Subject: Re: More fsck problems?
To: Martin Cernohorsky <email@example.com>
From: Luke Mewburn <firstname.lastname@example.org>
Date: 09/24/1997 19:24:28
"Martin Cernohorsky" writes:
> On Sep 24, 2:52, Dave Huang wrote:
> > I tried booting single user and fsck-ing /dev/rwd0a, and fsck asked "MARK
> > FILE SYSTEM CLEAN?" After I said yes, it says "FILE SYSTEM MARKED CLEAN".
> > So far so good... I try to fsck /dev/rwd0a again, and fsck says "File
> > system is clean; not checking". Still good. But when I "mount -u /", I get
> > the "file system not clean; please fsck(8)" message. Not good :(
> I see the same problem on my machine. Worth noting is, that fsck
> on the block device (ie. "fsck /dev/wd0a") fix it!
I've just committed a patch which #ifdef's out the bits which
were added in lite2 that are the likely culprits. they were:
file what the lite2 merge added that i've now disabled
main.c replaced ckfini(1) with hotroot() check and
then based args to ckfini() on whether fs == "/"
comment: This is probably the culprit...
setup.c sets skipclean = 0 if the superblock can't be read
(just before the alternate superblocks are tried).
comment: not sure about this
utilities.c saves and restores fsmodified flag around flushing
of superblock to disk
comment: again, not sure about this
each bit of code (wrapped in #ifdef LITE2BORKEN) can then be separately
verified by the FS gurus, and the appropriate code chunk used.
for now, functionality should be as per before the merge of lite-2.
once again, sorry about any inconvenience.