1
0
Fork 0
mirror of https://github.com/Ryujinx/Ryujinx.git synced 2024-10-01 12:30:00 +02:00
Ryujinx/Ryujinx.Graphics/Graphics3d/NvGpuFifo.cs
Alex Barney b2b736abc2 Misc cleanup (#708)
* Fix typos

* Remove unneeded using statements

* Enforce var style more

* Remove redundant qualifiers

* Fix some indentation

* Disable naming warnings on files with external enum names

* Fix build

* Mass find & replace for comments with no spacing

* Standardize todo capitalization and for/if spacing
2019-07-02 04:39:22 +02:00

176 lines
No EOL
5 KiB
C#

using Ryujinx.Graphics.Memory;
namespace Ryujinx.Graphics.Graphics3d
{
class NvGpuFifo
{
private const int MacrosCount = 0x80;
private const int MacroIndexMask = MacrosCount - 1;
// Note: The size of the macro memory is unknown, we just make
// a guess here and use 256kb as the size. Increase if needed.
private const int MmeWords = 256 * 256;
private NvGpu _gpu;
private NvGpuEngine[] _subChannels;
private struct CachedMacro
{
public int Position { get; private set; }
private bool _executionPending;
private int _argument;
private MacroInterpreter _interpreter;
public CachedMacro(NvGpuFifo pFifo, INvGpuEngine engine, int position)
{
Position = position;
_executionPending = false;
_argument = 0;
_interpreter = new MacroInterpreter(pFifo, engine);
}
public void StartExecution(int argument)
{
_argument = argument;
_executionPending = true;
}
public void Execute(NvGpuVmm vmm, int[] mme)
{
if (_executionPending)
{
_executionPending = false;
_interpreter?.Execute(vmm, mme, Position, _argument);
}
}
public void PushArgument(int argument)
{
_interpreter?.Fifo.Enqueue(argument);
}
}
private int _currMacroPosition;
private int _currMacroBindIndex;
private CachedMacro[] _macros;
private int[] _mme;
public NvGpuFifo(NvGpu gpu)
{
_gpu = gpu;
_subChannels = new NvGpuEngine[8];
_macros = new CachedMacro[MacrosCount];
_mme = new int[MmeWords];
}
public void CallMethod(NvGpuVmm vmm, GpuMethodCall methCall)
{
if ((NvGpuFifoMeth)methCall.Method == NvGpuFifoMeth.BindChannel)
{
NvGpuEngine engine = (NvGpuEngine)methCall.Argument;
_subChannels[methCall.SubChannel] = engine;
}
else
{
switch (_subChannels[methCall.SubChannel])
{
case NvGpuEngine._2d: Call2dMethod (vmm, methCall); break;
case NvGpuEngine._3d: Call3dMethod (vmm, methCall); break;
case NvGpuEngine.P2mf: CallP2mfMethod(vmm, methCall); break;
case NvGpuEngine.M2mf: CallM2mfMethod(vmm, methCall); break;
}
}
}
private void Call2dMethod(NvGpuVmm vmm, GpuMethodCall methCall)
{
_gpu.Engine2d.CallMethod(vmm, methCall);
}
private void Call3dMethod(NvGpuVmm vmm, GpuMethodCall methCall)
{
if (methCall.Method < 0x80)
{
switch ((NvGpuFifoMeth)methCall.Method)
{
case NvGpuFifoMeth.SetMacroUploadAddress:
{
_currMacroPosition = methCall.Argument;
break;
}
case NvGpuFifoMeth.SendMacroCodeData:
{
_mme[_currMacroPosition++] = methCall.Argument;
break;
}
case NvGpuFifoMeth.SetMacroBindingIndex:
{
_currMacroBindIndex = methCall.Argument;
break;
}
case NvGpuFifoMeth.BindMacro:
{
int position = methCall.Argument;
_macros[_currMacroBindIndex++] = new CachedMacro(this, _gpu.Engine3d, position);
break;
}
default: CallP2mfMethod(vmm, methCall); break;
}
}
else if (methCall.Method < 0xe00)
{
_gpu.Engine3d.CallMethod(vmm, methCall);
}
else
{
int macroIndex = (methCall.Method >> 1) & MacroIndexMask;
if ((methCall.Method & 1) != 0)
{
_macros[macroIndex].PushArgument(methCall.Argument);
}
else
{
_macros[macroIndex].StartExecution(methCall.Argument);
}
if (methCall.IsLastCall)
{
_macros[macroIndex].Execute(vmm, _mme);
}
}
}
private void CallP2mfMethod(NvGpuVmm vmm, GpuMethodCall methCall)
{
_gpu.EngineP2mf.CallMethod(vmm, methCall);
}
private void CallM2mfMethod(NvGpuVmm vmm, GpuMethodCall methCall)
{
_gpu.EngineM2mf.CallMethod(vmm, methCall);
}
}
}