New issue
Advanced search Search tips

Issue 591135 link

Starred by 2 users

Issue metadata

Status: WontFix
Owner:
Closed: Mar 2016
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Chrome
Pri: 1
Type: Bug



Sign in to add a comment

Chromebook has old copy of a window's contents "stuck" on the background

Project Member Reported by w...@chromium.org, Mar 1 2016

Issue description

Version: 49.0.2623.59
OS: ChromeOS

What steps will reproduce the problem?
No idea; I had switched the device to Tablet mode and then unlocked it, so that may be related?
System has a single window with two tabs in it.

What is the expected output? What do you see instead?

When I un-maximized my window, and moved it around, there was another, older, copy of it underneath - it looked like another window was still open, but it turns out to be just stale on-screen imagery.

Only seen this once, but I don't often use Tablet-mode, so it may still be a new regression.
 

Comment 1 by w...@chromium.org, Mar 1 2016

Screenshot 2016-03-01 at 11.34.10.png
29.9 KB View Download

Comment 2 by w...@chromium.org, Mar 1 2016

Interestingly, I have two user profiles signed-in, and this "stuck" window appears on both profiles.

Comment 3 by w...@chromium.org, Mar 3 2016

Components: -UI>Shell>WindowManager -Internals>Compositing Internals>Views>Desktop
Owner: tapted@chromium.org
Status: Assigned (was: Untriaged)
Sneakily assigning this to tapted@, since he grabbed 591535, and they seem likely related!
Cc: tapted@chromium.org
Owner: osh...@chromium.org
Oh weird. I don't think they're related - the stuff I regressed in  Issue 591535  was pretty specific to tab loading spinners.

There is some pretty crazy stuff that happens when you [un]maximize a window on CrOS though. Cleaning up the layers involved is the responsibility of the ash::CrossFadeObserver - https://code.google.com/p/chromium/codesearch#chromium/src/ash/wm/window_animations.cc&q=CrossFadeObserver&sq=package:chromium&l=265

I encountered it around the m49 timeframe in https://codereview.chromium.org/1474993003/ but I'm preeeetty sure that only fixed a bug.

I think oshima is the expert here and will have a much better clue than me what this might be.

Comment 5 by w...@chromium.org, Mar 5 2016

Labels: -Pri-3 Pri-1
FWIW I just signed-in to ChromeOS 49.0.2623.75 w/ my corp account, then signed-in a second account, and I now have the window from my corp account appearing on _both_ accounts' workspaces - the other account's windows only appear on one.

Bumping priority since this seems a more fundamental regression.

Comment 6 by osh...@chromium.org, Mar 15 2016

Status: Started (was: Assigned)

Comment 7 by osh...@chromium.org, Mar 15 2016

I think the  issue #5  is different issue. wez@, can you file a separate bug with more detailed description and log file?

Comment 8 by w...@chromium.org, Mar 16 2016

Re #7: If I'm able to repro the weird window behaviour again then I will do, yes!

Comment 9 by osh...@chromium.org, Mar 16 2016

Just wanted to confirm. Is this window that appeared on both accounts are functional on both desktop?

Comment 10 by w...@chromium.org, Mar 16 2016

This bug is about a window that was non-functional (and had no corresponding processes listed in Task Manager) - basically just a layer that was somehow "stuck" in the compositor tree.
So am I correct that stale window image stuck on background even when you switched the desktop, which has different background image?

If so, there is no need to file a separate bug.

Comment 12 by w...@chromium.org, Mar 17 2016

I'm sorry, I assumed that comment #9 was asking about this bug, rather than my comment #5 - in #5 the window appeared on both desktops and appeared to be interactive on both.
Cc: afakhry@chromium.org osh...@chromium.org
 Issue 577920  has been merged into this issue.
Sorry, I wasn't clear either. If #5 happened again, please file a separate bug.

If this (#1) bug happened again, could you try changing the resolution using ctrl/shift/+- and see if that clears the image, or sticks?
I couldn't reproduce this on ToT. I'll check if this is fixed in 50.
Status: WontFix (was: Started)
I tested this on 50.0.2661.39 and could not reproduce the issue. I'll close this, but please reopen if you have this issue on m50 beta.

Sign in to add a comment