ryujinx/ARMeilleure/Translation/ArmEmitterContext.cs
FICTURE7 9d7627af64
Add multi-level function table (#2228)
* Add AddressTable<T>

* Use AddressTable<T> for dispatch

* Remove JumpTable & co.

* Add fallback for out of range addresses

* Add PPTC support

* Add documentation to `AddressTable<T>`

* Make AddressTable<T> configurable

* Fix table walk

* Fix IsMapped check

* Remove CountTableCapacity

* Add PPTC support for fast path

* Rename IsMapped to IsValid

* Remove stale comment

* Change format of address in exception message

* Add TranslatorStubs

* Split DispatchStub

Avoids recompilation of stubs during tests.

* Add hint for 64bit or 32bit

* Add documentation to `Symbol`

* Add documentation to `TranslatorStubs`

Make `TranslatorStubs` disposable as well.

* Add documentation to `SymbolType`

* Add `AddressTableEventSource` to monitor function table size

Add an EventSource which measures the amount of unmanaged bytes
allocated by AddressTable<T> instances.

 dotnet-counters monitor -n Ryujinx --counters ARMeilleure

* Add `AllowLcqInFunctionTable` optimization toggle

This is to reduce the impact this change has on the test duration.
Before everytime a test was ran, the FunctionTable would be initialized
and populated so that the newly compiled test would get registered to
it.

* Implement unmanaged dispatcher

Uses the DispatchStub to dispatch into the next translation, which
allows execution to stay in unmanaged for longer and skips a
ConcurrentDictionary look up when the target translation has been
registered to the FunctionTable.

* Remove redundant null check

* Tune levels of FunctionTable

Uses 5 levels instead of 4 and change unit of AddressTableEventSource
from KB to MB.

* Use 64-bit function table

Improves codegen for direct branches:

    mov qword [rax+0x408],0x10603560
 -  mov rcx,sub_10603560_OFFSET
 -  mov ecx,[rcx]
 -  mov ecx,ecx
 -  mov rdx,JIT_CACHE_BASE
 -  add rdx,rcx
 +  mov rcx,sub_10603560
 +  mov rdx,[rcx]
    mov rcx,rax

Improves codegen for dispatch stub:

    and rax,byte +0x1f
 -  mov eax,[rcx+rax*4]
 -  mov eax,eax
 -  mov rcx,JIT_CACHE_BASE
 -  lea rax,[rcx+rax]
 +  mov rax,[rcx+rax*8]
    mov rcx,rbx

* Remove `JitCacheSymbol` & `JitCache.Offset`

* Turn `Translator.Translate` into an instance method

We do not have to add more parameter to this method and related ones as
new structures are added & needed for translation.

* Add symbol only when PTC is enabled

Address LDj3SNuD's feedback

* Change `NativeContext.Running` to a 32-bit integer

* Fix PageTable symbol for host mapped
2021-05-29 18:06:28 -03:00

199 lines
6.8 KiB
C#

using ARMeilleure.Common;
using ARMeilleure.Decoders;
using ARMeilleure.Diagnostics;
using ARMeilleure.Instructions;
using ARMeilleure.IntermediateRepresentation;
using ARMeilleure.Memory;
using ARMeilleure.State;
using ARMeilleure.Translation.PTC;
using System;
using System.Collections.Generic;
using System.Reflection;
using static ARMeilleure.IntermediateRepresentation.OperandHelper;
namespace ARMeilleure.Translation
{
class ArmEmitterContext : EmitterContext
{
private readonly Dictionary<ulong, Operand> _labels;
private OpCode _optOpLastCompare;
private OpCode _optOpLastFlagSet;
private Operand _optCmpTempN;
private Operand _optCmpTempM;
private Block _currBlock;
public Block CurrBlock
{
get
{
return _currBlock;
}
set
{
_currBlock = value;
ResetBlockState();
}
}
public OpCode CurrOp { get; set; }
public IMemoryManager Memory { get; }
public bool HasPtc { get; }
public EntryTable<uint> CountTable { get; }
public AddressTable<ulong> FunctionTable { get; }
public TranslatorStubs Stubs { get; }
public ulong EntryAddress { get; }
public bool HighCq { get; }
public Aarch32Mode Mode { get; }
public ArmEmitterContext(
IMemoryManager memory,
EntryTable<uint> countTable,
AddressTable<ulong> funcTable,
TranslatorStubs stubs,
ulong entryAddress,
bool highCq,
Aarch32Mode mode)
{
HasPtc = Ptc.State != PtcState.Disabled;
Memory = memory;
CountTable = countTable;
FunctionTable = funcTable;
Stubs = stubs;
EntryAddress = entryAddress;
HighCq = highCq;
Mode = mode;
_labels = new Dictionary<ulong, Operand>();
}
public override Operand Call(MethodInfo info, params Operand[] callArgs)
{
if (!HasPtc)
{
return base.Call(info, callArgs);
}
else
{
int index = Delegates.GetDelegateIndex(info);
IntPtr funcPtr = Delegates.GetDelegateFuncPtrByIndex(index);
OperandType returnType = GetOperandType(info.ReturnType);
Symbol symbol = new Symbol(SymbolType.DelegateTable, (ulong)index);
Symbols.Add((ulong)funcPtr.ToInt64(), info.Name);
return Call(Const(funcPtr.ToInt64(), symbol), returnType, callArgs);
}
}
public Operand GetLabel(ulong address)
{
if (!_labels.TryGetValue(address, out Operand label))
{
label = Label();
_labels.Add(address, label);
}
return label;
}
public void MarkComparison(Operand n, Operand m)
{
_optOpLastCompare = CurrOp;
_optCmpTempN = Copy(n);
_optCmpTempM = Copy(m);
}
public void MarkFlagSet(PState stateFlag)
{
// Set this only if any of the NZCV flag bits were modified.
// This is used to ensure that when emiting a direct IL branch
// instruction for compare + branch sequences, we're not expecting
// to use comparison values from an old instruction, when in fact
// the flags were already overwritten by another instruction further along.
if (stateFlag >= PState.VFlag)
{
_optOpLastFlagSet = CurrOp;
}
}
private void ResetBlockState()
{
_optOpLastCompare = null;
_optOpLastFlagSet = null;
}
public Operand TryGetComparisonResult(Condition condition)
{
if (_optOpLastCompare == null || _optOpLastCompare != _optOpLastFlagSet)
{
return null;
}
Operand n = _optCmpTempN;
Operand m = _optCmpTempM;
InstName cmpName = _optOpLastCompare.Instruction.Name;
if (cmpName == InstName.Subs)
{
switch (condition)
{
case Condition.Eq: return ICompareEqual (n, m);
case Condition.Ne: return ICompareNotEqual (n, m);
case Condition.GeUn: return ICompareGreaterOrEqualUI(n, m);
case Condition.LtUn: return ICompareLessUI (n, m);
case Condition.GtUn: return ICompareGreaterUI (n, m);
case Condition.LeUn: return ICompareLessOrEqualUI (n, m);
case Condition.Ge: return ICompareGreaterOrEqual (n, m);
case Condition.Lt: return ICompareLess (n, m);
case Condition.Gt: return ICompareGreater (n, m);
case Condition.Le: return ICompareLessOrEqual (n, m);
}
}
else if (cmpName == InstName.Adds && _optOpLastCompare is IOpCodeAluImm op)
{
// There are several limitations that needs to be taken into account for CMN comparisons:
// - The unsigned comparisons are not valid, as they depend on the
// carry flag value, and they will have different values for addition and
// subtraction. For addition, it's carry, and for subtraction, it's borrow.
// So, we need to make sure we're not doing a unsigned compare for the CMN case.
// - We can only do the optimization for the immediate variants,
// because when the second operand value is exactly INT_MIN, we can't
// negate the value as theres no positive counterpart.
// Such invalid values can't be encoded on the immediate encodings.
if (op.RegisterSize == RegisterSize.Int32)
{
m = Const((int)-op.Immediate);
}
else
{
m = Const(-op.Immediate);
}
switch (condition)
{
case Condition.Eq: return ICompareEqual (n, m);
case Condition.Ne: return ICompareNotEqual (n, m);
case Condition.Ge: return ICompareGreaterOrEqual(n, m);
case Condition.Lt: return ICompareLess (n, m);
case Condition.Gt: return ICompareGreater (n, m);
case Condition.Le: return ICompareLessOrEqual (n, m);
}
}
return null;
}
}
}