Re: [RFC PATCH 0/3] Perf persistent events

[Robert Richter <rric@kernel.org>]

On 18.03.13 09:46:38, Ingo Molnar wrote:
> 
> * Namhyung Kim <namhyung@kernel.org> wrote:
> 
> > So my question is how can an user know which persistent events are 
> > available in her system?
> 
> I think we need VFS enumeration for that: directories give a high level a 
> structure (allowing things like per user contexts) while readdir will give 
> list of specific persistent buffer contexts.
> 
> Sensible naming convention would be needed to make things easy to discover 
> and list - and for active buffers to not be forgotten about.
> 
> cgroups or a new 'eventfs' filesystem would be an option.

An option would be to attach the persistent events to a hosting pmu
(e.g. 'ras' in this case) and provide the events via sysfs as already
done by other pmus:

 /sys/bus/event_source/devices/ras/events/
 /sys/bus/event_source/devices/ras/events/mce_record
 ...

perf tools work then out-the-box with -e ras/mce_record/.

The event is selected by the 'ras' pmu and then routed to the original
pmu that might be e.g. 'tracepoint'. So we attach each persistent
event to a 'virtual' pmu which does the grouping in the perf sysfs and
the forwarding to its actual pmu.

-Robert