The following reply was made to PR port-m68k/30924; it has been noted by GNATS.

From: YAMAMOTO Takashi <yamt@mwd.biglobe.ne.jp>
To: cube@cubidou.net
Cc: gnats-bugs@netbsd.org
Subject: Re: port-m68k/30924: m68k copyinout_t breakage
Date: Sun, 07 Aug 2005 00:38:26 +0900

 > > i think it's better to have separate types for in and out,
 > > at least.
 > 
 > Why?  They have the same prototype, and I don't see any reason why
 > that would change.  They're basically the same function, and the name
 > decides the direction.
 
 IMO they are different enough.
 
 > OTOH, that would make prototypes that use the type clear about which
 > direction is used, although there would be no way of enforcing it.
 
 even if nothing prevents boolean_t from being used as pid_t,
 having separate types for them is a good idea, IMO. :)
 
 > > i read the XXX-cube comment but i'm still not sure
 > > why it can't be in systm.h.
 > 
 > Because of the clockframe nonsense on i386/amd64.  I don't know if any
 > other arch do that, I only use those two, but that would not be entirely
 > surprising.
 
 can you send me the diff which shows the "clockframe nonsense"?
 
 > > if it's due to circular dependencies among headers, how about
 > > fixing them instead?
 > 
 > It's really not my area.
 > 
 > > eg. move the in-kernel function prototypes including timer_create1
 > > out of time.h so that they won't be included by "primitive" ones?
 > 
 > That's a possibility.  I'd rather keep things simple, I will certainly
 > not be volunteering to fix all the kernel include crappiness just to
 > have copyinout_t declared in a decent way.
 
 i think that it's really your area because you are adding
 another complexity to the crappiness.
 
 YAMAMOTO Takashi