New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 604945 link

Starred by 20 users

Issue metadata

Status: Verified
Owner:
Closed: May 2016
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Linux
Pri: 2
Type: Bug



Sign in to add a comment

HTML5 Video not reproducing on Chrome

Reported by gonzalop...@gmail.com, Apr 19 2016

Issue description

UserAgent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/50.0.2661.75 Safari/537.36

Example URL:
https://www.youtube.com/watch?v=A3NE7z5_Oxs

Steps to reproduce the problem:
1. Any HTML5 Video doesn't play, it does play the audio though
2. 
3. 

What is the expected behavior?

What went wrong?
It is the hardware accelerator feature, when is turned on only the video doesn't play while the audio does.

Did this work before? Yes Before applying all the new Debian updates

Is it a problem with Flash or HTML5? HTML5

Does this work in other browsers? Yes 

Chrome version: 50.0.2661.75  Channel: stable
OS Version: Debian 8
Flash Version: Shockwave Flash 21.0 r0

When I turn off the hardware accelerator feature the video works
 
chromeDump.txt
8.7 KB View Download
I am seeing the same behaviour with a similar configuration.
Labels: Needs-Feedback
I couldn't repro this issue on the provided YouTube link. Can you confirm which OS are you using? it can't be Linux since Hardware accelerate decoding is not enabled in Linux. 
Not OP here (they say they're using Debian 8)

Mine is;

Chrome version Chrome version: 50.0.2661.75  Channel: stable
OS Version: Slackware 14.1 x86_64
Flash Version: Not installed

For me it is any Youtube or Vimeo HTML5 video. 

As odd as it sounds, unchecking Hardware Acceleration does make it work, and checking it again causes the bug. 

This is using a radeon 5450 GPU with open-source drivers. 
james@, can you provide logs in chrome://media-internals of both check and uncheck "Hardware acceleration" option?
Yes; I'll have to do it Monday as it's at the office.
Cc: j.iso...@samsung.com

Comment 7 by meta...@gmail.com, May 3 2016

I've got the same problem, no HTML5 video works; Chrome on Debian 8. I'm running Chrome over an X11 session over SSH. It used to work, until a couple of releases ago. Info dump:

Errors appear in console throughout attempted playback. Sample error:

[101:103:0503/102304:ERROR:video_layer_impl.cc(369)] Not implemented reached in virtual void cc::VideoLayerImpl::AppendQuads(cc::RenderPass *, cc::AppendQuadsData *)

media-internals player properties:

render_id: 12
player_id: 0
pipeline_state: kStopping
render_id: 17
player_id: 27
pipeline_state: kPlaying
event: PAUSE
url: blob:https%3A//www.youtube.com/2c3d118c-6337-4c2a-8f6c-f1c77c000a19
duration: 1988.641
found_video_stream: true
video_codec_name: vp9
found_audio_stream: true
audio_codec_name: opus
audio_dds: false
audio_decoder: OpusAudioDecoder
video_dds: false
video_decoder: VpxVideoDecoder

Google Chrome	50.0.2661.94 (Official Build) (64-bit)
Revision	0c319c380220abe3daf0277e40a2d0b22a7c4550-refs/branch-heads/2661@{#645}
OS	Linux

Linux 3.16.0-4-amd64 #1 SMP Debian 3.16.7-ckt20-1+deb8u3 (2016-01-17) x86_64 GNU/Linux

Comment 8 by meta...@gmail.com, May 3 2016

And after turning off hardware video acceleration, when it works again:

render_id: 2
player_id: 0
pipeline_state: kInitDemuxer
event: WEBMEDIAPLAYER_CREATED
url: blob:chrome-extension%3A//fahmaaghhglfmonjliepjlchgpgfmobi/cb21fae2-0ac7-4da6-8573-2269f125eb75
render_id: 5
player_id: 27
pipeline_state: kPlaying
event: PLAY
url: blob:https%3A//www.youtube.com/d96561f2-0e9e-43c5-bbc0-7aa27ff741fe
duration: 1988.641
found_video_stream: true
video_codec_name: vp9
found_audio_stream: true
audio_codec_name: opus
audio_dds: false
audio_decoder: OpusAudioDecoder
video_dds: false
video_decoder: VpxVideoDecoder
seek_target: 2

The video_layer_impl.cc errors also stop showing up in the console.
I diffed mine (between setting acceleration on and off) and couldn't see any difference apart from the time stamps; maybe I need to go back and do that again. 

Additionally, can confirm going to 50.0.2661.94 doesn't change anything. 

 
Also, I have a nearly identical software build (Slackware 14.1) on a different laptop with an Intel HD3000 GPU and I do not experience this bug, so it is *something* in the graphics stack. 
Cc: kbr@chromium.org zmo@chromium.org piman@chromium.org
Hi, isn't same issue as https://bugs.chromium.org/p/chromium/issues/detail?id=606282 ?

Could you provide output of chrome://gpu . Thx

Also please try one of the 2 solutions from https://codereview.chromium.org/1968933003/patch/1/10001 .

Comment 12 by piman@chromium.org, May 11 2016

 Issue 606282  has been merged into this issue.

Comment 13 by piman@chromium.org, May 11 2016

Cc: danakj@chromium.org dalecur...@chromium.org

Comment 14 by piman@chromium.org, May 11 2016

Sounds like https://code.google.com/p/chromium/codesearch#chromium/src/cc/layers/video_layer_impl.cc&q=video_layer_impl.cc&sq=package:chromium&type=cs&l=374 which would come from here: https://code.google.com/p/chromium/codesearch#chromium/src/cc/resources/video_resource_updater.cc&sq=package:chromium&type=cs&l=93

It sounds like we're getting a VideoFrame of a type we don't expect.

Are we trying to use accelerated decoding when it's not available?

Comment 15 by srack...@gmail.com, May 11 2016

@ j.iso...@samsung.com

chrome://gpu attached.

Ubuntu MATE 14.04 64-bit (Virtual box 5.0.16 VM) running on Win10 64-bit host.
gpu.html
48.5 KB View Download
Thx it confirms that this happens when using the GL_RENDERER Gallium 0.4 on llvmpipe driver. The CL I pointed in comment #11 should fix the problem. Please have a go.

Antoine, I investigated today and this is because video_frame->HasTextures() is true but context_provider_ is null.

See https://code.google.com/p/chromium/codesearch#chromium/src/cc/resources/video_resource_updater.cc&l=259

and
https://code.google.com/p/chromium/codesearch#chromium/src/cc/resources/video_resource_updater.cc&l=675 .

This situation happens because gpu_memory_buffer_pool_ is not null, see https://code.google.com/p/chromium/codesearch#chromium/src/content/renderer/media/media_stream_video_renderer_sink.cc&l=35 whereas gpu composited is off (because black listed).

Well at least this is the sequence that on my setup leads to "Not implemented reached in virtual void cc::VideoLayerImpl::AppendQuads(cc::RenderPass *, cc::AppendQuadsData *)" but maybe this can happen in other situations.
Cheers, Julien



Comment 17 by piman@chromium.org, May 11 2016

Cc: reve...@chromium.org dcasta...@chromium.org
Cc: sande...@chromium.org
 crbug.com/597406  might be related to this bug. https://bugs.chromium.org/p/chromium/issues/detail?id=606282#c2 describes similar steps to what described in https://bugs.chromium.org/p/chromium/issues/detail?id=597406#c10.










Comment 19 by piman@chromium.org, May 12 2016

 Issue 597406  has been merged into this issue.

Comment 20 by piman@chromium.org, May 12 2016

Owner: dcasta...@chromium.org
->dcastagna who has a CL for this - https://codereview.chromium.org/1972783002/

Comment 21 by tin...@google.com, May 12 2016

Cc: gov...@chromium.org
+govind@ as this is a potential Stable RB, as per dcastagna@ this is a big issue with videos on Linux on Stable and it's in today's 50 stable refresh push

Labels: Merge-Request-51 Merge-Request-50
Tina, seems like this bug is originally reported on Very first release of M50 Stable (50.0.2661.75),follow up releases 50.0.2661.86, 50.0.2661.94 and 50.0.2661.102 (today's release). 

Also is this ONLY on Linux?

Reply to #22, Before we approve merge to M50, Could you please confirm whether this bug is baked/verified in Canary and safe to merge?

Cc: tinazh@chromium.org
This issue affects only linux users with software compositing and GPUs not blacklisted.
I'm marking it as ReleaseBlock-Stable.
If we respin, we can include this patch, but we don't have to respin just for this.

crrev.com/1972783002 has been verified locally but hasn't landed yet, it is still going trough the CQ.
Labels: ReleaseBlock-Stable
Status: Started (was: Unconfirmed)

Comment 27 Deleted

Reply to #25:
Ok, thank you  dcastagna@. We will surely take it if we respin M50. Please update the bug once fix is landed, baked/verified in canary.
Project Member

Comment 29 by bugdroid1@chromium.org, May 12 2016

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/a82ed96acbda1350db7095bfdf944766a38084d2

commit a82ed96acbda1350db7095bfdf944766a38084d2
Author: dcastagna <dcastagna@chromium.org>
Date: Thu May 12 02:52:47 2016

Disable GMB VideoFrames if GPU compositing is disabled.

When GPU compositing is disabled, GpuMemoryBuffer backed VideoFrames
won't be displayed correctly.

This is most likely the cause of  crbug.com/604945 

BUG= 604945 

Review-Url: https://codereview.chromium.org/1972783002
Cr-Commit-Position: refs/heads/master@{#393160}

[modify] https://crrev.com/a82ed96acbda1350db7095bfdf944766a38084d2/content/renderer/render_thread_impl.cc

 Issue 611096  has been merged into this issue.

Comment 31 by tin...@google.com, May 13 2016

Labels: -Merge-Request-50 Merge-Review-50 Hotlist-Merge-Review
[Automated comment] Request affecting a post-stable build (M50), manual review required.

Comment 32 by tin...@google.com, May 13 2016

Labels: -Merge-Request-51 Merge-Approved-51 Hotlist-Merge-Approved
Your change meets the bar and is auto-approved for M51 (branch: 2704)

Comment 33 by tin...@google.com, May 13 2016

Labels: -Merge-Request-50 Merge-Review-50 Hotlist-Merge-Review
[Automated comment] Request affecting a post-stable build (M50), manual review required.
Project Member

Comment 34 by bugdroid1@chromium.org, May 13 2016

Labels: -merge-approved-51 merge-merged-2704
The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/b413ce5086f08e0aed334713a03fb885722b96c9

commit b413ce5086f08e0aed334713a03fb885722b96c9
Author: dcastagna <dcastagna@chromium.org>
Date: Fri May 13 17:47:40 2016

Disable GMB VideoFrames if GPU compositing is disabled.

When GPU compositing is disabled, GpuMemoryBuffer backed VideoFrames
won't be displayed correctly.

This is most likely the cause of  crbug.com/604945 

BUG= 604945 

Review-Url: https://codereview.chromium.org/1972783002
Cr-Commit-Position: refs/heads/master@{#393160}
(cherry picked from commit a82ed96acbda1350db7095bfdf944766a38084d2)

NOTRY=true
NOPRESUBMIT=true

Review-Url: https://codereview.chromium.org/1973853002
Cr-Commit-Position: refs/branch-heads/2704@{#539}
Cr-Branched-From: 6e53600def8f60d8c632fadc70d7c1939ccea347-refs/heads/master@{#386251}

[modify] https://crrev.com/b413ce5086f08e0aed334713a03fb885722b96c9/content/renderer/render_thread_impl.cc

I just discovered same issue (No video from HTML5 video sources - i.e. black video-port with audio playing fine)

- Problem only appeared after one of the last two Chrome Package updates (had not tried video on the '94-1 version; problem discovered after installing '102-1 version of Chrome.

Upgraded the following packages:
google-chrome-stable (50.0.2661.94-1) to 50.0.2661.102-1

Noted that problem DOES also occur on version 50.0.2661.94 of Chromium, which is currently still installed; Problem in Chromium is also resolved by disabling hardware acceleration from chrome://settings/ .

- problem also seen when playing from Incognito Window 

- 1st discovered on Facebook, then Youtube. 
- Disabling GPU acceleration via "google-chrome --disable-gpu" or from "chrome://settings/ - Advanced" resolves problem. (workaround found from this thread: https://productforums.google.com/forum/#!topic/chrome/9xjvBiS-CdI )

- Firefox 46.0.1 does not have the problem 
- also noted that the Chrome "Magic Actions" extension can be set to force Flash for video type and this does resolve problem for YouTube only (FB still has same symptoms)

System Configuration:
AMD Athlon(tm) II X2 250 Processor (64-bit;@ 1800MHz)
8GB Ram
running Debian GNU/Linux 8.4 (jessie)

Video Driver Info:

# lspci | grep VGA

	01:05.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] RS880 [Radeon HD 4200]

# find /dev -group video
# 

# glxinfo | grep -i vendor

server glx vendor string: SGI
client glx vendor string: Mesa Project and SGI





gpu.pdf
249 KB Download
FB_Example.png
151 KB View Download
YouTube_HTML5_example.png
222 KB View Download
I can no longer reproduce this on Beta (51.0.2704.54). The issue appears to have been fixed.
Status: Verified (was: Started)
Could someone please let me know how to fix the issue of viewing html 5 in youtube. Thank you.


Labels: -Hotlist-Merge-Review -Hotlist-Merge-Approved -Merge-Review-50

Sign in to add a comment