Hello,

Jumping into this thread late, sorry. Anyways, my proposal:

- I think we should stick with requiring 0x00 padding (I am avoiding the term 
'payload' here for a reason). This would prevent the abuse as a covert channel, 
is easily understood and implementable, and should not have any significant 
performance impact during creation of a message. 

- In the sense of "be conservative what you send, be liberal what to accept" 
I'm fine to change to text to require (recommend?) a Responder to ignore the 
contents. Which means that a malformed padding will never (rarely?) trigger an 
error response. However, my feeling is that we should still allow a "strict 
Responder" to indicate an error. That boils down to text like "a Responder 
MUST/SHOULD (?) ignore the contents of Padding" - I'd go for SHOULD,  and 
subsequently keep the definition to use FORMERR if a strict Responder wishes to 
indicate an error

- Further, I understand that TLS level compression is discouraged for DNS 
anyways ( I'm not an expert in this field, but that's what I understood what 
dkg said in our initial hallway discussion in Praha), so I'd stick with 0x00. 
If we find out this is really a problem,  we can always define a padding with 
more sophisticated contents which Updates or Obsoletes this document. Requiring 
that a Responder ignores the contents would create forward compatibility on the 
Responder end.

I'm more than happy to propose text if th WG agrees with the above summary :-)

Alex
(from mobile, sorry for brevity and typos)

---- Stephen Farrell schrieb ----

>
>
>On 17/11/15 01:43, Daniel Kahn Gillmor wrote:
>> On Mon 2015-11-16 16:20:21 -0500, Stephen Farrell wrote:
>>> I agree that that is clearly the opinion in the TLS WG today. I'm
>>> less sure I agree that's correct - my fear is that it's being driven
>>> maybe too much by browsers and the web, where the conclusion is
>>> valid.
>> 
>> Well, certainly for a system-wide (or application-wide) arbitrary DNS
>> client it's the same risk.
>
>Similar but not quite the same. Cookies are the main likely target
>of CRIME but a DPRIVE equivalent can only target the (non-inserted)
>QNAME (or is there more?), which hasn't quite got the same kind of
>sensitivity.
>
>>  We have no way of knowing how many avenues
>> there are for attackers to be able to make the user make certain DNS
>> requests to their preferred privacy-preserving resolver.
>
>Sure. I would guess the introduction of DPRIVE will change the
>attack surface and new attacks will be discovered. (Timing may
>well prove a fertile ground I would guess.)
>
>> And since we expect (D)TLS channels to be kept open and reused, that
>> attacker-controlled data will be touching the same deflate dictionary
>> that the ostensibly private queries are using.  This is the same
>> scenario as CRIME and BREACH, no?
>
>Depends. It is unclear to me why padding doesn't help here,
>if applied after compression. (Mind you, at that point I don't
>know why one is bothering with compressing;-)
>
>> 
>>>> Any attempt at compression needs to happen at the application layer
>>>> itself with full knowledge of the risks and tradeoffs. 
>>>
>>> FWIW, I don't agree with the above, I think it's generalising too
>>> much from the browser use-case.
>> 
>> The point here is that if you don't know how the application layer mixes
>> attacker-controlled data with data that needs to be confidential, TLS
>> can't possibly do compression safely.
>
>But what is the application layer for a system wide stub resolver?
>Your argument would also imply DNS/DPRIVE can't do compression either.
>
>> So while i agree with you that there could be (non-browser) applications
>> that clearly never mix any attacker-controlled data with
>> sensitive/private data on the same TLS channel, the only way to know
>> whether you're in such a state is by reasoning about it *at the
>> application layer*, which is why that's where the decisions about
>> compression need to be made.
>
>We don't agree about that. I don't think I've seen a convincing
>argument that's the *only* way to know. (E.g. always emitting
>fixed size packets in a single write and sending some cover
>traffic now and then when there's any compression in play would
>also seem to help, no?) But that's off topic for this list I
>guess, and may get too complicated before it'd get useful
>enough.
>
>> 
>> anyway, i'm glad we agree on the outcome :)
>
>Yep - isn't the entire discussion of compression for DNS
>pretty moot anyway? It's not clear to me what compressible
>plaintext there is where compression would be a real benefit.
>
>Cheers,
>S.
>
>
>> 
>>      --dkg
>> 
>> _______________________________________________
>> dns-privacy mailing list
>> [email protected]
>> https://www.ietf.org/mailman/listinfo/dns-privacy
>> 
_______________________________________________
dns-privacy mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dns-privacy

Reply via email to