FIPS 140-2 validation is mandated for encryption modules used by the US federal government when dealing with sensitive-confidential (but not classified) information. I believe it's also mandated for the Canada federal government, as well.
This mandate means that even if it takes 10 minutes to initialize, they are *required* to deal with it, even if they really could use a faster, stripped-down version. This also means that if you're in a situation that doesn't actually require FIPS validated cryptographic modules, you can simply use FIPS-compatible (but not FIPS-validated) algorithms for interoperability with them. Most notably, this means not using MD5 at all (except in conjunction with one of the SHA algorithms -- there's a situation in TLS where the keying material is produced by MD5+SHA1, and that has been held to be okay in that situation because the output is still not predictable and still not subject to an easy chosen-birthday attack). If you decide to try to get another FIPS validation, you should expect an interminable timeframe (not merely 6 months, but possibly several years) and a HUGE financial outlay (and I mean "in the orders of tens if not hundreds of thousands of US dollars"), and if you're not already an expert in navigating the system I would recommend avoiding it if at all possible. Steve Marquess is the one who could most likely explain the entire process, and if you look in the mailing list archives you can see several messages that he and John Weathersby (both of the Open Source Software Institute) have posted about the process. -Kyle H On Sun, Dec 21, 2008 at 11:26 PM, a_l t <avisar.li...@gmail.com> wrote: > If I want to validate a stripped down module (let's say for simplicity just > without the unwanted self tests), is there a fast way to do it, or I should > expect a 6 months process? > I also didn't quite understood what you meant in the last sentence: "Where > FIPS validation is mandated operations considerations take second place." > Thanks, > Avisar > > On Mon, Dec 22, 2008 at 1:46 AM, Steve Marquess <marqu...@oss-institute.org> > wrote: >> >> a_l t wrote: >>> >>> I'm running it on TI DSP (C6455) and it takes around 1 minute. >>> >>> On Sun, Dec 21, 2008 at 10:28 PM, Victor Duchovni >>> <victor.ducho...@morganstanley.com >>> <mailto:victor.ducho...@morganstanley.com>> wrote: >>> >>> On Sun, Dec 21, 2008 at 05:28:14PM +0200, a_l t wrote: >>> >>> > I'm using the FIPS approved OpenSSL. In the initialization it runs >>> > several self tests which take quite a long time. I use only several >>> > algorithms from the OpenSSL, is there a way to remove the self >>> > tests of the algorithms that I don't use (like DSA) without losing >>> > the FIPS certification. >>> >>> How long do the self-tests take? >> >> Ouch. If you must enable FIPS mode you don't have many options. Find >> some other faster product, if there is any; use a validated hardware device; >> hack and validate a stripped down derivative of the OpenSSL FIPS Object >> Module. Where FIPS validation is mandated operations considerations take >> second place. >> >> -Steve M. >> >> -- >> Steve Marquess >> Open Source Software institute >> marqu...@oss-institute.org >> >> ______________________________________________________________________ >> OpenSSL Project http://www.openssl.org >> User Support Mailing List openssl-users@openssl.org >> Automated List Manager majord...@openssl.org > > ______________________________________________________________________ OpenSSL Project http://www.openssl.org User Support Mailing List openssl-users@openssl.org Automated List Manager majord...@openssl.org
