I got it. You ran some program that uses libxul (yelp, galeon, epiphany), under root, which created /usr/lib/xulrunner/components/xpti.dat and compreg.dat. The latter, having been created before upgrade to 1.8.0.1-6, doesn't contain the references to the native uconv component, thus the failure. The debug builds may not stricly use these files at initialisation, so that it won't crash with these builds.
Remove /usr/lib/xulrunner/components/*.dat and you'll be happy users again. I'll do that automatically on upgrade of the package (in -7). Thanks a lot to Øystein for his precious help. Thanks also to Carsten. Mike On Wed, Mar 22, 2006 at 01:05:01PM +0100, Mike Hommey <[EMAIL PROTECTED]> wrote: > Hi everyone, > > Considering the changes between -5 and -6, it seemed pretty obvious that > the activation of the use of native uconv (ie using glibc's iconv) was > the root of the problem, but I could not see the link between iconv and > a crash in nsFrame::BoxReflow. Being unable to reproduce the bug myself > doesn't really help. > > After some code lookup and investigation based on Oystein's core dump > (thank again), I found out about this upstream bug[1] in which I found a > hint as to why the error occurs: "I think the real problem is with the > stylesheets failing to load." > > According to the straces you both provided, and compared to what strace > gives on my computer, it appears that no .css file is being loaded > successfully. At all. > > So the problem might be happening somewhere around line 1200 or 1214 of > layout/style/nsCSSLoader.cpp. I still don't understand why it fails on > some computers and not on some others. > > It would be really helpful to get Oystein's stdout/stderr output of yelp > running with debug enabled libxul0d, the one with which no crash occurs, > so that we can have an hint about what kind of assertions fail and where. > > In the meantime, I'm going to try to look further into the code to see > what can be happening. > > Thanks for your help > > Mike > > 1. https://bugzilla.mozilla.org/show_bug.cgi?id=292549 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
