mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-23 19:23:23 +01:00
de89474ce3
Push VRegUses/collectVRegUses() down the class hierarchy towards its only user ScheduleDAGMILive. NFCI: The initialization of the map happens at a later point but that should not matter. This is in preparation to allow DAG mutators to merge nodes, which relies on this map getting computed later. llvm-svn: 286654
354 lines
14 KiB
C++
354 lines
14 KiB
C++
//==- ScheduleDAGInstrs.h - MachineInstr Scheduling --------------*- C++ -*-==//
|
|
//
|
|
// The LLVM Compiler Infrastructure
|
|
//
|
|
// This file is distributed under the University of Illinois Open Source
|
|
// License. See LICENSE.TXT for details.
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
//
|
|
// This file implements the ScheduleDAGInstrs class, which implements
|
|
// scheduling for a MachineInstr-based dependency graph.
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
#ifndef LLVM_CODEGEN_SCHEDULEDAGINSTRS_H
|
|
#define LLVM_CODEGEN_SCHEDULEDAGINSTRS_H
|
|
|
|
#include "llvm/ADT/MapVector.h"
|
|
#include "llvm/ADT/SparseMultiSet.h"
|
|
#include "llvm/ADT/SparseSet.h"
|
|
#include "llvm/CodeGen/ScheduleDAG.h"
|
|
#include "llvm/CodeGen/TargetSchedule.h"
|
|
#include "llvm/Support/Compiler.h"
|
|
#include "llvm/Target/TargetRegisterInfo.h"
|
|
#include <list>
|
|
|
|
namespace llvm {
|
|
class MachineFrameInfo;
|
|
class MachineLoopInfo;
|
|
class MachineDominatorTree;
|
|
class RegPressureTracker;
|
|
class PressureDiffs;
|
|
|
|
/// An individual mapping from virtual register number to SUnit.
|
|
struct VReg2SUnit {
|
|
unsigned VirtReg;
|
|
LaneBitmask LaneMask;
|
|
SUnit *SU;
|
|
|
|
VReg2SUnit(unsigned VReg, LaneBitmask LaneMask, SUnit *SU)
|
|
: VirtReg(VReg), LaneMask(LaneMask), SU(SU) {}
|
|
|
|
unsigned getSparseSetIndex() const {
|
|
return TargetRegisterInfo::virtReg2Index(VirtReg);
|
|
}
|
|
};
|
|
|
|
/// Mapping from virtual register to SUnit including an operand index.
|
|
struct VReg2SUnitOperIdx : public VReg2SUnit {
|
|
unsigned OperandIndex;
|
|
|
|
VReg2SUnitOperIdx(unsigned VReg, LaneBitmask LaneMask,
|
|
unsigned OperandIndex, SUnit *SU)
|
|
: VReg2SUnit(VReg, LaneMask, SU), OperandIndex(OperandIndex) {}
|
|
};
|
|
|
|
/// Record a physical register access.
|
|
/// For non-data-dependent uses, OpIdx == -1.
|
|
struct PhysRegSUOper {
|
|
SUnit *SU;
|
|
int OpIdx;
|
|
unsigned Reg;
|
|
|
|
PhysRegSUOper(SUnit *su, int op, unsigned R): SU(su), OpIdx(op), Reg(R) {}
|
|
|
|
unsigned getSparseSetIndex() const { return Reg; }
|
|
};
|
|
|
|
/// Use a SparseMultiSet to track physical registers. Storage is only
|
|
/// allocated once for the pass. It can be cleared in constant time and reused
|
|
/// without any frees.
|
|
typedef SparseMultiSet<PhysRegSUOper, llvm::identity<unsigned>, uint16_t>
|
|
Reg2SUnitsMap;
|
|
|
|
/// Use SparseSet as a SparseMap by relying on the fact that it never
|
|
/// compares ValueT's, only unsigned keys. This allows the set to be cleared
|
|
/// between scheduling regions in constant time as long as ValueT does not
|
|
/// require a destructor.
|
|
typedef SparseSet<VReg2SUnit, VirtReg2IndexFunctor> VReg2SUnitMap;
|
|
|
|
/// Track local uses of virtual registers. These uses are gathered by the DAG
|
|
/// builder and may be consulted by the scheduler to avoid iterating an entire
|
|
/// vreg use list.
|
|
typedef SparseMultiSet<VReg2SUnit, VirtReg2IndexFunctor> VReg2SUnitMultiMap;
|
|
|
|
typedef SparseMultiSet<VReg2SUnitOperIdx, VirtReg2IndexFunctor>
|
|
VReg2SUnitOperIdxMultiMap;
|
|
|
|
typedef PointerUnion<const Value *, const PseudoSourceValue *> ValueType;
|
|
struct UnderlyingObject : PointerIntPair<ValueType, 1, bool> {
|
|
UnderlyingObject(ValueType V, bool MayAlias)
|
|
: PointerIntPair<ValueType, 1, bool>(V, MayAlias) {}
|
|
ValueType getValue() const { return getPointer(); }
|
|
bool mayAlias() const { return getInt(); }
|
|
};
|
|
typedef SmallVector<UnderlyingObject, 4> UnderlyingObjectsVector;
|
|
|
|
/// ScheduleDAGInstrs - A ScheduleDAG subclass for scheduling lists of
|
|
/// MachineInstrs.
|
|
class ScheduleDAGInstrs : public ScheduleDAG {
|
|
protected:
|
|
const MachineLoopInfo *MLI;
|
|
const MachineFrameInfo &MFI;
|
|
|
|
/// TargetSchedModel provides an interface to the machine model.
|
|
TargetSchedModel SchedModel;
|
|
|
|
/// True if the DAG builder should remove kill flags (in preparation for
|
|
/// rescheduling).
|
|
bool RemoveKillFlags;
|
|
|
|
/// The standard DAG builder does not normally include terminators as DAG
|
|
/// nodes because it does not create the necessary dependencies to prevent
|
|
/// reordering. A specialized scheduler can override
|
|
/// TargetInstrInfo::isSchedulingBoundary then enable this flag to indicate
|
|
/// it has taken responsibility for scheduling the terminator correctly.
|
|
bool CanHandleTerminators;
|
|
|
|
/// Whether lane masks should get tracked.
|
|
bool TrackLaneMasks;
|
|
|
|
/// State specific to the current scheduling region.
|
|
/// ------------------------------------------------
|
|
|
|
/// The block in which to insert instructions
|
|
MachineBasicBlock *BB;
|
|
|
|
/// The beginning of the range to be scheduled.
|
|
MachineBasicBlock::iterator RegionBegin;
|
|
|
|
/// The end of the range to be scheduled.
|
|
MachineBasicBlock::iterator RegionEnd;
|
|
|
|
/// Instructions in this region (distance(RegionBegin, RegionEnd)).
|
|
unsigned NumRegionInstrs;
|
|
|
|
/// After calling BuildSchedGraph, each machine instruction in the current
|
|
/// scheduling region is mapped to an SUnit.
|
|
DenseMap<MachineInstr*, SUnit*> MISUnitMap;
|
|
|
|
/// State internal to DAG building.
|
|
/// -------------------------------
|
|
|
|
/// Defs, Uses - Remember where defs and uses of each register are as we
|
|
/// iterate upward through the instructions. This is allocated here instead
|
|
/// of inside BuildSchedGraph to avoid the need for it to be initialized and
|
|
/// destructed for each block.
|
|
Reg2SUnitsMap Defs;
|
|
Reg2SUnitsMap Uses;
|
|
|
|
/// Tracks the last instruction(s) in this region defining each virtual
|
|
/// register. There may be multiple current definitions for a register with
|
|
/// disjunct lanemasks.
|
|
VReg2SUnitMultiMap CurrentVRegDefs;
|
|
/// Tracks the last instructions in this region using each virtual register.
|
|
VReg2SUnitOperIdxMultiMap CurrentVRegUses;
|
|
|
|
AliasAnalysis *AAForDep;
|
|
|
|
/// Remember a generic side-effecting instruction as we proceed.
|
|
/// No other SU ever gets scheduled around it (except in the special
|
|
/// case of a huge region that gets reduced).
|
|
SUnit *BarrierChain;
|
|
|
|
public:
|
|
|
|
/// A list of SUnits, used in Value2SUsMap, during DAG construction.
|
|
/// Note: to gain speed it might be worth investigating an optimized
|
|
/// implementation of this data structure, such as a singly linked list
|
|
/// with a memory pool (SmallVector was tried but slow and SparseSet is not
|
|
/// applicable).
|
|
typedef std::list<SUnit *> SUList;
|
|
protected:
|
|
/// A map from ValueType to SUList, used during DAG construction,
|
|
/// as a means of remembering which SUs depend on which memory
|
|
/// locations.
|
|
class Value2SUsMap;
|
|
|
|
/// Remove in FIFO order some SUs from huge maps.
|
|
void reduceHugeMemNodeMaps(Value2SUsMap &stores,
|
|
Value2SUsMap &loads, unsigned N);
|
|
|
|
/// Add a chain edge between SUa and SUb, but only if both AliasAnalysis
|
|
/// and Target fail to deny the dependency.
|
|
void addChainDependency(SUnit *SUa, SUnit *SUb,
|
|
unsigned Latency = 0);
|
|
|
|
/// Add dependencies as needed from all SUs in list to SU.
|
|
void addChainDependencies(SUnit *SU, SUList &sus, unsigned Latency) {
|
|
for (auto *su : sus)
|
|
addChainDependency(SU, su, Latency);
|
|
}
|
|
|
|
/// Add dependencies as needed from all SUs in map, to SU.
|
|
void addChainDependencies(SUnit *SU, Value2SUsMap &Val2SUsMap);
|
|
|
|
/// Add dependencies as needed to SU, from all SUs mapped to V.
|
|
void addChainDependencies(SUnit *SU, Value2SUsMap &Val2SUsMap,
|
|
ValueType V);
|
|
|
|
/// Add barrier chain edges from all SUs in map, and then clear
|
|
/// the map. This is equivalent to insertBarrierChain(), but
|
|
/// optimized for the common case where the new BarrierChain (a
|
|
/// global memory object) has a higher NodeNum than all SUs in
|
|
/// map. It is assumed BarrierChain has been set before calling
|
|
/// this.
|
|
void addBarrierChain(Value2SUsMap &map);
|
|
|
|
/// Insert a barrier chain in a huge region, far below current
|
|
/// SU. Add barrier chain edges from all SUs in map with higher
|
|
/// NodeNums than this new BarrierChain, and remove them from
|
|
/// map. It is assumed BarrierChain has been set before calling
|
|
/// this.
|
|
void insertBarrierChain(Value2SUsMap &map);
|
|
|
|
/// For an unanalyzable memory access, this Value is used in maps.
|
|
UndefValue *UnknownValue;
|
|
|
|
/// DbgValues - Remember instruction that precedes DBG_VALUE.
|
|
/// These are generated by buildSchedGraph but persist so they can be
|
|
/// referenced when emitting the final schedule.
|
|
typedef std::vector<std::pair<MachineInstr *, MachineInstr *> >
|
|
DbgValueVector;
|
|
DbgValueVector DbgValues;
|
|
MachineInstr *FirstDbgValue;
|
|
|
|
/// Set of live physical registers for updating kill flags.
|
|
BitVector LiveRegs;
|
|
|
|
public:
|
|
explicit ScheduleDAGInstrs(MachineFunction &mf,
|
|
const MachineLoopInfo *mli,
|
|
bool RemoveKillFlags = false);
|
|
|
|
~ScheduleDAGInstrs() override {}
|
|
|
|
/// \brief Get the machine model for instruction scheduling.
|
|
const TargetSchedModel *getSchedModel() const { return &SchedModel; }
|
|
|
|
/// \brief Resolve and cache a resolved scheduling class for an SUnit.
|
|
const MCSchedClassDesc *getSchedClass(SUnit *SU) const {
|
|
if (!SU->SchedClass && SchedModel.hasInstrSchedModel())
|
|
SU->SchedClass = SchedModel.resolveSchedClass(SU->getInstr());
|
|
return SU->SchedClass;
|
|
}
|
|
|
|
/// begin - Return an iterator to the top of the current scheduling region.
|
|
MachineBasicBlock::iterator begin() const { return RegionBegin; }
|
|
|
|
/// end - Return an iterator to the bottom of the current scheduling region.
|
|
MachineBasicBlock::iterator end() const { return RegionEnd; }
|
|
|
|
/// newSUnit - Creates a new SUnit and return a ptr to it.
|
|
SUnit *newSUnit(MachineInstr *MI);
|
|
|
|
/// getSUnit - Return an existing SUnit for this MI, or NULL.
|
|
SUnit *getSUnit(MachineInstr *MI) const;
|
|
|
|
/// startBlock - Prepare to perform scheduling in the given block.
|
|
virtual void startBlock(MachineBasicBlock *BB);
|
|
|
|
/// finishBlock - Clean up after scheduling in the given block.
|
|
virtual void finishBlock();
|
|
|
|
/// Initialize the scheduler state for the next scheduling region.
|
|
virtual void enterRegion(MachineBasicBlock *bb,
|
|
MachineBasicBlock::iterator begin,
|
|
MachineBasicBlock::iterator end,
|
|
unsigned regioninstrs);
|
|
|
|
/// Notify that the scheduler has finished scheduling the current region.
|
|
virtual void exitRegion();
|
|
|
|
/// buildSchedGraph - Build SUnits from the MachineBasicBlock that we are
|
|
/// input.
|
|
void buildSchedGraph(AliasAnalysis *AA,
|
|
RegPressureTracker *RPTracker = nullptr,
|
|
PressureDiffs *PDiffs = nullptr,
|
|
LiveIntervals *LIS = nullptr,
|
|
bool TrackLaneMasks = false);
|
|
|
|
/// addSchedBarrierDeps - Add dependencies from instructions in the current
|
|
/// list of instructions being scheduled to scheduling barrier. We want to
|
|
/// make sure instructions which define registers that are either used by
|
|
/// the terminator or are live-out are properly scheduled. This is
|
|
/// especially important when the definition latency of the return value(s)
|
|
/// are too high to be hidden by the branch or when the liveout registers
|
|
/// used by instructions in the fallthrough block.
|
|
void addSchedBarrierDeps();
|
|
|
|
/// schedule - Order nodes according to selected style, filling
|
|
/// in the Sequence member.
|
|
///
|
|
/// Typically, a scheduling algorithm will implement schedule() without
|
|
/// overriding enterRegion() or exitRegion().
|
|
virtual void schedule() = 0;
|
|
|
|
/// finalizeSchedule - Allow targets to perform final scheduling actions at
|
|
/// the level of the whole MachineFunction. By default does nothing.
|
|
virtual void finalizeSchedule() {}
|
|
|
|
void dumpNode(const SUnit *SU) const override;
|
|
|
|
/// Return a label for a DAG node that points to an instruction.
|
|
std::string getGraphNodeLabel(const SUnit *SU) const override;
|
|
|
|
/// Return a label for the region of code covered by the DAG.
|
|
std::string getDAGName() const override;
|
|
|
|
/// \brief Fix register kill flags that scheduling has made invalid.
|
|
void fixupKills(MachineBasicBlock *MBB);
|
|
protected:
|
|
void initSUnits();
|
|
void addPhysRegDataDeps(SUnit *SU, unsigned OperIdx);
|
|
void addPhysRegDeps(SUnit *SU, unsigned OperIdx);
|
|
void addVRegDefDeps(SUnit *SU, unsigned OperIdx);
|
|
void addVRegUseDeps(SUnit *SU, unsigned OperIdx);
|
|
|
|
/// \brief PostRA helper for rewriting kill flags.
|
|
void startBlockForKills(MachineBasicBlock *BB);
|
|
|
|
/// \brief Toggle a register operand kill flag.
|
|
///
|
|
/// Other adjustments may be made to the instruction if necessary. Return
|
|
/// true if the operand has been deleted, false if not.
|
|
bool toggleKillFlag(MachineInstr *MI, MachineOperand &MO);
|
|
|
|
/// Returns a mask for which lanes get read/written by the given (register)
|
|
/// machine operand.
|
|
LaneBitmask getLaneMaskForMO(const MachineOperand &MO) const;
|
|
};
|
|
|
|
/// newSUnit - Creates a new SUnit and return a ptr to it.
|
|
inline SUnit *ScheduleDAGInstrs::newSUnit(MachineInstr *MI) {
|
|
#ifndef NDEBUG
|
|
const SUnit *Addr = SUnits.empty() ? nullptr : &SUnits[0];
|
|
#endif
|
|
SUnits.emplace_back(MI, (unsigned)SUnits.size());
|
|
assert((Addr == nullptr || Addr == &SUnits[0]) &&
|
|
"SUnits std::vector reallocated on the fly!");
|
|
return &SUnits.back();
|
|
}
|
|
|
|
/// getSUnit - Return an existing SUnit for this MI, or NULL.
|
|
inline SUnit *ScheduleDAGInstrs::getSUnit(MachineInstr *MI) const {
|
|
DenseMap<MachineInstr*, SUnit*>::const_iterator I = MISUnitMap.find(MI);
|
|
if (I == MISUnitMap.end())
|
|
return nullptr;
|
|
return I->second;
|
|
}
|
|
} // namespace llvm
|
|
|
|
#endif
|