Help with Luks devices

Dan Nicholson dbn.lists at gmail.com
Mon Aug 7 16:00:21 PDT 2006


On 6/24/06, Dan Nicholson <dbn.lists at gmail.com> wrote:
>
> I've set up a Luks encrypted ext3 filesystem on a USB key.  Using
> luks-setup from Michael Petullo's luks-tools, I can get the volume
> mapped and mount the device in /dev/mapper.  My problem is that I
> don't have a /dev/dm-* device, and this seems to be what HAL is
> looking for to perform the equivalent actions.
>
> Since the /dev/dm-* nodes aren't handled by stock Udev, I was
> wondering if someone could explain the userspace tools needed.  I have
> current device-mapper, cryptsetup-luks and hal-0.5.7 on linux-2.17.
> I'm on a home brewed system, and I can't figure out what the
> additional steps are.  I've groveled Fedora Rawhide for a while, but
> can't find the magic bullet.

I figured out my problems with the HAL Luks setup a while back. Now to
see if I can get any feedback this time...

I changed my udev rules so that the /dev/dm-* nodes were created. Why
wasn't I doing this before? I have a Linux From Scratch system. The
guy who wrote our udev rules, who uses LVM and is a very smart guy,
says that creation of the /dev/dm-* nodes triggers a bug in LVM.
Normally this is fine since if you use the device-mapper tools on
their own, these nodes are unnecessary. Indeed, looking at the udev
rules distributed in the udev-096 tarball, some of the distros are
creating the nodes and some aren't. RedHat is one that allows the
nodes to be created.

Now, I know that the proper fix would be for the device-mapper people
to fix up their driver to work properly with udev and sysfs. In the
meantime, however, I'd like to ask what the drawback is to finding the
device-mapper info from sysfs /sys/block/dm-*? I fully admit that I
know very little about udev/sysfs/HAL, so I might be overlooking
something obvious.

I found these posts in the archives which may be relevant.

http://lists.freedesktop.org/archives/hal/2004-September/001047.html
http://lists.freedesktop.org/archives/hal/2004-December/001528.html

Thanks.

--
Dan


More information about the hal mailing list