idnits 2.17.1 draft-iab-identifier-comparison-06.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 1 instance of lines with non-RFC2606-compliant FQDNs in the document. == There are 2 instances of lines with private range IPv4 addresses in the document. If these are generic example addresses, they should be changed to use any of the ranges defined in RFC 6890 (or successor): 192.0.2.x, 198.51.100.x or 203.0.113.x. ** 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 418: '...dentity of an Internet host, it SHOULD...' RFC 2119 keyword, line 420: '...#.#.#.#") form. The host SHOULD check...' Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (December 14, 2012) is 4151 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Missing Reference: 'RFC5890' is mentioned on line 547, but not defined == Outdated reference: A later version (-09) exists of draft-ietf-precis-problem-statement-08 -- Obsolete informational reference (is this intentional?): RFC 3490 (Obsoleted by RFC 5890, RFC 5891) Summary: 1 error (**), 0 flaws (~~), 5 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group D. Thaler, Ed. 3 Internet-Draft Microsoft 4 Intended status: Informational December 14, 2012 5 Expires: June 17, 2013 7 Issues in Identifier Comparison for Security Purposes 8 draft-iab-identifier-comparison-06.txt 10 Abstract 12 Identifiers such as hostnames, URIs, and email addresses are often 13 used in security contexts to identify security principals and 14 resources. In such contexts, an identifier supplied via some 15 protocol is often compared against some policy to make security 16 decisions such as whether the principal may access the resource, what 17 level of authentication or encryption is required, etc. If the 18 parties involved in a security decision use different algorithms to 19 compare identifiers, then failure scenarios ranging from denial of 20 service to elevation of privilege can result. 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 June 17, 2013. 39 Copyright Notice 41 Copyright (c) 2012 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 1.1. Canonicalization . . . . . . . . . . . . . . . . . . . . . 4 58 2. Security Uses . . . . . . . . . . . . . . . . . . . . . . . . 5 59 2.1. Types of Identifiers . . . . . . . . . . . . . . . . . . . 6 60 2.2. False Positives and Negatives . . . . . . . . . . . . . . 7 61 2.3. Hypothetical Example . . . . . . . . . . . . . . . . . . . 8 62 3. Common Identifiers . . . . . . . . . . . . . . . . . . . . . . 9 63 3.1. Hostnames . . . . . . . . . . . . . . . . . . . . . . . . 9 64 3.1.1. IPv4 Literals . . . . . . . . . . . . . . . . . . . . 10 65 3.1.2. IPv6 Literals . . . . . . . . . . . . . . . . . . . . 11 66 3.1.3. Internationalization . . . . . . . . . . . . . . . . . 12 67 3.1.4. Resolution for comparison . . . . . . . . . . . . . . 12 68 3.2. Ports and Service Names . . . . . . . . . . . . . . . . . 13 69 3.3. URIs . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 70 3.3.1. Scheme component . . . . . . . . . . . . . . . . . . . 15 71 3.3.2. Authority component . . . . . . . . . . . . . . . . . 15 72 3.3.3. Path component . . . . . . . . . . . . . . . . . . . . 16 73 3.3.4. Query component . . . . . . . . . . . . . . . . . . . 16 74 3.3.5. Fragment component . . . . . . . . . . . . . . . . . . 16 75 3.3.6. Resolution for comparison . . . . . . . . . . . . . . 17 76 3.4. Email Address-like Identifiers . . . . . . . . . . . . . . 17 77 4. General Issues . . . . . . . . . . . . . . . . . . . . . . . . 18 78 4.1. Conflation . . . . . . . . . . . . . . . . . . . . . . . . 18 79 4.2. Internationalization . . . . . . . . . . . . . . . . . . . 18 80 4.3. Scope . . . . . . . . . . . . . . . . . . . . . . . . . . 19 81 4.4. Temporality . . . . . . . . . . . . . . . . . . . . . . . 20 82 5. Security Considerations . . . . . . . . . . . . . . . . . . . 20 83 6. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 21 84 7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 21 85 8. Informative References . . . . . . . . . . . . . . . . . . . . 21 86 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . . 24 88 1. Introduction 90 In computing and the Internet, various types of "identifiers" are 91 used to identify humans, devices, content, etc. Before discussing 92 security issues, we first give some background on some typical 93 processes involving identifiers. 95 As depicted in Figure 1, there are multiple processes relevant to our 96 discussion. 97 1. An identifier must first be generated. If the identifier is 98 intended to be unique, the generation process includes some 99 mechanism, such as allocation by a central authority, to help 100 ensure uniqueness. However the notion of "unique" involves 101 determining whether a putative identifier matches any other 102 already-allocated identifier. As we will see, for many types of 103 identifiers, this is not simply an exact binary match. 105 As a result of generating the identifier, it is often stored in 106 two locations: with the requester or "holder" of the identifier, 107 and with some repository of identifiers (e.g., DNS). For 108 example, if the identifier was allocated by a central authority, 109 the repository might be that authority. If the identifier 110 identifies a device or content on a device, the repository might 111 be that device. 112 2. The identifier must be distributed, either by the holder of the 113 identifier or by a repository of identifiers, to others who could 114 use the identifier. This distribution might be electronic, but 115 sometimes it is via other channels such as voice, business card, 116 billboard, or other form of advertisement. The identifier itself 117 might be distributed directly, or it might be used to generate a 118 portion of another type of identifier that is then distributed. 119 For example, a URI or email address might include a server name, 120 and hence distributing the URI or email address also inherently 121 distributes the server name. 122 3. The identifier must be used by some party. Generally the user 123 supplies the identifier which is (directly or indirectly) sent to 124 the repository of identifiers. For example, using an email 125 address to send email to the holder of an identifier may result 126 in the email arriving at the holder's email server which has 127 access to the mail stores. 129 The repository of identifiers must then attempt to match the 130 user-supplied identifier with an identifier in its repository. 132 +------------+ 133 | Holder of | 1. Generation 134 | identifier +<---------+ 135 +----+-------+ | 136 | | Match 137 | v/ 138 | +-------+-------+ 139 +----------+ Repository of | 140 | | identifiers | 141 | +-------+-------+ 142 2. Distribution | ^\ 143 | | Match 144 v | 145 +---------+-------+ | 146 | User of | | 147 | identifier +----------+ 148 +-----------------+ 3. Use 150 Typical Identifier Processes 152 Figure 1 154 One key aspect is that the identifier values passed in generation, 155 distribution, and use, may all be different forms. For example, 156 generation might be exchanged in printed form, distribution done via 157 voice, and use done electronically. As such, the match process can 158 be complicated. 160 Furthermore, in many uses, the relationship between holder, 161 repositories, and users may be more involved. For example, when a 162 hierarchy of web caches exist, each cache is itself a repository of a 163 sort, and the match process is usually intended to be the same as on 164 the origin server. 166 1.1. Canonicalization 168 Perhaps the most common algorithm for comparison involves first 169 converting each identifier to a canonical form (a process known as 170 "canonicalization" or "normalization"), and then testing the 171 resulting canonical representations for bitwise equality. In so 172 doing, it is thus critical that all entities involved agree on the 173 same canonical form and use the same canonicalization algorithm so 174 that the overall comparison process is also the same. 176 Note that in some contexts, such as in internationalization, the 177 terms "canonicalization" and "normalization" have a precise meaning. 178 In this document, however, we use these terms synonymously in their 179 more generic form, to mean conversion to some standard form. 181 While the most common method of comparison includes canonicalization, 182 comparison can also be done by defining an equivalence algorithm, 183 where no single form is canonical. However in most cases, a 184 canonical form is useful for other purposes, such as output, and so 185 in such cases defining a canonical form suffices to define a 186 comparison method. 188 2. Security Uses 190 Identifiers such as hostnames, URIs, and email addresses are used in 191 security contexts to identify principals and resources as well as 192 other security parameters such as types and values of claims. Those 193 identifiers are then used to make security decisions based on an 194 identifier supplied via some protocol. For example: 195 o Authentication: a protocol might match a security principal 196 identifier to look up expected keying material, and then match 197 keying material. 198 o Authorization: a protocol might match a resource name to look up 199 an access control list (ACL), and then look up the security 200 principal identifier (or a surrogate for it) in that ACL. 201 o Accounting: a system might create an accounting record for a 202 security principal identifier or resource name, and then might 203 later need to match a supplied identifier to (for example) add new 204 filtering rules based on the records in order to stop an attack. 206 If the parties involved in a security decision use different matching 207 algorithms for the same identifiers, then failure scenarios ranging 208 from denial of service to elevation of privilege can result, as we 209 will see. 211 This is especially complicated in cases involving multiple parties 212 and multiple protocols. For example, there are many scenarios where 213 some form of "security token service" is used to grant to a requester 214 permission to access a resource, where the resource is held by a 215 third party that relies on the security token service (see Figure 2). 216 The protocol used to request permission (e.g., Kerberos or OAuth) may 217 be different from the protocol used to access the resource (e.g., 218 HTTP). Opportunities for security problems arise when two protocols 219 define different comparison algorithms for the same type of 220 identifier, or when a protocol is ambiguously specified and two 221 endpoints (e.g., a security token service and a resource holder) 222 implement different algorithms within the same protocol. 224 +----------+ 225 | security | 226 | token | 227 | service | 228 +----------+ 229 ^ 230 | 1. supply credentials and 231 | get token for resource 232 | +--------+ 233 +----------+ 2. supply token and access resource |resource| 234 |requester |=------------------------------------->| holder | 235 +----------+ +--------+ 237 Simple Security Exchange 239 Figure 2 241 In many cases the situation is more complex. With certificates, the 242 name in a certificate gets compared against names in ACLs or other 243 things. In the case of web site security, the name in the 244 certificate gets compared to a portion of the URI that a user may 245 have typed into a browser. The fact that many different people are 246 doing the typing, on many different types of systems, complicates the 247 problem. 249 Add to this the certificate enrollment step, and the certificate 250 issuance step, and two more parties have an opportunity to adjust the 251 encoding or worse, the software that supports them might make changes 252 that the parties are unaware are happening. 254 2.1. Types of Identifiers 256 In this document we will refer to the following types of identifiers: 258 o Absolute: identifiers that can be compared byte-by-byte for 259 equality. Two identifiers that have different bytes are defined 260 to be different. For example, binary IP addresses are in this 261 class. 262 o Definite: identifiers that have a well-defined comparison 263 algorithm on which all parties agree. For example, URI scheme 264 names are required to be ASCII and are defined to match in a case- 265 insensitive way; the comparison is thus definite since all parties 266 agree on how to do a case-insensitive match among ASCII strings. 267 o Indefinite: identifiers that have no single comparison algorithm 268 on which all parties agree. For example, human names are in this 269 class. Everyone might want the comparison to be tailored for 270 their locale, for some definition of locale. In some cases, there 271 may be limited subsets of parties that might be able to agree 272 (e.g., ASCII users might all agree on a common comparison 273 algorithm whereas users of other Latin scripts, such as Turkish, 274 may not), but identifiers often tend to leak out of such limited 275 environments. 277 2.2. False Positives and Negatives 279 It is first worth discussing in more detail the effects of errors in 280 the comparison algorithm. A "false positive" results when two 281 identifiers compare as if they were equal, but in reality refer to 282 two different objects (e.g., security principals or resources). When 283 privilege is granted on a match, a false positive thus results in an 284 elevation of privilege, for example allowing execution of an 285 operation that should not have been permitted otherwise. When 286 privilege is denied on a match (e.g., matching an entry in a block/ 287 deny list or a revocation list), a permissible operation is denied. 288 At best, this can cause worse performance (e.g., a cache miss, or 289 forcing redundant authentication), and at worst can result in a 290 denial of service. 292 A "false negative" results when two identifiers that in reality refer 293 to the same thing compare as if they were different, and the effects 294 are the reverse of those for false positives. That is, when 295 privilege is granted on a match, the result is at best worse 296 performance and at worst a denial of service; when privilege is 297 denied on a match, elevation of privilege results. 299 Figure 3 summarizes these effects. 301 | "Grant on match" | "Deny on match" 302 ---------------+------------------------+----------------------- 303 False positive | Elevation of privilege | Denial of service 304 ---------------+------------------------+----------------------- 305 False negative | Denial of service | Elevation of privilege 306 ---------------+------------------------+----------------------- 308 Worst Effects of False Positives/Negatives 310 Figure 3 312 Elevation of privilege is almost always seen as far worse than denial 313 of service. Hence, for URIs for example, Section 6.1 of [RFC3986] 314 states: "comparison methods are designed to minimize false negatives 315 while strictly avoiding false positives". 317 Thus URIs were defined with a "grant privilege on match" paradigm in 318 mind, where it is critical to prevent elevation of privilege while 319 minimizing denial of service. Using URIs in a "deny privilege on 320 match" system can thus be problematic. 322 2.3. Hypothetical Example 324 In this example, both security principals and resources are 325 identified using URIs. Foo Corp has paid example.com for access to 326 the Stuff service. Foo Corp allows its employees to create accounts 327 on the Stuff service. Alice gets the account 328 "http://example.com/Stuff/FooCorp/alice" and Bob gets 329 "http://example.com/Stuff/FooCorp/bob". It turns out, however, that 330 Foo Corp's URI canonicalizer includes URI fragment components in 331 comparisons whereas example.com's does not, and Foo Corp does not 332 disallow the # character in the account name. So Chuck, who is a 333 malicious employee of Foo Corp, asks to create an account at 334 example.com with the name alice#stuff. Foo Corp's URI logic checks 335 its records for accounts it has created with stuff and sees that 336 there is no account with the name alice#stuff. Hence, in its 337 records, it associates the account alice#stuff with Chuck and will 338 only issue tokens good for use with 339 "http://example.com/Stuff/FooCorp/alice#stuff" to Chuck. 341 Chuck, the attacker, goes to a security token service at Foo Corp and 342 asks for a security token good for 343 "http://example.com/Stuff/FooCorp/alice#stuff". Foo Corp issues the 344 token since Chuck is the legitimate owner (in Foo Corp's view) of the 345 alice#stuff account. Chuck then submits the security token in a 346 request to "http://example.com/Stuff/FooCorp/alice". 348 But example.com uses a URI canonicalizer that, for the purposes of 349 checking equality, ignores fragments. So when example.com looks in 350 the security token to see if the requester has permission from Foo 351 Corp to access the given account it successfully matches the URI in 352 the security token, "http://example.com/Stuff/FooCorp/alice#stuff", 353 with the requested resource name 354 "http://example.com/Stuff/FooCorp/alice". 356 Leveraging the inconsistencies in the canonicalizers used by Foo Corp 357 and example.com, Chuck is able to successfully launch an elevation of 358 privilege attack and access Alice's resource. 360 Furthermore, consider an attacker using a similar corporation such as 361 "foocorp" (or any variation containing a non-ASCII character that 362 some humans might expect to represent the same corporation). If the 363 resource holder treats them as different, but the security token 364 service treats them as the same, then again elevation of privilege 365 can occur. 367 3. Common Identifiers 369 In this section, we walk through a number of common types of 370 identifiers and discuss various issues related to comparison that may 371 affect security whenever they are used to identify security 372 principals or resources. These examples illustrate common patterns 373 that may arise with other types of identifiers. 375 3.1. Hostnames 377 Hostnames (composed of dot-separated labels) are commonly used either 378 directly as identifiers, or as components in identifiers such as in 379 URIs and email addresses. Another example is in [RFC5280], sections 380 7.2 and 7.3 (and updated in section 3 of 381 [I-D.ietf-pkix-rfc5280-clarifications]), which specify use in 382 certificates. 384 In this section we discuss a number of issues in comparing strings 385 that appear to be some form of hostname. 387 It is first worth pointing out that the term itself is often 388 ambiguous, and hence it is important that any use clarify which 389 definition is intended. Some examples of definitions include: 390 a. A Fully-Qualified Domain Name (FQDN), 391 b. An FQDN that is associated with address records, 392 c. The leftmost label in an FQDN, or 393 d. The leftmost label in an FQDN that is associated with address 394 records. 396 The use of different definitions in different places results in 397 questions such as whether "example" and "example.com" are considered 398 equal or not. 400 Section 3 of [RFC6055] discusses the differences between a "hostname" 401 vs. a "DNS name", where the former is a subset of the latter by using 402 a restricted set of characters. If one canonicalizer uses the "DNS 403 name" definition whereas another uses a "hostname" definition, a name 404 might be valid in the former but invalid in the latter. As long as 405 invalid identifiers are denied privilege, this difference will not 406 result in elevation of privilege. 408 [IAB1123] briefly discusses issues with the ambiguity around whether 409 a label will be "alphabetic", including among other issues, how 410 "alphabetic" should be interpreted in an internationalized 411 environment, and whether a hostname can be interpreted as an IP 412 address. We explore this last issue in more detail below. 414 3.1.1. IPv4 Literals 416 [RFC1123] section 2.1 states: 418 Whenever a user inputs the identity of an Internet host, it SHOULD 419 be possible to enter either (1) a host domain name or (2) an IP 420 address in dotted-decimal ("#.#.#.#") form. The host SHOULD check 421 the string syntactically for a dotted-decimal number before 422 looking it up in the Domain Name System. 424 and 426 This last requirement is not intended to specify the complete 427 syntactic form for entering a dotted-decimal host number; that is 428 considered to be a user-interface issue. 430 In specifying the inet_addr() API, the POSIX standard [IEEE-1003.1] 431 defines "IPv4 dotted decimal notation" as allowing not only strings 432 of the form "10.0.1.2", but also allows octal and hexadecimal, and 433 addresses with less than four parts. For example, "10.0.258", 434 "0xA000001", and "012.0x102" all represent the same IPv4 address in 435 standard "IPv4 dotted decimal" notation. We will refer to this as 436 the "loose" syntax of an IPv4 address literal. 438 In section 6.1 of [RFC3493] getaddrinfo() is defined to support the 439 same (loose) syntax as inet_addr(): 441 If the specified address family is AF_INET or AF_UNSPEC, address 442 strings using Internet standard dot notation as specified in 443 inet_addr() are valid. 445 In contrast, section 6.3 of the same RFC states, specifying 446 inet_pton(): 448 If the af argument of inet_pton() is AF_INET, the src string shall 449 be in the standard IPv4 dotted-decimal form: ddd.ddd.ddd.ddd where 450 "ddd" is a one to three digit decimal number between 0 and 255. 451 The inet_pton() function does not accept other formats (such as 452 the octal numbers, hexadecimal numbers, and fewer than four 453 numbers that inet_addr() accepts). 455 As shown above, inet_pton() uses what we will refer to as the 456 "strict" form of an IPv4 address literal. Some platforms also use 457 the strict form with getaddrinfo() when the AI_NUMERICHOST flag is 458 passed to it. 460 Both the strict and loose forms are standard forms, and hence a 461 protocol specification is still ambiguous if it simply defines a 462 string to be in the "standard IPv4 dotted decimal form". And, as a 463 result of these differences, names such as "10.11.12" are ambiguous 464 as to whether they are an IP address or a hostname, and even 465 "10.11.12.13" can be ambiguous because of the "SHOULD" in RFC 1123 466 above making it optional whether to treat it as an address or a name. 468 Protocols and data formats that can use addresses in string form for 469 security purposes need to resolve these ambiguities. For example, 470 for the host component of URIs, section 3.2.2 of [RFC3986] resolves 471 the first ambiguity by only allowing the strict form, and the second 472 ambiguity by specifying that it is considered an IPv4 address 473 literal. New protocols and data formats should similarly consider 474 using the strict form rather than the loose form in order to better 475 match user expectations. 477 A string might be valid under the "loose" definition, but invalid 478 under the "strict" definition. As long as invalid identifiers are 479 denied privilege, this difference will not result in elevation of 480 privilege. Some protocols, however, use strings that can be either 481 an IP address literal or a hostname. Such strings are at best 482 Definite identifiers, and often turn out to be Indefinite 483 identifiers. (See Section 4.1 for more discussion.) 485 Furthermore, when strings can contain non-ASCII characters, they can 486 contain other characters that may look like dots or digits to a human 487 viewing and/or entering the identifier, especially to one who might 488 expect digits to appear in his or her native script. 490 3.1.2. IPv6 Literals 492 IPv6 addresses similarly have a wide variety of alternate but 493 semantically identical string representations, as defined in section 494 2.2 of [RFC4291] and section 2 of [I-D.ietf-6man-uri-zoneid]. As 495 discussed in section 3.2.5 of [RFC5952], this fact causes problems in 496 security contexts if comparison (such as in X.509 certificates), is 497 done between strings rather than between the binary representations 498 of addresses. 500 [RFC5952] recently specified a recommended canonical string format as 501 an attempt to solve this problem, but it may not be ubiquitously 502 supported at present. And, when strings can contain non-ASCII 503 characters, the same issues (and more, since hexadecimal and colons 504 are allowed) arise as with IPv4 literals. 506 Whereas (binary) IPv6 addresses are Absolute identifiers, IPv6 507 address literals are Definite identifiers, since string-to-address 508 conversion for IPv6 address literals is unambiguous. 510 3.1.3. Internationalization 512 The IETF policy on character sets and languages [RFC2277] requires 513 support for UTF-8 in protocols, and as a result many protocols now do 514 support non-ASCII characters. When a hostname is sent in a UTF-8 515 field, there are a number of ways it may be encoded. For example, 516 hostname labels might be encoded directly in UTF-8, or might first be 517 Punycode-encoded [RFC3492] or even percent-encoded from UTF-8. 519 For example, in URIs, [RFC3986] section 3.2.2 specifically allows for 520 the use of percent-encoded UTF-8 characters in the hostname, as well 521 as the use of IDNA encoding [RFC3490] using the Punycode algorithm. 523 Percent-encoding is unambiguous for hostnames since the percent 524 character cannot appear in the strict definition of a "hostname", 525 though it can appear in a DNS name. 527 Punycode-encoded labels (or "A-labels") on the other hand can be 528 ambiguous if hosts are actually allowed to be named with a name 529 starting with "xn--", and false positives can result. While this may 530 be extremely unlikely for normal scenarios, it nevertheless provides 531 a possible vector for an attacker. 533 A hostname comparator thus needs to decide whether a Punycode-encoded 534 label should or should not be considered a valid hostname label, and 535 if so, then whether it should match a label encoded in some other 536 form such as a percent-encoded Unicode label (U-label). 538 For example, Section 3 of "Transport Layer Security (TLS) Extensions" 539 [RFC6066], states: 541 "HostName" contains the fully qualified DNS hostname of the 542 server, as understood by the client. The hostname is represented 543 as a byte string using ASCII encoding without a trailing dot. 544 This allows the support of internationalized domain names through 545 the use of A-labels defined in [RFC5890]. DNS hostnames are case- 546 insensitive. The algorithm to compare hostnames is described in 547 [RFC5890], Section 2.3.2.4. 549 For some additional discussion of security issues that arise with 550 internationalization, see [TR36]. 552 3.1.4. Resolution for comparison 554 Some systems (specifically Java URLs [JAVAURL]) use the rule that if 555 two hostnames resolve to the same IP address(es) then the hostnames 556 are considered equal. That is, the canonicalization algorithm 557 involves name resolution with an IP address being the canonical form. 559 For example, if resolution was done via DNS, and DNS contained: 561 example.com. IN A 10.0.0.6 562 example.net. CNAME example.com. 563 example.org. IN A 10.0.0.6 565 then the algorithm might treat all three names as equal, even though 566 the third name might refer to a different entity. 568 With the introduction of dynamic IP addresses, private IP addresses, 569 multiple IP addresses per name, multiple address families (e.g., IPv4 570 vs. IPv6), devices that roam to new locations, commonly deployed DNS 571 tricks that result in the answer depending on factors such as the 572 requester's location and the load on the server whose address is 573 returned, etc., this method of comparison cannot be relied upon. 574 There is no guarantee that two names for the same host will resolve 575 the name to the same IP addresses, nor that the addresses resolved 576 refer to the same entity such as when the names resolve to private IP 577 addresses, nor even that the system has connectivity (and the 578 willingness to wait for the delay) to resolve names at the time the 579 answer is needed. 581 In addition, a comparison mechanism that relies on the ability to 582 resolve identifiers such as hostnames to other identifies such as IP 583 addresses leaks information about security decisions to outsiders if 584 these queries are publicly observable. 586 Finally, it is worth noting that resolving two identifiers to 587 determine if they refer to the same entity can be thought of as a use 588 of such identifiers, as opposed to actually comparing the identifiers 589 themselves, which is the focus of this document. 591 3.2. Ports and Service Names 593 Port numbers and service names are discussed in depth in [RFC6335]. 594 Historically, there were port numbers, service names used in SRV 595 records, and mnemonic identifiers for assigned port numbers (known as 596 port "keywords" at [IANA-PORT]). The latter two are now unified, and 597 various protocols use one or more of these types in strings. For 598 example, the common syntax used by many URI schemes allows port 599 numbers but not service names. Some implementations of the 600 getaddrinfo() API support strings that can be either port numbers or 601 port keywords (but not service names). 603 For protocols that use service names that must be resolved, the 604 issues are the same as those for resolution of addresses in 605 Section 3.1.4. In addition, Section 5.1 of [RFC6335] clarifies that 606 service names/port keywords must contain at least one letter. This 607 prevents confusion with port numbers in strings where both are 608 allowed. 610 3.3. URIs 612 This section looks at issues related to using URIs for security 613 purposes. For example, [RFC5280], section 7.4, specifies comparison 614 of URIs in certificates. Examples of URIs in security token-based 615 access control systems include WS-*, SAML-P and OAuth WRAP. In such 616 systems, a variety of participants in the security infrastructure are 617 identified by URIs. For example, requesters of security tokens are 618 sometimes identified with URIs. The issuers of security tokens and 619 the relying parties who are intended to consume security tokens are 620 frequently identified by URIs. Claims in security tokens often have 621 their types defined using URIs and the values of the claims can also 622 be URIs. 624 Also, when a URI is embedded in plain text (e.g., an email message), 625 there is an additional concern because there is no termination 626 criterion for a URI. For example, consider 627 http://unicode.org/cldr/utility/list-unicodeset.jsp?a=a&g=gc. 628 Some applications that detect URIs will stop before the first '.' in 629 the path, while others go to last '.', and yet others may stop at the 630 ';'. As another point of comparison, Section 2.37 of [EE] (a 631 standard for history citations) specifies the use of a space after a 632 URI and before the punctuation. 634 URIs are defined with multiple components, each of which has its own 635 rules. We cover each in turn below. However, it is also important 636 to note that there exist multiple comparison algorithms. [RFC3986] 637 section 6.2 states: 639 A variety of methods are used in practice to test URI equivalence. 640 These methods fall into a range, distinguished by the amount of 641 processing required and the degree to which the probability of 642 false negatives is reduced. As noted above, false negatives 643 cannot be eliminated. In practice, their probability can be 644 reduced, but this reduction requires more processing and is not 645 cost-effective for all applications. 646 If this range of comparison practices is considered as a ladder, 647 the following discussion will climb the ladder, starting with 648 practices that are cheap but have a relatively higher chance of 649 producing false negatives, and proceeding to those that have 650 higher computational cost and lower risk of false negatives. 652 The ladder approach has both pros and cons. On the pro side, it 653 allows some uses to optimize for security, and other uses to optimize 654 for cost, thus allowing URIs to be applicable to a wide range of 655 uses. A disadvantage is that when different approaches are taken by 656 different components in the same system using the same identifiers, 657 the inconsistencies can result in security issues. 659 3.3.1. Scheme component 661 [RFC3986] defines URI schemes as being case-insensitive ASCII and in 662 section 6.2.2.1 specifies that scheme names should be normalized to 663 lower-case characters. 665 New schemes can be defined over time. In general two URIs with an 666 unrecognized scheme cannot be safely compared, however. This is 667 because the canonicalization and comparison rules for the other 668 components may vary by scheme. For example, a new URI scheme might 669 have a default port of X, and without that knowledge, a comparison 670 algorithm cannot know whether "example.com" and "example.com:X" 671 should be considered to match in the authority component. Hence for 672 security purposes, it is safest for unrecognized schemes to be 673 treated as invalid identifiers. However, if the URIs are only used 674 with a "grant access on match" paradigm then unrecognized schemes can 675 be supported by doing a generic case-sensitive comparison, at the 676 expense of some false negatives. 678 3.3.2. Authority component 680 The authority component is scheme-specific, but many schemes follow a 681 common syntax that allows for userinfo, host, and port. 683 3.3.2.1. Host 685 Section 3.1 discussed issues with hostnames in general. In addition, 686 [RFC3986] section 3.2.2 allows future changes using the IPvFuture 687 production. As with IPv4 and IPv6 literals, IPvFuture formats may 688 have issues with multiple semantically identical string 689 representations, and may also be semantically identical to an IPv4 or 690 IPv6 address. As such, false negatives may be common if IPvFuture is 691 used. 693 3.3.2.2. Port 695 See discussion in Section 3.2. 697 3.3.2.3. Userinfo 699 [RFC3986] defines the userinfo production that allows arbitrary data 700 about the user of the URI to be placed before '@' signs in URIs. For 701 example: "http://alice:bob:chuck@example.com/bar" has the value 702 "alice:bob:chuck" as its userinfo. When comparing URIs in a security 703 context, one must decide whether to treat the userinfo as being 704 significant or not. Some URI comparison services for example treat 705 "http://alice:ick@example.com" and "http://example.com" as being 706 equal. 708 When the userinfo is treated as being significant, it has additional 709 considerations (e.g., whether it is case-sensitive or not) which we 710 cover in Section 3.4. 712 3.3.3. Path component 714 [RFC3986] supports the use of path segment values such as "./" or 715 "../" for relative URIs. Strictly speaking, including such path 716 segment values in a fully qualified URI is syntactically illegal but 717 [RFC3986] section 4.1 nevertheless defines an algorithm to remove 718 them. 720 Unless a scheme states otherwise, the path component is defined to be 721 case-sensitive. However, if the resource is stored and accessed 722 using a filesystem using case-insensitive paths, there will be many 723 paths that refer to the same resource. As such, false negatives can 724 be common in this case. 726 3.3.4. Query component 728 There is the question as to whether "http://example.com/foo", 729 "http://example.com/foo?", and "http://example.com/foo?bar" are each 730 considered equal or different. 732 Similarly, it is unspecified whether the order of values matters. 733 For example, should "http://example.com/blah?ick=bick&foo=bar" be 734 considered equal to "http://example.com/blah?foo=bar&ick=bick"? And 735 if a domain name is permitted to appear in a query component (e.g., 736 in a reference to another URI), the same issues in Section 3.1 apply. 738 3.3.5. Fragment component 740 Some URI formats include fragment identifiers. These are typically 741 handles to locations within a resource and are used for local 742 reference. A classic example is the use of fragments in HTTP URIs 743 where a URI of the form "http://example.com/blah.html#ick" means 744 retrieve the resource "http://example.com/blah.html" and, once it has 745 arrived locally, find the HTML anchor named ick and display that. 747 So, for example, when a user clicks on the link 748 "http://example.com/blah.html#baz" a browser will check its cache by 749 doing a URI comparison for "http://example.com/blah.html" and, if the 750 resource is present in the cache, a match is declared. 752 Hence comparisons for security purposes typically ignore the fragment 753 component and treat all fragments as equal to the full resource. 754 However, if one were actually trying to compare the piece of a 755 resource that was identified by the fragment identifier, ignoring it 756 would result in potential false positives. 758 3.3.6. Resolution for comparison 760 As with Section 3.1.4 for hostnames, it may be tempting to define a 761 URI comparison algorithm based on whether they resolve to the same 762 content. Similar problems exist, however, including content that 763 dynamically changes over time or based on factors such as the 764 requester's location, potential lack of external connectivity at the 765 time/place comparison is done, potentially undesirable delay 766 introduced, etc. 768 In addition, as noted in Section 3.1.4, resolution leaks information 769 about security decisions to outsiders if the queries are publicly 770 observable. 772 3.4. Email Address-like Identifiers 774 Section 3.4.1 of [RFC5322] defines the syntax of an email address- 775 like identifier, and Section 3.2 of [RFC6532] updates it to support 776 internationalization. [RFC5280], section 7.5, further discusses the 777 use of internationalized email addresses in certificates. 779 [RFC6532] use in certificates points to [RFC6530], where Section 13 780 of that document contains a discussion of many issues resulting from 781 internationalization. 783 Email address-like identifiers have a local part and a domain part. 784 The issues with the domain part are essentially the same as with 785 hostnames, covered earlier. 787 The local part is left for each domain to define. People quite 788 commonly use email addresses as usernames with web sites such as 789 banks or shopping sites, but the site doesn't know whether 790 foo@example.com is the same person as FOO@example.com. Thus email 791 address-like identifiers are typically Indefinite identifiers. 793 To avoid false positives, some security mechanisms (such as 794 [RFC5280]) compare the local part using an exact match. Hence, like 795 URIs, email address-like identifiers are designed for use in grant- 796 on-match security schemes, not in deny-on-match schemes. 798 Furthermore, if a mailbox is stored and accessed using a fileystem 799 using case-insensitive paths, there may be many paths that refer to 800 the same mailbox. As such, false negatives can be common in this 801 case. 803 4. General Issues 805 4.1. Conflation 807 There are a number of examples (some in the preceding sections) of 808 strings that conflate two types of identifiers, using some heuristic 809 to try to determine which type of identifier is given. Similarly, 810 two ways of encoding the same type of identifier might be conflated 811 within the same string. 813 Some examples include: 814 1. A string that might be an IPv4 address literal or an IPv6 address 815 literal 816 2. A string that might be an IP address literal or a hostname 817 3. A string that might be a port number or a service name 818 4. A DNS label that might be literal or be Punycode-encoded 820 Strings that allow such conflation can only be considered Definite if 821 there exists a well-defined rule to determine which identifier type 822 is meant. One way to do so is to ensure that the valid syntax for 823 the two is disjoint (e.g., distinguishing IPv4 vs. IPv6 address 824 literals by the use of colons in the latter). A second way to do so 825 is to define a precedence rule that results in some identifiers being 826 inaccessible via a conflated string (e.g., a host literally named 827 "xn--de-jg4avhby1noc0d" may be inaccessible due to the "xn--" prefix 828 denoting the use of Punycode encoding). In some cases, such 829 inaccessible space may be reserved so that the actual set of 830 identifiers in use are unambiguous. For example, Section 2.5.5.2 of 831 [RFC4291] defines a range of the IPv6 address space for representing 832 IPv4 addresses. 834 4.2. Internationalization 836 In addition to the issues with hostnames discussed in Section 3.1.3, 837 there are a number of internationalization issues that apply to many 838 types of Definite and Indefinite identifiers. 840 First, there is no DNS mechanism for identifying whether non- 841 identical strings would be seen by a human as being equivalent. 842 There are problematic examples even with ASCII (Basic Latin) strings 843 including regional spelling variations such as "color" and "colour" 844 and many non-English cases including partially-numeric strings in 845 Arabic script contexts, Chinese strings in Simplified and Traditional 846 forms, and so on. Attempts to produce such alternate forms 847 algorithmically could produce false positives and hence have an 848 adverse affect on security. 850 Second, some strings are visually confusable with others, and hence 851 if a security decision is made by a user based on visual inspection, 852 many opportunities for false positives exist. As such, using visual 853 inspection for security is unreliable. In addition to the security 854 issues, visual confusability also adversely affects the usability of 855 identifiers distributed via visual mediums. Similar issues can arise 856 with audible confusability when using audio (e.g., for radio 857 distribution, accessibility to the blind, etc.) in place of a visual 858 medium. 860 Determining whether a string is a valid identifier should typically 861 be done after, or as part of, canonicalization. Otherwise an 862 attacker might use the canonicalization algorithm to inject (e.g., 863 via percent encoding, NFKC, or non-shortest-form UTF-8) delimiters 864 such as '@' in an email address-like identifier, or a '.' in a 865 hostname. 867 Any case-insensitive comparisons need to define how comparison is 868 done, since such comparisons may vary by locale of the endpoint. As 869 such, using case-insensitive comparisons in general often result in 870 identifiers being either Indefinite or, if the legal character set is 871 restricted (e.g., to ASCII), then Definite. 873 See also [WEBER] for a more visual discussion of many of these 874 issues. 876 Finally, the set of permitted characters and the canonical form of 877 the characters (and hence the canonicalization algorithm) sometimes 878 varies by protocol today, even when the intent is to use the same 879 identifier, such as when one protocol passes identifiers to the 880 other. See [I-D.ietf-precis-problem-statement] for further 881 discussion. 883 4.3. Scope 885 Another issue arises when an identifier (e.g., "localhost", 886 "10.11.12.13", etc.) is not globally unique. [RFC3986] Section 1.1 887 states: 889 URIs have a global scope and are interpreted consistently 890 regardless of context, though the result of that interpretation 891 may be in relation to the end-user's context. For example, 892 "http://localhost/" has the same interpretation for every user of 893 that reference, even though the network interface corresponding to 894 "localhost" may be different for each end-user: interpretation is 895 independent of access. 897 Whenever a non-globally-unique identifier is passed to another entity 898 outside of the scope of uniqueness, it will refer to a different 899 resource, and can result in a false positive. This problem is often 900 addressed by using the identifier together with some other unique 901 identifier of the context. For example "alice" may uniquely identify 902 a user within a system, but must be used with "example.com" (as in 903 "alice@example.com") to uniquely identify the context outside of that 904 system. 906 It is also worth noting that non-globally-scoped IPv6 addresses can 907 be written with, or otherwise associated with, a "zone ID" to 908 identify the context (see [RFC4007] for more information). However, 909 zone IDs are only unique within a host, so they typically narrow, 910 rather than expand, the scope of uniqueness of the resulting 911 identifier. 913 4.4. Temporality 915 Often identifiers are not unique across all time, but have some 916 lifetime associated with them after which they may be reassigned to 917 another entity. For example, bob@example.com might go to to an 918 employee of the Example company, but if he leaves and another Bob is 919 later hired, the same identifier might be reused. As another 920 example, IP address 203.0.113.0 might be assigned to one subscriber, 921 and then later reassigned to another subscriber. If all entities 922 that store the identifier (e.g., in an access control list) are not 923 updated, security issues can arise. This issue is similar to the 924 issue of scope discussed in Section 4.3, except that the scope of 925 uniqueness is temporal rather than topological. 927 5. Security Considerations 929 This entire document is about security considerations. 931 To minimize elevation of privilege issues, any system that requires 932 the ability to use both deny and allow operations within the same 933 identifier space should avoid the use of Indefinite identifiers in 934 security comparisons. 936 To minimize future security risks, any new identifiers being designed 937 should specify an Absolute or Definite comparison algorithm, and if 938 extensibility is allowed (e.g., as new schemes in URIs allow) then 939 the comparison algorithm should remain invariant so that unrecognized 940 extensions can be compared. That is, security risks can be reduced 941 by specifying the comparison algorithm, making sure to resolve any 942 ambiguities pointed out in this document (e.g., "standard dotted 943 decimal"). 945 Some issues (such as unrecognized extensions) can be mitigated by 946 treating such identifiers as invalid. Validity checking of 947 identifiers is further discussed in [RFC3696]. 949 Perhaps the hardest issues arise when multiple protocols are used 950 together, such as in the figure in Section 2, where the two protocols 951 are defined or implemented using different comparison algorithms. 952 When constructing an architecture that uses multiple such protocols, 953 designers should pay attention to any differences in comparison 954 algorithms among the protocols, in order to fully understand the 955 security risks. An area for future work is how to deal with such 956 security risks in current systems. 958 6. Acknowledgements 960 Yaron Goland contributed to the discussion on URIs. Patrik Faltstrom 961 contributed to the background on identifiers. John Klensin 962 contributed text in a number of different sections. Additional 963 helpful feedback and suggestions came from Bernard Aboba, Leslie 964 Daigle, Mark Davis, Russ Housley, Christian Huitema, Magnus Nystrom, 965 and Chris Weber. 967 7. IANA Considerations 969 This document requires no actions by the IANA. 971 8. Informative References 973 [EE] Mills, E., "Evidence Explained: Citing History Sources 974 from Artifacts to Cyberspace", 2007. 976 [I-D.ietf-6man-uri-zoneid] 977 Carpenter, B., Cheshire, S., and R. Hinden, "Representing 978 IPv6 Zone Identifiers in Address Literals and Uniform 979 Resource Identifiers", draft-ietf-6man-uri-zoneid-06 (work 980 in progress), December 2012. 982 [I-D.ietf-pkix-rfc5280-clarifications] 983 Yee, P., "Updates to the Internet X.509 Public Key 984 Infrastructure Certificate and Certificate Revocation List 985 (CRL) Profile", draft-ietf-pkix-rfc5280-clarifications-11 986 (work in progress), November 2012. 988 [I-D.ietf-precis-problem-statement] 989 Blanchet, M. and A. Sullivan, "Stringprep Revision and 990 PRECIS Problem Statement", 991 draft-ietf-precis-problem-statement-08 (work in progress), 992 September 2012. 994 [IAB1123] IAB, "The interpretation of rules in the ICANN gTLD 995 Applicant Guidebook", February 2012, . 1000 [IANA-PORT] 1001 IANA, "PORT NUMBERS", June 2011, 1002 . 1004 [IEEE-1003.1] 1005 IEEE and The Open Group, "The Open Group Base 1006 Specifications, Issue 6 IEEE Std 1003.1, 2004 Edition", 1007 IEEE Std 1003.1, 2004. 1009 [JAVAURL] Oracle, "Class URL, Java(TM) Platform, Standard Ed. 7", 1010 2011, . 1013 [RFC1123] Braden, R., "Requirements for Internet Hosts - Application 1014 and Support", STD 3, RFC 1123, October 1989. 1016 [RFC2277] Alvestrand, H., "IETF Policy on Character Sets and 1017 Languages", BCP 18, RFC 2277, January 1998. 1019 [RFC3490] Faltstrom, P., Hoffman, P., and A. Costello, 1020 "Internationalizing Domain Names in Applications (IDNA)", 1021 RFC 3490, March 2003. 1023 [RFC3492] Costello, A., "Punycode: A Bootstring encoding of Unicode 1024 for Internationalized Domain Names in Applications 1025 (IDNA)", RFC 3492, March 2003. 1027 [RFC3493] Gilligan, R., Thomson, S., Bound, J., McCann, J., and W. 1028 Stevens, "Basic Socket Interface Extensions for IPv6", 1029 RFC 3493, February 2003. 1031 [RFC3696] Klensin, J., "Application Techniques for Checking and 1032 Transformation of Names", RFC 3696, February 2004. 1034 [RFC3986] Berners-Lee, T., Fielding, R., and L. Masinter, "Uniform 1035 Resource Identifier (URI): Generic Syntax", STD 66, 1036 RFC 3986, January 2005. 1038 [RFC4007] Deering, S., Haberman, B., Jinmei, T., Nordmark, E., and 1039 B. Zill, "IPv6 Scoped Address Architecture", RFC 4007, 1040 March 2005. 1042 [RFC4291] Hinden, R. and S. Deering, "IP Version 6 Addressing 1043 Architecture", RFC 4291, February 2006. 1045 [RFC5280] Cooper, D., Santesson, S., Farrell, S., Boeyen, S., 1046 Housley, R., and W. Polk, "Internet X.509 Public Key 1047 Infrastructure Certificate and Certificate Revocation List 1048 (CRL) Profile", RFC 5280, May 2008. 1050 [RFC5322] Resnick, P., Ed., "Internet Message Format", RFC 5322, 1051 October 2008. 1053 [RFC5952] Kawamura, S. and M. Kawashima, "A Recommendation for IPv6 1054 Address Text Representation", RFC 5952, August 2010. 1056 [RFC6055] Thaler, D., Klensin, J., and S. Cheshire, "IAB Thoughts on 1057 Encodings for Internationalized Domain Names", RFC 6055, 1058 February 2011. 1060 [RFC6066] Eastlake, D., "Transport Layer Security (TLS) Extensions: 1061 Extension Definitions", RFC 6066, January 2011. 1063 [RFC6335] Cotton, M., Eggert, L., Touch, J., Westerlund, M., and S. 1064 Cheshire, "Internet Assigned Numbers Authority (IANA) 1065 Procedures for the Management of the Service Name and 1066 Transport Protocol Port Number Registry", BCP 165, 1067 RFC 6335, August 2011. 1069 [RFC6530] Klensin, J. and Y. Ko, "Overview and Framework for 1070 Internationalized Email", RFC 6530, February 2012. 1072 [RFC6532] Yang, A., Steele, S., and N. Freed, "Internationalized 1073 Email Headers", RFC 6532, February 2012. 1075 [TR36] Unicode Consortium, "Unicode Security Considerations", 1076 Unicode Technical Report 36, August 2004. 1078 [WEBER] Weber, C., "Attacking Software Globalization", March 2010, 1079 . 1082 Author's Address 1084 Dave Thaler (editor) 1085 Microsoft Corporation 1086 One Microsoft Way 1087 Redmond, WA 98052 1088 USA 1090 Phone: +1 425 703 8835 1091 Email: dthaler@microsoft.com