[Swan-dev] [cryptography] Announcing Mozilla::PKIX, a New Certificate Verification Library (fwd)

Lennart Sorensen lsorense at csclub.uwaterloo.ca
Thu Apr 10 18:52:26 EEST 2014


On Thu, Apr 10, 2014 at 11:43:22AM -0400, Paul Wouters wrote:
> And use 15 year old cryptographic code that has seen no audit?

Well that was a different problem.

Almost everything uses openssl or gnutls.  Picking the 3rd but not very
common option is rather annoying for embedded systems.

> And having to extend that old code that only supports AES/3DES and
> MD5/SHA1 with newer algorithms to support SHA2, SHA3, AES-GCM, AES-CCM,
> AES-CTR, and IPsec suite B Elliptic Curves?
> 
> And than who will pay to audit/certify that code?
> 
> We had to switch to a library to do this work. As the codebase already
> supported using NSS instead of our old code, we opted to continue
> that path.

Hmm, I hadn't realized it already had some NSS support.

> The same is true for the X.509 support required, and adding to the code
> that deals with ASN.1/X.509 parsing of the above mentioned new crypto.
> 
> And this is true not only for the userland, but also applies to KLIPS
> versus NETKEY/XFRM.

We gave up on klips many years ago in favour of netkey.

> Now, we only use a very small portion of NSS, and perhaps we can talk to
> the NSS people about factoring that out into a separate smaller library.

It might reduce the size a bit, but I don't see it reducing the
verification/certification work.

Oh well I guess triplicate crypto is the future we are stuck with then.

> We understand the pain of having to add NSS to embedded platforms. But
> there is really no alternative. The only switching that is possible
> would be from NSS to openssl. It would make life easier on embedded
> platforms that already need openssl. But for us it adds the overhead
> of all the certificate loading/parsing code as openssl does not have
> the same concept as the NSS DB for a "store" of cryptographic information.

So any work on the openssl option that paul mentioned about a year ago
as a future option?  Of course openssl's crazy license makes trouble
for some projects too.

Of course I am wondering what kind of work will be involved in generating
the nss database each boot from the configuration database.  Probably not
too hard.

-- 
Len Sorensen


More information about the Swan-dev mailing list