1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-21 03:53:04 +02:00
llvm-mirror/include/llvm/CodeGen/LiveInterval.h

866 lines
30 KiB
C
Raw Normal View History

//===-- llvm/CodeGen/LiveInterval.h - Interval representation ---*- 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 LiveRange and LiveInterval classes. Given some
// numbering of each the machine instructions an interval [i, j) is said to be a
// live range for register v if there is no instruction with number j' >= j
// such that v is live at j' and there is no instruction with number i' < i such
// that v is live at i'. In this implementation ranges can have holes,
// i.e. a range might look like [1,20), [50,65), [1000,1001). Each
// individual segment is represented as an instance of LiveRange::Segment,
// and the whole range is represented as an instance of LiveRange.
//
//===----------------------------------------------------------------------===//
#ifndef LLVM_CODEGEN_LIVEINTERVAL_H
#define LLVM_CODEGEN_LIVEINTERVAL_H
#include "llvm/ADT/IntEqClasses.h"
#include "llvm/CodeGen/SlotIndexes.h"
#include "llvm/Support/AlignOf.h"
#include "llvm/Support/Allocator.h"
#include <cassert>
#include <climits>
[LiveIntervalAnalysis] Speed up creation of live ranges for physical registers by using a segment set. The patch addresses a compile-time performance regression in the LiveIntervals analysis pass (see http://llvm.org/bugs/show_bug.cgi?id=18580). This regression is especially critical when compiling long functions. Our analysis had shown that the most of time is taken for generation of live intervals for physical registers. Insertions in the middle of the array of live ranges cause quadratic algorithmic complexity, which is apparently the main reason for the slow-down. Overview of changes: - The patch introduces an additional std::set<Segment>* member in LiveRange for storing segments in the phase of initial creation. The set is used if this member is not NULL, otherwise everything works the old way. - The set of operations on LiveRange used during initial creation (i.e. used by createDeadDefs and extendToUses) have been reimplemented to use the segment set if it is available. - After a live range is created the contents of the set are flushed to the segment vector, because the set is not as efficient as the vector for the later uses of the live range. After the flushing, the set is deleted and cannot be used again. - The set is only for live ranges computed in LiveIntervalAnalysis::computeLiveInRegUnits() and getRegUnit() but not in computeVirtRegs(), because I did not bring any performance benefits to computeVirtRegs() and for some examples even brought a slow down. Patch by Vaidas Gasiunas <vaidas.gasiunas@sap.com> Differential Revision: http://reviews.llvm.org/D6013 llvm-svn: 228421
2015-02-06 19:42:41 +01:00
#include <set>
namespace llvm {
class CoalescerPair;
class LiveIntervals;
class MachineInstr;
class MachineRegisterInfo;
class TargetRegisterInfo;
class raw_ostream;
template <typename T, unsigned Small> class SmallPtrSet;
2009-06-17 23:01:20 +02:00
/// VNInfo - Value Number Information.
/// This class holds information about a machine level values, including
/// definition and use points.
///
class VNInfo {
public:
typedef BumpPtrAllocator Allocator;
2009-06-17 23:01:20 +02:00
/// The ID number of this value.
unsigned id;
2012-01-05 02:01:01 +01:00
/// The index of the defining instruction.
SlotIndex def;
2009-06-17 23:01:20 +02:00
/// VNInfo constructor.
VNInfo(unsigned i, SlotIndex d)
: id(i), def(d)
{ }
2009-06-17 23:01:20 +02:00
/// VNInfo construtor, copies values from orig, except for the value number.
VNInfo(unsigned i, const VNInfo &orig)
: id(i), def(orig.def)
{ }
/// Copy from the parameter into this VNInfo.
void copyFrom(VNInfo &src) {
def = src.def;
}
2009-06-17 23:01:20 +02:00
/// Returns true if this value is defined by a PHI instruction (or was,
/// PHI instructions may have been eliminated).
/// PHI-defs begin at a block boundary, all other defs begin at register or
/// EC slots.
bool isPHIDef() const { return def.isBlock(); }
2009-06-17 23:01:20 +02:00
/// Returns true if this value is unused.
bool isUnused() const { return !def.isValid(); }
/// Mark this value as unused.
void markUnused() { def = SlotIndex(); }
};
/// Result of a LiveRange query. This class hides the implementation details
/// of live ranges, and it should be used as the primary interface for
/// examining live ranges around instructions.
class LiveQueryResult {
VNInfo *const EarlyVal;
VNInfo *const LateVal;
const SlotIndex EndPoint;
const bool Kill;
public:
LiveQueryResult(VNInfo *EarlyVal, VNInfo *LateVal, SlotIndex EndPoint,
bool Kill)
: EarlyVal(EarlyVal), LateVal(LateVal), EndPoint(EndPoint), Kill(Kill)
{}
/// Return the value that is live-in to the instruction. This is the value
/// that will be read by the instruction's use operands. Return NULL if no
/// value is live-in.
VNInfo *valueIn() const {
return EarlyVal;
}
/// Return true if the live-in value is killed by this instruction. This
/// means that either the live range ends at the instruction, or it changes
/// value.
bool isKill() const {
return Kill;
}
/// Return true if this instruction has a dead def.
bool isDeadDef() const {
return EndPoint.isDead();
}
/// Return the value leaving the instruction, if any. This can be a
/// live-through value, or a live def. A dead def returns NULL.
VNInfo *valueOut() const {
return isDeadDef() ? nullptr : LateVal;
}
/// Returns the value alive at the end of the instruction, if any. This can
/// be a live-through value, a live def or a dead def.
VNInfo *valueOutOrDead() const {
return LateVal;
}
/// Return the value defined by this instruction, if any. This includes
/// dead defs, it is the value created by the instruction's def operands.
VNInfo *valueDefined() const {
return EarlyVal == LateVal ? nullptr : LateVal;
}
/// Return the end point of the last live range segment to interact with
/// the instruction, if any.
///
/// The end point is an invalid SlotIndex only if the live range doesn't
/// intersect the instruction at all.
///
/// The end point may be at or past the end of the instruction's basic
/// block. That means the value was live out of the block.
SlotIndex endPoint() const {
return EndPoint;
}
};
/// This class represents the liveness of a register, stack slot, etc.
/// It manages an ordered list of Segment objects.
/// The Segments are organized in a static single assignment form: At places
/// where a new value is defined or different values reach a CFG join a new
/// segment with a new value number is used.
class LiveRange {
public:
/// This represents a simple continuous liveness interval for a value.
/// The start point is inclusive, the end point exclusive. These intervals
/// are rendered as [start,end).
struct Segment {
SlotIndex start; // Start point of the interval (inclusive)
SlotIndex end; // End point of the interval (exclusive)
VNInfo *valno; // identifier for the value contained in this segment.
Segment() : valno(nullptr) {}
Segment(SlotIndex S, SlotIndex E, VNInfo *V)
: start(S), end(E), valno(V) {
assert(S < E && "Cannot create empty or backwards segment");
}
/// Return true if the index is covered by this segment.
bool contains(SlotIndex I) const {
return start <= I && I < end;
}
/// Return true if the given interval, [S, E), is covered by this segment.
bool containsInterval(SlotIndex S, SlotIndex E) const {
assert((S < E) && "Backwards interval?");
return (start <= S && S < end) && (start < E && E <= end);
}
bool operator<(const Segment &Other) const {
return std::tie(start, end) < std::tie(Other.start, Other.end);
}
bool operator==(const Segment &Other) const {
return start == Other.start && end == Other.end;
}
void dump() const;
};
typedef SmallVector<Segment,4> Segments;
typedef SmallVector<VNInfo*,4> VNInfoList;
Segments segments; // the liveness segments
VNInfoList valnos; // value#'s
[LiveIntervalAnalysis] Speed up creation of live ranges for physical registers by using a segment set. The patch addresses a compile-time performance regression in the LiveIntervals analysis pass (see http://llvm.org/bugs/show_bug.cgi?id=18580). This regression is especially critical when compiling long functions. Our analysis had shown that the most of time is taken for generation of live intervals for physical registers. Insertions in the middle of the array of live ranges cause quadratic algorithmic complexity, which is apparently the main reason for the slow-down. Overview of changes: - The patch introduces an additional std::set<Segment>* member in LiveRange for storing segments in the phase of initial creation. The set is used if this member is not NULL, otherwise everything works the old way. - The set of operations on LiveRange used during initial creation (i.e. used by createDeadDefs and extendToUses) have been reimplemented to use the segment set if it is available. - After a live range is created the contents of the set are flushed to the segment vector, because the set is not as efficient as the vector for the later uses of the live range. After the flushing, the set is deleted and cannot be used again. - The set is only for live ranges computed in LiveIntervalAnalysis::computeLiveInRegUnits() and getRegUnit() but not in computeVirtRegs(), because I did not bring any performance benefits to computeVirtRegs() and for some examples even brought a slow down. Patch by Vaidas Gasiunas <vaidas.gasiunas@sap.com> Differential Revision: http://reviews.llvm.org/D6013 llvm-svn: 228421
2015-02-06 19:42:41 +01:00
// The segment set is used temporarily to accelerate initial computation
// of live ranges of physical registers in computeRegUnitRange.
// After that the set is flushed to the segment vector and deleted.
typedef std::set<Segment> SegmentSet;
std::unique_ptr<SegmentSet> segmentSet;
[LiveIntervalAnalysis] Speed up creation of live ranges for physical registers by using a segment set. The patch addresses a compile-time performance regression in the LiveIntervals analysis pass (see http://llvm.org/bugs/show_bug.cgi?id=18580). This regression is especially critical when compiling long functions. Our analysis had shown that the most of time is taken for generation of live intervals for physical registers. Insertions in the middle of the array of live ranges cause quadratic algorithmic complexity, which is apparently the main reason for the slow-down. Overview of changes: - The patch introduces an additional std::set<Segment>* member in LiveRange for storing segments in the phase of initial creation. The set is used if this member is not NULL, otherwise everything works the old way. - The set of operations on LiveRange used during initial creation (i.e. used by createDeadDefs and extendToUses) have been reimplemented to use the segment set if it is available. - After a live range is created the contents of the set are flushed to the segment vector, because the set is not as efficient as the vector for the later uses of the live range. After the flushing, the set is deleted and cannot be used again. - The set is only for live ranges computed in LiveIntervalAnalysis::computeLiveInRegUnits() and getRegUnit() but not in computeVirtRegs(), because I did not bring any performance benefits to computeVirtRegs() and for some examples even brought a slow down. Patch by Vaidas Gasiunas <vaidas.gasiunas@sap.com> Differential Revision: http://reviews.llvm.org/D6013 llvm-svn: 228421
2015-02-06 19:42:41 +01:00
typedef Segments::iterator iterator;
iterator begin() { return segments.begin(); }
iterator end() { return segments.end(); }
typedef Segments::const_iterator const_iterator;
const_iterator begin() const { return segments.begin(); }
const_iterator end() const { return segments.end(); }
typedef VNInfoList::iterator vni_iterator;
vni_iterator vni_begin() { return valnos.begin(); }
vni_iterator vni_end() { return valnos.end(); }
typedef VNInfoList::const_iterator const_vni_iterator;
const_vni_iterator vni_begin() const { return valnos.begin(); }
const_vni_iterator vni_end() const { return valnos.end(); }
/// Constructs a new LiveRange object.
LiveRange(bool UseSegmentSet = false)
: segmentSet(UseSegmentSet ? llvm::make_unique<SegmentSet>()
: nullptr) {}
/// Constructs a new LiveRange object by copying segments and valnos from
/// another LiveRange.
LiveRange(const LiveRange &Other, BumpPtrAllocator &Allocator) {
[LiveIntervalAnalysis] Speed up creation of live ranges for physical registers by using a segment set. The patch addresses a compile-time performance regression in the LiveIntervals analysis pass (see http://llvm.org/bugs/show_bug.cgi?id=18580). This regression is especially critical when compiling long functions. Our analysis had shown that the most of time is taken for generation of live intervals for physical registers. Insertions in the middle of the array of live ranges cause quadratic algorithmic complexity, which is apparently the main reason for the slow-down. Overview of changes: - The patch introduces an additional std::set<Segment>* member in LiveRange for storing segments in the phase of initial creation. The set is used if this member is not NULL, otherwise everything works the old way. - The set of operations on LiveRange used during initial creation (i.e. used by createDeadDefs and extendToUses) have been reimplemented to use the segment set if it is available. - After a live range is created the contents of the set are flushed to the segment vector, because the set is not as efficient as the vector for the later uses of the live range. After the flushing, the set is deleted and cannot be used again. - The set is only for live ranges computed in LiveIntervalAnalysis::computeLiveInRegUnits() and getRegUnit() but not in computeVirtRegs(), because I did not bring any performance benefits to computeVirtRegs() and for some examples even brought a slow down. Patch by Vaidas Gasiunas <vaidas.gasiunas@sap.com> Differential Revision: http://reviews.llvm.org/D6013 llvm-svn: 228421
2015-02-06 19:42:41 +01:00
assert(Other.segmentSet == nullptr &&
"Copying of LiveRanges with active SegmentSets is not supported");
// Duplicate valnos.
for (const VNInfo *VNI : Other.valnos) {
createValueCopy(VNI, Allocator);
}
// Now we can copy segments and remap their valnos.
for (const Segment &S : Other.segments) {
segments.push_back(Segment(S.start, S.end, valnos[S.valno->id]));
}
}
/// advanceTo - Advance the specified iterator to point to the Segment
/// containing the specified position, or end() if the position is past the
/// end of the range. If no Segment contains this position, but the
/// position is in a hole, this method returns an iterator pointing to the
/// Segment immediately after the hole.
iterator advanceTo(iterator I, SlotIndex Pos) {
assert(I != end());
if (Pos >= endIndex())
return end();
while (I->end <= Pos) ++I;
return I;
}
const_iterator advanceTo(const_iterator I, SlotIndex Pos) const {
assert(I != end());
if (Pos >= endIndex())
return end();
while (I->end <= Pos) ++I;
return I;
}
/// find - Return an iterator pointing to the first segment that ends after
/// Pos, or end(). This is the same as advanceTo(begin(), Pos), but faster
/// when searching large ranges.
///
/// If Pos is contained in a Segment, that segment is returned.
/// If Pos is in a hole, the following Segment is returned.
/// If Pos is beyond endIndex, end() is returned.
iterator find(SlotIndex Pos);
const_iterator find(SlotIndex Pos) const {
return const_cast<LiveRange*>(this)->find(Pos);
}
void clear() {
valnos.clear();
segments.clear();
}
size_t size() const {
return segments.size();
}
bool hasAtLeastOneValue() const { return !valnos.empty(); }
bool containsOneValue() const { return valnos.size() == 1; }
unsigned getNumValNums() const { return (unsigned)valnos.size(); }
/// getValNumInfo - Returns pointer to the specified val#.
///
inline VNInfo *getValNumInfo(unsigned ValNo) {
return valnos[ValNo];
}
inline const VNInfo *getValNumInfo(unsigned ValNo) const {
return valnos[ValNo];
}
/// containsValue - Returns true if VNI belongs to this range.
bool containsValue(const VNInfo *VNI) const {
return VNI && VNI->id < getNumValNums() && VNI == getValNumInfo(VNI->id);
}
/// getNextValue - Create a new value number and return it. MIIdx specifies
/// the instruction that defines the value number.
VNInfo *getNextValue(SlotIndex def, VNInfo::Allocator &VNInfoAllocator) {
VNInfo *VNI =
new (VNInfoAllocator) VNInfo((unsigned)valnos.size(), def);
2009-06-17 23:01:20 +02:00
valnos.push_back(VNI);
return VNI;
}
/// createDeadDef - Make sure the range has a value defined at Def.
/// If one already exists, return it. Otherwise allocate a new value and
/// add liveness for a dead def.
VNInfo *createDeadDef(SlotIndex Def, VNInfo::Allocator &VNInfoAllocator);
2009-06-17 23:01:20 +02:00
/// Create a copy of the given value. The new value will be identical except
/// for the Value number.
VNInfo *createValueCopy(const VNInfo *orig,
VNInfo::Allocator &VNInfoAllocator) {
VNInfo *VNI =
new (VNInfoAllocator) VNInfo((unsigned)valnos.size(), *orig);
valnos.push_back(VNI);
return VNI;
}
/// RenumberValues - Renumber all values in order of appearance and remove
/// unused values.
void RenumberValues();
2013-09-06 18:19:22 +02:00
/// MergeValueNumberInto - This method is called when two value numbers
/// are found to be equivalent. This eliminates V1, replacing all
/// segments with the V1 value number with the V2 value number. This can
/// cause merging of V1/V2 values numbers and compaction of the value space.
VNInfo* MergeValueNumberInto(VNInfo *V1, VNInfo *V2);
/// Merge all of the live segments of a specific val# in RHS into this live
/// range as the specified value number. The segments in RHS are allowed
/// to overlap with segments in the current range, it will replace the
/// value numbers of the overlaped live segments with the specified value
/// number.
void MergeSegmentsInAsValue(const LiveRange &RHS, VNInfo *LHSValNo);
/// MergeValueInAsValue - Merge all of the segments of a specific val#
/// in RHS into this live range as the specified value number.
/// The segments in RHS are allowed to overlap with segments in the
/// current range, but only if the overlapping segments have the
/// specified value number.
void MergeValueInAsValue(const LiveRange &RHS,
const VNInfo *RHSValNo, VNInfo *LHSValNo);
bool empty() const { return segments.empty(); }
/// beginIndex - Return the lowest numbered slot covered.
SlotIndex beginIndex() const {
assert(!empty() && "Call to beginIndex() on empty range.");
return segments.front().start;
}
/// endNumber - return the maximum point of the range of the whole,
/// exclusive.
SlotIndex endIndex() const {
assert(!empty() && "Call to endIndex() on empty range.");
return segments.back().end;
}
bool expiredAt(SlotIndex index) const {
return index >= endIndex();
}
bool liveAt(SlotIndex index) const {
const_iterator r = find(index);
return r != end() && r->start <= index;
}
/// Return the segment that contains the specified index, or null if there
/// is none.
const Segment *getSegmentContaining(SlotIndex Idx) const {
const_iterator I = FindSegmentContaining(Idx);
return I == end() ? nullptr : &*I;
}
/// Return the live segment that contains the specified index, or null if
/// there is none.
Segment *getSegmentContaining(SlotIndex Idx) {
iterator I = FindSegmentContaining(Idx);
return I == end() ? nullptr : &*I;
}
/// getVNInfoAt - Return the VNInfo that is live at Idx, or NULL.
VNInfo *getVNInfoAt(SlotIndex Idx) const {
const_iterator I = FindSegmentContaining(Idx);
return I == end() ? nullptr : I->valno;
}
/// getVNInfoBefore - Return the VNInfo that is live up to but not
/// necessarilly including Idx, or NULL. Use this to find the reaching def
/// used by an instruction at this SlotIndex position.
VNInfo *getVNInfoBefore(SlotIndex Idx) const {
const_iterator I = FindSegmentContaining(Idx.getPrevSlot());
return I == end() ? nullptr : I->valno;
}
/// Return an iterator to the segment that contains the specified index, or
/// end() if there is none.
iterator FindSegmentContaining(SlotIndex Idx) {
iterator I = find(Idx);
return I != end() && I->start <= Idx ? I : end();
}
const_iterator FindSegmentContaining(SlotIndex Idx) const {
const_iterator I = find(Idx);
return I != end() && I->start <= Idx ? I : end();
}
/// overlaps - Return true if the intersection of the two live ranges is
/// not empty.
bool overlaps(const LiveRange &other) const {
if (other.empty())
return false;
return overlapsFrom(other, other.begin());
}
/// overlaps - Return true if the two ranges have overlapping segments
/// that are not coalescable according to CP.
///
/// Overlapping segments where one range is defined by a coalescable
/// copy are allowed.
bool overlaps(const LiveRange &Other, const CoalescerPair &CP,
const SlotIndexes&) const;
/// overlaps - Return true if the live range overlaps an interval specified
/// by [Start, End).
bool overlaps(SlotIndex Start, SlotIndex End) const;
/// overlapsFrom - Return true if the intersection of the two live ranges
/// is not empty. The specified iterator is a hint that we can begin
/// scanning the Other range starting at I.
bool overlapsFrom(const LiveRange &Other, const_iterator I) const;
/// Returns true if all segments of the @p Other live range are completely
/// covered by this live range.
/// Adjacent live ranges do not affect the covering:the liverange
/// [1,5](5,10] covers (3,7].
bool covers(const LiveRange &Other) const;
/// Add the specified Segment to this range, merging segments as
/// appropriate. This returns an iterator to the inserted segment (which
/// may have grown since it was inserted).
[LiveIntervalAnalysis] Speed up creation of live ranges for physical registers by using a segment set. The patch addresses a compile-time performance regression in the LiveIntervals analysis pass (see http://llvm.org/bugs/show_bug.cgi?id=18580). This regression is especially critical when compiling long functions. Our analysis had shown that the most of time is taken for generation of live intervals for physical registers. Insertions in the middle of the array of live ranges cause quadratic algorithmic complexity, which is apparently the main reason for the slow-down. Overview of changes: - The patch introduces an additional std::set<Segment>* member in LiveRange for storing segments in the phase of initial creation. The set is used if this member is not NULL, otherwise everything works the old way. - The set of operations on LiveRange used during initial creation (i.e. used by createDeadDefs and extendToUses) have been reimplemented to use the segment set if it is available. - After a live range is created the contents of the set are flushed to the segment vector, because the set is not as efficient as the vector for the later uses of the live range. After the flushing, the set is deleted and cannot be used again. - The set is only for live ranges computed in LiveIntervalAnalysis::computeLiveInRegUnits() and getRegUnit() but not in computeVirtRegs(), because I did not bring any performance benefits to computeVirtRegs() and for some examples even brought a slow down. Patch by Vaidas Gasiunas <vaidas.gasiunas@sap.com> Differential Revision: http://reviews.llvm.org/D6013 llvm-svn: 228421
2015-02-06 19:42:41 +01:00
iterator addSegment(Segment S);
/// If this range is live before @p Use in the basic block that starts at
/// @p StartIdx, extend it to be live up to @p Use, and return the value. If
/// there is no segment before @p Use, return nullptr.
VNInfo *extendInBlock(SlotIndex StartIdx, SlotIndex Use);
/// join - Join two live ranges (this, and other) together. This applies
/// mappings to the value numbers in the LHS/RHS ranges as specified. If
/// the ranges are not joinable, this aborts.
void join(LiveRange &Other,
const int *ValNoAssignments,
const int *RHSValNoAssignments,
SmallVectorImpl<VNInfo *> &NewVNInfo);
/// True iff this segment is a single segment that lies between the
/// specified boundaries, exclusively. Vregs live across a backedge are not
/// considered local. The boundaries are expected to lie within an extended
/// basic block, so vregs that are not live out should contain no holes.
bool isLocal(SlotIndex Start, SlotIndex End) const {
return beginIndex() > Start.getBaseIndex() &&
endIndex() < End.getBoundaryIndex();
}
/// Remove the specified segment from this range. Note that the segment
/// must be a single Segment in its entirety.
void removeSegment(SlotIndex Start, SlotIndex End,
bool RemoveDeadValNo = false);
void removeSegment(Segment S, bool RemoveDeadValNo = false) {
removeSegment(S.start, S.end, RemoveDeadValNo);
}
/// Remove segment pointed to by iterator @p I from this range. This does
/// not remove dead value numbers.
iterator removeSegment(iterator I) {
return segments.erase(I);
}
/// Query Liveness at Idx.
/// The sub-instruction slot of Idx doesn't matter, only the instruction
/// it refers to is considered.
LiveQueryResult Query(SlotIndex Idx) const {
// Find the segment that enters the instruction.
const_iterator I = find(Idx.getBaseIndex());
const_iterator E = end();
if (I == E)
return LiveQueryResult(nullptr, nullptr, SlotIndex(), false);
// Is this an instruction live-in segment?
// If Idx is the start index of a basic block, include live-in segments
// that start at Idx.getBaseIndex().
VNInfo *EarlyVal = nullptr;
VNInfo *LateVal = nullptr;
SlotIndex EndPoint;
bool Kill = false;
if (I->start <= Idx.getBaseIndex()) {
EarlyVal = I->valno;
EndPoint = I->end;
// Move to the potentially live-out segment.
if (SlotIndex::isSameInstr(Idx, I->end)) {
Kill = true;
if (++I == E)
return LiveQueryResult(EarlyVal, LateVal, EndPoint, Kill);
}
// Special case: A PHIDef value can have its def in the middle of a
// segment if the value happens to be live out of the layout
// predecessor.
// Such a value is not live-in.
if (EarlyVal->def == Idx.getBaseIndex())
EarlyVal = nullptr;
}
// I now points to the segment that may be live-through, or defined by
// this instr. Ignore segments starting after the current instr.
if (!SlotIndex::isEarlierInstr(Idx, I->start)) {
LateVal = I->valno;
EndPoint = I->end;
}
return LiveQueryResult(EarlyVal, LateVal, EndPoint, Kill);
}
/// removeValNo - Remove all the segments defined by the specified value#.
/// Also remove the value# from value# list.
void removeValNo(VNInfo *ValNo);
/// Returns true if the live range is zero length, i.e. no live segments
/// span instructions. It doesn't pay to spill such a range.
bool isZeroLength(SlotIndexes *Indexes) const {
for (const Segment &S : segments)
if (Indexes->getNextNonNullIndex(S.start).getBaseIndex() <
S.end.getBaseIndex())
return false;
return true;
}
bool operator<(const LiveRange& other) const {
const SlotIndex &thisIndex = beginIndex();
const SlotIndex &otherIndex = other.beginIndex();
return thisIndex < otherIndex;
}
[LiveIntervalAnalysis] Speed up creation of live ranges for physical registers by using a segment set. The patch addresses a compile-time performance regression in the LiveIntervals analysis pass (see http://llvm.org/bugs/show_bug.cgi?id=18580). This regression is especially critical when compiling long functions. Our analysis had shown that the most of time is taken for generation of live intervals for physical registers. Insertions in the middle of the array of live ranges cause quadratic algorithmic complexity, which is apparently the main reason for the slow-down. Overview of changes: - The patch introduces an additional std::set<Segment>* member in LiveRange for storing segments in the phase of initial creation. The set is used if this member is not NULL, otherwise everything works the old way. - The set of operations on LiveRange used during initial creation (i.e. used by createDeadDefs and extendToUses) have been reimplemented to use the segment set if it is available. - After a live range is created the contents of the set are flushed to the segment vector, because the set is not as efficient as the vector for the later uses of the live range. After the flushing, the set is deleted and cannot be used again. - The set is only for live ranges computed in LiveIntervalAnalysis::computeLiveInRegUnits() and getRegUnit() but not in computeVirtRegs(), because I did not bring any performance benefits to computeVirtRegs() and for some examples even brought a slow down. Patch by Vaidas Gasiunas <vaidas.gasiunas@sap.com> Differential Revision: http://reviews.llvm.org/D6013 llvm-svn: 228421
2015-02-06 19:42:41 +01:00
/// Flush segment set into the regular segment vector.
/// The method is to be called after the live range
/// has been created, if use of the segment set was
/// activated in the constructor of the live range.
void flushSegmentSet();
void print(raw_ostream &OS) const;
void dump() const;
/// \brief Walk the range and assert if any invariants fail to hold.
///
/// Note that this is a no-op when asserts are disabled.
#ifdef NDEBUG
void verify() const {}
#else
void verify() const;
#endif
protected:
/// Append a segment to the list of segments.
void append(const LiveRange::Segment S);
2009-07-14 04:17:17 +02:00
private:
[LiveIntervalAnalysis] Speed up creation of live ranges for physical registers by using a segment set. The patch addresses a compile-time performance regression in the LiveIntervals analysis pass (see http://llvm.org/bugs/show_bug.cgi?id=18580). This regression is especially critical when compiling long functions. Our analysis had shown that the most of time is taken for generation of live intervals for physical registers. Insertions in the middle of the array of live ranges cause quadratic algorithmic complexity, which is apparently the main reason for the slow-down. Overview of changes: - The patch introduces an additional std::set<Segment>* member in LiveRange for storing segments in the phase of initial creation. The set is used if this member is not NULL, otherwise everything works the old way. - The set of operations on LiveRange used during initial creation (i.e. used by createDeadDefs and extendToUses) have been reimplemented to use the segment set if it is available. - After a live range is created the contents of the set are flushed to the segment vector, because the set is not as efficient as the vector for the later uses of the live range. After the flushing, the set is deleted and cannot be used again. - The set is only for live ranges computed in LiveIntervalAnalysis::computeLiveInRegUnits() and getRegUnit() but not in computeVirtRegs(), because I did not bring any performance benefits to computeVirtRegs() and for some examples even brought a slow down. Patch by Vaidas Gasiunas <vaidas.gasiunas@sap.com> Differential Revision: http://reviews.llvm.org/D6013 llvm-svn: 228421
2015-02-06 19:42:41 +01:00
friend class LiveRangeUpdater;
void addSegmentToSet(Segment S);
void markValNoForDeletion(VNInfo *V);
};
inline raw_ostream &operator<<(raw_ostream &OS, const LiveRange &LR) {
LR.print(OS);
return OS;
}
/// LiveInterval - This class represents the liveness of a register,
/// or stack slot.
class LiveInterval : public LiveRange {
public:
typedef LiveRange super;
/// A live range for subregisters. The LaneMask specifies which parts of the
/// super register are covered by the interval.
/// (@sa TargetRegisterInfo::getSubRegIndexLaneMask()).
class SubRange : public LiveRange {
public:
SubRange *Next;
unsigned LaneMask;
/// Constructs a new SubRange object.
SubRange(unsigned LaneMask)
: Next(nullptr), LaneMask(LaneMask) {
}
/// Constructs a new SubRange object by copying liveness from @p Other.
SubRange(unsigned LaneMask, const LiveRange &Other,
BumpPtrAllocator &Allocator)
: LiveRange(Other, Allocator), Next(nullptr), LaneMask(LaneMask) {
}
};
private:
SubRange *SubRanges; ///< Single linked list of subregister live ranges.
public:
const unsigned reg; // the register or stack slot of this interval.
float weight; // weight of this interval
LiveInterval(unsigned Reg, float Weight)
: SubRanges(nullptr), reg(Reg), weight(Weight) {}
~LiveInterval() {
clearSubRanges();
}
template<typename T>
class SingleLinkedListIterator {
T *P;
public:
SingleLinkedListIterator<T>(T *P) : P(P) {}
SingleLinkedListIterator<T> &operator++() {
P = P->Next;
return *this;
}
SingleLinkedListIterator<T> &operator++(int) {
SingleLinkedListIterator res = *this;
++*this;
return res;
}
bool operator!=(const SingleLinkedListIterator<T> &Other) {
return P != Other.operator->();
}
bool operator==(const SingleLinkedListIterator<T> &Other) {
return P == Other.operator->();
}
T &operator*() const {
return *P;
}
T *operator->() const {
return P;
}
};
typedef SingleLinkedListIterator<SubRange> subrange_iterator;
subrange_iterator subrange_begin() {
return subrange_iterator(SubRanges);
}
subrange_iterator subrange_end() {
return subrange_iterator(nullptr);
}
typedef SingleLinkedListIterator<const SubRange> const_subrange_iterator;
const_subrange_iterator subrange_begin() const {
return const_subrange_iterator(SubRanges);
}
const_subrange_iterator subrange_end() const {
return const_subrange_iterator(nullptr);
}
iterator_range<subrange_iterator> subranges() {
return make_range(subrange_begin(), subrange_end());
}
iterator_range<const_subrange_iterator> subranges() const {
return make_range(subrange_begin(), subrange_end());
}
/// Creates a new empty subregister live range. The range is added at the
/// beginning of the subrange list; subrange iterators stay valid.
SubRange *createSubRange(BumpPtrAllocator &Allocator, unsigned LaneMask) {
SubRange *Range = new (Allocator) SubRange(LaneMask);
appendSubRange(Range);
return Range;
}
/// Like createSubRange() but the new range is filled with a copy of the
/// liveness information in @p CopyFrom.
SubRange *createSubRangeFrom(BumpPtrAllocator &Allocator, unsigned LaneMask,
const LiveRange &CopyFrom) {
SubRange *Range = new (Allocator) SubRange(LaneMask, CopyFrom, Allocator);
appendSubRange(Range);
return Range;
}
/// Returns true if subregister liveness information is available.
bool hasSubRanges() const {
return SubRanges != nullptr;
}
/// Removes all subregister liveness information.
void clearSubRanges();
/// Removes all subranges without any segments (subranges without segments
/// are not considered valid and should only exist temporarily).
void removeEmptySubRanges();
/// Construct main live range by merging the SubRanges of @p LI.
void constructMainRangeFromSubranges(const SlotIndexes &Indexes,
VNInfo::Allocator &VNIAllocator);
/// getSize - Returns the sum of sizes of all the LiveRange's.
///
unsigned getSize() const;
/// isSpillable - Can this interval be spilled?
bool isSpillable() const {
return weight != llvm::huge_valf;
}
/// markNotSpillable - Mark interval as not spillable
void markNotSpillable() {
weight = llvm::huge_valf;
}
bool operator<(const LiveInterval& other) const {
const SlotIndex &thisIndex = beginIndex();
const SlotIndex &otherIndex = other.beginIndex();
return std::tie(thisIndex, reg) < std::tie(otherIndex, other.reg);
}
void print(raw_ostream &OS) const;
void dump() const;
/// \brief Walks the interval and assert if any invariants fail to hold.
///
/// Note that this is a no-op when asserts are disabled.
#ifdef NDEBUG
void verify(const MachineRegisterInfo *MRI = nullptr) const {}
#else
void verify(const MachineRegisterInfo *MRI = nullptr) const;
#endif
private:
/// Appends @p Range to SubRanges list.
void appendSubRange(SubRange *Range) {
Range->Next = SubRanges;
SubRanges = Range;
}
/// Free memory held by SubRange.
void freeSubRange(SubRange *S);
};
inline raw_ostream &operator<<(raw_ostream &OS, const LiveInterval &LI) {
LI.print(OS);
return OS;
}
raw_ostream &operator<<(raw_ostream &OS, const LiveRange::Segment &S);
inline bool operator<(SlotIndex V, const LiveRange::Segment &S) {
return V < S.start;
}
inline bool operator<(const LiveRange::Segment &S, SlotIndex V) {
return S.start < V;
}
/// Helper class for performant LiveRange bulk updates.
///
/// Calling LiveRange::addSegment() repeatedly can be expensive on large
/// live ranges because segments after the insertion point may need to be
/// shifted. The LiveRangeUpdater class can defer the shifting when adding
/// many segments in order.
///
/// The LiveRange will be in an invalid state until flush() is called.
class LiveRangeUpdater {
LiveRange *LR;
SlotIndex LastStart;
LiveRange::iterator WriteI;
LiveRange::iterator ReadI;
SmallVector<LiveRange::Segment, 16> Spills;
void mergeSpills();
public:
/// Create a LiveRangeUpdater for adding segments to LR.
/// LR will temporarily be in an invalid state until flush() is called.
LiveRangeUpdater(LiveRange *lr = nullptr) : LR(lr) {}
~LiveRangeUpdater() { flush(); }
/// Add a segment to LR and coalesce when possible, just like
/// LR.addSegment(). Segments should be added in increasing start order for
/// best performance.
void add(LiveRange::Segment);
void add(SlotIndex Start, SlotIndex End, VNInfo *VNI) {
add(LiveRange::Segment(Start, End, VNI));
}
/// Return true if the LR is currently in an invalid state, and flush()
/// needs to be called.
bool isDirty() const { return LastStart.isValid(); }
/// Flush the updater state to LR so it is valid and contains all added
/// segments.
void flush();
/// Select a different destination live range.
void setDest(LiveRange *lr) {
if (LR != lr && isDirty())
flush();
LR = lr;
}
/// Get the current destination live range.
LiveRange *getDest() const { return LR; }
void dump() const;
void print(raw_ostream&) const;
};
inline raw_ostream &operator<<(raw_ostream &OS, const LiveRangeUpdater &X) {
X.print(OS);
return OS;
}
/// ConnectedVNInfoEqClasses - Helper class that can divide VNInfos in a
/// LiveInterval into equivalence clases of connected components. A
/// LiveInterval that has multiple connected components can be broken into
/// multiple LiveIntervals.
///
/// Given a LiveInterval that may have multiple connected components, run:
///
/// unsigned numComps = ConEQ.Classify(LI);
/// if (numComps > 1) {
/// // allocate numComps-1 new LiveIntervals into LIS[1..]
/// ConEQ.Distribute(LIS);
/// }
class ConnectedVNInfoEqClasses {
LiveIntervals &LIS;
IntEqClasses EqClass;
public:
explicit ConnectedVNInfoEqClasses(LiveIntervals &lis) : LIS(lis) {}
/// Classify - Classify the values in LI into connected components.
/// Return the number of connected components.
unsigned Classify(const LiveInterval *LI);
/// getEqClass - Classify creates equivalence classes numbered 0..N. Return
/// the equivalence class assigned the VNI.
unsigned getEqClass(const VNInfo *VNI) const { return EqClass[VNI->id]; }
/// Distribute values in \p LI into a separate LiveIntervals
/// for each connected component. LIV must have an empty LiveInterval for
/// each additional connected component. The first connected component is
/// left in \p LI.
void Distribute(LiveInterval &LI, LiveInterval *LIV[],
MachineRegisterInfo &MRI);
};
}
#endif