Re: physmem, simple containers

[Niels Möller]

Marcus Brinkmann <address@hidden> writes:

> On Tue, Jan 27, 2004 at 12:52:08PM +0100, Niels Möller wrote:
> > But what about WORTEL_MSG_GET_CAP_REPLY? Despite its name, it is sent
> > by physmem to wortel (like all the other wortel requests). It seems
> > intended for having physmem give some capability to wortel. Exactly
> > what is it that physmem should give wortel?

> Yes, but it is the handle for the other task, like the task server.

Should it be a plain reference, or a control handle? It's not at all
obvious to me why wortel should need any task handles at all, and in
particular not control handles. This is under the assumption that it's
physmem, not wortel, that is responsible for starting the bootstrap
tasks.

On the other hand, the task server will need a capability that lets it
talk to wortel. For this to be setup, the local capability id must go
from wortel to physmem to the taskserver, and the global taskid of the
task server must be passed back from physmem to wortel.

This is a capability transfer *from* wortel to the task-server, via
physmem.

(I'm tempted to not treat wortel as a proper hurd task in the task
server. Then the task server would just mark the first few values for
the taskid as "reserved". Threads with reserved ids, i.e. wortel,
would then not be able to take part in the standard capability
transfer protocol, and furthermore, the manager capability will not be
sufficient for killing or manipulating wortel. If it helps to let
wortel be real task in the task server, we can of course do that, but
I'm not sure that really is necessary).

/Niels