RFC 6265






Internet Engineering Task Force (IETF)                          A. Barth
Request for Comments: 6265                                 U.C. Berkeley
Obsoletes: 2965                                               April 2011
Category: Standards Track
ISSN: 2070-1721


                    HTTP State Management Mechanism

Abstract



   This document defines the HTTP Cookie and Set-Cookie header fields.
   These header fields can be used by HTTP servers to store state
   (called cookies) at HTTP user agents, letting the servers maintain a
   stateful session over the mostly stateless HTTP protocol.  Although
   cookies have many historical infelicities that degrade their security
   and privacy, the Cookie and Set-Cookie header fields are widely used
   on the Internet.  This document obsoletes RFC 2965.

Status of This Memo



   This is an Internet Standards Track document.

   This document is a product of the Internet Engineering Task Force
   (IETF).  It represents the consensus of the IETF community.  It has
   received public review and has been approved for publication by the
   Internet Engineering Steering Group (IESG).  Further information on
   Internet Standards is available in Section 2 of RFC 5741.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at
   http://www.rfc-editor.org/info/rfc6265.

Copyright Notice



   Copyright (c) 2011 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
   (http://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.  Code Components extracted from this document must
   include Simplified BSD License text as described in Section 4.e of
   the Trust Legal Provisions and are provided without warranty as
   described in the Simplified BSD License.




Barth                        Standards Track                    [Page 1]

RFC 6265             HTTP State Management Mechanism          April 2011


   This document may contain material from IETF Documents or IETF
   Contributions published or made publicly available before November
   10, 2008.  The person(s) controlling the copyright in some of this
   material may not have granted the IETF Trust the right to allow
   modifications of such material outside the IETF Standards Process.
   Without obtaining an adequate license from the person(s) controlling
   the copyright in such materials, this document may not be modified
   outside the IETF Standards Process, and derivative works of it may
   not be created outside the IETF Standards Process, except to format
   it for publication as an RFC or to translate it into languages other
   than English.

Table of Contents



   1. Introduction ....................................................3
   2. Conventions .....................................................4
      2.1. Conformance Criteria .......................................4
      2.2. Syntax Notation ............................................5
      2.3. Terminology ................................................5
   3. Overview ........................................................6
      3.1. Examples ...................................................6
   4. Server Requirements .............................................8
      4.1. Set-Cookie .................................................8
           4.1.1. Syntax ..............................................8
           4.1.2. Semantics (Non-Normative) ..........................10
      4.2. Cookie ....................................................13
           4.2.1. Syntax .............................................13
           4.2.2. Semantics ..........................................13
   5. User Agent Requirements ........................................14
      5.1. Subcomponent Algorithms ...................................14
           5.1.1. Dates ..............................................14
           5.1.2. Canonicalized Host Names ...........................16
           5.1.3. Domain Matching ....................................16
           5.1.4. Paths and Path-Match ...............................16
      5.2. The Set-Cookie Header .....................................17
           5.2.1. The Expires Attribute ..............................19
           5.2.2. The Max-Age Attribute ..............................20
           5.2.3. The Domain Attribute ...............................20
           5.2.4. The Path Attribute .................................21
           5.2.5. The Secure Attribute ...............................21
           5.2.6. The HttpOnly Attribute .............................21
      5.3. Storage Model .............................................21
      5.4. The Cookie Header .........................................25
   6. Implementation Considerations ..................................27
      6.1. Limits ....................................................27
      6.2. Application Programming Interfaces ........................27
      6.3. IDNA Dependency and Migration .............................27
   7. Privacy Considerations .........................................28



Barth                        Standards Track                    [Page 2]

RFC 6265             HTTP State Management Mechanism          April 2011


      7.1. Third-Party Cookies .......................................28
      7.2. User Controls .............................................28
      7.3. Expiration Dates ..........................................29
   8. Security Considerations ........................................29
      8.1. Overview ..................................................29
      8.2. Ambient Authority .........................................30
      8.3. Clear Text ................................................30
      8.4. Session Identifiers .......................................31
      8.5. Weak Confidentiality ......................................32
      8.6. Weak Integrity ............................................32
      8.7. Reliance on DNS ...........................................33
   9. IANA Considerations ............................................33
      9.1. Cookie ....................................................34
      9.2. Set-Cookie ................................................34
      9.3. Cookie2 ...................................................34
      9.4. Set-Cookie2 ...............................................34
   10. References ....................................................35
      10.1. Normative References .....................................35
      10.2. Informative References ...................................35
   Appendix A. Acknowledgements ......................................37

1.  Introduction



   This document defines the HTTP Cookie and Set-Cookie header fields.
   Using the Set-Cookie header field, an HTTP server can pass name/value
   pairs and associated metadata (called cookies) to a user agent.  When
   the user agent makes subsequent requests to the server, the user
   agent uses the metadata and other information to determine whether to
   return the name/value pairs in the Cookie header.

   Although simple on their surface, cookies have a number of
   complexities.  For example, the server indicates a scope for each
   cookie when sending it to the user agent.  The scope indicates the
   maximum amount of time in which the user agent should return the
   cookie, the servers to which the user agent should return the cookie,
   and the URI schemes for which the cookie is applicable.

   For historical reasons, cookies contain a number of security and
   privacy infelicities.  For example, a server can indicate that a
   given cookie is intended for "secure" connections, but the Secure
   attribute does not provide integrity in the presence of an active
   network attacker.  Similarly, cookies for a given host are shared
   across all the ports on that host, even though the usual "same-origin
   policy" used by web browsers isolates content retrieved via different
   ports.

   There are two audiences for this specification: developers of cookie-
   generating servers and developers of cookie-consuming user agents.



Barth                        Standards Track                    [Page 3]

RFC 6265             HTTP State Management Mechanism          April 2011


   To maximize interoperability with user agents, servers SHOULD limit
   themselves to the well-behaved profile defined in Section 4 when
   generating cookies.

   User agents MUST implement the more liberal processing rules defined
   in Section 5, in order to maximize interoperability with existing
   servers that do not conform to the well-behaved profile defined in
   Section 4.

   This document specifies the syntax and semantics of these headers as
   they are actually used on the Internet.  In particular, this document
   does not create new syntax or semantics beyond those in use today.
   The recommendations for cookie generation provided in Section 4
   represent a preferred subset of current server behavior, and even the
   more liberal cookie processing algorithm provided in Section 5 does
   not recommend all of the syntactic and semantic variations in use
   today.  Where some existing software differs from the recommended
   protocol in significant ways, the document contains a note explaining
   the difference.

   Prior to this document, there were at least three descriptions of
   cookies: the so-called "Netscape cookie specification" [Netscape],
   RFC 2109 [RFC2109], and RFC 2965 [RFC2965].  However, none of these
   documents describe how the Cookie and Set-Cookie headers are actually
   used on the Internet (see [Kri2001] for historical context).  In
   relation to previous IETF specifications of HTTP state management
   mechanisms, this document requests the following actions:

   1.  Change the status of [RFC2109] to Historic (it has already been
       obsoleted by [RFC2965]).

   2.  Change the status of [RFC2965] to Historic.



   3.  Indicate that [RFC2965] has been obsoleted by this document.



   In particular, in moving RFC 2965 to Historic and obsoleting it, this
   document deprecates the use of the Cookie2 and Set-Cookie2 header
   fields.

2.  Conventions

2.1.  Conformance Criteria



   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
   document are to be interpreted as described in [RFC2119].





Barth                        Standards Track                    [Page 4]

RFC 6265             HTTP State Management Mechanism          April 2011


   Requirements phrased in the imperative as part of algorithms (such as
   "strip any leading space characters" or "return false and abort these
   steps") are to be interpreted with the meaning of the key word
   ("MUST", "SHOULD", "MAY", etc.) used in introducing the algorithm.

   Conformance requirements phrased as algorithms or specific steps can
   be implemented in any manner, so long as the end result is
   equivalent.  In particular, the algorithms defined in this
   specification are intended to be easy to understand and are not
   intended to be performant.

2.2.  Syntax Notation



   This specification uses the Augmented Backus-Naur Form (ABNF)
   notation of [RFC5234].

   The following core rules are included by reference, as defined in
   [RFC5234], Appendix B.1: ALPHA (letters), CR (carriage return), CRLF
   (CR LF), CTLs (controls), DIGIT (decimal 0-9), DQUOTE (double quote),
   HEXDIG (hexadecimal 0-9/A-F/a-f), LF (line feed), NUL (null octet),
   OCTET (any 8-bit sequence of data except NUL), SP (space), HTAB
   (horizontal tab), CHAR (any [USASCII] character), VCHAR (any visible
   [USASCII] character), and WSP (whitespace).

   The OWS (optional whitespace) rule is used where zero or more linear
   whitespace characters MAY appear:

   OWS            = *( [ obs-fold ] WSP )
                    ; "optional" whitespace
   obs-fold       = CRLF

   OWS SHOULD either not be produced or be produced as a single SP
   character.

2.3.  Terminology



   The terms user agent, client, server, proxy, and origin server have
   the same meaning as in the HTTP/1.1 specification ([RFC2616], Section
   1.3).

   The request-host is the name of the host, as known by the user agent,
   to which the user agent is sending an HTTP request or from which it
   is receiving an HTTP response (i.e., the name of the host to which it
   sent the corresponding HTTP request).

   The term request-uri is defined in Section 5.1.2 of [RFC2616].





Barth                        Standards Track                    [Page 5]

RFC 6265             HTTP State Management Mechanism          April 2011


   Two sequences of octets are said to case-insensitively match each
   other if and only if they are equivalent under the i;ascii-casemap
   collation defined in [RFC4790].

   The term string means a sequence of non-NUL octets.

3.  Overview

   This section outlines a way for an origin server to send state
   information to a user agent and for the user agent to return the
   state information to the origin server.

   To store state, the origin server includes a Set-Cookie header in an
   HTTP response.  In subsequent requests, the user agent returns a
   Cookie request header to the origin server.  The Cookie header
   contains cookies the user agent received in previous Set-Cookie
   headers.  The origin server is free to ignore the Cookie header or
   use its contents for an application-defined purpose.

   Origin servers MAY send a Set-Cookie response header with any
   response.  User agents MAY ignore Set-Cookie headers contained in
   responses with 100-level status codes but MUST process Set-Cookie
   headers contained in other responses (including responses with 400-
   and 500-level status codes).  An origin server can include multiple
   Set-Cookie header fields in a single response.  The presence of a
   Cookie or a Set-Cookie header field does not preclude HTTP caches
   from storing and reusing a response.

   Origin servers SHOULD NOT fold multiple Set-Cookie header fields into
   a single header field.  The usual mechanism for folding HTTP headers
   fields (i.e., as defined in [RFC2616]) might change the semantics of
   the Set-Cookie header field because the %x2C (",") character is used
   by Set-Cookie in a way that conflicts with such folding.

3.1.  Examples



   Using the Set-Cookie header, a server can send the user agent a short
   string in an HTTP response that the user agent will return in future
   HTTP requests that are within the scope of the cookie.  For example,
   the server can send the user agent a "session identifier" named SID
   with the value 31d4d96e407aad42.  The user agent then returns the
   session identifier in subsequent requests.









Barth                        Standards Track                    [Page 6]

RFC 6265             HTTP State Management Mechanism          April 2011


   == Server -> User Agent ==

   Set-Cookie: SID=31d4d96e407aad42

   == User Agent -> Server ==

   Cookie: SID=31d4d96e407aad42

   The server can alter the default scope of the cookie using the Path
   and Domain attributes.  For example, the server can instruct the user
   agent to return the cookie to every path and every subdomain of
   example.com.

   == Server -> User Agent ==

   Set-Cookie: SID=31d4d96e407aad42; Path=/; Domain=example.com

   == User Agent -> Server ==

   Cookie: SID=31d4d96e407aad42

   As shown in the next example, the server can store multiple cookies
   at the user agent.  For example, the server can store a session
   identifier as well as the user's preferred language by returning two
   Set-Cookie header fields.  Notice that the server uses the Secure and
   HttpOnly attributes to provide additional security protections for
   the more sensitive session identifier (see Section 4.1.2.)

   == Server -> User Agent ==

   Set-Cookie: SID=31d4d96e407aad42; Path=/; Secure; HttpOnly
   Set-Cookie: lang=en-US; Path=/; Domain=example.com

   == User Agent -> Server ==

   Cookie: SID=31d4d96e407aad42; lang=en-US

   Notice that the Cookie header above contains two cookies, one named
   SID and one named lang.  If the server wishes the user agent to
   persist the cookie over multiple "sessions" (e.g., user agent
   restarts), the server can specify an expiration date in the Expires
   attribute.  Note that the user agent might delete the cookie before
   the expiration date if the user agent's cookie store exceeds its
   quota or if the user manually deletes the server's cookie.







Barth                        Standards Track                    [Page 7]

RFC 6265             HTTP State Management Mechanism          April 2011


   == Server -> User Agent ==

   Set-Cookie: lang=en-US; Expires=Wed, 09 Jun 2021 10:18:14 GMT

   == User Agent -> Server ==

   Cookie: SID=31d4d96e407aad42; lang=en-US

   Finally, to remove a cookie, the server returns a Set-Cookie header
   with an expiration date in the past.  The server will be successful
   in removing the cookie only if the Path and the Domain attribute in
   the Set-Cookie header match the values used when the cookie was
   created.

   == Server -> User Agent ==

   Set-Cookie: lang=; Expires=Sun, 06 Nov 1994 08:49:37 GMT

   == User Agent -> Server ==

   Cookie: SID=31d4d96e407aad42

4.  Server Requirements



   This section describes the syntax and semantics of a well-behaved
   profile of the Cookie and Set-Cookie headers.

4.1.  Set-Cookie



   The Set-Cookie HTTP response header is used to send cookies from the
   server to the user agent.

4.1.1.  Syntax



   Informally, the Set-Cookie response header contains the header name
   "Set-Cookie" followed by a ":" and a cookie.  Each cookie begins with
   a name-value-pair, followed by zero or more attribute-value pairs.
   Servers SHOULD NOT send Set-Cookie headers that fail to conform to
   the following grammar:












Barth                        Standards Track                    [Page 8]

RFC 6265             HTTP State Management Mechanism          April 2011


 set-cookie-header = "Set-Cookie:" SP set-cookie-string
 set-cookie-string = cookie-pair *( ";" SP cookie-av )
 cookie-pair       = cookie-name "=" cookie-value
 cookie-name       = token
 cookie-value      = *cookie-octet / ( DQUOTE *cookie-octet DQUOTE )
 cookie-octet      = %x21 / %x23-2B / %x2D-3A / %x3C-5B / %x5D-7E
                       ; US-ASCII characters excluding CTLs,
                       ; whitespace DQUOTE, comma, semicolon,
                       ; and backslash
 token             = <token, defined in [RFC2616], Section 2.2>

 cookie-av         = expires-av / max-age-av / domain-av /
                     path-av / secure-av / httponly-av /
                     extension-av
 expires-av        = "Expires=" sane-cookie-date
 sane-cookie-date  = <rfc1123-date, defined in [RFC2616], Section 3.3.1>
 max-age-av        = "Max-Age=" non-zero-digit *DIGIT
                       ; In practice, both expires-av and max-age-av
                       ; are limited to dates representable by the
                       ; user agent.
 non-zero-digit    = %x31-39
                       ; digits 1 through 9
 domain-av         = "Domain=" domain-value
 domain-value      = <subdomain>
                       ; defined in [RFC1034], Section 3.5, as
                       ; enhanced by [RFC1123], Section 2.1
 path-av           = "Path=" path-value
 path-value        = <any CHAR except CTLs or ";">
 secure-av         = "Secure"
 httponly-av       = "HttpOnly"
 extension-av      = <any CHAR except CTLs or ";">

   Note that some of the grammatical terms above reference documents
   that use different grammatical notations than this document (which
   uses ABNF from [RFC5234]).

   The semantics of the cookie-value are not defined by this document.

   To maximize compatibility with user agents, servers that wish to
   store arbitrary data in a cookie-value SHOULD encode that data, for
   example, using Base64 [RFC4648].

   The portions of the set-cookie-string produced by the cookie-av term
   are known as attributes.  To maximize compatibility with user agents,
   servers SHOULD NOT produce two attributes with the same name in the
   same set-cookie-string.  (See Section 5.3 for how user agents handle
   this case.)




Barth                        Standards Track                    [Page 9]

RFC 6265             HTTP State Management Mechanism          April 2011


   Servers SHOULD NOT include more than one Set-Cookie header field in
   the same response with the same cookie-name.  (See Section 5.2 for
   how user agents handle this case.)

   If a server sends multiple responses containing Set-Cookie headers
   concurrently to the user agent (e.g., when communicating with the
   user agent over multiple sockets), these responses create a "race
   condition" that can lead to unpredictable behavior.

   NOTE: Some existing user agents differ in their interpretation of
   two-digit years.  To avoid compatibility issues, servers SHOULD use
   the rfc1123-date format, which requires a four-digit year.

   NOTE: Some user agents store and process dates in cookies as 32-bit
   UNIX time_t values.  Implementation bugs in the libraries supporting
   time_t processing on some systems might cause such user agents to
   process dates after the year 2038 incorrectly.

4.1.2.  Semantics (Non-Normative)



   This section describes simplified semantics of the Set-Cookie header.
   These semantics are detailed enough to be useful for understanding
   the most common uses of cookies by servers.  The full semantics are
   described in Section 5.

   When the user agent receives a Set-Cookie header, the user agent
   stores the cookie together with its attributes.  Subsequently, when
   the user agent makes an HTTP request, the user agent includes the
   applicable, non-expired cookies in the Cookie header.

   If the user agent receives a new cookie with the same cookie-name,
   domain-value, and path-value as a cookie that it has already stored,
   the existing cookie is evicted and replaced with the new cookie.
   Notice that servers can delete cookies by sending the user agent a
   new cookie with an Expires attribute with a value in the past.

   Unless the cookie's attributes indicate otherwise, the cookie is
   returned only to the origin server (and not, for example, to any
   subdomains), and it expires at the end of the current session (as
   defined by the user agent).  User agents ignore unrecognized cookie
   attributes (but not the entire cookie).










Barth                        Standards Track                   [Page 10]

RFC 6265             HTTP State Management Mechanism          April 2011


4.1.2.1.  The Expires Attribute



   The Expires attribute indicates the maximum lifetime of the cookie,
   represented as the date and time at which the cookie expires.  The
   user agent is not required to retain the cookie until the specified
   date has passed.  In fact, user agents often evict cookies due to
   memory pressure or privacy concerns.

4.1.2.2.  The Max-Age Attribute



   The Max-Age attribute indicates the maximum lifetime of the cookie,
   represented as the number of seconds until the cookie expires.  The
   user agent is not required to retain the cookie for the specified
   duration.  In fact, user agents often evict cookies due to memory
   pressure or privacy concerns.

      NOTE: Some existing user agents do not support the Max-Age
      attribute.  User agents that do not support the Max-Age attribute
      ignore the attribute.

   If a cookie has both the Max-Age and the Expires attribute, the Max-
   Age attribute has precedence and controls the expiration date of the
   cookie.  If a cookie has neither the Max-Age nor the Expires
   attribute, the user agent will retain the cookie until "the current
   session is over" (as defined by the user agent).

4.1.2.3.  The Domain Attribute



   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.  (Note that a leading %x2E ("."), if present,
   is ignored even though that character is not permitted, but a
   trailing %x2E ("."), if present, will cause the user agent to ignore
   the attribute.)  If the server omits the Domain attribute, the user
   agent will return the cookie only to the origin server.

      WARNING: Some existing user agents treat an absent Domain
      attribute as if the Domain attribute were present and contained
      the current host name.  For example, if example.com returns a Set-
      Cookie header without a Domain attribute, these user agents will
      erroneously send the cookie to www.example.com as well.








Barth                        Standards Track                   [Page 11]

RFC 6265             HTTP State Management Mechanism          April 2011


   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.)

4.1.2.4.  The Path Attribute



   The scope of each cookie is limited to a set of paths, controlled by
   the Path attribute.  If the server omits the Path attribute, the user
   agent will use the "directory" of the request-uri's path component as
   the default value.  (See Section 5.1.4 for more details.)

   The user agent will include the cookie in an HTTP request only if the
   path portion of the request-uri matches (or is a subdirectory of) the
   cookie's Path attribute, where the %x2F ("/") character is
   interpreted as a directory separator.

   Although seemingly useful for isolating cookies between different
   paths within a given host, the Path attribute cannot be relied upon
   for security (see Section 8).

4.1.2.5.  The Secure Attribute



   The Secure attribute limits the scope of the cookie to "secure"
   channels (where "secure" is defined by the user agent).  When a
   cookie has the Secure attribute, the user agent will include the
   cookie in an HTTP request only if the request is transmitted over a
   secure channel (typically HTTP over Transport Layer Security (TLS)
   [RFC2818]).

   Although seemingly useful for protecting cookies from active network
   attackers, the Secure attribute protects only the cookie's
   confidentiality.  An active network attacker can overwrite Secure
   cookies from an insecure channel, disrupting their integrity (see
   Section 8.6 for more details).









Barth                        Standards Track                   [Page 12]

RFC 6265             HTTP State Management Mechanism          April 2011


4.1.2.6.  The HttpOnly Attribute



   The HttpOnly attribute limits the scope of the cookie to HTTP
   requests.  In particular, the attribute instructs the user agent to
   omit the cookie when providing access to cookies via "non-HTTP" APIs
   (such as a web browser API that exposes cookies to scripts).

   Note that the HttpOnly attribute is independent of the Secure
   attribute: a cookie can have both the HttpOnly and the Secure
   attribute.

4.2.  Cookie



4.2.1.  Syntax



   The user agent sends stored cookies to the origin server in the
   Cookie header.  If the server conforms to the requirements in
   Section 4.1 (and the user agent conforms to the requirements in
   Section 5), the user agent will send a Cookie header that conforms to
   the following grammar:

   cookie-header = "Cookie:" OWS cookie-string OWS
   cookie-string = cookie-pair *( ";" SP cookie-pair )

4.2.2.  Semantics



   Each cookie-pair represents a cookie stored by the user agent.  The
   cookie-pair contains the cookie-name and cookie-value the user agent
   received in the Set-Cookie header.

   Notice that the cookie attributes are not returned.  In particular,
   the server cannot determine from the Cookie header alone when a
   cookie will expire, for which hosts the cookie is valid, for which
   paths the cookie is valid, or whether the cookie was set with the
   Secure or HttpOnly attributes.

   The semantics of individual cookies in the Cookie header are not
   defined by this document.  Servers are expected to imbue these
   cookies with application-specific semantics.

   Although cookies are serialized linearly in the Cookie header,
   servers SHOULD NOT rely upon the serialization order.  In particular,
   if the Cookie header contains two cookies with the same name (e.g.,
   that were set with different Path or Domain attributes), servers
   SHOULD NOT rely upon the order in which these cookies appear in the
   header.





Barth                        Standards Track                   [Page 13]

RFC 6265             HTTP State Management Mechanism          April 2011


5.  User Agent Requirements



   This section specifies the Cookie and Set-Cookie headers in
   sufficient detail that a user agent implementing these requirements
   precisely can interoperate with existing servers (even those that do
   not conform to the well-behaved profile described in Section 4).

   A user agent could enforce more restrictions than those specified
   herein (e.g., for the sake of improved security); however,
   experiments have shown that such strictness reduces the likelihood
   that a user agent will be able to interoperate with existing servers.

5.1.  Subcomponent Algorithms



   This section defines some algorithms used by user agents to process
   specific subcomponents of the Cookie and Set-Cookie headers.

5.1.1.  Dates



   The user agent MUST use an algorithm equivalent to the following
   algorithm to parse a cookie-date.  Note that the various boolean
   flags defined as a part of the algorithm (i.e., found-time, found-
   day-of-month, found-month, found-year) are initially "not set".

   1.  Using the grammar below, divide the cookie-date into date-tokens.

   cookie-date     = *delimiter date-token-list *delimiter
   date-token-list = date-token *( 1*delimiter date-token )
   date-token      = 1*non-delimiter

   delimiter       = %x09 / %x20-2F / %x3B-40 / %x5B-60 / %x7B-7E
   non-delimiter   = %x00-08 / %x0A-1F / DIGIT / ":" / ALPHA / %x7F-FF
   non-digit       = %x00-2F / %x3A-FF

   day-of-month    = 1*2DIGIT ( non-digit *OCTET )
   month           = ( "jan" / "feb" / "mar" / "apr" /
                       "may" / "jun" / "jul" / "aug" /
                       "sep" / "oct" / "nov" / "dec" ) *OCTET
   year            = 2*4DIGIT ( non-digit *OCTET )
   time            = hms-time ( non-digit *OCTET )
   hms-time        = time-field ":" time-field ":" time-field
   time-field      = 1*2DIGIT

   2.  Process each date-token sequentially in the order the date-tokens
       appear in the cookie-date:






Barth                        Standards Track                   [Page 14]

RFC 6265             HTTP State Management Mechanism          April 2011


       1.  If the found-time flag is not set and the token matches the
           time production, set the found-time flag and set the hour-
           value, minute-value, and second-value to the numbers denoted
           by the digits in the date-token, respectively.  Skip the
           remaining sub-steps and continue to the next date-token.

       2.  If the found-day-of-month flag is not set and the date-token
           matches the day-of-month production, set the found-day-of-
           month flag and set the day-of-month-value to the number
           denoted by the date-token.  Skip the remaining sub-steps and
           continue to the next date-token.

       3.  If the found-month flag is not set and the date-token matches
           the month production, set the found-month flag and set the
           month-value to the month denoted by the date-token.  Skip the
           remaining sub-steps and continue to the next date-token.

       4.  If the found-year flag is not set and the date-token matches
           the year production, set the found-year flag and set the
           year-value to the number denoted by the date-token.  Skip the
           remaining sub-steps and continue to the next date-token.

   3.  If the year-value is greater than or equal to 70 and less than or
       equal to 99, increment the year-value by 1900.

   4.  If the year-value is greater than or equal to 0 and less than or
       equal to 69, increment the year-value by 2000.

       1.  NOTE: Some existing user agents interpret two-digit years
           differently.

   5.  Abort these steps and fail to parse the cookie-date if:

       *  at least one of the found-day-of-month, found-month, found-
          year, or found-time flags is not set,

       *  the day-of-month-value is less than 1 or greater than 31,

       *  the year-value is less than 1601,

       *  the hour-value is greater than 23,

       *  the minute-value is greater than 59, or

       *  the second-value is greater than 59.

       (Note that leap seconds cannot be represented in this syntax.)




Barth                        Standards Track                   [Page 15]

RFC 6265             HTTP State Management Mechanism          April 2011


   6.  Let the parsed-cookie-date be the date whose day-of-month, month,
       year, hour, minute, and second (in UTC) are the day-of-month-
       value, the month-value, the year-value, the hour-value, the
       minute-value, and the second-value, respectively.  If no such
       date exists, abort these steps and fail to parse the cookie-date.

   7.  Return the parsed-cookie-date as the result of this algorithm.



5.1.2.  Canonicalized Host Names



   A canonicalized host name is the string generated by the following
   algorithm:

   1.  Convert the host name to a sequence of individual domain name
       labels.

   2.  Convert each label that is not a Non-Reserved LDH (NR-LDH) label,
       to an A-label (see Section 2.3.2.1 of [RFC5890] for the former
       and latter), or to a "punycode label" (a label resulting from the
       "ToASCII" conversion in Section 4 of [RFC3490]), as appropriate
       (see Section 6.3 of this specification).

   3.  Concatenate the resulting labels, separated by a %x2E (".")
       character.

5.1.3.  Domain Matching



   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).

5.1.4.  Paths and Path-Match



   The user agent MUST use an algorithm equivalent to the following
   algorithm to compute the default-path of a cookie:




Barth                        Standards Track                   [Page 16]

RFC 6265             HTTP State Management Mechanism          April 2011


   1.  Let uri-path be the path portion of the request-uri if such a
       portion exists (and empty otherwise).  For example, if the
       request-uri contains just a path (and optional query string),
       then the uri-path is that path (without the %x3F ("?") character
       or query string), and if the request-uri contains a full
       absoluteURI, the uri-path is the path component of that URI.

   2.  If the uri-path is empty or if the first character of the uri-
       path is not a %x2F ("/") character, output %x2F ("/") and skip
       the remaining steps.

   3.  If the uri-path contains no more than one %x2F ("/") character,
       output %x2F ("/") and skip the remaining step.

   4.  Output the characters of the uri-path from the first character up
       to, but not including, the right-most %x2F ("/").

   A request-path path-matches a given cookie-path if at least one of
   the following conditions holds:

   o  The cookie-path and the request-path are identical.

   o  The cookie-path is a prefix of the request-path, and the last
      character of the cookie-path is %x2F ("/").

   o  The cookie-path is a prefix of the request-path, and the first
      character of the request-path that is not included in the cookie-
      path is a %x2F ("/") character.

5.2.  The Set-Cookie Header



   When a user agent receives a Set-Cookie header field in an HTTP
   response, the user agent MAY ignore the Set-Cookie header field in
   its entirety.  For example, the user agent might wish to block
   responses to "third-party" requests from setting cookies (see
   Section 7.1).

   If the user agent does not ignore the Set-Cookie header field in its
   entirety, the user agent MUST parse the field-value of the Set-Cookie
   header field as a set-cookie-string (defined below).

   NOTE: The algorithm below is more permissive than the grammar in
   Section 4.1.  For example, the algorithm strips leading and trailing
   whitespace from the cookie name and value (but maintains internal
   whitespace), whereas the grammar in Section 4.1 forbids whitespace in
   these positions.  User agents use this algorithm so as to
   interoperate with servers that do not follow the recommendations in
   Section 4.



Barth                        Standards Track                   [Page 17]

RFC 6265             HTTP State Management Mechanism          April 2011


   A user agent MUST use an algorithm equivalent to the following
   algorithm to parse a "set-cookie-string":

   1.  If the set-cookie-string contains a %x3B (";") character:

          The name-value-pair string consists of the characters up to,
          but not including, the first %x3B (";"), and the unparsed-
          attributes consist of the remainder of the set-cookie-string
          (including the %x3B (";") in question).

       Otherwise:

          The name-value-pair string consists of all the characters
          contained in the set-cookie-string, and the unparsed-
          attributes is the empty string.

   2.  If the name-value-pair string lacks a %x3D ("=") character,
       ignore the set-cookie-string entirely.

   3.  The (possibly empty) name string consists of the characters up
       to, but not including, the first %x3D ("=") character, and the
       (possibly empty) value string consists of the characters after
       the first %x3D ("=") character.

   4.  Remove any leading or trailing WSP characters from the name
       string and the value string.

   5.  If the name string is empty, ignore the set-cookie-string
       entirely.

   6.  The cookie-name is the name string, and the cookie-value is the
       value string.



   The user agent MUST use an algorithm equivalent to the following
   algorithm to parse the unparsed-attributes:

   1.  If the unparsed-attributes string is empty, skip the rest of
       these steps.

   2.  Discard the first character of the unparsed-attributes (which
       will be a %x3B (";") character).

   3.  If the remaining unparsed-attributes contains a %x3B (";")
       character:

          Consume the characters of the unparsed-attributes up to, but
          not including, the first %x3B (";") character.




Barth                        Standards Track                   [Page 18]

RFC 6265             HTTP State Management Mechanism          April 2011


       Otherwise:

          Consume the remainder of the unparsed-attributes.

       Let the cookie-av string be the characters consumed in this step.

   4.  If the cookie-av string contains a %x3D ("=") character:

          The (possibly empty) attribute-name string consists of the
          characters up to, but not including, the first %x3D ("=")
          character, and the (possibly empty) attribute-value string
          consists of the characters after the first %x3D ("=")
          character.

       Otherwise:

          The attribute-name string consists of the entire cookie-av
          string, and the attribute-value string is empty.

   5.  Remove any leading or trailing WSP characters from the attribute-
       name string and the attribute-value string.

   6.  Process the attribute-name and attribute-value according to the
       requirements in the following subsections.  (Notice that
       attributes with unrecognized attribute-names are ignored.)

   7.  Return to Step 1 of this algorithm.

   When the user agent finishes parsing the set-cookie-string, the user
   agent is said to "receive a cookie" from the request-uri with name
   cookie-name, value cookie-value, and attributes cookie-attribute-
   list.  (See Section 5.3 for additional requirements triggered by
   receiving a cookie.)

5.2.1.  The Expires Attribute



   If the attribute-name case-insensitively matches the string
   "Expires", the user agent MUST process the cookie-av as follows.

   Let the expiry-time be the result of parsing the attribute-value as
   cookie-date (see Section 5.1.1).

   If the attribute-value failed to parse as a cookie date, ignore the
   cookie-av.

   If the expiry-time is later than the last date the user agent can
   represent, the user agent MAY replace the expiry-time with the last
   representable date.



Barth                        Standards Track                   [Page 19]

RFC 6265             HTTP State Management Mechanism          April 2011


   If the expiry-time is earlier than the earliest date the user agent
   can represent, the user agent MAY replace the expiry-time with the
   earliest representable date.

   Append an attribute to the cookie-attribute-list with an attribute-
   name of Expires and an attribute-value of expiry-time.

5.2.2.  The Max-Age Attribute



   If the attribute-name case-insensitively matches the string "Max-
   Age", the user agent MUST process the cookie-av as follows.

   If the first character of the attribute-value is not a DIGIT or a "-"
   character, ignore the cookie-av.

   If the remainder of attribute-value contains a non-DIGIT character,
   ignore the cookie-av.

   Let delta-seconds be the attribute-value converted to an integer.

   If delta-seconds is less than or equal to zero (0), let expiry-time
   be the earliest representable date and time.  Otherwise, let the
   expiry-time be the current date and time plus delta-seconds seconds.

   Append an attribute to the cookie-attribute-list with an attribute-
   name of Max-Age and an attribute-value of expiry-time.

5.2.3.  The Domain Attribute



   If the attribute-name case-insensitively matches the string "Domain",
   the user agent MUST process the cookie-av as follows.

   If the attribute-value is empty, the behavior is undefined.  However,
   the user agent SHOULD ignore the cookie-av entirely.

   If the first character of the attribute-value string is %x2E ("."):

      Let cookie-domain be the attribute-value without the leading %x2E
      (".") character.

   Otherwise:

      Let cookie-domain be the entire attribute-value.

   Convert the cookie-domain to lower case.

   Append an attribute to the cookie-attribute-list with an attribute-
   name of Domain and an attribute-value of cookie-domain.



Barth                        Standards Track                   [Page 20]

RFC 6265             HTTP State Management Mechanism          April 2011


5.2.4.  The Path Attribute



   If the attribute-name case-insensitively matches the string "Path",
   the user agent MUST process the cookie-av as follows.

   If the attribute-value is empty or if the first character of the
   attribute-value is not %x2F ("/"):

      Let cookie-path be the default-path.

   Otherwise:

      Let cookie-path be the attribute-value.

   Append an attribute to the cookie-attribute-list with an attribute-
   name of Path and an attribute-value of cookie-path.

5.2.5.  The Secure Attribute



   If the attribute-name case-insensitively matches the string "Secure",
   the user agent MUST append an attribute to the cookie-attribute-list
   with an attribute-name of Secure and an empty attribute-value.

5.2.6.  The HttpOnly Attribute



   If the attribute-name case-insensitively matches the string
   "HttpOnly", the user agent MUST append an attribute to the cookie-
   attribute-list with an attribute-name of HttpOnly and an empty
   attribute-value.

5.3.  Storage Model



   The user agent stores the following fields about each cookie: name,
   value, expiry-time, domain, path, creation-time, last-access-time,
   persistent-flag, host-only-flag, secure-only-flag, and http-only-
   flag.

   When the user agent "receives a cookie" from a request-uri with name
   cookie-name, value cookie-value, and attributes cookie-attribute-
   list, the user agent MUST process the cookie as follows:

   1.   A user agent MAY ignore a received cookie in its entirety.  For
        example, the user agent might wish to block receiving cookies
        from "third-party" responses or the user agent might not wish to
        store cookies that exceed some size.






Barth                        Standards Track                   [Page 21]

RFC 6265             HTTP State Management Mechanism          April 2011


   2.   Create a new cookie with name cookie-name, value cookie-value.
        Set the creation-time and the last-access-time to the current
        date and time.

   3.   If the cookie-attribute-list contains an attribute with an
        attribute-name of "Max-Age":

           Set the cookie's persistent-flag to true.

           Set the cookie's expiry-time to attribute-value of the last
           attribute in the cookie-attribute-list with an attribute-name
           of "Max-Age".

        Otherwise, if the cookie-attribute-list contains an attribute
        with an attribute-name of "Expires" (and does not contain an
        attribute with an attribute-name of "Max-Age"):

           Set the cookie's persistent-flag to true.

           Set the cookie's expiry-time to attribute-value of the last
           attribute in the cookie-attribute-list with an attribute-name
           of "Expires".

        Otherwise:

           Set the cookie's persistent-flag to false.

           Set the cookie's expiry-time to the latest representable
           date.

   4.   If the cookie-attribute-list contains an attribute with an
        attribute-name of "Domain":

           Let the domain-attribute be the attribute-value of the last
           attribute in the cookie-attribute-list with an attribute-name
           of "Domain".

        Otherwise:

           Let the domain-attribute be the empty string.

   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.



Barth                        Standards Track                   [Page 22]

RFC 6265             HTTP State Management Mechanism          April 2011


           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.

   7.   If the cookie-attribute-list contains an attribute with an
        attribute-name of "Path", set the cookie's path to attribute-
        value of the last attribute in the cookie-attribute-list with an
        attribute-name of "Path".  Otherwise, set the cookie's path to
        the default-path of the request-uri.

   8.   If the cookie-attribute-list contains an attribute with an
        attribute-name of "Secure", set the cookie's secure-only-flag to
        true.  Otherwise, set the cookie's secure-only-flag to false.

   9.   If the cookie-attribute-list contains an attribute with an
        attribute-name of "HttpOnly", set the cookie's http-only-flag to
        true.  Otherwise, set the cookie's http-only-flag to false.





Barth                        Standards Track                   [Page 23]

RFC 6265             HTTP State Management Mechanism          April 2011


   10.  If the cookie was received from a "non-HTTP" API and the
        cookie's http-only-flag is set, abort these steps and ignore the
        cookie entirely.



   11.  If the cookie store contains a cookie with the same name,
        domain, and path as the newly created cookie:



        1.  Let old-cookie be the existing cookie with the same name,
            domain, and path as the newly created cookie.  (Notice that
            this algorithm maintains the invariant that there is at most
            one such cookie.)

        2.  If the newly created cookie was received from a "non-HTTP"
            API and the old-cookie's http-only-flag is set, abort these
            steps and ignore the newly created cookie entirely.

        3.  Update the creation-time of the newly created cookie to
            match the creation-time of the old-cookie.

        4.  Remove the old-cookie from the cookie store.

   12.  Insert the newly created cookie into the cookie store.



   A cookie is "expired" if the cookie has an expiry date in the past.

   The user agent MUST evict all expired cookies from the cookie store
   if, at any time, an expired cookie exists in the cookie store.

   At any time, the user agent MAY "remove excess cookies" from the
   cookie store if the number of cookies sharing a domain field exceeds
   some implementation-defined upper bound (such as 50 cookies).

   At any time, the user agent MAY "remove excess cookies" from the
   cookie store if the cookie store exceeds some predetermined upper
   bound (such as 3000 cookies).

   When the user agent removes excess cookies from the cookie store, the
   user agent MUST evict cookies in the following priority order:

   1.  Expired cookies.

   2.  Cookies that share a domain field with more than a predetermined
       number of other cookies.

   3.  All cookies.

   If two cookies have the same removal priority, the user agent MUST
   evict the cookie with the earliest last-access date first.



Barth                        Standards Track                   [Page 24]

RFC 6265             HTTP State Management Mechanism          April 2011


   When "the current session is over" (as defined by the user agent),
   the user agent MUST remove from the cookie store all cookies with the
   persistent-flag set to false.

5.4.  The Cookie Header



   The user agent includes stored cookies in the Cookie HTTP request
   header.

   When the user agent generates an HTTP request, the user agent MUST
   NOT
attach more than one Cookie header field.

   A user agent MAY omit the Cookie header in its entirety.  For
   example, the user agent might wish to block sending cookies during
   "third-party" requests from setting cookies (see Section 7.1).

   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 secure-only-flag is true, then the request-
          uri's scheme must denote a "secure" protocol (as defined by
          the user agent).

             NOTE: The notion of a "secure" protocol is not defined by
             this document.  Typically, user agents consider a protocol
             secure if the protocol makes use of transport-layer





Barth                        Standards Track                   [Page 25]

RFC 6265             HTTP State Management Mechanism          April 2011


             security, such as SSL or TLS.  For example, most user
             agents consider "https" to be a scheme that denotes a
             secure protocol.

       *  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).

   2.  The user agent SHOULD sort the cookie-list in the following
       order:

       *  Cookies with longer paths are listed before cookies with
          shorter paths.

       *  Among cookies that have equal-length path fields, cookies with
          earlier creation-times are listed before cookies with later
          creation-times.

       NOTE: Not all user agents sort the cookie-list in this order, but
       this order reflects common practice when this document was
       written, and, historically, there have been servers that
       (erroneously) depended on this order.

   3.  Update the last-access-time of each cookie in the cookie-list to
       the current date and time.

   4.  Serialize the cookie-list into a cookie-string by processing each
       cookie in the cookie-list in order:

       1.  Output the cookie's name, the %x3D ("=") character, and the
           cookie's value.

       2.  If there is an unprocessed cookie in the cookie-list, output
           the characters %x3B and %x20 ("; ").

   NOTE: Despite its name, the cookie-string is actually a sequence of
   octets, not a sequence of characters.  To convert the cookie-string
   (or components thereof) into a sequence of characters (e.g., for
   presentation to the user), the user agent might wish to try using the
   UTF-8 character encoding [RFC3629] to decode the octet sequence.
   This decoding might fail, however, because not every sequence of
   octets is valid UTF-8.









Barth                        Standards Track                   [Page 26]

RFC 6265             HTTP State Management Mechanism          April 2011


6.  Implementation Considerations

6.1.  Limits



   Practical user agent implementations have limits on the number and
   size of cookies that they can store.  General-use user agents SHOULD
   provide each of the following minimum capabilities:

   o  At least 4096 bytes per cookie (as measured by the sum of the
      length of the cookie's name, value, and attributes).

   o  At least 50 cookies per domain.

   o  At least 3000 cookies total.

   Servers SHOULD use as few and as small cookies as possible to avoid
   reaching these implementation limits and to minimize network
   bandwidth due to the Cookie header being included in every request.

   Servers SHOULD gracefully degrade if the user agent fails to return
   one or more cookies in the Cookie header because the user agent might
   evict any cookie at any time on orders from the user.

6.2.  Application Programming Interfaces



   One reason the Cookie and Set-Cookie headers use such esoteric syntax
   is that many platforms (both in servers and user agents) provide a
   string-based application programming interface (API) to cookies,
   requiring application-layer programmers to generate and parse the
   syntax used by the Cookie and Set-Cookie headers, which many
   programmers have done incorrectly, resulting in interoperability
   problems.

   Instead of providing string-based APIs to cookies, platforms would be
   well-served by providing more semantic APIs.  It is beyond the scope
   of this document to recommend specific API designs, but there are
   clear benefits to accepting an abstract "Date" object instead of a
   serialized date string.

6.3.  IDNA Dependency and Migration



   IDNA2008 [RFC5890] supersedes IDNA2003 [RFC3490].  However, there are
   differences between the two specifications, and thus there can be
   differences in processing (e.g., converting) domain name labels that
   have been registered under one from those registered under the other.
   There will be a transition period of some time during which IDNA2003-
   based domain name labels will exist in the wild.  User agents SHOULD
   implement IDNA2008 [RFC5890] and MAY implement [UTS46] or [RFC5895]



Barth                        Standards Track                   [Page 27]

RFC 6265             HTTP State Management Mechanism          April 2011


   in order to facilitate their IDNA transition.  If a user agent does
   not implement IDNA2008, the user agent MUST implement IDNA2003
   [RFC3490].

7.  Privacy Considerations

   Cookies are often criticized for letting servers track users.  For
   example, a number of "web analytics" companies use cookies to
   recognize when a user returns to a web site or visits another web
   site.  Although cookies are not the only mechanism servers can use to
   track users across HTTP requests, cookies facilitate tracking because
   they are persistent across user agent sessions and can be shared
   between hosts.

7.1.  Third-Party Cookies



   Particularly worrisome are so-called "third-party" cookies.  In
   rendering an HTML document, a user agent often requests resources
   from other servers (such as advertising networks).  These third-party
   servers can use cookies to track the user even if the user never
   visits the server directly.  For example, if a user visits a site
   that contains content from a third party and then later visits
   another site that contains content from the same third party, the
   third party can track the user between the two sites.

   Some user agents restrict how third-party cookies behave.  For
   example, some of these user agents refuse to send the Cookie header
   in third-party requests.  Others refuse to process the Set-Cookie
   header in responses to third-party requests.  User agents vary widely
   in their third-party cookie policies.  This document grants user
   agents wide latitude to experiment with third-party cookie policies
   that balance the privacy and compatibility needs of their users.
   However, this document does not endorse any particular third-party
   cookie policy.

   Third-party cookie blocking policies are often ineffective at
   achieving their privacy goals if servers attempt to work around their
   restrictions to track users.  In particular, two collaborating
   servers can often track users without using cookies at all by
   injecting identifying information into dynamic URLs.

7.2.  User Controls



   User agents SHOULD provide users with a mechanism for managing the
   cookies stored in the cookie store.  For example, a user agent might
   let users delete all cookies received during a specified time period





Barth                        Standards Track                   [Page 28]

RFC 6265             HTTP State Management Mechanism          April 2011


   or all the cookies related to a particular domain.  In addition, many
   user agents include a user interface element that lets users examine
   the cookies stored in their cookie store.

   User agents SHOULD provide users with a mechanism for disabling
   cookies.  When cookies are disabled, the user agent MUST NOT include
   a Cookie header in outbound HTTP requests and the user agent MUST NOT
   process Set-Cookie headers in inbound HTTP responses.

   Some user agents provide users the option of preventing persistent
   storage of cookies across sessions.  When configured thusly, user
   agents MUST treat all received cookies as if the persistent-flag were
   set to false.  Some popular user agents expose this functionality via
   "private browsing" mode [Aggarwal2010].

   Some user agents provide users with the ability to approve individual
   writes to the cookie store.  In many common usage scenarios, these
   controls generate a large number of prompts.  However, some privacy-
   conscious users find these controls useful nonetheless.

7.3.  Expiration Dates



   Although servers can set the expiration date for cookies to the
   distant future, most user agents do not actually retain cookies for
   multiple decades.  Rather than choosing gratuitously long expiration
   periods, servers SHOULD promote user privacy by selecting reasonable
   cookie expiration periods based on the purpose of the cookie.  For
   example, a typical session identifier might reasonably be set to
   expire in two weeks.

8.  Security Considerations



8.1.  Overview



   Cookies have a number of security pitfalls.  This section overviews a
   few of the more salient issues.

   In particular, cookies encourage developers to rely on ambient
   authority for authentication, often becoming vulnerable to attacks
   such as cross-site request forgery [CSRF].  Also, when storing
   session identifiers in cookies, developers often create session
   fixation vulnerabilities.

   Transport-layer encryption, such as that employed in HTTPS, is
   insufficient to prevent a network attacker from obtaining or altering
   a victim's cookies because the cookie protocol itself has various
   vulnerabilities (see "Weak Confidentiality" and "Weak Integrity",




Barth                        Standards Track                   [Page 29]

RFC 6265             HTTP State Management Mechanism          April 2011


   below).  In addition, by default, cookies do not provide
   confidentiality or integrity from network attackers, even when used
   in conjunction with HTTPS.

8.2.  Ambient Authority



   A server that uses cookies to authenticate users can suffer security
   vulnerabilities because some user agents let remote parties issue
   HTTP requests from the user agent (e.g., via HTTP redirects or HTML
   forms).  When issuing those requests, user agents attach cookies even
   if the remote party does not know the contents of the cookies,
   potentially letting the remote party exercise authority at an unwary
   server.

   Although this security concern goes by a number of names (e.g.,
   cross-site request forgery, confused deputy), the issue stems from
   cookies being a form of ambient authority.  Cookies encourage server
   operators to separate designation (in the form of URLs) from
   authorization (in the form of cookies).  Consequently, the user agent
   might supply the authorization for a resource designated by the
   attacker, possibly causing the server or its clients to undertake
   actions designated by the attacker as though they were authorized by
   the user.

   Instead of using cookies for authorization, server operators might
   wish to consider entangling designation and authorization by treating
   URLs as capabilities.  Instead of storing secrets in cookies, this
   approach stores secrets in URLs, requiring the remote entity to
   supply the secret itself.  Although this approach is not a panacea,
   judicious application of these principles can lead to more robust
   security.

8.3.  Clear Text



   Unless sent over a secure channel (such as TLS), the information in
   the Cookie and Set-Cookie headers is transmitted in the clear.

   1.  All sensitive information conveyed in these headers is exposed to
       an eavesdropper.

   2.  A malicious intermediary could alter the headers as they travel
       in either direction, with unpredictable results.

   3.  A malicious client could alter the Cookie header before
       transmission, with unpredictable results.






Barth                        Standards Track                   [Page 30]

RFC 6265             HTTP State Management Mechanism          April 2011


   Servers SHOULD encrypt and sign the contents of cookies (using
   whatever format the server desires) when transmitting them to the
   user agent (even when sending the cookies over a secure channel).
   However, encrypting and signing cookie contents does not prevent an
   attacker from transplanting a cookie from one user agent to another
   or from replaying the cookie at a later time.

   In addition to encrypting and signing the contents of every cookie,
   servers that require a higher level of security SHOULD use the Cookie
   and Set-Cookie headers only over a secure channel.  When using
   cookies over a secure channel, servers SHOULD set the Secure
   attribute (see Section 4.1.2.5) for every cookie.  If a server does
   not set the Secure attribute, the protection provided by the secure
   channel will be largely moot.

   For example, consider a webmail server that stores a session
   identifier in a cookie and is typically accessed over HTTPS.  If the
   server does not set the Secure attribute on its cookies, an active
   network attacker can intercept any outbound HTTP request from the
   user agent and redirect that request to the webmail server over HTTP.
   Even if the webmail server is not listening for HTTP connections, the
   user agent will still include cookies in the request.  The active
   network attacker can intercept these cookies, replay them against the
   server, and learn the contents of the user's email.  If, instead, the
   server had set the Secure attribute on its cookies, the user agent
   would not have included the cookies in the clear-text request.

8.4.  Session Identifiers



   Instead of storing session information directly in a cookie (where it
   might be exposed to or replayed by an attacker), servers commonly
   store a nonce (or "session identifier") in a cookie.  When the server
   receives an HTTP request with a nonce, the server can look up state
   information associated with the cookie using the nonce as a key.

   Using session identifier cookies limits the damage an attacker can
   cause if the attacker learns the contents of a cookie because the
   nonce is useful only for interacting with the server (unlike non-
   nonce cookie content, which might itself be sensitive).  Furthermore,
   using a single nonce prevents an attacker from "splicing" together
   cookie content from two interactions with the server, which could
   cause the server to behave unexpectedly.

   Using session identifiers is not without risk.  For example, the
   server SHOULD take care to avoid "session fixation" vulnerabilities.
   A session fixation attack proceeds in three steps.  First, the
   attacker transplants a session identifier from his or her user agent
   to the victim's user agent.  Second, the victim uses that session



Barth                        Standards Track                   [Page 31]

RFC 6265             HTTP State Management Mechanism          April 2011


   identifier to interact with the server, possibly imbuing the session
   identifier with the user's credentials or confidential information.
   Third, the attacker uses the session identifier to interact with
   server directly, possibly obtaining the user's authority or
   confidential information.

8.5.  Weak Confidentiality



   Cookies do not provide isolation by port.  If a cookie is readable by
   a service running on one port, the cookie is also readable by a
   service running on another port of the same server.  If a cookie is
   writable by a service on one port, the cookie is also writable by a
   service running on another port of the same server.  For this reason,
   servers SHOULD NOT both run mutually distrusting services on
   different ports of the same host and use cookies to store security-
   sensitive information.

   Cookies do not provide isolation by scheme.  Although most commonly
   used with the http and https schemes, the cookies for a given host
   might also be available to other schemes, such as ftp and gopher.
   Although this lack of isolation by scheme is most apparent in non-
   HTTP APIs that permit access to cookies (e.g., HTML's document.cookie
   API), the lack of isolation by scheme is actually present in
   requirements for processing cookies themselves (e.g., consider
   retrieving a URI with the gopher scheme via HTTP).

   Cookies do not always provide isolation by path.  Although the
   network-level protocol does not send cookies stored for one path to
   another, some user agents expose cookies via non-HTTP APIs, such as
   HTML's document.cookie API.  Because some of these user agents (e.g.,
   web browsers) do not isolate resources received from different paths,
   a resource retrieved from one path might be able to access cookies
   stored for another path.

8.6.  Weak Integrity



   Cookies do not provide integrity guarantees for sibling domains (and
   their subdomains).  For example, consider foo.example.com and
   bar.example.com.  The foo.example.com server can set a cookie with a
   Domain attribute of "example.com" (possibly overwriting an existing
   "example.com" cookie set by bar.example.com), and the user agent will
   include that cookie in HTTP requests to bar.example.com.  In the
   worst case, bar.example.com will be unable to distinguish this cookie
   from a cookie it set itself.  The foo.example.com server might be
   able to leverage this ability to mount an attack against
   bar.example.com.





Barth                        Standards Track                   [Page 32]

RFC 6265             HTTP State Management Mechanism          April 2011


   Even though the Set-Cookie header supports the Path attribute, the
   Path attribute does not provide any integrity protection because the
   user agent will accept an arbitrary Path attribute in a Set-Cookie
   header.  For example, an HTTP response to a request for
   http://example.com/foo/bar can set a cookie with a Path attribute of
   "/qux".  Consequently, servers SHOULD NOT both run mutually
   distrusting services on different paths of the same host and use
   cookies to store security-sensitive information.

   An active network attacker can also inject cookies into the Cookie
   header sent to https://example.com/ by impersonating a response from
   http://example.com/ and injecting a Set-Cookie header.  The HTTPS
   server at example.com will be unable to distinguish these cookies
   from cookies that it set itself in an HTTPS response.  An active
   network attacker might be able to leverage this ability to mount an
   attack against example.com even if example.com uses HTTPS
   exclusively.

   Servers can partially mitigate these attacks by encrypting and
   signing the contents of their cookies.  However, using cryptography
   does not mitigate the issue completely because an attacker can replay
   a cookie he or she received from the authentic example.com server in
   the user's session, with unpredictable results.

   Finally, an attacker might be able to force the user agent to delete
   cookies by storing a large number of cookies.  Once the user agent
   reaches its storage limit, the user agent will be forced to evict
   some cookies.  Servers SHOULD NOT rely upon user agents retaining
   cookies.

8.7.  Reliance on DNS



   Cookies rely upon the Domain Name System (DNS) for security.  If the
   DNS is partially or fully compromised, the cookie protocol might fail
   to provide the security properties required by applications.

9.  IANA Considerations



   The permanent message header field registry (see [RFC3864]) has been
   updated with the following registrations.











Barth                        Standards Track                   [Page 33]

RFC 6265             HTTP State Management Mechanism          April 2011


9.1.  Cookie



   Header field name: Cookie

   Applicable protocol: http

   Status: standard

   Author/Change controller: IETF

   Specification document: this specification (Section 5.4)

9.2.  Set-Cookie



   Header field name: Set-Cookie

   Applicable protocol: http

   Status: standard

   Author/Change controller: IETF

   Specification document: this specification (Section 5.2)

9.3.  Cookie2



   Header field name: Cookie2

   Applicable protocol: http

   Status: obsoleted

   Author/Change controller: IETF

   Specification document: [RFC2965]

9.4.  Set-Cookie2



   Header field name: Set-Cookie2

   Applicable protocol: http

   Status: obsoleted

   Author/Change controller: IETF

   Specification document: [RFC2965]




Barth                        Standards Track                   [Page 34]

RFC 6265             HTTP State Management Mechanism          April 2011


10.  References

10.1.  Normative References



   [RFC1034]  Mockapetris, P., "Domain names - concepts and facilities",
              STD 13, RFC 1034, November 1987.

   [RFC1123]  Braden, R., "Requirements for Internet Hosts - Application
              and Support", STD 3, RFC 1123, October 1989.

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119, March 1997.

   [RFC2616]  Fielding, R., Gettys, J., Mogul, J., Frystyk, H.,
              Masinter, L., Leach, P., and T. Berners-Lee, "Hypertext
              Transfer Protocol -- HTTP/1.1", RFC 2616, June 1999.

   [RFC3490]  Faltstrom, P., Hoffman, P., and A. Costello,
              "Internationalizing Domain Names in Applications (IDNA)",
              RFC 3490, March 2003.

              See Section 6.3 for an explanation why the normative
              reference to an obsoleted specification is needed.

   [RFC4790]  Newman, C., Duerst, M., and A. Gulbrandsen, "Internet
              Application Protocol Collation Registry", RFC 4790,
              March 2007.

   [RFC5234]  Crocker, D., Ed. and P. Overell, "Augmented BNF for Syntax
              Specifications: ABNF", STD 68, RFC 5234, January 2008.

   [RFC5890]  Klensin, J., "Internationalized Domain Names for
              Applications (IDNA): Definitions and Document Framework",
              RFC 5890, August 2010.

   [USASCII]  American National Standards Institute, "Coded Character
              Set -- 7-bit American Standard Code for Information
              Interchange", ANSI X3.4, 1986.

10.2.  Informative References



   [RFC2109]  Kristol, D. and L. Montulli, "HTTP State Management
              Mechanism", RFC 2109, February 1997.

   [RFC2965]  Kristol, D. and L. Montulli, "HTTP State Management
              Mechanism", RFC 2965, October 2000.





Barth                        Standards Track                   [Page 35]

RFC 6265             HTTP State Management Mechanism          April 2011


   [RFC2818]  Rescorla, E., "HTTP Over TLS", RFC 2818, May 2000.

   [Netscape] Netscape Communications Corp., "Persistent Client State --
              HTTP Cookies", 1999, <http://web.archive.org/web/
              20020803110822/http://wp.netscape.com/newsref/std/
              cookie_spec.html>.

   [Kri2001]  Kristol, D., "HTTP Cookies: Standards, Privacy, and
              Politics", ACM Transactions on Internet Technology Vol. 1,
              #2, November 2001, <http://arxiv.org/abs/cs.SE/0105018>.

   [RFC3629]  Yergeau, F., "UTF-8, a transformation format of ISO
              10646", STD 63, RFC 3629, November 2003.

   [RFC4648]  Josefsson, S., "The Base16, Base32, and Base64 Data
              Encodings", RFC 4648, October 2006.

   [RFC3864]  Klyne, G., Nottingham, M., and J. Mogul, "Registration
              Procedures for Message Header Fields", BCP 90, RFC 3864,
              September 2004.

   [RFC5895]  Resnick, P. and P. Hoffman, "Mapping Characters for
              Internationalized Domain Names in Applications (IDNA)
              2008", RFC 5895, September 2010.

   [UTS46]    Davis, M. and M. Suignard, "Unicode IDNA Compatibility
              Processing", Unicode Technical Standards # 46, 2010,
              <http://unicode.org/reports/tr46/>.

   [CSRF]     Barth, A., Jackson, C., and J. Mitchell, "Robust Defenses
              for Cross-Site Request Forgery", 2008,
              <http://portal.acm.org/citation.cfm?id=1455770.1455782>.

   [Aggarwal2010]
              Aggarwal, G., Burzstein, E., Jackson, C., and D. Boneh,
              "An Analysis of Private Browsing Modes in Modern
              Browsers", 2010, <http://www.usenix.org/events/sec10/tech/
              full_papers/Aggarwal.pdf>.













Barth                        Standards Track                   [Page 36]

RFC 6265             HTTP State Management Mechanism          April 2011


Appendix A.  Acknowledgements



   This document borrows heavily from RFC 2109 [RFC2109].  We are
   indebted to David M. Kristol and Lou Montulli for their efforts to
   specify cookies.  David M. Kristol, in particular, provided
   invaluable advice on navigating the IETF process.  We would also like
   to thank Thomas Broyer, Tyler Close, Alissa Cooper, Bil Corry,
   corvid, Lisa Dusseault, Roy T. Fielding, Blake Frantz, Anne van
   Kesteren, Eran Hammer-Lahav, Jeff Hodges, Bjoern Hoehrmann, Achim
   Hoffmann, Georg Koppen, Dean McNamee, Alexey Melnikov, Mark Miller,
   Mark Pauley, Yngve N. Pettersen, Julian Reschke, Peter Saint-Andre,
   Mark Seaborn, Maciej Stachowiak, Daniel Stenberg, Tatsuhiro
   Tsujikawa, David Wagner, Dan Winship, and Dan Witte for their
   valuable feedback on this document.

Author's Address



   Adam Barth
   University of California, Berkeley

   EMail: abarth@eecs.berkeley.edu
   URI:   http://www.adambarth.com/





























Barth                        Standards Track                   [Page 37]