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

Issue 647076 link

Starred by 4 users

Issue metadata

Status: WontFix
Owner:
Last visit > 30 days ago
Closed: Nov 2016
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Chrome
Pri: 2
Type: Bug



Sign in to add a comment

GPU crashes after stop GPU process from task manager

Project Member Reported by hsiangc@chromium.org, Sep 14 2016

Issue description

google Chrome	54.0.2840.24 (Official Build) dev   
Platform	8743.0.25 (Official Build) dev-channel 
JavaScript	V8 5.4.500.22
Flash	23.0.0.151 

What steps will reproduce the problem?
1. Play a HD Flash video from Hulu.com
2. Crash GPU process from task manager window.
3. Go to about:crashes page.

What is the expected output?
GPU should not crash

What do you see instead?
GPU crashes

Feedback:
Report ID: 21940071736

Crash ID: f126a3ee00000000

 
issue also see on HTML5 video from vimeo

Comment 2 by enne@chromium.org, Oct 7 2016

Mergedinto: 646643
Status: Duplicate (was: Untriaged)
From crash/ it looks like this is a browser crash, which indicates these two issues:
https://bugs.chromium.org/p/chromium/issues/detail?id=646643
https://bugs.chromium.org/p/chromium/issues/detail?id=621282

Duping into one of them.
Status: Untriaged (was: Duplicate)
Issue 646643 has OS=Windows, whereas this issue is for Chrome OS.  This issue has a specific cause of killing the GPU process manually via the task manager, which seems to be unrelated to issue 646643.  So the two issues should not be merged.

See issue 646960 for an explanation of why you should not merge issues just because they have the common crash signature "da39a3ee_5e6b4b0d_3255bfef_95601890_afd80709".
Labels: -Pri-2 Pri-1
Status: Assigned (was: Untriaged)
Labels: -ReleaseBlock-Stable Stability-Crash
Owner: posciak@chromium.org
Pawel, this seems video-related, can you look?
Sorry, is this crash a second crash, happening after the initial crash triggered from the task manager?
Labels: M-55
issue reproduce on M-55 8872.16.0/55.0.2883.20.
Crash ID: 71ad9f2b00000000
Could you please clarify per my question in #7? Thank you.
The crash is first crash after kill GPU process from task manager
So basically GPU process crashes twice? Once as expected, and then when trying to reload the page and play the video again, it unexpectedly crashes again? Thanks.
I just want to confirm this issue is always reproduce so after first crash,I reload the page and play the video, do kill GPU process and the crash happen again.
Just want to clarify this test scenario,when manually kill the GPU process from task manager, it should not have any crash at all.
It seems killing a process manually shouldn't result into a crash. 
For the exact same steps on a website/tab without a video playing, does the same behavior not occur?
yes no crash occurs if without a video playing

Comment 17 by enne@chromium.org, Nov 9 2016

posciak: ping on this p1 crash bug.
Labels: videoshortlist
Although I agree we should look into this, I'm not sure this should be a P1 bug, given that this is triggered by the user manually killing the GPU process first...
Labels: -videoshortlist
Owner: hsiangc@chromium.org
I could reproduce the crash on cyan 8987.0.0.
- Play a youtube video.
- Press "End process" on GPU Process in Task Manager.

Observed result:
Chrome showed "Chrome didn't shut down correctly." I didn't see anything in chrome://crashes though.

I disabled video acceleration in chrome://flags (Hardware-acclerated video decode). I still could reproduce this. So this doesn't look like a hardware decode issue.

- I could not reprouduce the crash on elm 8987.0.0.
- I don't think this is a P1 either because users need to manually kill GPU process.

hsiangc@ Can you check if you can reproduce after disabling video acceleration in chrome://flags (Hardware-acclerated video decode)? Thanks.



Owner: wuchengli@chromium.org
wuchengli@ issue no reproduce on both M-55 and M-56 now 
Labels: -Pri-1 Pri-2
Lowering priority as per discussion so far, please update if needed. @hsiangc, when you get a chance can you try a repro on stable? Want to see if the magical cure that affected 55 & 56 affected 54 as well. 
Owner: hsiangc@chromium.org
hsiangc@ Please try 54. If it doesn't repro in 54, we can close this.
Owner: wuchengli@chromium.org
issue still reproduce on M-54 8743.35.0/54.0.2840.101
Labels: -M-54
Status: WontFix (was: Assigned)
M54 is already stable. It's too late to fix this in M54. I'm closing this.

Sign in to add a comment