home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.unix.wizards
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!caen!uvaarpa!murdoch!murdoch.acc.virginia.edu!aj3u
- From: aj3u@larch.cs.virginia.edu (Asim Jalis)
- Subject: Does Unix Guarantee Limits on Delays
- Message-ID: <AJ3U.92Dec25152444@larch.cs.virginia.edu>
- Sender: usenet@murdoch.acc.Virginia.EDU
- Organization: University of Virginia, Charlottesville, VA 22903
- Distribution: comp
- Date: Fri, 25 Dec 1992 20:24:44 GMT
- Lines: 27
-
- The problem that I am working with is this. I want to time the
- interval between the occurrence of two mouse events. I want to
- have some confidence in the accuracy of my readings, independent
- of the number of other processes running on the system or their
- nature.
-
- Does Unix guarantee, either explicitly, or implicitly through the
- way it is designed, that a call to the clock daemon will return
- within a certain fixed upper limit amount of time? Possible
- sources of delays could be task-switching by the scheduler, or
- the clock daemon not responding immediately for some reason;
- there might be others that people here could suggest. Also,
- consider the same question but with this time querying about
- bounds on the execution of each ordinary instruction.
-
- If Unix does not provide such guarantees then I would like
- suggestions on how testing could establish a worst case upper
- bound on this delay. For example, I could have N1 infinite loop
- processes running and N2 processes that repeatedly call the clock
- daemon and N3 i/o processes etc. Are there any other classes of
- processes that could make a significant difference in how the
- scheduler and the clock daemon behave?
-
- Any help would be much appreciated.
-
-
- Asim Jalis.
-