[1621] in Kerberos-V5-bugs
[Wayne Schroeder: Kerberos Mods from SDSC]
daemon@ATHENA.MIT.EDU (Theodore Ts'o)
Sat Sep 9 00:03:59 1995
Date: Sat, 9 Sep 1995 00:03:54 -0400
From: Theodore Ts'o <tytso@MIT.EDU>
To: krb5-bugs@MIT.EDU
------- Forwarded Message
00472; Thu, 7 Sep 1995 16:38:43 -0700
From: Wayne Schroeder <schroede@SDSC.EDU>
Date: Thu, 7 Sep 95 16:38:43 PDT
To: tytso@MIT.EDU
Cc: schroede@franklin.sdsc.edu
Subject: Kerberos Mods from SDSC
Hi,
I was told that we should send our Kerberos (5 beta 4 rel 3) patches
to you for possible inclusion in the MIT base. Tom Hutton, also of
the San Diego Supercomputer Center, talked with Jeff at MIT and was
given your name.
A few of us have been working with Kerberos 5.4.3, off and on, for the
last 8 or 9 months getting the libraries and tools to work on many of
our various architectures. Mostly this has involved debug and
development for the Cray (ifdef _UNICOS or CRAY), SGI (sgi), DEC Alpha
(__alpha), and IBM RS6000 (_AIX or _IBMR2). We've been using it some
in friendly user production and hope to make more use of it soon.
We're excited about it as an important contribution to our (and other
NSF centers' and the community's) security.
I'm presenting a paper on it at a Cray User Group conference later
this month. If you want, you can view my current draft of the paper
at:
http://www.sdsc.edu/SDSC/Staff/schroede/kerberos_cug.html
We are a little disorganized now as Tom Hutton, who was leading the
project, is no longer able to do so. In his place, I've attempted to
generate a usable listing of our mods using the cvs context diff
command (ie 'cvs diff -c -r1.1'). I believe this is all, or at least
most, of our mods.
We also hacked a few Makefiles directly to work around a few problems.
Since you folks have changed the auto configure system, some of these
may be solved. When we get to 5.6, we may need to redo them and fit
the changes into your current scheme.
In a message following this, I'll send you the mods. If these are not
clear or usable for any reason, let me know and I'll try to send you
what you need. Using cvs diff was the easiest, but if you need more
context or something, I think I could run a series of diffs on some
reference files.
For some reason, the cvs diff command generated quite a few diffs for
readme and documentation files (such as "/admin/destroy/kdb5_destroy.M,v"),
which seemed to be caused by some problem we had in setting up CVS.
I've edited them out. The following is an example:
===================================================================
RCS file: /projects/security/kerberos/krb5B4-3/Master/krb5B4/admin/aname/kdb5_anadd.M,v
retrieving revision 1.1
diff -c -r1.1 kdb5_anadd.M
*** 1.1 1995/01/14 00:23:56
--- kdb5_anadd.M 1995/01/14 00:23:57
***************
*** 1,6 ****
.\" $Source: /projects/security/kerberos/krb5B4-3/Master/krb5B4/admin/aname/kdb5_anadd.M,v $
.\" $Author: hutton $
! .\" $Id: kdb5_anadd.M,v 1.1 1995/01/14 00:23:56 hutton Exp $
.\" Copyright 1990 by the Massachusetts Institute of Technology.
.\"
.\" Export of this software from the United States of America may
--- 1,6 ----
.\" $Source: /projects/security/kerberos/krb5B4-3/Master/krb5B4/admin/aname/kdb5_anadd.M,v $
.\" $Author: hutton $
! .\" $Id: kdb5_anadd.M,v 1.1.1.1 1995/01/14 00:23:57 hutton Exp $
.\" Copyright 1990 by the Massachusetts Institute of Technology.
.\"
.\" Export of this software from the United States of America may
===================================================================
The mods to "/util/autoconf/autoconf.info,v" and
"util/autoconf/texinfo.tex,v" are like these, but a bit different, so
I left them in. Don't know if they are relevant though.
There were some Cray ifdefs in 5 beta 4 release 3 for Unicos 6.1
(ifdef unicos61), some of which I needed for our Cray C90 (running
Unicos 8.1) so I changed those to ifdef _unicos.
For some of the Cray changes, I used some Cray Research source (like
10 or 20 lines a few times). For these, I have the included comment
"/* CRI based code */ " (just for reference). I have checked with CRI
management on releasing their code (it's not much code, but...) to
you and the world, and they have given us permission. I can forward
you the email if you want.
The DEBUG_MARK code came in handy when debugging, particularly on the
IBM, and we left it in (its ifdef controlled) for now.
I changed 'suspend' to 'zsuspend' to avoid a conflict with the Cray
definitions.
I've edited a few of the comments in the mods to make them more
accurately reflect the current status. For example, some of our
changes were made first for the Cray, and later used on other systems
too but without changing comments. In the mods I'll send you, I've
corrected some comments like these.
There are about 36 changed sections and, with context, it is about
2,970 lines.
I hope this is not too late for inclusion in your next release.
Let me know if you have any questions or problems with it.
Thanks,
Wayne Schroeder
San Diego Supercomputer Center
http://www.sdsc.edu/SDSC/Staff/schroede/Home.html
------- End Forwarded Message
