[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]
Re: [e2fsprogs-1.24] "fsck -A -a" fails on reboot
- From: Theodore Tso <tytso mit edu>
- To: Michael Renner <robe amd co at>
- Cc: Theodore Tso <tytso mit edu>, ext3-users redhat com
- Subject: Re: [e2fsprogs-1.24] "fsck -A -a" fails on reboot
- Date: Fri, 7 Sep 2001 19:18:56 -0400
On Fri, Sep 07, 2001 at 09:45:17PM +0200, Michael Renner wrote:
>
> Sorry if i explained my problem too fuzzy (eek, somebody give me a
> thesaurus), i never had any problems with the root-partition... it just
> complains that the /boot is mounted; here are my tst_ismounted results
> (which where executed right after the sulogin-login (eek again).
>
Ah, OK. People have been really bad about giving me bug reports. OK,
so the problem is Slackware's boot scripts. It *should* be doing one
of two things after checking the root filesystem:
(a) Remounting the root filesystem read-write, and then making
/etc/mtab sane.
(b) Mounting /proc
You can do either one of these things. Most distributions do one or
the other; some do both. I didn't anticipate that a distribution
could be quite that stupid to not do either.
The next question then is whether I want to work around this, or
simply tell Slackware users to file a bug with their distribution.....
Fundamentally, there's nothing I can do about this. If /etc/mtab
isn't accurate, and /proc is present, there's no way I can tell
whether or not a particular partition is mounted read-write or
read-only. It's possible that the root could be mounted read-only,
but other filesystems are mounted read-write. So the blanket check
that was in older e2fsprogs really was a dangerous thing.
If I put in the workaround at all, it will be as a configure option:
--enable-dangerous-workaround-for-stupid-distributions
:-)
- Ted
[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]