RE: more on malicious hardware
>>Adding a backdoor to chips is a different story, though, since that would require cutting a second set of masks. >>I am assuming that there must be no backdoor in the legitimately produced chips since the client would detect >>it as a slight violation of some of their timing simulations. The client also often inspects the masks before >>the chips are produced and basically reverse-engineers the whole chip on that level. A backdoor -- hardware or software -- in a smart card or TPM would be difficult to detect by either of these means. In the case that nation A is buying these from nation F, don't you think that nation F would be motivated to slip in a couple extra lines of code or a couple extra 100 gates just in case? If A got into a tangle with C, F would in a very strong position. - The Cryptography Mailing List Unsubscribe by sending "unsubscribe cryptography" to [EMAIL PROTECTED]
Re: more on malicious hardware
"Perry E. Metzger" <[EMAIL PROTECTED]> >It turns out that the counterfeit chips business is booming: > >http://www.eetimes.com/rss/showArticle.jhtml?articleID=207401126 If you look at the linked article you'll see the example they give of counterfeit chips is: >The fake chips are often rebranded parts claiming greater performance than >they actually have to earn the sellers a fast profit. which has been going on for about 15 years or so [0], back then they'd grind the tops off the ceramic heat spreader on P5s and print on a new speed rating, later when distributors got wise to this they went to laser-etched labels that were indistinguishable from the originals. The other example given in the article was chips for avionics/milspec use, which just means that they rebranded standard non-milspec parts as being for milspec use, and that scam predates CPU speed re-gradings by some time (in my device zoo I have some interesting not-really-milspec 7400s dating back to the early 80s, as well as other oddities like Apple-branded TTL and who knows what else). Neither of these are really counterfeits, they're genuine chips remarked for use outside their intended use parameters. In fact given the overclockability of many CPU binnings and the fact that manufacturers have in the past sold CPUs at lower speed grades than they tested for at manufacture in order to meet price points (in other words a CPU was tested for x MHz, branded for x - y MHz to meet a price point, and then had the branding ground off and was re-branded for its original binned speed by dodgy re- distributors), the result may be no loss at all [1]. It'd be interesting to see some figures for genuine faked-from-whole-cloth devices vs. basic re-brandings/re-binnings/whatever of original products, I would imagine there's vastly more re-branding and re-binning going on than someone actually cloning (say) a CMI8788 and selling it as the real thing. Peter. [0] And probably for a long time before that, AFAIK it first became a major issue when the high price differential between different speed binnings of P5s made it a lucrative business. Making a non-milspec 7404 into a milspec part was a lot less profitable. [1] That one should keep the lawyers busy: If I sell a device de-rated purely to meet a price point and someone else on-sells it at its original designed rating, with what would you prosecute them? - The Cryptography Mailing List Unsubscribe by sending "unsubscribe cryptography" to [EMAIL PROTECTED]
Re: more on malicious hardware
Perry E. Metzger wrote: It turns out that the counterfeit chips business is booming: http://www.eetimes.com/rss/showArticle.jhtml?articleID=207401126 In combination with the news about what as few as 1500 extra gates can do, this is especially worrisome. So when do the contests start by adventuresome minds to see how *few* gates are needed to compromise a chip's security much like the self replicating code referenced by Ken Thompson in his paper? Best, Allen - The Cryptography Mailing List Unsubscribe by sending "unsubscribe cryptography" to [EMAIL PROTECTED]
Re: more on malicious hardware
Perry E. Metzger wrote: It turns out that the counterfeit chips business is booming: http://www.eetimes.com/rss/showArticle.jhtml?articleID=207401126 In combination with the news about what as few as 1500 extra gates can do, this is especially worrisome. Chip pirating is a huge problem. Part of the reason we think it is ethical to release the Crypto-1 details is the fact that pirated Mifare chips have been sold for years. Pirating becomes more widespread as the chip manufacturing is outsourced. The cost of manufacturing a small volume of chips is dominated by cutting the "masks" used in the lithographic process steps. Once these masks are built, shelling out more chip copies is relatively cheap which enables manufacturers to overproduce and sell pirated copies. Adding a backdoor to chips is a different story, though, since that would require cutting a second set of masks. I am assuming that there must be no backdoor in the legitimately produced chips since the client would detect it as a slight violation of some of their timing simulations. The client also often inspects the masks before the chips are produced and basically reverse-engineers the whole chip on that level. - The Cryptography Mailing List Unsubscribe by sending "unsubscribe cryptography" to [EMAIL PROTECTED]
