Bug 183912 – Use libwebrtc ObjectiveC H264 encoder and decoder

• Component: WebRTC
• Status: RESOLVED FIXED

[youenn fablet]

This allows sticking closer to libwebrtc

[youenn fablet]

Created attachment 336297 [details]
Patch

[youenn fablet]

Created attachment 336306 [details]
Patch

[youenn fablet]

Created attachment 336314 [details]
Patch

[youenn fablet]

Created attachment 336318 [details]
Patch

[Eric Carlson]

Comment on attachment 336318 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=336318&action=review

> Source/ThirdParty/libwebrtc/Source/webrtc/sdk/WebKit/WebKitUtilities.mm:35
> +#if! defined(WEBRTC_IOS)

Nit: the space is on the wrong side of the "!"

> Source/ThirdParty/libwebrtc/Source/webrtc/sdk/objc/Framework/Classes/VideoToolbox/RTCVideoDecoderH264.mm:194
> +  Block_release(_callback);

Can _callback ever be NULL?

> Source/ThirdParty/libwebrtc/Source/webrtc/sdk/objc/Framework/Classes/VideoToolbox/RTCVideoEncoderH264.mm:485
> +  Block_release(_callback);

Ditto.

> Source/ThirdParty/libwebrtc/WebKit/0001-Adapting-libwebrtc-H264-codec.patch:6
> +From 0130a91f901af2c60c81c32565c531141bbd2b84 Mon Sep 17 00:00:00 2001
> +From: Youenn Fablet <[e-mail]>
> +Date: Thu, 22 Mar 2018 12:51:04 -0700
> +Subject: [PATCH] Adapting libwebrtc H264 codec
> +
> +---

Oops :-)

> Source/WebCore/platform/mediastream/mac/RealtimeIncomingVideoSourceCocoa.mm:83
> +        if (!m_blackFrame || m_blackFrameWidth != frame.width() || m_blackFrameHeight != frame.height()) {
> +            CVPixelBufferRef pixelBuffer = nullptr;
> +            auto status = CVPixelBufferCreate(kCFAllocatorDefault, frame.width(), frame.height(), kCVPixelFormatType_420YpCbCr8Planar, nullptr, &pixelBuffer);
> +            ASSERT_UNUSED(status, status == noErr);
> +
> +            m_blackFrame = adoptCF(pixelBuffer);
> +            m_blackFrameWidth = frame.width();
> +            m_blackFrameHeight = frame.height();
> +
> +            status = CVPixelBufferLockBaseAddress(pixelBuffer, 0);
> +            ASSERT(status == noErr);
> +            void* data = CVPixelBufferGetBaseAddress(pixelBuffer);
> +            size_t yLength = frame.width() * frame.height();
> +            memset(data, 0, yLength);
> +            memset(static_cast<uint8_t*>(data) + yLength, 128, yLength / 2);
> +
> +            status = CVPixelBufferUnlockBaseAddress(pixelBuffer, 0);
> +            ASSERT(!status);
> +        }
> +        return m_blackFrame.get();

Nit: I know this isn't new, but it would be cleaner if it was in a blackFrame method, e.g. "if (muted())  return blackFrame();"

[EWS Watchlist]

Comment on attachment 336318 [details]
Patch

Attachment 336318 [details] did not pass mac-ews (mac):
Output: http://webkit-queues.webkit.org/results/7068032

New failing tests:
media/modern-media-controls/tracks-panel/tracks-panel-position-and-size.html

[EWS Watchlist]

Created attachment 336326 [details]
Archive of layout-test-results from ews102 for mac-sierra

The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: ews102  Port: mac-sierra  Platform: Mac OS X 10.12.6

[youenn fablet]

Created attachment 336328 [details]
Patch for landing

[youenn fablet]

Created attachment 336333 [details]
Patch

[youenn fablet]

(In reply to youenn fablet from comment #9)
> Created attachment 336333 [details]
> Patch

Added Block_release in dealloc after discussing with Jer.

[WebKit Commit Bot]

Comment on attachment 336333 [details]
Patch

Clearing flags on attachment: 336333

Committed r229876: <https://trac.webkit.org/changeset/229876>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Radar WebKit Bug Importer]

<rdar://problem/38776207>

[WebKit Commit Bot]

Re-opened since this is blocked by bug 183929

[youenn fablet]

Created attachment 336382 [details]
Patch

[WebKit Commit Bot]

Comment on attachment 336382 [details]
Patch

Clearing flags on attachment: 336382

Committed r229908: <https://trac.webkit.org/changeset/229908>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.