2.4.0-prerelease problems (it corrupted my ext2 filesystem)

Hi

I was using 2.4.0-prerelease without extra patches and I experienced some heavy (ext2) file system corruption. I was grabbing some video using bttv at the time. Kernel didn't oops, but processess just started terminating.

Here is a the interesting part from my logs:

Bad swap file entry 5c5b6256
VM: killing process qtvidcap
swap_free: Trying to free nonexistent swap-page last message repeated 23 times
swap_free: Trying to free swap from unused swap-device
swap_free: Trying to free nonexistent swap-page last message repeated 266 times
Bad swap file entry 272c2e24
VM: killing process pppd
swap_free: Trying to free nonexistent swap-page last message repeated 30 times

Vedran Rodic
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/

Could you provide details of your configuration?

--
Daniel
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/

I put the complete kernel log of that session at http://quark.fsb.hr/~vrodic/kern.log

I scanned my swap device several times today with badblocks -w, and it didn't show any errors. I also did some RAM tests with memtest86, again with no errors.

If you need more details, just ask.

Vedran
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/

Are you still running 2.4.0-pre? Can you reproduce the problem? Does the problem occur only with v4l? Did you back up your files?

BTW, while spelunking the swap code I noticed this oddity:

pte_t pte_mkdirty(pte_t pte) { (pte).pte_low |= _PAGE_DIRTY; return
pte; }

and similarly for 10 or so other functions - these functions just return the passed pte, and reassign it to the pte orginally passed. :-/ This dates from version 1.2.13.

--
Daniel
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/

No. I am afraid :\, I think that I can't reproduce this problem easily. You see at the log files that it happened after some hours of uptime, and I did some bttv grabbing for an hour or so. I used 2.4.0-pre, pre12, pre11, pre9. pre12 was nice, I had no problems with pre11 either (well except for the instability of nvidia binary drivers, but I'm not using that for a longer period now). I did bttv video grabbing only on 2.4.0pre.

I didn't backup recently, but I was lucky:) The most damaged areas were /usr, /etc managed to get untouched, and my home dir is fine. e2fsck did clear a lot of inodes, but I managed to boot the system and reinstall most of the stuff. I'm backing up stuff now, and when I finish (10G will be a bit slow with a 4x burner), I'll start playing with fire again.

I just remembered that I booted 2.4.0-pre after the disaster, and it mentioned something about not being able to access sectors beyond the end of the device. I booted 2.2.17 after, and ran e2fsck.

Could it be that I experienced IDE DMA timeout problem?

Vedran
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/