Hey Micah,

On Wed, Apr 19, 2006 at 11:51:12AM -0400, Micah Anderson wrote:
> After you've built and installed a loop-aes modules package you
> should probably be adding information to /etc/modules to load the
> loop module on boot, as this allows for loading of unattended
> filesystems on boot (such as swap partitions). It is necessary to
> load the module regardless as you need it to mount any
> partitions.

Yes, it's a bit non-optimal at the moment. The autoloading of
loop used to work before udev. Now only /dev/loop/0 gets created
automatically and loop mounts or swap on any other loop device
can no longer trigger the autoload. This is assuming you use
udev, otherwise it should all just work and we are looking at a
different bug :-)

> I'd suggest either adding a note in the README.Debian about how
> to update /etc/modules, or if it is possible to have it done
> automatically in the resulting .deb, thats even better.

I agree. I've been meaning to talk with upstream about how
autoloading could be improved on systems that use udev. I'm
considering two options ATM: One would be to ask the udev people
to extend the special case for loop to create /dev/loop[0-7]
automatically. This would not work for max_loop= larger than the
default though. The other idea is to have losetup, mount and
swapon just open() an existing loop device to trigger the
autoload before doing the regular setup. 

It makes sense to mention /etc/modules in the documentation 
until this gets improved though. Hopefully it will not be
required in the long run, otherwise I will need to check if it's
OK for loop to be added automatically in postinst or if that
would be against policy. Anyways, thanks for you suggestion. 
This is probably a problem many people come across.

cheers,
Max


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Reply via email to