Collective resource limits status report #5
Gabor Kovesdan
gabor at FreeBSD.org
Sun Jul 4 16:38:47 UTC 2010
Since last report, I've resolved the problems that caused kernel panic.
The linked list handling now seems to be ok, I've checked it many times
but somehow still there's a bug somewhere. I store an entry there but
later the job isn't found. I've also added some Robert Watson locking
(see rwlock(9), last line :) to check if that helps but it caused
another kernel panics, so this is commented out at the moment. Now, I'm
working on resolving the locking problem and making the lists actually
work. I might try debugging with adding some temporary syscalls to
access internal kernel structures or printing out something with a
sysctl (this was pgj's idea) so that I can figure out what's going
wrong. If you have a better idea or you find something strange in the
patch, please let me know.
Another issue, I've tried to merge my work into p4 but I got strange
problems. I wrote to soc-admins@ and perforce-admin@ regarding this, I
hope I can also figure it out soon. For the meantime, I provide a full
patch, I think it's still easy to read because the main part of the code
resides in sys/kern/kern_jobs.c.
http://kovesdan.org/patches/jobs_current.diff
Gabor
More information about the soc-status
mailing list