[Vol-users] One-byte Modification for Breaking Memory Forensic Analysis.

Michael Cohen scudette at gmail.com
Mon Apr 30 14:11:38 CDT 2012


Hi George,

  Thanks for sharing this paper. I could not read the actual paper but
found a cached version of the presentation. From my reading these guys
simply broke some of the automatic functionality offered by the tools.
Clearly if an investigator has no idea of what they are doing and want
to just fire and forget the tool, this will stop them in their track
but for people who actually know how to use the tool this is not
really an obstacle. Certainly with volatility you can override many of
these parameters (eg. --dtb --profile) and so this is kind of
irrelevant.

For example, the first step was to prevent the tool autodetecting the
DTB by scanning for the idle process signature. The DTB should still
be correct, but the scanner might get confused due to changing some of
the header parameters. In reality scanning for any kernel process
should be able to yield the DTB and even a statistical technique can
be used by simply recovering all the DTBs of all the processes (the
kernel DTB is found in more than one process, but a process DTB is
pretty unique for each process).

In addition, the volatility approach is to ask the user to specify the
profile, since presumably the investigator already knows what OS they
captured. The automatic guessing functionality in volatility is not
really needed (it makes things easier) but ultimately it just offers a
bunch of suggestions. (And currently it is actually implemented by
recognizing the KDBG structure which is not actually critical to the
running of the system so the rootkit can very easily change it).

Of course the author could have done lots of antiforensic techniques,
like change pool tags etc, and this has been known for a long time. In
reading the presentation I didnt feel there was anything actually new
here - perhaps we need better documentation, to  help people to
overcome these simple antiforensic efforts, but they have been known
for ages and are not very effective. (I do not claim however that
there are not other more effective anti forensic measures - which of
course there are).

Disappointingly the author did not add to any efforts with
antiacquisition techniques, other than just talk about other efforts.
For example, the shadow walking technique is easy to bypass simply by
resetting CR3 (This flushes the TLBs essentially resyncing them).
Despite this there is a lot of further work to be done in the
antiacquisition space (For example its easy for a rootkit to exclude
entire regions from acquisition tools by simply adding them to the DMA
resource tree - which most acquisition tools avoid so as not to crash
the system).

Again thanks for sharing,
Michael.

On 30 April 2012 20:21, George M. Garner J.r (online)
<ggarner_online at gmgsystemsinc.com> wrote:
> https://media.blackhat.com/bh-eu-12/Haruyama/bh-eu-12-Haruyama-Memory_Forensic-Slides.pdf.
>
> Some people are having trouble accessing the link which I previously post
> (and for others it works just fine).  If anyone is unable to access the link
> and wants a copy of this paper they may contact me offline for assistance.
>
> Sorry for the inconvenience.
>
> _______________________________________________
> Vol-users mailing list
> Vol-users at volatilityfoundation.org
> http://lists.volatilityfoundation.org/mailman/listinfo/vol-users


More information about the Vol-users mailing list