Re: [AVTCORE] Comments on draft-ietf-avt-srtp-aes-gcm-01

[David McGrew <[e-mail]>]

Thanks for the comments, Michael.  More inline:

On Sep 12, 2011, at 10:33 AM, Peck, Michael A wrote:

> Hi,
>
> I support the publication of draft-ietf-avt-srtp-aes-gcm.
>
> I have a few minor comments on the document:
>
> Section 1.1 Crypto Context
> For clarity, suggest changing the text:
> which results in authentication of each participant in the SRTP  
> session and in their possession of a shared secret master key and a  
> shared master salt
> to:
> which results in authentication of each participant in the SRTP  
> session and possession by each participant of a shared secret master  
> key and a shared master salt
>
> s/maser key/master key
> 	
> Section 1.2.3:
> s/inititialition/initialization
>
> s/6-byte/6-octet
> (for consistency with the rest of the paragraph)
>
> Section 1.2.4:
> "12-octet SRTP session encryption salt" - should it say SRTCP  
> instead of SRTP here?
>

Good catch.

> Section 1.2.5:
> s/plaintext, which/plaintext and which
>
> Section 1.2.6:
> Missing a . at the end of the paragraph.
>
> Section 1.2.6.1:
> RFC 3711 section 3.4 states that the authentication tag is a  
> REQUIRED field in SRTCP.
> But draft-ietf-avt-srtp-aes-gcm-01 states in Figures 4 & 5 that the  
> authentication tag is NOT RECOMMENDED for SRTCP (because GCM already  
> places the authentication tag elsewhere).
> It may be appropriate to add some explanatory text about why the RFC  
> 3711 section 3.4 requirement for an authentication tag does not  
> apply here.
>
> "and eight octets and the encryption flag are treated as plaintext"  
> - I don't understand this wording - I don't know which eight octets  
> it is referring to, and the encryption flag is treated as AAD, not  
> plaintext.
> Perhaps just remove this whole sentence and rely on the diagram  
> instead - changing the paragraph text to just:
> "Figure 4 below shows how fields of an SRTCP packet are to be  
> treated when the encryption flag is set to 1."

Looks like a cut/paste error.  The "first 8 octets" was supposed to  
refer to the eight octets of the SRTCP header; it was supposed to have  
read "When the encryption flag is set to 1, the first 8-octets, the  
encryption flag, and the SRTCP index are treated as AAD.  All data  
between the SRTCP header and the encryption flag are treated as  
plaintext."

>
> Section 2.2:
> s/AEAD-AES-128-GCM-8/AEAD_AES_128_GCM_8
> s/AEAD-AES-256-GCM-8/AEAD_AES_256_GCM_8
>
> Section 4:
> Should the DTLS-SRTP SRTP Protection Profiles defined here follow  
> the example of RFC 5764 Section 4.1.2 and also include parameters?  
> (cipher, cipher_key_length, cipher_salt_length, maximum lifetime,  
> auth function, auth_key_length, auth_tag_length)
> We have a table in Section 7 of draft-peck-suiteb-dtls-srtp that may  
> be a good fit here.

That sounds like a good idea, thanks for the pointer.

David

>
> Thanks,
> Mike
>
> --
> Michael Peck
> The MITRE Corporation