Hello, we finally have a new version! RTnet 0.9.0 is waiting to be downloaded from the usual place: www.rts.uni-hannover.de/rtnet/download.html
This release marks another important milestone, although it's not yet
THE release, i.e. 1.0. Thanks to the great work of several contributors
we now have a first driver for Gigabit Ethernet, Ethernet over
RT-Firewire, even more flexible TDMA ("joint slots"), and other smaller
useful extensions. The Kernel-based configuration system as known from
countless other projects has been adopted, just try "make menuconfig".
An internal restructuring and cleanup improved the modularisation of
RTnet: if you don't need real-time UDP/IP, just switch it off now.
But the most notable change is likely the "outsourcing" of RTnet's RTOS
and user API layer, also known as the Real-Time Driver Model (RTDM). A
revised version of this layer is now part of Xenomai the former
RTAI/fusion. Due to the renaming of this project, also RTnet had to be
adapted so that support for RTAI/fusion before the Xenomai 2.0 release
was dropped.
On the RTAI side, we are waiting for a new release (3.3) which is going
to include RTDM support as well. In the meantime you can already test
this RTnet release against the magma CVS branch which should work on 2.6
kernels and is prepared to support 2.4 again when remaining issues in
RTAI are fixed. Due to the required maintenance effort we decided to
drop older RTAI version support. Anyone still depending on this is
welcome to contact us and discuss migration paths or back-ports of RTDM.
Let's have a "short" outlook on future development. We still hope to
reach 1.0 release this year. On that way, we are trying to include some
of the following ongoing improvements:
- Multicast support is waiting to be cleanly integrated. This addition
will first require intensive testing of modified core system
components.
- Update and test the PPC drivers with respect to Xenomai and kernel
2.6. (PPC support for RTAI appears to be discontinued after 3.2 -
correct me if I'm spreading wrong information). Wolfgang is on it.
- CANopen over RTnet is under development. The first approach will be
low-intrusive, i.e. built on top of the user API. A deeper
integration into the RTnet core is also under consideration, as well
as a generic CANopen API, not just for RTnet...
- In-kernel support for RTnet is imaginable now with the advent of
Xenomai's new 2.1 branch. This will bring the need to adapt the build
system, to which degree has to be analysed yet.
- WLAN adoption is being evaluated at the moment. The wide range of
products with different degrees of control they provide to the user
over packet transmission and reception makes it hard to predict what
kind or determinism we will gain in the end. Anyway, the goal is
again to use affordable commodity hardware!
- A study on a TDMA design tool has just been completed. The result is
still in a development stage and the integration of the XML output
this tool produces into RTnet's configuration system has to be
implemented first, but the general direction is promising to ease the
design process of larger networks.
- An evaluation of Ingo Molnar's RT kernel branch as a host for RTnet
is scheduled, but without a fixed date. RTDM will likely be easy to
adopt to the RT kernel API, but we first have to wait for that
approach to mature, both with respect to design and implementation.
RTnet over PREEMPT_RT will then be a good test case to see potential
performance differences between single and dual kernel real-time Linux
approaches.
So, let me say thank you again to all the people helping to improve
RTnet constantly, regarding this release especially to Klaus Keppler,
Bill Vareka, and Yuchen Zhang.
Jan
signature.asc
Description: OpenPGP digital signature
