Slashdot Mirror
Debian Bug Leaves Private SSL/SSH Keys Guessable
SecurityBob writes "Debian package maintainers tend to very often modify the source code of the package they are maintaining so that it better fits into the distribution itself. However, most of the time, their changes are not sent back to upstream for validation, which might cause some tension between upstream developers and Debian packagers. Today, a critical security advisory has been released: a Debian packager modified the source code of OpenSSL back in 2006 so as to remove the seeding of OpenSSL random number generator, which in turns makes cryptographic key material generated on a Debian system guessable. The solution? Upgrade OpenSSL and re-generate all your SSH and SSL keys. This problem not only affects Debian, but also all its derivatives, such as Ubuntu." Reader RichiH also points to Debian's announcement and Ubuntu's announcement.
Ubuntu got an updated advisory at http://www.ubuntu.com/usn/usn-612-2
I'm sure the problem will be fixed if the developers acknowledge that the problem exists. Not a big worry.
Yes, it is a big worry because any keys generated with this package are now potentially suspect. This means that anybody who's used Debian or a Debian derived distribution like Ubuntu needs to go back and destroy all host and personal keys generated since 2006. All of those keys are potentially guessable.
And that's a real vulnerability. Early versions of Netscape's SSL implementation (the first SSL implementation) were trivially crackable because of just such a vulnerability.
Need a Python, C++, Unix, Linux develop
It was reading from initialized memory to for the seed value, leading to valgrind warnings. See the original Debian bug report: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=363516
No, the random text your put in can be random or not. What they were talking about is a number generator that was used or applied to an algorithm that decided the prime of your key in so that the same algorithm wouldn't be used on two machines on the first then second keys. Without the random number generator, you could theoretically guess the algorithm used to generate the key based of the actions of another install and then decipher the key to gain access to whatever it was protecting.
To simplify and generalize it, if every machine uses X+1*256 to get a 256 bit key equal to 768, then you could reverse that and know X would =2 (3*256=768) and fake the credentials. The random number generator should change that to X+R*256 which make reversing the key harder because you can only solve to X+R=3 now. In practice, it will be a really larger number and a lot different process though. I can't say that I fully understand it but that simplification should show the difference well enough to give an Idea of where the problem is.
Not entirely true. Keys generated before the patch made it to the repos are safe - and I think quite a lot of debian boxes are old enough (I know I've got one.) There's a link in the advisory to a tool that checks if the keys are vulnerable.
This doesn't change the fact that this is a really serious fuckup. Debian lost quite some credibility in my eyes.
The original change was tracked in the debian Bugtracking system at: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=363516
The offending patch: http://svn.debian.org/viewsvn/pkg-openssl/openssl/trunk/rand/md_rand.c?rev=141&view=diff&r1=141&r2=140&p1=openssl/trunk/rand/md_rand.c&p2=/openssl/trunk/rand/md_rand.c
The fix: http://svn.debian.org/viewsvn/pkg-openssl/openssl/trunk/crypto/rand/md_rand.c?rev=300&view=diff&r1=300&r2=299&p1=openssl/trunk/crypto/rand/md_rand.c&p2=/openssl/trunk/crypto/rand/md_rand.c
The patch that broke it was checked in by Kurt Roeckx [kroeckx@debian.org]. Don't know if he broke it or if he was just the gatekeeper for checkins. See:
http://svn.debian.org/viewsvn/pkg-openssl/openssl/trunk/rand/md_rand.c?rev=141&view=diff&r1=141&r2=140&p1=openssl/trunk/rand/md_rand.c&p2=/openssl/trunk/rand/md_rand.c which shows the change that introduced the bug; and its parent node:
http://svn.debian.org/viewsvn/pkg-openssl/openssl/trunk/rand/?rev=141#dirlist which shows the maintainer responsible.
From looking at this patch, I think this is what happened. valgrind complained about a rather unusual coding convention in ssleay_rand_bytes. This is a function that returns random data into a buffer. However, before writing into the buffer, it reads from the buffer and incorporates the old contents into the internal random state. valgrind complained about this use of an output buffer for input. Normally you would never want to use potentially uninitialized data like this, but in this case it is OK as all that is being done is the data is being folded into the random state. In the worst case, this can't hurt, and maybe it will help, if the old data had some randomness.
Anyway, valgrind complained about it, and the maintainer commented out the use of the buffer. That would actually be OK, it is not a big deal. But the implementor made a mistake, and also commented out another similar usage, in a different function, ssleay_rand_add. This was a huge mistake, as the purpose of ssleay_rand_add is to add randomness into the random state. In that function, buf is an INPUT buffer, and adding it into the random state is perfectly legitimate, in fact it is the whole purpose of the function. But apparently because it looked similar to the questionable usage in ssleay_rand_bytes, the maintainer commented out the code in ssleay_rand_add at the same time. (I don't know whether valgrind also complained about this second usage, but if so, it was mistaken. I think it's more likely that the maintainer just got fooled and over-generalized from the valgrind complaint.)
So the whole thing was an attempt to clean up code and remove warnings, but the fix accidentally broke a crucial piece of functionality, rendering ssleay_rand_add completely non-functional. So any attempt to add randomness into the RNG state, such as for seeding purposes, is ineffective. The random state ends up with much less variability, and therefore all the crypto is weak. As Bruce Schneier points out, bad crypto looks much the same as good crypto, so it took this long to notice it.
Hats off to the reviewer who picked up on the problem. Don't know who it was, but the same Kurt Roeckx [kroeckx@debian.org] checked in the fix.
It's not seeding too often that's the problem, it's seeding with predictable data and expecting that data to be random. The time is very predictable, and contains very little entropy (randomness).
If you seed very often with data containing a lot of entropy (for example, radioactive decay), then there's no problem. It's also not a problem to add the time in the mix if you mark it as having almost no entropy.
Seriously guys, take a few minutes to understand the code before hacking on it. If it was written by a n00b, you need to understand it to make sure they didn't do other stuff wrong. If it was written by someone competent (which is probably the case for openssl), then you need to make sure you're not the one making the mistake. In this case, to make it work with valgrind, the patch should have been to debian/rules, simply changing the way Configure was called.
If I read the published "weak key detector" script correctly, Debian OpenSSHs will always generate one out of a fixed set of 262148 possible keypairs. Do the math yourself. If you know those keys, this is a 5-minute brute force attack.
The problem is, OpenSSL needed a buffer in which to XOR several sources of randomness, including /dev/random. OpenSSL didn't bother initializing the buffer, because doing so just eats CPU cycles, and a possibly random buffer is marginally better than a non-random one. However, the Debian patch removed *all* of the randomness except for the PID, leaving the user with one of just 262148 (or so) distinct keypairs. The tool that checks if your keys have been conpromised has a list of all of those keypairs and checks if your keypair appears.
Nothing for 6-digit uids?