> On Jul 11, 2024, at 9:00 AM, Orie Steele <[email protected]> wrote:
> 
> : ) Ilari you've got a lot of patience, you've reminded me of hpke info 
> length restrictions several times now.
> 
> I assume by setting info empty, HPKE internals are handling everything that 
> is needed here.
> 
> If they are not, is is ok for us to write something like:
> 
> For HPKE JWE Integrated Encryption and Key Encryption hpke-info SHOULD be 
> empty, however, if both sender and recipient agree, hpke-info can be set to 
> any value that is consistent with section ___ of RFC9180.

This would work for my use case, so I am in favor of it.  

> 
> Regards,
> 
> OS
> 
> On Thu, Jul 11, 2024 at 2:27 AM Ilari Liusvaara <[email protected] 
> <mailto:[email protected]>> wrote:
>> On Wed, Jul 10, 2024 at 06:36:28PM -0500, Orie Steele wrote:
>> > Hi Matt,
>> > 
>> > In the other threads, Ilari has suggested using setup info in a manner
>> > similar to how info is used with concat kdf in ECDH-ES.
>> 
>> Maybe I was unclear. What I suggested was:
>> 
>> - Key Encryption: 
>>   * HPKE aad: ECDH-ES FixedInfo
>>   * HPKE info: empty.
>> - Integrated Encryption:
>>   * HPKE aad: From JWE Section 5.1 step 14.
>>   * HPKE info: empty.
>> 
>> There is 16 bytes of fixed stuff in ECDH-ES FixedInfo, leaving 48 bytes
>> for enc, apu and apv. That is not enough.
>> 
>> 
>> > On Wed, Jul 10, 2024, 6:07 PM Matt Chanda <[email protected] 
>> > <mailto:[email protected]>> wrote:
>> > 
>> > > Hi Orie,
>> > >
>> > > The draft says: "The "Setup info" MUST NOT be used with either HPKE JWE
>> > > Integrated Encryption and HPKE JWE Key Encryption."  Could you provide 
>> > > some
>> > > more information about this part?
>> 
>> There are three problems with using setup info:
>> 
>> - Some HPKE implementations error if trying to use both info and aad.
>> - Info is limited to 64(!) bytes, which is not sufficient for JWE.
>> - If context does not match, it leads to mismatched keys, which are
>>   not guaranteed to be handled sanely by AEAD... Probably not
>>   significant (but that is not something one wants to hear in crypto).
>> 
>> 
>> > > RFC 9180 section 8.1 doesn't have this restriction unless using single
>> > > shot APIs.  I'm planning on using a nonce in the info field to make sure
>> > > the key material is specific to the transaction per the advice on NIST
>> > > 800-56A Appendix B.
>> 
>> The 64 byte restriction is still there. Aad has no length limit.
>> 
>> And aad works fine for binding context, because HPKE fails decryption on
>> authentication failure.
>> 
>> 
>> (And funkily enough, the recommendation is only on implementations that
>> do not support multi-shot at all. So if implementation exposes multi-
>> shot, it should let application use both info and aad in single-shot.)
>> 
>> 
>> 
>> 
>> -Ilari
>> 
>> _______________________________________________
>> jose mailing list -- [email protected] <mailto:[email protected]>
>> To unsubscribe send an email to [email protected] 
>> <mailto:[email protected]>
> 
> 
> --
> 
> ORIE STEELE
> Chief Technology Officer
> www.transmute.industries
>  <https://transmute.industries/>
> _______________________________________________
> jose mailing list -- [email protected]
> To unsubscribe send an email to [email protected]

_______________________________________________
jose mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to