Re: [Xenomai] Rescnt imbalance in rtdm_mutex_timedlock

[Gilles Chanteperdrix <gilles.chanteperdrix@xenomai.org>]

On 03/27/2014 07:24 PM, Philippe Gerum wrote:
> I'm unsure we really need to take a chance at giving such hint. Modern
> CPUs tend to have branch prediction right.

Not to mention the fact that the compiler may not use the information 
used by likely/unlikely, or may not be able use it in a sensible way.

Typically, for code such as

if (unlikely(condition))
	return 0;

I do not see that it makes sense for the compiler to "jump out of hot 
path", for a return 0, it will put the return 0 in place, there is 
nothing to win by jumping away to then just return.

-- 
					    Gilles.