[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#407495: marked as done (Linux 2.6.18-686 serious boot problem in multi-user-mode, single-user boots without problems. (Kernel or scripts in initrd?))

Your message dated Mon, 3 Aug 2009 19:44:47 +0200
with message-id <20090803174447.GH22735@inutil.org>
and subject line Re: Bug#407495: Linux 2.6.18-686 serious boot problem in multi-user-mode, single-user boots without problems. (Kernel or scripts in initrd?)
has caused the Debian Bug report #407495,
regarding Linux 2.6.18-686 serious boot problem in multi-user-mode, single-user boots without problems. (Kernel or scripts in initrd?)
to be marked as done.

This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.

(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact owner@bugs.debian.org
immediately.)


-- 
407495: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=407495
Debian Bug Tracking System
Contact owner@bugs.debian.org with problems
--- Begin Message --- 
Package: linux-image-2.6.18-3-686
Version: 2.6.18-7
Severity: grave


My hardware constallation:

It is a little hard to describe. I had done a fresh network-installation
on a rather old machine. The hardware is a little "nonstandard", so I
think I have to describe it.

The greatest part of the system is installed on a 30 GB harddisk, which
is not usable with my BIOS. But I this computer, there is a IDE
RAID-controller, which makes the disk available. The root file system,
including /boot is installed on a second hard disc, which is accessible
and bootable by the BIOS. I could not do a totally new partition of the
disk, because it contains a lot of data of a friend. ;-)

This system disk is /dev/hda, the data disk is /dev/hde. This are my
partitions and mountpoints:

/dev/hda1		Swap
/dev/hda2		(W95 extended)
/dev/hda3 on /      	(ext3, with bootable-flag)

/dev/hde1 		Swap
/dev/hde2 on /usr       (ext3)
/dev/hde3 on /home      (ext3)

The output of sed 1q /proc/ide/ide2/config:

pci bus 00 device 48 vendor 1095 device 0680 channel 0

I dont know the exact type of the the card. I am happy that it works...

My boot manager is grub.

The 2.6.8-3-386 kernel boots without any problem. The exact version of
the working kernel is (output of cat /proc/version):

Linux version 2.6.8-3-386 (root@lart) (gcc version 3.3.5 (Debian
1:3.3.5-13)) #1 Sat Jul 15 09:26:40 UTC 2006


How I experienced the problem:

After the internet installation I got a 2.6.8 kernel. I wanted to use a
webcam, but the spca5xx modules package required a 2.6.18 kernel, so I
decided to update the kernel. I used aptitude for the installation.

After the installation, I wanted to boot my new kernel. The kernel
booted, but after a short while strange error messages appeared on the
screen and I found myself in this little shell named busybox. The most
important messages here:


Messages:

Running /scripts/local-premount
[...]
name_to_dev_t(/dev/hda1) = hda1(3,1)
Trying to resume from /dev/hda1
Attempting manual resume
[...]
kinit: No resume image
[...]
Mounting /sys on /root/sys failed
[...]
/scripts/init-bottom
Mounting /root/dev on /dev/static/dev failed

(Sorry for not giving a whole transcript, I found no way to write it to
a file...)


Important additional information:

I read a little in the scripts in /scripts (but I didn't understand them
fully and I didn't try this). They seem to expect the root file system
(in my case /dev/hda3) mounted on /root - but as I looked in /root, I
found my /home directories on this machine. But the system seems to
"believe", that the /dev/hde disk is my /dev/hda disk, a "mount" command
shows me that /dev/hda was mounted. 

Really strange, and not what I expected...

But it can go much stranger.

I made a second try and booted the Kernel in recovery mode (single user,
with the single parameter to the kernel). And it booted without any
problems. This is really weird, but I can not analyze it further.
(Analyzing such problems is fun, but not the thing I boot a computer
for.)

Of course, it is reproduceable on my machine.

I think, there comes a subtle bug in the kernel or the scripts on the
initrd together with the unusual hardware.

--- End Message ---
--- Begin Message --- 
On Fri, Jan 19, 2007 at 05:19:09PM +0000, Stephen Gran wrote:
> This one time, at band camp, Elias Schwerdtfeger said:
> > Am Freitag, den 19.01.2007, 16:07 +0000 schrieb Stephen Gran:
> > 
> > > I'm glad to hear it's working for you.  This is unfortunately a known
> > > problem without much of an out of the box resolution.  Probably a line
> > > in the release notes would be good to help people who might stumble
> > > across this.  In the meantime, I am going to reduce the severity of the
> > > bug to normal, as it has a known workaround, and it is a known side
> > > effect of asynchronous device probing that will not be changed in the
> > > upstream kernel.
> > 
> > Yes, it works to boot the kernel. (And that's a lot in this situation.)
> > But /dev/hda and /dev/hde are still swapped, and the filesystem check
> > fails. To test the system, I mounted the disks manually in single user
> > mode and typed "init 2" to get a working system. 
> > 
> > I don't want to edit this in my fstab, because I still want to boot the
> > old kernel sometimes. 
> 
> Use labels for all your filesystems.  Then actual disk enumeration
> doesn't matter anymore.  Your old kernel will still run fine.  You can
> even use UUID for swap partitions instead of partition entries.
> 
> This is my fstab:
> LABEL=sda_swap  none            swap    sw              0       0
> LABEL=ROOT      /               ext3    defaults,errors=remount-ro 0       1
> LABEL=TMP       /tmp            ext3    defaults        0       2
> LABEL=VAR       /var            ext3    defaults        0       2
> LABEL=HOME      /home           ext3    defaults        0       2

Current d-i created filesystem labels out of the box.

I'm marking this bug as closed.

Cheers,
        Moritz

--- End Message ---
Reply to: