>"Completely"? I beg to differ, watch and you will see that on every
>new TCP connection, there will be 2 spurious and unnecessary wakeups,
Ah I didn't thought about this. Right.
Hmm maybe this is the reason I couldn't understand for which apache sleeps
in flock(LOCK_EX) instead of in accept(2)? If there was 10 apache task
sleeping in accpet(2) then there would been as worse 30 reschedule where
only 1 task would go ahead accepting the connection for real.
Anyway I rejected from my tree my version of the wake-one since with my
approch wake_up continue to browse the whole list of sleepers (it can
address only the overschedule problem). I don't think there will be
interest in my approch since 2.3.1 will do the better thing.
Andrea Arcangeli
-
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/