[uClinux-dev] realtime scheduling...

John Willlams jwilliams at itee.uq.edu.au
Fri Mar 14 04:35:17 EST 2003


Hi Lars,

>  Hmm, yes it can handle that, but I don't really understand why ? I
>think I am missing something, since you have an interrupt from an
>external source, you don't need to schedule that event, or do you mean
>if it can schedule an 'linux interrupt' at a lower priority later ? This
>it can do.

I was thinking it might be good to put a big wall between the real-time and
non real-time aspects of a system (and application).  So, by hanging
everything truly "real-time" off the NMI (using a 2nd interrupt controller),
and everying related to uClinux of the maskable interrupts, the two can be
pretty much completely ignorant of each other (except for a data tunnel from
one side of the wall to the other).

The recent discussions about RTLinux patches, then patching the patches for
uclinux, and so on, gives me headaches just thinking about it.  Every new
parallel branch in the source is a maintenence nightmare.  The absence of
RTLinux support for uClinux2.4 makes me think there must be a better,
cleaner way to add RT support without hacking into the bowels of uClinux.

The other thing is that since I'm living in the land of reconfigurable
computing on FPGAs, so it's very easy for me to say "ah just add another
interrupt controller, then create a FIFO peripheral to allow communication",
and so on.   Whether it's easy to do may be another matter! :)

Sorry if I'm not being very clear, it's just a lot of random ideas at the
moment, slowly forming into shape!

John




More information about the uClinux-dev mailing list