@Luc We are talking about one big file (18 GB). If two versions are encrypted with the same IV, you will know at which point they diverge, so it gives you more than just knowledge that they are not identical: it will show you where difference begins.
For an extra version where the file is split into 8 kB blocks encrypted separately, the leak will tell you which blocks have changed and which have not (and where is the first changed bit in each block), which can reveal a lot more.
All of this constitutes traffic analysis. My usual example is that traffic analysis allowed Americans to predict the Japanese attack on Midway, and they could stop their advance.
Without traffic analysis we would all speak German and/or Russian.
@ThomasPornin Though the purpose behind my question was more in general about remote backups. When you have to use a different IV every time you make a new backup set, you'll basically be uploading your entire disk (or anything in the backup anyway) every single time
In that case, leaking which files were changed (changing IV each time a file gets changed) might be a better option
@Simon Anyone with some knowledge in C could fix bugs. The source code is online. It may be a hurdle for beginners, but I hope there will be more persons contributing to it. For example there is this site phpinternalsbook.com to learn about the internals of PHP, this will boost your knowledge and should get you started.
Hate's a pretty strong word, really. I don't hate phpMyAdmin, but I do find it as an indicator. Generally speaking using phpMyAdmin to manage mysqld, as opposed to mysql, can show non-professionals in a professional role.
However, about half of the Linux system compromises I've investigated have come through phpmyadmin.
My favorite, though, was that we discovered a Windows Server 2003 box participating in a botnet. It was a departmental server with relatively low access rights. We brought it in and I was next up on the rotation.
It was running a system inventory service. Pretty not bad service, really, lots of cool features.
The Windows version of it is a single installer that, apparently, bundles XAMPP.
A full blown XAMPP installation. A version of XAMPP that was over a year old when this thing was installed (about a year before the compromise). Funny enough, that version of XAMPP also included a really fucking old version of phpMyAdmin.
SHA-512 truncated to 256 bits is as safe as SHA-256 as far as we know. The NIST did basically that with SHA-512/256 introduced March 2012 in FIPS 180-4 (because it is faster than SHA-256 when implemented in software on many 64-bit CPUs). SHA-224 is just as safe as using 224 bits of SHA-256, becau...
When I've read this article http://blog.ircmaxell.com/2013/07/disclosure-wordpress-wpdb-sql-injection.html I though damn, they were right. WP is crap. They still use mysql_ functions
The first: it must be disconnected from any network, to prevent leak.
For power supply should be used a special UPS with filters.
What about TEMPEST attacks, how should be enclosed crypto-station?
I imagine that all of construction or just a room closed by RF proof screen.
Is there way to s...
so... it appears I've bought a new toilet tank that comes in the homeopathic mode and doesn't wanna release enough water to actually flush, it just dilutes the solution to infinity... any idea what I need to move in it to flush? I've been trying to move the position of that lever and the "float ball" (or however it's called)... no luck :(
I almost don't want to answer, because everyone with an answer here is right to some degree. But I also feel that everyone is also wrong to some degree as well.
To those who say it's secure, they are correct in general. "Double" hashing (or the logical expansion of that, iterating a hash functio...
Problem / symptoms: I've bought a new water tank (nothing fancy inside, just what I'd describe as "the usual, run-off-the-mill mechanism"*), but it doesn't flush properly, as it doesn't release enough water and it only ends up mixing the water in the toilet, endlessly diluting it (thus my humorou...
@HamZa as far as I understand it, he's correct to point out the issue but incorrect as to its relevance: the number of iterations is negligible against the threshold where collisions become an issue. @CodesInChaos or @ThomasPornin would know better
@Gilles Well to be honest, I don't really understand the internals of hashing. What I do know is that I would never do something like that in a realworld application.
It appears I've been nominated for the mod position on Space Exploration... I'd really like to talk with one of the mods here, preferably @RoryAlsop that's also nominated there, so I get a better impression what all that involves and if I'd be able to deliver :?
@TerryChia Lol :P r[0] would work too, he's got two beta diamonds (to be fair, one of them is a late appointment so he got into the scene at a later stage)
@Gilles When you hash repeatedly with a n-bit hash function, the space of possible state values reduced to an internal cycle of average size 2^(n/2), but you reach it only after (on average) 2^(n/2) iterations. With MD5 (n = 128) this hardly matters. With SHA-256 (n = 256) you cannot even hope being subject to any real space reduction effect.
@TerryChia LOL still, as any mere mortal, I could only take a single successful attempt on my life, and I'd hope I wore brown trousers for the occasion :)
@ManishEarth could do with a bit of description what would mod duties be on site like Space Exploration yes... and I wouldn't mind your personal opinion either
CipherCloud just filed a DMCA notice with stack exchange to take down the question How is CipherCloud doing homomorphic encryption? (now deleted by stackexchange).
Since I obviously can't post the full question+answers here, a short summary of what it contained:
The question itself asks if/how...
@lynks To my knowledge, there is no known "small cycle" in MD5 or any of the other usual hash functions. This does not mean that small cycles don't exist, only that we do not know if they exist.
Note that if it would be quite hard to find a small cycle without having an efficient collision attack at the same time; so "small cycles", if known, ought to be widely known ("publish or perish").
the hash space clearly has cycles, and given that collisions exist, those cycles must have multiple 'entry points'. There will, however, be some values left over that do not act as entry points to the larger cycles. Eventually you should be able to get down to a small number of hashes that are either fixed, or are members of small cycles.
I'm just thinking aloud, clearly the numbers are too big to be feasible.
@ThomasPornin can I ask you a quick question, you know banks these days use card readers which can sign a transaction, but how are these actually implemented? Is the challenge which is generated to be signed created based on time+transaction id+some cryptographic random number?
@lynks The point, though, is that when you start from a random point, probability of reaching any other cycle than the "big cycle" of size 2^(n/2) is very small.
@adnan, ignoring the past few months of crypto fail? — D3C4FF21 mins ago
@D3C4FF I'm sorry, but your comment made me really angry. The only reason is that it conveys a certain amount of ignorance that I didn't expect from someone like you.
@LucasKauffman In France, smart cards use a "signature" which is in fact a MAC; the card has a secret key that the bank also has. The MAC is computed over a packet which includes the transaction details and (at least) some time stamps.
Note that a lot of the security comes from the assumption that the payment terminal is honest and follows the protocol. This is why these beasts are tamper-resistant, apply active defence mechanisms, and are expensive.
@Adnan "Professional manner" is the real point. Ultimately, what we need is that developers keep cool and don't deploy crypto "enhancements" without thinking them through.
While most software has some bugs, you can try to gauge the competency of the developer by them. My impression is that Nadim isn't the best coder for security stuff, but that he reacts well to vulnerabilities he's told about.
Cryptocat's developers have demonstrated in the past that they were quite overzealous on the creativity part, which is a worry. But I believe in redemption and they might amend their ways.
@Adnan The difference being this is an application which is entirely about the strength of it's encryption. facepalm level flaws don't really reassure me.
Will it get better in the future? Probably. I don't really have confidence in it at this very moment though.
@ThomasPornin I know you have to enter some details about the transaction followed by a number (which seems to be related to time) and then you need provide your pin after which you get a number and you then return that number to the bank
so I guess it might be similar to what you explained, I'm just wondering if it's actually secure :P
@LucasKauffman If the user has to type things back and forth, then it probably is symmetric-crypto only. Asymmetric crypto (signatures...) are too large for that.
@Adnan Fair enough, no offence intended. But people recommended using such software in an environment where 'peoples lives are at risk' (it was widely recommended for activist types etc). That's not really appropriate since its basically some dudes experimental pet-project.
@Adnan I get your point though. People make mistakes, no matter who they are, and they learn. And i agree, Nadim did the right thing and fixed the bugs and admitted his failings. Hopefully he learnt from it, and hopefully no-one was terribly impacted as a result. On the flip side, with revelations about the NSA and stuff recording 'everything' it worries me a little bit if the crypto is 'safe' until someone finds a flaw and the NSA can go back and decrypt all the things.
@D3C4FF FWIW and if you hadn't read it already, the security issue happened under SSL. So your messages are only at risk if Cryptocat's servers are compramised within an hour of the conversation. (Encrypted messages are removed from the server after one hour of the conversation end)
The first: it must be disconnected from any network, to prevent leak.
For power supply should be used a special UPS with filters.
What about TEMPEST attacks, how should be enclosed crypto-station?
I imagine that all of construction or just a room closed by RF proof screen.
Is there way to s...
The man behind PGP, Phil Zimmermann created Silent Circle which does just that.
CipherCloud just filed a DMCA notice with stack exchange to take down the question How is CipherCloud doing homomorphic encryption? (now deleted by stackexchange).
Since I obviously can't post the full question+answers here, a short summary of what it contained:
The question itself asks if/how...
