home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.realtime
- Path: sparky!uunet!spool.mu.edu!agate!linus!linus.mitre.org!pacific.mitre.org!drodman
- From: drodman@pacific.mitre.org (David B Rodman)
- Subject: Re: real_time systems
- Message-ID: <1992Dec16.213018.12351@linus.mitre.org>
- Sender: news@linus.mitre.org (News Service)
- Nntp-Posting-Host: pacific.mitre.org
- Organization: The MITRE Corporation, McLean, VA, USA
- References: <199216.4481.2332@dosgate>
- Distribution: comp
- Date: Wed, 16 Dec 1992 21:30:18 GMT
- Lines: 20
-
- >ZS>Is it true to say that, preemption is an essential characteristic of
- >ZS>a real_time system? if not how can we make sure that all tasks will
- >ZS>meet their deadlines?
-
- Elliot Chaplin>If a cooperative roundrobin won't fill the bill, a watchdog timer can be
- Elliot Chaplin>added. See an article by Jack Woehr in DrDobbs maybe a year ago.
- >---
-
- Is preemption an essential charateristic of a real time system? No, consider a
- system where the deadline of every task is longer than the utilization time of
- that task plus the latency of every other task executing... as in the roundrobin
- scheduling suggested by Chaplin.
-
- What you need to investigate is Rate Monotonic Analysis, it is a way to look at
- tasks utilization, their deadlines, and the blocking they have by equal priority
- (round robin scheduling) and higher priority tasks (preemptive priority scheduling).
-
- David Rodman
- drodman@mitre.org My opinions are my own, you can't sue my employer for them and
- I don't have any money.
-