Sender ID: Authenticating E-Mail
draft-lyon-senderid-core-01
The information below is for an old version of the document that is already published as an RFC.
Document | Type |
This is an older version of an Internet-Draft that was ultimately published as RFC 4406.
|
|
---|---|---|---|
Authors | Meng Weng Wong , Jim Lyon | ||
Last updated | 2020-02-12 (Latest revision 2005-05-19) | ||
RFC stream | Internet Engineering Task Force (IETF) | ||
Intended RFC status | Experimental | ||
Formats | |||
Stream | WG state | (None) | |
Document shepherd | (None) | ||
IESG | IESG state | Became RFC 4406 (Historic) | |
Action Holders |
(None)
|
||
Consensus boilerplate | Unknown | ||
Telechat date | (None) | ||
Responsible AD | Ted Hardie | ||
Send notices to | (None) |
draft-lyon-senderid-core-01
Internet Draft J. Lyon Category: Experimental Microsoft Corp Document: draft-lyon-senderid-core-01.txt M. Wong pobox.com May 2005 Sender ID: Authenticating E-Mail Status of this Memo By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she becomes aware will be disclosed, in accordance with Section 6 of BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet- Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than a "work in progress." The list of current Internet-Drafts can be accessed at http://www.ietf.org/1id-abstracts.html The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html Abstract Internet mail suffers from the fact that much unwanted mail is sent using spoofed addresses -- "spoofed" in this case means the address is used without the permission of the domain owner. This document describes a family of tests by which SMTP servers can determine whether an e-mail address in a received message was used with the permission of the owner of the domain contained in that e-mail address. Lyon, Wong Experimental [Page 1] quot;MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119]. All the mobility-related terms used in this document are to be interpreted as defined in the Proxy Mobile IPv6 specifications [RFC5213] and [RFC5844]. Additionally, this document uses the following abbreviations: Network Access Server (NAS): A function that provides authorization services for a device/user access to the network as defined in [RFC2865]. This document makes an assumption that the NAS function is co-located with the Xia, et al. Standards Track [Page 4] RFC 6572 RADIUS PMIPv6 June 2012 MAG. In scenarios where the NAS function and MAG are decoupled, the messaging interface needed between them for the operation of PMIP6 is beyond the scope of this document. Home AAA (HAAA): An Authentication, Authorization, and Accounting (AAA) server located in the MN's home network. This sever has access to the mobile node's policy profiles. Visited AAA (VAAA): An Authentication, Authorization, and Accounting (AAA) server located in the MN's visited network. The VAAA server takes the role of a proxy-server, forwarding the received AAA service request to the HAAA server in the mobile node's home network and relaying the response to the requesting node, after applying any local access network policies. Local AAA (LAAA): An Authentication, Authorization, and Accounting proxy located in the local network. In a roaming case, the local AAAA has the visited AAA role. 3. Solution Overview This document defines the RADIUS-based AAA interactions with the two mobility management elements in the Proxy Mobile IPv6 domain. o Interactions between a MAG and a RADIUS-based AAA server o Interactions between a LMA and a RADIUS-based AAA server The mobile node's policy profile [RFC5213] is present in a policy store and is needed by the PMIPv6 mobility management elements for authorizing the mobile node for mobility management service and for obtaining various service-related parameters. This policy store could be locally co-located with the mobility management agents enabling direct local access or could be available from a AAA server through a RADIUS-based AAA interface. When a mobile node attaches to an access network, the NAS on that access network may activate the network access authentication procedure. The choice of the authentication mechanism is specific to the access network deployment; however, it is typically based on the Extensible Authentication Protocol (EAP) [RFC3748]. The NAS performs the network access authentication and queries the HAAA using AAA Xia, et al. Standards Track [Page 5] RFC 6572 RADIUS PMIPv6 June 2012 protocol, such as RADIUS. If the network access authentication succeeds, the MN's policy profile is obtained as part of the RADIUS message exchange with the AAA server. The mobile node may be an IPv4-only node, IPv6-only node, or a dual- stack (IPv4/v6) node. Based on the policy specified in the policy profile, the network access authentication procedure SHOULD provide the unambiguous indication of the type of address(es) to be assigned for the MN in the network and with all other service-related and policy parameters relevant to the mobility service. After the successful network access authentication and obtaining the mobile node's policy profile, the MAG sends a Proxy Binding Update (PBU) to the LMA. Upon receiving the PBU, the LMA interacts with the HAAA to obtain the mobile node's policy profile, which is required for authorizing and activating mobility service. This document adds support for three distinct PMIPv6 mobility use cases, taking into account the administrative domains to which the MAG and the LMA belong. The following are the three relevant deployment models. 1. the MAG and LMA are both in the home network, 2. the MAG and LMA are both in the visited network, 3. the MAG is in the visited network while the LMA is in the home network. Figure 1 shows participating network entities for the PMIPv6 mobility session, which is located in the home network. The MAG and LMA interact only with the HAAA. Xia, et al. Standards Track [Page 6] RFC 6572 RADIUS PMIPv6 June 2012 +--------+ | HAAA & | RADIUS +-----+ | Policy |<-------->| LMA | | Profile| +-----+ +--------+ | <--- LMA-Address ^ | | // \\ +---|------------- //---\\----------------+ ( | IPv4/IPv6 // \\ ) ( | Network // \\ ) +---|-----------//---------\\-------------+ | // \\ RADIUS // <- Tunnel1 \\ <- Tunnel2 | // \\ | |- MAG1-Address |- MAG2-Address | +----+ +----+ +---->|MAG1| |MAG2| +----+ +----+ | | | | MN1 MN2 Figure 1: The MAG and LMA Are Both in the Home Network Figure 2 shows both the LMA and MAG are in the visited network. The MAG and LMA exchange signaling with the HAAA through the VAAA, which acts as a Proxy. The visited network may append additional information to the HAAA replies in order to reflect the local policy. Xia, et al. Standards Track [Page 7] RFC 6572 RADIUS PMIPv6 June 2012 +---------------+ | HAAA &Sender ID: Authenticating E-Mail May 2005 Table of Contents 1. Introduction...................................................3 2. Problem Statement..............................................3 3. SPF Records....................................................4 3.1 Version and Scope..........................................4 3.2 Multiple Records...........................................5 3.3 Positional Modifiers.......................................4 3.4 Compatibility..............................................7 4. Decision Model.................................................7 4.1 Arguments..................................................8 4.2 Results....................................................8 4.3 Record Lookup..............................................8 4.4 Record Selection...........................................8 5. Actions Based on the Decision..................................9 5.1 Neutral, None, SoftFail or PermError.......................9 5.2 Pass.......................................................9 5.3 Fail......................................................10 5.4 TempError.................................................10 6. Security Considerations.......................................11 6.1 DNS Attacks...............................................11 6.2 TCP Attacks...............................................11 6.3 Forged Sender Attacks.....................................11 6.4 Address Space Hijacking...................................12 6.5 Malicious DNS attacks on third-parties....................12 7. Implementation Guidance.......................................13 7.1 Simple E-mailers..........................................13 7.2 E-Mail Forwarders.........................................13 7.3 Mailing List Servers......................................14 7.4 Third-Party Mailers.......................................14 7.5 MUA Implementers..........................................15 8. IANA Considerations...........................................15 9. Acknowledgements..............................................16 10. References...................................................16 10.1 Normative References.....................................16 10.2 Informative References...................................16 11. Authors' Addresses...........................................17 Conventions used in this document The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119]. Lyon, Wong Experimental [Page 2] Sender ID: Authenticating E-Mail May 2005 1. Introduction Today, a huge majority of unwanted e-mail contains headers that lie about the origin of the mail. This is true of most spam and substantially all of the virus e-mail that is sent. This document describes a mechanism such that receiving MTAs, MDAs and/or MUAs can recognize mail in the above category and take appropriate action. For example, an MTA might refuse to accept a message, an MDA might discard a message rather than placing it into a mailbox, and an MUA might render that message in some distinctive fashion. In order to avoid further fragmentation of the Internet e-mail system, it is desirable that the Internet community as a whole come to a consensus as to what mail senders should do to make their mail appear non-spoofed, and how mail receivers should determine whether mail is spoofed. On the other hand, it is not necessary to reach a consensus regarding the actions that various parties take once a message has been determined to be spoofed. This can be done one unilaterally -- agent might decide to discard a spoofed message while another decides to add a disclaimer. This document defines a pair of closely-related tests. One validates a message's Purported Responsible Address (PRA) as defined in [PRA]. The other validates a message's Reverse-Path (also known as MAIL-FROM address) as defined in [SPF]. An e-mail sender complying with this specification SHOULD publish information for both tests, and SHOULD arrange that any mail that is sent will pass both tests. An e-mail receiver complying with this specification SHOULD perform at least one of these tests. 2. Problem Statement Briefly stated, the mechanisms of this document allow one to answer the following question: When a message is transferred via SMTP between two unrelated parties, does the SMTP client host have permission to send mail on behalf of a mailbox referenced by the message? As seen from the question, this mechanism applies to unrelated parties: it is useful at the point where a message passes across the Internet from one organization to another. It is beyond the scope of this document to describe authentication mechanisms that can be deployed within an organization. Lyon, Wong Experimental [Page 3] Sender ID: Authenticating E-Mail May 2005 The PRA version of the test seeks to authenticate the mailbox associated with the most recent introduction of a message into the mail delivery system. In simple cases, this is who the mail is from. However, in the case of a third-party mailer, a forwarder or a mailing list server, the address being authenticated is that of the third party, the forwarder or the mailing list. On the other hand, the MAIL-FROM version of the test seeks to authenticate the mailbox that would receive Delivery Status Notifications (DSNs, or bounces) for the message. In simple cases, this too is who the mail is from. However, third-party mailers, forwarders and mailing list servers MUST specify an address under their control, and SHOULD arrange that DSNs received at this address are forwarded to the original bounce address. In both cases, the domain associated with an e-mail address is what is authenticated; no attempt is made to authenticate the local-part. A domain owner gets to determine which SMTP clients speak on behalf of addresses within the domain; a responsible domain owner should not authorize SMTP clients that will lie about local parts. In the long run, once the domain of the sender is authenticated, it will be possible to use that domain as part of a mechanism to determine the likelihood that a given message is spam, using, for example, reputation and accreditation services. (These services are not the subject of the present mechanism, but it should enable them.) 3. SPF 2.0 Records Domains declare which hosts are and are not authorized to transmit e-mail messages on their behalf by publishing Sender Policy Framework records in the Domain Name System. [SPF] defines a format for these records identified by the version prefix "v=spf1". This section defines an amended format, identified by the version prefix "spf2.0", that allows sending domains to explicitly specify how their records should be interpreted, and provides for additional extensibility. Sending domains MAY publish either or both formats. Since the two formats are identical in most respects, the following sub-sections define the "spf2.0& | +----------| Policy Profile| | +---------------+ | +---------+ |[VL]AAA &| RADIUS +-----+ | Policy |<------->| LMA | | Profile | +-----+ +---------+ | <--- LMA-Address ^ // \\ +---|------------- //---\\----------------+ ( | IPv4/IPv6 // \\ ) ( | Network // \\ ) +---|-----------//---------\\-------------+ | // \\ RADIUS // <- Tunnel1 \\ <- Tunnel2 | // \\ | |- MAG1-Address |- MAG2-Address | +----+ +----+ +---->|MAG1| |MAG2| +----+ +----+ | | MN1 MN2 Figure 2: The MAG and LMA Are Both in the Visited/Local Network Figure 3 illustrates a topology where the MAG resides in the visited network while the associated LMA is in MN's home network. Any message between the MAG and the HAAA passes through the VAAA, which acts as a Proxy. During the network authentication, the visited network's specific policy may also be propagated from the VAAA to the MAG. The LMA has a direct access to the HAAA. Xia, et al. Standards Track [Page 8] RFC 6572 RADIUS PMIPv6 June 2012 +---------------+ | HAAA & | +----------| Policy Profile| | +---------------+ | | | RADIUS +---------+ | |[VL]AAA &| +-----+ | Policy | | LMA | | Profile | +-----+ +---------+ | <--- LMA-Address ^ // \\ +---|------------- //---\\----------------+ ( | IPv4/IPv6 // \\ ) ( | Network // \\ ) +---|-----------//---------\\-------------+ | // \\ RADIUS // <- Tunnel1 \\ <- Tunnel2 | // \\ | |- MAG1-Address |- MAG2-Address | +----+ +----+ +---->|MAG1| |MAG2| +----+ +----+ | | MN1 MN2 Figure 3: Visited MAG and Home LMA Topology 4. Attribute Definitions 4.1. MIP6-Feature-Vector Diameter [RFC3588] reserves AVP Code space 1-255 as RADIUS attribute compatibility space. The MIP6-Feature-Vector attribute (Type value 124) defined in [RFC5447] is of type OctetString and contains a 64-bit flags field of supported mobility capabilities. This document reserves two new capability bits according to the rules in [RFC5447], and reuses the PMIPv6 capability bits defined by [RFC5779]. The following capability flag bits are used or defined in this document: PMIP6_SUPPORTED (0x0000010000000000) This capability bit is used as defined in [RFC5779]. IP4_HOA_SUPPORTED (0x0000020000000000) This capability bit is used as defined in [RFC5779]. Assignment of the IPv4-HoA (Home Address) is defined by [RFC5844]. Xia, et al. Standards Track [Page 9] RFC 6572 RADIUS PMIPv6 June 2012 LOCAL_MAG_ROUTING_SUPPORTED (0x0000040000000000) This capability bit is used as defined in [RFC5779]. IP4_TRANSPORT_SUPPORTED (0x0000800000000000) This capability bit is used for negotiation of the IPv4 transport support between the MAG and AAA. When the MAG sets this flag bit in the MIP6-Feature-Vector, it indicates the ability of the MAG to provide IPv4 transport (i.e., IPv4-based encapsulation) for carrying IP traffic between the MAG and the LMA. If this flag bit is unset in the returned MIP6-Feature-Vector attribute, the AAA does not authorize the use of IPv4 transport on the MAG-to-LMA tunnel. IP4_HOA_ONLY_SUPPORTED (0x0001000000000000) This capability bit is used for determination of the authorized PMIPv6 mobility mode. When this bit is set by the AAA, it indicates PMIPv6 mobility with IPv4 support has only been authorized for the MN. As a result, the RADIUS Access-Accept SHOULD NOT carry the IPv6 Home Network Prefix (IPv6 HNP). When this bit is set, the PMIP6_SUPPORTED flag MUST also be set and the IP4_HOA_SUPPORTED flag MUST NOT be set. To summarize the use of the MIP6-Feature-Vector the following capability bit combination settings mean: PMIP6-SUPPORTED bit set - only IPv6 mobility is supported and authorized. PMIP6-SUPPORTED and IP4-ONLY-HOA-SUPPORTED bits set - only IPv4 mobility is supported and authorized. PMIP6-SUPPORTED and IP4-HOA-SUPPORTED bits set - both IPv6 and IPv4 mobility are supported and authorized. The MIP6-Feature-Vector attribute is also used on the LMA to the RADIUS AAA interface. This capability announcement attribute enables direct capability negotiation between the LMA and the AAA. The capabilities that are announced by both parties in the MIP6-Feature- Vector are known to be mutually supported. The LMA may use this mechanism during authorization of the received PBU against the AAA to check individual PMIPv6 feature permissions for a particular MN. If the RADIUS Access-Accept contains a contradicting combination of the capability flag bits such as both the IP4_HOA_ONLY_SUPPORTED and the IP4_HOA_SUPPORTED flags being set, then the RADIUS client MUST Xia, et al. Standards Track [Page 10] RFC 6572 RADIUS PMIPv6 June 2012 treat the Access-Accept as an Access-Reject and SHOULD log the event. Similarly, if the RADIUS Access-Request contains a contradicting combination of the capability flag bits, then the RADIUS server MUST reply with an Access-Reject message and SHOULD log the event. 4.2. Mobile-Node-Identifier The Mobile-Node-Identifier attribute (Type value 145) is of type String and contains the mobile node identifier (MN-Identifier), see [RFC5213], in a form of a Network Access Identifier (NAI) [RFC4282]. This identifier and the identifier used for access authentication may be different; however, there needs to be a mapping between the two identities as specified in Section 6.6 of [RFC5213]. This attribute is used on the interface between the MAG and the AAA server. The Mobile-Node-Identifier attribute is designed for deployments where the identity used during network access authentication and the identity used for mobility management is decoupled. It may also be the case where the MAG does not have means to find out the MN identity that could be used in subsequent PBU and Proxy Binding Acknowledgement (PBA) exchanges (e.g., due to identity hiding during the network access authentication) or when the HAAA wants to assign periodically changing identities to the MN. The Mobile-Node-Identifier attribute MAY be returned by the HAAA in the RADIUS Access-Accept message that completes a successful authentication and authorization exchange between the MAG and the HAAA. The MAG MUST use the received MN-Identifier. 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type | Length | Mobile Node Identifier... ~ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Type: Mobile-Node-Identifier 145. Length: In octets, including Type and Length fields (>= 3). Mobile Node Identifier: This field is of type String and contains the MN-Identifier of the MN to be used in the PBU/PBA exchange. Xia, et al. Standards Track [Page 11] RFC 6572 RADIUS PMIPv6 June 2012 4.3. Service-Selection The Service-Selection attribute (Type value 146) is of type UTF-8 text and contains the name of the service or the external network with which the mobility service for the particular MN SHOULD be associated [RFC5149]. The identifier MUST be unique within the PMIPv6 Domain when normalized using the selected normalization form [UNF] for the particular PMIPv6 Domain deployment. For instance, [RFC5149] uses the Normalization Form KC (NFKC). The MAG MUST include the Service-Selection attribute in the Access- Request sent to the AAA if the information was acquired, e.g., by operator-specific configuration. The AAA MAY include the Service- Selection attribute in the Access-Accept response message to the MAG even if it was not included in the Access-Request as a means of indicating the MN's default service. The Service Selection mobility option defined in [RFC5149] can be used in PBU/PBA messages between the MAG and LMA. On the LMA-to-AAA interface, the LMA MAY populate the Service-Selection attribute in the Access-Request message using the service information found in the received PBU, if such a mobility option were included. The Service- Selection identifier should be used to assist the PBU authorization, the assignment of the MN-HNP, and the IPv4-MN-HoA as described in [RFC5149] and [RFC5779]. 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type | Length | Service Identifier... ~ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Type: Service-Selection 146. Length: In octets, including Type and Length fields (>= 3). Text: This field is of type UTF-8 text and contains the Service Identifier with which the MN is associated. 4.4. PMIP6-Home-LMA-IPv6-Address The PMIP6-Home-LMA-IPv6-Address attribute (Type value 147) is of type IPv6 address and is used to deliver the IPv6 address of the LMA located in the home network. Xia, et al. Standards Track [Page 12] quot; format relative to [SPF]. 3.1 Version and Scope Under Sender ID, receiving domains may perform a check of either the PRA identity or the MAIL-FROM identity. Sending domains therefore require a method for declaring whether their published list of authorized outbound e-mail servers can be used for the PRA check, the MAIL-FROM check or both. Lyon, Wong Experimental [Page 4] Sender ID: Authenticating E-Mail May 2005 This section replaces section 4.5 of [SPF] and adds the concept of SPF record scopes. SPF records begin with a version identifier and may also include a scope: record = version terms *SP version = "v=spf1" | ( "spf2." ver-minor scope) ver-minor = 1*DIGIT scope = "/" scope-id *( "," scope-id ) scope-id = "mfrom" / "pra" / name For example, the SPF record: spf2.0/mfrom,pra +mx +ip4:192.168.0.100 -all defines an SPF record that can be used for either MAIL FROM or PRA checks. This document only defines the existence of two scopes: "mfrom" and "pra". The details of these two scopes are defined in other documents: "mfrom" is defined in [SPF], "pra" is defined in [PRA]. Other scopes may be defined by future documents only. There is no registry for scopes. A scope definition must define what it identifies as the sending mailbox for a message, how to extract that information from a message, how to determine the initial arguments for the check_host() function, and what the compliant responses to the result are. This ensures that domains with published records and mail receiver agree on the semantics of the scope. A compliant domain SHOULD publish authorizations for every defined scope. 3.1.1 Minor Version All published records that use the "spf2" version identifier MUST start with "spf2.0". This document only specifies records with a minor version of "0". Future versions of this document may define other minor versions to be used. 3.2 Multiple Records A domain MAY publish multiple SPF 2.0 records, provided that each scope appears in at most one SPF 2.0 record. In addition, a domain MAY also publish an SPF record that uses the "v=spf1" version identifier defined in [SPF]. The selection rules in Section 4.4 define the precedence of these records. Lyon, Wong Experimental [Page 5] Sender ID: Authenticating E-Mail May 2005 3.3 Positional Modifiers This section replaces section 4.6.3 of [SPF] and adds the concept of positional modifiers. Modifiers are key/value pairs that affect the evaluation of the check_host() function. Modifiers are either global or positional: Global modifiers MAY appear anywhere in the record, but SHOULD appear at the end, after all mechanisms and positional modifiers. Positional modifiers apply only to the mechanism they follow. It is a syntax error for a positional modifier to appear before the first mechanism. Modifiers of either type are also either singular or multiple: Singular modifiers may appear only once in the record if they are global, or once after each mechanism if they are positional. Multiple modifiers may appear multiple times in the record if they are global or multiple times after each mechanism if they are positional. A modifier is not allowed to be defined as both global and positional. The modifiers "redirect" and "exp" described in section 5 of [SPF] are global and singular. Ordering of modifiers does not matter, except: 1) positional modifiers must appear after the mechanism they affect and before any subsequent mechanisms. and 2) when a multiple modifier appears more than one time, the ordering of the appearances may be significant to the modifier. Other than these constraints, implementations MUST treat different orders of modifiers the same. An intended side effect of these rules is modifiers cannot be defined that modify other modifiers. These rules allow an implementation to correctly pre-parse a record. Furthermore, they are crafted to allow the parsing algorithm to be stable, even when new modifiers are introduced. Modifiers which are unrecognized MUST be ignored. This allows older implementations to handle records with modifiers that were defined after they were written. Lyon, Wong Experimental [Page 6] Sender ID: Authenticating E-Mail May 2005 3.4 Compatibility Domain administrators complying with this specification are required to publish information in DNS regarding their authorized outbound e-mail servers. [SPF] describes a format for this information identified by the version prefix "v=spf1". Many domains have published information in DNS using this format. In order to provide compatibility for these domains, Sender ID implementations SHOULD interpret the version prefix "v=spf1" as equivalent to "spf2.0/mfrom,pra", provided no record starting with "spf2.0" exists. Administrators who have already published "v=spf1" records SHOULD review these records to determine whether they are also valid for use with PRA checks. If the information in a "v=spf1" record is not correct for a PRA check, administrators SHOULD publish either an "spf2.0/pra" record with correct information, or an "spf2.0/pra ?all" record indicating that the result of a PRA check is explicitly inconclusive. 4. Decision Model Sender ID enables receiving e-mail systems to answer the question: Given an e-mail message, and given an IP address from which it has been (or will be) received, is the SMTP client at that IP address authorized to send that e-mail message? This question will usually be asked by an SMTP server as part of deciding whether to accept an incoming mail message. However, this question could also be asked later by a different party. An MUA, for example, could use the result of this question to determine how to file or present a message. There are three steps to answering this question: (1) From an e-mail message, extract the address to verify. The PRA variant of this test does so as specified in [PRA], or alternatively, using the submitter address as specified in [Submitter]. The MAIL FROM variant of this test does so as specified in [SPF]. (2) Extract the domain part of the address determined in step (1). (3) Call the check_host() function defined in [SPF] and modified by the following sub-sections. Lyon, Wong Experimental [Page 7] Sender ID: Authenticating E-Mail May 2005 If the Sender ID check is being performed by an MTA as part of receiving an e-mail message, and it cannot determine an address in step (1) above (because the message or address is malformed), then the message SHOULD be rejected with error "550 5.7.1 Missing Purported Responsible Address" or error "550 5.7.1 Missing Reverse- Path address". 4.1 Arguments Sender ID modifies the check_host() function by the addition of a scope parameter. Thus, for Sender ID the check_host() function is called passing the following parameters: a. A scope of "pra" (for the PRA variant of the test), or "mfrom" (for the MAIL FROM variant of the test). b. The IP address (either IPv4 or IPv6) from which the message is being or has been received. c. The domain from step (2) above. d. The address from step (1) above. 4.2 Results The result of the check_host() function is one of the values "Neutral", "Pass", "Fail", "SoftFail", "None", "TempError" or "PermError". Section 5 describes how these results are used by MTAs receiving messages. This specification imposes no requirements on parties performing this test in other environments. 4.3 Record Lookup SPF records are looked up in DNS in accordance with section 4.4 of [SPF}. When performing the PRA version of the test, if the DNS query returns "non-existent domain" (RCODE 3), then check_host() exits immediately with the result "Fail". 4.4 Record Selection This section replaces the record selection steps described in section 4.5 of [SPF]. Starting with the set of records that were returned by the lookup, record selection proceeds in these steps: 1. If any records of type SPF are in the set, then all records of type TXT are discarded. Lyon, Wong Experimental [Page 8] Sender ID: Authenticating E-Mail May 2005 2. Records that do not begin with proper version and scope sections are discarded. The version section for "spf2" records contains a ver-minor field that is for backward compatible future extensions. This field must be well-formed for a record to be retained, but is otherwise ignored. 3. Records that use the "spf2" version identifier and do not have a scope-id that matches <scope> are discarded. Note that this is a complete string match on the scope-id tokens: If <scope> is "pra", then the record starting "spf2.0/mfrom,prattle,fubar" would be discarded, but a record starting "spf2.0/mfrom,pra,fubar" would be retained. 4. If the lookup returned two records, one containing the "v=spf1" version identifier and the other containing the "spf2" version identifier, the "spf2" version takes precedence for the desired scope-id. If the "spf2" record does not contain the desired scope-id, then the "v=spf1" record is selected. 5. If an "spf2" record does not contain the desired scope-id and there is no "v=spf1" record for the domain, then no record is selected. After the above steps, there should be one record remaining and evaluation can proceed. If there are two or more records remaining, then check_host() exits immediately with the error "PermError". If the PRA version of the test is being performed and no records remain, the requirement in [SPF] to find the Zone Cut and repeat the above steps is OPTIONAL. If there are no matching records remaining after the initial DNS query or any subsequent optional DNS queries, then check_host() exits immediately with the result "None". 5. Actions Based on the Decision When the Sender ID test is used by an SMTP server as part of receiving a message, the server should take the actions described by this section. The check_host() function returns one of the following results. See [SPF] for the meaning of these results. 5.1 Neutral, None, SoftFail or PermError An SMTP server receiving one of these results SHOULD NOT reject the message for this reason alone, but MAY subject the message to Lyon, Wong Experimental [Page 9] Sender ID: Authenticating E-Mail May 2005 heightened scrutiny by other measures, and MAY reject the message as a result of this heightened scrutiny. Such additional security measures MAY take into account that a message for which the result is "SoftFail" is less likely to be authentic than a message for which the result is "Neutral". 5.2 Pass An SMTP server receiving this result SHOULD treat the message as authentic. It may accept or reject the message depending on other policies. 5.3 Fail When performing the Sender-ID test during an SMTP transaction, an MTA receiving this result SHOULD reject the message with a "550 5.7.1 Sender ID (xxx) yyy - zzz" SMTP error, where "xxx" is replaced with "PRA" or "MAIL FROM", "yyy" is replaced with the additional reason returned by the check_host() function and "zzz" is replaced with the explanation string returned by the check_host() function. When performing the Sender-ID test after accepting an e-mail message for delivery, an MTA receiving this result SHOULD not deliver the message. Instead, it should create a DSN message, consistent with the usual rules for DSN messages. 5.4 TempError An SMTP server receiving this result MAY reject the message with a "450 4.4.3 Sender ID check is temporarily unavailable" error code. Alternatively, an SMTP server receiving this result MAY accept a message and optionally subject it to heightened scrutiny by other anti-spam measures. Lyon, Wong Experimental [Page 10] Sender ID: Authenticating E-Mail May 2005 6. Security Considerations This entire document describes a new mechanism for mitigating spoofed e-mail, which is today a pervasive security problem in the Internet. Assuming that this mechanism is widely deployed, the following sections describe counter-attacks that could be used to defeat this mechanism. 6.1 DNS Attacks The new mechanism is entirely dependent on DNS lookups, and is therefore only as secure as DNS. An attacker bent on spoofing messages could attempt to get his messages accepted by sending forged answers to DNS queries. An MTA could largely defeat such an attack by using a properly paranoid DNS resolver. DNSSEC may ultimately provide a way to completely neutralize this class of attacks. 6.2 TCP Attacks This mechanism is designed to be used in conjunction with SMTP over TCP. A sufficiently resourceful attacker might be able to send TCP packets with forged from-addresses, and thus execute an entire SMTP session that appears to come from somewhere other than its true origin. Such an attack requires guessing what TCP sequence numbers an SMTP server will use. It also requires transmitting completely in the blind - the attack will be unable hear any of the server's side of the conversation. Attacks of this sort can be ameliorated if IP gateways refuse to forward packets when the source address is clearly bogus. 6.3 Forged Sender Attacks This mechanism chooses an address to validate either from one of a number of message headers or from the RFC2821 MAIL command, and then uses that address for validation. A message with a true Resent-From header or Return-Path, but a forged From header will be accepted. Since many MUAs do not display all of the headers of received messages, the message will appear to be forged when displayed. Lyon, Wong Experimental [Page 11] Sender ID: Authenticating E-Mail May 2005 In order to neutralize this attack, MUAs will need to start displaying at least the address that was verified. In addition MTAs could subject messages to heightened scrutiny when the validated address differs from the From header. 6.4 Address Space Hijacking This mechanism assumes the integrity of IP address space for determining whether a given client is authorized to send messages from a given PRA. In addition to the TCP attack given in section 6.2, a sufficiently resourceful attacker might be able to alter the IP routing structure to permit two-way communication using a specified IP address. It would then be possible to execute an SMTP session that appears to come from an authorized address, without the need to guess TCP sequence numbers or transmit in the blind. Such an attack might occur if the attacker obtained access to a router which participates in external BGP routing. Such a router could advertise a more specific route to a rogue SMTP client, temporarily overriding the legitimate owner of the address. 6.5 Malicious DNS attacks on third-parties There is class of attacks in which an attacker A can entice a participant P to send a malicious message to a victim V. These attacks are undertaken by A citing the address of V in the SMTP MAIL FROM request and then by causing P to generate (or invoke the generation of) a Delivery Status Notification 'bounce' message (RFC3464), which is sent to the victim V. The attacker relies upon it being common practice to copy the original message into the 'bounce' report, thereby causing the malice to be sent onwards to V. This mode of attack has the advantages (to the attacker) of obfuscating the location of the host from which the attack was mounted, and of possibly damaging the reputation of P by making it appear that P originated or was an active participant in the sending of the malicious message. In current practice, A causes P to cause the 'bounce' by addressing the original message to a non-existent recipient. Sender-ID enables a new variant of this attack. Lyon, Wong Experimental [Page 12] Sender ID: Authenticating E-Mail May 2005 In this variant the attacker A sends a message whose PRA (section 4) is selected by the attacker to be such that, when P undertakes the Sender-ID test, a 'Fail' will result (section 5.3). The message will be rejected (as the attacker intended) and a malicious 'bounce' message may be generated and sent to the victim V. 7. Implementation Guidance This section describes the actions that certain members of the Internet e-mail ecosystem must take to be compliant with this specification. 7.1 Simple E-mailers A domain that injects original e-mail into the Internet, using its own name in From headers, need do nothing to be compliant. However, such domains SHOULD publish records in DNS as defined by [SPF] and this specification. In the majority of cases, the domain's published information will be the same for both the PRA and MAIL FROM variants of this test. In this case, domains SHOULD publish their information using an SPF record with the prefix "v=spf1". Doing so will render their published information usable by the older SPF protocol, too. (See [SPF] for information on the SPF protocol.) 7.2 E-Mail Forwarders In order to pass the PRA variant of the test, a program that forwards received mail to other addresses MUST add an appropriate header that contains an e-mail address that it is authorized to use. Such programs SHOULD use the Resent-From header for this purpose. In order to pass the MAIL FROM variant of the test, a program that forwards received mail to other addresses MUST alter the MAIL FROM address to an address under its control. Should that address eventually receive a DSN relating to the original message, that DSN SHOULD be forwarded to the original MAIL FROM address. However, if this altered address receives any messages other than DSNs related to the original message, these messages MUST NOT be forwarded to the original MAIL FROM address; they SHOULD be refused during an SMTP transaction. Lyon, Wong Experimental [Page 13] Sender ID: Authenticating E-Mail May 2005 Additionally, e-mail forwarders SHOULD publish Sender ID records for their domains, and SHOULD use MTAs for which the Sender ID check yields a "pass" result. Some of today's forwarders already add an appropriate header (although many of them use Sender rather than Resent-From.) Most of them do not perform the address-rewriting specified above. Note that an e-mail forwarder might receive a single message for two or more recipients, each of whom requests forwarding to a new address. In this case, the forwarder's MTA SHOULD transmit the message to each new recipient individually, with each copy of the message containing a different newly inserted Resent-From header field. 7.3 Mailing List Servers In order to pass the PRA variant of the test, a mailing list server MUST add an appropriate header that contains an e-mail address that it is authorized to use. Such programs SHOULD use the Resent-From header for this purpose. In order to pass the MAIL FROM variant of the test, a mailing list server MUST alter the MAIL FROM address to an address under its control. Additionally, mailing list servers SHOULD publish Sender ID records for their domains, and SHOULD use MTAs for which the Sender ID check yields a "pass" result. Most of today's mailing list software already adds an appropriate header (although most of them use Sender rather than Resent-From), and most of them already alter the MAIL FROM address. 7.4 Third-Party Mailers In order to pass the PRA variant of this test, a program that sends mail on behalf of another user MUST add an appropriate header that contains an e-mail address that it is authorized to use. Such programs SHOULD use the Sender header for this purpose. In order to pass the MAIL FROM variant of this test, a program that sends mail on behalf of another user MUST use a MAIL FROM address that is under its control. Defining what the program does with any mail received at that address is beyond the scope of this document. Lyon, Wong Experimental [Page 14] Sender ID: Authenticating E-Mail May 2005 Additionally, third-party mailers servers SHOULD publish Sender ID records for their domains, and SHOULD use MTAs for which the Sender ID check yields a "pass" result. Many, but not all, of today's third-party mailers are already compliant with the PRA variant of the test. The extent to which mailers are already compliant with the MAIL FROM variant of this test is unknown. 7.5 MUA Implementers When displaying a received message, an MUA SHOULD display the purported responsible address as defined by this document whenever that address differs from the RFC 2822 From address. This display SHOULD be in addition to the RFC 2822 From address. When a received message contains multiple headers that might be used for the purported responsible address determination, an MUA should consider displaying all of them. That is, if a message contains several Resent-From's, a Sender and a From, an MUA should consider displaying all of them. Sender ID also does not validate the display name that may be transmitted along with an e-mail address. The display name is also vulnerable to spoofing and other forms of attacks. In order to reduce the occurrence and effectiveness of such attacks, MUA implementers should consider methods to safeguard the display name. This could include: * Not presenting the display name to the user at all, or not presenting the display name unless the corresponding e-mail address is listed in the user's address book. * Treating as suspicious any e-mail where the display name is itself in the form of an e-mail address, especially when it differs from the actual e-mail address in the header. * Making it clear to users that the e-mail address has been checked rather than the display name. 8. IANA Considerations This document contains no actions for IANA. Lyon, Wong Experimental [Page 15] Sender ID: Authenticating E-Mail May 2005 9. Acknowledgements This design is based on earlier work published in 2003 in [RMX] and [DMP] drafts (by Hadmut Danisch and Gordon Fecyk respectively). The idea of using a DNS record to check the legitimacy of an e-mail address traces its ancestry to "Repudiating Mail From" draft by Paul Vixie [Vixie] (based on suggestion by Jim Miller) and to "Domain- Authorized SMTP Mail" draft by David Green [Green] who first introduced this idea on namedroppers mailing list in 2002. The current document borrows heavily from each of the above, and incorporates ideas proposed by many members of the MARID working group. The contributions of each of the above are gratefully acknowledged. 10. References 10.1 Normative References [PRA] J. Lyon, "Purported Responsible Address in E-Mail Messages", draft-lyon-senderid-pra-01. Work in progress. [RFC2119] S. Bradner, "Key words for use in RFCs to Indicate Requirement Levels", RFC 2119. [SPF] M. Wong and W. Schlitt, "Sender Policy Framework: Authorizing Use of Domains in E-Mail", draft- schlitt-spf-classic-00. Work in progress. [Submitter] E. Allman and H. Katz, "SMTP Service Extension for Indicating the Responsible Submitter of an E-mail Message", draft-katz-submitter-00. Work in progress. 10.2 Informative References [CallerID] Microsoft Corporation, Caller ID for E-Mail Technical Specification, http://www.microsoft.com/mscorp/twc/privacy/spam_callerid .mspx. [Green} David Green, "Mail-Transmitter RR", http://ops.ietf.org/lists/namedroppers/namedroppers.2002/ msg00656.html, June 2002. [RMX] H. Danisch, "The RMX DNS RR and method for lightweight SMTP sender authorization", draft-danisch-dns-rr-smtp-04. Work in progress. Lyon, Wong Experimental [Page 16] Sender ID: Authenticating E-Mail May 2005 [Vixie] Paul Vixie, "Repudiating Mail From", http://ops.ietf.org/lists/namedroppers/namedroppers.2002/ msg00658.html, June 2002. 11. Authors' Addresses Jim Lyon Microsoft Corporation One Microsoft Way Redmond, WA 98052 USA jimlyon@microsoft.com Meng Weng Wong Singapore mengwong@dumbo.pobox.com Lyon, Wong Experimental [Page 17] Sender ID: Authenticating E-Mail May 2005 Intellectual Property Statement The IETF takes no position regarding the validity or scope of any Intellectual Property Rights or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; nor does it represent that it has made any independent effort to identify any such rights. Information on the procedures with respect to rights in RFC documents can be found in BCP 78 and BCP 79. Copies of IPR disclosures made to the IETF Secretariat and any assurances of licenses to be made available, or the result of an attempt made to obtain a general license or permission for the use of such proprietary rights by implementers or users of this specification can be obtained from the IETF on-line IPR repository at http://www.ietf.org/ipr. The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology that may be required to implement this standard. Please address the information to the IETF at ietf- ipr@ietf.org. Disclaimer of Validity This document and the information contained herein are provided on an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Copyright Statement Copyright (C) The Internet Society (2005). This document is subject to the rights, licenses and restrictions contained in BCP 78, and except as set forth therein, the authors retain all their rights. Acknowledgment Funding for the RFC Editor function is currently provided by the Internet Society. Lyon, Wong Experimental [Page 18]