Adam Roach has entered the following ballot position for draft-ietf-acme-star-09: Discuss
When responding, please keep the subject line intact and reply to all email addresses included in the To and CC lines. (Feel free to cut this introductory paragraph, however.) Please refer to https://www.ietf.org/iesg/statement/discuss-criteria.html for more information about IESG DISCUSS and COMMENT positions. The document, along with other ballot positions, can be found here: https://datatracker.ietf.org/doc/draft-ietf-acme-star/ ---------------------------------------------------------------------- DISCUSS: ---------------------------------------------------------------------- Thanks for the work that everyone put into this document! I have one item that needs discussion, and then a small number of very minor editorial nits. I plan to ballot YES once the issue I identify below has been addressed. 7.3: > In order to avoid correlation of certificates by account, if > unauthenticated GET is negotiated (Section 3.4) the recommendation in > Section 10.5 of [RFC8555] regarding the choice of URL structure > applies, i.e. servers SHOULD choose URLs of certificate resources in > a non-guessable way, for example using capability URLs > [W3C.WD-capability-urls-20140218]. Thanks for reinforcing the privacy point from 8555 here; however, I think the situation here is substantially more sensitive. With base 8555 behavior, the existence of a URL (as can be deduced by distinguishing between 401 and 404 responses) poses a privacy risk, if parties can deduce a pattern to the URL paths, allowing correlation of users' domains. With unauthenticated GET, the ability to guess the associated URL would be incredibly dire, allowing arbitrary unauthorized parties to download the cert in question. While this may be somewhat obvious to the authors and working groups, it's the kind of thing that really needs to be spelled out in the security section, to avoid any oversight on the part of implementors. For a similar reason, I'm pretty sure that "SHOULD" is not the right level of requirement for unguessability. An ACME STAR service with guessable URLs may as well not perform validation at all, since it's effectively handing out everyone's cert to everyone on the Internet. Unguessability for unauthenticated GET needs to be a MUST, and the document should explain why this is a hard requirement. ---------------------------------------------------------------------- COMMENT: ---------------------------------------------------------------------- §3.3: > The Server SHOULD include the "Cert-Not-Before" and "Cert-Not-After" > HTTP headers in the response. Nit: "...HTTP header fields..." > Following are further clarifications regarding usage of these > headers, as per [RFC7231] Sec. 8.3.1. All apply to both headers. Nit: "...header fields..." > > o This header is a single value, not a list. Nit: "...header field..." > o The header is used only in responses to GET, HEAD and POST-as-GET Nit: "...header field..." > requests, and only for MIME types that denote public key > certificates. > o Header semantics are independent of context. Nit: "Header field..." > o The header is not hop-by-hop. Nit: "...header field..." > o Intermediaries MAY insert or delete the value, but MUST ensure > that if present, the header value equals the corresponding value Nit: "...header field..." > within the credential. > o The header is not appropriate for a Vary field. Nit: "...header field..." > o The header is allowed within message trailers. Nit: "...header field..." > o The header is not appropriate within redirects. Nit: "...header field..." > o The header does not introduce additional security considerations. Nit: "...header field..." > It discloses in a simpler form information that is already > available inside the credential. --------------------------------------------------------------------------- §3.3: > o Intermediaries MAY insert or delete the value, but MUST ensure > that if present, the header value equals the corresponding value > within the credential. This probably isn't what you want to say. Read literally, this imposes a requirements on intermediaries who are neither removing nor adding these header fields to validate that they match the value in the certs. That's clearly an unrealistic expectation. I suspect the intention here is that any entity who inserts a value must ensure that the newly inserted value matches the corresponding value in the certificate. _______________________________________________ Acme mailing list Acme@ietf.org https://www.ietf.org/mailman/listinfo/acme