Philippe Gerum wrote:
> On 07/29/2015 10:49 AM, Nicolas SCHURANDO wrote:
> Would I be right to assume that, when the task gets
>> switched to secondary mode, and because the other tasks are already
>> requiring a lot of cpu time, there is not enough free cpu time to run
>> the
>> newly switched task which gets then killed by the watchdog ?
>
> Unlikely, but possible, if the CDL task takes ages to run the mode
> switching code, and gets caught by the watchdog tick before it had a
> chance to eventually suspend from primary mode.
>
Another possibility is a hole in Xenomai/I-pipe fpu logic which causes the
trap to trigger again and again when trying to relax because of the
bounce.
A trace from the I-pipe tracer would help here.
--
Gilles.
https://click-hack.org
_______________________________________________
Xenomai mailing list
[email protected]
http://xenomai.org/mailman/listinfo/xenomai
