Re: [PATCH v2 14/14] Change CPUACCT to default n

[Paul Turner]

On Thu, Nov 24, 2011 at 6:05 PM, Li Zefan <lizf@xxxxxxxxxxxxxx> wrote:
>> despite it being a not entirely natural fit.  Something I proposed at
>> Prague and that we could explore here is the idea of a co-mounted
>> controller.  In this example it would only be mountable with cpu so
>> you could always depend on the cpu hierarchy being there; likewise we
>> can put (jump-labeled) touchpoints within the cpu-subsystem to call
>> out for updates as appropriate when the co-mount exists.
>>
>
> IIUC, this co-mounting idea is something I implemented years ago:
>
> https://lkml.org/lkml/2008/6/18/389
>
> The use case and the reason it was rejected:
>
> https://lkml.org/lkml/2008/7/1/97
>

Rejection is a bit of a strong statement -- the idea seemed amenable
but lacking a strong use-case.  That said, taking a deep look at some
of what Glauber is trying to do in this series I don't think it's
something that would help here.

For this discussion the motivation for a co-mount would be to
piggy-back on the cpu sub-systems own hierarchy walks to reduce
overhead.  However, this is not structured in a way that can take
advantage of this, and, looking at what Glauber is attempting to
collect it's not clear that it can be.

I think this moves the discussion towards whether we should consider
deprecating some of the exported fields (namely usage and
usage_per_cpu) from cpuacct instead of the entire controller as we had
initially desired.  This would allow cpuacct to exist with a much
lower overhead, especially within the context-switch path.

- Paul

> --
> Li Zefan
>
--
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/