[Xenomai] rt_taks_bind() returns -EAGAIN for existing task

Philippe Gerum rpm at xenomai.org
Sun Jun 11 19:06:17 CEST 2017

On 06/11/2017 05:58 PM, Giulio Moro wrote:
>> Time which could be devoted to this subject
>> is too short, when other duties have to be carried out as well. Usual
>> dilemma, usual outcome so far.
> Anything I and the community (us) can do to help with this?
> Pull requests with documentation amendments?

Fixing the existing service-oriented documentation would certainly be nice.

Also, my understanding is that people direly need in-depth articles that
guide them for implementing typical Xenomai use cases and patterns,
explaining basic concepts in the same move; not necessarily a long and
boring general prose, but rather to-the-point material illustrated with
simple figures.


- Closed-loop systems with Cobalt anyone?
- Driving devices in real-time, an RTDM tale
- Can a dual kernel system do audio?
- Could Xenomai speak arinc653, how?
- Did Xenomai and Android meet?
- etc [1]

Diving into details, some articles could cover specific technical issues:

- IPC between Cobalt and the regular world
- Sessions: sharing objects between processes within a UN*X group
- Driving a DMA with Cobalt
- Practical example of using slackspot (fixing it if necessary).
- How to create a user-space driver with UDD
- Survival kit when debugging a boot hang with Cobalt
- Who lives in /proc/xenomai after all?
Cobalt schedulers do?
- Sharing IRQs between Linux and Cobalt considered funky.
- Why and how calling Linux services within a real-time loop may kill
the latencies?
- Solving the C++ constructor madness when that code requires Xenomai
core services, a 10 years long psychotherapy
- etc

A reference for me would be LWN (https://lwn.net). The way the editors
there manage to explain technical issues makes me feel like I was a
smart reader - which is in itself a sufficient proof that they are doing
a hell of a good job.



More information about the Xenomai mailing list