Debian Codegraph @

rfc7739.html

raw
1 <pre>Internet Engineering Task Force (IETF) F. Gont
2 Request for Comments: 7739 Huawei Technologies
3 Category: Informational February 2016
4 ISSN: 2070-1721
5
6
7 <span class="h1">Security Implications of Predictable Fragment Identification Values</span>
8
9 Abstract
10
11 IPv6 specifies the Fragment Header, which is employed for the
12 fragmentation and reassembly mechanisms. The Fragment Header
13 contains an "Identification" field that, together with the IPv6
14 Source Address and the IPv6 Destination Address of a packet,
15 identifies fragments that correspond to the same original datagram,
16 such that they can be reassembled together by the receiving host.
17 The only requirement for setting the Identification field is that the
18 corresponding value must be different than that employed for any
19 other fragmented datagram sent recently with the same Source Address
20 and Destination Address. Some implementations use a simple global
21 counter for setting the Identification field, thus leading to
22 predictable Identification values. This document analyzes the
23 security implications of predictable Identification values, and
24 provides implementation guidance for setting the Identification field
25 of the Fragment Header, such that the aforementioned security
26 implications are mitigated.
27
28 Status of This Memo
29
30 This document is not an Internet Standards Track specification; it is
31 published for informational purposes.
32
33 This document is a product of the Internet Engineering Task Force
34 (IETF). It represents the consensus of the IETF community. It has
35 received public review and has been approved for publication by the
36 Internet Engineering Steering Group (IESG). Not all documents
37 approved by the IESG are a candidate for any level of Internet
38 Standard; see <a href="./rfc5741#section-2">Section&nbsp;2 of RFC 5741</a>.
39
40 Information about the current status of this document, any errata,
41 and how to provide feedback on it may be obtained at
42 <a href="http://www.rfc-editor.org/info/rfc7739">http://www.rfc-editor.org/info/rfc7739</a>.
43
44
45
46
47
48
49
50
51
52 <span class="grey">Gont Informational [Page 1]</span></pre>
53 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-2" ></span>
54 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
55
56
57 Copyright Notice
58
59 Copyright (c) 2016 IETF Trust and the persons identified as the
60 document authors. All rights reserved.
61
62 This document is subject to <a href="https://www.rfc-editor.org/bcp/bcp78">BCP 78</a> and the IETF Trust's Legal
63 Provisions Relating to IETF Documents
64 (<a href="http://trustee.ietf.org/license-info">http://trustee.ietf.org/license-info</a>) in effect on the date of
65 publication of this document. Please review these documents
66 carefully, as they describe your rights and restrictions with respect
67 to this document. Code Components extracted from this document must
68 include Simplified BSD License text as described in Section 4.e of
69 the Trust Legal Provisions and are provided without warranty as
70 described in the Simplified BSD License.
71
72 Table of Contents
73
74 <a href="#section-1">1</a>. Introduction . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-3">3</a>
75 <a href="#section-2">2</a>. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-3">3</a>
76 3. Security Implications of Predictable Fragment Identification
77 Values . . . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-3">3</a>
78 4. Constraints for the Selection of Fragment Identification
79 Values . . . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-7">7</a>
80 <a href="#section-5">5</a>. Algorithms for Selecting Fragment Identification Values . . . <a href="#page-8">8</a>
81 5.1. Per-Destination Counter (Initialized to a Random Value) . 8
82 <a href="#section-5.2">5.2</a>. Randomized Identification Values . . . . . . . . . . . . <a href="#page-9">9</a>
83 5.3. Hash-Based Fragment Identification Selection Algorithm . 10
84 <a href="#section-6">6</a>. Security Considerations . . . . . . . . . . . . . . . . . . . <a href="#page-12">12</a>
85 <a href="#section-7">7</a>. References . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-13">13</a>
86 <a href="#section-7.1">7.1</a>. Normative References . . . . . . . . . . . . . . . . . . <a href="#page-13">13</a>
87 <a href="#section-7.2">7.2</a>. Informative References . . . . . . . . . . . . . . . . . <a href="#page-14">14</a>
88 <a href="#appendix-A">Appendix A</a>. Information Leakage Produced by Vulnerable
89 Implementations . . . . . . . . . . . . . . . . . . <a href="#page-16">16</a>
90 <a href="#appendix-B">Appendix B</a>. Survey of Fragment Identification Selection
91 Algorithms Employed by Popular IPv6 Implementations 18
92 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-20">20</a>
93 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-20">20</a>
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108 <span class="grey">Gont Informational [Page 2]</span></pre>
109 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-3" ></span>
110 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
111
112
113 <span class="h2"><a class="selflink" id="section-1" href="#section-1">1</a>. Introduction</span>
114
115 IPv6 specifies the Fragment Header, which is employed for the
116 fragmentation and reassembly mechanisms. The Fragment Header
117 contains an "Identification" field that, together with the IPv6
118 Source Address and the IPv6 Destination Address of a packet,
119 identifies fragments that correspond to the same original datagram,
120 such that they can be reassembled together by the receiving host.
121 The only requirement for setting the Identification field is that its
122 value must be different than that employed for any other fragmented
123 datagram sent recently with the same Source Address and Destination
124 Address.
125
126 The most trivial algorithm to avoid reusing Identification values too
127 quickly is to maintain a global counter that is incremented for each
128 fragmented datagram that is transmitted. However, this trivial
129 algorithm leads to predictable Identification values that can be
130 leveraged to perform a variety of attacks.
131
132 <a href="#section-3">Section 3</a> of this document analyzes the security implications of
133 predictable Identification values. <a href="#section-4">Section 4</a> discusses constraints
134 in the possible algorithms for selecting Identification values.
135 <a href="#section-5">Section 5</a> specifies a number of algorithms that could be used for
136 generating Identification values that mitigate the issues discussed
137 in this document. Finally, <a href="#appendix-B">Appendix B</a> contains a survey of the
138 algorithms employed by popular IPv6 implementations for generating
139 the Identification values.
140
141 <span class="h2"><a class="selflink" id="section-2" href="#section-2">2</a>. Terminology</span>
142
143 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
144 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
145 document are to be interpreted as described in <a href="./rfc2119">RFC 2119</a> [<a href="./rfc2119" title="&quot;Key words for use in RFCs to Indicate Requirement Levels&quot;">RFC2119</a>].
146
147 <span class="h2"><a class="selflink" id="section-3" href="#section-3">3</a>. Security Implications of Predictable Fragment Identification Values</span>
148
149 Predictable Identification values result in an information leakage
150 that can be exploited in a number of ways. Among others, they may
151 potentially be exploited to:
152
153 o determine the packet rate at which a given system is transmitting
154 information
155
156 o perform stealth port scans to a third party
157
158 o uncover the rules of a number of firewalls
159
160 o count the number of systems behind a middle-box
161
162
163
164 <span class="grey">Gont Informational [Page 3]</span></pre>
165 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-4" ></span>
166 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
167
168
169 o perform Denial-of-Service (DoS) attacks, or
170
171 o perform data injection attacks against transport or application
172 protocols
173
174 The security implications introduced by predictable Identification
175 values in IPv6 are very similar to those of predictable
176 Identification values in IPv4.
177
178 NOTE:
179 [<a href="#ref-Sanfilippo1998a">Sanfilippo1998a</a>] originally pointed out how the IPv4
180 Identification field could be examined to determine the packet
181 rate at which a given system is transmitting information. Later,
182 [<a href="#ref-Sanfilippo1998b">Sanfilippo1998b</a>] described how a system with such an
183 implementation could be used to perform a stealth port scan to a
184 third (victim) host. [<a href="#ref-Sanfilippo1999">Sanfilippo1999</a>] explained how to exploit
185 this implementation strategy to uncover the rules of a number of
186 firewalls. [<a href="#ref-Bellovin2002">Bellovin2002</a>] explained how the IPv4 Identification
187 field could be exploited to count the number of systems behind a
188 NAT. [<a href="#ref-Fyodor2004">Fyodor2004</a>] is an entire paper on most (if not all) the
189 ways to exploit the information provided by the Identification
190 field of the IPv4 header (and these results apply in a similar way
191 to IPv6). [<a href="#ref-Zalewski2003">Zalewski2003</a>] originally envisioned the exploitation
192 of IP fragmentation/reassembly for performing data injection
193 attacks against upper-layer protocols. [<a href="#ref-Herzberg2013">Herzberg2013</a>] explores
194 the use of IPv4/IPv6 fragmentation and predictable Identification
195 values for performing DNS cache poisoning attacks in great detail.
196 [<a href="./rfc6274" title="&quot;Security Assessment of the Internet Protocol Version 4&quot;">RFC6274</a>] covers the security implications of the IPv4 case in
197 detail.
198
199 One key difference between the IPv4 case and the IPv6 case is that,
200 in IPv4, the Identification field is part of the fixed IPv4 header
201 (and thus usually set for all packets), while in IPv6 the
202 Identification field is present only in those packets that carry a
203 Fragment Header. As a result, successful exploitation of the
204 Identification field depends on two different factors:
205
206 o vulnerable Identification generators, and
207
208 o the ability of an attacker to trigger the use of IPv6
209 fragmentation for packets sent from/to the victim node
210
211 The scenarios in which an attacker may successfully perform the
212 aforementioned attacks depend on the specific attack type. For
213 example, in order to perform a DoS attack on communications between
214 two hosts, an attacker would need to know the IPv6 addresses employed
215 by the aforementioned two nodes. Such knowledge may be readily
216 available if the target of the attack is the communication between
217
218
219
220 <span class="grey">Gont Informational [Page 4]</span></pre>
221 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-5" ></span>
222 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
223
224
225 two specific BGP peers, two specific SMTP servers, or one specific
226 primary DNS server and one of its secondary DNS servers, but may not
227 be easily available if the goal is a DoS attack on all communications
228 between arbitrary IPv6 hosts (e.g., the goal is to perform a DoS
229 attack on all communications involving one specific node with
230 arbitrary/unknown hosts). Other attacks, such as performing stealth
231 port scans to a third party or determining the packet rate at which a
232 given system is transmitting information, only require the attacker
233 to know the IPv6 address of a vulnerable implementation.
234
235 As noted in <a href="#section-1">Section 1</a>, some implementations have been known to use
236 predictable Identification values. For instance, <a href="#appendix-B">Appendix B</a> of this
237 document shows that recent versions of a number of popular IPv6
238 implementations employ predictable values for the Identification
239 field of the Fragment Header.
240
241 Additionally, we note that [<a href="./rfc2460" title="&quot;Internet Protocol, Version 6 (IPv6) Specification&quot;">RFC2460</a>] states that when an ICMPv6
242 Packet Too Big (PTB) error message advertising a Maximum Transfer
243 Unit (MTU) smaller than 1280 bytes is received, the receiving host is
244 not required to reduce the Path-MTU for the corresponding Destination
245 Address, but must simply include a Fragment Header in all subsequent
246 packets sent to that destination. This triggers the use of the so-
247 called IPv6 "atomic fragments" [<a href="./rfc6946" title="&quot;Processing of IPv6 &quot;">RFC6946</a>]: IPv6 fragments with a
248 Fragment Offset equal to 0, and the "M" ("More fragments") bit clear.
249 [<a href="#ref-DEPGEN" title="&quot;Generation of IPv6 Atomic Fragments Considered Harmful&quot;">DEPGEN</a>] documents the motivation of deprecating the generation of
250 IPv6 atomic fragments in [<a href="./rfc2460" title="&quot;Internet Protocol, Version 6 (IPv6) Specification&quot;">RFC2460</a>].
251
252 Thus, an attacker can usually cause a victim host to "fragment" its
253 outgoing packets by sending it a forged ICMPv6 Packet Too Big (PTB)
254 error message that advertises an MTU smaller than 1280 bytes.
255
256 There are a number of aspects that should be considered, though:
257
258 o All the implementations the author is aware of record the Path-MTU
259 information on a per-destination basis. Thus, an attacker can
260 only cause the victim to enable fragmentation for those packets
261 sent to the Source Address of IPv6 packet embedded in the payload
262 of the ICMPv6 PTB message. However, we note that <a href="./rfc1981#section-5.2">Section&nbsp;5.2 of
263 [RFC1981]</a> notes that an implementation could maintain a single
264 system-wide Path MTU (PMTU) value to be used for all packets sent
265 to that node. Clearly, such implementations would exacerbate the
266 problem of any attacks based on Path MTU Discovery (PMTUD)
267 [<a href="./rfc5927" title="&quot;ICMP Attacks against TCP&quot;">RFC5927</a>] or IPv6 fragmentation.
268
269 o If the victim node implements some of the counter-measures for
270 ICMP attacks described in <a href="./rfc5927">RFC 5927</a> [<a href="./rfc5927" title="&quot;ICMP Attacks against TCP&quot;">RFC5927</a>], it might be
271 difficult for an attacker to cause the victim node to employ
272 fragmentation for its outgoing packets. However, many current
273
274
275
276 <span class="grey">Gont Informational [Page 5]</span></pre>
277 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-6" ></span>
278 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
279
280
281 implementations fail to enforce these validation checks. For
282 example, Linux 2.6.38-8 does not even require received ICMPv6
283 error messages to correspond to an ongoing communication instance.
284
285 o Some implementations (notably Linux) have already been updated
286 according to [<a href="#ref-DEPGEN" title="&quot;Generation of IPv6 Atomic Fragments Considered Harmful&quot;">DEPGEN</a>] such that ICMPv6 PTB messages do not result
287 in the generation of IPv6 atomic fragments.
288
289 Implementations that employ predictable Identification values and
290 also fail to enforce validation checks on ICMPv6 error messages
291 become vulnerable to the same type of attacks that can be exploited
292 with IPv4 fragmentation, discussed earlier in this section.
293
294 One possible way in which predictable Identification values could be
295 leveraged for performing a DoS attack is as follows: Let us assume
296 that Host A is communicating with Host B, and that an attacker wants
297 to perform a DoS attack such communication. The attacker would learn
298 the Identification value currently in use by Host A, possibly by
299 sending any packet that would elicit a fragmented response (e.g., an
300 ICPMv6 echo request with a large payload). The attacker would then
301 send a forged ICMPv6 PTB error message to Host A (with the IPv6
302 Source Address of the embedded IPv6 packet set to the IPv6 address of
303 Host A, and the Destination Address of the embedded IPv6 packet set
304 to the IPv6 address of a Host B), such that any subsequent packets
305 sent by Host A to Host B include a Fragment Header. Finally, the
306 attacker would send forged IPv6 fragments to Host B, with their IPv6
307 Source Address set to that of Host A, and Identification values that
308 would result in collisions with the Identification values employed
309 for the legitimate traffic sent by Host A to Host B. If Host B
310 discards fragments that result in collisions of Identification values
311 (e.g., such fragments overlap, and the host implements [<a href="./rfc5722" title="&quot;Handling of Overlapping IPv6 Fragments&quot;">RFC5722</a>]),
312 the attacker could simply trash the Identification space by sending
313 multiple forged fragments with different Identification values, such
314 that any subsequent packets from Host A to Host B are discarded at
315 Host B as a result of the malicious fragments sent by the attacker.
316
317 NOTE:
318 For example, Linux 2.6.38-10 is vulnerable to the aforementioned
319 issue.
320
321 [<a id="ref-RFC6946">RFC6946</a>] describes an improved processing of these packets that
322 would eliminate this specific attack vector, at least in the case
323 of TCP connections that employ the Path-MTU Discovery mechanism.
324
325 The aforementioned attack scenario is simply included to illustrate
326 the problem of employing predictable Identification values. We note
327 that regardless of the attacker's ability to cause a victim host to
328
329
330
331
332 <span class="grey">Gont Informational [Page 6]</span></pre>
333 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-7" ></span>
334 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
335
336
337 employ fragmentation when communicating with third parties, use of
338 predictable Identification values makes communication flows that
339 employ fragmentation vulnerable to any fragmentation-based attacks.
340
341 <span class="h2"><a class="selflink" id="section-4" href="#section-4">4</a>. Constraints for the Selection of Fragment Identification Values</span>
342
343 The Identification field of the Fragment Header is 32-bits long.
344 However, when translators (e.g. [<a href="./rfc6145" title="&quot;IP/ICMP Translation Algorithm&quot;">RFC6145</a>]) are employed, the high-
345 order 16 bits of the Identification field are effectively ignored.
346
347 NOTE:
348 [<a href="./rfc6145" title="&quot;IP/ICMP Translation Algorithm&quot;">RFC6145</a>] notes that, when translating in the IPv6-to-IPv4
349 direction, "if there is a Fragment Header in the IPv6 packet, the
350 last 16 bits of its value MUST be used for the IPv4 identification
351 value".
352
353 Additionally, <a href="./rfc6052#section-3.3">Section&nbsp;3.3 of [RFC6052]</a> encourages operators to use
354 a Network-Specific Prefix (NSP) that maps the IPv4 address space
355 into IPv6. Thus, when an NSP is being used, IPv6 addresses
356 representing IPv4 nodes (reached through a stateless translator)
357 are indistinguishable from native IPv6 addresses.
358
359 Thus, when translators are employed, the "effective" length of the
360 Identification field is 16 bits and, as a result, at least during the
361 IPv6/IPv4 transition/co-existence phase, it is probably safer to
362 assume that only the low-order 16 bits of the Identification field
363 are of use to the destination system.
364
365 Regarding the selection of Identification values, the only
366 requirement specified in [<a href="./rfc2460" title="&quot;Internet Protocol, Version 6 (IPv6) Specification&quot;">RFC2460</a>] is that the Identification value
367 must be different than that of any other fragmented packet sent
368 recently with the same Source Address and Destination Address.
369 Failure to comply with this requirement could lead to the
370 interoperability problems discussed in [<a href="./rfc4963" title="&quot;IPv4 Reassembly Errors at High Data Rates&quot;">RFC4963</a>].
371
372 From a security standpoint, unpredictable Identification values are
373 desirable. However, this is somewhat at odds with the "reuse"
374 requirements specified in [<a href="./rfc2460" title="&quot;Internet Protocol, Version 6 (IPv6) Specification&quot;">RFC2460</a>], that specifies that an
375 Identification value must be different than that employed for any
376 other fragmented packet sent recently with the same Source Address
377 and Destination Address.
378
379 Finally, since Identification values need to be selected for each
380 outgoing datagram that requires fragmentation, the performance impact
381 should be considered when choosing an algorithm for the selection of
382 Identification values.
383
384
385
386
387
388 <span class="grey">Gont Informational [Page 7]</span></pre>
389 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-8" ></span>
390 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
391
392
393 <span class="h2"><a class="selflink" id="section-5" href="#section-5">5</a>. Algorithms for Selecting Fragment Identification Values</span>
394
395 There are a number of algorithms that may be used for setting the
396 Identification field such that the security issues discussed in this
397 document are avoided. This section presents three of those.
398
399 The algorithm in <a href="#section-5.1">Section 5.1</a> typically leads to a low Identification
400 reuse frequency at the expense of keeping per-destination state; this
401 algorithm only uses a Pseudorandom Number Generator (PNRG) when the
402 host communicates with a new destination. The algorithm in
403 <a href="#section-5.2">Section 5.2</a> may result in a higher Identification reuse frequency.
404 It also uses a PRNG for each datagram that needs to be fragmented.
405 Hence, the algorithm in <a href="#section-5.1">Section 5.1</a> will likely result in better
406 performance properties. Finally, the algorithm in <a href="#section-5.3">Section 5.3</a>
407 achieves a similar Identification reuse frequency to that of the
408 algorithm in <a href="#section-5.1">Section 5.1</a> without the need of keeping state, but
409 possibly at the expense of lower per-packet performance.
410
411 NOTE:
412 Since the specific algorithm to be employed for the PRNGs in
413 <a href="#section-5.1">Section 5.1</a> and <a href="#section-5.2">Section 5.2</a>, and the specific algorithms to be
414 employed for the hash functions in <a href="#section-5.3">Section 5.3</a> have not been
415 specified, it is impossible to provide a quantitative performance
416 comparison of the algorithms described in this section.
417
418 <span class="h3"><a class="selflink" id="section-5.1" href="#section-5.1">5.1</a>. Per-Destination Counter (Initialized to a Random Value)</span>
419
420 This algorithm consists of the following steps:
421
422 1. Whenever a packet must be sent with a Fragment Header, the
423 sending host should look up in the Destination Cache an entry
424 corresponding to the Destination Address of the packet.
425
426 2. If such an entry exists, it contains the last Identification
427 value used for that Destination Address. Therefore, such a value
428 should be incremented by 1 and used for setting the
429 Identification field of the outgoing packet. Additionally, the
430 updated value should be recorded in the corresponding entry of
431 the Destination Cache [<a href="./rfc4861" title="&quot;Neighbor Discovery for IP version 6 (IPv6)&quot;">RFC4861</a>].
432
433 3. If such an entry does not exist, it should be created, and the
434 Identification value for that destination should be initialized
435 with a random value (e.g., with a Pseudorandom Number Generator),
436 and used for setting the Identification field of the Fragment
437 Header of the outgoing fragmented datagram.
438
439
440
441
442
443
444 <span class="grey">Gont Informational [Page 8]</span></pre>
445 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-9" ></span>
446 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
447
448
449 The advantages of this algorithm are:
450
451 o It is simple to implement, with the only complexity residing in
452 the PRNG used to initialize the Identification value contained in
453 each entry of the Destination Cache.
454
455 o The Identification reuse frequency will typically be lower than
456 that achieved by a global counter (when sending traffic to
457 multiple destinations), since this algorithm uses per-destination
458 counters (rather than a single system-wide counter).
459
460 o It has good performance properties (once the corresponding entry
461 in the Destination Cache has been created and initialized, each
462 subsequent Identification value simply involves the increment of a
463 counter).
464
465 The possible drawbacks of this algorithm are:
466
467 o If, as a result of resource management, an entry of the
468 Destination Cache must be removed, the last Identification value
469 used for that Destination will be lost. Thus, subsequent traffic
470 to that destination would cause that entry to be recreated and
471 reinitialized to random value, thus possibly leading to
472 Identification "collisions".
473
474 o Since the Identification values are predictable by the destination
475 host, a vulnerable host might possibly leak to third parties the
476 Identification values used by other hosts to send traffic to it
477 (i.e., Host B could leak to Host C the Identification values that
478 Host A is using to send packets to Host B). <a href="#appendix-A">Appendix A</a> describes
479 one possible scenario for such leakage in detail.
480
481 <span class="h3"><a class="selflink" id="section-5.2" href="#section-5.2">5.2</a>. Randomized Identification Values</span>
482
483 Clearly, use of a Pseudorandom Number Generator for selecting the
484 Identification would be desirable from a security standpoint. With
485 such a scheme, the Identification of each fragmented datagram would
486 be selected as:
487
488 Identification = random()
489
490 where "random()" is the PRNG.
491
492 The specific properties of such scheme would clearly depend on the
493 specific PRNG employed. For example, some PRNGs may result in higher
494 Identification reuse frequencies than others, in the same way that
495 some PRNGs may be more expensive (in terms of processing requirements
496 and/or implementation complexity) than others.
497
498
499
500 <span class="grey">Gont Informational [Page 9]</span></pre>
501 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-10" ></span>
502 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
503
504
505 Discussion of the properties of possible PRNGs is considered out of
506 the scope of this document. However, we do note that some PRNGs
507 employed in the past by some implementations have been found to be
508 predictable [<a href="#ref-Klein2007">Klein2007</a>]. Please see [<a href="./rfc4086" title="&quot;Randomness Requirements for Security&quot;">RFC4086</a>] for randomness
509 requirements for security.
510
511 <span class="h3"><a class="selflink" id="section-5.3" href="#section-5.3">5.3</a>. Hash-Based Fragment Identification Selection Algorithm</span>
512
513 Another alternative is to implement a hash-based algorithm similar to
514 that specified in [<a href="./rfc6056" title="&quot;Recommendations for Transport- Protocol Port Randomization&quot;">RFC6056</a>] for the selection of transport port
515 numbers. With such a scheme, the Identification value of each
516 fragmented datagram would be selected with the expression:
517
518 Identification = F(Src IP, Dst IP, secret1) +
519 counter[G(Src IP, Dst Pref, secret2)]
520
521 where:
522
523 Identification:
524 Identification value to be used for the fragmented datagram.
525
526 F():
527 Hash function.
528
529 Src IP:
530 IPv6 Source Address of the datagram to be fragmented.
531
532 Dst IP:
533 IPv6 Destination Address of the datagram to be fragmented.
534
535 secret1:
536 Secret data unknown to the attacker. This value can be
537 initialized to a pseudo-random value during the system
538 bootstrapping sequence. It should remain constant at least while
539 there could be previously sent fragments still in the network or
540 at the fragment reassembly buffer of the corresponding destination
541 system(s).
542
543 counter[]:
544 System-wide array of 32-bit counters (e.g. with 8K elements or
545 more). Each counter should be initialized to a pseudo-random
546 value during the system bootstrapping sequence.
547
548 G():
549 Hash function. It may or may not be the same hash function as
550 that used for F().
551
552
553
554
555
556 <span class="grey">Gont Informational [Page 10]</span></pre>
557 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-11" ></span>
558 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
559
560
561 Dst Pref:
562 IPv6 "Destination Prefix" of the datagram to be fragmented (can be
563 assumed to be the first eight bytes of the Destination Address of
564 such packet). Note: the "Destination Prefix" (rather than
565 Destination Address) is used, such that the ability of an attacker
566 of searching the "increments" space by using multiple addresses of
567 the same subnet is reduced.
568
569 secret2:
570 Secret data unknown to the attacker. This value can be
571 initialized to a pseudo-random value during the system
572 bootstrapping sequence. It should remain constant at least while
573 there could be previously sent fragments still in the network or
574 at the fragment reassembly buffer of the corresponding destination
575 system(s).
576
577 NOTE:
578 counter[G(src IP, Dst Pref, secret2)] should be incremented by one
579 each time an Identification value is selected.
580
581 The output of F() will be constant for each (Src IP, Dst IP) pair.
582 Similarly, the output of G() will be constant for each (Src IP, Dst
583 Pref) pair. Thus, the resulting Identification value will be the
584 result of a random offset plus a linear function (provided by
585 counter[]), therefore resulting in a monotonically increasing
586 sequence of Identification values for each (src IP, Dst IP) pair.
587
588 NOTE:
589 F() essentially provides the unpredictability (by off-path
590 attackers) of the resulting Identification values, while counter[]
591 provides a linear function such that the Identification values are
592 different for each fragmented packet while the Identification
593 reuse frequency is minimized.
594
595 The advantages of this algorithm are:
596
597 o The Identification reuse frequency will typically be lower than
598 that achieved by a global counter (when sending traffic to
599 multiple destinations), since this algorithm uses multiple system-
600 wide counters (rather than a single system-wide counter). The
601 extent to which the reuse frequency will be lower depends on the
602 number of elements in counter[], and the number of other active
603 flows that result in the same value of G() (and hence cause the
604 same counter to be incremented for each datagram that is
605 fragmented).
606
607
608
609
610
611
612 <span class="grey">Gont Informational [Page 11]</span></pre>
613 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-12" ></span>
614 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
615
616
617 o It is possible to implement the algorithm such that good
618 performance is achieved. For example, the result of F() could be
619 stored in the Destination Cache (such that it need not be
620 recomputed for each packet that must be sent) along with the
621 computed index/argument for counter[].
622
623 NOTE:
624 If this implementation approach is followed, and an entry of
625 the Destination Cache must be removed as a result of resource
626 management, the last Identification value used for that
627 Destination will *not* be lost. This is an improvement over
628 the algorithm specified in <a href="#section-5.1">Section 5.1</a>.
629
630 The possible drawbacks of this algorithm are:
631
632 o Since the Identification values are predictable by the destination
633 host, a vulnerable host could possibly leak to third parties the
634 Identification values used by other hosts to send traffic to it
635 (i.e., Host B could leak to Host C the Identification values that
636 Host A is using to send packets to Host B). <a href="#appendix-A">Appendix A</a> describes
637 a possible scenario in which that information leakage could take
638 place. We note, however, that this algorithm makes the
639 aforementioned attack less reliable for the attacker, since each
640 counter could be possibly shared by multiple traffic flows (i.e.,
641 packets destined to other destinations might cause the same
642 counter to be incremented).
643
644 This algorithm might be preferable (over the one specified in
645 <a href="#section-5.1">Section 5.1</a>) in those scenarios in which a node is expected to
646 communicate with a large number of destinations, and thus it is
647 desirable to limit the amount of information to be maintained in
648 memory.
649
650 NOTE:
651 In such scenarios, if the algorithm specified in <a href="#section-5.1">Section 5.1</a> were
652 implemented, entries from the Destination Cache might need to be
653 pruned frequently, thus increasing the risk of Identification
654 "collisions".
655
656 <span class="h2"><a class="selflink" id="section-6" href="#section-6">6</a>. Security Considerations</span>
657
658 This document discusses the security implications of predictable
659 Identification values, and provides implementation guidance such that
660 the aforementioned security implications can be mitigated.
661
662
663
664
665
666
667
668 <span class="grey">Gont Informational [Page 12]</span></pre>
669 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-13" ></span>
670 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
671
672
673 A number of possible algorithms are described, to provide some
674 implementation alternatives to implementers. We note that the
675 selection of such an algorithm usually implies a number of trade-offs
676 (security, performance, implementation complexity, interoperability
677 properties, etc.).
678
679 <span class="h2"><a class="selflink" id="section-7" href="#section-7">7</a>. References</span>
680
681 <span class="h3"><a class="selflink" id="section-7.1" href="#section-7.1">7.1</a>. Normative References</span>
682
683 [<a id="ref-RFC1981">RFC1981</a>] McCann, J., Deering, S., and J. Mogul, "Path MTU Discovery
684 for IP version 6", <a href="./rfc1981">RFC 1981</a>, DOI 10.17487/RFC1981, August
685 1996, &lt;<a href="http://www.rfc-editor.org/info/rfc1981">http://www.rfc-editor.org/info/rfc1981</a>&gt;.
686
687 [<a id="ref-RFC2119">RFC2119</a>] Bradner, S., "Key words for use in RFCs to Indicate
688 Requirement Levels", <a href="https://www.rfc-editor.org/bcp/bcp14">BCP 14</a>, <a href="./rfc2119">RFC 2119</a>,
689 DOI 10.17487/RFC2119, March 1997,
690 &lt;<a href="http://www.rfc-editor.org/info/rfc2119">http://www.rfc-editor.org/info/rfc2119</a>&gt;.
691
692 [<a id="ref-RFC2460">RFC2460</a>] Deering, S. and R. Hinden, "Internet Protocol, Version 6
693 (IPv6) Specification", <a href="./rfc2460">RFC 2460</a>, DOI 10.17487/RFC2460,
694 December 1998, &lt;<a href="http://www.rfc-editor.org/info/rfc2460">http://www.rfc-editor.org/info/rfc2460</a>&gt;.
695
696 [<a id="ref-RFC4086">RFC4086</a>] Eastlake 3rd, D., Schiller, J., and S. Crocker,
697 "Randomness Requirements for Security", <a href="https://www.rfc-editor.org/bcp/bcp106">BCP 106</a>, <a href="./rfc4086">RFC 4086</a>,
698 DOI 10.17487/RFC4086, June 2005,
699 &lt;<a href="http://www.rfc-editor.org/info/rfc4086">http://www.rfc-editor.org/info/rfc4086</a>&gt;.
700
701 [<a id="ref-RFC4861">RFC4861</a>] Narten, T., Nordmark, E., Simpson, W., and H. Soliman,
702 "Neighbor Discovery for IP version 6 (IPv6)", <a href="./rfc4861">RFC 4861</a>,
703 DOI 10.17487/RFC4861, September 2007,
704 &lt;<a href="http://www.rfc-editor.org/info/rfc4861">http://www.rfc-editor.org/info/rfc4861</a>&gt;.
705
706 [<a id="ref-RFC5722">RFC5722</a>] Krishnan, S., "Handling of Overlapping IPv6 Fragments",
707 <a href="./rfc5722">RFC 5722</a>, DOI 10.17487/RFC5722, December 2009,
708 &lt;<a href="http://www.rfc-editor.org/info/rfc5722">http://www.rfc-editor.org/info/rfc5722</a>&gt;.
709
710 [<a id="ref-RFC6052">RFC6052</a>] Bao, C., Huitema, C., Bagnulo, M., Boucadair, M., and X.
711 Li, "IPv6 Addressing of IPv4/IPv6 Translators", <a href="./rfc6052">RFC 6052</a>,
712 DOI 10.17487/RFC6052, October 2010,
713 &lt;<a href="http://www.rfc-editor.org/info/rfc6052">http://www.rfc-editor.org/info/rfc6052</a>&gt;.
714
715 [<a id="ref-RFC6056">RFC6056</a>] Larsen, M. and F. Gont, "Recommendations for Transport-
716 Protocol Port Randomization", <a href="https://www.rfc-editor.org/bcp/bcp156">BCP 156</a>, <a href="./rfc6056">RFC 6056</a>,
717 DOI 10.17487/RFC6056, January 2011,
718 &lt;<a href="http://www.rfc-editor.org/info/rfc6056">http://www.rfc-editor.org/info/rfc6056</a>&gt;.
719
720
721
722
723
724 <span class="grey">Gont Informational [Page 13]</span></pre>
725 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-14" ></span>
726 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
727
728
729 [<a id="ref-RFC6145">RFC6145</a>] Li, X., Bao, C., and F. Baker, "IP/ICMP Translation
730 Algorithm", <a href="./rfc6145">RFC 6145</a>, DOI 10.17487/RFC6145, April 2011,
731 &lt;<a href="http://www.rfc-editor.org/info/rfc6145">http://www.rfc-editor.org/info/rfc6145</a>&gt;.
732
733 [<a id="ref-RFC6946">RFC6946</a>] Gont, F., "Processing of IPv6 "Atomic" Fragments",
734 <a href="./rfc6946">RFC 6946</a>, DOI 10.17487/RFC6946, May 2013,
735 &lt;<a href="http://www.rfc-editor.org/info/rfc6946">http://www.rfc-editor.org/info/rfc6946</a>&gt;.
736
737 <span class="h3"><a class="selflink" id="section-7.2" href="#section-7.2">7.2</a>. Informative References</span>
738
739 [<a id="ref-RFC4963">RFC4963</a>] Heffner, J., Mathis, M., and B. Chandler, "IPv4 Reassembly
740 Errors at High Data Rates", <a href="./rfc4963">RFC 4963</a>,
741 DOI 10.17487/RFC4963, July 2007,
742 &lt;<a href="http://www.rfc-editor.org/info/rfc4963">http://www.rfc-editor.org/info/rfc4963</a>&gt;.
743
744 [<a id="ref-RFC5927">RFC5927</a>] Gont, F., "ICMP Attacks against TCP", <a href="./rfc5927">RFC 5927</a>,
745 DOI 10.17487/RFC5927, July 2010,
746 &lt;<a href="http://www.rfc-editor.org/info/rfc5927">http://www.rfc-editor.org/info/rfc5927</a>&gt;.
747
748 [<a id="ref-RFC6274">RFC6274</a>] Gont, F., "Security Assessment of the Internet Protocol
749 Version 4", <a href="./rfc6274">RFC 6274</a>, DOI 10.17487/RFC6274, July 2011,
750 &lt;<a href="http://www.rfc-editor.org/info/rfc6274">http://www.rfc-editor.org/info/rfc6274</a>&gt;.
751
752 [<a id="ref-DEPGEN">DEPGEN</a>] Gont, F., Liu, S., and T. Anderson, "Generation of IPv6
753 Atomic Fragments Considered Harmful", Work in Progress,
754 <a href="./draft-ietf-6man-deprecate-atomfrag-generation-05">draft-ietf-6man-deprecate-atomfrag-generation-05</a>, January
755 2016.
756
757 [<a id="ref-Bellovin2002">Bellovin2002</a>]
758 Bellovin, S., "A Technique for Counting NATted Hosts",
759 IMW'02 Nov. 6-8, 2002, Marseille, France,
760 DOI 10.1145/637201.637243, 2002.
761
762 [<a id="ref-Fyodor2004">Fyodor2004</a>]
763 Lyon, G., "TCP Idle Scan", from Chapter 5 of "Nmap Network
764 Scanning", 2004,
765 &lt;<a href="http://www.insecure.org/nmap/idlescan.html">http://www.insecure.org/nmap/idlescan.html</a>&gt;.
766
767 [<a id="ref-Herzberg2013">Herzberg2013</a>]
768 Herzberg, A. and H. Shulman, "Fragmentation Considered
769 Poisonous", Technical Report 13-03, March 2013,
770 &lt;<a href="http://u.cs.biu.ac.il/~herzbea/security/13-03-frag.pdf">http://u.cs.biu.ac.il/~herzbea/security/13-03-frag.pdf</a>&gt;.
771
772 [<a id="ref-Klein2007">Klein2007</a>]
773 Klein, A., "OpenBSD DNS Cache Poisoning and Multiple O/S
774 Predictable IP ID Vulnerability", 2007,
775 &lt;<a href="http://www.trusteer.com/files/OpenBSD_DNS_Cache_Poisoning_and_Multiple_OS_Predictable_IP_ID_Vulnerability.pdf">http://www.trusteer.com/files/OpenBSD_DNS_Cache_Poisoning</a>
776 <a href="http://www.trusteer.com/files/OpenBSD_DNS_Cache_Poisoning_and_Multiple_OS_Predictable_IP_ID_Vulnerability.pdf">_and_Multiple_OS_Predictable_IP_ID_Vulnerability.pdf</a>&gt;.
777
778
779
780 <span class="grey">Gont Informational [Page 14]</span></pre>
781 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-15" ></span>
782 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
783
784
785 [<a id="ref-Sanfilippo1998a">Sanfilippo1998a</a>]
786 Sanfilippo, S., "Subject: about the ip header id", message
787 to Bugtraq mailing list, 14 December 1998,
788 &lt;<a href="http://diswww.mit.edu/menelaus.mit.edu/bt/8704">http://diswww.mit.edu/menelaus.mit.edu/bt/8704</a>&gt;.
789
790 [<a id="ref-Sanfilippo1998b">Sanfilippo1998b</a>]
791 Sanfilippo, S., "Subject: new tcp scan method", message
792 to Bugtraq mailing list, 18 December 1998,
793 &lt;<a href="http://diswww.mit.edu/menelaus.mit.edu/bt/8736">http://diswww.mit.edu/menelaus.mit.edu/bt/8736</a>&gt;.
794
795 [<a id="ref-Sanfilippo1999">Sanfilippo1999</a>]
796 Sanfilippo, S., "Subject: more about IP ID", message
797 to Bugtraq mailing list, 20 November 1999,
798 &lt;<a href="http://diswww.mit.edu/menelaus.mit.edu/bt/12686">http://diswww.mit.edu/menelaus.mit.edu/bt/12686</a>&gt;.
799
800 [<a id="ref-SI6-IPv6">SI6-IPv6</a>] SI6 Networks, "SI6 Networks' IPv6 Toolkit",
801 &lt;<a href="http://www.si6networks.com/tools/ipv6toolkit">http://www.si6networks.com/tools/ipv6toolkit</a>&gt;.
802
803 [<a id="ref-Zalewski2003">Zalewski2003</a>]
804 Zalewski, M., "Subject: A new TCP/IP blind data injection
805 technique?", message to Bugtraq mailing list, 11 December
806 2003, &lt;<a href="http://lcamtuf.coredump.cx/ipfrag.txt">http://lcamtuf.coredump.cx/ipfrag.txt</a>&gt;.
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836 <span class="grey">Gont Informational [Page 15]</span></pre>
837 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-16" ></span>
838 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
839
840
841 <span class="h2"><a class="selflink" id="appendix-A" href="#appendix-A">Appendix A</a>. Information Leakage Produced by Vulnerable Implementations</span>
842
843 <a href="#section-3">Section 3</a> provides a number of references describing a number of ways
844 in which a vulnerable implementation may reveal the Identification
845 values to be used in subsequent packets, thus opening the door to a
846 number of attacks. In all of those scenarios, a vulnerable
847 implementation leaks/reveals its own Identification number.
848
849 This section presents a different attack scenario, in which a
850 vulnerable implementation leaks/reveals the Identification number of
851 a non-vulnerable implementation. That is, a vulnerable
852 implementation (Host A) leaks the current Identification value in use
853 by a third-party host (Host B) to send fragmented datagrams from Host
854 B to Host A.
855
856 NOTE:
857 For the most part, this section is included to illustrate how a
858 vulnerable implementation might be leveraged to leak out the
859 Identification value of an otherwise non-vulnerable
860 implementation.
861
862 The following scenarios assume:
863
864 Host A:
865 An IPv6 host that implements the algorithm specified in
866 <a href="#section-5.1">Section 5.1</a>, implements [<a href="./rfc5722" title="&quot;Handling of Overlapping IPv6 Fragments&quot;">RFC5722</a>], but does not implement
867 [<a href="./rfc6946" title="&quot;Processing of IPv6 &quot;">RFC6946</a>].
868
869 Host B:
870 Victim node. Selects the Identification values from a global
871 counter.
872
873 Host C:
874 Attacker. Can forge the IPv6 Source Address of his packets at
875 will.
876
877 In the following scenarios, large ICMPv6 Echo Request packets are
878 employed to "sample" the Identification value of a host. We note
879 that while the figures show only one packet for the ICMPv6 Echo
880 Request and the ICMPv6 Echo Reply packets, each of those packets will
881 typically comprise two fragments, such that the corresponding
882 unfragmented datagram is larger than the MTU of the networks to which
883 Host B and Host C are attached. Additionally, the following
884 scenarios assume that Host A employs a Fragment Header when sending
885 traffic to Host B (typically the so-called "IPv6 atomic fragments"
886 [<a href="./rfc6946" title="&quot;Processing of IPv6 &quot;">RFC6946</a>]): this behavior may be triggered by forged ICMPv6 PTB
887 messages that advertise an MTU smaller than 1280 bytes (assuming the
888 victim still generates atomic fragments [<a href="#ref-DEPGEN" title="&quot;Generation of IPv6 Atomic Fragments Considered Harmful&quot;">DEPGEN</a>]).
889
890
891
892 <span class="grey">Gont Informational [Page 16]</span></pre>
893 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-17" ></span>
894 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
895
896
897 In lines #1-#2 (and lines #7-#8), the attacker samples the current
898 Identification value at Host B. In line #3, the attacker sends a
899 forged TCP SYN segment to Host A. In line 4, the attacker sends a
900 forged TCP segment to Host B as an incomplete IPv6 fragmented
901 datagram (e.g., a single fragment with Fragment Offset=0, More
902 fragments=1). If corresponding TCP port is closed, and the attacker
903 fails when trying to produce a collision of Identification values
904 (see line #4), the following packet exchange might take place:
905
906 A B C
907
908 #1 &lt;------ Echo Req #1 -----------
909 #2 --- Echo Repl #1, FID=5000 ---&gt;
910 #3 &lt;------------------- SYN #1, src= B -----------------------
911 #4 &lt;--- SYN/ACK, FID=42 src=A ----
912 #5 ---- SYN/ACK, FID=9000 ---&gt;
913 #6 &lt;----- RST, FID= 5001 -----
914 #7 &lt;-------- Echo Req #2 ---------
915 #8 --- Echo Repl #2, FID=5002 ---&gt;
916
917 The RST segment in line #6 is elicited by the SYN/ACK segment from
918 line #5 (illegitimately elicited by the SYN segment from line #3).
919 The packet from line #4, sent as an incomplete IPv6 datagram,
920 eventually times out.
921
922 On the other hand, if the attacker succeeds to produce a collision of
923 Identification values, the following packet exchange could take
924 place:
925
926 A B C
927
928 #1 &lt;------- Echo Req #1 ----------
929 #2 --- Echo Repl #1, FID=5000 ---&gt;
930 #3 &lt;------------------- SYN #1, src= B -----------------------
931 #4 &lt;-- SYN/ACK, FID=9000 src=A ---
932 #5 ---- SYN/ACK, FID=9000 ---&gt;
933 ... (<a href="./rfc5722">RFC5722</a>) ...
934 #6 &lt;------- Echo Req #2 ----------
935 #7 ---- Echo Repl #2, FID=5001 --&gt;
936
937 Clearly, the Identification value sampled from the second ICMPv6 Echo
938 Reply packet ("Echo Repl #2") implicitly indicates whether the
939 Identification value in the forged SYN/ACK (see line #4 in both
940 figures) was the current Identification value in use by Host A.
941
942 As a result, the attacker could employ this technique to learn the
943 current Identification value used by host A to send packets to host
944 B, even when Host A itself has a non-vulnerable implementation.
945
946
947
948 <span class="grey">Gont Informational [Page 17]</span></pre>
949 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-18" ></span>
950 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
951
952
953 <span class="h2"><a class="selflink" id="appendix-B" href="#appendix-B">Appendix B</a>. Survey of Fragment Identification Selection Algorithms</span>
954 Employed by Popular IPv6 Implementations
955
956 This section includes a survey of the Identification selection
957 algorithms employed by some popular operating systems.
958
959 NOTE:
960 The survey was produced with the SI6 Networks' IPv6 toolkit
961 [<a href="#ref-SI6-IPv6" title="&quot;SI6 Networks' IPv6 Toolkit&quot;">SI6-IPv6</a>].
962
963 +------------------------------+------------------------------------+
964 | Operating System | Algorithm |
965 +------------------------------+------------------------------------+
966 | Cisco IOS 15.3 | Predictable (Global Counter, |
967 | | Init=0, Incr=1) |
968 +------------------------------+------------------------------------+
969 | FreeBSD 9.0 | Unpredictable (Random) |
970 +------------------------------+------------------------------------+
971 | Linux 3.0.0-15 | Predictable (Global Counter, |
972 | | Init=0, Incr=1) |
973 +------------------------------+------------------------------------+
974 | Linux-current | Unpredictable (Per-dest Counter, |
975 | | Init=random, Incr=1) |
976 +------------------------------+------------------------------------+
977 | NetBSD 5.1 | Unpredictable (Random) |
978 +------------------------------+------------------------------------+
979 | OpenBSD-current | Unpredictable (Random, SKIP32) |
980 +------------------------------+------------------------------------+
981 | Solaris 10 | Predictable (Per-dst Counter, |
982 | | Init=0, Incr=1) |
983 +------------------------------+------------------------------------+
984 | Windows XP SP2 | Predictable (Global Counter, |
985 | | Init=0, Incr=2) |
986 +------------------------------+------------------------------------+
987 | Windows XP Professional | Predictable (Global Counter, |
988 | 32bit, SP3 | Init=0, Incr=2) |
989 +------------------------------+------------------------------------+
990 | Windows Vista (Build 6000) | Predictable (Global Counter, |
991 | | Init=0, Incr=2) |
992 +------------------------------+------------------------------------+
993 | Windows Vista Business | Predictable (Global Counter, |
994 | 64bit, SP1 | Init=0, Incr=2) |
995 +------------------------------+------------------------------------+
996 | Windows 7 Home Premium | Predictable (Global Counter, |
997 | | Init=0, Incr=2) |
998 +------------------------------+------------------------------------+
999 | Windows Server 2003 R2 | Predictable (Global Counter, |
1000 | Standard 64bit, SP2 | Init=0, Incr=2) |
1001
1002
1003
1004 <span class="grey">Gont Informational [Page 18]</span></pre>
1005 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-19" ></span>
1006 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
1007
1008
1009 +------------------------------+------------------------------------+
1010 | Windows Server 2008 Standard | Predictable (Global Counter, |
1011 | 32bit, SP1 | Init=0, Incr=2) |
1012 +------------------------------+------------------------------------+
1013 | Windows Server 2008 R2 | Predictable (Global Counter, |
1014 | Standard 64bit, SP1 | Init=0, Incr=2) |
1015 +------------------------------+------------------------------------+
1016 | Windows Server 2012 Standard | Predictable (Global Counter, |
1017 | 64bit | Init=0, Incr=2) |
1018 +------------------------------+------------------------------------+
1019 | Windows 7 Home Premium | Predictable (Global Counter, |
1020 | 32bit, SP1 | Init=0, Incr=2) |
1021 +------------------------------+------------------------------------+
1022 | Windows 7 Ultimate 32bit, | Predictable (Global Counter, |
1023 | SP1 | Init=0, Incr=2) |
1024 +------------------------------+------------------------------------+
1025 | Windows 8 Enterprise 32 bit | Unpredictable (Alg. from Section |
1026 | | 5.3) |
1027 +------------------------------+------------------------------------+
1028
1029 Table 1: Fragment Identification algorithms employed by different OSs
1030
1031 NOTE:
1032 In the text above, "predictable" should be taken as "easily
1033 guessable by an off-path attacker, by sending a few probe
1034 packets".
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060 <span class="grey">Gont Informational [Page 19]</span></pre>
1061 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-20" ></span>
1062 <span class="grey"><a href="./rfc7739">RFC 7739</a> Implications of Predictable Fragment IDs February 2016</span>
1063
1064
1065 Acknowledgements
1066
1067 The author would like to thank Ivan Arce for proposing the attack
1068 scenario described in <a href="#appendix-A">Appendix A</a>.
1069
1070 The author would like to thank Ivan Arce, Stephen Bensley, Ron
1071 Bonica, Tassos Chatzithomaoglou, Guillermo Gont, Brian Haberman, Bob
1072 Hinden, Sheng Jiang, Tatuya Jinmei, Merike Kaeo, Will Liu, Juan
1073 Antonio Matos, Simon Perreault, Hosnieh Rafiee, Meral Shirazipour,
1074 Mark Smith, Dave Thaler, and Klaas Wierenga, for providing valuable
1075 comments on earlier draft versions of this document.
1076
1077 This document is based on work performed by Fernando Gont on behalf
1078 of the UK Centre for the Protection of National Infrastructure
1079 (CPNI).
1080
1081 The author would like to thank Buffy for her love and support.
1082
1083 Author's Address
1084
1085 Fernando Gont
1086 Huawei Technologies
1087 Evaristo Carriego 2644
1088 Haedo, Provincia de Buenos Aires 1706
1089 Argentina
1090
1091 Phone: +54 11 4650 8472
1092 Email: [email protected]
1093 URI: <a href="http://www.si6networks.com">http://www.si6networks.com</a>
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116 Gont Informational [Page 20]
1117 </pre>