gecko-dev/dom/media/fmp4/wmf/WMFVideoMFTManager.cpp

520 lines
16 KiB
C++

/* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
/* vim:set ts=2 sw=2 sts=2 et cindent: */
/* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
#include <algorithm>
#include "WMFVideoMFTManager.h"
#include "MediaDecoderReader.h"
#include "WMFUtils.h"
#include "ImageContainer.h"
#include "VideoUtils.h"
#include "DXVA2Manager.h"
#include "nsThreadUtils.h"
#include "Layers.h"
#include "mozilla/layers/LayersTypes.h"
#include "mp4_demuxer/AnnexB.h"
#include "mp4_demuxer/DecoderData.h"
#include "prlog.h"
#include "gfx2DGlue.h"
#include "gfxWindowsPlatform.h"
#include "IMFYCbCrImage.h"
#ifdef PR_LOGGING
PRLogModuleInfo* GetDemuxerLog();
#define LOG(...) PR_LOG(GetDemuxerLog(), PR_LOG_DEBUG, (__VA_ARGS__))
#else
#define LOG(...)
#endif
using mozilla::gfx::ToIntRect;
using mozilla::layers::Image;
using mozilla::layers::IMFYCbCrImage;
using mozilla::layers::LayerManager;
using mozilla::layers::LayersBackend;
const GUID MFVideoFormat_VP80 =
{
0x30385056,
0x0000,
0x0010,
{0x80, 0x00, 0x00, 0xaa, 0x00, 0x38, 0x9b, 0x71}
};
const GUID MFVideoFormat_VP90 =
{
0x30395056,
0x0000,
0x0010,
{0x80, 0x00, 0x00, 0xaa, 0x00, 0x38, 0x9b, 0x71}
};
const CLSID CLSID_WebmMfVp8Dec =
{
0x451e3cb7,
0x2622,
0x4ba5,
{0x8e, 0x1d, 0x44, 0xb3, 0xc4, 0x1d, 0x09, 0x24}
};
const CLSID CLSID_WebmMfVp9Dec =
{
0x7ab4bd2,
0x1979,
0x4fcd,
{0xa6, 0x97, 0xdf, 0x9a, 0xd1, 0x5b, 0x34, 0xfe}
};
namespace mozilla {
WMFVideoMFTManager::WMFVideoMFTManager(
const mp4_demuxer::VideoDecoderConfig& aConfig,
mozilla::layers::LayersBackend aLayersBackend,
mozilla::layers::ImageContainer* aImageContainer,
bool aDXVAEnabled)
: mImageContainer(aImageContainer)
, mDXVAEnabled(aDXVAEnabled)
, mLayersBackend(aLayersBackend)
// mVideoStride, mVideoWidth, mVideoHeight, mUseHwAccel are initialized in
// Init().
{
NS_ASSERTION(!NS_IsMainThread(), "Should not be on main thread.");
MOZ_ASSERT(mImageContainer);
MOZ_COUNT_CTOR(WMFVideoMFTManager);
// Need additional checks/params to check vp8/vp9
if (aConfig.mime_type.EqualsLiteral("video/mp4") ||
aConfig.mime_type.EqualsLiteral("video/avc")) {
mStreamType = H264;
} else if (aConfig.mime_type.EqualsLiteral("video/webm; codecs=vp8")) {
mStreamType = VP8;
} else if (aConfig.mime_type.EqualsLiteral("video/webm; codecs=vp9")) {
mStreamType = VP9;
} else {
mStreamType = Unknown;
}
}
WMFVideoMFTManager::~WMFVideoMFTManager()
{
MOZ_COUNT_DTOR(WMFVideoMFTManager);
// Ensure DXVA/D3D9 related objects are released on the main thread.
if (mDXVA2Manager) {
DeleteOnMainThread(mDXVA2Manager);
}
}
const GUID&
WMFVideoMFTManager::GetMFTGUID()
{
MOZ_ASSERT(mStreamType != Unknown);
switch (mStreamType) {
case H264: return CLSID_CMSH264DecoderMFT;
case VP8: return CLSID_WebmMfVp8Dec;
case VP9: return CLSID_WebmMfVp9Dec;
default: return GUID_NULL;
};
}
const GUID&
WMFVideoMFTManager::GetMediaSubtypeGUID()
{
MOZ_ASSERT(mStreamType != Unknown);
switch (mStreamType) {
case H264: return MFVideoFormat_H264;
case VP8: return MFVideoFormat_VP80;
case VP9: return MFVideoFormat_VP90;
default: return GUID_NULL;
};
}
class CreateDXVAManagerEvent : public nsRunnable {
public:
NS_IMETHOD Run() {
NS_ASSERTION(NS_IsMainThread(), "Must be on main thread.");
mDXVA2Manager = DXVA2Manager::Create();
return NS_OK;
}
nsAutoPtr<DXVA2Manager> mDXVA2Manager;
};
bool
WMFVideoMFTManager::InitializeDXVA()
{
MOZ_ASSERT(!mDXVA2Manager);
// If we use DXVA but aren't running with a D3D layer manager then the
// readback of decoded video frames from GPU to CPU memory grinds painting
// to a halt, and makes playback performance *worse*.
if (!mDXVAEnabled ||
(mLayersBackend != LayersBackend::LAYERS_D3D9 &&
mLayersBackend != LayersBackend::LAYERS_D3D10 &&
mLayersBackend != LayersBackend::LAYERS_D3D11)) {
return false;
}
// The DXVA manager must be created on the main thread.
nsRefPtr<CreateDXVAManagerEvent> event(new CreateDXVAManagerEvent());
NS_DispatchToMainThread(event, NS_DISPATCH_SYNC);
mDXVA2Manager = event->mDXVA2Manager;
return mDXVA2Manager != nullptr;
}
TemporaryRef<MFTDecoder>
WMFVideoMFTManager::Init()
{
mUseHwAccel = false; // default value; changed if D3D setup succeeds.
bool useDxva = InitializeDXVA();
RefPtr<MFTDecoder> decoder(new MFTDecoder());
HRESULT hr = decoder->Create(GetMFTGUID());
NS_ENSURE_TRUE(SUCCEEDED(hr), nullptr);
if (useDxva) {
RefPtr<IMFAttributes> attr(decoder->GetAttributes());
UINT32 aware = 0;
if (attr) {
attr->GetUINT32(MF_SA_D3D_AWARE, &aware);
}
if (aware) {
// TODO: Test if I need this anywhere... Maybe on Vista?
//hr = attr->SetUINT32(CODECAPI_AVDecVideoAcceleration_H264, TRUE);
//NS_ENSURE_TRUE(SUCCEEDED(hr), hr);
MOZ_ASSERT(mDXVA2Manager);
ULONG_PTR manager = ULONG_PTR(mDXVA2Manager->GetDXVADeviceManager());
hr = decoder->SendMFTMessage(MFT_MESSAGE_SET_D3D_MANAGER, manager);
if (SUCCEEDED(hr)) {
mUseHwAccel = true;
}
}
}
// Setup the input/output media types.
RefPtr<IMFMediaType> inputType;
hr = wmf::MFCreateMediaType(byRef(inputType));
NS_ENSURE_TRUE(SUCCEEDED(hr), nullptr);
hr = inputType->SetGUID(MF_MT_MAJOR_TYPE, MFMediaType_Video);
NS_ENSURE_TRUE(SUCCEEDED(hr), nullptr);
hr = inputType->SetGUID(MF_MT_SUBTYPE, GetMediaSubtypeGUID());
NS_ENSURE_TRUE(SUCCEEDED(hr), nullptr);
hr = inputType->SetUINT32(MF_MT_INTERLACE_MODE, MFVideoInterlace_MixedInterlaceOrProgressive);
NS_ENSURE_TRUE(SUCCEEDED(hr), nullptr);
RefPtr<IMFMediaType> outputType;
hr = wmf::MFCreateMediaType(byRef(outputType));
NS_ENSURE_TRUE(SUCCEEDED(hr), nullptr);
hr = outputType->SetGUID(MF_MT_MAJOR_TYPE, MFMediaType_Video);
NS_ENSURE_TRUE(SUCCEEDED(hr), nullptr);
GUID outputSubType = mUseHwAccel ? MFVideoFormat_NV12 : MFVideoFormat_YV12;
hr = outputType->SetGUID(MF_MT_SUBTYPE, outputSubType);
NS_ENSURE_TRUE(SUCCEEDED(hr), nullptr);
hr = decoder->SetMediaTypes(inputType, outputType);
NS_ENSURE_TRUE(SUCCEEDED(hr), nullptr);
mDecoder = decoder;
LOG("Video Decoder initialized, Using DXVA: %s", (mUseHwAccel ? "Yes" : "No"));
// Just in case ConfigureVideoFrameGeometry() does not set these
mVideoInfo = VideoInfo();
mVideoStride = 0;
mVideoWidth = 0;
mVideoHeight = 0;
mPictureRegion.SetEmpty();
return decoder.forget();
}
HRESULT
WMFVideoMFTManager::Input(mp4_demuxer::MP4Sample* aSample)
{
if (!mDecoder) {
// This can happen during shutdown.
return E_FAIL;
}
if (mStreamType != VP8 && mStreamType != VP9) {
// We must prepare samples in AVC Annex B.
if (!mp4_demuxer::AnnexB::ConvertSampleToAnnexB(aSample)) {
return E_FAIL;
}
}
// Forward sample data to the decoder.
const uint8_t* data = reinterpret_cast<const uint8_t*>(aSample->data);
uint32_t length = aSample->size;
return mDecoder->Input(data, length, aSample->composition_timestamp);
}
HRESULT
WMFVideoMFTManager::ConfigureVideoFrameGeometry()
{
RefPtr<IMFMediaType> mediaType;
HRESULT hr = mDecoder->GetOutputMediaType(mediaType);
NS_ENSURE_TRUE(SUCCEEDED(hr), hr);
// Verify that the video subtype is what we expect it to be.
// When using hardware acceleration/DXVA2 the video format should
// be NV12, which is DXVA2's preferred format. For software decoding
// we use YV12, as that's easier for us to stick into our rendering
// pipeline than NV12. NV12 has interleaved UV samples, whereas YV12
// is a planar format.
GUID videoFormat;
hr = mediaType->GetGUID(MF_MT_SUBTYPE, &videoFormat);
NS_ENSURE_TRUE(videoFormat == MFVideoFormat_NV12 || !mUseHwAccel, E_FAIL);
NS_ENSURE_TRUE(videoFormat == MFVideoFormat_YV12 || mUseHwAccel, E_FAIL);
nsIntRect pictureRegion;
hr = GetPictureRegion(mediaType, pictureRegion);
NS_ENSURE_TRUE(SUCCEEDED(hr), hr);
UINT32 width = 0, height = 0;
hr = MFGetAttributeSize(mediaType, MF_MT_FRAME_SIZE, &width, &height);
NS_ENSURE_TRUE(SUCCEEDED(hr), hr);
uint32_t aspectNum = 0, aspectDenom = 0;
hr = MFGetAttributeRatio(mediaType,
MF_MT_PIXEL_ASPECT_RATIO,
&aspectNum,
&aspectDenom);
NS_ENSURE_TRUE(SUCCEEDED(hr), hr);
// Calculate and validate the picture region and frame dimensions after
// scaling by the pixel aspect ratio.
nsIntSize frameSize = nsIntSize(width, height);
nsIntSize displaySize = nsIntSize(pictureRegion.width, pictureRegion.height);
ScaleDisplayByAspectRatio(displaySize, float(aspectNum) / float(aspectDenom));
if (!IsValidVideoRegion(frameSize, pictureRegion, displaySize)) {
// Video track's frame sizes will overflow. Ignore the video track.
return E_FAIL;
}
// Success! Save state.
mVideoInfo.mDisplay = displaySize;
mVideoInfo.mHasVideo = true;
GetDefaultStride(mediaType, &mVideoStride);
mVideoWidth = width;
mVideoHeight = height;
mPictureRegion = pictureRegion;
LOG("WMFVideoMFTManager frame geometry frame=(%u,%u) stride=%u picture=(%d, %d, %d, %d) display=(%d,%d) PAR=%d:%d",
width, height,
mVideoStride,
mPictureRegion.x, mPictureRegion.y, mPictureRegion.width, mPictureRegion.height,
displaySize.width, displaySize.height,
aspectNum, aspectDenom);
return S_OK;
}
HRESULT
WMFVideoMFTManager::CreateBasicVideoFrame(IMFSample* aSample,
int64_t aStreamOffset,
VideoData** aOutVideoData)
{
NS_ENSURE_TRUE(aSample, E_POINTER);
NS_ENSURE_TRUE(aOutVideoData, E_POINTER);
*aOutVideoData = nullptr;
HRESULT hr;
RefPtr<IMFMediaBuffer> buffer;
// Must convert to contiguous buffer to use IMD2DBuffer interface.
hr = aSample->ConvertToContiguousBuffer(byRef(buffer));
NS_ENSURE_TRUE(SUCCEEDED(hr), hr);
// Try and use the IMF2DBuffer interface if available, otherwise fallback
// to the IMFMediaBuffer interface. Apparently IMF2DBuffer is more efficient,
// but only some systems (Windows 8?) support it.
BYTE* data = nullptr;
LONG stride = 0;
RefPtr<IMF2DBuffer> twoDBuffer;
hr = buffer->QueryInterface(static_cast<IMF2DBuffer**>(byRef(twoDBuffer)));
if (SUCCEEDED(hr)) {
hr = twoDBuffer->Lock2D(&data, &stride);
NS_ENSURE_TRUE(SUCCEEDED(hr), hr);
} else {
hr = buffer->Lock(&data, nullptr, nullptr);
NS_ENSURE_TRUE(SUCCEEDED(hr), hr);
stride = mVideoStride;
}
// YV12, planar format: [YYYY....][VVVV....][UUUU....]
// i.e., Y, then V, then U.
VideoData::YCbCrBuffer b;
// Y (Y') plane
b.mPlanes[0].mData = data;
b.mPlanes[0].mStride = stride;
b.mPlanes[0].mHeight = mVideoHeight;
b.mPlanes[0].mWidth = mVideoWidth;
b.mPlanes[0].mOffset = 0;
b.mPlanes[0].mSkip = 0;
// The V and U planes are stored 16-row-aligned, so we need to add padding
// to the row heights to ensure the Y'CbCr planes are referenced properly.
uint32_t padding = 0;
if (mVideoHeight % 16 != 0) {
padding = 16 - (mVideoHeight % 16);
}
uint32_t y_size = stride * (mVideoHeight + padding);
uint32_t v_size = stride * (mVideoHeight + padding) / 4;
uint32_t halfStride = (stride + 1) / 2;
uint32_t halfHeight = (mVideoHeight + 1) / 2;
uint32_t halfWidth = (mVideoWidth + 1) / 2;
// U plane (Cb)
b.mPlanes[1].mData = data + y_size + v_size;
b.mPlanes[1].mStride = halfStride;
b.mPlanes[1].mHeight = halfHeight;
b.mPlanes[1].mWidth = halfWidth;
b.mPlanes[1].mOffset = 0;
b.mPlanes[1].mSkip = 0;
// V plane (Cr)
b.mPlanes[2].mData = data + y_size;
b.mPlanes[2].mStride = halfStride;
b.mPlanes[2].mHeight = halfHeight;
b.mPlanes[2].mWidth = halfWidth;
b.mPlanes[2].mOffset = 0;
b.mPlanes[2].mSkip = 0;
Microseconds pts = GetSampleTime(aSample);
Microseconds duration = GetSampleDuration(aSample);
nsRefPtr<layers::PlanarYCbCrImage> image =
new IMFYCbCrImage(buffer, twoDBuffer);
VideoData::SetVideoDataToImage(image,
mVideoInfo,
b,
ToIntRect(mPictureRegion),
false);
nsRefPtr<VideoData> v =
VideoData::CreateFromImage(mVideoInfo,
mImageContainer,
aStreamOffset,
std::max(0LL, pts),
duration,
image.forget(),
false,
-1,
ToIntRect(mPictureRegion));
v.forget(aOutVideoData);
return S_OK;
}
HRESULT
WMFVideoMFTManager::CreateD3DVideoFrame(IMFSample* aSample,
int64_t aStreamOffset,
VideoData** aOutVideoData)
{
NS_ENSURE_TRUE(aSample, E_POINTER);
NS_ENSURE_TRUE(aOutVideoData, E_POINTER);
NS_ENSURE_TRUE(mDXVA2Manager, E_ABORT);
NS_ENSURE_TRUE(mUseHwAccel, E_ABORT);
*aOutVideoData = nullptr;
HRESULT hr;
nsRefPtr<Image> image;
hr = mDXVA2Manager->CopyToImage(aSample,
mPictureRegion,
mImageContainer,
getter_AddRefs(image));
NS_ENSURE_TRUE(SUCCEEDED(hr), hr);
NS_ENSURE_TRUE(image, E_FAIL);
Microseconds pts = GetSampleTime(aSample);
Microseconds duration = GetSampleDuration(aSample);
nsRefPtr<VideoData> v = VideoData::CreateFromImage(mVideoInfo,
mImageContainer,
aStreamOffset,
pts,
duration,
image.forget(),
false,
-1,
ToIntRect(mPictureRegion));
NS_ENSURE_TRUE(v, E_FAIL);
v.forget(aOutVideoData);
return S_OK;
}
// Blocks until decoded sample is produced by the deoder.
HRESULT
WMFVideoMFTManager::Output(int64_t aStreamOffset,
nsRefPtr<MediaData>& aOutData)
{
RefPtr<IMFSample> sample;
HRESULT hr;
aOutData = nullptr;
// Loop until we decode a sample, or an unexpected error that we can't
// handle occurs.
while (true) {
hr = mDecoder->Output(&sample);
if (hr == MF_E_TRANSFORM_NEED_MORE_INPUT) {
return MF_E_TRANSFORM_NEED_MORE_INPUT;
}
if (hr == MF_E_TRANSFORM_STREAM_CHANGE) {
// Video stream output type change. Probably a geometric apperature
// change. Reconfigure the video geometry, so that we output the
// correct size frames.
MOZ_ASSERT(!sample);
hr = ConfigureVideoFrameGeometry();
NS_ENSURE_TRUE(SUCCEEDED(hr), hr);
// Loop back and try decoding again...
continue;
}
if (SUCCEEDED(hr)) {
break;
}
// Else unexpected error, assert, and bail.
NS_WARNING("WMFVideoMFTManager::Output() unexpected error");
return E_FAIL;
}
nsRefPtr<VideoData> frame;
if (mUseHwAccel) {
hr = CreateD3DVideoFrame(sample, aStreamOffset, getter_AddRefs(frame));
} else {
hr = CreateBasicVideoFrame(sample, aStreamOffset, getter_AddRefs(frame));
}
// Frame should be non null only when we succeeded.
MOZ_ASSERT((frame != nullptr) == SUCCEEDED(hr));
NS_ENSURE_TRUE(SUCCEEDED(hr), hr);
NS_ENSURE_TRUE(frame, E_FAIL);
aOutData = frame;
return S_OK;
}
void
WMFVideoMFTManager::Shutdown()
{
mDecoder = nullptr;
DeleteOnMainThread(mDXVA2Manager);
}
bool
WMFVideoMFTManager::IsHardwareAccelerated() const
{
return mDecoder && mUseHwAccel;
}
} // namespace mozilla