Counter vs Btimer for tmtests

[Sebastian Huber]

On 07/11/2018 19:05, Joel Sherrill wrote:
> Thanks.
>
> A ticket needs to be filed to convert the tests if the counter works 
> on all BSPs.

https://devel.rtems.org/ticket/2847

> What are the BSP requirements for counter?

The BSP needs a free running counter which is not affected by power 
saving states of the processor.

>
> Also, it likely means that results need to be recalibrated and cannot 
> be compared
> for historical purposes.

The CPU counter will likely use the same device as the timer driver.

>
> At this point, I consider the two tests using counter future looking 
> and outliers.
>
> --joel
>
> On Wed, Nov 7, 2018 at 12:00 PM Sebastian Huber 
> <sebastian.huber at embedded-brains.de 
> <mailto:sebastian.huber at embedded-brains.de>> wrote:
>
>     ----- Am 7. Nov 2018 um 18:13 schrieb joel joel at rtems.org
>     <mailto:joel at rtems.org>:
>
>     > Hi
>     >
>     > Do these report the same units?
>
>     Not necessarily.
>
>     >
>     > If not, there are only *tmtests which use rtems/counter.h and
>     thus would be
>     > reporting numbers with different units than the btimer.
>
>     You can convert the counter value to nanoseconds:
>
>     https://docs.rtems.org/doxygen/branches/master/group__ClassicCounter.html#ga3e50a042e60bfcafdff5dbd46cad8947
>
>     This is used by some tests for the XML output which can be used to
>     plot some graphs:
>
>     https://git.rtems.org/rtems/tree/testsuites/tmtests/tmcontext01/tmcontext01.scn
>     https://git.rtems.org/rtems/tree/testsuites/tmtests/tmcontext01/plot.py
>

-- 
Sebastian Huber, embedded brains GmbH

Address : Dornierstr. 4, D-82178 Puchheim, Germany
Phone   : +49 89 189 47 41-16
Fax     : +49 89 189 47 41-09
E-Mail  : sebastian.huber at embedded-brains.de
PGP     : Public key available on request.

Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG.