udev rule ruins boot process w/kernel-2.6.14.3--2 [SOLVED]

kozaki · 2005-12-11 18:38:08

Hi, I'm stucked since last upgrade this morning (from testing as I need kernel-2.6.14 for mobo Ethernet controller)

[12/11/05 02:39] starting full system upgrade
[12/11/05 02:39] installed libcap (1.10-1)
[12/11/05 02:39] installed libdaemon (0.10-1)
[12/11/05 02:39] installed nss-mdns (0.7-1)
[12/11/05 02:39] installed avahi (0.6.1-2)
[12/11/05 02:39] upgraded gnome-vfs (2.12.2-2 -> 2.12.2-3)
[12/11/05 02:39] upgraded gnutls (1.3.0-2 -> 1.2.9-2)
[12/11/05 02:39] upgraded mkinitrd (1.01-18 -> 1.01-21)
[12/11/05 02:40] upgraded kernel26 (2.6.14.3-1 -> 2.6.14.3-2)

Boot goes up a few seconds until it prints :

EXT2-fs warning (device sda6): ext2_fill_super: mounting ext3 filesystem as ext2
(...)
::: checking filesystem
/dev/sda6 clean
/dev/sda9 is mounted. e2fsck: cannot continue, aborting
Type root password for (...)

both /dev/sda6 ("/" partition) & /dev/sda9 ("/home") come with ext3 FS.
Note that after each start, Win XP launched checkdisk.

Whatever I tried did change nothing.

- booting with Knoppix, e2fsck both partitions show they're clean :!:

- Booting from Arch install CD, I triple-checked /etc/fstab & ~/rc.conf are allright.

- Then i edited Grub to boot on initrd26-full.img
Also checked /etc/mkinitrd.conf* and launched mkinitrd auto

Now I dunno what to do :?:

I found nothing on this bbs related to such a problem. Googling shows many posts from guys triying to boot an ext3 FS without ext3 support in their system, which IMHO cannot be the case here ?...

Will now try to boot again after disabling sda9 (/home) in fstab.

ps : I'm writing on Arch forum from a windoze OS :x

---
* /etc/mkinitrd.conf :

HOSTCONTROLLER_IDE=ide-cd
HOSTCONTROLLER_SCSI=sd_mod
HOSTCONTROLLER_SATA=sata_uli
HOSTCONTROLLER_USB=
FILESYSTEMS=ext3

kozaki · 2005-12-11 19:11:24

Wrinting from arch, but

- edited /etc/fstab, adding "noauto" to /dev/sda9 (/home) line, then rebooted, with same message (...) /dev/sda9 is mounted. e2fsck: cannot continue, aborting (...) :?:

- Only when I commented /dev/sda9 then I could finally boot into my box.
/home wasn't mounted but e2fsck told me it was :?:
Launched e2fsck -nfv /dev/sda9

Warning!  /dev/sda9 is mounted.
Warning: skipping journal recovery because doing a read-only filesystem check.
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information

   10448 inodes used (4%)
     441 non-contiguous inodes (4.2%)
         # of inodes with ind/dind/tind blocks: 598/25/0
  209591 blocks used (40%)
       0 bad blocks
       0 large files

    9399 regular files
    1008 directories
       0 character device files
       0 block device files
       0 fifos
       0 links
      24 symbolic links (18 fast symbolic links)
       8 sockets
--------
   10439 files

Then e2fsck -fv /dev/sda9 with clean results
Then mounted /dev/sda9, --> Everything's working fine :? :?: :?:

Please have you got a clue why all that stuff about my /home partition ?

Lone_Wolf · 2005-12-11 19:24:02

Note that after each start, Win XP launched checkdisk.

This would worry me a lot if it were my pc.

I think it points to a HD failure.

Make sure you backup important stuff, just in case it is indeed a hardware problem.

kozaki · 2005-12-11 19:28:00

Sure I will backup. & will activate SMART in Bios, then lauchn HDD tests with apropriate tools.

I also did notice that e2fsck results are clean. HDD is Hitachi SATA2, 2 weeks old

kozaki · 2005-12-12 12:57:49

Well, while I disabled /dev/sda9 (as well as /dev/sda12 jfs stockage partition) in /etc/fstab, I now can boot & mount them manually without warning.

SMART activated show : OK for the Sata HDD (single HDD anyway)
Also noticed PCI Latency in the Bios is set to 64. could that be a problem ?

Now, where can that message comes from

$ dmesg | grep ext2_fill
EXT2-fs warning (device sda6): ext2_fill_super: mounting ext3 filesystem as ext2

:?:

Snowman · 2005-12-12 16:30:19

kozaki wrote:

Now, where can that message comes from
$ dmesg | grep ext2_fill
EXT2-fs warning (device sda6): ext2_fill_super: mounting ext3 filesystem as ext2
:?:

Try googling for the error message:
google:EXT2-fs warning (device sda6): ext2_fill_super: mounting ext3 filesystem as ext2

kozaki · 2005-12-12 23:36:42

snowman > done (i may have wrote too much but this is in 1st post )

kozaki wrote:

Googling shows many posts from guys triying to boot an ext3 FS without ext3 support in their system, which IMHO cannot be the case in this location ?...

As soon as i enable /dev/sda9 /home in /etc/fstab, boot hang up.
Typing the root password, mounting "/" rw then commenting that fstab line, then rebooting --> boot goes fine.

# e2fsck -v /dev/sda9 --> allright.
# mount /dev/sda9 /home --> allright (note i don't add any FS option as it is ext3)

/me dunno what to do next :?:

Now the whole new system is getting sloooooooow as soon as there's HDD writing access (like p2p and a lot of other apps) :x
/dev/sda9, or Sata related problem (p2p writes on a separate jfs partition) ?

Snowman · 2005-12-13 01:22:21

kozaki wrote:

snowman > done (i may have wrote too much but this is in 1st post )
kozaki wrote:
Googling shows many posts from guys triying to boot an ext3 FS without ext3 support in their system, which IMHO cannot be the case in this location ?...

Sorry. Sometime, I don't fully read each post in a thread before posting.

About your problem, you could downgrade back to kernel-2.6.14.3-1 to make sure that the problem is due to kernel-2.6.14.3-2 and not a config/hardware problem.

kozaki · 2005-12-13 01:32:59

snowman > no problem at all, i know you're trying to help
yep, i will try downgrade to kernel26-2.6.14.3-1.
Actually, see what happends with the JFS partition (defined in fstab but unmounted @ boot, then) :

$ su
Password:
# mount /dev/sda12 /mnt/sda12
mount: type erroné de système de fichiers, option erronée, super bloc erroné sur
/dev/sda12,
codepage manquante ou autre erreur
Dans quelques cas certaines informations sont utiles dans syslog - essayez
dmesg | tail ou quelque chose du genre
# mount -t jfs /dev/sda12 /mnt/sda12
mount: type erroné de système de fichiers, option erronée, super bloc erroné sur
/dev/sda12,
codepage manquante ou autre erreur
Dans quelques cas certaines informations sont utiles dans syslog - essayez
dmesg | tail ou quelque chose du genre
# jfs_fsck /dev/sda12
jfs_fsck version 1.1.10, 19-Oct-2005
processing started: 12/13/2005 3.25.6
Using default parameter: -p
The current device is: /dev/sda12
Block size in bytes: 4096
Filesystem size in blocks: 9797634
**Phase 0 - Replay Journal Log
Filesystem is clean.
# mount -t jfs /dev/sda12 /mnt/sda12
#

jfs_fsck finds nothing (it takes 5" to finish) then I can mount it flawlessly :?:

Snowman · 2005-12-13 02:52:31

Try passing devfs=nomount to the kernel.
Ref: http://mandrake-mandriva.news-view.co.uk/topic-516.html

apeiro · 2005-12-13 06:56:49

Hi kozaki,

Please try regenerating your init ramdisk with the latest version of mkinitrd, 1.01-23. I think we've fixed the VFS errors you see.

kozaki · 2005-12-13 12:22:18

As for 'devfs=nomount' I have this line in grub/menu/lst since I installed Arch on this box
Also, despite the warning at boot :

Dec 13 03:23:03 llewellyn EXT2-fs warning (device sda6): ext2_fill_super: mounting ext3 filesystem as ext2

$ cat /proc/mounts shows it is mounted as ext3 :

rootfs / rootfs rw 0 0
/dev/sda6 / ext3 rw,data=ordered 0 0

There's nothing /me understand in this issue.

Upgraded to latest kernel, mkinitrd & filesystem from testing --> will give next reboot all my attention :?

kozaki · 2005-12-13 13:14:15

Actually found where did this strange messages & behaviours come from

I did not test enough an Udev rule i tried to (u)mount my external drives & keys on a 100% Udev base. And this made Udev be confused between the internal SATA HDD (/dev/sda) & the external IDE USB HDD (also called /dev/sda !)
Checking cat /proc/mounts after boot process didn't give any usefull infos, but last time I booted & was stucked then typed root passwd for maintenance, cat /proc/mounts revealed that external IDE USB HDD was mounted in /dev/sdaN

Removed this rule while i found a new one that I can use with SATA HDD + external USB drives, and now my box returned to KISS

# Udev rules for sd* automounting without /etc/fstab
# By lanrat (Rafal Szczepaniak)

# rules & ACTION that automatically (u)mount N external USB drives :

KERNEL=="sd[a-z]", NAME="%k", SYMLINK+="usbhd-%k", GROUP="users", OPTIONS="last_rule"
ACTION=="add", KERNEL=="sd[a-z][0-9]", SYMLINK+="usbhd-%k", GROUP="users", NAME="%k"
ACTION=="add", KERNEL=="sd[a-z][0-9]", RUN+="/bin/mkdir -p /media/usbhd-%k"
ACTION=="add", KERNEL=="sd[a-z][0-9]", RUN+="/bin/ln -s /media/usbhd-%k /mnt/usbhd-%k"
ACTION=="add", KERNEL=="sd[a-z][0-9]", PROGRAM=="/sbin/vol_id -t %N", RESULT=="vfat", RUN+="/bin/mount -t vfat -o rw,noauto,sync,dirsync,noexec,nodev,noatime,dmask=000,fmask=111 /dev/%k /media/usbhd-%k", OPTIONS="last_rule"
ACTION=="add", KERNEL=="sd[a-z][0-9]", RUN+="/bin/mount -t auto -o rw,noauto,sync,dirsync,noexec,nodev,noatime /dev/%k /media/usbhd-%k", OPTIONS="last_rule"
ACTION=="remove", KERNEL=="sd[a-z][0-9]", RUN+="/bin/rm -f /mnt/usbhd-%k"
ACTION=="remove", KERNEL=="sd[a-z][0-9]", RUN+="/bin/umount -l /media/usbhd-%k"
ACTION=="remove", KERNEL=="sd[a-z][0-9]", RUN+="/bin/rmdir /media/usbhd-%k", OPTIONS="last_rule"

lanrat > I still consider your hacked rule as a great one. Only Sata sucks my life in some way (think udev, hdparm, etc)

lanrat · 2005-12-13 15:56:10

Oh, you're right I was testing the rules on my IDE system only :-). The easy way is to change [a-z] to something like [b-z] and leave your internal sata hdd (/dev/sda) to be handled by standard arch udev rules.

Arch Linux

#1 2005-12-11 18:38:08

udev rule ruins boot process w/kernel-2.6.14.3--2 [SOLVED]

#2 2005-12-11 19:11:24

Re: udev rule ruins boot process w/kernel-2.6.14.3--2 [SOLVED]

#3 2005-12-11 19:24:02

Re: udev rule ruins boot process w/kernel-2.6.14.3--2 [SOLVED]

#4 2005-12-11 19:28:00

Re: udev rule ruins boot process w/kernel-2.6.14.3--2 [SOLVED]

#5 2005-12-12 12:57:49

Re: udev rule ruins boot process w/kernel-2.6.14.3--2 [SOLVED]

#6 2005-12-12 16:30:19

Re: udev rule ruins boot process w/kernel-2.6.14.3--2 [SOLVED]

#7 2005-12-12 23:36:42

Re: udev rule ruins boot process w/kernel-2.6.14.3--2 [SOLVED]

#8 2005-12-13 01:22:21

Re: udev rule ruins boot process w/kernel-2.6.14.3--2 [SOLVED]

#9 2005-12-13 01:32:59

Re: udev rule ruins boot process w/kernel-2.6.14.3--2 [SOLVED]

#10 2005-12-13 02:52:31

Re: udev rule ruins boot process w/kernel-2.6.14.3--2 [SOLVED]

#11 2005-12-13 06:56:49

Re: udev rule ruins boot process w/kernel-2.6.14.3--2 [SOLVED]

#12 2005-12-13 12:22:18

Re: udev rule ruins boot process w/kernel-2.6.14.3--2 [SOLVED]

#13 2005-12-13 13:14:15

Re: udev rule ruins boot process w/kernel-2.6.14.3--2 [SOLVED]

#14 2005-12-13 15:56:10

Re: udev rule ruins boot process w/kernel-2.6.14.3--2 [SOLVED]

Board footer