log☇︎
5134 entries in 0.848s
davout: it occurs to me it would be really nice to have a way to easily dice-generate an rsa keypair as to not rely on whatever prng
STRML: gpg: imported: 1 (RSA: 1)
fluffypony: gpg: imported: 1 (RSA: 1)
asciilifeform: artifexd: there are considerably subtler ways to diddle rsa key generation.
assbot: 5 results for 'rsa backdoor' : http://s.b-a.link/?q=rsa+backdoor
artifexd: !s rsa backdoor
assbot: Backdoor in a Public RSA Key / Information Security / Kukuruku / Technology Hub ... ( http://bit.ly/1CgrzZd )
assbot: backdoored rsa key generation ... ( http://bit.ly/1CgrAwv )
punkman: https://gist.github.com/ryancdotorg/18235723e926be0afbdd http://kukuruku.co/hub/infosec/backdoor-in-a-public-rsa-key
asciilifeform: how many would recognize an 'rsa pill' if saw.
mircea_popescu: i been looking for it magically can't find it. the one that had the rsa break in the last fascicle
mircea_popescu: even if you use rsa-only keys, you can rsa-encrypt a message so that it's easier to break but still decryptable by destination.
mircea_popescu: peterl btw, know who invented rsa ?
asciilifeform: (e.g., i may know that it is a machine at ip p.q.r.s with rsa pubkey K)
asciilifeform: 576 can hold a 2048-bit (ephemeral rsa key!) sig, a reasonably long key fp, and small payload.
decimation: I donno, what would you name your rsa/udp/wot idea?
asciilifeform: here's a very rough description of above experiment. to start a transmission a --> b, 'a' sends a hello (udp) packet, which consists of an ephemeral rsa key, hash and length of (coming) payload, and all of the preceding signed with his wot key. this fits handily below the traditional 576 byte mtu.
asciilifeform: 'there is like 30 papers each year published in cryptographic literature in which cryptosystems fail exactly because they use small integers...' << ahahaha. what a brazen and deliberate confusion. bringing up rsa coppersmith attack in thread on unrelated cryptosystem.
mircea_popescu: in fact the reason why we favour rsa is quite long.
decimation: yeah this is why RSA seems more compelling
decimation: yeah, it was a noob implementation. But it brought up a question in my mind: for RSA, private key and public key are derived from a prime number. Is there a similar procedure for ECDSA?
BingoBoingo: Well, I dunno of any better to gen keys with atm than RSA.
iang: seems to work better if I give it an RSA key ;-)
BingoBoingo: iang: Current best practice seems to be gen RSA/RSA and just make the it huge
assbot: Logged on 23-11-2014 05:11:53; asciilifeform: a 4096-bit rsa signature and key fp fit handily in a udp minimal packet.
assbot: Logged on 23-11-2014 05:04:09; asciilifeform: mircea_popescu: this is why, in my unofficial wonderland, you can't even open a socket without transmitting an rsa-signed a 'this is me, and my wot' breath of life packet.
decimation: asciilifeform: but apparently they broke a 512-bit rsa cert!
decimation: right, which is why no one uses 1024 rsa anymore (at least anyone who matters)
asciilifeform: a 4096-bit rsa signature and key fp fit handily in a udp minimal packet. ☟︎
asciilifeform: mircea_popescu: this is why, in my unofficial wonderland, you can't even open a socket without transmitting an rsa-signed a 'this is me, and my wot' breath of life packet. ☟︎
mircea_popescu: anwyay : you'll also notice we quite strictly use 4kb rsa sigs in our gpg
mircea_popescu: full rsa.
PeterL: have there been private rsa keys linked to extortion?
decimation: I wonder if usg has attempted to jail someone for possessing a private rsa key that is directly linked with extortion?
mircea_popescu: gpg: Signature made Sun 02 Nov 2014 11:29:39 PM EET using RSA key ID F3251143
asciilifeform: possibly these will have to travel armoured and rsa'd to turdatron's pubkey ☟︎
jurov: rsa is only possible because we have probabilistic primality tests for bignums up to arbitrary certainty
mircea_popescu: you telling me the rsa e must be a fermat prime ? why ?
assbot: cryptanalysis - Definition of Textbook RSA - Cryptography Stack Exchange
asciilifeform: http://crypto.stackexchange.com/questions/1448/definition-of-textbook-rsa << for schoolchildren
asciilifeform: but with pure-rsa, enciphering on a generic pc will be a suicidal act.
asciilifeform: mircea_popescu probably recalls our convo re: pure-rsa. we'll have it.
bounce: why do you want pure rsa?
mircea_popescu: you know, thinking about this, i think i actyally want pgp reimplemented. usgavin's speshul maths about how things improve make me realise that having a pure rsa based code rather than the current encode a symmetric chypher method is perhaps feasible.
jurov: to get 1024bits you'll need 400 dice throws. and lesser rsa is vulnerable
jurov: i had in mind RSA something.. multiplying 200digit numbers would be tedious but doable
pete_dushenski: interestingly, kryptokit keys are only 2,047 instead of 2,048 rsa keys. odd, no?
asciilifeform: i still can't fathom why they're not having their muppets push the 'all this derp is because they really broke rsa in the '90s' angle.
asciilifeform: kakobrekla: RSA pen & paper << about the same as muscle-airplane. yes, possible.
kakobrekla: well lucky for her, RSA can be done with pen and paper ?
decimation: actually that was flash exploited via excel http://gcn.com/articles/2011/04/04/rsa-hack-securid-adobe-flash.aspx
decimation: like that rsa keyfob scam
asciilifeform: ulbricht << unless he's learned to grind rsa in his head, how is this to happen? his jailers will otherwise have the key at their leisure. ☟︎
asciilifeform: http://cryptome.org/2014/10/nsa-eci-the-intercept-14-1010.pdf << now we know what drawer in the indiana jones vault contains rsa pill.
mats_cd03: interesting to me, since i know about zero things re: rsa.
assbot: security - Why is this commit that sets the RSA public exponent to 1 problematic? - Stack Overflow
mats_cd03: http://stackoverflow.com/questions/17490282/why-is-this-commit-that-sets-the-rsa-public-exponent-to-1-problematic
kuzetsa: what's wrong with using openssl to generate RSA keys?
kuzetsa: gnupg uses libgcrypt for the RSA key generation :(
kuzetsa: and then when you select RSA it's done via openssl
xmj: kuzetsa: hopefully your RSA key uses OpenSSH
kuzetsa: well... I mean my GPG key for authing with gribble is an RSA key but other than that
kuzetsa: I haven't got anything using RSA in production at this time
assbot: MFSA 2014-73: RSA Signature Forgery in NSS
asciilifeform: hand-cranking rsa, brass gears...
assbot: Prime Number Hide-and-Seek: How the RSA Cipher Works
mats_cd03: http://www.muppetlabs.com/~breadbox/software/tiny/teensy.html and http://www.muppetlabs.com/~breadbox/txt/rsa.html worth a read
mircea_popescu: if he can get paid to listen to rsa keys buyt then for "technical reasons" can make that baloon 10x,
asciilifeform: only want to hear rsa key. (or aes, whatever.)
asciilifeform: mircea_popescu: fundamental misunderstanding. machine does not know (unless it is fiendishly complicated, and elaborately trapped, perhaps) that you are doing, e.g., rsa.
asciilifeform: next year, suppose, they include a gigantic multiplier and special instructions for rsa. want to use these ?
decimation: asciilifeform: I thought that the usg signed stuff with rsa certs generally
mircea_popescu: "Anything ECC is currently highly suspicious, not in the least because the math is complex." <<< ah i'm so flattered. so a year or two ago, schneier and the "consensus" i nthe community was that rsa bad, ecc good ; mp was exactly on the other position.
asciilifeform: see also earlier discussion of swallowing rsa signatures of people you don't know
asciilifeform: decimation: popular bot << no need for 'bitsquat' here. most of the clever sort of botnet use domains generated weekly, based on time of day. but, clever folks also use rsa signatures for payload auth.
asciilifeform is always entertained when people come and ask, e.g., 'can you break rsa?'
asciilifeform: as they did on rsa's.
asciilifeform: decimation: not that we're necessarily stuck with rsa for all time, but the burden of proof on anyone suggesting an alternative is bowl-loosening, if you apprehend it.
decimation: asciilifeform: it does seem that mental rsa is more likely to be useful than a totally new system
asciilifeform merely argues that mental rsa is not the total absurdity it appears to be on first glance
asciilifeform: 'mental rsa' is mostly useful as a gedankenexperiment.
asciilifeform: decimation: i and perhaps a few other people did suggest, half-seriously, that people could learn to carry out, e.g., rsa, using mental arithmetic shortcuts of some variety
asciilifeform: decimation: and the keys << nope. go encipher an rsa message to me and then break it without my priv key
xmj: fluffypony: what exactly do you expect for RSA's future?
assbot: A Tiny RSA Cryptosystem based on Arduino Microcontroller Useful for Small Scale Networks
asciilifeform: this is almost convincing re: the impossibility argument. but suppose the goods isn't a generic rsa pill, but something more akin to nuke. that is, oppenheimer & friends didn't run off to an island and form own kingdom, because their 'jewel' required astonishing resources to put into action (industrial empire) and not very useful on their own, on paper
asciilifeform: well then. suppose, bush sr., or whoever, circa '92, whenever, was informed of the discovery, in his fuhrerbunker. then he proclaims: 'let's get the people with valuable secrets to use rsa. how to do this?'
asciilifeform: jurov: the hypothetical master doesn't need to. he just sorta sits there quietly and bakes pills against rsa; creates diddled but internationally-appealing block ciphers; etc
xmj: fluffypony: in english, cross jurisdictions? or RSA specific?
assbot: Bleichenbacher's e=3 RSA Attack
asciilifeform remembers the rsa key challenge and its quiet disappearance
Jason: peterl: tl;dr: i'd pick rsa. dsa has some hash weaknesses.
assbot: security - What is better for GPG keys - RSA or DSA? - Super User
Jason: peterl: https://superuser.com/questions/13164/what-is-better-for-gpg-keys-rsa-or-dsa
peterl: which is better, DSA or RSA key?
ascii_at_beach: it's like rsa. you can communicate it to an educated person in five minutes.
mircea_popescu: by zed.gendns.com with esmtps (UNKNOWN:DHE-RSA-AES256-GCM-SHA384:256)
asciilifeform: mircea_popescu: if you're willing to ditch ecdsa and use traditional rsa, and eat the cost of longer keys, this is trivial.
mircea_popescu: "have a 4096 rsa key made out of 8 parts! it may not be as strong as the proper key, but that's good : if you forget your passphrase you can break the key and retreive your stuff!"
mircea_popescu: but even so, 10 bits a throw, a 4k rsa key needs what, 7-800 ?