Limesdr mini on Orange Pi Zero

[Harald Welte]

Hi Gullik,

thanks a lot for your excellent analysis and for sharing it here.

It is by far the most in-depth study of LimeSDR<->osmo-trx related
problems that I've seen. Great.

Now the interesting question of course is how to anylyze this further.

Linux has received *tons* of tracing mechanisms during the last decade,
[un]fortuantely I've only read docs/tutorials and never had to use any
of it so far.  But there for sure are tools that allow to analyze where
the latency is coming from.

It may be some particularly nasty driver on your platform.  It may be
thermal throttling due to insufficient cooling of the CPU, ...

I guess you already deactivated virtually anything that's not required
on the system and running in the background?  I'm not so much worried
about other background tasks at normal priorities, but more about what
kind of peripherals they might interact through which device drivers

Another idea is to exclude any influence of the block layer by ensuring
that there are no log files written either directly from osmo-trx, nor
indirectly via stdio redirection / syslog or the like.  Otherwise the
occasional flush to the block device might be a possible cause.

On Tue, Feb 05, 2019 at 04:54:49PM +0100, Gullik Webjorn wrote:
> No on to finding the culprit....or just change HW / OS.....this combo has
> lousy RT characteristics.....

People have observed the same pattern also on other hardware, so finding
a way to systematically debug this kind of issue will help not only on your
particular platform today.

Regards,
	Harald
-- 
- Harald Welte <laforge at gnumonks.org>           http://laforge.gnumonks.org/
============================================================================
"Privacy in residential applications is a desirable marketing option."
                                                  (ETSI EN 300 175-7 Ch. A6)