Jan Kiszka wrote:
> Philippe,
> having a Xenomai heap mapped and then forking / exiting the fork is not
> a good idea, is it? I mean for the sake of archdep.numaps...

OK, think I found the gremlin: xnheap_vmclose must be paired with a
xnheap_vmopen for proper reference counting, not simple mapping
operations. Will test my theory (actually: Linux Device Drivers'
explanation) and post a patch afterwards.


Siemens AG, Corporate Technology, CT SE 2
Corporate Competence Center Embedded Linux

Xenomai-core mailing list

Reply via email to