[nptl] Re: Advance on bug with SCHED_RR thread

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

Thank for your patience :)

Here are the results (both runs on dual-i686)

-- Conf 1 --
kernel: 2.6.9 vanilla
glibc : from Fedora Core 2
build : gcc -lpthread -lrt test_schedrr3.c -o test
Result: the problem appears:
# ./test 10 50
[...]
testdrv> Wait that all threads run...
Thread     1: created but NOT STARTED
Thread     2: created but NOT STARTED
Thread     3: started for 29.463012000 sec, sched. 3729722 times
Thread     4: started for 29.463015000 sec, sched. 3729735 times
Thread     5: started for 29.910831000 sec, sched. 3804873 times
Thread     6: started for 29.910834000 sec, sched. 3804876 times
Thread     7: started for 29.910837000 sec, sched. 3804879 times
Thread     8: started for 29.910840000 sec, sched. 3804877 times
Thread     9: started for 29.910843000 sec, sched. 3804882 times
Thread    10: started for 29.910853000 sec, sched. 3804884 times
TIMED OUT


-- Conf 2 --
kernel: 2.6.9 vanilla
glibc : fresh CVS from monday
build : gcc -B/home1/cvs/root -B/home1/cvs/root/lib
-Wl,-rpath,/home1/cvs/root/lib,-dynamic-linker,/home1/cvs/root/lib/ld-linux.so.2  -L/home1/cvs/root/lib -lc -lpthread -lrt test_schedrr3.c -o test2
Result: same as above:
# ./test2 10 50
[...]
testdrv> Wait that all threads run...
Thread     1: created but NOT STARTED
Thread     2: created but NOT STARTED
Thread     3: created but NOT STARTED
Thread     4: created but NOT STARTED
Thread     5: started for 29.928881000 sec, sched. 5126832 times
Thread     6: started for 29.992817000 sec, sched. 5139456 times
Thread     7: started for 29.992820000 sec, sched. 5139461 times
Thread     8: started for 29.992822000 sec, sched. 5139465 times
Thread     9: started for 29.992825000 sec, sched. 5139462 times
Thread    10: started for 29.992833000 sec, sched. 5139462 times
TIMED OUT


So, I think you can safely report this to OSDL :)

Best regards,
Seb.

On Tue, 2004-11-16 at 23:46, Loic Domaigne wrote:
> > Yes, you're right. I've been busy lately so I somehow forgot this bug,
> > but anyway it has to be processed! 
> > 
> > I'll have to test again with a 2.6.9 kernel, but it was here with the
> > 2.6.8.1 vanilla one.
> > 
> > I should be able to give it some time in the end of this week.
> 
> Ok, I'll wait for a confirumation on 2.6.9 vanilla kernel. If it is 
> confirmed, I shall post a bugzilla on OSDL.
> 
> Cheers,
> Loic.
> Liste de diffusion nptl
> Pour se dsinscrire : mailto:nptl_request@bullopensource.org?subject=unsubscribe
-------------------------------
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