Re: Reliability of RPC services

[Jonathan S. Shapiro]

On Tue, 2006-04-25 at 18:58 +0200, Pierre THIERRY wrote:
> Scribit Jonathan S. Shapiro dies 25/04/2006 hora 12:29:
> > You do not even need to do that. If you have used the reply capability
> > protocol as intended, all you need to do is increment the protected
> > payload that is stored inside the FCRB.
> 
> But it has two drawbacks:
> 
> - checking that the capability is still usable seems to me to be much
>   complicated, if possible at all

This check is already necessary in the specification.

> - IIUC, this check would need to page in the FCRB, which is bad

So does severing the FCRB.

> With the severing option, the check is lightweight and can be done while
> the FCRB remain paged out, if it has been.

Actually, severing requires an interprocess call to the storage
allocator. It is *much* more expensive!

shap