Re: [websec] #58: Should we pin only SPKI, or also names

[Trevor Perrin <[e-mail]>]

On Wed, Aug 7, 2013 at 2:35 AM, Gervase Markham <[e-mail]> wrote:
> On 07/08/13 10:28, Trevor Perrin wrote:
>> This is relevant to the .well-known URI vs HTTP Header discussion.
>> Pinning CA names would reduce HPKP size, making frequent sending of
>> headers more palatable.
>
> Surely it would also significantly reduce flexibility?

I wouldn't remove key pinning.  Just allow pinning to CA names as well.

So an HPKP header could list keys both "directly" (by hash value) and
"indirectly" (by name).

Presumably the browser would have a table mapping names to key lists,
and would evaluate a site's cert chain by checking for an intersection
with all relevant keys in the pin (direct or indirect).


> At the moment, I can pin to a particular leaf, or a particular
> intermediate, or a particular root, or to a set of any of the above. I
> can decide where in the chain to pin depending on my analysis of the
> cost/benefit.

I'm not suggesting disallowing that.

That sounds a bit complicated for most sites, though.  I'm imagining
that listing several trustworthy CAs, by name, would be a safer and
easier pin for most?


Trevor