Subject: Re: Using splsched()
To: Paul Kranenburg <pk@cs.few.eur.nl>
From: Charles M. Hannum <mycroft@mit.edu>
List: tech-kern
Date: 11/14/1997 05:30:54
When I originally proposed splsched(), the idea was to have it used
for all activity which modified scheduler state.  It was to be used in
cases where we currently use splhigh(), so that there could be
interrupts outside of splsched() (e.g. hard serial interrupts).

You're effectively using splstatclock() for that purpose (which is
probably fine, as it at least resolves the serial issues, and in
practice a separation between splstatclock() and my splsched() isn't
all that useful).

What you're using splsched() for is specifically things which modify
sleep/wakeup queues.  The name splsched() doesn't make any sense for
this, because it actually has nothing to do with the scheduler.