On Thu, Sep 06, 2007 at 06:05:28PM +0200, Matthias Koenig wrote:
> 1. Setting up the loop device does not canonicalize the file name
> # losetup /dev/loop0 /tmp/volume
> # cd /tmp; losetup /dev/loop1 volume
> # cd /tmp; mkdir foo; cd foo; losetup /dev/loop2 ../volume
> # losetup -a
> /dev/loop0: [0803]:523558 (/tmp/volume)
> /dev/loop1: [0803]:523558 (volume)
> /dev/loop2: [0803]:523558 (../volume)
Yeah, canonicalize() makes sense. Go ahead.
> # mount -oloop /tmp/volume /mnt/
> # mount -oloop /tmp/volume /mnt/
> # mount
> /tmp/volume on /mnt type ext2 (rw,loop=/dev/loop0)
> /tmp/volume on /mnt type ext2 (rw,loop=/dev/loop1)
I know about this problem and it's "fixed" in Fedora util-linux
package (almost for two years)...
... but I didn't port the patch to util-linux-ng, because I'm still
not sure that this is a good idea.
Thanks for your note about offsets -- it's a good reason to drop the
stupid Fedora patch.
Now other problem comes to mind -- hardlinks... Max's hint with
lo_node is good.
> Any thoughts about this or something I have missed?
Doesn't have kernel all information? Why this is not disabled by
kernel? Why this issues should be resolved by an userspace util when
kernel supports it? Is it real problem for all filesystems? Doesn't
exist same issue with multipath devices (SANs)?
Maybe we can add a warning message to losetup only. The losetup
shouldn't be more "smart" than kernel or end-user -- I'd like to
follow kernel in this case.
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
