Best Practices

[David Shaw]

On Dec 12, 2010, at 11:21 AM, Robert J. Hansen wrote:

> On 12/12/2010 10:23 AM, Daniel Kahn Gillmor wrote:
>> What part of OpenPGP certificates require SHA-1?
> 
> ... At first blush, V4 certificate checksums, symmetrically encrypted
> integrity protected data packets, the MDC system in general, certificate
> fingerprints, etc.  I just grepped through the RFC looking for any
> hardcoded SHA-1; David is probably a much better reference than I am on
> this.
> 
> Probably the most annoying -- to me, at least -- is the fingerprint
> requirement.  If a preimage collision is discovered in SHA-1 then it's
> all over.  I can take your signature on my enemy's key, graft it onto my
> own impersonator of my enemy's key, and then get others to believe it.

Yes.  The other uses of SHA-1 are not nearly as significant as the fingerprint (and thus key ID).  For example, it's true that the MDC uses SHA-1, but no big deal - just make a new MDC that uses whatever you like, and repeat as needed.  Virtually all deployed code will handle this correctly (for example, a feature flag indicating the existence of the "mdc2" capability), and use it only when all participants can handle it.

The fingerprint issue is more than just making a new packet for a new MDC or revocation subpacket, though.  There is no concept in OpenPGP of a flag telling an implementation how to calculate the fingerprint - or rather there IS a flag: the version field, but its hardcoded :)

David