created
/// (split, spill) during the process and that must be assigned.
/// \p FixedRegisters contains all the virtual registers that cannot be
/// recolored.
///
/// \p RecolorStack tracks the original assignments of successfully recolored
/// registers.
///
/// \p Depth gives the current depth of the last chance recoloring.
/// \return a physical register that can be used for VirtReg or ~0u if none
/// exists.
unsigned RAGreedy::tryLastChanceRecoloring(const LiveInterval &VirtReg,
AllocationOrder &Order,
SmallVectorImpl<Register> &NewVRegs,
SmallVirtRegSet &FixedRegisters,
RecoloringStack &RecolorStack,
unsigned Depth) {
if (!TRI->shouldUseLastChanceRecoloringForVirtReg(*MF, VirtReg))
return ~0u;
LLVM_DEBUG(dbgs() << "Try last chance recoloring for " << VirtReg << '\n');
const ssize_t EntryStackSize = RecolorStack.size();
// Ranges must be Done.
assert((ExtraInfo->getStage(VirtReg) >= RS_Done || !VirtReg.isSpillable()) &&
"Last chance recoloring should really be last chance");
// Set the max depth to LastChanceRecoloringMaxDepth.
// We may want to reconsider that if we end up with a too large search space
// for target with hundreds of registers.
// Indeed, in that case we may want to cut the search space earlier.
if (Depth >= LastChanceRecoloringMaxDepth && !ExhaustiveSearch) {
LLVM_DEBUG(dbgs() << "Abort because max depth has been reached.\n");
CutOffInfo |= CO_Depth;
return ~0u;
}
// Set of Live intervals that will need to be recolored.
SmallLISet RecoloringCandidates;
// Mark VirtReg as fixed, i.e., it will not be recolored pass this point in
// this recoloring "session".
assert(!FixedRegisters.count(VirtReg.reg()));
FixedRegisters.insert(VirtReg.reg());
SmallVector<Register, 4> CurrentNewVRegs;
for (MCRegister PhysReg : Order) {
assert(PhysReg.isValid());
LLVM_DEBUG(dbgs() << "Try to assign: " << VirtReg << " to "
<< printReg(PhysReg, TRI) << '\n');
RecoloringCandidates.clear();
CurrentNewVRegs.clear();
// It is only possible to recolor virtual register interference.
if (Matrix->checkInterference(VirtReg, PhysReg) >
LiveRegMatrix::IK_VirtReg) {
LLVM_DEBUG(
dbgs() << "Some interferences are not with virtual registers.\n");
continue;
}
// Early give up on this PhysReg if it is obvious we cannot recolor all
// the interferences.
if (!mayRecolorAllInterferences(PhysReg, VirtReg, RecoloringCandidates,
FixedRegisters)) {
LLVM_DEBUG(dbgs() << "Some interferences cannot be recolored.\n");
continue;
}
// RecoloringCandidates contains all the virtual registers that interfere
// with VirtReg on PhysReg (or one of its aliases). Enqueue them for
// recoloring and perform the actual recoloring.
PQueue RecoloringQueue;
for (const LiveInterval *RC : RecoloringCandidates) {
Register ItVirtReg = RC->reg();
enqueue(RecoloringQueue, RC);
assert(VRM->hasPhys(ItVirtReg) &&
"Interferences are supposed to be with allocated variables");
// Record the current allocation.
RecolorStack.push_back(std::make_pair(RC, VRM->getPhys(ItVirtReg)));
// unset the related struct.
Matrix->unassign(*RC);
}
// Do as if VirtReg was assigned to PhysReg so that the underlying
// recoloring has the right information about the interferes and
// available colors.
Matrix->assign(VirtReg, PhysReg);
// Save the current recoloring state.
// If we cannot recolor all the interferences, we will have to start again
// at this point for the next physical register.
SmallVirtRegSet SaveFixedRegisters(FixedRegisters);
if (tryRecoloringCandidates(RecoloringQueue, CurrentNewVRegs,
FixedRegisters, RecolorStack, Depth)) {
// Push the queued vregs into the main queue.
for (Register NewVReg : CurrentNewVRegs)
NewVRegs.push_back(NewVReg);
// Do not mess up with the global assignment process.
// I.e., VirtReg must be unassigned.
Matrix->unassign(VirtReg);
return PhysReg;
}
LLVM_DEBUG(dbgs() << "Fail to assign: " << VirtReg << " to "
<< printReg(PhysReg, TRI) << '\n');
// The recoloring attempt failed, undo the changes.
FixedRegisters = SaveFixedRegisters;
Matrix->unassign(VirtReg);
// For a newly created vreg which is also in RecoloringCandidates,
// don't add it to NewVRegs because its physical register will be restored
// below. Other vregs in CurrentNewVRegs are created by calling
// selectOrSplit and should be added into NewVRegs.
for (Register &R : CurrentNewVRegs) {
if (RecoloringCandidates.count(&LIS->getInterval(R)))
continue;
NewVRegs.push_back(R);
}
// Roll back our unsuccessful recoloring. Also roll back any successful
// recolorings in any recursive recoloring attempts, since it's possible
// they would have introduced conflicts with assignments we will be
// restoring further up the stack. Perform all unassignments prior to
// reassigning, since sub-recolorings may have conflicted with the registers
// we are going to restore to their original assignments.
for (ssize_t I = RecolorStack.size() - 1; I >= EntryStackSize; --I) {
const LiveInterval *LI;
MCRegister PhysReg;
std::tie(LI, PhysReg) = RecolorStack[I];
if (VRM->hasPhys(LI->reg()))
Matrix->unassign(*LI);
}
for (size_t I = EntryStackSize; I != RecolorStack.size(); ++I) {
const LiveInterval *LI;
MCRegister PhysReg;
std::tie(LI, PhysReg) = RecolorStack[I];
if (!LI->empty() && !MRI->reg_nodbg_empty(LI->reg()))
Matrix->assign(*LI, PhysReg);
}
// Pop the stack of recoloring attempts.
RecolorStack.resize(EntryStackSize);
}
// Last chance recoloring did not worked either, give up.
return ~0u;
}
/// tryRecoloringCandidates - Try to assign a new color to every register
/// in \RecoloringQueue.
/// \p NewRegs will contain any new virtual register created during the
/// recoloring process.
/// \p FixedRegisters[in/out] contains all the registers that have been
/// recolored.
/// \return true if all virtual registers in RecoloringQueue were successfully
/// recolored, false otherwise.
bool RAGreedy::tryRecoloringCandidates(PQueue &RecoloringQueue,
SmallVectorImpl<Register> &NewVRegs,
SmallVirtRegSet &FixedRegisters,
RecoloringStack &RecolorStack,
unsigned Depth) {
while (!RecoloringQueue.empty()) {
const LiveInterval *LI = dequeue(RecoloringQueue);
LLVM_DEBUG(dbgs() << "Try to recolor: " << *LI << '\n');
MCRegister PhysReg = selectOrSplitImpl(*LI, NewVRegs, FixedRegisters,
RecolorStack, Depth + 1);
// When splitting happens, the live-range may actually be empty.
// In that case, this is okay to continue the recoloring even
// if we did not find an alternative color for it. Indeed,
// there will not be anything to color for LI in the end.
if (PhysReg == ~0u || (!PhysReg && !LI->empty()))
return false;
if (!PhysReg) {
assert(LI->empty() && "Only empty live-range do not require a register");
LLVM_DEBUG(dbgs() << "Recoloring of " << *LI
<< " succeeded. Empty LI.\n");
continue;
}
LLVM_DEBUG(dbgs() << "Recoloring of " << *LI
<< " succeeded with: " << printReg(PhysReg, TRI) << '\n');
Matrix->assign(*LI, PhysReg);
FixedRegisters.insert(LI->reg());
}
return true;
}
//===----------------------------------------------------------------------===//
// Main Entry Point
//===----------------------------------------------------------------------===//
MCRegister RAGreedy::selectOrSplit(const LiveInterval &VirtReg,
SmallVectorImpl<Register> &NewVRegs) {
CutOffInfo = CO_None;
LLVMContext &Ctx = MF->getFunction().getContext();
SmallVirtRegSet FixedRegisters;
RecoloringStack RecolorStack;
MCRegister Reg =
selectOrSplitImpl(VirtReg, NewVRegs, FixedRegisters, RecolorStack);
if (Reg == ~0U && (CutOffInfo != CO_None)) {
uint8_t CutOffEncountered = CutOffInfo & (CO_Depth | CO_Interf);
if (CutOffEncountered == CO_Depth)
Ctx.emitError("register allocation failed: maximum depth for recoloring "
"reached. Use -fexhaustive-register-search to skip "
"cutoffs");
else if (CutOffEncountered == CO_Interf)
Ctx.emitError("register allocation failed: maximum interference for "
"recoloring reached. Use -fexhaustive-register-search "
"to skip cutoffs");
else if (CutOffEncountered == (CO_Depth | CO_Interf))
Ctx.emitError("register allocation failed: maximum interference and "
"depth for recoloring reached. Use "
"-fexhaustive-register-search to skip cutoffs");
}
return Reg;
}
/// Using a CSR for the first time has a cost because it causes push|pop
/// to be added to prologue|epilogue. Splitting a cold section of the live
/// range can have lower cost than using the CSR for the first time;
/// Spilling a live range in the cold path can have lower cost than using
/// the CSR for the first time. Returns the physical register if we decide
/// to use the CSR; otherwise return 0.
MCRegister RAGreedy::tryAssignCSRFirstTime(
const LiveInterval &VirtReg, AllocationOrder &Order, MCRegister PhysReg,
uint8_t &CostPerUseLimit, SmallVectorImpl<Register> &NewVRegs) {
if (ExtraInfo->getStage(VirtReg) == RS_Spill && VirtReg.isSpillable()) {
// We choose spill over using the CSR for the first time if the spill cost
// is lower than CSRCost.
SA->analyze(&VirtReg);
if (calcSpillCost() >= CSRCost)
return PhysReg;
// We are going to spill, set CostPerUseLimit to 1 to make sure that
// we will not use a callee-saved register in tryEvict.
CostPerUseLimit = 1;
return 0;
}
if (ExtraInfo->getStage(VirtReg) < RS_Split) {
// We choose pre-splitting over using the CSR for the first time if
// the cost of splitting is lower than CSRCost.
SA->analyze(&VirtReg);
unsigned NumCands = 0;
BlockFrequency BestCost = CSRCost; // Don't modify CSRCost.
unsigned BestCand = calculateRegionSplitCost(VirtReg, Order, BestCost,
NumCands, true /*IgnoreCSR*/);
if (BestCand == NoCand)
// Use the CSR if we can't find a region split below CSRCost.
return PhysReg;
// Perform the actual pre-splitting.
doRegionSplit(VirtReg, BestCand, false/*HasCompact*/, NewVRegs);
return 0;
}
return PhysReg;
}
void RAGreedy::aboutToRemoveInterval(const LiveInterval &LI) {
// Do not keep invalid information around.
SetOfBrokenHints.remove(&LI);
}
void RAGreedy::initializeCSRCost() {
// We use the larger one out of the command-line option and the value report
// by TRI.
CSRCost = BlockFrequency(
std::max((unsigned)CSRFirstTimeCost, TRI->getCSRFirstUseCost()));
if (!CSRCost.getFrequency())
return;
// Raw cost is relative to Entry == 2^14; scale it appropriately.
uint64_t ActualEntry = MBFI->getEntryFreq();
if (!ActualEntry) {
CSRCost = 0;
return;
}
uint64_t FixedEntry = 1 << 14;
if (ActualEntry < FixedEntry)
CSRCost *= BranchProbability(ActualEntry, FixedEntry);
else if (ActualEntry <= UINT32_MAX)
// Invert the fraction and divide.
CSRCost /= BranchProbability(FixedEntry, ActualEntry);
else
// Can't use BranchProbability in general, since it takes 32-bit numbers.
CSRCost = CSRCost.getFrequency() * (ActualEntry / FixedEntry);
}
/// Collect the hint info for \p Reg.
/// The results are stored into \p Out.
/// \p Out is not cleared before being populated.
void RAGreedy::collectHintInfo(Register Reg, HintsInfo &Out) {
for (const MachineInstr &Instr : MRI->reg_nodbg_instructions(Reg)) {
if (!Instr.isFullCopy())
continue;
// Look for the other end of the copy.
Register OtherReg = Instr.getOperand(0).getReg();
if (OtherReg == Reg) {
OtherReg = Instr.getOperand(1).getReg();
if (OtherReg == Reg)
continue;
}
// Get the current assignment.
MCRegister OtherPhysReg =
OtherReg.isPhysical() ? OtherReg.asMCReg() : VRM->getPhys(OtherReg);
// Push the collected information.
Out.push_back(HintInfo(MBFI->getBlockFreq(Instr.getParent()), OtherReg,
OtherPhysReg));
}
}
/// Using the given \p List, compute the cost of the broken hints if
/// \p PhysReg was used.
/// \return The cost of \p List for \p PhysReg.
BlockFrequency RAGreedy::getBrokenHintFreq(const HintsInfo &List,
MCRegister PhysReg) {
BlockFrequency Cost = 0;
for (const HintInfo &Info : List) {
if (Info.PhysReg != PhysReg)
Cost += Info.Freq;
}
return Cost;
}
/// Using the register assigned to \p VirtReg, try to recolor
/// all the live ranges that are copy-related with \p VirtReg.
/// The recoloring is then propagated to all the live-ranges that have
/// been recolored and so on, until no more copies can be coalesced or
/// it is not profitable.
/// For a given live range, profitability is determined by the sum of the
/// frequencies of the non-identity copies it would introduce with the old
/// and new register.
void RAGreedy::tryHintRecoloring(const LiveInterval &VirtReg) {
// We have a broken hint, check if it is possible to fix it by
// reusing PhysReg for the copy-related live-ranges. Indeed, we evicted
// some register and PhysReg may be available for the other live-ranges.
SmallSet<Register, 4> Visited;
SmallVector<unsigned, 2> RecoloringCandidates;
HintsInfo Info;
Register Reg = VirtReg.reg();
MCRegister PhysReg = VRM->getPhys(Reg);
// Start the recoloring algorithm from the input live-interval, then
// it will propagate to the ones that are copy-related with it.
Visited.insert(Reg);
RecoloringCandidates.push_back(Reg);
LLVM_DEBUG(dbgs() << "Trying to reconcile hints for: " << printReg(Reg, TRI)
<< '(' << printReg(PhysReg, TRI) << ")\n");
do {
Reg = RecoloringCandidates.pop_back_val();
// We cannot recolor physical register.
if (Reg.isPhysical())
continue;
// This may be a skipped class
if (!VRM->hasPhys(Reg)) {
assert(!ShouldAllocateClass(*TRI, *MRI->getRegClass(Reg)) &&
"We have an unallocated variable which should have been handled");
continue;
}
// Get the live interval mapped with this virtual register to be able
// to check for the interference with the new color.
LiveInterval &LI = LIS->getInterval(Reg);
MCRegister CurrPhys = VRM->getPhys(Reg);
// Check that the new color matches the register class constraints and
// that it is free for this live range.
if (CurrPhys != PhysReg && (!MRI->getRegClass(Reg)->contains(PhysReg) ||
Matrix->checkInterference(LI, PhysReg)))
continue;
LLVM_DEBUG(dbgs() << printReg(Reg, TRI) << '(' << printReg(CurrPhys, TRI)
<< ") is recolorable.\n");
// Gather the hint info.
Info.clear();
collectHintInfo(Reg, Info);
// Check if recoloring the live-range will increase the cost of the
// non-identity copies.
if (CurrPhys != PhysReg) {
LLVM_DEBUG(dbgs() << "Checking profitability:\n");
BlockFrequency OldCopiesCost = getBrokenHintFreq(Info, CurrPhys);
BlockFrequency NewCopiesCost = getBrokenHintFreq(Info, PhysReg);
LLVM_DEBUG(dbgs() << "Old Cost: " << OldCopiesCost.getFrequency()
<< "\nNew Cost: " << NewCopiesCost.getFrequency()
<< '\n');
if (OldCopiesCost < NewCopiesCost) {
LLVM_DEBUG(dbgs() << "=> Not profitable.\n");
continue;
}
// At this point, the cost is either cheaper or equal. If it is
// equal, we consider this is profitable because it may expose
// more recoloring opportunities.
LLVM_DEBUG(dbgs() << "=> Profitable.\n");
// Recolor the live-range.
Matrix->unassign(LI);
Matrix->assign(LI, PhysReg);
}
// Push all copy-related live-ranges to keep reconciling the broken
// hints.
for (const HintInfo &HI : Info) {
if (Visited.insert(HI.Reg).second)
RecoloringCandidates.push_back(HI.Reg);
}
} while (!RecoloringCandidates.empty());
}
/// Try to recolor broken hints.
/// Broken hints may be repaired by recoloring when an evicted variable
/// freed up a register for a larger live-range.
/// Consider the following example:
/// BB1:
/// a =
/// b =
/// BB2:
/// ...
/// = b
/// = a
/// Let us assume b gets split:
/// BB1:
/// a =
/// b =
/// BB2:
/// c = b
/// ...
/// d = c
/// = d
/// = a
/// Because of how the allocation work, b, c, and d may be assigned different
/// colors. Now, if a gets evicted later:
/// BB1:
/// a =
/// st a, SpillSlot
/// b =
/// BB2:
/// c = b
/// ...
/// d = c
/// = d
/// e = ld SpillSlot
/// = e
/// This is likely that we can assign the same register for b, c, and d,
/// getting rid of 2 copies.
void RAGreedy::tryHintsRecoloring() {
for (const LiveInterval *LI : SetOfBrokenHints) {
assert(LI->reg().isVirtual() &&
"Recoloring is possible only for virtual registers");
// Some dead defs may be around (e.g., because of debug uses).
// Ignore those.
if (!VRM->hasPhys(LI->reg()))
continue;
tryHintRecoloring(*LI);
}
}
MCRegister RAGreedy::selectOrSplitImpl(const LiveInterval &VirtReg,
SmallVectorImpl<Register> &NewVRegs,
SmallVirtRegSet &FixedRegisters,
RecoloringStack &RecolorStack,
unsigned Depth) {
uint8_t CostPerUseLimit = uint8_t(~0u);
// First try assigning a free register.
auto Order =
AllocationOrder::create(VirtReg.reg(), *VRM, RegClassInfo, Matrix);
if (MCRegister PhysReg =
tryAssign(VirtReg, Order, NewVRegs, FixedRegisters)) {
// When NewVRegs is not empty, we may have made decisions such as evicting
// a virtual register, go with the earlier decisions and use the physical
// register.
if (CSRCost.getFrequency() &&
EvictAdvisor->isUnusedCalleeSavedReg(PhysReg) && NewVRegs.empty()) {
MCRegister CSRReg = tryAssignCSRFirstTime(VirtReg, Order, PhysReg,
CostPerUseLimit, NewVRegs);
if (CSRReg || !NewVRegs.empty())
// Return now if we decide to use a CSR or create new vregs due to
// pre-splitting.
return CSRReg;
} else
return PhysReg;
}
LiveRangeStage Stage = ExtraInfo->getStage(VirtReg);
LLVM_DEBUG(dbgs() << StageName[Stage] << " Cascade "
<< ExtraInfo->getCascade(VirtReg.reg()) << '\n');
// Try to evict a less worthy live range, but only for ranges from the primary
// queue. The RS_Split ranges already failed to do this, and they should not
// get a second chance until they have been split.
if (Stage != RS_Split)
if (Register PhysReg =
tryEvict(VirtReg, Order, NewVRegs, CostPerUseLimit,
FixedRegisters)) {
Register Hint = MRI->getSimpleHint(VirtReg.reg());
// If VirtReg has a hint and that hint is broken record this
// virtual register as a recoloring candidate for broken hint.
// Indeed, since we evicted a variable in its neighborhood it is
// likely we can at least partially recolor some of the
// copy-related live-ranges.
if (Hint && Hint != PhysReg)
SetOfBrokenHints.insert(&VirtReg);
return PhysReg;
}
assert((NewVRegs.empty() || Depth) && "Cannot append to existing NewVRegs");
// The first time we see a live range, don't try to split or spill.
// Wait until the second time, when all smaller ranges have been allocated.
// This gives a better picture of the interference to split around.
if (Stage < RS_Split) {
ExtraInfo->setStage(VirtReg, RS_Split);
LLVM_DEBUG(dbgs() << "wait for second round\n");
NewVRegs.push_back(VirtReg.reg());
return 0;
}
if (Stage < RS_Spill) {
// Try splitting VirtReg or interferences.
unsigned NewVRegSizeBefore = NewVRegs.size();
Register PhysReg = trySplit(VirtReg, Order, NewVRegs, FixedRegisters);
if (PhysReg || (NewVRegs.size() - NewVRegSizeBefore))
return PhysReg;
}
// If we couldn't allocate a register from spilling, there is probably some
// invalid inline assembly. The base class will report it.
if (Stage >= RS_Done || !VirtReg.isSpillable()) {
return tryLastChanceRecoloring(VirtReg, Order, NewVRegs, FixedRegisters,
RecolorStack, Depth);
}
// Finally spill VirtReg itself.
if ((EnableDeferredSpilling ||
TRI->shouldUseDeferredSpillingForVirtReg(*MF, VirtReg)) &&
ExtraInfo->getStage(VirtReg) < RS_Memory) {
// TODO: This is experimental and in particular, we do not model
// the live range splitting done by spilling correctly.
// We would need a deep integration with the spiller to do the
// right thing here. Anyway, that is still good for early testing.
ExtraInfo->setStage(VirtReg, RS_Memory);
LLVM_DEBUG(dbgs() << "Do as if this register is in memory\n");
NewVRegs.push_back(VirtReg.reg());
} else {
NamedRegionTimer T("spill", "Spiller", TimerGroupName,
TimerGroupDescription, TimePassesIsEnabled);
LiveRangeEdit LRE(&VirtReg, NewVRegs, *MF, *LIS, VRM, this, &DeadRemats);
spiller().spill(LRE);
ExtraInfo->setStage(NewVRegs.begin(), NewVRegs.end(), RS_Done);
// Tell LiveDebugVariables about the new ranges. Ranges not being covered by
// the new regs are kept in LDV (still mapping to the old register), until
// we rewrite spilled locations in LDV at a later stage.
DebugVars->splitRegister(VirtReg.reg(), LRE.regs(), *LIS);
if (VerifyEnabled)
MF->verify(this, "After spilling");
}
// The live virtual register requesting allocation was spilled, so tell
// the caller not to allocate anything during this round.
return 0;
}
void RAGreedy::RAGreedyStats::report(MachineOptimizationRemarkMissed &R) {
using namespace ore;
if (Spills) {
R << NV("NumSpills", Spills) << " spills ";
R << NV("TotalSpillsCost", SpillsCost) << " total spills cost ";
}
if (FoldedSpills) {
R << NV("NumFoldedSpills", FoldedSpills) << " folded spills ";
R << NV("TotalFoldedSpillsCost", FoldedSpillsCost)
<< " total folded spills cost ";
}
if (Reloads) {
R << NV("NumReloads", Reloads) << " reloads ";
R << NV("TotalReloadsCost", ReloadsCost) << " total reloads cost ";
}
if (FoldedReloads) {
R << NV("NumFoldedReloads", FoldedReloads) << " folded reloads ";
R << NV("TotalFoldedReloadsCost", FoldedReloadsCost)
<< " total folded reloads cost ";
}
if (ZeroCostFoldedReloads)
R << NV("NumZeroCostFoldedReloads", ZeroCostFoldedReloads)
<< " zero cost folded reloads ";
if (Copies) {
R << NV("NumVRCopies", Copies) << " virtual registers copies ";
R << NV("TotalCopiesCost", CopiesCost) << " total copies cost ";
}
}
RAGreedy::RAGreedyStats RAGreedy::computeStats(MachineBasicBlock &MBB) {
RAGreedyStats Stats;
const MachineFrameInfo &MFI = MF->getFrameInfo();
int FI;
auto isSpillSlotAccess = [&MFI](const MachineMemOperand *A) {
return MFI.isSpillSlotObjectIndex(cast<FixedStackPseudoSourceValue>(
A->getPseudoValue())->getFrameIndex());
};
auto isPatchpointInstr = [](const MachineInstr &MI) {
return MI.getOpcode() == TargetOpcode::PATCHPOINT ||
MI.getOpcode() == TargetOpcode::STACKMAP ||
MI.getOpcode() == TargetOpcode::STATEPOINT;
};
for (MachineInstr &MI : MBB) {
if (MI.isCopy()) {
const MachineOperand &Dest = MI.getOperand(0);
const MachineOperand &Src = MI.getOperand(1);
Register SrcReg = Src.getReg();
Register DestReg = Dest.getReg();
// Only count `COPY`s with a virtual register as source or destination.
if (SrcReg.isVirtual() || DestReg.isVirtual()) {
if (SrcReg.isVirtual()) {
SrcReg = VRM->getPhys(SrcReg);
if (Src.getSubReg())
SrcReg = TRI->getSubReg(SrcReg, Src.getSubReg());
}
if (DestReg.isVirtual()) {
DestReg = VRM->getPhys(DestReg);
if (Dest.getSubReg())
DestReg = TRI->getSubReg(DestReg, Dest.getSubReg());
}
if (SrcReg != DestReg)
++Stats.Copies;
}
continue;
}
SmallVector<const MachineMemOperand *, 2> Accesses;
if (TII->isLoadFromStackSlot(MI, FI) && MFI.isSpillSlotObjectIndex(FI)) {
++Stats.Reloads;
continue;
}
if (TII->isStoreToStackSlot(MI, FI) && MFI.isSpillSlotObjectIndex(FI)) {
++Stats.Spills;
continue;
}
if (TII->hasLoadFromStackSlot(MI, Accesses) &&
llvm::any_of(Accesses, isSpillSlotAccess)) {
if (!isPatchpointInstr(MI)) {
Stats.FoldedReloads += Accesses.size();
continue;
}
// For statepoint there may be folded and zero cost folded stack reloads.
std::pair<unsigned, unsigned> NonZeroCostRange =
TII->getPatchpointUnfoldableRange(MI);
SmallSet<unsigned, 16> FoldedReloads;
SmallSet<unsigned, 16> ZeroCostFoldedReloads;
for (unsigned Idx = 0, E = MI.getNumOperands(); Idx < E; ++Idx) {
MachineOperand &MO = MI.getOperand(Idx);
if (!MO.isFI() || !MFI.isSpillSlotObjectIndex(MO.getIndex()))
continue;
if (Idx >= NonZeroCostRange.first && Idx < NonZeroCostRange.second)
FoldedReloads.insert(MO.getIndex());
else
ZeroCostFoldedReloads.insert(MO.getIndex());
}
// If stack slot is used in folded reload it is not zero cost then.
for (unsigned Slot : FoldedReloads)
ZeroCostFoldedReloads.erase(Slot);
Stats.FoldedReloads += FoldedReloads.size();
Stats.ZeroCostFoldedReloads += ZeroCostFoldedReloads.size();
continue;
}
Accesses.clear();
if (TII->hasStoreToStackSlot(MI, Accesses) &&
llvm::any_of(Accesses, isSpillSlotAccess)) {
Stats.FoldedSpills += Accesses.size();
}
}
// Set cost of collected statistic by multiplication to relative frequency of
// this basic block.
float RelFreq = MBFI->getBlockFreqRelativeToEntryBlock(&MBB);
Stats.ReloadsCost = RelFreq * Stats.Reloads;
Stats.FoldedReloadsCost = RelFreq * Stats.FoldedReloads;
Stats.SpillsCost = RelFreq * Stats.Spills;
Stats.FoldedSpillsCost = RelFreq * Stats.FoldedSpills;
Stats.CopiesCost = RelFreq * Stats.Copies;
return Stats;
}
RAGreedy::RAGreedyStats RAGreedy::reportStats(MachineLoop *L) {
RAGreedyStats Stats;
// Sum up the spill and reloads in subloops.
for (MachineLoop *SubLoop : *L)
Stats.add(reportStats(SubLoop));
for (MachineBasicBlock *MBB : L->getBlocks())
// Handle blocks that were not included in subloops.
if (Loops->getLoopFor(MBB) == L)
Stats.add(computeStats(*MBB));
if (!Stats.isEmpty()) {
using namespace ore;
ORE->emit([&]() {
MachineOptimizationRemarkMissed R(DEBUG_TYPE, "LoopSpillReloadCopies",
L->getStartLoc(), L->getHeader());
Stats.report(R);
R << "generated in loop";
return R;
});
}
return Stats;
}
void RAGreedy::reportStats() {
if (!ORE->allowExtraAnalysis(DEBUG_TYPE))
return;
RAGreedyStats Stats;
for (MachineLoop *L : *Loops)
Stats.add(reportStats(L));
// Process non-loop blocks.
for (MachineBasicBlock &MBB : *MF)
if (!Loops->getLoopFor(&MBB))
Stats.add(computeStats(MBB));
if (!Stats.isEmpty()) {
using namespace ore;
ORE->emit([&]() {
DebugLoc Loc;
if (auto *SP = MF->getFunction().getSubprogram())
Loc = DILocation::get(SP->getContext(), SP->getLine(), 1, SP);
MachineOptimizationRemarkMissed R(DEBUG_TYPE, "SpillReloadCopies", Loc,
&MF->front());
Stats.report(R);
R << "generated in function";
return R;
});
}
}
bool RAGreedy::hasVirtRegAlloc() {
for (unsigned I = 0, E = MRI->getNumVirtRegs(); I != E; ++I) {
Register Reg = Register::index2VirtReg(I);
if (MRI->reg_nodbg_empty(Reg))
continue;
const TargetRegisterClass *RC = MRI->getRegClass(Reg);
if (!RC)
continue;
if (ShouldAllocateClass(*TRI, *RC))
return true;
}
return false;
}
bool RAGreedy::runOnMachineFunction(MachineFunction &mf) {
LLVM_DEBUG(dbgs() << "********** GREEDY REGISTER ALLOCATION **********\n"
<< "********** Function: " << mf.getName() << '\n');
MF = &mf;
TII = MF->getSubtarget().getInstrInfo();
if (VerifyEnabled)
MF->verify(this, "Before greedy register allocator");
RegAllocBase::init(getAnalysis<VirtRegMap>(),
getAnalysis<LiveIntervals>(),
getAnalysis<LiveRegMatrix>());
// Early return if there is no virtual register to be allocated to a
// physical register.
if (!hasVirtRegAlloc())
return false;
Indexes = &getAnalysis<SlotIndexes>();
MBFI = &getAnalysis<MachineBlockFrequencyInfo>();
DomTree = &getAnalysis<MachineDominatorTree>();
ORE = &getAnalysis<MachineOptimizationRemarkEmitterPass>().getORE();
Loops = &getAnalysis<MachineLoopInfo>();
Bundles = &getAnalysis<EdgeBundles>();
SpillPlacer = &getAnalysis<SpillPlacement>();
DebugVars = &getAnalysis<LiveDebugVariables>();
initializeCSRCost();
RegCosts = TRI->getRegisterCosts(*MF);
RegClassPriorityTrumpsGlobalness =
GreedyRegClassPriorityTrumpsGlobalness.getNumOccurrences()
? GreedyRegClassPriorityTrumpsGlobalness
: TRI->regClassPriorityTrumpsGlobalness(*MF);
ReverseLocalAssignment = GreedyReverseLocalAssignment.getNumOccurrences()
? GreedyReverseLocalAssignment
: TRI->reverseLocalAssignment();
ExtraInfo.emplace();
EvictAdvisor =
getAnalysis<RegAllocEvictionAdvisorAnalysis>().getAdvisor(*MF, *this);
PriorityAdvisor =
getAnalysis<RegAllocPriorityAdvisorAnalysis>().getAdvisor(*MF, *this);
VRAI = std::make_unique<VirtRegAuxInfo>(*MF, *LIS, *VRM, *Loops, *MBFI);
SpillerInstance.reset(createInlineSpiller(*this, *MF, *VRM, *VRAI));
VRAI->calculateSpillWeightsAndHints();
LLVM_DEBUG(LIS->dump());
SA.reset(new SplitAnalysis(*VRM, *LIS, *Loops));
SE.reset(new SplitEditor(*SA, *LIS, *VRM, *DomTree, *MBFI, *VRAI));
IntfCache.init(MF, Matrix->getLiveUnions(), Indexes, LIS, TRI);
GlobalCand.resize(32); // This will grow as needed.
SetOfBrokenHints.clear();
allocatePhysRegs();
tryHintsRecoloring();
if (VerifyEnabled)
MF->verify(this, "Before post optimization");
postOptimization();
reportStats();
releaseMemory();
return true;
}