Bug 1495025 - P5. Add Windows P010 and P016 support for software decoder r=cpearce

As we do not have an IMF nor D3D11 NV12 image, we always require a full copy of the data that will deinterleave the chroma channels.

Depends on D7316

Differential Revision: https://phabricator.services.mozilla.com/D7318

dom/media/MediaData.cpp

@@ -323,14 +323,18 @@ VideoData::CreateAndCopyData(const VideoInfo& aInfo,
 #if XP_WIN
   // We disable this code path on Windows version earlier of Windows 8 due to
   // intermittent crashes with old drivers. See bug 1405110.
-  if (IsWin8OrLater() && !XRE_IsParentProcess() &&
+  // D3D11YCbCrImage can only handle YCbCr images using 3 non-interleaved planes
-      aAllocator && aAllocator->SupportsD3D11()) {
+  // non-zero mSkip value indicates that one of the plane would be interleaved.
+  if (IsWin8OrLater() && !XRE_IsParentProcess() && aAllocator &&
+      aAllocator->SupportsD3D11() && aBuffer.mPlanes[0].mSkip == 0 &&
+      aBuffer.mPlanes[1].mSkip == 0 && aBuffer.mPlanes[2].mSkip == 0) {
     RefPtr<layers::D3D11YCbCrImage> d3d11Image = new layers::D3D11YCbCrImage();
     PlanarYCbCrData data = ConstructPlanarYCbCrData(aInfo, aBuffer, aPicture);
     if (d3d11Image->SetData(layers::ImageBridgeChild::GetSingleton()
-                            ? layers::ImageBridgeChild::GetSingleton().get()
+                              ? layers::ImageBridgeChild::GetSingleton().get()
-                            : aAllocator,
+                              : aAllocator,
-                            aContainer, data)) {
+                            aContainer,
+                            data)) {
       v->mImage = d3d11Image;
       return v.forget();
     }

dom/media/platforms/wmf/MFTDecoder.cpp

@@ -88,7 +88,6 @@ MFTDecoder::SetMediaTypes(IMFMediaType* aInputType,
                           std::function<HRESULT(IMFMediaType*)>&& aCallback)
 {
   MOZ_ASSERT(mscom::IsCurrentThreadMTA());
-  mOutputType = aOutputType;
 
   // Set the input type to the one the caller gave us...
   HRESULT hr = mDecoder->SetInputType(0, aInputType, 0);
@@ -132,10 +131,7 @@ MFTDecoder::FindDecoderOutputType(bool aMatchAllAttributes)
   MOZ_ASSERT(mscom::IsCurrentThreadMTA());
   MOZ_ASSERT(mOutputType, "SetDecoderTypes must have been called once");
 
-  GUID currentSubtype = {0};
+  return FindDecoderOutputTypeWithSubtype(mOutputSubType, aMatchAllAttributes);
-  HRESULT hr = mOutputType->GetGUID(MF_MT_SUBTYPE, &currentSubtype);
-  NS_ENSURE_TRUE(SUCCEEDED(hr), hr);
-  return FindDecoderOutputTypeWithSubtype(currentSubtype, aMatchAllAttributes);
 }
 
 HRESULT
@@ -191,6 +187,7 @@ MFTDecoder::SetDecoderOutputType(
                                             MFT_OUTPUT_STREAM_PROVIDES_SAMPLES);
 
       mOutputType = outputType;
+      mOutputSubType = outSubtype;
 
       return S_OK;
     }

dom/media/platforms/wmf/MFTDecoder.h

@@ -49,6 +49,7 @@ public:
   // Retrieves the media type being output. This may not be valid until
   //  the first sample is decoded.
   HRESULT GetOutputMediaType(RefPtr<IMFMediaType>& aMediaType);
+  const GUID& GetOutputMediaSubType() const { return mOutputSubType; }
 
   // Submits data into the MFT for processing.
   //
@@ -105,6 +106,7 @@ private:
   RefPtr<IMFTransform> mDecoder;
 
   RefPtr<IMFMediaType> mOutputType;
+  GUID mOutputSubType;
 
   // True if the IMFTransform allocates the samples that it returns.
   bool mMFTProvidesOutputSamples = false;

dom/media/platforms/wmf/WMFVideoMFTManager.cpp

@@ -899,8 +899,18 @@ WMFVideoMFTManager::CreateBasicVideoFrame(IMFSample* aSample,
     stride = mVideoStride;
   }
 
-  // YV12, planar format: [YYYY....][VVVV....][UUUU....]
+  const GUID& subType = mDecoder->GetOutputMediaSubType();
+  MOZ_DIAGNOSTIC_ASSERT(subType == MFVideoFormat_YV12 ||
+                        subType == MFVideoFormat_P010 ||
+                        subType == MFVideoFormat_P016);
+  const gfx::ColorDepth colorDepth = subType == MFVideoFormat_YV12
+                                       ? gfx::ColorDepth::COLOR_8
+                                       : gfx::ColorDepth::COLOR_16;
+
+  // YV12, planar format (3 planes): [YYYY....][VVVV....][UUUU....]
   // i.e., Y, then V, then U.
+  // P010, P016 planar format (2 planes) [YYYY....][UVUV...]
+  // See https://docs.microsoft.com/en-us/windows/desktop/medfound/10-bit-and-16-bit-yuv-video-formats
   VideoData::YCbCrBuffer b;
 
   uint32_t videoWidth = mImageSize.width;
@@ -922,24 +932,43 @@ WMFVideoMFTManager::CreateBasicVideoFrame(IMFSample* aSample,
   uint32_t halfHeight = (videoHeight + 1) / 2;
   uint32_t halfWidth = (videoWidth + 1) / 2;
 
-  // U plane (Cb)
+  if (subType == MFVideoFormat_YV12) {
-  b.mPlanes[1].mData = data + y_size + v_size;
+    // U plane (Cb)
-  b.mPlanes[1].mStride = halfStride;
+    b.mPlanes[1].mData = data + y_size + v_size;
-  b.mPlanes[1].mHeight = halfHeight;
+    b.mPlanes[1].mStride = halfStride;
-  b.mPlanes[1].mWidth = halfWidth;
+    b.mPlanes[1].mHeight = halfHeight;
-  b.mPlanes[1].mOffset = 0;
+    b.mPlanes[1].mWidth = halfWidth;
-  b.mPlanes[1].mSkip = 0;
+    b.mPlanes[1].mOffset = 0;
+    b.mPlanes[1].mSkip = 0;
 
-  // V plane (Cr)
+    // V plane (Cr)
-  b.mPlanes[2].mData = data + y_size;
+    b.mPlanes[2].mData = data + y_size;
-  b.mPlanes[2].mStride = halfStride;
+    b.mPlanes[2].mStride = halfStride;
-  b.mPlanes[2].mHeight = halfHeight;
+    b.mPlanes[2].mHeight = halfHeight;
-  b.mPlanes[2].mWidth = halfWidth;
+    b.mPlanes[2].mWidth = halfWidth;
-  b.mPlanes[2].mOffset = 0;
+    b.mPlanes[2].mOffset = 0;
-  b.mPlanes[2].mSkip = 0;
+    b.mPlanes[2].mSkip = 0;
+  } else {
+    // U plane (Cb)
+    b.mPlanes[1].mData = data + y_size;
+    b.mPlanes[1].mStride = stride;
+    b.mPlanes[1].mHeight = halfHeight;
+    b.mPlanes[1].mWidth = halfWidth;
+    b.mPlanes[1].mOffset = 0;
+    b.mPlanes[1].mSkip = 1;
+
+    // V plane (Cr)
+    b.mPlanes[2].mData = data + y_size + sizeof(short);
+    b.mPlanes[2].mStride = stride;
+    b.mPlanes[2].mHeight = halfHeight;
+    b.mPlanes[2].mWidth = halfWidth;
+    b.mPlanes[2].mOffset = 0;
+    b.mPlanes[2].mSkip = 1;
+  }
 
   // YuvColorSpace
   b.mYUVColorSpace = mYUVColorSpace;
+  b.mColorDepth = colorDepth;
 
   TimeUnit pts = GetSampleTime(aSample);
   NS_ENSURE_TRUE(pts.IsValid(), E_FAIL);
@@ -948,7 +977,8 @@ WMFVideoMFTManager::CreateBasicVideoFrame(IMFSample* aSample,
   gfx::IntRect pictureRegion =
     mVideoInfo.ScaledImageRect(videoWidth, videoHeight);
 
-  if (!mKnowsCompositor || !mKnowsCompositor->SupportsD3D11() || !mIMFUsable) {
+  if (colorDepth != gfx::ColorDepth::COLOR_8 || !mKnowsCompositor ||
+      !mKnowsCompositor->SupportsD3D11() || !mIMFUsable) {
     RefPtr<VideoData> v =
       VideoData::CreateAndCopyData(mVideoInfo,
                                    mImageContainer,
@@ -1065,15 +1095,14 @@ WMFVideoMFTManager::Output(int64_t aStreamOffset,
 
       // Attempt to find an appropriate OutputType, trying in order:
       // if HW accelerated: NV12, P010, P016
-      // if SW: YV12
+      // if SW: YV12, P010, P016
       if (FAILED((hr = (mDecoder->FindDecoderOutputTypeWithSubtype(
                     mUseHwAccel ? MFVideoFormat_NV12 : MFVideoFormat_YV12,
                     false)))) &&
-          (!mUseHwAccel ||
+          FAILED((hr = mDecoder->FindDecoderOutputTypeWithSubtype(
-           (FAILED((hr = mDecoder->FindDecoderOutputTypeWithSubtype(
+                    MFVideoFormat_P010, false))) &&
-                      MFVideoFormat_P010, false))) &&
+          FAILED((hr = mDecoder->FindDecoderOutputTypeWithSubtype(
-            FAILED((hr = mDecoder->FindDecoderOutputTypeWithSubtype(
+                    MFVideoFormat_P016, false)))) {
-                      MFVideoFormat_P016, false)))))) {
         LOG("No suitable output format found");
         return hr;
       }

gfx/layers/client/TextureClient.cpp

@@ -1895,13 +1895,6 @@ MappedYCbCrChannelData::CopyInto(MappedYCbCrChannelData& aDst)
   if (bytesPerPixel == 1) {
     copyData(aDst.data, aDst, data, *this);
   } else if (bytesPerPixel == 2) {
-    if (skip != 0) {
-      // The skip value definition doesn't specify if it's in bytes, or in
-      // "pixels". We will assume the later. There are currently no decoders
-      // returning HDR content with a skip value different than zero anyway.
-      NS_WARNING("skip value non zero for HDR content, please verify code "
-                 "(see bug 1421187)");
-    }
     copyData(reinterpret_cast<uint16_t*>(aDst.data),
              aDst,
              reinterpret_cast<uint16_t*>(data),