Re: kqueue microbenchmark resul

From: Dan Kegel (dank@alumni.caltech.edu)
Date: Sat Oct 28 2000 - 18:51:59 EST


> >In fact, if you did leave the read queued in a daemon using select()
> >before, you'd keep looping endlessly taking all CPU and never idle
> >because there would always be read data available.

That would be a programming error on the part of the application.
Any application using a level-triggered interface like select
or poll must of course mask off events it is not interested in,
to avoid getting them endlessly. That's just 'select 101'.

> Also, level triggered notifications would also seem to cause
> multiple thread wakeups and thundering herd problems when
> there are multiple worker threads reading from the same queue.
>
> How does (?) kevent avoid this from happening?

Easy - applications which have multiple threads reading from the
same queue would use oneshot events instead of level-triggered events.
Level-triggered events are only for applications where a single thread
is reading from the queue.

- Dan
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Tue Oct 31 2000 - 21:00:24 EST