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
gpu / ipc / common / dxgi_helpers.cc [blame]
// Copyright 2021 The Chromium Authors
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
#ifdef UNSAFE_BUFFERS_BUILD
// TODO(crbug.com/40285824): Remove this and convert code to safer constructs.
#pragma allow_unsafe_buffers
#endif
#include "gpu/ipc/common/dxgi_helpers.h"
#include "base/check.h"
#include "base/debug/crash_logging.h"
#include "base/debug/dump_without_crashing.h"
#include "base/logging.h"
#include "base/metrics/histogram_functions.h"
#include "base/time/time.h"
#include "third_party/libyuv/include/libyuv/planar_functions.h"
namespace {
constexpr char kStagingTextureLabel[] = "DxgiGmb_Map_StagingTexture";
Microsoft::WRL::ComPtr<ID3D11Texture2D> CreateStagingTexture(
ID3D11Device* d3d11_device,
D3D11_TEXTURE2D_DESC input_desc) {
D3D11_TEXTURE2D_DESC staging_desc = {};
staging_desc.Width = input_desc.Width;
staging_desc.Height = input_desc.Height;
staging_desc.Format = input_desc.Format;
staging_desc.MipLevels = 1;
staging_desc.ArraySize = 1;
staging_desc.SampleDesc.Count = 1;
staging_desc.Usage = D3D11_USAGE_STAGING;
staging_desc.CPUAccessFlags = D3D11_CPU_ACCESS_READ;
Microsoft::WRL::ComPtr<ID3D11Texture2D> staging_texture;
HRESULT hr =
d3d11_device->CreateTexture2D(&staging_desc, nullptr, &staging_texture);
if (FAILED(hr)) {
DLOG(ERROR) << "Failed to create staging texture. hr=" << std::hex << hr;
return nullptr;
}
// Add debug label to the long lived texture.
staging_texture->SetPrivateData(WKPDID_D3DDebugObjectName,
strlen(kStagingTextureLabel),
kStagingTextureLabel);
return staging_texture;
}
} // namespace
namespace gpu {
D3D11ScopedTextureUnmap::D3D11ScopedTextureUnmap(
Microsoft::WRL::ComPtr<ID3D11DeviceContext> context,
Microsoft::WRL::ComPtr<ID3D11Texture2D> texture)
: context_(std::move(context)), texture_(std::move(texture)) {}
D3D11ScopedTextureUnmap::~D3D11ScopedTextureUnmap() {
context_->Unmap(texture_.Get(), 0);
}
DXGIScopedReleaseKeyedMutex::DXGIScopedReleaseKeyedMutex(
Microsoft::WRL::ComPtr<IDXGIKeyedMutex> keyed_mutex,
UINT64 key)
: keyed_mutex_(std::move(keyed_mutex)), key_(key) {
DCHECK(keyed_mutex_);
}
DXGIScopedReleaseKeyedMutex::~DXGIScopedReleaseKeyedMutex() {
HRESULT hr = keyed_mutex_->ReleaseSync(key_);
DCHECK(SUCCEEDED(hr));
}
bool CopyDXGIBufferToShMem(
HANDLE dxgi_handle,
base::span<uint8_t> shared_memory,
ID3D11Device* d3d11_device,
Microsoft::WRL::ComPtr<ID3D11Texture2D>* staging_texture) {
DCHECK(d3d11_device);
uint8_t* dest_buffer = shared_memory.data();
size_t dst_buffer_size = shared_memory.size_bytes();
Microsoft::WRL::ComPtr<ID3D11Device1> device1;
HRESULT hr = d3d11_device->QueryInterface(IID_PPV_ARGS(&device1));
if (FAILED(hr)) {
DLOG(ERROR) << "Failed to open D3D11_1 device. hr=" << std::hex << hr;
return false;
}
Microsoft::WRL::ComPtr<ID3D11Texture2D> texture;
// Open texture on device using shared handle
hr = device1->OpenSharedResource1(dxgi_handle, IID_PPV_ARGS(&texture));
if (FAILED(hr)) {
DLOG(ERROR) << "Failed to open shared texture. hr=" << std::hex << hr;
return false;
}
return CopyD3D11TexToMem(texture.Get(), dest_buffer, dst_buffer_size,
d3d11_device, staging_texture);
}
bool CopyD3D11TexToMem(
ID3D11Texture2D* src_texture,
uint8_t* dst_buffer,
size_t buffer_size,
ID3D11Device* d3d11_device,
Microsoft::WRL::ComPtr<ID3D11Texture2D>* staging_texture) {
DCHECK(d3d11_device);
DCHECK(staging_texture);
DCHECK(dst_buffer);
DCHECK(src_texture);
D3D11_TEXTURE2D_DESC texture_desc = {};
src_texture->GetDesc(&texture_desc);
if (texture_desc.Format != DXGI_FORMAT_NV12) {
DLOG(ERROR) << "Can't copy non-NV12 texture. format="
<< static_cast<int>(texture_desc.Format);
return false;
}
size_t copy_size = texture_desc.Height * texture_desc.Width * 3 / 2;
if (buffer_size < copy_size) {
DLOG(ERROR) << "Invalid buffer size for copy.";
return false;
}
// The texture isn't accessible for CPU reads, thus a staging texture is used.
bool create_staging_texture = !*staging_texture;
if (*staging_texture) {
D3D11_TEXTURE2D_DESC staging_texture_desc;
(*staging_texture)->GetDesc(&staging_texture_desc);
create_staging_texture =
(staging_texture_desc.Width != texture_desc.Width ||
staging_texture_desc.Height != texture_desc.Height ||
staging_texture_desc.Format != texture_desc.Format);
}
if (create_staging_texture) {
*staging_texture = CreateStagingTexture(d3d11_device, texture_desc);
if (!*staging_texture)
return false;
}
Microsoft::WRL::ComPtr<ID3D11DeviceContext> device_context;
d3d11_device->GetImmediateContext(&device_context);
HRESULT hr = S_OK;
if (texture_desc.MiscFlags & D3D11_RESOURCE_MISC_SHARED_KEYEDMUTEX) {
Microsoft::WRL::ComPtr<IDXGIKeyedMutex> keyed_mutex;
hr = src_texture->QueryInterface(IID_PPV_ARGS(&keyed_mutex));
if (FAILED(hr)) {
DLOG(ERROR) << "Failed to get keyed mutex. Error msg: "
<< logging::SystemErrorCodeToString(hr);
return false;
}
// Key equal to 0 is also used by the producer. Therefore, this keyed
// mutex acts purely as a regular mutex.
// 300ms is long enough to get the mutex in 99.999% of cases. Yet we
// don't want to stall the callee indefinitely if the mutex is held by
// e.g. GpuMain thread while it's blocked on driver waiting for shader
// compilation.
// It's better to drop a frame in this case.
hr = keyed_mutex->AcquireSync(0, 300);
// Can't check FAILED(hr), because AcquireSync may return e.g. WAIT_TIMEOUT
// value.
if (hr != S_OK) {
DLOG(ERROR) << "Failed to acquire keyed mutex. Error msg: "
<< logging::SystemErrorCodeToString(hr);
return false;
}
DXGIScopedReleaseKeyedMutex release_keyed_mutex(keyed_mutex, 0);
device_context->CopySubresourceRegion(staging_texture->Get(), 0, 0, 0, 0,
src_texture, 0, nullptr);
} else {
device_context->CopySubresourceRegion(staging_texture->Get(), 0, 0, 0, 0,
src_texture, 0, nullptr);
}
D3D11_MAPPED_SUBRESOURCE mapped_resource = {};
hr = device_context->Map(staging_texture->Get(), 0, D3D11_MAP_READ, 0,
&mapped_resource);
if (FAILED(hr)) {
DLOG(ERROR) << "Failed to map texture for read. Error msg: "
<< logging::SystemErrorCodeToString(hr);
return false;
}
D3D11ScopedTextureUnmap scoped_unmap(device_context, *staging_texture);
const uint8_t* source_buffer = static_cast<uint8_t*>(mapped_resource.pData);
const uint32_t source_stride = mapped_resource.RowPitch;
const uint32_t dest_stride = texture_desc.Width;
return libyuv::NV12Copy(source_buffer, source_stride,
source_buffer + texture_desc.Height * source_stride,
source_stride, dst_buffer, dest_stride,
dst_buffer + texture_desc.Height * dest_stride,
dest_stride, texture_desc.Width,
texture_desc.Height) == 0;
}
GPU_EXPORT bool CopyShMemToDXGIBuffer(base::span<uint8_t> shared_memory,
HANDLE dxgi_handle,
ID3D11Device* d3d11_device) {
CHECK(d3d11_device);
uint8_t* src_buffer = shared_memory.data();
size_t src_buffer_size = shared_memory.size_bytes();
Microsoft::WRL::ComPtr<ID3D11Device1> device1;
HRESULT hr = d3d11_device->QueryInterface(IID_PPV_ARGS(&device1));
if (FAILED(hr)) {
DLOG(ERROR) << "Failed to open D3D11_1 device. hr=" << std::hex << hr;
return false;
}
Microsoft::WRL::ComPtr<ID3D11Texture2D> texture;
// Open texture on device using shared handle
hr = device1->OpenSharedResource1(dxgi_handle, IID_PPV_ARGS(&texture));
if (FAILED(hr)) {
DLOG(ERROR) << "Failed to open shared texture. hr=" << std::hex << hr;
return false;
}
return CopyMemToD3D11Tex(src_buffer, src_buffer_size, texture.Get(),
d3d11_device);
}
GPU_EXPORT bool CopyMemToD3D11Tex(uint8_t* src_buffer,
size_t buffer_size,
ID3D11Texture2D* output_texture,
ID3D11Device* d3d11_device) {
CHECK(d3d11_device);
CHECK(src_buffer);
CHECK(output_texture);
D3D11_TEXTURE2D_DESC texture_desc = {};
output_texture->GetDesc(&texture_desc);
if (texture_desc.Format != DXGI_FORMAT_NV12) {
DLOG(ERROR) << "Can't copy non-NV12 texture. format="
<< static_cast<int>(texture_desc.Format);
return false;
}
size_t copy_size = texture_desc.Height * texture_desc.Width * 3 / 2;
if (buffer_size < copy_size) {
DLOG(ERROR) << "Invalid buffer size for copy.";
return false;
}
Microsoft::WRL::ComPtr<ID3D11DeviceContext> device_context;
d3d11_device->GetImmediateContext(&device_context);
HRESULT hr = S_OK;
if (texture_desc.MiscFlags & D3D11_RESOURCE_MISC_SHARED_KEYEDMUTEX) {
Microsoft::WRL::ComPtr<IDXGIKeyedMutex> keyed_mutex;
hr = output_texture->QueryInterface(IID_PPV_ARGS(&keyed_mutex));
if (FAILED(hr)) {
DLOG(ERROR) << "Failed to get keyed mutex. Error msg: "
<< logging::SystemErrorCodeToString(hr);
return false;
}
// Key equal to 0 is also used by the producer. Therefore, this keyed
// mutex acts purely as a regular mutex.
hr = keyed_mutex->AcquireSync(0, INFINITE);
// Can't check FAILED(hr), because AcquireSync may return e.g. WAIT_TIMEOUT
// value.
if (hr != S_OK) {
DLOG(ERROR) << "Failed to acquire keyed mutex. Error msg: "
<< logging::SystemErrorCodeToString(hr);
return false;
}
DXGIScopedReleaseKeyedMutex release_keyed_mutex(keyed_mutex, 0);
device_context->UpdateSubresource(output_texture, 0, nullptr, src_buffer,
texture_desc.Width, copy_size);
} else {
device_context->UpdateSubresource(output_texture, 0, nullptr, src_buffer,
texture_desc.Width, copy_size);
}
return true;
}
} // namespace gpu