TEEP Monday March 27 15:30-17:00 JST, Session III

Chairs: Nancy Cam-Winget, Tiru Reddy

Notetakers:
Kohei Isobe, Haiguang Wang

Agenda Bash & Logistics - Chairs (5min)

Hackathon Report – Akira Tsukamoto (15 min)

• Clarification of the CNF only containing hash value of TEEP agent
• Compromised TEEP agent: TEEP Broker running on SGX should be
  protected by SGX. No action fom the TEEP Agent as that is not the
  one compromised which is the consideration in the main draft.
• Clarification of token and challenge TEEP message. No changes
  required in the draft.
• CDDL and decision to use CBOR required updates to the Makefile in
  the implementation
• No new issues from this hackathon.

TEEP HTTP Transport – Dave Thaler (10 min)

•
Draft: https://datatracker.ietf.org/doc/html/draft-ietf-teep-otrp-over-http

• Expect to have draft version 15 to cover comments to address IESG
  comments
• Comments from INT AD (Erik Kline) to address errors around timeout
  considerations.
• All responses have been posted to the ADs all who have approved
• Stean Santesson (Secdir review) on normative vs. use of "strongly"
  now reviewed
• Sec AD: missing "message buffer" usage now clarified. Also URI
  guidance and now references RFC 3986
• Transport AD (Zaheduzzaman Sarker) commnts addressed from
  introduction (using normative language)
• TSV AD: POST messages without freshness are unreached. Adding
  Cache-Control header MUST NOT be used.
• ART AD (Francesca Palombini): requiring clarifications and
  referencnes to BCP195 and RFC9205
• ART AD: re-echos similar comments from other AD
• RTG AD (John Scudder) claificaion on client initiation
• Will publish v15 unless there's any other comments

TEEP Protocol and transport drafts – Dave Thaler (40 minutes)

• Draft: https://datatracker.ietf.org/doc/draft-ietf-teep-protocol/ •
Issues: https://github.com/ietf-teep/teep-protocol/issues

• Issue 297: Error return for QueryResponse, added update message for
  symmerty with QueryResponse.
• Issue 310: threat of compromised agent trying to use Attestation
  results from a healthy agent. Mitigation is to require a "cnf" claim
  as defined in S3.1 of rFC 8747
• Updated security consideration to ensure mitigation on the attack.
• Issue 281: EAT profile didnd't have any mandatory claims. Update now
  includes required claims
• Issue 289: relationship of TEEP to AR4si; informative reference with
  inclusion of 3 cases that are applicable (line between the TAM and
  verifier). Category 1: asks verifier to give both ar4si and teep
  profile; category 2: can only give 1 without nonce, send evidence in
  2 formats, category 3: can only ask one per evidence have TEEP agent
  talk directly to verifier to get ar4si
• Issue 285: update TEEP EAT profile : adding sw-name claim and
  manifests cialm to convey software information
• Issue 286: summary of crypto algorithm use cases. Clarifications on
  how to deal with sensitive information in EAT and SUIT reports. To
  ensure AEAD algorithms were used
• Add EAT/SUIT cipher negotiation supported-eat-suit-ciphersuites in
  QueryRequest : as like supported-teep-cipher-suites. Added new field
  covers both EAT and SUIT encryption. Should we split?
• Current draft may need a change as MTI document to be consistant
  with SUIT draft.
• WGLC on next step? Dependency of HPKE, EAT, SUIT.

Firmware Encryption - Hannes Tschofenig and Russ Housley (20 min)

• Major change made on the HPKE key distribution.
• Hannes: not sure if there's a way to remove the normative dependency
  but will have to review
• Dave Thaler: believes the TEEP document can make it an informative
  reference (but SUIT manifest and others do need to stay normative)
• Hannes: prefers that as the trust domains may fall into the same
  categorization
• Based on SUIT meeting outcome, can make the chanage to make FW
  encryption informative, but not sure can shift the others as they
  are required (e.g. normative)

AOB

cnf Implementation

• Hannes asked the details of cnf implementation in this hackathon.
• Kohei shared Ken's and his implementation. He couldn't find the
  concreate caluculate way of public key's hash in COSE compared with
  JWK. He'll post more details in TEEP mailing list.
• Akira Tsukamoto: the hash of public key in CNF is already on Github
• Dave: There is some concern on the KID value and Dave trying to
  anser the question.
• An issue need to be raised to address the cddl file updates.