>
> How about this approach to measuring latency? This approach focuses on
> measuring latency caused by interrupt masking but with some more knowledge
> about processor registers, can measure absolute latency as well.
>
> Assume that the Hz timer interrupt occurs at regular intervals derived from
> the processor clock, also assume that the processor has an continuously
> incrementing counter running at the processor clock speed (Intel's RTSC
> instruction as I recall?). Then simply read the counter during the Hz timer
> interrupt service routine and subtract from it the previously read value.
> With no -variation- in latency, the difference will be 1/Hz. Have the Hz
> timer routine add each difference measurement to a histogram, the result
will
> closely correspond to the variation in latency caused by interrupt mask
> times. For extra credit, tag minimum and maximum latency values with some
> internal kernel data that provides hints about what might have caused the
> latency.
-
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