4 ms·

The NSA has tried to backdoor Linux three times (2016)

18 points by MrXOR 7y ago

MrXOR 7y ago
What about SELinux?
- sofaofthedamned 7y ago
  Don't think it matters, though I was wondering the same. SEL is kernel, nothing to do with userland.
NikolaeVarius 7y ago
Good ol' Reddit, making a statement implying its factual, with linked sources saying anything but.
Analemma_ 7y ago
This is a bad headline. We have no idea who was behind the 2003 incident, and the linked article even admits that the 2013 Debian SSL vulnerability was probably an honest mistake. Saying that both of these were definitely the work of the NSA is an accusation not supported by any evidence.
- mtgx 7y ago
  If I were a Project Lead at NSA's backdoor division, I'd also make it look like when the backdoor is discovered (vulnerabilities are always discovered, even if it happens 2 decades later) it will look like a "honest mistake" or even a "dumb mistake" that a developer made. Definitely wouldn't call backdoor variables something like "rear_entry", "remote_auth", etc. I don't assume all NSA spies are smart enough to think like that and sometimes NSA code may be a dead-giveaway, but I do assume that at least some of them are, for the more important/bigger backdoor projects. You can also look at NSA's history with IPSec, or the more recent Simon & Speck situation, where they pushed forward their own design that at first glance might seem fine, but if you tried to push them on the details, they would stonewall you.
  - yodon 7y ago
    The NSA also didn't want to explain their reasoning for shortening the DES key size to 56 bits. It took many years for the public crypto community to discover the technique called differential cryptography, at which point it became clear that the actual theoretical maximum strength of the DES crypto system was 56 bits and the originally proposed longer key sizes didn't actually offer any true strength over 56 bit key variants. It remains important to be distrustful of all things in cryptography, but it also remains important that the one time the NSA had a huge and apparently negative impact on the strength of a crypto system, they were actually not weakening the system at all they were instead providing the public community with a significantly more accurate measure of the actual strength of the then proposed method (and given the accuracy with which differential cryptography computed 56 bits, I remain surprised they were able to provide that information, as the recommendation clearly communicated to a skilled adversary that they had developed and were leveraging this significant unpublished technique at that time).
    - dogma1138 7y ago
      Differential cryptanalysis was discovered initially by IBM while working on DES. The NSA made design alterations to DES to make it more resistant to DCA. However they didn’t publish those design decisions because they would reveal the existence of the technique, which at the time was applicable to many other ciphers including nearly all Soviet ciphers. The NSA didn’t backdoor DES the key shortening and the addition of parity bits was an attempt to make it more resilient to DCA while making it only slightly more susceptible to brute force attacks. It’s also important to note that DES was a 1970’s cipher which survived long enough to be relevant, even a 1080ti would take 2 or more days to crack a single DES key and its considerably more powerful than likely all the super computers in the world combined in say 1983. The NSA isn’t your friend for sure but they know exactly what they are doing.
AdmiralAsshat 7y ago
So we have one actual backdoor attempt, one OpenSSL bug that may have been a backdoor attempt but was more likely an honest bug, one tongue-in-cheek attempt mentioned by Linus, and that same attempt repeated seriously by his father. Assuming Linus was being serious and not facetious, I would say that's one attempt. And if the NSA really wanted Linus to backdoor his kernel, I imagine they would have approached him in-person and asked (as they have been known to do), so I'm not convinced that the two other attempts were them either.
mtgx 7y ago
That we know of.

18 points by MrXOR 7y ago

MrXOR 7y ago

What about SELinux?

sofaofthedamned 7y ago
Don't think it matters, though I was wondering the same. SEL is kernel, nothing to do with userland.

NikolaeVarius 7y ago

Good ol' Reddit, making a statement implying its factual, with linked sources saying anything but.

Analemma_ 7y ago

This is a bad headline. We have no idea who was behind the 2003 incident, and the linked article even admits that the 2013 Debian SSL vulnerability was probably an honest mistake. Saying that both of these were definitely the work of the NSA is an accusation not supported by any evidence.

mtgx 7y ago
If I were a Project Lead at NSA's backdoor division, I'd also make it look like when the backdoor is discovered (vulnerabilities are always discovered, even if it happens 2 decades later) it will look like a "honest mistake" or even a "dumb mistake" that a developer made. Definitely wouldn't call backdoor variables something like "rear_entry", "remote_auth", etc. I don't assume all NSA spies are smart enough to think like that and sometimes NSA code may be a dead-giveaway, but I do assume that at least some of them are, for the more important/bigger backdoor projects. You can also look at NSA's history with IPSec, or the more recent Simon & Speck situation, where they pushed forward their own design that at first glance might seem fine, but if you tried to push them on the details, they would stonewall you.
- yodon 7y ago
  The NSA also didn't want to explain their reasoning for shortening the DES key size to 56 bits. It took many years for the public crypto community to discover the technique called differential cryptography, at which point it became clear that the actual theoretical maximum strength of the DES crypto system was 56 bits and the originally proposed longer key sizes didn't actually offer any true strength over 56 bit key variants. It remains important to be distrustful of all things in cryptography, but it also remains important that the one time the NSA had a huge and apparently negative impact on the strength of a crypto system, they were actually not weakening the system at all they were instead providing the public community with a significantly more accurate measure of the actual strength of the then proposed method (and given the accuracy with which differential cryptography computed 56 bits, I remain surprised they were able to provide that information, as the recommendation clearly communicated to a skilled adversary that they had developed and were leveraging this significant unpublished technique at that time).
  - dogma1138 7y ago
    Differential cryptanalysis was discovered initially by IBM while working on DES. The NSA made design alterations to DES to make it more resistant to DCA. However they didn’t publish those design decisions because they would reveal the existence of the technique, which at the time was applicable to many other ciphers including nearly all Soviet ciphers. The NSA didn’t backdoor DES the key shortening and the addition of parity bits was an attempt to make it more resilient to DCA while making it only slightly more susceptible to brute force attacks. It’s also important to note that DES was a 1970’s cipher which survived long enough to be relevant, even a 1080ti would take 2 or more days to crack a single DES key and its considerably more powerful than likely all the super computers in the world combined in say 1983. The NSA isn’t your friend for sure but they know exactly what they are doing.

AdmiralAsshat 7y ago

So we have one actual backdoor attempt, one OpenSSL bug that may have been a backdoor attempt but was more likely an honest bug, one tongue-in-cheek attempt mentioned by Linus, and that same attempt repeated seriously by his father. Assuming Linus was being serious and not facetious, I would say that's one attempt. And if the NSA really wanted Linus to backdoor his kernel, I imagine they would have approached him in-person and asked (as they have been known to do), so I'm not convinced that the two other attempts were them either.

mtgx 7y ago

That we know of.