Re: L4-Hurd; wortel's missing modules; was: (autoreconf errors)

[Niels Möller]

Johan Rydberg <address@hidden> writes:

> When it comes to the other core services, there has only been done
> work on wortel (the root server.)  Niels has said that he wants to
> write the task server (but I'm not sure that is correct any longer.)

I'm still interested in the task-server, but since I haven't been
doing any hacking on it lately I don't want to stop anybody else from
continuing that work, in particular as I still think that the
task-server is needed early on the boot-sequence, and is important for
getting other work done. (Most of my free hacking time the last few
months have been spent on a new subquadratic gcd implementation for
GMP).

> : As far as how to setup the initial tasks, why has there not been
> : agreement? Is it political, or is it that no one really has a good idea?

> What I was talking about is what information wortel should provide to
> the bootstrap servers, so that they can get on their feet.  It is not
> trivial task to boostrap a multiserver operating system.  For example,
> physmem can not create any threads before the task server has been
> started, and the task server can not operate without physmem (since it
> is in control of all physical memory.)

To me, it seems logical that wortel first starts the task server. For
this to work, the task server has to do with statically allocated
memory, at least initially, which shouldn't be too hard (just have
statically allocated structures that suffice for 5 tasks and 17
threads or whatever is needed).

Next, wortel can use the task server to start the physmemserver.

Finally, wortel has to do give the available memory to physmem, and do
some handshaking to let the taskserver use the physmem server.

Some if this may get easier if the task-server or physmem-server are
threads in wortels address space, or if the task server is written to
only ever use only static memory allocation.

/Niels