On Fri, 16 Mar 2007 10:27:57 +0100
Franklin PIAT <fpiat@bigfoot.com> wrote:
> I recently tested Etch on a few laptops and I noticed the uswsusp resume_device was always
> set to long device (devfs like) names.
>
> I first assumed It was a well known bug for Etch installation, but it doesn't seems so. Is it?
> Do you think it means that it will prevent hibernation to work by default for Etch on laptop ?
>
> My understanding is that :
> - uswsusp is installed under the D-I environement ('Laptop' task depends on it).
> - The D-I environnement uses devfs long names.
> - so uswsusp configures swap device using devfs.
> - therefore s2disk doesn't work.
when booted to the regular debian kernel which uses `normal' device names
This is correct. In only recently found this out through a bugreport.
Because we're already so far in the freeze and this behavior is going
away in the future anyway (I've been told) and didn't want to fix this.
Also I though it would involve some kludge which was likely to break, but ..
> Would it be better to:
> - add a hook in the current package, in order to detect that we're in D-I, then use di's mapdevfs.
.. I didn't know mapdevfs existed... I think I can prepare a simple
workaround based on this. Something along the lines of
if [ -x /bin/mapdevfs ]; then
newpath=`mapdevfs $oldpath`
fi
That is if mapdevfs does what I think it does;)
Release team: would an upload to t-p-u with a interdiff which will
roughly look like this be acceptable? It would basically hit everybody
that installs etch on a laptop, so could be interpreted as RC.
grts Tim
Attachment:
signature.asc
Description: PGP signature