Dear Developers!
I have introdued a new SCHEDULING POLICY in my linux kerenel-2.4.26.
My scheduling algorithm is different than SCHED_OTHER policy in following ways?
1. Time slice is 100 ms in place of 50 ms. (kernel/sched.c TICK_SCALE
is redefined)
2. Stack limit is 16 mb in place of 8mb.(linux/sched.h #define
_STK_LIM (16*1024*1024) )
3.Max_COUNTER is for 400ms (linux/sched.h)
4.I am giving more weight to process running in current data segment.
( if (p->mm == this_mm || !p->mm)
weight += 5; /* I am giving significant advantage to the current MM */ )
5.i am giving significant weight to SCHED_MY process so even a real
time process may loose slight advantage.so in forking processes change
scheduling policy of one process as SCHED_RR.It may be good to
generate similar programme with four different SCHED algorithm and
compare their time got by struct rusage fields.
6.It leads to preemption even the current process priority is as good
as next process. (preemption_goodness subroutine)
7.Timeout is also defined in case of
SCHED_MY(#define MAX_SCHEDULE_TIMEOUT 2*LONG_MAX /linux/sched.h)
Make the current task sleep until @timeout jiffies have elapsed.It
also assures that Task will be in TASK_RUNNING after
schedule_timeout() has occured.
8. move an exhausted process to be last in runqueue.Inspired by SCHED_RR.
I am n't clear about what should
be the effect of following changes on actual scheduling.Please explain
me the importance of these parameters and, also suggest some
application running which I can notice the changes. I am usung
"getrusage( )" function to see different parametter values.
Eagerly waiting for your reply!
Thanking You
Abhijit Kumar
