DBus in the kernel?

[Daniel P. Berrange]

On Tue, Jan 05, 2010 at 03:51:21PM +0200, Ville M. Vainio wrote:
> 2010/1/5 Kimmo Hämäläinen <kimmo.hamalainen at nokia.com>:
> 
> >> That being said, can't we get zero-copy already with peer-to-peer dbus
> >> connection? It seems to be somewhat unsupported (or not widely used at
> >> least), but I guess it has problems of its own.
> >
> > I think that works so that the other client is actually working as a
> > D-Bus server, or maybe there is some more convenient way?
> 
> I think (relying on random docs here, I've never used it, because I
> assumed there is a good reason for nobody using it ;-) peer-to-peer
> dbus connection just sets the connection up using the local socket
> path directly.
> 
> The way you'd *want* dbus to operate would be direct writing to a
> peer-to-peer socket at all times, and only going for the daemon when
> that socket dies (to get a new name mapping & socket).

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

Daniel
-- 
|: http://berrange.com/     -o-    http://www.flickr.com/photos/dberrange/ :|
|: http://libvirt.org  -o-  http://virt-manager.org  -o-  http://ovirt.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: http://freshmeat.net/~danielpb/    -o-   http://gtk-vnc.sourceforge.net :|