See upcoming testcase, you need to test the testcase locally, because of the use of enhanced privileges. It creates a very large amount of tthreads, after which Mozilla gives an out of memory error, which is fine. But afterwards, nothing of the ui gets repainted anymore. Maybe this bug is invalid, but it seems weird to me that things are not painted anymore in this case, that's why I filed the bug.

Ok, this was fixed somehow between 2006-08-09 and 2006-08-10: http://bonsai.mozilla.org/cvsquery.cgi?treeid=default&module=all&branch=HEAD&branchtype=match&dir=&file=&filetype=match&who=&whotype=match&sortby=Date&hours=2&date=explicit&mindate=2006-08-09+03&maxdate=2006-08-10+06&cvsroot=%2Fcvsroot I guess it was fixed by bug 343655 or bug 342366. I also guess that this wasn't a problem on linux then? And finally, I guess this bug can be marked worksforme/fixed?

I filed bug 413866 on a particular case where it seems to have regressed. (although I'm still not sure whether this is maybe invalid completely)

I'm going to reopen this. I still got this "stop painting of ui" when viewing a divx video like this one (you need to have the divx web player installed): http://www.stage6.com/user/SwissPortraiture/video/2219903/Swiss-Portraiture Also, after loading the testcase, I seem to crash after a while, after I opened and closed a bunch of tabs. I'll attach the stack, it seems to crash in the cycle collector.

For some reason, crash reports that are sent are somehow not handled, because they don't show up in the list of reported crashes.

(In reply to comment #4) > I'll attach the stack, it seems to crash in the cycle collector. Looks like an out of memory exception.

(In reply to comment #6) > (In reply to comment #4) > > I'll attach the stack, it seems to crash in the cycle collector. > > Looks like an out of memory exception. > I see a bunch more out of memory exceptions coming from the graph builder in just the last week, see http://crash-stats.mozilla.com/report/list?range_unit=weeks&query_search=stack&query_type=contains&platform=windows&signature=RaiseException&query=NodePool&range_value=1 I think that needs a new bug, maybe blocking...

Ok, I've file it as bug 420145 now, so I'll leave this bug now open for the "ui not painting anymore" isse, although I haven't been able to reproduce it with the testcase anymore (but I did with the stage6 divx video).

Lack of painting when out of memory is entirely unsurprising. What's surprising is that you didn't crash for long enough to notice something else go wrong.