Messages in this thread | | | Date | Thu, 3 Sep 1998 12:28:59 +0200 (CEST) | From | Andrea Arcangeli <> | Subject | Re: Closure Re: 2us Interrupt latency's for Linux 2.0.xx |
| |
On Thu, 3 Sep 1998, Manoj Apte wrote:
>Andreas posted a release of his lil package on Aug 27. He notes that linux >shows periodic high interrupt latency due to large amount of time >(~15us) spent in the bottom handler processing in IRQ0.
Nonono. It' s not the time spent in the bottom_half. It' s the time spent in the timer interrupt code that runs cli()ed. The first __sti() is in do_bottom_half() a bit _before_ start running the bottom half handlers.
>I shall ofcourse do a lil more math to satisfy myself.. if anyone is >interested .. I can keep you posted.
I am interested of course ;-). Using floating point to calc the mean is pretty trivial.
There was a suggestion also to hack the timer irq to get:
irq0-time1 = x1 cycles irq0-time2 = x2 cycles irq0-time3 = x3 cycles
and then do the math over the xN cycles.
Andrea[s] 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.altern.org/andrebalsa/doc/lkml-faq.html
| |