Re: [RFC][PATCH 1/2 v2] tracing: Add TRACE_EVENT_CONDITIONAL()

[Mathieu Desnoyers]

* Steven Rostedt (rostedt@xxxxxxxxxxx) wrote:
> On Fri, 2010-12-03 at 10:27 -0500, Mathieu Desnoyers wrote:
> 
> > > >   TP_CONDITION(unlikely(someparam)),
> > > 
> > > I actually think this is an abuse of "unlikely".
> > 
> > Why are you considering this an abuse ?
> 
> Because it is overused. I would rather get rid of most unlikely()'s
> because they are mostly meaningless. Just run the unlikely profiler, and
> you will see a large number of them are just plain incorrect.
> 
> Adding them here probably doesn't do any good. The only reason for this
> TP_CONDITION() is to ignore those cases that it just does not make sense
> to trace. Like a wake up tracepoint that does not wake anything up. No
> need for "unlikely" or "likely", by trying to do that, you will most
> likely get it wrong.
> 
> 	unlikely(use_likely_correctly)

Ah OK. You are afraid that people will misuse it, not saying that it would be
technically incorrect. Fair enough. It sounds like a good enough reason for not
documenting this use-case.

Thanks,

Mathieu

-- 
Mathieu Desnoyers
Operating System Efficiency R&D Consultant
EfficiOS Inc.
http://www.efficios.com
--
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/