Date: Wed, 29 Mar 1995 04:39:49 -0800 (PST) From: Fernando Astorga <fernando@cea.Berkeley.EDU> To: freebsd-hackers@FreeBSD.org Subject: FreeBSD support for realtime systems Message-ID: <199503291239.EAA04745@halfmoon.cea.berkeley.edu>
next in thread | raw e-mail | index | archive | help
I was wondering if anyone out there has had previous experience with running real-time systems on a UNIX environment. But first, here's a description: A group in Mech. Eng. department here has developed a real-time scheduler to control a set of machines in the lab. Well, since they needed control of interrupts to execute tasks at a set time, their first implementation was in DOS and used Visual C++. As you can imagine, there's been several problems with this set-up, especially with DOS crashes as students try writing new programs to control these machines. The first problem is VC++, but that is being fixed by switching to Borland C++ which is a better compiler. The next is a bit more difficult. They want to move this system to run on a more robust OS, so they've been considering OS/2 and NT. But these systems have fixed time slices, making the job of the scheduler more difficult, if not impossible, i.e. if the interval for a task is less than the time slice. I think OS/2 has a time-slice of 20ms, way too slow for some of these applications. I'm pushing for porting this scheduler to UNIX, preferably FreeBSD, since it will run on the lab's PCs. But there are some issues that I'm not too familiar with, esp. UNIX's time-slice/process priority system. Can someone who has had experience with real-time systems on UNIX provide some info on how it can be done? Thanks. Fernando Astorga fernando@cea.berkeley.edu
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199503291239.EAA04745>