On Mon, 2006-04-24 at 22:09 +0200, Tom Bachmann wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Marcus Brinkmann wrote:
> > I don't think there is a general
> > answer for M to the question when it needs to recover.
> > 
> 
> can't C just tell M?

I think that there must have been a private message here, or perhaps I
have not received Marcus's message yet.

The answer in general is "no", for two reasons:

1. C may not be entitled to know what M will do. It is hard to predict
how to recover from unknown actions.

2. It would make the interface impossibly complicated.


This is why I was very careful in my description to say that "the
recovery boundary is between M and S". This means that C trusts M fully
to recover in whatever way is appropriate (i.e. C and M fail as a unit).

If we are also concerned that M may fail to respond to C, then it is the
obligation of C to implement a recovery strategy.

The point of using three parties in my example was to illustrate that
all recovery boundaries are IPC boundaries, but not all IPC boundaries
are recovery boundaries.

shap



_______________________________________________
L4-hurd mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/l4-hurd

Reply via email to