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

Issue 660452 link

Starred by 2 users

Issue metadata

Status: WontFix
Owner: ----
Closed: Jan 2017
Cc:
EstimatedDays: ----
NextAction: ----
OS: Windows , Mac
Pri: 1
Type: Bug-Regression



Sign in to add a comment

Removing Video elements from page does not release memory in Chrome's GPU Process

Reported by ross.and...@radn.org, Oct 28 2016

Issue description

Chrome Version       : Windows 54.0.2840.71 m
URLs (if applicable) : http://plnkr.co/edit/8zNSdWwjxx1X7vBqEB5g

What steps will reproduce the problem?
(1) Open Chrome's task manager
(2) Visit http://plnkr.co/edit/8zNSdWwjxx1X7vBqEB5g
(3) Run the Plunkr
(4) Click the Update button a few times, it will clear all <video> elements from the page and add them again
(5) Notice that the GPU Process memory allocation increases with each click
(6) GPU Process appears to eventually crash and restart causing all Chrome tabs to repaint.
(7) GPU process will release memory on tab close, tab reload, or navigation to another url

What is the expected result?

Expect that GPU Process memory will tend towards a steady level, not continue to increase until it crashes

What happens instead?

GPU Process memory allocation increases with each cycle of removing <video> elements and re-adding them until it appears to crash

 
Labels: Pri-1
Cc: kkaluri@chromium.org
Labels: -Type-Bug hasbisect-per-revision M-56 OS-Mac OS-Windows Type-Bug-Regression
Owner: x...@chromium.org
Status: Assigned (was: Unconfirmed)
Able to reproduce this issue on windows 10 and Mac 10.11.6 on chrome stable version 54.0.2840.71 
Issue is broken in M52. 

Note:
-----
Issue is not reproduced on Ubuntu 14.04 machine.

Bisect Info:
===========

Good build : 52.0.2739.0 ,  Revision Range(393996)
Bad build  : 52.0.2740.0 ,  Revision Range(394284)

After executing the per-revision-bisect script, i got the following CL's between good and bad build versions
===========================================
https://chromium.googlesource.com/chromium/src/+log/b74909cbced955d38dd10db3dbb0252d32a5f91b..54e747470a4de1c2437556581c3a1abd63e6a423

The suspecting Change Log is :
-----------
https://chromium.googlesource.com/chromium/src/+/0d83958df92063615a84019f689985c74da54f8b

From the above CL suspecting the below change
---------------------------
 https://codereview.chromium.org/1984953005


xdai@- Could you please look into this issue, if it's related to your change?  if not could you please help us to reassign this issue to the right owner.

Comment 3 by hdodda@chromium.org, Nov 22 2016

Issue is still reproduced on chrome latest stable M54 #54.0.2840.98 on mac os 10.11.6 .

@xdai -- could you please look into this .

Thanks!
Cc: tkonch...@chromium.org
Labels: -M-56 M-57
Still able to reproduce the issue on mac 10.11.6 chrome version 57.0.2935.0

xdai@, Could you please take a look

Comment 5 by x...@chromium.org, Nov 28 2016

Owner: ----
Sorry for late response. Just came back from a long vacation. 
It should not be related to my change. My CL is just a revert of a bad CL, so it should not introduce new issues as reported here.

Comment 6 by ajha@chromium.org, Jan 20 2017

Cc: ajha@chromium.org
Labels: Needs-Feedback
Status: Unconfirmed (was: Assigned)
Somehow I was unable to encounter the crash on Windows-10 on the latest chrome stable version: 55.0.2883.87.

ross.anderson@: Could you please check the same on the latest stable and confirm if the crash is seen there. Please attach any crash id from chrome://crashes if this is generate.

Thanks in advance!
Hey there - I can't repro it on Windows 10 with 55.0.2883.87 either.  GPU Process memory no longer increases out of control - looks like it's being released correctly now.

Comment 8 by ajha@chromium.org, Jan 20 2017

Labels: -Needs-Feedback
Status: WontFix (was: Unconfirmed)
Thanks for the update. Closing this as per C#7 for now. Feel free to file a new issue with crash id if this is encountered again.

Sign in to add a comment