Re: [PATCH] perf lock: clean the options for perf record

[Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp>]

On 2011年03月04日 23:41, Frederic Weisbecker wrote:
> On Fri, Mar 04, 2011 at 09:37:18AM -0500, Steven Rostedt wrote:
>> On Fri, 2011-03-04 at 18:41 +0900, Hitoshi Mitake wrote:
>>
>>> BTW, how do you think about the idea of exporting data in
>>> python (or other neutral) expression from procfs? I feel it is a
>>> good idea. Communicating with unified format between user space and
>>> kernel space will reduce lots of parsing overhead. Is this too
>>> aggressive or insane?
>>
>> As I mentioned in another email, I have no problem with an easy to parse
>> file. But I will aggressively NAK any "python" or other scripting
>> language. I'm sure I would get the same response if I were to have the
>> kernel outputting perl language ;)
>
> Same for me.
>
> But even before talking about that, I wonder if doing this is actually needed.
>

With analyzing lock_stat, grasping the rough trend of lock usage might
be possible. I'm imagining the tool like top, e.g. read the data from
lock_stat periodically, analyze the difference between unit time, and
print the result.