Re: Hurd on L4

[Marcus Brinkmann]

On Mon, May 12, 2003 at 05:29:29PM -0300, Fortes Marcelo wrote:
> 1) have some server of Hurd running on top of L4?
> or it needs to be translated to synchronous message
> passing yet?

So far there are only design concepts, and partial implementations of some
features we will need.  We don't need asynchronous messages, but there have
been other details to be worked out.  Most issues have been clarified by
now (at least roughly), though.  We are far from having anything useful
running on it.

> 2) Hurd is almost using 100% Pthreads instead cthreads
> in source code how does it impact over you work?

The Hurd uses cthreads, but people worked on porting it over to pthreads. 
The Hurd on top of L4 will use pthreads.  Pthreads will be the only high
level thread library on the Hurd in the future.
 
> 3)There are some ideas of Hurd on top of Mach that can
> be subverted by port it to run on top of L4?

Is that a question?  Mach IPC is almost network transparent, while L4 IPC is
highly not.  So I guess that getting RPCs over a network will be a tad more
difficult on L4 to get right.  There are other things like that, but nothing
fundamental.
 
> 4)About Unix compatibility how much it will be
> Unix-Like OS running "SCO Unix" binaries for exemple?

We have no interest in running proprietary software or having compatibility
with proprietary systems ourselves.  What could happen is GNU/Linux
compatibility at some time because that would be quite easy because we use
the same tools and glibc.

In general, L4 seems to allow system call emulation by the exception handler.
So as long as you get exceptions for everything you want to emulate, you
will be, at least conceptionally, be able to do so.  This is not different
from Mach.
 
> 5)Can Someone please explain-me about the Hurd policy
> of "user ring 0" or a non root administrator OS with
> full power over the system? icannot understand a Unix
> system with user 100% independent of super user...
> (maybe i have not understooded this concept)

I hope you have read the documentation on our web site.  The authentication
mechanism allows communication between untrusted users and untrusted
servers.  All system functionality is implemented by cooperating, but
independent servers.  Users can add their own servers, and use them instead
the system servers if they want.  This way they can change the personality
of the system without affecting anybody else.  The authentication mechanism
guarantees that a user can not compromise the security of the overall system
or other users in particular while this is done.

User ring 0 is taken by L4.  The Hurd itself does not run with any kernel
priviledges (it runs entirely in user space).  However, some components of
the Hurd needs to run with privileges that a normal user doesn't have (like
the privileges to control arbirtary threads and tasks, and to access the
devices).  This functionality is provided by the kernel (Mach had special
privileged ports, L4 has the rootserver threads).

Thanks,
Marcus


-- 
`Rhubarb is no Egyptian god.' GNU      http://www.gnu.org    address@hidden
Marcus Brinkmann              The Hurd http://www.gnu.org/software/hurd/
address@hidden
http://www.marcus-brinkmann.de/