[etherlab-users] UNMATCHED and SKIPPED datagrams

[Erwin Burgstaller]

* Thu, Jan 22, 2009 - 10:22:48 +0100, Florian Pose - fp at igh-essen.com:
> What do you see in that case? Are all datagrams skipped? Does the same
> number of datagrams appear as unmatched?

No, not all of them. There were usually four or five per second and yes,
the number of unmatched was the same as of skipped.

But the problem has completely gone now. Setting the other rtai tasks
priority to 2, while the kernel module has 1, did help.

Sorry, I thought I did already post about that success, but I've seen it
has only been the plan of doing any changes to priorities.

> The only case that I could explain this with, is if the cycle time of
> your application is in the same size as the frame round-trip time (which
> is usually 10-20 usec. But your cycle time isn't that fast, is it?

No it isn't. It does 1000 loops per second and ecrt_domain_process() is
called every loop of course.

Thanks for your help,

       Erwin

-- 
Erwin Burgstaller