Hi,
On an old-fashioned Unix scheduler, each timer interrupt increments
the value p_cpu in the process structure of the running process. For
each time slice, a new priority p_pri is computed as a function of
mainly p_cpu and p_nice, the cpu usage and nice value.
On SVR4 systems this follows a table with one entry per priority and a
value telling what priorities the process is given if it uses up all of
the quantum assigned at this level (punished, then) or if it yields the
processor before end of the quantum(rewarded, then).
It is my opinion that a timesharing process that waits long, then runs
on the processor for a period of time that is less than the quantum will
not be preempted by other timesharing processes. Its timesharing
priority will be high as it waited long.
Is this true ?
How can I get to know the value of the timesharing quanta on DU (since
they dont have the dispadmin(8) command) ?
--
Lars Bro lbr_at_dksin.dk
ABB Daimler-Benz Transportation Signal A/S +45 36 39 01 77
Received on Thu Sep 17 1998 - 08:22:04 NZST