idnits 2.17.1 draft-ietf-sasl-scram-04.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- ** The document seems to lack a License Notice according IETF Trust Provisions of 28 Dec 2009, Section 6.b.i or Provisions of 12 Sep 2009 Section 6.b -- however, there's a paragraph with a matching beginning. Boilerplate error? (You're using the IETF Trust Provisions' Section 6.b License Notice from 12 Feb 2009 rather than one of the newer Notices. See https://trustee.ietf.org/license-info/.) 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 : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (July 31, 2009) is 5376 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Missing Reference: 'I-D.ietf-sasl-gs2' is mentioned on line 1084, but not defined == Missing Reference: 'RFC2743' is mentioned on line 1089, but not defined == Missing Reference: 'RFC4121' is mentioned on line 1098, but not defined == Missing Reference: 'RFC3962' is mentioned on line 1092, but not defined == Missing Reference: 'RFC4401' is mentioned on line 1103, but not defined == Missing Reference: 'RFC4402' is mentioned on line 1107, but not defined ** Obsolete undefined reference: RFC 4402 (Obsoleted by RFC 7802) == Missing Reference: 'RFC4086' is mentioned on line 1095, but not defined == Missing Reference: 'RFCXXXX' is mentioned on line 895, but not defined ** Downref: Normative reference to an Informational RFC: RFC 2104 ** Downref: Normative reference to an Informational RFC: RFC 3174 ** Obsolete normative reference: RFC 3454 (Obsoleted by RFC 7564) ** Obsolete normative reference: RFC 4013 (Obsoleted by RFC 7613) == Outdated reference: A later version (-10) exists of draft-altman-tls-channel-bindings-05 == Outdated reference: A later version (-04) exists of draft-melnikov-sasl-scram-ldap-02 -- Obsolete informational reference (is this intentional?): RFC 2898 (Obsoleted by RFC 8018) -- Obsolete informational reference (is this intentional?): RFC 5246 (Obsoleted by RFC 8446) Summary: 6 errors (**), 0 flaws (~~), 11 warnings (==), 3 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 NETWORK WORKING GROUP A. Menon-Sen 3 Internet-Draft Oryx Mail Systems GmbH 4 Intended status: Standards Track A. Melnikov 5 Expires: February 1, 2010 Isode Ltd 6 C. Newman 7 N. Williams 8 Sun Microsystems 9 July 31, 2009 11 Salted Challenge Response (SCRAM) SASL Mechanism 12 draft-ietf-sasl-scram-04.txt 14 Status of this Memo 16 This Internet-Draft is submitted to IETF in full conformance with the 17 provisions of BCP 78 and BCP 79. 19 Internet-Drafts are working documents of the Internet Engineering 20 Task Force (IETF), its areas, and its working groups. Note that 21 other groups may also distribute working documents as Internet- 22 Drafts. 24 Internet-Drafts are draft documents valid for a maximum of six months 25 and may be updated, replaced, or obsoleted by other documents at any 26 time. It is inappropriate to use Internet-Drafts as reference 27 material or to cite them other than as "work in progress." 29 The list of current Internet-Drafts can be accessed at 30 http://www.ietf.org/ietf/1id-abstracts.txt. 32 The list of Internet-Draft Shadow Directories can be accessed at 33 http://www.ietf.org/shadow.html. 35 This Internet-Draft will expire on February 1, 2010. 37 Copyright Notice 39 Copyright (c) 2009 IETF Trust and the persons identified as the 40 document authors. All rights reserved. 42 This document is subject to BCP 78 and the IETF Trust's Legal 43 Provisions Relating to IETF Documents in effect on the date of 44 publication of this document (http://trustee.ietf.org/license-info). 45 Please review these documents carefully, as they describe your rights 46 and restrictions with respect to this document. 48 Abstract 50 The secure authentication mechanism most widely deployed and used by 51 Internet application protocols is the transmission of clear-text 52 passwords over a channel protected by Transport Layer Security (TLS). 53 There are some significant security concerns with that mechanism, 54 which could be addressed by the use of a challenge response 55 authentication mechanism protected by TLS. Unfortunately, the 56 challenge response mechanisms presently on the standards track all 57 fail to meet requirements necessary for widespread deployment, and 58 have had success only in limited use. 60 This specification describes a family of Simple Authentication and 61 Security Layer (SASL, RFC 4422) authentication mechanisms called the 62 Salted Challenge Response Authentication Mechanism (SCRAM), which 63 addresses the security concerns and meets the deployability 64 requirements. When used in combination with TLS or an equivalent 65 security layer, a mechanism from this family could improve the 66 status-quo for application protocol authentication and provide a 67 suitable choice for a mandatory-to-implement mechanism for future 68 application protocol standards. 70 Table of Contents 72 1. Conventions Used in This Document . . . . . . . . . . 4 73 1.1. Terminology . . . . . . . . . . . . . . . . . . . . . 4 74 1.2. Notation . . . . . . . . . . . . . . . . . . . . . . . 5 75 2. Introduction . . . . . . . . . . . . . . . . . . . . . 7 76 3. SCRAM Algorithm Overview . . . . . . . . . . . . . . . 9 77 4. SCRAM Mechanism Names . . . . . . . . . . . . . . . . 10 78 5. SCRAM Authentication Exchange . . . . . . . . . . . . 11 79 5.1. SCRAM Attributes . . . . . . . . . . . . . . . . . . . 12 80 6. Channel Binding . . . . . . . . . . . . . . . . . . . 15 81 6.1. Default Channel Binding . . . . . . . . . . . . . . . 16 82 7. Formal Syntax . . . . . . . . . . . . . . . . . . . . 17 83 8. SCRAM as a GSS-API Mechanism . . . . . . . . . . . . . 20 84 8.1. GSS-API Principal Name Types for SCRAM . . . . . . . . 20 85 8.2. GSS-API Per-Message Tokens for SCRAM . . . . . . . . . 20 86 8.3. GSS_Pseudo_random() for SCRAM . . . . . . . . . . . . 21 87 9. Security Considerations . . . . . . . . . . . . . . . 22 88 10. IANA Considerations . . . . . . . . . . . . . . . . . 24 89 11. Acknowledgements . . . . . . . . . . . . . . . . . . . 26 90 Appendix A. Other Authentication Mechanisms . . . . . . . . . . . 27 91 Appendix B. Design Motivations . . . . . . . . . . . . . . . . . . 28 92 Appendix C. Internet-Draft Change History . . . . . . . . . . . . 29 93 12. References . . . . . . . . . . . . . . . . . . . . . . 31 94 12.1. Normative References . . . . . . . . . . . . . . . . . 31 95 12.2. Normative References for GSS-API implementors . . . . 31 96 12.3. Informative References . . . . . . . . . . . . . . . . 32 97 Authors' Addresses . . . . . . . . . . . . . . . . . . 34 99 1. Conventions Used in This Document 101 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 102 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 103 document are to be interpreted as described in [RFC2119]. 105 Formal syntax is defined by [RFC5234] including the core rules 106 defined in Appendix B of [RFC5234]. 108 Example lines prefaced by "C:" are sent by the client and ones 109 prefaced by "S:" by the server. If a single "C:" or "S:" label 110 applies to multiple lines, then the line breaks between those lines 111 are for editorial clarity only, and are not part of the actual 112 protocol exchange. 114 1.1. Terminology 116 This document uses several terms defined in [RFC4949] ("Internet 117 Security Glossary") including the following: authentication, 118 authentication exchange, authentication information, brute force, 119 challenge-response, cryptographic hash function, dictionary attack, 120 eavesdropping, hash result, keyed hash, man-in-the-middle, nonce, 121 one-way encryption function, password, replay attack and salt. 122 Readers not familiar with these terms should use that glossary as a 123 reference. 125 Some clarifications and additional definitions follow: 127 o Authentication information: Information used to verify an identity 128 claimed by a SCRAM client. The authentication information for a 129 SCRAM identity consists of salt, iteration count, the "StoredKey" 130 and "ServerKey" (as defined in the algorithm overview) for each 131 supported cryptographic hash function. 133 o Authentication database: The database used to look up the 134 authentication information associated with a particular identity. 135 For application protocols, LDAPv3 (see [RFC4510]) is frequently 136 used as the authentication database. For network-level protocols 137 such as PPP or 802.11x, the use of RADIUS is more common. 139 o Base64: An encoding mechanism defined in [RFC4648] which converts 140 an octet string input to a textual output string which can be 141 easily displayed to a human. The use of base64 in SCRAM is 142 restricted to the canonical form with no whitespace. 144 o Octet: An 8-bit byte. 146 o Octet string: A sequence of 8-bit bytes. 148 o Salt: A random octet string that is combined with a password 149 before applying a one-way encryption function. This value is used 150 to protect passwords that are stored in an authentication 151 database. 153 1.2. Notation 155 The pseudocode description of the algorithm uses the following 156 notations: 158 o ":=": The variable on the left hand side represents the octet 159 string resulting from the expression on the right hand side. 161 o "+": Octet string concatenation. 163 o "[ ]": A portion of an expression enclosed in "[" and "]" may not 164 be included in the result under some circumstances. See the 165 associated text for a description of those circumstances. 167 o HMAC(key, str): Apply the HMAC keyed hash algorithm (defined in 168 [RFC2104]) using the octet string represented by "key" as the key 169 and the octet string "str" as the input string. The size of the 170 result is the hash result size for the hash function in use. For 171 example, it is 20 octets for SHA-1 (see [RFC3174]). 173 o H(str): Apply the cryptographic hash function to the octet string 174 "str", producing an octet string as a result. The size of the 175 result depends on the hash result size for the hash function in 176 use. 178 o XOR: Apply the exclusive-or operation to combine the octet string 179 on the left of this operator with the octet string on the right of 180 this operator. The length of the output and each of the two 181 inputs will be the same for this use. 183 o Hi(str, salt): 185 U0 := HMAC(str, salt + INT(1)) 186 U1 := HMAC(str, U0) 187 U2 := HMAC(str, U1) 188 ... 189 Ui-1 := HMAC(str, Ui-2) 190 Ui := HMAC(str, Ui-1) 192 Hi := U0 XOR U1 XOR U2 XOR ... XOR Ui 193 where "i" is the iteration count, "+" is the string concatenation 194 operator and INT(g) is a four-octet encoding of the integer g, 195 most significant octet first. 197 o This is, essentially, PBKDF2 [RFC2898] with HMAC() as the PRF and 198 with dkLen == output length of HMAC() == output length of H(). 200 2. Introduction 202 This specification describes a family of authentication mechanisms 203 called the Salted Challenge Response Authentication Mechanism (SCRAM) 204 which addresses the requirements necessary to deploy a challenge- 205 response mechanism more widely than past attempts. When used in 206 combination with Transport Layer Security (TLS, see [RFC5246]) or an 207 equivalent security layer, a mechanism from this family could improve 208 the status-quo for application protocol authentication and provide a 209 suitable choice for a mandatory-to-implement mechanism for future 210 application protocol standards. 212 For simplicity, this family of mechanisms does not presently include 213 negotiation of a security layer [RFC4422]. It is intended to be used 214 with an external security layer such as that provided by TLS or SSH, 215 with optional channel binding [RFC5056] to the external security 216 layer. 218 SCRAM is specified herein as a pure Simple Authentication and 219 Security Layer (SASL) [RFC4422] mechanism, but it conforms to the new 220 bridge between SASL and the Generic Security Services Application 221 Programming Interface (GSS-API) called "GS2" [I-D.ietf-sasl-gs2]. 222 This means that this document defines both, a SASL mechanism and a 223 GSS-API mechanism. 225 SCRAM provides the following protocol features: 227 o The authentication information stored in the authentication 228 database is not sufficient by itself to impersonate the client. 229 The information is salted to prevent a pre-stored dictionary 230 attack if the database is stolen. 232 o The server does not gain the ability to impersonate the client to 233 other servers (with an exception for server-authorized proxies). 235 o The mechanism permits the use of a server-authorized proxy without 236 requiring that proxy to have super-user rights with the back-end 237 server. 239 o Mutual authentication is supported, but only the client is named 240 (i.e., the server has no name). 242 A separate document defines a standard LDAPv3 [RFC4510] attribute 243 that enables storage of the SCRAM authentication information in LDAP. 244 See [I-D.melnikov-sasl-scram-ldap]. 246 For an in-depth discussion of why other challenge response mechanisms 247 are not considered sufficient, see appendix A. For more information 248 about the motivations behind the design of this mechanism, see 249 appendix B. 251 Comments regarding this draft may be sent either to the 252 ietf-sasl@imc.org mailing list or to the authors. 254 3. SCRAM Algorithm Overview 256 Note that this section omits some details, such as client and server 257 nonces. See Section 5 for more details. 259 To begin with, the SCRAM client is in possession of a username and 260 password. It sends the username to the server, which retrieves the 261 corresponding authentication information, i.e. a salt, StoredKey, 262 ServerKey and the iteration count i. (Note that a server 263 implementation may chose to use the same iteration count for all 264 accounts.) The server sends the salt and the iteration count to the 265 client, which then computes the following values and sends a 266 ClientProof to the server: 268 SaltedPassword := Hi(password, salt) 269 ClientKey := HMAC(SaltedPassword, "Client Key") 270 StoredKey := H(ClientKey) 271 AuthMessage := client-first-message-bare + "," + 272 server-first-message + "," + 273 client-final-message-without-proof 274 ClientSignature := HMAC(StoredKey, AuthMessage) 275 ClientProof := ClientKey XOR ClientSignature 276 ServerKey := HMAC(SaltedPassword, "Server Key") 277 ServerSignature := HMAC(ServerKey, AuthMessage) 279 The server authenticates the client by computing the ClientSignature, 280 exclusive-ORing that with the ClientProof to recover the ClientKey 281 and verifying the correctness of the ClientKey by applying the hash 282 function and comparing the result to the StoredKey. If the ClientKey 283 is correct, this proves that the client has access to the user's 284 password. 286 Similarly, the client authenticates the server by computing the 287 ServerSignature and comparing it to the value sent by the server. If 288 the two are equal, it proves that the server had access to the user's 289 ServerKey. 291 The AuthMessage is computed by concatenating messages from the 292 authentication exchange. The format of these messages is defined in 293 Section 7. 295 4. SCRAM Mechanism Names 297 A SCRAM mechanism name is a string "SCRAM-" followed by the 298 uppercased name of the underlying hash function taken from the IANA 299 "Hash Function Textual Names" registry (see http://www.iana.org), 300 optionally followed by the suffix "-PLUS" (see below). Note that 301 SASL mechanism names are limited to 20 characters, which means that 302 only hash function names with lengths shorter or equal to 9 303 characters (20-length("SCRAM-")-length("-PLUS") can be used. For 304 cases when the underlying hash function name is longer than 9 305 characters, an alternative 9 character (or shorter) name can be used 306 to construct the corresponding SCRAM mechanism name, as long as this 307 alternative name doesn't conflict with any other hash function name 308 from the IANA "Hash Function Textual Names" registry. 310 For interoperability, all SCRAM clients and servers MUST implement 311 the SCRAM-SHA-1 authentication mechanism, i.e. an authentication 312 mechanism from the SCRAM family that uses the SHA-1 hash function as 313 defined in [RFC3174]. 315 The "-PLUS" suffix is used only when the server supports channel 316 binding to the external channel. If the server supports channel 317 binding, it will advertise both the "bare" and "plus" versions of 318 whatever mechanisms it supports (e.g., if the server supports only 319 SCRAM with SHA-1 then it will advertise support for both SCRAM-SHA-1 320 and SCRAM-SHA-1-PLUS); if the server does not support channel 321 binding, then it will advertise only the "bare" version of the 322 mechanism (e.g., only SCRAM-SHA-1). The "-PLUS" exists to allow 323 negotiation of the use of channel binding. See Section 6. 325 5. SCRAM Authentication Exchange 327 SCRAM is a SASL mechanism whose client response and server challenge 328 messages are text-based messages containing one or more attribute- 329 value pairs separated by commas. Each attribute has a one-letter 330 name. The messages and their attributes are described in 331 Section 5.1, and defined in Section 7. 333 This is a simple example of a SCRAM-SHA-1 authentication exchange 334 when the client doesn't support channel bindings: 336 C: n,,n=Chris Newman,r=ClientNonce 337 S: r=ClientNonceServerNonce,s=PxR/wv+epq,i=128 338 C: c=biwsCg==,r=ClientNonceServerNonce,p=WxPv/siO5l+qxN4 339 S: v=WxPv/siO5l+qxN4 341 [[anchor5: Note that the all hashes above are fake and will be fixed 342 during AUTH48.]] 344 With channel-binding data sent by the client this might look like 345 this (see [tls-server-end-point] for the definition of tls-server- 346 end-point TLS channel binding): 348 C: p=tls-server-end-point,,n=Chris Newman,r=ClientNonce 349 S: r=ClientNonceServerNonce,s=PxR/wv+epq,i=128 350 C: c=cD10bHMtc2VydmVyLWVuZC1wb2ludCwsy1hFtXOnZ+ySrQM6srFp 351 l/77uqvtxrg7nBY1BetEr/g=,r=ClientNonceServerNonce,p=Wx 352 Pv/siO5l+qxN4 353 S: v=WxPv/siO5l+qxN4 355 [[anchor6: Note that all hashes above are fake and will be fixed 356 during AUTH48.]] 358 First, the client sends a message containing: 360 o a GS2 header consisting of a flag indicating whether channel 361 binding is supported-but-not-used, not supported, or used, and an 362 optional SASL authorization identity; 364 o SCRAM username and a random, unique nonce attributes. 366 Note that the client's first message will always start with "n", "y" 367 or "p", otherwise the message is invalid and authentication MUST 368 fail. This is important, as it allows for GS2 extensibility (e.g., 369 to add support for security layers). 371 In response, the server sends the user's iteration count i, the 372 user's salt, and appends its own nonce to the client-specified one. 373 The client then responds with the same nonce and a ClientProof 374 computed using the selected hash function as explained earlier. The 375 server verifies the nonce and the proof, verifies that the 376 authorization identity (if supplied by the client in the first 377 message) is authorized to act as the authentication identity, and, 378 finally, it responds with a ServerSignature, concluding the 379 authentication exchange. The client then authenticates the server by 380 computing the ServerSignature and comparing it to the value sent by 381 the server. If the two are different, the client MUST consider the 382 authentication exchange to be unsuccessful and it might have to drop 383 the connection. 385 5.1. SCRAM Attributes 387 This section describes the permissible attributes, their use, and the 388 format of their values. All attribute names are single US-ASCII 389 letters and are case-sensitive. 391 Note that the order of attributes in client or server messages is 392 fixed, with the exception of extension attributes (described by the 393 "extensions" ABNF production), which can appear in any order in the 394 designated positions. See the ABNF section for authoritative 395 reference. 397 o a: This is an optional attribute, and is part of the GS2 398 [I-D.ietf-sasl-gs2] bridge between the GSS-API and SASL. This 399 attribute specifies an authorization identity. A client may 400 include it in its first message to the server if it wants to 401 authenticate as one user, but subsequently act as a different 402 user. This is typically used by an administrator to perform some 403 management task on behalf of another user, or by a proxy in some 404 situations. 406 Upon the receipt of this value the server verifies its 407 correctness according to the used SASL protocol profile. 408 Failed verification results in failed authentication exchange. 410 If this attribute is omitted (as it normally would be), the 411 authorization identity is assumed to be derived from the 412 username specified with the (required) "n" attribute. 414 The server always authenticates the user specified by the "n" 415 attribute. If the "a" attribute specifies a different user, 416 the server associates that identity with the connection after 417 successful authentication and authorization checks. 419 The syntax of this field is the same as that of the "n" field 420 with respect to quoting of '=' and ','. 422 o n: This attribute specifies the name of the user whose password is 423 used for authentication (a.k.a. "authentication identity" 424 [RFC4422]). A client MUST include it in its first message to the 425 server. If the "a" attribute is not specified (which would 426 normally be the case), this username is also the identity which 427 will be associated with the connection subsequent to 428 authentication and authorization. 430 Before sending the username to the server, the client MUST 431 prepare the username using the "SASLPrep" profile [RFC4013] of 432 the "stringprep" algorithm [RFC3454]. If the preparation of 433 the username fails or results in an empty string, the client 434 SHOULD abort the authentication exchange (*). 436 (*) An interactive client can request a repeated entry of the 437 username value. 439 Upon receipt of the username by the server, the server SHOULD 440 prepare it using the "SASLPrep" profile [RFC4013] of the 441 "stringprep" algorithm [RFC3454]. If the preparation of the 442 username fails or results in an empty string, the server SHOULD 443 abort the authentication exchange. 445 The characters ',' or '=' in usernames are sent as '=2C' and 446 '=3D' respectively. If the server receives a username which 447 contains '=' not followed by either '2C' or '3D', then the 448 server MUST fail the authentication. 450 o m: This attribute is reserved for future extensibility. In this 451 version of SCRAM, its presence in a client or a server message 452 MUST cause authentication failure when the attribute is parsed by 453 the other end. 455 o r: This attribute specifies a sequence of random printable 456 characters excluding ',' which forms the nonce used as input to 457 the hash function. No quoting is applied to this string. As 458 described earlier, the client supplies an initial value in its 459 first message, and the server augments that value with its own 460 nonce in its first response. It is important that this value be 461 different for each authentication. The client MUST verify that 462 the initial part of the nonce used in subsequent messages is the 463 same as the nonce it initially specified. The server MUST verify 464 that the nonce sent by the client in the second message is the 465 same as the one sent by the server in its first message. 467 o c: This REQUIRED attribute specifies base64-encoded of a header 468 and the channel-binding data. It is sent by the client in its 469 second authentication message. The header consist of: 471 * the GS2 header from the client's first message (recall: a 472 channel binding flag and an optional authzid). This header is 473 going to include channel binding type prefix (see [RFC5056]), 474 if and only if the client is using channel binding; 476 * followed by the external channel's channel binding data, if and 477 only if the client is using channel binding. 479 o s: This attribute specifies the base64-encoded salt used by the 480 server for this user. It is sent by the server in its first 481 message to the client. 483 o i: This attribute specifies an iteration count for the selected 484 hash function and user, and MUST be sent by the server along with 485 the user's salt. 487 For SCRAM-SHA-1/SCRAM-SHA-1-PLUS SASL mechanism servers SHOULD 488 announce a hash iteration-count of at least 4096. Note that a 489 client implementation MAY cache SaltedPassword/ClientKey for 490 later reauthentication to the same service, as it is likely 491 that the server is going to advertise the same salt value upon 492 reauthentication. This might be useful for mobile clients 493 where CPU usage is a concern. 495 o p: This attribute specifies a base64-encoded ClientProof. The 496 client computes this value as described in the overview and sends 497 it to the server. 499 o v: This attribute specifies a base64-encoded ServerSignature. It 500 is sent by the server in its final message, and is used by the 501 client to verify that the server has access to the user's 502 authentication information. This value is computed as explained 503 in the overview. 505 6. Channel Binding 507 SCRAM supports channel binding to external secure channels, such as 508 TLS. Clients and servers may or may not support channel binding, 509 therefore the use of channel binding is negotiable. SCRAM does not 510 provide security layers, however, therefore it is imperative that 511 SCRAM provide integrity protection for the negotiation of channel 512 binding. 514 Use of channel binding is negotiated as follows: 516 o Servers SHOULD advertise both non-PLUS (SCRAM-) and 517 the PLUS-variant (SCRAM--PLUS) SASL mechanism 518 names. If the server cannot support channel binding, it MAY 519 advertise only the non-PLUS variant. If the server would never 520 succeed authentication of the non-PLUS variant due to policy 521 reasons, it MAY advertise only the PLUS-variant. 523 o If the client negotiates mechanisms then the client MUST select 524 SCRAM--PLUS if offered by the server and the client 525 wants to select SCRAM with the given hash function. Otherwise 526 (the client does not negotiate mechanisms), if the client has no 527 prior knowledge about mechanisms supported by the server and 528 wasn't explicitly configured to use a particular variant of the 529 SCRAM mechanism, then it MUST select only SCRAM- 530 (not suffixed with "-PLUS"). 532 o If the client supports channel binding and the server appears to 533 support it (i.e., the client sees SCRAM--PLUS), or 534 if the client wishes to use channel binding but the client does 535 not negotiate mechanisms, then the client MUST set the GS2 channel 536 binding flag to "p" in order to indicate the channel binding type 537 it is using and it MUST include the channel binding data for the 538 external channel in the computation of the "c=" attribute (see 539 Section 5.1). 541 o If the client supports channel binding but the server does not 542 appear to (i.e., the client did not see SCRAM-- 543 PLUS) then the client MUST either fail authentication or it MUST 544 choose the non-PLUS mechanism and set the GS2 channel binding flag 545 to "y" and MUST NOT include channel binding data for the external 546 channel in the computation of the "c=" attribute (see 547 Section 5.1). 549 o If the client does not support channel binding then the client 550 MUST set the GS2 channel binding flag to "n" and MUST NOT include 551 channel binding data for the external channel in the computation 552 of the "c=" attribute (see Section 5.1). 554 o Upon receipt of the client first message the server checks the GS2 555 channel binding flag (gs2-cb-flag). 557 * If the flag is set to "y" and the server supports channel 558 binding the server MUST fail authentication. This is because 559 if the client sets the GS2 channel binding flag set to "y" then 560 the client must have believed that the server did not support 561 channel binding -- if the server did in fact support channel 562 binding then this is an indication that there has been a 563 downgrade attack (e.g., an attacker changed the server's 564 mechanism list to exclude the -PLUS suffixed SCRAM mechanism 565 name(s)). 567 * If the channel binding flag was "p" and the server does not 568 support the indicated channel binding type then the server MUST 569 fail authentication. 571 The server MUST always validate the client's "c=" field. The server 572 does this by constructing the value of the "c=" attribute and then 573 checking that it matches the client's c= attribute value. 575 For more discussions of channel bindings, and the syntax of the 576 channel binding data for various security protocols, see [RFC5056]. 578 6.1. Default Channel Binding 580 A default channel binding type agreement process for all SASL 581 application protocols that do not provide their own channel binding 582 type agreement is provided as follows. 584 'tls-unique' is the default channel binding type for any application 585 that doesn't specify one. 587 Servers MUST implement the "tls-unique" [tls-unique] 588 [I-D.altman-tls-channel-bindings] channel binding type, if they 589 implement any channel binding. Clients SHOULD implement the "tls- 590 unique" [tls-unique] [I-D.altman-tls-channel-bindings] channel 591 binding type, if they implement any channel binding. Clients and 592 servers SHOULD choose the highest- layer/innermost end-to-end TLS 593 channel as the channel to bind to. 595 Servers MUST choose the channel binding type indicated by the client, 596 or fail authentication if they don't support it. 598 7. Formal Syntax 600 The following syntax specification uses the Augmented Backus-Naur 601 Form (ABNF) notation as specified in [RFC5234]. "UTF8-2", "UTF8-3" 602 and "UTF8-4" non-terminal are defined in [RFC3629]. 604 ALPHA = 605 DIGIT = 606 UTF8-2 = 607 UTF8-3 = 608 UTF8-4 = 610 attr-val = ALPHA "=" value 611 ;; Generic syntax of any attribute sent 612 ;; by server or client 614 value = 1*value-char 616 value-safe-char = %x01-2B / %x2D-3C / %x3E-7F / 617 UTF8-2 / UTF8-3 / UTF8-4 618 ;; UTF8-char except NUL, "=", and ",". 620 value-char = value-safe-char / "=" 622 base64-char = ALPHA / DIGIT / "/" / "+" 624 base64-4 = 4base64-char 626 base64-3 = 3base64-char "=" 628 base64-2 = 2base64-char "==" 630 base64 = *base64-4 [base64-3 / base64-2] 632 posit-number = %x31-39 *DIGIT 633 ;; A positive number 635 saslname = 1*(value-safe-char / "=2C" / "=3D") 636 ;; Conforms to 638 authzid = "a=" saslname 639 ;; Protocol specific. 641 cb-name = 1*(ALPHA / DIGIT / "." / "-") 642 ;; See RFC 5056 section 7. 643 ;; E.g. "tls-server-end-point" or 644 ;; "tls-unique" 646 gs2-cbind-flag = "p=" cb-name / "n" / "y" 647 ;; "n" -> client doesn't support channel binding 648 ;; "y" -> client does support channel binding 649 ;; but thinks the server does not. 650 ;; "p" -> client requires channel binding. 651 ;; The selected channel binding follows "p=". 653 gs2-header = gs2-cbind-flag "," [ authzid ] "," 654 ;; GS2 header for SCRAM 655 ;; (the actual GS2 header includes an optional 656 ;; flag to indicate that the GSS mechanism is not 657 ;; "standard" but since SCRAM is "standard" we 658 ;; don't include that flag). 660 username = "n=" saslname 661 ;; Usernames are prepared using SASLPrep. 663 reserved-mext = "m=" 1*(value-char) 664 ;; Reserved for signalling mandatory extensions. 665 ;; The exact syntax will be defined in 666 ;; the future. 668 channel-binding = "c=" base64 669 ;; base64 encoding of cbind-input 671 proof = "p=" base64 673 nonce = "r=" c-nonce [s-nonce] 674 ;; Second part provided by server. 676 c-nonce = value 678 s-nonce = value 680 salt = "s=" base64 682 verifier = "v=" base64 683 ;; base-64 encoded ServerSignature. 685 iteration-count = "i=" posit-number 686 ;; A positive number 688 client-first-message-bare = 689 [reserved-mext ","] 690 username "," nonce ["," extensions] 692 client-first-message = 693 gs2-header client-first-message-bare 695 server-first-message = 696 [reserved-mext ","] nonce "," salt "," 697 iteration-count ["," extensions] 699 client-final-message-without-proof = 700 channel-binding "," nonce ["," 701 extensions] 703 client-final-message = 704 client-final-message-without-proof "," proof 706 gss-server-error = "e=" value 707 server-final-message = gss-server-error / 708 verifier ["," extensions] 709 ;; The error message is only for the GSS-API 710 ;; form of SCRAM, and it is OPTIONAL to 711 ;; implement it. 713 extensions = attr-val *("," attr-val) 714 ;; All extensions are optional, 715 ;; i.e. unrecognized attributes 716 ;; not defined in this document 717 ;; MUST be ignored. 719 cbind-data = 1*OCTET 721 cbind-input = gs2-header [ cbind-data ] 722 ;; cbind-data MUST be present for 723 ;; gs2-cbind-flag of "p" and MUST be absent 724 ;; for "y" or "n". 726 8. SCRAM as a GSS-API Mechanism 728 This section and its sub-sections and all normative references of it 729 not referenced elsewhere in this document are INFORMATIONAL for SASL 730 implementors, but they are NORMATIVE for GSS-API implementors. 732 SCRAM is actually also GSS-API mechanism. The messages are the same, 733 but a) the GS2 header on the client's first message and channel 734 binding data is excluded when SCRAM is used as a GSS-API mechanism, 735 and b) the RFC2743 section 3.1 initial context token header is 736 prefixed to the client's first authentication message (context 737 token). 739 The GSS-API mechanism OID for SCRAM is (see Section 10). 741 8.1. GSS-API Principal Name Types for SCRAM 743 SCRAM does not name acceptors. Therefore only GSS_C_NO_NAME and 744 names of type GSS_C_NT_ANONYMOUS shall be allowed as the target name 745 input of GSS_Init_sec_context() when using a SCRAM mechanism. 747 SCRAM supports only a single name type for initiators: 748 GSS_C_NT_USER_NAME. GSS_C_NT_USER_NAME is the default name type for 749 SCRAM. 751 There is no name canonicalization procedure for SCRAM beyond applying 752 SASLprep as described in Section 5.1. 754 The query, display and exported name syntax for SCRAM principal names 755 is the same: there is no syntax -- SCRAM principal names are free- 756 form. (The exported name token does, of course, conform to [RFC2743] 757 section 3.2, but the "NAME" part of the token is just a SCRAM user 758 name.) 760 8.2. GSS-API Per-Message Tokens for SCRAM 762 The per-message tokens for SCRAM as a GSS-API mechanism SHALL be the 763 same as those for the Kerberos V GSS-API mechanism [RFC4121], using 764 the Kerberos V "aes128-cts-hmac-sha1-96" enctype [RFC3962]. 766 The 128-bit session key SHALL be derived by using the least 767 significant (right-most) 128 bits of HMAC(StoredKey, "GSS-API session 768 key" || ClientKey || AuthMessage). 770 SCRAM does support PROT_READY, and is PROT_READY on the initiator 771 side first upon receipt of the server's reply to the initial security 772 context token. 774 8.3. GSS_Pseudo_random() for SCRAM 776 The GSS_Pseudo_random() [RFC4401] for SCRAM SHALL be the same as for 777 the Kerberos V GSS-API mechanism [RFC4402]. There is no acceptor- 778 asserted sub-session key for SCRAM, thus GSS_C_PRF_KEY_FULL and 779 GSS_C_PRF_KEY_PARTIAL are equivalent for SCRAM's GSS_Pseudo_random(). 781 9. Security Considerations 783 If the authentication exchange is performed without a strong security 784 layer, then a passive eavesdropper can gain sufficient information to 785 mount an offline dictionary or brute-force attack which can be used 786 to recover the user's password. The amount of time necessary for 787 this attack depends on the cryptographic hash function selected, the 788 strength of the password and the iteration count supplied by the 789 server. An external security layer with strong encryption will 790 prevent this attack. 792 If the external security layer used to protect the SCRAM exchange 793 uses an anonymous key exchange, then the SCRAM channel binding 794 mechanism can be used to detect a man-in-the-middle attack on the 795 security layer and cause the authentication to fail as a result. 796 However, the man-in-the-middle attacker will have gained sufficient 797 information to mount an offline dictionary or brute-force attack. 798 For this reason, SCRAM includes the ability to increase the iteration 799 count over time. 801 If the authentication information is stolen from the authentication 802 database, then an offline dictionary or brute-force attack can be 803 used to recover the user's password. The use of salt mitigates this 804 attack somewhat by requiring a separate attack on each password. 805 Authentication mechanisms which protect against this attack are 806 available (e.g., the EKE class of mechanisms). RFC 2945 [RFC2945] is 807 an example of such technology. There are IPR disclosures at 808 http://datatracker.ietf.org/ipr/ that mention RFC 2945. 810 If an attacker obtains the authentication information from the 811 authentication repository and either eavesdrops on one authentication 812 exchange or impersonates a server, the attacker gains the ability to 813 impersonate that user to all servers providing SCRAM access using the 814 same hash function, password, iteration count and salt. For this 815 reason, it is important to use randomly-generated salt values. 817 SCRAM does not negotiate a hash function to use. Hash function 818 negotiation is left to the SASL mechanism negotiation. It is 819 important that clients be able to sort a locally available list of 820 mechanisms by preference so that the client may pick the most 821 preferred of a server's advertised mechanism list. This preference 822 order is not specified here as it is a local matter. The preference 823 order should include objective and subjective notions of mechanism 824 cryptographic strength (e.g., SCRAM with a successor to SHA-1 may be 825 preferred over SCRAM with SHA-1). 827 Note that to protect the SASL mechanism negotiation applications 828 normally must list the server mechs twice: once before and once after 829 authentication, the latter using security layers. Since SCRAM does 830 not provide security layers the only ways to protect the mechanism 831 negotiation are: a) use channel binding to an external channel, or b) 832 use an external channel that authenticates a user-provided server 833 name. 835 SCRAM does not protect against downgrade attacks of channel binding 836 types. The complexities of negotiation a channel binding type, and 837 handling down-grade attacks in that negotiation, was intentionally 838 left out of scope for this document. 840 A hostile server can perform a computational denial-of-service attack 841 on clients by sending a big iteration count value. 843 See [RFC4086] for more information about generating randomness. 845 10. IANA Considerations 847 IANA is requested to add the following family of SASL mechanisms to 848 the SASL Mechanism registry established by [RFC4422]: 850 To: iana@iana.org 851 Subject: Registration of a new SASL family SCRAM 853 SASL mechanism name (or prefix for the family): SCRAM-* 854 Security considerations: Section 7 of [RFCXXXX] 855 Published specification (optional, recommended): [RFCXXXX] 856 Person & email address to contact for further information: 857 IETF SASL WG 858 Intended usage: COMMON 859 Owner/Change controller: IESG 860 Note: Members of this family must be explicitly registered 861 using the "IETF Consensus" registration procedure. 862 Reviews must be requested on the SASL WG mailing list. 864 "IETF Consensus" registration procedure MUST be used for registering 865 new mechanisms in this family. The SASL mailing list 866 (or a successor designated by the responsible 867 Security AD) MUST be used for soliciting reviews on such 868 registrations. 870 Note to future SCRAM- mechanism designers: each new SCRAM- SASL 871 mechanism MUST be explicitly registered with IANA and MUST comply 872 with SCRAM- mechanism naming convention defined in Section 4 of this 873 document. 875 IANA is requested to add the following entries to the SASL Mechanism 876 registry established by [RFC4422]: 878 To: iana@iana.org 879 Subject: Registration of a new SASL mechanism SCRAM-SHA-1 881 SASL mechanism name (or prefix for the family): SCRAM-SHA-1 882 Security considerations: Section 7 of [RFCXXXX] 883 Published specification (optional, recommended): [RFCXXXX] 884 Person & email address to contact for further information: 885 IETF SASL WG 886 Intended usage: COMMON 887 Owner/Change controller: IESG 888 Note: 890 To: iana@iana.org 891 Subject: Registration of a new SASL mechanism SCRAM-SHA-1-PLUS 893 SASL mechanism name (or prefix for the family): SCRAM-SHA-1-PLUS 894 Security considerations: Section 7 of [RFCXXXX] 895 Published specification (optional, recommended): [RFCXXXX] 896 Person & email address to contact for further information: 897 IETF SASL WG 898 Intended usage: COMMON 899 Owner/Change controller: IESG 900 Note: 902 This document also requests IANA to assign a GSS-API mechanism OID 903 for SCRAM. 905 11. Acknowledgements 907 This document benefited from discussions on the SASL WG mailing list. 908 The authors would like to specially thank Dave Cridland, Simon 909 Josefsson and Jeffrey Hutzelman for their contributions to this 910 document. 912 Appendix A. Other Authentication Mechanisms 914 The DIGEST-MD5 [I-D.ietf-sasl-digest-to-historic] mechanism has 915 proved to be too complex to implement and test, and thus has poor 916 interoperability. The security layer is often not implemented, and 917 almost never used; everyone uses TLS instead. For a more complete 918 list of problems with DIGEST-MD5 which lead to the creation of SCRAM 919 see [I-D.ietf-sasl-digest-to-historic]. 921 The CRAM-MD5 SASL mechanism, while widely deployed has also some 922 problems, in particular it is missing some modern SASL features such 923 as support for internationalized usernames and passwords, support for 924 passing of authorization identity, support for channel bindings. It 925 also doesn't support server authentication. For a more complete list 926 of problems with CRAM-MD5 see [I-D.ietf-sasl-crammd5-to-historic]. 928 The PLAIN [RFC4616] SASL mechanism allows a malicious server or 929 eavesdropper to impersonate the authenticating user to any other 930 server for which the user has the same password. It also sends the 931 password in the clear over the network, unless TLS is used. Server 932 authentication is not supported. 934 Appendix B. Design Motivations 936 The following design goals shaped this document. Note that some of 937 the goals have changed since the initial version of the document. 939 o The SASL mechanism has all modern SASL features: support for 940 internationalized usernames and passwords, support for passing of 941 authorization identity, support for channel bindings. 943 o The protocol supports mutual authentication. 945 o The authentication information stored in the authentication 946 database is not sufficient by itself to impersonate the client. 948 o The server does not gain the ability to impersonate the client to 949 other servers (with an exception for server-authorized proxies), 950 unless such other servers allow SCRAM authentication and use the 951 same salt and iteration count for the user. 953 o The mechanism is extensible, but [hopefully] not overengineered in 954 this respect. 956 o Easier to implement than DIGEST-MD5 in both clients and servers. 958 Appendix C. Internet-Draft Change History 960 (RFC Editor: Please delete everything after this point) 962 Changes since -10 964 o Converted the source for this I-D to XML. 966 o Added text to make SCRAM compliant with the new GS2 design. 968 o Added text on channel binding negotiation. 970 o Added text on channel binding, including a reference to RFC5056. 972 o Added text on SCRAM as a GSS-API mechanism. This noted as not 973 relevant to SASL-only implementors -- the normative references for 974 SCRAM as a GSS-API mechanism are segregated as well. 976 Changes since -07 978 o Updated References. 980 o Clarified purpose of the m= attribute. 982 o Fixed a problem with authentication/authorization identity's ABNF 983 not allowing for some characters. 985 o Updated ABNF for nonce to show client-generated and server- 986 generated parts. 988 o Only register SCRAM-SHA-1 with IANA and require explicit 989 registrations of all other SCRAM- mechanisms. 991 Changes since -06 993 o Removed hash negotiation from SCRAM and turned it into a family of 994 SASL mechanisms. 996 o Start using "Hash Function Textual Names" IANA registry for SCRAM 997 mechanism naming. 999 o Fixed definition of Hi(str, salt) to be consistent with [RFC2898]. 1001 o Clarified extensibility of SCRAM: added m= attribute (for future 1002 mandatory extensions) and specified that all unrecognized 1003 attributes must be ignored. 1005 Changes since -05 1006 o Changed the mandatory to implement hash algorithm to SHA-1 (as per 1007 WG consensus). 1009 o Added text about use of SASLPrep for username canonicalization/ 1010 validation. 1012 o Clarified that authorization identity is canonicalized/verified 1013 according to SASL protocol profile. 1015 o Clarified that iteration count is per-user. 1017 o Clarified how clients select the authentication function. 1019 o Added IANA registration for the new mechanism. 1021 o Added missing normative references (UTF-8, SASLPrep). 1023 o Various editorial changes based on comments from Hallvard B 1024 Furuseth, Nico William and Simon Josefsson. 1026 Changes since -04 1028 o Update Base64 and Security Glossary references. 1030 o Add Formal Syntax section. 1032 o Don't bother with "v=". 1034 o Make MD5 mandatory to implement. Suggest i=128. 1036 Changes since -03 1038 o Seven years have passed, in which it became clear that DIGEST-MD5 1039 suffered from unacceptably bad interoperability, so SCRAM-MD5 is 1040 now back from the dead. 1042 o Be hash agnostic, so MD5 can be replaced more easily. 1044 o General simplification. 1046 12. References 1048 12.1. Normative References 1050 [RFC2104] Krawczyk, H., Bellare, M., and R. Canetti, "HMAC: Keyed- 1051 Hashing for Message Authentication", RFC 2104, 1052 February 1997. 1054 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1055 Requirement Levels", BCP 14, RFC 2119, March 1997. 1057 [RFC3174] Eastlake, D. and P. Jones, "US Secure Hash Algorithm 1 1058 (SHA1)", RFC 3174, September 2001. 1060 [RFC3454] Hoffman, P. and M. Blanchet, "Preparation of 1061 Internationalized Strings ("stringprep")", RFC 3454, 1062 December 2002. 1064 [RFC3629] Yergeau, F., "UTF-8, a transformation format of ISO 1065 10646", STD 63, RFC 3629, November 2003. 1067 [RFC4013] Zeilenga, K., "SASLprep: Stringprep Profile for User Names 1068 and Passwords", RFC 4013, February 2005. 1070 [RFC4422] Melnikov, A. and K. Zeilenga, "Simple Authentication and 1071 Security Layer (SASL)", RFC 4422, June 2006. 1073 [RFC4648] Josefsson, S., "The Base16, Base32, and Base64 Data 1074 Encodings", RFC 4648, October 2006. 1076 [RFC5056] Williams, N., "On the Use of Channel Bindings to Secure 1077 Channels", RFC 5056, November 2007. 1079 [RFC5234] Crocker, D. and P. Overell, "Augmented BNF for Syntax 1080 Specifications: ABNF", STD 68, RFC 5234, January 2008. 1082 12.2. Normative References for GSS-API implementors 1084 [I-D.ietf-sasl-gs2] 1085 Josefsson, S. and N. Williams, "Using GSS-API Mechanisms 1086 in SASL: The GS2 Mechanism Family", draft-ietf-sasl-gs2-12 1087 (work in progress), April 2009. 1089 [RFC2743] Linn, J., "Generic Security Service Application Program 1090 Interface Version 2, Update 1", RFC 2743, January 2000. 1092 [RFC3962] Raeburn, K., "Advanced Encryption Standard (AES) 1093 Encryption for Kerberos 5", RFC 3962, February 2005. 1095 [RFC4086] Eastlake, D., Schiller, J., and S. Crocker, "Randomness 1096 Requirements for Security", BCP 106, RFC 4086, June 2005. 1098 [RFC4121] Zhu, L., Jaganathan, K., and S. Hartman, "The Kerberos 1099 Version 5 Generic Security Service Application Program 1100 Interface (GSS-API) Mechanism: Version 2", RFC 4121, 1101 July 2005. 1103 [RFC4401] Williams, N., "A Pseudo-Random Function (PRF) API 1104 Extension for the Generic Security Service Application 1105 Program Interface (GSS-API)", RFC 4401, February 2006. 1107 [RFC4402] Williams, N., "A Pseudo-Random Function (PRF) for the 1108 Kerberos V Generic Security Service Application Program 1109 Interface (GSS-API) Mechanism", RFC 4402, February 2006. 1111 [tls-unique] 1112 Zhu, L., "Registration of TLS unique channel binding 1113 (generic)", IANA http://www.iana.org/assignments/ 1114 channel-binding-types/tls-unique, July 2008. 1116 12.3. Informative References 1118 [I-D.altman-tls-channel-bindings] 1119 Altman, J., Williams, N., and L. Zhu, "Channel Bindings 1120 for TLS", draft-altman-tls-channel-bindings-05 (work in 1121 progress), June 2009. 1123 [I-D.ietf-sasl-crammd5-to-historic] 1124 Zeilenga, K., "CRAM-MD5 to Historic", 1125 draft-ietf-sasl-crammd5-to-historic-00 (work in progress), 1126 November 2008. 1128 [I-D.ietf-sasl-digest-to-historic] 1129 Melnikov, A., "Moving DIGEST-MD5 to Historic", 1130 draft-ietf-sasl-digest-to-historic-00 (work in progress), 1131 July 2008. 1133 [I-D.melnikov-sasl-scram-ldap] 1134 Melnikov, A., "LDAP schema for storing SCRAM secrets", 1135 draft-melnikov-sasl-scram-ldap-02 (work in progress), 1136 July 2009. 1138 [RFC2898] Kaliski, B., "PKCS #5: Password-Based Cryptography 1139 Specification Version 2.0", RFC 2898, September 2000. 1141 [RFC2945] Wu, T., "The SRP Authentication and Key Exchange System", 1142 RFC 2945, September 2000. 1144 [RFC4510] Zeilenga, K., "Lightweight Directory Access Protocol 1145 (LDAP): Technical Specification Road Map", RFC 4510, 1146 June 2006. 1148 [RFC4616] Zeilenga, K., "The PLAIN Simple Authentication and 1149 Security Layer (SASL) Mechanism", RFC 4616, August 2006. 1151 [RFC4949] Shirey, R., "Internet Security Glossary, Version 2", 1152 RFC 4949, August 2007. 1154 [RFC5246] Dierks, T. and E. Rescorla, "The Transport Layer Security 1155 (TLS) Protocol Version 1.2", RFC 5246, August 2008. 1157 [tls-server-end-point] 1158 Zhu, L., "Registration of TLS server end-point channel 1159 bindings", IANA http://www.iana.org/assignments/ 1160 channel-binding-types/tls-server-end-point, July 2008. 1162 Authors' Addresses 1164 Abhijit Menon-Sen 1165 Oryx Mail Systems GmbH 1167 Email: ams@oryx.com 1169 Alexey Melnikov 1170 Isode Ltd 1172 Email: Alexey.Melnikov@isode.com 1174 Chris Newman 1175 Sun Microsystems 1176 1050 Lakes Drive 1177 West Covina, CA 91790 1178 USA 1180 Email: chris.newman@sun.com 1182 Nicolas Williams 1183 Sun Microsystems 1184 5300 Riata Trace Ct 1185 Austin, TX 78727 1186 USA 1188 Email: Nicolas.Williams@sun.com