[rescue] SPARCstation 1 boot woes

[Mouse]

> I'm certainly interested in your offer for a 1.4T netboot setup!

On ftp.rodents-montreal.org (also available from the same host over
HTTP, though I'm not sure what MIME type they'll be served as), there's
a directory /mouse/misc/1.4T-sparc-netboot/ which contains three
things:

-r--r--r--  1 mouse         62008 Feb 26  2002 boot.net
-rw-r--r--  1 mouse     155633028 Oct 21 11:14 root.tar.bz2
-rw-r--r--  1 mouse        263690 Oct 21 11:15 root.tvf.bz2

They're on the wrong end of a fairly slow DSL line, so root.tar.bz2 is
likely to take a while to fetch.

boot.net is the TFTP-loaded booter; this goes in your /tftpboot or
local equivalent.  For my SS20s, this needs to be named with the client
IP address in uppercase hex, with .SUN4M apppended (the 20 is a sun4m
machine) - for example, my diskless-sparc setup puts the machine at
10.0.1.11, so the file is named 0A00010B.SUN4M.  (I have the same file
also present at 0A00010B.SUN4 and 0A00010B.SUN4C, though, as I remarked
upthread, I can't test those at the moment.  If you've netbooted
anything, though, you presumably know what name to put this under.)p

root.tar.bz2 is, as the name implies, a bzip2ed tarball of the
filesystem tree, in this case a directory named sparc-root.  To help
you plan disk usage, du -s on the directory I tarred up says it takes
473108K; allowing half a gig should be plenty.  It contains a kernel,
named "netbsd" (boot.net's default name).

root.tvf.bz2 is (bzip2 compressed) tar tvf output for the tarball.

Checksums:

[Tangent] 83> md5 *
a62b770cb486bdc8f20c97b9fa2e8a2f boot.net
b6ce9b1f1d5d43a6f75f58ad7a415006 root.tar.bz2
8648b985883db652a98ac917975dc8ae root.tvf.bz2
[Tangent] 84> sha256 *
7b620a7b6d6c63cd9db9f91c043b8d897f1e9ecd4b1072de7bd593b3c9ec146a boot.net
818949a2e3e6051b35f16e2683cb5b5f6b6387ace3526e50571762cabba6fb9c root.tar.bz2
f2b7bcb35f7f22d64dee14e78b54e63b4674977e84053e378be107c00370c56e root.tvf.bz2
[Tangent] 85> 

In case you're interested, the tarball includes /usr/src, including my
mutations, though I suspect that's not of much interest for your
purposes.  (As its .git/config implies, the same mutations are
available by cloning
git://git.rodents-montreal.org/Mouse/netbsd-fork/1.4T/src.)

I test-booted that root area with an SS20 and it booted fine (to
single-user - the embedded rc.conf is set rc_configured=NO, so it won't
try to go multi-user even if you don't boot with -s).  As I said
upthread, I don't have a sun4c machine to test with at the moment,
though I hope to get one of my SLC/ELC boards set up before too long.

>> Indeed.  I'm not sure how much point there is in trying to boot an
>> OS until it passes POST.
> Just for the sake of refreshing context, I have both OBP versions 1.0
> and 2.9.  1.0 is failing POST, but 2.9 appears fully functional
> except for this issue with booting.

That _is_ disturbing.  If the hardware has failed, it's failed really
subtly.

Anyway, I hope this helps in some way.

/~\ The ASCII				  Mouse
\ / Ribbon Campaign
 X  Against HTML		mouse at rodents-montreal.org
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B