To: kaashoek@lcs.mit.edu
Subject: Thesis update and questions
From: nathanw@mit.edu (Nathan J. Williams)
Date: 30 Aug 2000 15:36:59 -0400
Message-ID: <mtu7l8yee1w.fsf@contents-vnder-pressvre.mit.edu>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Lines: 51
Xref: contents-vnder-pressvre.mit.edu mit.thesis:23


Hi. I realized we haven't touched base all summer, but I've been
working along on the scheduler activations implementation for NetBSD
and making substantial progress. 

 * General status update

   - I defined a user/kernel interface for scheduler activations,
     based principally on the Mach interface. It went through a couple
     of rounds of revisions with some other NetBSD developers who are
     doing work with threading and schedulers. 

   - I went through the NetBSD kernel and implemented a LWP/process
     separation, distinguishing process state from execution context
     state, laying the groundwork for having multiple execution states
     (LWPs or SAs) per process. (In the interest of time, I have only
     implemented the machine-dependent portion of that code for Intel
     systems). That code is currently running and surprisingly stable.

   - I am currently testing the multiple-LWP code, integrating the
     code for user-level representation of thread context
     (ucontext_t), and preparing to implement the scheduler
     activations system calls themselves.

 * Office space?

   Since I'm no longer working for anyone in the AI lab, they'd like
   to reclaim the space I'm in for incoming students. I'm not sure
   exactly when I have to be out; the first round of new students
   seems to leave space for me here.

   Is there office space available with your group? It'd be good to
   be around your group more, too.


 * USENIX

   I'd definitely like to present a paper on my work at next summer's
   main Usenix conference, on the Freenix track, and the summary
   submission deadline will come up quickly (November 27). 

   Also, I saw that OSDI is coming up and that you're involved with
   it. I'd like your advice as to whether it's worth attending and
   perhaps giving a work-in-progress presentation. (I even know a
   dirt-cheap place to stay in San Diego, from my Usenix trip this
   last June).


What would be a good time to get together and talk about these things?

        - Nathan
