Subject: Re: populating section 9.
To: Darren Reed <>
From: Greywolf <>
List: tech-kern
Date: 02/20/2000 10:27:28
n Sun, 20 Feb 100, Darren Reed wrote:

# There are a number of man pages, missing, from section 9, for functions
# provided by libkern which also appear in section 3.
# I've included a script, below, which goes to /sys/lib, looks at what's
# there and then attempts to find man pages in /usr/share/man/man3 for
# what's in /sys/lib and `converts' the man page from (3) to (9).  It
# dumps all the various section 9 man pages in /sys/lib, presently :-)
# Whilst some people might want to say "no need for duplicate man pages",
# unless functions have man pages are in section 9, they (a) don't have
# a publicly documented presence in the kernel as being part of the
# interface available and (b) there's no other good way to tell what is
# and isn't in libca vs the kernel.  I think it is good for us to include
# things like bcmp, etc, in section 9 as well as 3.
# ...[elided]...
# Comments ?

Yes.  I thought the purpose of section 9 was not for stuff that existed
in the kernel as much for device driver interfaces (DDI) and driver-kernel
interfaces (DKI); i.e. how to get the kernel and a driver to play nice.
Not quite a tutorial, but interface-level stuff.  I'm not clear on what
I'm saying, as a kernel hacker I am not, but hopefully someone else will
clue in on this and be able to better explain this.

I don't think that section 3 stuff which happens to exist in the kernel
should be gratuitously duplicated in section 9, unless >_*MAYBE*_< the
kernel interface to those functions is sufficiently different from the
userland version, or, i.e. the kernel version does not provide bounds
checking on the fly(!), et al...

NetBSD: better for your uptime than Viagra.