Slashdot Mirror
Microsoft Admits XP Has Same Bug As Win2K
Arashtamere sends in a Computerworld story on a security flaw in the Windows 2000 pseudo-random number generator published by Israeli researchers earlier this month. Microsoft has now admitted that the flaw is present in XP too. Microsoft denies that the bug is a security vulnerability, since an attacker would have to have gained administrative access to a system before exploiting it. (The Israeli researchers point out that many common exploits provide admin access.) This stance apparently lets them off the hook for patching Win2K, which is in "extended support" mode, though it powers about 9% of US and EU business computers. Microsoft said that XP SP3, due in the first half of next year, will fix the bug. The company said that Vista, Windows Server 2003 SP2, and the new Windows Server 2008 are not vulnerable.
A reason to upgrade to Vista! ;)
Religion is regarded by the common people as true, by the wise as false, and by rulers as useful.
But to say that is to deny our ability to flame MS! Clearly it's an example of MS' incompetence that a random number generator that's 7+ years old has been broken by recent maths and it can be exploited to gain full access when you already have full access!
Here is the original article on the ACM.
Very brief summary of article
Each process has their own instance of the generator, and the refresh of the internal state is done after 128 kbs of output from the generator (roughly 600-1200 SSL connections with IE). Not only that, it is run in the userspace so it is not a security violation to examine the internal state of the generator. The function used is not one-way which provides a means looking at past transactions of a user (within the 128 kbs of data).
Lack of planning on your part does not constitute an emergency on mine.
While in general I think open-source and closed-source software can coexist, I think this is a pretty good example of why anything related to crypto should be open. All of public key cryptography relies on the secrecy of private keys, not on the secrecy of the algorithm itself. And while they might have faithfully implemented the algorithm, who knows what kinds of arguments/whatever to the crypto functions might cause undesired results -- it's just too hard to test.
In any case, the thing that surprised me most from the article was that Windows 2000 users would be left out in the cold: "Because the company has determined that the PRNG problem is not a security vulnerability, it is unlikely to provide a patch [for Win2K]." Wow. Especially when it's something this easy to fix. This bug also solves any attacker's problem of trying to sort valuable from non-valuable information, since presumably any valuable information (credit cards used online, etc) will use encryption. And while someone suggested that a program should use its own random number generator, there is a problem because, in general, your application (not running as Admin) shouldn't have access to nearly the same amount of entropy sources (like network activity, GUI inputs, etc).
--
Educational microcontroller kits for the digital generation -- great gift!
If you truly agree with MSFT, then you should quit working in computers right now, for everybody's sake.
Many corporate computers have local admin accounts that are likely to share a user/password combo across large numbers of machines. A keylogger might not get you these credentials, but the ability to crack these credentials could get you admin access to a huge number of other computers.
It is people like you who make sure that security consultants will never want for work.