> yep, although i think this does not solve William's problem: if TIMER_BH
> gets marked by a timer interrupt _after_ it has been run, then we miss the
> event.
If BHs run for whole tick, they deserved to be lost and the problem
is not a problem.
The case is when timer marked during _NET_BH_ is active, that is
the case which should be handled correclty. Different people
blamed on this for long time, it results in massive loses
on serial links.
BTW in softnet TIMER_BH is not softirq (I am waiting for per-cpu timers 8)8)
Hint, hint!), so that scheduleing TIMER_BH is the same as in current 2.3.
I meaned scheduling of genuine softirqs.
Alexey
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/