> On Jan 18, 2023, at 1:11 PM, Ilari Liusvaara <[email protected]> wrote:
> 
> On Wed, Jan 18, 2023 at 11:15:10AM -0700, Laurence Lundblade wrote:
>> HPKE integrates into COSE in two ways:
>> 1) As a COSE_Recipient where it encrypts the CEK  — two-layer mode
>> 2) As the content encryption “algorithm” — one-layer mode
>> 
>> A more detailed description is here 
>> <https://mailarchive.ietf.org/arch/msg/cose/8Ga_-k_whir8Z4lzdpmPG6KDav4/>.
>> 
>> draft-ietf-cose-aes-ctr-and-cbc works it for two-layer mode, but not
>> for one-layer mode.
>> 
>> To use non-AEAD for one-layer mode, it seems like HPKE will have to
>> be modified. There will have to be a definition of algorithm IDs for
>> non-AEAD algorithms in the HPKE algorithm ID space and such.
> 
> I do not think such modifications are acceptable.
> 
> 
> One technically could use HPKE in exporter-only to derive an encryption
> key and then use that.

That would be a two-layer solution with a COSE_Recipient, right? I think it 
would work, but it doesn’t have the ability to support multiple recipients like 
the current two-layer proposal. The current two-layer solution seems preferred.

So it seems the path forward for use of aes-ctr and aes-cbc is the current 
two-layer solution. If extra bytes on the wire for the COSE_Recipient structure 
are OK with everyone, there’s nothing to do.

LL
_______________________________________________
COSE mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/cose

Reply via email to