Jan Kiszka wrote: > ... > So, as long as RTAI's RTDM is not heavily used, please consider testing > RTnet against Xenomai when in doubt.
To put this statement in a clear light (I received some complaint): We have two implementations of RTDM over Linux theses days (more may follow soon). For practical reasons, the Xenomai version is the reference implementation as well as playground for extensions. It's been during everyday work for various drivers, including RTnet. The second implementation is a port over RTAI 3.3 and magma, keeping most of the original RTDM code, but wrapping the lower interface from the Xenomai nucleus to the RTAI core. AFAIK, this RTDM version primarily serves for attaching RTnet, and has additionally been tested against the 16550A driver as well as the timer benchmark driver. Now RTAI users start using it for further RTDM drivers as well, like Sebastian's SJA1000. That's good news! Anyway, the reference implementation as well as the wrapping can still contain bugs, no question. Those might be typos (as this one likely was), problems in the specification of the interface, or whatever. To track them down, we first have to exclude as much trouble sources as possible. In this light, it is obviously helpful to have a failing scenario checked against the reference implementation as well. That's all, that's common. Jan
signature.asc
Description: OpenPGP digital signature
