[coyotos-dev] Need for revised invocation

[Jonathan S. Shapiro]

On Mon, 2006-02-06 at 10:23 +0100, Valerio Bellizzomi wrote:
> On 05/02/2006, at 17.21, Charles Landau wrote:
> 
> (snip)
> >
> >None of the above requires preemption. Preemption is evil and should 
> >be avoided at almost any cost.
> >
> 
> There shouldn't be any preemption at all if the system runs user jobs under
> a precomputed schedule.
> The only one impediment that I see now, is that the **verification** of the
> precomputed schedule is NP-complete.
> More on this later.

Val:

First, precomputed schedules preempt all the time: every time your slice
ends.

Second: I believe that by "verification" you mean that the schedule
construction problem is NP-complete. This is the admission control
problem. Given a proposed schedule, checking that it meets all
constraints is *not* NP complete.