Re: [patch] Real-Time Preemption, -RT-2.6.10-rc2-mm2-V0.7.29-0

[Ingo Molnar]

> * Florian Schmidt <mista.tapas@xxxxxxx> wrote:
> 
> > Hmm, the max jitter rtc_wakeup shows at 1024hz is around 150us. Which
> > seems a tiny bit large, too, as the rtc histogram shows a max wakeup
> > latency of 16us..
> 
> yep, that's a bit too large too. What type of load does it need to
> trigger such a 150 usec delay reliably?

on a 2 GHz UP box the worst-case max jitter i can trigger via rtc_wakeup
is 11 usecs, using the -5 kernel. The workload i used was 40 parallel
copies of LTP plus a few hackbench runs. This is how i started
rtc_wakeup:

 chrt -f 80 -p `pidof 'IRQ 0'`
 chrt -f 98 -p `pidof 'IRQ 8'`

 cd rtc_wakeup
 ./rtc_wakeup -f 1024 -t 100000

i.e. IRQ0 is below IRQ8 and the rtc_wakeup threads, but above every
other IRQ thread. Here's the histogram of a short (~5 minutes) run:

  1 247383
  2 34842
  3 1488
  4 3188
  5 125
  6 1

so this a 6 usecs max delay measured by /dev/rtc. So on your box, if the
max histogram delay was 16 usecs, i'd not expect a worse than ~30 usecs
jitter measured by rtc_wakeup. Can you reproduce the 150 usecs jitter
with the above IRQ setup?

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