This article is more than 1 year old

Microsoft denies NT failed crypto tests

But some form of compliant module appears to be in the works anyway...

Microsoft has hotly denied claims that Windows NT 4.0 has failed US government cryptography tests (see earlier story), and has posted a response on the NTbugtraq mailing list. According to Microsoft, the module submitted for evaluation has not yet undergone final testing, and -- QED -- cannot have failed. The Microsoft riposte takes the form of a FAQ produced by Microsoft NT Security product manager Jason Garms. The FAQ denies failure of US government FIPS 140-1 testing, and points out that it is the module Microsoft Enhanced DSS/Diffie-Hellman Cryptographic Provider (CP) which has been submitted for testing, not NT. This is the module which has not undergone final testing. He also points out that the testing lab was not contracted to examine the secuirty of NT 4.0 or the Microsoft CryptoAPI, and says that "while accredited CMVP testing laboratories do make design and implementations back to the vendor to maximise the probability that a cryptomodule will achieve FIPS 140-1 validation, no redesign or change in the Windows NT product was required". But we've not completed testing yet, one might observe. As regards applications, Garms claims that there will be no disruption, the difference being that some apps such as Internet Explorer 4.0 or Outlook 98 won't be able to take advantage of the new cryptomodule’s features, while IE 5.0 will be able to. Finally, he says that the "shipping vehicle" for the new cryptomodule will be determined once it completes evaluation. So, you might wonder, what has this little lot been all about then? On the one hand we've got claims that NT has failed cryptographic tests, and on the other we have absolute denials. It would appear that someone is being less than candid. If we rewind to the original Network World story then we can note that one Patrick Arnold, program manager at Microsoft Federal Systems, is quoted as saying a fix will be offered in Q1, and that "only Internet Explorer 5.0 will know how to work in FIPS mode". So Arnold clearly thinks there's a problem of some sort that warrants a fix. And maybe we can dovetail that neatly into what Garms is talking about -- a FIPS module for NT that is currently undergoing testing, and that will quite possibly ship in this quarter. So maybe Garms FIPS 140-1 cryptomodule is Arnold's fix. We'll leave you to draw your own conclusions about who's being candid here, and about the -- allegedly untested -- CryptoAPIs of NT 4.0. ®

More about

TIP US OFF

Send us news


Other stories you might like