Reputation
111,439
Next tag badge:
95/100 score
25/20 answers
Badges
13 197 323
Newest
 Good Answer
Impact
~3.9m people reached

1d
comment Difficulty of breaking private key password
Yes, it is based on GNFS; but though the algorithm is quite old (the late 1980s), technology has evolved and many minor but cumulative optimizations were found, that estimates try to take into account. To find the numbers, you have to explore the keylength.com site and play with the various simulators, and possibly read the relevant articles (but that's rather heavy on mathematics).
Feb
4
comment Why OpenSSH deprecated DSA keys
No, not really a duplicate. The underlying question has not been answered yet.
Feb
3
comment My school wants to keep the details of our door authentication system a secret. Is that a good idea?
@JohannesKuhn: it was the method used where I was schooled; the sysadmin observed the students' activities, and gave the root password to the most advanced. It worked marvels. (And, on a more general basis, this is called a "meritocracy" and is an efficient management method -- in the 13th century, it allowed Genghis Khan to conquer the World.)
Feb
2
comment My school wants to keep the details of our door authentication system a secret. Is that a good idea?
@PyRulez: the operational notion is "review". You want some extra analysis by other people. An open publication can help a lot in getting free reviews.
Feb
2
comment Will typing my password twice make it more secure? Or typing each character twice?
Also, the field is very dynamic. E.g. if most attackers try passwords in alphabetical order, then users begin to start their passwords with 'zzz'. Pretty soon, attackers adapt and start doing enumeration in reverse order. Or in random order, which guarantees attackers against worst cases. Trying to play tricks depending on average attacker behaviour tends to backfire. The notion of entropy is what remains when the attacker knows perfectly how you generate passwords -- thus, entropy is the only robust foundation for password security.
Feb
2
comment Will typing my password twice make it more secure? Or typing each character twice?
@immibis: ah, that's the catch. Doing something that "most people won't do" is not the same as doing something that "the attacker won't try first because most people don't do it". Claiming that you gain more than one bit through password doubling is making a bet on the attacker's indifference or incompetence. Unfortunately, this does not hold against the most dangerous attackers, who are after you, personally.
Feb
1
comment Can browsers connect HTTPS w/ the NONE cipher?
RFC 7525 is a "best practice" document that talks about TLS usage, and it explicitly forbids ("MUST NOT") negotiating a cipher suite that does not include encryption. So not only do current servers and browsers not support them, but it is quite unlikely to change in the future.
Jan
5
comment What can be done to prevent disruption of hardware in the event of a powerful solar flare?
@ShaneDiDona: depends on "big enough". Conceptually, a solar flare could simply terminally irradiate any living organism on an hemisphere, leaving the other one intact, except for the severe disruption in ecosystems. This has been proposed as a possible mechanism for some past global extinction events, with no definitive proof. A much bigger flare, really the Sun going nova, could wrap the Earth in very hot matter ejected from the Sun, effectively cooking everybody -- it is unclear whether the Sun can actually do that before reaching its old age, in about 4 or 5 billions of years.
Dec
7
comment What is a good practical (and sane) way to manage all your passwords for online sites?
The main things it buys me is that most of my passwords really fit in my brain and are thus usable even on machines where KeePass is not installed. I do not trust many machines and I certainly won't enter my password on a shared machine anywhere, but there are a couple of machines that I trust and yet do not manage directly.
Dec
7
comment Is publishing CRLs over HTTP a potential vulnerability?
@JRL: I read it on some documentation on MSDN, but I do not have the reference on hand. At one point, I tested it explicitly (but such things may change over time, depending on the whims of Microsoft).
Nov
18
comment when to escape user input
"Escaping" is related to interpretation of data. In most text-based languages (e.g. SQL, HTML...), there are literal values which are strings of characters; most of the characters just stand for themselves, but a few trigger special processing. For instance, in a string delimited by double-quotes, the " character triggers the end of string. If you want to include a double-quote in the string contents, then that special behaviour must be deactivated. "Escaping" is replacing a character x into a sequence (starting with another special char) that results in a non-special x (e.g. \").
Nov
17
comment Secure Configuration of Ciphers/MACs/Kex available in SSH
@pioto: not to my knowledge. There is still no known attack that would apply to a SSH connection. OpenSSH's defaults may have changed, though, because they must not only choose only secure algorithm, but they must also contend with the feelings of their user base, which are notoriously subject to drift over times.
Nov
3
comment What DH Group size do TLS Ephemeral DHE ciphers use?
The server sends the modulus, generator, and its public DH value in the ServerKeyExchange. The client then learns all of these; the client will send its public DH value in the ClientKeyExchange, which occurs later on in the handshake. (Note that the client must know the complete modulus, not just its size. In DH, all computations must be done modulo the same prime p.)
Oct
23
comment In message signing, is the digest encrypted with a generated symmetric key or with the sender's private key?
No, it is not "encrypted with the private key". This is a flawed analogy that does not work for algorithms other than RSA, and, when looked at closely, does not work for RSA either. It is an old way of explaining signatures that, in practice, spreads more confusion than enlightenment.
Oct
23
comment In message signing, is the digest encrypted with a generated symmetric key or with the sender's private key?
Yes it is. The notion about "too many CPU cycles" is just the traditional way to explain things, but, as many traditional things, it is wrong. It can be traced back to the first PGP implementation, at a time where the security of chaining modes was not well understood, which explains the weirdness of what PGP does. In the head of Zimmerman, it was a problem of CPU, but since then we learned a lot and now know better; you can always add more CPU, but making a secure RSA-based chaining is a lot harder.
Oct
23
comment AES_GCM and TLS sequence number
"A sequence number is incremented after each record" (end of section 6.1). Though this could be clearer, "incremented" really means "add 1" ("incremented by 1" would be a pleonasm). In any case, that's what all existing SSL/TLS implementations do.
Oct
22
comment In message signing, is the digest encrypted with a generated symmetric key or with the sender's private key?
Your diagram explains the signature as being "encryption with the private key". For this offence against logic and pedagogy, I shall be forced to extract your spleen with a blunt spoon.
Oct
21
comment Is there any way to safely examine the contents of a USB memory stick?
SPARC systems have a BIOS; they just call it "firmware". But all computers boot on some code in ROM/Flash, and that code, like every single piece of software, may have bugs. Of course, this is not the same code as a BIOS for an x86-based computer, so one may hope that the attacker "won't have thought about SPARC machines".
Oct
6
comment Isn't ssh-copy-id exposing to a brute force attack?
OK, I added a provision for that. Maybe entering the password would not have worked. Anyway, the important point is that ssh-copy-id is just an automation script that uses the normal SSH protocol and configuration; it is not an additional mechanism. The server is not even aware of the existence of ssh-copy-id.
Sep
29
comment SNI (server name indication) with SSL
Usually, clients that know about extensions like SNI first try connecting with the extension, and fallback to not sending the extension only as a second try; not the other way round. If you have a SSL-3.0-only client that does not send a SNI, then chances are that the client will never send a SNI (and it should have been upgraded more than a decade ago).