mirror of
https://github.com/Ryujinx/Ryujinx.git
synced 2024-10-01 12:30:00 +02:00
d904706fc0
* Implement Jump Table for Native Calls NOTE: this slows down rejit considerably! Not recommended to be used without codegen optimisation or AOT. - Does not work on Linux - A32 needs an additional commit. * A32 Support (WIP) * Actually write Direct Call pointers to the table That would help. * Direct Calls: Rather than returning to the translator, attempt to keep within the native stack frame. A return to the translator can still happen, but only by exceptionally bubbling up to it. Also: - Always translate lowCq as a function. Faster interop with the direct jumps, and this will be useful in future if we want to do speculative translation. - Tail Call Detection: after the decoding stage, detect if we do a tail call, and avoid translating into it. Detected if a jump is made to an address outwith the contiguous sequence of blocks surrounding the entry point. The goal is to reduce code touched by jit and rejit. * A32 Support * Use smaller max function size for lowCq, fix exceptional returns When a return has an unexpected value and there is no code block following this one, we now return the value rather than continuing. * CompareAndSwap (buggy) * Ensure CompareAndSwap does not get optimized away. * Use CompareAndSwap to make the dynamic table thread safe. * Tail call for linux, throw on too many arguments. * Combine CompareAndSwap 128 and 32/64. They emit different IR instructions since their PreAllocator behaviour is different, but now they just have one function on EmitterContext. * Fix issues separating from optimisations. * Use a stub to find and execute missing functions. This allows us to skip doing many runtime comparisons and branches, and reduces the amount of code we need to emit significantly. For the indirect call table, this stub also does the work of moving in the highCq address to the table when one is found. * Make Jump Tables and Jit Cache dynmically resize Reserve virtual memory, commit as needed. * Move TailCallRemover to its own class. * Multithreaded Translation (based on heuristic) A poor one, at that. Need to get core count for a better one, which means a lot of OS specific garbage. * Better priority management for background threads. * Bound core limit a bit more Past a certain point the load is not paralellizable and starts stealing from the main thread. Likely due to GC, memory, heap allocation thread contention. Reduce by one core til optimisations come to improve the situation. * Fix memory management on linux. * Temporary solution to some sync problems. This will make sure threads exit correctly, most of the time. There is a potential race where setting the sync counter to 0 does nothing (counter stays at what it was before, thread could take too long to exit), but we need to find a better way to do this anyways. Synchronization frequency has been tightened as we never enter blockwise segments of code. Essentially this means, check every x functions or loop iterations, before lowcq blocks existed and were worth just as much. Ideally it should be done in a better way, since functions can be anywhere from 1 to 5000 instructions. (maybe based on host timer, or an interrupt flag from a scheduler thread) * Address feedback minus CompareAndSwap change. * Use default ReservedRegion granularity. * Merge CompareAndSwap with its V128 variant. * We already got the source, no need to do it again. * Make sure all background translation threads exit. * Fix CompareAndSwap128 Detection criteria was a bit scuffed. * Address Comments.
278 lines
No EOL
9.7 KiB
C#
278 lines
No EOL
9.7 KiB
C#
using ARMeilleure.Decoders;
|
|
using ARMeilleure.Diagnostics;
|
|
using ARMeilleure.Instructions;
|
|
using ARMeilleure.IntermediateRepresentation;
|
|
using ARMeilleure.Memory;
|
|
using ARMeilleure.State;
|
|
using System;
|
|
using System.Collections.Concurrent;
|
|
using System.Threading;
|
|
|
|
using static ARMeilleure.IntermediateRepresentation.OperandHelper;
|
|
|
|
namespace ARMeilleure.Translation
|
|
{
|
|
public class Translator
|
|
{
|
|
private const ulong CallFlag = InstEmitFlowHelper.CallFlag;
|
|
|
|
private const bool AlwaysTranslateFunctions = true; // If false, only translates a single block for lowCq.
|
|
|
|
private MemoryManager _memory;
|
|
|
|
private ConcurrentDictionary<ulong, TranslatedFunction> _funcs;
|
|
|
|
private JumpTable _jumpTable;
|
|
|
|
private PriorityQueue<RejitRequest> _backgroundQueue;
|
|
|
|
private AutoResetEvent _backgroundTranslatorEvent;
|
|
|
|
private volatile int _threadCount;
|
|
|
|
public Translator(MemoryManager memory)
|
|
{
|
|
_memory = memory;
|
|
|
|
_funcs = new ConcurrentDictionary<ulong, TranslatedFunction>();
|
|
|
|
_jumpTable = JumpTable.Instance;
|
|
|
|
_backgroundQueue = new PriorityQueue<RejitRequest>(2);
|
|
|
|
_backgroundTranslatorEvent = new AutoResetEvent(false);
|
|
|
|
DirectCallStubs.InitializeStubs();
|
|
}
|
|
|
|
private void TranslateQueuedSubs()
|
|
{
|
|
while (_threadCount != 0)
|
|
{
|
|
if (_backgroundQueue.TryDequeue(out RejitRequest request))
|
|
{
|
|
TranslatedFunction func = Translate(request.Address, request.Mode, highCq: true);
|
|
|
|
_funcs.AddOrUpdate(request.Address, func, (key, oldFunc) => func);
|
|
_jumpTable.RegisterFunction(request.Address, func);
|
|
}
|
|
else
|
|
{
|
|
_backgroundTranslatorEvent.WaitOne();
|
|
}
|
|
}
|
|
_backgroundTranslatorEvent.Set(); // Wake up any other background translator threads, to encourage them to exit.
|
|
}
|
|
|
|
public void Execute(State.ExecutionContext context, ulong address)
|
|
{
|
|
if (Interlocked.Increment(ref _threadCount) == 1)
|
|
{
|
|
// Simple heuristic, should be user configurable in future. (1 for 4 core/ht or less, 2 for 6 core+ht etc).
|
|
// All threads are normal priority except from the last, which just fills as much of the last core as the os lets it with a low priority.
|
|
// If we only have one rejit thread, it should be normal priority as highCq code is performance critical.
|
|
// TODO: Use physical cores rather than logical. This only really makes sense for processors with hyperthreading. Requires OS specific code.
|
|
int unboundedThreadCount = Math.Max(1, (Environment.ProcessorCount - 6) / 3);
|
|
int threadCount = Math.Min(3, unboundedThreadCount);
|
|
for (int i = 0; i < threadCount; i++)
|
|
{
|
|
bool last = i != 0 && i == unboundedThreadCount - 1;
|
|
Thread backgroundTranslatorThread = new Thread(TranslateQueuedSubs)
|
|
{
|
|
Name = "CPU.BackgroundTranslatorThread." + i,
|
|
Priority = last ? ThreadPriority.Lowest : ThreadPriority.Normal
|
|
};
|
|
|
|
backgroundTranslatorThread.Start();
|
|
}
|
|
}
|
|
|
|
Statistics.InitializeTimer();
|
|
|
|
NativeInterface.RegisterThread(context, _memory, this);
|
|
|
|
do
|
|
{
|
|
address = ExecuteSingle(context, address);
|
|
}
|
|
while (context.Running && (address & ~1UL) != 0);
|
|
|
|
NativeInterface.UnregisterThread();
|
|
|
|
if (Interlocked.Decrement(ref _threadCount) == 0)
|
|
{
|
|
_backgroundTranslatorEvent.Set();
|
|
}
|
|
}
|
|
|
|
public ulong ExecuteSingle(State.ExecutionContext context, ulong address)
|
|
{
|
|
TranslatedFunction func = GetOrTranslate(address, context.ExecutionMode);
|
|
|
|
Statistics.StartTimer();
|
|
|
|
ulong nextAddr = func.Execute(context);
|
|
|
|
Statistics.StopTimer(address);
|
|
|
|
return nextAddr;
|
|
}
|
|
|
|
internal TranslatedFunction GetOrTranslate(ulong address, ExecutionMode mode)
|
|
{
|
|
// TODO: Investigate how we should handle code at unaligned addresses.
|
|
// Currently, those low bits are used to store special flags.
|
|
bool isCallTarget = (address & CallFlag) != 0;
|
|
|
|
address &= ~CallFlag;
|
|
|
|
if (!_funcs.TryGetValue(address, out TranslatedFunction func))
|
|
{
|
|
func = Translate(address, mode, highCq: false);
|
|
|
|
_funcs.TryAdd(address, func);
|
|
}
|
|
else if (isCallTarget && func.ShouldRejit())
|
|
{
|
|
_backgroundQueue.Enqueue(0, new RejitRequest(address, mode));
|
|
|
|
_backgroundTranslatorEvent.Set();
|
|
}
|
|
|
|
return func;
|
|
}
|
|
|
|
private TranslatedFunction Translate(ulong address, ExecutionMode mode, bool highCq)
|
|
{
|
|
ArmEmitterContext context = new ArmEmitterContext(_memory, _jumpTable, (long)address, highCq, Aarch32Mode.User);
|
|
|
|
Logger.StartPass(PassName.Decoding);
|
|
|
|
Block[] blocks = AlwaysTranslateFunctions
|
|
? Decoder.DecodeFunction (_memory, address, mode, highCq)
|
|
: Decoder.DecodeBasicBlock(_memory, address, mode);
|
|
|
|
Logger.EndPass(PassName.Decoding);
|
|
|
|
Logger.StartPass(PassName.Translation);
|
|
|
|
EmitSynchronization(context);
|
|
|
|
if (blocks[0].Address != address)
|
|
{
|
|
context.Branch(context.GetLabel(address));
|
|
}
|
|
|
|
ControlFlowGraph cfg = EmitAndGetCFG(context, blocks);
|
|
|
|
Logger.EndPass(PassName.Translation);
|
|
|
|
Logger.StartPass(PassName.RegisterUsage);
|
|
|
|
RegisterUsage.RunPass(cfg, mode, isCompleteFunction: false);
|
|
|
|
Logger.EndPass(PassName.RegisterUsage);
|
|
|
|
OperandType[] argTypes = new OperandType[] { OperandType.I64 };
|
|
|
|
CompilerOptions options = highCq
|
|
? CompilerOptions.HighCq
|
|
: CompilerOptions.None;
|
|
|
|
GuestFunction func = Compiler.Compile<GuestFunction>(cfg, argTypes, OperandType.I64, options);
|
|
|
|
return new TranslatedFunction(func, rejit: !highCq);
|
|
}
|
|
|
|
private static ControlFlowGraph EmitAndGetCFG(ArmEmitterContext context, Block[] blocks)
|
|
{
|
|
for (int blkIndex = 0; blkIndex < blocks.Length; blkIndex++)
|
|
{
|
|
Block block = blocks[blkIndex];
|
|
|
|
context.CurrBlock = block;
|
|
|
|
context.MarkLabel(context.GetLabel(block.Address));
|
|
|
|
for (int opcIndex = 0; opcIndex < block.OpCodes.Count; opcIndex++)
|
|
{
|
|
OpCode opCode = block.OpCodes[opcIndex];
|
|
|
|
context.CurrOp = opCode;
|
|
|
|
bool isLastOp = opcIndex == block.OpCodes.Count - 1;
|
|
|
|
if (isLastOp && block.Branch != null && block.Branch.Address <= block.Address)
|
|
{
|
|
EmitSynchronization(context);
|
|
}
|
|
|
|
Operand lblPredicateSkip = null;
|
|
|
|
if (opCode is OpCode32 op && op.Cond < Condition.Al)
|
|
{
|
|
lblPredicateSkip = Label();
|
|
|
|
InstEmitFlowHelper.EmitCondBranch(context, lblPredicateSkip, op.Cond.Invert());
|
|
}
|
|
|
|
if (opCode.Instruction.Emitter != null)
|
|
{
|
|
opCode.Instruction.Emitter(context);
|
|
}
|
|
else
|
|
{
|
|
throw new InvalidOperationException($"Invalid instruction \"{opCode.Instruction.Name}\".");
|
|
}
|
|
|
|
if (lblPredicateSkip != null)
|
|
{
|
|
context.MarkLabel(lblPredicateSkip);
|
|
|
|
// If this is the last op on the block, and there's no "next" block
|
|
// after this one, then we have to return right now, with the address
|
|
// of the next instruction to be executed (in the case that the condition
|
|
// is false, and the branch was not taken, as all basic blocks should end
|
|
// with some kind of branch).
|
|
if (isLastOp && block.Next == null)
|
|
{
|
|
InstEmitFlowHelper.EmitTailContinue(context, Const(opCode.Address + (ulong)opCode.OpCodeSizeInBytes));
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
return context.GetControlFlowGraph();
|
|
}
|
|
|
|
private static void EmitSynchronization(EmitterContext context)
|
|
{
|
|
long countOffs = NativeContext.GetCounterOffset();
|
|
|
|
Operand countAddr = context.Add(context.LoadArgument(OperandType.I64, 0), Const(countOffs));
|
|
|
|
Operand count = context.Load(OperandType.I32, countAddr);
|
|
|
|
Operand lblNonZero = Label();
|
|
Operand lblExit = Label();
|
|
|
|
context.BranchIfTrue(lblNonZero, count);
|
|
|
|
Operand running = context.Call(new _Bool(NativeInterface.CheckSynchronization));
|
|
|
|
context.BranchIfTrue(lblExit, running);
|
|
|
|
context.Return(Const(0L));
|
|
|
|
context.Branch(lblExit);
|
|
|
|
context.MarkLabel(lblNonZero);
|
|
|
|
count = context.Subtract(count, Const(1));
|
|
|
|
context.Store(countAddr, count);
|
|
|
|
context.MarkLabel(lblExit);
|
|
}
|
|
}
|
|
} |