Web content is non-responsive after exiting VR mode |
||||||||
Issue descriptionChrome Version: 61.0.3155.0 OS: Android What steps will reproduce the problem? (1) Navigate to https://www.google.com and go into VR mode (2) Exit VR mode by taking the phone out of the headset What is the expected result? The webpage continues to work. What happens instead? The tab seems hung; can't scroll or click on any links.
,
Jul 14 2017
Can't seem to repro on ToT. There were some big-ish bugs yesterday that made testing chrome on android tricky in general. Those fixes have landed and hopefully have resolved this issue as well.
,
Jul 14 2017
I can still repro on 61.0.3156.4. I have Chrome Home (bottom omnibox enabled), do you? Wonder if that could be related.
,
Jul 17 2017
Also cannot repro on ToT, with Chrome Home on google.com. When you said you exited VR by taking your phone out of the headset, what did you mean by that? Taking your phone out of the headset won't actually do anything unless you were being shown the DOFF (Device Off) screen that asks you to take your headset off. Is that what you were seeing, or did you use the 'x' or system back button to exit VR?
,
Jul 17 2017
I believe I took the phone out of the headset and then used the 'x' to exit VR. Is it possible there's some other flag or field trial that might be relevant? I think a VC might be the best way to demo this. I'm traveling this week but left my pixel and headset at my desk; one of my coworkers might be able to grab them and demo this over VC. If that sounds good, please send me an email with who I should invite and I'll set something up.
,
Jul 19 2017
When this occurred, was it just input being ignored (ie - you couldn't click on things or scroll), or was layout/compositing frozen too (changing orientation also didn't respond)? We do freeze the compositor in WebVR mode, but that shouldn't occur here. We also have some wait loops in the render process main thread for WebVR, but again shouldn't be hitting any issues here. I wouldn't be surprised if some sequence of input events that is missing paired events, (like begin scroll with unpaired end scroll) could get into an inconsistent state and cause some input to be ignored. I've so far been unable to repro, but can hit a DCHECK in input handling on debug builds. estark, was this very easy to reproduce for you (like 100% of the time?)
,
Jul 19 2017
I don't think I tried changing orientation. +meacer who might be able to try to repro this while he is trying to repro another VR bug this week. (I'm on the road without a pixel or headset.) This reproduced pretty easily for me, I think it was like 3 or 4 times in a row.
,
Jul 24 2017
I can still repro reliably on 61.0.3163.0. Changing orientation works.
,
Jul 27 2017
,
Oct 11 2017
Moving to M64, though sadly I've still not been able to reproduce the issue. cc'ing asimjour who has been looking at our input routing and may have some ideas if it's something like what billorr@ suggested. Otherwise, I'm wondering if this could be thermal throttling?
,
Oct 12 2017
This wouldn't be thermal throttling.
,
Oct 13 2017
Amir, could you please take a look in that case to see if it's related to any input issues you're aware of?
,
Nov 3 2017
I can't reproduce this. dbbrooks@ could you please take a look and see if you can reproduce this one.
,
Nov 3 2017
I cannot reproduce this either. Chrome: 64.0.3256.0 VRCore: 1.10.172754103
,
Nov 6 2017
Since we can't reproduce this issue, I'm closing it. We can reopen the bug in case that we see a similar problem in the future. |
||||||||
►
Sign in to add a comment |
||||||||
Comment 1 by est...@chromium.org
, Jul 14 2017