idnits 2.17.1 draft-bagnulo-mptcp-attacks-01.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- == There are 3 instances of lines with non-RFC2606-compliant FQDNs in the document. ** The document seems to lack a both a reference to RFC 2119 and the recommended RFC 2119 boilerplate, even if it appears to use RFC 2119 keywords. RFC 2119 keyword, line 630: '...n other words, MPTCP MUST work through...' RFC 2119 keyword, line 686: '... MUST be used with a preconfigure...' Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (October 1, 2013) is 3859 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Missing Reference: 'RFC4895' is mentioned on line 685, but not defined ** Obsolete normative reference: RFC 793 (Obsoleted by RFC 9293) ** Obsolete normative reference: RFC 6528 (Obsoleted by RFC 9293) -- Obsolete informational reference (is this intentional?): RFC 6824 (Obsoleted by RFC 8684) -- Obsolete informational reference (is this intentional?): RFC 4960 (Obsoleted by RFC 9260) -- Obsolete informational reference (is this intentional?): RFC 3775 (Obsoleted by RFC 6275) == Outdated reference: A later version (-04) exists of draft-bittau-tcp-crypt-03 Summary: 3 errors (**), 0 flaws (~~), 4 warnings (==), 4 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group M. Bagnulo 3 Internet-Draft UC3M 4 Intended status: Informational C. Paasch 5 Expires: April 4, 2014 UCLouvain 6 F. Gont 7 SI6 Networks / UTN-FRH 8 O. Bonaventure 9 UCLouvain 10 C. Raiciu 11 UPB 12 October 1, 2013 14 Analysis of MPTCP residual threats and possible fixes 15 draft-bagnulo-mptcp-attacks-01 17 Abstract 19 This documents performs an analysis of the residual threats for MPTCP 20 and explores possible solutions to them. 22 Status of this Memo 24 This Internet-Draft is submitted in full conformance with the 25 provisions of BCP 78 and BCP 79. 27 Internet-Drafts are working documents of the Internet Engineering 28 Task Force (IETF). Note that other groups may also distribute 29 working documents as Internet-Drafts. The list of current Internet- 30 Drafts is at http://datatracker.ietf.org/drafts/current/. 32 Internet-Drafts are draft documents valid for a maximum of six months 33 and may be updated, replaced, or obsoleted by other documents at any 34 time. It is inappropriate to use Internet-Drafts as reference 35 material or to cite them other than as "work in progress." 37 This Internet-Draft will expire on April 4, 2014. 39 Copyright Notice 41 Copyright (c) 2013 IETF Trust and the persons identified as the 42 document authors. All rights reserved. 44 This document is subject to BCP 78 and the IETF Trust's Legal 45 Provisions Relating to IETF Documents 46 (http://trustee.ietf.org/license-info) in effect on the date of 47 publication of this document. Please review these documents 48 carefully, as they describe your rights and restrictions with respect 49 to this document. Code Components extracted from this document must 50 include Simplified BSD License text as described in Section 4.e of 51 the Trust Legal Provisions and are provided without warranty as 52 described in the Simplified BSD License. 54 Table of Contents 56 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 57 2. ADD_ADDR attack . . . . . . . . . . . . . . . . . . . . . . . 5 58 2.1. Possible security enhancements to prevent this attack . . 10 59 3. DoS attack on MP_JOIN . . . . . . . . . . . . . . . . . . . . 12 60 3.1. Possible security enhancements to prevent this attack . . 13 61 4. SYN flooding amplification . . . . . . . . . . . . . . . . . . 14 62 4.1. Possible security enhancements to prevent this attack . . 14 63 5. Eavesdropper in the initial handshake . . . . . . . . . . . . 15 64 5.1. Possible security enhancements to prevent this attack . . 15 65 6. SYN/JOIN attack . . . . . . . . . . . . . . . . . . . . . . . 16 66 6.1. Possible security enhancements to prevent this attack . . 16 67 7. Reccomendation . . . . . . . . . . . . . . . . . . . . . . . . 17 68 8. Security considerations . . . . . . . . . . . . . . . . . . . 19 69 9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 20 70 10. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 21 71 11. References . . . . . . . . . . . . . . . . . . . . . . . . . . 22 72 11.1. Normative References . . . . . . . . . . . . . . . . . . . 22 73 11.2. Informative References . . . . . . . . . . . . . . . . . . 22 74 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 24 76 1. Introduction 78 This document provides a complement to the threat analysis for 79 Multipath TCP (MPTCP) [RFC6824] documented in RFC 6181 [RFC6181]. 80 RFC 6181 provided a threat analysis for the general solution space of 81 extending TCP to operate with multiple IP addresses per connection. 82 Its main goal was to leverage previous experience acquired during the 83 design of other multi-address protocols, notably SHIM6 [RFC5533], 84 SCTP [RFC4960] and MIPv6 [RFC3775] during the design of MPTCP. Thus, 85 RFC 6181 was produced before the actual MPTCP specification (RFC6824) 86 was completed, and documented a set of recommendations that were 87 considered during the production of such specification. 89 This document complements RFC 6181 with a vulnerability analysis of 90 the specific mechanisms specified in RFC 6824. The motivation for 91 this analysis is to identify possible security issues with MPTCP as 92 currently specified and propose security enhancements to address the 93 identified security issues. 95 The goal of the security mechanisms defined in RFC 6824 were to make 96 MPTCP no worse than currently available single-path TCP. We believe 97 that this goal is still valid, so we will perform our analysis on the 98 same grounds. 100 Types of attackers: for all attacks considered in this document, we 101 identify the type of attacker. We can classify the attackers based 102 on their location as follows: 104 o Off-path attacker. This is an attacker that does not need to be 105 located in any of the paths of the MPTCP session at any point in 106 time during the lifetime of the MPTCP session. This means that 107 the Off-path attacker cannot eavesdrop any of the packets of the 108 MPTCP session. 110 o Partial time On-path attacker. This is an attacker that needs to 111 be in at least one of the paths during part but not during the 112 entire lifetime of the MPTCP session. The attacker can be in the 113 forward and/or backward directions, for the initial subflow and/or 114 other subflows. The specific needs of the attacker will be made 115 explicit in the attack description. 117 o On-path attacker. This attacker needs to be on at least one of 118 the paths during the whole duration of the MPTCP session. The 119 attacker can be in the forward and/or backward directions, for the 120 initial subflow and/or other subflows. The specific needs of the 121 attacker will be made explicit in the attack description. 123 We can also classify the attackers based on their actions as follows: 125 o Eavesdropper. The attacker is able to capture some of the packets 126 of the MPTCP session to perform the attack, but it is not capable 127 of changing, discarding or delaying any packet of the MPTCP 128 session. The attacker can be in the forward and/or backward 129 directions, for the initial subflow and/or other subflows. The 130 specific needs of the attacker will be made explicit in the attack 131 description. 133 o Active attacker. The attacker is able to change, discard or delay 134 some of the packets of the MPTCP session. The attacker can be in 135 the forward and/or backward directions, for the initial subflow 136 and/or other subflows. The specific needs of the attacker will be 137 made explicit in the attack description. 139 In this document, we consider the following possible combinations of 140 attackers: 142 o an On-path eavesdropper 144 o an On-path active attacker 146 o an Off-path active attacker 148 o a Partial-time On-path eavesdropper 150 o a Partial-time On-path active attacker 152 In the rest of the document we describe different attacks that are 153 possible against the MPTCP protocol specified in RFC6824 and we 154 propose possible security enhancements to address them. 156 2. ADD_ADDR attack 158 Summary of the attack: 160 Type of attack: MPTCP session hijack enabling Man-in-the-Middle. 162 Type of attacker: Off-path, active attacker. 164 Threat: Medium 166 Description: 168 In this attack, the attacker uses the ADD_ADDR option defined in 169 RFC6824 to hijack an ongoing MPTCP session and enables himself to 170 perform a Man-in-the-Middle attack on the MPTCP session. 172 Consider the following scenario. Host A with address IPA has one 173 MPTCP session with Host B with address IPB. The MPTCP subflow 174 between IPA and IPB is using port PA on host A and port PB on host B. 175 The tokens for the MPTCP session are TA and TB for Host A and Host B 176 respectively. Host C is the attacker. It owns address IPC. The 177 attack is executed as follows: 179 1. Host C sends a forged packet with source address IPA, destination 180 address IPB, source port PA and destination port PB. The packet 181 has the ACK flag set. The TCP sequence number for the segment is 182 i and the ACK sequence number is j. We will assume all these are 183 valid, we discuss what the attacker needs to figure these ones 184 later on. The packet contains the ADD_ADDR option. The ADD_ADDR 185 option announces IPC as an alternative address for the 186 connection. It also contains an eight bit address identifier 187 which does not bring any strong security benefit. 189 2. Host B receives the ADD_ADDR message and it replies by sending a 190 TCP SYN packet. (Note: the MPTCP specification states that the 191 host receiving the ADD_ADDR option may initiate a new subflow. 192 If the host is configured so that it does not initiate a new 193 subflow the attack will not succeed. For example, on the Linux 194 implementation, the server does not create subflows. Only the 195 client does so.) The source address for the packet is IPB, the 196 destination address for the packet is IPC, the source port is PB' 197 and the destination port is PA' (It is not required that PA=PA' 198 nor that PB=PB'). The sequence number for this packet is the new 199 initial sequence number for this subflow. The ACK sequence 200 number is not relevant as the ACK flag is not set. The packet 201 carries an MP_JOIN option and it carries the token TA. It also 202 carries a random nonce generated by Host B called RB. 204 3. Host C receives the SYN+MP_JOIN packet from Host B, and it alters 205 it in the following way. It changes the source address to IPC 206 and the destination address to IPA. It sends the modified packet 207 to Host A, impersonating Host B. 209 4. Host A receives the SYN+MP_JOIN message and it replies with a 210 SYN/ACK+MP_JOIN message. The packet has source address IPA and 211 destination address IPC, as well as all the other needed 212 parameters. In particular, Host A computes a valid HMAC and 213 places it in the MP_JOIN option. 215 5. Host C receives the SYN/ACK+MP_JOIN message and it changes the 216 source address to IPC and the destination address to IPB. It 217 sends the modified packet to IPB impersonating Host A. 219 6. Host B receives the SYN/ACK+MP_JOIN message. Host B verifies the 220 HMAC of the MP_JOIN option and confirms its validity. It replies 221 with an ACK+MP_JOIN packet. The packet has source address IPB 222 and destination address IPC, as well as all the other needed 223 parameters. The returned MP_JOIN option contains a valid HMAC 224 computed by Host B. 226 7. Host C receives the ACK+MP_JOIN message from B and it alters it 227 in the following way. It changes the source address to IPC and 228 the destination address to IPA. It sends the modified packet to 229 Host A impersonating Host B. 231 8. Host A receives the ACK+MP_JOIN message and creates the new 232 subflow. 234 At this point the attacker has managed to place itself as a 235 MitM for one subflow for the existing MPTCP session. It 236 should be noted that there still exists the subflow between 237 address IPA and IPB that does not flow through the attacker, 238 so the attacker has not completely intercepted all the packets 239 in the communication (yet). If the attacker wishes to 240 completely intercept the MPTCP session it can do the following 241 additional step. 243 9. Host C sends two TCP RST messages. One TCP RST packet is sent to 244 Host B, with source address IPA and destination address IPB and 245 source and destination ports PA and PB, respectively. The other 246 TCP RST message is sent to Host A, with source address IPB and 247 destination address IPA and source and destination ports PB and 248 PA, respectively. Both RST messages must contain a valid 249 sequence number. Note that figuring the sequence numbers to be 250 used here for subflow A is the same difficulty as being able to 251 send the initial ADD_ADDR option with valid Sequence number and 252 ACK value. If there are more subflows, then the attacker needs 253 to find the Sequence Number and ACK for each subflow. 255 At this point the attacker has managed to fully hijack the 256 MPTCP session. 258 Information required by the attacker to perform the described attack: 260 In order to perform this attack the attacker needs to guess or know 261 the following pieces of information: (The attacker need this 262 information for one of the subflows belonging to the MPTCP session.) 264 o the four-tuple {Client-side IP Address, Client-side Port, Server- 265 side Address, Servcer-side Port} that identifies the target TCP 266 connection 268 o a valid sequence number for the subflow 270 o a valid ACK sequence number for the subflow 272 o a valid address identifier for IPC 274 TCP connections are uniquely identified by the four-tuple {Source 275 Address, Source Port, Destination Address, Destination Port}. Thus, 276 in order to attack a TCP connection, an attacker needs to know or be 277 able to guess each of the values in that four-tuple. Assuming the 278 two peers of the target TCP connection are known, the Source Address 279 and the Destination Address can be assumed to be known. 281 We note that in order to be able to successfully perform this 282 attack, the attacker needs to be able to send packets with a 283 forged source address. This means that the attacker cannot be 284 located in a network where techniques like ingress filtering 285 [RFC2827] or source address validation [I-D.ietf-savi-framework] 286 are deployed. However, ingress filtering is not as widely 287 implemented as one would expect, and hence cannot be relied upon 288 as a mitigation for this kind of attack. 290 Assuming the attacker knows the application protocol for which the 291 TCP connection is being employed, the server-side port can also be 292 assumed to be known. Finally, the client-side port will generally 293 not be known, and will need to be guessed by the attacker. The 294 chances of an attacker guessing the client-side port will depend on 295 the ephemeral port range employed by the client, and whether the 296 client implements port randomization [RFC6056]. 298 Assuming TCP sequence number randomization is in place (see e.g. 299 [RFC6528]), an attacker would have to blindly guess a valid TCP 300 sequence number. That is, 302 RCV.NXT =< SEG.SEQ < RCV.NXT+RCV.WND or RCV.NXT =< SEG.SEQ+ 303 SEG.LEN-1 < RCV.NXT+RCV.WND 305 As a result, the chances of an attacker to succeed will depend on the 306 TCP receive window size at the target TCP peer. 308 We note that automatic TCP buffer tuning mechanisms have been 309 become common for popular TCP implementations, and hence very 310 large TCP window sizes of values up to 2 MB could end up being 311 employed by such TCP implementations. 313 According to [RFC0793], the Acknowledgement Number is considered 314 valid as long as it does not acknowledge the receipt of data that has 315 not yet been sent. That is, the following expression must be true: 317 SEG.ACK <= SND.NXT 319 However, for implementations that support [RFC5961], the following 320 (stricter) validation check is enforced: 322 SND.UNA - SND.MAX.WND <= SEG.ACK <= SND.NXT 324 Finally, in order for the address identifier to be valid, the only 325 requirement is that it needs to be different than the ones already 326 being used by Host A in that MPTCP session, so a random identifier is 327 likely to work. 329 Given that a large number of factors affect the chances of an 330 attacker of successfully performing the aforementioned off-path 331 attacks, we provide two general expressions for the expected number 332 of packets the attacker needs to send to succeed in the attack: one 333 for MPTCP implementations that support [RFC5961], and another for 334 MPTCP implementations that do not. 336 Implementations that do not support RFC 5961 338 Packets = (2^32/(RCV_WND)) * 2 * EPH_PORT_SIZE/2 * 1/MSS 340 Where the new : 342 Packets: 343 Maximum number of packets required to successfully perform an off- 344 path (blind) attack. 346 RCV_WND: 347 TCP receive window size (RCV.WND) at the target node. 349 EPH_PORT_SIZE: 350 Number of ports comprising the ephemeral port range at the 351 "client" system. 353 MSS: 354 Maximum Segment Size, assuming the attacker will send full 355 segments to maximize the chances to get a hit. 357 Notes: 358 The value "2^32" represents the size of the TCP sequence number 359 space. 360 The value "2" accounts for 2 different ACK numbers (separated by 361 2^31) that should be employed to make sure the ACK number is 362 valid. 364 The following table contains some sample results for the number of 365 required packets, based on different values of RCV_WND and 366 EPH_PORT_SIZE for a MSS of 1500 bytes. 368 +-------------+---------+---------+--------+---------+ 369 | Ports \ Win | 16 KB | 128 KB | 256 KB | 2048 KB | 370 +-------------+---------+---------+--------+---------+ 371 | 4000 | 699050 | 87381 | 43690 | 5461 | 372 +-------------+---------+---------+--------+---------+ 373 | 10000 | 1747626 | 218453 | 109226 | 13653 | 374 +-------------+---------+---------+--------+---------+ 375 | 50000 | 8738133 | 1092266 | 546133 | 68266 | 376 +-------------+---------+---------+--------+---------+ 378 Table 1: Max. Number of Packets for Successful Attack 380 Implementations that do support RFC 5961 382 Packets = (2^32/(RCV_WND)) * (2^32/(2 * SND_MAX_WND)) * 383 EPH_PORT_SIZE/2 * 1/MSS 385 Where: 387 Packets: 388 Maximum number of packets required to successfully perform an off- 389 path (blind) attack. 391 RCV_WND: 392 TCP receive window size (RCV.WND) at the target MPTCP endpoint. 394 SND_MAX_WND: 395 Maximum TCP send window size ever employed by the target MPTCP 396 end-point (SND.MAX.WND). 398 EPH_PORT_SIZE: 399 Number of ports comprising the ephemeral port range at the 400 "client" system. 402 Notes: 403 The value "2^32" represents the size of the TCP sequence number 404 space. 405 The parameter "SND_MAX_WND" is specified in [RFC5961]. 406 The value "2*SND_MAX_WND" results from the expresion "SND.NXT - 407 SND.UNA - MAX.SND.WND", assuming that, for connections that 408 perform bulk data transfers, "SND.NXT - SND.UNA == MAX.SND.WND". 409 If an attacker targets a TCP endpoint that is not actively 410 transferring data, "2 * SND_MAX_WND" would become "SND_MAX_WND" 411 (and hence a successful attack would typically require more 412 packets). 414 The following table contains some sample results for the number of 415 required packets, based on different values of RCV_WND, SND_MAX_WND, 416 and EPH_PORT_SIZE. For these implementations, only a limited number 417 of sample results are provided, just as an indication of how 418 [RFC5961] increases the difficulty of performing these attacks. 420 +-------------+-------------+-----------+-----------+---------+ 421 | Ports \ Win | 16 KB | 128 KB | 256 KB | 2048 KB | 422 +-------------+-------------+-----------+-----------+---------+ 423 | 4000 | 45812984490 | 715827882 | 178956970 | 2796202 | 424 +-------------+-------------+-----------+-----------+---------+ 426 Table 2: Max. Number of Packets for Successful Attack 428 Note: 429 In the aforementioned table, all values are computed with RCV_WND 430 equal to SND_MAX_WND. 432 2.1. Possible security enhancements to prevent this attack 434 1. To include the token of the connection in the ADD_ADDR option. 435 This would make it harder for the attacker to launch the attack, 436 since he needs to either eavesdrop the token (so this can no 437 longer be a blind attack) or to guess it, but a random 32 bit 438 number is not so easy to guess. However, this would imply that 439 any eavesdropper that is able to see the token, would be able to 440 launch this attack. This solution then increases the 441 vulnerability window against eavesdroppers from the initial 3-way 442 handshake for the MPTCP session to any exchange of the ADD_ADDR 443 messages. 445 2. To include the HMAC of the address contained in the ADD_ADDR 446 option concatenated with the key of the receiver of the ADD-ADDR 447 message. This makes it much more secure, since it requires the 448 attacker to have both keys (either by eavesdropping it in the 449 first exchange or by guessing it). Because this solution relies 450 on the key used in the MPTCP session, the protection of this 451 solution would increase if new key generation methods are defined 452 for MPTCP (e.g. using SSL keys as has been proposed). 454 3. To include the destination address of the ADD_ADDR message in the 455 HMAC. This would certainly make the attack harder (the attacker 456 would need to know the key). It wouldn't allow hosts behind NATs 457 to be reached by an address in the ADD_ADDR option, even with 458 static NAT bindings (like a web server at home). Probably it 459 would make sense to combine it with option 2) (i.e. to have the 460 HMAC of the address in the ADD_ADDR option and the destination 461 address of the packet). 463 4. To include the destination address of the SYN packet in the HMAC 464 of the MP_JOIN message. This has the same problems than option 465 3) in the presence of NATs. 467 3. DoS attack on MP_JOIN 469 Summary of the attack: 471 Type of attack: MPTCP Denial-of-Service attack, preventing the 472 hosts from creating new subflows. 474 Type of attacker: Off-path, active attacker 476 Threat: Low (? - as it is hard to guess the 32-bit token and still 477 then the attacker only prevents the creation of new subflows) 479 Description: 481 As currently specified, the initial SYN+MP_JOIN message of the 3-way 482 handshake for additional subflows creates state in the host receiving 483 the message. This, because the SYN+MP_JOIN contains the 32-bit token 484 that allows the receiver to identify the MPTCP-session and the 32-bit 485 random nonce, used in the HMAC calculation. As this information is 486 not resent in the third ACK of the 3-way handshake, a host must 487 create state upon reception of a SYN+MP_JOIN. 489 Assume that there exists an MPTCP-session between host A and host B, 490 with token Ta and Tb. An attacker, sending a SYN+MP_JOIN to host B, 491 with the valid token Tb, will trigger the creation of state on host 492 B. The number of these half-open connections a host can store per 493 MPTCP-session is limited by a certain number, and it is 494 implementation-dependent. The attacker can simply exhaust this limit 495 by sending multiple SYN+MP_JOINs with different 5-tuples. The 496 (possibly forged) source address of the attack packets will typically 497 correspond to an address that is not in use, or else the SYN/ACK sent 498 by Host B would elicit a RST from the impersonated node, thus 499 removing the corresponding state at Host B. Further discussion of 500 traditional SYN-flod attacks and common mitigations can be found in 501 [RFC4987] 503 This effectively prevents the host A from sending any more SYN+ 504 MP_JOINs to host B, as the number of acceptable half-open connections 505 per MPTCP-session on host B has been exhausted. 507 The attacker needs to know the token Tb in order to perform the 508 described attack. This can be achieved if it is a partial on-time 509 eavesdropper, observing the 3-way handshake of the establishment of 510 an additional subflow between host A and host B. If the attacker is 511 never on-path, it has to guess the 32-bit token. 513 3.1. Possible security enhancements to prevent this attack 515 The third packet of the 3-way handshake could be extended to contain 516 also the 32-bit token and the random nonce that has been sent in the 517 SYN+MP_JOIN. Further, host B will have to generate its own random 518 nonce in a reproducible fashion (e.g., a Hash of the 5-tuple + 519 initial sequence-number + local secret). This will allow host B to 520 reply to a SYN+MP_JOIN without having to create state. Upon the 521 reception of the third ACK, host B can then verify the correctness of 522 the HMAC and create the state. 524 4. SYN flooding amplification 526 Summary of the attack: 528 Type of attack: The attacker can use the SYN+MP_JOIN messages to 529 amplify the SYN flooding attack. 531 Type of attacker: Off-path, active attacker 533 Threat: Medium 535 Description: 537 SYN flooding attacks [RFC4987] use SYN messages to exhaust the 538 server's resources and prevent new TCP connections. A common 539 mitigation is the use of SYN cookies [RFC4987] that allow the 540 stateless processing of the initial SYN message. 542 With MPTCP, the initial SYN can be processed in a stateless fashion 543 using the aforementioned SYN cookies. However, as we described in 544 the previous section, as currently specified, the SYN+MP_JOIN 545 messages are not processed in a stateless manner. This opens a new 546 attack vector. The attacker can now open a MPTCP session by sending 547 a regular SYN and creating the associated state but then send as many 548 SYN+MP_JOIN messages as supported by the server with different source 549 address source port combinations, consuming server's resources 550 without having to create state in the attacker. This is an 551 amplification attack, where the cost on the attacker side is only the 552 cost of the state associated with the initial SYN while the cost on 553 the server side is the state for the initial SYN plus all the state 554 associated to all the following SYN+MP_JOIN. 556 4.1. Possible security enhancements to prevent this attack 558 1. The solution described for the previous DoS attack on MP_JOIN 559 would also prevent this attack. 561 2. Limiting the number of half open subflows to a low number (like 562 3) would also limit the impact of this attack. 564 5. Eavesdropper in the initial handshake 566 Summary of the attack 568 Type of attack: An eavesdropper present in the initial handshake 569 where the keys are exchanged can hijack the MPTCP session at any 570 time in the future. 572 Type of attacker: a Partial-time On-path eavesdropper 574 Threat: Low 576 Description: 578 In this case, the attacker is present along the path when the initial 579 3-way handshake takes place, and therefore is able to learn the keys 580 used in the MPTCP session. This allows the attacker to move away 581 from the MPTCP session path and still be able to hijack the MPTCP 582 session in the future. This vulnerability was readily identified at 583 the moment of the design of the MPTCP security solution and the 584 threat was considered acceptable. 586 5.1. Possible security enhancements to prevent this attack 588 There are many techniques that can be used to prevent this attack and 589 each of them represents different tradeoffs. At this point, we limit 590 ourselves to enumerate them and provide useful pointers. 592 1. Use of hash-chains. The use of hash chains for MPTCP has been 593 explored in [hash-chains] 595 2. Use of SSL keys for MPTCP security as described in 596 [I-D.paasch-mptcp-ssl] 598 3. Use of Cryptographically-Generated Addresses (CGAs) for MPTCP 599 security. CGAs [RFC3972] have been used in the past to secure 600 multi addressed protocols like SHIM6 [RFC5533]. 602 4. Use of TCPCrypt [I-D.bittau-tcp-crypt] 604 5. Use DNSSEC. DNSSEC has been proposed to secure the Mobile IP 605 protocol [dnssec] 607 6. SYN/JOIN attack 609 Summary of the attack 611 Type of attack: An attacker that can intercept the SYN/JOIN 612 message can alter the source address being added. 614 Type of attacker: a Partial-time On-path eavesdropper 616 Threat: Low 618 Description: 620 The attacker is present along the path when the SYN/JOIN exchange 621 takes place, and this allows the attacker to add any new address it 622 wants to by simply substituting the source address of the SYN/JOIN 623 packet for one it chooses. This vulnerability was readily identified 624 at the moment of the design of the MPTCP security solution and the 625 threat was considered acceptable. 627 6.1. Possible security enhancements to prevent this attack 629 It should be noted that this vulnerability is fundamental due to the 630 NAT support requirement. In other words, MPTCP MUST work through 631 NATS in order to be deployable in the current Internet. NAT behavior 632 is unfortunately indistinguishable from this attack. It is 633 impossible to secure the source address, since doing so would prevent 634 MPTCP to work though NATS. This basically implies that the solution 635 cannot rely on securing the address. A more promising approach would 636 be then to look into securing the payload exchanged, limiting the 637 impact that the attack would have in the communication (e.g. 638 TCPCrypt or similar). 640 7. Reccomendation 642 Current MPTCP specification [RFC6824] is experimental. There is an 643 ongoing effort to move it to Standards track. We believe that the 644 work on MPTCP security should follow two treads: 646 o The work on improving MPTCP security so that is enough to become a 647 Standard Track document. 649 o The work on analyzing possible additional security enhancements to 650 provide a more secure version of MPTCP. 652 We will expand on these two next. 654 MPTCP security for a Standard Track specification. 656 We believe that in order for MPTCP to progress to Standard Track, the 657 ADD-ADDR attack must be addressed. We believe that the solution that 658 should be adopted in order to deal with this attack is to include an 659 HMAC to the ADD ADDR message (with the address being added used as 660 input to the HMAC, as well as the key). This would make the ADD ADDR 661 message as secure as the JOIN message. In addition, this implies 662 that if we implement a more secure way to create the key used in the 663 MPTCP connection, it can be used to improve the security of both the 664 JOIN and the ADD ADDR message automatically (since both use the same 665 key in the HMAC). 667 We believe that this is enough for MPTCP to progress as a Standard 668 track document. This is so because the security level is similar to 669 single path TCP, as results from our previous analysis. Moreover, 670 the security level achieved with these changes is exactly the same as 671 other Standard Track documents. In particular, this would be the 672 same security level that SCPT with dynamic addresses as defined in 673 [RFC5061]. The Security Consideration section of RFC5061 (which is a 674 Standard Track document) reads: 676 The addition and or deletion of an IP address to an existing 677 association does provide an additional mechanism by which existing 678 associations can be hijacked. Therefore, this document requires 679 the use of the authentication mechanism defined in [RFC4895] to 680 limit the ability of an attacker to hijack an association. 681 Hijacking an association by using the addition and deletion of an 682 IP address is only possible for an attacker who is able to 683 intercept the initial two packets of the association setup when 684 the SCTP-AUTH extension is used without pre-shared keys. If such 685 a threat is considered a possibility, then the [RFC4895] extension 686 MUST be used with a preconfigured shared endpoint pair key to 687 mitigate this threat. 689 This is the same security level that would be achieved by MPTCP plus 690 the ADD ADDR security measure recommended. 692 Security enhancements for MPTCP 694 We also believe that is worthwhile exploring alternatives to secure 695 MPTCP. As we identified earlier, the problem is securing JOIN 696 messages is fundamentally incompatible with NAT support, so it is 697 likely that a solution to this problem involves the protection of the 698 data itself. Exploring the integration of MPTCP and approaches like 699 TCPCrypt seems a promising venue. 701 8. Security considerations 703 This whole document is about security considerations for MPTCP. 705 9. IANA Considerations 707 There are no IANA considerations in this memo. 709 10. Acknowledgments 711 We would like to thank Mark Handley for his comments on the attacks 712 and countermeasures discussed in this document. 714 11. References 716 11.1. Normative References 718 [RFC0793] Postel, J., "Transmission Control Protocol", STD 7, 719 RFC 793, September 1981. 721 [RFC3972] Aura, T., "Cryptographically Generated Addresses (CGA)", 722 RFC 3972, March 2005. 724 [RFC5961] Ramaiah, A., Stewart, R., and M. Dalal, "Improving TCP's 725 Robustness to Blind In-Window Attacks", RFC 5961, 726 August 2010. 728 [RFC6056] Larsen, M. and F. Gont, "Recommendations for Transport- 729 Protocol Port Randomization", BCP 156, RFC 6056, 730 January 2011. 732 [RFC6528] Gont, F. and S. Bellovin, "Defending against Sequence 733 Number Attacks", RFC 6528, February 2012. 735 [RFC5061] Stewart, R., Xie, Q., Tuexen, M., Maruyama, S., and M. 736 Kozuka, "Stream Control Transmission Protocol (SCTP) 737 Dynamic Address Reconfiguration", RFC 5061, 738 September 2007. 740 11.2. Informative References 742 [RFC6824] Ford, A., Raiciu, C., Handley, M., and O. Bonaventure, 743 "TCP Extensions for Multipath Operation with Multiple 744 Addresses", RFC 6824, January 2013. 746 [RFC6181] Bagnulo, M., "Threat Analysis for TCP Extensions for 747 Multipath Operation with Multiple Addresses", RFC 6181, 748 March 2011. 750 [RFC5533] Nordmark, E. and M. Bagnulo, "Shim6: Level 3 Multihoming 751 Shim Protocol for IPv6", RFC 5533, June 2009. 753 [RFC4960] Stewart, R., "Stream Control Transmission Protocol", 754 RFC 4960, September 2007. 756 [RFC3775] Johnson, D., Perkins, C., and J. Arkko, "Mobility Support 757 in IPv6", RFC 3775, June 2004. 759 [RFC2827] Ferguson, P. and D. Senie, "Network Ingress Filtering: 760 Defeating Denial of Service Attacks which employ IP Source 761 Address Spoofing", BCP 38, RFC 2827, May 2000. 763 [I-D.ietf-savi-framework] 764 Wu, J., Bi, J., Bagnulo, M., Baker, F., and C. Vogt, 765 "Source Address Validation Improvement Framework", 766 draft-ietf-savi-framework-06 (work in progress), 767 January 2012. 769 [RFC4987] Eddy, W., "TCP SYN Flooding Attacks and Common 770 Mitigations", RFC 4987, August 2007. 772 [I-D.paasch-mptcp-ssl] 773 Paasch, C. and O. Bonaventure, "Securing the MultiPath TCP 774 handshake with external keys", draft-paasch-mptcp-ssl-00 775 (work in progress), October 2012. 777 [I-D.bittau-tcp-crypt] 778 Bittau, A., Boneh, D., Hamburg, M., Handley, M., Mazieres, 779 D., and Q. Slack, "Cryptographic protection of TCP Streams 780 (tcpcrypt)", draft-bittau-tcp-crypt-03 (work in progress), 781 September 2012. 783 [hash-chains] 784 Diez, J., Bagnulo, M., Valera, F., and I. Vidal, "Security 785 for multipath TCP: a constructive approach", International 786 Journal of Internet Protocol Technology 6, 2011. 788 [dnssec] Kukec, A., Bagnulo, M., Ayaz, S., Bauer, C., and W. Eddy, 789 "OAM-DNSSEC: Route Optimization for Aeronautical Mobility 790 using DNSSEC", 4th ACM International Workshop on Mobility 791 in the Evolving Internet Architecture MobiArch 2009, 2009. 793 Authors' Addresses 795 Marcelo Bagnulo 796 Universidad Carlos III de Madrid 797 Av. Universidad 30 798 Leganes, Madrid 28911 799 SPAIN 801 Phone: 34 91 6249500 802 Email: marcelo@it.uc3m.es 803 URI: http://www.it.uc3m.es 805 Christoph Paasch 806 UCLouvain 807 Place Sainte Barbe, 2 808 Louvain-la-Neuve, 1348 809 Belgium 811 Email: christoph.paasch@uclouvain.be 813 Fernando Gont 814 SI6 Networks / UTN-FRH 815 Evaristo Carriego 2644 816 Haedo, Provincia de Buenos Aires 1706 817 Argentina 819 Phone: +54 11 4650 8472 820 Email: fgont@si6networks.com 821 URI: http://www.si6networks.com 823 Olivier Bonaventure 824 UCLouvain 825 Place Sainte Barbe, 2 826 Louvain-la-Neuve, 1348 827 Belgium 829 Email: olivier.bonaventure@uclouvain.be 830 Costin Raiciu 831 Universitatea Politehnica Bucuresti 832 Splaiul Independentei 313a 833 Bucuresti 834 Romania 836 Email: costin.raiciu@cs.pub.ro