On Wed, 12 Aug 1998, Neil A. Carson wrote:

> Yep we've confirmed tonight that it seems to have gone away in more
> recent kernels than my relatively prehistoric tree. I think Mark's
> embedded in graphing IRQ latencies right now. Figures don't seem too bad
> on the whole.
> 
> Did it help your warble at all?

As Neil said I have been graphing the latency of hard clock interrupts
under different conditions.
As a result I have made some changes to the cpu_switch() and switch_exit()
routines to improve interrupt latencies.

Hopefully there should be some more changes coming shortly.

Cheers,
				Mark