Path: utzoo!attcan!uunet!aplcen!haven!umd5!cgs
From: cgs@umd5.umd.edu (Chris G. Sylvain)
Newsgroups: comp.os.minix
Subject: Re: readclock (v1.5.10)
Keywords: character device, no v1.5 mkproto
Message-ID: <7385@umd5.umd.edu>
Date: 1 Oct 90 18:58:17 GMT
Organization: University of Maryland, College Park
Lines: 20


I had created /dev/port when I began looking into why readclock was always
returning '-q' .. I took a wrong turn when I found peek() reported the
correct id byte (0xFC) when I cast CPU_TYPE_SEGMENT to (unsigned) ..

The real answer was:  /dev/port is created by 'mknod port c 1 4'
(a _character_ special device) NOT by 'mknod port b 1 4' (a _block_ special
device). Then make readclock Set-UID of the owner of /dev/port ('bin' for me).
No cast of CPU_TYPE_SEGMENT is then necessary ..

I note no cdiff against the file system prototype file supplied with
v1.2 MINIX (and not changed for v1.3) was included in the v1.5.10 upgrade kit.
Also, no mention of the special case for /dev/port being a character special
instead of a block special device is made in kernel/memory.c .. Clues and
hints are always appreciated when they do appear !!
-- 
--==---==---==--
Wabe: It goes a long way beyond a sundial on either side
--   ARPA: cgs@umd5.UMD.EDU     BITNET: cgs%umd5@umd2   --
--   UUCP: ..!uunet!umd5.umd.edu!cgs                    --