Rendering engine freezes
Reported by
wali...@gmail.com,
Mar 21 2017
|
||||||
Issue descriptionUserAgent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_12_3) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/56.0.2924.87 Safari/537.36 Steps to reproduce the problem: 1. visit https://mappingfestival2017.herokuapp.com/beta 2. enter borisisathug 3. navigate in the website, going back and forth several times Specifically, it seems to happen when going back to 'program' What is the expected behavior? The page should keep rendering What went wrong? After navigatin in the site for a while, the tab stops rendering. You can still click around, open the inspector, navigate, but you see a frozen tab. Reloading doesn't start the rendering again, you need to open a new tab. Did this work before? N/A Chrome version: 56.0.2924.87 Channel: n/a OS Version: OS X 10.12.3 Flash Version: old issue: https://bugs.chromium.org/p/chromium/issues/detail?id=228948 I've seen this behavior happen on various computers, although only Mac OS X. Also, the page navigation is handled by turbolinks(https://github.com/turbolinks/turbolinks ), with their cache mechanism disabled.
,
Mar 23 2017
Thanks for the report! By "freeze" or "frozen tab", what do you mean specifically? "You can still click around, open the inspector, navigate" indicates that Chrome is still active. I tried the page but have not yet seen freeze-ish behavior. Please help us finding the problem. Other than you can click and navigate, what you *cannot* do? e.g. Can you scroll, or see hover effect? The web page is quite complicated and probably hard to minimize the reproduction - so we would like to reproduce the case to see what is the internal state.
,
Mar 23 2017
,
Mar 25 2017
Hi! By Frozen, I mean that the rendering stops entirely. I can still click on links, which triggers a visit and is handled as expected(ajax loaded, history pushed and location bar updated), I can still scroll(and see the logs related to scrolling in the console). However, the page remains in the exact same state, ie no pixel change. Resizing the window also doesn't adapt the content. Some screenshots are white, which happened after switching to another tab and coming back to it. Some screengrabs: https://cl.ly/2I2T090V1B19 https://cl.ly/2C1s310S2O2z https://cl.ly/0Q2F3q2H3T2q I cannot reproduce systematically, but navigating the site seems to eventually trigger it(again, have seen on various computers). I suspect(on no particular basis): - turbolinks, which does the ajax handling, and has a caching system(mostly used to handle ~history.back) - some DOM elements stored in memory between the pages, and it gets too much at some point, particularly in terms of rendering contexts(the /program page has many elements, each in their own z-plane) - the DOM element that is removed on page navigation has a z-index of 60px when it comes out. Could this be source of the error? Once I get a frozen tab, what could i check to see if there is a problem? Thanks for your input!
,
Mar 25 2017
Thank you for providing more feedback. Adding requester "kochi@chromium.org" to the cc list and removing "Needs-Feedback" label. For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
,
Mar 27 2017
,
Mar 27 2017
Thanks for more info. Blink has an event loop that handles events, animation, etc., and if it gets in a completely blocked state, even mouse event cannot happen - so in your case, only the screen rendering (in Blink's terminology it may be "painting") stopped and event loop seems working. I watched the videos, but haven't got any clue on why it stopped rendering. (seeing lots of errors/warnings on the console, but it is not an issue?) ~5 minutes of navigation didn't get a repro. (I tried on Macbook Pro FYI, as "I've seen this behavior happen on various computers, although only Mac OS X." in the original comment.) As we haven't reproduced the problem locally, we have little to make progress on this - except for spending more time on playing around the page to hope it reproduces. If you could minimize the problem (minimal set of HTML, JS and CSS)- it would be most helpful. But if the size (e.g. number of elements used in the page) matters to reproduce the issue, minimizing would be very hard. Anyway, without reproduction it's too hard to identify the issue :-(
,
Mar 27 2017
,
May 5 2017
Feedback timeout |
||||||
►
Sign in to add a comment |
||||||
Comment 1 by erikc...@chromium.org
, Mar 22 2017