(Fixed Kerberos 5 Telnet Daemon is Released) Re: Telnet Daemons May Give Remote Users Root Level Access Privileges
SecurityTracker Alert ID: 1002121|
SecurityTracker URL: http://securitytracker.com/id/1002121
(Links to External Site)
Date: Aug 1 2001
Execution of arbitrary code via network, Root access via network, User access via network|
Fix Available: Yes Vendor Confirmed: Yes |
TESO reported that many BSD-derived Telnet daemons (servers) contain a vulnerability that may allow a remote user to obtain root level access on the server.|
The vulnerability is reportedly due to a buffer overflow in the telnet option handling.
The following systems are reported to be vulnerable:
BSDI 4.x default, FreeBSD .x default, IRIX 6.5, Linux netkit-telnetd < 0.14, NetBSD 1.x default, OpenBSD 2.x, Solaris 2.x sparc, and "almost any other vendor's telnetd".
A remote user can send a specially formatted option string to the remote telnet server and overwrite sensitive memory, causing arbitrary code to be executed with the privileges of the telnet server (which is typically root level privileges).
Telnet options are reportedly processed by the 'telrcv' function. The results of the parsing, which are to be send back to the client, are stored in the 'netobuf' buffer. It is apparently assumed that the reply data is smaller than the buffer size, so no bounds checking is performed. By using a combination of options, especially the 'AYT' Are You There option, it is possible for a remote user to append data to the buffer. It is reported that the characters that can be written to the buffer are limited, which makes constructing a successful exploit difficult.
The report states that a working exploit has been developed for BSDI, NetBSD and FreeBSD. However, the exploit was not released.
A remote user can execute arbitrary code on the server with the privileges of the telnet server, which is typically root level privileges.|
There is a fix for the telnet daemon distributed with Kerberos 5. See the Source Message for the update information.|
|Underlying OS: Linux (Any), UNIX (Any)|
|Underlying OS Comments: many Linux and Unix OSs are vulnerable, but not all - see the Alert text for more information|
This archive entry is a follow-up to the message listed below.|
Source Message Contents
Subject: security advisory: krb5 telnetd buffer overflows|
-----BEGIN PGP SIGNED MESSAGE-----
KRB5 TELNETD BUFFER OVERFLOWS
Buffer overflows exist in the telnet daemon included with MIT krb5.
Exploits are believed to exist for various operating systems on at
least the i386 architecture.
If telnetd is running, a remote user may gain unauthorized root
* MIT Kerberos 5, all releases to date.
The recommended approach is to apply the appropriate patches and to
rebuild your telnetd. Patches for the krb5-1.2.2 release may be found
The associated detached PGP signature is at:
These patches might apply successfully to older releases with some
amount of fuzz.
Please note that if you are using GNU make to build your krb5 sources,
the build system may attempt to rebuild the configure script from the
changed configure.in. This may cause trouble if you don't have
autoconf installed properly. To prevent this, you should use the
touch command or some similar means to ensure that the file
modification time on the configure script is newer than that of the
If you are unable to patch your telnetd, you may should disable the
telnet service altogether.
This announcement and code patches related to it may be found on the
MIT Kerberos security advisory page at:
The main MIT Kerberos web page is at:
Thanks to TESO for the original alert / Bugtraq posting.
Thanks to Jeffrey Altman for assistance in developing these patches.
A buffer overflow bug was discovered in telnet daemons derived from
BSD source code. Since the telnet daemon in MIT krb5 uses code
largely derived originally from BSD sources, it too is vulnerable.
By carefully constructing a series of telnet options to send to a
telnet server, a remote attacker may exercise a bug relating to lack
of bounds-checking, causing an overflow of a fixed-size buffer. This
overflow may possibly force the execution of malicious code.
It is not known how difficult this vulnerability is to exploit, since
the buffer is not on the stack. Some discussion seems to indicate
that exploits exist for this vulnerability that are believed to work
against various operating systems for i386-based machines. It is not
known whether these existing exploits have been successfully ported to
-----BEGIN PGP SIGNATURE-----
Version: PGP 6.5.8
-----END PGP SIGNATURE-----