amd64/105155: Firfox hangs on a recently upgraded 6.2-PRERELEASE.

[Roland Smith]

On Sat, Nov 04, 2006 at 11:04:27PM +0100, O. Hartmann wrote:
> Roland Smith wrote:
> >> Number:         105155
> >> Category:       amd64
> >> Synopsis:       Firfox hangs on a recently upgraded 6.2-PRERELEASE.
> >> Confidential:   no
> >> Severity:       serious
> >> Priority:       medium
> >> Responsible:    freebsd-amd64
> >> State:          open
> >> Quarter:        
> >> Keywords:       
> >> Date-Required:
> >> Class:          sw-bug
> >> Submitter-Id:   current-users
> >> Arrival-Date:   Sat Nov 04 21:00:31 GMT 2006
> >> Closed-Date:
> >> Last-Modified:
> >> Originator:     Roland Smith
> >> Release:        FreeBSD 6.2-PRERELEASE amd64
> >> Organization:
> >> Environment:
> >>     
> > System: FreeBSD slackbox.xs4all.nl 6.2-PRERELEASE FreeBSD 6.2-PRERELEASE #0: Sat Nov 4 12:09:58 CET 2006 rsmith at slackbox.xs4all.nl:/usr/obj/usr/src/sys/RFS amd64
> >
> >
> > 	
> >   
> >> Description:
> >>     
> >
> > After upgrading to 6.2-PRERELEASE, firefox hangs when trying to set
> > some of the preferences, or when opening certain websites.
> >
> > Upgrading all ports didn't help.
> >
> > Starting firfox from a terminal gives the following error:
> >
> > GConf Error: Failed to launch configuration server: Failed to execute child process "/usr/local/libexec/gconfd-2" (Invalid argument)
> >
> >
> >   
> >> How-To-Repeat:
> >>     
> > Update to 6.2-PRERELEASE on amd64, start firefox. Open the "preferences"
> > dialog, select the "Content" tab, and press the "Manage..." button. Firefox
> > will hang.
> >

> I can do the recommended 'task for crash', but it takes a while, say,
> about 30 seconds, before Firefox is bak.
> 
> But I also realized a more often hang when closing Firefox, the
> process is still there, eating up 100% CPU time.

I get instant hangs. Usually there is no CPU usage at all. If I try to
attach gdb to a firefox-bin process, it barfs.

One time I got a hang at startup with 100% CPU usage.

It was possible to to a ktrace on it. But the ktrace.out is 50 MB, so I
won't post it here. :-) I can mail it if anybody is interested.

One thing that struck me when going over the tracedump was this:

 23905 firefox-bin CALL  kse_thr_interrupt(0,0x6,0x7fffffffd120)
 23905 firefox-bin RET   kse_thr_interrupt -1 errno 22 Invalid argument

It happens on all the firefox-bin processes. It looks like is being
called with a NULL pointer as its first argument. This looks fishy to
me. If it is suspect, the trouble might be in the pthread library?

BTW, the prototype for kse_thr_interrupt in the kse(2) manpage doesn't
match /usr/include/sys/kse.h

Roland
-- 
R.F.Smith                                   http://www.xs4all.nl/~rsmith/
[plain text _non-HTML_ PGP/GnuPG encrypted/signed email much appreciated]
pgp: 1A2B 477F 9970 BA3C 2914  B7CE 1277 EFB0 C321 A725 (KeyID: C321A725)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-amd64/attachments/20061104/8270babe/attachment.pgp