From owner-freebsd-hackers Sat Dec 24 04:56:47 1994 Return-Path: hackers-owner Received: (from root@localhost) by freefall.cdrom.com (8.6.9/8.6.6) id EAA27521 for hackers-outgoing; Sat, 24 Dec 1994 04:56:47 -0800 Received: from hda.com (hda.com [199.232.40.182]) by freefall.cdrom.com (8.6.9/8.6.6) with ESMTP id MAA27515 for ; Sat, 24 Dec 1994 12:56:45 GMT Received: (dufault@localhost) by hda.com (8.6.9/8.3) id HAA18559; Sat, 24 Dec 1994 07:58:25 -0500 From: Peter Dufault Message-Id: <199412241258.HAA18559@hda.com> Subject: Real Time Extensions on -current To: freebsd-hackers@freefall.cdrom.com Date: Sat, 24 Dec 1994 07:58:25 -0500 (EST) X-Mailer: ELM [version 2.4 PL24] Content-Type: text Content-Length: 1868 Sender: hackers-owner@freebsd.org Precedence: bulk Is anyone else working, or thinking about working, or wanting to work, on real time extensions for FreeBSD or running a Mach kernel in a FreeBSD context? I can identify two paths to follow: 1. Use the mk83i real time mach distribution together with the Helsinki University of Technology Lites work to get a real time mach environment working in the FreeBSD context. 2. Use orthogonal kernel extensions to implement enough of a real time environment to do useful work, and gradually extend it. The kind of thing I have in mind here is two level interrupts to reduce latency, seeking out anywhere interrupts are off, adding a real time schedular and a real time process class, add a "real time fault" that would kick in whenever a process did anything not real time safe (non-real time system safe system call, single level interrupt, alarm call during real time, etc). Initially you could write real time drivers that would execute in the kernel preempting normal kernel activity (thus requiring the orthogonality) and able to signal and ready real time preocesses. I'm working on "1.". The knowledge required is beyond what most BSD hackers have and the startup overhead just to get to the point of knowing everything that is involved is fairly great. "2" is a hack but easier to get your hands around and you can get to the point that you have something usable quickly (I actually did a hack version back in the patchkit days and used it successfully in some control experiments we did). I'd appreciate thoughts, knowledgable comments, and information from anyone who is interested in this subject. I'll save up names. Peter -- Peter Dufault Real Time Machine Control and Simulation HD Associates, Inc. Voice: 508 433 6936 dufault@hda.com Fax: 508 433 5267 ++++ New e-mail address. E-mail problems? Tell hdslip@iii.net