Penetration Testing mailing list archives

RE: Client/Server application that does not authenticate users


From: "Dinis Cruz" <dinis () ddplus net>
Date: Sat, 14 Aug 2004 00:49:37 +0100

Then you can just write a little script to highjack (i.e. 'patch') that
function in the OS (or in the application's exe) and you will be able to
impersonate who ever you want without rebooting into another user.

This could also be used to do an automated brute force username attack
(since you don't need passwords)

Dinis Cruz
.Net Security Consultant 
DDPlus 

-----Original Message-----
From: Brian Erdelyi [mailto:brian_erdelyi () yahoo com]
Sent: 13 August 2004 23:41
To: Dinis Cruz
Subject: RE: Client/Server application that does not authenticate users

I had thought of this as well consider the application
makes use of environment viarables.  I did test and
confirmed it is not using the Windows environment
variable "USERNAME".

The vendor has reported thay use a WIN32 API called
GetUserName.


--- Dinis Cruz <dinis () ddplus net> wrote:

I knew of an web app that got the username for the
user variable "Username"

Guess what would happen in you typed in the client
workstation "Set
Username=Admin" :)

For guidelines check out the OWASP documents: Top 10
(http://www.owasp.org/documentation/topten.html),
Testing guide
(http://www.owasp.org/documentation/testing.html),
the ISO 17799 Project
(http://www.owasp.org/standards/iso17799.html) and
the app sec FAQ
(http://www.owasp.org/documentation/faq.html)

Hope this helps

Best regards

Dinis Cruz
.Net Security Consultant
DDPlus


-----Original Message-----
From: Brian Erdelyi
[mailto:brian_erdelyi () yahoo com]
Sent: 13 August 2004 11:58
To: Dinis Cruz; pen-test () securityfocus com
Subject: RE: Client/Server application that does
not authenticate users

I am working with the vendor on this.
Unfortunately,
I was assured by the cendor that the application
does
authenticate users and uses accesscontrol lists to
assign permissions.  They claimed I was was using
an
uncommon interpretation of the term
"authentication".
The next level of support disagreed with my use of
the
term "vulnerability".

The server does ask for a username (the client
automatically forwards the Windows username of the
currently logged on computer) but no password is
requested or sent at any point.  This is by design
of
the application (which from my perspective is
seriously flawed for an application that allows
users
to sell and trade millions of dollars worth of
bonds).

I will give the vendor some time to analyse the
description I have provided to them and respond.

I'd like to provide some very specific suggestions
and
guidance on how other applications are designed
and
coded to authenticate users.

Is there an RFC on secure programming?



--- Dinis Cruz <dinis () ddplus net> wrote:

Quite common.

The other major mistake that most do is to rely
on
the Client's GUI to
enforce the 'security boundaries' of the client
application (for example:
they rely on the fact that the user's GUI
doesn't
have the functionality to
change passwords (including the administrators),
so
if such a request is
made it must be from a valid source....)

But, the big question is: "what happens next?"

Are they going to tell their customers that
their
data could had been (or
was) compromised?

Dinis Cruz
.Net Security Consultant
DDPlus




__________________________________
Do you Yahoo!?
Yahoo! Mail is new and improved - Check it out!
http://promotions.yahoo.com/new_mail







__________________________________
Do you Yahoo!?
Yahoo! Mail - 50x more storage than other providers!
http://promotions.yahoo.com/new_mail



Current thread: