Re: [Cbor] To be signed with packed CBOR

[Brendan Moran <[e-mail]>]

Hi Jim,

> On 2 Aug 2020, at 00:29, Jim Schaad <[e-mail]> wrote:
>
> Brendan,
>
> I know that this is currently an academic question, but something you said
> in the SUIT meeting has me slightly worried.  The problem with being the
> note taker is that you never have anytime to mentally go down the channels
> of ideas that pop up.
>
> One of the things that you said is that the use of packed CBOR would make
> things better when there are multiple authentication structures on what is
> basically the same data.  The problem is that if you do the packing prior to
> doing the signature, then you end up in the situation where you need somehow
> to also include the packing dictionary in the data that is being signed.  I
> don't have enough knowledge of how the manifest works but if after a
> signing, one can add more information to the manifest and then apply packing
> to that section which could add more items to the packed lookup tables thus
> breaking the signatures.

I think we probably got our wires crossed here. There’s a few points to this:

1. SUIT is organised in such a way that one would probably pack the COSE objects and the non-COSE objects in separate, non-overlapping packed CBOR structures. This should mean that you do not require a signature over the dictionary for COSE packing.

2. With nested packed CBOR, we might need a way to specify whether the packer recursed into an already-packed object or not. That would bring us back to the situation in 1.

As a rule, I suspect we should assume that any dictionary should apply to either unauthenticated objects OR authenticated objects, but NEVER both. This will require us to distinguish between recursively packed (nested object contains references to enclosing object dictionary) and simply packed (nested object contains no references to enclosing object dictionary) objects. This may require a tag.

Brendan

>
> All,
>
> This is one of the reasons that I was asking questions about doing the
> signature on a packed or unpacked version of the data.  The idea of adding
> things which might alter the lookup tables at the root means that you have
> problems knowing that what was signed is what is validated without doing
> things like walking the content to extract a new pair of lookup tables to
> include as part of the signed content.
>
> Jim
>
>

IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.