[731] in NetBSD-Development
Re: New 1.0A installation procedure
daemon@ATHENA.MIT.EDU (ghudson@MIT.EDU)
Tue Apr 25 15:27:53 1995
From: ghudson@MIT.EDU
Date: Tue, 25 Apr 1995 15:26:09 -0400
To: Greg Hudson <ghudson@MIT.EDU>
Cc: netbsd-dev@MIT.EDU
In-Reply-To: "[729] in NetBSD-Development"
> - At boot time, the script hangs during or after
> config_afs (I haven't determined exactly where). If
> you hit ^C, it skips the rest of rc.athena. I have
> the same problem on glacier right now, and I haven't
> figured out what the cause is yet.
The cause was a faulty version of awk; I'm not sure how it got
generated in my build tree, but it had at least two problems (in this
case, it was waiting for input even though it had a filename argument;
it also got an internal error in another invocation), and when I
rebuilt awk, it worked fine.
> - xlogin can't find its resources correctly, and looks
> kind of funky.
I had forgotten to put the contents of /etc/athena/login/bitmaps in
the athena packing list. Fixed.
> - attach will dump core with the attach.conf we
> distribute.
This was due to someone doing "#define re_comp regcomp" etc. in
attach's config.c. I nuked those defines and had it link with
-lcompat, and it works fine now.
> - Even after nuking attach.conf, when I tried to log
> in as sipbtest, it couldn't attach sipbtest's home
> directory.
I had forgotten to make attach setuid in the Athena install script.
Other problems fixed: I had made "MAKEDEV all" generate ttyv0 through
ttyv5, but not ttyv6; I had the wrong versions of libafs.o; the NetBSD
distribution makes /root/.klogin and /etc/hosts.equiv files with bogus
stuff in them (I nuked hosts.equiv and made .klogin an empty file);
the floppy devices aren't world-readable and -writable after the
Athena installation (they still aren't unless you do the Athen
installation).
We should have a working distribution now. I'm rebooting the
installation and athena kernels to include an autodetecting ed0 driver
(in addition to ed1, ed2, and ed3 at the old ports and IRQs, since
some ed devices aren't autoprobed) as well as fe0, fe1, and fe2
devices at three common ports (the fe driver doesn't appear to
autoprobe at multiple ports).
The next step is to write some installation documentation. We should
be able to focus more on partitioning and X configuration this time,
since the actual installation procedure is less complicated.
