Subject: kernel filesys/fsck warnings
To: None <>
From: Robert Elz <kre@munnari.OZ.AU>
List: current-users
Date: 01/28/2004 02:46:30

If I boot a 1.6ZH kernel (and I assume ZI) on my system, it moans about ...

  WARNING: possible botched superblock upgrade detected
  on filesystem previously mounted on /var/tmp
  fs_bsize == fs_maxbsize (0x00002000) but FS_FLAGS_UPDATED is not set
  Test your filesystem by running fsck_ffs -n -f on it.
  If it reports:
  you should be able to recover with fsck_ffs -b 32 -c 4
  See the file src/UPDATING or  
  for more details

(For essentially all the filesystems I believe, here I just picked one).

Running fsck (fsck_ffs) as instructed says nothing at all, no comments
about superblock disagreements, nor any other problems.

This is a 1.6X fsck - the -current fsck won't compile, I assume
because I haven't done "make includes" (because I don't really want
to upgrade to 1.6ZH or later - this kernel give me the infamous
"lost interrupt" stuff on the IDE controller that many people have
noted, so while I've tested this one, I'm not planning on running it
very much).

UPDATING doesn't say much more than the kernel does.

The e-mail reference from the archive claims ...

  This should only affect people who migrated from a kernel before
  20030402 directly to a kernel from the time period
  between 20030913 (revision 1.120 of ffs_vfsops.c)
  and my changes of 20040109 (revision 1.131 of ffs_vfsops.c)

which isn't my system.   This system didn't exist 2003-04-02, it
was just recently installed using a 1.6X base system (the -current
system I'm happiest with at the minute, it is very stable - I assume
1.6Y would have been as good, but it only lasted a day, and I missed it).

To handle the hardware I have properly, it is running a 1.6ZF kernel
(most of the time) though (that one doesn't complain).

The 1.6X is from 2003-08-29, which isn't before 20030402, but is
before the "bad" (or perhaps worst) ffs_vfsops.c (according to the
above note).   The 1.6ZF was from 2003-11-23 which is right in the
middle of the "bad period".

Do the created filesystems need some kind of fix, or is the kernel
warning bogus?   If I were able to compile a new fsck (such things
can be managed if really required), would that detect the problem
and fix things?   In that case, I assume I'd need a new newfs as well?