Re: [Qemu-devel] [PATCH 3/3] ui: fix VNC websockets TLS integration

[Daniel P. Berrange]

On Mon, Mar 16, 2015 at 01:35:02PM +0000, Daniel P. Berrange wrote:
> On Mon, Mar 16, 2015 at 01:17:16PM +0000, Alex Bennée wrote:
> > 
> > Daniel P. Berrange <address@hidden> writes:
> > 
> > > The way the websockets TLS code was integrated into the VNC server
> > > made it insecure and essentially useless. The only time that the
> > > websockets TLS support could be used is if the primary VNC server
> > <snip>
> > >
> > > With this patch applied a number of things change
> > >
> > >  - TLS is not activated for websockets unless the 'tls' flag is
> > >    actually given.
> > >  - Non-TLS websockets connections are dropped if TLS is active
> > >  - The client certificate is validated after handshake completes
> > >    if the 'x509verify' flag is given
> > >  - Separate VNC auth scheme is tracked for websockets server,
> > >    since it makes no sense to try to use VeNCrypt over a TLS
> > >    enabled websockets connection.
> > >  - The separate "VncDisplayTLS ws_tls" field is dropped, since
> > >    the auth setup ensures we can never have multiple TLS sessions.
> > 
> > I wonder if the mechanical changes to the tls field could be separated
> > from the logic changes to the handling of authentication and certificate
> > checking?
> 
> They are rather intertwined, because the need for this duplicated
> TLS field was a result of the way auth was mishandled. So cleaning
> up one implies cleaning up the other & vica-verca.

I've actually realized I can split it, if I do the auth scheme clean
up first.


Regards,
Daniel
-- 
|: http://berrange.com      -o-    http://www.flickr.com/photos/dberrange/ :|
|: http://libvirt.org              -o-             http://virt-manager.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: http://entangle-photo.org       -o-       http://live.gnome.org/gtk-vnc :|