Subject: Re: splraise?
To: Chris Torek <torek@BSDI.COM>
From: Scott Reynolds <scottr@Plexus.COM>
List: tech-kern
Date: 03/06/1997 17:08:30
On Wed, 5 Mar 1997, Chris Torek wrote:

> If you simply think of all other splN's as `raise IPL to at least N',
> everything should work out.

This is exactly what splraise() (or as Jason appropriately points out,
_splraise()) accomplishes, so that this assumption holds true even if the
various hardware IPLs are `upside down'.  This can be particularly
beneficial on systems where the splxxx() levels are computed at boot time. 
By examining the interrupt levels on the appropriate hardware, you can
arrange it so that no more interrupts are blocked than necessary at any
given time.