[Pkg-sysvinit-devel] Bug#552248: initscripts: checkfs.sh fails with "device doesn't exist" on ext4 /home

[thveillon.debian]

>Petter Reinholdtsen wrote:
> [Thveillon]
>> They have different sata controllers, and they don't display this
>> behavior in neither Fedora11 (tested on the Desktop) and Ubuntu
>> Karmic (on the laptop).
> 
> Both use event based boot, as far as I know, which solve the
> fundamental problem with the changed kernel behaviour.  Did it work
> with Debian Lenny?

The laptop is too new to be used easily with Lenny, but it was working
properly with Squeeze prior to the "insserv" migration. The desktop has
been running Lenny during most of it Testing cycle, and beginning of the
 Stable one. It ran Squeeze before the "insserv" migration too.


>> Let me know if I can provide any other information, or provide logs from
>> a boot on a standard Debian kernel (those attached are from a custom one).
> 
> Logs from a standard kernel would be nice, to see if device timings or
> driver load order is different.  The output from 'ls /etc/rc?.d',
> 'ls -l $(grep -v '#' /etc/fstab|awk '/dev/ {print $1}') /dev/rtc*' and
> 'cat /etc/default/rcS' would be nice too.  The system seem to be
> without udev.  Is that correct?  I noticed from the boot log that the
> hardware clock is not set correctly either, because the rtc devices
> are missing.

Following your message I did various tests, it seems that the failing
fsck is just a symptom after all, not the cause of the problem.

The clock not being set correctly because /dev/rtc doesn't exist is only
happening during a failed boot, not a normal one. As you'll see in the
output of the commands the /dev/rtc* are there.

After the failed boots I have various and semi-random other problems if
I try to go past the rescue shell. /home isn't mounted, and not
mountable, but I also sometime lose the keyboard/mouse(trackpad), the
wifi card doesn't come up, and such...

Both machines are *with* udev, but it doesn't work during the failed
boots and I believe udev is the root of all other problems. There's so
many bugs open right now that it won't come has a big surprise.


> Why do you make your own kernel?  Anything special about its
> configuration?

I use custom kernels for various reasons, curiosity being one. On the
laptop the sound is only (kind of) working with 2.6.30* and up, which
wasn't available when I first installed Debian on it, and is now fully
working with 2.6.31*.

On both I use "desktop" settings (1000Mhz, voluntary preemption), set
the "Core2 or newer" cpu type, remove debug or devel related options,
remove Xen related options, remove the "v4l1" (using "v4l1-compat" only)
and use only the "old" IEEE1394 stack (I think it's the default too on
Debian kernels now) since some of my video and scanner equipment doesn't
work properly otherwise.
I can provide a "diff" if desired, but I had the same behavior with both
custom and Debian kernels.

Since tomorrow (Monday here) I'll need working machines I ended up
restoring older system from backups, it now runs fine again (with the
same kernels). I guess I'll wait until next week-end to upgrade again...
Showing up in front of the classroom to display on a large screen a
"black screen of death" isn't exactly a good way to promote Debian ! (My
wife's a teacher and use the laptop at work).

Sorry for the waste of time, you can dismiss the bug report since I
"worked around" the problem for now, and I think it's udev related in
the first place anyway. I'll reopen it if it happens again on next upgrade.


> Happy hacking,


Have a nice week, and a nice Skolelinux gathering/conference.

Thank you for your attention.
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: output_desktop.txt
URL: <http://lists.alioth.debian.org/pipermail/pkg-sysvinit-devel/attachments/20091025/e82573f5/attachment-0002.txt>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: output_laptop.txt
URL: <http://lists.alioth.debian.org/pipermail/pkg-sysvinit-devel/attachments/20091025/e82573f5/attachment-0003.txt>