Subject: ATAPI ZIP success
To: Port-i386 <port-i386@NetBSD.ORG>
From: Jeff Mitchell <skeezix@skeleton.org>
List: port-i386
Date: 04/08/1998 00:54:26
	Thought I'd recap my meagre findings, in case anyone else gets
stuck with the same problem I had (couldn't get anything other than
"device not configured" errors, bad secotrs, etc, on the ZIP use)

uname machine details:

NetBSD freebase-tofu.interlog.com 1.3E NetBSD 1.3E
(netbsd13current-atapi) #2: Sun Apr  5 23:26:56 EDT 1998
skeezix@freebase-tofu.interlog.com:/usr/src/sys/arch/i386/compile/netbsd13current-atapi
i386

	Machine is a 486dx-80; EIDE controller with 2 drives on one chain,
and a ZIP and CDROM on the other. The CDRom is set to slave, and the ZIP
drive is set to thr unknown jumper -- its labeled "master, slave,
reserved" for the three positions. Until I had the sudden revelation to
set the jumper to reserved, I got no useful behaviour. Looks like the ZIP
drive has its master/slave/reserved label reversed... (as when I had it in
its master setting, I got only the ZIP drive, and no cdrom at all, during
probes). Fine now :/

	In no order, because I'm whipped :)

	1) With various revisions of the kernel sources, I couldn't get
the atapi patch to take correctly. Various chunk failures, or odd
filenames produced, etc. I solved this by going -current, which is always
a happy thing to do.

	2) -current has worked well; I'm running 1.3BETA binaries et al,
with the 1.3E current kernel. It includes atapi support already, making
the patch irrelevant. (1.3BETA included suport for ATAPI CDroms, however,
already, but not for removable media).

	3) Once the drive was present and probed, I coudl only make its
light flicker with disklabel; newfs et al was no luck. When I created a
disktab and attempted to write it to the zip disks, it would seem to
"take", excepting the partition definitions. Either I've botched them and
several of us missed it, or else they just wouldn't go onto the disks. I
rebooted to ensure the label "took". I toyed with various labels et al for
ages, and just couldnt' get the desired effect. The drive light flickjered
though, and probed nicely, so seemed correctly installed. 

	4) Brian C. Grayson mentioned the obvious, which I'd discounted.
Instead of using disktab's, I manually created a disklabel with "disklabel
-e", and created some partitions and set the number of partitions. This
"took", and when "newfs /dev/sd0a"'d, everything was cool. Took seconds,
then I mounted it, and all was fine. Thansk *MUCHLY* Brian!

	5) Aftyer I got it all going, I replugged in my extra drive (too
much power draw :/) and booted up Windows, and it used the ZIP drive
without a problem. This added to the satisfaction level in the drive. Of
course I unplugged this evil drive again ;)

	The lesson of the story; Listen to Kehlog Albrahn.

		Jeff

	For those in need, my disktab is pretty generic:

zipdisk|ZIP 100MB:\
        dt=IDE:ty=removable:se#512:ns#96:nt#64:nc#32: \
        pa#196608:oa#0:ta=4.2BSD:ba#2048:fa#1024:\
        pc#196608:oc#0:\
        pd#196608:od#0:


--
"I'm CONSING NOW!" "Nature is for weenies." -- Hozhead