From owner-freebsd-hackers  Mon Feb  3 07:15:03 1997
Return-Path: <owner-hackers>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id HAA02558
          for hackers-outgoing; Mon, 3 Feb 1997 07:15:03 -0800 (PST)
Received: from hda.hda.com (ip95-max1-fitch.ziplink.net [199.232.245.95])
          by freefall.freebsd.org (8.8.5/8.8.5) with SMTP id HAA02546
          for <hackers@FreeBSD.ORG>; Mon, 3 Feb 1997 07:14:57 -0800 (PST)
Received: (from dufault@localhost) by hda.hda.com (8.6.12/8.6.12) id JAA19438; Mon, 3 Feb 1997 09:48:58 -0500
From: Peter Dufault <dufault@hda.com>
Message-Id: <199702031448.JAA19438@hda.hda.com>
Subject: Re: Jukka Ukkonen: POSIX.4 - scheduler once more (as you requested)
In-Reply-To: <199702030234.NAA01680@genesis.atrad.adelaide.edu.au> from Michael Smith at "Feb 3, 97 01:04:14 pm"
To: msmith@atrad.adelaide.edu.au (Michael Smith)
Date: Mon, 3 Feb 1997 09:48:58 -0500 (EST)
Cc: dufault@hda.com, jkh@time.cdrom.com, hackers@FreeBSD.ORG
X-Mailer: ELM [version 2.4ME+ PL25 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-hackers@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

> Peter Dufault stands accused of saying:
> > One quick observation:
> > 
> > > X * 5. The source code must be available for anyone who wishes to have it.
> > 
> > What do people think of packaging this up as a user library against
> > an LKM'd pseudo /dev/realtime driver?  I have the skeleton to do
> 
> How does this help the posix.4 scheduling model though?  Does having it as
> an LKM still allow this to work "as expected"?

I have an LKM skeleton that includes scheduling.  I have to decide if I like
it or not.  Everything funnels through an ioctl, though it is set up to be
installed as system calls if that is preferred.  I've been thinking about a
dispatch through system calls that is redirected from the default if you have
the device open since I think that works as the best of both worlds.

> > that.  My reasoning is I'd like to be able to have different realtime
> > facilities, for example, process migration to an attached embedded
> > processor that would "fault" back as soon as you tried to do
> > something in that environment.  It also gives you a way to have
> > realtime user or group protection.
> 
> Sounds reasonable (and interesting).  Do you have more words on this 
> somewhere?

I have too many words on it.  Maybe at the end of the week I'll proofread it and
put it someplace public.

-- 
Peter Dufault (dufault@hda.com)   Realtime Machine Control and Simulation
HD Associates, Inc.               Voice: 508 433 6936