Re: [PATCH 8/9] per-cgroup boot time

[Peter Zijlstra <a.p.zijlstra@chello.nl>]

On Tue, 2011-09-20 at 10:06 -0300, Glauber Costa wrote:
> On 09/20/2011 10:04 AM, Peter Zijlstra wrote:
> > On Tue, 2011-09-20 at 09:37 -0300, Glauber Costa wrote:
> >
> >>>> +	getboottime(&boottime);
> >>>> +	ts = timespec_add(boottime, ca->start_time);
> >>>> +	jif = ts.tv_sec;
> >>>>
> >>>>    	for_each_possible_cpu(i) {
> >>>>    		cpustat = per_cpu_ptr(ca->cpustat, i);
> >>>
> >>>
> >>> I'm confused, what does it do? You take a boot time timestamp at cgroup
> >>> creation, add that to all boot-time readings and print the result. How
> >>> does that make sense? Subtracting the start_time, maybe, that would make
> >>> the cgroup creation time 0, adding, not so much.
> >>
> >> Boot time represent at which times the machine was booted. In this
> >> context, at which time the container/cgroup was created. So it have to
> >> be an addition.... don't really understand your question
> >
> > I think we're all properly confused now :-)
> >
> > getboottime() gives a time since boot, right?
> >
> > You take stamp at cgroup creation: say 50s after boot.
> >
> > Then on usage you take a new getboottime() reading (which per definition
> > is>  50s) and add your 50s that you read previous. This results in the
> > cgroup having 50s of boot-time _MORE_ than the machine. Say you read at
> > 123s, you then add your 50s timestamp, resulting in 173s to report.
> >
> > If instead you did a subtraction: 123-50=73, you would have reported the
> > time since cgroup creation.
> >
> no.
> 
> /**
>   * getboottime - Return the real time of system boot.
>   * @ts:         pointer to the timespec to be set
>   *
>   * Returns the wall-time of boot in a timespec.
> ...

Argh,. ok.