elhacker.net cabecera Bienvenido(a), Visitante. Por favor Ingresar o Registrarse
¿Perdiste tu email de activación?.

 

 


Tema destacado: Trabajando con las ramas de git (tercera parte)


+  Foro de elhacker.net
|-+  Programación
| |-+  Ingeniería Inversa (Moderadores: karmany, .:UND3R:., MCKSys Argentina)
| | |-+  Nueva generacion de antidebuggers?
0 Usuarios y 1 Visitante están viendo este tema.
Páginas: [1] Ir Abajo Respuesta Imprimir
Autor Tema: Nueva generacion de antidebuggers?  (Leído 3,928 veces)
TaU

Desconectado Desconectado

Mensajes: 184



Ver Perfil
Nueva generacion de antidebuggers?
« en: 22 Julio 2004, 22:19 pm »

Me he kedado asin :o al leer lo esto:

Citar
[FMADV] - OllyDbg Format String Bug

* Introduction:
There exists a format string bug in the code that handles Debugger
Messages in OllyDbg. This means any traced application can crash OllyDbg
and execute machine code.

* About (From the Webpage):
OllyDbg is a 32-bit assembler level analysing debugger for Microsoft
Windows. Emphasis on binary code analysis makes it particularly useful in
cases where source is unavailable.
 
OllyDbg is seen as an industry standard when it comes to analysing
vulnerabilties on win32 and it's easy to understand makes it a must for
anyone developing exploits on windows. Many people have sung the praises
of OllyDbg, including some very high profile engineers and exploit
developers.
 
* Technical details:
Typically OllyDbg attaches to a process and allows the user how to
customize the session; wether they trace, or they breakpoint some stuff or
whatever. The windows API is actually very debugger friendly and has many
functions to interact with debuggers (most likely built for their own
(safe) debugger WinDbg). One of these functions, OutputDebugString sends a
string directly to the debugger for interpretation, which OllyDbg displays to
the user via a status line along the bottom, sans a format specifier,
which means the user supplied string is used as the format specifier.

To reproduce this excellent bug, these steps can be taken:

1. Download Python (http://python.org) and win32com
(http://starship.python.net/crew/mhammond/win32/Downloads.html). These
two are _essential_ to any hacker's windows box.

2. Run 'python' so you get an interactive shell.

3. Attach to the 'python' process with OllyDbg, press 'F9' to continue
execution.

4. Type 'import win32api' and press enter in the python screen.

5. Type 'win32api.OutputDebugString("%s" * 50)' to crash OllyDbg.
Typically, if you have OllyDbg set as the JIT Debugger, another OllyDbg
screen will pop up ;) OR

6. Type 'win32api.OutputDebugString("%8.8x" * 15)' to view whats on the
stack!

7. The python process will now have died since OllyDbg died, so do the
process again!
 
If this is all too hard, or you don't believe ;) Then a screenshot for
your viewing pleasure is availiable at:
http://felinemenace.org/~nd/ollyfmt.png

Andrewg of FelineMenace managed to create a python script to exploit this
vulnerability, albeit with some shellcode problems :)
 
* Conclusion:
It certainly opens up the possibly for binaries to start owning their
debuggers, in an anti-reversing sense. GDB is a huge project too, with
multiple public/unpublished bugs. Because Debuggers work with the
executable in a state of execution, disassemblers such as IDA could be
vulnerable to a static attack of a malformed binary, much like the 
executable handling in the OpenBSD kernel i suppose. The possibilities are
endless! However there is a definate need for disclosure of these bugs, as
debuggers/disassembler are the first defense against the malicious.
 
* Greets:
TFM (Team FelineMenace), Greg + rootkit.com, people who spend their day
making sure imported beer is actually imported, peach.gotdns.org.

----
http://felinemenace.org/~nd

Origen

Desde luego si lo que dice es cierto habrá que ir con extremo cuidado a partir de ahora con los archivos que keramos examinar con el Olly.

Lo que no acabo de entender del todo como consigue el exploit que el Olly le ejecute código arbitrario....

En fin, que piensan del asunto?

Salu2
« Última modificación: 22 Julio 2004, 22:22 pm por TaU » En línea

"Si no se vive como se piensa, se acabará pensando como se vive", Pep Figueres
Revolucionario / Presidente de Costa Rica / Primer jefe de estado de la historia en abolir el ejército / Catalán.
www.wadalbertia.org  -<|¡^P
TaU

Desconectado Desconectado

Mensajes: 184



Ver Perfil
Re: Nueva generacion de antidebuggers?
« Respuesta #1 en: 28 Julio 2004, 20:03 pm »

Bueno, bueno, no hablen todos a la vez... de uno en uno, please...

¬_¬
En línea

"Si no se vive como se piensa, se acabará pensando como se vive", Pep Figueres
Revolucionario / Presidente de Costa Rica / Primer jefe de estado de la historia en abolir el ejército / Catalán.
www.wadalbertia.org  -<|¡^P
el-brujo
ehn
***
Desconectado Desconectado

Mensajes: 21.585


La libertad no se suplica, se conquista


Ver Perfil WWW
Re: Nueva generacion de antidebuggers?
« Respuesta #2 en: 29 Julio 2004, 01:20 am »

xDDDDDD
En línea

byebye


Desconectado Desconectado

Mensajes: 5.093



Ver Perfil
Re: Nueva generacion de antidebuggers?
« Respuesta #3 en: 29 Julio 2004, 01:57 am »

no veo pq hay que opinar, es simplemente un fallo que se tendria que solucionar en nuevas versiones.
En línea

heap

Desconectado Desconectado

Mensajes: 273



Ver Perfil WWW
Re: Nueva generacion de antidebuggers?
« Respuesta #4 en: 15 Febrero 2005, 05:30 am »

Citar
OutputDebugString sends a
string directly to the debugger for interpretation, which OllyDbg displays to
the user via a status line along the bottom, sans a format specifier,
which means the user supplied string is used as the format specifier.
Mira la cosa es ahi, cuando se le envia el string al debbuger y el debbuger(ollydbg) lo va a imprimir en la pantalla, necesita un specificador de fomato de string %s, al parecer no esta presente lo que abre la posibilidad a los desarrolladores de aplicacion de especificarle el especificador de formato, por ejemplo %n que escribe en la memoria, y asi redirigir la ejecucion a algun lado, hay mil cosas que pueden hacer, pero igual este bug se soluciona en 1 segundito y ademas los que desarrollan aplicaciones no van a utilizar un exploit, eso seria invertir en algo useless............a mi no me parece tan interesante ya que para llegar a explotar el bug, necesitan que el debbugeador haga ciertos pasos.....Pero bueno ahi esta tu explicacion de porque pueden correr codigo arbitrario
En línea

Páginas: [1] Ir Arriba Respuesta Imprimir 

Ir a:  

WAP2 - Aviso Legal - Powered by SMF 1.1.21 | SMF © 2006-2008, Simple Machines