Originally posted by: STaSh
Yes, but what I was talking about was an actual software defect. Booting to the Recovery Console, on any system, is supposed to prompt your for the Admin password. The fact that it apparently does not, when booting a W2K Recovery CD on a system with an XP NTFS partition, is a flaw, because it is not functioning as designed.
It's not a flaw, it is working as designed. The 2000 recovery console does not recognize the XP SAM, which is why you don't get prompted. I can put a knoppix CD in a 2000 or XP system and it won't prompt for any password either, because knoppix could care less about the SAM. Is this a flaw too?
The key difference, and what makes it a
bug, is that the system was designed to prompt for a password before access.
If you walked up to an NT workstation, and hit C-A-D to access the login dialog, and were able to hit CTRL-ALT-SHIFT-WinKey-F12, and Windows would magically let you login without a password, is that not a flaw? I mean, hell,
you had physical access to the machine, why should there be any attempt at secure authentication at all? (Btw, that was a hypothetical made-up example, I have no idea what that key sequence does - hopefully nothing special.)
Originally posted by: STaSh
Likewise, I can put the same knoppix CD in a linux system and have the same access. This is not a flaw! This is security 101.
It *is* security 101, but that does *not* mean that it isn't a flaw - it is.
Originally posted by: STaSh
It wasn't FUD, you just mis-understood what I was saying - that MS's "most secure OS ever" had a silly little defect, that let people trivially bypass the password (as opposed to making them work for it by having to hack the SAM and bypassing it that way).
It is FUD, or at least ignorance. Why would any one with physical access to your drives hack a SAM file? If you had physical access to a linux box, would you be wasting your time trying to brute force the root password? Sensationalist journalism that crows "HUGE NEW XP HOLE! RUN FOR YOUR LIVES!!" does not make this an actual vulnerability.
It most certainly was not ignorance, and it was truthful, hence not FUD either. How is it ignorance to point out that the software is apparently NOT working "as designed". The security implications of that are another matter entirely.
(However, in your above hypothetical example, if Linux supported EFS in the same manner that W2K/XP does, then yes, you would still need to brute-force the user's pasword, in order to decrypt their EFS cert., in order to decrypt the file keys, in order to decrypt the files, in order to access them. But please don't misunderstand me again - crypto is about the
only step that you can take to protect your data, when the attacker has physical access.)
But while a wooden door may not be able to prevent a determined attacker with a crowbar from entering, it's still better than no lock on the door at all.
Or should car doors not have locks either, because it's simple for someone to just smash the window and let themself in to the vehicle?
Security is not about absolutes, it never was, there is nothing that is "absolutely secure". It is a spectrum of tradeoffs, between the cost of sucessfully breaching security for the attacker, and the inconveniences during operation for the defender. That software bug lowers the cost to the potential attacker to
zero. I would say that is still more-or-less a security vulnerability, in the same manner as choosing a weak password. Except in this case, it's MS's doing, and you can't change that.