[coyotos-dev] Problems with notification words

[Jonathan S. Shapiro]

On Tue, 2007-02-13 at 09:14 -0500, Jonathan S. Shapiro wrote:
> A solution that we considered in EROS was to introduce an "notification
> word" into the process state. If the notification word is non-zero when
> a process becomes available, a notification message is immediately
> synthesized by the kernel and delivered to the application. I will
> describe in my next note the problems with the notification word idea.

There are two basic problems with notification words:

1. You can only receive them when you are in the "available" state. In
   particular, you *can't* receive them in the "ready" state, because
   this would invalidate the outstanding reply capability.

2. They are non-preemptive and delivered in-band. This means that they
   cannot provide a good simulation environment for (e.g.) UNIX signals.

Point (1) is really the non-preemptive issue in disguise. What we really
want here is a mechanism for preemption.

Actually, point (1) raises another advantage of FCRBs: the whole issue
of resume capabilities really is simplified by first-class message
buffers. The fact that preempting a "wait for reply" made the reply
non-deliverable was really very awkward.

-- 
Jonathan S. Shapiro, Ph.D.
Managing Director
The EROS Group, LLC
+1 443 927 1719 x5100