idnits 2.17.1 draft-ietf-v6ops-mech-v2-02.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- ** Looks like you're using RFC 2026 boilerplate. This must be updated to follow RFC 3978/3979, as updated by RFC 4748. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- == No 'Intended status' indicated for this document; assuming Proposed Standard Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The document seems to lack an IANA Considerations section. (See Section 2.2 of https://www.ietf.org/id-info/checklist for how to handle the case when there are no actions for IANA.) ** There is 1 instance of too long lines in the document, the longest one being 2 characters in excess of 72. Miscellaneous warnings: ---------------------------------------------------------------------------- == Line 1197 has weird spacing: '...arified that ...' == The document seems to lack the recommended RFC 2119 boilerplate, even if it appears to use RFC 2119 keywords. (The document does seem to have the reference to RFC 2119 which the ID-Checklist requires). -- The document seems to lack a disclaimer for pre-RFC5378 work, but may have content which was first submitted before 10 November 2008. If you have contacted all the original authors and they are all willing to grant the BCP78 rights to the IETF Trust, then this is fine, and you can ignore this comment. If not, you may need to add the pre-RFC5378 disclaimer. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- Couldn't find a document date in the document -- date freshness check skipped. Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Missing Reference: 'RFC2401' is mentioned on line 917, but not defined ** Obsolete undefined reference: RFC 2401 (Obsoleted by RFC 4301) == Unused Reference: 'RFC1122' is defined on line 1000, but no explicit reference was found in the text == Unused Reference: 'RFC1812' is defined on line 1009, but no explicit reference was found in the text == Unused Reference: 'RFC2462' is defined on line 1015, but no explicit reference was found in the text ** Obsolete normative reference: RFC 1981 (Obsoleted by RFC 8201) ** Obsolete normative reference: RFC 2460 (Obsoleted by RFC 8200) ** Obsolete normative reference: RFC 2463 (Obsoleted by RFC 4443) == Outdated reference: A later version (-12) exists of draft-ietf-dnsop-ipv6-dns-issues-04 -- Obsolete informational reference (is this intentional?): RFC 2401 (ref. 'RFC1812') (Obsoleted by RFC 4301) -- Obsolete informational reference (is this intentional?): RFC 2461 (Obsoleted by RFC 4861) -- Obsolete informational reference (is this intentional?): RFC 2462 (Obsoleted by RFC 4862) -- Obsolete informational reference (is this intentional?): RFC 2667 (Obsoleted by RFC 4087) -- Obsolete informational reference (is this intentional?): RFC 3484 (Obsoleted by RFC 6724) -- Obsolete informational reference (is this intentional?): RFC 3513 (Obsoleted by RFC 4291) Summary: 7 errors (**), 0 flaws (~~), 8 warnings (==), 8 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 INTERNET-DRAFT E. Nordmark 3 January 30, 2004 Sun Microsystems, Inc. 4 Obsoletes: 2893 R. E. Gilligan 5 Intransa, Inc. 7 Basic Transition Mechanisms for IPv6 Hosts and Routers 8 10 Status of this Memo 12 This document is an Internet-Draft and is subject to all provisions 13 of Section 10 of RFC2026. 15 Internet-Drafts are working documents of the Internet Engineering 16 Task Force (IETF), its areas, and its working groups. Note that 17 other groups may also distribute working documents as Internet- 18 Drafts. 20 Internet-Drafts are draft documents valid for a maximum of six months 21 and may be updated, replaced, or obsoleted by other documents at any 22 time. It is inappropriate to use Internet-Drafts as reference 23 material or to cite them other than as "work in progress." 25 The list of current Internet-Drafts can be accessed at 26 http://www.ietf.org/ietf/1id-abstracts.txt 28 The list of Internet-Draft Shadow Directories can be accessed at 29 http://www.ietf.org/shadow.html. 31 This draft expires on July 30, 2004. 33 Abstract 35 This document specifies IPv4 compatibility mechanisms that can be 36 implemented by IPv6 hosts and routers. Two mechanisms are specified, 37 "dual stack" and configured tunneling. Dual stack implies providing 38 complete implementations of both versions of the Internet Protocol 39 (IPv4 and IPv6) and configured tunneling provides a means to carry 40 IPv6 packets over unmodified IPv4 routing infrastructures. 42 This document obsoletes RFC 2893. 44 Contents 46 Status of this Memo.......................................... 1 48 1. Introduction............................................. 3 49 1.1. Terminology......................................... 3 51 2. Dual IP Layer Operation.................................. 5 52 2.1. Address Configuration............................... 5 53 2.2. DNS................................................. 5 55 3. Configured Tunneling Mechanisms.......................... 7 56 3.1. Encapsulation....................................... 8 57 3.2. Tunnel MTU and Fragmentation........................ 9 58 3.2.1. Static Tunnel MTU.............................. 10 59 3.2.2. Dynamic Tunnel MTU............................. 10 60 3.3. Hop Limit........................................... 12 61 3.4. Handling ICMPv4 errors.............................. 12 62 3.5. IPv4 Header Construction............................ 14 63 3.6. Decapsulation....................................... 15 64 3.7. Link-Local Addresses................................ 18 65 3.8. Neighbor Discovery over Tunnels..................... 18 67 4. Threat Related to Source Address Spoofing................ 19 69 5. Security Considerations.................................. 20 71 6. Acknowledgments.......................................... 22 73 7. References............................................... 22 74 7.1. Normative References................................ 22 75 7.2. Non-normative References............................ 22 77 8. Authors' Addresses....................................... 24 79 9. Changes from RFC 2893.................................... 25 80 9.1. Changes from draft-ietf-v6ops-mech-v2-00............ 27 81 9.2. Changes from draft-ietf-v6ops-mech-v2-01............ 28 83 1. Introduction 85 The key to a successful IPv6 transition is compatibility with the 86 large installed base of IPv4 hosts and routers. Maintaining 87 compatibility with IPv4 while deploying IPv6 will streamline the task 88 of transitioning the Internet to IPv6. This specification defines 89 two mechanisms that IPv6 hosts and routers may implement in order to 90 be compatible with IPv4 hosts and routers. 92 The mechanisms in this document are designed to be employed by IPv6 93 hosts and routers that need to interoperate with IPv4 hosts and 94 utilize IPv4 routing infrastructures. We expect that most nodes in 95 the Internet will need such compatibility for a long time to come, 96 and perhaps even indefinitely. 98 The mechanisms specified here are: 100 - Dual IP layer (also known as Dual Stack): A technique for 101 providing complete support for both Internet protocols -- IPv4 102 and IPv6 -- in hosts and routers. 104 - Configured tunneling of IPv6 over IPv4: A technique for 105 establishing point-to-point tunnels by encapsulating IPv6 106 packets within IPv4 headers to carry them over IPv4 routing 107 infrastructures. 109 The mechanisms defined here are intended to be the core of a 110 "transition toolbox" -- a growing collection of techniques which 111 implementations and users may employ to ease the transition. The 112 tools may be used as needed. Implementations and sites decide which 113 techniques are appropriate to their specific needs. 115 This document defines the basic set of transition mechanisms, but 116 these are not the only tools available. Additional transition and 117 compatibility mechanisms are specified in other documents. 119 1.1. Terminology 121 The following terms are used in this document: 123 Types of Nodes 125 IPv4-only node: 127 A host or router that implements only IPv4. An IPv4- 128 only node does not understand IPv6. The installed base 129 of IPv4 hosts and routers existing before the transition 130 begins are IPv4-only nodes. 132 IPv6/IPv4 node: 134 A host or router that implements both IPv4 and IPv6. 136 IPv6-only node: 138 A host or router that implements IPv6, and does not 139 implement IPv4. The operation of IPv6-only nodes is not 140 addressed in this memo. 142 IPv6 node: 144 Any host or router that implements IPv6. IPv6/IPv4 and 145 IPv6-only nodes are both IPv6 nodes. 147 IPv4 node: 149 Any host or router that implements IPv4. IPv6/IPv4 and 150 IPv4-only nodes are both IPv4 nodes. 152 Techniques Used in the Transition 154 IPv6-over-IPv4 tunneling: 156 The technique of encapsulating IPv6 packets within IPv4 157 so that they can be carried across IPv4 routing 158 infrastructures. 160 Configured tunneling: 162 IPv6-over-IPv4 tunneling where the IPv4 tunnel endpoint 163 address is determined by configuration information on 164 the encapsulator. All tunnels are assumed to be 165 bidirectional, behaving as virtual point-to-point links. 167 Other transition mechanisms, including other tunneling mechanisms, 168 are outside the scope of this document. 170 The keywords MUST, MUST NOT, REQUIRED, SHALL, SHALL NOT, SHOULD, 171 SHOULD NOT, RECOMMENDED, MAY, and OPTIONAL, when they appear in this 172 document, are to be interpreted as described in [RFC2119]. 174 2. Dual IP Layer Operation 176 The most straightforward way for IPv6 nodes to remain compatible with 177 IPv4-only nodes is by providing a complete IPv4 implementation. IPv6 178 nodes that provide a complete IPv4 and IPv6 implementations are 179 called "IPv6/IPv4 nodes." IPv6/IPv4 nodes have the ability to send 180 and receive both IPv4 and IPv6 packets. They can directly 181 interoperate with IPv4 nodes using IPv4 packets, and also directly 182 interoperate with IPv6 nodes using IPv6 packets. 184 Even though a node may be equipped to support both protocols, one or 185 the other stack may be disabled for operational reasons. Here we use 186 a rather loose notion of "stack". A stack being enabled has IP 187 addresses assigned etc, but whether or not any particular application 188 is available on the stacks is explicitly not defined. Thus IPv6/IPv4 189 nodes may be operated in one of three modes: 191 - With their IPv4 stack enabled and their IPv6 stack disabled. 193 - With their IPv6 stack enabled and their IPv4 stack disabled. 195 - With both stacks enabled. 197 IPv6/IPv4 nodes with their IPv6 stack disabled will operate like 198 IPv4-only nodes. Similarly, IPv6/IPv4 nodes with their IPv4 stacks 199 disabled will operate like IPv6-only nodes. IPv6/IPv4 nodes MAY 200 provide a configuration switch to disable either their IPv4 or IPv6 201 stack. 203 The configured tunneling technique, which is described in section 3, 204 may or may not be used in addition to the dual IP layer operation. 206 2.1. Address Configuration 208 Because the nodes support both protocols, IPv6/IPv4 nodes may be 209 configured with both IPv4 and IPv6 addresses. IPv6/IPv4 nodes use 210 IPv4 mechanisms (e.g., DHCP) to acquire their IPv4 addresses, and 211 IPv6 protocol mechanisms (e.g., stateless address autoconfiguration 212 and/or DHCPv6) to acquire their IPv6 addresses. 214 2.2. DNS 216 The Domain Naming System (DNS) is used in both IPv4 and IPv6 to map 217 between hostnames and IP addresses. A new resource record type named 218 "AAAA" has been defined for IPv6 addresses [RFC3596]. Since 219 IPv6/IPv4 nodes must be able to interoperate directly with both IPv4 220 and IPv6 nodes, they must provide resolver libraries capable of 221 dealing with IPv4 "A" records as well as IPv6 "AAAA" records. Note 222 that the lookup of A versus AAAA records is independent of whether 223 the DNS packets are carried in IPv4 or IPv6 packets, and that there 224 is no assumption that the DNS server know the IPv4/IPv6 capabilities 225 of the requesting node. 227 The issues and operational guidelines for using IPv6 with DNS are 228 described at more length in other documents [DNSOPV6]. 230 DNS resolver libraries on IPv6/IPv4 nodes MUST be capable of handling 231 both AAAA and A records. However, when a query locates an AAAA 232 record holding an IPv6 address, and an A record holding an IPv4 233 address, the resolver library MAY filter or order the results 234 returned to the application in order to influence the version of IP 235 packets used to communicate with that node. In terms of filtering, 236 the resolver library has three alternatives: 238 - Return only the IPv6 address(es) to the application. 240 - Return only the IPv4 address(es) to the application. 242 - Return both types of addresses to the application. 244 If it returns only the IPv6 address(es), the application will 245 communicate with the node using IPv6. If it returns only the IPv4 246 address(es), the application will communicate with the node using 247 IPv4. If it returns both types of addresses, the application will 248 have the choice which address to use, and thus which IP protocol to 249 employ. 251 If it returns both, the resolver MAY elect to order the addresses -- 252 IPv6 first, or IPv4 first. Since most applications try the addresses 253 in the order they are returned by the resolver, this can affect the 254 IP version "preference" of applications. 256 A resolver library performing filtering or ordering of addresses 257 might also want to take into account external factors such as, 258 whether IPv6 interfaces have been configured on the node. 260 The decision to filter or order DNS results is implementation 261 specific. IPv6/IPv4 nodes MAY provide policy configuration to 262 control filtering or ordering of addresses returned by the resolver 263 -- i.e., which addresses to filter or which order to sort -- or leave 264 the decision entirely up to the application. 266 An implementation MUST allow the application to control whether or 267 not such filtering takes place. 269 More details on the relative preferences of IPv4 and IPv6 addresses 270 are specified in the default address selection document [RFC3484]. 272 3. Configured Tunneling Mechanisms 274 In most deployment scenarios, the IPv6 routing infrastructure will be 275 built up over time. While the IPv6 infrastructure is being deployed, 276 the existing IPv4 routing infrastructure can remain functional, and 277 can be used to carry IPv6 traffic. Tunneling provides a way to 278 utilize an existing IPv4 routing infrastructure to carry IPv6 279 traffic. 281 IPv6/IPv4 hosts and routers can tunnel IPv6 datagrams over regions of 282 IPv4 routing topology by encapsulating them within IPv4 packets. 283 Tunneling can be used in a variety of ways: 285 - Router-to-Router. IPv6/IPv4 routers interconnected by an IPv4 286 infrastructure can tunnel IPv6 packets between themselves. In 287 this case, the tunnel spans one segment of the end-to-end path 288 that the IPv6 packet takes. 290 - Host-to-Router. IPv6/IPv4 hosts can tunnel IPv6 packets to an 291 intermediary IPv6/IPv4 router that is reachable via an IPv4 292 infrastructure. This type of tunnel spans the first segment of 293 the packet's end-to-end path. 295 - Host-to-Host. IPv6/IPv4 hosts that are interconnected by an 296 IPv4 infrastructure can tunnel IPv6 packets between themselves. 297 In this case, the tunnel spans the entire end-to-end path that 298 the packet takes. 300 - Router-to-Host. IPv6/IPv4 routers can tunnel IPv6 packets to 301 their final destination IPv6/IPv4 host. This tunnel spans only 302 the last segment of the end-to-end path. 304 Configured tunneling can be used in all of the above cases, but is 305 most likely to be used router-to-router due to the need to explicitly 306 configure the tunneling endpoints. 308 The underlying mechanisms for tunneling are: 310 - The entry node of the tunnel (the encapsulator) creates an 311 encapsulating IPv4 header and transmits the encapsulated packet. 313 - The exit node of the tunnel (the decapsulator) receives the 314 encapsulated packet, reassembles the packet if needed, removes 315 the IPv4 header, and processes the received IPv6 packet. 317 - The encapsulator may need to maintain soft state information for 318 each tunnel recording such parameters as the MTU of the tunnel 319 in order to process IPv6 packets forwarded into the tunnel. 321 In configured tunneling, the tunnel endpoint address is determined 322 from configuration information in the encapsulator. For each tunnel, 323 the encapsulator must store the tunnel endpoint address. When an 324 IPv6 packet is transmitted over a tunnel, the tunnel endpoint address 325 configured for that tunnel is used as the destination address for the 326 encapsulating IPv4 header. 328 The determination of which packets to tunnel is usually made by 329 routing information on the encapsulator. This is usually done via a 330 routing table, which directs packets based on their destination 331 address using the prefix mask and match technique. 333 3.1. Encapsulation 335 The encapsulation of an IPv6 datagram in IPv4 is shown below: 337 +-------------+ 338 | IPv4 | 339 | Header | 340 +-------------+ +-------------+ 341 | IPv6 | | IPv6 | 342 | Header | | Header | 343 +-------------+ +-------------+ 344 | Transport | | Transport | 345 | Layer | ===> | Layer | 346 | Header | | Header | 347 +-------------+ +-------------+ 348 | | | | 349 ~ Data ~ ~ Data ~ 350 | | | | 351 +-------------+ +-------------+ 353 Encapsulating IPv6 in IPv4 355 In addition to adding an IPv4 header, the encapsulator also has to 356 handle some more complex issues: 358 - Determine when to fragment and when to report an ICMPv6 "packet 359 too big" error back to the source. 361 - How to reflect ICMPv4 errors from routers along the tunnel path 362 back to the source as ICMPv6 errors. 364 Those issues are discussed in the following sections. 366 3.2. Tunnel MTU and Fragmentation 368 Naively the encapsulator could view encapsulation as IPv6 using IPv4 369 as a link layer with a very large MTU (65535-20 bytes to be exact; 20 370 bytes "extra" are needed for the encapsulating IPv4 header). The 371 encapsulator would only need to report ICMPv6 "packet too big" errors 372 back to the source for packets that exceed this MTU. However, such a 373 scheme would be inefficient or non-interoperable for three reasons 374 and therefore MUST NOT be used: 376 1) It would result in more fragmentation than needed. IPv4 layer 377 fragmentation should be avoided due to the performance problems 378 caused by the loss unit being smaller than the retransmission 379 unit [KM97]. 381 2) Any IPv4 fragmentation occurring inside the tunnel, i.e. between 382 the encapsulator and the decapsulator, would have to be 383 reassembled at the tunnel endpoint. For tunnels that terminate 384 at a router, this would require additional memory and other 385 resources to reassemble the IPv4 fragments into a complete IPv6 386 packet before that packet could be forwarded onward. 388 3) The encapsulator has no way of knowing that the decapsulator is 389 able to defragment such IPv4 packets (see Section 3.7 for 390 details), and has no way of knowing that the decapsulator is 391 able to handle such a large IPv6 Maximum Receive Unit (MRU). 393 Hence, the encapsulator MUST NOT treat the tunnel as an interface 394 with an MTU of 64 kilobytes, but instead either use the fixed static 395 MTU or OPTIONAL dynamic MTU determination based on the IPv4 path MTU 396 to the tunnel endpoint. 398 If both the mechanisms are implemented, the decision which to use 399 SHOULD be configurable on a per-tunnel endpoint basis. 401 3.2.1. Static Tunnel MTU 403 A node using static tunnel MTU treats the tunnel interface as having 404 a fixed interface MTU. By default, the MTU MUST be between 1280 and 405 1480 bytes (inclusive), but it SHOULD be 1280 bytes. If the default 406 is not 1280 bytes, the implementation MUST have a configuration knob 407 which can be used to change the MTU value. 409 A node must be able to accept a fragmented IPv6 packet that, after 410 reassembly, is as large as 1500 octets [RFC2460]. This memo also 411 includes requirements (see Section 3.6) for the amount of IPv4 412 reassembly and IPv6 MRU that MUST be supported by all the 413 decapsulators. These ensure correct interoperability with any fixed 414 MTUs between 1280 and 1480 bytes. 416 A larger fixed MTU than supported by these requirements, must not be 417 configured unless it has been administratively ensured that the 418 decapsulator can reassemble or receive packets of that size. 420 The selection of a good tunnel MTU depends on many factors; at least: 422 - Whether the IPv4 protocol-41 packets will be transported over 423 media which may have a lower path MTU (e.g., IPv4 Virtual 424 Private Networks); then picking too high a value might lead to 425 IPv4 fragmentation. 427 - Whether the tunnel is used to transport IPv6 tunneled packets 428 (e.g., a mobile node with an IPv4-in-IPv6 configured tunnel, and 429 an IPv6-in-IPv6 tunnel interface); then picking too low a value 430 might lead to IPv6 fragmentation. 432 If layered encapsulation is believed to be present, it may be prudent 433 to consider supporting dynamic MTU determination instead as it is 434 able to minimize fragmentation and optimize packet sizes. 436 When using the static tunnel MTU the Don't Fragment bit MUST NOT be 437 set in the encapsulating IPv4 header. As a result the encapsulator 438 should not receive any ICMPv4 "packet too big" messages as a result 439 of the packets it has encapsulated. 441 3.2.2. Dynamic Tunnel MTU 443 The dynamic MTU determination is OPTIONAL. However, if it is 444 implemented, it SHOULD have the behavior described in this document. 446 The fragmentation inside the tunnel can be reduced to a minimum by 447 having the encapsulator track the IPv4 Path MTU across the tunnel, 448 using the IPv4 Path MTU Discovery Protocol [RFC1191] and recording 449 the resulting path MTU. The IPv6 layer in the encapsulator can then 450 view a tunnel as a link layer with an MTU equal to the IPv4 path MTU, 451 minus the size of the encapsulating IPv4 header. 453 Note that this does not eliminate IPv4 fragmentation in the case when 454 the IPv4 path MTU would result in an IPv6 MTU less than 1280 bytes. 455 (Any link layer used by IPv6 has to have an MTU of at least 1280 456 bytes [RFC2460].) In this case the IPv6 layer has to "see" a link 457 layer with an MTU of 1280 bytes and the encapsulator has to use IPv4 458 fragmentation in order to forward the 1280 byte IPv6 packets. 460 The encapsulator SHOULD employ the following algorithm to determine 461 when to forward an IPv6 packet that is larger than the tunnel's path 462 MTU using IPv4 fragmentation, and when to return an ICMPv6 "packet 463 too big" message per [RFC1981]: 465 if (IPv4 path MTU - 20) is less than 1280 466 if packet is larger than 1280 bytes 467 Send ICMPv6 "packet too big" with MTU = 1280. 468 Drop packet. 469 else 470 Encapsulate but do not set the Don't Fragment 471 flag in the IPv4 header. The resulting IPv4 472 packet might be fragmented by the IPv4 layer on 473 the encapsulator or by some router along 474 the IPv4 path. 475 endif 476 else 477 if packet is larger than (IPv4 path MTU - 20) 478 Send ICMPv6 "packet too big" with 479 MTU = (IPv4 path MTU - 20). 480 Drop packet. 481 else 482 Encapsulate and set the Don't Fragment flag 483 in the IPv4 header. 484 endif 485 endif 487 Encapsulators that have a large number of tunnels may choose between 488 dynamic versus static tunnel MTU on a per-tunnel endpoint basis. In 489 cases where the number of tunnels that any one node is using is 490 large, it is helpful to observe that this state information can be 491 cached and discarded when not in use. 493 Note that using dynamic tunnel MTU is subject to IPv4 PMTU blackholes 494 should the ICMPv4 "packet too big" messages be dropped by firewalls 495 or not generated by the routers. [RFC1435, RFC2923] 497 3.3. Hop Limit 499 IPv6-over-IPv4 tunnels are modeled as "single-hop" from the IPv6 500 perspective. The tunnel is opaque to users of the network, and is not 501 detectable by network diagnostic tools such as traceroute. 503 The single-hop model is implemented by having the encapsulators and 504 decapsulators process the IPv6 hop limit field as they would if they 505 were forwarding a packet on to any other datalink. That is, they 506 decrement the hop limit by 1 when forwarding an IPv6 packet. (The 507 originating node and final destination do not decrement the hop 508 limit.) 510 The TTL of the encapsulating IPv4 header is selected in an 511 implementation dependent manner. The current suggested value is 512 published in the "Assigned Numbers" RFC [RFC3232][ASSIGNED]. The 513 implementations MAY also consider using the value 255, as it could be 514 used as a hint in the decapsulation checks in the future [GTSM]. 515 Implementations MAY provide a mechanism to allow the administrator to 516 configure the IPv4 TTL as the IP Tunnel MIB [RFC2667]. 518 3.4. Handling ICMPv4 errors 520 In response to encapsulated packets it has sent into the tunnel, the 521 encapsulator might receive ICMPv4 error messages from IPv4 routers 522 inside the tunnel. These packets are addressed to the encapsulator 523 because it is the IPv4 source of the encapsulated packet. 525 ICMPv4 error handling is only applicable to dynamic MTU 526 determination, even though the functions could be used with static 527 MTU tunnels as well. 529 The ICMPv4 "packet too big" error messages are handled according to 530 IPv4 Path MTU Discovery [RFC1191] and the resulting path MTU is 531 recorded in the IPv4 layer. The recorded path MTU is used by IPv6 to 532 determine if an ICMPv6 "packet too big" error has to be generated as 533 described in section 3.2.2. 535 The handling of other types of ICMPv4 error messages depends on how 536 much information is included in the "packet in error" field, which 537 holds the encapsulated packet that caused the error. 539 Many older IPv4 routers return only 8 bytes of data beyond the IPv4 540 header of the packet in error, which is not enough to include the 541 address fields of the IPv6 header. More modern IPv4 routers are 542 likely to return enough data beyond the IPv4 header to include the 543 entire IPv6 header and possibly even the data beyond that. 545 If the offending packet includes enough data, the encapsulator MAY 546 extract the encapsulated IPv6 packet and use it to generate an ICMPv6 547 message directed back to the originating IPv6 node, as shown below: 549 +--------------+ 550 | IPv4 Header | 551 | dst = encaps | 552 | node | 553 +--------------+ 554 | ICMPv4 | 555 | Header | 556 - - +--------------+ 557 | IPv4 Header | 558 | src = encaps | 559 IPv4 | node | 560 +--------------+ - - 561 Packet | IPv6 | 562 | Header | Original IPv6 563 in +--------------+ Packet - 564 | Transport | Can be used to 565 Error | Header | generate an 566 +--------------+ ICMPv6 567 | | error message 568 ~ Data ~ back to the source. 569 | | 570 - - +--------------+ - - 572 ICMPv4 Error Message Returned to Encapsulating Node 574 When receiving ICMPv4 errors as above and the errors are not "packet 575 too big" it would be useful to log the error as an error related to 576 the tunnel. Also, if sufficient headers are included in the error, 577 then the originating node MAY send an ICMPv6 error of type 578 "unreachable" with code "address unreachable" to the IPv6 source. 579 (The "address unreachable" code is appropriate since, from the 580 perspective of IPv6, the tunnel is a link and that code is used for 581 link-specific errors [RFC2463]). 583 Note that when IPv4 path MTU is exceeded, and ICMPv4 errors of only 8 584 bytes of payload are generated, or ICMPv4 errors do not cause the 585 generation of ICMPv6 errors in case there is enough payload, there 586 will be at least two packet drops instead of at least one (the case 587 of a single layer of MTU discovery). Consider a case where an IPv6 588 host is connected to an IPv4/IPv6 router, which is connected to a 589 network where an ICMPv4 error about too big packet size is generated. 590 First the router needs to learn the tunnel (IPv4) MTU which causes at 591 least one packet loss, and then the host needs to learn the (IPv6) 592 MTU from the router which causes at least one packet loss. Still, in 593 all cases there can be more than one packet loss if there are 594 multiple large packets in flight at the same time. 596 3.5. IPv4 Header Construction 598 When encapsulating an IPv6 packet in an IPv4 datagram, the IPv4 599 header fields are set as follows: 601 Version: 603 4 605 IP Header Length in 32-bit words: 607 5 (There are no IPv4 options in the encapsulating 608 header.) 610 Type of Service: 612 0 unless otherwise specified. (See [RFC2983] and 613 [RFC3168] section 9.1 for issues relating to the Type- 614 of-Service byte and tunneling.) 616 Total Length: 618 Payload length from IPv6 header plus length of IPv6 and 619 IPv4 headers (i.e., IPv6 payload length plus a constant 620 60 bytes). 622 Identification: 624 Generated uniquely as for any IPv4 packet transmitted by 625 the system. 627 Flags: 629 Set the Don't Fragment (DF) flag as specified in section 630 3.2. Set the More Fragments (MF) bit as necessary if 631 fragmenting. 633 Fragment offset: 635 Set as necessary if fragmenting. 637 Time to Live: 639 Set in an implementation-specific manner, as described 640 in section 3.3. 642 Protocol: 644 41 (Assigned payload type number for IPv6). 646 Header Checksum: 648 Calculate the checksum of the IPv4 header. [RFC791] 650 Source Address: 652 IPv4 address of outgoing interface of the encapsulator 653 or an administratively specified address as described 654 below. 656 Destination Address: 658 IPv4 address of the tunnel endpoint. 660 When encapsulating the packets, the nodes must ensure that they will 661 use the source address that the tunnel peer has configured, so that 662 the source addresses are acceptable to the decapsulator. This may be 663 a problem with multi-addressed, and in particular, multi-interface 664 nodes, especially when the routing is changed from a stable 665 condition, as the source address selection may be adversely affected. 666 Therefore, it SHOULD be possible to administratively specify the 667 source address of a tunnel. 669 3.6. Decapsulation 671 When an IPv6/IPv4 host or a router receives an IPv4 datagram that is 672 addressed to one of its own IPv4 addresses, and the value of the 673 protocol field is 41, the packet is potentially part of a tunnel and 674 needs to be verified to belong to one of the configured tunnel 675 interfaces (by checking source/destination addresses), reassembled 676 (if fragmented at the IPv4 level), have the IPv4 header removed and 677 the resulting IPv6 datagram be submitted to the IPv6 layer code on 678 the node. 680 The decapsulator MUST verify that the tunnel source address is 681 correct before further processing packets, to mitigate the problems 682 with address spoofing (see section 4). This check also applies to 683 packets which are delivered to transport protocols on the 684 decapsulator. This is done by verifying that the source address is 685 the IPv4 address of the other end of a tunnel configured on the node. 686 Packets for which the IPv4 source address does not match MUST be 687 discarded and an ICMP message SHOULD NOT be generated; however, if 688 the implementation normally sends an ICMP message when receiving an 689 unknown protocol packet, such an error message MAY be sent (e.g., 690 ICMPv4 Protocol 41 Unreachable). 692 A side effect of this address verification is that the node will 693 silently discard packets with a wrong source address, and packets 694 which were received by the node but not directly addressed to it 695 (e.g., broadcast addresses). 697 In addition, the node MAY perform ingress filtering [RFC2827] on the 698 IPv4 source address, i.e., check that the packet is arriving from the 699 interface in the direction of the route towards the tunnel end-point, 700 similar to a Strict Reverse Path Forwarding (RPF) check [BCP38UPD]. 701 If done, it is RECOMMENDED that this check is disabled by default. 702 The packets caught by this check SHOULD be discarded; an ICMP message 703 SHOULD NOT be generated by default. 705 The decapsulator MUST be capable of having, on the tunnel interfaces, 706 an IPv6 MRU of at least the maximum of of 1500 bytes and the largest 707 (IPv6) interface MTU on the decapsulator. 709 The decapsulator MUST be capable of reassembling an IPv4 packet that 710 is (after the reassembly) the maximum of 1500 bytes and the largest 711 (IPv4) interface MTU on the decapsulator. The 1500 byte number is a 712 result of encapsulators that use the static MTU scheme in section 713 3.2.1, while encapsulators that use the dynamic scheme in section 714 3.2.2 can cause up to the largest interface MTU on the decapsulator 715 to be received. (Note that it is strictly the interface MTU on the 716 last IPv4 router *before* the decapsulator that matters, but for most 717 links the MTU is the same between all neighbors.) 719 This reassembly limit allows dynamic tunnel MTU determination by the 720 encapsulator to take advantage of larger IPv4 path MTUs. An 721 implementation MAY have a configuration knob which can be used to set 722 a larger value of the tunnel reassembly buffers than the above 723 number, but it MUST NOT be set below the above number. 725 The decapsulation is shown below: 727 +-------------+ 728 | IPv4 | 729 | Header | 730 +-------------+ +-------------+ 731 | IPv6 | | IPv6 | 732 | Header | | Header | 733 +-------------+ +-------------+ 734 | Transport | | Transport | 735 | Layer | ===> | Layer | 736 | Header | | Header | 737 +-------------+ +-------------+ 738 | | | | 739 ~ Data ~ ~ Data ~ 740 | | | | 741 +-------------+ +-------------+ 743 Decapsulating IPv6 from IPv4 745 When decapsulating the packet, the IPv6 header is not modified. 746 (However, see [RFC2983] and [RFC3168] section 9.1 for issues relating 747 to the Type of Service byte and tunneling.) If the packet is 748 subsequently forwarded, its hop limit is decremented by one. 750 The decapsulator performs IPv4 reassembly before decapsulating the 751 IPv6 packet. 753 The encapsulating IPv4 header is discarded. When reconstructing the 754 IPv6 packet the length MUST be determined from the IPv6 payload 755 length since the IPv4 packet might be padded (thus have a length 756 which is larger than the IPv6 packet plus the IPv4 header being 757 removed). 759 After the decapsulation the node MUST silently discard a packet with 760 an invalid IPv6 source address. The list of invalid source addresses 761 SHOULD include at least: 763 - all multicast addresses (FF00::/8) 765 - the loopback address (::1) 767 - all the IPv4-compatible IPv6 addresses [RFC3513] (::/96), 768 excluding the unspecified address for Duplicate Address 769 Detection (::/128) 771 - all the IPv4-mapped IPv6 addresses (::ffff:0:0/96) 773 In addition, the node should perform ingress filtering [RFC2827] on 774 the IPv6 source address, similar to on any of its interfaces, e.g.: 776 - if the tunnel is towards the Internet, check that the site's 777 IPv6 prefixes are not used as the source addresses, or 779 - if the tunnel is towards an edge network, check that the source 780 address belongs to that edge network. 782 3.7. Link-Local Addresses 784 The configured tunnels are IPv6 interfaces (over the IPv4 "link 785 layer") and thus MUST have link-local addresses. The link-local 786 addresses are used by, e.g., routing protocols operating over the 787 tunnels. 789 The interface identifier [RFC3513] for such an interface may be based 790 on the 32-bit IPv4 address of an underlying interface, or formed 791 using some other means, as long as it's unique from the other tunnel 792 endpoint with a reasonably high probability. 794 If an IPv4 address is used for forming the IPv6 link-local address, 795 the interface identifier is the IPv4 address, prepended by zeros. 796 Note that the "Universal/Local" bit is zero, indicating that the 797 interface identifier is not globally unique. The link-local address 798 is formed by appending the interface identifier to the prefix 799 FE80::/64. 801 When the host has more than one IPv4 address in use on the physical 802 interface concerned, an administrative choice of one of these IPv4 803 addresses is made when forming the link-local address. 805 +-------+-------+-------+-------+-------+-------+------+------+ 806 | FE 80 00 00 00 00 00 00 | 807 +-------+-------+-------+-------+-------+-------+------+------+ 808 | 00 00 00 00 | IPv4 Address | 809 +-------+-------+-------+-------+-------+-------+------+------+ 811 3.8. Neighbor Discovery over Tunnels 813 Configured tunnel implementations MUST at least accept and respond to 814 the probe packets used by Neighbor Unreachability Detection (NUD) 816 [RFC2461]. The implementations SHOULD also send NUD probe packets to 817 detect when the configured tunnel fails at which point the 818 implementation can use an alternate path to reach the destination. 819 Note that Neighbor Discovery allows that the sending of NUD probes be 820 omitted for router to router links if the routing protocol tracks 821 bidirectional reachability. 823 For the purposes of Neighbor Discovery the configured tunnels 824 specified in this document are assumed to NOT have a link-layer 825 address, even though the link-layer (IPv4) does have an address. 826 This means that: 828 - the sender of Neighbor Discovery packets SHOULD NOT include 829 Source Link Layer Address options or Target Link Layer Address 830 options on the tunnel link. 832 - the receiver MUST, while otherwise processing the Neighbor 833 Discovery packet, silently ignore the content of any Source Link 834 Layer Address options or Target Link Layer Address options 835 received on the tunnel link. 837 Not using a link layer address options is consistent with how 838 Deighbor Discovery is used on other point-to-point links. 840 4. Threat Related to Source Address Spoofing 842 The specification above contains rules that apply tunnel source 843 address verification in particular and ingress filtering 844 [RFC2827][BCP38UPD] in general to packets before they are 845 decapsulated. When IP-in-IP tunneling (independent of IP versions) 846 is used it is important that this can not be used to bypass any 847 ingress filtering in use for non-tunneled packets. Thus the rules in 848 this document are derived based on should ingress filtering be used 849 for IPv4 and IPv6, the use of tunneling should not provide an easy 850 way to circumvent the filtering. 852 In this case, without specific ingress filtering checks in the 853 decapsulator, it would be possible for an attacker to inject a packet 854 with: 856 - Outer IPv4 source: real IPv4 address of attacker 858 - Outer IPv4 destination: IPv4 address of decapsulator 860 - Inner IPv6 source: Alice which is either the decapsulator or a 861 node close to it. 863 - Inner IPv6 destination: Bob 865 Even if all IPv4 routers between the attacker and the decapsulator 866 implement IPv4 ingress filtering, and all IPv6 routers between the 867 decapsulator and Bob implement IPv6 ingress filtering, the above 868 spoofed packets will not be filtered out. As a result Bob will 869 receive a packet that looks like it was sent from Alice even though 870 the sender was some unrelated node. 872 The solution to this is to have the decapsulator only accept 873 encapsulated packets from the explicitly configured source address 874 (i.e., the other end of the tunnel) as specified in section 3.6. 875 While this does not provide complete protection in the case ingress 876 filtering has not been deployed, it does provide a significant 877 increase in security. The issue and the remainder threats are 878 discussed at more length in Security Considerations. 880 5. Security Considerations 882 An implementation of tunneling needs to be aware that while a tunnel 883 is a link (as defined in [RFC2460]), the threat model for a tunnel 884 might be rather different than for other links, since the tunnel 885 potentially includes all of the Internet. 887 Several mechanisms (e.g., Neighbor Discovery) depend on Hop Count 888 being 255 and/or the addresses being link-local for ensuring that a 889 packet originated on-link, in a semi-trusted environment. Tunnels 890 are more vulnerable to a breach of this assumption than physical 891 links, as an attacker anywhere in the Internet can send an IPv6-in- 892 IPv4 packet to the tunnel decapsulator, causing injection of an 893 encapsulted IPv6 packet to the configured tunnel interface unless the 894 decapsulation checks are able to discard packets injected in such a 895 manner. 897 Therefore, this memo specifies strict checks to mitigate this threat: 899 - IPv4 source address of the packet MUST be the same as configured 900 for the tunnel end-point, 902 - IPv4 ingress filtering MAY be implemented to check that the IPv4 903 packets are received from an expected interface, 905 - IPv6 packets with several, obviously invalid IPv6 source 906 addresses MUST be discarded (see Section 3.6 for details), and 908 - IPv6 ingress filtering should be performed, to check that the 909 IPv6 packets are received from an expected interface. 911 Especially the first verification is vital: to avoid this check, the 912 attacker must be able to know the source of the tunnel (difficult) 913 and be able to spoof it (easier). 915 If the remainder threats of tunnel source verification are considered 916 to be significant, a tunneling scheme with authentication should be 917 used instead, for example IPsec [RFC2401] (preferable) or Generic 918 Routing Encapsulation with a pre-configured secret key [RFC2890]. As 919 the configured tunnels are set up more or less manually, setting up 920 the keying material is probably not a problem. 922 If the tunneling is done inside an administrative domain, proper 923 ingress filtering at the edge of the domain can also eliminate the 924 threat from outside of the domain. Therefore shorter tunnels are 925 preferable to longer ones, possibly spanning the whole Internet. 927 Additionally, an implementation must treat interfaces to different 928 links as separate e.g. to ensure that Neighbor Discovery packets 929 arriving on one link does not effect other links. This is especially 930 important for tunnel links. 932 When dropping packets due to failing to match the allowed IPv4 source 933 addresses for a tunnel the node should not "acknowledge" the 934 existence of a tunnel, otherwise this could be used to probe the 935 acceptable tunnel endpoint addresses. For that reason, the 936 specification says that such packets MUST be discarded, and an ICMP 937 error message SHOULD NOT be generated, unless the implementation 938 normally sends ICMP destination unreachable messages for unknown 939 protocols; in such a case, the same code MAY be sent. As should be 940 obvious, the not returning the same ICMP code if an error is returned 941 for other protocols may hint that the IPv6 stack (or the protocol 41 942 tunneling processing) has been enabled -- the behaviour should be 943 consistent on how the implementation otherwise behaves to be 944 transparent to probing. 946 6. Acknowledgments 948 We would like to thank the members of the IPv6 working group, the 949 Next Generation Transition (ngtrans) working group, and the v6ops 950 working group for their many contributions and extensive review of 951 this document. Special thanks are due to Jim Bound, Ross Callon, Bob 952 Hinden, Bill Manning, John Moy, Mohan Parthasarathy, Pekka Savola, 953 Fred Templin, Chirayu Patel, and Tim Chown for many helpful 954 suggestions. Pekka Savola helped in editing the final revisions of 955 the specification. 957 7. References 959 7.1. Normative References 961 [RFC791] J. Postel, "Internet Protocol", RFC 791, September 1981. 963 [RFC1191] Mogul, J., and S. Deering., "Path MTU Discovery", RFC 1191, 964 November 1990. 966 [RFC1981] McCann, J., S. Deering, and J. Mogul. "Path MTU Discovery 967 for IP version 6", RFC 1981, August 1996. 969 [RFC2119] S. Bradner, "Key words for use in RFCs to Indicate 970 Requirement Levels", RFC 2119, March 1997. 972 [RFC2460] Deering, S., and Hinden, R. "Internet Protocol, Version 6 973 (IPv6) Specification", RFC 2460, December 1998. 975 [RFC2463] A. Conta, S. Deering, "Internet Control Message Protocol 976 (ICMPv6) for the Internet Protocol Version 6 (IPv6) 977 Specification", RFC 2463, December 1998. 979 7.2. Non-normative References 981 [ASSIGNED] IANA, "Assigned numbers online database", 982 http://www.iana.org/numbers.html 984 [BCP38UPD] Baker, F., and Savola P., "Ingress Filtering for Multihomed 985 Networks", draft-savola-bcp38-multihoming-update-03.txt, 986 work-in-progress, December 2003. 988 [DNSOPV6] Durand, A., Ihren, J., and Savola P., "Operational 989 Considerations and Issues with IPv6 DNS", draft-ietf-dnsop- 990 ipv6-dns-issues-04.txt, work-in-progress, January 2004. 992 [GTSM] Gill, V., Heasley, J., and D. Meyer, "The Generalized TTL 993 Security Mechanism (GTSM)", draft-gill-gtsh-04.txt, work- 994 in-progress, October 2003. 996 [KM97] Kent, C., and J. Mogul, "Fragmentation Considered Harmful". 997 In Proc. SIGCOMM '87 Workshop on Frontiers in Computer 998 Communications Technology. August 1987. 1000 [RFC1122] Braden, R., "Requirements for Internet Hosts - Communication 1001 Layers", STD 3, RFC 1122, October 1989. 1003 [RFC1435] S. Knowles, "IESG Advice from Experience with Path MTU 1004 Discovery", RFC 1435, March 1993. 1006 [RFC1812] F. Baker, "Requirements for IP Version 4 Routers", RFC 1812, 1007 June 1995. 1009 [RFC1812] Kent, S., Atkinson, R., "Security Architecture for the 1010 Internet Protocol", RFC 2401, November 1998. 1012 [RFC2461] Narten, T., Nordmark, E., and Simpson, W. "Neighbor 1013 Discovery for IP Version 6 (IPv6)", RFC 2461, December 1998. 1015 [RFC2462] Thomson, S., and Narten, T. "IPv6 Stateless Address 1016 Autoconfiguration," RFC 2462, December 1998. 1018 [RFC2667] D. Thaler, "IP Tunnel MIB", RFC 2667, August 1999. 1020 [RFC2827] Ferguson, P., and Senie, D., "Network Ingress Filtering: 1021 Defeating Denial of Service Attacks which employ IP Source 1022 Address Spoofing", RFC 2827, May 2000. 1024 [RFC2890] Dommety, G., "Key and Sequence Number Extensions to GRE", 1025 RFC 2890, September 2000. 1027 [RFC2923] K. Lahey, "TCP Problems with Path MTU Discovery", RFC 2923, 1028 September 2000. 1030 [RFC2983] D. Black, "Differentiated Services and Tunnels", RFC 2983, 1031 October 2000. 1033 [RFC3056] B. Carpenter, and K. Moore, "Connection of IPv6 Domains via 1034 IPv4 Clouds", RFC 3056, February 2001. 1036 [RFC3168] K. Ramakrishnan, S. Floyd, D. Black, "The Addition of 1037 Explicit Congestion Notification (ECN) to IP", RFC 3168, 1038 September 2001. 1040 [RFC3232] Reynolds, J., "Assigned Numbers: RFC 1700 is Replaced by an 1041 On-line Database", RFC 3232, January 2002. 1043 [RFC3484] R. Draves, "Default Address Selection for IPv6", RFC 3484, 1044 February 2003. 1046 [RFC3513] Hinden, R., and S. Deering, "IP Version 6 Addressing 1047 Architecture", RFC 3513, April 2003. 1049 [RFC3596] Thomson, S., C. Huitema, V. Ksinant, and M. Souissi, "DNS 1050 Extensions to support IP version 6", RFC 3596, October 2003. 1052 8. Authors' Addresses 1054 Erik Nordmark 1055 Sun Microsystems Laboratories 1056 180, avenue de l'Europe 1057 38334 SAINT ISMIER Cedex, France 1058 Tel : +33 (0)4 76 18 88 03 1059 Fax : +33 (0)4 76 18 88 88 1060 Email : erik.nordmark@sun.com 1062 Robert E. Gilligan 1063 Intransa, Inc. 1064 2870 Zanker Rd., Suite 100 1065 San Jose, CA 95134 1067 Tel : +1 408 678 8600 1068 Fax : +1 408 678 8800 1069 Email : gilligan@intransa.com, gilligan@leaf.com 1071 9. Changes from RFC 2893 1073 The motivation for the bulk of these changes are to simplify the 1074 document to only contain the mechanisms of wide-spread use. 1076 RFC 2893 contains a mechanism called automatic tunneling. But a much 1077 more general mechanism is specified in RFC 3056 [RFC3056] which gives 1078 each node with a (global) IPv4 address a /48 IPv6 prefix i.e., enough 1079 for a whole site. 1081 The following changes have been performed since RFC 2893: 1083 - Removed references to A6 and retained AAAA. 1085 - Removed automatic tunneling and use of IPv4-compatible 1086 addresses. 1088 - Removed default Configured Tunnel using IPv4 "Anycast Address" 1090 - Removed Source Address Selection section since this is now 1091 covered by another document ([RFC3484]). 1093 - Removed brief mention of 6over4. 1095 - Split into normative and non-normative references and other 1096 reference cleanup. 1098 - Dropped "or equal" in if (IPv4 path MTU - 20) is less than or 1099 equal to 1280 1101 - Dropped this: However, IPv6 may be used in some environments 1102 where interoperability with IPv4 is not required. IPv6 nodes 1103 that are designed to be used in such environments need not use 1104 or even implement these mechanisms. 1106 - Described Static MTU and Dynamic MTU cases separately; clarified 1107 that the dynamic path MTU mechanism is OPTIONAL but if it is 1108 implemented it should follow the rules in section 3.2.2. 1110 - Specified Static MTU to default to a MTU of 1280 to 1480 bytes, 1111 and that this may be configurable. Discussed the issues with 1112 using Static MTU at more length. 1114 - Specified minimal rules for IPv4 reassembly and IPv6 MRU to 1115 enhance interoperability and to minimize blacholes. 1117 - Restated the "currently underway" language about Type-of- 1118 Service, and loosely point at [RFC2983] and [RFC3168]. 1120 - Fixed reference to Assigned Numbers to be to online version 1121 (with proper pointer to "Assigned Numbers is obsolete" RFC). 1123 - Clarified text about ingress filtering e.g. that it applies to 1124 packet delivered to transport protocols on the decapsulator as 1125 well as packets being forwarded by the decapsulator, and how the 1126 decapsulator's checks help when IPv4 and IPv6 ingress filtering 1127 is in place. 1129 - Removed unidirectional tunneling; assume all tunnels are 1130 bidirectional. 1132 - Removed the guidelines for advertising addresses in DNS as 1133 slightly out of scope, referring to another document for the 1134 details. 1136 - Removed the SHOULD requirement that the link-local addresses 1137 should be formed based on IPv4 addresses. 1139 - Added a SHOULD for implementing a knob to be able to set the 1140 source address of the tunnel, and add discussion why this is 1141 useful. 1143 - Added stronger wording for source address checks: both IPv4 and 1144 IPv6 source addresses MUST be checked, and RPF-like ingress 1145 filtering is optional. 1147 - Rewrote security considerations to be more precise about the 1148 threats of tunneling. 1150 - Added a note that using TTL=255 when encapsulating might be 1151 useful for decapsulation security checks later on. 1153 - Added more discussion in Section 3.2 why using an "infinite" 1154 IPv6 MTU leads to likely interoperability problems. 1156 - Added an explicit requirement that if both MTU determination 1157 methods are used, choosing one should be possible on a per- 1158 tunnel basis. 1160 Clarified that ICMPv4 error handling is only applicable to 1161 dynamic MTU determination. 1163 - Made a lot of miscellaneous editorial cleanups. 1165 9.1. Changes from draft-ietf-v6ops-mech-v2-00 1167 [[ RFC-Editor note: remove the change history between the drafts 1168 before publication. ]] 1170 - Clarified in section 2.2 that there is no assumption that the 1171 DNS server knows the IPv4/IPv6 capabilities of the requesting 1172 node. 1174 - Clarified in section 2.2 that a filtering resolver might want to 1175 take into account external factors e.g., whether IPv6 interfaces 1176 have been configured on the node. 1178 - Clarified in section 2.3 that part of the motivation for the 1179 section is that this is the opposite of common DNS practices in 1180 IPv4; advertising unreachable IPv4 addresses in the DNS is 1181 common. 1183 - Removed the now artificial separation in a section on "common 1184 tunneling techniques" and "configured tunneling" to make one 1185 section on "configured tunneling". 1187 - Restructured the section on tunnel MTU to make the relationship 1188 between static tunnel MTU and dynamic tunnel MTU more clear. 1189 This includes fixing the unclear language about "must be 1280 1190 but may be configurable". 1192 - Added warning about manually configuring large tunnel MTUs 1193 causing excessive fragmentation. 1195 - Added warning about IPv4 PMTU blackholes when using dynamic MTU. 1197 - Clarified that when decapsulating the receiver must be liberal 1198 and allow for padding of the encapsulated packet. 1200 - Added example that when reflecting ICMPv4 errors as ICMPv6 1201 errors it would be appropriate to use ICMPv6 unreachable type 1202 with code "address unreachable" since an error from inside the 1203 tunnel is in effect a link specific problem from IPv6's 1204 perspective. 1206 - Consolidated the text on ingress filtering and created a 1207 separate section on the threat related to source address 1208 spoofing through open decapsulators. 1210 - Clarified "martian" filtering as follows: 0.0.0.0 should be 1211 0.0.0.0/8, same for 127. (per RFC1812), and elaborated that the 1212 broadcast address check includes both the 255.255.255.255 1213 address and all the broadcast addresses of the decapsulator. 1215 - Clarified that packets which fail the checks (such as verifying 1216 the IPv4 source address, martian, and ingress filtering) on the 1217 decapsulator should be silently dropped. 1219 - Clarified that while source link layer address options and 1220 target link layer address options are ignored in received ND 1221 packets, the ND packets themselves are processed as normal. 1223 9.2. Changes from draft-ietf-v6ops-mech-v2-01 1225 - Removed unidirectional tunnels; assume all the tunnels are 1226 bidirectional. 1228 - Removed the definition of IPv4-compatible IPv6 addresses. 1230 - Removed redundant text in the Hop Limit processing rules. 1232 - Removed the guidelines for advertising addresses in DNS as 1233 slightly out of scope, referring to another document for the 1234 details. 1236 - Removed the SHOULD requirement that the link-local addresses 1237 should be formed based on IPv4 addresses. 1239 - Added more discussion on the ICMPv4/6 Path MTU Discovery and the 1240 required number of packet drops. 1242 - Added a SHOULD for implementing a knob to be able to set the 1243 source address of the tunnel, and add discussion why this is 1244 useful. 1246 - Added stronger wording for source address checks: both IPv4 and 1247 IPv6 source addresses MUST be checked, and RPF-like ingress 1248 filtering is optional. 1250 - Rewrote security considerations to be more precise about the 1251 threats of tunneling. 1253 - Added a note that using TTL=255 when encapsulating might be 1254 useful for decapsulation security checks later on. 1256 - Added more discussion in Section 3.2 why using an "infinite" 1257 IPv6 MTU leads to likely interoperability problems. 1259 - Added an explicit requirement that if both MTU determination 1260 methods are used, choosing one should be possible on a per- 1261 tunnel basis. 1263 Clarified that ICMPv4 error handling is only applicable to 1264 dynamic MTU determination. 1266 - Specified Static MTU to default to a MTU of 1280 to 1480 bytes, 1267 and that this may be configurable. Discussed the issues with 1268 using Static MTU at more length. 1270 - Specified minimal rules for IPv4 reassembly and IPv6 MRU to 1271 enhance interoperability and to minimize blacholes. 1273 - 1275 - Made a lot of miscellaneous editorial cleanups.