Re: [PATCH 3/6] perf record: add time-of-day option

[Thomas Gleixner <tglx@linutronix.de>]

On Wed, 2 Mar 2011, David Ahern wrote:
> On 03/02/2011 07:16 AM, Thomas Gleixner wrote:
> > How does all that deal with CLOCK_REALTIME being affected by NTP and
> > settimeofday? Not really, as far as I can tell. It somehow works, but
> > that depends on the frequency of your event injection.
> 
> It is sampled at some periodic rate to get NTP changes. Right now it is
> hardcoded at once an hour. The frequency option can be added to the
> --tod parameter.

Once an hour. Brilliant. And how do you correlated that when
settimeofday was called 5 times in between the samples. Not at all.

> > 
> > To be honest, that's just too much churn for a feature which is single
> > purpose and questionable functionality.
> > 
> > It would be far more interesting to have trace points in the
> > timekeeping code, which are useful for other things as well
> > (e.g. precise monitoring of NTP) and provide always a correct
> > association with the trace clock. That would cover everything from
> > frequency adjustments, clock setting including the information about
> > the relation to clock monotonic.
> 
> I'm not looking to track ntp changes. I want time-of-day strings on
> sample dumps. That is not a questionable feature request; it is a
> usability enhancement.

Sigh, did you even read what I wrote? 

I did not say, that YOU have to track NTP changes.

And I fully understood that you want to correlate perf time
stamps with CLOCK_REALTIME.

Neither did I say, that this is a questionable feature request.

I merily objected to add functionality which is half baken and only
particular useful.

Further I made a suggestion to add trace points to the time keeping
code instead, which

      - solve your problem
      - are useful for other purposes
      - do not require ioctls, timer driven sw events

Thanks,

	tglx