Ryujinx/ARMeilleure/CodeGen/RegisterAllocators/LiveInterval.cs
gdkchan a731ab3a2a Add a new JIT compiler for CPU code (#693)
* Start of the ARMeilleure project

* Refactoring around the old IRAdapter, now renamed to PreAllocator

* Optimize the LowestBitSet method

* Add CLZ support and fix CLS implementation

* Add missing Equals and GetHashCode overrides on some structs, misc small tweaks

* Implement the ByteSwap IR instruction, and some refactoring on the assembler

* Implement the DivideUI IR instruction and fix 64-bits IDIV

* Correct constant operand type on CSINC

* Move division instructions implementation to InstEmitDiv

* Fix destination type for the ConditionalSelect IR instruction

* Implement UMULH and SMULH, with new IR instructions

* Fix some issues with shift instructions

* Fix constant types for BFM instructions

* Fix up new tests using the new V128 struct

* Update tests

* Move DIV tests to a separate file

* Add support for calls, and some instructions that depends on them

* Start adding support for SIMD & FP types, along with some of the related ARM instructions

* Fix some typos and the divide instruction with FP operands

* Fix wrong method call on Clz_V

* Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes

* Implement SIMD logical instructions and more misc. fixes

* Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations

* Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes

* Implement SIMD shift instruction and fix Dup_V

* Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table

* Fix check with tolerance on tester

* Implement FP & SIMD comparison instructions, and some fixes

* Update FCVT (Scalar) encoding on the table to support the Half-float variants

* Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes

* Use old memory access methods, made a start on SIMD memory insts support, some fixes

* Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes

* Fix arguments count with struct return values, other fixes

* More instructions

* Misc. fixes and integrate LDj3SNuD fixes

* Update tests

* Add a faster linear scan allocator, unwinding support on windows, and other changes

* Update Ryujinx.HLE

* Update Ryujinx.Graphics

* Fix V128 return pointer passing, RCX is clobbered

* Update Ryujinx.Tests

* Update ITimeZoneService

* Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks

* Use generic GetFunctionPointerForDelegate method and other tweaks

* Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics

* Remove some unused code on the assembler

* Fix REX.W prefix regression on float conversion instructions, add some sort of profiler

* Add hardware capability detection

* Fix regression on Sha1h and revert Fcm** changes

* Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator

* Fix silly mistake introduced on last commit on CpuId

* Generate inline stack probes when the stack allocation is too large

* Initial support for the System-V ABI

* Support multiple destination operands

* Fix SSE2 VectorInsert8 path, and other fixes

* Change placement of XMM callee save and restore code to match other compilers

* Rename Dest to Destination and Inst to Instruction

* Fix a regression related to calls and the V128 type

* Add an extra space on comments to match code style

* Some refactoring

* Fix vector insert FP32 SSE2 path

* Port over the ARM32 instructions

* Avoid memory protection races on JIT Cache

* Another fix on VectorInsert FP32 (thanks to LDj3SNuD

* Float operands don't need to use the same register when VEX is supported

* Add a new register allocator, higher quality code for hot code (tier up), and other tweaks

* Some nits, small improvements on the pre allocator

* CpuThreadState is gone

* Allow changing CPU emulators with a config entry

* Add runtime identifiers on the ARMeilleure project

* Allow switching between CPUs through a config entry (pt. 2)

* Change win10-x64 to win-x64 on projects

* Update the Ryujinx project to use ARMeilleure

* Ensure that the selected register is valid on the hybrid allocator

* Allow exiting on returns to 0 (should fix test regression)

* Remove register assignments for most used variables on the hybrid allocator

* Do not use fixed registers as spill temp

* Add missing namespace and remove unneeded using

* Address PR feedback

* Fix types, etc

* Enable AssumeStrictAbiCompliance by default

* Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 21:56:22 +03:00

390 lines
11 KiB
C#

using ARMeilleure.IntermediateRepresentation;
using System;
using System.Collections.Generic;
using System.Diagnostics;
using System.Linq;
namespace ARMeilleure.CodeGen.RegisterAllocators
{
class LiveInterval : IComparable<LiveInterval>
{
public const int NotFound = -1;
private LiveInterval _parent;
private SortedSet<int> _usePositions;
public int UsesCount => _usePositions.Count;
private List<LiveRange> _ranges;
private SortedList<int, LiveInterval> _childs;
public bool IsSplit => _childs.Count != 0;
public Operand Local { get; }
public Register Register { get; set; }
public int SpillOffset { get; private set; }
public bool IsSpilled => SpillOffset != -1;
public bool IsFixed { get; }
public bool IsEmpty => _ranges.Count == 0;
public LiveInterval(Operand local = null, LiveInterval parent = null)
{
Local = local;
_parent = parent ?? this;
_usePositions = new SortedSet<int>();
_ranges = new List<LiveRange>();
_childs = new SortedList<int, LiveInterval>();
SpillOffset = -1;
}
public LiveInterval(Register register) : this()
{
IsFixed = true;
Register = register;
}
public void SetStart(int position)
{
if (_ranges.Count != 0)
{
Debug.Assert(position != _ranges[0].End);
_ranges[0] = new LiveRange(position, _ranges[0].End);
}
else
{
_ranges.Add(new LiveRange(position, position + 1));
}
}
public int GetStart()
{
if (_ranges.Count == 0)
{
throw new InvalidOperationException("Empty interval.");
}
return _ranges[0].Start;
}
public void SetEnd(int position)
{
if (_ranges.Count != 0)
{
int lastIdx = _ranges.Count - 1;
Debug.Assert(position != _ranges[lastIdx].Start);
_ranges[lastIdx] = new LiveRange(_ranges[lastIdx].Start, position);
}
else
{
_ranges.Add(new LiveRange(position, position + 1));
}
}
public int GetEnd()
{
if (_ranges.Count == 0)
{
throw new InvalidOperationException("Empty interval.");
}
return _ranges[_ranges.Count - 1].End;
}
public void AddRange(int start, int end)
{
if (start >= end)
{
throw new ArgumentException("Invalid range start position " + start + ", " + end);
}
int index = _ranges.BinarySearch(new LiveRange(start, end));
if (index >= 0)
{
// New range insersects with an existing range, we need to remove
// all the intersecting ranges before adding the new one.
// We also extend the new range as needed, based on the values of
// the existing ranges being removed.
int lIndex = index;
int rIndex = index;
while (lIndex > 0 && _ranges[lIndex - 1].End >= start)
{
lIndex--;
}
while (rIndex + 1 < _ranges.Count && _ranges[rIndex + 1].Start <= end)
{
rIndex++;
}
if (start > _ranges[lIndex].Start)
{
start = _ranges[lIndex].Start;
}
if (end < _ranges[rIndex].End)
{
end = _ranges[rIndex].End;
}
_ranges.RemoveRange(lIndex, (rIndex - lIndex) + 1);
InsertRange(lIndex, start, end);
}
else
{
InsertRange(~index, start, end);
}
}
private void InsertRange(int index, int start, int end)
{
// Here we insert a new range on the ranges list.
// If possible, we extend an existing range rather than inserting a new one.
// We can extend an existing range if any of the following conditions are true:
// - The new range starts right after the end of the previous range on the list.
// - The new range ends right before the start of the next range on the list.
// If both cases are true, we can extend either one. We prefer to extend the
// previous range, and then remove the next one, but theres no specific reason
// for that, extending either one will do.
int? extIndex = null;
if (index > 0 && _ranges[index - 1].End == start)
{
start = _ranges[index - 1].Start;
extIndex = index - 1;
}
if (index < _ranges.Count && _ranges[index].Start == end)
{
end = _ranges[index].End;
if (extIndex.HasValue)
{
_ranges.RemoveAt(index);
}
else
{
extIndex = index;
}
}
if (extIndex.HasValue)
{
_ranges[extIndex.Value] = new LiveRange(start, end);
}
else
{
_ranges.Insert(index, new LiveRange(start, end));
}
}
public void AddUsePosition(int position)
{
_usePositions.Add(position);
}
public bool Overlaps(int position)
{
return _ranges.BinarySearch(new LiveRange(position, position + 1)) >= 0;
}
public bool Overlaps(LiveInterval other)
{
foreach (LiveRange range in other._ranges)
{
if (_ranges.BinarySearch(range) >= 0)
{
return true;
}
}
return false;
}
public int GetOverlapPosition(LiveInterval other)
{
foreach (LiveRange range in other._ranges)
{
int overlapIndex = _ranges.BinarySearch(range);
if (overlapIndex >= 0)
{
// It's possible that we have multiple overlaps within a single interval,
// in this case, we pick the one with the lowest start position, since
// we return the first overlap position.
while (overlapIndex > 0 && _ranges[overlapIndex - 1].End > range.Start)
{
overlapIndex--;
}
LiveRange overlappingRange = _ranges[overlapIndex];
return overlappingRange.Start;
}
}
return NotFound;
}
public IEnumerable<LiveInterval> SplitChilds()
{
return _childs.Values;
}
public IEnumerable<int> UsePositions()
{
return _usePositions;
}
public int FirstUse()
{
if (_usePositions.Count == 0)
{
return NotFound;
}
return _usePositions.First();
}
public int NextUseAfter(int position)
{
foreach (int usePosition in _usePositions)
{
if (usePosition >= position)
{
return usePosition;
}
}
return NotFound;
}
public LiveInterval Split(int position)
{
LiveInterval right = new LiveInterval(Local, _parent);
int splitIndex = 0;
for (; splitIndex < _ranges.Count; splitIndex++)
{
LiveRange range = _ranges[splitIndex];
if (position > range.Start && position <= range.End)
{
right._ranges.Add(new LiveRange(position, range.End));
range = new LiveRange(range.Start, position);
_ranges[splitIndex++] = range;
break;
}
if (range.Start >= position)
{
break;
}
}
if (splitIndex < _ranges.Count)
{
int count = _ranges.Count - splitIndex;
right._ranges.AddRange(_ranges.GetRange(splitIndex, count));
_ranges.RemoveRange(splitIndex, count);
}
foreach (int usePosition in _usePositions.Where(x => x >= position))
{
right._usePositions.Add(usePosition);
}
_usePositions.RemoveWhere(x => x >= position);
Debug.Assert(_ranges.Count != 0, "Left interval is empty after split.");
Debug.Assert(right._ranges.Count != 0, "Right interval is empty after split.");
AddSplitChild(right);
return right;
}
private void AddSplitChild(LiveInterval child)
{
Debug.Assert(!child.IsEmpty, "Trying to insert a empty interval.");
_parent._childs.Add(child.GetStart(), child);
}
public LiveInterval GetSplitChild(int position)
{
if (Overlaps(position))
{
return this;
}
foreach (LiveInterval splitChild in _childs.Values)
{
if (splitChild.Overlaps(position))
{
return splitChild;
}
}
return null;
}
public bool TrySpillWithSiblingOffset()
{
foreach (LiveInterval splitChild in _parent._childs.Values)
{
if (splitChild.IsSpilled)
{
Spill(splitChild.SpillOffset);
return true;
}
}
return false;
}
public void Spill(int offset)
{
SpillOffset = offset;
}
public int CompareTo(LiveInterval other)
{
if (_ranges.Count == 0 || other._ranges.Count == 0)
{
return _ranges.Count.CompareTo(other._ranges.Count);
}
return _ranges[0].Start.CompareTo(other._ranges[0].Start);
}
public override string ToString()
{
return string.Join("; ", _ranges);
}
}
}