Re: [PATCHv4 08/17] pps: add async PPS event handler

[Rodolfo Giometti <giometti@enneenne.com>]

On Sun, Nov 21, 2010 at 02:23:02AM +0300, Alexander Gordeev wrote:
> ?? Sat, 20 Nov 2010 17:08:51 +0100
> Rodolfo Giometti <giometti@enneenne.com> ??????????:
> 
> > On Thu, Nov 18, 2010 at 07:01:01PM +0300, Alexander Gordeev wrote:
> > > This handler should be called from an IRQ handler. It uses per-device
> > > workqueue internally.
> > 
> > Can you please explain to me why do you need this patch? Maybe you can
> > add a verbose patch's description? :)
> 
> Well, it's all about optimizing latencies on rt-preempt kernel: if
> everything is done in a process context than we can use mutexes
> (haven't done that yet) and don't disable interrupts, which is better
> for hard real time. I measured that pps_event with kernel consumer
> enabled takes 1-2us on my test machine. Not a big deal, of course...
> 
> Sorry, I've completely forgotten about an echo function. However quick
> look to current clients shows that it is only used in pps-ktimer.c
> for debug printing... Maybe it's not needed at all? I mean, have you
> ever got any user request for this feature? If yes, it can be removed
> IMHO since RFC-2783 says that it's optional. I can handle the removal in
> the next version of the patchset.
> 
> If you don't want it I can have this patch on my rt branch only and
> don't try to push it into mainline.

The echo function is not used as far as I know, so it's ok for me to
remove echo support if you get better performance.

In case we can study how to reimplement it! ;)

Acked-by: Rodolfo Giometti <giometti@linux.it>

-- 

GNU/Linux Solutions                  e-mail: giometti@enneenne.com
Linux Device Driver                          giometti@linux.it
Embedded Systems                     phone:  +39 349 2432127
UNIX programming                     skype:  rodolfo.giometti
Freelance ICT Italia - Consulente ICT Italia - www.consulenti-ict.it