Re: interrupt latency

Larry McVoy (lm@bitmover.com)
Thu, 13 Aug 1998 10:12:10 -0600


Oliver Xymoron <oxymoron@waste.org>:
: Something just occurred to me: you can measure latency by making a broken
: interrupt handler.

I think I need to be a little more specific about why I would like to
measure interrupts and get some more focussed advice.

The thing I'm trying to figure out is:

How long is it before a user process gets the info when an external
event occurs?

An example, the one I care about, is a network packet receive interrupt.
I'd really like to know how long it takes from the time a packet hits
the card to the time that a user process can see that packet.

In former lives on former OS's, a big chunk of that was the interrupt
handling itself.

By the way, I'm not interested so much in how fast it currently is - I'm
interested in how fast it could be. So the fact that the driver may be
crap and the networking stack is too slow, etc., isn't the point.

So what I need is some simple way of mimicing what a network card does. I
was considering wiring up two machines with crossed over parallel ports and
writing a little hot potato test that went across the parallel port wires.
But I suspect that isn't really comparable to what I really want to figure
out.

Comments?

-
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.altern.org/andrebalsa/doc/lkml-faq.html