1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
|
using Ryujinx.Graphics.Device;
using Ryujinx.Graphics.Gpu.Engine.Compute;
using Ryujinx.Graphics.Gpu.Engine.Dma;
using Ryujinx.Graphics.Gpu.Engine.InlineToMemory;
using Ryujinx.Graphics.Gpu.Engine.Threed;
using Ryujinx.Graphics.Gpu.Engine.Twod;
using Ryujinx.Graphics.Gpu.Image;
using Ryujinx.Graphics.Gpu.Memory;
using System;
using System.Runtime.CompilerServices;
namespace Ryujinx.Graphics.Gpu.Engine.GPFifo
{
/// <summary>
/// Represents a GPU General Purpose FIFO command processor.
/// </summary>
class GPFifoProcessor : IDisposable
{
private const int MacrosCount = 0x80;
private const int MacroIndexMask = MacrosCount - 1;
private const int LoadInlineDataMethodOffset = 0x6d;
private const int UniformBufferUpdateDataMethodOffset = 0x8e4;
private readonly GpuChannel _channel;
/// <summary>
/// Channel memory manager.
/// </summary>
public MemoryManager MemoryManager => _channel.MemoryManager;
/// <summary>
/// Channel texture manager.
/// </summary>
public TextureManager TextureManager => _channel.TextureManager;
/// <summary>
/// 3D Engine.
/// </summary>
public ThreedClass ThreedClass => _3dClass;
/// <summary>
/// Internal GPFIFO state.
/// </summary>
private struct DmaState
{
public int Method;
public int SubChannel;
public int MethodCount;
public bool NonIncrementing;
public bool IncrementOnce;
}
private DmaState _state;
private readonly ThreedClass _3dClass;
private readonly ComputeClass _computeClass;
private readonly InlineToMemoryClass _i2mClass;
private readonly TwodClass _2dClass;
private readonly DmaClass _dmaClass;
private readonly GPFifoClass _fifoClass;
/// <summary>
/// Creates a new instance of the GPU General Purpose FIFO command processor.
/// </summary>
/// <param name="context">GPU context</param>
/// <param name="channel">Channel that the GPFIFO processor belongs to</param>
public GPFifoProcessor(GpuContext context, GpuChannel channel)
{
_channel = channel;
_fifoClass = new GPFifoClass(context, this);
_3dClass = new ThreedClass(context, channel, _fifoClass);
_computeClass = new ComputeClass(context, channel, _3dClass);
_i2mClass = new InlineToMemoryClass(context, channel);
_2dClass = new TwodClass(channel);
_dmaClass = new DmaClass(context, channel, _3dClass);
}
/// <summary>
/// Processes a command buffer.
/// </summary>
/// <param name="baseGpuVa">Base GPU virtual address of the command buffer</param>
/// <param name="commandBuffer">Command buffer</param>
public void Process(ulong baseGpuVa, ReadOnlySpan<int> commandBuffer)
{
for (int index = 0; index < commandBuffer.Length; index++)
{
int command = commandBuffer[index];
ulong gpuVa = baseGpuVa + (ulong)index * 4;
if (_state.MethodCount != 0)
{
if (TryFastI2mBufferUpdate(commandBuffer, ref index))
{
continue;
}
Send(gpuVa, _state.Method, command, _state.SubChannel, _state.MethodCount <= 1);
if (!_state.NonIncrementing)
{
_state.Method++;
}
if (_state.IncrementOnce)
{
_state.NonIncrementing = true;
}
_state.MethodCount--;
}
else
{
CompressedMethod meth = Unsafe.As<int, CompressedMethod>(ref command);
if (TryFastUniformBufferUpdate(meth, commandBuffer, index))
{
index += meth.MethodCount;
continue;
}
switch (meth.SecOp)
{
case SecOp.IncMethod:
case SecOp.NonIncMethod:
case SecOp.OneInc:
_state.Method = meth.MethodAddress;
_state.SubChannel = meth.MethodSubchannel;
_state.MethodCount = meth.MethodCount;
_state.IncrementOnce = meth.SecOp == SecOp.OneInc;
_state.NonIncrementing = meth.SecOp == SecOp.NonIncMethod;
break;
case SecOp.ImmdDataMethod:
Send(gpuVa, meth.MethodAddress, meth.ImmdData, meth.MethodSubchannel, true);
break;
}
}
}
_3dClass.FlushUboDirty();
}
/// <summary>
/// Tries to perform a fast Inline-to-Memory data update.
/// If successful, all data will be copied at once, and <see cref="DmaState.MethodCount"/>
/// command buffer entries will be consumed.
/// </summary>
/// <param name="commandBuffer">Command buffer where the data is contained</param>
/// <param name="offset">Offset at <paramref name="commandBuffer"/> where the data is located, auto-incremented on success</param>
/// <returns>True if the fast copy was successful, false otherwise</returns>
[MethodImpl(MethodImplOptions.AggressiveInlining)]
private bool TryFastI2mBufferUpdate(ReadOnlySpan<int> commandBuffer, ref int offset)
{
if (_state.Method == LoadInlineDataMethodOffset && _state.NonIncrementing && _state.SubChannel <= 2)
{
int availableCount = commandBuffer.Length - offset;
int consumeCount = Math.Min(_state.MethodCount, availableCount);
var data = commandBuffer.Slice(offset, consumeCount);
if (_state.SubChannel == 0)
{
_3dClass.LoadInlineData(data);
}
else if (_state.SubChannel == 1)
{
_computeClass.LoadInlineData(data);
}
else /* if (_state.SubChannel == 2) */
{
_i2mClass.LoadInlineData(data);
}
offset += consumeCount - 1;
_state.MethodCount -= consumeCount;
return true;
}
return false;
}
/// <summary>
/// Tries to perform a fast constant buffer data update.
/// If successful, all data will be copied at once, and <see cref="CompressedMethod.MethodCount"/> + 1
/// command buffer entries will be consumed.
/// </summary>
/// <param name="meth">Compressed method to be checked</param>
/// <param name="commandBuffer">Command buffer where <paramref name="meth"/> is contained</param>
/// <param name="offset">Offset at <paramref name="commandBuffer"/> where <paramref name="meth"/> is located</param>
/// <returns>True if the fast copy was successful, false otherwise</returns>
[MethodImpl(MethodImplOptions.AggressiveInlining)]
private bool TryFastUniformBufferUpdate(CompressedMethod meth, ReadOnlySpan<int> commandBuffer, int offset)
{
int availableCount = commandBuffer.Length - offset;
if (meth.MethodAddress == UniformBufferUpdateDataMethodOffset &&
meth.MethodCount < availableCount &&
meth.SecOp == SecOp.NonIncMethod)
{
_3dClass.ConstantBufferUpdate(commandBuffer.Slice(offset + 1, meth.MethodCount));
return true;
}
return false;
}
/// <summary>
/// Sends a uncompressed method for processing by the graphics pipeline.
/// </summary>
/// <param name="gpuVa">GPU virtual address where the command word is located</param>
/// <param name="meth">Method to be processed</param>
private void Send(ulong gpuVa, int offset, int argument, int subChannel, bool isLastCall)
{
if (offset < 0x60)
{
_fifoClass.Write(offset * 4, argument);
}
else if (offset < 0xe00)
{
offset *= 4;
switch (subChannel)
{
case 0:
_3dClass.Write(offset, argument);
break;
case 1:
_computeClass.Write(offset, argument);
break;
case 2:
_i2mClass.Write(offset, argument);
break;
case 3:
_2dClass.Write(offset, argument);
break;
case 4:
_dmaClass.Write(offset, argument);
break;
}
}
else
{
IDeviceState state = subChannel switch
{
0 => _3dClass,
3 => _2dClass,
_ => null,
};
if (state != null)
{
int macroIndex = (offset >> 1) & MacroIndexMask;
if ((offset & 1) != 0)
{
_fifoClass.MmePushArgument(macroIndex, gpuVa, argument);
}
else
{
_fifoClass.MmeStart(macroIndex, argument);
}
if (isLastCall)
{
_fifoClass.CallMme(macroIndex, state);
_3dClass.PerformDeferredDraws();
}
}
}
}
/// <summary>
/// Writes data directly to the state of the specified class.
/// </summary>
/// <param name="classId">ID of the class to write the data into</param>
/// <param name="offset">State offset in bytes</param>
/// <param name="value">Value to be written</param>
public void Write(ClassId classId, int offset, int value)
{
switch (classId)
{
case ClassId.Threed:
_3dClass.Write(offset, value);
break;
case ClassId.Compute:
_computeClass.Write(offset, value);
break;
case ClassId.InlineToMemory:
_i2mClass.Write(offset, value);
break;
case ClassId.Twod:
_2dClass.Write(offset, value);
break;
case ClassId.Dma:
_dmaClass.Write(offset, value);
break;
case ClassId.GPFifo:
_fifoClass.Write(offset, value);
break;
}
}
/// <summary>
/// Sets the shadow ram control value of all sub-channels.
/// </summary>
/// <param name="control">New shadow ram control value</param>
public void SetShadowRamControl(int control)
{
_3dClass.SetShadowRamControl(control);
}
/// <summary>
/// Forces a full host state update by marking all state as modified,
/// and also requests all GPU resources in use to be rebound.
/// </summary>
public void ForceAllDirty()
{
_3dClass.ForceStateDirty();
_channel.BufferManager.Rebind();
_channel.TextureManager.Rebind();
}
/// <summary>
/// Perform any deferred draws.
/// </summary>
public void PerformDeferredDraws()
{
_3dClass.PerformDeferredDraws();
}
protected virtual void Dispose(bool disposing)
{
if (disposing)
{
_3dClass.Dispose();
}
}
public void Dispose()
{
Dispose(true);
GC.SuppressFinalize(this);
}
}
}
|