[nptl] Re: SCHED_RR issue

[Sebastien Decugis <sebastien.decugis@ext.bull.net>]

> But we need:
> 2. verify that the program doesn't have any "BUG/RT-issue". 

I've read the source and the program looks fine for me -- but once
again, I'm not an expert.

> 1. gather more data.

Here is a first result. More data will follow.

#### architecture: 2x IA32 (xeon), 2GB Ram, FC2 + kernel 2.6.8.1 ####

[root@nptl test]# time ./test_schedrr 1 98
Starting 1 threads SCHED_RR with prio 98
The 1 SCHED_RR threads running at prio 98 have been created
Wait that all threads run...
PASSED
 
real    0m0.024s
user    0m0.022s
sys     0m0.002s


[root@nptl test]# time ./test_schedrr 2 98
Starting 2 threads SCHED_RR with prio 98
The 2 SCHED_RR threads running at prio 98 have been created
Wait that all threads run...
PASSED
 
real    0m0.026s
user    0m0.024s
sys     0m0.002s



[root@nptl test]# time ./test_schedrr 3 98
Starting 3 threads SCHED_RR with prio 98
The 3 SCHED_RR threads running at prio 98 have been created
Wait that all threads run...
Thread 1: started=1, counter=14641217
Thread 2: started=1, counter=14641215
Thread 3: started=0, counter=0
 
real    0m30.127s
user    0m35.365s
sys     0m24.852s


** Note that with 3 threads, the result is quite random, it hangs only
from time to time. When I increase the # of threads, the hang happens
more often.

** I've run the same test and got the exact same behavior when using
prio = 1 instead of 98 -- so it looks like the prio parameter has no
effect.


I'll now experiment with other architectures / OS.

Seb.




-------------------------------
Sebastien DECUGIS
NPTL Test & Trace Project
http://nptl.bullopensource.org/

"You may fail if you try.
You -will- fail if you don't."


Liste de diffusion nptl
Pour se désinscrire : mailto:nptl_request@bullopensource.org?subject=unsubscribe