Hello All,
He's some major holes that I have found in the Netmanage Chameleon
tools. Forwarded the info to Netmanange a few weeks ago, but no
response from them on patches and such.
All seem to exist in the older Chameleon 4.5 as well as the newer
Unixlink 97 tools. Most of the testing was done with NetCat for NT
on NT 3.51 and 4.0
Notes: Anything listed as a 'Buffer Overflow' means that a NT Dr.Watson
message was produced with the 'Exception: access violation' message. This
may or may not be an exploitable buffer overflow condition, but it
definitely looks like the programs are not always doing sanity checks on
user input.
1 - FTP server. You must have at least one user defined on the server.
-- Buffer overflows with username. Username needs more than 150 chars
to overrun. Very similar to the WAR FTPd probs.
-- passwd with lots of chars causes a 'local error processing' to
scroll on the screen.
2 - HTTP server [personal web server]. Not sure what exactly is
happening here, but if a URL request longer than 519 chars is
submitted to the server, it spontaneously unloads.....never produces
an error message.
example: GET more_than_519_characters<cr><cr>
3 - Email/Zmail -- The email package comes with both client and server
functions. POP3d and SMTPd are enabled while the email client is active.
POP3d
-- buffer overflow with 'USER username' and username over 152 chars
-- buffer overflow with 'PASS passwd' and password over 104 chars
-- buffer overflows with all of the commands [list, retr, dele, quit].
Don't even have to log in. Even QUIT with a bunch of garbage after
it will cause the POP3d to crash..........
SMTPd
-- buffer overflow with 'HELO hostname' and hostname over 471 chars.
-- buffer overflow with 'HELP topic' and topic over 514 chars.
4 - Finger client -- If you setup netcat to listen on the finger port,
and send back a reply of over 257 chars to any finger request from the
Chameleon client, an overflow will occur at the finger
client....strange, but who really uses finger anyway.... ;)
These are the only utilities that I have really looked into -- But
they all seem to have problems with validity checks. I have not
built test exploits yet, but there is a definate possibility that
some of these bugs are exploitable. They are definately DOS bugs.
Regards,
Anton Rager
arager@McGraw-Hill.com
This archive was generated by hypermail 2b30 : Fri Apr 13 2001 - 13:52:11 PDT