Subject: NetBSD inconsistencies and odditites
To: None <>
From: L. Todd Masco <>
List: amiga-dev
Date: 08/20/1994 22:23:02
Okay, now that sun-lamp seems to be okay...

I've been hacking around a lot these last few days, using various
 IPC techniques... and NetBSD is not pretty from this perspective,
 especially on the man page/functionality synchronization level.

Got a list:

- First of all, as I've mentioned before, there's the System V-style
 ipc (semaphores, shared memory via shmat()) that seems to be somewhat
 supported (IE, my programs run properly when I #define KERNEL) are
 #ifdef KERNEL'd out.  ipcs and ipcrm work properly.

- MAP_FILE is referenced in the man page for mmap(2), but is not defined
 anywhere (certainly not in the headers referred to by the man page).
 The man page says an EINVAL will be delivered if this parameter
 is not set, but luckily that doesn't seem to be the case.  The man
 page for mmap is the only place I've ever seen this parameter

- F_SETLK, etc, (the record-locking features of fcntl(2)) appear in
 the header <fcntl.h> but not in the man page.  Is record-locking
 supported or not?   According to Stevens[1], everything after the initial
 release of BSD4.3  (Reno and later) is supposed to support this.

[1] Stevens, W. Richard: Advanced Programming in the UNIX(r) Environment,
	5th printing, p. 367, Addison Wesley.
L. Todd Masco  | "Cowboy politicians sucking up to the aristocracy, not  |  even sure if they like democracy..."  - TR-I