Re: [TLS] Collisions (Re: Nico's suggestions - Re: Consensus Call:

[Marsh Ray <[e-mail]>]

On 5/11/2010 3:07 PM, Nicolas Williams wrote:
> On Tue, May 11, 2010 at 03:04:38PM -0500, Marsh Ray wrote:
>> On 5/11/2010 2:46 PM, Nicolas Williams wrote:
>>> On Tue, May 11, 2010 at 02:32:12PM -0500, Marsh Ray wrote:
>>>> So how much better is this caching scheme than what you could have today
>>>> with session resumption?
>>>
>>> I recall this question being asked long ago.  IIRC the idea is that you
>>> could cache a cert chain for many servers' certs
>>
>> Isn't one of the implications of the choice of a non-collision-resistant
>> hash that you must use a completely separate cache for each hostname?
> 
> I don't think that's clear yet.  However, it is clear that making this
> extension safe will require more care than it seemed to me the first
> time I glanced at it.  It may well be that the answer to your question
> is "yes", in which case I think you'd be right that session resumption
> is plenty good enough.

Note: I didn't say precisely that I thought session resumption was good
enough, nor did I say that cached info isn't the greatest thing since
sliced bread. I just wanted to ask some direct questions.

Alternatively, if we determine that indeed the non-collision-resistance
of the hash function is the root of all remaining concerns that would be
very positive. We could solve them all in one stroke with
s/FNV-1a/SHA-256/g.

*ducks*

- Marsh