Skip Menu |
 

Subject: alpha-osf 1.3-branch nightly build test failures: rpc_test test suite errors out
=== rpc_test tests ===

Schedule of variations:
unix

Running target unix
Using /usr/local/share/dejagnu/baseboards/unix.exp as board description
file for target.
Using /usr/local/share/dejagnu/config/unix.exp as generic interface file
for target.
Using ./config/unix.exp as tool-and-target-specific interface file.
ERROR: tcl error sourcing library file ./lib/helpers.exp.
child process exited abnormally
child process exited abnormally
while executing
"exec $env(QUALNAME)"
invoked from within
"if ![info exists CANON_HOST] {
set CANON_HOST [exec $env(QUALNAME)]
setup_database
file delete $env(RPC_TEST_SRVTAB)
exec $env(MAKE_KE..."
(file "./lib/helpers.exp" line 116)
invoked from within
"source ./lib/helpers.exp"
("uplevel" body line 1)
invoked from within
"uplevel #0 source ./lib/helpers.exp"
invoked from within
"catch "uplevel #0 source ${dir}/${initfile}" error"

The program we're running during setup seems to be bombing out fairly
reliably....
That's in the shared-library builds only; the static-library builds seem
to be doing fine.
Quick analysis:

The QUALNAME environment variable points to a perl script from
kadmin/testing/scripts, modified at configuration or build time to point
to the local perl executable. On Tru64, that perl executable is set to
load a dynamic library libperl.so from a non-standard directory.

In the shared library tests, the environment is tweaked to force our own
shared libraries to be found first. I believe the way this is done has
the side effect of preventing libperl.so from being found at all.

So we probably need either (1) a quick hack to add the perl library
directory to the library search path, or (2) a proper fix to make our
libraries be found first, but still permit any directories configured
into executables to be searched after ours.
To: rt-comment@krbdev.mit.edu
Subject: Re: [krbdev.mit.edu #1793]
From: Tom Yu <tlyu@mit.edu>
Date: Wed, 03 Sep 2003 17:32:15 -0400
RT-Send-Cc:
Download (untitled)
message/rfc822 4.3KiB
Return-Path: <krb5-bugs-bounces@MIT.EDU>
Received: from po9.mit.edu (po9.mit.edu [18.7.21.65])
by po9.mit.edu (Cyrus v2.1.5) with LMTP; Sat, 30 Aug 2003 19:56:51 -0400
X-Sieve: CMU Sieve 2.2
Received: from pacific-carrier-annex.mit.edu by po9.mit.edu (8.12.4/4.7) id
h7UNupOS015077; Sat, 30 Aug 2003 19:56:51 -0400 (EDT)
Received: from pch.mit.edu (PCH.MIT.EDU [18.7.21.90])
by pacific-carrier-annex.mit.edu (8.12.4/8.9.2) with ESMTP id
h7UNujGr004399;
Sat, 30 Aug 2003 19:56:45 -0400 (EDT)
Received: from pch.mit.edu (localhost [127.0.0.1])
by pch.mit.edu (8.12.8p1/8.12.8) with ESMTP id h7UNuXk3010273;
Sat, 30 Aug 2003 19:56:34 -0400 (EDT)
Received: from pacific-carrier-annex.mit.edu (PACIFIC-CARRIER-ANNEX.MIT.EDU
[18.7.21.83])
by pch.mit.edu (8.12.8p1/8.12.8) with ESMTP id h7UNuUk0010269
for <krb5-bugs@PCH.mit.edu>; Sat, 30 Aug 2003 19:56:30 -0400 (EDT)
Received: from hermes.ctd.anl.gov (hermes.ctd.anl.gov [130.202.113.27])
h7UNuTGr004310
for <krb5-prs@mit.edu>; Sat, 30 Aug 2003 19:56:30 -0400 (EDT)
Received: from hermes.ctd.anl.gov (localhost [127.0.0.1])
by hermes.ctd.anl.gov (8.9.1a/8.9.1) with ESMTP id SAA06143
for <krb5-prs@mit.edu>; Sat, 30 Aug 2003 18:56:29 -0500 (CDT)
Received: from anl.gov (atalanta.ctd.anl.gov [146.137.194.4])
by hermes.ctd.anl.gov (8.9.1a/8.9.1) with ESMTP id SAA06134;
Sat, 30 Aug 2003 18:56:28 -0500 (CDT)
Message-ID: <3F513978.F3BAC1F4@anl.gov>
Date: Sat, 30 Aug 2003 18:55:36 -0500
From: "Douglas E. Engert" <deengert@anl.gov>
X-Mailer: Mozilla 4.79 [en] (Windows NT 5.0; U)
X-Accept-Language: en
To: rt-comment@krbdev.mit.edu
References: <rt-1793-8083.11.4004234894577@krbdev.mit.edu>
cc: krb5-prs@mit.edu
Subject: Re: [krbdev.mit.edu #1793] alpha-osf 1.3-branch nightly build
testfailures: rpc_test test suite errors out
X-BeenThere: krb5-bugs@mit.edu
X-Mailman-Version: 2.1
Precedence: list
List-Id: People interested in Kerberos 5 bugs <krb5-bugs.mit.edu>
List-Help: <mailto:krb5-bugs-request@mit.edu?subject=help>
List-Post: <mailto:krb5-bugs@mit.edu>
List-Subscribe: <http://mailman.mit.edu/mailman/listinfo/krb5-bugs>,
<mailto:krb5-bugs-request@mit.edu?subject=subscribe>
List-Archive: <http://mailman.mit.edu/pipermail/krb5-bugs>
List-Unsubscribe: <http://mailman.mit.edu/mailman/listinfo/krb5-bugs>,
<mailto:krb5-bugs-request@mit.edu?subject=unsubscribe>
Sender: krb5-bugs-bounces@MIT.EDU
Errors-To: krb5-bugs-bounces@MIT.EDU
X-Spam-Score: -3.9
X-Spam-Flag: NO
X-Scanned-By: MIMEDefang 2.28 (www . roaringpenguin . com / mimedefang)
Lines: 50
MIME-Version: 1.0

This brings up a nit I have had with almost all packages that build
shared libs. They all assume that they can derived the -rpath from
the -L or they don't support -rpath. (or -R or what ever.)
This is not always the case, and it may leave build directories in the
-rpath which could be a security risk.

If you are making changes in this area, consider adding a configure
option to allow one to specify the -rpath to be used.

This in not as critical with Kerberos, as it does not depend on
any other packages. But packages like OpenSSH depend on OpenSSL,
Kerberos, wrapper, etc. And it is common to build and install them
as a group.

Just an idea.


Ken Raeburn via RT wrote:
Show quoted text
>
> Quick analysis:
>
> The QUALNAME environment variable points to a perl script from
> kadmin/testing/scripts, modified at configuration or build time to point
> to the local perl executable. On Tru64, that perl executable is set to
> load a dynamic library libperl.so from a non-standard directory.
>
> In the shared library tests, the environment is tweaked to force our own
> shared libraries to be found first. I believe the way this is done has
> the side effect of preventing libperl.so from being found at all.
>
> So we probably need either (1) a quick hack to add the perl library
> directory to the library search path, or (2) a proper fix to make our
> libraries be found first, but still permit any directories configured
> into executables to be searched after ours.
> _______________________________________________
> krb5-bugs mailing list
> krb5-bugs@mit.edu
> http://mailman.mit.edu/mailman/listinfo/krb5-bugs

--

Douglas E. Engert <DEEngert@anl.gov>
Argonne National Laboratory
9700 South Cass Avenue
Argonne, Illinois 60439
(630) 252-5444
Show quoted text
_______________________________________________
krb5-bugs mailing list
krb5-bugs@mit.edu
http://mailman.mit.edu/mailman/listinfo/krb5-bugs
To: rt@krbdev.mit.edu
Subject: Re: [krbdev.mit.edu #1793]
From: Tom Yu <tlyu@mit.edu>
Date: Wed, 03 Sep 2003 17:33:46 -0400
RT-Send-Cc:
This probably needs the same sort of _RLD_ROOT hack I was proposing
earlier for Irix... basically, ensure that "/" is in _RLD_ROOT
following a different directory that we construct that looks like the
installed tree.
To: "Douglas E. Engert" <deengert@anl.gov>
Cc: rt-comment@krbdev.mit.edu
Subject: Re: [krbdev.mit.edu #1793] alpha-osf ...
From: Ken Raeburn <raeburn@MIT.EDU>
Date: Wed, 03 Sep 2003 22:38:28 -0400
RT-Send-Cc:
[subject truncated because RT is broken]

"Douglas E. Engert" <deengert@anl.gov> writes:
Show quoted text
> This brings up a nit I have had with almost all packages that build
> shared libs. They all assume that they can derived the -rpath from
> the -L or they don't support -rpath. (or -R or what ever.)
> This is not always the case, and it may leave build directories in the
> -rpath which could be a security risk.

Is this actually a problem with the MIT krb5 code? I think we're
using -rpath/-R to specify the library install directory should be
used at run time, and not the build directory. At least, if there's a
platform where we could do it and we're not, it's a bug.

Hm, I wonder how to best test that.... Shouldn't be too complicated,
we can probably even work it into a configure script or "make check"
or something.

Ken
From: tlyu@mit.edu
Subject: CVS Commit
Implement hack for faking up _RLD_ROOT with a shadow of the directory
tree up to the installed "lib" directory. This helps with running
tests on Tru64 and Irix.


To generate a diff of this commit:



cvs diff -r5.191 -r5.192 krb5/src/config/ChangeLog
cvs diff -r1.92 -r1.93 krb5/src/config/pre.in
cvs diff -r5.14 -r5.15 krb5/src/config/shlib.conf
cvs diff -r1.48 -r1.49 krb5/src/kadmin/testing/scripts/ChangeLog
cvs diff -r1.19 -r1.20
krb5/src/kadmin/testing/scripts/env-setup.shin
cvs diff -r1.128 -r1.129 krb5/src/util/ChangeLog
cvs diff -r1.38 -r1.39 krb5/src/util/Makefile.in
From: tlyu@mit.edu
Subject: CVS Commit
pullup from trunk


To generate a diff of this commit:



cvs diff -r5.180.2.7 -r5.180.2.8 krb5/src/config/ChangeLog
cvs diff -r1.90.2.2 -r1.90.2.3 krb5/src/config/pre.in
cvs diff -r5.11.2.1 -r5.11.2.2 krb5/src/config/shlib.conf
cvs diff -r1.46 -r1.46.2.1
krb5/src/kadmin/testing/scripts/ChangeLog
cvs diff -r1.18 -r1.18.2.1
krb5/src/kadmin/testing/scripts/env-setup.shin
cvs diff -r1.121.2.4 -r1.121.2.5 krb5/src/util/ChangeLog
cvs diff -r1.37 -r1.37.2.1 krb5/src/util/Makefile.in
From: tlyu@mit.edu
Subject: CVS Commit
pull up missed change due to merge botch


To generate a diff of this commit:



cvs diff -r1.37.2.1 -r1.37.2.2 krb5/src/util/Makefile.in