[792] in Kerberos-V5-bugs
arrrrgh...
daemon@ATHENA.MIT.EDU (*Hobbit*)
Thu Sep 29 05:04:37 1994
Date: Thu, 29 Sep 1994 05:04:28 -0400
From: *Hobbit* <hobbit@asylum.sf.ca.us>
To: krb5-bugs@MIT.EDU
I'm completely thrashing over here, and am wondering if anyone cares about
this sort of bug report or not. I am probably doing something bizarre
in the interest of experimentation, and certainly to avoid interfering with
normal system operations here until things are proven and stable -- but my
"experimental" setup seems to be wreaking havoc with beta4 patch 2. In
particular, I'm getting the KDC to dump core in the following way:
core file = core -- program ``krb5kdc''
SIGSEGV 11: segmentation violation
_find_alternate_tgs(0x5ae00,0xefffe7b4,0xefffe6ff,0xefffe700,0x1,0x0) + d0
_process_tgs_req(0xeffff910,0x5ae00,0x0,0xeffff90c,0xefffe90c,0x6c) + 3a0
_dispatch(0xeffff910,0xeffff938,0x0,0xeffff90c,0xefffe90c,0xf95) + bc
_process_packet(0x6,0xeffffb46,0x0,0x0,0x0,0x40) + 114
_listen_and_process(0x40,0x7,0x0,0x1,0x80,0x0) + 144
_main(0x2,0xeffffa9c,0xeffffaa8,0x48400,0x0,0x0) + 16c
This happens when I try to get a ticket for any service with instance
"asylum.sf.ca.us" -- thus, "rlogin -x asylum.sf.ca.us" causes it, and
"sserver 9999 & ; sclient asylum.sf.ca.us 9999" causes it.
The environment is krb5-beta4-patchlevel-2 on a Sunos 4.1.3 SS10, built with
krb4 compatibility included. I've included copies of various files below in
case they're relevant. Again, these are experimental, so they aren't going to
look like a live system.
Other random bitching, and there's a lot more where this came from:
Would someone please provide me [and maybe the newsgroup] with a list
of all the necessary v5 application server principal names, so that I don't
have to dig them out of the source code and possibly get it wrong? Trying
to create a srvtab in the dark is really annoying, and I still haven't got
it quite right.
What do people do when they want to extract a largish srvtab for a given
host? Do they honestly have to type "xst hostname svca svcb svcc svcd svce
svcf svcg" and hope they got them all, or is there a way to pull all of
them out using just the instance name?
I would also appreciate knowing where to find the file containing the true
definition of "mit_des_debug", so I can build the entire tree with -DDEBUG
turned on. [Try it. You'll see where it breaks.]
_H*
=== krb.conf ===
FUCKME.EDU
FUCKME.EDU asylum.sf.ca.us admin server
ATHENA.MIT.EDU kerberos.mit.edu admin server
ATHENA.MIT.EDU kerberos-1.mit.edu
ATHENA.MIT.EDU kerberos-2.mit.edu
ATHENA.MIT.EDU kerberos-3.mit.edu
LCS.MIT.EDU kerberos.lcs.mit.edu admin server
IFS.UMICH.EDU kerberos.ifs.umich.edu
CS.WASHINGTON.EDU hawk.cs.washington.edu
CS.WASHINGTON.EDU aspen.cs.washington.edu
CS.BERKELEY.EDU okeeffe.berkeley.edu
MEDIA.MIT.EDU kerberos.media.mit.edu
NEAR.NET kerberos.near.net
CATS.UCSC.EDU mehitabel.ucsc.edu admin server
CATS.UCSC.EDU ucsch.ucsc.edu
=== krb.realms ===
.com FUCKME.EDU
.us FUCKME.EDU
localhost FUCKME.EDU
=== kdb5_edit output ===
Enter KDC database master key:
entry: rcmd/asylum.sf.ca.us@FUCKME.EDU
entry: kcmd/asylum.sf.ca.us@FUCKME.EDU
entry: krbtgt/FUCKME.EDU@FUCKME.EDU
entry: hobbit@FUCKME.EDU
entry: K/M@FUCKME.EDU
entry: kadmin/FUCKME.EDU@FUCKME.EDU
entry: a/b@FUCKME.EDU
