[coyotos-dev] IPC Redesign

[Jonathan S. Shapiro]

On Tue, 2007-05-22 at 12:24 -0700, Charles Landau wrote:
> >Setting that aside, fully transparent forwarding is not possible in
> >principle, because you need to know what the call/response protocol is,
> 
> Excuse me, but didn't the long discussion on cap-talk about membranes 
> and such come to the opposite conclusion? And doesn't the Horton 
> protocol involve a transparent forwarder? The only thing you need to 
> know about the protocol is where the capabilities are.

No. The long discussion on cap-talk came to the conclusion that E-style
membrane wrapping is not feasible in an OS-style capability system,
because every act of wrapping requires storage allocation

> >All that being said, my view is that the adoption of IDL strongly
> >reduces the importance of this issue. Given an IDL specification, it is
> >possible to automatically generate a forwarder for any given interface.
> 
> Yes, but for a *transparent* forwarder (assuming such a thing is 
> possible), I don't see how CapIDL can help. And that is a pity, 
> though perhaps not such a large one.

Let me restate: in the presence of IDL, I do not consider a transparent
forwarder to be a high-priority objective.

That said, it is *definitely* important that a forwarder which knows the
alleged IDL be able to detect messages that fail to conform to the
proper payload shape. This is necessary (e.g.) for MLS enforcement.


shap