Ryujinx-uplift/ARMeilleure/Instructions/InstEmitFlowHelper.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

236 lines
7.6 KiB
C#

using ARMeilleure.Decoders;
using ARMeilleure.IntermediateRepresentation;
using ARMeilleure.State;
using ARMeilleure.Translation;
using ARMeilleure.Translation.Cache;
using ARMeilleure.Translation.PTC;
using static ARMeilleure.Instructions.InstEmitHelper;
using static ARMeilleure.IntermediateRepresentation.OperandHelper;
namespace ARMeilleure.Instructions
{
static class InstEmitFlowHelper
{
public static void EmitCondBranch(ArmEmitterContext context, Operand target, Condition cond)
{
if (cond != Condition.Al)
{
context.BranchIfTrue(target, GetCondTrue(context, cond));
}
else
{
context.Branch(target);
}
}
public static Operand GetCondTrue(ArmEmitterContext context, Condition condition)
{
Operand cmpResult = context.TryGetComparisonResult(condition);
if (cmpResult != null)
{
return cmpResult;
}
Operand value = Const(1);
Operand Inverse(Operand val)
{
return context.BitwiseExclusiveOr(val, Const(1));
}
switch (condition)
{
case Condition.Eq:
value = GetFlag(PState.ZFlag);
break;
case Condition.Ne:
value = Inverse(GetFlag(PState.ZFlag));
break;
case Condition.GeUn:
value = GetFlag(PState.CFlag);
break;
case Condition.LtUn:
value = Inverse(GetFlag(PState.CFlag));
break;
case Condition.Mi:
value = GetFlag(PState.NFlag);
break;
case Condition.Pl:
value = Inverse(GetFlag(PState.NFlag));
break;
case Condition.Vs:
value = GetFlag(PState.VFlag);
break;
case Condition.Vc:
value = Inverse(GetFlag(PState.VFlag));
break;
case Condition.GtUn:
{
Operand c = GetFlag(PState.CFlag);
Operand z = GetFlag(PState.ZFlag);
value = context.BitwiseAnd(c, Inverse(z));
break;
}
case Condition.LeUn:
{
Operand c = GetFlag(PState.CFlag);
Operand z = GetFlag(PState.ZFlag);
value = context.BitwiseOr(Inverse(c), z);
break;
}
case Condition.Ge:
{
Operand n = GetFlag(PState.NFlag);
Operand v = GetFlag(PState.VFlag);
value = context.ICompareEqual(n, v);
break;
}
case Condition.Lt:
{
Operand n = GetFlag(PState.NFlag);
Operand v = GetFlag(PState.VFlag);
value = context.ICompareNotEqual(n, v);
break;
}
case Condition.Gt:
{
Operand n = GetFlag(PState.NFlag);
Operand z = GetFlag(PState.ZFlag);
Operand v = GetFlag(PState.VFlag);
value = context.BitwiseAnd(Inverse(z), context.ICompareEqual(n, v));
break;
}
case Condition.Le:
{
Operand n = GetFlag(PState.NFlag);
Operand z = GetFlag(PState.ZFlag);
Operand v = GetFlag(PState.VFlag);
value = context.BitwiseOr(z, context.ICompareNotEqual(n, v));
break;
}
}
return value;
}
public static void EmitCall(ArmEmitterContext context, ulong immediate)
{
bool isRecursive = immediate == context.EntryAddress;
if (isRecursive)
{
context.Branch(context.GetLabel(immediate));
}
else
{
EmitTableBranch(context, Const(immediate), isJump: false);
}
}
public static void EmitVirtualCall(ArmEmitterContext context, Operand target)
{
EmitTableBranch(context, target, isJump: false);
}
public static void EmitVirtualJump(ArmEmitterContext context, Operand target, bool isReturn)
{
if (isReturn)
{
context.Return(target);
}
else
{
EmitTableBranch(context, target, isJump: true);
}
}
private static void EmitTableBranch(ArmEmitterContext context, Operand guestAddress, bool isJump)
{
context.StoreToContext();
if (guestAddress.Type == OperandType.I32)
{
guestAddress = context.ZeroExtend32(OperandType.I64, guestAddress);
}
// Store the target guest address into the native context. The stubs uses this address to dispatch into the
// next translation.
Operand nativeContext = context.LoadArgument(OperandType.I64, 0);
Operand dispAddressAddr = context.Add(nativeContext, Const((ulong)NativeContext.GetDispatchAddressOffset()));
context.Store(dispAddressAddr, guestAddress);
Operand hostAddress;
// If address is mapped onto the function table, we can skip the table walk. Otherwise we fallback
// onto the dispatch stub.
if (guestAddress.Kind == OperandKind.Constant && context.FunctionTable.IsValid(guestAddress.Value))
{
Operand hostAddressAddr = !context.HasPtc ?
Const(ref context.FunctionTable.GetValue(guestAddress.Value)) :
Const(ref context.FunctionTable.GetValue(guestAddress.Value), new Symbol(SymbolType.FunctionTable, guestAddress.Value));
hostAddress = context.Load(OperandType.I64, hostAddressAddr);
}
else
{
hostAddress = !context.HasPtc ?
Const((long)context.Stubs.DispatchStub) :
Const((long)context.Stubs.DispatchStub, Ptc.DispatchStubSymbol);
}
if (isJump)
{
context.Tailcall(hostAddress, nativeContext);
}
else
{
OpCode op = context.CurrOp;
Operand returnAddress = context.Call(hostAddress, OperandType.I64, nativeContext);
context.LoadFromContext();
// Note: The return value of a translated function is always an Int64 with the address execution has
// returned to. We expect this address to be immediately after the current instruction, if it isn't we
// keep returning until we reach the dispatcher.
Operand nextAddr = Const((long)op.Address + op.OpCodeSizeInBytes);
// Try to continue within this block.
// If the return address isn't to our next instruction, we need to return so the JIT can figure out
// what to do.
Operand lblContinue = context.GetLabel(nextAddr.Value);
context.BranchIf(lblContinue, returnAddress, nextAddr, Comparison.Equal, BasicBlockFrequency.Cold);
context.Return(returnAddress);
}
}
}
}