QuicksearchGoogle the SiteCC Licensesupersized.org |
Sunday, April 8. 2007Making it Clear Just Why Protected Processes are a Bad IdeaComments
Display comments as
(Linear | Threaded)
I assume the reason why MS didn't attempt to implement this as a standard, but specially privileged, built-in user, is that there already exists a large body of documentation describing how to write authentication modules, e.g. for smartcards--obviously, anyone having read this could write a module to allow authentication as the protected process user and thereby exercise its permissions. Nevertheless, I completely agree that the whole approach is a nonsense in terms of its lack of self-consistency and overcomplication of what should be an easy problem within the existing security model. Even outside the security model such as it is, the administrator's power doesn't come only from their having certain privileges under the Administrator account: it comes from having physical access to the machine, which can be used to gain access to any function or property of it (no matter how awkward it might be to do so). I don't actually know this, but I'd expect LSA modules to be protected the same way as kernel-mode code (i.e. have to be signed by Microsft for 64-bit). So that using an LSA module here would be the same as using a driver, as Alex did. And it's notable that even before the Vista DRM push, Microsoft used security through obscurity wherever they could and didn't give the admin account access to everything. For instance, how do I as an admin extract the private key created on my machine with CryptoAPI and marked non-extractable? For that matter, how do I extract the symmetric key used by by computer's domain machine account for Kerberos login? Problem is Microsoft has never had a clue how to do true security through obscurity. True security through obscurity person has be guessing. Single line of security is not obscurity. Not providing the paper does not make obscurity. Linux systems can have true obscurity if configured right. Attacker is not sure what software or security on the other end. Since the attacker is guessing and cannot be exactly sure what is on the other end its true obscurity. True obscurity causes fear in attackers and reduces attacks. I think Microsoft's goal for PPs being off-limits, is that their DRM requirements must not allow any "Premium Content" data to be accessible to any unauthorized processes. If even a legitimate A/V process was allowed to read memory space containing "Premium Content", then who's to say other malicious processes couldn't sneak in too. Microsoft is all wrong with this, as what they are really trying to do is turn a PC running Vista into a half-baked computer electronic (CE) device. |
Tagged entriesThis Entry's LinksReferring links
|
Tracked: Nov 05, 04:55