Re: How to Test HDD Encryption

[Ansgar -59cobalt- Wiechers <bugtraq () planetcobalt net>]

On 2007-11-14 jfvanmeter () comcast net wrote:
> I recently completed a pen test for a client and discoveried a new
> directory traversal in a web enable application. The target for the
> test was FDE enabled , once the target was booted and the OS mounted I
> could exploit the directory traversal to read any file on the system. 
>
> After I informed my client of the problem, they asked me to do
> additional testing, I found if I encrypted the file, and ran the
> directory traversal, I could no longer read the file. 
>
> So until the ventor patched there software, a mitigating step to lower
> the risk was to encrypted the files the client felt were the greatest
> risk if someone was to run the exploit.

First, hosts that are world-accessible should not hold sensitive data
in the first place. Second, services running on a world-accessible hosts
should be running under their own user account. Third, a directory
traversal will expose only those files that are accessible by the user
running the exploited service.

If encrypting the files mitigated the problem, then the user running the
service apparently didn't need access to those files. Thus revoking read
access for that user would have mitigated the problem just as well
without the additional overhead of file-level encryption.

> I believe you need both, FDE to protect the data at rest, and file
> encryption to protect the data when it is active.

I still fail to see an advantage of file-level encryption.

Regards
Ansgar Wiechers
-- 
"All vulnerabilities deserve a public fear period prior to patches
becoming available."
--Jason Coombs on Bugtraq