Benchmark for sem_timedwait v1.1

[Ondřej Bílka]

On Mon, Feb 17, 2014 at 04:05:51PM +0530, Siddhesh Poyarekar wrote:
> On Mon, Feb 17, 2014 at 11:19:00AM +0100, Ondřej Bílka wrote:
> > 
> > So test-skeleton is apparently ok. If it is not then you should fix that
> > first.
> > 
> 
> Those are all the string tests I ported from string/*.  I haven't
> changed them yet because they have a very different output format and
> hence, won't fit into the generic benchmark skeleton yet.  I should
> probably elaborate on my request: if all your benchmark does is print
> out timing information for inputs, then it should be adapted to use
> bench-skeleton.c.  If there are other parameters you wish to capture
> in the output, then pretty much anything goes, altough it would be
> nice if a framework was written around them to reduce code
> duplication, which means use test-skeleton.c until you/I/someone-else
> comes up with something nicer.
> 
There are no sensible inputs. There may be different usage patterns to
measure which may involve spawning threads which is out of bench-skeleton scope.