RFC 6376

Internet Engineering Task Force (IETF)                   D. Crocker, Ed.
Request for Comments: 6376                   Brandenburg InternetWorking
Obsoletes: 4871, 5672                                     T. Hansen, Ed.
Category: Standards Track                              AT&T Laboratories
ISSN: 2070-1721                                        M. Kucherawy, Ed.
                                                          September 2011

              DomainKeys Identified Mail (DKIM) Signatures


   DomainKeys Identified Mail (DKIM) permits a person, role, or
   organization that owns the signing domain to claim some
   responsibility for a message by associating the domain with the
   message.  This can be an author's organization, an operational relay,
   or one of their agents.  DKIM separates the question of the identity
   of the Signer of the message from the purported author of the
   message.  Assertion of responsibility is validated through a
   cryptographic signature and by querying the Signer's domain directly
   to retrieve the appropriate public key.  Message transit from author
   to recipient is through relays that typically make no substantive
   change to the message content and thus preserve the DKIM signature.

   This memo obsoletes RFC 4871 and RFC 5672.

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

Copyright Notice

   Copyright (c) 2011 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

Crocker, et al.              Standards Track                    [Page 1]

RFC 6376                     DKIM Signatures              September 2011

   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.

   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 . . . . . . . . . . . . . . . . . . . . . . . . .  4
     1.1.  DKIM Architecture Documents  . . . . . . . . . . . . . . .  5
     1.2.  Signing Identity . . . . . . . . . . . . . . . . . . . . .  5
     1.3.  Scalability  . . . . . . . . . . . . . . . . . . . . . . .  5
     1.4.  Simple Key Management  . . . . . . . . . . . . . . . . . .  6
     1.5.  Data Integrity . . . . . . . . . . . . . . . . . . . . . .  6
   2.  Terminology and Definitions  . . . . . . . . . . . . . . . . .  6
     2.1.  Signers  . . . . . . . . . . . . . . . . . . . . . . . . .  6
     2.2.  Verifiers  . . . . . . . . . . . . . . . . . . . . . . . .  7
     2.3.  Identity . . . . . . . . . . . . . . . . . . . . . . . . .  7
     2.4.  Identifier . . . . . . . . . . . . . . . . . . . . . . . .  7
     2.5.  Signing Domain Identifier (SDID) . . . . . . . . . . . . .  7
     2.6.  Agent or User Identifier (AUID)  . . . . . . . . . . . . .  7
     2.7.  Identity Assessor  . . . . . . . . . . . . . . . . . . . .  7
     2.8.  Whitespace . . . . . . . . . . . . . . . . . . . . . . . .  8
     2.9.  Imported ABNF Tokens . . . . . . . . . . . . . . . . . . .  8
     2.10. Common ABNF Tokens . . . . . . . . . . . . . . . . . . . .  9
     2.11. DKIM-Quoted-Printable  . . . . . . . . . . . . . . . . . .  9
   3.  Protocol Elements  . . . . . . . . . . . . . . . . . . . . . . 10
     3.1.  Selectors  . . . . . . . . . . . . . . . . . . . . . . . . 10
     3.2.  Tag=Value Lists  . . . . . . . . . . . . . . . . . . . . . 12
     3.3.  Signing and Verification Algorithms  . . . . . . . . . . . 13
     3.4.  Canonicalization . . . . . . . . . . . . . . . . . . . . . 14
     3.5.  The DKIM-Signature Header Field  . . . . . . . . . . . . . 18

Crocker, et al.              Standards Track                    [Page 2]

RFC 6376                     DKIM Signatures              September 2011

     3.6.  Key Management and Representation  . . . . . . . . . . . . 26
     3.7.  Computing the Message Hashes . . . . . . . . . . . . . . . 29
     3.8.  Input Requirements . . . . . . . . . . . . . . . . . . . . 32
     3.9.  Output Requirements  . . . . . . . . . . . . . . . . . . . 32
     3.10. Signing by Parent Domains  . . . . . . . . . . . . . . . . 33
     3.11. Relationship between SDID and AUID . . . . . . . . . . . . 33
   4.  Semantics of Multiple Signatures . . . . . . . . . . . . . . . 34
     4.1.  Example Scenarios  . . . . . . . . . . . . . . . . . . . . 34
     4.2.  Interpretation . . . . . . . . . . . . . . . . . . . . . . 35
   5.  Signer Actions . . . . . . . . . . . . . . . . . . . . . . . . 36
     5.1.  Determine Whether the Email Should Be Signed and by
           Whom . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
     5.2.  Select a Private Key and Corresponding Selector
           Information  . . . . . . . . . . . . . . . . . . . . . . . 37
     5.3.  Normalize the Message to Prevent Transport Conversions . . 37
     5.4.  Determine the Header Fields to Sign  . . . . . . . . . . . 38
     5.5.  Compute the Message Hash and Signature . . . . . . . . . . 43
     5.6.  Insert the DKIM-Signature Header Field . . . . . . . . . . 43
   6.  Verifier Actions . . . . . . . . . . . . . . . . . . . . . . . 43
     6.1.  Extract Signatures from the Message  . . . . . . . . . . . 44
     6.2.  Communicate Verification Results . . . . . . . . . . . . . 49
     6.3.  Interpret Results/Apply Local Policy . . . . . . . . . . . 50
   7.  IANA Considerations  . . . . . . . . . . . . . . . . . . . . . 51
     7.1.  Email Authentication Methods Registry  . . . . . . . . . . 51
     7.2.  DKIM-Signature Tag Specifications  . . . . . . . . . . . . 51
     7.3.  DKIM-Signature Query Method Registry . . . . . . . . . . . 52
     7.4.  DKIM-Signature Canonicalization Registry . . . . . . . . . 52
     7.5.  _domainkey DNS TXT Resource Record Tag Specifications  . . 53
     7.6.  DKIM Key Type Registry . . . . . . . . . . . . . . . . . . 53
     7.7.  DKIM Hash Algorithms Registry  . . . . . . . . . . . . . . 54
     7.8.  DKIM Service Types Registry  . . . . . . . . . . . . . . . 54
     7.9.  DKIM Selector Flags Registry . . . . . . . . . . . . . . . 55
     7.10. DKIM-Signature Header Field  . . . . . . . . . . . . . . . 55
   8.  Security Considerations  . . . . . . . . . . . . . . . . . . . 55
     8.1.  ASCII Art Attacks  . . . . . . . . . . . . . . . . . . . . 55
     8.2.  Misuse of Body Length Limits ("l=" Tag)  . . . . . . . . . 55
     8.3.  Misappropriated Private Key  . . . . . . . . . . . . . . . 56
     8.4.  Key Server Denial-of-Service Attacks . . . . . . . . . . . 56
     8.5.  Attacks against the DNS  . . . . . . . . . . . . . . . . . 57
     8.6.  Replay/Spam Attacks  . . . . . . . . . . . . . . . . . . . 57
     8.7.  Limits on Revoking Keys  . . . . . . . . . . . . . . . . . 58
     8.8.  Intentionally Malformed Key Records  . . . . . . . . . . . 58
     8.9.  Intentionally Malformed DKIM-Signature Header Fields . . . 58
     8.10. Information Leakage  . . . . . . . . . . . . . . . . . . . 58
     8.11. Remote Timing Attacks  . . . . . . . . . . . . . . . . . . 59
     8.12. Reordered Header Fields  . . . . . . . . . . . . . . . . . 59
     8.13. RSA Attacks  . . . . . . . . . . . . . . . . . . . . . . . 59
     8.14. Inappropriate Signing by Parent Domains  . . . . . . . . . 59

Crocker, et al.              Standards Track                    [Page 3]

RFC 6376                     DKIM Signatures              September 2011

     8.15. Attacks Involving Extra Header Fields  . . . . . . . . . . 60
   9.  References . . . . . . . . . . . . . . . . . . . . . . . . . . 61
     9.1.  Normative References . . . . . . . . . . . . . . . . . . . 61
     9.2.  Informative References . . . . . . . . . . . . . . . . . . 62
   Appendix A.  Example of Use (INFORMATIVE)  . . . . . . . . . . . . 64
     A.1.  The User Composes an Email . . . . . . . . . . . . . . . . 64
     A.2.  The Email is Signed  . . . . . . . . . . . . . . . . . . . 65
     A.3.  The Email Signature is Verified  . . . . . . . . . . . . . 66
   Appendix B.  Usage Examples (INFORMATIVE)  . . . . . . . . . . . . 67
     B.1.  Alternate Submission Scenarios . . . . . . . . . . . . . . 67
     B.2.  Alternate Delivery Scenarios . . . . . . . . . . . . . . . 69
   Appendix C.  Creating a Public Key (INFORMATIVE) . . . . . . . . . 71
     C.1.  Compatibility with DomainKeys Key Records  . . . . . . . . 72
     C.2.  RFC 4871 Compatibility . . . . . . . . . . . . . . . . . . 73
   Appendix D.  MUA Considerations (INFORMATIVE)  . . . . . . . . . . 73
   Appendix E.  Changes since RFC 4871  . . . . . . . . . . . . . . . 73
   Appendix F.  Acknowledgments . . . . . . . . . . . . . . . . . . . 75

1.  Introduction

   DomainKeys Identified Mail (DKIM) permits a person, role, or
   organization to claim some responsibility for a message by
   associating a domain name [RFC1034] with the message [RFC5322], which
   they are authorized to use.  This can be an author's organization, an
   operational relay, or one of their agents.  Assertion of
   responsibility is validated through a cryptographic signature and by
   querying the Signer's domain directly to retrieve the appropriate
   public key.  Message transit from author to recipient is through
   relays that typically make no substantive change to the message
   content and thus preserve the DKIM signature.  A message can contain
   multiple signatures, from the same or different organizations
   involved with the message.

   The approach taken by DKIM differs from previous approaches to
   message signing (e.g., Secure/Multipurpose Internet Mail Extensions
   (S/MIME) [RFC5751], OpenPGP [RFC4880]) in that:

   o  the message signature is written as a message header field so that
      neither human recipients nor existing MUA (Mail User Agent)
      software is confused by signature-related content appearing in the
      message body;

   o  there is no dependency on public- and private-key pairs being
      issued by well-known, trusted certificate authorities;

   o  there is no dependency on the deployment of any new Internet
      protocols or services for public-key distribution or revocation;

Crocker, et al.              Standards Track                    [Page 4]

RFC 6376                     DKIM Signatures              September 2011

   o  signature verification failure does not force rejection of the

   o  no attempt is made to include encryption as part of the mechanism;

   o  message archiving is not a design goal.


   o  is compatible with the existing email infrastructure and
      transparent to the fullest extent possible;

   o  requires minimal new infrastructure;

   o  can be implemented independently of clients in order to reduce
      deployment time;

   o  can be deployed incrementally; and

   o  allows delegation of signing to third parties.

1.1.  DKIM Architecture Documents

   Readers are advised to be familiar with the material in [RFC4686],
   [RFC5585], and [RFC5863], which provide the background for the
   development of DKIM, an overview of the service, and deployment and
   operations guidance and advice, respectively.

1.2.  Signing Identity

   DKIM separates the question of the identity of the Signer of the
   message from the purported author of the message.  In particular, a
   signature includes the identity of the Signer.  Verifiers can use the
   signing information to decide how they want to process the message.
   The signing identity is included as part of the signature header

      INFORMATIVE RATIONALE: The signing identity specified by a DKIM
      signature is not required to match an address in any particular
      header field because of the broad methods of interpretation by
      recipient mail systems, including MUAs.

1.3.  Scalability

   DKIM is designed to support the extreme scalability requirements that
   characterize the email identification problem.  There are many
   millions of domains and a much larger number of individual addresses.

Crocker, et al.              Standards Track                    [Page 5]

RFC 6376                     DKIM Signatures              September 2011

   DKIM seeks to preserve the positive aspects of the current email
   infrastructure, such as the ability for anyone to communicate with
   anyone else without introduction.

1.4.  Simple Key Management

   DKIM differs from traditional hierarchical public-key systems in that
   no certificate authority infrastructure is required; the Verifier
   requests the public key from a repository in the domain of the
   claimed Signer directly rather than from a third party.

   The DNS is proposed as the initial mechanism for the public keys.
   Thus, DKIM currently depends on DNS administration and the security
   of the DNS system.  DKIM is designed to be extensible to other key
   fetching services as they become available.

1.5.  Data Integrity

   A DKIM signature associates the "d=" name with the computed hash of
   some or all of the message (see Section 3.7) in order to prevent the
   reuse of the signature with different messages.  Verifying the
   signature asserts that the hashed content has not changed since it
   was signed and asserts nothing else about "protecting" the end-to-end
   integrity of the message.

2.  Terminology and Definitions

   This section defines terms used in the rest of the document.

   DKIM is designed to operate within the Internet Mail service, as
   defined in [RFC5598].  Basic email terminology is taken from that

   Syntax descriptions use Augmented BNF (ABNF) [RFC5234].

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "OPTIONAL" in this document are to be interpreted as described in
   [RFC2119].  These words take their normative meanings only when they
   are presented in ALL UPPERCASE.

2.1.  Signers

   Elements in the mail system that sign messages on behalf of a domain
   are referred to as Signers.  These may be MUAs (Mail User Agents),
   MSAs (Mail Submission Agents), MTAs (Mail Transfer Agents), or other
   agents such as mailing list exploders.  In general, any Signer will

Crocker, et al.              Standards Track                    [Page 6]

RFC 6376                     DKIM Signatures              September 2011

   be involved in the injection of a message into the message system in
   some way.  The key issue is that a message must be signed before it
   leaves the administrative domain of the Signer.

2.2.  Verifiers

   Elements in the mail system that verify signatures are referred to as
   Verifiers.  These may be MTAs, Mail Delivery Agents (MDAs), or MUAs.
   In most cases, it is expected that Verifiers will be close to an end
   user (reader) of the message or some consuming agent such as a
   mailing list exploder.

2.3.  Identity

   A person, role, or organization.  In the context of DKIM, examples
   include the author, the author's organization, an ISP along the
   handling path, an independent trust assessment service, and a mailing
   list operator.

2.4.  Identifier

   A label that refers to an identity.

2.5.  Signing Domain Identifier (SDID)

   A single domain name that is the mandatory payload output of DKIM and
   that refers to the identity claiming some responsibility for the
   message by signing it.  It is specified in Section 3.5.

2.6.  Agent or User Identifier (AUID)

   A single identifier that refers to the agent or user on behalf of
   whom the Signing Domain Identifier (SDID) has taken responsibility.
   The AUID comprises a domain name and an optional <local-part>.  The
   domain name is the same as that used for the SDID or is a subdomain
   of it.  For DKIM processing, the domain name portion of the AUID has
   only basic domain name semantics; any possible owner-specific
   semantics are outside the scope of DKIM.  It is specified in
   Section 3.5.

   Note that acceptable values for the AUID may be constrained via a
   flag in the public-key record.  (See Section 3.6.1.)

2.7.  Identity Assessor

   An element in the mail system that consumes DKIM's payload, which is
   the responsible Signing Domain Identifier (SDID).  The Identity
   Assessor is dedicated to the assessment of the delivered identifier.

Crocker, et al.              Standards Track                    [Page 7]

RFC 6376                     DKIM Signatures              September 2011

   Other DKIM (and non-DKIM) values can also be used by the Identity
   Assessor (if they are available) to provide a more general message
   evaluation filtering engine.  However, this additional activity is
   outside the scope of this specification.

2.8.  Whitespace

   There are three forms of whitespace:

   o  WSP represents simple whitespace, i.e., a space or a tab character
      (formal definition in [RFC5234]).

   o  LWSP is linear whitespace, defined as WSP plus CRLF (formal
      definition in [RFC5234]).

   o  FWS is folding whitespace.  It allows multiple lines separated by
      CRLF followed by at least one whitespace, to be joined.

   The formal ABNF for these are (WSP and LWSP are given for information

   WSP =   SP / HTAB
   LWSP =  *(WSP / CRLF WSP)
   FWS =   [*WSP CRLF] 1*WSP

   The definition of FWS is identical to that in [RFC5322] except for
   the exclusion of obs-FWS.

2.9.  Imported ABNF Tokens

   The following tokens are imported from other RFCs as noted.  Those
   RFCs should be considered definitive.

   The following tokens are imported from [RFC5321]:

   o  "local-part" (implementation warning: this permits quoted strings)

   o  "sub-domain"

   The following tokens are imported from [RFC5322]:

   o  "field-name" (name of a header field)

   o  "dot-atom-text" (in the local-part of an email address)

   The following tokens are imported from [RFC2045]:

   o  "qp-section" (a single line of quoted-printable-encoded text)

Crocker, et al.              Standards Track                    [Page 8]

RFC 6376                     DKIM Signatures              September 2011

   o  "hex-octet" (a quoted-printable encoded octet)

      INFORMATIVE NOTE: Be aware that the ABNF in [RFC2045] does not
      obey the rules of [RFC5234] and must be interpreted accordingly,
      particularly as regards case folding.

   Other tokens not defined herein are imported from [RFC5234].  These
   are intuitive primitives such as SP, HTAB, WSP, ALPHA, DIGIT, CRLF,

2.10.  Common ABNF Tokens

   The following ABNF tokens are used elsewhere in this document:

   hyphenated-word =  ALPHA [ *(ALPHA / DIGIT / "-") (ALPHA / DIGIT) ]
   ALPHADIGITPS    =  (ALPHA / DIGIT / "+" / "/")
   base64string    =  ALPHADIGITPS *([FWS] ALPHADIGITPS)
                      [ [FWS] "=" [ [FWS] "=" ] ]
   hdr-name        =  field-name
   qp-hdr-value    =  dkim-quoted-printable    ; with "|" encoded

2.11.  DKIM-Quoted-Printable

   The DKIM-Quoted-Printable encoding syntax resembles that described in
   Quoted-Printable [RFC2045], Section 6.7: any character MAY be encoded
   as an "=" followed by two hexadecimal digits from the alphabet
   "0123456789ABCDEF" (no lowercase characters permitted) representing
   the hexadecimal-encoded integer value of that character.  All control
   characters (those with values < %x20), 8-bit characters (values >
   %x7F), and the characters DEL (%x7F), SPACE (%x20), and semicolon
   (";", %x3B) MUST be encoded.  Note that all whitespace, including
   SPACE, CR, and LF characters, MUST be encoded.  After encoding, FWS
   MAY be added at arbitrary locations in order to avoid excessively
   long lines; such whitespace is NOT part of the value, and MUST be
   removed before decoding.  Use of characters not listed as "mail-safe"
   in [RFC2049] is NOT RECOMMENDED.


   dkim-quoted-printable =  *(FWS / hex-octet / dkim-safe-char)
                               ; hex-octet is from RFC2045
   dkim-safe-char        =  %x21-3A / %x3C / %x3E-7E
                               ; '!' - ':', '<', '>' - '~'

Crocker, et al.              Standards Track                    [Page 9]

RFC 6376                     DKIM Signatures              September 2011

      INFORMATIVE NOTE: DKIM-Quoted-Printable differs from Quoted-
      Printable as defined in [RFC2045] in several important ways:

      1.  Whitespace in the input text, including CR and LF, must be
          encoded.  [RFC2045] does not require such encoding, and does
          not permit encoding of CR or LF characters that are part of a
          CRLF line break.

      2.  Whitespace in the encoded text is ignored.  This is to allow
          tags encoded using DKIM-Quoted-Printable to be wrapped as
          needed.  In particular, [RFC2045] requires that line breaks in
          the input be represented as physical line breaks; that is not
          the case here.

      3.  The "soft line break" syntax ("=" as the last non-whitespace
          character on the line) does not apply.

      4.  DKIM-Quoted-Printable does not require that encoded lines be
          no more than 76 characters long (although there may be other
          requirements depending on the context in which the encoded
          text is being used).

3.  Protocol Elements

   Protocol Elements are conceptual parts of the protocol that are not
   specific to either Signers or Verifiers.  The protocol descriptions
   for Signers and Verifiers are described in later sections ("Signer
   Actions" (Section 5) and "Verifier Actions" (Section 6)).  NOTE: This
   section must be read in the context of those sections.

3.1.  Selectors

   To support multiple concurrent public keys per signing domain, the
   key namespace is subdivided using "selectors".  For example,
   selectors might indicate the names of office locations (e.g.,
   "sanfrancisco", "coolumbeach", and "reykjavik"), the signing date
   (e.g., "january2005", "february2005", etc.), or even an individual

   Selectors are needed to support some important use cases.  For

   o  Domains that want to delegate signing capability for a specific
      address for a given duration to a partner, such as an advertising
      provider or other outsourced function.

   o  Domains that want to allow frequent travelers to send messages
      locally without the need to connect with a particular MSA.

Crocker, et al.              Standards Track                   [Page 10]

RFC 6376                     DKIM Signatures              September 2011

   o  "Affinity" domains (e.g., college alumni associations) that
      provide forwarding of incoming mail, but that do not operate a
      mail submission agent for outgoing mail.

   Periods are allowed in selectors and are component separators.  When
   keys are retrieved from the DNS, periods in selectors define DNS
   label boundaries in a manner similar to the conventional use in
   domain names.  Selector components might be used to combine dates
   with locations, for example, "march2005.reykjavik".  In a DNS
   implementation, this can be used to allow delegation of a portion of
   the selector namespace.


   selector =   sub-domain *( "." sub-domain )

   The number of public keys and corresponding selectors for each domain
   is determined by the domain owner.  Many domain owners will be
   satisfied with just one selector, whereas administratively
   distributed organizations can choose to manage disparate selectors
   and key pairs in different regions or on different email servers.

   Beyond administrative convenience, selectors make it possible to
   seamlessly replace public keys on a routine basis.  If a domain
   wishes to change from using a public key associated with selector
   "january2005" to a public key associated with selector
   "february2005", it merely makes sure that both public keys are
   advertised in the public-key repository concurrently for the
   transition period during which email may be in transit prior to
   verification.  At the start of the transition period, the outbound
   email servers are configured to sign with the "february2005" private
   key.  At the end of the transition period, the "january2005" public
   key is removed from the public-key repository.

      INFORMATIVE NOTE: A key may also be revoked as described below.
      The distinction between revoking and removing a key selector
      record is subtle.  When phasing out keys as described above, a
      signing domain would probably simply remove the key record after
      the transition period.  However, a signing domain could elect to
      revoke the key (but maintain the key record) for a further period.
      There is no defined semantic difference between a revoked key and
      a removed key.

   While some domains may wish to make selector values well-known,
   others will want to take care not to allocate selector names in a way
   that allows harvesting of data by outside parties.  For example, if
   per-user keys are issued, the domain owner will need to decide

Crocker, et al.              Standards Track                   [Page 11]

RFC 6376                     DKIM Signatures              September 2011

   whether to associate this selector directly with the name of a
   registered end user or make it some unassociated random value, such
   as a fingerprint of the public key.

      INFORMATIVE OPERATIONS NOTE: Reusing a selector with a new key
      (for example, changing the key associated with a user's name)
      makes it impossible to tell the difference between a message that
      didn't verify because the key is no longer valid and a message
      that is actually forged.  For this reason, Signers are ill-advised
      to reuse selectors for new keys.  A better strategy is to assign
      new keys to new selectors.

3.2.  Tag=Value Lists

   DKIM uses a simple "tag=value" syntax in several contexts, including
   in messages and domain signature records.

   Values are a series of strings containing either plain text, "base64"
   text (as defined in [RFC2045], Section 6.8), "qp-section" (ibid,
   Section 6.7), or "dkim-quoted-printable" (as defined in
   2.11">Section 2.11).  The name of the tag will determine the encoding of
   each value.  Unencoded semicolon (";") characters MUST NOT occur in
   the tag value, since that separates tag-specs.

      INFORMATIVE IMPLEMENTATION NOTE: Although the "plain text" defined
      below (as "tag-value") only includes 7-bit characters, an
      implementation that wished to anticipate future standards would be
      advised not to preclude the use of UTF-8-encoded ([RFC3629]) text
      in tag=value lists.

   Formally, the ABNF syntax rules are as follows:

   tag-list  =  tag-spec *( ";" tag-spec ) [ ";" ]
   tag-spec  =  [FWS] tag-name [FWS] "=" [FWS] tag-value [FWS]
   tag-name  =  ALPHA *ALNUMPUNC
   tag-value =  [ tval *( 1*(WSP / FWS) tval ) ]
                     ; Prohibits WSP and FWS at beginning and end
   tval      =  1*VALCHAR
   VALCHAR   =  %x21-3A / %x3C-7E
                     ; EXCLAMATION to TILDE except SEMICOLON

   Note that WSP is allowed anywhere around tags.  In particular, any
   WSP after the "=" and any WSP before the terminating ";" is not part
   of the value; however, WSP inside the value is significant.

Crocker, et al.              Standards Track                   [Page 12]

RFC 6376                     DKIM Signatures              September 2011

   Tags MUST be interpreted in a case-sensitive manner.  Values MUST be
   processed as case sensitive unless the specific tag description of
   semantics specifies case insensitivity.

   Tags with duplicate names MUST NOT occur within a single tag-list; if
   a tag name does occur more than once, the entire tag-list is invalid.

   Whitespace within a value MUST be retained unless explicitly excluded
   by the specific tag description.

   Tag=value pairs that represent the default value MAY be included to
   aid legibility.

   Unrecognized tags MUST be ignored.

   Tags that have an empty value are not the same as omitted tags.  An
   omitted tag is treated as having the default value; a tag with an
   empty value explicitly designates the empty string as the value.

3.3.  Signing and Verification Algorithms

   DKIM supports multiple digital signature algorithms.  Two algorithms
   are defined by this specification at this time: rsa-sha1 and rsa-
   sha256.  Signers MUST implement and SHOULD sign using rsa-sha256.
   Verifiers MUST implement both rsa-sha1 and rsa-sha256.

      INFORMATIVE NOTE: Although rsa-sha256 is strongly encouraged, some
      senders might prefer to use rsa-sha1 when balancing security
      strength against performance, complexity, or other needs.  In
      general, however, rsa-sha256 should always be used whenever

3.3.1.  The rsa-sha1 Signing Algorithm

   The rsa-sha1 Signing Algorithm computes a message hash as described
   in Section 3.7 using SHA-1 [FIPS-180-3-2008] as the hash-alg.  That
   hash is then signed by the Signer using the RSA algorithm (defined in
   Public-Key Cryptography Standards (PKCS) #1 version 1.5 [RFC3447]) as
   the crypt-alg and the Signer's private key.  The hash MUST NOT be
   truncated or converted into any form other than the native binary
   form before being signed.  The signing algorithm SHOULD use a public
   exponent of 65537.

3.3.2.  The rsa-sha256 Signing Algorithm

   The rsa-sha256 Signing Algorithm computes a message hash as described
   in Section 3.7 using SHA-256 [FIPS-180-3-2008] as the hash-alg.  That
   hash is then signed by the Signer using the RSA algorithm (defined in

Crocker, et al.              Standards Track                   [Page 13]

RFC 6376                     DKIM Signatures              September 2011

   PKCS#1 version 1.5 [RFC3447]) as the crypt-alg and the Signer's
   private key.  The hash MUST NOT be truncated or converted into any
   form other than the native binary form before being signed.  The
   signing algorithm SHOULD use a public exponent of 65537.

3.3.3.  Key Sizes

   Selecting appropriate key sizes is a trade-off between cost,
   performance, and risk.  Since short RSA keys more easily succumb to
   off-line attacks, Signers MUST use RSA keys of at least 1024 bits for
   long-lived keys.  Verifiers MUST be able to validate signatures with
   keys ranging from 512 bits to 2048 bits, and they MAY be able to
   validate signatures with larger keys.  Verifier policies may use the
   length of the signing key as one metric for determining whether a
   signature is acceptable.

   Factors that should influence the key size choice include the

   o  The practical constraint that large (e.g., 4096-bit) keys might
      not fit within a 512-byte DNS UDP response packet

   o  The security constraint that keys smaller than 1024 bits are
      subject to off-line attacks

   o  Larger keys impose higher CPU costs to verify and sign email

   o  Keys can be replaced on a regular basis; thus, their lifetime can
      be relatively short

   o  The security goals of this specification are modest compared to
      typical goals of other systems that employ digital signatures

   See [RFC3766] for further discussion on selecting key sizes.

3.3.4.  Other Algorithms

   Other algorithms MAY be defined in the future.  Verifiers MUST ignore
   any signatures using algorithms that they do not implement.

3.4.  Canonicalization

   Some mail systems modify email in transit, potentially invalidating a
   signature.  For most Signers, mild modification of email is
   immaterial to validation of the DKIM domain name's use.  For such
   Signers, a canonicalization algorithm that survives modest in-transit
   modification is preferred.

Crocker, et al.              Standards Track                   [Page 14]

RFC 6376                     DKIM Signatures              September 2011

   Other Signers demand that any modification of the email, however
   minor, result in a signature verification failure.  These Signers
   prefer a canonicalization algorithm that does not tolerate in-transit
   modification of the signed email.

   Some Signers may be willing to accept modifications to header fields
   that are within the bounds of email standards such as [RFC5322], but
   are unwilling to accept any modification to the body of messages.

   To satisfy all requirements, two canonicalization algorithms are
   defined for each of the header and the body: a "simple" algorithm
   that tolerates almost no modification and a "relaxed" algorithm that
   tolerates common modifications such as whitespace replacement and
   header field line rewrapping.  A Signer MAY specify either algorithm
   for header or body when signing an email.  If no canonicalization
   algorithm is specified by the Signer, the "simple" algorithm defaults
   for both header and body.  Verifiers MUST implement both
   canonicalization algorithms.  Note that the header and body may use
   different canonicalization algorithms.  Further canonicalization
   algorithms MAY be defined in the future; Verifiers MUST ignore any
   signatures that use unrecognized canonicalization algorithms.

   Canonicalization simply prepares the email for presentation to the
   signing or verification algorithm.  It MUST NOT change the
   transmitted data in any way.  Canonicalization of header fields and
   body are described below.

   NOTE: This section assumes that the message is already in "network
   normal" format (text is ASCII encoded, lines are separated with CRLF
   characters, etc.).  See also Section 5.3 for information about
   normalizing the message.

3.4.1.  The "simple" Header Canonicalization Algorithm

   The "simple" header canonicalization algorithm does not change header
   fields in any way.  Header fields MUST be presented to the signing or
   verification algorithm exactly as they are in the message being
   signed or verified.  In particular, header field names MUST NOT be
   case folded and whitespace MUST NOT be changed.

3.4.2.  The "relaxed" Header Canonicalization Algorithm

   The "relaxed" header canonicalization algorithm MUST apply the
   following steps in order:

   o  Convert all header field names (not the header field values) to
      lowercase.  For example, convert "SUBJect: AbC" to "subject: AbC".

Crocker, et al.              Standards Track                   [Page 15]

RFC 6376                     DKIM Signatures              September 2011

   o  Unfold all header field continuation lines as described in
      [RFC5322]; in particular, lines with terminators embedded in
      continued header field values (that is, CRLF sequences followed by
      WSP) MUST be interpreted without the CRLF.  Implementations MUST
remove the CRLF at the end of the header field value.

   o  Convert all sequences of one or more WSP characters to a single SP
      character.  WSP characters here include those before and after a
      line folding boundary.

   o  Delete all WSP characters at the end of each unfolded header field

   o  Delete any WSP characters remaining before and after the colon
      separating the header field name from the header field value.  The
      colon separator MUST be retained.

3.4.3.  The "simple" Body Canonicalization Algorithm

   The "simple" body canonicalization algorithm ignores all empty lines
   at the end of the message body.  An empty line is a line of zero
   length after removal of the line terminator.  If there is no body or
   no trailing CRLF on the message body, a CRLF is added.  It makes no
   other changes to the message body.  In more formal terms, the
   "simple" body canonicalization algorithm converts "*CRLF" at the end
   of the body to a single "CRLF".

   Note that a completely empty or missing body is canonicalized as a
   single "CRLF"; that is, the canonicalized length will be 2 octets.

   The SHA-1 value (in base64) for an empty body (canonicalized to a
   "CRLF") is:


   The SHA-256 value is:


3.4.4.  The "relaxed" Body Canonicalization Algorithm

   The "relaxed" body canonicalization algorithm MUST apply the
   following steps (a) and (b) in order:

   a.  Reduce whitespace:

       *  Ignore all whitespace at the end of lines.  Implementations
          MUST NOT remove the CRLF at the end of the line.

Crocker, et al.              Standards Track                   [Page 16]

RFC 6376                     DKIM Signatures              September 2011

       *  Reduce all sequences of WSP within a line to a single SP

   b.  Ignore all empty lines at the end of the message body.  "Empty
       line" is defined in Section 3.4.3.  If the body is non-empty but
       does not end with a CRLF, a CRLF is added.  (For email, this is
       only possible when using extensions to SMTP or non-SMTP transport

   The SHA-1 value (in base64) for an empty body (canonicalized to a
   null input) is:


   The SHA-256 value is:


3.4.5.  Canonicalization Examples (INFORMATIVE)

   In the following examples, actual whitespace is used only for
   clarity.  The actual input and output text is designated using
   bracketed descriptors: "<SP>" for a space character, "<HTAB>" for a
   tab character, and "<CRLF>" for a carriage-return/line-feed sequence.
   For example, "X <SP> Y" and "X<SP>Y" represent the same three

   Example 1: A message reading:

   A: <SP> X <CRLF>
   B <SP> : <SP> Y <HTAB><CRLF>
                   <HTAB> Z <SP><SP><CRLF>
   <SP> C <SP><CRLF>
   D <SP><HTAB><SP> E <CRLF>

   when canonicalized using relaxed canonicalization for both header and
   body results in a header reading:

   a:X <CRLF>
   b:Y <SP> Z <CRLF>

   and a body reading:

   <SP> C <CRLF>
   D <SP> E <CRLF>

Crocker, et al.              Standards Track                   [Page 17]

RFC 6376                     DKIM Signatures              September 2011

   Example 2: The same message canonicalized using simple
   canonicalization for both header and body results in a header

   A: <SP> X <CRLF>
   B <SP> : <SP> Y <HTAB><CRLF>
          <HTAB> Z <SP><SP><CRLF>

   and a body reading:

   <SP> C <SP><CRLF>
   D <SP><HTAB><SP> E <CRLF>

   Example 3: When processed using relaxed header canonicalization and
   simple body canonicalization, the canonicalized version has a header

   a:X <CRLF>
   b:Y <SP> Z <CRLF>

   and a body reading:

   <SP> C <SP><CRLF>
   D <SP><HTAB><SP> E <CRLF>

3.5.  The DKIM-Signature Header Field

   The signature of the email is stored in the DKIM-Signature header
   field.  This header field contains all of the signature and key-
   fetching data.  The DKIM-Signature value is a tag-list as described
   in Section 3.2.

   The DKIM-Signature header field SHOULD be treated as though it were a
   trace header field as defined in Section 3.6 of [RFC5322] and hence
   SHOULD NOT be reordered and SHOULD be prepended to the message.

   The DKIM-Signature header field being created or verified is always
   included in the signature calculation, after the rest of the header
   fields being signed; however, when calculating or verifying the
   signature, the value of the "b=" tag (signature value) of that DKIM-
   Signature header field MUST be treated as though it were an empty
   string.  Unknown tags in the DKIM-Signature header field MUST be
   included in the signature calculation but MUST be otherwise ignored
   by Verifiers.  Other DKIM-Signature header fields that are included
   in the signature should be treated as normal header fields; in
   particular, the "b=" tag is not treated specially.

Crocker, et al.              Standards Track                   [Page 18]

RFC 6376                     DKIM Signatures              September 2011

   The encodings for each field type are listed below.  Tags described
   as qp-section are encoded as described in Section 6.7 of MIME Part
   One [RFC2045], with the additional conversion of semicolon characters
   to "=3B"; intuitively, this is one line of quoted-printable encoded
   text.  The dkim-quoted-printable syntax is defined in Section 2.11.

   Tags on the DKIM-Signature header field along with their type and
   requirement status are shown below.  Unrecognized tags MUST be

   v= Version (plain-text; REQUIRED).  This tag defines the version of
      this specification that applies to the signature record.  It MUST
      have the value "1" for implementations compliant with this version
      of DKIM.


      sig-v-tag       = %x76 [FWS] "=" [FWS] 1*DIGIT

         INFORMATIVE NOTE: DKIM-Signature version numbers may increase
         arithmetically as new versions of this specification are

   a= The algorithm used to generate the signature (plain-text;
      REQUIRED).  Verifiers MUST support "rsa-sha1" and "rsa-sha256";
      Signers SHOULD sign using "rsa-sha256".  See Section 3.3 for a
      description of the algorithms.


      sig-a-tag       = %x61 [FWS] "=" [FWS] sig-a-tag-alg
      sig-a-tag-alg   = sig-a-tag-k "-" sig-a-tag-h
      sig-a-tag-k     = "rsa" / x-sig-a-tag-k
      sig-a-tag-h     = "sha1" / "sha256" / x-sig-a-tag-h
      x-sig-a-tag-k   = ALPHA *(ALPHA / DIGIT)
                           ; for later extension
      x-sig-a-tag-h   = ALPHA *(ALPHA / DIGIT)
                           ; for later extension

   b= The signature data (base64; REQUIRED).  Whitespace is ignored in
      this value and MUST be ignored when reassembling the original
      signature.  In particular, the signing process can safely insert
      FWS in this value in arbitrary places to conform to line-length
      limits.  See "Signer Actions" (Section 5) for how the signature is

Crocker, et al.              Standards Track                   [Page 19]

RFC 6376                     DKIM Signatures              September 2011


      sig-b-tag       = %x62 [FWS] "=" [FWS] sig-b-tag-data
      sig-b-tag-data  = base64string

   bh=  The hash of the canonicalized body part of the message as
      limited by the "l=" tag (base64; REQUIRED).  Whitespace is ignored
      in this value and MUST be ignored when reassembling the original
      signature.  In particular, the signing process can safely insert
      FWS in this value in arbitrary places to conform to line-length
      limits.  See Section 3.7 for how the body hash is computed.


      sig-bh-tag      = %x62 %x68 [FWS] "=" [FWS] sig-bh-tag-data
      sig-bh-tag-data = base64string

   c= Message canonicalization (plain-text; OPTIONAL, default is
      "simple/simple").  This tag informs the Verifier of the type of
      canonicalization used to prepare the message for signing.  It
      consists of two names separated by a "slash" (%d47) character,
      corresponding to the header and body canonicalization algorithms,
      respectively.  These algorithms are described in Section 3.4.  If
      only one algorithm is named, that algorithm is used for the header
      and "simple" is used for the body.  For example, "c=relaxed" is
      treated the same as "c=relaxed/simple".


      sig-c-tag       = %x63 [FWS] "=" [FWS] sig-c-tag-alg
                        ["/" sig-c-tag-alg]
      sig-c-tag-alg   = "simple" / "relaxed" / x-sig-c-tag-alg
      x-sig-c-tag-alg = hyphenated-word    ; for later extension

   d= The SDID claiming responsibility for an introduction of a message
      into the mail stream (plain-text; REQUIRED).  Hence, the SDID
      value is used to form the query for the public key.  The SDID MUST
      correspond to a valid DNS name under which the DKIM key record is
      published.  The conventions and semantics used by a Signer to
      create and use a specific SDID are outside the scope of this
      specification, as is any use of those conventions and semantics.
      When presented with a signature that does not meet these
      requirements, Verifiers MUST consider the signature invalid.

      Internationalized domain names MUST be encoded as A-labels, as
      described in Section 2.3 of [RFC5890].

Crocker, et al.              Standards Track                   [Page 20]

RFC 6376                     DKIM Signatures              September 2011


      sig-d-tag       = %x64 [FWS] "=" [FWS] domain-name
      domain-name     = sub-domain 1*("." sub-domain)
                        ; from [RFC5321] Domain,
                        ; excluding address-literal

   h= Signed header fields (plain-text, but see description; REQUIRED).
      A colon-separated list of header field names that identify the
      header fields presented to the signing algorithm.  The field MUST
      contain the complete list of header fields in the order presented
      to the signing algorithm.  The field MAY contain names of header
      fields that do not exist when signed; nonexistent header fields do
      not contribute to the signature computation (that is, they are
      treated as the null input, including the header field name, the
      separating colon, the header field value, and any CRLF
      terminator).  The field MAY contain multiple instances of a header
      field name, meaning multiple occurrences of the corresponding
      header field are included in the header hash.  The field MUST NOT
      include the DKIM-Signature header field that is being created or
      verified but may include others.  Folding whitespace (FWS) MAY be
      included on either side of the colon separator.  Header field
      names MUST be compared against actual header field names in a
      case-insensitive manner.  This list MUST NOT be empty.  See
      Section 5.4 for a discussion of choosing header fields to sign and
      Section 5.4.2 for requirements when signing multiple instances of
      a single field.


      sig-h-tag       = %x68 [FWS] "=" [FWS] hdr-name
                         *( [FWS] ":" [FWS] hdr-name )

         INFORMATIVE EXPLANATION: By "signing" header fields that do not
         actually exist, a Signer can allow a Verifier to detect
         insertion of those header fields after signing.  However, since
         a Signer cannot possibly know what header fields might be
         defined in the future, this mechanism cannot be used to prevent
         the addition of any possible unknown header fields.

         INFORMATIVE NOTE: "Signing" fields that are not present at the
         time of signing not only prevents fields and values from being
         added but also prevents adding fields with no values.

   i= The Agent or User Identifier (AUID) on behalf of which the SDID is
      taking responsibility (dkim-quoted-printable; OPTIONAL, default is
      an empty local-part followed by an "@" followed by the domain from
      the "d=" tag).

Crocker, et al.              Standards Track                   [Page 21]

RFC 6376                     DKIM Signatures              September 2011

      The syntax is a standard email address where the local-part MAY be
      omitted.  The domain part of the address MUST be the same as, or a
      subdomain of, the value of the "d=" tag.

      Internationalized domain names MUST be encoded as A-labels, as
      described in Section 2.3 of [RFC5890].


      sig-i-tag       = %x69 [FWS] "=" [FWS] [ Local-part ]
                                 "@" domain-name

      The AUID is specified as having the same syntax as an email
      address but it need not have the same semantics.  Notably, the
      domain name need not be registered in the DNS -- so it might not
      resolve in a query -- and the local-part MAY be drawn from a
      namespace unrelated to any mailbox.  The details of the structure
      and semantics for the namespace are determined by the Signer.  Any
      knowledge or use of those details by Verifiers or Assessors is
      outside the scope of this specification.  The Signer MAY choose to
      use the same namespace for its AUIDs as its users' email addresses
      or MAY choose other means of representing its users.  However, the
      Signer SHOULD use the same AUID for each message intended to be
      evaluated as being within the same sphere of responsibility, if it
      wishes to offer receivers the option of using the AUID as a stable
      identifier that is finer grained than the SDID.

         INFORMATIVE NOTE: The local-part of the "i=" tag is optional
         because in some cases a Signer may not be able to establish a
         verified individual identity.  In such cases, the Signer might
         wish to assert that although it is willing to go as far as
         signing for the domain, it is unable or unwilling to commit to
         an individual user name within the domain.  It can do so by
         including the domain part but not the local-part of the

         INFORMATIVE DISCUSSION: This specification does not require the
         value of the "i=" tag to match the identity in any message
         header fields.  This is considered to be a Verifier policy
         issue.  Constraints between the value of the "i=" tag and other
         identities in other header fields seek to apply basic
         authentication into the semantics of trust associated with a
         role such as content author.  Trust is a broad and complex
         topic, and trust mechanisms are subject to highly creative
         attacks.  The real-world efficacy of any but the most basic
         bindings between the "i=" value and other identities is not
         well established, nor is its vulnerability to subversion by an
         attacker.  Hence, reliance on the use of these options should

Crocker, et al.              Standards Track                   [Page 22]

RFC 6376                     DKIM Signatures              September 2011

         be strictly limited.  In particular, it is not at all clear to
         what extent a typical end-user recipient can rely on any
         assurances that might be made by successful use of the "i="

   l= Body length count (plain-text unsigned decimal integer; OPTIONAL,
      default is entire body).  This tag informs the Verifier of the
      number of octets in the body of the email after canonicalization
      included in the cryptographic hash, starting from 0 immediately
      following the CRLF preceding the body.  This value MUST NOT be
      larger than the actual number of octets in the canonicalized
      message body.  See further discussion in Section 8.2.

         INFORMATIVE NOTE: The value of the "l=" tag is constrained to
         76 decimal digits.  This constraint is not intended to predict
         the size of future messages or to require implementations to
         use an integer representation large enough to represent the
         maximum possible value but is intended to remind the
         implementer to check the length of this and all other tags
         during verification and to test for integer overflow when
         decoding the value.  Implementers may need to limit the actual
         value expressed to a value smaller than 10^76, e.g., to allow a
         message to fit within the available storage space.


      sig-l-tag    = %x6c [FWS] "=" [FWS]

   q= A colon-separated list of query methods used to retrieve the
      public key (plain-text; OPTIONAL, default is "dns/txt").  Each
      query method is of the form "type[/options]", where the syntax and
      semantics of the options depend on the type and specified options.
      If there are multiple query mechanisms listed, the choice of query
      mechanism MUST NOT change the interpretation of the signature.
      Implementations MUST use the recognized query mechanisms in the
      order presented.  Unrecognized query mechanisms MUST be ignored.

      Currently, the only valid value is "dns/txt", which defines the
      DNS TXT resource record (RR) lookup algorithm described elsewhere
      in this document.  The only option defined for the "dns" query
      type is "txt", which MUST be included.  Verifiers and Signers MUST
      support "dns/txt".


      sig-q-tag        = %x71 [FWS] "=" [FWS] sig-q-tag-method
                            *([FWS] ":" [FWS] sig-q-tag-method)

Crocker, et al.              Standards Track                   [Page 23]

RFC 6376                     DKIM Signatures              September 2011

      sig-q-tag-method = "dns/txt" / x-sig-q-tag-type
                         ["/" x-sig-q-tag-args]
      x-sig-q-tag-type = hyphenated-word  ; for future extension
      x-sig-q-tag-args = qp-hdr-value

   s= The selector subdividing the namespace for the "d=" (domain) tag
      (plain-text; REQUIRED).

      Internationalized selector names MUST be encoded as A-labels, as
      described in Section 2.3 of [RFC5890].


      sig-s-tag    = %x73 [FWS] "=" [FWS] selector

   t= Signature Timestamp (plain-text unsigned decimal integer;
      RECOMMENDED, default is an unknown creation time).  The time that
      this signature was created.  The format is the number of seconds
      since 00:00:00 on January 1, 1970 in the UTC time zone.  The value
      is expressed as an unsigned integer in decimal ASCII.  This value
      is not constrained to fit into a 31- or 32-bit integer.
      Implementations SHOULD be prepared to handle values up to at least
      10^12 (until approximately AD 200,000; this fits into 40 bits).
      To avoid denial-of-service attacks, implementations MAY consider
      any value longer than 12 digits to be infinite.  Leap seconds are
      not counted.  Implementations MAY ignore signatures that have a
      timestamp in the future.


      sig-t-tag    = %x74 [FWS] "=" [FWS] 1*12DIGIT

   x= Signature Expiration (plain-text unsigned decimal integer;
      RECOMMENDED, default is no expiration).  The format is the same as
      in the "t=" tag, represented as an absolute date, not as a time
      delta from the signing timestamp.  The value is expressed as an
      unsigned integer in decimal ASCII, with the same constraints on
      the value in the "t=" tag.  Signatures MAY be considered invalid
      if the verification time at the Verifier is past the expiration
      date.  The verification time should be the time that the message
      was first received at the administrative domain of the Verifier if
      that time is reliably available; otherwise, the current time
      should be used.  The value of the "x=" tag MUST be greater than
      the value of the "t=" tag if both are present.

         INFORMATIVE NOTE: The "x=" tag is not intended as an anti-
         replay defense.

Crocker, et al.              Standards Track                   [Page 24]

RFC 6376                     DKIM Signatures              September 2011

         INFORMATIVE NOTE: Due to clock drift, the receiver's notion of
         when to consider the signature expired may not exactly match
         what the sender is expecting.  Receivers MAY add a 'fudge
         factor' to allow for such possible drift.


      sig-x-tag    = %x78 [FWS] "=" [FWS]

   z= Copied header fields (dkim-quoted-printable, but see description;
      OPTIONAL, default is null).  A vertical-bar-separated list of
      selected header fields present when the message was signed,
      including both the field name and value.  It is not required to
      include all header fields present at the time of signing.  This
      field need not contain the same header fields listed in the "h="
      tag.  The header field text itself must encode the vertical bar
      ("|", %x7C) character (i.e., vertical bars in the "z=" text are
      meta-characters, and any actual vertical bar characters in a
      copied header field must be encoded).  Note that all whitespace
      must be encoded, including whitespace between the colon and the
      header field value.  After encoding, FWS MAY be added at arbitrary
      locations in order to avoid excessively long lines; such
      whitespace is NOT part of the value of the header field and MUST
      be removed before decoding.

      The header fields referenced by the "h=" tag refer to the fields
      in the [RFC5322] header of the message, not to any copied fields
      in the "z=" tag.  Copied header field values are for diagnostic


      sig-z-tag      = %x7A [FWS] "=" [FWS] sig-z-tag-copy
                       *( "|" [FWS] sig-z-tag-copy )
      sig-z-tag-copy = hdr-name [FWS] ":" qp-hdr-value

         INFORMATIVE EXAMPLE of a signature header field spread across
         multiple continuation lines:

   DKIM-Signature: v=1; a=rsa-sha256; d=example.net; s=brisbane;
      c=simple; q=dns/txt; i=@eng.example.net;
      t=1117574938; x=1118006938;

Crocker, et al.              Standards Track                   [Page 25]

RFC 6376                     DKIM Signatures              September 2011

3.6.  Key Management and Representation

   Signature applications require some level of assurance that the
   verification public key is associated with the claimed Signer.  Many
   applications achieve this by using public-key certificates issued by
   a trusted third party.  However, DKIM can achieve a sufficient level
   of security, with significantly enhanced scalability, by simply
   having the Verifier query the purported Signer's DNS entry (or some
   security-equivalent) in order to retrieve the public key.

   DKIM keys can potentially be stored in multiple types of key servers
   and in multiple formats.  The storage and format of keys are
   irrelevant to the remainder of the DKIM algorithm.

   Parameters to the key lookup algorithm are the type of the lookup
   (the "q=" tag), the domain of the Signer (the "d=" tag of the DKIM-
   Signature header field), and the selector (the "s=" tag).

   public_key = dkim_find_key(q_val, d_val, s_val)

   This document defines a single binding, using DNS TXT RRs to
   distribute the keys.  Other bindings may be defined in the future.

3.6.1.  Textual Representation

   It is expected that many key servers will choose to present the keys
   in an otherwise unstructured text format (for example, an XML form
   would not be considered to be unstructured text for this purpose).
   The following definition MUST be used for any DKIM key represented in
   an otherwise unstructured textual form.

   The overall syntax is a tag-list as described in Section 3.2.  The
   current valid tags are described below.  Other tags MAY be present
   and MUST be ignored by any implementation that does not understand

   v= Version of the DKIM key record (plain-text; RECOMMENDED, default
      is "DKIM1").  If specified, this tag MUST be set to "DKIM1"
      (without the quotes).  This tag MUST be the first tag in the
      record.  Records beginning with a "v=" tag with any other value
      MUST be discarded.  Note that Verifiers must do a string
      comparison on this value; for example, "DKIM1" is not the same as


      key-v-tag    = %x76 [FWS] "=" [FWS] %x44.4B.49.4D.31

Crocker, et al.              Standards Track                   [Page 26]

RFC 6376                     DKIM Signatures              September 2011

   h= Acceptable hash algorithms (plain-text; OPTIONAL, defaults to
      allowing all algorithms).  A colon-separated list of hash
      algorithms that might be used.  Unrecognized algorithms MUST be
      ignored.  Refer to Section 3.3 for a discussion of the hash
      algorithms implemented by Signers and Verifiers.  The set of
      algorithms listed in this tag in each record is an operational
      choice made by the Signer.


      key-h-tag       = %x68 [FWS] "=" [FWS] key-h-tag-alg
                        *( [FWS] ":" [FWS] key-h-tag-alg )
      key-h-tag-alg   = "sha1" / "sha256" / x-key-h-tag-alg
      x-key-h-tag-alg = hyphenated-word   ; for future extension

   k= Key type (plain-text; OPTIONAL, default is "rsa").  Signers and
      Verifiers MUST support the "rsa" key type.  The "rsa" key type
      indicates that an ASN.1 DER-encoded [ITU-X660-1997] RSAPublicKey
      (see [RFC3447], Sections 3.1 and A.1.1) is being used in the "p="
      tag.  (Note: the "p=" tag further encodes the value using the
      base64 algorithm.)  Unrecognized key types MUST be ignored.


      key-k-tag        = %x76 [FWS] "=" [FWS] key-k-tag-type
      key-k-tag-type   = "rsa" / x-key-k-tag-type
      x-key-k-tag-type = hyphenated-word   ; for future extension

   n= Notes that might be of interest to a human (qp-section; OPTIONAL,
      default is empty).  No interpretation is made by any program.
      This tag should be used sparingly in any key server mechanism that
      has space limitations (notably DNS).  This is intended for use by
      administrators, not end users.


      key-n-tag    = %x6e [FWS] "=" [FWS] qp-section

   p= Public-key data (base64; REQUIRED).  An empty value means that
      this public key has been revoked.  The syntax and semantics of
      this tag value before being encoded in base64 are defined by the
      "k=" tag.

         INFORMATIVE RATIONALE: If a private key has been compromised or
         otherwise disabled (e.g., an outsourcing contract has been
         terminated), a Signer might want to explicitly state that it
         knows about the selector, but all messages using that selector

Crocker, et al.              Standards Track                   [Page 27]

RFC 6376                     DKIM Signatures              September 2011

         should fail verification.  Verifiers SHOULD return an error
         code for any DKIM-Signature header field with a selector
         referencing a revoked key.  (See Section 6.1.2 for details.)


      key-p-tag    = %x70 [FWS] "=" [ [FWS] base64string]

         INFORMATIVE NOTE: A base64string is permitted to include
         whitespace (FWS) at arbitrary places; however, any CRLFs must
         be followed by at least one WSP character.  Implementers and
         administrators are cautioned to ensure that selector TXT RRs
         conform to this specification.

   s= Service Type (plain-text; OPTIONAL; default is "*").  A colon-
      separated list of service types to which this record applies.
      Verifiers for a given service type MUST ignore this record if the
      appropriate type is not listed.  Unrecognized service types MUST
      be ignored.  Currently defined service types are as follows:

      *  matches all service types

      email   electronic mail (not necessarily limited to SMTP)

      This tag is intended to constrain the use of keys for other
      purposes, should use of DKIM be defined by other services in the


      key-s-tag        = %x73 [FWS] "=" [FWS] key-s-tag-type
                         *( [FWS] ":" [FWS] key-s-tag-type )
      key-s-tag-type   = "email" / "*" / x-key-s-tag-type
      x-key-s-tag-type = hyphenated-word   ; for future extension

   t= Flags, represented as a colon-separated list of names (plain-
      text; OPTIONAL, default is no flags set).  Unrecognized flags MUST
      be ignored.  The defined flags are as follows:

      y  This domain is testing DKIM.  Verifiers MUST NOT treat messages
         from Signers in testing mode differently from unsigned email,
         even should the signature fail to verify.  Verifiers MAY wish
         to track testing mode results to assist the Signer.

Crocker, et al.              Standards Track                   [Page 28]

RFC 6376                     DKIM Signatures              September 2011

      s  Any DKIM-Signature header fields using the "i=" tag MUST have
         the same domain value on the right-hand side of the "@" in the
         "i=" tag and the value of the "d=" tag.  That is, the "i="
         domain MUST NOT be a subdomain of "d=".  Use of this flag is
         RECOMMENDED unless subdomaining is required.


      key-t-tag        = %x74 [FWS] "=" [FWS] key-t-tag-flag
                         *( [FWS] ":" [FWS] key-t-tag-flag )
      key-t-tag-flag   = "y" / "s" / x-key-t-tag-flag
      x-key-t-tag-flag = hyphenated-word   ; for future extension

3.6.2.  DNS Binding

   A binding using DNS TXT RRs as a key service is hereby defined.  All
   implementations MUST support this binding.  Namespace

   All DKIM keys are stored in a subdomain named "_domainkey".  Given a
   DKIM-Signature field with a "d=" tag of "example.com" and an "s=" tag
   of "foo.bar", the DNS query will be for
   "foo.bar._domainkey.example.com".  Resource Record Types for Key Storage

   The DNS Resource Record type used is specified by an option to the
   query-type ("q=") tag.  The only option defined in this base
   specification is "txt", indicating the use of a TXT RR.  A later
   extension of this standard may define another RR type.

   Strings in a TXT RR MUST be concatenated together before use with no
   intervening whitespace.  TXT RRs MUST be unique for a particular
   selector name; that is, if there are multiple records in an RRset,
   the results are undefined.

   TXT RRs are encoded as described in Section 3.6.1.

3.7.  Computing the Message Hashes

   Both signing and verifying message signatures start with a step of
   computing two cryptographic hashes over the message.  Signers will
   choose the parameters of the signature as described in "Signer
   Actions" (Section 5); Verifiers will use the parameters specified in
   the DKIM-Signature header field being verified.  In the following
   discussion, the names of the tags in the DKIM-Signature header field
   that either exists (when verifying) or will be created (when signing)

Crocker, et al.              Standards Track                   [Page 29]

RFC 6376                     DKIM Signatures              September 2011

   are used.  Note that canonicalization (Section 3.4) is only used to
   prepare the email for signing or verifying; it does not affect the
   transmitted email in any way.

   The Signer/Verifier MUST compute two hashes: one over the body of the
   message and one over the selected header fields of the message.

   Signers MUST compute them in the order shown.  Verifiers MAY compute
   them in any order convenient to the Verifier, provided that the
   result is semantically identical to the semantics that would be the
   case had they been computed in this order.

   In hash step 1, the Signer/Verifier MUST hash the message body,
   canonicalized using the body canonicalization algorithm specified in
   the "c=" tag and then truncated to the length specified in the "l="
   tag.  That hash value is then converted to base64 form and inserted
   into (Signers) or compared to (Verifiers) the "bh=" tag of the DKIM-
   Signature header field.

   In hash step 2, the Signer/Verifier MUST pass the following to the
   hash algorithm in the indicated order.

   1.  The header fields specified by the "h=" tag, in the order
       specified in that tag, and canonicalized using the header
       canonicalization algorithm specified in the "c=" tag.  Each
       header field MUST be terminated with a single CRLF.

   2.  The DKIM-Signature header field that exists (verifying) or will
       be inserted (signing) in the message, with the value of the "b="
       tag (including all surrounding whitespace) deleted (i.e., treated
       as the empty string), canonicalized using the header
       canonicalization algorithm specified in the "c=" tag, and without
       a trailing CRLF.

   All tags and their values in the DKIM-Signature header field are
   included in the cryptographic hash with the sole exception of the
   value portion of the "b=" (signature) tag, which MUST be treated as
   the null string.  All tags MUST be included even if they might not be
   understood by the Verifier.  The header field MUST be presented to
   the hash algorithm after the body of the message rather than with the
   rest of the header fields and MUST be canonicalized as specified in
   the "c=" (canonicalization) tag.  The DKIM-Signature header field
   MUST NOT be included in its own "h=" tag, although other DKIM-
   Signature header fields MAY be signed (see Section 4).

   When calculating the hash on messages that will be transmitted using
   base64 or quoted-printable encoding, Signers MUST compute the hash
   after the encoding.  Likewise, the Verifier MUST incorporate the

Crocker, et al.              Standards Track                   [Page 30]

RFC 6376                     DKIM Signatures              September 2011

   values into the hash before decoding the base64 or quoted-printable
   text.  However, the hash MUST be computed before transport-level
   encodings such as SMTP "dot-stuffing" (the modification of lines
   beginning with a "." to avoid confusion with the SMTP end-of-message
   marker, as specified in [RFC5321]).

   With the exception of the canonicalization procedure described in
   Section 3.4, the DKIM signing process treats the body of messages as
   simply a string of octets.  DKIM messages MAY be either in plain-text
   or in MIME format; no special treatment is afforded to MIME content.
   Message attachments in MIME format MUST be included in the content
   that is signed.

   More formally, pseudo-code for the signature algorithm is:

   body-hash    =  hash-alg (canon-body, l-param)
   data-hash    =  hash-alg (h-headers, D-SIG, body-hash)
   signature    =  sig-alg (d-domain, selector, data-hash)


   body-hash:  is the output from hashing the body, using hash-alg.

   hash-alg:   is the hashing algorithm specified in the "a" parameter.

   canon-body: is a canonicalized representation of the body, produced
               using the body algorithm specified in the "c" parameter,
               as defined in Section 3.4 and excluding the
               DKIM-Signature field.

   l-param:    is the length-of-body value of the "l" parameter.

   data-hash:  is the output from using the hash-alg algorithm, to hash
               the header including the DKIM-Signature header, and the
               body hash.

   h-headers:  is the list of headers to be signed, as specified in the
               "h" parameter.

   D-SIG:      is the canonicalized DKIM-Signature field itself without
               the signature value portion of the parameter, that is, an
               empty parameter value.

   signature:  is the signature value produced by the signing algorithm.

   sig-alg:    is the signature algorithm specified by the "a"

Crocker, et al.              Standards Track                   [Page 31]

RFC 6376                     DKIM Signatures              September 2011

   d-domain:   is the domain name specified in the "d" parameter.

   selector:   is the selector value specified in the "s" parameter.

      NOTE: Many digital signature APIs provide both hashing and
      application of the RSA private key using a single "sign()"
      primitive.  When using such an API, the last two steps in the
      algorithm would probably be combined into a single call that would
      perform both the "a-hash-alg" and the "sig-alg".

3.8.  Input Requirements

   A message that is not compliant with [RFC5322], [RFC2045], and
   [RFC2047] can be subject to attempts by intermediaries to correct or
   interpret such content.  See Section 8 of [RFC4409] for examples of
   changes that are commonly made.  Such "corrections" may invalidate
   DKIM signatures or have other undesirable effects, including some
   that involve changes to the way a message is presented to an end

   Accordingly, DKIM's design is predicated on valid input.  Therefore,
   Signers and Verifiers SHOULD take reasonable steps to ensure that the
   messages they are processing are valid according to [RFC5322],
   [RFC2045], and any other relevant message format standards.

   See Section 8.15 for additional discussion.

3.9.  Output Requirements

   The evaluation of each signature ends in one of three states, which
   this document refers to as follows:

   SUCCESS:  a successful verification

   PERMFAIL:  a permanent, non-recoverable error such as a signature
      verification failure

   TEMPFAIL:  a temporary, recoverable error such as a DNS query timeout

   For each signature that verifies successfully or produces a TEMPFAIL
   result, output of the DKIM algorithm MUST include the set of:

   o  The domain name, taken from the "d=" signature tag; and

   o  The result of the verification attempt for that signature.

Crocker, et al.              Standards Track                   [Page 32]

RFC 6376                     DKIM Signatures              September 2011

   The output MAY include other signature properties or result meta-
   data, including PERMFAILed or otherwise ignored signatures, for use
   by modules that consume those results.

   See Section 6.1 for discussion of signature validation result codes.

3.10.  Signing by Parent Domains

   In some circumstances, it is desirable for a domain to apply a
   signature on behalf of any of its subdomains without the need to
   maintain separate selectors (key records) in each subdomain.  By
   default, private keys corresponding to key records can be used to
   sign messages for any subdomain of the domain in which they reside;
   for example, a key record for the domain example.com can be used to
   verify messages where the AUID ("i=" tag of the signature) is
   sub.example.com, or even sub1.sub2.example.com.  In order to limit
   the capability of such keys when this is not intended, the "s" flag
   MAY be set in the "t=" tag of the key record, to constrain the
   validity of the domain of the AUID.  If the referenced key record
   contains the "s" flag as part of the "t=" tag, the domain of the AUID
   ("i=" flag) MUST be the same as that of the SDID (d=) domain.  If
   this flag is absent, the domain of the AUID MUST be the same as, or a
   subdomain of, the SDID.

3.11.  Relationship between SDID and AUID

   DKIM's primary task is to communicate from the Signer to a recipient-
   side Identity Assessor a single Signing Domain Identifier (SDID) that
   refers to a responsible identity.  DKIM MAY optionally provide a
   single responsible Agent or User Identifier (AUID).

   Hence, DKIM's mandatory output to a receive-side Identity Assessor is
   a single domain name.  Within the scope of its use as DKIM output,
   the name has only basic domain name semantics; any possible owner-
   specific semantics are outside the scope of DKIM.  That is, within
   its role as a DKIM identifier, additional semantics cannot be assumed
   by an Identity Assessor.

   Upon successfully verifying the signature, a receive-side DKIM
   Verifier MUST communicate the Signing Domain Identifier (d=) to a
   consuming Identity Assessor module and MAY communicate the Agent or
   User Identifier (i=) if present.

   To the extent that a receiver attempts to intuit any structured
   semantics for either of the identifiers, this is a heuristic function
   that is outside the scope of DKIM's specification and semantics.

Crocker, et al.              Standards Track                   [Page 33]

RFC 6376                     DKIM Signatures              September 2011

   Hence, it is relegated to a higher-level service, such as a delivery-
   handling filter that integrates a variety of inputs and performs
   heuristic analysis of them.

      INFORMATIVE DISCUSSION: This document does not require the value
      of the SDID or AUID to match an identifier in any other message
      header field.  This requirement is, instead, an Assessor policy
      issue.  The purpose of such a linkage would be to authenticate the
      value in that other header field.  This, in turn, is the basis for
      applying a trust assessment based on the identifier value.  Trust
      is a broad and complex topic, and trust mechanisms are subject to
      highly creative attacks.  The real-world efficacy of any but the
      most basic bindings between the SDID or AUID and other identities
      is not well established, nor is its vulnerability to subversion by
      an attacker.  Hence, reliance on the use of such bindings should
      be strictly limited.  In particular, it is not at all clear to
      what extent a typical end-user recipient can rely on any
      assurances that might be made by successful use of the SDID or

4.  Semantics of Multiple Signatures

4.1.  Example Scenarios

   There are many reasons why a message might have multiple signatures.
   For example, suppose SHA-256 is in the future found to be
   insufficiently strong, and DKIM usage transitions to SHA-1024.  A
   Signer might immediately sign using the newer algorithm but also
   continue to sign using the older algorithm for interoperability with
   Verifiers that had not yet upgraded.  The Signer would do this by
   adding two DKIM-Signature header fields, one using each algorithm.
   Older Verifiers that did not recognize SHA-1024 as an acceptable
   algorithm would skip that signature and use the older algorithm;
   newer Verifiers could use either signature at their option and, all
   other things being equal, might not even attempt to verify the other

   Similarly, a Signer might sign a message including all header fields
   and no "l=" tag (to satisfy strict Verifiers) and a second time with
   a limited set of header fields and an "l=" tag (in anticipation of
   possible message modifications en route to other Verifiers).
   Verifiers could then choose which signature they prefer.

   Of course, a message might also have multiple signatures because it
   passed through multiple Signers.  A common case is expected to be
   that of a signed message that passes through a mailing list that also

Crocker, et al.              Standards Track                   [Page 34]

RFC 6376                     DKIM Signatures              September 2011

   signs all messages.  Assuming both of those signatures verify, a
   recipient might choose to accept the message if either of those
   signatures were known to come from trusted sources.

   In particular, recipients might choose to whitelist mailing lists to
   which they have subscribed and that have acceptable anti-abuse
   policies so as to accept messages sent to that list even from unknown
   authors.  They might also subscribe to less trusted mailing lists
   (e.g., those without anti-abuse protection) and be willing to accept
   all messages from specific authors but insist on doing additional
   abuse scanning for other messages.

   Another related example of multiple Signers might be forwarding
   services, such as those commonly associated with academic alumni
   sites.  For example, a recipient might have an address at
   members.example.org, a site that has anti-abuse protection that is
   somewhat less effective than the recipient would prefer.  Such a
   recipient might have specific authors whose messages would be trusted
   absolutely, but messages from unknown authors that had passed the
   forwarder's scrutiny would have only medium trust.

4.2.  Interpretation

   A Signer that is adding a signature to a message merely creates a new
   DKIM-Signature header, using the usual semantics of the "h=" option.
   A Signer MAY sign previously existing DKIM-Signature header fields
   using the method described in Section 5.4 to sign trace header

   Note that Signers should be cognizant that signing DKIM-Signature
   header fields may result in signature failures with intermediaries
   that do not recognize that DKIM-Signature header fields are trace
   header fields and unwittingly reorder them, thus breaking such
   signatures.  For this reason, signing existing DKIM-Signature header
   fields is unadvised, albeit legal.

      INFORMATIVE NOTE: If a header field with multiple instances is
      signed, those header fields are always signed from the bottom up.
      Thus, it is not possible to sign only specific DKIM-Signature
      header fields.  For example, if the message being signed already
      contains three DKIM-Signature header fields A, B, and C, it is
      possible to sign all of them, B and C only, or C only, but not A
      only, B only, A and B only, or A and C only.

   A Signer MAY add more than one DKIM-Signature header field using
   different parameters.  For example, during a transition period, a
   Signer might want to produce signatures using two different hash

Crocker, et al.              Standards Track                   [Page 35]

RFC 6376                     DKIM Signatures              September 2011

   Signers SHOULD NOT remove any DKIM-Signature header fields from
   messages they are signing, even if they know that the signatures
   cannot be verified.

   When evaluating a message with multiple signatures, a Verifier SHOULD
   evaluate signatures independently and on their own merits.  For
   example, a Verifier that by policy chooses not to accept signatures
   with deprecated cryptographic algorithms would consider such
   signatures invalid.  Verifiers MAY process signatures in any order of
   their choice; for example, some Verifiers might choose to process
   signatures corresponding to the From field in the message header
   before other signatures.  See Section 6.1 for more information about
   signature choices.

      INFORMATIVE IMPLEMENTATION NOTE: Verifier attempts to correlate
      valid signatures with invalid signatures in an attempt to guess
      why a signature failed are ill-advised.  In particular, there is
      no general way that a Verifier can determine that an invalid
      signature was ever valid.

   Verifiers SHOULD continue to check signatures until a signature
   successfully verifies to the satisfaction of the Verifier.  To limit
   potential denial-of-service attacks, Verifiers MAY limit the total
   number of signatures they will attempt to verify.

   If a Verifier module reports signatures whose evaluations produced
   PERMFAIL results, Identity Assessors SHOULD ignore those signatures
   (see Section 6.1), acting as though they were not present in the

5.  Signer Actions

   The following steps are performed in order by Signers.

5.1.  Determine Whether the Email Should Be Signed and by Whom

   A Signer can obviously only sign email for domains for which it has a
   private key and the necessary knowledge of the corresponding public
   key and selector information.  However, there are a number of other
   reasons beyond the lack of a private key why a Signer could choose
   not to sign an email.

      INFORMATIVE NOTE: A Signer can be implemented as part of any
      portion of the mail system as deemed appropriate, including an
      MUA, a SUBMISSION server, or an MTA.  Wherever implemented,
      Signers should beware of signing (and thereby asserting
      responsibility for) messages that may be problematic.  In
      particular, within a trusted enclave, the signing domain might be

Crocker, et al.              Standards Track                   [Page 36]

RFC 6376                     DKIM Signatures              September 2011

      derived from the header according to local policy; SUBMISSION
      servers might only sign messages from users that are properly
      authenticated and authorized.

      Received header fields if the outgoing gateway MTA obfuscates
      Received header fields, for example, to hide the details of
      internal topology.

   If an email cannot be signed for some reason, it is a local policy
   decision as to what to do with that email.

5.2.  Select a Private Key and Corresponding Selector Information

   This specification does not define the basis by which a Signer should
   choose which private key and selector information to use.  Currently,
   all selectors are equal as far as this specification is concerned, so
   the decision should largely be a matter of administrative
   convenience.  Distribution and management of private keys is also
   outside the scope of this document.

      INFORMATIVE OPERATIONS ADVICE: A Signer should not sign with a
      private key when the selector containing the corresponding public
      key is expected to be revoked or removed before the Verifier has
      an opportunity to validate the signature.  The Signer should
      anticipate that Verifiers can choose to defer validation, perhaps
      until the message is actually read by the final recipient.  In
      particular, when rotating to a new key pair, signing should
      immediately commence with the new private key, and the old public
      key should be retained for a reasonable validation interval before
      being removed from the key server.

5.3.  Normalize the Message to Prevent Transport Conversions

   Some messages, particularly those using 8-bit characters, are subject
   to modification during transit, notably conversion to 7-bit form.
   Such conversions will break DKIM signatures.  In order to minimize
   the chances of such breakage, Signers SHOULD convert the message to a
   suitable MIME content-transfer encoding such as quoted-printable or
   base64 as described in [RFC2045] before signing.  Such conversion is
   outside the scope of DKIM; the actual message SHOULD be converted to
   7-bit MIME by an MUA or MSA prior to presentation to the DKIM

   If the message is submitted to the Signer with any local encoding
   that will be modified before transmission, that modification to
   canonical [RFC5322] form MUST be done before signing.  In particular,
   bare CR or LF characters (used by some systems as a local line

Crocker, et al.              Standards Track                   [Page 37]

RFC 6376                     DKIM Signatures              September 2011

   separator convention) MUST be converted to the SMTP-standard CRLF
   sequence before the message is signed.  Any conversion of this sort
   SHOULD be applied to the message actually sent to the recipient(s),
   not just to the version presented to the signing algorithm.

   More generally, the Signer MUST sign the message as it is expected to
   be received by the Verifier rather than in some local or internal

5.3.1.  Body Length Limits

   A body length count MAY be specified to limit the signature
   calculation to an initial prefix of the body text, measured in
   octets.  If the body length count is not specified, the entire
   message body is signed.

      INFORMATIVE RATIONALE: This capability is provided because it is
      very common for mailing lists to add trailers to messages (e.g.,
      instructions on how to get off the list).  Until those messages
      are also signed, the body length count is a useful tool for the
      Verifier since it can, as a matter of policy, accept messages
      having valid signatures with extraneous data.

   The length actually hashed should be inserted in the "l=" tag of the
   DKIM-Signature header field.  (See Section 3.5.)

   The body length count allows the Signer of a message to permit data
   to be appended to the end of the body of a signed message.  The body
   length count MUST be calculated following the canonicalization
   algorithm; for example, any whitespace ignored by a canonicalization
   algorithm is not included as part of the body length count.

   A body length count of zero means that the body is completely

   Signers wishing to ensure that no modification of any sort can occur
   should specify the "simple" canonicalization algorithm for both
   header and body and omit the body length count.

   See Section 8.2 for further discussion.

5.4.  Determine the Header Fields to Sign

   The From header field MUST be signed (that is, included in the "h="
   tag of the resulting DKIM-Signature header field).  Signers SHOULD
sign an existing header field likely to be legitimately modified
   or removed in transit.  In particular, [RFC5321] explicitly permits

Crocker, et al.              Standards Track                   [Page 38]

RFC 6376                     DKIM Signatures              September 2011

   modification or removal of the Return-Path header field in transit.
   Signers MAY include any other header fields present at the time of
   signing at the discretion of the Signer.

      INFORMATIVE OPERATIONS NOTE: The choice of which header fields to
      sign is non-obvious.  One strategy is to sign all existing, non-
      repeatable header fields.  An alternative strategy is to sign only
      header fields that are likely to be displayed to or otherwise be
      likely to affect the processing of the message at the receiver.  A
      third strategy is to sign only "well-known" headers.  Note that
      Verifiers may treat unsigned header fields with extreme
      skepticism, including refusing to display them to the end user or
      even ignoring the signature if it does not cover certain header
      fields.  For this reason, signing fields present in the message
      such as Date, Subject, Reply-To, Sender, and all MIME header
      fields are highly advised.

   The DKIM-Signature header field is always implicitly signed and MUST
be included in the "h=" tag except to indicate that other
   preexisting signatures are also signed.

   Signers MAY claim to have signed header fields that do not exist
   (that is, Signers MAY include the header field name in the "h=" tag
   even if that header field does not exist in the message).  When
   computing the signature, the nonexisting header field MUST be treated
   as the null string (including the header field name, header field
   value, all punctuation, and the trailing CRLF).

      INFORMATIVE RATIONALE: This allows Signers to explicitly assert
      the absence of a header field; if that header field is added
      later, the signature will fail.

      INFORMATIVE NOTE: A header field name need only be listed once
      more than the actual number of that header field in a message at
      the time of signing in order to prevent any further additions.
      For example, if there is a single Comments header field at the
      time of signing, listing Comments twice in the "h=" tag is
      sufficient to prevent any number of Comments header fields from
      being appended; it is not necessary (but is legal) to list
      Comments three or more times in the "h=" tag.

   Refer to Section 5.4.2 for a discussion of the procedure to be
   followed when canonicalizing a header with more than one instance of
   a particular header field name.

   Signers need to be careful of signing header fields that might have
   additional instances added later in the delivery process, since such
   header fields might be inserted after the signed instance or

Crocker, et al.              Standards Track                   [Page 39]

RFC 6376                     DKIM Signatures              September 2011

   otherwise reordered.  Trace header fields (such as Received) and
   Resent-* blocks are the only fields prohibited by [RFC5322] from
   being reordered.  In particular, since DKIM-Signature header fields
   may be reordered by some intermediate MTAs, signing existing DKIM-
   Signature header fields is error-prone.

      INFORMATIVE ADMONITION: Despite the fact that [RFC5322] does not
      prohibit the reordering of header fields, reordering of signed
      header fields with multiple instances by intermediate MTAs will
      cause DKIM signatures to be broken; such antisocial behavior
      should be avoided.

      INFORMATIVE IMPLEMENTER'S NOTE: Although not required by this
      specification, all end-user visible header fields should be signed
      to avoid possible "indirect spamming".  For example, if the
      Subject header field is not signed, a spammer can resend a
      previously signed mail, replacing the legitimate subject with a
      one-line spam.

5.4.1.  Recommended Signature Content

   The purpose of the DKIM cryptographic algorithm is to affix an
   identifier to the message in a way that is both robust against normal
   transit-related changes and resistant to kinds of replay attacks.  An
   essential aspect of satisfying these requirements is choosing what
   header fields to include in the hash and what fields to exclude.

   The basic rule for choosing fields to include is to select those
   fields that constitute the "core" of the message content.  Hence, any
   replay attack will have to include these in order to have the
   signature succeed; however, with these included, the core of the
   message is valid, even if sent on to new recipients.

   Common examples of fields with addresses and fields with textual
   content related to the body are:

   o  From (REQUIRED; see Section 5.4)

   o  Reply-To

   o  Subject

   o  Date

   o  To, Cc

   o  Resent-Date, Resent-From, Resent-To, Resent-Cc

Crocker, et al.              Standards Track                   [Page 40]

RFC 6376                     DKIM Signatures              September 2011

   o  In-Reply-To, References

   o  List-Id, List-Help, List-Unsubscribe, List-Subscribe, List-Post,
      List-Owner, List-Archive

   If the "l=" signature tag is in use (see Section 3.5), the Content-
   Type field is also a candidate for being included as it could be
   replaced in a way that causes completely different content to be
   rendered to the receiving user.

   There are trade-offs in the decision of what constitutes the "core"
   of the message, which for some fields is a subjective concept.
   Including fields such as "Message-ID", for example, is useful if one
   considers a mechanism for being able to distinguish separate
   instances of the same message to be core content.  Similarly, "In-
   Reply-To" and "References" might be desirable to include if one
   considers message threading to be a core part of the message.

   Another class of fields that may be of interest are those that convey
   security-related information about the message, such as
   Authentication-Results [RFC5451].

   The basic rule for choosing fields to exclude is to select those
   fields for which there are multiple fields with the same name and
   fields that are modified in transit.  Examples of these are:

   o  Return-Path

   o  Received

   o  Comments, Keywords

   Note that the DKIM-Signature field is also excluded from the header
   hash because its handling is specified separately.

   Typically, it is better to exclude other optional fields because of
   the potential that additional fields of the same name will be
   legitimately added or reordered prior to verification.  There are
   likely to be legitimate exceptions to this rule because of the wide
   variety of application-specific header fields that might be applied
   to a message, some of which are unlikely to be duplicated, modified,
   or reordered.

   Signers SHOULD choose canonicalization algorithms based on the types
   of messages they process and their aversion to risk.  For example,
   e-commerce sites sending primarily purchase receipts, which are not
   expected to be processed by mailing lists or other software likely to
   modify messages, will generally prefer "simple" canonicalization.

Crocker, et al.              Standards Track                   [Page 41]

RFC 6376                     DKIM Signatures              September 2011

   Sites sending primarily person-to-person email will likely prefer to
   be more resilient to modification during transport by using "relaxed"

   Unless mail is processed through intermediaries, such as mailing
   lists that might add "unsubscribe" instructions to the bottom of the
   message body, the "l=" tag is likely to convey no additional benefit
   while providing an avenue for unauthorized addition of text to a
   message.  The use of "l=0" takes this to the extreme, allowing
   complete alteration of the text of the message without invalidating
   the signature.  Moreover, a Verifier would be within its rights to
   consider a partly signed message body as unacceptable.  Judicious use
   is advised.

5.4.2.  Signatures Involving Multiple Instances of a Field

   Signers choosing to sign an existing header field that occurs more
   than once in the message (such as Received) MUST sign the physically
   last instance of that header field in the header block.  Signers
   wishing to sign multiple instances of such a header field MUST
   include the header field name multiple times in the "h=" tag of the
   DKIM-Signature header field and MUST sign such header fields in order
   from the bottom of the header field block to the top.  The Signer MAY
   include more instances of a header field name in "h=" than there are
   actual corresponding header fields so that the signature will not
   verify if additional header fields of that name are added.


      If the Signer wishes to sign two existing Received header fields,
      and the existing header contains:

      Received: <A>
      Received: <B>
      Received: <C>

      then the resulting DKIM-Signature header field should read:

      DKIM-Signature: ... h=Received : Received :...

      and Received header fields <C> and <B> will be signed in that

Crocker, et al.              Standards Track                   [Page 42]

RFC 6376                     DKIM Signatures              September 2011

5.5.  Compute the Message Hash and Signature

   The Signer MUST compute the message hash as described in Section 3.7
   and then sign it using the selected public-key algorithm.  This will
   result in a DKIM-Signature header field that will include the body
   hash and a signature of the header hash, where that header includes
   the DKIM-Signature header field itself.

   Entities such as mailing list managers that implement DKIM and that
   modify the message or a header field (for example, inserting
   unsubscribe information) before retransmitting the message SHOULD
   check any existing signature on input and MUST make such
   modifications before re-signing the message.

5.6.  Insert the DKIM-Signature Header Field

   Finally, the Signer MUST insert the DKIM-Signature header field
   created in the previous step prior to transmitting the email.  The
   DKIM-Signature header field MUST be the same as used to compute the
   hash as described above, except that the value of the "b=" tag MUST
   be the appropriately signed hash computed in the previous step,
   signed using the algorithm specified in the "a=" tag of the DKIM-
   Signature header field and using the private key corresponding to the
   selector given in the "s=" tag of the DKIM-Signature header field, as
   chosen above in Section 5.2.

   The DKIM-Signature header field MUST be inserted before any other
   DKIM-Signature fields in the header block.

      INFORMATIVE IMPLEMENTATION NOTE: The easiest way to achieve this
      is to insert the DKIM-Signature header field at the beginning of
      the header block.  In particular, it may be placed before any
      existing Received header fields.  This is consistent with treating
      DKIM-Signature as a trace header field.

6.  Verifier Actions

   Since a Signer MAY remove or revoke a public key at any time, it is
   advised that verification occur in a timely manner.  In many
   configurations, the most timely place is during acceptance by the
   border MTA or shortly thereafter.  In particular, deferring
   verification until the message is accessed by the end user is

   A border or intermediate MTA MAY verify the message signature(s).  An
   MTA who has performed verification MAY communicate the result of that
   verification by adding a verification header field to incoming
   messages.  This simplifies things considerably for the user, who can

Crocker, et al.              Standards Track                   [Page 43]

RFC 6376                     DKIM Signatures              September 2011

   now use an existing mail user agent.  Most MUAs have the ability to
   filter messages based on message header fields or content; these
   filters would be used to implement whatever policy the user wishes
   with respect to unsigned mail.

   A verifying MTA MAY implement a policy with respect to unverifiable
   mail, regardless of whether or not it applies the verification header
   field to signed messages.

   Verifiers MUST produce a result that is semantically equivalent to
   applying the steps listed in Sections 6.1, 6.1.1, and 6.1.2 in order.
   In practice, several of these steps can be performed in parallel in
   order to improve performance.

6.1.  Extract Signatures from the Message

   The order in which Verifiers try DKIM-Signature header fields is not
   defined; Verifiers MAY try signatures in any order they like.  For
   example, one implementation might try the signatures in textual
   order, whereas another might try signatures by identities that match
   the contents of the From header field before trying other signatures.
   Verifiers MUST NOT attribute ultimate meaning to the order of
   multiple DKIM-Signature header fields.  In particular, there is
   reason to believe that some relays will reorder the header fields in
   potentially arbitrary ways.

      INFORMATIVE IMPLEMENTATION NOTE: Verifiers might use the order as
      a clue to signing order in the absence of any other information.
      However, other clues as to the semantics of multiple signatures
      (such as correlating the signing host with Received header fields)
      might also be considered.

   Survivability of signatures after transit is not guaranteed, and
   signatures can fail to verify through no fault of the Signer.
   Therefore, a Verifier SHOULD NOT treat a message that has one or more
   bad signatures and no good signatures differently from a message with
   no signature at all.

   When a signature successfully verifies, a Verifier will either stop
   processing or attempt to verify any other signatures, at the
   discretion of the implementation.  A Verifier MAY limit the number of
   signatures it tries, in order to avoid denial-of-service attacks (see
   Section 8.4 for further discussion).

   In the following description, text reading "return status
   (explanation)" (where "status" is one of "PERMFAIL" or "TEMPFAIL")
   means that the Verifier MUST immediately cease processing that
   signature.  The Verifier SHOULD proceed to the next signature, if one

Crocker, et al.              Standards Track                   [Page 44]

RFC 6376                     DKIM Signatures              September 2011

   is present, and completely ignore the bad signature.  If the status
   is "PERMFAIL", the signature failed and should not be reconsidered.
   If the status is "TEMPFAIL", the signature could not be verified at
   this time but may be tried again later.  A Verifier MAY either
   arrange to defer the message for later processing or try another
   signature; if no good signature is found and any of the signatures
   resulted in a TEMPFAIL status, the Verifier MAY arrange to defer the
   message for later processing.  The "(explanation)" is not normative
   text; it is provided solely for clarification.

   Verifiers that are prepared to validate multiple signature header
   fields SHOULD proceed to the next signature header field, if one
   exists.  However, Verifiers MAY make note of the fact that an invalid
   signature was present for consideration at a later step.

      INFORMATIVE NOTE: The rationale of this requirement is to permit
      messages that have invalid signatures but also a valid signature
      to work.  For example, a mailing list exploder might opt to leave
      the original submitter signature in place even though the exploder
      knows that it is modifying the message in some way that will break
      that signature, and the exploder inserts its own signature.  In
      this case, the message should succeed even in the presence of the
      known-broken signature.

   For each signature to be validated, the following steps should be
   performed in such a manner as to produce a result that is
   semantically equivalent to performing them in the indicated order.

6.1.1.  Validate the Signature Header Field

   Implementers MUST meticulously validate the format and values in the
   DKIM-Signature header field; any inconsistency or unexpected values
   MUST cause the header field to be completely ignored and the Verifier
   to return PERMFAIL (signature syntax error).  Being "liberal in what
   you accept" is definitely a bad strategy in this security context.
   Note, however, that this does not include the existence of unknown
   tags in a DKIM-Signature header field, which are explicitly
   permitted.  Verifiers MUST return PERMFAIL (incompatible version)
   when presented a DKIM-Signature header field with a "v=" tag that is
   inconsistent with this specification.

      INFORMATIVE IMPLEMENTATION NOTE: An implementation may, of course,
      choose to also verify signatures generated by older versions of
      this specification.

Crocker, et al.              Standards Track                   [Page 45]

RFC 6376                     DKIM Signatures              September 2011

   If any tag listed as "required" in Section 3.5 is omitted from the
   DKIM-Signature header field, the Verifier MUST ignore the DKIM-
   Signature header field and return PERMFAIL (signature missing
   required tag).

      INFORMATIVE NOTE: The tags listed as required in Section 3.5 are
      "v=", "a=", "b=", "bh=", "d=", "h=", and "s=".  Should there be a
      conflict between this note and Section 3.5, Section 3.5 is

   If the DKIM-Signature header field does not contain the "i=" tag, the
   Verifier MUST behave as though the value of that tag were "@d", where
   "d" is the value from the "d=" tag.

   Verifiers MUST confirm that the domain specified in the "d=" tag is
   the same as or a parent domain of the domain part of the "i=" tag.
   If not, the DKIM-Signature header field MUST be ignored, and the
   Verifier should return PERMFAIL (domain mismatch).

   If the "h=" tag does not include the From header field, the Verifier
   MUST ignore the DKIM-Signature header field and return PERMFAIL (From
   field not signed).

   Verifiers MAY ignore the DKIM-Signature header field and return
   PERMFAIL (signature expired) if it contains an "x=" tag and the
   signature has expired.

   Verifiers MAY ignore the DKIM-Signature header field if the domain
   used by the Signer in the "d=" tag is not associated with a valid
   signing entity.  For example, signatures with "d=" values such as
   "com" and "co.uk" could be ignored.  The list of unacceptable domains
   SHOULD be configurable.

   Verifiers MAY ignore the DKIM-Signature header field and return
   PERMFAIL (unacceptable signature header) for any other reason, for
   example, if the signature does not sign header fields that the
   Verifier views to be essential.  As a case in point, if MIME header
   fields are not signed, certain attacks may be possible that the
   Verifier would prefer to avoid.

6.1.2.  Get the Public Key

   The public key for a signature is needed to complete the verification
   process.  The process of retrieving the public key depends on the
   query type as defined by the "q=" tag in the DKIM-Signature header
   field.  Obviously, a public key need only be retrieved if the process
   of extracting the signature information is completely successful.

Crocker, et al.              Standards Track                   [Page 46]

RFC 6376                     DKIM Signatures              September 2011

   Details of key management and representation are described in
   Section 3.6.  The Verifier MUST validate the key record and MUST
   ignore any public-key records that are malformed.

      NOTE: The use of a wildcard TXT RR that covers a queried DKIM
      domain name will produce a response to a DKIM query that is
      unlikely to be a valid DKIM key record.  This problem is not
      specific to DKIM and applies to many other types of queries.
      Client software that processes DNS responses needs to take this
      problem into account.

   When validating a message, a Verifier MUST perform the following
   steps in a manner that is semantically the same as performing them in
   the order indicated; in some cases, the implementation may
   parallelize or reorder these steps, as long as the semantics remain

   1.  The Verifier retrieves the public key as described in Section 3.6
       using the algorithm in the "q=" tag, the domain from the "d="
       tag, and the selector from the "s=" tag.

   2.  If the query for the public key fails to respond, the Verifier
       MAY seek a later verification attempt by returning TEMPFAIL (key

   3.  If the query for the public key fails because the corresponding
       key record does not exist, the Verifier MUST immediately return
       PERMFAIL (no key for signature).

   4.  If the query for the public key returns multiple key records, the
       Verifier can choose one of the key records or may cycle through
       the key records, performing the remainder of these steps on each
       record at the discretion of the implementer.  The order of the
       key records is unspecified.  If the Verifier chooses to cycle
       through the key records, then the "return ..." wording in the
       remainder of this section means "try the next key record, if any;
       if none, return to try another signature in the usual way".

   5.  If the result returned from the query does not adhere to the
       format defined in this specification, the Verifier MUST ignore
       the key record and return PERMFAIL (key syntax error).  Verifiers
       are urged to validate the syntax of key records carefully to
       avoid attempted attacks.  In particular, the Verifier MUST ignore
       keys with a version code ("v=" tag) that they do not implement.

Crocker, et al.              Standards Track                   [Page 47]

RFC 6376                     DKIM Signatures              September 2011

   6.  If the "h=" tag exists in the public-key record and the hash
       algorithm implied by the "a=" tag in the DKIM-Signature header
       field is not included in the contents of the "h=" tag, the
       Verifier MUST ignore the key record and return PERMFAIL
       (inappropriate hash algorithm).

   7.  If the public-key data (the "p=" tag) is empty, then this key has
       been revoked and the Verifier MUST treat this as a failed
       signature check and return PERMFAIL (key revoked).  There is no
       defined semantic difference between a key that has been revoked
       and a key record that has been removed.

   8.  If the public-key data is not suitable for use with the algorithm
       and key types defined by the "a=" and "k=" tags in the DKIM-
       Signature header field, the Verifier MUST immediately return
       PERMFAIL (inappropriate key algorithm).

6.1.3.  Compute the Verification

   Given a Signer and a public key, verifying a signature consists of
   actions semantically equivalent to the following steps.

   1.  Based on the algorithm defined in the "c=" tag, the body length
       specified in the "l=" tag, and the header field names in the "h="
       tag, prepare a canonicalized version of the message as is
       described in Section 3.7 (note that this canonicalized version
       does not actually replace the original content).  When matching
       header field names in the "h=" tag against the actual message
       header field, comparisons MUST be case-insensitive.

   2.  Based on the algorithm indicated in the "a=" tag, compute the
       message hashes from the canonical copy as described in
       Section 3.7.

   3.  Verify that the hash of the canonicalized message body computed
       in the previous step matches the hash value conveyed in the "bh="
       tag.  If the hash does not match, the Verifier SHOULD ignore the
       signature and return PERMFAIL (body hash did not verify).

   4.  Using the signature conveyed in the "b=" tag, verify the
       signature against the header hash using the mechanism appropriate
       for the public-key algorithm described in the "a=" tag.  If the
       signature does not validate, the Verifier SHOULD ignore the
       signature and return PERMFAIL (signature did not verify).

Crocker, et al.              Standards Track                   [Page 48]

RFC 6376                     DKIM Signatures              September 2011

   5.  Otherwise, the signature has correctly verified.

      INFORMATIVE IMPLEMENTER'S NOTE: Implementations might wish to
      initiate the public-key query in parallel with calculating the
      hash as the public key is not needed until the final decryption is
      calculated.  Implementations may also verify the signature on the
      message header before validating that the message hash listed in
      the "bh=" tag in the DKIM-Signature header field matches that of
      the actual message body; however, if the body hash does not match,
      the entire signature must be considered to have failed.

   A body length specified in the "l=" tag of the signature limits the
   number of bytes of the body passed to the verification algorithm.
   All data beyond that limit is not validated by DKIM.  Hence,
   Verifiers might treat a message that contains bytes beyond the
   indicated body length with suspicion and can choose to treat the
   signature as if it were invalid (e.g., by returning PERMFAIL
   (unsigned content)).

   Should the algorithm reach this point, the verification has
   succeeded, and DKIM reports SUCCESS for this signature.

6.2.  Communicate Verification Results

   Verifiers wishing to communicate the results of verification to other
   parts of the mail system may do so in whatever manner they see fit.
   For example, implementations might choose to add an email header
   field to the message before passing it on.  Any such header field
   SHOULD be inserted before any existing DKIM-Signature or preexisting
   authentication status header fields in the header field block.  The
   Authentication-Results: header field ([RFC5451]) MAY be used for this

      INFORMATIVE ADVICE to MUA filter writers: Patterns intended to
      search for results header fields to visibly mark authenticated
      mail for end users should verify that such a header field was
      added by the appropriate verifying domain and that the verified
      identity matches the author identity that will be displayed by the
      MUA.  In particular, MUA filters should not be influenced by bogus
      results header fields added by attackers.  To circumvent this
      attack, Verifiers MAY wish to request deletion of existing results
      header fields after verification and before arranging to add a new
      header field.

Crocker, et al.              Standards Track                   [Page 49]

RFC 6376                     DKIM Signatures              September 2011

6.3.  Interpret Results/Apply Local Policy

   It is beyond the scope of this specification to describe what actions
   an Identity Assessor can make, but mail carrying a validated SDID
   presents an opportunity to an Identity Assessor that unauthenticated
   email does not.  Specifically, an authenticated email creates a
   predictable identifier by which other decisions can reliably be
   managed, such as trust and reputation.  Conversely, unauthenticated
   email lacks a reliable identifier that can be used to assign trust
   and reputation.  It is reasonable to treat unauthenticated email as
   lacking any trust and having no positive reputation.

   In general, modules that consume DKIM verification output SHOULD NOT
   determine message acceptability based solely on a lack of any
   signature or on an unverifiable signature; such rejection would cause
   severe interoperability problems.  If an MTA does wish to reject such
   messages during an SMTP session (for example, when communicating with
   a peer who, by prior agreement, agrees to only send signed messages),
   and a signature is missing or does not verify, the handling MTA
   SHOULD use a 550/5.7.x reply code.

   Where the Verifier is integrated within the MTA and it is not
   possible to fetch the public key, perhaps because the key server is
   not available, a temporary failure message MAY be generated using a
   451/4.7.5 reply code, such as:

   451 4.7.5 Unable to verify signature - key server unavailable

   Temporary failures such as inability to access the key server or
   other external service are the only conditions that SHOULD use a 4xx
   SMTP reply code.  In particular, cryptographic signature verification
   failures MUST NOT provoke 4xx SMTP replies.

   Once the signature has been verified, that information MUST be
   conveyed to the Identity Assessor (such as an explicit allow/
   whitelist and reputation system) and/or to the end user.  If the SDID
   is not the same as the address in the From: header field, the mail
   system SHOULD take pains to ensure that the actual SDID is clear to
   the reader.

   While the symptoms of a failed verification are obvious -- the
   signature doesn't verify -- establishing the exact cause can be more
   difficult.  If a selector cannot be found, is that because the
   selector has been removed, or was the value changed somehow in
   transit?  If the signature line is missing, is that because it was
   never there, or was it removed by an overzealous filter?  For
   diagnostic purposes, the exact reason why the verification fails
   SHOULD be made available and possibly recorded in the system logs.

Crocker, et al.              Standards Track                   [Page 50]

RFC 6376                     DKIM Signatures              September 2011

   If the email cannot be verified, then it SHOULD be treated the same
   as all unverified email, regardless of whether or not it looks like
   it was signed.

   See Section 8.15 for additional discussion.

7.  IANA Considerations

   DKIM has registered namespaces with IANA.  In all cases, new values
   are assigned only for values that have been documented in a published
   RFC that has IETF Consensus [RFC5226].

   This memo updates these registries as described below.  Of note is
   the addition of a new "status" column.  All registrations into these
   namespaces MUST include the name being registered, the document in
   which it was registered or updated, and an indication of its current
   status, which MUST be one of "active" (in current use) or "historic"
   (no longer in current use).

   No new tags are defined in this specification compared to [RFC4871],
   but one has been designated as "historic".

   Also, the "Email Authentication Methods" registry is revised to refer
   to this update.

7.1.  Email Authentication Methods Registry

   The "Email Authentication Methods" registry is updated to indicate
   that "dkim" is defined in this memo.

7.2.  DKIM-Signature Tag Specifications

   A DKIM-Signature provides for a list of tag specifications.  IANA has
   established the "DKIM-Signature Tag Specifications" registry for tag
   specifications that can be used in DKIM-Signature fields.

Crocker, et al.              Standards Track                   [Page 51]

RFC 6376                     DKIM Signatures              September 2011

                    | TYPE | REFERENCE       | STATUS |
                    |   v  | (this document) | active |
                    |   a  | (this document) | active |
                    |   b  | (this document) | active |
                    |  bh  | (this document) | active |
                    |   c  | (this document) | active |
                    |   d  | (this document) | active |
                    |   h  | (this document) | active |
                    |   i  | (this document) | active |
                    |   l  | (this document) | active |
                    |   q  | (this document) | active |
                    |   s  | (this document) | active |
                    |   t  | (this document) | active |
                    |   x  | (this document) | active |
                    |   z  | (this document) | active |

    Table 1: DKIM-Signature Tag Specifications Registry Updated Values

7.3.  DKIM-Signature Query Method Registry

   The "q=" tag-spec (specified in Section 3.5) provides for a list of
   query methods.

   IANA has established the "DKIM-Signature Query Method" registry for
   mechanisms that can be used to retrieve the key that will permit
   validation processing of a message signed using DKIM.

               | TYPE | OPTION | REFERENCE       | STATUS |
               |  dns |   txt  | (this document) | active |

       Table 2: DKIM-Signature Query Method Registry Updated Values

7.4.  DKIM-Signature Canonicalization Registry

   The "c=" tag-spec (specified in Section 3.5) provides for a specifier
   for canonicalization algorithms for the header and body of the

   IANA has established the "DKIM-Signature Canonicalization Header"
   Registry for algorithms for converting a message into a canonical
   form before signing or verifying using DKIM.

Crocker, et al.              Standards Track                   [Page 52]

RFC 6376                     DKIM Signatures              September 2011

                  |   TYPE  | REFERENCE       | STATUS |
                  |  simple | (this document) | active |
                  | relaxed | (this document) | active |

     Table 3: DKIM-Signature Canonicalization Header Registry Updated

                  |   TYPE  | REFERENCE       | STATUS |
                  |  simple | (this document) | active |
                  | relaxed | (this document) | active |

   Table 4: DKIM-Signature Canonicalization Body Registry Updated Values

7.5.  _domainkey DNS TXT Resource Record Tag Specifications

   A _domainkey DNS TXT RR provides for a list of tag specifications.
   IANA has established the DKIM "_domainkey DNS TXT Record Tag
   Specifications" registry for tag specifications that can be used in
   DNS TXT resource records.

                   | TYPE | REFERENCE       | STATUS   |
                   |   v  | (this document) | active   |
                   |   g  | [RFC4871]       | historic |
                   |   h  | (this document) | active   |
                   |   k  | (this document) | active   |
                   |   n  | (this document) | active   |
                   |   p  | (this document) | active   |
                   |   s  | (this document) | active   |
                   |   t  | (this document) | active   |

      Table 5: _domainkey DNS TXT Record Tag Specifications Registry
                              Updated Values

7.6.  DKIM Key Type Registry

   The "k=" <key-k-tag> (specified in Section 3.6.1) and the "a=" <sig-
   a-tag-k> (specified in Section 3.5) tags provide for a list of
   mechanisms that can be used to decode a DKIM signature.

Crocker, et al.              Standards Track                   [Page 53]

RFC 6376                     DKIM Signatures              September 2011

   IANA has established the "DKIM Key Type" registry for such

                       | TYPE | REFERENCE | STATUS |
                       |  rsa | [RFC3447] | active |

              Table 6: DKIM Key Type Registry Updated Values

7.7.  DKIM Hash Algorithms Registry

   The "h=" <key-h-tag> (specified in Section 3.6.1) and the "a=" <sig-
   a-tag-h> (specified in Section 3.5) tags provide for a list of
   mechanisms that can be used to produce a digest of message data.

   IANA has established the "DKIM Hash Algorithms" registry for such

                  |  TYPE  | REFERENCE         | STATUS |
                  |  sha1  | [FIPS-180-3-2008] | active |
                  | sha256 | [FIPS-180-3-2008] | active |

           Table 7: DKIM Hash Algorithms Registry Updated Values

7.8.  DKIM Service Types Registry

   The "s=" <key-s-tag> tag (specified in Section 3.6.1) provides for a
   list of service types to which this selector may apply.

   IANA has established the "DKIM Service Types" registry for service

                   |  TYPE | REFERENCE       | STATUS |
                   | email | (this document) | active |
                   |   *   | (this document) | active |

            Table 8: DKIM Service Types Registry Updated Values

Crocker, et al.              Standards Track                   [Page 54]

RFC 6376                     DKIM Signatures              September 2011

7.9.  DKIM Selector Flags Registry

   The "t=" <key-t-tag> tag (specified in Section 3.6.1) provides for a
   list of flags to modify interpretation of the selector.

   IANA has established the "DKIM Selector Flags" registry for
   additional flags.

                    | TYPE | REFERENCE       | STATUS |
                    |   y  | (this document) | active |
                    |   s  | (this document) | active |

           Table 9: DKIM Selector Flags Registry Updated Values

7.10.  DKIM-Signature Header Field

   IANA has added DKIM-Signature to the "Permanent Message Header Field
   Names" registry (see [RFC3864]) for the "mail" protocol, using this
   document as the reference.

8.  Security Considerations

   It has been observed that any introduced mechanism that attempts to
   stem the flow of spam is subject to intensive attack.  DKIM needs to
   be carefully scrutinized to identify potential attack vectors and the
   vulnerability to each.  See also [RFC4686].

8.1.  ASCII Art Attacks

   The relaxed body canonicalization algorithm may enable certain types
   of extremely crude "ASCII Art" attacks where a message may be
   conveyed by adjusting the spacing between words.  If this is a
   concern, the "simple" body canonicalization algorithm should be used

8.2.  Misuse of Body Length Limits ("l=" Tag)

   Use of the "l=" tag might allow display of fraudulent content without
   appropriate warning to end users.  The "l=" tag is intended for
   increasing signature robustness when sending to mailing lists that
   both modify their content and do not sign their modified messages.
   However, using the "l=" tag enables attacks in which an intermediary
   with malicious intent can modify a message to include content that
   solely benefits the attacker.  It is possible for the appended

Crocker, et al.              Standards Track                   [Page 55]

RFC 6376                     DKIM Signatures              September 2011

   content to completely replace the original content in the end
   recipient's eyes and to defeat duplicate message detection

   An example of such an attack includes altering the MIME structure,
   exploiting lax HTML parsing in the MUA, and defeating duplicate
   message detection algorithms.

   To avoid this attack, Signers should be extremely wary of using this
   tag, and Assessors might wish to ignore signatures that use the tag.

8.3.  Misappropriated Private Key

   As with any other security application that uses private- or public-
   key pairs, DKIM requires caution around the handling and protection
   of keys.  A compromised private key or access to one means an
   intruder or malware can send mail signed by the domain that
   advertises the matching public key.

   Thus, private keys issued to users, rather than one used by an
   ADministrative Management Domain (ADMD) itself, create the usual
   problem of securing data stored on personal resources that can affect
   the ADMD.

   A more secure architecture involves sending messages through an
   outgoing MTA that can authenticate the submitter using existing
   techniques (e.g., SMTP Authentication), possibly validate the message
   itself (e.g., verify that the header is legitimate and that the
   content passes a spam content check), and sign the message using a
   key appropriate for the submitter address.  Such an MTA can also
   apply controls on the volume of outgoing mail each user is permitted
   to originate in order to further limit the ability of malware to
   generate bulk email.

8.4.  Key Server Denial-of-Service Attacks

   Since the key servers are distributed (potentially separate for each
   domain), the number of servers that would need to be attacked to
   defeat this mechanism on an Internet-wide basis is very large.
   Nevertheless, key servers for individual domains could be attacked,
   impeding the verification of messages from that domain.  This is not
   significantly different from the ability of an attacker to deny
   service to the mail exchangers for a given domain, although it
   affects outgoing, not incoming, mail.

   A variation on this attack involves a very large amount of mail being
   sent using spoofed signatures from a given domain: the key servers
   for that domain could be overwhelmed with requests in a denial-of-

Crocker, et al.              Standards Track                   [Page 56]

RFC 6376                     DKIM Signatures              September 2011

   service attack (see [RFC4732]).  However, given the low overhead of
   verification compared with handling of the email message itself, such
   an attack would be difficult to mount.

8.5.  Attacks against the DNS

   Since the DNS is a required binding for key services, specific
   attacks against the DNS must be considered.

   While the DNS is currently insecure [RFC3833], these security
   problems are the motivation behind DNS Security (DNSSEC) [RFC4033],
   and all users of the DNS will reap the benefit of that work.

   DKIM is only intended as a "sufficient" method of proving
   authenticity.  It is not intended to provide strong cryptographic
   proof about authorship or contents.  Other technologies such as
   OpenPGP [RFC4880] and S/MIME [RFC5751] address those requirements.

   A second security issue related to the DNS revolves around the
   increased DNS traffic as a consequence of fetching selector-based
   data as well as fetching signing domain policy.  Widespread
   deployment of DKIM will result in a significant increase in DNS
   queries to the claimed signing domain.  In the case of forgeries on a
   large scale, DNS servers could see a substantial increase in queries.

   A specific DNS security issue that should be considered by DKIM
   Verifiers is the name chaining attack described in Section 2.3 of
   [RFC3833].  A DKIM Verifier, while verifying a DKIM-Signature header
   field, could be prompted to retrieve a key record of an attacker's
   choosing.  This threat can be minimized by ensuring that name
   servers, including recursive name servers, used by the Verifier
   enforce strict checking of "glue" and other additional information in
   DNS responses and are therefore not vulnerable to this attack.

8.6.  Replay/Spam Attacks

   In this attack, a spammer sends a piece of spam through an MTA that
   signs it, banking on the reputation of the signing domain (e.g., a
   large popular mailbox provider) rather than its own, and then re-
   sends that message to a large number of intended recipients.  The
   recipients observe the valid signature from the well-known domain,
   elevating their trust in the message and increasing the likelihood of
   delivery and presentation to the user.

   Partial solutions to this problem involve the use of reputation
   services to convey the fact that the specific email address is being
   used for spam and that messages from that Signer are likely to be
   spam.  This requires a real-time detection mechanism in order to

Crocker, et al.              Standards Track                   [Page 57]

RFC 6376                     DKIM Signatures              September 2011

   react quickly enough.  However, such measures might be prone to
   abuse, if, for example, an attacker re-sent a large number of
   messages received from a victim in order to make the victim appear to
   be a spammer.

   Large Verifiers might be able to detect unusually large volumes of
   mails with the same signature in a short time period.  Smaller
   Verifiers can get substantially the same volume of information via
   existing collaborative systems.

8.7.  Limits on Revoking Keys

   When a large domain detects undesirable behavior on the part of one
   of its users, it might wish to revoke the key used to sign that
   user's messages in order to disavow responsibility for messages that
   have not yet been verified or that are the subject of a replay
   attack.  However, the ability of the domain to do so can be limited
   if the same key, for scalability reasons, is used to sign messages
   for many other users.  Mechanisms for explicitly revoking keys on a
   per-address basis have been proposed but require further study as to
   their utility and the DNS load they represent.

8.8.  Intentionally Malformed Key Records

   It is possible for an attacker to publish key records in DNS that are
   intentionally malformed, with the intent of causing a denial-of-
   service attack on a non-robust Verifier implementation.  The attacker
   could then cause a Verifier to read the malformed key record by
   sending a message to one of its users referencing the malformed
   record in a (not necessarily valid) signature.  Verifiers MUST
   thoroughly verify all key records retrieved from the DNS and be
   robust against intentionally as well as unintentionally malformed key

8.9.  Intentionally Malformed DKIM-Signature Header Fields

   Verifiers MUST be prepared to receive messages with malformed DKIM-
   Signature header fields and thoroughly verify the header field before
   depending on any of its contents.

8.10.  Information Leakage

   An attacker could determine when a particular signature was verified
   by using a per-message selector and then monitoring their DNS traffic
   for the key lookup.  This would act as the equivalent of a "web bug"
   for verification time rather than the time the message was read.

Crocker, et al.              Standards Track                   [Page 58]

RFC 6376                     DKIM Signatures              September 2011

8.11.  Remote Timing Attacks

   In some cases, it may be possible to extract private keys using a
   remote timing attack [BONEH03].  Implementations should consider
   obfuscating the timing to prevent such attacks.

8.12.  Reordered Header Fields

   Existing standards allow intermediate MTAs to reorder header fields.
   If a Signer signs two or more header fields of the same name, this
   can cause spurious verification errors on otherwise legitimate
   messages.  In particular, Signers that sign any existing DKIM-
   Signature fields run the risk of having messages incorrectly fail to

8.13.  RSA Attacks

   An attacker could create a large RSA signing key with a small
   exponent, thus requiring that the verification key have a large
   exponent.  This will force Verifiers to use considerable computing
   resources to verify the signature.  Verifiers might avoid this attack
   by refusing to verify signatures that reference selectors with public
   keys having unreasonable exponents.

   In general, an attacker might try to overwhelm a Verifier by flooding
   it with messages requiring verification.  This is similar to other
   MTA denial-of-service attacks and should be dealt with in a similar

8.14.  Inappropriate Signing by Parent Domains

   The trust relationship described in Section 3.10 could conceivably be
   used by a parent domain to sign messages with identities in a
   subdomain not administratively related to the parent.  For example,
   the ".com" registry could create messages with signatures using an
   "i=" value in the example.com domain.  There is no general solution
   to this problem, since the administrative cut could occur anywhere in
   the domain name.  For example, in the domain "example.podunk.ca.us",
   there are three administrative cuts (podunk.ca.us, ca.us, and us),
   any of which could create messages with an identity in the full

      INFORMATIVE NOTE: This is considered an acceptable risk for the
      same reason that it is acceptable for domain delegation.  For
      example, in the case above, any of the domains could potentially
      simply delegate "example.podunk.ca.us" to a server of their choice

Crocker, et al.              Standards Track                   [Page 59]

RFC 6376                     DKIM Signatures              September 2011

      and completely replace all DNS-served information.  Note that a
      Verifier MAY ignore signatures that come from an unlikely domain
      such as ".com", as discussed in Section 6.1.1.

8.15.  Attacks Involving Extra Header Fields

   Many email components, including MTAs, MSAs, MUAs, and filtering
   modules, implement message format checks only loosely.  This is done
   out of years of industry pressure to be liberal in what is accepted
   into the mail stream for the sake of reducing support costs;
   improperly formed messages are often silently fixed in transit,
   delivered unrepaired, or displayed inappropriately (e.g., by showing
   only the first of multiple From: fields).

   Agents that evaluate or apply DKIM output need to be aware that a
   DKIM Signer can sign messages that are malformed (e.g., violate
   [RFC5322], such as by having multiple instances of a field that is
   only permitted once), that become malformed in transit, or that
   contain header or body content that is not true or valid.  Use of
   DKIM on such messages might constitute an attack against a receiver,
   especially where additional credence is given to a signed message
   without adequate evaluation of the Signer.

   These can represent serious attacks, but they have nothing to do with
   DKIM; they are attacks on the recipient or on the wrongly identified

   Moreover, an agent would be incorrect to infer that all instances of
   a header field are signed just because one is.

   A genuine signature from the domain under attack can be obtained by
   legitimate means, but extra header fields can then be added, either
   by interception or by replay.  In this scenario, DKIM can aid in
   detecting addition of specific fields in transit.  This is done by
   having the Signer list the field name(s) in the "h=" tag an extra
   time (e.g., "h=from:from:..." for a message with one From field), so
   that addition of an instance of that field downstream will render the
   signature unable to be verified.  (See Section 3.5 for details.)
   This, in essence, is an explicit indication that the Signer
   repudiates responsibility for such a malformed message.

   DKIM signs and validates the data it is told to and works correctly.
   So in this case, DKIM has done its job of delivering a validated
   domain (the "d=" value) and, given the semantics of a DKIM signature,
   essentially the Signer has taken some responsibility for a
   problematic message.  It is up to the Identity Assessor or some other

Crocker, et al.              Standards Track                   [Page 60]

RFC 6376                     DKIM Signatures              September 2011

   subsequent agent to act on such messages as needed, such as degrading
   the trust of the message (or, indeed, of the Signer), warning the
   recipient, or even refusing delivery.

   All components of the mail system that perform loose enforcement of
   other mail standards will need to revisit that posture when
   incorporating DKIM, especially when considering matters of potential
   attacks such as those described.

9.  References

9.1.  Normative References

              U.S. Department of Commerce, "Secure Hash Standard", FIPS
              PUB 180-3, October 2008.

              "Information Technology - ASN.1 encoding rules:
              Specification of Basic Encoding Rules (BER), Canonical
              Encoding Rules (CER) and Distinguished Encoding Rules
              (DER)", 1997.

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

   [RFC2045]  Freed, N. and N. Borenstein, "Multipurpose Internet Mail
              Extensions (MIME) Part One: Format of Internet Message
              Bodies", RFC 2045, November 1996.

   [RFC2049]  Freed, N. and N. Borenstein, "Multipurpose Internet Mail
              Extensions (MIME) Part Five: Conformance Criteria and
              Examples", RFC 2049, November 1996.

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

   [RFC3447]  Jonsson, J. and B. Kaliski, "Public-Key Cryptography
              Standards (PKCS) #1: RSA Cryptography Specifications
              Version 2.1", RFC 3447, February 2003.

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

   [RFC5321]  Klensin, J., "Simple Mail Transfer Protocol", RFC 5321,
              October 2008.

Crocker, et al.              Standards Track                   [Page 61]

RFC 6376                     DKIM Signatures              September 2011

   [RFC5322]  Resnick, P., Ed., "Internet Message Format", RFC 5322,
              October 2008.

   [RFC5598]  Crocker, D., "Internet Mail Architecture", RFC 5598,
              July 2009.

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

9.2.  Informative References

   [BONEH03]  "Remote Timing Attacks are Practical", Proceedings 12th
              USENIX Security Symposium, 2003.

   [RFC2047]  Moore, K., "MIME (Multipurpose Internet Mail Extensions)
              Part Three: Message Header Extensions for Non-ASCII Text",
              RFC 2047, November 1996.

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

   [RFC3766]  Orman, H. and P. Hoffman, "Determining Strengths For
              Public Keys Used For Exchanging Symmetric Keys", BCP 86,
              RFC 3766, April 2004.

   [RFC3833]  Atkins, D. and R. Austein, "Threat Analysis of the Domain
              Name System (DNS)", RFC 3833, August 2004.

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

   [RFC4033]  Arends, R., Austein, R., Larson, M., Massey, D., and S.
              Rose, "DNS Security Introduction and Requirements",
              RFC 4033, March 2005.

   [RFC4409]  Gellens, R. and J. Klensin, "Message Submission for Mail",
              RFC 4409, April 2006.

   [RFC4686]  Fenton, J., "Analysis of Threats Motivating DomainKeys
              Identified Mail (DKIM)", RFC 4686, September 2006.

   [RFC4732]  Handley, M., Rescorla, E., and IAB, "Internet Denial-of-
              Service Considerations", RFC 4732, December 2006.

Crocker, et al.              Standards Track                   [Page 62]

RFC 6376                     DKIM Signatures              September 2011

   [RFC4870]  Delany, M., "Domain-Based Email Authentication Using
              Public Keys Advertised in the DNS (DomainKeys)", RFC 4870,
              May 2007.

   [RFC4871]  Allman, E., Callas, J., Delany, M., Libbey, M., Fenton,
              J., and M. Thomas, "DomainKeys Identified Mail (DKIM)
              Signatures", RFC 4871, May 2007.

   [RFC4880]  Callas, J., Donnerhacke, L., Finney, H., Shaw, D., and R.
              Thayer, "OpenPGP Message Format", RFC 4880, November 2007.

   [RFC5226]  Narten, T. and H. Alvestrand, "Guidelines for Writing an
              IANA Considerations Section in RFCs", BCP 26, RFC 5226,
              May 2008.

   [RFC5451]  Kucherawy, M., "Message Header Field for Indicating
              Message Authentication Status", RFC 5451, April 2009.

   [RFC5585]  Hansen, T., Crocker, D., and P. Hallam-Baker, "DomainKeys
              Identified Mail (DKIM) Service Overview", RFC 5585,
              July 2009.

   [RFC5672]  Crocker, D., "RFC 4871 DomainKeys Identified Mail (DKIM)
              Signatures -- Update", RFC 5672, August 2009.

   [RFC5751]  Ramsdell, B. and S. Turner, "Secure/Multipurpose Internet
              Mail Extensions (S/MIME) Version 3.2 Message
              Specification", RFC 5751, January 2010.

   [RFC5863]  Hansen, T., Siegel, E., Hallam-Baker, P., and D. Crocker,
              "DomainKeys Identified Mail (DKIM) Development,
              Deployment, and Operations", RFC 5863, May 2010.

   [RFC6377]  Kucherawy, M., "DomainKeys Identified Mail (DKIM) and
              Mailing Lists", RFC 6377, September 2011.

Crocker, et al.              Standards Track                   [Page 63]

RFC 6376                     DKIM Signatures              September 2011

Appendix A.  Example of Use (INFORMATIVE)

   This section shows the complete flow of an email from submission to
   final delivery, demonstrating how the various components fit
   together.  The key used in this example is shown in Appendix C.

A.1.  The User Composes an Email

   From: Joe SixPack <joe@football.example.com>
   To: Suzie Q <suzie@shopping.example.net>
   Subject: Is dinner ready?
   Date: Fri, 11 Jul 2003 21:00:37 -0700 (PDT)
   Message-ID: <20030712040037.46341.5F8J@football.example.com>


   We lost the game.  Are you hungry yet?


                   Figure 1: The User Composes an Email

Crocker, et al.              Standards Track                   [Page 64]

RFC 6376                     DKIM Signatures              September 2011

A.2.  The Email is Signed

   This email is signed by the example.com outbound email server and now
   looks like this:

   DKIM-Signature: v=1; a=rsa-sha256; s=brisbane; d=example.com;
        c=simple/simple; q=dns/txt; i=joe@football.example.com;
        h=Received : From : To : Subject : Date : Message-ID;
   Received: from client1.football.example.com  []
        by submitserver.example.com with SUBMISSION;
        Fri, 11 Jul 2003 21:01:54 -0700 (PDT)
   From: Joe SixPack <joe@football.example.com>
   To: Suzie Q <suzie@shopping.example.net>
   Subject: Is dinner ready?
   Date: Fri, 11 Jul 2003 21:00:37 -0700 (PDT)
   Message-ID: <20030712040037.46341.5F8J@football.example.com>


   We lost the game.  Are you hungry yet?


                       Figure 2: The Email is Signed

   The signing email server requires access to the private key
   associated with the "brisbane" selector to generate this signature.

Crocker, et al.              Standards Track                   [Page 65]

RFC 6376                     DKIM Signatures              September 2011

A.3.  The Email Signature is Verified

   The signature is normally verified by an inbound SMTP server or
   possibly the final delivery agent.  However, intervening MTAs can
   also perform this verification if they choose to do so.  The
   verification process uses the domain "example.com" extracted from the
   "d=" tag and the selector "brisbane" from the "s=" tag in the DKIM-
   Signature header field to form the DNS DKIM query for:

   Signature verification starts with the physically last Received
   header field, the From header field, and so forth, in the order
   listed in the "h=" tag.  Verification follows with a single CRLF
   followed by the body (starting with "Hi.").  The email is canonically
   prepared for verifying with the "simple" method.  The result of the
   query and subsequent verification of the signature is stored (in this
   example) in the X-Authentication-Results header field line.  After
   successful verification, the email looks like this:

   X-Authentication-Results: shopping.example.net
     header.from=joe@football.example.com; dkim=pass
   Received: from mout23.football.example.com (
     by shopping.example.net with SMTP;
     Fri, 11 Jul 2003 21:01:59 -0700 (PDT)
   DKIM-Signature: v=1; a=rsa-sha256; s=brisbane; d=example.com;
     c=simple/simple; q=dns/txt; i=joe@football.example.com;
     h=Received : From : To : Subject : Date : Message-ID;
   Received: from client1.football.example.com  []
     by submitserver.example.com with SUBMISSION;
     Fri, 11 Jul 2003 21:01:54 -0700 (PDT)
   From: Joe SixPack <joe@football.example.com>
   To: Suzie Q <suzie@shopping.example.net>
   Subject: Is dinner ready?
   Date: Fri, 11 Jul 2003 21:00:37 -0700 (PDT)
   Message-ID: <20030712040037.46341.5F8J@football.example.com>


   We lost the game.  Are you hungry yet?


                     Figure 3: Successful Verification

Crocker, et al.              Standards Track                   [Page 66]

RFC 6376                     DKIM Signatures              September 2011

Appendix B.  Usage Examples (INFORMATIVE)

   DKIM signing and validating can be used in different ways, for
   different operational scenarios.  This Appendix discusses some common

      NOTE: Descriptions in this Appendix are for informational purposes
      only.  They describe various ways that DKIM can be used, given
      particular constraints and needs.  In no case are these examples
      intended to be taken as providing explanation or guidance
      concerning DKIM specification details when creating an

B.1.  Alternate Submission Scenarios

   In the most simple scenario, a user's MUA, MSA, and Internet
   (boundary) MTA are all within the same administrative environment,
   using the same domain name.  Therefore, all of the components
   involved in submission and initial transfer are related.  However, it
   is common for two or more of the components to be under independent
   administrative control.  This creates challenges for choosing and
   administering the domain name to use for signing and for its
   relationship to common email identity header fields.

B.1.1.  Delegated Business Functions

   Some organizations assign specific business functions to discrete
   groups, inside or outside the organization.  The goal, then, is to
   authorize that group to sign some mail but to constrain what
   signatures they can generate.  DKIM selectors (the "s=" signature
   tag) facilitate this kind of restricted authorization.  Examples of
   these outsourced business functions are legitimate email marketing
   providers and corporate benefits providers.

   Here, the delegated group needs to be able to send messages that are
   signed, using the email domain of the client company.  At the same
   time, the client often is reluctant to register a key for the
   provider that grants the ability to send messages for arbitrary
   addresses in the domain.

   There are multiple ways to administer these usage scenarios.  In one
   case, the client organization provides all of the public query
   service (for example, DNS) administration, and in another, it uses
   DNS delegation to enable all ongoing administration of the DKIM key
   record by the delegated group.

Crocker, et al.              Standards Track                   [Page 67]

RFC 6376                     DKIM Signatures              September 2011

   If the client organization retains responsibility for all of the DNS
   administration, the outsourcing company can generate a key pair,
   supplying the public key to the client company, which then registers
   it in the query service using a unique selector.  The client company
   retains control over the use of the delegated key because it retains
   the ability to revoke the key at any time.

   If the client wants the delegated group to do the DNS administration,
   it can have the domain name that is specified with the selector point
   to the provider's DNS server.  The provider then creates and
   maintains all of the DKIM signature information for that selector.
   Hence, the client cannot provide constraints on the local-part of
   addresses that get signed, but it can revoke the provider's signing
   rights by removing the DNS delegation record.

B.1.2.  PDAs and Similar Devices

   PDAs demonstrate the need for using multiple keys per domain.
   Suppose that John Doe wants to be able to send messages using his
   corporate email address, jdoe@example.com, and his email device does
   not have the ability to make a Virtual Private Network (VPN)
   connection to the corporate network, either because the device is
   limited or because there are restrictions enforced by his Internet
   access provider.  If the device is equipped with a private key
   registered for jdoe@example.com by the administrator of the
   example.com domain and appropriate software to sign messages, John
   could sign the message on the device itself before transmission
   through the outgoing network of the access service provider.

B.1.3.  Roaming Users

   Roaming users often find themselves in circumstances where it is
   convenient or necessary to use an SMTP server other than their home
   server; examples are conferences and many hotels.  In such
   circumstances, a signature that is added by the submission service
   will use an identity that is different from the user's home system.

   Ideally, roaming users would connect back to their home server using
   either a VPN or a SUBMISSION server running with SMTP AUTHentication
   on port 587.  If the signing can be performed on the roaming user's
   laptop, then they can sign before submission, although the risk of
   further modification is high.  If neither of these are possible,
   these roaming users will not be able to send mail signed using their
   own domain key.

Crocker, et al.              Standards Track                   [Page 68]

RFC 6376                     DKIM Signatures              September 2011

B.1.4.  Independent (Kiosk) Message Submission

   Stand-alone services, such as walk-up kiosks and web-based
   information services, have no enduring email service relationship
   with the user, but users occasionally request that mail be sent on
   their behalf.  For example, a website providing news often allows the
   reader to forward a copy of the article to a friend.  This is
   typically done using the reader's own email address, to indicate who
   the author is.  This is sometimes referred to as the "Evite" problem,
   named after the website of the same name that allows a user to send
   invitations to friends.

   A common way this is handled is to continue to put the reader's email
   address in the From header field of the message but put an address
   owned by the email posting site into the Sender header field.  The
   posting site can then sign the message, using the domain that is in
   the Sender field.  This provides useful information to the receiving
   email site, which is able to correlate the signing domain with the
   initial submission email role.

   Receiving sites often wish to provide their end users with
   information about mail that is mediated in this fashion.  Although
   the real efficacy of different approaches is a subject for human
   factors usability research, one technique that is used is for the
   verifying system to rewrite the From header field to indicate the
   address that was verified, for example: From: John Doe via
   news@news-site.example <jdoe@example.com>.  (Note that such rewriting
   will break a signature, unless it is done after the verification pass
   is complete.)

B.2.  Alternate Delivery Scenarios

   Email is often received at a mailbox that has an address different
   from the one used during initial submission.  In these cases, an
   intermediary mechanism operates at the address originally used, and
   it then passes the message on to the final destination.  This
   mediation process presents some challenges for DKIM signatures.

B.2.1.  Affinity Addresses

   "Affinity addresses" allow a user to have an email address that
   remains stable, even as the user moves among different email
   providers.  They are typically associated with college alumni
   associations, professional organizations, and recreational
   organizations with which they expect to have a long-term
   relationship.  These domains usually provide forwarding of incoming
   email, and they often have an associated Web application that
   authenticates the user and allows the forwarding address to be

Crocker, et al.              Standards Track                   [Page 69]

RFC 6376                     DKIM Signatures              September 2011

   changed.  However, these services usually depend on users sending
   outgoing messages through their own service provider's MTAs.  Hence,
   mail that is signed with the domain of the affinity address is not
   signed by an entity that is administered by the organization owning
   that domain.

   With DKIM, affinity domains could use the Web application to allow
   users to register per-user keys to be used to sign messages on behalf
   of their affinity address.  The user would take away the secret half
   of the key pair for signing, and the affinity domain would publish
   the public half in DNS for access by Verifiers.

   This is another application that takes advantage of user-level
   keying, and domains used for affinity addresses would typically have
   a very large number of user-level keys.  Alternatively, the affinity
   domain could handle outgoing mail, operating a mail submission agent
   that authenticates users before accepting and signing messages for
   them.  This is, of course, dependent on the user's service provider
   not blocking the relevant TCP ports used for mail submission.

B.2.2.  Simple Address Aliasing (.forward)

   In some cases, a recipient is allowed to configure an email address
   to cause automatic redirection of email messages from the original
   address to another, such as through the use of a Unix .forward file.
   In this case, messages are typically redirected by the mail handling
   service of the recipient's domain, without modification, except for
   the addition of a Received header field to the message and a change
   in the envelope recipient address.  In this case, the recipient at
   the final address' mailbox is likely to be able to verify the
   original signature since the signed content has not changed, and DKIM
   is able to validate the message signature.

B.2.3.  Mailing Lists and Re-Posters

   There is a wide range of behaviors in services that take delivery of
   a message and then resubmit it.  A primary example is with mailing
   lists (collectively called "forwarders" below), ranging from those
   that make no modification to the message itself, other than to add a
   Received header field and change the envelope information, to those
   that add header fields, change the Subject header field, add content
   to the body (typically at the end), or reformat the body in some
   manner.  The simple ones produce messages that are quite similar to
   the automated alias services.  More elaborate systems essentially
   create a new message.

Crocker, et al.              Standards Track                   [Page 70]

RFC 6376                     DKIM Signatures              September 2011

   A Forwarder that does not modify the body or signed header fields of
   a message is likely to maintain the validity of the existing
   signature.  It also could choose to add its own signature to the

   Forwarders that modify a message in a way that could make an existing
   signature invalid are particularly good candidates for adding their
   own signatures (e.g., mailing-list-name@example.net).  Since
   (re-)signing is taking responsibility for the content of the message,
   these signing forwarders are likely to be selective and forward or
   re-sign a message only if it is received with a valid signature or if
   they have some other basis for knowing that the message is not

   A common practice among systems that are primarily redistributors of
   mail is to add a Sender header field to the message to identify the
   address being used to sign the message.  This practice will remove
   any preexisting Sender header field as required by [RFC5322].  The
   forwarder applies a new DKIM-Signature header field with the
   signature, public key, and related information of the forwarder.

   See [RFC6377] for additional related topics and discussion.

Appendix C.  Creating a Public Key (INFORMATIVE)

   The default signature is an RSA-signed SHA-256 digest of the complete
   email.  For ease of explanation, the openssl command is used to
   describe the mechanism by which keys and signatures are managed.  One
   way to generate a 1024-bit, unencrypted private key suitable for DKIM
   is to use openssl like this:

   $ openssl genrsa -out rsa.private 1024

   For increased security, the "-passin" parameter can also be added to
   encrypt the private key.  Use of this parameter will require entering
   a password for several of the following steps.  Servers may prefer to
   use hardware cryptographic support.

   The "genrsa" step results in the file rsa.private containing the key
   information similar to this:

Crocker, et al.              Standards Track                   [Page 71]

RFC 6376                     DKIM Signatures              September 2011

   -----END RSA PRIVATE KEY-----

   To extract the public-key component from the private key, use openssl
   like this:

   $ openssl rsa -in rsa.private -out rsa.public -pubout -outform PEM

   This results in the file rsa.public containing the key information
   similar to this:

   -----BEGIN PUBLIC KEY-----
   -----END PUBLIC KEY-----

   This public-key data (without the BEGIN and END tags) is placed in
   the DNS:

   $ORIGIN _domainkey.example.org.

C.1.  Compatibility with DomainKeys Key Records

   DKIM key records were designed to be backward compatible in many
   cases with key records used by DomainKeys [RFC4870] (sometimes
   referred to as "selector records" in the DomainKeys context).  One
   area of incompatibility warrants particular attention.  The "g=" tag
   value may be used in DomainKeys and [RFC4871] key records to provide

Crocker, et al.              Standards Track                   [Page 72]

RFC 6376                     DKIM Signatures              September 2011

   finer granularity of the validity of the key record to a specific
   local-part.  A null "g=" value in DomainKeys is valid for all
   addresses in the domain.  This differs from the usage in the original
   DKIM specification ([RFC4871]), where a null "g=" value is not valid
   for any address.  In particular, see the example public-key record in
   Section 3.2.3 of [RFC4870].

C.2.  RFC 4871 Compatibility

   Although the "g=" tag has been deprecated in this version of the DKIM
   specification (and thus MUST now be ignored), Signers are advised not
   to include the "g=" tag in key records because some [RFC4871]-
   compliant Verifiers will be in use for a considerable period to come.

Appendix D.  MUA Considerations (INFORMATIVE)

   When a DKIM signature is verified, the processing system sometimes
   makes the result available to the recipient user's MUA.  How to
   present this information to users in a way that helps them is a
   matter of continuing human factors usability research.  The tendency
   is to have the MUA highlight the SDID, in an attempt to show the user
   the identity that is claiming responsibility for the message.  An MUA
   might do this with visual cues such as graphics, might include the
   address in an alternate view, or might even rewrite the original From
   address using the verified information.  Some MUAs might indicate
   which header fields were protected by the validated DKIM signature.
   This could be done with a positive indication on the signed header
   fields, with a negative indication on the unsigned header fields, by
   visually hiding the unsigned header fields, or some combination of
   these.  If an MUA uses visual indications for signed header fields,
   the MUA probably needs to be careful not to display unsigned header
   fields in a way that might be construed by the end user as having
   been signed.  If the message has an "l=" tag whose value does not
   extend to the end of the message, the MUA might also hide or mark the
   portion of the message body that was not signed.

   The aforementioned information is not intended to be exhaustive.  The
   MUA can choose to highlight, accentuate, hide, or otherwise display
   any other information that may, in the opinion of the MUA author, be
   deemed important to the end user.

Appendix E.  Changes since RFC 4871

   o  Abstract and introduction refined based on accumulated experience.

   o  Various references updated.

Crocker, et al.              Standards Track                   [Page 73]

RFC 6376                     DKIM Signatures              September 2011

   o  Several errata resolved (see http://www.rfc-editor.org/):

      *  1376 applied

      *  1377 applied

      *  1378 applied

      *  1379 applied

      *  1380 applied

      *  1381 applied

      *  1382 applied

      *  1383 discarded (no longer applies)

      *  1384 applied

      *  1386 applied

      *  1461 applied

      *  1487 applied

      *  1532 applied

      *  1596 applied

   o  Introductory section enumerating relevant architectural documents

   o  Introductory section briefly discussing the matter of data
      integrity added.

   o  Allowed tolerance of some clock drift.

   o  Dropped "g=" tag from key records.  The implementation report
      indicates that it is not in use.

   o  Removed errant note about wildcards in the DNS.

   o  Removed SMTP-specific advice in most places.

   o  Reduced (non-normative) recommended signature content list, and
      reworked the text in that section.

Crocker, et al.              Standards Track                   [Page 74]

RFC 6376                     DKIM Signatures              September 2011

   o  Clarified signature generation algorithm by rewriting its pseudo-

   o  Numerous terminology subsections added, imported from [RFC5672].
      Also, began using these terms throughout the document (e.g., SDID,

   o  Sections added that specify input and output requirements.  Input
      requirements address a security concern raised by the working
      group (see also new sections in Security Considerations).  Output
      requirements are imported from [RFC5672].

   o  Appendix subsection added discussing compatibility with DomainKeys
      ([RFC4870]) records.

   o  Referred to [RFC5451] as an example method of communicating the
      results of DKIM verification.

   o  Removed advice about possible uses of the "l=" signature tag.

   o  IANA registry updated.

   o  Added two new Security Considerations sections talking about
      malformed message attacks.

   o  Various copy editing.

Appendix F.  Acknowledgments

   The previous IETF version of DKIM [RFC4871] was edited by Eric
   Allman, Jon Callas, Mark Delany, Miles Libbey, Jim Fenton, and
   Michael Thomas.

   That specification was the result of an extended collaborative
   effort, including participation by Russ Allbery, Edwin Aoki, Claus
   Assmann, Steve Atkins, Rob Austein, Fred Baker, Mark Baugher, Steve
   Bellovin, Nathaniel Borenstein, Dave Crocker, Michael Cudahy, Dennis
   Dayman, Jutta Degener, Frank Ellermann, Patrik Faeltstroem, Mark
   Fanto, Stephen Farrell, Duncan Findlay, Elliot Gillum, Olafur
   Gudmundsson, Phillip Hallam-Baker, Tony Hansen, Sam Hartman, Arvel
   Hathcock, Amir Herzberg, Paul Hoffman, Russ Housley, Craig Hughes,
   Cullen Jennings, Don Johnsen, Harry Katz, Murray S. Kucherawy, Barry
   Leiba, John Levine, Charles Lindsey, Simon Longsdale, David Margrave,
   Justin Mason, David Mayne, Thierry Moreau, Steve Murphy, Russell
   Nelson, Dave Oran, Doug Otis, Shamim Pirzada, Juan Altmayer Pizzorno,
   Sanjay Pol, Blake Ramsdell, Christian Renaud, Scott Renfro, Neil

Crocker, et al.              Standards Track                   [Page 75]

RFC 6376                     DKIM Signatures              September 2011

   Rerup, Eric Rescorla, Dave Rossetti, Hector Santos, Jim Schaad, the
   Spamhaus.org team, Malte S. Stretz, Robert Sanders, Rand Wacker, Sam
   Weiler, and Dan Wing.

   The earlier DomainKeys was a primary source from which DKIM was
   derived.  Further information about DomainKeys is at [RFC4870].

   This revision received contributions from Steve Atkins, Mark Delany,
   J.D. Falk, Jim Fenton, Michael Hammer, Barry Leiba, John Levine,
   Charles Lindsey, Jeff Macdonald, Franck Martin, Brett McDowell, Doug
   Otis, Bill Oxley, Hector Santos, Rolf Sonneveld, Michael Thomas, and
   Alessandro Vesely.

Authors' Addresses

   Dave Crocker (editor)
   Brandenburg InternetWorking
   675 Spruce Dr.
   Sunnyvale, CA  94086

   Phone: +1.408.246.8253
   EMail: dcrocker@bbiw.net
   URI:   http://bbiw.net

   Tony Hansen (editor)
   AT&T Laboratories
   200 Laurel Ave. South
   Middletown, NJ  07748

   EMail: tony+dkimsig@maillennium.att.com

   Murray S. Kucherawy (editor)
   128 King St., 2nd Floor
   San Francisco, CA  94107

   EMail: msk@cloudmark.com

Crocker, et al.              Standards Track                   [Page 76]