> Greetings... > I have some test code that sends a fifo message > from an RTLinux module to a Redhat process, which > returns it. I find that the routrip time is > typically around 10 usec, which is fine. But... > I get periods of time where the roundrip time spikes > up to 100 msec (note the 'm'). Can anyone explain > this? Is 100 msec perhaps a settable timeslice? > Thanks for any help. > thats fine - your linux task is a non-rt process and that can be delayed conciderably by any system process, lets say updatedb or the like, if you want to improve the average behavior you can run the process with higher priority (SCHED_FIFO,SCHED_RR) as root, but the worst case will not improve. If you need the linux process to answer within a given time with a worst case response then you splitt your task along the wrong line - all processes wich require guaranteed worst case response time should be rt-threads not linux processes.
hofrat _______________________________________________ Rtl mailing list [EMAIL PROTECTED] http://www2.fsmlabs.com/mailman/listinfo.cgi/rtl
