Issue metadata
Sign in to add a comment
|
hardware video decode broken on lumpy/butterfly |
||||||||||||||||||||||
Issue descriptionChrome Version : 60.0.3100.0 OS Version: 9554.1.0 URLs (if applicable) : https://imgur.com/gallery/2E3c3 What steps will reproduce the problem? 1. visit https://imgur.com/gallery/2E3c3 2. see a black screen 3. disable chrome://flags/#disable-accelerated-video-decode 4. log out & log in 5. visit URL again and now video decode works going to attach the mp4 directly that fails. downloading and trying to play locally also does not work. https://i.imgur.com/BjYfSUj.mp4 UserAgentString: Mozilla/5.0 (X11; CrOS x86_64 9554.1.0) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/60.0.3100.0 Safari/537.36
,
May 22 2017
Unable to repo issue on 60.0.3105.0 :9574.0.0 on daisy and one intel device. vapier@ Can you please provide device details and is it observed atfer new update. Also please submit feedback report using alt + shit +i and add bug number. Thanks!
,
May 23 2017
there is nothing newer than 60.0.3100.0 available for lumpy so i can't test that daisy is an arm platform with vastly different hardware than lumpy, and i suspect your "one intel device" is also not a lumpy (or sandybridge based at least). you need to get a comparable platform before trying to reproduce. i've posted feedback now.
,
May 24 2017
vapier@ Thanks for the device info. Observed issue on 9578.0.0 / 60.0.3105.0 on lumpy device. Not observed issue by disabling h/w decode from about:flags Feedback Id : 61464778137
,
May 24 2017
This video is working on 53.0.2785.154/8530.96.0 and 58.0.3029.140/9334.72.0. But not working on 59.0.3071.57/9460.42.0. https://bugs.chromium.org/p/chromium/issues/detail?id=724876 is also similar issue ?
,
May 24 2017
thanks, that bug sounds related, prob even the same. i'll let Paweł decide how he wants to manage these :).
,
May 24 2017
,
May 25 2017
this issue also effecting vimeo videos and local MP4 videos. Build used : 9460.50.0, 59.0.3071.71, device : lumpy
,
May 26 2017
Good build : 9434.0.0/59.0.3055.0 Bad build : 9435.0.0/59.0.3063.0 Chrome OS diff: https://crosland.corp.google.com/log/9434.0.0..9435.0.0 Chrome diff: https://chromium.googlesource.com/chromium/src/+log/59.0.3055.0..59.0.3063.0?pretty=fuller&n=10000
,
May 26 2017
none of those CrOS changes look relevant (applies to repos/boards not used by lumpy), or interesting (doesn't look video related). the Chrome pin/unpin are a little interesting, but if the Chrome bisect is accurate, there's plenty to go through there :).
,
May 26 2017
This issue also happening on butterfly.
,
May 26 2017
,
May 30 2017
,
May 30 2017
,
May 31 2017
I could not reproduce this on other platforms, and I was not able to secure a lumpy or butterfly locally here. I tested instead on a remote device on ToT and the decoding was working fine on this stream, the stream is also correct running it through the reference decoder. This could suggest an issue in rendering rather then decoding. Probably the easiest would be to bisect Chrome, given that we have a short regression range, but for that I will need to coordinate with someone that could actually see the screen, given that decoding seems to be fine and there seem to have been some video rendering-related changes in the bisect diff.
,
May 31 2017
i guess i wasn't explicit before ... playing the video seems to work in the sense that i hear audio and the progress bar moves steadily. but the window stays black. if need be, i could put my lumpy into dev mode and test out builds you have.
,
Jun 1 2017
Bisected to have been caused by crrev.com/517d0115d3c435383f79a0b730b6514a6cc97f44.
,
Jun 1 2017
dcastagna@: Looks like your intention in crrev.com/517d0115d3c435383f79a0b730b6514a6cc97f44 has been to use RGBA_RESOURCE everywhere but on Android. However, on lumpy, returning RGBA_RESOURCE instead of STREAM_TEXTURE_RESOURCE seems to be causing this rendering issue. video_frame->metadata()->IsTrue(media::VideoFrameMetadata::COPY_REQUIRED) is false on lumpy and before your CL we used STREAM_TEXTURE_RESOURCE there. Could you suggest what should we do here please? Thanks.
,
Jun 1 2017
Can this bug be caught be thumbnail testcase in VDA unittest?
,
Jun 1 2017
No, unfortunately this is an issue in rendering path in Chrome, which is not used in the codec unittest.
,
Jun 2 2017
I'm going to revert this CL for now, since we need to unblock beta.
,
Jun 2 2017
Revert in CQ: https://codereview.chromium.org/2920893003/.
,
Jun 2 2017
Revert is in. Requesting merge to 59 and 60 please.
,
Jun 2 2017
This bug requires manual review: We are only 3 days from stable. Please contact the milestone owner if you have questions. Owners: amineer@(Android), cmasso@(iOS), gkihumba@(ChromeOS), Abdul Syed@(Desktop) For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
,
Jun 2 2017
,
Jun 2 2017
,
Jun 2 2017
,
Jun 3 2017
Your change meets the bar and is auto-approved for M60. Please go ahead and merge the CL to branch 3112 manually. Please contact milestone owner if you have questions. Owners: amineer@(Android), cmasso@(iOS), josafat@(ChromeOS), bustamante@(Desktop) For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
,
Jun 5 2017
,
Jun 5 2017
,
Jun 7 2017
Verified on 9460.60.0, 59.0.3071.91. I'll close the issue after testing on M60.
,
Jun 14 2017
Verified on butterfly on 9592.20.0/60.0.3112.31
,
Jun 16 2017
|
|||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||
Comment 1 by marc...@chromium.org
, May 22 2017Status: Assigned (was: Unconfirmed)