1328545 - Crash [@ OOM | small ] with FragmentOrElement.cpp involved

Status: RESOLVED WONTFIX

(Core :: DOM: Core & HTML, defect, P5)

Description

[Carsten Book [:Tomcat]]

Found via bughunter and reproduced with latest tinderbox opt windows builds on beta and trunk

Steps to reproduce:
-> Load http://results.jukola.com/tulokset/results_j2016_ju.xml
---> Crash in opt builds after 10 -> 30 seconds

affects trunk to beta builds 

Crash ID: https://crash-stats.mozilla.com/report/index/4bdddf42-cd3f-48f2-8adb-d8d962170104

Comment 1

[Carsten Book [:Tomcat]]

[Tracking Requested - why for this release]:

Tobias is this something for you ? Or do you know who could take a look, thanks!

Comment 2

[Liz Henry (:lizzard) (relman/hg->git project)]

I can go ahead and track this for 51 and up, if we come up with a patch we can certainly consider uplifting since improving the OOM crash rate would be a good thing. 

Andrew can you help find someone to investigate? Thanks!

Comment 3

[Olli Pettay [:smaug][[e-mail]]]

That crash-stat is showing us crashing deep inside DOM when doing a tiny allocation. Making that particular case fallible wouldn't be reasonable.

/me wonders how Tomcat managed to find that page. Results in a massive xml file from a small-ish Finnish sport happening :)

Comment 4

[Olli Pettay [:smaug][[e-mail]]]

FWIW, the page crashes the child process on Chrome too.

Comment 5

[Andrew Overholt [:overholt]]

I'm not sure how important XML issues are (and thus need to be tracked). Are these crashes frequent?

Comment 7

[Carsten Book [:Tomcat]]

(In reply to Andrew Overholt [:overholt] from comment #5)
> I'm not sure how important XML issues are (and thus need to be tracked). Are
> these crashes frequent?

well i have also no idea how important just found this as real-world-website example for this crash.

Comment 8

[Gerry Chang [:gchang]]

Too late for 51, mark 51 as won't fix.

Comment 9

[Julien Cristau [:jcristau]]

Too late for 52 as well.

Comment 10

[Liz Henry (:lizzard) (relman/hg->git project)]

Dropping this for 53 since it does not seem actionable.
Is this anything you want to dig deeper into or shall we leave it alone?

Comment 11

[Olli Pettay [:smaug][[e-mail]]]

It is trivial to crash any browser (child process) by just creating more and more elements, so I don't see there anything we can do.

Comment 12

[Wayne Mery (:wsmwk)]

(In reply to Olli Pettay [:smaug][[e-mail]] from comment #11)

It is trivial to crash any browser (child process) by just creating more and
more elements, so I don't see there anything we can do.

=> wontfix