Hi.... On Wed, Oct 20, 2010 at 01:30, moussa ba <mus...@gmail.com> wrote: > I see a lot of drivers and filesystems using cond_resched. I understand that > this is for voluntary preemption and that this helps interactive > performance. Can someone shed a bit of light on why that is the case and > more importantly, how and where do you decide to call cond_resched (or > schedule...) in your code.
I think there's no such rule of thumb here, but I think everybody follow the same sense: a. is this code path long enough to complete? b. Is this code path safe to be rescheduled? for example, atomic section is a big no no c. in this code path, are we likely to see another process coming to be runnable? probably more..the idea is, whenever you see a chance to improve throughput while reducing latency at the same time, re scheduling is worth to call IMHO -- regards, Mulyadi Santosa Freelance Linux trainer and consultant blog: the-hydra.blogspot.com training: mulyaditraining.blogspot.com -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to ecar...@nl.linux.org Please read the FAQ at http://kernelnewbies.org/FAQ
