Re: [http-state] HTTP cookie processing wrt "public suffixes"

[Adam Barth <[e-mail]>]

On Sun, Nov 9, 2014 at 3:23 PM, =JeffH <[e-mail]> wrote:
> This is a (draft) community-service posting: The purpose is to unambiguously
> state the specification of "cookie processing wrt public suffixes".
>
> It is somewhat difficult to tease this out of the requisite specification(s)
> and associated documents, e.g., [RFC6265] and the Public Suffix List [PSL],
> and so here it is (corrections/comments welcome).
>
> =JeffH
> ------
> HTTP cookie processing wrt "public suffixes" aka "effective TLDs"
>
> Contents:
>   1. Terminology
>   2. Summarized Cookie-setting and -returning Algorithms wrt eTLDs
>   3. Detailed Cookie-setting and -returning Algorithms wrt eTLDs
>   4. Excerpts from [RFC6265]
>   6. References
>
>
> 1. Terminology:
>
> Cookie - a unit of HTTP state management data, sent by the server to the UA
> as part of (some) HTTP responses. The UA returns cookie(s) to the server on
> all subsequent requests, modulo various rules (some of which are discussed
> herein, the full story is given by [RFC6265]).
>
> effective TLD (eTLD) - a term closely related to "public suffix". It is more
> general than the term "public suffix" -- i.e., because it implies that the
> domain in question ought to be treated as effectively a TLD, but it may
> actually not be an official TLD, nor might it allow for "public" subdomain
> registrations (as the term "public suffix" specifically implies).
>
> Origin - {scheme, host, port} tuple of a webapp [RFC6454].
>
> Public Suffix - A "public suffix" is a DNS domain [RFC1034] under which
> Internet users can directly register names [modulo some policies]. Some
> examples of public suffixes are .com, .co.uk and pvt.k12.ma.us [PSL].
> [However, this definition does not adequately address various subtleties in
> practice, hence, in part, the [DBOUND] effort. Note that the term "effective
> TLD (eTLD)" is closely related, and that the domains listed in [PSL] are a
> superset of TLDs.]
>
> subdomain - a child domain of some given domain. E.g., foo.example.org is a
> subdomain of example.org.
>
> superdomain - the parent domain of some given domain. E.g., example.org is
> the superdomain of foo.example.org.
>
> TLD / Top-level Domain - TLDs are those single-label DNS domains directly
> under the DNS root [RFC1591]. See [IANA-TLDs] for the current list of TLDs.

There's another layer of subtly here surrounding single-label DNS
names that are not in [IANA-TLDs].  For example, inside many corporate
networks, "www" is a single-label DNS name that resolves to a web
server (and many times is interested in setting cookies).  It's
unclear whether you mean to include such things in this definition of
TLD.

> UA - a (web) User Agent, e.g. a web browser.
>
> webapp - a web application, which has both a server-side and client-side
> instances. The client-side instances, which are emitted by the server-side
> over HTTP, are effectively mobile code (typically composed of
> HTML+JavaScript+CSS+etc), and execute within UAs (as "web pages"), and have
> an origin derived from the server-side's URI.
>
>
>
> 2. Summarized Cookie-setting and -returning Algorithms wrt eTLDs:
>
> (a) A server-side webapp, whose origin's host component [RFC6454] (aka
> domain name) IS NOT a "public suffix"/eTLD, can "set cookies" (on UAs) for
> its own domain name, or for superdomains -- unless the targeted superdomain
> is a eTLD. In the latter case, the set-cookie attempt is ignored.
> (b) The UA will return cookies set for a given host (aka domain) to only
> that host, or, to that host and all subdomains thereof (depending on
> server-specified particulars of the cookie itself).
>
> (c) Conversely, a server-side webapp whose origin's host component IS
> denoted as a "public suffix"/eTLD (i.e., it appears in [PSL] or similar
> compendiums) can set cookies for itself (modulo UA configuration), but its
> subdomains can not set cookies for the eTLD domain.
> (d) The UA will not return any of the eTLD's cookies to subdomain  servers.
> Effectively, this depends upon UA configuration -- i.e., a UA might be
> configured to not accept cookies from an eTLD server at all.

Are there popular UAs that are configured that way?  Presumably they'd
break in corporate networks that use DNS names like "www"...

> 3. Detailed Cookie-setting and -returning Algorithms wrt eTLDs:
>
> Excerpts from [RFC6265] expressing the specifics underlying the
> cookie-setting and -returning algorithms, and in which the effects of
> "public suffixes"/eTLDs are specified, are found in the section after this
> one. They are referred to (using an "S" prefix) in the following
> explanations of the algorithms summarized in the preceding section:
>
> In terms of (a), a server that is NOT a eTLD doing cookie-setting: the 2nd
> portion of S 4.1.2.3 describes the cookie-setting semantics, step 5 of S 5.3
> handles cookie-setting attempts for "public suffixes", step 6 of S 5.3
> defines the determination of the cookie's applicable domain and its notation
> in the cookie's cookie store entry, e.g. setting the host-only-flag. The
> host-only-flag governs whether the cookie will be returned to the origin
> server only, or whether it will be returned to the origin server as well as
> its subdomains.
>
> In terms of (b), a UA returning cookies to non-eTLD servers: S 5.4 governs
> the construction of the Cookie header for the HTTP request, which depends
> directly upon the cookie entry's host-only-flag, and the domain-match
> algorithm (S 5.1.3).
>
>
> In terms of (c), a "public suffix"/eTLD webapp doing cookie-setting: the 2nd
> portion of S 4.1.2.3 describes again applies, and refers directly to S 5.3
> in this case. S 5.3 steps 5 and 6 apply and the path through them depends on
> whether the UA is config'd to reject "public suffixes" or not. If not, and
> the domain-attribute is empty, then in step 6 the cookie is stored in the
> cookie store with its host-only-flag set to true and its domain set to the
> canonicalized request-host.
>
> In terms of (d), a UA returning cookies to a "public suffix"/eTLD server:
> if the UA does accept cookies set by eTLD servers, then step 1 of S 5.4 will
> gather only such cookies return to ONLY the eTLD server, due to the manner
> that they were set in the cookie store.
>
>
> 4. Excerpts from [RFC6265]:
>
>   [Note: I've included the hierarchy of [RFC6265] section titles in
>    order to provide context.]
>
>   4. Server Requirements
>   4.1. Set-Cookie
>   4.1.2. Semantics (Non-Normative)
>   4.1.2.3. The Domain Attribute
>   https://tools.ietf.org/html/rfc6265#section-4.1.2.3
>
>    The Domain attribute specifies those hosts to which the cookie will
>    be sent.  For example, if the value of the Domain attribute is
>    "example.com", the user agent will include the cookie in the Cookie
>    header when making HTTP requests to example.com, www.example.com, and
>    www.corp.example.com. ... If the server omits the Domain attribute,
>    the user agent will return the cookie only to the origin server.
>    ...
>    The user agent will reject cookies unless the Domain attribute
>    specifies a scope for the cookie that would include the origin
>    server.  For example, the user agent will accept a cookie with a
>    Domain attribute of "example.com" or of "foo.example.com" from
>    foo.example.com, but the user agent will not accept a cookie with a
>    Domain attribute of "bar.example.com" or of "baz.foo.example.com".
>
>    NOTE: For security reasons, many user agents are configured to reject
>    Domain attributes that correspond to "public suffixes".  For example,
>    some user agents will reject Domain attributes of "com" or "co.uk".
>    (See Section 5.3 for more information.)
>    ...
>
>   5. User Agent Requirements
>   5.1. Subcomponent Algorithms
>   5.1.3. Domain Matching
>   https://tools.ietf.org/html/rfc6265#section-5.1.3
>
>    A string domain-matches a given domain string if at least one of the
>    following conditions hold:
>
>    o  The domain string and the string are identical.  (Note that both
>       the domain string and the string will have been canonicalized to
>       lower case at this point.)
>
>    o  All of the following conditions hold:
>
>       *  The domain string is a suffix of the string.
>
>       *  The last character of the string that is not included in the
>          domain string is a %x2E (".") character.
>
>       *  The string is a host name (i.e., not an IP address).
>
>
>   5.3. Storage Model
>   https://tools.ietf.org/html/rfc6265#section-5.3
>   ...
>    5.   If the user agent is configured to reject "public suffixes" and
>         the domain-attribute is a public suffix:
>
>            If the domain-attribute is identical to the canonicalized
>            request-host:
>
>               Let the domain-attribute be the empty string.
>
>            Otherwise:
>
>               Ignore the cookie entirely and abort these steps.
>
>            NOTE: A "public suffix" is a domain that is controlled by a
>            public registry, such as "com", "co.uk", and "pvt.k12.wy.us".
>            This step is essential for preventing attacker.com from
>            disrupting the integrity of example.com by setting a cookie
>            with a Domain attribute of "com".  Unfortunately, the set of
>            public suffixes (also known as "registry controlled domains")
>            changes over time.  If feasible, user agents SHOULD use an
>            up-to-date public suffix list, such as the one maintained by
>            the Mozilla project at <http://publicsuffix.org/>.
>
>    6.   If the domain-attribute is non-empty:
>
>            If the canonicalized request-host does not domain-match the
>            domain-attribute:
>
>               Ignore the cookie entirely and abort these steps.
>
>            Otherwise:
>
>               Set the cookie's host-only-flag to false.
>
>               Set the cookie's domain to the domain-attribute.
>
>         Otherwise:
>
>            Set the cookie's host-only-flag to true.
>
>            Set the cookie's domain to the canonicalized request-host.
>    ...
>
>    5.4. The Cookie Header
>    https://tools.ietf.org/html/rfc6265#section-5.4
>
>    The user agent includes stored cookies in the Cookie HTTP request
>    header.
>    ...
>    If the user agent does attach a Cookie header field to an HTTP
>    request, the user agent MUST send the cookie-string (defined below)
>    as the value of the header field.
>
>    The user agent MUST use an algorithm equivalent to the following
>    algorithm to compute the "cookie-string" from a cookie store and a
>    request-uri:
>
>    1.  Let cookie-list be the set of cookies from the cookie store that
>        meets all of the following requirements:
>
>        *  Either:
>
>              The cookie's host-only-flag is true and the canonicalized
>              request-host is identical to the cookie's domain.
>
>           Or:
>
>              The cookie's host-only-flag is false and the canonicalized
>              request-host domain-matches the cookie's domain.
>
>        *  The request-uri's path path-matches the cookie's path.
>
>    ...
>
>        *  If the cookie's http-only-flag is true, then exclude the
>           cookie if the cookie-string is being generated for a "non-
>           HTTP" API (as defined by the user agent).
>    ...
>
>
> 6. References:
>
> [DBOUND] https://www.ietf.org/mailman/listinfo/dbound
>
> [IANA-TLDs] https://www.iana.org/domains/root/db
>
> [PSL] Public Suffix List
>       https://publicsuffix.org/
>
> [RFC1034] DOMAIN NAMES - CONCEPTS AND FACILITIES
>           https://tools.ietf.org/html/rfc1034
>
> [RFC1591] Domain Name System Structure and Delegation
>           http://tools.ietf.org/html/rfc1591
>           see also: https://en.wikipedia.org/wiki/Top-level_domain
>
> [RFC6265] HTTP State Management Mechanism
>           https://tools.ietf.org/html/rfc6265
>
> [RFC6454] The Web Origin Concept
>           https://tools.ietf.org/html/rfc6454
>
>
>
> end
>
> _______________________________________________
> http-state mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/http-state