Independent Submission E. Fokschaner Request for Comments: 8565 1 April 2019 Category: Informational ISSN: 2070-1721 Hypertext Jeopardy Protocol (HTJP/1.0) Abstract The Hypertext Jeopardy Protocol (HTJP) inverts the request/response semantics of the Hypertext Transfer Protocol (HTTP). Using conventional HTTP, one connects to a server, asks a question, and expects a correct answer. Using HTJP, one connects to a server, sends an answer, and expects a correct question. This document specifies the semantics of HTJP. Status of This Memo This document is not an Internet Standards Track specification; it is published for informational purposes. This is a contribution to the RFC Series, independently of any other RFC stream. The RFC Editor has chosen to publish this document at its discretion and makes no statement about its value for implementation or deployment. Documents approved for publication by the RFC Editor are not candidates for any level of Internet Standard; see Section 2 of RFC 7841. Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at https://www.rfc-editor.org/info/rfc8565. Copyright Notice Copyright (c) 2019 IETF Trust and the persons identified as the document authors. All rights reserved. This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document.
Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 2. Conventions Used in This Document . . . . . . . . . . . . . . 3 3. Comparison with HTTP . . . . . . . . . . . . . . . . . . . . 3 4. Response and Request Semantics . . . . . . . . . . . . . . . 4 4.1. Applicability of Postel's Robustness Principle . . . . . 4 4.2. Identifying the Server Associated with an HTJP Response . 5 4.3. Temporal Considerations . . . . . . . . . . . . . . . . . 5 4.4. Pseudo-Valid HTJP Messages . . . . . . . . . . . . . . . 6 4.5. HTTP Responses That Are Not Requestable . . . . . . . . . 6 5. Caches and Proxies . . . . . . . . . . . . . . . . . . . . . 7 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 7 7. Security Considerations . . . . . . . . . . . . . . . . . . . 7 7.1. Securing HTTP against HTJP . . . . . . . . . . . . . . . 7 7.1.1. Anti-HTJP-Nonce Header . . . . . . . . . . . . . . . 8 7.2. HTJPS . . . . . . . . . . . . . . . . . . . . . . . . . . 8 8. References . . . . . . . . . . . . . . . . . . . . . . . . . 9 8.1. Normative References . . . . . . . . . . . . . . . . . . 9 8.2. Informative References . . . . . . . . . . . . . . . . . 10 Appendix A. Hypertext Double Jeopardy Protocol . . . . . . . . . 11 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . 11 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . 11 1. Introduction The Hypertext Jeopardy Protocol (HTJP) 1.0 is a stateless application-level response/request protocol that functions as the semantic inverse of the Hypertext Transfer Protocol (HTTP) 1.1 . It can roughly be specified in relation to HTTP by the following rules: o Where an HTTP client would send an HTTP request message, an HTJP client would send an HTTP response message. o Where an HTTP server would send an HTTP response message, an HTJP server would send an HTTP request message. o The HTTP request sent as an HTJP response should be an HTTP request that (if sent to the appropriate HTTP server) would elicit the HTTP response sent in the HTJP request. HTJP is compatible with the HTTP/1.1 specification, at least in spirit, if not in letter.
HTJP has novel applications in all the following areas: o Generative automated testing of HTTP implementations and HTTP- based applications. o Monitoring of HTTP-based applications in production. o Forensic and diagnostic reconstruction of HTTP requests from HTTP response logs. o Discovery of first-party and third-party security vulnerabilities. 2. Conventions Used in This Document The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here. 3. Comparison with HTTP It is a little-known fact that HTTP/1.1 already defines itself to be its own inverse mode of operation. Section 3.1 of RFC 7230 [RFC7230], which describes the start line of the HTTP message format, states: In theory, a client could receive requests and a server could receive responses, distinguishing them by their different start- line formats, but, in practice, servers are implemented to only expect a request [...] and clients are implemented to only expect a response. It is only convention that HTTP clients send HTTP requests and that HTTP servers send HTTP responses. Therefore, HTJP is just HTTP with some alternative conventions. It is not a distinct protocol. Furthermore, since all messages in HTJP are indistinguishable from HTTP messages, an HTJP peer would have no way of identifying itself explicitly as using HTJP rather than HTTP. Therefore, we describe HTJP as a "pseudo-protocol" in order to distinguish clients, servers, and conversations that are using the HTTP conventions laid out in this document from those that use conventions that are more commonly associated with HTTP.
4. Response and Request Semantics An HTJP request MUST be an HTTP response message. An HTJP response message MUST be an HTTP request message that, if issued to the appropriate HTTP server, would elicit the HTTP response specified by the HTJP request being replied to. As described in Section 3, HTJP is unconventional but valid HTTP, and so the entirety of the HTTP specification (as defined in [RFC7230], [RFC7231], [RFC7232], [RFC7233], [RFC7234], and [RFC7235]) MUST be respected when doing so is consistent with HTJP's unique semantics. The following example illustrates a typical message exchange for an HTJP request concerning the same hypothetical server from Section 2.1 of RFC 7230 [RFC7230]. Client request: HTTP/1.1 200 OK Date: Mon, 27 Jul 2009 12:28:53 GMT Server: Apache Last-Modified: Wed, 22 Jul 2009 19:15:56 GMT ETag: "34aa387-d-1568eb00" Accept-Ranges: bytes Content-Length: 51 Vary: Accept-Encoding Content-Type: text/plain Hello World! My payload includes a trailing CRLF. Server response: GET /hello.txt HTTP/1.1 Host: www.example.com 4.1. Applicability of Postel's Robustness Principle Implementations of HTJP SHOULD respect Postel's Robustness Principle [IAB-PROTOCOL-MAINTENANCE]. Applied to HTJP, Postel's Robustness Principle implies that, given the choice of multiple valid HTJP responses for an HTJP request, one SHOULD prefer a response that is more adherent to the HTTP standard or uses fewer HTTP features. For example, sometimes a User-Agent header has no bearing on the HTTP response from an HTTP server. On seeing such a response in an HTJP request, an HTJP server could validly respond with a practically unlimited number of permutations on the User-Agent header value. However, it SHOULD prefer to respond
with an HTTP request that has no User-Agent header whatsoever, in keeping with Postel's Robustness Principle. The same consideration applies when encountering an HTJP request for which there are both valid and "pseudo-valid" (Section 4.4) HTJP responses available. 4.2. Identifying the Server Associated with an HTJP Response It may be of interest to a user of HTJP to try issuing an HTJP response as an HTTP request to the appropriate server. This brings up the issue of correctly identifying the host to which the HTJP response should be sent. Much of the time this will be able to be determined from the Host header field of the HTJP response. The Host header is required by conformant HTTP/1.1 requests. In the case that the Host header is not present (for example, if the HTJP response is an HTTP/1.0 request rather than HTTP/1.1), an HTJP response MAY use the absolute URI form in the HTTP request line, to add clarity about the target host if it would be validly accepted by the server. This specific example is complicated by the fact that prior to HTTP/1.1 it was not required that implementations accept the absolute URI form. For this reason, it is also possible to phrase the HTJP response as an HTTP request to a Forward Proxy server, which would have accepted, indeed needed, the absolute URI request line prior to and after HTTP/1.1. As a last resort, it may be possible to heuristically derive the target host of an HTJP response from the HTJP request; for example, the HTJP request may have absolute references to other HTTP resources that seem to come from the same host. 4.3. Temporal Considerations When an HTJP response is issued, there is no guarantee that, by the time the response is received by an HTJP client, the HTTP server that is associated with said response will still reply with it. Providing any guarantee about "when" an HTTP server would reply with said response is obviously a theoretically unsolvable problem and therefore outside the scope of this HTJP specification. It is only required that the HTJP response be correct at some point in the range of the 32-bit Unix Timestamp; see "Seconds Since the Epoch" (Section 4.16) of Unix General Concepts [UNIX-General-Concepts]. HTJP servers that are responding with an HTTP request for a volatile resource, and with high confidence in the time range at which the resource would be in the state described by the HTJP request, MAY add a Date header to the HTJP response. This is in conformance with the ability for HTTP requests to carry a Date header; see Section 126.96.36.199 of [RFC7231].
HTJP clients can try to demand more temporal certainty by adding a Date header to their HTTP response, embedding criteria for the time of the HTTP response in the HTTP response itself. Of course, the client might still only receive that exact HTTP response if it manages to deliver the HTTP request at the precise time of the previously requested Date header, and even then it is still not guaranteed due to HTTP caching et cetera. 4.4. Pseudo-Valid HTJP Messages In the wild, HTTP clients and servers have been known to occasionally exchange HTTP messages that are not conformant to any HTTP specification. For this reason, we will identify a class of messages that are, on the one hand, invalid HTTP messages, yet at the same time, correctly answerable HTJP requests or correct answers to an HTJP request, as "pseudo-valid" HTJP messages. Consider, for example, an HTTP server that erroneously reports a Content-Length header field of zero when sending an HTTP payload of non-zero length. Despite this HTTP message violating the HTTP specification, it is possible for an HTJP server to receive such a message and correctly respond to it, satisfying the HTJP semantics in doing so. This applies to both HTJP requests and HTJP responses. There may be times when the only valid HTJP response is an invalid HTTP request. When there are both valid and invalid HTTP requests that could satisfy the HTJP request, Postel's Robustness Principle SHOULD be applied, as described in Section 4.1. 4.5. HTTP Responses That Are Not Requestable Given that an HTJP response MUST be an HTTP request, and that HTTP requests do not have a status field (such as a status code), there is no way for an HTJP response to signal a failure in response to an HTJP request, via a status code or otherwise. The correct HTJP response to an HTJP request when a server cannot determine an HTTP request that elicits the HTTP response is to not respond at all. The HTJP responder MAY close the connection; however, the HTJP requester MUST NOT interpret the closing of the connection as a response. This can have issues when HTJP servers are hosted behind non-HTJP-aware HTTP proxies, as the proxy may inject a 5xx HTTP response, which could be misinterpreted as an HTJP response. See more on proxies in Section 5.
5. Caches and Proxies Despite being valid HTTP traffic, support for caching and proxying of HTJP traffic is unfortunately not widespread. It is estimated that currently approximately 0.0% of all HTTP-aware intermediaries gracefully handle HTJP traffic. For this reason, it is currently RECOMMENDED that HTJP should not be deployed behind an HTTP proxying or caching layer. Support will likely increase proportionally to HTJP's popularity in production. 6. IANA Considerations In order that implementations of this specification can operate correctly, and in accordance with [BCP90], the following new HTTP header field would need to be registered in the "Message Headers" registry maintained at <https://www.iana.org/assignments/message-headers/>. To find out the details of this registration, an application should send an HTJP request to a suitable server specifying the Anti-HTJP-Nonce value that the server would have supplied had the IANA registration been made. +-------------------+----------+---------------+---------------+ | Header Field Name | Protocol | Status | Reference | +-------------------+----------+---------------+---------------+ | Anti-HTJP-Nonce | http | informational | Section 7.1.1 | +-------------------+----------+---------------+---------------+ 7. Security Considerations 7.1. Securing HTTP against HTJP An incomplete implementation of HTJP is inadvisable from a security perspective. A complete implementation of HTJP may have interesting security features that are worthy of detailed examination. Due to its semantics, the issuing of a successfully authorized HTTP response to an HTJP server will result in a reply containing the HTTP request that elicits said response, including any credentials, tokens, cookies, or other authorization materials that were required to elicit that response.
As an example: Client request: HTTP/1.1 200 OK Date: Mon, 27 Jul 2009 12:28:53 GMT Content-Length: 61 Content-Type: text/plain Some predictable information accessed using authorization. Server response: (line breaks in the Authorization header are for RFC formatting) GET /information.txt HTTP/1.1 Host: authorised-usage-service.example.com Authorization: Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9. eyJzdWIiOiJodGpwIiwibmFtZSI6IkV2ZXJ5b25lIiwiaWF0IjowfQ. JOL-kIObgTI0MzFfm1yVFFkIo1xf7DZGjY_oedRBZW0 Given that we cannot prevent anyone from attempting to implement HTJP, it is RECOMMENDED to consider how HTJP impacts security when using HTTP. Note that it was only possible to query for the credentialed HTTP request because the response to the authorized request was predictable. HTTP servers could mitigate this vulnerability exposed by HTJP by only serving a response that is at least as secret as the credentials themselves in response to an authorized request. 7.1.1. Anti-HTJP-Nonce Header A generic solution to this problem is to use an "Anti-HTJP-Nonce" HTTP header in HTTP responses. The value of an "Anti-HTJP-Nonce" header SHOULD be a cryptographically secure random number in any encoding that is valid for an HTTP header value. The length of this number SHOULD be determined by the producer of the HTTP response, accounting for their method of random number generation and their threat model. 7.2. HTJPS HTJP, being just HTTP, has most of the same security concerns and features as HTTP itself. For example, the use of HTJP over an encrypted connection, such as TLS 1.3 [RFC8446], similar to HTTP Secure (HTTPS), is referred to as HTJP Secure (HTJPS). However, it is important to note that, unlike with HTTPS, it is not expected that the hostname you are securely communicating with is the same hostname
as featured in the Host headers or absolute URIs of the HTJP messages themselves, as HTJP messages are typically referring to other HTTP hosts. This excludes the case of a server that supports both conventional HTTP and HTJP, where it is possible to make HTJP requests securely to the same host that is also the subject of the HTJP requests being made. 8. References 8.1. Normative References [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, DOI 10.17487/RFC2119, March 1997, <https://www.rfc-editor.org/info/rfc2119>. [RFC7230] Fielding, R., Ed. and J. Reschke, Ed., "Hypertext Transfer Protocol (HTTP/1.1): Message Syntax and Routing", RFC 7230, DOI 10.17487/RFC7230, June 2014, <https://www.rfc-editor.org/info/rfc7230>. [RFC7231] Fielding, R., Ed. and J. Reschke, Ed., "Hypertext Transfer Protocol (HTTP/1.1): Semantics and Content", RFC 7231, DOI 10.17487/RFC7231, June 2014, <https://www.rfc-editor.org/info/rfc7231>. [RFC7232] Fielding, R., Ed. and J. Reschke, Ed., "Hypertext Transfer Protocol (HTTP/1.1): Conditional Requests", RFC 7232, DOI 10.17487/RFC7232, June 2014, <https://www.rfc-editor.org/info/rfc7232>. [RFC7233] Fielding, R., Ed., Lafon, Y., Ed., and J. Reschke, Ed., "Hypertext Transfer Protocol (HTTP/1.1): Range Requests", RFC 7233, DOI 10.17487/RFC7233, June 2014, <https://www.rfc-editor.org/info/rfc7233>. [RFC7234] Fielding, R., Ed., Nottingham, M., Ed., and J. Reschke, Ed., "Hypertext Transfer Protocol (HTTP/1.1): Caching", RFC 7234, DOI 10.17487/RFC7234, June 2014, <https://www.rfc-editor.org/info/rfc7234>. [RFC7235] Fielding, R., Ed. and J. Reschke, Ed., "Hypertext Transfer Protocol (HTTP/1.1): Authentication", RFC 7235, DOI 10.17487/RFC7235, June 2014, <https://www.rfc-editor.org/info/rfc7235>.
[RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, May 2017, <https://www.rfc-editor.org/info/rfc8174>. [UNIX-General-Concepts] "General Concepts", Chapter 4 of "The Open Group Base Specifications, Issue 7", 2018 edition, IEEE Std 1003.1-2017, 2018, <http://pubs.opengroup.org/ onlinepubs/9699919799/basedefs/V1_chap04.html>. 8.2. Informative References [BCP90] Klyne, G., Nottingham, M., and J. Mogul, "Registration Procedures for Message Header Fields", BCP 90, RFC 3864, September 2004, <https://www.rfc-editor.org/info/bcp90>. [IAB-PROTOCOL-MAINTENANCE] Thomson, M., "The Harmful Consequences of the Robustness Principle", Work in Progress, draft-iab-protocol- maintenance-02, March 2019. [RFC8446] Rescorla, E., "The Transport Layer Security (TLS) Protocol Version 1.3", RFC 8446, DOI 10.17487/RFC8446, August 2018, <https://www.rfc-editor.org/info/rfc8446>.
Appendix A. Hypertext Double Jeopardy Protocol Also worth mentioning, in case one encounters it in the wild, is the Hypertext Double Jeopardy Protocol (HTJ2P). The Hypertext Double Jeopardy Protocol 1.0 is a stateless application-level request/ response protocol that functions as the inverse of the Hypertext Jeopardy Protocol (HTJP) 1.0 . An HTJ2P response MUST be an HTTP response which would be issued for an HTTP request delivered as the HTJ2P request. Implementations of HTJ2P have groundbreaking potential in the fields of HTTP caching, and in the implementation of HTJP. Acknowledgements The author thanks Alex Trebek for his distinguished contributions to culture and society. The author thanks Peter Phillips for the suggestion of the Anti-HTJP-Nonce header. The author also wishes to thank anyone who has ever built a tool or a technology that made people ask "Why?". Author's Address Edmund Fokschaner Email: email@example.com