Time for a bis document ?


On Tue, 31 Mar 2026, 23:26 Matthias Gierlings, <matthias.gierlings=
[email protected]> wrote:

> Dear DNSOP-WG Members,
>
> we are a team of researchers at Ruhr-University Bochum and University
> of Wuppertal who uncovered vulnerabilities in the DNS Queries over
> HTTPS Standard (RFC 8484)[1]. Those vulnerabilities enable exploits
> against DoH services that inject malicious, attacker controlled
> content into web-origins of victim sites. This novel class of
> Cross-Site-Scripting (XSS) attacks is called XSS-over-DoH and
> consists of:
>
> (1) *Direct XSS over DoH*, which directly delivers malicious markup
>      containing JavaScript, that is subsequently rendered and executed
>      in the origin of a vulnerable DoH server.
> (2) *CSP-Bypass over DoH* bypasses strong CSPs
>      (Content-Security-Policies) and defeats robust browser protection
>      mechanisms designed to suppress script execution in the face of
>      markup injection.
>
> Please refer to the attached document for further details on both
> attacks, the associated techniques and their impact on victims.
>
> The current version RFC 8484 has various issues, e.g.:
>
> - RFC 8484, p. 10, §5.4. "Content Negotiation" explicitly permits
>    media types other than "application/dns-message" to be used for DoH
>    communication. This enables (1) Direct-XSS-over-DoH. Further §5.4.
>    as well as §9. "Security Considerations" miss to highlight the
>    dangers of using other media types especially those that are
>    renderable by clients.
>
> - Even in cases where services set the more common media type
>    "application/dns-message" MIME-sniffing in browsers causes them
>    to treat JavaScript payloads as script code despite conflicting
>    information present in the "Content-Type:
>    application/dns-message" header. This enables (2) CSP-Bypass over
>    DoH.
>
> - From RFC 8484, p. 12 8.1. On the Wire:
>    > DNS over TLS [RFC7858] provides similar protections, while direct
>    > UDP- and TCP-based transports are vulnerable to this class of
>    > attack An experimental effort to offer guidance on choosing the
>    > padding length can be found in [RFC8467].
>
>    > Additionally, the use of the HTTPS default port 443 and the
>    > ability to mix DoH traffic with other HTTPS traffic on the same
>    > connection can deter unprivileged on-path devices from
>    > interfering with DNS operations and make DNS traffic analysis
>    > more difficult.
>
>   This section suggests that, due to the use of TLS, DoH achieves a
>   protection level similar to that of DoT. However it fails to
>   recognize the severe consequences that result from protocol
>   confusion between HTTP(S) and DoH (please refer to the attached
>   document for details). This confusion is only possible because
>   DoH reuses the HTTP(S) well-known port and the HTTP(S) message format
>   to encapsulate DNS. With DNS over TLS (DoT) a confusion is impossible
>   because in default configurations the web-origin of DoT services is
>   different from that of a website using the same domain due to the use
>   of a different service port (853).
>
> We would like to open the discussion on how to improve the current
> DoH standard and propose to thoroughly cover the aforementioned
> issues in the section "Security Considerations". We also suggest a
> countermeasure called "Explicit Content Type Negotiation (ECTN)" to
> become part of the standard. ECTN prevents confusion of HTTP(S) and
> DoH communication contexts at the server level and stops any
> attacker controlled content from being served in context of plain
> HTTP(S) communication. A detailed description of ECTN and secondary
> countermeasures is also part of the attached document.
>
> We are looking forward to work with you towards improving RFC 8484
> "DNS-Queries over HTTPS" and make the protocol more secure for service
> operators and its users. We are happy to help clarifying in further
> or upcoming questions.
>
> Kind regards
>
> Matthias Gierlings
>
> [1] https://datatracker.ietf.org/doc/html/rfc8484
> _______________________________________________
> DNSOP mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
>
_______________________________________________
DNSOP mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to