Re: NMOUNT/OMOUNT midway status

From: Ruslan Ermilov <ru_at_freebsd.org>
Date: Tue, 7 Dec 2004 00:54:33 +0200
Hi,

On Mon, Dec 06, 2004 at 10:10:13PM +0100, Poul-Henning Kamp wrote:
> 
> I have now converted all of the filesystems which are not rootfs
> candidates under the old world order[1] (nfs, cd9660 and ufs).
> 
> There shouldn't be any visible effects of this work seen from
> userland: the old mount_*(8) binaries and mount(2) system calls
> should all just work like they used to.
> 
> If they don't, try to take a peek yourself, it's pretty simple
> after all.  If you can't spot the mistake I've made, send me
> email.
> 
> The next step involved converting the three 'big' filesystems and
> the root filesystem mount code, and I have not quite found out
> how to partition that into a sequence of sensible commits.
> 
> If you want to help me test I keep a patch relative to -current
> up to date:
> 
> 	http://phk.freebsd.dk/patch/nmount.patch
> 
> Poul-Henning
> 
> 
> [1] In the new worldorder all filesystems which understand a "from"
> argument is a root filesystem candidate, that means that msdosfs,
> ntfs, hpfs and ext2fs with no or only minor tweaking can be used
> for root filesystem.
> 
Are you aware of the following problem?  On a recent
-CURRENT system, remounting / by fsck(8) is not possible:

	mount reload of '/' failed: Invalid argument

This has been triggered by an unclean shutdown due to a
panic, but can also be reproduced easily by booting or
shutting down into single-user mode, having / mounted
read-only, and typing "fsck /".


Cheers,
-- 
Ruslan Ermilov
ru_at_FreeBSD.org
FreeBSD committer

Received on Mon Dec 06 2004 - 21:54:36 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:24 UTC