Jonathan Stone <jonathan@DSG.Stanford.EDU> writes:

> 
> However, rom a software engineering perspective, I really, really
> *dont* want to see these changes commited until the interrupt
> architecture is documented well enough for, say, a new portmaster to
> understand it and be sure that a given port actually *meets* the
> design assumptions.

`That's why I explained what it's supposed to do.'  It's true that in
an ideal world this would be documented in gory detail, but I don't
currently have time to do that, and I need this done several months
ago.

If you're worried about the pmax port, you can just make splsched()
equivalent to splhigh() for now; that's an acceptable implemenation,
although it won't give you the performance benefit I alluded to.
Jason can DTRT with it when he cleans up the pmax interrupt code.

> Is it necessary to commit the changes this week?

It's been necessary to do it for a long time.  I can't wait any
longer.