itojun@iijlab.net writes:

> >> > itojun@itojun.org (Jun-ichiro itojun Hagino) writes:
> >> > > 	i see the following on dmesg (for / and /usr partition),
> >> > > 	however, i do not see "VALUES IN SUPER BLOCK ..." message on
> >> > > 	fsck_ffs -n -f.  fsck_ffs -b 32 -c 4 had no effect.
> >> > > 	how can i suppress the message?  do i have to start over from scratch
> >> > > 	(i.e. newfs)?
> >
> >> 	i can supply any info you would need, so just let me know what you need
> >> 	to know about my filesystem.

> >Ok.  I don't understand why 'fsck_ffs -b 32 -c 4' didn't set the
> >FS_FLAGS_UPDATED bit in fs_old_flags, which should cause the kernel
> >to quiet the error.
> >
> >Can you please send me the output of a -current 'dumpfs -as' and
> >a transcript of running 'fsck_ffs -d -b 32 -c 4'
> >(The -d option just provides a little extra debugging information)
> 
> 	somehow dmesg warning went away.  i dunno why.
> 	i'll send you dumpfs output anyways.
> 

I've looked at the dumpfs output you sent me, and there is still a
problem.  I don't know why, but all of your alternate superblocks have
been updated but the standard superblock is not. This could have
happened if you used the -p flag or if you answered no to
the UPDATE STANDARD SUPERBLOCK question.  Alternately, I wonder
if there is a missing ckfini() call somewhere in fsck_ffs.

We should probably fix fsck_ffs to disallow combining -p and -b.
Alternately, we could fix fsck to just do the standard superblock
update if both -b and -p are specified.  Skipping it completely
is wrong.

You might be able to fix it with 'fsck_ffs -b 16 -c 4' which will
tell it to pretend that the normal superblock is the alternate
superblock and use that to do the update.

Hm.  It would be useful for me to reproduce this case myself
so I can take a close look at how the filesystem got in this state.

Darrin