On Fri, Jun 15, 2007 at 09:54:13AM +0200, Karel Zak wrote:
> On Fri, Jun 15, 2007 at 12:28:18AM +0100, Nix wrote:
> >
> > lrwxrwxrwx 1 root root 3 2007-06-14 22:53 /dev/cdrom -> hdf
> >
> > /dev/cdrom /mnt/cdrom iso9660 defaults,ro,user,nosuid,nodev,noauto 0 0
> >
> > Now, we see this:
> >
> > $ mount /mnt/cdrom
> > $ grep cdrom /etc/mtab
> > /dev/hdf /mnt/cdrom iso9660 ro,noexec,nosuid,nodev 0 0
> > $ umount /mnt/cdrom
> > umount: /mnt/cdrom mount disagrees with the fstab
> >
> > Those new /etc/mtab contents are hardly ideal: they're the same as
> > /proc/mounts now, but it's definitely a bit of a loss of information,
>
> Yes, the new mount version uses canonicalized paths in mtab.
>
> > It's late and I'm hay-fevery so I haven't tracked down the actual bug
> > yet. I'll do it tomorrow if nobody else does (but honestly I expect
> > someone else will get to this first, it's probably really easy to fix).
>
> The problem is that fstab.c: getfs_by_specdir() doesn't convert
> paths from fstab to canonical paths.
No.. there was my typo:
streq(canonicalize(mc->m.mnt_fsname), dir)
^^^^
must be 'spec'
I'm going to rewrite all this function. There is too much memory
leaks for my eyes...
I've added a new regression test for this issue. Thanks again! We
need more and more testers.
Karel
--
Karel Zak <kzak@xxxxxxxxxx>
-
To unsubscribe from this list: send the line "unsubscribe util-linux-ng" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
