Re: [PATCH 0/6] Optimize the cpu hotplug locking -v2

[Srivatsa S. Bhat]

On 10/10/2013 10:24 PM, Oleg Nesterov wrote:
> On 10/10, Andrew Morton wrote:
>>
>> On Thu, 10 Oct 2013 17:26:12 +0200 Oleg Nesterov <oleg@xxxxxxxxxx> wrote:
>>
>>> On 10/10, Ingo Molnar wrote:
>>>>
>>>> So ... why not make it _really_ cheap, i.e. the read lock costing nothing,
>>>> and tie CPU hotplug to freezing all tasks in the system?
>>>>
>>>> Actual CPU hot unplugging and repluggin is _ridiculously_ rare in a
>>>> system, I don't understand how we tolerate _any_ overhead from this utter
>>>> slowpath.
>>>
>>> Well, iirc Srivatsa (cc'ed) pointed out that some systems do cpu_down/up
>>> quite often to save the power.
>>
>> cpu hotremove already uses stop_machine,
> 
> And Srivatsa wants to remove it from cpu_down().
> 

Yes, I have worked on several designs to remove stop_machine() from the cpu_down
path.

http://lwn.net/Articles/538819/
http://lwn.net/Articles/556727/

>> so such an approach shouldn't
>> actually worsen things (a lot) for them?
> 
> this depends on what this "freezing all tasks" actually means.
> I understood it as try_to_freeze_tasks/etc, looks too heavy...
> 
> But my only point was, I am not sure we can assume that cpu_down/up
> is extremly rare and its cost does not matter.
> 
>> use stop_machine() on the add/remove
>> (ie, "writer") side and nothing at all on the "reader" side.  Is there
>> anything which fundamentally prevents cpu hotplug from doing the same?
> 

Its certainly possible to remove stop_machine() from CPU hotplug, as I've
demonstrated in the patchsets mentioned above. And there were pretty good
performance improvements too, arising out of that change, as described here:

http://article.gmane.org/gmane.linux.ports.ppc.embedded/56122

Regards,
Srivatsa S. Bhat



--
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/