Schedulability proof for message passing systems
It is not really any reason why we do not have schedulability
proofs for messagepassing systems. Certainly messagepassing systems
are associated with network communication protocols with
bad timing guarantees, but it is as relevant to look at communication
as a time-less, conceptual information flow from one thread to
the other.
Another, removable, obstacle is that messagepassing systems often
motivates leaving the "one thread per timing demand" assumtion.
Do a litterature study; Start with
Martin
Korsgaards PhD Thesis, find a suitably conservative set of
assumptions and attempt at a proof strategy.
|