[Devel] Re: [PATCH v3 3/6] expose fine-grained per-cpu data for cpuacct stats

[Glauber Costa]

On 05/30/2012 02:43 PM, Peter Zijlstra wrote:
> On Wed, 2012-05-30 at 14:34 +0400, Glauber Costa wrote:
>> On 05/30/2012 02:34 PM, Peter Zijlstra wrote:
>>> On Wed, 2012-05-30 at 13:48 +0400, Glauber Costa wrote:
>>>
>>>> +static int cpuacct_stats_percpu_show(struct cgroup *cgrp, struct cftype *cft,
>>>> +				     struct cgroup_map_cb *cb)
>>>> +{
>>>> +	struct cpuacct *ca = cgroup_ca(cgrp);
>>>> +	int cpu;
>>>> +
>>>> +	for_each_online_cpu(cpu) {
>>>> +		do_fill_cb(cb, ca, "user", cpu, CPUTIME_USER);
>>>> +		do_fill_cb(cb, ca, "nice", cpu, CPUTIME_NICE);
>>>> +		do_fill_cb(cb, ca, "system", cpu, CPUTIME_SYSTEM);
>>>> +		do_fill_cb(cb, ca, "irq", cpu, CPUTIME_IRQ);
>>>> +		do_fill_cb(cb, ca, "softirq", cpu, CPUTIME_SOFTIRQ);
>>>> +		do_fill_cb(cb, ca, "guest", cpu, CPUTIME_GUEST);
>>>> +		do_fill_cb(cb, ca, "guest_nice", cpu, CPUTIME_GUEST_NICE);
>>>> +	}
>>>> +
>>>> +	return 0;
>>>> +}
>>>
>>> Uhm, hotplug anyone?
>> What's with hotplug ?
>
>
> Who's to say all cpus are online at this point? If you don't want to
> make the files come and go with hotplug, one should use
> for_each_possible_cpu().
>
I don't oppose that.
But I don't really parse what you mean by "make the files go away". We 
have just one file, with multiple entries. The entries are in key:value 
form, so I don't see a huge problem in having some entries disappearing.

As a matter of fact, this is exactly what happens in /proc/stat when you 
offline a cpu: the correspondent line will stop showing up.