ammar_secret
07-23-2004, 04:57 PM
* ADM CONFIDENTIAL -- (ADM Confidential Restricted when
* combined with the aggregated modules for this product)
* OBJECT CODE ONLY SOURCE MATERIALS
* (C) COPYRIGHT ADM Crew. 1999
* All Rights Reserved
*
* This module may not be used, published, distributed or archived without
* the written permission of the ADM Crew. Please contact your local sales
* representative.
*
* ADM named 8.2/8.2.1 NXT remote overflow - horizon/plaguez
*
* "a misanthropic anthropoid with nothing to say"
*
* thanks to stran9er for sdnsofw.c
*
* Intel exploitation is pretty straightforward.. should give you a remote
* shell. The shellcode will break chroot, do a getpeername on all open
* sockets, and dup to the first one that returns AFINET. It also forks and
* runs a command in case the fd duping doesn't go well. Solaris/SPARC is a
* bit more complicated.. we are going through a well trodden part of the
* code, so we don't get the context switch we need to have it populate the
* register windows from the stack. However, if you just hammer the service
* with requests, you will quickly get a context switch at the right time.
* Thus, the SPARC shellcode currently only breaks chroot, closes current
* fd's and runs a command.
* Also, the NetBSD shellcode doesn't break chroot because they stop the
* dir tricks. Of course, they allow mknods in chrooted environments, so
* if named is running as root, then it still might be expoitable.
* The non-exec stack patch version returns into a malloc'ed buffer, whose
* address can vary quite alot. Thus, it may not be as reliable as the other
* versions..
Thanks to Behr00z_ice ([Only registered and activated users can see links])
* combined with the aggregated modules for this product)
* OBJECT CODE ONLY SOURCE MATERIALS
* (C) COPYRIGHT ADM Crew. 1999
* All Rights Reserved
*
* This module may not be used, published, distributed or archived without
* the written permission of the ADM Crew. Please contact your local sales
* representative.
*
* ADM named 8.2/8.2.1 NXT remote overflow - horizon/plaguez
*
* "a misanthropic anthropoid with nothing to say"
*
* thanks to stran9er for sdnsofw.c
*
* Intel exploitation is pretty straightforward.. should give you a remote
* shell. The shellcode will break chroot, do a getpeername on all open
* sockets, and dup to the first one that returns AFINET. It also forks and
* runs a command in case the fd duping doesn't go well. Solaris/SPARC is a
* bit more complicated.. we are going through a well trodden part of the
* code, so we don't get the context switch we need to have it populate the
* register windows from the stack. However, if you just hammer the service
* with requests, you will quickly get a context switch at the right time.
* Thus, the SPARC shellcode currently only breaks chroot, closes current
* fd's and runs a command.
* Also, the NetBSD shellcode doesn't break chroot because they stop the
* dir tricks. Of course, they allow mknods in chrooted environments, so
* if named is running as root, then it still might be expoitable.
* The non-exec stack patch version returns into a malloc'ed buffer, whose
* address can vary quite alot. Thus, it may not be as reliable as the other
* versions..
Thanks to Behr00z_ice ([Only registered and activated users can see links])