Re: 2.1.57: trouble reading ext2 superblocks

Trevor Johnson (
Mon, 29 Sep 1997 20:59:34 -0400 (EDT)

Thank you for your response. I meant to post the results of "fdisk -l"
but forgot. It didn't show anything unusual, as far as I could tell.
Also (and I meant to point this out too) my /var partition is on hdb, and
fsck gave it a clean bill of health. The reason I posted the problem to
linux-kernel is that I have never seen these errors with the pre-2.0.31-x
(x = 3 or 9, IIRC) and 2.1.29 kernels I run every day, nor with the other
development kernels up to 2.1.42 which I have tried, but have seen them
with 2.1.51 and 2.1.57. I hadn't applied any patches to 2.1.51, and
changed nothing but the sound driver in 2.1.57.

I hadn't thought of it before, but hda has only been in use for a short
time. I probably used mke2fs 1.10 to format all the partitions on it.
The reason I used an older mke2fs yesterday is that that's what I happened
to have on the Xdenu setup I use for stuff like this. The other
partitions on hdb were probably formatted with mke2fs 1.06. I don't
understand the details of what fs/ext2/super.c is doing, but I hypothesize
that it is only compatible with recent versions of your ext2 utilities
(thank you). I will try reformatting /dev/hdb1 with a recent mke2fs and
report on how it goes.
Trevor Johnson

Theodore Y. Ts'o wrote:

> Trevor,
> Given that both e2fsck and mount were comaplaining that all of
> your /dev/hdb* partitions didn't contain a valid e2fsck filesystem
> (at least until you reformatted /dev/hdb1), the first thing that I would
> be suspicious of is that your partition table might have gotten
> scrambled.