//===- MachineScheduler.cpp - Machine Instruction Scheduler ---------------===// // // 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 // //===----------------------------------------------------------------------===// // // MachineScheduler schedules machine instructions after phi elimination. It // preserves LiveIntervals so it can be invoked before register allocation. // //===----------------------------------------------------------------------===// #include "llvm/CodeGen/MachineScheduler.h" #include "llvm/ADT/ArrayRef.h" #include "llvm/ADT/BitVector.h" #include "llvm/ADT/DenseMap.h" #include "llvm/ADT/PriorityQueue.h" #include "llvm/ADT/STLExtras.h" #include "llvm/ADT/SmallVector.h" #include "llvm/ADT/Statistic.h" #include "llvm/ADT/iterator_range.h" #include "llvm/Analysis/AliasAnalysis.h" #include "llvm/CodeGen/LiveInterval.h" #include "llvm/CodeGen/LiveIntervals.h" #include "llvm/CodeGen/MachineBasicBlock.h" #include "llvm/CodeGen/MachineDominators.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/MachinePassRegistry.h" #include "llvm/CodeGen/MachineRegisterInfo.h" #include "llvm/CodeGen/Passes.h" #include "llvm/CodeGen/RegisterClassInfo.h" #include "llvm/CodeGen/RegisterPressure.h" #include "llvm/CodeGen/ScheduleDAG.h" #include "llvm/CodeGen/ScheduleDAGInstrs.h" #include "llvm/CodeGen/ScheduleDAGMutation.h" #include "llvm/CodeGen/ScheduleDFS.h" #include "llvm/CodeGen/ScheduleHazardRecognizer.h" #include "llvm/CodeGen/SlotIndexes.h" #include "llvm/CodeGen/TargetFrameLowering.h" #include "llvm/CodeGen/TargetInstrInfo.h" #include "llvm/CodeGen/TargetLowering.h" #include "llvm/CodeGen/TargetPassConfig.h" #include "llvm/CodeGen/TargetRegisterInfo.h" #include "llvm/CodeGen/TargetSchedule.h" #include "llvm/CodeGen/TargetSubtargetInfo.h" #include "llvm/Config/llvm-config.h" #include "llvm/InitializePasses.h" #include "llvm/MC/LaneBitmask.h" #include "llvm/Pass.h" #include "llvm/Support/CommandLine.h" #include "llvm/Support/Compiler.h" #include "llvm/Support/Debug.h" #include "llvm/Support/ErrorHandling.h" #include "llvm/Support/GraphWriter.h" #include "llvm/Support/MachineValueType.h" #include "llvm/Support/raw_ostream.h" #include #include #include #include #include #include #include #include #include #include using namespace llvm; #define DEBUG_TYPE "machine-scheduler" STATISTIC(NumClustered, "Number of load/store pairs clustered"); namespace llvm { cl::opt ForceTopDown("misched-topdown", cl::Hidden, cl::desc("Force top-down list scheduling")); cl::opt ForceBottomUp("misched-bottomup", cl::Hidden, cl::desc("Force bottom-up list scheduling")); cl::opt DumpCriticalPathLength("misched-dcpl", cl::Hidden, cl::desc("Print critical path length to stdout")); cl::opt VerifyScheduling( "verify-misched", cl::Hidden, cl::desc("Verify machine instrs before and after machine scheduling")); } // end namespace llvm #ifndef NDEBUG static cl::opt ViewMISchedDAGs("view-misched-dags", cl::Hidden, cl::desc("Pop up a window to show MISched dags after they are processed")); /// In some situations a few uninteresting nodes depend on nearly all other /// nodes in the graph, provide a cutoff to hide them. static cl::opt ViewMISchedCutoff("view-misched-cutoff", cl::Hidden, cl::desc("Hide nodes with more predecessor/successor than cutoff")); static cl::opt MISchedCutoff("misched-cutoff", cl::Hidden, cl::desc("Stop scheduling after N instructions"), cl::init(~0U)); static cl::opt SchedOnlyFunc("misched-only-func", cl::Hidden, cl::desc("Only schedule this function")); static cl::opt SchedOnlyBlock("misched-only-block", cl::Hidden, cl::desc("Only schedule this MBB#")); static cl::opt PrintDAGs("misched-print-dags", cl::Hidden, cl::desc("Print schedule DAGs")); #else static const bool ViewMISchedDAGs = false; static const bool PrintDAGs = false; #endif // NDEBUG /// Avoid quadratic complexity in unusually large basic blocks by limiting the /// size of the ready lists. static cl::opt ReadyListLimit("misched-limit", cl::Hidden, cl::desc("Limit ready list to N instructions"), cl::init(256)); static cl::opt EnableRegPressure("misched-regpressure", cl::Hidden, cl::desc("Enable register pressure scheduling."), cl::init(true)); static cl::opt EnableCyclicPath("misched-cyclicpath", cl::Hidden, cl::desc("Enable cyclic critical path analysis."), cl::init(true)); static cl::opt EnableMemOpCluster("misched-cluster", cl::Hidden, cl::desc("Enable memop clustering."), cl::init(true)); // DAG subtrees must have at least this many nodes. static const unsigned MinSubtreeSize = 8; // Pin the vtables to this file. void MachineSchedStrategy::anchor() {} void ScheduleDAGMutation::anchor() {} //===----------------------------------------------------------------------===// // Machine Instruction Scheduling Pass and Registry //===----------------------------------------------------------------------===// MachineSchedContext::MachineSchedContext() { RegClassInfo = new RegisterClassInfo(); } MachineSchedContext::~MachineSchedContext() { delete RegClassInfo; } namespace { /// Base class for a machine scheduler class that can run at any point. class MachineSchedulerBase : public MachineSchedContext, public MachineFunctionPass { public: MachineSchedulerBase(char &ID): MachineFunctionPass(ID) {} void print(raw_ostream &O, const Module* = nullptr) const override; protected: void scheduleRegions(ScheduleDAGInstrs &Scheduler, bool FixKillFlags); }; /// MachineScheduler runs after coalescing and before register allocation. class MachineScheduler : public MachineSchedulerBase { public: MachineScheduler(); void getAnalysisUsage(AnalysisUsage &AU) const override; bool runOnMachineFunction(MachineFunction&) override; static char ID; // Class identification, replacement for typeinfo protected: ScheduleDAGInstrs *createMachineScheduler(); }; /// PostMachineScheduler runs after shortly before code emission. class PostMachineScheduler : public MachineSchedulerBase { public: PostMachineScheduler(); void getAnalysisUsage(AnalysisUsage &AU) const override; bool runOnMachineFunction(MachineFunction&) override; static char ID; // Class identification, replacement for typeinfo protected: ScheduleDAGInstrs *createPostMachineScheduler(); }; } // end anonymous namespace char MachineScheduler::ID = 0; char &llvm::MachineSchedulerID = MachineScheduler::ID; INITIALIZE_PASS_BEGIN(MachineScheduler, DEBUG_TYPE, "Machine Instruction Scheduler", false, false) INITIALIZE_PASS_DEPENDENCY(AAResultsWrapperPass) INITIALIZE_PASS_DEPENDENCY(MachineDominatorTree) INITIALIZE_PASS_DEPENDENCY(MachineLoopInfo) INITIALIZE_PASS_DEPENDENCY(SlotIndexes) INITIALIZE_PASS_DEPENDENCY(LiveIntervals) INITIALIZE_PASS_END(MachineScheduler, DEBUG_TYPE, "Machine Instruction Scheduler", false, false) MachineScheduler::MachineScheduler() : MachineSchedulerBase(ID) { initializeMachineSchedulerPass(*PassRegistry::getPassRegistry()); } void MachineScheduler::getAnalysisUsage(AnalysisUsage &AU) const { AU.setPreservesCFG(); AU.addRequired(); AU.addRequired(); AU.addRequired(); AU.addRequired(); AU.addRequired(); AU.addPreserved(); AU.addRequired(); AU.addPreserved(); MachineFunctionPass::getAnalysisUsage(AU); } char PostMachineScheduler::ID = 0; char &llvm::PostMachineSchedulerID = PostMachineScheduler::ID; INITIALIZE_PASS(PostMachineScheduler, "postmisched", "PostRA Machine Instruction Scheduler", false, false) PostMachineScheduler::PostMachineScheduler() : MachineSchedulerBase(ID) { initializePostMachineSchedulerPass(*PassRegistry::getPassRegistry()); } void PostMachineScheduler::getAnalysisUsage(AnalysisUsage &AU) const { AU.setPreservesCFG(); AU.addRequired(); AU.addRequired(); AU.addRequired(); AU.addRequired(); MachineFunctionPass::getAnalysisUsage(AU); } MachinePassRegistry MachineSchedRegistry::Registry; /// A dummy default scheduler factory indicates whether the scheduler /// is overridden on the command line. static ScheduleDAGInstrs *useDefaultMachineSched(MachineSchedContext *C) { return nullptr; } /// MachineSchedOpt allows command line selection of the scheduler. static cl::opt> MachineSchedOpt("misched", cl::init(&useDefaultMachineSched), cl::Hidden, cl::desc("Machine instruction scheduler to use")); static MachineSchedRegistry DefaultSchedRegistry("default", "Use the target's default scheduler choice.", useDefaultMachineSched); static cl::opt EnableMachineSched( "enable-misched", cl::desc("Enable the machine instruction scheduling pass."), cl::init(true), cl::Hidden); static cl::opt EnablePostRAMachineSched( "enable-post-misched", cl::desc("Enable the post-ra machine instruction scheduling pass."), cl::init(true), cl::Hidden); /// Decrement this iterator until reaching the top or a non-debug instr. static MachineBasicBlock::const_iterator priorNonDebug(MachineBasicBlock::const_iterator I, MachineBasicBlock::const_iterator Beg) { assert(I != Beg && "reached the top of the region, cannot decrement"); while (--I != Beg) { if (!I->isDebugInstr()) break; } return I; } /// Non-const version. static MachineBasicBlock::iterator priorNonDebug(MachineBasicBlock::iterator I, MachineBasicBlock::const_iterator Beg) { return priorNonDebug(MachineBasicBlock::const_iterator(I), Beg) .getNonConstIterator(); } /// If this iterator is a debug value, increment until reaching the End or a /// non-debug instruction. static MachineBasicBlock::const_iterator nextIfDebug(MachineBasicBlock::const_iterator I, MachineBasicBlock::const_iterator End) { for(; I != End; ++I) { if (!I->isDebugInstr()) break; } return I; } /// Non-const version. static MachineBasicBlock::iterator nextIfDebug(MachineBasicBlock::iterator I, MachineBasicBlock::const_iterator End) { return nextIfDebug(MachineBasicBlock::const_iterator(I), End) .getNonConstIterator(); } /// Instantiate a ScheduleDAGInstrs that will be owned by the caller. ScheduleDAGInstrs *MachineScheduler::createMachineScheduler() { // Select the scheduler, or set the default. MachineSchedRegistry::ScheduleDAGCtor Ctor = MachineSchedOpt; if (Ctor != useDefaultMachineSched) return Ctor(this); // Get the default scheduler set by the target for this function. ScheduleDAGInstrs *Scheduler = PassConfig->createMachineScheduler(this); if (Scheduler) return Scheduler; // Default to GenericScheduler. return createGenericSchedLive(this); } /// Instantiate a ScheduleDAGInstrs for PostRA scheduling that will be owned by /// the caller. We don't have a command line option to override the postRA /// scheduler. The Target must configure it. ScheduleDAGInstrs *PostMachineScheduler::createPostMachineScheduler() { // Get the postRA scheduler set by the target for this function. ScheduleDAGInstrs *Scheduler = PassConfig->createPostMachineScheduler(this); if (Scheduler) return Scheduler; // Default to GenericScheduler. return createGenericSchedPostRA(this); } /// Top-level MachineScheduler pass driver. /// /// Visit blocks in function order. Divide each block into scheduling regions /// and visit them bottom-up. Visiting regions bottom-up is not required, but is /// consistent with the DAG builder, which traverses the interior of the /// scheduling regions bottom-up. /// /// This design avoids exposing scheduling boundaries to the DAG builder, /// simplifying the DAG builder's support for "special" target instructions. /// At the same time the design allows target schedulers to operate across /// scheduling boundaries, for example to bundle the boundary instructions /// without reordering them. This creates complexity, because the target /// scheduler must update the RegionBegin and RegionEnd positions cached by /// ScheduleDAGInstrs whenever adding or removing instructions. A much simpler /// design would be to split blocks at scheduling boundaries, but LLVM has a /// general bias against block splitting purely for implementation simplicity. bool MachineScheduler::runOnMachineFunction(MachineFunction &mf) { if (skipFunction(mf.getFunction())) return false; if (EnableMachineSched.getNumOccurrences()) { if (!EnableMachineSched) return false; } else if (!mf.getSubtarget().enableMachineScheduler()) return false; LLVM_DEBUG(dbgs() << "Before MISched:\n"; mf.print(dbgs())); // Initialize the context of the pass. MF = &mf; MLI = &getAnalysis(); MDT = &getAnalysis(); PassConfig = &getAnalysis(); AA = &getAnalysis().getAAResults(); LIS = &getAnalysis(); if (VerifyScheduling) { LLVM_DEBUG(LIS->dump()); MF->verify(this, "Before machine scheduling."); } RegClassInfo->runOnMachineFunction(*MF); // Instantiate the selected scheduler for this target, function, and // optimization level. std::unique_ptr Scheduler(createMachineScheduler()); scheduleRegions(*Scheduler, false); LLVM_DEBUG(LIS->dump()); if (VerifyScheduling) MF->verify(this, "After machine scheduling."); return true; } bool PostMachineScheduler::runOnMachineFunction(MachineFunction &mf) { if (skipFunction(mf.getFunction())) return false; if (EnablePostRAMachineSched.getNumOccurrences()) { if (!EnablePostRAMachineSched) return false; } else if (!mf.getSubtarget().enablePostRAMachineScheduler()) { LLVM_DEBUG(dbgs() << "Subtarget disables post-MI-sched.\n"); return false; } LLVM_DEBUG(dbgs() << "Before post-MI-sched:\n"; mf.print(dbgs())); // Initialize the context of the pass. MF = &mf; MLI = &getAnalysis(); PassConfig = &getAnalysis(); AA = &getAnalysis().getAAResults(); if (VerifyScheduling) MF->verify(this, "Before post machine scheduling."); // Instantiate the selected scheduler for this target, function, and // optimization level. std::unique_ptr Scheduler(createPostMachineScheduler()); scheduleRegions(*Scheduler, true); if (VerifyScheduling) MF->verify(this, "After post machine scheduling."); return true; } /// Return true of the given instruction should not be included in a scheduling /// region. /// /// MachineScheduler does not currently support scheduling across calls. To /// handle calls, the DAG builder needs to be modified to create register /// anti/output dependencies on the registers clobbered by the call's regmask /// operand. In PreRA scheduling, the stack pointer adjustment already prevents /// scheduling across calls. In PostRA scheduling, we need the isCall to enforce /// the boundary, but there would be no benefit to postRA scheduling across /// calls this late anyway. static bool isSchedBoundary(MachineBasicBlock::iterator MI, MachineBasicBlock *MBB, MachineFunction *MF, const TargetInstrInfo *TII) { return MI->isCall() || TII->isSchedulingBoundary(*MI, MBB, *MF); } /// A region of an MBB for scheduling. namespace { struct SchedRegion { /// RegionBegin is the first instruction in the scheduling region, and /// RegionEnd is either MBB->end() or the scheduling boundary after the /// last instruction in the scheduling region. These iterators cannot refer /// to instructions outside of the identified scheduling region because /// those may be reordered before scheduling this region. MachineBasicBlock::iterator RegionBegin; MachineBasicBlock::iterator RegionEnd; unsigned NumRegionInstrs; SchedRegion(MachineBasicBlock::iterator B, MachineBasicBlock::iterator E, unsigned N) : RegionBegin(B), RegionEnd(E), NumRegionInstrs(N) {} }; } // end anonymous namespace using MBBRegionsVector = SmallVector; static void getSchedRegions(MachineBasicBlock *MBB, MBBRegionsVector &Regions, bool RegionsTopDown) { MachineFunction *MF = MBB->getParent(); const TargetInstrInfo *TII = MF->getSubtarget().getInstrInfo(); MachineBasicBlock::iterator I = nullptr; for(MachineBasicBlock::iterator RegionEnd = MBB->end(); RegionEnd != MBB->begin(); RegionEnd = I) { // Avoid decrementing RegionEnd for blocks with no terminator. if (RegionEnd != MBB->end() || isSchedBoundary(&*std::prev(RegionEnd), &*MBB, MF, TII)) { --RegionEnd; } // The next region starts above the previous region. Look backward in the // instruction stream until we find the nearest boundary. unsigned NumRegionInstrs = 0; I = RegionEnd; for (;I != MBB->begin(); --I) { MachineInstr &MI = *std::prev(I); if (isSchedBoundary(&MI, &*MBB, MF, TII)) break; if (!MI.isDebugInstr()) { // MBB::size() uses instr_iterator to count. Here we need a bundle to // count as a single instruction. ++NumRegionInstrs; } } // It's possible we found a scheduling region that only has debug // instructions. Don't bother scheduling these. if (NumRegionInstrs != 0) Regions.push_back(SchedRegion(I, RegionEnd, NumRegionInstrs)); } if (RegionsTopDown) std::reverse(Regions.begin(), Regions.end()); } /// Main driver for both MachineScheduler and PostMachineScheduler. void MachineSchedulerBase::scheduleRegions(ScheduleDAGInstrs &Scheduler, bool FixKillFlags) { // Visit all machine basic blocks. // // TODO: Visit blocks in global postorder or postorder within the bottom-up // loop tree. Then we can optionally compute global RegPressure. for (MachineFunction::iterator MBB = MF->begin(), MBBEnd = MF->end(); MBB != MBBEnd; ++MBB) { Scheduler.startBlock(&*MBB); #ifndef NDEBUG if (SchedOnlyFunc.getNumOccurrences() && SchedOnlyFunc != MF->getName()) continue; if (SchedOnlyBlock.getNumOccurrences() && (int)SchedOnlyBlock != MBB->getNumber()) continue; #endif // Break the block into scheduling regions [I, RegionEnd). RegionEnd // points to the scheduling boundary at the bottom of the region. The DAG // does not include RegionEnd, but the region does (i.e. the next // RegionEnd is above the previous RegionBegin). If the current block has // no terminator then RegionEnd == MBB->end() for the bottom region. // // All the regions of MBB are first found and stored in MBBRegions, which // will be processed (MBB) top-down if initialized with true. // // The Scheduler may insert instructions during either schedule() or // exitRegion(), even for empty regions. So the local iterators 'I' and // 'RegionEnd' are invalid across these calls. Instructions must not be // added to other regions than the current one without updating MBBRegions. MBBRegionsVector MBBRegions; getSchedRegions(&*MBB, MBBRegions, Scheduler.doMBBSchedRegionsTopDown()); for (MBBRegionsVector::iterator R = MBBRegions.begin(); R != MBBRegions.end(); ++R) { MachineBasicBlock::iterator I = R->RegionBegin; MachineBasicBlock::iterator RegionEnd = R->RegionEnd; unsigned NumRegionInstrs = R->NumRegionInstrs; // Notify the scheduler of the region, even if we may skip scheduling // it. Perhaps it still needs to be bundled. Scheduler.enterRegion(&*MBB, I, RegionEnd, NumRegionInstrs); // Skip empty scheduling regions (0 or 1 schedulable instructions). if (I == RegionEnd || I == std::prev(RegionEnd)) { // Close the current region. Bundle the terminator if needed. // This invalidates 'RegionEnd' and 'I'. Scheduler.exitRegion(); continue; } LLVM_DEBUG(dbgs() << "********** MI Scheduling **********\n"); LLVM_DEBUG(dbgs() << MF->getName() << ":" << printMBBReference(*MBB) << " " << MBB->getName() << "\n From: " << *I << " To: "; if (RegionEnd != MBB->end()) dbgs() << *RegionEnd; else dbgs() << "End"; dbgs() << " RegionInstrs: " << NumRegionInstrs << '\n'); if (DumpCriticalPathLength) { errs() << MF->getName(); errs() << ":%bb. " << MBB->getNumber(); errs() << " " << MBB->getName() << " \n"; } // Schedule a region: possibly reorder instructions. // This invalidates the original region iterators. Scheduler.schedule(); // Close the current region. Scheduler.exitRegion(); } Scheduler.finishBlock(); // FIXME: Ideally, no further passes should rely on kill flags. However, // thumb2 size reduction is currently an exception, so the PostMIScheduler // needs to do this. if (FixKillFlags) Scheduler.fixupKills(*MBB); } Scheduler.finalizeSchedule(); } void MachineSchedulerBase::print(raw_ostream &O, const Module* m) const { // unimplemented } #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP) LLVM_DUMP_METHOD void ReadyQueue::dump() const { dbgs() << "Queue " << Name << ": "; for (const SUnit *SU : Queue) dbgs() << SU->NodeNum << " "; dbgs() << "\n"; } #endif //===----------------------------------------------------------------------===// // ScheduleDAGMI - Basic machine instruction scheduling. This is // independent of PreRA/PostRA scheduling and involves no extra book-keeping for // virtual registers. // ===----------------------------------------------------------------------===/ // Provide a vtable anchor. ScheduleDAGMI::~ScheduleDAGMI() = default; /// ReleaseSucc - Decrement the NumPredsLeft count of a successor. When /// NumPredsLeft reaches zero, release the successor node. /// /// FIXME: Adjust SuccSU height based on MinLatency. void ScheduleDAGMI::releaseSucc(SUnit *SU, SDep *SuccEdge) { SUnit *SuccSU = SuccEdge->getSUnit(); if (SuccEdge->isWeak()) { --SuccSU->WeakPredsLeft; if (SuccEdge->isCluster()) NextClusterSucc = SuccSU; return; } #ifndef NDEBUG if (SuccSU->NumPredsLeft == 0) { dbgs() << "*** Scheduling failed! ***\n"; dumpNode(*SuccSU); dbgs() << " has been released too many times!\n"; llvm_unreachable(nullptr); } #endif // SU->TopReadyCycle was set to CurrCycle when it was scheduled. However, // CurrCycle may have advanced since then. if (SuccSU->TopReadyCycle < SU->TopReadyCycle + SuccEdge->getLatency()) SuccSU->TopReadyCycle = SU->TopReadyCycle + SuccEdge->getLatency(); --SuccSU->NumPredsLeft; if (SuccSU->NumPredsLeft == 0 && SuccSU != &ExitSU) SchedImpl->releaseTopNode(SuccSU); } /// releaseSuccessors - Call releaseSucc on each of SU's successors. void ScheduleDAGMI::releaseSuccessors(SUnit *SU) { for (SDep &Succ : SU->Succs) releaseSucc(SU, &Succ); } /// ReleasePred - Decrement the NumSuccsLeft count of a predecessor. When /// NumSuccsLeft reaches zero, release the predecessor node. /// /// FIXME: Adjust PredSU height based on MinLatency. void ScheduleDAGMI::releasePred(SUnit *SU, SDep *PredEdge) { SUnit *PredSU = PredEdge->getSUnit(); if (PredEdge->isWeak()) { --PredSU->WeakSuccsLeft; if (PredEdge->isCluster()) NextClusterPred = PredSU; return; } #ifndef NDEBUG if (PredSU->NumSuccsLeft == 0) { dbgs() << "*** Scheduling failed! ***\n"; dumpNode(*PredSU); dbgs() << " has been released too many times!\n"; llvm_unreachable(nullptr); } #endif // SU->BotReadyCycle was set to CurrCycle when it was scheduled. However, // CurrCycle may have advanced since then. if (PredSU->BotReadyCycle < SU->BotReadyCycle + PredEdge->getLatency()) PredSU->BotReadyCycle = SU->BotReadyCycle + PredEdge->getLatency(); --PredSU->NumSuccsLeft; if (PredSU->NumSuccsLeft == 0 && PredSU != &EntrySU) SchedImpl->releaseBottomNode(PredSU); } /// releasePredecessors - Call releasePred on each of SU's predecessors. void ScheduleDAGMI::releasePredecessors(SUnit *SU) { for (SDep &Pred : SU->Preds) releasePred(SU, &Pred); } void ScheduleDAGMI::startBlock(MachineBasicBlock *bb) { ScheduleDAGInstrs::startBlock(bb); SchedImpl->enterMBB(bb); } void ScheduleDAGMI::finishBlock() { SchedImpl->leaveMBB(); ScheduleDAGInstrs::finishBlock(); } /// enterRegion - Called back from MachineScheduler::runOnMachineFunction after /// crossing a scheduling boundary. [begin, end) includes all instructions in /// the region, including the boundary itself and single-instruction regions /// that don't get scheduled. void ScheduleDAGMI::enterRegion(MachineBasicBlock *bb, MachineBasicBlock::iterator begin, MachineBasicBlock::iterator end, unsigned regioninstrs) { ScheduleDAGInstrs::enterRegion(bb, begin, end, regioninstrs); SchedImpl->initPolicy(begin, end, regioninstrs); } /// This is normally called from the main scheduler loop but may also be invoked /// by the scheduling strategy to perform additional code motion. void ScheduleDAGMI::moveInstruction( MachineInstr *MI, MachineBasicBlock::iterator InsertPos) { // Advance RegionBegin if the first instruction moves down. if (&*RegionBegin == MI) ++RegionBegin; // Update the instruction stream. BB->splice(InsertPos, BB, MI); // Update LiveIntervals if (LIS) LIS->handleMove(*MI, /*UpdateFlags=*/true); // Recede RegionBegin if an instruction moves above the first. if (RegionBegin == InsertPos) RegionBegin = MI; } bool ScheduleDAGMI::checkSchedLimit() { #ifndef NDEBUG if (NumInstrsScheduled == MISchedCutoff && MISchedCutoff != ~0U) { CurrentTop = CurrentBottom; return false; } ++NumInstrsScheduled; #endif return true; } /// Per-region scheduling driver, called back from /// MachineScheduler::runOnMachineFunction. This is a simplified driver that /// does not consider liveness or register pressure. It is useful for PostRA /// scheduling and potentially other custom schedulers. void ScheduleDAGMI::schedule() { LLVM_DEBUG(dbgs() << "ScheduleDAGMI::schedule starting\n"); LLVM_DEBUG(SchedImpl->dumpPolicy()); // Build the DAG. buildSchedGraph(AA); postprocessDAG(); SmallVector TopRoots, BotRoots; findRootsAndBiasEdges(TopRoots, BotRoots); LLVM_DEBUG(dump()); if (PrintDAGs) dump(); if (ViewMISchedDAGs) viewGraph(); // Initialize the strategy before modifying the DAG. // This may initialize a DFSResult to be used for queue priority. SchedImpl->initialize(this); // Initialize ready queues now that the DAG and priority data are finalized. initQueues(TopRoots, BotRoots); bool IsTopNode = false; while (true) { LLVM_DEBUG(dbgs() << "** ScheduleDAGMI::schedule picking next node\n"); SUnit *SU = SchedImpl->pickNode(IsTopNode); if (!SU) break; assert(!SU->isScheduled && "Node already scheduled"); if (!checkSchedLimit()) break; MachineInstr *MI = SU->getInstr(); if (IsTopNode) { assert(SU->isTopReady() && "node still has unscheduled dependencies"); if (&*CurrentTop == MI) CurrentTop = nextIfDebug(++CurrentTop, CurrentBottom); else moveInstruction(MI, CurrentTop); } else { assert(SU->isBottomReady() && "node still has unscheduled dependencies"); MachineBasicBlock::iterator priorII = priorNonDebug(CurrentBottom, CurrentTop); if (&*priorII == MI) CurrentBottom = priorII; else { if (&*CurrentTop == MI) CurrentTop = nextIfDebug(++CurrentTop, priorII); moveInstruction(MI, CurrentBottom); CurrentBottom = MI; } } // Notify the scheduling strategy before updating the DAG. // This sets the scheduled node's ReadyCycle to CurrCycle. When updateQueues // runs, it can then use the accurate ReadyCycle time to determine whether // newly released nodes can move to the readyQ. SchedImpl->schedNode(SU, IsTopNode); updateQueues(SU, IsTopNode); } assert(CurrentTop == CurrentBottom && "Nonempty unscheduled zone."); placeDebugValues(); LLVM_DEBUG({ dbgs() << "*** Final schedule for " << printMBBReference(*begin()->getParent()) << " ***\n"; dumpSchedule(); dbgs() << '\n'; }); } /// Apply each ScheduleDAGMutation step in order. void ScheduleDAGMI::postprocessDAG() { for (auto &m : Mutations) m->apply(this); } void ScheduleDAGMI:: findRootsAndBiasEdges(SmallVectorImpl &TopRoots, SmallVectorImpl &BotRoots) { for (SUnit &SU : SUnits) { assert(!SU.isBoundaryNode() && "Boundary node should not be in SUnits"); // Order predecessors so DFSResult follows the critical path. SU.biasCriticalPath(); // A SUnit is ready to top schedule if it has no predecessors. if (!SU.NumPredsLeft) TopRoots.push_back(&SU); // A SUnit is ready to bottom schedule if it has no successors. if (!SU.NumSuccsLeft) BotRoots.push_back(&SU); } ExitSU.biasCriticalPath(); } /// Identify DAG roots and setup scheduler queues. void ScheduleDAGMI::initQueues(ArrayRef TopRoots, ArrayRef BotRoots) { NextClusterSucc = nullptr; NextClusterPred = nullptr; // Release all DAG roots for scheduling, not including EntrySU/ExitSU. // // Nodes with unreleased weak edges can still be roots. // Release top roots in forward order. for (SUnit *SU : TopRoots) SchedImpl->releaseTopNode(SU); // Release bottom roots in reverse order so the higher priority nodes appear // first. This is more natural and slightly more efficient. for (SmallVectorImpl::const_reverse_iterator I = BotRoots.rbegin(), E = BotRoots.rend(); I != E; ++I) { SchedImpl->releaseBottomNode(*I); } releaseSuccessors(&EntrySU); releasePredecessors(&ExitSU); SchedImpl->registerRoots(); // Advance past initial DebugValues. CurrentTop = nextIfDebug(RegionBegin, RegionEnd); CurrentBottom = RegionEnd; } /// Update scheduler queues after scheduling an instruction. void ScheduleDAGMI::updateQueues(SUnit *SU, bool IsTopNode) { // Release dependent instructions for scheduling. if (IsTopNode) releaseSuccessors(SU); else releasePredecessors(SU); SU->isScheduled = true; } /// Reinsert any remaining debug_values, just like the PostRA scheduler. void ScheduleDAGMI::placeDebugValues() { // If first instruction was a DBG_VALUE then put it back. if (FirstDbgValue) { BB->splice(RegionBegin, BB, FirstDbgValue); RegionBegin = FirstDbgValue; } for (std::vector>::iterator DI = DbgValues.end(), DE = DbgValues.begin(); DI != DE; --DI) { std::pair P = *std::prev(DI); MachineInstr *DbgValue = P.first; MachineBasicBlock::iterator OrigPrevMI = P.second; if (&*RegionBegin == DbgValue) ++RegionBegin; BB->splice(++OrigPrevMI, BB, DbgValue); if (OrigPrevMI == std::prev(RegionEnd)) RegionEnd = DbgValue; } DbgValues.clear(); FirstDbgValue = nullptr; } #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP) LLVM_DUMP_METHOD void ScheduleDAGMI::dumpSchedule() const { for (MachineBasicBlock::iterator MI = begin(), ME = end(); MI != ME; ++MI) { if (SUnit *SU = getSUnit(&(*MI))) dumpNode(*SU); else dbgs() << "Missing SUnit\n"; } } #endif //===----------------------------------------------------------------------===// // ScheduleDAGMILive - Base class for MachineInstr scheduling with LiveIntervals // preservation. //===----------------------------------------------------------------------===// ScheduleDAGMILive::~ScheduleDAGMILive() { delete DFSResult; } void ScheduleDAGMILive::collectVRegUses(SUnit &SU) { const MachineInstr &MI = *SU.getInstr(); for (const MachineOperand &MO : MI.operands()) { if (!MO.isReg()) continue; if (!MO.readsReg()) continue; if (TrackLaneMasks && !MO.isUse()) continue; Register Reg = MO.getReg(); if (!Register::isVirtualRegister(Reg)) continue; // Ignore re-defs. if (TrackLaneMasks) { bool FoundDef = false; for (const MachineOperand &MO2 : MI.operands()) { if (MO2.isReg() && MO2.isDef() && MO2.getReg() == Reg && !MO2.isDead()) { FoundDef = true; break; } } if (FoundDef) continue; } // Record this local VReg use. VReg2SUnitMultiMap::iterator UI = VRegUses.find(Reg); for (; UI != VRegUses.end(); ++UI) { if (UI->SU == &SU) break; } if (UI == VRegUses.end()) VRegUses.insert(VReg2SUnit(Reg, LaneBitmask::getNone(), &SU)); } } /// enterRegion - Called back from MachineScheduler::runOnMachineFunction after /// crossing a scheduling boundary. [begin, end) includes all instructions in /// the region, including the boundary itself and single-instruction regions /// that don't get scheduled. void ScheduleDAGMILive::enterRegion(MachineBasicBlock *bb, MachineBasicBlock::iterator begin, MachineBasicBlock::iterator end, unsigned regioninstrs) { // ScheduleDAGMI initializes SchedImpl's per-region policy. ScheduleDAGMI::enterRegion(bb, begin, end, regioninstrs); // For convenience remember the end of the liveness region. LiveRegionEnd = (RegionEnd == bb->end()) ? RegionEnd : std::next(RegionEnd); SUPressureDiffs.clear(); ShouldTrackPressure = SchedImpl->shouldTrackPressure(); ShouldTrackLaneMasks = SchedImpl->shouldTrackLaneMasks(); assert((!ShouldTrackLaneMasks || ShouldTrackPressure) && "ShouldTrackLaneMasks requires ShouldTrackPressure"); } // Setup the register pressure trackers for the top scheduled and bottom // scheduled regions. void ScheduleDAGMILive::initRegPressure() { VRegUses.clear(); VRegUses.setUniverse(MRI.getNumVirtRegs()); for (SUnit &SU : SUnits) collectVRegUses(SU); TopRPTracker.init(&MF, RegClassInfo, LIS, BB, RegionBegin, ShouldTrackLaneMasks, false); BotRPTracker.init(&MF, RegClassInfo, LIS, BB, LiveRegionEnd, ShouldTrackLaneMasks, false); // Close the RPTracker to finalize live ins. RPTracker.closeRegion(); LLVM_DEBUG(RPTracker.dump()); // Initialize the live ins and live outs. TopRPTracker.addLiveRegs(RPTracker.getPressure().LiveInRegs); BotRPTracker.addLiveRegs(RPTracker.getPressure().LiveOutRegs); // Close one end of the tracker so we can call // getMaxUpward/DownwardPressureDelta before advancing across any // instructions. This converts currently live regs into live ins/outs. TopRPTracker.closeTop(); BotRPTracker.closeBottom(); BotRPTracker.initLiveThru(RPTracker); if (!BotRPTracker.getLiveThru().empty()) { TopRPTracker.initLiveThru(BotRPTracker.getLiveThru()); LLVM_DEBUG(dbgs() << "Live Thru: "; dumpRegSetPressure(BotRPTracker.getLiveThru(), TRI)); }; // For each live out vreg reduce the pressure change associated with other // uses of the same vreg below the live-out reaching def. updatePressureDiffs(RPTracker.getPressure().LiveOutRegs); // Account for liveness generated by the region boundary. if (LiveRegionEnd != RegionEnd) { SmallVector LiveUses; BotRPTracker.recede(&LiveUses); updatePressureDiffs(LiveUses); } LLVM_DEBUG(dbgs() << "Top Pressure:\n"; dumpRegSetPressure(TopRPTracker.getRegSetPressureAtPos(), TRI); dbgs() << "Bottom Pressure:\n"; dumpRegSetPressure(BotRPTracker.getRegSetPressureAtPos(), TRI);); assert((BotRPTracker.getPos() == RegionEnd || (RegionEnd->isDebugInstr() && BotRPTracker.getPos() == priorNonDebug(RegionEnd, RegionBegin))) && "Can't find the region bottom"); // Cache the list of excess pressure sets in this region. This will also track // the max pressure in the scheduled code for these sets. RegionCriticalPSets.clear(); const std::vector &RegionPressure = RPTracker.getPressure().MaxSetPressure; for (unsigned i = 0, e = RegionPressure.size(); i < e; ++i) { unsigned Limit = RegClassInfo->getRegPressureSetLimit(i); if (RegionPressure[i] > Limit) { LLVM_DEBUG(dbgs() << TRI->getRegPressureSetName(i) << " Limit " << Limit << " Actual " << RegionPressure[i] << "\n"); RegionCriticalPSets.push_back(PressureChange(i)); } } LLVM_DEBUG(dbgs() << "Excess PSets: "; for (const PressureChange &RCPS : RegionCriticalPSets) dbgs() << TRI->getRegPressureSetName(RCPS.getPSet()) << " "; dbgs() << "\n"); } void ScheduleDAGMILive:: updateScheduledPressure(const SUnit *SU, const std::vector &NewMaxPressure) { const PressureDiff &PDiff = getPressureDiff(SU); unsigned CritIdx = 0, CritEnd = RegionCriticalPSets.size(); for (const PressureChange &PC : PDiff) { if (!PC.isValid()) break; unsigned ID = PC.getPSet(); while (CritIdx != CritEnd && RegionCriticalPSets[CritIdx].getPSet() < ID) ++CritIdx; if (CritIdx != CritEnd && RegionCriticalPSets[CritIdx].getPSet() == ID) { if ((int)NewMaxPressure[ID] > RegionCriticalPSets[CritIdx].getUnitInc() && NewMaxPressure[ID] <= (unsigned)std::numeric_limits::max()) RegionCriticalPSets[CritIdx].setUnitInc(NewMaxPressure[ID]); } unsigned Limit = RegClassInfo->getRegPressureSetLimit(ID); if (NewMaxPressure[ID] >= Limit - 2) { LLVM_DEBUG(dbgs() << " " << TRI->getRegPressureSetName(ID) << ": " << NewMaxPressure[ID] << ((NewMaxPressure[ID] > Limit) ? " > " : " <= ") << Limit << "(+ " << BotRPTracker.getLiveThru()[ID] << " livethru)\n"); } } } /// Update the PressureDiff array for liveness after scheduling this /// instruction. void ScheduleDAGMILive::updatePressureDiffs( ArrayRef LiveUses) { for (const RegisterMaskPair &P : LiveUses) { unsigned Reg = P.RegUnit; /// FIXME: Currently assuming single-use physregs. if (!Register::isVirtualRegister(Reg)) continue; if (ShouldTrackLaneMasks) { // If the register has just become live then other uses won't change // this fact anymore => decrement pressure. // If the register has just become dead then other uses make it come // back to life => increment pressure. bool Decrement = P.LaneMask.any(); for (const VReg2SUnit &V2SU : make_range(VRegUses.find(Reg), VRegUses.end())) { SUnit &SU = *V2SU.SU; if (SU.isScheduled || &SU == &ExitSU) continue; PressureDiff &PDiff = getPressureDiff(&SU); PDiff.addPressureChange(Reg, Decrement, &MRI); LLVM_DEBUG(dbgs() << " UpdateRegP: SU(" << SU.NodeNum << ") " << printReg(Reg, TRI) << ':' << PrintLaneMask(P.LaneMask) << ' ' << *SU.getInstr(); dbgs() << " to "; PDiff.dump(*TRI);); } } else { assert(P.LaneMask.any()); LLVM_DEBUG(dbgs() << " LiveReg: " << printVRegOrUnit(Reg, TRI) << "\n"); // This may be called before CurrentBottom has been initialized. However, // BotRPTracker must have a valid position. We want the value live into the // instruction or live out of the block, so ask for the previous // instruction's live-out. const LiveInterval &LI = LIS->getInterval(Reg); VNInfo *VNI; MachineBasicBlock::const_iterator I = nextIfDebug(BotRPTracker.getPos(), BB->end()); if (I == BB->end()) VNI = LI.getVNInfoBefore(LIS->getMBBEndIdx(BB)); else { LiveQueryResult LRQ = LI.Query(LIS->getInstructionIndex(*I)); VNI = LRQ.valueIn(); } // RegisterPressureTracker guarantees that readsReg is true for LiveUses. assert(VNI && "No live value at use."); for (const VReg2SUnit &V2SU : make_range(VRegUses.find(Reg), VRegUses.end())) { SUnit *SU = V2SU.SU; // If this use comes before the reaching def, it cannot be a last use, // so decrease its pressure change. if (!SU->isScheduled && SU != &ExitSU) { LiveQueryResult LRQ = LI.Query(LIS->getInstructionIndex(*SU->getInstr())); if (LRQ.valueIn() == VNI) { PressureDiff &PDiff = getPressureDiff(SU); PDiff.addPressureChange(Reg, true, &MRI); LLVM_DEBUG(dbgs() << " UpdateRegP: SU(" << SU->NodeNum << ") " << *SU->getInstr(); dbgs() << " to "; PDiff.dump(*TRI);); } } } } } } void ScheduleDAGMILive::dump() const { #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP) if (EntrySU.getInstr() != nullptr) dumpNodeAll(EntrySU); for (const SUnit &SU : SUnits) { dumpNodeAll(SU); if (ShouldTrackPressure) { dbgs() << " Pressure Diff : "; getPressureDiff(&SU).dump(*TRI); } dbgs() << " Single Issue : "; if (SchedModel.mustBeginGroup(SU.getInstr()) && SchedModel.mustEndGroup(SU.getInstr())) dbgs() << "true;"; else dbgs() << "false;"; dbgs() << '\n'; } if (ExitSU.getInstr() != nullptr) dumpNodeAll(ExitSU); #endif } /// schedule - Called back from MachineScheduler::runOnMachineFunction /// after setting up the current scheduling region. [RegionBegin, RegionEnd) /// only includes instructions that have DAG nodes, not scheduling boundaries. /// /// This is a skeletal driver, with all the functionality pushed into helpers, /// so that it can be easily extended by experimental schedulers. Generally, /// implementing MachineSchedStrategy should be sufficient to implement a new /// scheduling algorithm. However, if a scheduler further subclasses /// ScheduleDAGMILive then it will want to override this virtual method in order /// to update any specialized state. void ScheduleDAGMILive::schedule() { LLVM_DEBUG(dbgs() << "ScheduleDAGMILive::schedule starting\n"); LLVM_DEBUG(SchedImpl->dumpPolicy()); buildDAGWithRegPressure(); postprocessDAG(); SmallVector TopRoots, BotRoots; findRootsAndBiasEdges(TopRoots, BotRoots); // Initialize the strategy before modifying the DAG. // This may initialize a DFSResult to be used for queue priority. SchedImpl->initialize(this); LLVM_DEBUG(dump()); if (PrintDAGs) dump(); if (ViewMISchedDAGs) viewGraph(); // Initialize ready queues now that the DAG and priority data are finalized. initQueues(TopRoots, BotRoots); bool IsTopNode = false; while (true) { LLVM_DEBUG(dbgs() << "** ScheduleDAGMILive::schedule picking next node\n"); SUnit *SU = SchedImpl->pickNode(IsTopNode); if (!SU) break; assert(!SU->isScheduled && "Node already scheduled"); if (!checkSchedLimit()) break; scheduleMI(SU, IsTopNode); if (DFSResult) { unsigned SubtreeID = DFSResult->getSubtreeID(SU); if (!ScheduledTrees.test(SubtreeID)) { ScheduledTrees.set(SubtreeID); DFSResult->scheduleTree(SubtreeID); SchedImpl->scheduleTree(SubtreeID); } } // Notify the scheduling strategy after updating the DAG. SchedImpl->schedNode(SU, IsTopNode); updateQueues(SU, IsTopNode); } assert(CurrentTop == CurrentBottom && "Nonempty unscheduled zone."); placeDebugValues(); LLVM_DEBUG({ dbgs() << "*** Final schedule for " << printMBBReference(*begin()->getParent()) << " ***\n"; dumpSchedule(); dbgs() << '\n'; }); } /// Build the DAG and setup three register pressure trackers. void ScheduleDAGMILive::buildDAGWithRegPressure() { if (!ShouldTrackPressure) { RPTracker.reset(); RegionCriticalPSets.clear(); buildSchedGraph(AA); return; } // Initialize the register pressure tracker used by buildSchedGraph. RPTracker.init(&MF, RegClassInfo, LIS, BB, LiveRegionEnd, ShouldTrackLaneMasks, /*TrackUntiedDefs=*/true); // Account for liveness generate by the region boundary. if (LiveRegionEnd != RegionEnd) RPTracker.recede(); // Build the DAG, and compute current register pressure. buildSchedGraph(AA, &RPTracker, &SUPressureDiffs, LIS, ShouldTrackLaneMasks); // Initialize top/bottom trackers after computing region pressure. initRegPressure(); } void ScheduleDAGMILive::computeDFSResult() { if (!DFSResult) DFSResult = new SchedDFSResult(/*BottomU*/true, MinSubtreeSize); DFSResult->clear(); ScheduledTrees.clear(); DFSResult->resize(SUnits.size()); DFSResult->compute(SUnits); ScheduledTrees.resize(DFSResult->getNumSubtrees()); } /// Compute the max cyclic critical path through the DAG. The scheduling DAG /// only provides the critical path for single block loops. To handle loops that /// span blocks, we could use the vreg path latencies provided by /// MachineTraceMetrics instead. However, MachineTraceMetrics is not currently /// available for use in the scheduler. /// /// The cyclic path estimation identifies a def-use pair that crosses the back /// edge and considers the depth and height of the nodes. For example, consider /// the following instruction sequence where each instruction has unit latency /// and defines an eponymous virtual register: /// /// a->b(a,c)->c(b)->d(c)->exit /// /// The cyclic critical path is a two cycles: b->c->b /// The acyclic critical path is four cycles: a->b->c->d->exit /// LiveOutHeight = height(c) = len(c->d->exit) = 2 /// LiveOutDepth = depth(c) + 1 = len(a->b->c) + 1 = 3 /// LiveInHeight = height(b) + 1 = len(b->c->d->exit) + 1 = 4 /// LiveInDepth = depth(b) = len(a->b) = 1 /// /// LiveOutDepth - LiveInDepth = 3 - 1 = 2 /// LiveInHeight - LiveOutHeight = 4 - 2 = 2 /// CyclicCriticalPath = min(2, 2) = 2 /// /// This could be relevant to PostRA scheduling, but is currently implemented /// assuming LiveIntervals. unsigned ScheduleDAGMILive::computeCyclicCriticalPath() { // This only applies to single block loop. if (!BB->isSuccessor(BB)) return 0; unsigned MaxCyclicLatency = 0; // Visit each live out vreg def to find def/use pairs that cross iterations. for (const RegisterMaskPair &P : RPTracker.getPressure().LiveOutRegs) { unsigned Reg = P.RegUnit; if (!Register::isVirtualRegister(Reg)) continue; const LiveInterval &LI = LIS->getInterval(Reg); const VNInfo *DefVNI = LI.getVNInfoBefore(LIS->getMBBEndIdx(BB)); if (!DefVNI) continue; MachineInstr *DefMI = LIS->getInstructionFromIndex(DefVNI->def); const SUnit *DefSU = getSUnit(DefMI); if (!DefSU) continue; unsigned LiveOutHeight = DefSU->getHeight(); unsigned LiveOutDepth = DefSU->getDepth() + DefSU->Latency; // Visit all local users of the vreg def. for (const VReg2SUnit &V2SU : make_range(VRegUses.find(Reg), VRegUses.end())) { SUnit *SU = V2SU.SU; if (SU == &ExitSU) continue; // Only consider uses of the phi. LiveQueryResult LRQ = LI.Query(LIS->getInstructionIndex(*SU->getInstr())); if (!LRQ.valueIn()->isPHIDef()) continue; // Assume that a path spanning two iterations is a cycle, which could // overestimate in strange cases. This allows cyclic latency to be // estimated as the minimum slack of the vreg's depth or height. unsigned CyclicLatency = 0; if (LiveOutDepth > SU->getDepth()) CyclicLatency = LiveOutDepth - SU->getDepth(); unsigned LiveInHeight = SU->getHeight() + DefSU->Latency; if (LiveInHeight > LiveOutHeight) { if (LiveInHeight - LiveOutHeight < CyclicLatency) CyclicLatency = LiveInHeight - LiveOutHeight; } else CyclicLatency = 0; LLVM_DEBUG(dbgs() << "Cyclic Path: SU(" << DefSU->NodeNum << ") -> SU(" << SU->NodeNum << ") = " << CyclicLatency << "c\n"); if (CyclicLatency > MaxCyclicLatency) MaxCyclicLatency = CyclicLatency; } } LLVM_DEBUG(dbgs() << "Cyclic Critical Path: " << MaxCyclicLatency << "c\n"); return MaxCyclicLatency; } /// Release ExitSU predecessors and setup scheduler queues. Re-position /// the Top RP tracker in case the region beginning has changed. void ScheduleDAGMILive::initQueues(ArrayRef TopRoots, ArrayRef BotRoots) { ScheduleDAGMI::initQueues(TopRoots, BotRoots); if (ShouldTrackPressure) { assert(TopRPTracker.getPos() == RegionBegin && "bad initial Top tracker"); TopRPTracker.setPos(CurrentTop); } } /// Move an instruction and update register pressure. void ScheduleDAGMILive::scheduleMI(SUnit *SU, bool IsTopNode) { // Move the instruction to its new location in the instruction stream. MachineInstr *MI = SU->getInstr(); if (IsTopNode) { assert(SU->isTopReady() && "node still has unscheduled dependencies"); if (&*CurrentTop == MI) CurrentTop = nextIfDebug(++CurrentTop, CurrentBottom); else { moveInstruction(MI, CurrentTop); TopRPTracker.setPos(MI); } if (ShouldTrackPressure) { // Update top scheduled pressure. RegisterOperands RegOpers; RegOpers.collect(*MI, *TRI, MRI, ShouldTrackLaneMasks, false); if (ShouldTrackLaneMasks) { // Adjust liveness and add missing dead+read-undef flags. SlotIndex SlotIdx = LIS->getInstructionIndex(*MI).getRegSlot(); RegOpers.adjustLaneLiveness(*LIS, MRI, SlotIdx, MI); } else { // Adjust for missing dead-def flags. RegOpers.detectDeadDefs(*MI, *LIS); } TopRPTracker.advance(RegOpers); assert(TopRPTracker.getPos() == CurrentTop && "out of sync"); LLVM_DEBUG(dbgs() << "Top Pressure:\n"; dumpRegSetPressure( TopRPTracker.getRegSetPressureAtPos(), TRI);); updateScheduledPressure(SU, TopRPTracker.getPressure().MaxSetPressure); } } else { assert(SU->isBottomReady() && "node still has unscheduled dependencies"); MachineBasicBlock::iterator priorII = priorNonDebug(CurrentBottom, CurrentTop); if (&*priorII == MI) CurrentBottom = priorII; else { if (&*CurrentTop == MI) { CurrentTop = nextIfDebug(++CurrentTop, priorII); TopRPTracker.setPos(CurrentTop); } moveInstruction(MI, CurrentBottom); CurrentBottom = MI; BotRPTracker.setPos(CurrentBottom); } if (ShouldTrackPressure) { RegisterOperands RegOpers; RegOpers.collect(*MI, *TRI, MRI, ShouldTrackLaneMasks, false); if (ShouldTrackLaneMasks) { // Adjust liveness and add missing dead+read-undef flags. SlotIndex SlotIdx = LIS->getInstructionIndex(*MI).getRegSlot(); RegOpers.adjustLaneLiveness(*LIS, MRI, SlotIdx, MI); } else { // Adjust for missing dead-def flags. RegOpers.detectDeadDefs(*MI, *LIS); } if (BotRPTracker.getPos() != CurrentBottom) BotRPTracker.recedeSkipDebugValues(); SmallVector LiveUses; BotRPTracker.recede(RegOpers, &LiveUses); assert(BotRPTracker.getPos() == CurrentBottom && "out of sync"); LLVM_DEBUG(dbgs() << "Bottom Pressure:\n"; dumpRegSetPressure( BotRPTracker.getRegSetPressureAtPos(), TRI);); updateScheduledPressure(SU, BotRPTracker.getPressure().MaxSetPressure); updatePressureDiffs(LiveUses); } } } //===----------------------------------------------------------------------===// // BaseMemOpClusterMutation - DAG post-processing to cluster loads or stores. //===----------------------------------------------------------------------===// namespace { /// Post-process the DAG to create cluster edges between neighboring /// loads or between neighboring stores. class BaseMemOpClusterMutation : public ScheduleDAGMutation { struct MemOpInfo { SUnit *SU; SmallVector BaseOps; int64_t Offset; unsigned Width; MemOpInfo(SUnit *SU, ArrayRef BaseOps, int64_t Offset, unsigned Width) : SU(SU), BaseOps(BaseOps.begin(), BaseOps.end()), Offset(Offset), Width(Width) {} static bool Compare(const MachineOperand *const &A, const MachineOperand *const &B) { if (A->getType() != B->getType()) return A->getType() < B->getType(); if (A->isReg()) return A->getReg() < B->getReg(); if (A->isFI()) { const MachineFunction &MF = *A->getParent()->getParent()->getParent(); const TargetFrameLowering &TFI = *MF.getSubtarget().getFrameLowering(); bool StackGrowsDown = TFI.getStackGrowthDirection() == TargetFrameLowering::StackGrowsDown; return StackGrowsDown ? A->getIndex() > B->getIndex() : A->getIndex() < B->getIndex(); } llvm_unreachable("MemOpClusterMutation only supports register or frame " "index bases."); } bool operator<(const MemOpInfo &RHS) const { // FIXME: Don't compare everything twice. Maybe use C++20 three way // comparison instead when it's available. if (std::lexicographical_compare(BaseOps.begin(), BaseOps.end(), RHS.BaseOps.begin(), RHS.BaseOps.end(), Compare)) return true; if (std::lexicographical_compare(RHS.BaseOps.begin(), RHS.BaseOps.end(), BaseOps.begin(), BaseOps.end(), Compare)) return false; if (Offset != RHS.Offset) return Offset < RHS.Offset; return SU->NodeNum < RHS.SU->NodeNum; } }; const TargetInstrInfo *TII; const TargetRegisterInfo *TRI; bool IsLoad; public: BaseMemOpClusterMutation(const TargetInstrInfo *tii, const TargetRegisterInfo *tri, bool IsLoad) : TII(tii), TRI(tri), IsLoad(IsLoad) {} void apply(ScheduleDAGInstrs *DAGInstrs) override; protected: void clusterNeighboringMemOps(ArrayRef MemOps, ScheduleDAGInstrs *DAG); void collectMemOpRecords(std::vector &SUnits, SmallVectorImpl &MemOpRecords); }; class StoreClusterMutation : public BaseMemOpClusterMutation { public: StoreClusterMutation(const TargetInstrInfo *tii, const TargetRegisterInfo *tri) : BaseMemOpClusterMutation(tii, tri, false) {} }; class LoadClusterMutation : public BaseMemOpClusterMutation { public: LoadClusterMutation(const TargetInstrInfo *tii, const TargetRegisterInfo *tri) : BaseMemOpClusterMutation(tii, tri, true) {} }; } // end anonymous namespace namespace llvm { std::unique_ptr createLoadClusterDAGMutation(const TargetInstrInfo *TII, const TargetRegisterInfo *TRI) { return EnableMemOpCluster ? std::make_unique(TII, TRI) : nullptr; } std::unique_ptr createStoreClusterDAGMutation(const TargetInstrInfo *TII, const TargetRegisterInfo *TRI) { return EnableMemOpCluster ? std::make_unique(TII, TRI) : nullptr; } } // end namespace llvm // Sorting all the loads/stores first, then for each load/store, checking the // following load/store one by one, until reach the first non-dependent one and // call target hook to see if they can cluster. void BaseMemOpClusterMutation::clusterNeighboringMemOps( ArrayRef MemOpRecords, ScheduleDAGInstrs *DAG) { // Keep track of the current cluster length and bytes for each SUnit. DenseMap> SUnit2ClusterInfo; // At this point, `MemOpRecords` array must hold atleast two mem ops. Try to // cluster mem ops collected within `MemOpRecords` array. for (unsigned Idx = 0, End = MemOpRecords.size(); Idx < (End - 1); ++Idx) { // Decision to cluster mem ops is taken based on target dependent logic auto MemOpa = MemOpRecords[Idx]; // Seek for the next load/store to do the cluster. unsigned NextIdx = Idx + 1; for (; NextIdx < End; ++NextIdx) // Skip if MemOpb has been clustered already or has dependency with // MemOpa. if (!SUnit2ClusterInfo.count(MemOpRecords[NextIdx].SU->NodeNum) && !DAG->IsReachable(MemOpRecords[NextIdx].SU, MemOpa.SU) && !DAG->IsReachable(MemOpa.SU, MemOpRecords[NextIdx].SU)) break; if (NextIdx == End) continue; auto MemOpb = MemOpRecords[NextIdx]; unsigned ClusterLength = 2; unsigned CurrentClusterBytes = MemOpa.Width + MemOpb.Width; if (SUnit2ClusterInfo.count(MemOpa.SU->NodeNum)) { ClusterLength = SUnit2ClusterInfo[MemOpa.SU->NodeNum].first + 1; CurrentClusterBytes = SUnit2ClusterInfo[MemOpa.SU->NodeNum].second + MemOpb.Width; } if (!TII->shouldClusterMemOps(MemOpa.BaseOps, MemOpb.BaseOps, ClusterLength, CurrentClusterBytes)) continue; SUnit *SUa = MemOpa.SU; SUnit *SUb = MemOpb.SU; if (SUa->NodeNum > SUb->NodeNum) std::swap(SUa, SUb); // FIXME: Is this check really required? if (!DAG->addEdge(SUb, SDep(SUa, SDep::Cluster))) continue; LLVM_DEBUG(dbgs() << "Cluster ld/st SU(" << SUa->NodeNum << ") - SU(" << SUb->NodeNum << ")\n"); ++NumClustered; if (IsLoad) { // Copy successor edges from SUa to SUb. Interleaving computation // dependent on SUa can prevent load combining due to register reuse. // Predecessor edges do not need to be copied from SUb to SUa since // nearby loads should have effectively the same inputs. for (const SDep &Succ : SUa->Succs) { if (Succ.getSUnit() == SUb) continue; LLVM_DEBUG(dbgs() << " Copy Succ SU(" << Succ.getSUnit()->NodeNum << ")\n"); DAG->addEdge(Succ.getSUnit(), SDep(SUb, SDep::Artificial)); } } else { // Copy predecessor edges from SUb to SUa to avoid the SUnits that // SUb dependent on scheduled in-between SUb and SUa. Successor edges // do not need to be copied from SUa to SUb since no one will depend // on stores. // Notice that, we don't need to care about the memory dependency as // we won't try to cluster them if they have any memory dependency. for (const SDep &Pred : SUb->Preds) { if (Pred.getSUnit() == SUa) continue; LLVM_DEBUG(dbgs() << " Copy Pred SU(" << Pred.getSUnit()->NodeNum << ")\n"); DAG->addEdge(SUa, SDep(Pred.getSUnit(), SDep::Artificial)); } } SUnit2ClusterInfo[MemOpb.SU->NodeNum] = {ClusterLength, CurrentClusterBytes}; LLVM_DEBUG(dbgs() << " Curr cluster length: " << ClusterLength << ", Curr cluster bytes: " << CurrentClusterBytes << "\n"); } } void BaseMemOpClusterMutation::collectMemOpRecords( std::vector &SUnits, SmallVectorImpl &MemOpRecords) { for (auto &SU : SUnits) { if ((IsLoad && !SU.getInstr()->mayLoad()) || (!IsLoad && !SU.getInstr()->mayStore())) continue; const MachineInstr &MI = *SU.getInstr(); SmallVector BaseOps; int64_t Offset; bool OffsetIsScalable; unsigned Width; if (TII->getMemOperandsWithOffsetWidth(MI, BaseOps, Offset, OffsetIsScalable, Width, TRI)) { MemOpRecords.push_back(MemOpInfo(&SU, BaseOps, Offset, Width)); LLVM_DEBUG(dbgs() << "Num BaseOps: " << BaseOps.size() << ", Offset: " << Offset << ", OffsetIsScalable: " << OffsetIsScalable << ", Width: " << Width << "\n"); } #ifndef NDEBUG for (auto *Op : BaseOps) assert(Op); #endif } } /// Callback from DAG postProcessing to create cluster edges for loads/stores. void BaseMemOpClusterMutation::apply(ScheduleDAGInstrs *DAG) { // Collect all the clusterable loads/stores SmallVector MemOpRecords; collectMemOpRecords(DAG->SUnits, MemOpRecords); if (MemOpRecords.size() < 2) return; // Sorting the loads/stores, so that, we can stop the cluster as early as // possible. llvm::sort(MemOpRecords); // Trying to cluster all the neighboring loads/stores. clusterNeighboringMemOps(MemOpRecords, DAG); } //===----------------------------------------------------------------------===// // CopyConstrain - DAG post-processing to encourage copy elimination. //===----------------------------------------------------------------------===// namespace { /// Post-process the DAG to create weak edges from all uses of a copy to /// the one use that defines the copy's source vreg, most likely an induction /// variable increment. class CopyConstrain : public ScheduleDAGMutation { // Transient state. SlotIndex RegionBeginIdx; // RegionEndIdx is the slot index of the last non-debug instruction in the // scheduling region. So we may have RegionBeginIdx == RegionEndIdx. SlotIndex RegionEndIdx; public: CopyConstrain(const TargetInstrInfo *, const TargetRegisterInfo *) {} void apply(ScheduleDAGInstrs *DAGInstrs) override; protected: void constrainLocalCopy(SUnit *CopySU, ScheduleDAGMILive *DAG); }; } // end anonymous namespace namespace llvm { std::unique_ptr createCopyConstrainDAGMutation(const TargetInstrInfo *TII, const TargetRegisterInfo *TRI) { return std::make_unique(TII, TRI); } } // end namespace llvm /// constrainLocalCopy handles two possibilities: /// 1) Local src: /// I0: = dst /// I1: src = ... /// I2: = dst /// I3: dst = src (copy) /// (create pred->succ edges I0->I1, I2->I1) /// /// 2) Local copy: /// I0: dst = src (copy) /// I1: = dst /// I2: src = ... /// I3: = dst /// (create pred->succ edges I1->I2, I3->I2) /// /// Although the MachineScheduler is currently constrained to single blocks, /// this algorithm should handle extended blocks. An EBB is a set of /// contiguously numbered blocks such that the previous block in the EBB is /// always the single predecessor. void CopyConstrain::constrainLocalCopy(SUnit *CopySU, ScheduleDAGMILive *DAG) { LiveIntervals *LIS = DAG->getLIS(); MachineInstr *Copy = CopySU->getInstr(); // Check for pure vreg copies. const MachineOperand &SrcOp = Copy->getOperand(1); Register SrcReg = SrcOp.getReg(); if (!Register::isVirtualRegister(SrcReg) || !SrcOp.readsReg()) return; const MachineOperand &DstOp = Copy->getOperand(0); Register DstReg = DstOp.getReg(); if (!Register::isVirtualRegister(DstReg) || DstOp.isDead()) return; // Check if either the dest or source is local. If it's live across a back // edge, it's not local. Note that if both vregs are live across the back // edge, we cannot successfully contrain the copy without cyclic scheduling. // If both the copy's source and dest are local live intervals, then we // should treat the dest as the global for the purpose of adding // constraints. This adds edges from source's other uses to the copy. unsigned LocalReg = SrcReg; unsigned GlobalReg = DstReg; LiveInterval *LocalLI = &LIS->getInterval(LocalReg); if (!LocalLI->isLocal(RegionBeginIdx, RegionEndIdx)) { LocalReg = DstReg; GlobalReg = SrcReg; LocalLI = &LIS->getInterval(LocalReg); if (!LocalLI->isLocal(RegionBeginIdx, RegionEndIdx)) return; } LiveInterval *GlobalLI = &LIS->getInterval(GlobalReg); // Find the global segment after the start of the local LI. LiveInterval::iterator GlobalSegment = GlobalLI->find(LocalLI->beginIndex()); // If GlobalLI does not overlap LocalLI->start, then a copy directly feeds a // local live range. We could create edges from other global uses to the local // start, but the coalescer should have already eliminated these cases, so // don't bother dealing with it. if (GlobalSegment == GlobalLI->end()) return; // If GlobalSegment is killed at the LocalLI->start, the call to find() // returned the next global segment. But if GlobalSegment overlaps with // LocalLI->start, then advance to the next segment. If a hole in GlobalLI // exists in LocalLI's vicinity, GlobalSegment will be the end of the hole. if (GlobalSegment->contains(LocalLI->beginIndex())) ++GlobalSegment; if (GlobalSegment == GlobalLI->end()) return; // Check if GlobalLI contains a hole in the vicinity of LocalLI. if (GlobalSegment != GlobalLI->begin()) { // Two address defs have no hole. if (SlotIndex::isSameInstr(std::prev(GlobalSegment)->end, GlobalSegment->start)) { return; } // If the prior global segment may be defined by the same two-address // instruction that also defines LocalLI, then can't make a hole here. if (SlotIndex::isSameInstr(std::prev(GlobalSegment)->start, LocalLI->beginIndex())) { return; } // If GlobalLI has a prior segment, it must be live into the EBB. Otherwise // it would be a disconnected component in the live range. assert(std::prev(GlobalSegment)->start < LocalLI->beginIndex() && "Disconnected LRG within the scheduling region."); } MachineInstr *GlobalDef = LIS->getInstructionFromIndex(GlobalSegment->start); if (!GlobalDef) return; SUnit *GlobalSU = DAG->getSUnit(GlobalDef); if (!GlobalSU) return; // GlobalDef is the bottom of the GlobalLI hole. Open the hole by // constraining the uses of the last local def to precede GlobalDef. SmallVector LocalUses; const VNInfo *LastLocalVN = LocalLI->getVNInfoBefore(LocalLI->endIndex()); MachineInstr *LastLocalDef = LIS->getInstructionFromIndex(LastLocalVN->def); SUnit *LastLocalSU = DAG->getSUnit(LastLocalDef); for (const SDep &Succ : LastLocalSU->Succs) { if (Succ.getKind() != SDep::Data || Succ.getReg() != LocalReg) continue; if (Succ.getSUnit() == GlobalSU) continue; if (!DAG->canAddEdge(GlobalSU, Succ.getSUnit())) return; LocalUses.push_back(Succ.getSUnit()); } // Open the top of the GlobalLI hole by constraining any earlier global uses // to precede the start of LocalLI. SmallVector GlobalUses; MachineInstr *FirstLocalDef = LIS->getInstructionFromIndex(LocalLI->beginIndex()); SUnit *FirstLocalSU = DAG->getSUnit(FirstLocalDef); for (const SDep &Pred : GlobalSU->Preds) { if (Pred.getKind() != SDep::Anti || Pred.getReg() != GlobalReg) continue; if (Pred.getSUnit() == FirstLocalSU) continue; if (!DAG->canAddEdge(FirstLocalSU, Pred.getSUnit())) return; GlobalUses.push_back(Pred.getSUnit()); } LLVM_DEBUG(dbgs() << "Constraining copy SU(" << CopySU->NodeNum << ")\n"); // Add the weak edges. for (SmallVectorImpl::const_iterator I = LocalUses.begin(), E = LocalUses.end(); I != E; ++I) { LLVM_DEBUG(dbgs() << " Local use SU(" << (*I)->NodeNum << ") -> SU(" << GlobalSU->NodeNum << ")\n"); DAG->addEdge(GlobalSU, SDep(*I, SDep::Weak)); } for (SmallVectorImpl::const_iterator I = GlobalUses.begin(), E = GlobalUses.end(); I != E; ++I) { LLVM_DEBUG(dbgs() << " Global use SU(" << (*I)->NodeNum << ") -> SU(" << FirstLocalSU->NodeNum << ")\n"); DAG->addEdge(FirstLocalSU, SDep(*I, SDep::Weak)); } } /// Callback from DAG postProcessing to create weak edges to encourage /// copy elimination. void CopyConstrain::apply(ScheduleDAGInstrs *DAGInstrs) { ScheduleDAGMI *DAG = static_cast(DAGInstrs); assert(DAG->hasVRegLiveness() && "Expect VRegs with LiveIntervals"); MachineBasicBlock::iterator FirstPos = nextIfDebug(DAG->begin(), DAG->end()); if (FirstPos == DAG->end()) return; RegionBeginIdx = DAG->getLIS()->getInstructionIndex(*FirstPos); RegionEndIdx = DAG->getLIS()->getInstructionIndex( *priorNonDebug(DAG->end(), DAG->begin())); for (SUnit &SU : DAG->SUnits) { if (!SU.getInstr()->isCopy()) continue; constrainLocalCopy(&SU, static_cast(DAG)); } } //===----------------------------------------------------------------------===// // MachineSchedStrategy helpers used by GenericScheduler, GenericPostScheduler // and possibly other custom schedulers. //===----------------------------------------------------------------------===// static const unsigned InvalidCycle = ~0U; SchedBoundary::~SchedBoundary() { delete HazardRec; } /// Given a Count of resource usage and a Latency value, return true if a /// SchedBoundary becomes resource limited. /// If we are checking after scheduling a node, we should return true when /// we just reach the resource limit. static bool checkResourceLimit(unsigned LFactor, unsigned Count, unsigned Latency, bool AfterSchedNode) { int ResCntFactor = (int)(Count - (Latency * LFactor)); if (AfterSchedNode) return ResCntFactor >= (int)LFactor; else return ResCntFactor > (int)LFactor; } void SchedBoundary::reset() { // A new HazardRec is created for each DAG and owned by SchedBoundary. // Destroying and reconstructing it is very expensive though. So keep // invalid, placeholder HazardRecs. if (HazardRec && HazardRec->isEnabled()) { delete HazardRec; HazardRec = nullptr; } Available.clear(); Pending.clear(); CheckPending = false; CurrCycle = 0; CurrMOps = 0; MinReadyCycle = std::numeric_limits::max(); ExpectedLatency = 0; DependentLatency = 0; RetiredMOps = 0; MaxExecutedResCount = 0; ZoneCritResIdx = 0; IsResourceLimited = false; ReservedCycles.clear(); ReservedCyclesIndex.clear(); #ifndef NDEBUG // Track the maximum number of stall cycles that could arise either from the // latency of a DAG edge or the number of cycles that a processor resource is // reserved (SchedBoundary::ReservedCycles). MaxObservedStall = 0; #endif // Reserve a zero-count for invalid CritResIdx. ExecutedResCounts.resize(1); assert(!ExecutedResCounts[0] && "nonzero count for bad resource"); } void SchedRemainder:: init(ScheduleDAGMI *DAG, const TargetSchedModel *SchedModel) { reset(); if (!SchedModel->hasInstrSchedModel()) return; RemainingCounts.resize(SchedModel->getNumProcResourceKinds()); for (SUnit &SU : DAG->SUnits) { const MCSchedClassDesc *SC = DAG->getSchedClass(&SU); RemIssueCount += SchedModel->getNumMicroOps(SU.getInstr(), SC) * SchedModel->getMicroOpFactor(); for (TargetSchedModel::ProcResIter PI = SchedModel->getWriteProcResBegin(SC), PE = SchedModel->getWriteProcResEnd(SC); PI != PE; ++PI) { unsigned PIdx = PI->ProcResourceIdx; unsigned Factor = SchedModel->getResourceFactor(PIdx); RemainingCounts[PIdx] += (Factor * PI->Cycles); } } } void SchedBoundary:: init(ScheduleDAGMI *dag, const TargetSchedModel *smodel, SchedRemainder *rem) { reset(); DAG = dag; SchedModel = smodel; Rem = rem; if (SchedModel->hasInstrSchedModel()) { unsigned ResourceCount = SchedModel->getNumProcResourceKinds(); ReservedCyclesIndex.resize(ResourceCount); ExecutedResCounts.resize(ResourceCount); unsigned NumUnits = 0; for (unsigned i = 0; i < ResourceCount; ++i) { ReservedCyclesIndex[i] = NumUnits; NumUnits += SchedModel->getProcResource(i)->NumUnits; } ReservedCycles.resize(NumUnits, InvalidCycle); } } /// Compute the stall cycles based on this SUnit's ready time. Heuristics treat /// these "soft stalls" differently than the hard stall cycles based on CPU /// resources and computed by checkHazard(). A fully in-order model /// (MicroOpBufferSize==0) will not make use of this since instructions are not /// available for scheduling until they are ready. However, a weaker in-order /// model may use this for heuristics. For example, if a processor has in-order /// behavior when reading certain resources, this may come into play. unsigned SchedBoundary::getLatencyStallCycles(SUnit *SU) { if (!SU->isUnbuffered) return 0; unsigned ReadyCycle = (isTop() ? SU->TopReadyCycle : SU->BotReadyCycle); if (ReadyCycle > CurrCycle) return ReadyCycle - CurrCycle; return 0; } /// Compute the next cycle at which the given processor resource unit /// can be scheduled. unsigned SchedBoundary::getNextResourceCycleByInstance(unsigned InstanceIdx, unsigned Cycles) { unsigned NextUnreserved = ReservedCycles[InstanceIdx]; // If this resource has never been used, always return cycle zero. if (NextUnreserved == InvalidCycle) return 0; // For bottom-up scheduling add the cycles needed for the current operation. if (!isTop()) NextUnreserved += Cycles; return NextUnreserved; } /// Compute the next cycle at which the given processor resource can be /// scheduled. Returns the next cycle and the index of the processor resource /// instance in the reserved cycles vector. std::pair SchedBoundary::getNextResourceCycle(unsigned PIdx, unsigned Cycles) { unsigned MinNextUnreserved = InvalidCycle; unsigned InstanceIdx = 0; unsigned StartIndex = ReservedCyclesIndex[PIdx]; unsigned NumberOfInstances = SchedModel->getProcResource(PIdx)->NumUnits; assert(NumberOfInstances > 0 && "Cannot have zero instances of a ProcResource"); for (unsigned I = StartIndex, End = StartIndex + NumberOfInstances; I < End; ++I) { unsigned NextUnreserved = getNextResourceCycleByInstance(I, Cycles); if (MinNextUnreserved > NextUnreserved) { InstanceIdx = I; MinNextUnreserved = NextUnreserved; } } return std::make_pair(MinNextUnreserved, InstanceIdx); } /// Does this SU have a hazard within the current instruction group. /// /// The scheduler supports two modes of hazard recognition. The first is the /// ScheduleHazardRecognizer API. It is a fully general hazard recognizer that /// supports highly complicated in-order reservation tables /// (ScoreboardHazardRecognizer) and arbitrary target-specific logic. /// /// The second is a streamlined mechanism that checks for hazards based on /// simple counters that the scheduler itself maintains. It explicitly checks /// for instruction dispatch limitations, including the number of micro-ops that /// can dispatch per cycle. /// /// TODO: Also check whether the SU must start a new group. bool SchedBoundary::checkHazard(SUnit *SU) { if (HazardRec->isEnabled() && HazardRec->getHazardType(SU) != ScheduleHazardRecognizer::NoHazard) { return true; } unsigned uops = SchedModel->getNumMicroOps(SU->getInstr()); if ((CurrMOps > 0) && (CurrMOps + uops > SchedModel->getIssueWidth())) { LLVM_DEBUG(dbgs() << " SU(" << SU->NodeNum << ") uops=" << SchedModel->getNumMicroOps(SU->getInstr()) << '\n'); return true; } if (CurrMOps > 0 && ((isTop() && SchedModel->mustBeginGroup(SU->getInstr())) || (!isTop() && SchedModel->mustEndGroup(SU->getInstr())))) { LLVM_DEBUG(dbgs() << " hazard: SU(" << SU->NodeNum << ") must " << (isTop() ? "begin" : "end") << " group\n"); return true; } if (SchedModel->hasInstrSchedModel() && SU->hasReservedResource) { const MCSchedClassDesc *SC = DAG->getSchedClass(SU); for (const MCWriteProcResEntry &PE : make_range(SchedModel->getWriteProcResBegin(SC), SchedModel->getWriteProcResEnd(SC))) { unsigned ResIdx = PE.ProcResourceIdx; unsigned Cycles = PE.Cycles; unsigned NRCycle, InstanceIdx; std::tie(NRCycle, InstanceIdx) = getNextResourceCycle(ResIdx, Cycles); if (NRCycle > CurrCycle) { #ifndef NDEBUG MaxObservedStall = std::max(Cycles, MaxObservedStall); #endif LLVM_DEBUG(dbgs() << " SU(" << SU->NodeNum << ") " << SchedModel->getResourceName(ResIdx) << '[' << InstanceIdx - ReservedCyclesIndex[ResIdx] << ']' << "=" << NRCycle << "c\n"); return true; } } } return false; } // Find the unscheduled node in ReadySUs with the highest latency. unsigned SchedBoundary:: findMaxLatency(ArrayRef ReadySUs) { SUnit *LateSU = nullptr; unsigned RemLatency = 0; for (SUnit *SU : ReadySUs) { unsigned L = getUnscheduledLatency(SU); if (L > RemLatency) { RemLatency = L; LateSU = SU; } } if (LateSU) { LLVM_DEBUG(dbgs() << Available.getName() << " RemLatency SU(" << LateSU->NodeNum << ") " << RemLatency << "c\n"); } return RemLatency; } // Count resources in this zone and the remaining unscheduled // instruction. Return the max count, scaled. Set OtherCritIdx to the critical // resource index, or zero if the zone is issue limited. unsigned SchedBoundary:: getOtherResourceCount(unsigned &OtherCritIdx) { OtherCritIdx = 0; if (!SchedModel->hasInstrSchedModel()) return 0; unsigned OtherCritCount = Rem->RemIssueCount + (RetiredMOps * SchedModel->getMicroOpFactor()); LLVM_DEBUG(dbgs() << " " << Available.getName() << " + Remain MOps: " << OtherCritCount / SchedModel->getMicroOpFactor() << '\n'); for (unsigned PIdx = 1, PEnd = SchedModel->getNumProcResourceKinds(); PIdx != PEnd; ++PIdx) { unsigned OtherCount = getResourceCount(PIdx) + Rem->RemainingCounts[PIdx]; if (OtherCount > OtherCritCount) { OtherCritCount = OtherCount; OtherCritIdx = PIdx; } } if (OtherCritIdx) { LLVM_DEBUG( dbgs() << " " << Available.getName() << " + Remain CritRes: " << OtherCritCount / SchedModel->getResourceFactor(OtherCritIdx) << " " << SchedModel->getResourceName(OtherCritIdx) << "\n"); } return OtherCritCount; } void SchedBoundary::releaseNode(SUnit *SU, unsigned ReadyCycle, bool InPQueue, unsigned Idx) { assert(SU->getInstr() && "Scheduled SUnit must have instr"); #ifndef NDEBUG // ReadyCycle was been bumped up to the CurrCycle when this node was // scheduled, but CurrCycle may have been eagerly advanced immediately after // scheduling, so may now be greater than ReadyCycle. if (ReadyCycle > CurrCycle) MaxObservedStall = std::max(ReadyCycle - CurrCycle, MaxObservedStall); #endif if (ReadyCycle < MinReadyCycle) MinReadyCycle = ReadyCycle; // Check for interlocks first. For the purpose of other heuristics, an // instruction that cannot issue appears as if it's not in the ReadyQueue. bool IsBuffered = SchedModel->getMicroOpBufferSize() != 0; bool HazardDetected = (!IsBuffered && ReadyCycle > CurrCycle) || checkHazard(SU) || (Available.size() >= ReadyListLimit); if (!HazardDetected) { Available.push(SU); if (InPQueue) Pending.remove(Pending.begin() + Idx); return; } if (!InPQueue) Pending.push(SU); } /// Move the boundary of scheduled code by one cycle. void SchedBoundary::bumpCycle(unsigned NextCycle) { if (SchedModel->getMicroOpBufferSize() == 0) { assert(MinReadyCycle < std::numeric_limits::max() && "MinReadyCycle uninitialized"); if (MinReadyCycle > NextCycle) NextCycle = MinReadyCycle; } // Update the current micro-ops, which will issue in the next cycle. unsigned DecMOps = SchedModel->getIssueWidth() * (NextCycle - CurrCycle); CurrMOps = (CurrMOps <= DecMOps) ? 0 : CurrMOps - DecMOps; // Decrement DependentLatency based on the next cycle. if ((NextCycle - CurrCycle) > DependentLatency) DependentLatency = 0; else DependentLatency -= (NextCycle - CurrCycle); if (!HazardRec->isEnabled()) { // Bypass HazardRec virtual calls. CurrCycle = NextCycle; } else { // Bypass getHazardType calls in case of long latency. for (; CurrCycle != NextCycle; ++CurrCycle) { if (isTop()) HazardRec->AdvanceCycle(); else HazardRec->RecedeCycle(); } } CheckPending = true; IsResourceLimited = checkResourceLimit(SchedModel->getLatencyFactor(), getCriticalCount(), getScheduledLatency(), true); LLVM_DEBUG(dbgs() << "Cycle: " << CurrCycle << ' ' << Available.getName() << '\n'); } void SchedBoundary::incExecutedResources(unsigned PIdx, unsigned Count) { ExecutedResCounts[PIdx] += Count; if (ExecutedResCounts[PIdx] > MaxExecutedResCount) MaxExecutedResCount = ExecutedResCounts[PIdx]; } /// Add the given processor resource to this scheduled zone. /// /// \param Cycles indicates the number of consecutive (non-pipelined) cycles /// during which this resource is consumed. /// /// \return the next cycle at which the instruction may execute without /// oversubscribing resources. unsigned SchedBoundary:: countResource(unsigned PIdx, unsigned Cycles, unsigned NextCycle) { unsigned Factor = SchedModel->getResourceFactor(PIdx); unsigned Count = Factor * Cycles; LLVM_DEBUG(dbgs() << " " << SchedModel->getResourceName(PIdx) << " +" << Cycles << "x" << Factor << "u\n"); // Update Executed resources counts. incExecutedResources(PIdx, Count); assert(Rem->RemainingCounts[PIdx] >= Count && "resource double counted"); Rem->RemainingCounts[PIdx] -= Count; // Check if this resource exceeds the current critical resource. If so, it // becomes the critical resource. if (ZoneCritResIdx != PIdx && (getResourceCount(PIdx) > getCriticalCount())) { ZoneCritResIdx = PIdx; LLVM_DEBUG(dbgs() << " *** Critical resource " << SchedModel->getResourceName(PIdx) << ": " << getResourceCount(PIdx) / SchedModel->getLatencyFactor() << "c\n"); } // For reserved resources, record the highest cycle using the resource. unsigned NextAvailable, InstanceIdx; std::tie(NextAvailable, InstanceIdx) = getNextResourceCycle(PIdx, Cycles); if (NextAvailable > CurrCycle) { LLVM_DEBUG(dbgs() << " Resource conflict: " << SchedModel->getResourceName(PIdx) << '[' << InstanceIdx - ReservedCyclesIndex[PIdx] << ']' << " reserved until @" << NextAvailable << "\n"); } return NextAvailable; } /// Move the boundary of scheduled code by one SUnit. void SchedBoundary::bumpNode(SUnit *SU) { // Update the reservation table. if (HazardRec->isEnabled()) { if (!isTop() && SU->isCall) { // Calls are scheduled with their preceding instructions. For bottom-up // scheduling, clear the pipeline state before emitting. HazardRec->Reset(); } HazardRec->EmitInstruction(SU); // Scheduling an instruction may have made pending instructions available. CheckPending = true; } // checkHazard should prevent scheduling multiple instructions per cycle that // exceed the issue width. const MCSchedClassDesc *SC = DAG->getSchedClass(SU); unsigned IncMOps = SchedModel->getNumMicroOps(SU->getInstr()); assert( (CurrMOps == 0 || (CurrMOps + IncMOps) <= SchedModel->getIssueWidth()) && "Cannot schedule this instruction's MicroOps in the current cycle."); unsigned ReadyCycle = (isTop() ? SU->TopReadyCycle : SU->BotReadyCycle); LLVM_DEBUG(dbgs() << " Ready @" << ReadyCycle << "c\n"); unsigned NextCycle = CurrCycle; switch (SchedModel->getMicroOpBufferSize()) { case 0: assert(ReadyCycle <= CurrCycle && "Broken PendingQueue"); break; case 1: if (ReadyCycle > NextCycle) { NextCycle = ReadyCycle; LLVM_DEBUG(dbgs() << " *** Stall until: " << ReadyCycle << "\n"); } break; default: // We don't currently model the OOO reorder buffer, so consider all // scheduled MOps to be "retired". We do loosely model in-order resource // latency. If this instruction uses an in-order resource, account for any // likely stall cycles. if (SU->isUnbuffered && ReadyCycle > NextCycle) NextCycle = ReadyCycle; break; } RetiredMOps += IncMOps; // Update resource counts and critical resource. if (SchedModel->hasInstrSchedModel()) { unsigned DecRemIssue = IncMOps * SchedModel->getMicroOpFactor(); assert(Rem->RemIssueCount >= DecRemIssue && "MOps double counted"); Rem->RemIssueCount -= DecRemIssue; if (ZoneCritResIdx) { // Scale scheduled micro-ops for comparing with the critical resource. unsigned ScaledMOps = RetiredMOps * SchedModel->getMicroOpFactor(); // If scaled micro-ops are now more than the previous critical resource by // a full cycle, then micro-ops issue becomes critical. if ((int)(ScaledMOps - getResourceCount(ZoneCritResIdx)) >= (int)SchedModel->getLatencyFactor()) { ZoneCritResIdx = 0; LLVM_DEBUG(dbgs() << " *** Critical resource NumMicroOps: " << ScaledMOps / SchedModel->getLatencyFactor() << "c\n"); } } for (TargetSchedModel::ProcResIter PI = SchedModel->getWriteProcResBegin(SC), PE = SchedModel->getWriteProcResEnd(SC); PI != PE; ++PI) { unsigned RCycle = countResource(PI->ProcResourceIdx, PI->Cycles, NextCycle); if (RCycle > NextCycle) NextCycle = RCycle; } if (SU->hasReservedResource) { // For reserved resources, record the highest cycle using the resource. // For top-down scheduling, this is the cycle in which we schedule this // instruction plus the number of cycles the operations reserves the // resource. For bottom-up is it simply the instruction's cycle. for (TargetSchedModel::ProcResIter PI = SchedModel->getWriteProcResBegin(SC), PE = SchedModel->getWriteProcResEnd(SC); PI != PE; ++PI) { unsigned PIdx = PI->ProcResourceIdx; if (SchedModel->getProcResource(PIdx)->BufferSize == 0) { unsigned ReservedUntil, InstanceIdx; std::tie(ReservedUntil, InstanceIdx) = getNextResourceCycle(PIdx, 0); if (isTop()) { ReservedCycles[InstanceIdx] = std::max(ReservedUntil, NextCycle + PI->Cycles); } else ReservedCycles[InstanceIdx] = NextCycle; } } } } // Update ExpectedLatency and DependentLatency. unsigned &TopLatency = isTop() ? ExpectedLatency : DependentLatency; unsigned &BotLatency = isTop() ? DependentLatency : ExpectedLatency; if (SU->getDepth() > TopLatency) { TopLatency = SU->getDepth(); LLVM_DEBUG(dbgs() << " " << Available.getName() << " TopLatency SU(" << SU->NodeNum << ") " << TopLatency << "c\n"); } if (SU->getHeight() > BotLatency) { BotLatency = SU->getHeight(); LLVM_DEBUG(dbgs() << " " << Available.getName() << " BotLatency SU(" << SU->NodeNum << ") " << BotLatency << "c\n"); } // If we stall for any reason, bump the cycle. if (NextCycle > CurrCycle) bumpCycle(NextCycle); else // After updating ZoneCritResIdx and ExpectedLatency, check if we're // resource limited. If a stall occurred, bumpCycle does this. IsResourceLimited = checkResourceLimit(SchedModel->getLatencyFactor(), getCriticalCount(), getScheduledLatency(), true); // Update CurrMOps after calling bumpCycle to handle stalls, since bumpCycle // resets CurrMOps. Loop to handle instructions with more MOps than issue in // one cycle. Since we commonly reach the max MOps here, opportunistically // bump the cycle to avoid uselessly checking everything in the readyQ. CurrMOps += IncMOps; // Bump the cycle count for issue group constraints. // This must be done after NextCycle has been adjust for all other stalls. // Calling bumpCycle(X) will reduce CurrMOps by one issue group and set // currCycle to X. if ((isTop() && SchedModel->mustEndGroup(SU->getInstr())) || (!isTop() && SchedModel->mustBeginGroup(SU->getInstr()))) { LLVM_DEBUG(dbgs() << " Bump cycle to " << (isTop() ? "end" : "begin") << " group\n"); bumpCycle(++NextCycle); } while (CurrMOps >= SchedModel->getIssueWidth()) { LLVM_DEBUG(dbgs() << " *** Max MOps " << CurrMOps << " at cycle " << CurrCycle << '\n'); bumpCycle(++NextCycle); } LLVM_DEBUG(dumpScheduledState()); } /// Release pending ready nodes in to the available queue. This makes them /// visible to heuristics. void SchedBoundary::releasePending() { // If the available queue is empty, it is safe to reset MinReadyCycle. if (Available.empty()) MinReadyCycle = std::numeric_limits::max(); // Check to see if any of the pending instructions are ready to issue. If // so, add them to the available queue. for (unsigned I = 0, E = Pending.size(); I < E; ++I) { SUnit *SU = *(Pending.begin() + I); unsigned ReadyCycle = isTop() ? SU->TopReadyCycle : SU->BotReadyCycle; if (ReadyCycle < MinReadyCycle) MinReadyCycle = ReadyCycle; if (Available.size() >= ReadyListLimit) break; releaseNode(SU, ReadyCycle, true, I); if (E != Pending.size()) { --I; --E; } } CheckPending = false; } /// Remove SU from the ready set for this boundary. void SchedBoundary::removeReady(SUnit *SU) { if (Available.isInQueue(SU)) Available.remove(Available.find(SU)); else { assert(Pending.isInQueue(SU) && "bad ready count"); Pending.remove(Pending.find(SU)); } } /// If this queue only has one ready candidate, return it. As a side effect, /// defer any nodes that now hit a hazard, and advance the cycle until at least /// one node is ready. If multiple instructions are ready, return NULL. SUnit *SchedBoundary::pickOnlyChoice() { if (CheckPending) releasePending(); // Defer any ready instrs that now have a hazard. for (ReadyQueue::iterator I = Available.begin(); I != Available.end();) { if (checkHazard(*I)) { Pending.push(*I); I = Available.remove(I); continue; } ++I; } for (unsigned i = 0; Available.empty(); ++i) { // FIXME: Re-enable assert once PR20057 is resolved. // assert(i <= (HazardRec->getMaxLookAhead() + MaxObservedStall) && // "permanent hazard"); (void)i; bumpCycle(CurrCycle + 1); releasePending(); } LLVM_DEBUG(Pending.dump()); LLVM_DEBUG(Available.dump()); if (Available.size() == 1) return *Available.begin(); return nullptr; } #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP) // This is useful information to dump after bumpNode. // Note that the Queue contents are more useful before pickNodeFromQueue. LLVM_DUMP_METHOD void SchedBoundary::dumpScheduledState() const { unsigned ResFactor; unsigned ResCount; if (ZoneCritResIdx) { ResFactor = SchedModel->getResourceFactor(ZoneCritResIdx); ResCount = getResourceCount(ZoneCritResIdx); } else { ResFactor = SchedModel->getMicroOpFactor(); ResCount = RetiredMOps * ResFactor; } unsigned LFactor = SchedModel->getLatencyFactor(); dbgs() << Available.getName() << " @" << CurrCycle << "c\n" << " Retired: " << RetiredMOps; dbgs() << "\n Executed: " << getExecutedCount() / LFactor << "c"; dbgs() << "\n Critical: " << ResCount / LFactor << "c, " << ResCount / ResFactor << " " << SchedModel->getResourceName(ZoneCritResIdx) << "\n ExpectedLatency: " << ExpectedLatency << "c\n" << (IsResourceLimited ? " - Resource" : " - Latency") << " limited.\n"; } #endif //===----------------------------------------------------------------------===// // GenericScheduler - Generic implementation of MachineSchedStrategy. //===----------------------------------------------------------------------===// void GenericSchedulerBase::SchedCandidate:: initResourceDelta(const ScheduleDAGMI *DAG, const TargetSchedModel *SchedModel) { if (!Policy.ReduceResIdx && !Policy.DemandResIdx) return; const MCSchedClassDesc *SC = DAG->getSchedClass(SU); for (TargetSchedModel::ProcResIter PI = SchedModel->getWriteProcResBegin(SC), PE = SchedModel->getWriteProcResEnd(SC); PI != PE; ++PI) { if (PI->ProcResourceIdx == Policy.ReduceResIdx) ResDelta.CritResources += PI->Cycles; if (PI->ProcResourceIdx == Policy.DemandResIdx) ResDelta.DemandedResources += PI->Cycles; } } /// Compute remaining latency. We need this both to determine whether the /// overall schedule has become latency-limited and whether the instructions /// outside this zone are resource or latency limited. /// /// The "dependent" latency is updated incrementally during scheduling as the /// max height/depth of scheduled nodes minus the cycles since it was /// scheduled: /// DLat = max (N.depth - (CurrCycle - N.ReadyCycle) for N in Zone /// /// The "independent" latency is the max ready queue depth: /// ILat = max N.depth for N in Available|Pending /// /// RemainingLatency is the greater of independent and dependent latency. /// /// These computations are expensive, especially in DAGs with many edges, so /// only do them if necessary. static unsigned computeRemLatency(SchedBoundary &CurrZone) { unsigned RemLatency = CurrZone.getDependentLatency(); RemLatency = std::max(RemLatency, CurrZone.findMaxLatency(CurrZone.Available.elements())); RemLatency = std::max(RemLatency, CurrZone.findMaxLatency(CurrZone.Pending.elements())); return RemLatency; } /// Returns true if the current cycle plus remaning latency is greater than /// the critical path in the scheduling region. bool GenericSchedulerBase::shouldReduceLatency(const CandPolicy &Policy, SchedBoundary &CurrZone, bool ComputeRemLatency, unsigned &RemLatency) const { // The current cycle is already greater than the critical path, so we are // already latency limited and don't need to compute the remaining latency. if (CurrZone.getCurrCycle() > Rem.CriticalPath) return true; // If we haven't scheduled anything yet, then we aren't latency limited. if (CurrZone.getCurrCycle() == 0) return false; if (ComputeRemLatency) RemLatency = computeRemLatency(CurrZone); return RemLatency + CurrZone.getCurrCycle() > Rem.CriticalPath; } /// Set the CandPolicy given a scheduling zone given the current resources and /// latencies inside and outside the zone. void GenericSchedulerBase::setPolicy(CandPolicy &Policy, bool IsPostRA, SchedBoundary &CurrZone, SchedBoundary *OtherZone) { // Apply preemptive heuristics based on the total latency and resources // inside and outside this zone. Potential stalls should be considered before // following this policy. // Compute the critical resource outside the zone. unsigned OtherCritIdx = 0; unsigned OtherCount = OtherZone ? OtherZone->getOtherResourceCount(OtherCritIdx) : 0; bool OtherResLimited = false; unsigned RemLatency = 0; bool RemLatencyComputed = false; if (SchedModel->hasInstrSchedModel() && OtherCount != 0) { RemLatency = computeRemLatency(CurrZone); RemLatencyComputed = true; OtherResLimited = checkResourceLimit(SchedModel->getLatencyFactor(), OtherCount, RemLatency, false); } // Schedule aggressively for latency in PostRA mode. We don't check for // acyclic latency during PostRA, and highly out-of-order processors will // skip PostRA scheduling. if (!OtherResLimited && (IsPostRA || shouldReduceLatency(Policy, CurrZone, !RemLatencyComputed, RemLatency))) { Policy.ReduceLatency |= true; LLVM_DEBUG(dbgs() << " " << CurrZone.Available.getName() << " RemainingLatency " << RemLatency << " + " << CurrZone.getCurrCycle() << "c > CritPath " << Rem.CriticalPath << "\n"); } // If the same resource is limiting inside and outside the zone, do nothing. if (CurrZone.getZoneCritResIdx() == OtherCritIdx) return; LLVM_DEBUG(if (CurrZone.isResourceLimited()) { dbgs() << " " << CurrZone.Available.getName() << " ResourceLimited: " << SchedModel->getResourceName(CurrZone.getZoneCritResIdx()) << "\n"; } if (OtherResLimited) dbgs() << " RemainingLimit: " << SchedModel->getResourceName(OtherCritIdx) << "\n"; if (!CurrZone.isResourceLimited() && !OtherResLimited) dbgs() << " Latency limited both directions.\n"); if (CurrZone.isResourceLimited() && !Policy.ReduceResIdx) Policy.ReduceResIdx = CurrZone.getZoneCritResIdx(); if (OtherResLimited) Policy.DemandResIdx = OtherCritIdx; } #ifndef NDEBUG const char *GenericSchedulerBase::getReasonStr( GenericSchedulerBase::CandReason Reason) { switch (Reason) { case NoCand: return "NOCAND "; case Only1: return "ONLY1 "; case PhysReg: return "PHYS-REG "; case RegExcess: return "REG-EXCESS"; case RegCritical: return "REG-CRIT "; case Stall: return "STALL "; case Cluster: return "CLUSTER "; case Weak: return "WEAK "; case RegMax: return "REG-MAX "; case ResourceReduce: return "RES-REDUCE"; case ResourceDemand: return "RES-DEMAND"; case TopDepthReduce: return "TOP-DEPTH "; case TopPathReduce: return "TOP-PATH "; case BotHeightReduce:return "BOT-HEIGHT"; case BotPathReduce: return "BOT-PATH "; case NextDefUse: return "DEF-USE "; case NodeOrder: return "ORDER "; }; llvm_unreachable("Unknown reason!"); } void GenericSchedulerBase::traceCandidate(const SchedCandidate &Cand) { PressureChange P; unsigned ResIdx = 0; unsigned Latency = 0; switch (Cand.Reason) { default: break; case RegExcess: P = Cand.RPDelta.Excess; break; case RegCritical: P = Cand.RPDelta.CriticalMax; break; case RegMax: P = Cand.RPDelta.CurrentMax; break; case ResourceReduce: ResIdx = Cand.Policy.ReduceResIdx; break; case ResourceDemand: ResIdx = Cand.Policy.DemandResIdx; break; case TopDepthReduce: Latency = Cand.SU->getDepth(); break; case TopPathReduce: Latency = Cand.SU->getHeight(); break; case BotHeightReduce: Latency = Cand.SU->getHeight(); break; case BotPathReduce: Latency = Cand.SU->getDepth(); break; } dbgs() << " Cand SU(" << Cand.SU->NodeNum << ") " << getReasonStr(Cand.Reason); if (P.isValid()) dbgs() << " " << TRI->getRegPressureSetName(P.getPSet()) << ":" << P.getUnitInc() << " "; else dbgs() << " "; if (ResIdx) dbgs() << " " << SchedModel->getProcResource(ResIdx)->Name << " "; else dbgs() << " "; if (Latency) dbgs() << " " << Latency << " cycles "; else dbgs() << " "; dbgs() << '\n'; } #endif namespace llvm { /// Return true if this heuristic determines order. bool tryLess(int TryVal, int CandVal, GenericSchedulerBase::SchedCandidate &TryCand, GenericSchedulerBase::SchedCandidate &Cand, GenericSchedulerBase::CandReason Reason) { if (TryVal < CandVal) { TryCand.Reason = Reason; return true; } if (TryVal > CandVal) { if (Cand.Reason > Reason) Cand.Reason = Reason; return true; } return false; } bool tryGreater(int TryVal, int CandVal, GenericSchedulerBase::SchedCandidate &TryCand, GenericSchedulerBase::SchedCandidate &Cand, GenericSchedulerBase::CandReason Reason) { if (TryVal > CandVal) { TryCand.Reason = Reason; return true; } if (TryVal < CandVal) { if (Cand.Reason > Reason) Cand.Reason = Reason; return true; } return false; } bool tryLatency(GenericSchedulerBase::SchedCandidate &TryCand, GenericSchedulerBase::SchedCandidate &Cand, SchedBoundary &Zone) { if (Zone.isTop()) { // Prefer the candidate with the lesser depth, but only if one of them has // depth greater than the total latency scheduled so far, otherwise either // of them could be scheduled now with no stall. if (std::max(TryCand.SU->getDepth(), Cand.SU->getDepth()) > Zone.getScheduledLatency()) { if (tryLess(TryCand.SU->getDepth(), Cand.SU->getDepth(), TryCand, Cand, GenericSchedulerBase::TopDepthReduce)) return true; } if (tryGreater(TryCand.SU->getHeight(), Cand.SU->getHeight(), TryCand, Cand, GenericSchedulerBase::TopPathReduce)) return true; } else { // Prefer the candidate with the lesser height, but only if one of them has // height greater than the total latency scheduled so far, otherwise either // of them could be scheduled now with no stall. if (std::max(TryCand.SU->getHeight(), Cand.SU->getHeight()) > Zone.getScheduledLatency()) { if (tryLess(TryCand.SU->getHeight(), Cand.SU->getHeight(), TryCand, Cand, GenericSchedulerBase::BotHeightReduce)) return true; } if (tryGreater(TryCand.SU->getDepth(), Cand.SU->getDepth(), TryCand, Cand, GenericSchedulerBase::BotPathReduce)) return true; } return false; } } // end namespace llvm static void tracePick(GenericSchedulerBase::CandReason Reason, bool IsTop) { LLVM_DEBUG(dbgs() << "Pick " << (IsTop ? "Top " : "Bot ") << GenericSchedulerBase::getReasonStr(Reason) << '\n'); } static void tracePick(const GenericSchedulerBase::SchedCandidate &Cand) { tracePick(Cand.Reason, Cand.AtTop); } void GenericScheduler::initialize(ScheduleDAGMI *dag) { assert(dag->hasVRegLiveness() && "(PreRA)GenericScheduler needs vreg liveness"); DAG = static_cast(dag); SchedModel = DAG->getSchedModel(); TRI = DAG->TRI; if (RegionPolicy.ComputeDFSResult) DAG->computeDFSResult(); Rem.init(DAG, SchedModel); Top.init(DAG, SchedModel, &Rem); Bot.init(DAG, SchedModel, &Rem); // Initialize resource counts. // Initialize the HazardRecognizers. If itineraries don't exist, are empty, or // are disabled, then these HazardRecs will be disabled. const InstrItineraryData *Itin = SchedModel->getInstrItineraries(); if (!Top.HazardRec) { Top.HazardRec = DAG->MF.getSubtarget().getInstrInfo()->CreateTargetMIHazardRecognizer( Itin, DAG); } if (!Bot.HazardRec) { Bot.HazardRec = DAG->MF.getSubtarget().getInstrInfo()->CreateTargetMIHazardRecognizer( Itin, DAG); } TopCand.SU = nullptr; BotCand.SU = nullptr; } /// Initialize the per-region scheduling policy. void GenericScheduler::initPolicy(MachineBasicBlock::iterator Begin, MachineBasicBlock::iterator End, unsigned NumRegionInstrs) { const MachineFunction &MF = *Begin->getMF(); const TargetLowering *TLI = MF.getSubtarget().getTargetLowering(); // Avoid setting up the register pressure tracker for small regions to save // compile time. As a rough heuristic, only track pressure when the number of // schedulable instructions exceeds half the integer register file. RegionPolicy.ShouldTrackPressure = true; for (unsigned VT = MVT::i32; VT > (unsigned)MVT::i1; --VT) { MVT::SimpleValueType LegalIntVT = (MVT::SimpleValueType)VT; if (TLI->isTypeLegal(LegalIntVT)) { unsigned NIntRegs = Context->RegClassInfo->getNumAllocatableRegs( TLI->getRegClassFor(LegalIntVT)); RegionPolicy.ShouldTrackPressure = NumRegionInstrs > (NIntRegs / 2); } } // For generic targets, we default to bottom-up, because it's simpler and more // compile-time optimizations have been implemented in that direction. RegionPolicy.OnlyBottomUp = true; // Allow the subtarget to override default policy. MF.getSubtarget().overrideSchedPolicy(RegionPolicy, NumRegionInstrs); // After subtarget overrides, apply command line options. if (!EnableRegPressure) { RegionPolicy.ShouldTrackPressure = false; RegionPolicy.ShouldTrackLaneMasks = false; } // Check -misched-topdown/bottomup can force or unforce scheduling direction. // e.g. -misched-bottomup=false allows scheduling in both directions. assert((!ForceTopDown || !ForceBottomUp) && "-misched-topdown incompatible with -misched-bottomup"); if (ForceBottomUp.getNumOccurrences() > 0) { RegionPolicy.OnlyBottomUp = ForceBottomUp; if (RegionPolicy.OnlyBottomUp) RegionPolicy.OnlyTopDown = false; } if (ForceTopDown.getNumOccurrences() > 0) { RegionPolicy.OnlyTopDown = ForceTopDown; if (RegionPolicy.OnlyTopDown) RegionPolicy.OnlyBottomUp = false; } } void GenericScheduler::dumpPolicy() const { // Cannot completely remove virtual function even in release mode. #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP) dbgs() << "GenericScheduler RegionPolicy: " << " ShouldTrackPressure=" << RegionPolicy.ShouldTrackPressure << " OnlyTopDown=" << RegionPolicy.OnlyTopDown << " OnlyBottomUp=" << RegionPolicy.OnlyBottomUp << "\n"; #endif } /// Set IsAcyclicLatencyLimited if the acyclic path is longer than the cyclic /// critical path by more cycles than it takes to drain the instruction buffer. /// We estimate an upper bounds on in-flight instructions as: /// /// CyclesPerIteration = max( CyclicPath, Loop-Resource-Height ) /// InFlightIterations = AcyclicPath / CyclesPerIteration /// InFlightResources = InFlightIterations * LoopResources /// /// TODO: Check execution resources in addition to IssueCount. void GenericScheduler::checkAcyclicLatency() { if (Rem.CyclicCritPath == 0 || Rem.CyclicCritPath >= Rem.CriticalPath) return; // Scaled number of cycles per loop iteration. unsigned IterCount = std::max(Rem.CyclicCritPath * SchedModel->getLatencyFactor(), Rem.RemIssueCount); // Scaled acyclic critical path. unsigned AcyclicCount = Rem.CriticalPath * SchedModel->getLatencyFactor(); // InFlightCount = (AcyclicPath / IterCycles) * InstrPerLoop unsigned InFlightCount = (AcyclicCount * Rem.RemIssueCount + IterCount-1) / IterCount; unsigned BufferLimit = SchedModel->getMicroOpBufferSize() * SchedModel->getMicroOpFactor(); Rem.IsAcyclicLatencyLimited = InFlightCount > BufferLimit; LLVM_DEBUG( dbgs() << "IssueCycles=" << Rem.RemIssueCount / SchedModel->getLatencyFactor() << "c " << "IterCycles=" << IterCount / SchedModel->getLatencyFactor() << "c NumIters=" << (AcyclicCount + IterCount - 1) / IterCount << " InFlight=" << InFlightCount / SchedModel->getMicroOpFactor() << "m BufferLim=" << SchedModel->getMicroOpBufferSize() << "m\n"; if (Rem.IsAcyclicLatencyLimited) dbgs() << " ACYCLIC LATENCY LIMIT\n"); } void GenericScheduler::registerRoots() { Rem.CriticalPath = DAG->ExitSU.getDepth(); // Some roots may not feed into ExitSU. Check all of them in case. for (const SUnit *SU : Bot.Available) { if (SU->getDepth() > Rem.CriticalPath) Rem.CriticalPath = SU->getDepth(); } LLVM_DEBUG(dbgs() << "Critical Path(GS-RR ): " << Rem.CriticalPath << '\n'); if (DumpCriticalPathLength) { errs() << "Critical Path(GS-RR ): " << Rem.CriticalPath << " \n"; } if (EnableCyclicPath && SchedModel->getMicroOpBufferSize() > 0) { Rem.CyclicCritPath = DAG->computeCyclicCriticalPath(); checkAcyclicLatency(); } } namespace llvm { bool tryPressure(const PressureChange &TryP, const PressureChange &CandP, GenericSchedulerBase::SchedCandidate &TryCand, GenericSchedulerBase::SchedCandidate &Cand, GenericSchedulerBase::CandReason Reason, const TargetRegisterInfo *TRI, const MachineFunction &MF) { // If one candidate decreases and the other increases, go with it. // Invalid candidates have UnitInc==0. if (tryGreater(TryP.getUnitInc() < 0, CandP.getUnitInc() < 0, TryCand, Cand, Reason)) { return true; } // Do not compare the magnitude of pressure changes between top and bottom // boundary. if (Cand.AtTop != TryCand.AtTop) return false; // If both candidates affect the same set in the same boundary, go with the // smallest increase. unsigned TryPSet = TryP.getPSetOrMax(); unsigned CandPSet = CandP.getPSetOrMax(); if (TryPSet == CandPSet) { return tryLess(TryP.getUnitInc(), CandP.getUnitInc(), TryCand, Cand, Reason); } int TryRank = TryP.isValid() ? TRI->getRegPressureSetScore(MF, TryPSet) : std::numeric_limits::max(); int CandRank = CandP.isValid() ? TRI->getRegPressureSetScore(MF, CandPSet) : std::numeric_limits::max(); // If the candidates are decreasing pressure, reverse priority. if (TryP.getUnitInc() < 0) std::swap(TryRank, CandRank); return tryGreater(TryRank, CandRank, TryCand, Cand, Reason); } unsigned getWeakLeft(const SUnit *SU, bool isTop) { return (isTop) ? SU->WeakPredsLeft : SU->WeakSuccsLeft; } /// Minimize physical register live ranges. Regalloc wants them adjacent to /// their physreg def/use. /// /// FIXME: This is an unnecessary check on the critical path. Most are root/leaf /// copies which can be prescheduled. The rest (e.g. x86 MUL) could be bundled /// with the operation that produces or consumes the physreg. We'll do this when /// regalloc has support for parallel copies. int biasPhysReg(const SUnit *SU, bool isTop) { const MachineInstr *MI = SU->getInstr(); if (MI->isCopy()) { unsigned ScheduledOper = isTop ? 1 : 0; unsigned UnscheduledOper = isTop ? 0 : 1; // If we have already scheduled the physreg produce/consumer, immediately // schedule the copy. if (Register::isPhysicalRegister(MI->getOperand(ScheduledOper).getReg())) return 1; // If the physreg is at the boundary, defer it. Otherwise schedule it // immediately to free the dependent. We can hoist the copy later. bool AtBoundary = isTop ? !SU->NumSuccsLeft : !SU->NumPredsLeft; if (Register::isPhysicalRegister(MI->getOperand(UnscheduledOper).getReg())) return AtBoundary ? -1 : 1; } if (MI->isMoveImmediate()) { // If we have a move immediate and all successors have been assigned, bias // towards scheduling this later. Make sure all register defs are to // physical registers. bool DoBias = true; for (const MachineOperand &Op : MI->defs()) { if (Op.isReg() && !Register::isPhysicalRegister(Op.getReg())) { DoBias = false; break; } } if (DoBias) return isTop ? -1 : 1; } return 0; } } // end namespace llvm void GenericScheduler::initCandidate(SchedCandidate &Cand, SUnit *SU, bool AtTop, const RegPressureTracker &RPTracker, RegPressureTracker &TempTracker) { Cand.SU = SU; Cand.AtTop = AtTop; if (DAG->isTrackingPressure()) { if (AtTop) { TempTracker.getMaxDownwardPressureDelta( Cand.SU->getInstr(), Cand.RPDelta, DAG->getRegionCriticalPSets(), DAG->getRegPressure().MaxSetPressure); } else { if (VerifyScheduling) { TempTracker.getMaxUpwardPressureDelta( Cand.SU->getInstr(), &DAG->getPressureDiff(Cand.SU), Cand.RPDelta, DAG->getRegionCriticalPSets(), DAG->getRegPressure().MaxSetPressure); } else { RPTracker.getUpwardPressureDelta( Cand.SU->getInstr(), DAG->getPressureDiff(Cand.SU), Cand.RPDelta, DAG->getRegionCriticalPSets(), DAG->getRegPressure().MaxSetPressure); } } } LLVM_DEBUG(if (Cand.RPDelta.Excess.isValid()) dbgs() << " Try SU(" << Cand.SU->NodeNum << ") " << TRI->getRegPressureSetName(Cand.RPDelta.Excess.getPSet()) << ":" << Cand.RPDelta.Excess.getUnitInc() << "\n"); } /// Apply a set of heuristics to a new candidate. Heuristics are currently /// hierarchical. This may be more efficient than a graduated cost model because /// we don't need to evaluate all aspects of the model for each node in the /// queue. But it's really done to make the heuristics easier to debug and /// statistically analyze. /// /// \param Cand provides the policy and current best candidate. /// \param TryCand refers to the next SUnit candidate, otherwise uninitialized. /// \param Zone describes the scheduled zone that we are extending, or nullptr // if Cand is from a different zone than TryCand. void GenericScheduler::tryCandidate(SchedCandidate &Cand, SchedCandidate &TryCand, SchedBoundary *Zone) const { // Initialize the candidate if needed. if (!Cand.isValid()) { TryCand.Reason = NodeOrder; return; } // Bias PhysReg Defs and copies to their uses and defined respectively. if (tryGreater(biasPhysReg(TryCand.SU, TryCand.AtTop), biasPhysReg(Cand.SU, Cand.AtTop), TryCand, Cand, PhysReg)) return; // Avoid exceeding the target's limit. if (DAG->isTrackingPressure() && tryPressure(TryCand.RPDelta.Excess, Cand.RPDelta.Excess, TryCand, Cand, RegExcess, TRI, DAG->MF)) return; // Avoid increasing the max critical pressure in the scheduled region. if (DAG->isTrackingPressure() && tryPressure(TryCand.RPDelta.CriticalMax, Cand.RPDelta.CriticalMax, TryCand, Cand, RegCritical, TRI, DAG->MF)) return; // We only compare a subset of features when comparing nodes between // Top and Bottom boundary. Some properties are simply incomparable, in many // other instances we should only override the other boundary if something // is a clear good pick on one boundary. Skip heuristics that are more // "tie-breaking" in nature. bool SameBoundary = Zone != nullptr; if (SameBoundary) { // For loops that are acyclic path limited, aggressively schedule for // latency. Within an single cycle, whenever CurrMOps > 0, allow normal // heuristics to take precedence. if (Rem.IsAcyclicLatencyLimited && !Zone->getCurrMOps() && tryLatency(TryCand, Cand, *Zone)) return; // Prioritize instructions that read unbuffered resources by stall cycles. if (tryLess(Zone->getLatencyStallCycles(TryCand.SU), Zone->getLatencyStallCycles(Cand.SU), TryCand, Cand, Stall)) return; } // Keep clustered nodes together to encourage downstream peephole // optimizations which may reduce resource requirements. // // This is a best effort to set things up for a post-RA pass. Optimizations // like generating loads of multiple registers should ideally be done within // the scheduler pass by combining the loads during DAG postprocessing. const SUnit *CandNextClusterSU = Cand.AtTop ? DAG->getNextClusterSucc() : DAG->getNextClusterPred(); const SUnit *TryCandNextClusterSU = TryCand.AtTop ? DAG->getNextClusterSucc() : DAG->getNextClusterPred(); if (tryGreater(TryCand.SU == TryCandNextClusterSU, Cand.SU == CandNextClusterSU, TryCand, Cand, Cluster)) return; if (SameBoundary) { // Weak edges are for clustering and other constraints. if (tryLess(getWeakLeft(TryCand.SU, TryCand.AtTop), getWeakLeft(Cand.SU, Cand.AtTop), TryCand, Cand, Weak)) return; } // Avoid increasing the max pressure of the entire region. if (DAG->isTrackingPressure() && tryPressure(TryCand.RPDelta.CurrentMax, Cand.RPDelta.CurrentMax, TryCand, Cand, RegMax, TRI, DAG->MF)) return; if (SameBoundary) { // Avoid critical resource consumption and balance the schedule. TryCand.initResourceDelta(DAG, SchedModel); if (tryLess(TryCand.ResDelta.CritResources, Cand.ResDelta.CritResources, TryCand, Cand, ResourceReduce)) return; if (tryGreater(TryCand.ResDelta.DemandedResources, Cand.ResDelta.DemandedResources, TryCand, Cand, ResourceDemand)) return; // Avoid serializing long latency dependence chains. // For acyclic path limited loops, latency was already checked above. if (!RegionPolicy.DisableLatencyHeuristic && TryCand.Policy.ReduceLatency && !Rem.IsAcyclicLatencyLimited && tryLatency(TryCand, Cand, *Zone)) return; // Fall through to original instruction order. if ((Zone->isTop() && TryCand.SU->NodeNum < Cand.SU->NodeNum) || (!Zone->isTop() && TryCand.SU->NodeNum > Cand.SU->NodeNum)) { TryCand.Reason = NodeOrder; } } } /// Pick the best candidate from the queue. /// /// TODO: getMaxPressureDelta results can be mostly cached for each SUnit during /// DAG building. To adjust for the current scheduling location we need to /// maintain the number of vreg uses remaining to be top-scheduled. void GenericScheduler::pickNodeFromQueue(SchedBoundary &Zone, const CandPolicy &ZonePolicy, const RegPressureTracker &RPTracker, SchedCandidate &Cand) { // getMaxPressureDelta temporarily modifies the tracker. RegPressureTracker &TempTracker = const_cast(RPTracker); ReadyQueue &Q = Zone.Available; for (SUnit *SU : Q) { SchedCandidate TryCand(ZonePolicy); initCandidate(TryCand, SU, Zone.isTop(), RPTracker, TempTracker); // Pass SchedBoundary only when comparing nodes from the same boundary. SchedBoundary *ZoneArg = Cand.AtTop == TryCand.AtTop ? &Zone : nullptr; tryCandidate(Cand, TryCand, ZoneArg); if (TryCand.Reason != NoCand) { // Initialize resource delta if needed in case future heuristics query it. if (TryCand.ResDelta == SchedResourceDelta()) TryCand.initResourceDelta(DAG, SchedModel); Cand.setBest(TryCand); LLVM_DEBUG(traceCandidate(Cand)); } } } /// Pick the best candidate node from either the top or bottom queue. SUnit *GenericScheduler::pickNodeBidirectional(bool &IsTopNode) { // Schedule as far as possible in the direction of no choice. This is most // efficient, but also provides the best heuristics for CriticalPSets. if (SUnit *SU = Bot.pickOnlyChoice()) { IsTopNode = false; tracePick(Only1, false); return SU; } if (SUnit *SU = Top.pickOnlyChoice()) { IsTopNode = true; tracePick(Only1, true); return SU; } // Set the bottom-up policy based on the state of the current bottom zone and // the instructions outside the zone, including the top zone. CandPolicy BotPolicy; setPolicy(BotPolicy, /*IsPostRA=*/false, Bot, &Top); // Set the top-down policy based on the state of the current top zone and // the instructions outside the zone, including the bottom zone. CandPolicy TopPolicy; setPolicy(TopPolicy, /*IsPostRA=*/false, Top, &Bot); // See if BotCand is still valid (because we previously scheduled from Top). LLVM_DEBUG(dbgs() << "Picking from Bot:\n"); if (!BotCand.isValid() || BotCand.SU->isScheduled || BotCand.Policy != BotPolicy) { BotCand.reset(CandPolicy()); pickNodeFromQueue(Bot, BotPolicy, DAG->getBotRPTracker(), BotCand); assert(BotCand.Reason != NoCand && "failed to find the first candidate"); } else { LLVM_DEBUG(traceCandidate(BotCand)); #ifndef NDEBUG if (VerifyScheduling) { SchedCandidate TCand; TCand.reset(CandPolicy()); pickNodeFromQueue(Bot, BotPolicy, DAG->getBotRPTracker(), TCand); assert(TCand.SU == BotCand.SU && "Last pick result should correspond to re-picking right now"); } #endif } // Check if the top Q has a better candidate. LLVM_DEBUG(dbgs() << "Picking from Top:\n"); if (!TopCand.isValid() || TopCand.SU->isScheduled || TopCand.Policy != TopPolicy) { TopCand.reset(CandPolicy()); pickNodeFromQueue(Top, TopPolicy, DAG->getTopRPTracker(), TopCand); assert(TopCand.Reason != NoCand && "failed to find the first candidate"); } else { LLVM_DEBUG(traceCandidate(TopCand)); #ifndef NDEBUG if (VerifyScheduling) { SchedCandidate TCand; TCand.reset(CandPolicy()); pickNodeFromQueue(Top, TopPolicy, DAG->getTopRPTracker(), TCand); assert(TCand.SU == TopCand.SU && "Last pick result should correspond to re-picking right now"); } #endif } // Pick best from BotCand and TopCand. assert(BotCand.isValid()); assert(TopCand.isValid()); SchedCandidate Cand = BotCand; TopCand.Reason = NoCand; tryCandidate(Cand, TopCand, nullptr); if (TopCand.Reason != NoCand) { Cand.setBest(TopCand); LLVM_DEBUG(traceCandidate(Cand)); } IsTopNode = Cand.AtTop; tracePick(Cand); return Cand.SU; } /// Pick the best node to balance the schedule. Implements MachineSchedStrategy. SUnit *GenericScheduler::pickNode(bool &IsTopNode) { if (DAG->top() == DAG->bottom()) { assert(Top.Available.empty() && Top.Pending.empty() && Bot.Available.empty() && Bot.Pending.empty() && "ReadyQ garbage"); return nullptr; } SUnit *SU; do { if (RegionPolicy.OnlyTopDown) { SU = Top.pickOnlyChoice(); if (!SU) { CandPolicy NoPolicy; TopCand.reset(NoPolicy); pickNodeFromQueue(Top, NoPolicy, DAG->getTopRPTracker(), TopCand); assert(TopCand.Reason != NoCand && "failed to find a candidate"); tracePick(TopCand); SU = TopCand.SU; } IsTopNode = true; } else if (RegionPolicy.OnlyBottomUp) { SU = Bot.pickOnlyChoice(); if (!SU) { CandPolicy NoPolicy; BotCand.reset(NoPolicy); pickNodeFromQueue(Bot, NoPolicy, DAG->getBotRPTracker(), BotCand); assert(BotCand.Reason != NoCand && "failed to find a candidate"); tracePick(BotCand); SU = BotCand.SU; } IsTopNode = false; } else { SU = pickNodeBidirectional(IsTopNode); } } while (SU->isScheduled); if (SU->isTopReady()) Top.removeReady(SU); if (SU->isBottomReady()) Bot.removeReady(SU); LLVM_DEBUG(dbgs() << "Scheduling SU(" << SU->NodeNum << ") " << *SU->getInstr()); return SU; } void GenericScheduler::reschedulePhysReg(SUnit *SU, bool isTop) { MachineBasicBlock::iterator InsertPos = SU->getInstr(); if (!isTop) ++InsertPos; SmallVectorImpl &Deps = isTop ? SU->Preds : SU->Succs; // Find already scheduled copies with a single physreg dependence and move // them just above the scheduled instruction. for (SDep &Dep : Deps) { if (Dep.getKind() != SDep::Data || !Register::isPhysicalRegister(Dep.getReg())) continue; SUnit *DepSU = Dep.getSUnit(); if (isTop ? DepSU->Succs.size() > 1 : DepSU->Preds.size() > 1) continue; MachineInstr *Copy = DepSU->getInstr(); if (!Copy->isCopy() && !Copy->isMoveImmediate()) continue; LLVM_DEBUG(dbgs() << " Rescheduling physreg copy "; DAG->dumpNode(*Dep.getSUnit())); DAG->moveInstruction(Copy, InsertPos); } } /// Update the scheduler's state after scheduling a node. This is the same node /// that was just returned by pickNode(). However, ScheduleDAGMILive needs to /// update it's state based on the current cycle before MachineSchedStrategy /// does. /// /// FIXME: Eventually, we may bundle physreg copies rather than rescheduling /// them here. See comments in biasPhysReg. void GenericScheduler::schedNode(SUnit *SU, bool IsTopNode) { if (IsTopNode) { SU->TopReadyCycle = std::max(SU->TopReadyCycle, Top.getCurrCycle()); Top.bumpNode(SU); if (SU->hasPhysRegUses) reschedulePhysReg(SU, true); } else { SU->BotReadyCycle = std::max(SU->BotReadyCycle, Bot.getCurrCycle()); Bot.bumpNode(SU); if (SU->hasPhysRegDefs) reschedulePhysReg(SU, false); } } /// Create the standard converging machine scheduler. This will be used as the /// default scheduler if the target does not set a default. ScheduleDAGMILive *llvm::createGenericSchedLive(MachineSchedContext *C) { ScheduleDAGMILive *DAG = new ScheduleDAGMILive(C, std::make_unique(C)); // Register DAG post-processors. // // FIXME: extend the mutation API to allow earlier mutations to instantiate // data and pass it to later mutations. Have a single mutation that gathers // the interesting nodes in one pass. DAG->addMutation(createCopyConstrainDAGMutation(DAG->TII, DAG->TRI)); return DAG; } static ScheduleDAGInstrs *createConveringSched(MachineSchedContext *C) { return createGenericSchedLive(C); } static MachineSchedRegistry GenericSchedRegistry("converge", "Standard converging scheduler.", createConveringSched); //===----------------------------------------------------------------------===// // PostGenericScheduler - Generic PostRA implementation of MachineSchedStrategy. //===----------------------------------------------------------------------===// void PostGenericScheduler::initialize(ScheduleDAGMI *Dag) { DAG = Dag; SchedModel = DAG->getSchedModel(); TRI = DAG->TRI; Rem.init(DAG, SchedModel); Top.init(DAG, SchedModel, &Rem); BotRoots.clear(); // Initialize the HazardRecognizers. If itineraries don't exist, are empty, // or are disabled, then these HazardRecs will be disabled. const InstrItineraryData *Itin = SchedModel->getInstrItineraries(); if (!Top.HazardRec) { Top.HazardRec = DAG->MF.getSubtarget().getInstrInfo()->CreateTargetMIHazardRecognizer( Itin, DAG); } } void PostGenericScheduler::registerRoots() { Rem.CriticalPath = DAG->ExitSU.getDepth(); // Some roots may not feed into ExitSU. Check all of them in case. for (const SUnit *SU : BotRoots) { if (SU->getDepth() > Rem.CriticalPath) Rem.CriticalPath = SU->getDepth(); } LLVM_DEBUG(dbgs() << "Critical Path: (PGS-RR) " << Rem.CriticalPath << '\n'); if (DumpCriticalPathLength) { errs() << "Critical Path(PGS-RR ): " << Rem.CriticalPath << " \n"; } } /// Apply a set of heuristics to a new candidate for PostRA scheduling. /// /// \param Cand provides the policy and current best candidate. /// \param TryCand refers to the next SUnit candidate, otherwise uninitialized. void PostGenericScheduler::tryCandidate(SchedCandidate &Cand, SchedCandidate &TryCand) { // Initialize the candidate if needed. if (!Cand.isValid()) { TryCand.Reason = NodeOrder; return; } // Prioritize instructions that read unbuffered resources by stall cycles. if (tryLess(Top.getLatencyStallCycles(TryCand.SU), Top.getLatencyStallCycles(Cand.SU), TryCand, Cand, Stall)) return; // Keep clustered nodes together. if (tryGreater(TryCand.SU == DAG->getNextClusterSucc(), Cand.SU == DAG->getNextClusterSucc(), TryCand, Cand, Cluster)) return; // Avoid critical resource consumption and balance the schedule. if (tryLess(TryCand.ResDelta.CritResources, Cand.ResDelta.CritResources, TryCand, Cand, ResourceReduce)) return; if (tryGreater(TryCand.ResDelta.DemandedResources, Cand.ResDelta.DemandedResources, TryCand, Cand, ResourceDemand)) return; // Avoid serializing long latency dependence chains. if (Cand.Policy.ReduceLatency && tryLatency(TryCand, Cand, Top)) { return; } // Fall through to original instruction order. if (TryCand.SU->NodeNum < Cand.SU->NodeNum) TryCand.Reason = NodeOrder; } void PostGenericScheduler::pickNodeFromQueue(SchedCandidate &Cand) { ReadyQueue &Q = Top.Available; for (SUnit *SU : Q) { SchedCandidate TryCand(Cand.Policy); TryCand.SU = SU; TryCand.AtTop = true; TryCand.initResourceDelta(DAG, SchedModel); tryCandidate(Cand, TryCand); if (TryCand.Reason != NoCand) { Cand.setBest(TryCand); LLVM_DEBUG(traceCandidate(Cand)); } } } /// Pick the next node to schedule. SUnit *PostGenericScheduler::pickNode(bool &IsTopNode) { if (DAG->top() == DAG->bottom()) { assert(Top.Available.empty() && Top.Pending.empty() && "ReadyQ garbage"); return nullptr; } SUnit *SU; do { SU = Top.pickOnlyChoice(); if (SU) { tracePick(Only1, true); } else { CandPolicy NoPolicy; SchedCandidate TopCand(NoPolicy); // Set the top-down policy based on the state of the current top zone and // the instructions outside the zone, including the bottom zone. setPolicy(TopCand.Policy, /*IsPostRA=*/true, Top, nullptr); pickNodeFromQueue(TopCand); assert(TopCand.Reason != NoCand && "failed to find a candidate"); tracePick(TopCand); SU = TopCand.SU; } } while (SU->isScheduled); IsTopNode = true; Top.removeReady(SU); LLVM_DEBUG(dbgs() << "Scheduling SU(" << SU->NodeNum << ") " << *SU->getInstr()); return SU; } /// Called after ScheduleDAGMI has scheduled an instruction and updated /// scheduled/remaining flags in the DAG nodes. void PostGenericScheduler::schedNode(SUnit *SU, bool IsTopNode) { SU->TopReadyCycle = std::max(SU->TopReadyCycle, Top.getCurrCycle()); Top.bumpNode(SU); } ScheduleDAGMI *llvm::createGenericSchedPostRA(MachineSchedContext *C) { return new ScheduleDAGMI(C, std::make_unique(C), /*RemoveKillFlags=*/true); } //===----------------------------------------------------------------------===// // ILP Scheduler. Currently for experimental analysis of heuristics. //===----------------------------------------------------------------------===// namespace { /// Order nodes by the ILP metric. struct ILPOrder { const SchedDFSResult *DFSResult = nullptr; const BitVector *ScheduledTrees = nullptr; bool MaximizeILP; ILPOrder(bool MaxILP) : MaximizeILP(MaxILP) {} /// Apply a less-than relation on node priority. /// /// (Return true if A comes after B in the Q.) bool operator()(const SUnit *A, const SUnit *B) const { unsigned SchedTreeA = DFSResult->getSubtreeID(A); unsigned SchedTreeB = DFSResult->getSubtreeID(B); if (SchedTreeA != SchedTreeB) { // Unscheduled trees have lower priority. if (ScheduledTrees->test(SchedTreeA) != ScheduledTrees->test(SchedTreeB)) return ScheduledTrees->test(SchedTreeB); // Trees with shallower connections have have lower priority. if (DFSResult->getSubtreeLevel(SchedTreeA) != DFSResult->getSubtreeLevel(SchedTreeB)) { return DFSResult->getSubtreeLevel(SchedTreeA) < DFSResult->getSubtreeLevel(SchedTreeB); } } if (MaximizeILP) return DFSResult->getILP(A) < DFSResult->getILP(B); else return DFSResult->getILP(A) > DFSResult->getILP(B); } }; /// Schedule based on the ILP metric. class ILPScheduler : public MachineSchedStrategy { ScheduleDAGMILive *DAG = nullptr; ILPOrder Cmp; std::vector ReadyQ; public: ILPScheduler(bool MaximizeILP) : Cmp(MaximizeILP) {} void initialize(ScheduleDAGMI *dag) override { assert(dag->hasVRegLiveness() && "ILPScheduler needs vreg liveness"); DAG = static_cast(dag); DAG->computeDFSResult(); Cmp.DFSResult = DAG->getDFSResult(); Cmp.ScheduledTrees = &DAG->getScheduledTrees(); ReadyQ.clear(); } void registerRoots() override { // Restore the heap in ReadyQ with the updated DFS results. std::make_heap(ReadyQ.begin(), ReadyQ.end(), Cmp); } /// Implement MachineSchedStrategy interface. /// ----------------------------------------- /// Callback to select the highest priority node from the ready Q. SUnit *pickNode(bool &IsTopNode) override { if (ReadyQ.empty()) return nullptr; std::pop_heap(ReadyQ.begin(), ReadyQ.end(), Cmp); SUnit *SU = ReadyQ.back(); ReadyQ.pop_back(); IsTopNode = false; LLVM_DEBUG(dbgs() << "Pick node " << "SU(" << SU->NodeNum << ") " << " ILP: " << DAG->getDFSResult()->getILP(SU) << " Tree: " << DAG->getDFSResult()->getSubtreeID(SU) << " @" << DAG->getDFSResult()->getSubtreeLevel( DAG->getDFSResult()->getSubtreeID(SU)) << '\n' << "Scheduling " << *SU->getInstr()); return SU; } /// Scheduler callback to notify that a new subtree is scheduled. void scheduleTree(unsigned SubtreeID) override { std::make_heap(ReadyQ.begin(), ReadyQ.end(), Cmp); } /// Callback after a node is scheduled. Mark a newly scheduled tree, notify /// DFSResults, and resort the priority Q. void schedNode(SUnit *SU, bool IsTopNode) override { assert(!IsTopNode && "SchedDFSResult needs bottom-up"); } void releaseTopNode(SUnit *) override { /*only called for top roots*/ } void releaseBottomNode(SUnit *SU) override { ReadyQ.push_back(SU); std::push_heap(ReadyQ.begin(), ReadyQ.end(), Cmp); } }; } // end anonymous namespace static ScheduleDAGInstrs *createILPMaxScheduler(MachineSchedContext *C) { return new ScheduleDAGMILive(C, std::make_unique(true)); } static ScheduleDAGInstrs *createILPMinScheduler(MachineSchedContext *C) { return new ScheduleDAGMILive(C, std::make_unique(false)); } static MachineSchedRegistry ILPMaxRegistry( "ilpmax", "Schedule bottom-up for max ILP", createILPMaxScheduler); static MachineSchedRegistry ILPMinRegistry( "ilpmin", "Schedule bottom-up for min ILP", createILPMinScheduler); //===----------------------------------------------------------------------===// // Machine Instruction Shuffler for Correctness Testing //===----------------------------------------------------------------------===// #ifndef NDEBUG namespace { /// Apply a less-than relation on the node order, which corresponds to the /// instruction order prior to scheduling. IsReverse implements greater-than. template struct SUnitOrder { bool operator()(SUnit *A, SUnit *B) const { if (IsReverse) return A->NodeNum > B->NodeNum; else return A->NodeNum < B->NodeNum; } }; /// Reorder instructions as much as possible. class InstructionShuffler : public MachineSchedStrategy { bool IsAlternating; bool IsTopDown; // Using a less-than relation (SUnitOrder) for the TopQ priority // gives nodes with a higher number higher priority causing the latest // instructions to be scheduled first. PriorityQueue, SUnitOrder> TopQ; // When scheduling bottom-up, use greater-than as the queue priority. PriorityQueue, SUnitOrder> BottomQ; public: InstructionShuffler(bool alternate, bool topdown) : IsAlternating(alternate), IsTopDown(topdown) {} void initialize(ScheduleDAGMI*) override { TopQ.clear(); BottomQ.clear(); } /// Implement MachineSchedStrategy interface. /// ----------------------------------------- SUnit *pickNode(bool &IsTopNode) override { SUnit *SU; if (IsTopDown) { do { if (TopQ.empty()) return nullptr; SU = TopQ.top(); TopQ.pop(); } while (SU->isScheduled); IsTopNode = true; } else { do { if (BottomQ.empty()) return nullptr; SU = BottomQ.top(); BottomQ.pop(); } while (SU->isScheduled); IsTopNode = false; } if (IsAlternating) IsTopDown = !IsTopDown; return SU; } void schedNode(SUnit *SU, bool IsTopNode) override {} void releaseTopNode(SUnit *SU) override { TopQ.push(SU); } void releaseBottomNode(SUnit *SU) override { BottomQ.push(SU); } }; } // end anonymous namespace static ScheduleDAGInstrs *createInstructionShuffler(MachineSchedContext *C) { bool Alternate = !ForceTopDown && !ForceBottomUp; bool TopDown = !ForceBottomUp; assert((TopDown || !ForceTopDown) && "-misched-topdown incompatible with -misched-bottomup"); return new ScheduleDAGMILive( C, std::make_unique(Alternate, TopDown)); } static MachineSchedRegistry ShufflerRegistry( "shuffle", "Shuffle machine instructions alternating directions", createInstructionShuffler); #endif // !NDEBUG //===----------------------------------------------------------------------===// // GraphWriter support for ScheduleDAGMILive. //===----------------------------------------------------------------------===// #ifndef NDEBUG namespace llvm { template<> struct GraphTraits< ScheduleDAGMI*> : public GraphTraits {}; template<> struct DOTGraphTraits : public DefaultDOTGraphTraits { DOTGraphTraits(bool isSimple = false) : DefaultDOTGraphTraits(isSimple) {} static std::string getGraphName(const ScheduleDAG *G) { return std::string(G->MF.getName()); } static bool renderGraphFromBottomUp() { return true; } static bool isNodeHidden(const SUnit *Node) { if (ViewMISchedCutoff == 0) return false; return (Node->Preds.size() > ViewMISchedCutoff || Node->Succs.size() > ViewMISchedCutoff); } /// If you want to override the dot attributes printed for a particular /// edge, override this method. static std::string getEdgeAttributes(const SUnit *Node, SUnitIterator EI, const ScheduleDAG *Graph) { if (EI.isArtificialDep()) return "color=cyan,style=dashed"; if (EI.isCtrlDep()) return "color=blue,style=dashed"; return ""; } static std::string getNodeLabel(const SUnit *SU, const ScheduleDAG *G) { std::string Str; raw_string_ostream SS(Str); const ScheduleDAGMI *DAG = static_cast(G); const SchedDFSResult *DFS = DAG->hasVRegLiveness() ? static_cast(G)->getDFSResult() : nullptr; SS << "SU:" << SU->NodeNum; if (DFS) SS << " I:" << DFS->getNumInstrs(SU); return SS.str(); } static std::string getNodeDescription(const SUnit *SU, const ScheduleDAG *G) { return G->getGraphNodeLabel(SU); } static std::string getNodeAttributes(const SUnit *N, const ScheduleDAG *G) { std::string Str("shape=Mrecord"); const ScheduleDAGMI *DAG = static_cast(G); const SchedDFSResult *DFS = DAG->hasVRegLiveness() ? static_cast(G)->getDFSResult() : nullptr; if (DFS) { Str += ",style=filled,fillcolor=\"#"; Str += DOT::getColorString(DFS->getSubtreeID(N)); Str += '"'; } return Str; } }; } // end namespace llvm #endif // NDEBUG /// viewGraph - Pop up a ghostview window with the reachable parts of the DAG /// rendered using 'dot'. void ScheduleDAGMI::viewGraph(const Twine &Name, const Twine &Title) { #ifndef NDEBUG ViewGraph(this, Name, false, Title); #else errs() << "ScheduleDAGMI::viewGraph is only available in debug builds on " << "systems with Graphviz or gv!\n"; #endif // NDEBUG } /// Out-of-line implementation with no arguments is handy for gdb. void ScheduleDAGMI::viewGraph() { viewGraph(getDAGName(), "Scheduling-Units Graph for " + getDAGName()); }