Re: Strange interaction between latencytop and the scheduler

[Arjan van de Ven <arjan@linux.intel.com>]

Török Edwin wrote:
> 
> Latencytop userspace tool shows latencies > 0.1 msec, thus capturing
> backtraces for latencies <0.1msec could be avoided.
> If I apply the patch below, then enabling latencytop doesn't freeze X
> when running the "10-threads doing infloop usleep(1)" test.

ok I like the idea; I would propose though to make the 0.1 msec a sysctl or something,
so that people who really care about latencies lower than that can just set that lower.
(and the tool can then read the value and use that)

> Still, I don't want to loose track of the latencies we didn't collect
> backtraces for, so I added a special "untraced" category, reported as
> first line in /proc/latency_stats. If needed, instead of hardcoding the
> threshold, it could be made a sysctl, or set via writing to
> /proc/latency_stats,...

yeah for the total it makes sense; I'll do the sysctl thing for the threshold
and integrate this idea as well. Thanks a lot!