DBus in the kernel?

[Kimmo Hämäläinen]

On Tue, 2010-01-05 at 15:28 +0100, ext Ville M. Vainio wrote:
> On Tue, Jan 5, 2010 at 3:55 PM, Daniel P. Berrange <dan at berrange.com> wrote:
> 
> > NB, the daemon does more than just pass data between the clients, in
> > particular it enforces the security policy for what signals/methods/
> > interfaces/ etc each client is allowed to access. In a peer-to-peer
> > model you'd need to push the security policy out to each client
> 
> Would handling that level of security policy on dbus client process
> (through libdbus) really be that bad? We'd assume, of course, that
> we'd already have received sufficient peer credential information from
> the dbus-daemon that handed us the socket in the first place.
> 
> I guess there might be a document somewhere explaining why the current
> approach is necessary, but I can't help thinking the current approach
> trades performance for simplicity and proven code.

I think this would be doable, but the idea behind the kernel approach is
to boost the performance to a new level also. Having it inside kernel
brings up new possibilities of communicating with drivers and stuff, I
guess.  Maybe the next step is to have hardware sending D-Bus messages
around :D

-Kimmo