Re: [PATCH] sched: Provide iowait counters

[Arjan van de Ven]

Peter Zijlstra wrote:
> On Sat, 2009-07-25 at 09:42 -0700, Arjan van de Ven wrote:
> > In case you wonder why you need both tools, and can't just use one: we
> > wondered the same, but it turns out that if you only build
> > Tool Two, you don't get a good overview of what is going on on a
> > higher level. It's like looking at the world via a microscope all the
> > time,
>
> I'd be thinking you could compose your ms based picture on the sample
> data. For instance, if you sample on cpu-clock at 100kHz, then a 100
> samples get you a full blue slice, 50 get you a 50% blue slice, etc.
>
> Of course, since its sample based, you can miss some detail, but I think
> you're looking at the large picture anyway.
>
> Would this work?

ok thinking some more, in principle we can make a hybrid, that only uses the
perf data, as long as the perf data includes run/iowait/woken events.
(and process exit, name change. ideally also exec).
We'll just poll /proc only for the io throughput data, and collect that on a 1 msec
scale.
--
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/