December 2003 - cypherpunks-legacy

Random noise from disk drives
by Rich Salz 17 Dec '03

17 Dec '03

Don Davis has done some interesting/important/widely-referenced work on using disk latencies as a random number source. We were talking about some of his techniques, and he kindly gave me permission to forward along his email. He's not on the list, so I cc'd him. /r$ --------------------forwarded message----------------- well, if you're willing to accept a temporary peak load on your machine, then a paging rng is easy to write. the basic idea that worked is: * allocate a little more memory than is available in ram; * walk through it in steps of a prime # near 4kb or 5kb; * time each access with whatever system clock is easiest; * throw away accesses that take less than 5 millisec or so; * keep only the parity of each access that remains. naturally, on the first passes through the array, you'll get mostly fast accesses; thereafter, most accesses will cause page-faults. mostly a page-fault's access-time reflects where the page was on the disk, where the head and spindle were before the page-fault, etc. that's why you keep only the parity. however, you're not going to get a bit of entropy from every access, by any means. the reason for a prime-valued step is that you want to be sure of visiting every page, but you don't want to have to know the page-size. (i haven't checked this trick yet; i used 2kb or 4kb steps, if i remember correctly). the reason for the 5 msec cutoff is that it's rare for a disk access to happen so fast, and you want to screen out other causes of non-immediate memory accesses. you'll only get a real access that fast, if at the moment of the page-fault, the head's already in the right track, and if the block you want is less than a third of a rev away from the head. i used a byte-indexed table to get the parity quickly. you don't really need to bother packing the parity bits into bytes; store the 1's and 0's as chars into the md5 buffer, & hash it; then, put the hash itself into the buffer, xor more 1's & 0's back in on top, rehash, etc. running md5 8 times as often as necessary doesn't matter. note that this algorithm should defeat caching disk controllers, but i haven't checked for sure. if it doesn't, the symptom will be long stretches of fast access-times. any fancy stuff you put in, like feeding noise-bits back into your path through the array, is a bad idea; first, it really only adds pseudo-randomness; second, it's liable to reduce the frequency of page-faults. though i knew it was a bad idea, i tried it anyway, lots of ways, and that's what i found. simple is best. really, the only complicated code should be outside the rng: reallocating when memory availability changes, minimizing the ui, stuff like that. it is worthwhile to run find, or some other filesys- traversing program, while you're running the paging-rng. the benefit is not from the changed paging-times, but from the extra head-motion, which perturbs the spindle speed. if you choose to measure the rng's quality, study the raw parity-bits, not the hashed output. the hashes would look perfectly random, even if the parity bits were periodic. look for periodicity (fft), long-term autocorrelation, and measure the entropy with an entropy estimator. don't bother with runs tests and the others; the hashing will produce nice output bits, even if the parity inputs are periodic or are highly correlated in the long-run. but if they're not periodic and uncorrelated, the output bits will be random, instead of pseudo-random. on the only machine i studied closely, i got 100 bits/min, with a 1 khz interrupt-clock; if your interrupt schedule is more frequent, then you can expect proportionally more entropy. remember that even for long rsa keys, you only need enough entropy to prevent exhaustive key-search. use 100 bits or so to seed the prng, then use the prng to generate a prime. reseed for each prime, and as often as you can for symmetric session-keys. i don't trust pseudo-random key-generation (it's no surprise that i don't, i suppose). please let me know if anything comes of this, like if someone builds it properly before i get around to it. i suppose i ought to build it, test it, and write a paper about how much fun it was, but i have too much work, and too many papers to write. -don davis, boston -----------------egassem dedrawrof--------------------

1 0

FORSALE: HP48GX
by teg 17 Dec '03

17 Dec '03

-----BEGIN PGP SIGNED MESSAGE----- Given the usefulness of an HP calculator as a secure point of encryption, someone might be interested in my old calculator. I haven't used it since I graduated, except for balancing my old business books, way overkill. I'd love to give it a good home. HP 48 GX Calculator The most powerful calculator in the world Currently loaded with many Applied Math and Engineering programs written by myself (yours free with the calculator ;-) About 2 years old, excellent condition $125, negotiable HP IR Calculator Printer Infrared printer for the HP line of calculators Very light, battery operated About 2 years old, excellent condition $75, negotiable Please contact me if you are interested. My apologies if you consider this post selfinterest and off topic. - --- [This message has been signed by an auto-signing service. A valid signature means only that it has been received at the address corresponding to the signature and forwarded.] -----BEGIN PGP SIGNATURE----- Version: 2.6.2 Comment: Gratis auto-signing service iQBFAwUBMQRl3CoZzwIn1bdtAQGoYQF9GrClhiKQZgx9RApC2G7fwRZllDV8sNcx sX+Vr8tk1FqXRqZkCHa6UrZRS3xm/91o =pzQ7 -----END PGP SIGNATURE-----

1 0

Re: Random noise from disk drives
by Rich Salz 17 Dec '03

17 Dec '03

**I blew it.** Don's address is don(a)cam.ov.com. Sorry.

1 0

Query over PZ
by tallpaul＠pipeline.com 17 Dec '03

17 Dec '03

Anonymous User had a question over PZ's politics. PZ has posted to the list and thus has, I believe, his address available to list subscribers. How did PZ answer the question privately when Anonymous User posed the question to PZ on private e-mail? If Anonymous User did not first pose the question to PZ privately, why did he post it to the list publically? -- tallpaul PS: I'm likely to have some followup on these questions, but wanted to pose them to AU first.

1 0

FW: Veriphone and Netscape Team to Provide Internet Payment Solutions
by Pete Loshin 17 Dec '03

17 Dec '03

Here's the Netscape press release on their Verifone collaboration; note that they explain nothing about 128 bits or 1024 bits--but Netscape now appears to claim to be the sole developers of SEPP. It looks like Verifone will be building their approval/verification mechanism into a Netscape product, and selling it themselves. I've removed contact info and the source at Netscape (in their PR dept), just to avoid unnecessarily nettling anyone. Presumably Netscape shouldn't be able to complain about it getting passed around since it will likely be the basis of reports in various trade/news rags by next week anyway. Enjoy. -Pete Loshin pete(a)loshin.com ---------- COPITHORNE & BELLOWS (for VeriFone) [contact info deleted] VERIFONE, INC. [contact info deleted] NETSCAPE [contact info deleted] VeriFone and Netscape Team To Provide Internet Payment Solutions Worldwide Leaders To Accelerate Commerce on the Internet REDWOOD CITY, Calif. - January 22, 1996 - VeriFone, Inc. (NYSE: VFI), and Netscape Communications Corporation (NASDAQ: NSCP) today announced an agreement to create comprehensive Internet payment solutions, enabling merchants, banks, credit card companies and consumers to expand the use of the Internet for electronic commerce. The agreement between VeriFone, the world leader in retail electronic payment systems, and Netscape, the premier provider of open software for linking people and information over enterprise networks and the Internet, is designed to accelerate acceptance of the Internet as a mainstream vehicle for commerce. Under terms of the agreement, the two companies will join forces to address the transaction processing needs of both financial institutions and merchants. For financial institutions, Netscape's gateway software and VeriFone's payment technology will be combined to create a seamless interface between the Internet and existing financial networks. VeriFone will also provide customization, installation and support services for these customers. To serve merchants, VeriFone will resell Netscape's server and commerce products bundled with VeriFone payment functionality to deliver complete software solutions for merchants who want to broaden their reach by establishing a commercial presence on the Internet. VeriFone will provide merchants with these solutions through its existing customers and channels. Netscape and VeriFone create a combination that is uniquely qualified to fulfill the promise of global electronic commerce. VeriFone's point-of-sale systems for online payment processing and credit card verification are among standards recognized by thousands of banks, credit card and debit card processors, and millions of merchants and consumers throughout the world. VeriFone's solutions today address the technology standards, payment methods and government regulations of over 90 countries around the world, enabling the two companies to accelerate their goal of deploying Internet payment systems worldwide. "Netscape's agreement with VeriFone will bring a comprehensive Internet payment solution to merchants, banks, credit card companies and consumers which will continue to fuel electronic commerce worldwide," said James Barksdale, chief executive officer at Netscape. "By combining VeriFone's electronic payment technology with Netscape's complete line of server and commerce software products, financial institutions and merchants can implement a complete payment solution." Creating Comprehensive Payment Solutions The new solutions will be designed to be compatible with current and future credit card payment protocols, including Secure Electronic Payment Protocol (SEPP), developed by Netscape. The two companies are working with leading card associations and processors to include additional protocols, supporting future payments instruments such as micropayments and debit transactions. "To truly enable Internet commerce, banks, credit card companies, merchants and consumers need a trusted interoperable system as accessible and secure as current payment systems," said Hatim A. Tyabji, chairman, president and chief executive officer of VeriFone. "Consumers need to know they can safely pay for something electronically, merchants need to know they can safely accept it, and banks and other financial organizations need to be able to process it. Our goal is to give merchants the same 'plug and play' payment access on the Internet that they have come to expect in their retail stores." VeriFone expects the integrated Netscape/VeriFone Internet gateway systems for financial institutions and the merchant systems to become available in the second to third quarters of 1996 from VeriFone, with increasing payment protocols and enhancements becoming available throughout the year. Company Backgrounds Netscape Communications Corporation (http://www.netscape.com) is a premier provider of open software for linking people and information over enterprise networks and the Internet. The company offers a full line of Netscape Navigator clients, servers, development tools, and Netscape Internet Applications to create a complete platform for next-generation, live online applications. Traded on NASDAQ under the symbol "NSCP," Netscape Communications Corporation is based in Mountain View, California. VeriFone, Inc. (http://www.verifone.com) is a leading global provider of Transaction Automation and Internet commerce solutions used to deliver electronic payment services to financial institutions, retail merchants and consumers, as well as government agencies, healthcare providers and benefits recipients. The company's more than 30 facilities -- including regional offices, development centers, and manufacturing and distribution centers -- are located throughout North and South America, Europe, Asia, Africa and Australia. To date, VeriFone has shipped more than 4.8 million Transaction Automation systems, which have been installed in over 90 countries. The company's 1994 net revenues totaled $309.1 million. In 1995, VeriFone acquired Enterprise Integration Technologies (http://www.eit.com) -- a leading provider of software and consulting services for electronic commerce on the Internet. (30) VeriFone is a registered trademark of VeriFone, Inc. All other brand names and trademarks are the property of their respective owners.

2 1

RE: IPSEC == end of firewalls (was Re: (fwd) e$: PBS NewsHour, Path Dependency, IPSEC, Cyberdog, and the Melting of Mr.)
by Pete Loshin 17 Dec '03

17 Dec '03

More to the point, I don't think it's possible to trust the security of the network software USERS in any case. -Pete Loshin pete(a)loshin.com Perry Metzger wrote: >Nelson Minar writes: >> I'm all for the end of ridiculous non-TCP/IP protocols, but does >> anyone believe this point about encrypted IP traffic eliminating the >> need for firewalls? > >There is division in the IETF community on this point. > >Phil Karn (who I have the greatest respect for) thinks IPSEC means we >can get rid of the firewalls. I, for one, don't -- they are there >largely because people don't trust that their networking software is >free of security holes, and cryptography doesn't fix security holes >for the most part. > >Perry

1 0

Re: New Nym Server at nym.alias.net
by John Perry 17 Dec '03

17 Dec '03

The following message is a courtesy copy of an article that has been posted as well. Hello Everyone, [sorry about that last post, I hit send before I was ready. Here it is with more information.] I'm proud to announce the new nym server at nym.alias.net. The address for the nymserver is: nymrod(a)nym.alias.net (Catchy name huh?) Anyway nymrod is running Matt Ghio's latest nymserver code. It's been tested and I think (cross your fingers) all the bugs have been worked out. It works the same at the nym server at c2.org and gondonym. If you need a quick tutorial, send email to: help(a)nym.alias.net Here's the PGP public key for nymrod(a)nym.alias.net: -----BEGIN PGP PUBLIC KEY BLOCK----- Version: 2.6.2 mQCNAzEEgxQAAAEEAKSv0YzZpc/4Kgn+L9fzzTlCp1LLdSEnu5YnU3AB068XwsF9 rLEn6C7AT8UP0JZAOl/bbhVir5/oCDg93l1iVIdSm3LWis/FXo1r3pVdAFd6PZrl V/gibZpeAA6sthXRR1OdatuesW11zvs/jee8yUl8rcHfsEbbpE969vAzW/8FAAUR tDVOeW0gU2VydmVyIGF0IFZpc2hudS5BbGlhcy5OZXQgPG55bXJvZEBueW0uYWxp YXMubmV0Pg== =W87x -----END PGP PUBLIC KEY BLOCK----- Enjoy!! John Perry - KG5RG - perry(a)vishnu.alias.net - PGP-encrypted e-mail welcome! WWW - http://www.alias.net PGP 2.62 key for perry(a)vishnu.alias.net is on the keyservers.

1 0

RE: Netscape + Verifone
by Steve Makrecky 17 Dec '03

17 Dec '03

What is the frequency on the back of your pager? Some people would like to monitor this type of information. ---------- From: owner-cypherpunks(a)toad.com on behalf of Perry E. Metzger Sent: Monday, January 22, 1996 11:13 To: cypherpunks(a)toad.com Subject: Netscape + Verifone My pager delivers me selected miniaturized Reuters news stories. I just got one that reads: REDWOOD CITY, CA - Netscape Communications Corp. and Verifone Inc. will devise a system to make electronic payments on the internet more secure. Anyone know anything about this? I'm away from my normal sources of such things (like Bloomberg terminals)... .pm

2 1

SS Obergruppenfuhrer Zimmermann (NOT!)
by mpd＠netcom.com 17 Dec '03

17 Dec '03

Anonymous User <nobody(a)c2.org> writes: > "Private communications between neo-Nazis on the network are > effected under a program called "Pretty Good Privacy", > devised by an American neo-Nazi sympathiser." > Robin Gedye (in Bonn) p.23 of "The Sunday Telegraph" > January 21, 1996 Before anyone takes this nonsense too seriously, one should realize that the exhibition of such microscopic views of technology by journalists and politicians is fairly common. Who can forget Caspar Weinburger's stirring speech in front of an illegally exported low-end VAX, explaining that the machine was a sophisticated electronic device for the tracking of missiles and troop movements, now in the hands of America's enemies. Then there was the newspaper article which explained in perfect seriousness that "GIF" was a secret computer code used by child molesters to encode images of their victims. Characterizing PGP as a neo-Nazi tool for private communications written by a sympathizer, while absurd in a larger sense, is hardly sillier than the prior examples. And all such scenarios contain a microscopic grain of truth as seen by someone somewhere with a severe case of tunnel vision. The reporter no doubt reports correctly that some neo-Nazis use PGP to communicate privately. Doubtless PKZ supports the right of all people, including those with diverse political views, to conduct legal private conversations which cannot be overheard by their governments, as do most of the people on this list. I suppose in some obtuse sense this is sympathy. It is highly unlikely that anyone who uses Cypherpunk technology is as ignorant as this reporter. So let's just mail the poor guy a clue and move on. Things like this happen often, and it's not really worth a prolongued debate. -- Mike Duvos $ PGP 2.6 Public Key available $ mpd(a)netcom.com $ via Finger. $

7 6

Re: You want to read MY e-mail?
by Alan Pugh 17 Dec '03

17 Dec '03

=snip= >A bill by Sen. Paul Weissmann, R-Louisville, to make e-mail as privileged as >telephone conversations was killed in a Senate committee. > >A somewhat stronger bill still rests in the House. =snip= >Jim Carpenter, press secretary for Romer, said his office was "looking at all >the issues, all the options." > >Duke said he has always complied with requests for public records, but said the >request for e- mail messages goes too far. > >"I'm disturbed by it," Duke said. > >(John Sanko writes for the Rocky Mountain News in Denver.) > > OK, if _I_ can't read your e-mail Mr. Legislator, why should you >be able to read _mine_? a good question. perhaps _now_ they'll get serious about recognising email as eqivalent to postal mail. this would be an excellent way of pushing this elsewhere. when _their_ ox is gored, they'll scream like stuck pigs. one interesting aspect of this would be whether regular mail is considered a public record under the law as far as legislooters mail goes. if so, then she should get it. amp

1 0