163 reputation
10
bio website mirbsd.org/jupp.htm
location Germany
age
visits member for 1 year, 1 month
seen Jun 13 at 3:56

MirBSD founder (this includes mksh and jupp development), FreeWRT core team member (and casual developer for its partial successor OpenADK and µClibc-ng), Debian Developer (and Grml contributor), and doing lots of OSS work in general (virtually all C libraries, lots of porting, etc). Member of the Teckids society for promoting IT and science (with additional focus on Open Source, privacy, and data security) in youngsters.

I hereby offer all content I author on the SO/StackExchange network also under The MirOS Licence (HTML version with less WTF-8) in addition to “cc-wiki” (CC-BY-SA 2.5 or 3.0, depending on the SO/SE network site, apparently). I urge everyone else contributing here to do the same.

English is only my third language, so please excuse any mistakes related to that I may make, e.g. when not fully awake. I do try.

Shortlink: Qs in my tags


Jun
13
awarded  Announcer
Feb
8
awarded  Quorum
Jan
7
comment Relation between key size and PRNG state size
I’m not doing anything wrt. breaking crypto, my areas of interest lie elsewhere.
Jan
7
comment Relation between key size and PRNG state size
“I have also no idea why you say "160 bits of state is definitely not (enough) for a 1024-bit DSA key".” ← That is because DSA is weak in the face of bad entropy.
Jan
7
comment Relation between key size and PRNG state size
“that are effectively indistinguishable from random” does not mean that every bit is fully (1.000000) random
Jan
7
comment Relation between key size and PRNG state size
“Statements like "the quality of the output of the PRNG is debatable" is questionable; there exist high quality CSPRNGs” yes, but not all PRNGs are “high quality CSPRNGs”, that's what I meant, that the choice of RNG affects the answer.
Jan
7
awarded  Critic
Jan
6
comment Relation between key size and PRNG state size
@downvote: care to comment?
Jan
6
comment Random Number Generation with a Entropy pool versus Seed
You normally use both: a pool, to throw incoming events into, mix, and generate output from, and a seed, which is initially thrown into the pool (or just a dump of the initial pool, as in the MirBSD kernel).
Jan
6
answered Relation between key size and PRNG state size
Jan
6
comment At what stage is DHE and RSA used during the SSL/TLS handshake?
The problem here is that DHE needs good entropy on both sides, and that the DHE part is not secured by at least the server's key (or both parties' if they have some), to protect from casual eavesdroppers or the bad-entropy case. I'm currently considering switching to ephemeral RSA keys (short-lived, maybe a day or so) due to this, but have not yet tested whether this is portable enough.
Jan
2
awarded  Commentator
Jan
2
awarded  Scholar
Jan
2
accepted Need 32-bit mixing function that has perfect avalanche between octets
Jan
2
comment Need 32-bit mixing function that has perfect avalanche between octets
If anyone has got any comments left on the algorithm, or my implementation, of my use of it (to distribute entropy across multiple hashes), I will welcome feedback. I've credited this SE question in the CVS commit. (It's not yet “live” as I need to validate the “copy bytes around” part of my algorithm, but that is not the hard part. Just tedious.) Thanks!
Jan
2
comment Need 32-bit mixing function that has perfect avalanche between octets
I've not done a full test of my other criteria, only a random sample each, for the “keep 3 bytes the same and iterate the other in [00‥FF]” and “keep 2 bytes the same and iterate the other two in [0000‥FFFF]”. In both cases, every output byte had a perfect distribution across all 256 possible values, and in the second case, both the first and the second words had a perfect distribution across all 65536 possible values. I think that meets my criteria (except I didn't do it for all possible combinations).
Jan
2
comment Need 32-bit mixing function that has perfect avalanche between octets
OK, I've tested it a bit. It's (obviously) bijective and thus can also replace the original Finish part of the 32-bit hash (though it's 19 opcodes as opposed to 13, on i386, though we could save 2 redundant ones GCC introduces). Every output byte is dependent on every input byte.
Jan
2
comment Need 32-bit mixing function that has perfect avalanche between octets
@nightcracker I converted this into a pure 32-bit implementation (does not need to access the individual bytes ⇒ endian-neutral) as ix.io/9DF • the algorithm itself has test vectors that make me wonder a bit, but I'll continue having a look at it. Thanks already anyway, and sorry for being slow in evaluating this (I fell ill before christmas). GCC's asm output on this code is much better, even.
Dec
6
answered Can insecure algorithms be combined to form a secure algorithm?
Dec
6
comment Advantages of combined PRNGs
This will most likely only make specific attacks targetting the algorithms you use harder. Do make sure the two PRNGs are unrelated; then, it’s a sound scheme, but will not help much if one of the PRNGs’ output is not distributed well.