> So, by usleeping 500 times, the running time increases by 20%, and
> the accounted running time goes from 4.02 to 0.02....
>
> Also note that 2.2 sleeps for 2ms (incorrect?) while 2.0.x sleeps for
> 12 ms at a time....
As a side remark, one should note that you do not need to write malicious
programs to observe this limitation of the CPU time accounting in Linux.
I'm working on a realtime video processing application right now and it is
quite hard to time, because of the time accounting limitations. Frame
capture blocks the process until the framegrabber gets one frame, which
occurs 25 times per second. the time accounting runs in the timer
interrupt, which occurs 100 times per second. So my process is reported
sometimes as taking 0% of the CPU, and sometimes as taking 25%, and it is
quite hard to figure out what the real number is.
--
Michel "Walken" LESPINASSE - Development Engineer at Wind River Systems
walken@wrs.com - http://www.via.ecp.fr/~walken/
DNA is the software of Life.
Did you realize you can have that much fun for a code merge ?
-
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/