Problem With Inodes At The Boot Time. [solved], Detection of pb. with the filesystem Options

[randy11]

Hello,

I'm running a Beta3 on a SL-C1000. At the boot time I have

Cramfs: wrong magic
FAT: bogus logical sector size 381
Inode #1666 was a directory with children - removong those too...
Inode #2442 was a directory with children - removong those too...
Inode #2464 was a directory with children - removong those too...
Inode #2468 was a directory with children - removong those too...
Inode #11343 was a directory with children - removong those too...

If there is no CF or SD, it's the same thing.
The NAND is good : Menu Service (D+M).
I've tried

# tune2fs -c 2 -C 3 /dev/mtdblock3
tune2fs 1.34 (25-Jul-2003)
tune2fs: Bad magic number in super-block while trying to open /dev/mtdblock3
Couldn't find valid filesystem superblock.
#

When I make a "mount" :

rootfs on / type rootfs (rw)
/dev/root on / type jffs2 (rw)
/proc on /proc type proc (rw)
none on /dev type tmpfs (rw)
/dev/mtdblock3 on /home type jffs2 (rw,noatime)
none on /dev/pts type devpts (rw)
/dev/mtdblock3 on /mnt/user type jffs2 (rw,noatime)
/dev/mmcda1 on /mnt/card type ext2 (rw)
/dev/hda1 on /mnt/cf type vfat (rw,noatime)
/dev/hda2 on /mnt/cf2 type ext2 (rw,nosuid,nodev)

I've also tried a Backup/Restore (D+M) : no modification. The backup reproduce the problem (the filesystem keep the bug) : it's a normal situation because de NAND has no default, only the filesystem.

How to obtain a normal filesystem ?

Thanks for your help.

Randy11

[karlto]

/dev/mmcda1 on /mnt/card type ext2 (rw)
/dev/hda2 on /mnt/cf2 type ext2 (rw,nosuid,nodev)

Are these correctly unmounted at shutdown (especially the second one)?

[pelrun]

Don't worry about it - it's normal behaviour and there is nothing wrong with your filesystem. It's just an artifact of how jffs2 works.

There's a detailed explanation at http://www.oesf.org/forums/index.php?showt...=20773&hl=jffs2

[karlto]

Don't worry about it - it's normal behaviour and there is nothing wrong with your filesystem. It's just an artifact of how jffs2 works.

There's a detailed explanation at http://www.oesf.org/forums/index.php?showt...=20773&hl=jffs2

Heh, of course... (for some reason I decided they were ext2 messages). Good thing someone around here knows what they're talking about!

[randy11]

Are these correctly unmounted at shutdown (especially the second one)?

No , a HARD RESET was done... But I've the same message without the CF (/dev/hd2) or SD (/dev/mmcda1).

[karlto]

Are these correctly unmounted at shutdown (especially the second one)?

No , a HARD RESET was done... But I've the same message without the CF (/dev/hd2) or SD (/dev/mmcda1).

Sorry, disregard that comment - as pelrun said above, there is nothing wrong. The filesystem leaves all that stuff to clean up later to save too many writes to memory.

[randy11]

Sorry, disregard that comment - as pelrun said above, there is nothing wrong. The filesystem leaves all that stuff to clean up later to save too many writes to memory.

Thanks, for the replies. Well, if my understanding is good:
- no problem with the hardware;
- nothing will be done by the system, except if there is too many errors.

I'm right or not ?

For the fun, I will suppose that the corrections could be done :
- make a backup of the faulting partition on a true ext2 partition;
- make the corrections : "wrong magic", ...
- build an image for the jffs2 filesystem;
- replace the faulting file (jffs2 image) with the new.

(At this time it's not necessary in my particular case.)

What do you think about this idea ?

[pelrun]

Ignore the "wrong magic" - that message is only there because the system is autodetecting the partition format. It tries to mount using cramfs, then FAT, then jffs2 - the first two fail (understandably) and the jffs2 attempt succeeds. You will *always* see these messages.

In the event something does corrupt the data on your partition, you don't want to restore the whole filesystem anyway - you don't know how many files are affected. Backup your important files, do a fresh reinstall of the OS, then restore them. But this is very unlikely. jffs2 is very robust; it'd have to be a program illegally writing directly to the partition or a hardware fault and not just a crash.

But this is a last resort - you *don't* want to flush the journal or rewrite the partition unnecessarily, as it just wastes more of the limited writes your flash can do before it dies! And as soon as any of the journal is written out the warnings will return.

[karlto]

Thanks, for the replies. Well, if my understanding is good:
- no problem with the hardware;
- nothing will be done by the system, except if there is too many errors.

I'm right or not ?

1) Yes, there is no problem with the hardware
2) Just to clarify, what you posted above didn't show any errors at all - it was just letting you know that everything it planned to do is now done

(although tune2fs gave an error because it's built for ext2 and you tried to use it on jffs2)

[randy11]

1) Yes, there is no problem with the hardware
2) Just to clarify, what you posted above didn't show any errors at all - it was just letting you know that everything it planned to do is now done

(although tune2fs gave an error because it's built for ext2 and you tried to use it on jffs2)

Hourra ! Thanks everybody for your help.

Have a good day.

Randy11