1238275 - Firefox freezes when visiting the website: http://mgmt.cmb.ac.lk/

Status: RESOLVED DUPLICATE of bug 1308876

(Core :: Layout, defect)

Description

[Chamath Mc]

User Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:43.0) Gecko/20100101 Firefox/43.0
Build ID: 20160105164030

Steps to reproduce:

Go to the following website:
http://mgmt.cmb.ac.lk/


Actual results:

Firefox froze


Expected results:

Firefox should not have frozen

Comment 1

[Chamath Mc]

Request: The website given in the test plan is a university website. I kindly request to use techniques which uses minimal amount of its bandwidth during testing etc. (maybe by using a separate similar source code etc.?) Thank you.

Comment 2

[[:philipp]]

i can reproduce the issue, that during loading the page it is freezing for a considerate amount of time before it is becoming responsive.

Comment 3

[YF (Yang)]

The entire program or tab (e10s) be no response and a longer time wait (>1min).
It can be reproduce in Fx44b7 and Fx28.0, works fine in Chrome 47.

Comment 4

[YF (Yang)]

Attachment: testcase (hang)

It leads to hang (non-unlimited) with simple div and css only.

Comment 5

[Timothy Nikkel (:tnikkel)]

I'm guessing this is exponential blow up in number of reflows caused by the nested scroll frames trying to layout with/without scrollbars.

Comment 6

[Lucius Chiaraviglio]

Confirmed freeze on Windows 7 SP1 Professional 64-bit edition, with Firefox 43.0.4.  It eventually unfreezes briefly and then re-freezes a couple of times (on the originally posted site above) before finally working normally.  It does not seem to freeze the rest of the operating system, although I have not tested this exhaustively.

Could this be a duplicate of Bug 874756?  Unfortunately cannot test the above site on Windows Vista (on which that bug was found), because that computer fried.

Comment 7

[John Daggett (:jtd)]

With textperf logging, the URL in the description runs three enormous reflows:

With textperf logging, reflows are taking a total of 95 secs (!!). Three big reflows, each taking roughly 23 secs.

[00C19140]: W/textperf (textperf-loaddone) 144F9E70 time-ms:   95953 reflow: 48 chars: 12610 [http://mgmt.cmb.ac.lk/] content-textruns: 654 chrome-textruns: 0 max-textrun-len: 323 word-cache-lookups: 2013 word-cache-hit-ratio: 0.731 word-cache-space: 0 word-cache-long: 0 pref-fallbacks: 10002 system-fallbacks: 0 textruns-const: 954 textruns-destr: 769 generic-lookups: 19 cumulative-textruns-destr: 769

Comment 8

[Florencia Di Ciocco, NI to vbalducci]

Hi,

I get the impression that we can close this bug. Have you had this issue lately? It takes a while to load but so does on Chrome, so I get the impression that the page is at fault.

Regards, Flor.

Comment 9

[Daniel Holbert [:dholbert]]

Yeah -- the attached testcase loads just fine in current Nightly, and the original URL takes a little while to load but doesn't seem to hang (and apparently is now on par with other browsers per comment 8).

I confirmed that older builds (e.g. Nightly 2017-01-01) do hang on the attached testcase, so I ran mozregression --find-fix, which gave me this fix range: https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=30ea2905130e85f9e1d8d56fa3097901eec6514b&tochange=67cd1ee26f2661fa5efe3d952485ab3c89af4271

In that range, bug 1308876 seems most likely related ("Nested inline-blocks with matching width locks up browser due to O(2^depth) reflow performance"). --> Duping to that bug.