7 Internet Engineering Task Force (IETF) C. Kaufman
8 Request for Comments: 5996 Microsoft
9 Obsoletes: 4306, 4718 P. Hoffman
10 Category: Standards Track VPN Consortium
11 ISSN: 2070-1721 Y. Nir
18 Internet Key Exchange Protocol Version 2 (IKEv2)
22 This document describes version 2 of the Internet Key Exchange (IKE)
23 protocol. IKE is a component of IPsec used for performing mutual
24 authentication and establishing and maintaining Security Associations
25 (SAs). This document replaces and updates RFC 4306, and includes all
26 of the clarifications from RFC 4718.
30 This is an Internet Standards Track document.
32 This document is a product of the Internet Engineering Task Force
33 (IETF). It represents the consensus of the IETF community. It has
34 received public review and has been approved for publication by the
35 Internet Engineering Steering Group (IESG). Further information on
36 Internet Standards is available in Section 2 of RFC 5741.
38 Information about the current status of this document, any errata,
39 and how to provide feedback on it may be obtained at
40 http://www.rfc-editor.org/info/rfc5996.
58 Kaufman, et al. Standards Track [Page 1]
60 RFC 5996 IKEv2bis September 2010
65 Copyright (c) 2010 IETF Trust and the persons identified as the
66 document authors. All rights reserved.
68 This document is subject to BCP 78 and the IETF Trust's Legal
69 Provisions Relating to IETF Documents
70 (http://trustee.ietf.org/license-info) in effect on the date of
71 publication of this document. Please review these documents
72 carefully, as they describe your rights and restrictions with respect
73 to this document. Code Components extracted from this document must
74 include Simplified BSD License text as described in Section 4.e of
75 the Trust Legal Provisions and are provided without warranty as
76 described in the Simplified BSD License.
78 This document may contain material from IETF Documents or IETF
79 Contributions published or made publicly available before November
80 10, 2008. The person(s) controlling the copyright in some of this
81 material may not have granted the IETF Trust the right to allow
82 modifications of such material outside the IETF Standards Process.
83 Without obtaining an adequate license from the person(s) controlling
84 the copyright in such materials, this document may not be modified
85 outside the IETF Standards Process, and derivative works of it may
86 not be created outside the IETF Standards Process, except to format
87 it for publication as an RFC or to translate it into languages other
92 1. Introduction ....................................................5
93 1.1. Usage Scenarios ............................................6
94 1.1.1. Security Gateway to Security Gateway in
95 Tunnel Mode .........................................7
96 1.1.2. Endpoint-to-Endpoint Transport Mode .................7
97 1.1.3. Endpoint to Security Gateway in Tunnel Mode .........8
98 1.1.4. Other Scenarios .....................................9
99 1.2. The Initial Exchanges ......................................9
100 1.3. The CREATE_CHILD_SA Exchange ..............................13
101 1.3.1. Creating New Child SAs with the
102 CREATE_CHILD_SA Exchange ...........................14
103 1.3.2. Rekeying IKE SAs with the CREATE_CHILD_SA
104 Exchange ...........................................15
105 1.3.3. Rekeying Child SAs with the CREATE_CHILD_SA
106 Exchange ...........................................16
107 1.4. The INFORMATIONAL Exchange ................................17
108 1.4.1. Deleting an SA with INFORMATIONAL Exchanges ........17
109 1.5. Informational Messages outside of an IKE SA ...............18
110 1.6. Requirements Terminology ..................................19
114 Kaufman, et al. Standards Track [Page 2]
116 RFC 5996 IKEv2bis September 2010
119 1.7. Significant Differences between RFC 4306 and This
120 Document ..................................................20
121 2. IKE Protocol Details and Variations ............................22
122 2.1. Use of Retransmission Timers ..............................23
123 2.2. Use of Sequence Numbers for Message ID ....................24
124 2.3. Window Size for Overlapping Requests ......................25
125 2.4. State Synchronization and Connection Timeouts .............26
126 2.5. Version Numbers and Forward Compatibility .................28
127 2.6. IKE SA SPIs and Cookies ...................................30
128 2.6.1. Interaction of COOKIE and INVALID_KE_PAYLOAD .......33
129 2.7. Cryptographic Algorithm Negotiation .......................34
130 2.8. Rekeying ..................................................34
131 2.8.1. Simultaneous Child SA Rekeying .....................36
132 2.8.2. Simultaneous IKE SA Rekeying .......................39
133 2.8.3. Rekeying the IKE SA versus Reauthentication ........40
134 2.9. Traffic Selector Negotiation ..............................40
135 2.9.1. Traffic Selectors Violating Own Policy .............43
136 2.10. Nonces ...................................................44
137 2.11. Address and Port Agility .................................44
138 2.12. Reuse of Diffie-Hellman Exponentials .....................44
139 2.13. Generating Keying Material ...............................45
140 2.14. Generating Keying Material for the IKE SA ................46
141 2.15. Authentication of the IKE SA .............................47
142 2.16. Extensible Authentication Protocol Methods ...............50
143 2.17. Generating Keying Material for Child SAs .................52
144 2.18. Rekeying IKE SAs Using a CREATE_CHILD_SA Exchange ........53
145 2.19. Requesting an Internal Address on a Remote Network .......53
146 2.20. Requesting the Peer's Version ............................55
147 2.21. Error Handling ...........................................56
148 2.21.1. Error Handling in IKE_SA_INIT .....................56
149 2.21.2. Error Handling in IKE_AUTH ........................57
150 2.21.3. Error Handling after IKE SA is Authenticated ......58
151 2.21.4. Error Handling Outside IKE SA .....................58
152 2.22. IPComp ...................................................59
153 2.23. NAT Traversal ............................................60
154 2.23.1. Transport Mode NAT Traversal ......................64
155 2.24. Explicit Congestion Notification (ECN) ...................68
156 2.25. Exchange Collisions ......................................68
157 2.25.1. Collisions while Rekeying or Closing Child SAs ....69
158 2.25.2. Collisions while Rekeying or Closing IKE SAs ......69
159 3. Header and Payload Formats .....................................69
160 3.1. The IKE Header ............................................70
161 3.2. Generic Payload Header ....................................73
162 3.3. Security Association Payload ..............................75
163 3.3.1. Proposal Substructure ..............................78
164 3.3.2. Transform Substructure .............................79
165 3.3.3. Valid Transform Types by Protocol ..................82
166 3.3.4. Mandatory Transform IDs ............................83
170 Kaufman, et al. Standards Track [Page 3]
172 RFC 5996 IKEv2bis September 2010
175 3.3.5. Transform Attributes ...............................84
176 3.3.6. Attribute Negotiation ..............................86
177 3.4. Key Exchange Payload ......................................87
178 3.5. Identification Payloads ...................................87
179 3.6. Certificate Payload .......................................90
180 3.7. Certificate Request Payload ...............................93
181 3.8. Authentication Payload ....................................95
182 3.9. Nonce Payload .............................................96
183 3.10. Notify Payload ...........................................97
184 3.10.1. Notify Message Types ..............................98
185 3.11. Delete Payload ..........................................101
186 3.12. Vendor ID Payload .......................................102
187 3.13. Traffic Selector Payload ................................103
188 3.13.1. Traffic Selector .................................105
189 3.14. Encrypted Payload .......................................107
190 3.15. Configuration Payload ...................................109
191 3.15.1. Configuration Attributes .........................110
192 3.15.2. Meaning of INTERNAL_IP4_SUBNET and
193 INTERNAL_IP6_SUBNET ..............................113
194 3.15.3. Configuration Payloads for IPv6 ..................115
195 3.15.4. Address Assignment Failures ......................116
196 3.16. Extensible Authentication Protocol (EAP) Payload ........117
197 4. Conformance Requirements ......................................118
198 5. Security Considerations .......................................120
199 5.1. Traffic Selector Authorization ...........................123
200 6. IANA Considerations ...........................................124
201 7. Acknowledgements ..............................................125
202 8. References ....................................................126
203 8.1. Normative References .....................................126
204 8.2. Informative References ...................................127
205 Appendix A. Summary of Changes from IKEv1 ........................132
206 Appendix B. Diffie-Hellman Groups ................................133
207 B.1. Group 1 - 768-bit MODP ....................................133
208 B.2. Group 2 - 1024-bit MODP ...................................133
209 Appendix C. Exchanges and Payloads ..............................134
210 C.1. IKE_SA_INIT Exchange .....................................134
211 C.2. IKE_AUTH Exchange without EAP .............................135
212 C.3. IKE_AUTH Exchange with EAP ...............................136
213 C.4. CREATE_CHILD_SA Exchange for Creating or Rekeying
214 Child SAs .................................................137
215 C.5. CREATE_CHILD_SA Exchange for Rekeying the IKE SA ..........137
216 C.6. INFORMATIONAL Exchange ....................................137
226 Kaufman, et al. Standards Track [Page 4]
228 RFC 5996 IKEv2bis September 2010
233 IP Security (IPsec) provides confidentiality, data integrity, access
234 control, and data source authentication to IP datagrams. These
235 services are provided by maintaining shared state between the source
236 and the sink of an IP datagram. This state defines, among other
237 things, the specific services provided to the datagram, which
238 cryptographic algorithms will be used to provide the services, and
239 the keys used as input to the cryptographic algorithms.
241 Establishing this shared state in a manual fashion does not scale
242 well. Therefore, a protocol to establish this state dynamically is
243 needed. This document describes such a protocol -- the Internet Key
244 Exchange (IKE). Version 1 of IKE was defined in RFCs 2407 [DOI],
245 2408 [ISAKMP], and 2409 [IKEV1]. IKEv2 replaced all of those RFCs.
246 IKEv2 was defined in [IKEV2] (RFC 4306) and was clarified in [Clarif]
247 (RFC 4718). This document replaces and updates RFC 4306 and RFC
248 4718. IKEv2 was a change to the IKE protocol that was not backward
249 compatible. In contrast, the current document not only provides a
250 clarification of IKEv2, but makes minimum changes to the IKE
251 protocol. A list of the significant differences between RFC 4306 and
252 this document is given in Section 1.7.
254 IKE performs mutual authentication between two parties and
255 establishes an IKE security association (SA) that includes shared
256 secret information that can be used to efficiently establish SAs for
257 Encapsulating Security Payload (ESP) [ESP] or Authentication Header
258 (AH) [AH] and a set of cryptographic algorithms to be used by the SAs
259 to protect the traffic that they carry. In this document, the term
260 "suite" or "cryptographic suite" refers to a complete set of
261 algorithms used to protect an SA. An initiator proposes one or more
262 suites by listing supported algorithms that can be combined into
263 suites in a mix-and-match fashion. IKE can also negotiate use of IP
264 Compression (IPComp) [IP-COMP] in connection with an ESP or AH SA.
265 The SAs for ESP or AH that get set up through that IKE SA we call
268 All IKE communications consist of pairs of messages: a request and a
269 response. The pair is called an "exchange", and is sometimes called
270 a "request/response pair". The first exchange of messages
271 establishing an IKE SA are called the IKE_SA_INIT and IKE_AUTH
272 exchanges; subsequent IKE exchanges are called the CREATE_CHILD_SA or
273 INFORMATIONAL exchanges. In the common case, there is a single
274 IKE_SA_INIT exchange and a single IKE_AUTH exchange (a total of four
275 messages) to establish the IKE SA and the first Child SA. In
276 exceptional cases, there may be more than one of each of these
277 exchanges. In all cases, all IKE_SA_INIT exchanges MUST complete
278 before any other exchange type, then all IKE_AUTH exchanges MUST
282 Kaufman, et al. Standards Track [Page 5]
284 RFC 5996 IKEv2bis September 2010
287 complete, and following that, any number of CREATE_CHILD_SA and
288 INFORMATIONAL exchanges may occur in any order. In some scenarios,
289 only a single Child SA is needed between the IPsec endpoints, and
290 therefore there would be no additional exchanges. Subsequent
291 exchanges MAY be used to establish additional Child SAs between the
292 same authenticated pair of endpoints and to perform housekeeping
295 An IKE message flow always consists of a request followed by a
296 response. It is the responsibility of the requester to ensure
297 reliability. If the response is not received within a timeout
298 interval, the requester needs to retransmit the request (or abandon
301 The first exchange of an IKE session, IKE_SA_INIT, negotiates
302 security parameters for the IKE SA, sends nonces, and sends Diffie-
305 The second exchange, IKE_AUTH, transmits identities, proves knowledge
306 of the secrets corresponding to the two identities, and sets up an SA
307 for the first (and often only) AH or ESP Child SA (unless there is
308 failure setting up the AH or ESP Child SA, in which case the IKE SA
309 is still established without the Child SA).
311 The types of subsequent exchanges are CREATE_CHILD_SA (which creates
312 a Child SA) and INFORMATIONAL (which deletes an SA, reports error
313 conditions, or does other housekeeping). Every request requires a
314 response. An INFORMATIONAL request with no payloads (other than the
315 empty Encrypted payload required by the syntax) is commonly used as a
316 check for liveness. These subsequent exchanges cannot be used until
317 the initial exchanges have completed.
319 In the description that follows, we assume that no errors occur.
320 Modifications to the flow when errors occur are described in
325 IKE is used to negotiate ESP or AH SAs in a number of different
326 scenarios, each with its own special requirements.
338 Kaufman, et al. Standards Track [Page 6]
340 RFC 5996 IKEv2bis September 2010
343 1.1.1. Security Gateway to Security Gateway in Tunnel Mode
345 +-+-+-+-+-+ +-+-+-+-+-+
347 Protected |Tunnel | tunnel |Tunnel | Protected
348 Subnet <-->|Endpoint |<---------->|Endpoint |<--> Subnet
350 +-+-+-+-+-+ +-+-+-+-+-+
352 Figure 1: Security Gateway to Security Gateway Tunnel
354 In this scenario, neither endpoint of the IP connection implements
355 IPsec, but network nodes between them protect traffic for part of the
356 way. Protection is transparent to the endpoints, and depends on
357 ordinary routing to send packets through the tunnel endpoints for
358 processing. Each endpoint would announce the set of addresses
359 "behind" it, and packets would be sent in tunnel mode where the inner
360 IP header would contain the IP addresses of the actual endpoints.
362 1.1.2. Endpoint-to-Endpoint Transport Mode
364 +-+-+-+-+-+ +-+-+-+-+-+
365 | | IPsec transport | |
366 |Protected| or tunnel mode SA |Protected|
367 |Endpoint |<---------------------------------------->|Endpoint |
369 +-+-+-+-+-+ +-+-+-+-+-+
371 Figure 2: Endpoint to Endpoint
373 In this scenario, both endpoints of the IP connection implement
374 IPsec, as required of hosts in [IPSECARCH]. Transport mode will
375 commonly be used with no inner IP header. A single pair of addresses
376 will be negotiated for packets to be protected by this SA. These
377 endpoints MAY implement application-layer access controls based on
378 the IPsec authenticated identities of the participants. This
379 scenario enables the end-to-end security that has been a guiding
380 principle for the Internet since [ARCHPRINC], [TRANSPARENCY], and a
381 method of limiting the inherent problems with complexity in networks
382 noted by [ARCHGUIDEPHIL]. Although this scenario may not be fully
383 applicable to the IPv4 Internet, it has been deployed successfully in
384 specific scenarios within intranets using IKEv1. It should be more
385 broadly enabled during the transition to IPv6 and with the adoption
394 Kaufman, et al. Standards Track [Page 7]
396 RFC 5996 IKEv2bis September 2010
399 It is possible in this scenario that one or both of the protected
400 endpoints will be behind a network address translation (NAT) node, in
401 which case the tunneled packets will have to be UDP encapsulated so
402 that port numbers in the UDP headers can be used to identify
403 individual endpoints "behind" the NAT (see Section 2.23).
405 1.1.3. Endpoint to Security Gateway in Tunnel Mode
407 +-+-+-+-+-+ +-+-+-+-+-+
408 | | IPsec | | Protected
409 |Protected| tunnel |Tunnel | Subnet
410 |Endpoint |<------------------------>|Endpoint |<--- and/or
412 +-+-+-+-+-+ +-+-+-+-+-+
414 Figure 3: Endpoint to Security Gateway Tunnel
416 In this scenario, a protected endpoint (typically a portable roaming
417 computer) connects back to its corporate network through an IPsec-
418 protected tunnel. It might use this tunnel only to access
419 information on the corporate network, or it might tunnel all of its
420 traffic back through the corporate network in order to take advantage
421 of protection provided by a corporate firewall against Internet-based
422 attacks. In either case, the protected endpoint will want an IP
423 address associated with the security gateway so that packets returned
424 to it will go to the security gateway and be tunneled back. This IP
425 address may be static or may be dynamically allocated by the security
426 gateway. In support of the latter case, IKEv2 includes a mechanism
427 (namely, configuration payloads) for the initiator to request an IP
428 address owned by the security gateway for use for the duration of its
431 In this scenario, packets will use tunnel mode. On each packet from
432 the protected endpoint, the outer IP header will contain the source
433 IP address associated with its current location (i.e., the address
434 that will get traffic routed to the endpoint directly), while the
435 inner IP header will contain the source IP address assigned by the
436 security gateway (i.e., the address that will get traffic routed to
437 the security gateway for forwarding to the endpoint). The outer
438 destination address will always be that of the security gateway,
439 while the inner destination address will be the ultimate destination
442 In this scenario, it is possible that the protected endpoint will be
443 behind a NAT. In that case, the IP address as seen by the security
444 gateway will not be the same as the IP address sent by the protected
450 Kaufman, et al. Standards Track [Page 8]
452 RFC 5996 IKEv2bis September 2010
455 endpoint, and packets will have to be UDP encapsulated in order to be
456 routed properly. Interaction with NATs is covered in detail in
459 1.1.4. Other Scenarios
461 Other scenarios are possible, as are nested combinations of the
462 above. One notable example combines aspects of Sections 1.1.1 and
463 1.1.3. A subnet may make all external accesses through a remote
464 security gateway using an IPsec tunnel, where the addresses on the
465 subnet are routed to the security gateway by the rest of the
466 Internet. An example would be someone's home network being virtually
467 on the Internet with static IP addresses even though connectivity is
468 provided by an ISP that assigns a single dynamically assigned IP
469 address to the user's security gateway (where the static IP addresses
470 and an IPsec relay are provided by a third party located elsewhere).
472 1.2. The Initial Exchanges
474 Communication using IKE always begins with IKE_SA_INIT and IKE_AUTH
475 exchanges (known in IKEv1 as Phase 1). These initial exchanges
476 normally consist of four messages, though in some scenarios that
477 number can grow. All communications using IKE consist of request/
478 response pairs. We'll describe the base exchange first, followed by
479 variations. The first pair of messages (IKE_SA_INIT) negotiate
480 cryptographic algorithms, exchange nonces, and do a Diffie-Hellman
483 The second pair of messages (IKE_AUTH) authenticate the previous
484 messages, exchange identities and certificates, and establish the
485 first Child SA. Parts of these messages are encrypted and integrity
486 protected with keys established through the IKE_SA_INIT exchange, so
487 the identities are hidden from eavesdroppers and all fields in all
488 the messages are authenticated. See Section 2.14 for information on
489 how the encryption keys are generated. (A man-in-the-middle attacker
490 who cannot complete the IKE_AUTH exchange can nonetheless see the
491 identity of the initiator.)
493 All messages following the initial exchange are cryptographically
494 protected using the cryptographic algorithms and keys negotiated in
495 the IKE_SA_INIT exchange. These subsequent messages use the syntax
496 of the Encrypted payload described in Section 3.14, encrypted with
497 keys that are derived as described in Section 2.14. All subsequent
498 messages include an Encrypted payload, even if they are referred to
499 in the text as "empty". For the CREATE_CHILD_SA, IKE_AUTH, or
500 INFORMATIONAL exchanges, the message following the header is
501 encrypted and the message including the header is integrity protected
502 using the cryptographic algorithms negotiated for the IKE SA.
506 Kaufman, et al. Standards Track [Page 9]
508 RFC 5996 IKEv2bis September 2010
511 Every IKE message contains a Message ID as part of its fixed header.
512 This Message ID is used to match up requests and responses, and to
513 identify retransmissions of messages.
515 In the following descriptions, the payloads contained in the message
516 are indicated by names as listed below.
519 -----------------------------------------
522 CERTREQ Certificate Request
525 EAP Extensible Authentication
526 HDR IKE header (not a payload)
527 IDi Identification - Initiator
528 IDr Identification - Responder
532 SA Security Association
533 SK Encrypted and Authenticated
534 TSi Traffic Selector - Initiator
535 TSr Traffic Selector - Responder
538 The details of the contents of each payload are described in section
539 3. Payloads that may optionally appear will be shown in brackets,
540 such as [CERTREQ]; this indicates that a Certificate Request payload
541 can optionally be included.
543 The initial exchanges are as follows:
546 -------------------------------------------------------------------
547 HDR, SAi1, KEi, Ni -->
549 HDR contains the Security Parameter Indexes (SPIs), version numbers,
550 and flags of various sorts. The SAi1 payload states the
551 cryptographic algorithms the initiator supports for the IKE SA. The
552 KE payload sends the initiator's Diffie-Hellman value. Ni is the
555 <-- HDR, SAr1, KEr, Nr, [CERTREQ]
562 Kaufman, et al. Standards Track [Page 10]
564 RFC 5996 IKEv2bis September 2010
567 The responder chooses a cryptographic suite from the initiator's
568 offered choices and expresses that choice in the SAr1 payload,
569 completes the Diffie-Hellman exchange with the KEr payload, and sends
570 its nonce in the Nr payload.
572 At this point in the negotiation, each party can generate SKEYSEED,
573 from which all keys are derived for that IKE SA. The messages that
574 follow are encrypted and integrity protected in their entirety, with
575 the exception of the message headers. The keys used for the
576 encryption and integrity protection are derived from SKEYSEED and are
577 known as SK_e (encryption) and SK_a (authentication, a.k.a. integrity
578 protection); see Sections 2.13 and 2.14 for details on the key
579 derivation. A separate SK_e and SK_a is computed for each direction.
580 In addition to the keys SK_e and SK_a derived from the Diffie-Hellman
581 value for protection of the IKE SA, another quantity SK_d is derived
582 and used for derivation of further keying material for Child SAs.
583 The notation SK { ... } indicates that these payloads are encrypted
584 and integrity protected using that direction's SK_e and SK_a.
586 HDR, SK {IDi, [CERT,] [CERTREQ,]
590 The initiator asserts its identity with the IDi payload, proves
591 knowledge of the secret corresponding to IDi and integrity protects
592 the contents of the first message using the AUTH payload (see
593 Section 2.15). It might also send its certificate(s) in CERT
594 payload(s) and a list of its trust anchors in CERTREQ payload(s). If
595 any CERT payloads are included, the first certificate provided MUST
596 contain the public key used to verify the AUTH field.
598 The optional payload IDr enables the initiator to specify to which of
599 the responder's identities it wants to talk. This is useful when the
600 machine on which the responder is running is hosting multiple
601 identities at the same IP address. If the IDr proposed by the
602 initiator is not acceptable to the responder, the responder might use
603 some other IDr to finish the exchange. If the initiator then does
604 not accept the fact that responder used an IDr different than the one
605 that was requested, the initiator can close the SA after noticing the
608 The Traffic Selectors (TSi and TSr) are discussed in Section 2.9.
610 The initiator begins negotiation of a Child SA using the SAi2
611 payload. The final fields (starting with SAi2) are described in the
612 description of the CREATE_CHILD_SA exchange.
618 Kaufman, et al. Standards Track [Page 11]
620 RFC 5996 IKEv2bis September 2010
623 <-- HDR, SK {IDr, [CERT,] AUTH,
626 The responder asserts its identity with the IDr payload, optionally
627 sends one or more certificates (again with the certificate containing
628 the public key used to verify AUTH listed first), authenticates its
629 identity and protects the integrity of the second message with the
630 AUTH payload, and completes negotiation of a Child SA with the
631 additional fields described below in the CREATE_CHILD_SA exchange.
633 Both parties in the IKE_AUTH exchange MUST verify that all signatures
634 and Message Authentication Codes (MACs) are computed correctly. If
635 either side uses a shared secret for authentication, the names in the
636 ID payload MUST correspond to the key used to generate the AUTH
639 Because the initiator sends its Diffie-Hellman value in the
640 IKE_SA_INIT, it must guess the Diffie-Hellman group that the
641 responder will select from its list of supported groups. If the
642 initiator guesses wrong, the responder will respond with a Notify
643 payload of type INVALID_KE_PAYLOAD indicating the selected group. In
644 this case, the initiator MUST retry the IKE_SA_INIT with the
645 corrected Diffie-Hellman group. The initiator MUST again propose its
646 full set of acceptable cryptographic suites because the rejection
647 message was unauthenticated and otherwise an active attacker could
648 trick the endpoints into negotiating a weaker suite than a stronger
649 one that they both prefer.
651 If creating the Child SA during the IKE_AUTH exchange fails for some
652 reason, the IKE SA is still created as usual. The list of Notify
653 message types in the IKE_AUTH exchange that do not prevent an IKE SA
654 from being set up include at least the following: NO_PROPOSAL_CHOSEN,
655 TS_UNACCEPTABLE, SINGLE_PAIR_REQUIRED, INTERNAL_ADDRESS_FAILURE, and
658 If the failure is related to creating the IKE SA (for example, an
659 AUTHENTICATION_FAILED Notify error message is returned), the IKE SA
660 is not created. Note that although the IKE_AUTH messages are
661 encrypted and integrity protected, if the peer receiving this Notify
662 error message has not yet authenticated the other end (or if the peer
663 fails to authenticate the other end for some reason), the information
664 needs to be treated with caution. More precisely, assuming that the
665 MAC verifies correctly, the sender of the error Notify message is
666 known to be the responder of the IKE_SA_INIT exchange, but the
667 sender's identity cannot be assured.
674 Kaufman, et al. Standards Track [Page 12]
676 RFC 5996 IKEv2bis September 2010
679 Note that IKE_AUTH messages do not contain KEi/KEr or Ni/Nr payloads.
680 Thus, the SA payloads in the IKE_AUTH exchange cannot contain
681 Transform Type 4 (Diffie-Hellman group) with any value other than
682 NONE. Implementations SHOULD omit the whole transform substructure
683 instead of sending value NONE.
685 1.3. The CREATE_CHILD_SA Exchange
687 The CREATE_CHILD_SA exchange is used to create new Child SAs and to
688 rekey both IKE SAs and Child SAs. This exchange consists of a single
689 request/response pair, and some of its function was referred to as a
690 Phase 2 exchange in IKEv1. It MAY be initiated by either end of the
691 IKE SA after the initial exchanges are completed.
693 An SA is rekeyed by creating a new SA and then deleting the old one.
694 This section describes the first part of rekeying, the creation of
695 new SAs; Section 2.8 covers the mechanics of rekeying, including
696 moving traffic from old to new SAs and the deletion of the old SAs.
697 The two sections must be read together to understand the entire
700 Either endpoint may initiate a CREATE_CHILD_SA exchange, so in this
701 section the term initiator refers to the endpoint initiating this
702 exchange. An implementation MAY refuse all CREATE_CHILD_SA requests
705 The CREATE_CHILD_SA request MAY optionally contain a KE payload for
706 an additional Diffie-Hellman exchange to enable stronger guarantees
707 of forward secrecy for the Child SA. The keying material for the
708 Child SA is a function of SK_d established during the establishment
709 of the IKE SA, the nonces exchanged during the CREATE_CHILD_SA
710 exchange, and the Diffie-Hellman value (if KE payloads are included
711 in the CREATE_CHILD_SA exchange).
713 If a CREATE_CHILD_SA exchange includes a KEi payload, at least one of
714 the SA offers MUST include the Diffie-Hellman group of the KEi. The
715 Diffie-Hellman group of the KEi MUST be an element of the group the
716 initiator expects the responder to accept (additional Diffie-Hellman
717 groups can be proposed). If the responder selects a proposal using a
718 different Diffie-Hellman group (other than NONE), the responder MUST
719 reject the request and indicate its preferred Diffie-Hellman group in
720 the INVALID_KE_PAYLOAD Notify payload. There are two octets of data
721 associated with this notification: the accepted Diffie-Hellman group
722 number in big endian order. In the case of such a rejection, the
723 CREATE_CHILD_SA exchange fails, and the initiator will probably retry
724 the exchange with a Diffie-Hellman proposal and KEi in the group that
725 the responder gave in the INVALID_KE_PAYLOAD Notify payload.
730 Kaufman, et al. Standards Track [Page 13]
732 RFC 5996 IKEv2bis September 2010
735 The responder sends a NO_ADDITIONAL_SAS notification to indicate that
736 a CREATE_CHILD_SA request is unacceptable because the responder is
737 unwilling to accept any more Child SAs on this IKE SA. This
738 notification can also be used to reject IKE SA rekey. Some minimal
739 implementations may only accept a single Child SA setup in the
740 context of an initial IKE exchange and reject any subsequent attempts
743 1.3.1. Creating New Child SAs with the CREATE_CHILD_SA Exchange
745 A Child SA may be created by sending a CREATE_CHILD_SA request. The
746 CREATE_CHILD_SA request for creating a new Child SA is:
749 -------------------------------------------------------------------
750 HDR, SK {SA, Ni, [KEi],
753 The initiator sends SA offer(s) in the SA payload, a nonce in the Ni
754 payload, optionally a Diffie-Hellman value in the KEi payload, and
755 the proposed Traffic Selectors for the proposed Child SA in the TSi
758 The CREATE_CHILD_SA response for creating a new Child SA is:
760 <-- HDR, SK {SA, Nr, [KEr],
763 The responder replies (using the same Message ID to respond) with the
764 accepted offer in an SA payload, and a Diffie-Hellman value in the
765 KEr payload if KEi was included in the request and the selected
766 cryptographic suite includes that group.
768 The Traffic Selectors for traffic to be sent on that SA are specified
769 in the TS payloads in the response, which may be a subset of what the
770 initiator of the Child SA proposed.
772 The USE_TRANSPORT_MODE notification MAY be included in a request
773 message that also includes an SA payload requesting a Child SA. It
774 requests that the Child SA use transport mode rather than tunnel mode
775 for the SA created. If the request is accepted, the response MUST
776 also include a notification of type USE_TRANSPORT_MODE. If the
777 responder declines the request, the Child SA will be established in
778 tunnel mode. If this is unacceptable to the initiator, the initiator
779 MUST delete the SA. Note: Except when using this option to negotiate
780 transport mode, all Child SAs will use tunnel mode.
786 Kaufman, et al. Standards Track [Page 14]
788 RFC 5996 IKEv2bis September 2010
791 The ESP_TFC_PADDING_NOT_SUPPORTED notification asserts that the
792 sending endpoint will not accept packets that contain Traffic Flow
793 Confidentiality (TFC) padding over the Child SA being negotiated. If
794 neither endpoint accepts TFC padding, this notification is included
795 in both the request and the response. If this notification is
796 included in only one of the messages, TFC padding can still be sent
797 in the other direction.
799 The NON_FIRST_FRAGMENTS_ALSO notification is used for fragmentation
800 control. See [IPSECARCH] for a fuller explanation. Both parties
801 need to agree to sending non-first fragments before either party does
802 so. It is enabled only if NON_FIRST_FRAGMENTS_ALSO notification is
803 included in both the request proposing an SA and the response
804 accepting it. If the responder does not want to send or receive non-
805 first fragments, it only omits NON_FIRST_FRAGMENTS_ALSO notification
806 from its response, but does not reject the whole Child SA creation.
808 An IPCOMP_SUPPORTED notification, covered in Section 2.22, can also
809 be included in the exchange.
811 A failed attempt to create a Child SA SHOULD NOT tear down the IKE
812 SA: there is no reason to lose the work done to set up the IKE SA.
813 See Section 2.21 for a list of error messages that might occur if
814 creating a Child SA fails.
816 1.3.2. Rekeying IKE SAs with the CREATE_CHILD_SA Exchange
818 The CREATE_CHILD_SA request for rekeying an IKE SA is:
821 -------------------------------------------------------------------
822 HDR, SK {SA, Ni, KEi} -->
824 The initiator sends SA offer(s) in the SA payload, a nonce in the Ni
825 payload, and a Diffie-Hellman value in the KEi payload. The KEi
826 payload MUST be included. A new initiator SPI is supplied in the SPI
827 field of the SA payload. Once a peer receives a request to rekey an
828 IKE SA or sends a request to rekey an IKE SA, it SHOULD NOT start any
829 new CREATE_CHILD_SA exchanges on the IKE SA that is being rekeyed.
831 The CREATE_CHILD_SA response for rekeying an IKE SA is:
833 <-- HDR, SK {SA, Nr, KEr}
835 The responder replies (using the same Message ID to respond) with the
836 accepted offer in an SA payload, and a Diffie-Hellman value in the
837 KEr payload if the selected cryptographic suite includes that group.
838 A new responder SPI is supplied in the SPI field of the SA payload.
842 Kaufman, et al. Standards Track [Page 15]
844 RFC 5996 IKEv2bis September 2010
847 The new IKE SA has its message counters set to 0, regardless of what
848 they were in the earlier IKE SA. The first IKE requests from both
849 sides on the new IKE SA will have Message ID 0. The old IKE SA
850 retains its numbering, so any further requests (for example, to
851 delete the IKE SA) will have consecutive numbering. The new IKE SA
852 also has its window size reset to 1, and the initiator in this rekey
853 exchange is the new "original initiator" of the new IKE SA.
855 Section 2.18 also covers IKE SA rekeying in detail.
857 1.3.3. Rekeying Child SAs with the CREATE_CHILD_SA Exchange
859 The CREATE_CHILD_SA request for rekeying a Child SA is:
862 -------------------------------------------------------------------
863 HDR, SK {N(REKEY_SA), SA, Ni, [KEi],
866 The initiator sends SA offer(s) in the SA payload, a nonce in the Ni
867 payload, optionally a Diffie-Hellman value in the KEi payload, and
868 the proposed Traffic Selectors for the proposed Child SA in the TSi
871 The notifications described in Section 1.3.1 may also be sent in a
872 rekeying exchange. Usually, these will be the same notifications
873 that were used in the original exchange; for example, when rekeying a
874 transport mode SA, the USE_TRANSPORT_MODE notification will be used.
876 The REKEY_SA notification MUST be included in a CREATE_CHILD_SA
877 exchange if the purpose of the exchange is to replace an existing ESP
878 or AH SA. The SA being rekeyed is identified by the SPI field in the
879 Notify payload; this is the SPI the exchange initiator would expect
880 in inbound ESP or AH packets. There is no data associated with this
881 Notify message type. The Protocol ID field of the REKEY_SA
882 notification is set to match the protocol of the SA we are rekeying,
883 for example, 3 for ESP and 2 for AH.
885 The CREATE_CHILD_SA response for rekeying a Child SA is:
887 <-- HDR, SK {SA, Nr, [KEr],
890 The responder replies (using the same Message ID to respond) with the
891 accepted offer in an SA payload, and a Diffie-Hellman value in the
892 KEr payload if KEi was included in the request and the selected
893 cryptographic suite includes that group.
898 Kaufman, et al. Standards Track [Page 16]
900 RFC 5996 IKEv2bis September 2010
903 The Traffic Selectors for traffic to be sent on that SA are specified
904 in the TS payloads in the response, which may be a subset of what the
905 initiator of the Child SA proposed.
907 1.4. The INFORMATIONAL Exchange
909 At various points during the operation of an IKE SA, peers may desire
910 to convey control messages to each other regarding errors or
911 notifications of certain events. To accomplish this, IKE defines an
912 INFORMATIONAL exchange. INFORMATIONAL exchanges MUST ONLY occur
913 after the initial exchanges and are cryptographically protected with
914 the negotiated keys. Note that some informational messages, not
915 exchanges, can be sent outside the context of an IKE SA. Section
916 2.21 also covers error messages in great detail.
918 Control messages that pertain to an IKE SA MUST be sent under that
919 IKE SA. Control messages that pertain to Child SAs MUST be sent
920 under the protection of the IKE SA that generated them (or its
921 successor if the IKE SA was rekeyed).
923 Messages in an INFORMATIONAL exchange contain zero or more
924 Notification, Delete, and Configuration payloads. The recipient of
925 an INFORMATIONAL exchange request MUST send some response; otherwise,
926 the sender will assume the message was lost in the network and will
927 retransmit it. That response MAY be an empty message. The request
928 message in an INFORMATIONAL exchange MAY also contain no payloads.
929 This is the expected way an endpoint can ask the other endpoint to
930 verify that it is alive.
932 The INFORMATIONAL exchange is defined as:
935 -------------------------------------------------------------------
938 <-- HDR, SK {[N,] [D,]
941 The processing of an INFORMATIONAL exchange is determined by its
944 1.4.1. Deleting an SA with INFORMATIONAL Exchanges
946 ESP and AH SAs always exist in pairs, with one SA in each direction.
947 When an SA is closed, both members of the pair MUST be closed (that
948 is, deleted). Each endpoint MUST close its incoming SAs and allow
949 the other endpoint to close the other SA in each pair. To delete an
950 SA, an INFORMATIONAL exchange with one or more Delete payloads is
954 Kaufman, et al. Standards Track [Page 17]
956 RFC 5996 IKEv2bis September 2010
959 sent listing the SPIs (as they would be expected in the headers of
960 inbound packets) of the SAs to be deleted. The recipient MUST close
961 the designated SAs. Note that one never sends Delete payloads for
962 the two sides of an SA in a single message. If there are many SAs to
963 delete at the same time, one includes Delete payloads for the inbound
964 half of each SA pair in the INFORMATIONAL exchange.
966 Normally, the response in the INFORMATIONAL exchange will contain
967 Delete payloads for the paired SAs going in the other direction.
968 There is one exception. If, by chance, both ends of a set of SAs
969 independently decide to close them, each may send a Delete payload
970 and the two requests may cross in the network. If a node receives a
971 delete request for SAs for which it has already issued a delete
972 request, it MUST delete the outgoing SAs while processing the request
973 and the incoming SAs while processing the response. In that case,
974 the responses MUST NOT include Delete payloads for the deleted SAs,
975 since that would result in duplicate deletion and could in theory
978 Similar to ESP and AH SAs, IKE SAs are also deleted by sending an
979 Informational exchange. Deleting an IKE SA implicitly closes any
980 remaining Child SAs negotiated under it. The response to a request
981 that deletes the IKE SA is an empty INFORMATIONAL response.
983 Half-closed ESP or AH connections are anomalous, and a node with
984 auditing capability should probably audit their existence if they
985 persist. Note that this specification does not specify time periods,
986 so it is up to individual endpoints to decide how long to wait. A
987 node MAY refuse to accept incoming data on half-closed connections
988 but MUST NOT unilaterally close them and reuse the SPIs. If
989 connection state becomes sufficiently messed up, a node MAY close the
990 IKE SA, as described above. It can then rebuild the SAs it needs on
991 a clean base under a new IKE SA.
993 1.5. Informational Messages outside of an IKE SA
995 There are some cases in which a node receives a packet that it cannot
996 process, but it may want to notify the sender about this situation.
998 o If an ESP or AH packet arrives with an unrecognized SPI. This
999 might be due to the receiving node having recently crashed and
1000 lost state, or because of some other system malfunction or attack.
1002 o If an encrypted IKE request packet arrives on port 500 or 4500
1003 with an unrecognized IKE SPI. This might be due to the receiving
1004 node having recently crashed and lost state, or because of some
1005 other system malfunction or attack.
1010 Kaufman, et al. Standards Track [Page 18]
1012 RFC 5996 IKEv2bis September 2010
1015 o If an IKE request packet arrives with a higher major version
1016 number than the implementation supports.
1018 In the first case, if the receiving node has an active IKE SA to the
1019 IP address from whence the packet came, it MAY send an INVALID_SPI
1020 notification of the wayward packet over that IKE SA in an
1021 INFORMATIONAL exchange. The Notification Data contains the SPI of
1022 the invalid packet. The recipient of this notification cannot tell
1023 whether the SPI is for AH or ESP, but this is not important because
1024 the SPIs are supposed to be different for the two. If no suitable
1025 IKE SA exists, the node MAY send an informational message without
1026 cryptographic protection to the source IP address, using the source
1027 UDP port as the destination port if the packet was UDP (UDP-
1028 encapsulated ESP or AH). In this case, it should only be used by the
1029 recipient as a hint that something might be wrong (because it could
1030 easily be forged). This message is not part of an INFORMATIONAL
1031 exchange, and the receiving node MUST NOT respond to it because doing
1032 so could cause a message loop. The message is constructed as
1033 follows: there are no IKE SPI values that would be meaningful to the
1034 recipient of such a notification; using zero values or random values
1035 are both acceptable, this being the exception to the rule in
1036 Section 3.1 that prohibits zero IKE Initiator SPIs. The Initiator
1037 flag is set to 1, the Response flag is set to 0, and the version
1038 flags are set in the normal fashion; these flags are described in
1041 In the second and third cases, the message is always sent without
1042 cryptographic protection (outside of an IKE SA), and includes either
1043 an INVALID_IKE_SPI or an INVALID_MAJOR_VERSION notification (with no
1044 notification data). The message is a response message, and thus it
1045 is sent to the IP address and port from whence it came with the same
1046 IKE SPIs and the Message ID and Exchange Type are copied from the
1047 request. The Response flag is set to 1, and the version flags are
1048 set in the normal fashion.
1050 1.6. Requirements Terminology
1052 Definitions of the primitive terms in this document (such as Security
1053 Association or SA) can be found in [IPSECARCH]. It should be noted
1054 that parts of IKEv2 rely on some of the processing rules in
1055 [IPSECARCH], as described in various sections of this document.
1057 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
1058 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
1059 document are to be interpreted as described in [MUSTSHOULD].
1066 Kaufman, et al. Standards Track [Page 19]
1068 RFC 5996 IKEv2bis September 2010
1071 1.7. Significant Differences between RFC 4306 and This Document
1073 This document contains clarifications and amplifications to IKEv2
1074 [IKEV2]. Many of the clarifications are based on [Clarif]. The
1075 changes listed in that document were discussed in the IPsec Working
1076 Group and, after the Working Group was disbanded, on the IPsec
1077 mailing list. That document contains detailed explanations of areas
1078 that were unclear in IKEv2, and is thus useful to implementers of
1081 The protocol described in this document retains the same major
1082 version number (2) and minor version number (0) as was used in RFC
1083 4306. That is, the version number is *not* changed from RFC 4306.
1084 The small number of technical changes listed here are not expected to
1085 affect RFC 4306 implementations that have already been deployed at
1086 the time of publication of this document.
1088 This document makes the figures and references a bit more consistent
1089 than they were in [IKEV2].
1091 IKEv2 developers have noted that the SHOULD-level requirements in RFC
1092 4306 are often unclear in that they don't say when it is OK to not
1093 obey the requirements. They also have noted that there are MUST-
1094 level requirements that are not related to interoperability. This
1095 document has more explanation of some of these requirements. All
1096 non-capitalized uses of the words SHOULD and MUST now mean their
1097 normal English sense, not the interoperability sense of [MUSTSHOULD].
1099 IKEv2 (and IKEv1) developers have noted that there is a great deal of
1100 material in the tables of codes in Section 3.10.1 in RFC 4306. This
1101 leads to implementers not having all the needed information in the
1102 main body of the document. Much of the material from those tables
1103 has been moved into the associated parts of the main body of the
1106 This document removes discussion of nesting AH and ESP. This was a
1107 mistake in RFC 4306 caused by the lag between finishing RFC 4306 and
1108 RFC 4301. Basically, IKEv2 is based on RFC 4301, which does not
1109 include "SA bundles" that were part of RFC 2401. While a single
1110 packet can go through IPsec processing multiple times, each of these
1111 passes uses a separate SA, and the passes are coordinated by the
1112 forwarding tables. In IKEv2, each of these SAs has to be created
1113 using a separate CREATE_CHILD_SA exchange.
1115 This document removes discussion of the INTERNAL_ADDRESS_EXPIRY
1116 configuration attribute because its implementation was very
1117 problematic. Implementations that conform to this document MUST
1122 Kaufman, et al. Standards Track [Page 20]
1124 RFC 5996 IKEv2bis September 2010
1127 ignore proposals that have configuration attribute type 5, the old
1128 value for INTERNAL_ADDRESS_EXPIRY. This document also removed
1129 INTERNAL_IP6_NBNS as a configuration attribute.
1131 This document removes the allowance for rejecting messages in which
1132 the payloads were not in the "right" order; now implementations MUST
1133 NOT reject them. This is due to the lack of clarity where the orders
1134 for the payloads are described.
1136 The lists of items from RFC 4306 that ended up in the IANA registry
1137 were trimmed to only include items that were actually defined in RFC
1138 4306. Also, many of those lists are now preceded with the very
1139 important instruction to developers that they really should look at
1140 the IANA registry at the time of development because new items have
1141 been added since RFC 4306.
1143 This document adds clarification on when notifications are and are
1144 not sent encrypted, depending on the state of the negotiation at the
1147 This document discusses more about how to negotiate combined-mode
1150 In Section 1.3.2, "The KEi payload SHOULD be included" was changed to
1151 be "The KEi payload MUST be included". This also led to changes in
1154 In Section 2.1, there is new material covering how the initiator's
1155 SPI and/or IP is used to differentiate if this is a "half-open" IKE
1156 SA or a new request.
1158 This document clarifies the use of the critical flag in Section 2.5.
1160 In Section 2.8, "Note that, when rekeying, the new Child SA MAY have
1161 different Traffic Selectors and algorithms than the old one" was
1162 changed to "Note that, when rekeying, the new Child SA SHOULD NOT
1163 have different Traffic Selectors and algorithms than the old one".
1165 The new Section 2.8.2 covers simultaneous IKE SA rekeying.
1167 The new Section 2.9.2 covers Traffic Selectors in rekeying.
1169 This document adds the restriction in Section 2.13 that all
1170 pseudorandom functions (PRFs) used with IKEv2 MUST take variable-
1171 sized keys. This should not affect any implementations because there
1172 were no standardized PRFs that have fixed-size keys.
1178 Kaufman, et al. Standards Track [Page 21]
1180 RFC 5996 IKEv2bis September 2010
1183 Section 2.18 requires doing a Diffie-Hellman exchange when rekeying
1184 the IKE_SA. In theory, RFC 4306 allowed a policy where the Diffie-
1185 Hellman exchange was optional, but this was not useful (or
1186 appropriate) when rekeying the IKE_SA.
1188 Section 2.21 has been greatly expanded to cover the different cases
1189 where error responses are needed and the appropriate responses to
1192 Section 2.23 clarified that, in NAT traversal, now both UDP-
1193 encapsulated IPsec packets and non-UDP-encapsulated IPsec packets
1194 need to be understood when receiving.
1196 Added Section 2.23.1 to describe NAT traversal when transport mode is
1199 Added Section 2.25 to explain how to act when there are timing
1200 collisions when deleting and/or rekeying SAs, and two new error
1201 notifications (TEMPORARY_FAILURE and CHILD_SA_NOT_FOUND) were
1204 In Section 3.6, "Implementations MUST support the HTTP method for
1205 hash-and-URL lookup. The behavior of other URL methods is not
1206 currently specified, and such methods SHOULD NOT be used in the
1207 absence of a document specifying them" was added.
1209 In Section 3.15.3, a pointer to a new document that is related to
1210 configuration of IPv6 addresses was added.
1212 Appendix C was expanded and clarified.
1214 2. IKE Protocol Details and Variations
1216 IKE normally listens and sends on UDP port 500, though IKE messages
1217 may also be received on UDP port 4500 with a slightly different
1218 format (see Section 2.23). Since UDP is a datagram (unreliable)
1219 protocol, IKE includes in its definition recovery from transmission
1220 errors, including packet loss, packet replay, and packet forgery.
1221 IKE is designed to function so long as (1) at least one of a series
1222 of retransmitted packets reaches its destination before timing out;
1223 and (2) the channel is not so full of forged and replayed packets so
1224 as to exhaust the network or CPU capacities of either endpoint. Even
1225 in the absence of those minimum performance requirements, IKE is
1226 designed to fail cleanly (as though the network were broken).
1228 Although IKEv2 messages are intended to be short, they contain
1229 structures with no hard upper bound on size (in particular, digital
1230 certificates), and IKEv2 itself does not have a mechanism for
1234 Kaufman, et al. Standards Track [Page 22]
1236 RFC 5996 IKEv2bis September 2010
1239 fragmenting large messages. IP defines a mechanism for fragmentation
1240 of oversized UDP messages, but implementations vary in the maximum
1241 message size supported. Furthermore, use of IP fragmentation opens
1242 an implementation to denial-of-service (DoS) attacks [DOSUDPPROT].
1243 Finally, some NAT and/or firewall implementations may block IP
1246 All IKEv2 implementations MUST be able to send, receive, and process
1247 IKE messages that are up to 1280 octets long, and they SHOULD be able
1248 to send, receive, and process messages that are up to 3000 octets
1249 long. IKEv2 implementations need to be aware of the maximum UDP
1250 message size supported and MAY shorten messages by leaving out some
1251 certificates or cryptographic suite proposals if that will keep
1252 messages below the maximum. Use of the "Hash and URL" formats rather
1253 than including certificates in exchanges where possible can avoid
1254 most problems. Implementations and configuration need to keep in
1255 mind, however, that if the URL lookups are possible only after the
1256 Child SA is established, recursion issues could prevent this
1257 technique from working.
1259 The UDP payload of all packets containing IKE messages sent on port
1260 4500 MUST begin with the prefix of four zeros; otherwise, the
1261 receiver won't know how to handle them.
1263 2.1. Use of Retransmission Timers
1265 All messages in IKE exist in pairs: a request and a response. The
1266 setup of an IKE SA normally consists of two exchanges. Once the IKE
1267 SA is set up, either end of the Security Association may initiate
1268 requests at any time, and there can be many requests and responses
1269 "in flight" at any given moment. But each message is labeled as
1270 either a request or a response, and for each exchange, one end of the
1271 Security Association is the initiator and the other is the responder.
1273 For every pair of IKE messages, the initiator is responsible for
1274 retransmission in the event of a timeout. The responder MUST never
1275 retransmit a response unless it receives a retransmission of the
1276 request. In that event, the responder MUST ignore the retransmitted
1277 request except insofar as it causes a retransmission of the response.
1278 The initiator MUST remember each request until it receives the
1279 corresponding response. The responder MUST remember each response
1280 until it receives a request whose sequence number is larger than or
1281 equal to the sequence number in the response plus its window size
1282 (see Section 2.3). In order to allow saving memory, responders are
1283 allowed to forget the response after a timeout of several minutes.
1284 If the responder receives a retransmitted request for which it has
1285 already forgotten the response, it MUST ignore the request (and not,
1286 for example, attempt constructing a new response).
1290 Kaufman, et al. Standards Track [Page 23]
1292 RFC 5996 IKEv2bis September 2010
1295 IKE is a reliable protocol: the initiator MUST retransmit a request
1296 until it either receives a corresponding response or deems the IKE SA
1297 to have failed. In the latter case, the initiator discards all state
1298 associated with the IKE SA and any Child SAs that were negotiated
1299 using that IKE SA. A retransmission from the initiator MUST be
1300 bitwise identical to the original request. That is, everything
1301 starting from the IKE header (the IKE SA initiator's SPI onwards)
1302 must be bitwise identical; items before it (such as the IP and UDP
1303 headers) do not have to be identical.
1305 Retransmissions of the IKE_SA_INIT request require some special
1306 handling. When a responder receives an IKE_SA_INIT request, it has
1307 to determine whether the packet is a retransmission belonging to an
1308 existing "half-open" IKE SA (in which case the responder retransmits
1309 the same response), or a new request (in which case the responder
1310 creates a new IKE SA and sends a fresh response), or it belongs to an
1311 existing IKE SA where the IKE_AUTH request has been already received
1312 (in which case the responder ignores it).
1314 It is not sufficient to use the initiator's SPI and/or IP address to
1315 differentiate between these three cases because two different peers
1316 behind a single NAT could choose the same initiator SPI. Instead, a
1317 robust responder will do the IKE SA lookup using the whole packet,
1318 its hash, or the Ni payload.
1320 The retransmission policy for one-way messages is somewhat different
1321 from that for regular messages. Because no acknowledgement is ever
1322 sent, there is no reason to gratuitously retransmit one-way messages.
1323 Given that all these messages are errors, it makes sense to send them
1324 only once per "offending" packet, and only retransmit if further
1325 offending packets are received. Still, it also makes sense to limit
1326 retransmissions of such error messages.
1328 2.2. Use of Sequence Numbers for Message ID
1330 Every IKE message contains a Message ID as part of its fixed header.
1331 This Message ID is used to match up requests and responses and to
1332 identify retransmissions of messages. Retransmission of a message
1333 MUST use the same Message ID as the original message.
1335 The Message ID is a 32-bit quantity, which is zero for the
1336 IKE_SA_INIT messages (including retries of the message due to
1337 responses such as COOKIE and INVALID_KE_PAYLOAD), and incremented for
1338 each subsequent exchange. Thus, the first pair of IKE_AUTH messages
1339 will have an ID of 1, the second (when EAP is used) will be 2, and so
1340 on. The Message ID is reset to zero in the new IKE SA after the IKE
1346 Kaufman, et al. Standards Track [Page 24]
1348 RFC 5996 IKEv2bis September 2010
1351 Each endpoint in the IKE Security Association maintains two "current"
1352 Message IDs: the next one to be used for a request it initiates and
1353 the next one it expects to see in a request from the other end.
1354 These counters increment as requests are generated and received.
1355 Responses always contain the same Message ID as the corresponding
1356 request. That means that after the initial exchange, each integer n
1357 may appear as the Message ID in four distinct messages: the nth
1358 request from the original IKE initiator, the corresponding response,
1359 the nth request from the original IKE responder, and the
1360 corresponding response. If the two ends make a very different number
1361 of requests, the Message IDs in the two directions can be very
1362 different. There is no ambiguity in the messages, however, because
1363 the Initiator and Response flags in the message header specify which
1364 of the four messages a particular one is.
1366 Throughout this document, "initiator" refers to the party who
1367 initiated the exchange being described. The "original initiator"
1368 always refers to the party who initiated the exchange that resulted
1369 in the current IKE SA. In other words, if the "original responder"
1370 starts rekeying the IKE SA, that party becomes the "original
1371 initiator" of the new IKE SA.
1373 Note that Message IDs are cryptographically protected and provide
1374 protection against message replays. In the unlikely event that
1375 Message IDs grow too large to fit in 32 bits, the IKE SA MUST be
1378 2.3. Window Size for Overlapping Requests
1380 The SET_WINDOW_SIZE notification asserts that the sending endpoint is
1381 capable of keeping state for multiple outstanding exchanges,
1382 permitting the recipient to send multiple requests before getting a
1383 response to the first. The data associated with a SET_WINDOW_SIZE
1384 notification MUST be 4 octets long and contain the big endian
1385 representation of the number of messages the sender promises to keep.
1386 The window size is always one until the initial exchanges complete.
1388 An IKE endpoint MUST wait for a response to each of its messages
1389 before sending a subsequent message unless it has received a
1390 SET_WINDOW_SIZE Notify message from its peer informing it that the
1391 peer is prepared to maintain state for multiple outstanding messages
1392 in order to allow greater throughput.
1394 After an IKE SA is set up, in order to maximize IKE throughput, an
1395 IKE endpoint MAY issue multiple requests before getting a response to
1396 any of them, up to the limit set by its peer's SET_WINDOW_SIZE.
1397 These requests may pass one another over the network. An IKE
1398 endpoint MUST be prepared to accept and process a request while it
1402 Kaufman, et al. Standards Track [Page 25]
1404 RFC 5996 IKEv2bis September 2010
1407 has a request outstanding in order to avoid a deadlock in this
1408 situation. An IKE endpoint may also accept and process multiple
1409 requests while it has a request outstanding.
1411 An IKE endpoint MUST NOT exceed the peer's stated window size for
1412 transmitted IKE requests. In other words, if the responder stated
1413 its window size is N, then when the initiator needs to make a request
1414 X, it MUST wait until it has received responses to all requests up
1415 through request X-N. An IKE endpoint MUST keep a copy of (or be able
1416 to regenerate exactly) each request it has sent until it receives the
1417 corresponding response. An IKE endpoint MUST keep a copy of (or be
1418 able to regenerate exactly) the number of previous responses equal to
1419 its declared window size in case its response was lost and the
1420 initiator requests its retransmission by retransmitting the request.
1422 An IKE endpoint supporting a window size greater than one ought to be
1423 capable of processing incoming requests out of order to maximize
1424 performance in the event of network failures or packet reordering.
1426 The window size is normally a (possibly configurable) property of a
1427 particular implementation, and is not related to congestion control
1428 (unlike the window size in TCP, for example). In particular, what
1429 the responder should do when it receives a SET_WINDOW_SIZE
1430 notification containing a smaller value than is currently in effect
1431 is not defined. Thus, there is currently no way to reduce the window
1432 size of an existing IKE SA; you can only increase it. When rekeying
1433 an IKE SA, the new IKE SA starts with window size 1 until it is
1434 explicitly increased by sending a new SET_WINDOW_SIZE notification.
1436 The INVALID_MESSAGE_ID notification is sent when an IKE Message ID
1437 outside the supported window is received. This Notify message MUST
1438 NOT be sent in a response; the invalid request MUST NOT be
1439 acknowledged. Instead, inform the other side by initiating an
1440 INFORMATIONAL exchange with Notification data containing the four-
1441 octet invalid Message ID. Sending this notification is OPTIONAL, and
1442 notifications of this type MUST be rate limited.
1444 2.4. State Synchronization and Connection Timeouts
1446 An IKE endpoint is allowed to forget all of its state associated with
1447 an IKE SA and the collection of corresponding Child SAs at any time.
1448 This is the anticipated behavior in the event of an endpoint crash
1449 and restart. It is important when an endpoint either fails or
1450 reinitializes its state that the other endpoint detect those
1451 conditions and not continue to waste network bandwidth by sending
1452 packets over discarded SAs and having them fall into a black hole.
1458 Kaufman, et al. Standards Track [Page 26]
1460 RFC 5996 IKEv2bis September 2010
1463 The INITIAL_CONTACT notification asserts that this IKE SA is the only
1464 IKE SA currently active between the authenticated identities. It MAY
1465 be sent when an IKE SA is established after a crash, and the
1466 recipient MAY use this information to delete any other IKE SAs it has
1467 to the same authenticated identity without waiting for a timeout.
1468 This notification MUST NOT be sent by an entity that may be
1469 replicated (e.g., a roaming user's credentials where the user is
1470 allowed to connect to the corporate firewall from two remote systems
1471 at the same time). The INITIAL_CONTACT notification, if sent, MUST
1472 be in the first IKE_AUTH request or response, not as a separate
1473 exchange afterwards; receiving parties MAY ignore it in other
1476 Since IKE is designed to operate in spite of DoS attacks from the
1477 network, an endpoint MUST NOT conclude that the other endpoint has
1478 failed based on any routing information (e.g., ICMP messages) or IKE
1479 messages that arrive without cryptographic protection (e.g., Notify
1480 messages complaining about unknown SPIs). An endpoint MUST conclude
1481 that the other endpoint has failed only when repeated attempts to
1482 contact it have gone unanswered for a timeout period or when a
1483 cryptographically protected INITIAL_CONTACT notification is received
1484 on a different IKE SA to the same authenticated identity. An
1485 endpoint should suspect that the other endpoint has failed based on
1486 routing information and initiate a request to see whether the other
1487 endpoint is alive. To check whether the other side is alive, IKE
1488 specifies an empty INFORMATIONAL message that (like all IKE requests)
1489 requires an acknowledgement (note that within the context of an IKE
1490 SA, an "empty" message consists of an IKE header followed by an
1491 Encrypted payload that contains no payloads). If a cryptographically
1492 protected (fresh, i.e., not retransmitted) message has been received
1493 from the other side recently, unprotected Notify messages MAY be
1494 ignored. Implementations MUST limit the rate at which they take
1495 actions based on unprotected messages.
1497 The number of retries and length of timeouts are not covered in this
1498 specification because they do not affect interoperability. It is
1499 suggested that messages be retransmitted at least a dozen times over
1500 a period of at least several minutes before giving up on an SA, but
1501 different environments may require different rules. To be a good
1502 network citizen, retransmission times MUST increase exponentially to
1503 avoid flooding the network and making an existing congestion
1504 situation worse. If there has only been outgoing traffic on all of
1505 the SAs associated with an IKE SA, it is essential to confirm
1506 liveness of the other endpoint to avoid black holes. If no
1507 cryptographically protected messages have been received on an IKE SA
1508 or any of its Child SAs recently, the system needs to perform a
1509 liveness check in order to prevent sending messages to a dead peer.
1510 (This is sometimes called "dead peer detection" or "DPD", although it
1514 Kaufman, et al. Standards Track [Page 27]
1516 RFC 5996 IKEv2bis September 2010
1519 is really detecting live peers, not dead ones.) Receipt of a fresh
1520 cryptographically protected message on an IKE SA or any of its Child
1521 SAs ensures liveness of the IKE SA and all of its Child SAs. Note
1522 that this places requirements on the failure modes of an IKE
1523 endpoint. An implementation needs to stop sending over any SA if
1524 some failure prevents it from receiving on all of the associated SAs.
1525 If a system creates Child SAs that can fail independently from one
1526 another without the associated IKE SA being able to send a delete
1527 message, then the system MUST negotiate such Child SAs using separate
1530 There is a DoS attack on the initiator of an IKE SA that can be
1531 avoided if the initiator takes the proper care. Since the first two
1532 messages of an SA setup are not cryptographically protected, an
1533 attacker could respond to the initiator's message before the genuine
1534 responder and poison the connection setup attempt. To prevent this,
1535 the initiator MAY be willing to accept multiple responses to its
1536 first message, treat each as potentially legitimate, respond to it,
1537 and then discard all the invalid half-open connections when it
1538 receives a valid cryptographically protected response to any one of
1539 its requests. Once a cryptographically valid response is received,
1540 all subsequent responses should be ignored whether or not they are
1541 cryptographically valid.
1543 Note that with these rules, there is no reason to negotiate and agree
1544 upon an SA lifetime. If IKE presumes the partner is dead, based on
1545 repeated lack of acknowledgement to an IKE message, then the IKE SA
1546 and all Child SAs set up through that IKE SA are deleted.
1548 An IKE endpoint may at any time delete inactive Child SAs to recover
1549 resources used to hold their state. If an IKE endpoint chooses to
1550 delete Child SAs, it MUST send Delete payloads to the other end
1551 notifying it of the deletion. It MAY similarly time out the IKE SA.
1552 Closing the IKE SA implicitly closes all associated Child SAs. In
1553 this case, an IKE endpoint SHOULD send a Delete payload indicating
1554 that it has closed the IKE SA unless the other endpoint is no longer
1557 2.5. Version Numbers and Forward Compatibility
1559 This document describes version 2.0 of IKE, meaning the major version
1560 number is 2 and the minor version number is 0. This document is a
1561 replacement for [IKEV2]. It is likely that some implementations will
1562 want to support version 1.0 and version 2.0, and in the future, other
1570 Kaufman, et al. Standards Track [Page 28]
1572 RFC 5996 IKEv2bis September 2010
1575 The major version number should be incremented only if the packet
1576 formats or required actions have changed so dramatically that an
1577 older version node would not be able to interoperate with a newer
1578 version node if it simply ignored the fields it did not understand
1579 and took the actions specified in the older specification. The minor
1580 version number indicates new capabilities, and MUST be ignored by a
1581 node with a smaller minor version number, but used for informational
1582 purposes by the node with the larger minor version number. For
1583 example, it might indicate the ability to process a newly defined
1584 Notify message type. The node with the larger minor version number
1585 would simply note that its correspondent would not be able to
1586 understand that message and therefore would not send it.
1588 If an endpoint receives a message with a higher major version number,
1589 it MUST drop the message and SHOULD send an unauthenticated Notify
1590 message of type INVALID_MAJOR_VERSION containing the highest
1591 (closest) version number it supports. If an endpoint supports major
1592 version n, and major version m, it MUST support all versions between
1593 n and m. If it receives a message with a major version that it
1594 supports, it MUST respond with that version number. In order to
1595 prevent two nodes from being tricked into corresponding with a lower
1596 major version number than the maximum that they both support, IKE has
1597 a flag that indicates that the node is capable of speaking a higher
1598 major version number.
1600 Thus, the major version number in the IKE header indicates the
1601 version number of the message, not the highest version number that
1602 the transmitter supports. If the initiator is capable of speaking
1603 versions n, n+1, and n+2, and the responder is capable of speaking
1604 versions n and n+1, then they will negotiate speaking n+1, where the
1605 initiator will set a flag indicating its ability to speak a higher
1606 version. If they mistakenly (perhaps through an active attacker
1607 sending error messages) negotiate to version n, then both will notice
1608 that the other side can support a higher version number, and they
1609 MUST break the connection and reconnect using version n+1.
1611 Note that IKEv1 does not follow these rules, because there is no way
1612 in v1 of noting that you are capable of speaking a higher version
1613 number. So an active attacker can trick two v2-capable nodes into
1614 speaking v1. When a v2-capable node negotiates down to v1, it should
1615 note that fact in its logs.
1617 Also, for forward compatibility, all fields marked RESERVED MUST be
1618 set to zero by an implementation running version 2.0, and their
1619 content MUST be ignored by an implementation running version 2.0 ("Be
1620 conservative in what you send and liberal in what you receive" [IP]).
1621 In this way, future versions of the protocol can use those fields in
1622 a way that is guaranteed to be ignored by implementations that do not
1626 Kaufman, et al. Standards Track [Page 29]
1628 RFC 5996 IKEv2bis September 2010
1631 understand them. Similarly, payload types that are not defined are
1632 reserved for future use; implementations of a version where they are
1633 undefined MUST skip over those payloads and ignore their contents.
1635 IKEv2 adds a "critical" flag to each payload header for further
1636 flexibility for forward compatibility. If the critical flag is set
1637 and the payload type is unrecognized, the message MUST be rejected
1638 and the response to the IKE request containing that payload MUST
1639 include a Notify payload UNSUPPORTED_CRITICAL_PAYLOAD, indicating an
1640 unsupported critical payload was included. In that Notify payload,
1641 the notification data contains the one-octet payload type. If the
1642 critical flag is not set and the payload type is unsupported, that
1643 payload MUST be ignored. Payloads sent in IKE response messages MUST
1644 NOT have the critical flag set. Note that the critical flag applies
1645 only to the payload type, not the contents. If the payload type is
1646 recognized, but the payload contains something that is not (such as
1647 an unknown transform inside an SA payload, or an unknown Notify
1648 Message Type inside a Notify payload), the critical flag is ignored.
1650 Although new payload types may be added in the future and may appear
1651 interleaved with the fields defined in this specification,
1652 implementations SHOULD send the payloads defined in this
1653 specification in the order shown in the figures in Sections 1 and 2;
1654 implementations MUST NOT reject as invalid a message with those
1655 payloads in any other order.
1657 2.6. IKE SA SPIs and Cookies
1659 The initial two eight-octet fields in the header, called the "IKE
1660 SPIs", are used as a connection identifier at the beginning of IKE
1661 packets. Each endpoint chooses one of the two SPIs and MUST choose
1662 them so as to be unique identifiers of an IKE SA. An SPI value of
1663 zero is special: it indicates that the remote SPI value is not yet
1664 known by the sender.
1666 Incoming IKE packets are mapped to an IKE SA only using the packet's
1667 SPI, not using (for example) the source IP address of the packet.
1669 Unlike ESP and AH where only the recipient's SPI appears in the
1670 header of a message, in IKE the sender's SPI is also sent in every
1671 message. Since the SPI chosen by the original initiator of the IKE
1672 SA is always sent first, an endpoint with multiple IKE SAs open that
1673 wants to find the appropriate IKE SA using the SPI it assigned must
1674 look at the Initiator flag in the header to determine whether it
1675 assigned the first or the second eight octets.
1682 Kaufman, et al. Standards Track [Page 30]
1684 RFC 5996 IKEv2bis September 2010
1687 In the first message of an initial IKE exchange, the initiator will
1688 not know the responder's SPI value and will therefore set that field
1689 to zero. When the IKE_SA_INIT exchange does not result in the
1690 creation of an IKE SA due to INVALID_KE_PAYLOAD, NO_PROPOSAL_CHOSEN,
1691 or COOKIE (see Section 2.6), the responder's SPI will be zero also in
1692 the response message. However, if the responder sends a non-zero
1693 responder SPI, the initiator should not reject the response for only
1696 Two expected attacks against IKE are state and CPU exhaustion, where
1697 the target is flooded with session initiation requests from forged IP
1698 addresses. These attacks can be made less effective if a responder
1699 uses minimal CPU and commits no state to an SA until it knows the
1700 initiator can receive packets at the address from which it claims to
1703 When a responder detects a large number of half-open IKE SAs, it
1704 SHOULD reply to IKE_SA_INIT requests with a response containing the
1705 COOKIE notification. The data associated with this notification MUST
1706 be between 1 and 64 octets in length (inclusive), and its generation
1707 is described later in this section. If the IKE_SA_INIT response
1708 includes the COOKIE notification, the initiator MUST then retry the
1709 IKE_SA_INIT request, and include the COOKIE notification containing
1710 the received data as the first payload, and all other payloads
1711 unchanged. The initial exchange will then be as follows:
1714 -------------------------------------------------------------------
1715 HDR(A,0), SAi1, KEi, Ni -->
1716 <-- HDR(A,0), N(COOKIE)
1717 HDR(A,0), N(COOKIE), SAi1,
1719 <-- HDR(A,B), SAr1, KEr,
1721 HDR(A,B), SK {IDi, [CERT,]
1722 [CERTREQ,] [IDr,] AUTH,
1724 <-- HDR(A,B), SK {IDr, [CERT,]
1725 AUTH, SAr2, TSi, TSr}
1727 The first two messages do not affect any initiator or responder state
1728 except for communicating the cookie. In particular, the message
1729 sequence numbers in the first four messages will all be zero and the
1730 message sequence numbers in the last two messages will be one. 'A'
1731 is the SPI assigned by the initiator, while 'B' is the SPI assigned
1738 Kaufman, et al. Standards Track [Page 31]
1740 RFC 5996 IKEv2bis September 2010
1743 An IKE implementation can implement its responder cookie generation
1744 in such a way as to not require any saved state to recognize its
1745 valid cookie when the second IKE_SA_INIT message arrives. The exact
1746 algorithms and syntax used to generate cookies do not affect
1747 interoperability and hence are not specified here. The following is
1748 an example of how an endpoint could use cookies to implement limited
1751 A good way to do this is to set the responder cookie to be:
1753 Cookie = <VersionIDofSecret> | Hash(Ni | IPi | SPIi | <secret>)
1755 where <secret> is a randomly generated secret known only to the
1756 responder and periodically changed and | indicates concatenation.
1757 <VersionIDofSecret> should be changed whenever <secret> is
1758 regenerated. The cookie can be recomputed when the IKE_SA_INIT
1759 arrives the second time and compared to the cookie in the received
1760 message. If it matches, the responder knows that the cookie was
1761 generated since the last change to <secret> and that IPi must be the
1762 same as the source address it saw the first time. Incorporating SPIi
1763 into the calculation ensures that if multiple IKE SAs are being set
1764 up in parallel they will all get different cookies (assuming the
1765 initiator chooses unique SPIi's). Incorporating Ni in the hash
1766 ensures that an attacker who sees only message 2 can't successfully
1767 forge a message 3. Also, incorporating SPIi in the hash prevents an
1768 attacker from fetching one cookie from the other end, and then
1769 initiating many IKE_SA_INIT exchanges all with different initiator
1770 SPIs (and perhaps port numbers) so that the responder thinks that
1771 there are a lot of machines behind one NAT box that are all trying to
1774 If a new value for <secret> is chosen while there are connections in
1775 the process of being initialized, an IKE_SA_INIT might be returned
1776 with other than the current <VersionIDofSecret>. The responder in
1777 that case MAY reject the message by sending another response with a
1778 new cookie or it MAY keep the old value of <secret> around for a
1779 short time and accept cookies computed from either one. The
1780 responder should not accept cookies indefinitely after <secret> is
1781 changed, since that would defeat part of the DoS protection. The
1782 responder should change the value of <secret> frequently, especially
1785 When one party receives an IKE_SA_INIT request containing a cookie
1786 whose contents do not match the value expected, that party MUST
1787 ignore the cookie and process the message as if no cookie had been
1788 included; usually this means sending a response containing a new
1789 cookie. The initiator should limit the number of cookie exchanges it
1790 tries before giving up, possibly using exponential back-off. An
1794 Kaufman, et al. Standards Track [Page 32]
1796 RFC 5996 IKEv2bis September 2010
1799 attacker can forge multiple cookie responses to the initiator's
1800 IKE_SA_INIT message, and each of those forged cookie replies will
1801 cause two packets to be sent: one packet from the initiator to the
1802 responder (which will reject those cookies), and one response from
1803 responder to initiator that includes the correct cookie.
1805 A note on terminology: the term "cookies" originates with Karn and
1806 Simpson [PHOTURIS] in Photuris, an early proposal for key management
1807 with IPsec, and it has persisted. The Internet Security Association
1808 and Key Management Protocol (ISAKMP) [ISAKMP] fixed message header
1809 includes two eight-octet fields called "cookies", and that syntax is
1810 used by both IKEv1 and IKEv2, although in IKEv2 they are referred to
1811 as the "IKE SPI" and there is a new separate field in a Notify
1812 payload holding the cookie.
1814 2.6.1. Interaction of COOKIE and INVALID_KE_PAYLOAD
1816 There are two common reasons why the initiator may have to retry the
1817 IKE_SA_INIT exchange: the responder requests a cookie or wants a
1818 different Diffie-Hellman group than was included in the KEi payload.
1819 If the initiator receives a cookie from the responder, the initiator
1820 needs to decide whether or not to include the cookie in only the next
1821 retry of the IKE_SA_INIT request, or in all subsequent retries as
1824 If the initiator includes the cookie only in the next retry, one
1825 additional round trip may be needed in some cases. An additional
1826 round trip is needed also if the initiator includes the cookie in all
1827 retries, but the responder does not support this. For instance, if
1828 the responder includes the KEi payloads in cookie calculation, it
1829 will reject the request by sending a new cookie.
1831 If both peers support including the cookie in all retries, a slightly
1832 shorter exchange can happen.
1835 -----------------------------------------------------------
1836 HDR(A,0), SAi1, KEi, Ni -->
1837 <-- HDR(A,0), N(COOKIE)
1838 HDR(A,0), N(COOKIE), SAi1, KEi, Ni -->
1839 <-- HDR(A,0), N(INVALID_KE_PAYLOAD)
1840 HDR(A,0), N(COOKIE), SAi1, KEi', Ni -->
1841 <-- HDR(A,B), SAr1, KEr, Nr
1843 Implementations SHOULD support this shorter exchange, but MUST NOT
1844 fail if other implementations do not support this shorter exchange.
1850 Kaufman, et al. Standards Track [Page 33]
1852 RFC 5996 IKEv2bis September 2010
1855 2.7. Cryptographic Algorithm Negotiation
1857 The payload type known as "SA" indicates a proposal for a set of
1858 choices of IPsec protocols (IKE, ESP, or AH) for the SA as well as
1859 cryptographic algorithms associated with each protocol.
1861 An SA payload consists of one or more proposals. Each proposal
1862 includes one protocol. Each protocol contains one or more transforms
1863 -- each specifying a cryptographic algorithm. Each transform
1864 contains zero or more attributes (attributes are needed only if the
1865 Transform ID does not completely specify the cryptographic
1868 This hierarchical structure was designed to efficiently encode
1869 proposals for cryptographic suites when the number of supported
1870 suites is large because multiple values are acceptable for multiple
1871 transforms. The responder MUST choose a single suite, which may be
1872 any subset of the SA proposal following the rules below.
1874 Each proposal contains one protocol. If a proposal is accepted, the
1875 SA response MUST contain the same protocol. The responder MUST
1876 accept a single proposal or reject them all and return an error. The
1877 error is given in a notification of type NO_PROPOSAL_CHOSEN.
1879 Each IPsec protocol proposal contains one or more transforms. Each
1880 transform contains a Transform Type. The accepted cryptographic
1881 suite MUST contain exactly one transform of each type included in the
1882 proposal. For example: if an ESP proposal includes transforms
1883 ENCR_3DES, ENCR_AES w/keysize 128, ENCR_AES w/keysize 256,
1884 AUTH_HMAC_MD5, and AUTH_HMAC_SHA, the accepted suite MUST contain one
1885 of the ENCR_ transforms and one of the AUTH_ transforms. Thus, six
1886 combinations are acceptable.
1888 If an initiator proposes both normal ciphers with integrity
1889 protection as well as combined-mode ciphers, then two proposals are
1890 needed. One of the proposals includes the normal ciphers with the
1891 integrity algorithms for them, and the other proposal includes all
1892 the combined-mode ciphers without the integrity algorithms (because
1893 combined-mode ciphers are not allowed to have any integrity algorithm
1898 IKE, ESP, and AH Security Associations use secret keys that should be
1899 used only for a limited amount of time and to protect a limited
1900 amount of data. This limits the lifetime of the entire Security
1901 Association. When the lifetime of a Security Association expires,
1902 the Security Association MUST NOT be used. If there is demand, new
1906 Kaufman, et al. Standards Track [Page 34]
1908 RFC 5996 IKEv2bis September 2010
1911 Security Associations MAY be established. Reestablishment of
1912 Security Associations to take the place of ones that expire is
1913 referred to as "rekeying".
1915 To allow for minimal IPsec implementations, the ability to rekey SAs
1916 without restarting the entire IKE SA is optional. An implementation
1917 MAY refuse all CREATE_CHILD_SA requests within an IKE SA. If an SA
1918 has expired or is about to expire and rekeying attempts using the
1919 mechanisms described here fail, an implementation MUST close the IKE
1920 SA and any associated Child SAs and then MAY start new ones.
1921 Implementations may wish to support in-place rekeying of SAs, since
1922 doing so offers better performance and is likely to reduce the number
1923 of packets lost during the transition.
1925 To rekey a Child SA within an existing IKE SA, create a new,
1926 equivalent SA (see Section 2.17 below), and when the new one is
1927 established, delete the old one. Note that, when rekeying, the new
1928 Child SA SHOULD NOT have different Traffic Selectors and algorithms
1931 To rekey an IKE SA, establish a new equivalent IKE SA (see
1932 Section 2.18 below) with the peer to whom the old IKE SA is shared
1933 using a CREATE_CHILD_SA within the existing IKE SA. An IKE SA so
1934 created inherits all of the original IKE SA's Child SAs, and the new
1935 IKE SA is used for all control messages needed to maintain those
1936 Child SAs. After the new equivalent IKE SA is created, the initiator
1937 deletes the old IKE SA, and the Delete payload to delete itself MUST
1938 be the last request sent over the old IKE SA.
1940 SAs should be rekeyed proactively, i.e., the new SA should be
1941 established before the old one expires and becomes unusable. Enough
1942 time should elapse between the time the new SA is established and the
1943 old one becomes unusable so that traffic can be switched over to the
1946 A difference between IKEv1 and IKEv2 is that in IKEv1 SA lifetimes
1947 were negotiated. In IKEv2, each end of the SA is responsible for
1948 enforcing its own lifetime policy on the SA and rekeying the SA when
1949 necessary. If the two ends have different lifetime policies, the end
1950 with the shorter lifetime will end up always being the one to request
1951 the rekeying. If an SA has been inactive for a long time and if an
1952 endpoint would not initiate the SA in the absence of traffic, the
1953 endpoint MAY choose to close the SA instead of rekeying it when its
1954 lifetime expires. It can also do so if there has been no traffic
1955 since the last time the SA was rekeyed.
1962 Kaufman, et al. Standards Track [Page 35]
1964 RFC 5996 IKEv2bis September 2010
1967 Note that IKEv2 deliberately allows parallel SAs with the same
1968 Traffic Selectors between common endpoints. One of the purposes of
1969 this is to support traffic quality of service (QoS) differences among
1970 the SAs (see [DIFFSERVFIELD], [DIFFSERVARCH], and Section 4.1 of
1971 [DIFFTUNNEL]). Hence unlike IKEv1, the combination of the endpoints
1972 and the Traffic Selectors may not uniquely identify an SA between
1973 those endpoints, so the IKEv1 rekeying heuristic of deleting SAs on
1974 the basis of duplicate Traffic Selectors SHOULD NOT be used.
1976 There are timing windows -- particularly in the presence of lost
1977 packets -- where endpoints may not agree on the state of an SA. The
1978 responder to a CREATE_CHILD_SA MUST be prepared to accept messages on
1979 an SA before sending its response to the creation request, so there
1980 is no ambiguity for the initiator. The initiator MAY begin sending
1981 on an SA as soon as it processes the response. The initiator,
1982 however, cannot receive on a newly created SA until it receives and
1983 processes the response to its CREATE_CHILD_SA request. How, then, is
1984 the responder to know when it is OK to send on the newly created SA?
1986 From a technical correctness and interoperability perspective, the
1987 responder MAY begin sending on an SA as soon as it sends its response
1988 to the CREATE_CHILD_SA request. In some situations, however, this
1989 could result in packets unnecessarily being dropped, so an
1990 implementation MAY defer such sending.
1992 The responder can be assured that the initiator is prepared to
1993 receive messages on an SA if either (1) it has received a
1994 cryptographically valid message on the other half of the SA pair, or
1995 (2) the new SA rekeys an existing SA and it receives an IKE request
1996 to close the replaced SA. When rekeying an SA, the responder
1997 continues to send traffic on the old SA until one of those events
1998 occurs. When establishing a new SA, the responder MAY defer sending
1999 messages on a new SA until either it receives one or a timeout has
2000 occurred. If an initiator receives a message on an SA for which it
2001 has not received a response to its CREATE_CHILD_SA request, it
2002 interprets that as a likely packet loss and retransmits the
2003 CREATE_CHILD_SA request. An initiator MAY send a dummy ESP message
2004 on a newly created ESP SA if it has no messages queued in order to
2005 assure the responder that the initiator is ready to receive messages.
2007 2.8.1. Simultaneous Child SA Rekeying
2009 If the two ends have the same lifetime policies, it is possible that
2010 both will initiate a rekeying at the same time (which will result in
2011 redundant SAs). To reduce the probability of this happening, the
2012 timing of rekeying requests SHOULD be jittered (delayed by a random
2013 amount of time after the need for rekeying is noticed).
2018 Kaufman, et al. Standards Track [Page 36]
2020 RFC 5996 IKEv2bis September 2010
2023 This form of rekeying may temporarily result in multiple similar SAs
2024 between the same pairs of nodes. When there are two SAs eligible to
2025 receive packets, a node MUST accept incoming packets through either
2026 SA. If redundant SAs are created though such a collision, the SA
2027 created with the lowest of the four nonces used in the two exchanges
2028 SHOULD be closed by the endpoint that created it. "Lowest" means an
2029 octet-by-octet comparison (instead of, for instance, comparing the
2030 nonces as large integers). In other words, start by comparing the
2031 first octet; if they're equal, move to the next octet, and so on. If
2032 you reach the end of one nonce, that nonce is the lower one. The
2033 node that initiated the surviving rekeyed SA should delete the
2034 replaced SA after the new one is established.
2036 The following is an explanation on the impact this has on
2037 implementations. Assume that hosts A and B have an existing Child SA
2038 pair with SPIs (SPIa1,SPIb1), and both start rekeying it at the same
2042 -------------------------------------------------------------------
2043 send req1: N(REKEY_SA,SPIa1),
2044 SA(..,SPIa2,..),Ni1,.. -->
2045 <-- send req2: N(REKEY_SA,SPIb1),
2049 At this point, A knows there is a simultaneous rekeying happening.
2050 However, it cannot yet know which of the exchanges will have the
2051 lowest nonce, so it will just note the situation and respond as
2054 send resp2: SA(..,SPIa3,..),
2058 Now B also knows that simultaneous rekeying is going on. It responds
2061 <-- send resp1: SA(..,SPIb3,..),
2066 At this point, there are three Child SA pairs between A and B (the
2067 old one and two new ones). A and B can now compare the nonces.
2068 Suppose that the lowest nonce was Nr1 in message resp2; in this case,
2069 B (the sender of req2) deletes the redundant new SA, and A (the node
2070 that initiated the surviving rekeyed SA), deletes the old one.
2074 Kaufman, et al. Standards Track [Page 37]
2076 RFC 5996 IKEv2bis September 2010
2079 send req3: D(SPIa1) -->
2080 <-- send req4: D(SPIb2)
2082 <-- send resp3: D(SPIb1)
2084 send resp4: D(SPIa3) -->
2086 The rekeying is now finished.
2088 However, there is a second possible sequence of events that can
2089 happen if some packets are lost in the network, resulting in
2090 retransmissions. The rekeying begins as usual, but A's first packet
2094 -------------------------------------------------------------------
2095 send req1: N(REKEY_SA,SPIa1),
2098 <-- send req2: N(REKEY_SA,SPIb1),
2101 send resp2: SA(..,SPIa3,..),
2104 <-- send req3: D(SPIb1)
2106 send resp3: D(SPIa1) -->
2109 From B's point of view, the rekeying is now completed, and since it
2110 has not yet received A's req1, it does not even know that there was
2111 simultaneous rekeying. However, A will continue retransmitting the
2112 message, and eventually it will reach B.
2117 To B, it looks like A is trying to rekey an SA that no longer exists;
2118 thus, B responds to the request with something non-fatal such as
2121 <-- send resp1: N(CHILD_SA_NOT_FOUND)
2124 When A receives this error, it already knows there was simultaneous
2125 rekeying, so it can ignore the error message.
2130 Kaufman, et al. Standards Track [Page 38]
2132 RFC 5996 IKEv2bis September 2010
2135 2.8.2. Simultaneous IKE SA Rekeying
2137 Probably the most complex case occurs when both peers try to rekey
2138 the IKE_SA at the same time. Basically, the text in Section 2.8
2139 applies to this case as well; however, it is important to ensure that
2140 the Child SAs are inherited by the correct IKE_SA.
2142 The case where both endpoints notice the simultaneous rekeying works
2143 the same way as with Child SAs. After the CREATE_CHILD_SA exchanges,
2144 three IKE SAs exist between A and B: the old IKE SA and two new IKE
2145 SAs. The new IKE SA containing the lowest nonce SHOULD be deleted by
2146 the node that created it, and the other surviving new IKE SA MUST
2147 inherit all the Child SAs.
2149 In addition to normal simultaneous rekeying cases, there is a special
2150 case where one peer finishes its rekey before it even notices that
2151 other peer is doing a rekey. If only one peer detects a simultaneous
2152 rekey, redundant SAs are not created. In this case, when the peer
2153 that did not notice the simultaneous rekey gets the request to rekey
2154 the IKE SA that it has already successfully rekeyed, it SHOULD return
2155 TEMPORARY_FAILURE because it is an IKE SA that it is currently trying
2156 to close (whether or not it has already sent the delete notification
2157 for the SA). If the peer that did notice the simultaneous rekey gets
2158 the delete request from the other peer for the old IKE SA, it knows
2159 that the other peer did not detect the simultaneous rekey, and the
2160 first peer can forget its own rekey attempt.
2163 -------------------------------------------------------------------
2165 SA(..,SPIa1,..),Ni1,.. -->
2166 <-- send req2: SA(..,SPIb1,..),Ni2,..
2168 <-- send resp1: SA(..,SPIb2,..),Nr2,..
2173 At this point, host B sees a request to close the IKE_SA. There's
2174 not much more to do than to reply as usual. However, at this point
2175 host B should stop retransmitting req2, since once host A receives
2176 resp3, it will delete all the state associated with the old IKE_SA
2177 and will not be able to reply to it.
2181 The TEMPORARY_FAILURE notification was not included in RFC 4306, and
2182 support of the TEMPORARY_FAILURE notification is not negotiated.
2186 Kaufman, et al. Standards Track [Page 39]
2188 RFC 5996 IKEv2bis September 2010
2191 Thus, older peers that implement RFC 4306 but not this document may
2192 receive these notifications. In that case, they will treat it the
2193 same as any other unknown error notification, and will stop the
2194 exchange. Because the other peer has already rekeyed the exchange,
2195 doing so does not have any ill effects.
2197 2.8.3. Rekeying the IKE SA versus Reauthentication
2199 Rekeying the IKE SA and reauthentication are different concepts in
2200 IKEv2. Rekeying the IKE SA establishes new keys for the IKE SA and
2201 resets the Message ID counters, but it does not authenticate the
2202 parties again (no AUTH or EAP payloads are involved).
2204 Although rekeying the IKE SA may be important in some environments,
2205 reauthentication (the verification that the parties still have access
2206 to the long-term credentials) is often more important.
2208 IKEv2 does not have any special support for reauthentication.
2209 Reauthentication is done by creating a new IKE SA from scratch (using
2210 IKE_SA_INIT/IKE_AUTH exchanges, without any REKEY_SA Notify
2211 payloads), creating new Child SAs within the new IKE SA (without
2212 REKEY_SA Notify payloads), and finally deleting the old IKE SA (which
2213 deletes the old Child SAs as well).
2215 This means that reauthentication also establishes new keys for the
2216 IKE SA and Child SAs. Therefore, while rekeying can be performed
2217 more often than reauthentication, the situation where "authentication
2218 lifetime" is shorter than "key lifetime" does not make sense.
2220 While creation of a new IKE SA can be initiated by either party
2221 (initiator or responder in the original IKE SA), the use of EAP
2222 and/or Configuration payloads means in practice that reauthentication
2223 has to be initiated by the same party as the original IKE SA. IKEv2
2224 does not currently allow the responder to request reauthentication in
2225 this case; however, there are extensions that add this functionality
2228 2.9. Traffic Selector Negotiation
2230 When an RFC4301-compliant IPsec subsystem receives an IP packet that
2231 matches a "protect" selector in its Security Policy Database (SPD),
2232 the subsystem protects that packet with IPsec. When no SA exists
2233 yet, it is the task of IKE to create it. Maintenance of a system's
2234 SPD is outside the scope of IKE, although some implementations might
2235 update their SPD in connection with the running of IKE (for an
2236 example scenario, see Section 1.1.3).
2242 Kaufman, et al. Standards Track [Page 40]
2244 RFC 5996 IKEv2bis September 2010
2247 Traffic Selector (TS) payloads allow endpoints to communicate some of
2248 the information from their SPD to their peers. These must be
2249 communicated to IKE from the SPD (for example, the PF_KEY API [PFKEY]
2250 uses the SADB_ACQUIRE message). TS payloads specify the selection
2251 criteria for packets that will be forwarded over the newly set up SA.
2252 This can serve as a consistency check in some scenarios to assure
2253 that the SPDs are consistent. In others, it guides the dynamic
2256 Two TS payloads appear in each of the messages in the exchange that
2257 creates a Child SA pair. Each TS payload contains one or more
2258 Traffic Selectors. Each Traffic Selector consists of an address
2259 range (IPv4 or IPv6), a port range, and an IP protocol ID.
2261 The first of the two TS payloads is known as TSi (Traffic Selector-
2262 initiator). The second is known as TSr (Traffic Selector-responder).
2263 TSi specifies the source address of traffic forwarded from (or the
2264 destination address of traffic forwarded to) the initiator of the
2265 Child SA pair. TSr specifies the destination address of the traffic
2266 forwarded to (or the source address of the traffic forwarded from)
2267 the responder of the Child SA pair. For example, if the original
2268 initiator requests the creation of a Child SA pair, and wishes to
2269 tunnel all traffic from subnet 198.51.100.* on the initiator's side
2270 to subnet 192.0.2.* on the responder's side, the initiator would
2271 include a single Traffic Selector in each TS payload. TSi would
2272 specify the address range (198.51.100.0 - 198.51.100.255) and TSr
2273 would specify the address range (192.0.2.0 - 192.0.2.255). Assuming
2274 that proposal was acceptable to the responder, it would send
2275 identical TS payloads back.
2277 IKEv2 allows the responder to choose a subset of the traffic proposed
2278 by the initiator. This could happen when the configurations of the
2279 two endpoints are being updated but only one end has received the new
2280 information. Since the two endpoints may be configured by different
2281 people, the incompatibility may persist for an extended period even
2282 in the absence of errors. It also allows for intentionally different
2283 configurations, as when one end is configured to tunnel all addresses
2284 and depends on the other end to have the up-to-date list.
2286 When the responder chooses a subset of the traffic proposed by the
2287 initiator, it narrows the Traffic Selectors to some subset of the
2288 initiator's proposal (provided the set does not become the null set).
2289 If the type of Traffic Selector proposed is unknown, the responder
2290 ignores that Traffic Selector, so that the unknown type is not
2291 returned in the narrowed set.
2298 Kaufman, et al. Standards Track [Page 41]
2300 RFC 5996 IKEv2bis September 2010
2303 To enable the responder to choose the appropriate range in this case,
2304 if the initiator has requested the SA due to a data packet, the
2305 initiator SHOULD include as the first Traffic Selector in each of TSi
2306 and TSr a very specific Traffic Selector including the addresses in
2307 the packet triggering the request. In the example, the initiator
2308 would include in TSi two Traffic Selectors: the first containing the
2309 address range (198.51.100.43 - 198.51.100.43) and the source port and
2310 IP protocol from the packet and the second containing (198.51.100.0 -
2311 198.51.100.255) with all ports and IP protocols. The initiator would
2312 similarly include two Traffic Selectors in TSr. If the initiator
2313 creates the Child SA pair not in response to an arriving packet, but
2314 rather, say, upon startup, then there may be no specific addresses
2315 the initiator prefers for the initial tunnel over any other. In that
2316 case, the first values in TSi and TSr can be ranges rather than
2319 The responder performs the narrowing as follows:
2321 o If the responder's policy does not allow it to accept any part of
2322 the proposed Traffic Selectors, it responds with a TS_UNACCEPTABLE
2325 o If the responder's policy allows the entire set of traffic covered
2326 by TSi and TSr, no narrowing is necessary, and the responder can
2327 return the same TSi and TSr values.
2329 o If the responder's policy allows it to accept the first selector
2330 of TSi and TSr, then the responder MUST narrow the Traffic
2331 Selectors to a subset that includes the initiator's first choices.
2332 In this example above, the responder might respond with TSi being
2333 (198.51.100.43 - 198.51.100.43) with all ports and IP protocols.
2335 o If the responder's policy does not allow it to accept the first
2336 selector of TSi and TSr, the responder narrows to an acceptable
2337 subset of TSi and TSr.
2339 When narrowing is done, there may be several subsets that are
2340 acceptable but their union is not. In this case, the responder
2341 arbitrarily chooses one of them, and MAY include an
2342 ADDITIONAL_TS_POSSIBLE notification in the response. The
2343 ADDITIONAL_TS_POSSIBLE notification asserts that the responder
2344 narrowed the proposed Traffic Selectors but that other Traffic
2345 Selectors would also have been acceptable, though only in a separate
2346 SA. There is no data associated with this Notify type. This case
2347 will occur only when the initiator and responder are configured
2348 differently from one another. If the initiator and responder agree
2349 on the granularity of tunnels, the initiator will never request a
2350 tunnel wider than the responder will accept.
2354 Kaufman, et al. Standards Track [Page 42]
2356 RFC 5996 IKEv2bis September 2010
2359 It is possible for the responder's policy to contain multiple smaller
2360 ranges, all encompassed by the initiator's Traffic Selector, and with
2361 the responder's policy being that each of those ranges should be sent
2362 over a different SA. Continuing the example above, the responder
2363 might have a policy of being willing to tunnel those addresses to and
2364 from the initiator, but might require that each address pair be on a
2365 separately negotiated Child SA. If the initiator didn't generate its
2366 request based on the packet, but (for example) upon startup, there
2367 would not be the very specific first Traffic Selectors helping the
2368 responder to select the correct range. There would be no way for the
2369 responder to determine which pair of addresses should be included in
2370 this tunnel, and it would have to make a guess or reject the request
2371 with a SINGLE_PAIR_REQUIRED Notify message.
2373 The SINGLE_PAIR_REQUIRED error indicates that a CREATE_CHILD_SA
2374 request is unacceptable because its sender is only willing to accept
2375 Traffic Selectors specifying a single pair of addresses. The
2376 requestor is expected to respond by requesting an SA for only the
2377 specific traffic it is trying to forward.
2379 Few implementations will have policies that require separate SAs for
2380 each address pair. Because of this, if only some parts of the TSi
2381 and TSr proposed by the initiator are acceptable to the responder,
2382 responders SHOULD narrow the selectors to an acceptable subset rather
2383 than use SINGLE_PAIR_REQUIRED.
2385 2.9.1. Traffic Selectors Violating Own Policy
2387 When creating a new SA, the initiator needs to avoid proposing
2388 Traffic Selectors that violate its own policy. If this rule is not
2389 followed, valid traffic may be dropped. If you use decorrelated
2390 policies from [IPSECARCH], this kind of policy violations cannot
2393 This is best illustrated by an example. Suppose that host A has a
2394 policy whose effect is that traffic to 198.51.100.66 is sent via host
2395 B encrypted using AES, and traffic to all other hosts in
2396 198.51.100.0/24 is also sent via B, but must use 3DES. Suppose also
2397 that host B accepts any combination of AES and 3DES.
2399 If host A now proposes an SA that uses 3DES, and includes TSr
2400 containing (198.51.100.0-198.51.100.255), this will be accepted by
2401 host B. Now, host B can also use this SA to send traffic from
2402 198.51.100.66, but those packets will be dropped by A since it
2403 requires the use of AES for this traffic. Even if host A creates a
2404 new SA only for 198.51.100.66 that uses AES, host B may freely
2405 continue to use the first SA for the traffic. In this situation,
2410 Kaufman, et al. Standards Track [Page 43]
2412 RFC 5996 IKEv2bis September 2010
2415 when proposing the SA, host A should have followed its own policy,
2416 and included a TSr containing ((198.51.100.0-
2417 198.51.100.65),(198.51.100.67-198.51.100.255)) instead.
2419 In general, if (1) the initiator makes a proposal "for traffic X
2420 (TSi/TSr), do SA", and (2) for some subset X' of X, the initiator
2421 does not actually accept traffic X' with SA, and (3) the initiator
2422 would be willing to accept traffic X' with some SA' (!=SA), valid
2423 traffic can be unnecessarily dropped since the responder can apply
2424 either SA or SA' to traffic X'.
2428 The IKE_SA_INIT messages each contain a nonce. These nonces are used
2429 as inputs to cryptographic functions. The CREATE_CHILD_SA request
2430 and the CREATE_CHILD_SA response also contain nonces. These nonces
2431 are used to add freshness to the key derivation technique used to
2432 obtain keys for Child SA, and to ensure creation of strong
2433 pseudorandom bits from the Diffie-Hellman key. Nonces used in IKEv2
2434 MUST be randomly chosen, MUST be at least 128 bits in size, and MUST
2435 be at least half the key size of the negotiated pseudorandom function
2436 (PRF). However, the initiator chooses the nonce before the outcome
2437 of the negotiation is known. Because of that, the nonce has to be
2438 long enough for all the PRFs being proposed. If the same random
2439 number source is used for both keys and nonces, care must be taken to
2440 ensure that the latter use does not compromise the former.
2442 2.11. Address and Port Agility
2444 IKE runs over UDP ports 500 and 4500, and implicitly sets up ESP and
2445 AH associations for the same IP addresses over which it runs. The IP
2446 addresses and ports in the outer header are, however, not themselves
2447 cryptographically protected, and IKE is designed to work even through
2448 Network Address Translation (NAT) boxes. An implementation MUST
2449 accept incoming requests even if the source port is not 500 or 4500,
2450 and MUST respond to the address and port from which the request was
2451 received. It MUST specify the address and port at which the request
2452 was received as the source address and port in the response. IKE
2453 functions identically over IPv4 or IPv6.
2455 2.12. Reuse of Diffie-Hellman Exponentials
2457 IKE generates keying material using an ephemeral Diffie-Hellman
2458 exchange in order to gain the property of "perfect forward secrecy".
2459 This means that once a connection is closed and its corresponding
2460 keys are forgotten, even someone who has recorded all of the data
2461 from the connection and gets access to all of the long-term keys of
2466 Kaufman, et al. Standards Track [Page 44]
2468 RFC 5996 IKEv2bis September 2010
2471 the two endpoints cannot reconstruct the keys used to protect the
2472 conversation without doing a brute force search of the session key
2475 Achieving perfect forward secrecy requires that when a connection is
2476 closed, each endpoint MUST forget not only the keys used by the
2477 connection but also any information that could be used to recompute
2480 Because computing Diffie-Hellman exponentials is computationally
2481 expensive, an endpoint may find it advantageous to reuse those
2482 exponentials for multiple connection setups. There are several
2483 reasonable strategies for doing this. An endpoint could choose a new
2484 exponential only periodically though this could result in less-than-
2485 perfect forward secrecy if some connection lasts for less than the
2486 lifetime of the exponential. Or it could keep track of which
2487 exponential was used for each connection and delete the information
2488 associated with the exponential only when some corresponding
2489 connection was closed. This would allow the exponential to be reused
2490 without losing perfect forward secrecy at the cost of maintaining
2493 Whether and when to reuse Diffie-Hellman exponentials are private
2494 decisions in the sense that they will not affect interoperability.
2495 An implementation that reuses exponentials MAY choose to remember the
2496 exponential used by the other endpoint on past exchanges and if one
2497 is reused to avoid the second half of the calculation. See [REUSE]
2498 for a security analysis of this practice and for additional security
2499 considerations when reusing ephemeral Diffie-Hellman keys.
2501 2.13. Generating Keying Material
2503 In the context of the IKE SA, four cryptographic algorithms are
2504 negotiated: an encryption algorithm, an integrity protection
2505 algorithm, a Diffie-Hellman group, and a pseudorandom function (PRF).
2506 The PRF is used for the construction of keying material for all of
2507 the cryptographic algorithms used in both the IKE SA and the Child
2510 We assume that each encryption algorithm and integrity protection
2511 algorithm uses a fixed-size key and that any randomly chosen value of
2512 that fixed size can serve as an appropriate key. For algorithms that
2513 accept a variable-length key, a fixed key size MUST be specified as
2514 part of the cryptographic transform negotiated (see Section 3.3.5 for
2515 the definition of the Key Length transform attribute). For
2516 algorithms for which not all values are valid keys (such as DES or
2517 3DES with key parity), the algorithm by which keys are derived from
2518 arbitrary values MUST be specified by the cryptographic transform.
2522 Kaufman, et al. Standards Track [Page 45]
2524 RFC 5996 IKEv2bis September 2010
2527 For integrity protection functions based on Hashed Message
2528 Authentication Code (HMAC), the fixed key size is the size of the
2529 output of the underlying hash function.
2531 It is assumed that PRFs accept keys of any length, but have a
2532 preferred key size. The preferred key size MUST be used as the
2533 length of SK_d, SK_pi, and SK_pr (see Section 2.14). For PRFs based
2534 on the HMAC construction, the preferred key size is equal to the
2535 length of the output of the underlying hash function. Other types of
2536 PRFs MUST specify their preferred key size.
2538 Keying material will always be derived as the output of the
2539 negotiated PRF algorithm. Since the amount of keying material needed
2540 may be greater than the size of the output of the PRF, the PRF is
2541 used iteratively. The term "prf+" describes a function that outputs
2542 a pseudorandom stream based on the inputs to a pseudorandom function
2545 In the following, | indicates concatenation. prf+ is defined as:
2547 prf+ (K,S) = T1 | T2 | T3 | T4 | ...
2550 T1 = prf (K, S | 0x01)
2551 T2 = prf (K, T1 | S | 0x02)
2552 T3 = prf (K, T2 | S | 0x03)
2553 T4 = prf (K, T3 | S | 0x04)
2556 This continues until all the material needed to compute all required
2557 keys has been output from prf+. The keys are taken from the output
2558 string without regard to boundaries (e.g., if the required keys are a
2559 256-bit Advanced Encryption Standard (AES) key and a 160-bit HMAC
2560 key, and the prf function generates 160 bits, the AES key will come
2561 from T1 and the beginning of T2, while the HMAC key will come from
2562 the rest of T2 and the beginning of T3).
2564 The constant concatenated to the end of each prf function is a single
2565 octet. The prf+ function is not defined beyond 255 times the size of
2566 the prf function output.
2568 2.14. Generating Keying Material for the IKE SA
2570 The shared keys are computed as follows. A quantity called SKEYSEED
2571 is calculated from the nonces exchanged during the IKE_SA_INIT
2572 exchange and the Diffie-Hellman shared secret established during that
2573 exchange. SKEYSEED is used to calculate seven other secrets: SK_d
2574 used for deriving new keys for the Child SAs established with this
2578 Kaufman, et al. Standards Track [Page 46]
2580 RFC 5996 IKEv2bis September 2010
2583 IKE SA; SK_ai and SK_ar used as a key to the integrity protection
2584 algorithm for authenticating the component messages of subsequent
2585 exchanges; SK_ei and SK_er used for encrypting (and of course
2586 decrypting) all subsequent exchanges; and SK_pi and SK_pr, which are
2587 used when generating an AUTH payload. The lengths of SK_d, SK_pi,
2588 and SK_pr MUST be the preferred key length of the PRF agreed upon.
2590 SKEYSEED and its derivatives are computed as follows:
2592 SKEYSEED = prf(Ni | Nr, g^ir)
2594 {SK_d | SK_ai | SK_ar | SK_ei | SK_er | SK_pi | SK_pr }
2595 = prf+ (SKEYSEED, Ni | Nr | SPIi | SPIr )
2597 (indicating that the quantities SK_d, SK_ai, SK_ar, SK_ei, SK_er,
2598 SK_pi, and SK_pr are taken in order from the generated bits of the
2599 prf+). g^ir is the shared secret from the ephemeral Diffie-Hellman
2600 exchange. g^ir is represented as a string of octets in big endian
2601 order padded with zeros if necessary to make it the length of the
2602 modulus. Ni and Nr are the nonces, stripped of any headers. For
2603 historical backward-compatibility reasons, there are two PRFs that
2604 are treated specially in this calculation. If the negotiated PRF is
2605 AES-XCBC-PRF-128 [AESXCBCPRF128] or AES-CMAC-PRF-128 [AESCMACPRF128],
2606 only the first 64 bits of Ni and the first 64 bits of Nr are used in
2607 calculating SKEYSEED, but all the bits are used for input to the prf+
2610 The two directions of traffic flow use different keys. The keys used
2611 to protect messages from the original initiator are SK_ai and SK_ei.
2612 The keys used to protect messages in the other direction are SK_ar
2615 2.15. Authentication of the IKE SA
2617 When not using extensible authentication (see Section 2.16), the
2618 peers are authenticated by having each sign (or MAC using a padded
2619 shared secret as the key, as described later in this section) a block
2620 of data. In these calculations, IDi' and IDr' are the entire ID
2621 payloads excluding the fixed header. For the responder, the octets
2622 to be signed start with the first octet of the first SPI in the
2623 header of the second message (IKE_SA_INIT response) and end with the
2624 last octet of the last payload in the second message. Appended to
2625 this (for the purposes of computing the signature) are the
2626 initiator's nonce Ni (just the value, not the payload containing it),
2627 and the value prf(SK_pr, IDr'). Note that neither the nonce Ni nor
2628 the value prf(SK_pr, IDr') are transmitted. Similarly, the initiator
2629 signs the first message (IKE_SA_INIT request), starting with the
2630 first octet of the first SPI in the header and ending with the last
2634 Kaufman, et al. Standards Track [Page 47]
2636 RFC 5996 IKEv2bis September 2010
2639 octet of the last payload. Appended to this (for purposes of
2640 computing the signature) are the responder's nonce Nr, and the value
2641 prf(SK_pi, IDi'). It is critical to the security of the exchange
2642 that each side sign the other side's nonce.
2644 The initiator's signed octets can be described as:
2646 InitiatorSignedOctets = RealMessage1 | NonceRData | MACedIDForI
2647 GenIKEHDR = [ four octets 0 if using port 4500 ] | RealIKEHDR
2648 RealIKEHDR = SPIi | SPIr | . . . | Length
2649 RealMessage1 = RealIKEHDR | RestOfMessage1
2650 NonceRPayload = PayloadHeader | NonceRData
2651 InitiatorIDPayload = PayloadHeader | RestOfInitIDPayload
2652 RestOfInitIDPayload = IDType | RESERVED | InitIDData
2653 MACedIDForI = prf(SK_pi, RestOfInitIDPayload)
2655 The responder's signed octets can be described as:
2657 ResponderSignedOctets = RealMessage2 | NonceIData | MACedIDForR
2658 GenIKEHDR = [ four octets 0 if using port 4500 ] | RealIKEHDR
2659 RealIKEHDR = SPIi | SPIr | . . . | Length
2660 RealMessage2 = RealIKEHDR | RestOfMessage2
2661 NonceIPayload = PayloadHeader | NonceIData
2662 ResponderIDPayload = PayloadHeader | RestOfRespIDPayload
2663 RestOfRespIDPayload = IDType | RESERVED | RespIDData
2664 MACedIDForR = prf(SK_pr, RestOfRespIDPayload)
2666 Note that all of the payloads are included under the signature,
2667 including any payload types not defined in this document. If the
2668 first message of the exchange is sent multiple times (such as with a
2669 responder cookie and/or a different Diffie-Hellman group), it is the
2670 latest version of the message that is signed.
2672 Optionally, messages 3 and 4 MAY include a certificate, or
2673 certificate chain providing evidence that the key used to compute a
2674 digital signature belongs to the name in the ID payload. The
2675 signature or MAC will be computed using algorithms dictated by the
2676 type of key used by the signer, and specified by the Auth Method
2677 field in the Authentication payload. There is no requirement that
2678 the initiator and responder sign with the same cryptographic
2679 algorithms. The choice of cryptographic algorithms depends on the
2680 type of key each has. In particular, the initiator may be using a
2681 shared key while the responder may have a public signature key and
2682 certificate. It will commonly be the case (but it is not required)
2683 that, if a shared secret is used for authentication, the same key is
2684 used in both directions.
2690 Kaufman, et al. Standards Track [Page 48]
2692 RFC 5996 IKEv2bis September 2010
2695 Note that it is a common but typically insecure practice to have a
2696 shared key derived solely from a user-chosen password without
2697 incorporating another source of randomness. This is typically
2698 insecure because user-chosen passwords are unlikely to have
2699 sufficient unpredictability to resist dictionary attacks and these
2700 attacks are not prevented in this authentication method.
2701 (Applications using password-based authentication for bootstrapping
2702 and IKE SA should use the authentication method in Section 2.16,
2703 which is designed to prevent off-line dictionary attacks.) The pre-
2704 shared key needs to contain as much unpredictability as the strongest
2705 key being negotiated. In the case of a pre-shared key, the AUTH
2706 value is computed as:
2709 AUTH = prf( prf(Shared Secret, "Key Pad for IKEv2"),
2710 <InitiatorSignedOctets>)
2712 AUTH = prf( prf(Shared Secret, "Key Pad for IKEv2"),
2713 <ResponderSignedOctets>)
2715 where the string "Key Pad for IKEv2" is 17 ASCII characters without
2716 null termination. The shared secret can be variable length. The pad
2717 string is added so that if the shared secret is derived from a
2718 password, the IKE implementation need not store the password in
2719 cleartext, but rather can store the value prf(Shared Secret,"Key Pad
2720 for IKEv2"), which could not be used as a password equivalent for
2721 protocols other than IKEv2. As noted above, deriving the shared
2722 secret from a password is not secure. This construction is used
2723 because it is anticipated that people will do it anyway. The
2724 management interface by which the shared secret is provided MUST
2725 accept ASCII strings of at least 64 octets and MUST NOT add a null
2726 terminator before using them as shared secrets. It MUST also accept
2727 a hex encoding of the shared secret. The management interface MAY
2728 accept other encodings if the algorithm for translating the encoding
2729 to a binary string is specified.
2731 There are two types of EAP authentication (described in
2732 Section 2.16), and each type uses different values in the AUTH
2733 computations shown above. If the EAP method is key-generating,
2734 substitute master session key (MSK) for the shared secret in the
2735 computation. For non-key-generating methods, substitute SK_pi and
2736 SK_pr, respectively, for the shared secret in the two AUTH
2746 Kaufman, et al. Standards Track [Page 49]
2748 RFC 5996 IKEv2bis September 2010
2751 2.16. Extensible Authentication Protocol Methods
2753 In addition to authentication using public key signatures and shared
2754 secrets, IKE supports authentication using methods defined in RFC
2755 3748 [EAP]. Typically, these methods are asymmetric (designed for a
2756 user authenticating to a server), and they may not be mutual. For
2757 this reason, these protocols are typically used to authenticate the
2758 initiator to the responder and MUST be used in conjunction with a
2759 public-key-signature-based authentication of the responder to the
2760 initiator. These methods are often associated with mechanisms
2761 referred to as "Legacy Authentication" mechanisms.
2763 While this document references [EAP] with the intent that new methods
2764 can be added in the future without updating this specification, some
2765 simpler variations are documented here. [EAP] defines an
2766 authentication protocol requiring a variable number of messages.
2767 Extensible Authentication is implemented in IKE as additional
2768 IKE_AUTH exchanges that MUST be completed in order to initialize the
2771 An initiator indicates a desire to use EAP by leaving out the AUTH
2772 payload from the first message in the IKE_AUTH exchange. (Note that
2773 the AUTH payload is required for non-EAP authentication, and is thus
2774 not marked as optional in the rest of this document.) By including
2775 an IDi payload but not an AUTH payload, the initiator has declared an
2776 identity but has not proven it. If the responder is willing to use
2777 an EAP method, it will place an Extensible Authentication Protocol
2778 (EAP) payload in the response of the IKE_AUTH exchange and defer
2779 sending SAr2, TSi, and TSr until initiator authentication is complete
2780 in a subsequent IKE_AUTH exchange. In the case of a minimal EAP
2781 method, the initial SA establishment will appear as follows:
2784 -------------------------------------------------------------------
2785 HDR, SAi1, KEi, Ni -->
2786 <-- HDR, SAr1, KEr, Nr, [CERTREQ]
2787 HDR, SK {IDi, [CERTREQ,]
2790 <-- HDR, SK {IDr, [CERT,] AUTH,
2793 <-- HDR, SK {EAP (success)}
2795 <-- HDR, SK {AUTH, SAr2, TSi, TSr }
2802 Kaufman, et al. Standards Track [Page 50]
2804 RFC 5996 IKEv2bis September 2010
2807 As described in Section 2.2, when EAP is used, each pair of IKE SA
2808 initial setup messages will have their message numbers incremented;
2809 the first pair of AUTH messages will have an ID of 1, the second will
2812 For EAP methods that create a shared key as a side effect of
2813 authentication, that shared key MUST be used by both the initiator
2814 and responder to generate AUTH payloads in messages 7 and 8 using the
2815 syntax for shared secrets specified in Section 2.15. The shared key
2816 from EAP is the field from the EAP specification named MSK. This
2817 shared key generated during an IKE exchange MUST NOT be used for any
2820 EAP methods that do not establish a shared key SHOULD NOT be used, as
2821 they are subject to a number of man-in-the-middle attacks [EAPMITM]
2822 if these EAP methods are used in other protocols that do not use a
2823 server-authenticated tunnel. Please see the Security Considerations
2824 section for more details. If EAP methods that do not generate a
2825 shared key are used, the AUTH payloads in messages 7 and 8 MUST be
2826 generated using SK_pi and SK_pr, respectively.
2828 The initiator of an IKE SA using EAP needs to be capable of extending
2829 the initial protocol exchange to at least ten IKE_AUTH exchanges in
2830 the event the responder sends notification messages and/or retries
2831 the authentication prompt. Once the protocol exchange defined by the
2832 chosen EAP authentication method has successfully terminated, the
2833 responder MUST send an EAP payload containing the Success message.
2834 Similarly, if the authentication method has failed, the responder
2835 MUST send an EAP payload containing the Failure message. The
2836 responder MAY at any time terminate the IKE exchange by sending an
2837 EAP payload containing the Failure message.
2839 Following such an extended exchange, the EAP AUTH payloads MUST be
2840 included in the two messages following the one containing the EAP
2843 When the initiator authentication uses EAP, it is possible that the
2844 contents of the IDi payload is used only for Authentication,
2845 Authorization, and Accounting (AAA) routing purposes and selecting
2846 which EAP method to use. This value may be different from the
2847 identity authenticated by the EAP method. It is important that
2848 policy lookups and access control decisions use the actual
2849 authenticated identity. Often the EAP server is implemented in a
2850 separate AAA server that communicates with the IKEv2 responder. In
2851 this case, the authenticated identity, if different from that in the
2852 IDi payload, has to be sent from the AAA server to the IKEv2
2858 Kaufman, et al. Standards Track [Page 51]
2860 RFC 5996 IKEv2bis September 2010
2863 2.17. Generating Keying Material for Child SAs
2865 A single Child SA is created by the IKE_AUTH exchange, and additional
2866 Child SAs can optionally be created in CREATE_CHILD_SA exchanges.
2867 Keying material for them is generated as follows:
2869 KEYMAT = prf+(SK_d, Ni | Nr)
2871 Where Ni and Nr are the nonces from the IKE_SA_INIT exchange if this
2872 request is the first Child SA created or the fresh Ni and Nr from the
2873 CREATE_CHILD_SA exchange if this is a subsequent creation.
2875 For CREATE_CHILD_SA exchanges including an optional Diffie-Hellman
2876 exchange, the keying material is defined as:
2878 KEYMAT = prf+(SK_d, g^ir (new) | Ni | Nr )
2880 where g^ir (new) is the shared secret from the ephemeral Diffie-
2881 Hellman exchange of this CREATE_CHILD_SA exchange (represented as an
2882 octet string in big endian order padded with zeros in the high-order
2883 bits if necessary to make it the length of the modulus).
2885 A single CHILD_SA negotiation may result in multiple Security
2886 Associations. ESP and AH SAs exist in pairs (one in each direction),
2887 so two SAs are created in a single Child SA negotiation for them.
2888 Furthermore, Child SA negotiation may include some future IPsec
2889 protocol(s) in addition to, or instead of, ESP or AH (for example,
2890 ROHC_INTEG as described in [ROHCV2]). In any case, keying material
2891 for each Child SA MUST be taken from the expanded KEYMAT using the
2894 o All keys for SAs carrying data from the initiator to the responder
2895 are taken before SAs going from the responder to the initiator.
2897 o If multiple IPsec protocols are negotiated, keying material for
2898 each Child SA is taken in the order in which the protocol headers
2899 will appear in the encapsulated packet.
2901 o If an IPsec protocol requires multiple keys, the order in which
2902 they are taken from the SA's keying material needs to be described
2903 in the protocol's specification. For ESP and AH, [IPSECARCH]
2904 defines the order, namely: the encryption key (if any) MUST be
2905 taken from the first bits and the integrity key (if any) MUST be
2906 taken from the remaining bits.
2914 Kaufman, et al. Standards Track [Page 52]
2916 RFC 5996 IKEv2bis September 2010
2919 Each cryptographic algorithm takes a fixed number of bits of keying
2920 material specified as part of the algorithm, or negotiated in SA
2921 payloads (see Section 2.13 for description of key lengths, and
2922 Section 3.3.5 for the definition of the Key Length transform
2925 2.18. Rekeying IKE SAs Using a CREATE_CHILD_SA Exchange
2927 The CREATE_CHILD_SA exchange can be used to rekey an existing IKE SA
2928 (see Sections 1.3.2 and 2.8). New initiator and responder SPIs are
2929 supplied in the SPI fields in the Proposal structures inside the
2930 Security Association (SA) payloads (not the SPI fields in the IKE
2931 header). The TS payloads are omitted when rekeying an IKE SA.
2932 SKEYSEED for the new IKE SA is computed using SK_d from the existing
2935 SKEYSEED = prf(SK_d (old), g^ir (new) | Ni | Nr)
2937 where g^ir (new) is the shared secret from the ephemeral Diffie-
2938 Hellman exchange of this CREATE_CHILD_SA exchange (represented as an
2939 octet string in big endian order padded with zeros if necessary to
2940 make it the length of the modulus) and Ni and Nr are the two nonces
2941 stripped of any headers.
2943 The old and new IKE SA may have selected a different PRF. Because
2944 the rekeying exchange belongs to the old IKE SA, it is the old IKE
2945 SA's PRF that is used to generate SKEYSEED.
2947 The main reason for rekeying the IKE SA is to ensure that the
2948 compromise of old keying material does not provide information about
2949 the current keys, or vice versa. Therefore, implementations MUST
2950 perform a new Diffie-Hellman exchange when rekeying the IKE SA. In
2951 other words, an initiator MUST NOT propose the value "NONE" for the
2952 Diffie-Hellman transform, and a responder MUST NOT accept such a
2953 proposal. This means that a successful exchange rekeying the IKE SA
2954 always includes the KEi/KEr payloads.
2956 The new IKE SA MUST reset its message counters to 0.
2958 SK_d, SK_ai, SK_ar, SK_ei, and SK_er are computed from SKEYSEED as
2959 specified in Section 2.14, using SPIi, SPIr, Ni, and Nr from the new
2960 exchange, and using the new IKE SA's PRF.
2962 2.19. Requesting an Internal Address on a Remote Network
2964 Most commonly occurring in the endpoint-to-security-gateway scenario,
2965 an endpoint may need an IP address in the network protected by the
2966 security gateway and may need to have that address dynamically
2970 Kaufman, et al. Standards Track [Page 53]
2972 RFC 5996 IKEv2bis September 2010
2975 assigned. A request for such a temporary address can be included in
2976 any request to create a Child SA (including the implicit request in
2977 message 3) by including a CP payload. Note, however, it is usual to
2978 only assign one IP address during the IKE_AUTH exchange. That
2979 address persists at least until the deletion of the IKE SA.
2981 This function provides address allocation to an IPsec Remote Access
2982 Client (IRAC) trying to tunnel into a network protected by an IPsec
2983 Remote Access Server (IRAS). Since the IKE_AUTH exchange creates an
2984 IKE SA and a Child SA, the IRAC MUST request the IRAS-controlled
2985 address (and optionally other information concerning the protected
2986 network) in the IKE_AUTH exchange. The IRAS may procure an address
2987 for the IRAC from any number of sources such as a DHCP/BOOTP
2988 (Bootstrap Protocol) server or its own address pool.
2991 -------------------------------------------------------------------
2992 HDR, SK {IDi, [CERT,]
2993 [CERTREQ,] [IDr,] AUTH,
2994 CP(CFG_REQUEST), SAi2,
2996 <-- HDR, SK {IDr, [CERT,] AUTH,
2997 CP(CFG_REPLY), SAr2,
3000 In all cases, the CP payload MUST be inserted before the SA payload.
3001 In variations of the protocol where there are multiple IKE_AUTH
3002 exchanges, the CP payloads MUST be inserted in the messages
3003 containing the SA payloads.
3005 CP(CFG_REQUEST) MUST contain at least an INTERNAL_ADDRESS attribute
3006 (either IPv4 or IPv6) but MAY contain any number of additional
3007 attributes the initiator wants returned in the response.
3026 Kaufman, et al. Standards Track [Page 54]
3028 RFC 5996 IKEv2bis September 2010
3031 For example, message from initiator to responder:
3035 TSi = (0, 0-65535,0.0.0.0-255.255.255.255)
3036 TSr = (0, 0-65535,0.0.0.0-255.255.255.255)
3038 NOTE: Traffic Selectors contain (protocol, port range, address
3041 Message from responder to initiator:
3044 INTERNAL_ADDRESS(192.0.2.202)
3045 INTERNAL_NETMASK(255.255.255.0)
3046 INTERNAL_SUBNET(192.0.2.0/255.255.255.0)
3047 TSi = (0, 0-65535,192.0.2.202-192.0.2.202)
3048 TSr = (0, 0-65535,192.0.2.0-192.0.2.255)
3050 All returned values will be implementation dependent. As can be seen
3051 in the above example, the IRAS MAY also send other attributes that
3052 were not included in CP(CFG_REQUEST) and MAY ignore the non-
3053 mandatory attributes that it does not support.
3055 The responder MUST NOT send a CFG_REPLY without having first received
3056 a CP(CFG_REQUEST) from the initiator, because we do not want the IRAS
3057 to perform an unnecessary configuration lookup if the IRAC cannot
3060 In the case where the IRAS's configuration requires that CP be used
3061 for a given identity IDi, but IRAC has failed to send a
3062 CP(CFG_REQUEST), IRAS MUST fail the request, and terminate the Child
3063 SA creation with a FAILED_CP_REQUIRED error. The FAILED_CP_REQUIRED
3064 is not fatal to the IKE SA; it simply causes the Child SA creation to
3065 fail. The initiator can fix this by later starting a new
3066 Configuration payload request. There is no associated data in the
3067 FAILED_CP_REQUIRED error.
3069 2.20. Requesting the Peer's Version
3071 An IKE peer wishing to inquire about the other peer's IKE software
3072 version information MAY use the method below. This is an example of
3073 a configuration request within an INFORMATIONAL exchange, after the
3074 IKE SA and first Child SA have been created.
3082 Kaufman, et al. Standards Track [Page 55]
3084 RFC 5996 IKEv2bis September 2010
3087 An IKE implementation MAY decline to give out version information
3088 prior to authentication or even after authentication in case some
3089 implementation is known to have some security weakness. In that
3090 case, it MUST either return an empty string or no CP payload if CP is
3094 -------------------------------------------------------------------
3095 HDR, SK{CP(CFG_REQUEST)} -->
3096 <-- HDR, SK{CP(CFG_REPLY)}
3099 APPLICATION_VERSION("")
3101 CP(CFG_REPLY) APPLICATION_VERSION("foobar v1.3beta, (c) Foo Bar
3104 2.21. Error Handling
3106 There are many kinds of errors that can occur during IKE processing.
3107 The general rule is that if a request is received that is badly
3108 formatted, or unacceptable for reasons of policy (such as no matching
3109 cryptographic algorithms), the response contains a Notify payload
3110 indicating the error. The decision whether or not to send such a
3111 response depends whether or not there is an authenticated IKE SA.
3113 If there is an error parsing or processing a response packet, the
3114 general rule is to not send back any error message because responses
3115 should not generate new requests (and a new request would be the only
3116 way to send back an error message). Such errors in parsing or
3117 processing response packets should still cause the recipient to clean
3118 up the IKE state (for example, by sending a Delete for a bad SA).
3120 Only authentication failures (AUTHENTICATION_FAILED and EAP failure)
3121 and malformed messages (INVALID_SYNTAX) lead to a deletion of the IKE
3122 SA without requiring an explicit INFORMATIONAL exchange carrying a
3123 Delete payload. Other error conditions MAY require such an exchange
3124 if policy dictates that this is needed. If the exchange is
3125 terminated with EAP Failure, an AUTHENTICATION_FAILED notification is
3128 2.21.1. Error Handling in IKE_SA_INIT
3130 Errors that occur before a cryptographically protected IKE SA is
3131 established need to be handled very carefully. There is a trade-off
3132 between wanting to help the peer to diagnose a problem and thus
3133 responding to the error and wanting to avoid being part of a DoS
3134 attack based on forged messages.
3138 Kaufman, et al. Standards Track [Page 56]
3140 RFC 5996 IKEv2bis September 2010
3143 In an IKE_SA_INIT exchange, any error notification causes the
3144 exchange to fail. Note that some error notifications such as COOKIE,
3145 INVALID_KE_PAYLOAD or INVALID_MAJOR_VERSION may lead to a subsequent
3146 successful exchange. Because all error notifications are completely
3147 unauthenticated, the recipient should continue trying for some time
3148 before giving up. The recipient should not immediately act based on
3149 the error notification unless corrective actions are defined in this
3150 specification, such as for COOKIE, INVALID_KE_PAYLOAD, and
3151 INVALID_MAJOR_VERSION.
3153 2.21.2. Error Handling in IKE_AUTH
3155 All errors that occur in an IKE_AUTH exchange, causing the
3156 authentication to fail for whatever reason (invalid shared secret,
3157 invalid ID, untrusted certificate issuer, revoked or expired
3158 certificate, etc.) SHOULD result in an AUTHENTICATION_FAILED
3159 notification. If the error occurred on the responder, the
3160 notification is returned in the protected response, and is usually
3161 the only payload in that response. Although the IKE_AUTH messages
3162 are encrypted and integrity protected, if the peer receiving this
3163 notification has not authenticated the other end yet, that peer needs
3164 to treat the information with caution.
3166 If the error occurs on the initiator, the notification MAY be
3167 returned in a separate INFORMATIONAL exchange, usually with no other
3168 payloads. This is an exception for the general rule of not starting
3169 new exchanges based on errors in responses.
3171 Note, however, that request messages that contain an unsupported
3172 critical payload, or where the whole message is malformed (rather
3173 than just bad payload contents), MUST be rejected in their entirety,
3174 and MUST only lead to an UNSUPPORTED_CRITICAL_PAYLOAD or
3175 INVALID_SYNTAX Notification sent as a response. The receiver should
3176 not verify the payloads related to authentication in this case.
3178 If authentication has succeeded in the IKE_AUTH exchange, the IKE SA
3179 is established; however, establishing the Child SA or requesting
3180 configuration information may still fail. This failure does not
3181 automatically cause the IKE SA to be deleted. Specifically, a
3182 responder may include all the payloads associated with authentication
3183 (IDr, CERT, and AUTH) while sending error notifications for the
3184 piggybacked exchanges (FAILED_CP_REQUIRED, NO_PROPOSAL_CHOSEN, and so
3185 on), and the initiator MUST NOT fail the authentication because of
3186 this. The initiator MAY, of course, for reasons of policy later
3187 delete such an IKE SA.
3194 Kaufman, et al. Standards Track [Page 57]
3196 RFC 5996 IKEv2bis September 2010
3199 In an IKE_AUTH exchange, or in the INFORMATIONAL exchange immediately
3200 following it (in case an error happened when processing a response to
3201 IKE_AUTH), the UNSUPPORTED_CRITICAL_PAYLOAD, INVALID_SYNTAX, and
3202 AUTHENTICATION_FAILED notifications are the only ones to cause the
3203 IKE SA to be deleted or not created, without a Delete payload.
3204 Extension documents may define new error notifications with these
3205 semantics, but MUST NOT use them unless the peer has been shown to
3206 understand them, such as by using the Vendor ID payload.
3208 2.21.3. Error Handling after IKE SA is Authenticated
3210 After the IKE SA is authenticated, all requests having errors MUST
3211 result in a response notifying about the error.
3213 In normal situations, there should not be cases where a valid
3214 response from one peer results in an error situation in the other
3215 peer, so there should not be any reason for a peer to send error
3216 messages to the other end except as a response. Because sending such
3217 error messages as an INFORMATIONAL exchange might lead to further
3218 errors that could cause loops, such errors SHOULD NOT be sent. If
3219 errors are seen that indicate that the peers do not have the same
3220 state, it might be good to delete the IKE SA to clean up state and
3223 If a peer parsing a request notices that it is badly formatted (after
3224 it has passed the message authentication code checks and window
3225 checks) and it returns an INVALID_SYNTAX notification, then this
3226 error notification is considered fatal in both peers, meaning that
3227 the IKE SA is deleted without needing an explicit Delete payload.
3229 2.21.4. Error Handling Outside IKE SA
3231 A node needs to limit the rate at which it will send messages in
3232 response to unprotected messages.
3234 If a node receives a message on UDP port 500 or 4500 outside the
3235 context of an IKE SA known to it (and the message is not a request to
3236 start an IKE SA), this may be the result of a recent crash of the
3237 node. If the message is marked as a response, the node can audit the
3238 suspicious event but MUST NOT respond. If the message is marked as a
3239 request, the node can audit the suspicious event and MAY send a
3240 response. If a response is sent, the response MUST be sent to the IP
3241 address and port from where it came with the same IKE SPIs and the
3242 Message ID copied. The response MUST NOT be cryptographically
3243 protected and MUST contain an INVALID_IKE_SPI Notify payload. The
3244 INVALID_IKE_SPI notification indicates an IKE message was received
3245 with an unrecognized destination SPI; this usually indicates that the
3246 recipient has rebooted and forgotten the existence of an IKE SA.
3250 Kaufman, et al. Standards Track [Page 58]
3252 RFC 5996 IKEv2bis September 2010
3255 A peer receiving such an unprotected Notify payload MUST NOT respond
3256 and MUST NOT change the state of any existing SAs. The message might
3257 be a forgery or might be a response that a genuine correspondent was
3258 tricked into sending. A node should treat such a message (and also a
3259 network message like ICMP destination unreachable) as a hint that
3260 there might be problems with SAs to that IP address and should
3261 initiate a liveness check for any such IKE SA. An implementation
3262 SHOULD limit the frequency of such tests to avoid being tricked into
3263 participating in a DoS attack.
3265 If an error occurs outside the context of an IKE request (e.g., the
3266 node is getting ESP messages on a nonexistent SPI), the node SHOULD
3267 initiate an INFORMATIONAL exchange with a Notify payload describing
3270 A node receiving a suspicious message from an IP address (and port,
3271 if NAT traversal is used) with which it has an IKE SA SHOULD send an
3272 IKE Notify payload in an IKE INFORMATIONAL exchange over that SA.
3273 The recipient MUST NOT change the state of any SAs as a result, but
3274 may wish to audit the event to aid in diagnosing malfunctions.
3278 Use of IP Compression [IP-COMP] can be negotiated as part of the
3279 setup of a Child SA. While IP Compression involves an extra header
3280 in each packet and a compression parameter index (CPI), the virtual
3281 "compression association" has no life outside the ESP or AH SA that
3282 contains it. Compression associations disappear when the
3283 corresponding ESP or AH SA goes away. It is not explicitly mentioned
3284 in any Delete payload.
3286 Negotiation of IP Compression is separate from the negotiation of
3287 cryptographic parameters associated with a Child SA. A node
3288 requesting a Child SA MAY advertise its support for one or more
3289 compression algorithms through one or more Notify payloads of type
3290 IPCOMP_SUPPORTED. This Notify message may be included only in a
3291 message containing an SA payload negotiating a Child SA and indicates
3292 a willingness by its sender to use IPComp on this SA. The response
3293 MAY indicate acceptance of a single compression algorithm with a
3294 Notify payload of type IPCOMP_SUPPORTED. These payloads MUST NOT
3295 occur in messages that do not contain SA payloads.
3297 The data associated with this Notify message includes a two-octet
3298 IPComp CPI followed by a one-octet Transform ID optionally followed
3299 by attributes whose length and format are defined by that Transform
3300 ID. A message proposing an SA may contain multiple IPCOMP_SUPPORTED
3301 notifications to indica
projects / strongswan.git / blob