FAQ Search Today's Posts Mark Forums Read
» Video Reviews

» Linux Archive

Linux-archive is a website aiming to archive linux email lists and to make them easily accessible for linux users/developers.


» Sponsor

» Partners

» Sponsor

Go Back   Linux Archive > Redhat > Fedora User

 
 
LinkBack Thread Tools
 
Old 03-24-2008, 09:10 AM
Michael Schwendt
 
Default kmail dead in water, using tbird I hope

On Mon, 24 Mar 2008 00:01:52 -0400, Gene Heskett wrote:

> I've had several hard lockups over the last day, possibly related to a
> tickless kernel2.6.24.1 I could keep running long enough to recompile it
> with the usual kilohertz tick, so ATM I'm on one of the older 2.6.23
> kernels from the F8 repo's, and kde is 3.5.9
>
> Anyway, kmail (and kontact but I don't use it) is now doing a nearly
> instant Signal 11. The kmail.kcrash report sort of points a finger at
> libkmailprivate.so, but I've stopped X, had yum rip out kdepim and
> kdepim-libs, and then re-install them with no effect on the crash.
>
> So here is the kmail.kcrash text:
> Using host libthread_db library "/lib/libthread_db.so.1".
> (no debugging symbols found)
> [snip many lines oof this]
> (no debugging symbols found)
> [Thread debugging using libthread_db enabled]
> [New Thread -1208654128 (LWP 25523)]
> (no debugging symbols found)
> [snip many more lines]
> (no debugging symbols found)

This means that you should install the corresponding -debuginfo rpms to
fill in the missing details in the backtrace.

> [KCrash handler]
> #6 0x4f406736 in KMMsgDict::insert () from /usr/lib/libkmailprivate.so
> #7 0x4f58414e in KMFolderIndex::fillMessageDict ()
> from /usr/lib/libkmailprivate.so

Do you realise that it could be a segfault caused by corrupted input data
and not by corrupted packages? Depending on what libkmailprivate does in
this area of the code, it could be that it doesn't verify the input data
enough as would be necessary to prevent it from crashing.

--
fedora-list mailing list
fedora-list@redhat.com
To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
 
Old 03-24-2008, 12:48 PM
Gene Heskett
 
Default kmail dead in water, using tbird I hope

On Monday 24 March 2008, Michael Schwendt wrote:
>On Mon, 24 Mar 2008 00:01:52 -0400, Gene Heskett wrote:
>> I've had several hard lockups over the last day, possibly related to a
>> tickless kernel2.6.24.1 I could keep running long enough to recompile it
>> with the usual kilohertz tick, so ATM I'm on one of the older 2.6.23
>> kernels from the F8 repo's, and kde is 3.5.9
>>
>> Anyway, kmail (and kontact but I don't use it) is now doing a nearly
>> instant Signal 11. The kmail.kcrash report sort of points a finger at
>> libkmailprivate.so, but I've stopped X, had yum rip out kdepim and
>> kdepim-libs, and then re-install them with no effect on the crash.
>>
>> So here is the kmail.kcrash text:
>> Using host libthread_db library "/lib/libthread_db.so.1".
>> (no debugging symbols found)
>> [snip many lines oof this]
>> (no debugging symbols found)
>> [Thread debugging using libthread_db enabled]
>> [New Thread -1208654128 (LWP 25523)]
>> (no debugging symbols found)
>> [snip many more lines]
>> (no debugging symbols found)
>
>This means that you should install the corresponding -debuginfo rpms to
>fill in the missing details in the backtrace.
>
Where might those be found?, yum shows nothing.

I have found a bad set of .index files for the debian-users mailing list
folder, and it appears kmail crashed while regenerating it. Now its a name
only, not even an inode attached that I can find, and un-deletable. HOWEVER,
kmail is now running again after I manually deleted all other references to
the debian-user folder except this non-existent .debian-user.index file.

I had tried to delete all the *.index files after kmail first crashed because
that was the trigger 2 or 3 times before, so when I ram kmail this time, it
was several minutes, showing as high as 50% cpu in an htop display, before it
ever opened its operating screens.

I tried booting the Dec-2007 respin in rescue mode, but then cannot
umount /mnt/sysimage so that I can safely run a session of e2fsck
against /dev/sda1, but its busy and cannot be unmounted, so:

How can I force an fsck on /dev/VolGroupVol00 on the next boot?

In the meantime I have built, and rebooted to, a 2.6.24.1 kernel with
that 'tickless' BS turned off, and while its a bit early to state it without
reservation, the system does feel a heck of a lot more responsive, I was
having pauses in the whole system for 30 seconds at a time like some IRQ
wasn't being serviced in a timely manner. I am purposely making this message
longer to see if it hangs while I'm typing, and so far in this messages
editing it has not hung once like it was before.

Now its off to go read up on tune2fs & see if it can set that check me flag
on /dev/Vol* or /dev/sda1

Thanks for the reply, this has been an interesting trip indeed...

>Do you realise that it could be a segfault caused by corrupted input data
>and not by corrupted packages? Depending on what libkmailprivate does in
>this area of the code, it could be that it doesn't verify the input data
>enough as would be necessary to prevent it from crashing.

Should I, when things settle, file a bugzilla against libkmailprivate.so? Or
whatever it is that manages the .index files?

--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
It has been said that man is a rational animal. All my life I have
been searching for evidence which could support this.
-- Bertrand Russell

--
fedora-list mailing list
fedora-list@redhat.com
To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
 
Old 03-24-2008, 03:52 PM
Michael Schwendt
 
Default kmail dead in water, using tbird I hope

On Mon, 24 Mar 2008 09:48:16 -0400, Gene Heskett wrote:

> On Monday 24 March 2008, Michael Schwendt wrote:
> >On Mon, 24 Mar 2008 00:01:52 -0400, Gene Heskett wrote:
> >> I've had several hard lockups over the last day, possibly related to a
> >> tickless kernel2.6.24.1 I could keep running long enough to recompile it
> >> with the usual kilohertz tick, so ATM I'm on one of the older 2.6.23
> >> kernels from the F8 repo's, and kde is 3.5.9
> >>
> >> Anyway, kmail (and kontact but I don't use it) is now doing a nearly
> >> instant Signal 11. The kmail.kcrash report sort of points a finger at
> >> libkmailprivate.so, but I've stopped X, had yum rip out kdepim and
> >> kdepim-libs, and then re-install them with no effect on the crash.
> >>
> >> So here is the kmail.kcrash text:
> >> Using host libthread_db library "/lib/libthread_db.so.1".
> >> (no debugging symbols found)
> >> [snip many lines oof this]
> >> (no debugging symbols found)
> >> [Thread debugging using libthread_db enabled]
> >> [New Thread -1208654128 (LWP 25523)]
> >> (no debugging symbols found)
> >> [snip many more lines]
> >> (no debugging symbols found)
> >
> >This means that you should install the corresponding -debuginfo rpms to
> >fill in the missing details in the backtrace.
> >
> Where might those be found?, yum shows nothing.

In the [fedora-debuginfo] repo, and you need not enable that one,
because there are tools to install them:

$ rpm -qf $(which debuginfo-install)
yum-utils-1.1.11-1.fc8

The Wiki has additional information about them and backtraces:
http://fedoraproject.org/wiki/StackTraces

> How can I force an fsck on /dev/VolGroupVol00 on the next boot?

Add "forcefsck" to the command line.

> >Do you realise that it could be a segfault caused by corrupted input data
> >and not by corrupted packages? Depending on what libkmailprivate does in
> >this area of the code, it could be that it doesn't verify the input data
> >enough as would be necessary to prevent it from crashing.
>
> Should I, when things settle, file a bugzilla against libkmailprivate.so? Or
> whatever it is that manages the .index files?

In the KDE bug tracker, yes. It's an error condition that ought to be
handled IMO.

--
fedora-list mailing list
fedora-list@redhat.com
To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
 
Old 03-26-2008, 06:37 AM
Gene Heskett
 
Default kmail dead in water, using tbird I hope

On Monday 24 March 2008, Michael Schwendt wrote:
>On Mon, 24 Mar 2008 09:48:16 -0400, Gene Heskett wrote:
>> On Monday 24 March 2008, Michael Schwendt wrote:

>> >Do you realise that it could be a segfault caused by corrupted input data
>> >and not by corrupted packages? Depending on what libkmailprivate does in
>> >this area of the code, it could be that it doesn't verify the input data
>> >enough as would be necessary to prevent it from crashing.
>>
>> Should I, when things settle, file a bugzilla against libkmailprivate.so?
>> Or whatever it is that manages the .index files?
>
>In the KDE bug tracker, yes. It's an error condition that ought to be
>handled IMO.

Update, sorry it wasn't quicker. It was bad data, not kmail. I had a totally
bogus debian-lists.index that wasn't a file, an ls -l showed it as the name
and everything else as ???? and it couldn't be deleted. I did kill the rest
of the debian-users folder and data, and that let kmail run again. Then
someone else said I should touch /forcefsck and reboot, and that fixed it
right up. Took 40 minutes and a lot of y's but fixed it.

At this point, I am very solidly convinced that the 2.6.24 kernel I had been
running for a week or so, and which had crashed outright several times, or
gone on vacation for 30 seconds or more at a time, sometimes killing x, or
any number of other oddities, was the problem. I managed to get another copy
of it built but without the 'tickless' option in the .config, and the machine
is a whole new machine, no hangs since. I've also built 2.6.24.4 and that is
also working well ATM. I posted a message on lkml that I didn't think
tickless was quite ready for prime time, at least on my hardware.

Thanks.

--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Windows 95: Proof that P. T. Barnum was right.

--
fedora-list mailing list
fedora-list@redhat.com
To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
 

Thread Tools




All times are GMT. The time now is 05:11 AM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.
Copyright 2007 - 2008, www.linux-archive.org