>I understand that it may make sense that one would not need O_NONBLOCKING on
>/dev/null but I don't see why it should throw an error.  In other words there's
>nothing wrong with it technically (that I know of) and I don't see anything in
>the spec that hints to this fact.  In even plainer words, what implimentation is
>correct?

it would be easy to tweak sys/conf.h around the line where it #defines
cdev_mm_init() so that mmioctl didn't get #defined to enodev(), and
instead got defined to nullop().

the correctness of such a fix, however, isn't something i can comment
on.  you have to keep in mind that the following devices:

	/dev/io		(i386 only?)
	/dev/mem
	/dev/kmem
	/dev/null
	/dev/zero

are all backed by the same set of functions in arch/.../.../mem.c

-- 
|-----< "CODE WARRIOR" >-----|
codewarrior@daemon.org             * "ah!  i see you have the internet
twofsonet@graffiti.com (Andrew Brown)                that goes *ping*!"
andrew@crossbar.com       * "information is power -- share the wealth."