KEY RECOVERY STANDARD (ADVISORY COMMITTEE WORKING DRAFT) Foreword The Federal Information Processing Standards Publication Series of the National Institute of Standards and Technology (NIST) is the official publication relating to standards and guidelines adopted and promulgated under the provisions of Section 5131 of the Information Technology Management Reform Act of 1996, and the Computer Security Act of 1987, Public Law 104-106. Under these mandates, the Secretary of Commerce promulgates standards and guidance pertaining to the efficiency, security and privacy of Federal computer systems. The National Institute of Standards and Technology, through its Information Technology Laboratory, has the mission of developing standards, guidelines and associated methods and techniques for computer systems, and providing technical assistance to industry and government in the implementation of standards. Comments concerning Federal Information Processing Standards Publications are welcomed and should be addressed to the Director, Information Technology Laboratory, National Institute of Standards and Technology, Gaithersburg, MD 20899. Shukri Wakid, Director Information Technology Laboratory Abstract This standard specifies requirements to be met by government Key Recovery Systems. Such systems provide for the decryption of stored or communicated data when access to the data is properly authorized. ALTERNATIVE TO THE ABOVE: This standard specifies requirements to be met by key recovery components used by Federal government agencies. These components provide for the recovery of keys which will be used for the decryption of stored or communicated data when access to the data is properly authorized. Key words: ADP security, computer security, Key Recovery, Federal Information Processing Standard. Federal Information Processing Standards Publication XXX (Date) Announcing the KEY RECOVERY STANDARD Federal Information Processing Standards Publications (FIPS PUBS) are issued by the National Institute of Standards and Technology (NIST) after approval by the Secretary of Commerce pursuant to Section 5131 of the Information Technology Management Reform Act of 1996, and the Computer Security Act of 1987, Public Law 104-106. 1. Name of Standard. Key Recovery Standard. 2. Category of Standard. Computer Security, Cryptography. 3. Explanation. This Standard specifies requirements for key recovery components. These components provide for the recovery of keys to be used for the decryption of stored or communicated ciphertext when the decryption keys are not otherwise available. Key recovery is motivated by three primary scenarios: 1. recovery of stored data on behalf of an organization (or individual) e.g., in response to accidental loss of keys; 2. recovery of stored or communicated data on behalf of an organization (e.g., for the purposes of monitoring or auditing activities); and 3. recovery of communicated or stored data by lawfully authorized authorities. The first scenario supports the ability to regain access to data that would otherwise be lost. The second scenario encompasses internal investigation authorized by an organization. The final scenario encompasses data acquired under the authorization of court orders for wiretaps, search and seizure orders, civil suit subpoenas, etc A Key Recovery System (KRS) manages cryptographic keys in support of data recovery when normal key access mechanisms fail. These systems must be carefully designed so that plaintext may be recovered in a timely manner, and so that only authorized recoveries are permitted. Therefore, security is a critical factor in any KRS design. The purpose of this standard is to specify requirements for key recovery components, and to enable the validation of components claiming conformance. The standard encompasses the security (from an implementation, managerial and operation perspective) and the availability of key recovery components, as well as defining interoperability requirements. 4. Approving Authority. Secretary of Commerce. 5. Maintenance Agency. U.S. Department of Commerce, National Institute of Standards and Technology (NIST), Information Technology Laboratory. 6. Cross Index. a. FIPS PUB 46-2, Data Encryption Standard. b. FIPS PUB 81, DES Modes of Operation. c. FIPS PUB 140-1, Security Requirements for Cryptographic Modules, January 1994. d. DOD 5200.28-STD, Department of Defense Trusted Computer System Evaluation Criteria (TCSEC) ("The Orange Book"), National Computer Security Center, December 1985. e. SC 27 N1953 Evaluation Criteria for IT Security, Part 3 - Security Assurance Requirements Other NIST publications may be applicable to the implementation and use of this standard. A list (NIST Publications List 91) of currently available computer security publications, including ordering information, can be obtained from NIST. 7. Applicability. To be supplied by the Federal Government. 8. Applications. This standard is appropriate for use in a variety of applications, including: 1. When computer files are encrypted for secure storage or transmission; 2. When electronic mail is encrypted before transmission among communicating entities; and 3. When electronic voice communications are encrypted for privacy. 9. Specifications. Federal Information Processing Standard (FIPS xyz) Key Recovery Standard (affixed). 10. Implementations. System components, or parts thereof, conforming to this standard may be implemented in software, firmware, hardware, or any combination thereof. All cryptographic modules employed in components of such systems shall comply with FIPS 140-1. FIPS approved encryption algorithms (e.g., DES) shall be used in Federal applications of systems conforming to this standard. The use of new encryption algorithms which are FIPS approved after the date of the standard is also permitted. Information about the validation of implementations conforming to this standard may be found in Section X of the attached Specification. Additional information may be obtained from the National Institute of Standards and Technology, Information Technology Laboratory, Attn: Key Recovery Validation, Gaithersburg, MD 20899. 11. Export Control. Implementations of this standard are subject to export controls as specified in Title 15, Code of Federal Regulations, Parts 730-774 and Title 22, Code of Federal Regulations, Parts 120-130. Exporters are advised to contact the Encryption Policy Controls Division at the Department of Commerce, Bureau of Export Administration for more information. 12. Patents. Implementations of this standard may be covered by U.S. and foreign patents. 13. Implementation Schedule. The effective date of this standard is . From approval of this FIPS by the Secretary of Commerce to its effective date, agencies may purchase components that have been affirmed in writing from the manufacturer as complying with this standard. From the effective date until six months after the establishment of the validation program by NIST, agencies that have determined a need for key recovery components shall purchase components that have been affirmed in writing by the manufacturer as complying with this standard. A copy of the written affirmation shall have been sent to the Director, Information Technology Laboratory, National Institute of Standards and Technology, Gaithersburg, MD 20899. For a one year period following the six month period after the establishment of the validation program, agencies shall purchase validated key recovery components, or components that have been submitted for validation. After this period, only validated key recovery components will be considered as meeting the provisions of this standard. 14. Glossary. The following terms are used as defined below for purposes of this standard: Abstract Machine The underlying hardware or firmware abstraction to which the software is written. Accountability Assurance (1) Confidence that an entity meets its security objectives. (2) The degree of confidence that a product correctly implements the security policy. Auditable Events Events within a key recovery product which may appear in an audit log (see Section 4). Authentication Data Information used to authenticate an entity, e.g., a password, PIN, biometric, or response to a challenge. Authentication Information See "Authentication Data". Authentication Mechanism A technique used to authenticate an entity, e.g., user ID and password, token, biometric or challenge-response. Authentic Public Key Source Used to provide a certificate infrastructure to support the use of public key cryptography within the Key Recovery System. Authorized key recovery Key recovery either with the permission of the owner of the data or as otherwise permitted by law. Authorized Request A request based on a legal and lawful right for access by a data owner or other authorized entity. Authorized User A user who is authorized to access a system to perform one or more actions. Common Criteria (CC) An international standard for security in information security components. Common Criteria Evaluation Assurance Level (EAL) A predefined set of assurance components that represents a point on the CC assurance scale. Common Criteria Protection Profile An implementation-independent set of security requirements for a category of components which meet specific consumer needs. Confidentiality (1) Assurance that the information is not disclosed to unauthorized entities or processes. (2) The property that sensitive information is not disclosed to unauthorized individuals, entities or processes. (3) The property that information is not made available or disclosed to an unauthorized user, process or object. Configurable Capability A capability feature that is available but need not be selected for use. Configuration Item Items (e.g., documents, software, hardware) which are under configuration control. Configuration Management (CM) The management of security features and assurances through the control of changes made to a system's hardware, software, firmware, documentation set, test, test fixtures and test documentation throughout the development and operational life of the system. Cryptographic End System A system containing encryption and decryption mechanisms. Data Voice, facsimile, computer files, electronic mail, and other stored or communicated information. Data Key A symmetric key used to encrypt data. Data Recovery System The system/subsystem used to recover encrypted data using a recovered target key obtained by the Key Recovery Requestor System. Decryption (1) Transformation of ciphertext form of data to plaintext form. (2) The process of changing ciphertext into plaintext. Encryption (1) Transformation of plaintext form of data to ciphertext form. (2) A process of transforming plaintext into ciphertext for the purpose of security or privacy. (3) Transforming text into code in order to conceal its meaning. The process of transforming data to an unintelligible form in such a way that the original data either cannot be obtained (one-way encryption), or cannot be obtained without using the inverse decryption process. (3) Conversion of plaintext to ciphertext through the use of a cryptographic algorithm. Evidence of Origin (1) A proof of the origin of information that cannot be refuted by the originator, e.g., by using a digital signature. (2) Non-repudiation. Evidence of Receipt Should be "Proof of Receipt"? FIPS 140-1 Level 1 Security Requirements Specify basic security requirements for a cryptomodule. No physical security mechanisms are required in the module beyond the requirement for production- grade equipment. Software cryptographic functions may be performed in a general purpose personal computer. FIPS 140-1 Level 2 Security Requirements Improve upon the physical security of a Level 1 cryptomodule by (1) requiring tamper evident coatings or seals, or for pick-resistent locks, (2) requiring role-based authentication and (3) allowing software cryptography in multi-user timeshared systems when used in conjunction with a C21 or equivalent operating system. FIPS 140-1 Level 3 Security Requirements Improve upon the Level 1 and 2 requirements for cryptomodules by (1) requiring tamper detection mechanisms, (2) requiring identity-based authentication, (3) specifying stronger requirements for entering and outputting critical security parameters, and (4) allowing software cryptography in multi-user timeshared systems when a B1 or equivalent trusted operating system is employed along with a trusted path for the entry and output of critical security parameters. FIPS Compliant Meeting all requirements of a specified level of this standard. Flaw Remediation The correction of discovered security flaws in a product or system. Functional Requirements A high level description of the requirements for a system. Functional Specification High level description of the user-visible interface and behavior of a system. Implementation Representation A description of the implementation (e.g., source code when the implementation is software or firmware; or drawings and schematics, if the system is hardware). Independent Testing Testing performed by persons other than the developers. Informal Security Policy Model An accurate and concise statement of system security policy expressed informally (i.e., in natural language; e.g., English). Informal (1) Expressed in natural language. (2) Written as prose in natural language. Informal style/presentation Written in normal language, e.g., English. Integrity The property that sensitive data has not been modified or deleted in an unauthorized and undetected manner. Interactive Communication Two-way communication between end users. Interoperability The ability of components or systems to communicate with one another. Key Escrow (1) The processes of managing (e.g., generating, storing, transferring, auditing) the cryptographic keys or key components by one or more entities. (2) A key recovery technique that employs one or more Key Recovery Agents who hold (i.e., cache) keys or key components for their subscribers. (3) A method of key recovery where the secret or private keys, key parts or key related information to be recovered is stored by one or more Key Recovery Agents. Other Key Recovery Information may be available elsewhere. Key Recovery Access to information sufficient to recover encrypted data. Key Recovery Agent (KRA) A Key Recovery Agent System. Key Recovery Agent (KRA) System A key recovery system that performs a recovery service in response to an authorized request by a requestor system on behalf of a requestor. Key Recovery Capable System A cryptographic end system with either a KRI Generation Function or a Key Recovery Validation Function or both. Key Recovery Component An element within a Key Recovery System that provides key recovery functionality (e.g., KRI generation, KRI management, and/or key recovery function). Key Recovery Field A field, output by the key recovery mechanism of a product, that identifies key recovery agents and enables key recovery agents to identify the key(s) required to decrypt corresponding ciphertext output by the product. Key Recovery Function Recovers a target key using Key Recovery Information. Composed of the KRI Requestor Function and KRA Function. Key Recovery Information (KRI) Information that is used in the recovery of a key. The KRI does not include a plaintext key. Key Recovery Information Field (KRIF) Key recovery information which is specific to a single key recovery scheme. Key Recovery Block (KRB) A stream of bytes that serves as a container for a single key recovery scheme- specific KRIF and associates the KRIF with a set of standard fields in a predefined format. Key Recovery Policy A policy which specifies the conditions under which key recovery information must be created and conditions under which and to whom the key recovery information may be released; may also indicate the allowable Key Recovery Agent(s) and how or where key recovery information must be maintained. Key Recovery Requestor System The system/subsystem used by the requestor to request keys. Key Recovery Service An authorized key recovery as performed by a Key Recovery Agent. Key Recovery System (KRS) Consists of the KRI Generation Function, the KRI Management Function and the Key Recovery Function. Includes software, hardware, procedures and infrastructure. KRA Key Recovery Agent KRB Key Recovery Block KRI Acquisition Function Enables a receiver to determine the key recovery information (KRI) which is available and acquire KRI for subsequent processing. KRI Delivery Function Assembles and formats the key recovery information (KRI) and makes the KRI available. KRI Encapsulation A method of key recovery in which keys, key parts or key related information is maintained outside a Key Recovery Agent. KRI Generation Function Generates the key recovery information (KRI) needed to recover the target key and provides the KRI to the KRI Delivery Function. KRI Management Function Manages the KRI after generation for potential recovery. Composed of the KRI Delivery Function, KRI Receive Function and KRI Validation Function. KRI Providers Those entities provide Key Recovery Information (KRI) using a KRI Generation Function. KRI Validation Function Checks, authenticates, validates or verifies the available key recovery information. KRR Key Recovery Requestor System. KRS Key Recovery System Layered Product A product in which security functions are layered. For example, a secure application which is implemented on top of a secure operating system is a layered product. Least Abstract Representation (1) The most concrete representation of an implementation (e.g., source code). (2) The representation that is closest to the implementation, e.g., source code. Licensing Agent Authorizes Key Recovery Agents after an evaluation against the FIPS. Masquerading An attempt to gain access to a system by posing as an authorized user. Message Security Protocol (MSP) A data format that cryptographically binds data sensitivity and provides public key cryptography based security services for the data, including confidentiality, integrity, etc. MIME Multipurpose Internet Mail Extension as defined in RFC 2045. Monolithic Product A product in which security functions are not layered. See "Layered Product". Non-Key Recovery Product An encryption product whose encryption output is not recoverable through key recovery. Presentation of Evidence Providing the information necessary to carrying out the assurance activity. Private Key (1) In an asymmetric (public) key cryptosystem, that key of an entity's key pair which is known only by that entity. (2) A cryptographic key used with a public key cryptographic algorithm, uniquely associated with an entity, and not made public. Proof of Receipt A proof of the receipt of information so that the recipient cannot deny having received the information, e.g., using a digital signature by the recipient on the received message. Public Key (1) In an asymmetric key system, that key of an entity's key pair which is publicly known. (2) A cryptographic key used with a public key cryptographic algorithm, uniquely associated with an entity, and which may be made public. Registration Agent Archives vendor-specific information in order to find, acquire and parse recovery information. Representation Correspondence An accurate and complete mapping from a higher level representation to a lower level representation (e.g., from functional requirements to a functional specification, from a functional specification to a high level design, from a high level design to a low level design, from a low level design to source code, etc.). Requestor An entity that is authorized to request a key recovery. Requestor Function Interacts with one or more Key Recovery Agents using Key Recovery Information to recover a data encrypting key. Secret Key A cryptographic key used with a secret key [symmetric] cryptographic algorithm, uniquely associated with one or more entities, and which shall not be made public. Security Domain (1) A set of objects , a security policy , a security authority and a set of relevant activities inwhich the set of elements are subject to the security policy , administered by the security authority , for the specified activities. (2) A set of security-related services, mechanisms, and policies. Security Policy (1) A precise specification of the security rules under which a cryptographic module may operate, including the security rules derived from the requirements of this standard and the additional security rules imposed by the manufacturer. (2) A set of rules and procedures regulating the use of information including its processing, storage, distribution and presentation. Security Policy Model A formal representation of the security policy enforced by the product. Security Target A set of security requirements ad specifications to be used as the basis for evaluation of an identified product. Session-based Protocols Interactive communications. Session Key A key that is used to encrypt and/or decrypt data for a single communications session. Session Key Recovery Recovery of the Data Encryption Key. S/MIME Secure MIME as defined by RFC XXX. Source Authentication The ability to authenticate the identity of the source of a information. Standard Communication Protocol Any communication protocol adopted by a generally recognized standards organization. Store-and-Forward Communications One way communications (i.e., from a sender to a receiver) without the involvement of the receiver. The receiver may acquire the communication at a time which is significantly later than the time the communication is sent. System Includes software, hardware, procedures. Target key The key recovered by a Key Recovery System. Testing laboratory A laboratory which has been accredited by NIST to test systems, subsystems, key recovery agents, or components for conformance to this standard. Transaction-based Protocols Store-and-forward communications. Trusted Path A mechanism by which a person or process can communicate directly with a cryptographic module and which can only be activated by the person, process or module, and cannot be imitated by untrusted software within the module. Trusted Third Party An entity which is trusted by the parties performing the encryption or decryption processes, but are not identical with those parties. Unwrap Decryption of an encrypted key by another key. Vulnerability Analysis The determination of the vulnerabilities of a product or system. Wrap Encryption of a cryptographic key by another key. 15. Qualifications. The security requirements specified in this standard are based upon information provided by many sources within the Federal government and private industry. The requirements are designed to protect against adversaries mounting cost-effective attacks on unclassified government or commercial data. The primary goal in defining effective security for a system is to make the cost of any attack greater than the possible payoff. While the security requirements specified in this standard are intended to maintain the security of a key recovery component, conformance to this standard does not guarantee that a particular component is secure. It is the responsibility of the manufacturer of a key recovery component to build the component in a secure manner. Similarly, the use of a key recovery component that conforms to this standard in an overall system does not guarantee the security of the overall system. The responsible authority in each agency shall assure that an overall system provides an acceptable level of security. Since a standard of this nature must be flexible enough to adapt to advancements and innovations in key recovery technology, this standard will be initially reviewed in two years in order to consider new or revised requirements that may be needed to meet technological changes. 16. Waiver Procedure. Under certain exceptional circumstances, the heads of Federal departments and agencies may approve waivers to Federal Information Processing Standards (FIPS). The head of such agency may redelegate such authority only to a senior official designated pursuant to section 3506(b) of Title 44, United States Code. Waivers shall be granted only when: a. Compliance with a standard would adversely affect the accomplishment of the mission of an operator of a Federal computer system; or b. Cause a major adverse financial impact on the operator which is not offset by Government wide savings. Agency heads may act upon a written waiver request containing the information detailed above. Agency heads may also act without a written waiver request when they determine that conditions for meeting the standard cannot be met. Agency heads may approve waivers only by a written decision which explains the basis on which the agency head made the required finding(s). A copy of each such decision, with procurement sensitive or classified portions clearly identified, shall be sent to: National Institute of Standards and Technology; ATTN: FIPS Waiver Decisions, Building 225 Building, Room A-231, Gaithersburg, MD 20899. In addition, a notice of each waiver granted and each delegation of authority to approve waivers shall be sent promptly to the Committee on Government Operations of the House of Representatives and the Committee on Governmental Affairs of the Senate and shall be published promptly in the Federal Register. When the determination on a waiver applies to the procurement of equipment and/or services, a notice of the waiver determination must be published in the Commerce Business Daily as a part of the notice of solicitation for offers of an acquisition or, if the waiver determination is made after that notice is published, by amendment to such notice. A copy of the waiver, any supporting documents, the document approving the waiver and any supporting and accompanying documents, with such deletions as the agency is authorized and decides to make under 5 U.S.C. Sec. 552(b), shall be part of the procurement documentation and retained by the agency. 17. Where to Obtain Copies of the Standard. Copies of this publication are for sale by the National Technical Information Service (NTIS), U.S. Department of Commerce, Springfield, VA 22161. Publication and ordering information may be found at http://www.ntis.gov. When ordering, refer to Federal Information Processing Standards Publication xyz (FIPS PUB XXX), and identify the title. When microfiche is desired, this should be specified. Prices are published by NTIS in current catalogs and other issuances. Payment may be made by check, money order, credit card or deposit account. TABLE OF CONTENTS 1 OVERVIEW 1 1.1 Scope of the Standard 1 1.2 Road Map for the Standard 3 2 KEY RECOVERY MODEL 5 2.1 Key Recovery Information (KRI) Generation Function 7 2.2 KRI Delivery Function 9 2.3 KRI Acquisition Function 10 2.4 KRI Validation Function 10 2.5 Requestor Function 11 2.6 Key Recovery Agent(s) 12 2.7 Interoperability 13 2.8 Mapping Components to the Key Recovery Model 15 2.9 Key Recovery Techniques 16 2.9.1 KRI Encapsulation 16 2.9.2 Key Escrow 17 2.9.3 Interactions Between Systems Using Different Key Recovery Techniques 18 2.9.3.1 Interactions Between KRI Encapsulation and Key Escrow Techniques 18 2.9.3.2 Interactions Between KRI Encapsulation and Systems with No Key Recovery 19 2.9.3.3 Interaction Between Key Escrow and Systems with No Key Recovery 19 3 SECURITY REQUIREMENTS 21 3.1 Key Recovery Agent Requirements 21 3.1.1 Level 1 - Medium Assurance 21 3.1.1.1 Cryptographic Functions 21 3.1.1.2 Cryptographic Algorithms 21 3.1.1.3 Confidentiality 21 3.1.1.4 Integrity 22 3.1.1.5 Audit 22 3.1.1.6 Identification and Authentication 24 3.1.1.7 Access Control 26 3.1.1.8 Authentication of Received Transactions 27 3.1.1.9 Non-Repudiation 27 3.1.1.10 Protection of Trusted Security Functions 28 3.1.2 Level 2 - High Assurance 28 3.1.2.1 Cryptographic Functions 28 3.1.2.2 Cryptographic Algorithms 28 3.1.2.3 Confidentiality 29 3.1.2.4 Integrity 29 3.1.2.5 Audit 29 3.1.2.6 Identification and Authentication 30 3.1.2.7 Access Control 30 3.1.2.8 Authentication of Received Transactions 32 3.1.2.9 Non Repudiation 32 3.1.2.10 Protection of Trusted Security Functions 32 3.2 Key Recovery Information Generator 33 3.2.1 Level 1 - Medium Assurance Key Recovery Information Generator 33 3.2.1.1 Cryptographic Functions 33 3.2.1.2 Cryptographic Algorithms 33 3.2.1.3 Confidentiality 33 3.2.1.4 Integrity 33 3.2.1.5 Identification and Authentication 34 3.2.1.6 Access Control 34 3.2.2 Level 2 - High Assurance Key Recovery Information Generator 34 3.2.2.1 Cryptographic Functions 34 3.2.2.2 Cryptographic Algorithms 35 3.2.2.3 Confidentiality 35 3.2.2.4 Integrity 35 3.2.2.5 Identification and Authentication 35 3.2.2.6 Access Control 35 3.3 Key Recovery Information Delivery 35 3.4 Key Recovery Information Acquisition 35 3.5 Key Recovery Information Validator 35 3.5.1 Level 1 - Medium Assurance Key Recovery Information Validator 36 3.5.1.1 Cryptographic Functions 36 3.5.1.2 Cryptographic Algorithms 36 3.5.1.3 Integrity 36 3.5.2 Level 2 - High Assurance Key Recovery Information Validator 36 3.5.2.1 Cryptographic Functions 37 3.5.2.2 Cryptographic Algorithms 37 3.5.2.3 Integrity 37 3.6 Key Recovery Requestor 37 3.6.1 Level 1 - Medium Assurance 38 3.6.1.1 Cryptographic Functions 38 3.6.1.2 Cryptographic Algorithms 38 3.6.1.3 Confidentiality 38 3.6.1.4 Integrity 39 3.6.1.5 Audit 39 3.6.1.6 Identification and Authentication 41 3.6.1.7 Access Control 42 3.6.1.8 Authentication of Received Transactions 42 3.6.1.9 Non-Repudiation 43 3.6.1.10 Protection of Trusted Security Functions 43 3.6.2 Level 2 - High Assurance 43 3.6.2.1 Cryptographic Functions 44 3.6.2.2 Cryptographic Algorithms 44 3.6.2.3 Confidentiality 44 3.6.2.4 Integrity 44 3.6.2.5 Audit 44 3.6.2.6 Identification and Authentication 45 3.6.2.7 Access Control 45 3.6.2.8 Authentication of Received Transactions 45 3.6.2.9 Non Repudiation 45 3.6.2.10 Protection of Trusted Security Functions 46 KRA Availability 47 The KRA facility should be required to have the capability to securely replicate any key recovery information stored in order to support continued on-line access in case of a facility failure. 47 4 ASSURANCE REQUIREMENTS 49 4.1 Configuration Management 52 4.1.1 Configuration Management ACM_CAP - CM Capabilities 52 4.1.1.1 ACM_CAP.1 Minimal Support 52 4.1.2 Configuration Management ACM_SCP - CM Scope 53 4.1.2.1 ACM_SCP.2 Problem Tracking CM Coverage 54 4.2 Delivery and Operation 54 4.2.1 Delivery and Operation ADO_DEL - Delivery 54 4.2.1.1 ADO_DEL.1 Delivery Procedures 55 4.2.1.2 ADO_DEL.2 Detection of Modification 55 4.2.2 Delivery and Operation ADO_IGS - Installation, Generation, and Start-up 56 4.2.2.1 ADO_IGS.1 Installation, Generation, and Start-up Procedures 56 4.3 Development 57 4.3.1 Development ADV_FSP - Functional Specification 57 4.3.1.1 ADV_FSP.1 Functional Specification and Security Policy 58 4.3.1.2 ADV_FSP.2 Functional Specification, Security Policy and Informal Security Policy Model 59 4.3.2 Development ADV_HLD - High-Level Design 60 4.3.2.1 ADV_HLD.1 Descriptive High-Level Design 61 4.3.2.2 ADV_HLD.2 Security Enforcing High-Level Design 62 4.3.3 Development ADV_IMP - Implementation Representation 63 4.3.3.1 ADV_IMP.1 Subset of the Implementation 64 4.3.4 Development ADV_LLD - Low-Level Design 65 4.3.4.1 ADV_LLD.1 Descriptive Low-Level Design 65 4.3.5 Development ADV_RCR - Representation Correspondence 67 4.3.5.1 ADV_RCR.1 Informal Correspondence Demonstration 67 4.4 Guidance Documents 68 4.4.1 Guidance Documents AGD_ADM Administrator Guidance 68 4.4.1.1 AGD_ADM.1 Administrator Guidance 68 4.4.2 Guidance Documents AGD_USR - User Guidance 70 4.4.2.1 AGD_USR.1 User Guidance 70 4.5 Life Cycle Support 71 4.5.1 Life Cycle Support ALC_FLR - Flaw Remediation 71 4.5.1.1 ALC_FLR.1 Basic Flaw Remediation 71 4.5.1.2 ALC_FLR.2 Flaw Reporting Procedures 72 4.6 Tests 73 4.6.1 Tests ATE_COV - Coverage 73 4.6.1.1 ATE_COV.1 Complete Coverage - Informal 74 4.6.2 Tests ATE_DPT - Depth 74 4.6.2.1 ATE_DPT.1 Testing - Functional Specification 75 4.6.3 Tests ATE_FUN - Functional Tests 75 4.6.3.1 ATE_FUN.1 Functional Testing 76 4.6.4 Tests ATE_IND - Independent Testing 77 4.6.4.1 ATE_IND.2 Independent Testing - Sample 77 4.6.4.2 ATE_IND.3 Independent Testing - Complete 78 4.7 Vulnerability Assessment 79 4.7.1 Vulnerability Assessment AVA_VLA - Vulnerability Analysis 79 4.7.1.1 AVA_VLA.1 Developer Vulnerability Analysis 80 4.8 Excluded Assurance Requirements 81 APPENDIX A: EXAMPLES 82 APPENDIX B: KEY RECOVERY BLOCK 88 APPENDIX C: CERTIFICATE EXTENSIONS 94 APPENDIX D: INTEROPERABILITY EXAMPLES 98 1 Overview Federal Agencies have a right and a responsibility to protect the information and data contained in, processed by, and transmitted between their IT systems. Ownership of the information is often shared with individuals, companies, and organizations and therefore requires that the government protect that information on its own behalf and on behalf of those co-owners. That protection must meet or exceed Federal Government standards and the standards of those co- owners. Encryption is an important tool for protecting the confidentiality of communicated or stored data. When suitably strong encryption algorithms are employed and implemented with appropriate assurance, encryption can prevent the disclosure of communicated or stored data to unauthorized parties. However, the unavailability, loss, or corruption of the keys needed to decrypt encrypted data will prevent disclosure to authorized parties. To ensure authorized access to encrypted data in the face of such failures, it is necessary to provide methods for recovering decryption keys by authorized parties. This Standard establishes requirements for implementations of Key Recovery System (KRS) technology. 1.1 Scope of the Standard This Standard neither requires not endorses any specific technology for use in a KRS. It endeavors to be technology independent, so as to not to unduly impede innovation in this new area. However, it is not the case that every conceivable key recovery technology will be amenable to successful evaluation under this Standard, e.g., intrinsically insecure KRS technologies may not be able to be evaluated. This Standard establishes technical, security and interoperability requirements for the components of a KRS. This Standard presents a general model for a KRS. The model identifies functions that are intrinsic to any KRS: the generation of Key Recovery Information (KRI), the management of KRI, requests for key recovery, and the satisfaction of such requests by one or more Key Recovery Agents (KRAs). The Standard establishes functional, interoperability, security, and security assurance requirements that apply to an implementation of each KRS function. A product claiming compliance with the Standard must be mappable to one or more of the KRS functions defined in this Standard. There is no requirement that a product offered for evaluation embody all of the defined functions; a compliant product may not constitute a complete KRS. There is no requirement that a single product or a suite of products from a single vendor embody all of the functions needed to provide a complete KRS. Thus, the Standard permits the modular implementation of a KRS, based on the assembly of components from one or more sources. Since a Federal department or agency will require a complete KRS, additional guidance will be provided via other documents to assist in evaluating the security of a system assembled from components (from one or more vendors) that have been evaluated against this standard. (Option 1) A product may be submitted for the evaluation of a subset of the KRS functions it provides. If each KRS function in the subset selected is certified as compliant, the product shall be certified compliant with the Standard. This allows a product to offer both compliant and non-compliant KRS functions, and receive certification only for the compliant ones. In this case, those assembling a complete compliant KRS must procure the products necessary to provide a complete set of compliant KRS functions. (Option 2) A product may be submitted for the evaluation of a subset of the KRS functions it provides. If each KRS function in the subset selected is certified as compliant, and the product disables any non-compliant KRS functions it provides when operating in compliant mode, then the product shall be certified compliant with the Standard. This allows a product to offer both compliant and non-compliant KRS functions, and operate in one of two modes: 1) compliant with the Standard by enabling compliant KRS functions and disabling non-compliant KRS functions or 2) non-compliant with the Standard by enabling non-compliant KRS functions. Those assembling a complete compliant KRS must procure the products necessary to provide a complete set of compliant KRS functions. (Option 3) A product shall be submitted for the evaluation of all of the KRS functions it provides. If each KRS function in the product is certified as compliant in one of its operating modes, and all KRS functions can be configured in a compliant mode simultaneously, the product shall be certified compliant with the Standard. This allows a product to offer KRS functions with both compliant and non-compliant modes, as long as each KRS function has a compliant ode and the product can operate with all its KRS functions configured in a compliant mode. Note that it is incumbent upon the vendor to accurately claim the KRS functions the product provides. (Option 4) A product shall be submitted for the evaluation of all the KRS functions it provides. If each KRS function in the product is certified as compliant, and each KRS function in the product can only operate a compliant mode, the product shall be certified compliant with the Standard. This prohibits a product from providing non-compliant KRS functions, or KRS functions that can be configured in both compliant and non-compliant modes. (Option 5) A product shall be submitted for the evaluation of its KRS functions according to one of the options above. If the product is certified compliant according to the criteria in the option chosen above, and there exist products providing certified compliant KRS functions needed for a complete KRS that are not included in the product under evaluation, then the product under evaluation shall be certified compliant with the Standard. This means that a product cannot be certified compliant until there exist products (including the one under evaluation) which together can form a complete compliant KRS. The products may be from one or multiple vendors. This also implies that the first KRS product to be evaluated must provide compliant KRS functions for a complete KRS. The security of a KRS is dependent on a mix of security disciplines, including computer, communication, procedural, physical, and personnel security. This Standard addresses only the computer and communication aspects of KRS security. Other critical aspects of KRS operation are outside the scope of this Standard. For example, a KRS must be available and survivable if it is to ensure authorized access to encrypted data, but this Standard does not address such concerns. Thus, compliance with this standard represents a set of necessary but not sufficient conditions for overall KRS security and utility. If key recovery is offered as a service by a trusted third party, that party could employ components (e.g., a KRA) that comply with this Standard. However, the use of compliant components does not ensure the security for a KRS as a whole, nor does it ensure available or survivable KRS operations, as noted above. Hence, a KRS service cannot be said to comply with this Standard. 1.2 Road Map for the Standard Section 2 of this Standard defines the abstract model for a KRS and defines the functions essential to KRS operation. Any product claiming compliance must identify which KRS functions are embodied in the product. Section 2 establishes functional and interoperability requirements for identified KRS functions. A product submitted for certification relative to this FIPS will be evaluated against the functional and interoperability requirements applicable to the functions that a vendor asserts are embodied in the product. Section 3 defines the security requirements for KRS functions. Two levels of compliance are defined: Level 1 and Level 2. An implementation of a function at Level 1 provides basic security functionality, whereas Level 2 offers a higher level of security functionality. The choice of level for an application or environment is context sensitive, a function of many factors, and this Standard provides no guidance to prospective users in this regard. Section 4 defines security assurance requirements for the implementation of KRS functions. These requirements are derived from the Common Criteria, and represent a profile of that security assurance evaluation criteria for use in this context. Three levels of (increasing) security assurance are defined: A, B and C. For each KRS function defined in Section 2, and each security functionality level defined in Section 3, one of these three assurance levels apply. Thus, there is a one-to-one correspondence between security functionality and assurance levels, on a per-function basis. Appendix A contains illustrative examples of how to map the functions defined in the model in Section 2 to sample KRS components in the context of common applications. It also includes examples of how to map several key recovery system technologies to these functions. These examples are provided to assist vendors and evaluators in understanding the KRS functional model, but are not normative. Appendix B defines a Key Recovery Block (KRB) format based on work by the Key Recovery Alliance. The adoption of this format would facilitate the encapsulation of KRI from different key recovery schemes and allow validation of the integrity of KRI in a KRS, in support of requirements specified in Section 2. Appendix C defines an extension for X.509 v3 certificates and a profile for other extensions employed in such certificates. Many KRS designs make use of public key certificates. The extension defined here provides a standard means of representing certain data supportive of several KRS requirements. This appendix provides guidance for KRS designers and standards bodies who choose to make use of X.509 v3 certificates in support of key recovery. Appendix D contains illustrative examples of how key recovery enabled systems can be designed to maximize interoperability, both with systems that do not implement key recovery, and with systems that implement different key recovery schemes. 2 Key Recovery Model A Key Recovery System (KRS) enables authorized persons to recover plaintext from encrypted data when the decryption key is not otherwise available. Key Recovery is a broad term that applies to many different key recovery techniques. Each technique will result in the recovery of a key - herein called the target key. The target key may be either: * the data key that can be used to decrypt the data, or * a key that can be used to decrypt the encrypted data key. The information required by each key recovery technique to recover the target key may be different for each technique. The term "key recovery information" (KRI) will be used to refer to the aggregate of information needed by a key recovery technique to recover the target key. The key recovery information can be managed or handled in a variety of ways. It may exist for only a brief time during electronic transmission, or it may exist for a relatively long time on a storage device. The KRI may be distributed among multiple location(s) (e.g., at one or more Key Recovery Agents (KRAs), with a registration authority, associated with or attached to a message or file, stored with a third party which is separate from a KRA, in end user systems, in third party systems, at a CA, in a certificate, or in a requestor facility). Figure 1 presents a generalized model for a Key Recovery System, consisting of a KRI Generation Function, A KRI Management Function and a Key Recovery Function. The model addresses the creation of KRI for the recovery of the target key, the management of the KRI, and the recovery of the target key from that KRI. The KRI Management Function is decomposed into a KRI Delivery Function, a KRI Acquisition Function and a KRI Validation Function. The Key Recovery Function is further divided into a Requestor Function and a KRA Function. The resulting six functions are shown in Figure 2. The key recovery model addresses multiple key recovery techniques (see Section 2.8) and supports a wide variety of data applications, including: * Interactive communication sessions, * Store-and-forward communications, and * Data storage. [TO BE MOVED TO SECTION 2.8] The key recovery model addresses both key escrowing (i.e., key backup) and encapsulated KRI techniques, as well as hybrids of these two techniques. A key escrowing technique employs one or more Key Recovery Agents (KRAs) who hold (i.e., escrow) keys or key components for their subscribers. In an encapsulated KRI technique, the KRA(s) do not hold keys belonging to their subscribers. Instead, a structure is created which contains information which will allow the Key Recovery Function to recover the subscriber's key. With both key escrowing and KRI encapsulation, a KRA may be operated by an organization as an integral part of its own security infrastructure, or a KRA may be operated by a third party organization. In the case of key escrowing by a third party organization, the key escrowing system is often called a key escrow system. A Key Recovery System (KRS) may exist over multiple "locations" (e.g., cryptographic end systems, KRA systems, Requestor system, and storage or transmission media). The normal key exchange mechanism is not inherently affected by any key recovery mechanisms. However, key exchange mechanisms may be used to support the creation and distribution of key recovery information (e.g., the integration of KRI into existing key exchange mechanisms is not precluded). In the future, key exchange protocol designers may find it beneficial to integrate key recovery into the base design of the protocol. Appendix A provides examples of the distribution of functions of the model within products implementing a Key Recovery System. The functions of the Key Recovery Model specified in this standard must be implemented in components or products which, when used together with a key recovery policy and procedures, form a Key Recovery System. A key recovery policy specifies the conditions under which key recovery information must be created and the conditions under which key recovery information may be released. The policy may also indicate the allowable Key Recovery Agent(s), how or where key recovery information must be maintained, and whether or not the received encrypted information should be processed when key recovery information is not available. The key recovery policy could be "hardwired" (e.g., implemented in a manner which does not allow key recovery to be bypassed), selectable by a user, or implemented in policy management tables or modules. The remainder of this section identifies functional and interoperability requirements for key recovery products which are designed to be conformant with this standard. Requirements are designated by "Req" numbers, and the requirement and its number are presented in a bold font. Explanatory text is provided in subsequent paragraphs. (Req. 1) There shall be a well-defined mapping from the key recovery functions of a product to the functions of the key recovery model. A product claiming compliance with the Standard must be mappable to one or more of the KRS functions defined in this Standard. There is no requirement that a product offered for evaluation embody all of the defined functions, nor is there a requirement that a single vendor provide a complete KRS. The modular implementation of a KRS, based on the assembly of components from one or more sources, is allowed. 2.1 Key Recovery Information (KRI) Generation Function (Req. 2) The KRI Generation Function shall generate all or part of the KRI. The KRI Generation Function consists of one or more KRI-generating entities, also called KRI providers. A KRI provider could, for example, be the sender or receiver of a communication, a Certification Authority (CA), a Key Distribution Center, a Registration Authority, or a component vendor. The KRI may include the identity of a KRA, the identity of a key, a date and time, authorization information, an indication of the key recovery type and manufacturer, an algorithm identifier, an encrypted key, or pointer information (e.g., information that points to the location or holder of a key). The KRI Generation Function may be distributed over multiple locations (e.g., systems, or hardware or software components) - all KRI required to recover a given data key/ciphertext set may not be created by the same generating entity. For example, the entity generating an encryption key pair may be different than the entity using that key pair to secure the data key which was used to encrypt the ciphertext data. See Appendix A for further examples. During an initialization or configuration stage, and at times of periodic updates, the KRI-generating entities obtain initialization information and cryptographic parameters, or otherwise are configured to establish shared information as necessary with the KRA(s) to allow key recovery. For example, for key escrowing systems (see Section 2.X), initialization and configuration may involve setting parameters which will allow a secure communication channel to be established between a cryptographic end system and a KRA for the escrowing of private keys. For KRI encapsulation systems (see Section 2.Y), initialization may involve obtaining authentic copies of the KRA public key(s) for subsequent use in encapsulating the KRI by the cryptographic end system. These are critical aspects of the overall Key Recovery System, but their definition is beyond the scope of this document. (Req. 3) The KRI Generation Function shall assemble and format the KRI for use by other key recovery functions. The KRI Generation Function generates, assembles and formats the KRI, as appropriate, for consumption by the KRI Validation Function, the Key Recovery Requestor Function and the KRA Function. The format of the KRI and its availability may be specific to a key recovery technique. Information may be acquired from multiple sources (e.g., one or more CA certificates, a key generation device or a time stamping device) in order to generate the required KRI necessary for a given key recovery technique. A method is required for associating encrypted data with the KRI which can be used to recover that data. This may be accomplished in a product by (1) providing plaintext information pointing to the KRI within a structure containing the encrypted data, or (2) providing plaintext information pointing to the encrypted data within a structure containing the KRI, or (3) by a well- defined placement of the KRI and the encrypted data (e.g., within the same message), or (4) by acquiring information from another source (e.g., by examining a certificate to determine that a key is escrowed), or (5) by a combination of such techniques. The KRI Generation Function needs to determine that all information required to generate the KRI associated with given ciphertext is valid. This includes all information generated by itself, as well as information generated by other sources (e.g., another KRI Generation Function, a CA, time stamping authority, etc.) which are used in the assembly and format process. Verification by the KRI Validation Function may be supported, for example, by the generation of an integrity check value for the KRI. (Req. 4) The KRI Generation Function shall provide the generated KRI to the KRI Delivery Function. 2.2 KRI Delivery Function The KRI Delivery Function makes the generated KRI available for validation and recovery (e.g., by storing or transmitting the KRI). The KRI Delivery Function may be distributed over multiple locations (e.g., systems, or hardware or software components). (Req. 5) When KRI is delivered in conjunction with a standard communication protocol, the transmission format shall be determined by that protocol standard. There are a number of standard communication protocols that allow the use of encryption to protect the data carried by that protocol. When KRI is introduced into one of these communication protocols, it must be done in a manner which preserves the ability to communicate (see Section 2.7, Interoperability). (Req. 6) The KRI Delivery Function shall ensure that the persistance and availability of the KRI is commensurate with that of the corresponding transmitted or stored ciphertext data. For example, in a communication context, KRI need not be transmitted with every packet, but may have to be periodically retransmitted to facilitate requestor access for a very long lived communication session. (Note that the underlying communication or storage system is not considered to be part of the KRI Delivery Function.) KRI for a given data key/ciphertext pair must be available for the duration of time that the given ciphertext exists. If the ciphertext is decrypted and subsequently not available in its original ciphertext form (e.g., stored in plaintext or re-encrypted with a different data key), then the original KRI is no longer required. (Req. 7) The KRI Delivery Function shall make the KRI available to the key recovery function. The KRI Delivery Function shall make the KRI available to the key recovery function (requestors or KRAs or a combination of both). The term "make available" is system dependent and includes sending the KRI to the Key Recovery Requestor directly, or depositing the KRI in location(s) known to and accessible to the Key Recovery Requestor (i.e., the requestor(s)). (Req. 8) The KRI Delivery Function shall make the KRI available to the KRI Validation Function. The KRI Validation Function is responsible for verifying the validity of the KRI. The KRI Delivery Function must provide the KRI produced by the KRI Generation Function to the KRI Validation Function via the KRI Acquisition Function. The method of delivery may be via a communication channel, storage device or directly between modules within the same system. [MOVE TO SECTION 2.8?]For KRI encapsulation techniques, the target key (or portion of the target key) or key related information is encrypted by a key associated with a KRA. This key encrypting key is typically a public encryption key for the KRA. Additional KRI may accompany the data key, depending on the key recovery technique. [NOTE THAT A KEY EXCHANGE MECHANISM MAY CREATE ENCAPSULATED KRI]. When a key escrowing technique is employed for key recovery, the data key is typically encrypted by a key for which the corresponding decryption key (e.g., the private key exchange decryption key) has been stored (i.e., escrowed), in whole or in part, by one or more KRAs. 2.3 KRI Acquisition Function (Req. 9) The KRI Acquisition Function shall acquire the KRI needed for subsequent processing. The acquisition function may be in a receiving system which could, for example, be a cryptographic end system, a requestor system or a KRA. The KRI Acquisition Function could be distributed over multiple locations (e.g., systems, or hardware or software components). 2.4 KRI Validation Function (Req. 10) The KRI Validation Function shall provide a capability for ensuring the availability of KRI. The intent of this function is to provide assurance that a key requestor can successfully recover encrypted data using the KRI. Several degrees of validation may be performed, including: * Checking certificates for the presence of KRI (e.g., KRA identities, key recovery technique), * Checking that KRI is available for a KRA (e.g., in a recipient list or a key recovery block), * Authenticating the source of the KRI, * Validating the integrity of KRI associated with the encrypted data (e.g., received in the same message), and * Verifying that the KRI can actually be used to recover the data key needed to decrypt the encrypted data (e.g., the correct target key can be produced). * Creating KRI, either when no KRI is received or in lieu of accepting and verifying KRI that is received. In this case, a KRI Generation Function must be available. 2.5 Requestor Function The Requestor Function authenticates the entity making the request to the Key Recovery Agent. The Requestor Function consists of the requestor and a Requestor System (see Figure 3). The requestor is an entity who seeks to recover information that will allow the decryption of encrypted data. A request for a key recovery service, made by a requestor using a Requestor System to interact with one or more Key Recovery Agents, must be an authorized request -- the requestor and the Requestor System which issues a request for a key recovery service must have a legal and lawful right to access the data that can be decrypted using the recovered target key. Furthermore, the requestor and the Requestor System must establish their right to access that data. The authentication and authorization process is beyond the scope of this standard. (Req. 11) For given KRI, the Requestor Function shall have the ability to recover a target key by interacting with one or more Key Recovery Agents. The requestor provides key recovery information to the requestor system(s). The requestor system(s) interacts with one or more KRAs to obtain either the target key, or multiple key parts or key related information which will allow the reconstruction of the target key. The target key can then be used to recover the data using a Data Recovery System which is not specified in this standard. KRI may be designed so that one KRA may not be able to provide all the information necessary to recover a target key. For example, each KRA may be able to provide key components which are then combined to reconstruct the target key. 2.6 Key Recovery Agent(s) A Key Recovery Agent system, hereafter called a Key Recovery Agent (KRA), is a trusted entity who performs a key recovery service in response to an authorized request made by a Requestor System on behalf of a requestor. (Req. 12) A Key Recovery Agent shall have the ability to store the KRI provided by the KRI Delivery Function if such storage is required to satisfy a key recovery request. The KRA may need to store keys, key components or other information which will allow the recovery of a target key. A Key Recovery Agent needs to authenticate the entity requesting key recovery and verify the entity's authorization to access the encrypted data. Before honoring such a key recovery request, the KRA must authenticate the Requestor System's right to receive the requested key recovery service. (Req. 13) A Key Recovery Agent shall have the ability to process the KRI provided by the Requestor Function. Processing by the Key Recovery Agent shall yield some or all of the information (e.g., key or key component) required to decrypt data acquired by a requestor. The key recovery service performed by a KRA consists of processing all or part of the key recovery information provided to the KRA by the Requestor System, and returning an output value to the Requestor System. The output value may be either the target key, or multiple key parts or key related information which will allow the reconstruction of the target key. 2.7 Interoperability This standard establishes interoperability requirements for several types of key recovery system components {products?}: cryptographic end systems (System A and System B), Key Recovery Agents and Key Recovery Requestors. No interoperability requirements are imposed on supporting (ancillary) components, nor are interoperability requirements imposed on communication between a cryptographic end system and a Key Recovery Agent (KRA). In both of these latter cases, the imposition of interoperability requirements is viewed as potentially too restrictive in light of the wide range of key recovery technologies that the standard attempts to embrace. This standard defines a syntax for communication between a Key Recovery Requestor (KRR) and a KRA. This syntax applies only to electronic key recovery transactions effected via a communication medium (e.g., telephone, LAN or Internet). Key recovery transactions effected via storage media (e.g., diskette or tape) or via direct interaction (e.g., self recovery on a PC) are not covered by these requirements. These syntactic requirements have been established to reduce life cycle costs for users of key recovery systems and because it appears to be feasible to do so without introducing undue constraints on technology options. Section XX defines the syntax for this communication syntax. No interoperability requirements are imposed on communication among KRAs from different vendors. Interoperability requirements for cryptographic end systems apply only to the use of key recovery for communicated data, not for data storage. With regard to such systems (i.e., System A and System B, where A and B are distinct), interoperability requirements apply only in the context of systems that communicate in an interoperable, encrypted fashion, exclusive of the use of key recovery technology. Such systems fall into two categories: those that make use of "standard" communication protocols and those that make use of "proprietary" protocols. For this standard, the phrase "standard communication protocol" encompasses any communication protocol that has been adopted by a generally recognized protocol standards organization, including the International Telecommunication Union (ITU), the American National Standards Institute (ANSI), the Institute of Electrical and Electronics Engineers (IEEE), the Asynchronous Transfer Mode (ATM) Forum and the Internet Engineering Task Force (IETF). No interoperability requirements are established for cryptographic end systems that engage in encrypted communications using proprietary communication protocols. Such systems typically exhibit limited interoperability (except within individual vendor product lines) due to the use of non-standard protocols. Vendors are encouraged to embody the key recovery technology in their products in a fashion that minimizes disruption to the installed product base in order to facilitate communication between key recovery products and non-key recovery products. When key recovery is introduced into a system using a standard (encryption) communication protocol, it must be done in a fashion that preserves interoperability. Interoperability is required when communicating with other key recovery capable systems, and when communicating with non-key recovery capable systems. (Some key recovery capable systems may be configured so that they will refuse to communicate with other systems unless it can be determined that the other systems are employing key recovery. If this feature is activated, it may prevent interoperability between otherwise interoperable systems. However, the presence of this configurable feature does not exempt a system from meeting the interoperability requirements detailed below.) There are two general approaches to meeting this requirement. If a private key escrow scheme is employed, the (extant) secure communication protocol employed by the cryptographic end systems need not be modified to carry any key recovery information, and thus, interoperability is preserved. Note that in this case, interoperability is preserved both among key recovery capable systems, and between key recovery capable and non-key recovery capable systems. If no changes are made to the secure communication protocol, including supporting key/certificate management protocols, then it may or may not be possible for communicating systems to determine if key recovery is being employed. If a private key escrow scheme elects to transmit some information in a secure communication protocol to indicate that key recovery is enabled, then it must do so in a fashion that does not impair interoperability. For example, if X.509 public key certificates are employed to support secure communication, an extension could be added to each certificate specifying the KRA(s) for the subject. If such an extension were employed and not marked "critical", this would comply with the interoperability requirement established here. However, if such an extension were employed and marked "critical", this would not be compliant, as it would inhibit interoperability with non-key recovery aware systems. See Appendix C for certificate extensions. If a KRI encapsulation scheme is employed, the key recovery information is carried in the secure communication protocol. In some standard, secure communication protocols, it is possible to carry this information in a fashion that preserves interoperability without modifying the protocol. For example, in a secure e-mail protocol (e.g., MSP, PGP, S/MIME and X.411), an additional recipient, representing a KRA, could be added to the per-recipient token list to provide for key recovery on a per message basis. In ISAKMP, a participant in the key management exchange could generate and transmit a NOTIFY message with a payload containing per-session key recovery information. Even though no standard format for such a NOTIFY payload has been defined, a compliant implementation is required to silently discard an unrecognized payload, and this technique preserves interoperability. Both of these approaches to providing key recovery would be compliant with the interoperability requirements established here. If it is necessary to transport key recovery information and there is no provision in a standard communication protocol for doing so in an interoperable fashion, then it will be necessary to modify/extend the protocol to carry such information. It is outside the scope of this standard to specify how key recovery information should be transported in the context of such protocols. The definition of an interoperable means of carrying such information is solely the purview of the cognizant standards body for each affected protocol. Thus, a vendor of a cryptographic end system in this category must provide documentation demonstrating that the product transports key recovery information in a fashion consistent with the specification developed and adopted by the cognizant standards body for the protocol in question. Only then can a cryptographic end system with key recovery be considered compliant with this interoperability requirement. [ADDITIONAL MATERIAL TO BE PROVIDED FROM THE INTEROPERABILITY WORKING GROUP] 2.8 Mapping Components to the Key Recovery Model The functions of the Key Recovery Model specified in this standard must be implemented in components or products which, when used together with a key recovery policy and procedures, form a Key Recovery System. The key recovery functions within the model may be distributed across these components as appropriate for the specific key recovery technique and the key recovery policy adopted for an organization. This section examines how a cryptographic end system component can map to the Key Recovery Model. Cryptographic end systems encrypt and decrypt data. In order to recover encrypted data, the key recovery information must be generated in order to allow the recovery of data keys used by that system. The necessary information needed to recover the data key may be made available, for example, as encapsulated information which may be stored or communicated with the encrypted data, or as cached data or both. The model does not specify which system or systems generate the KRI. When KRI is generated by cryptographic end systems, the KRI could be generated by the entity that encrypts data (e.g., the sender) or the entity that decrypts data (e.g., the receiver). A cryptographic end system generates and processes KRI in accordance with a specified key recovery policy. Figure 4 depicts a cryptographic end system application with an encryption/decryption capability. The application has a send/receive function for sending and receiving data, a key management function for generating and distributing keys, and an encrypt/decrypt function to encrypt and decrypt data. The Cryptographic End System Application processes cleartext and produces ciphertext and vice versa. Note that the communications end point which is generating and processing the cleartext need not be co-located with the Cryptographic End System Application. The key recovery functions within the cryptographic end system include a KRI Generation Function, a KRI Delivery Function, a KRI Acquisition Function, and a KRI Validation Function. Note that cryptographic end system products need not contain this specific set of key recovery functions (see Appendix A). The use of the functions within a cryptographic end system can depend on which key recovery technique is being used and whether the system is acting as a sender or receiver system. When a key encapsulation application is acting as a sender, it would typically perform the KRI Generate and Delivery Functions, whereas when acting as a receiver, it would often perform the KRI Acquisition and Validation Functions. In a key escrow- based application, however, the sender may perform the KRI Acquisition and Validation Functions, rather than the receiver. 2.9 Key Recovery Techniques Cryptographic end systems that satisfy this key recovery standard use key recovery techniques which may be broadly categorized into two types, KRI encapsulation and key escrow. The KRI encapsulation technique associates key recovery information with the encrypted data in a manner which allows the KRA to recover the data key. The key escrow technique makes the cryptographic end system's key, usually a long term key such as a public/private keypair, directly accessible by a KRA. This section provides an overview of these two techniques. 2.9.1 KRI Encapsulation Figure 5 illustrates the interaction of two cryptographic end systems that share or communicate encrypted data using a KRI encapsulation technique for key recovery. To make the data key recoverable, the KRI Generation Function within the Cryptographic End System A, hereinafter referred to as System A, first generates (or acquires) and encapsulates key recovery information (KRI) corresponding to the data key. Then, the KRI is provided to the KRI Delivery Function. Cryptographic End System B, hereinafter referred to as System B, may receive the KRI as well as the encrypted data and key exchange information. The KRI received by the KRI Acquisition Function may be provided to a KRI Validation Function. Whether and what type of validation is performed is dependent on the structure and content of the KRI, the key recovery technique used, and the validation policy of the receiving cryptographic end system. Validation may include checking that the key recovery technique used by System A is appropriate (e.g., by examining System A's certificate). This method works equally well where System A and System B are actually the same system, as would be the case in a storage application. 2.9.2 Key Escrow Figure 6 illustrates the interaction of two cryptographic end systems that share or communicate encrypted data using a key escrow technique for key recovery. For each cryptographic end system, keys, key parts or key related information to be recovered are delivered to and stored at the KRA. In this technique, a third party or a cryptographic end system acts as a KRI Provider, generating and delivering KRI to KRA(s). In an environment where System A is encrypting data and sending it to System B, there are at least two mechanisms whereby System A can make the data key recoverable. First, System A can determine that System B is using an acceptable key escrow technique for key recovery by acquiring this information from some source (e.g., a certificate) using its KRI Acquisition and Validation Functions. In this case, System A's normal performance of the key exchange/negotiation protocol is sufficient to make the data key recoverable. Secondly, knowing that its own private key has been escrowed, System A could create encapsulate KRI which will allow itself to recover the data key (e.g., by wrapping (encrypting) the data key using its own key pair and placing it in a recipient list or key recovery block), in addition to the normal performance of the key exchange/negotiation protocol. System A would then transmit the wrapped data key with the communication or stored data. This method only works, however, in instances where the key exchange/negotiation mechanism permits such operations (e.g. RSA, but not Diffie-Hellman. If required to do so, System B may verify recoverability for either of the two mechanisms described above. In the first mechanism noted above, System B merely verifies that its own public key has been escrowed. This allows the normal performance of the key exchange/negotiation protocol to make the data key recoverable. In the second mechanism noted above, System B may acquire information from some source (e.g., a certificate) using its KRI Acquisition and Validation Functions that System A's wrapped data key suitably provides for recoverability. 2.9.3 Interactions Between Systems Using Different Key Recovery Techniques Cryptographic end systems which interact with systems using different key recovery techniques may still provide for key recovery. Furthermore, compliant cryptographic end systems may provide for key recovery even when communicating with systems with no key recovery capability. 2.9.3.1 Interactions Between KRI Encapsulation and Key Escrow Techniques In Figure 7 System A uses a KRI encapsulation technique to provide for key recovery, whereas System B uses a key escrow technique. System A can use its KRI Acquisition and Validation Functions to determine that System B uses key escrow (e.g., by examining System B's certificate). System A can create encapsulated KRI using its KRI Generation Function and provide it to its KRI Delivery Function. System B's KRI Provider must independently provide KRI to System B's KRA prior to any possible recovery of System B's key. In this case, System B does not need to validate the encapsulated KRI since System B's key has been escrowed, though may optionally choose to do so. In Figure 8, System A uses a key escrow technique to provide for key recovery, whereas System B uses a KRI encapsulation technique. For System A to provide for key recovery, encapsulated information must be provided (e.g., by encrypting a copy of the data key for System A and placing it in a recipient list or in a key recovery block) using the KRI Generation and Delivery Functions. Note that for some key exchange schemes, normal performance of the key exchange mechanism may provide for the KRI generation and delivery functions. System B can use its KRI Acquisition and Validate Functions to determine the type of key recovery employed by System A (e.g., by examining System A's certificate) and check for the presence of encapsulated KRI. If System B must either validate or provide for the data key's recoverability, System B can itself generate and deliver encapsulated KRI in accordance with its key recovery technique provided that the key exchange/negotiation protocol allows. 2.9.3.2 Interactions Between KRI Encapsulation and Systems with No Key Recovery In Figure 5, if System A uses KRI encapsulation and System B has no key recovery capability, System A can provide encapsulated KRI even though System B cannot attempt to verify its recoverability. The encapsulated KRI received from System A must not cause interoperability problems with System B, however. If the roles are reversed and System B initiates a communication, System A's KRI Acquisition and Validate functions will detect that System B has not provided suitable KRI. If System A must either validate or provide for the data key's recoverability, System A can itself generate and deliver encapsulated KRI provided that the key exchange/negotiation protocol allows. 2.9.3.3 Interaction Between Key Escrow and Systems with No Key Recovery In Figure 8, if System A uses Key Escrow and System B has no key recovery capability, System A can recover only if encapsulated information is created by its own KRI Generation and Delivery Functions (e.g., by encrypting a copy of the data key for System A and placing it in a recipient list or in a key recovery block). System A must ensure that System B will be able to ignore the presence of the KRI in order to permit interoperability. If the roles are reversed and System B sends encrypted data to System A, System A can recover if the data key is recoverable using System A's escrowed key. 3 Security Requirements The security requirements for the KRA and for the Requestor functions have been defined to allow a variety of product architectures. These include using a monolithic product on which no other software/firmware can be loaded, using a monolithic product on which other software/firmware can be loaded, or using a layered product that has a distinct operating system, application, and cryptographic module. The requirements for the KRA and the Requestor functions have been defined so that all of these architectures can be evaluated. This is especially true of the requirements in the following areas: Audit, Identification and Authentication, some of the Access Control, and Protection of Trusted Security Functions. Furthermore, the product architecture may imply that some of the requirements do not apply, e.g., if the threat a requirement is supposed to mitigate does not arise in a particular implementation model. For example, if the product is a monolith on which no other software/firmware can be loaded, the domain separation, trusted path, and reference validation mechanism requirements do not apply since the untrusted software threat does not exist. 3.1 Key Recovery Agent Requirements 3.1.1 Level 1 - Medium Assurance 3.1.1.1 Cryptographic Functions (Req. 14) All cryptographic modules shall be compliant with FIPS 140-1, Level 2 or higher. 3.1.1.2 Cryptographic Algorithms (Req. 15) The key recovery scheme shall be implemented so that only FIPS approved algorithms are required to use it. The implementation of these algorithms shall conform to the applicable FIPS. 3.1.1.3 Confidentiality These requirements are intended to protect against both the outsider and insider threats. The only insider threat addressed is the unauthorized user. The authorized insider threat is handled elsewhere using audit, role separation, and multi-person control. (Req. 16) The KRA shall protect all key recovery information stored against disclosure to unauthorized individuals. (Req. 17) The KRA shall protect key recovery information transmitted - electronically or physically communicated against disclosure to parties other than the requestor(s). (Req. 18) The strength of the algorithm used to protect the key recovery information shall be greater than or equal to the maximum strength of the encryption or key management algorithms employed for data encryption or for generation of the keys being recovered. (Req. 19) The product shall apply confidentiality services to all outgoing transactions. The strength of the algorithm used for confidentiality shall be greater than or equal to the maximum strength of the encryption or key management algorithms employed for data encryption or for generation of the keys being recovered. 3.1.1.4 Integrity (Req. 20) The product shall protect all KRI stored against modification. (Req. 21) The product shall apply source authentication to all outgoing transactions (i.e., requests and responses). The strength of the algorithm used for authentication shall be greater than or equal to the maximum strength of the encryption or key management algorithms employed for data encryption or for generation of the keys being recovered. (Req. 22) The product shall apply integrity services to all outgoing transactions. The strength of the algorithm used for integrity shall be greater than or equal to the maximum strength of the encryption or key management algorithms employed for data encryption or for generation of the keys being recovered. 3.1.1.5 Audit These requirements are used to create a log of information to allow oversight by a security officer to detect unauthorized operations by a key recovery agent. (Req. 23) The product shall be capable of generating an audit record for the following events: * Start-up and shutdown of the audit functions, and * All auditable events as defined in the functional components included in this standard. (Req. 24) The following actions shall be auditable: * Any specific operation performed to process audit data stored in the audit trail. (Note: This include backup and deletion of audit trail); * Any attempt to read, modify or destroy the audit trail; * All requests to use authentication data management mechanisms; * All modifications to the audit configuration that occur while the audit collection functions are operating; * All requests to access user authentication data; * Any use of an authentication mechanism. (e.g. login) The authentication information shall not be stored in the audit trail; * All attempts to use the user identification mechanism, including the user identity provided; * Any attempt to perform an operation on the audit trail (including emptying the audit trail); * Use of a security-relevant administrative function; * Explicit requests to assume the security administrative role; * The allocation of a function to a security administrative role; * The addition or deletion of a user to/from a security administrative role; and * The association of a security-relevant administrative function with a specific security administrative role. (Req. 25) The following actions shall be audited. * The keys shall not be included in the audit; * Requests, responses, and other transactions generated by the product, including key recovery responses; and * Requests, responses, and other transactions received by the product, including key recovery requests. (Req. 26) The product shall record at least the following information within each audit record: * Date and time of the event, type of event, subject (user) identity, and success or failure of the event; and * Other information, as described in the functional component audit events. (Req. 27) The product shall be able to generate a human understandable presentation of any audit data stored in the permanent audit trail. (Req. 28) The audit trail shall not store the old or new authentication information (e.g., password). (Req. 29) The product shall be able to associate each auditable event with the identity of the user that caused the event. (Req. 30) The product shall provide the authorized administrator with the ability to empty the audit trail. (Note: emptying the audit trail means backup and delete) (Req. 31) The product shall be able to include or exclude auditable events from the set of audited events based on the following attributes: User identity, and/or Event Type (Note: The requirement applies to the auditable (i.e. optionally audited) events only. The mandatory audited events must never be excluded or excludable from the set of audited events.) (Req. 32) The product shall store generated audit records in a permanent audit trail. (Req. 33) The product shall restrict access to the audit trail to the authorized administrator. (Note: This requirement implies providing integrity to the audit trail.) 3.1.1.6 Identification and Authentication These requirements permit the unique identification of key recovery agent personnel. This facilitates individual accountability via audit functions and access controls. Requirements are levied on the strength of the authentication mechanism and the robustness of the authentication mechanism against attacks by rogue key recovery agent personnel. These requirements do not apply to electronic transactions (requests and responses). The electronic transactions may be identified and authenticated (if the scheme permits) using the access control policy. (Req. 34) The product shall provide functions for initializing and modifying KRA personnel authentication data. (Req. 35) The product shall restrict the use of these functions on the KRA personnel authentication data to the authorized administrator. (Note: These functions shall be assigned to the security administrator.) (Req. 36) The product shall allow authorized KRA personnel to use these functions to modify their own authentication. (Req. 37) The product shall protect authentication data that is stored in the product from unauthorized observation, modification, and destruction. (Req. 38) The product shall protect authentication information from unauthorized reuse, including replay. (Note: This and the previous requirement provide a capability for secure remote login.) (Req. 39) The product shall be able to terminate the KRA personnel session establishment process after five unsuccessful authentication attempts. (Note: If the product terminates the session after fewer than five unsuccessful attempts, the product meets this requirement.) (Req. 40) After the termination of a KRA user session establishment process, the product shall be able to disable the user account until the account is enabled by an authorized administrator (i.e., security administrator).) (Req. 41) The product shall authenticate any KRA user's claimed identity prior to performing any functions on the user's behalf. (Req. 42) The product shall uniquely identify each KRA user before performing any actions requested by the user. (Req. 43) The product shall provide a mechanism to verify that secrets (i.e., authentication information such as passwords) meet the FIPS PUB 112 "Password Usage" and Department of Defense Password Management Guideline (CSC-STD-002-85). (Note: This requirement implies that the product (as opposed to procedural means and management instructions) will enforce the password length, aging, etc. type requirements.) 3.1.1.7 Access Control These requirements provide countermeasures against an entity masquerading as an authorized requestor or key recovery information generator. The requirements in this section address the security of electronic communication between the KRA and the requestor or information generator. If these interactions are not electronic, then physical and procedural means may be used to secure the transactions. These procedural and physical measures are beyond the scope the standard. (Req. 44) The product shall unambiguously associate the received response to an outstanding request. The strength of the algorithm used for the association shall be greater than or equal to the maximum strength of the encryption or key management algorithms employed for user traffic encryption or for generation of the keys being recovered. (Req. 45) The product shall release the keys only to authorized users. (Req. 46) The product shall release a key only if the requester is authorized to receive the key associated with the user specified in the request and for the validity period (time interval). (Req. 47) The product shall ensure that security features are always invoked and cannot be bypassed. (Req. 48) The product shall maintain a security domain for its own execution that protects it from interference and tampering by untrusted subjects. (Req. 49) The product shall enforce separation between the security domains of subjects in the system. (Req. 50) The product shall distinguish between security-relevant administrative functions from other functions. (Req. 51) The set of security-relevant administrative functions shall include all functions necessary to install, configure, and manage the product; minimally, this set shall include the assignment/deletion of authorized users from security administrative roles, the association of security-relevant administrative commands with security administrative roles, the assignment/deletion of subjects whose keys are held, the assignment/deletion of parties who may be provided the keys, product cryptographic key management, actions on the audit log, audit profile management, and changes to the system configuration. (Req. 52) The product shall restrict the ability to perform security-relevant administrative functions to a security administrative role that has a specific set of authorized functions and responsibilities. (Note: The term "security administrative role" refers to generic trusted administrative roles. The system administrator role is one, but not the only one, of these security administrative roles.) (Req. 53) The product shall be capable of distinguishing the set of KRA operators authorized for administrative functions from the set of all other users. (Req. 54) The product shall allow only specifically authorized KRA operators to assume the security administrative role (Req. 55) The product shall require an explicit request to be made in order for an authorized KRA operator to assume the security administrative role. 3.1.1.8 Authentication of Received Transactions (Req. 56) The product shall verify the source of received transactions. (Req. 57) The product shall verify the integrity of received transactions. (Req. 58) The product shall decrypt the received transactions which are encrypted. 3.1.1.9 Non-Repudiation These capabilities facilitate the use of a trusted time source to further support accountability. (Req. 59) The product shall be able to provide reliable time stamps for its own use. (Req. 60) The product shall be able to generate evidence of receipt for received transactions. (Note: The above requirement means using the reliable time stamp to put a trusted time stamp on the receipt. Furthermore, this requirement means that the product shall be able to generate evidence of receipt of the registration or deposit of key recovery information from users, and requests from a requestor) 3.1.1.10 Protection of Trusted Security Functions (Req. 61) Before establishing a session with a KRA operator, the product shall display an advisory warning message regarding unauthorized use of the product. (Req. 62) The default advisory warning message displayed by the product shall be as follows: "This system shall be used only by authorized personnel and only for authorized key recovery purposes. Violation shall result in criminal prosecution and civil penalties". (Req. 63) The product shall restrict the capability to modify the warning message to the authorized administrator. (Req. 64) Upon successful session establishment, the product shall display the date, time, method, and port of the last successful session establishment to the KRA operator. (Req. 65) Upon successful session establishment, the product shall display the date, time, method, and location of the last unsuccessful attempt to session establishment and the number of unsuccessful attempts since the last successful session establishment. (Req. 66) The data specified above shall not be removed without KRA operator intervention. 3.1.2 Level 2 - High Assurance 3.1.2.1 Cryptographic Functions (Req. 67) All cryptographic modules shall be compliant with FIPS 140-1, Level 3 or higher. 3.1.2.2 Cryptographic Algorithms Same as Level 1. 3.1.2.3 Confidentiality Level 2 requires additional protection against the insider threat of a rogue key recovery agent by requiring multi-party control on access to the key recovery information. All level 1 requirements and the following: (Req. 68) The system shall be designed for multiple KRAs. Two or more KRAs shall be required to obtain the key recovery information. 3.1.2.4 Integrity Same as Level 1. 3.1.2.5 Audit Level 2 adds a real time alarm to the security officer in the event of the audit trail becoming full to prevent audit data from being lost. It also requires the auditing of additional events consistent with the additional security requirements (i.e. trusted path) added at Level 2. Includes all the requirements of Level 1 and (Req. 69) The product shall generate an alarm to the authorized administrator if the size of the audit data in the audit trail exceeds a pre-defined limit. (Req. 70) The product shall provide the authorized administrator with the ability to manage the audit trail at any time during the operation of the product. (Req. 71) The following actions shall be auditable: * Execution of the tests of the underlying machine and the results of the tests; * All attempted uses of the trusted path functions; * Identification of the initiator and target of the trusted path; * Attempts to provide invalid inputs for administrative functions; and * The invocation of the non-repudiation service. The audit event shall include the identification of the information, the destination, and a copy of the evidence provided. The event shall exclude all private and secret keys in encrypted or unencrypted form. 3.1.2.6 Identification and Authentication Level 2 enhances assurance by supporting a hardware token. This provides an additional countermeasure to the threat of an attack on the authentication mechanism and the subsequent unauthorized access to key recovery information or critical functions. All Level 1 requirements and the following: (Req. 72) The product shall support a token based authentication. The token shall meet FIPS 140-1 Level 2 requirements. 3.1.2.7 Access Control Level 2 requires multi-party access controls for the release of key recovery information, and establishes roles and responsibilities for key recovery facility personnel as additional countermeasures to the threat of a single rogue key recovery agent. All Level 1 requirements and the following: (Req. 73) The KRA shall embody a facility for multi-party (at least 2) authorization in support of the release of key material. The following requirements are to provide for strict role separation (Req. 74) The product shall distinguish security-relevant administrative functions from other functions. (Req. 75) The set of security-relevant administrative functions shall include all functions necessary to install, configure, and manage the product; minimally, this set shall include the assignment/deletion of authorized users from security administrative roles, the association of security-relevant administrative commands with security administrative roles, the assignment/deletion of subjects whose keys are held, the assignment/deletion of parties who may be provided the keys, product cryptographic key management, actions on the audit log, audit profile management, and changes to the system configuration. (Req. 76) The product shall restrict the ability to perform a security-relevant administrative function to the security administrative role(s) authorized to use that function. (Req. 77) The product shall be capable of distinguishing the set of users authorized for administrative functions from the set of all other users. (Req. 78) The product shall allow only specifically authorized users to assume only those security administrative roles for which they have been authorized. (Req. 79) The product shall require an explicit request to assume a specific security administrative role to be made in order for an authorized user to assume that system administrator role. (Req. 80) The product shall define a set of security administrative roles that minimally includes system administrator, system operator, crypto officer and audit administrator. (Req. 81) The system administrator shall perform the following functions: the assignment/deletion of authorized users from system administrative roles, the association of security-relevant administrative commands with security administrative roles, the assignment/deletion of subjects whose keys are held, the assignment/deletion of parties who may be provided the keys. (Req. 82) The system operator shall change the system configuration and run the system. (Req. 83) The crypto officer shall manage the cryptographic keys. (Req. 84) The audit administrator shall manage the audit log and audit profiles. (Req. 85) The product shall associate each security-relevant administrative function with at least one security administrative role. (Req. 86) The product shall enforce checks for valid input values for security- relevant administrative functions as described in the Administrative guidance. 3.1.2.8 Authentication of Received Transactions Same as Level 1. 3.1.2.9 Non Repudiation Level 2 requires additional capabilities to prove the origin of transmissions to allow recipients to counter the threat of an adversary spoofing as a Key Recovery Agent. All Level 1 requirements and the following: (Req. 87) The product shall generate evidence of origin for transmitted key recovery requests or responses. (Note: The above requirement shall also [include?] the reliable time stamp service to include the time in the evidence.) (Req. 88) The product shall provide a capability to verify the evidence of origin of information to the recipient. (Req. 89) The product shall provide a capability to verify the evidence of receipt of proof of receipt to the originator of message (i.e., recipient of proof of receipt). (Req. 90) The product shall provide the originator with the ability to request evidence of receipt on information. 3.1.2.10 Protection of Trusted Security Functions All Level 1 requirements and the following: (Req. 91) The product shall provide a communication path between itself and local human users that is logically distinct from other communication paths and provides assured identification of its endpoints. (This communication path is generally called a trusted path.) (Req. 92) The local human users shall have the ability to initiate communication via the trusted path. (Req. 93) The product shall require the use of the trusted path for initial user (i.e., KRA operator) authentication. (Req. 94) The product shall provide the authorized administrator with the capability to demonstrate the correct operation of the security-relevant functions provided by the underlying abstract machine. (Req. 95) The product shall preserve a secure state when the abstract machine tests fail. 3.2 Key Recovery Information Generator 3.2.1 Level 1 - Medium Assurance Key Recovery Information Generator 3.2.1.1 Cryptographic Functions (Req. 96) All cryptographic modules shall be FIPS 140-1, Level 1 compliant. 3.2.1.2 Cryptographic Algorithms (Req. 97) The key recovery scheme shall be implemented so that only FIPS approved algorithms are required to use it. The implementation of these algorithms shall conform to the applicable FIPS standard(s). 3.2.1.3 Confidentiality This requirement is intended to minimize the vulnerability created by the key recovery mechanism. The key recovery mechanism should not be weaker and thus easier to attack than the original encryption mechanism. (Req. 98) The strength of the algorithm used to protect the key recovery information shall be greater than or equal to the maximum strength of the encryption or key management algorithms employed for data encryption or for generation of the keys being recovered. 3.2.1.4 Integrity These requirements counter the threat of an outsider corrupting the key recovery information. (Req. 99) The KRI Generator shall generate an integrity value for the KRI. (Req. 100) The KRI Generator shall associate the key recovery information with the encrypted data. (Req. 101) The KRI Generator shall generate an integrity value for the association of the key recovery information to the data. As an example, a key recovery scheme that includes a keyed message digest computed on the KRI using the data encryption key meets all of the above three integrity requirements. Requirement 99 is met since the keyed message digest provides integrity. Requirement 100 is met by the unambiguous placement of KRI and encrypted data as defined by the protocol (e.g., fixed location, pointer, tagged information, etc.). Requirement 101 is met since the same key is used to calculate or verify the keyed message digest and to decrypt the data, which ensures the integrity of the association between the KRI and the encrypted data. 3.2.1.5 Identification and Authentication (Req. 102) All cryptographic modules shall implement role-based authentication. (Req. 103) One of the roles shall be the system administrator role. 3.2.1.6 Access Control (Req. 104) The KRI Generator shall allow only the system administrator to configure the key recovery mechanism. (Req. 105) At a minimum, the configurations shall include activation and deactivation of the key recovery mechanism. 3.2.2 Level 2 - High Assurance Key Recovery Information Generator 3.2.2.1 Cryptographic Functions (Req. 106) All cryptographic modules shall be FIPS 140-1, Level 2 compliant. 3.2.2.2 Cryptographic Algorithms Same as Level 1. 3.2.2.3 Confidentiality Same as Level 1. 3.2.2.4 Integrity All of Level 1 requirements and the following: (Req. 107) The product shall provide the capability for the decryptor to verify that the key recovery information can be successfully used to decrypt the data. 3.2.2.5 Identification and Authentication Same as Level 1. 3.2.2.6 Access Control Same as Level 1. 3.3 Key Recovery Information Delivery No Security requirements. 3.4 Key Recovery Information Acquisition No security requirements. 3.5 Key Recovery Information Validator 3.5.1 Level 1 - Medium Assurance Key Recovery Information Validator 3.5.1.1 Cryptographic Functions (Req. 108) All cryptographic modules shall be FIPS 140-1, Level 1 compliant. 3.5.1.2 Cryptographic Algorithms (Req. 109) The key recovery scheme shall be implemented so that only FIPS approved algorithms are required to use it. The implementation of these algorithms shall conform to the applicable FIPS standard(s). 3.5.1.3 Integrity The purpose of the integrity requirements is to ensure that the key recovery information can be used to successfully decrypt the communication when the receiver can successfully decrypt the communication. Level 1 requirements counter the threat of an outsider corrupting the key recovery information. Level 2 requirements counter the threat of the sender corrupting the key recovery information. (Req. 110) Prior to decrypting the data, the KRI validator shall verify that the KRI acquired was that intended by the KRI generator. (Req. 111) Prior to decrypting the data, the KRI validator shall verify that the association of the key recovery information with the encrypted data was that intended by the KRI Generator. (Req. 112) Prior to decrypting the data, the KRI validator shall verify the integrity of the association of the key recovery information to the encrypted data. See Section 3.2.1.4 "Key Recovery Information Generator - Integrity" for an example of how the above integrity requirements can be satisfied. 3.5.2 Level 2 - High Assurance Key Recovery Information Validator 3.5.2.1 Cryptographic Functions (Req. 113) All cryptographic modules shall be FIPS 140-1, Level 2 compliant. 3.5.2.2 Cryptographic Algorithms Same as Level 1. 3.5.2.3 Integrity The product shall meet at least one of the following (i.e., is required to meet only one of the following, but may meet more than one) integrity requirements: (Req. 114) The validator shall ensure that the KRI received is accurate, i.e., the information can be used to perform key recovery successfully. The validtor only needs to meet the Level 1 integrity requirements when interoperating with a product supporting a different scheme. (Req. 115) A KRI Generator Function in the receiving cryptographic end system shall generate accurate key recovery information for received encrypted data. (Req. 116) The receiving cryptographic end system shall not be able to obtain the correct data decryption key if the received key recovery information is not accurate. 3.6 Key Recovery Requestor The security requirements for the KRA and for the Requestor functions have been defined to allow a variety of product architectures. These include using a monolithic product on which no other software/firmware can be loaded, using a monolithic product on which other software/firmware can be loaded, or using a layered product that has a distinct operating system, application, and cryptographic module. The requirements for the KRA and the Requestor functions have been defined so that all of these architectures can be evaluated. This is especially true of the requirements in the following areas: Audit, Identification and Authentication, some of the Access Control, and Protection of Trusted Security Functions. Furthermore, the product architecture may imply that some of the requirements do not apply, e.g., if the threat a requirement is supposed to mitigate does not arise in a particular implementation model. For example, if the product is a monolith on which no other software/firmware can be loaded, the domain separation, trusted path, and reference validation mechanism requirements do not apply since the untrusted software threat does not exist. 3.6.1 Level 1 - Medium Assurance 3.6.1.1 Cryptographic Functions (Req. 117) All cryptographic modules shall be compliant with FIPS 140-1, Level 2 or higher. 3.6.1.2 Cryptographic Algorithms (Req. 118) The key recovery requests and responses shall be implemented so that only FIPS approved algorithms are required to use them. The implementation of these algorithms shall conform to the applicable FIPS standard(s). 3.6.1.3 Confidentiality (Req. 119) The requestor shall protect key recovery information received and/or stored against disclosure to unauthorized individuals. (Note: Storing the data encrypted and implementing access controls is one way to meet this requiremen.t) (Req. 120) The requestor shall protect the key recovery request (specially the identities of subjects and time periods, if applicable) transmitted against disclosure to parties other than the KRA. (Note: Encryption of the request is one way to meet this requirement.) (Req. 121) The product shall apply confidentiality services to all requests. The strength of the algorithm used for confidentiality shall be greater than or equal to the maximum strength of the encryption or key management algorithms employed for data encryption or for generation of the keys being recovered. 3.6.1.4 Integrity (Req. 122) The product shall apply source authentication to all requests. The strength of the algorithm used for authentication shall be greater than or equal to the maximum strength of the encryption or key management algorithms employed for data encryption or for generation of the keys being recovered. (Req. 123) The product shall apply integrity services to all requests. The strength of the algorithm used for integrity shall be greater than or equal to the maximum strength of the encryption or key management algorithms employed for data encryption or for generation of the keys being recovered. 3.6.1.5 Audit (Req. 124) The product shall be capable of generating an audit record for the following events: * Start-up and shutdown of the audit functions; and * All auditable events as defined in the functional components included in this standard. (Req. 125) The following actions shall be auditable: * Any specific operation performed to process audit data stored in the audit trail. (Note: This include backup and deletion of audit trail); * Any attempt to read, modify or destroy the audit trail; * All requests to use authentication data management mechanisms; * All modifications to the audit configuration that occur while the audit collection functions are operating; * All requests to access user authentication data; * Any use of an authentication mechanism. (e.g. login) The authentication information shall not be stored in the audit trail; * All attempts to use the user identification mechanism, including the user identity provided; * Any attempt to perform an operation on the audit trail; * Use of a security-relevant administrative function; * Explicit requests to assume the security administrative role; * The allocation of a function to a security administrative role; * The addition or deletion of a user to/from a security administrative role; and * The association of a security-relevant administrative function with a specific security administrative role. (Req. 126) The following actions shall be audited. The keys shall not be included in the audit: * Requests, responses, and other transactions generated by the product, including key recovery responses; and * Requests, responses, and other transactions received by the product, including key recovery requests. (Req. 127) The product shall record within each audit record at least the following information: * Date and time of the event, type of event, subject (user) identity, and success or failure of the event; and * Other information, as described in the functional component audit events (Req. 128) The product shall be able to generate a human understandable presentation of any audit data stored in the permanent audit trail. (Req. 129) The audit trail shall not store the old or new authentication information (e.g., password) (Req. 130) The product shall be able to associate each auditable event with the identity of the user that caused the event. (Req. 131) The product shall provide the authorized administrator with the ability to empty the audit trail. (Note: emptying the audit trail means backup and delete). (Req. 132) The product shall be able to include or exclude auditable events from the set of audited events based on the following attributes: User identity, and/or Event Type (Note: The requirement applies to the auditable events only. The always audited events must never be excluded or excludable from the set of audited events.) (Req. 133) The product shall store generated audit records in a permanent audit trail. (Req. 134) The product shall restrict access to the audit trail to the authorized administrator. (Note: This requirement implies providing integrity to the audit trail) 3.6.1.6 Identification and Authentication The requirements in this section are for the identification and authentication of the various requestor personnel. These requirements do not apply to electronic transactions (requests and responses). (Req. 135) The product shall provide functions for initializing and modifying user authentication data. (Req. 136) The product shall restrict the use of these functions on the user personnel authentication data for any user to the authorized administrator. (Req. 137) The product shall allow authorized users to use these functions to modify their own authentication data in accordance with the identification and authentication policy. (Note: The identification and authentication policy should permit each person to change his/her authentication information.) (Req. 138) The product shall protect authentication data that is stored in the product from unauthorized observation, modification, and destruction. (Req. 139) The product shall be able to terminate the user session establishment process after five unsuccessful authentication attempts. (Note: If the product terminates the session after less than five unsuccessful attempts, the product meets the above requirement.) (Req. 140) After the termination of a user session establishment process, the product shall be able to disable the user account until the account is enabled by an authorized administrator (i.e., security administrator). (Req. 141) The product shall authenticate any user's claimed identity prior to performing any functions on the user's behalf. (Req. 142) The product shall uniquely identify each user before performing any actions requested by the user. (Req. 143) The product shall provide a mechanism to verify that secrets (i.e., authentication information such as passwords) meet the FIPS PUB 112 "Password Usage" and Department of Defense Password Management Guideline (CSC-STD-002-85). (Note: This requirement implies that the product (as opposed to procedural means and management instructions) will enforce the password length, aging, etc. type requirements.) 3.6.1.7 Access Control These requirements provide countermeasures against an entity masquerading as an authorized requestor. The requirements in this section address the security of electronic communication between the KRA and the requestor. If these interactions are not electronic, then physical and procedural means may be used to secure the transactions. These procedural and physical measures are beyond the scope the standard. (Req. 144) The product shall verify the association of the response to an outstanding request. (Req. 145) The product shall ensure that the key recovery information is destroyed (e.g., by zeroizing) when it is no longer required, when it is no longer valid (e.g., time expiry), when the KRA requires its deletion, or when the legal authority to it expires, whichever occurs first. (Req. 146) The product shall ensure that security features are always invoked and cannot be bypassed. (Req. 147) The product shall maintain a security domain for its own execution that protects it from interference and tampering by untrusted subjects. (Req. 148) The product shall enforce separation between the security domains of subjects in the system. 3.6.1.8 Authentication of Received Transactions (Req. 149) The product shall verify the source of received transactions. (Req. 150) The product shall verify the integrity of received transactions. (Req. 151) The product shall decrypt the received transactions which are encrypted. 3.6.1.9 Non-Repudiation (Req. 152) The product shall be able to provide reliable time stamps for its own use. (Note: We want to rely more on the KRA for time. But, having a requestor time stamp does not hurt.) (Req. 153) The product shall be able to generate evidence of receipt for received transactions. (Note: This requirement means using the reliable time stamp to put a trusted time stamp on the receipt. Furthermore, this requirement means that the product shall be able to generate evidence of receipt of registration or deposit of key recovery information from users, and requests from a requestor.) 3.6.1.10 Protection of Trusted Security Functions (Req. 154) Before establishing a session with a user, the product shall display an advisory warning message regarding unauthorized use of the product. (Req. 155) The default advisory warning message displayed by the product shall be as follows: "This system shall be used only by authorized personnel and only for authorized key recovery purposes. Violation shall result in criminal prosecution and civil penalties". (Req. 156) The product shall restrict the capability to modify the warning message to the authorized administrator. (Req. 157) Upon successful session establishment, the product shall display the date, time, method, and location of the last successful session establishment to the user. (Req. 158) Upon successful session establishment, the product shall display the date, time, method, and location of the last unsuccessful attempt to session establishment and the number of unsuccessful attempts since the last successful session establishment. (Req. 159) The data specified above shall not be removed without user intervention. 3.6.2 Level 2 - High Assurance 3.6.2.1 Cryptographic Functions (Req. 160) All cryptographic modules shall be compliant with FIPS 140-1, Level 3 or higher. 3.6.2.2 Cryptographic Algorithms Same as Level 1. 3.6.2.3 Confidentiality Same as level 1. 3.6.2.4 Integrity Same as level 1. 3.6.2.5 Audit Includes all the requirements of Level 1 and the following: (Req. 161) The product shall generate an alarm to the authorized administrator if the size of the audit data in the audit trail exceeds a pre-defined limit. (Req. 162) The product shall provide the authorized administrator with the ability to manage the audit trail at any time during the operation of the product. (Req. 163) The following actions shall be auditable: * Execution of the tests of the underlying machine and the results of the tests; * All attempted uses of the trusted path functions; * Identification of the initiator and target of the trusted path; * Attempts to provide invalid inputs for administrative functions; and * The invocation of the non-repudiation service. The audit event shall include identification of the information, the destination, and a copy of the evidence provided. The event shall exclude all private and secret keys in encrypted or unencrypted form. 3.6.2.6 Identification and Authentication All Level 1 requirements and the following: (Req. 164) The product shall support and token based authentication. The token shall meet FIPS 140-1 Level 2 requirements. 3.6.2.7 Access Control All Level 1 requirements and the following: (Req. 165) Two or more users shall be required to request the recovery information from a KRA. (Req. 166) The product shall enforce checks for valid input values for security- relevant administrative functions as described in the Administrative guidance. 3.6.2.8 Authentication of Received Transactions Same as Level 1. 3.6.2.9 Non Repudiation Includes the Level 1 requirements and the following: (Req. 167) The product shall generate evidence of origin for transmitted key recovery requests or responses. (Note: This requirement shall also support the reliable time stamp service by including the time in the evidence.) (Req. 168) The product shall provide a capability to verify the evidence of origin of information to the recipient. (Req. 169) The product shall provide a capability to verify the evidence of receipt of proof of receipt to the originator of message (i.e., recipient of proof of receipt). (Req. 170) The product shall provide the originator with the ability to request evidence of receipt on information. 3.6.2.10 Protection of Trusted Security Functions All Level 1 requirements and the following: (Req. 171) The product shall provide a communication path between itself and local human users that is logically distinct from other communication paths and provides assured identification of its endpoints. (Req. 172) The local human users shall have the ability to initiate communication via the trusted path. (Req. 173) The product shall require the use of the trusted path for initial user authentication. (Req. 174) The product shall provide the authorized administrator with the capability to demonstrate the correct operation of the security-relevant functions provided by the underlying abstract machine. (Req. 175) The product shall preserve a secure state when abstract machine tests fail. THE FOLLOWING ARE NOT REQUIREMENTS, BUT INFORMATIVE TEXT TO BE USED SOMEWHERE OR TO BE DELETED. KRA Availability These suggestions are intended to provide the capability for a key recovery agent to recover in the event of a system failure or compromise. They act as a counter to the threat of the unauthorized destruction of the key recovery information or capabilities at the key recovery agent facility. The KRA facility should be required to have the capability to securely replicate any key recovery information stored in order to support continued on-line access in case of a facility failure. The KRA facility should have a secure backup of the key recovery information stored in order to rebuild the key recovery database in case of KRA system failure. Ancillary Components Registration Agent Registration Agents maintain information on key recovery products and corresponding key recovery protocol (schemes). The registration agent should be able to ensure the accuracy and maintain the integrity of the product information. Integrity/Authenticity These features counter the threat of an adversary spoofing as the registration agent and of unauthorized access to the information and critical functions at the registration agent. The Registration Agent should verify authentication and integrity services for the received product information. The Registration Agent should apply authentication and integrity services to the product information it transmits. The Registration Agent should ensure that the product information it maintains is not modified by unauthorized parties. Licensing Agent Licensing Agents perform compliance audits of the KRAs to ensure that the KRAs operate in accordance with the KRA's stated policy. Authentic Public Key Source (APKS AKA Public Key Infrastructure (PKI)) Standards The APKS should carry out transactions in accordance with the Minimum Interoperability Specifications for PKI Components (MISPC) Security/Certificate Policy: The security of PKI and the degree to which the binding between an entity (subject or subscriber) and public key can be trusted, is determined by the Certificate Policy. Certificate Policy is defined and described in Certificate Policy Framework. Using this Framework, NIST has developed Baseline Security Requirements. NIST plans to enhance these for up to three more strictly superior policies. Thus, in order to define the security requirements for the APKS, we only need to select the proper certificate policy. Please note the certificate policy security requirements are quite comprehensive. For details, see IETF PKIX Part IV: Certificate Policy Framework. For Level 1, the APKS should meet the medium Certificate Policy. For Level 2, the APKS should meet the high Certificate Policy. 4 Assurance Requirements The assurance in a KRS compliant product can be achieved using the Common Criteria Evaluation Assurance Levels (EAL). The Common Criteria (CC) defines seven hierarchical assurance levels EAL1 through EAL7. The Common Criteria assurance levels may be overkill for the KRS compliance validation program. Thus, this section contains a tailored list of assurance requirements. These requirements are derived from the Common Criteria Part 3 (Assurance Requirements). Specifying assurance requirements in the common criteria language will help in converting the FIPS into a Common Criteria Protection Profile and in validating KRS compliant products under the Common Criteria (CC) Evaluation Methodology. For the sake of clarity, it should be noted that the CC structure for assurance requirements is hierarchical as follows. At the highest level, the requirements are categorized into classes. The classes are further decomposed into families. The families are decomposed into components. Each component has three sets of elements. The first set of elements is the list of developer (vendor) requirements which must be satisfied for the component. The second set of elements is a list of contents and presentation requirements for the assurance evidence for that element. The third and last set of elements is what an independent evaluator should do to assess the contents and presentation items which are provided. A later section of this report also explains why the remaining Common Criteria assurance requirements are not recommended. Three Assurance Levels (AL) are defined for this standard. These levels are somewhat related to the Common Criteria assurance levels, but are not derived from the Common Criteria assurance levels. The assurance levels for the classes, families and components of key recovery products are listed in Table 1. Subsequent sections provide further detail. (Req. 176) The KRA and Key Recovery Requestor Functions shall be required to meet the assurance requirements for AL B and AL C for Security Levels 1 and 2, respectively, as defined in Tables 1 and 2. (Req. 177) The KRI Generation and Validation Functions shall be required to meet the assurance requirements for AL A and AL B for Security Levels 1 and 2, respectively, as defined in Tables 1 and 2. Table 2 provides a summary of assurance level requirements for the various KRS functions. It should be noted that the assurance requirements are applied to test the product functionality and security features. Assurance Concept The assurance concepts and notations in this standard are based on the Common Criteria. The assurance concept consists of a hierarchical refinement of the requirements. At the top-level, the assurance requirements are broken down into classes. The classes include, configuration management, delivery and operation, development, guidance documents, life-cycle support, testing, and vulnerability analysis. Each class is broken down into families. For example, the development class contains families such as functional specification, high-level design, low-level design, implementation representation, etc. Each family consists of one or more components. Each component contains three sets of elements. The first set is the product developer actions. The second set is the requirements for content and presentation of information. The third and final set contains the evaluator actions. Assurance Notations The notation used for assurance requirements is based on the Common Criteria. Each class is defined as three characters; the first character is always "A" for assurance; the remaining two characters are meaningful for the class; e.g., CM for configuration management, DV for development, TE for testing, etc. The three letter assurance class is followed by an underscore "_" and a three letter meaningful name for the family, e.g., FSP for functional specification. The family is followed by "." and a component number. The component number is followed by a "." and a two character element indicator. Each component contains three sets of numbered elements. The first character of the element indicator is a sequential number within the set. The second character indicates the set : "D" for a developer action, "C" for content and presentation, and "E" for an evaluator action. In Table 1 below, the numbers in the last three columns identify the components of each family that must be satisfied in order to provide the appropriate assurance level. For example, for assurance family ADV_FSP (see Section 4.3.1), assurance level A specifies that component 1 applies. Component 1 is identified as ADV_FSP.1, and the requirements are listed in Section 4.2.1.1, ADV_FSP.1 Functional Specification and Security Policy. For assurance levels B and C, component 2 applies. Component 2 is identified as ADV_FSP.2, and the requirements are listed in Section 4.3.1.2, ADV_FSP.2 Informal Security Policy Model. Table 1: KRS Assurance Levels Assurance Class Assurance Family AL A AL B AL C Configuration Management ACM_CAP CM Capabilities 1 1 ACM_SCP CM Scope 2 Delivery and Operation ADO_DEL Delivery 1 2 ADO_IGS Installation, Generation and Start-up 1 1 1 ADV_FSP Functional Specification 1 2 2 ADV_HLD High-Level Design 1 2 2 Development ADV_IMP Implementation Representation 1 ADV_LLD Low-Level Design 1 ADV_RCR Representation Correspondence 1 Guidance Documents AGD_ADM Administrator Guidance 1 1 1 AGD_USR User Guidance 1 1 1 Life Cycle Support ALC_FLR Flaw Remediation 1 2 2 ATE_COV Coverage 1 1 1 ATE_DPT Depth 1 1 1 Tests ATE_FUN Functional Tests 1 1 1 ATE_IND Independent Testing 2 2 3 Vulnerability Assessment AVA_VLA Vulnerability Analysis 1 1 Table 2: Assurance Levels for KRS Functions KRS Function Security Level 1 Security Level 2 KRA AL B AL C Key Recovery Requestor AL B AL C KRI Generation AL A AL B KRI Delivery AL A AL B KRI Acquisition AL A AL B KRI Validation AL A AL B 4.1 Configuration Management Configuration management (CM) is an aspect of establishing that the functional requirements and specifications are realized in the implementation. CM meets these objectives by requiring discipline and control in the processes of refinement and modification of the product. CM systems are put in place to ensure the integrity of the configuration items that they control, by providing a method of tracking these configuration items, and by ensuring that only authorized users are capable of changing the items. 4.1.1 Configuration Management ACM_CAP - CM Capabilities Objectives The capabilities of the CM system address the likelihood that accidental or unauthorized modifications of the configuration items will occur. The CM system should ensure the integrity of the product from the early design stages through all subsequent maintenance efforts. The objectives of this assurance requirement include the following: 1. ensuring that the product is correct and complete before it is sent to the consumer; and 2. ensuring that no configuration items are missed during evaluation. Clear identification of the product is required to determine those items under evaluation that are subject to the criteria requirements. Application notes There is a requirement that a configuration list be provided. The configuration list contains all configuration items which are maintained by the CM system. 4.1.1.1 ACM_CAP.1 Minimal Support Developer action elements: (Req. 178) ACM_CAP.1.1D: The developer shall use a CM system. (Req. 179) ACM_CAP.1.2D: The developer shall provide CM documentation. Content and presentation of evidence elements: (Req. 180) ACM_CAP.1.1C: The CM documentation shall include a configuration list. (Req. 181) ACM_CAP.1.2C: The configuration list shall describe the configuration items that comprise the product. (Req. 182) ACM_CAP.1.3C: The CM documentation shall describe the method used to uniquely identify the product configuration items. Evaluator action elements: (Req. 183) ACM_CAP.1.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. 4.1.2 Configuration Management ACM_SCP - CM Scope Objectives The objective is to ensure that all necessary configuration items are tracked by the CM system. This helps to ensure that the integrity of these configuration items is protected through the capabilities of the CM system. The objectives of this assurance requirement include the following: 1. ensuring that the implementation representation (i.e., code) is tracked; and 2. ensuring that all necessary documentation, including problem reports, are tracked during development and operation. A CM system can control changes only to those items that have been placed under CM. The implementation representation, design, tests, user and administrator documentation, security flaws, and CM documentation should be placed under CM. The ability to track security flaws under CM ensures that security flaw reports are not lost or forgotten, and allows a developer to track security flaws to their resolution. Application notes There is a requirement that the implementation representation be tracked by the CM system. The implementation representation refers to all hardware, software, and firmware that comprise the physical product. In the case of a software-only product, the implementation representation may consist solely of source and object code, but in other cases, the implementation representation may refer to a combination of software, hardware, and firmware. There is a requirement that security flaws be tracked by the CM system. This requires that information regarding previous security flaws and their resolution be maintained, as well as details regarding current security flaws. 4.1.2.1 ACM_SCP.2 Problem Tracking CM Coverage Developer action elements: (Req. 184) ACM_SCP.2.1D: The developer shall provide CM documentation. Content and presentation of evidence elements: (Req. 185) ACM_SCP.2.1C: As a minimum, the following shall be tracked by the CM system: the implementation representation, design documentation, test documentation, user documentation, administrator documentation, CM documentation, and security flaws. (Req. 186) ACM_SCP.2.2C: The CM documentation shall describe how configuration items are tracked by the CM system. Evaluator action elements: (Req. 187) ACM_SCP.2.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. 4.2 Delivery and Operation 4.2.1 Delivery and Operation ADO_DEL - Delivery Objectives The requirements for delivery call for system control and distribution facilities and procedures that provide assurance that the recipient receives the product that the sender intended to send, without any modifications. For a valid delivery, what is received must correspond precisely to the master copy, thus avoiding any tampering with the actual version, or substitution of a false version. Application notes This assurance requirement should be applied to sensitive components whose modification can compromise security. 4.2.1.1 ADO_DEL.1 Delivery Procedures Developer action elements: (Req. 188) ADO_DEL.1.1D: The developer shall provide documentation about the procedures for the delivery of the product or parts of the product to the user. (Req. 189) ADO_DEL.1.2D: The developer shall use the delivery procedures. [NOTE: IS THIS TESTABLE?] Content and presentation of evidence elements: (Req. 190) ADO_DEL.1.1C: The delivery documentation shall describe the procedures to be employed when distributing versions of the product to a user's site. Evaluator action elements: (Req. 191) ADO_DEL.1.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. 4.2.1.2 ADO_DEL.2 Detection of Modification Developer action elements: (Req. 192) ADO_DEL.2.1D: The developer shall provide documentation about the procedures for the delivery of the product or parts of the product to the user. (Req. 193) ADO_DEL.2.2D: The developer shall use the delivery procedures. [NOTE: IS THIS TESTABLE?] Content and presentation of evidence elements: (Req. 194) ADO_DEL.2.1C: The delivery documentation shall describe the procedures to be employed when distributing versions of the product to a user's site. (Req. 195) ADO_DEL.2.2C: The delivery documentation shall state how the procedures are to be employed to detect modifications. (Req. 196) ADO_DEL.2.3C: The delivery documentation shall describe how the various procedures and technical measures provide for the detection of modifications, or any discrepancy between the developer's master copy and the version received at the user site. (Req. 197) ADO_DEL.2.4C: The delivery documentation shall describe how the various procedures allow the detection of attempted masquerading even in cases in which the developer has sent nothing to the user's site. Evaluator action elements: (Req. 198) ADO_DEL.2.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. 4.2.2 Delivery and Operation ADO_IGS - Installation, Generation, and Start-up Objectives Installation, generation, and start-up procedures are useful for ensuring that the product has been installed, generated, and started in a secure manner as intended by the developer. Application notes The generation requirements are applicable only to the products that provide the ability to generate an operational product from source or object code. The installation, generation, and start-up procedures may exist as a separate document, but would typically be grouped with other administrative guidance. 4.2.2.1 ADO_IGS.1 Installation, Generation, and Start-up Procedures Developer action elements: (Req. 199) ADO_IGS.1.1D: The developer shall document procedures to be used for the secure installation, generation, and start-up of the product. Content and presentation of evidence elements: (Req. 200) ADO_IGS.1.1C: The documentation shall describe the steps necessary for secure installation, generation, and start-up of the product. Evaluator action elements: (Req. 201) ADO_IGS.1.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. 4.3 Development 4.3.1 Development ADV_FSP - Functional Specification Objectives The functional specification is a high-level description of the user-visible interface and behavior of the product. It is a refinement of the statement of functional requirements for the product. The functional specification must show that all defined functional requirements are addressed, and that the security policy is enforced by the product. Application notes In addition to the content indicated in the following requirements, the functional specification shall also include any additional specific detail specified by the documentation notes in the related functional components. For example, the functional specification shall contain the specification of the interaction (protocol) among various product components. The developer must provide evidence that the product is completely represented by the functional specification. While a functional specification for the entire product would allow an evaluator to determine the product boundary, it is not necessary to require the specification of the boundary when other evidence could be provided to demonstrate the product boundary. The evaluator of the product is expected to make determinations regarding the relevance of the functional specification to the functional requirements. In the course of the functional specification evaluation, there are essentially three types of evaluator determination: specific functional requirements are met and no further work (e.g., with a less abstract representation of the product) is necessary; specific functional requirements are violated and the product fails to meet its requirements; and specific functional requirements have not been addressed and further analysis (of another product representation) is necessary. Whenever additional analysis is necessary, the evaluator is expected to carry that information forward to the analysis of other product representations. If requirements are not addressed after the analysis of the last provided product representation, this also represents a failure of the product evaluation. In all cases, it is important that the evaluator evaluate the product as a unit since, in many cases, the security functions must cooperate to meet specific functional requirements, and each security function must not interfere with the operation of any other security function. An informal security policy model can be a representation of the security policy in any notation, including a series of statements in the English Language. 4.3.1.1 ADV_FSP.1 Functional Specification and Security Policy Developer action elements: (Req. 202) ADV_FSP.1.1D: The developer shall provide a functional specification. (Req. 203) ADV_FSP.1.2D: The developer shall provide a product security policy. Content and presentation of evidence elements: (Req. 204) ADV_FSP.1.1C: The functional specification shall describe the product using an informal style. (Req. 205) ADV_FSP.1.2C: The functional specification shall include an informal presentation of syntax and semantics of all external product interfaces. (Req. 206) ADV_FSP.1.3C: The functional specification shall include evidence that demonstrates that the product is completely represented. Evaluator action elements: (Req. 207) ADV_FSP.1.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. (Req. 208) ADV_FSP.1.2E: The evaluator shall determine that the functional specification is consistent with the product security policy. (Req. 209) ADV_FSP.1.3E: The evaluator shall determine if the functional requirements are addressed by the representation of the product, i.e., the functional specification. 4.3.1.2 ADV_FSP.2 Functional Specification, Security Policy and Informal Security Policy Model Developer action elements: (Req. 210) ADV_FSP.2.1D: The developer shall provide a functional specification. (Req. 211) ADV_FSP.2.2D: The developer shall provide a product security policy. (Req. 212) ADV_FSP.2.3D: The developer shall provide an informal security policy model. (Req. 213) ADV_FSP.2.4D: The developer shall provide a demonstration of the correspondence between the informal security policy model and the functional specification. Content and presentation of evidence elements: (Req. 214) ADV_FSP.2.1C: The functional specification shall describe the product using an informal style. (Req. 215) ADV_FSP.2.2C: The functional specification shall include an informal presentation of the syntax and semantics of all external product interfaces. (Req. 216) ADV_FSP.2.3C: The functional specification shall include evidence that demonstrates that the product is completely represented. (Req. 217) ADV_FSP.2.4C: The demonstration of correspondence between the informal security policy model and the functional specification shall describe how the functional specification satisfies the informal security policy model. (Req. 218) ADV_FSP.2.5C: The demonstration of correspondence between the informal security policy model and the functional specification shall show that there are no security functions in the functional specification that conflict with the informal security policy model. (Req. 219) ADV_FSP.2.6C: The informal security policy model shall describe the rules and characteristics of all policies of the product that can be modeled. (Req. 220) ADV_FSP.2.7C: The informal security policy model shall include a rationale that demonstrates that policies that are modeled are satisfied by the informal security policy model. (Req. 221) ADV_FSP.2.8C: The informal security policy model shall justify that all policies that can be modeled are represented in the informal security policy model. Evaluator action elements: (Req. 222) ADV_FSP.2.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. (Req. 223) ADV_FSP.2.2E: The evaluator shall determine that the functional specification is consistent with the product security policy. (Req. 224) ADV_FSP.2.3E: The evaluator shall determine if the functional requirements are addressed by the representation of the product, i.e., the functional specification. 4.3.2 Development ADV_HLD - High-Level Design Objectives The high-level design of a product provides a description of the product in terms of major structural units (i.e., modules) and relates these units to the functions that they contain. The high-level design provides assurance that the product provides an architecture appropriate to implement the claimed functional requirements. The high-level design refines the functional specification into modules. For each module of the product, the high-level design describes its purpose and function and identifies the security functions enforced by the module. The interrelationships of all modules are also defined in the high-level design. These interrelationships will be represented as external interfaces for data flow, control flow, etc., as appropriate. Application notes In addition to the content indicated in the following requirements, (Req. 225) The high-level design shall also include any additional specific detail specified by the documentation notes in the related functional components. The developer is expected to describe the design of the product in terms of modules. The term ``module'' is used here to express the idea of decomposing the product into a relatively small number of parts. While the developer is not required to actually have ``modules'', the developer is expected to represent a similar level of decomposition. For example, a design may be similarly decomposed using ``layers'', ``domains'', or ``servers''. The evaluator of the product is expected to make determinations regarding the functional requirements in the product relevant to the high-level design. In the course of the high-level design evaluation, there are essentially three types of evaluator determination: specific functional requirements are met and no further work (e.g., with a less abstract representation of the product) is necessary; specific functional requirements are violated and the product fails to meet its requirements; and specific functional requirements have not been addressed and further analysis (of another product representation) is necessary. Whenever more analysis is necessary, the evaluator is expected to carry that information forward to the analysis of other product representations. If requirements are not addressed after the analysis of the last provided product representation, this also represents a failure of the product evaluation. In all cases, it is important that the evaluator evaluate the product as a unit since in many cases the security functions must cooperate to meet specific functional requirements and also each security function must not interfere with the operation of any other security function. The term ``security functionality'' is used to represent operations that a module performs that have some effect on the security functions implemented by the product. This distinction is made because modules do not necessarily relate to specific security functions. While a given module may correspond directly to a security function, or even multiple security functions, it is also possible that many modules must be combined to implement a single security function. The term ``security policy enforcing modules'' refers to a module that contributes to the enforcement of the security policy. 4.3.2.1 ADV_HLD.1 Descriptive High-Level Design Developer action elements : (Req. 226) ADV_HLD.1.1D: The developer shall provide the high-level design of the product. Content and presentation of evidence elements: (Req. 227) ADV_HLD.1.1C: The presentation of the high-level design shall be informal. (Req. 228) ADV_HLD.1.2C: The high-level design shall describe the structure of the product in terms of modules. (Req. 229) ADV_HLD.1.3C: The high-level design shall describe the security functionality provided by each module of the product. (Req. 230) ADV_HLD.1.4C: The high-level design shall identify the interfaces of the modules of the product. (Req. 231) ADV_HLD.1.5C: The high-level design shall identify any underlying hardware, firmware, and/or software required by the product with a presentation of the functions provided by the supporting protection mechanisms implemented in that hardware, firmware, or software. Evaluator action elements: (Req. 232) ADV_HLD.1.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. (Req. 233) ADV_HLD.1.2E: The evaluator shall determine if the functional requirements in the product are addressed by the design. 4.3.2.2 ADV_HLD.2 Security Enforcing High-Level Design Developer action elements : (Req. 234) ADV_HLD.2.1D: The developer shall provide the high-level design of the product. Content and presentation of evidence elements: (Req. 235) ADV_HLD.2.1C: The presentation of the high-level design shall be informal. (Req. 236) ADV_HLD.2.2C: The high-level design shall describe the structure of the product in terms of modules. (Req. 237) ADV_HLD.2.3C: The high-level design shall describe the security functionality provided by each module of the product. (Req. 238) ADV_HLD.2.4C: The high-level design shall identify the interfaces of the modules of the product. (Req. 239) ADV_HLD.2.5C: The high-level design shall identify any underlying hardware, firmware, and/or software required by the product with a presentation of the functions provided by the supporting protection mechanisms implemented in that hardware, firmware, or software. (Req. 240) ADV_HLD.2.6C: The high-level design shall describe the separation of the product into security policy enforcing modules and other modules. Evaluator action elements: (Req. 241) ADV_HLD.2.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. (Req. 242) ADV_HLD.2.2E: The evaluator shall determine if the functional requirements in the product are addressed by the design. 4.3.3 Development ADV_IMP - Implementation Representation Objectives The description of the implementation in the form of source code, firmware, hardware drawings, etc. captures the detailed internal workings of the product in support of analysis. Application notes The implementation representation is used to express the notion of the least abstract representation of the product, specifically the one that is used to create the product itself without further design refinement. Source code which is then compiled or a hardware drawing which is used to build the actual hardware are examples of parts of an implementation representation. The evaluator of the product is expected to make determinations regarding the functional requirements in the security target relevant to the implementation. In the course of the implementation evaluation, there are essentially three types of evaluator determination: specific functional requirements are met and no further work (e.g., with a more abstract representation of the product) is necessary; specific functional requirements are violated and the product fails to meet its requirements; and specific functional requirements have not been addressed and further analysis is necessary. However, since the implementation is the least abstract representation it is likely that further analysis cannot be performed unless the product representations have not been evaluated in the usual order (i.e., most abstract to least abstract). If requirements are not addressed after the analysis of all product representations, this represents a failure of the product evaluation. Note that this more comprehensive failure determination requirement is realized in the Representation correspondence (ADV_RCR) family. In all cases, it is important that the evaluator evaluates the product as a unit since, in many cases, the security functions must cooperate to meet specific functional requirements and each security function must not interfere with the operation of any other security function. It is expected that evaluators will use the implementation to directly support other evaluation activities (e.g., vulnerability analysis, test coverage analysis). 4.3.3.1 ADV_IMP.1 Subset of the Implementation Application notes The implementation representation needs to be provided for the security relevant functions of the product. Any hardware, software, and/or firmware that does not contribute to the security need not be provided, analyzed, or tested. However, an explanation must be provided, and the evaluator must agree that the excluded items are not security relevant. Developer action elements: (Req. 243) ADV_IMP.1.1D: The developer shall provide the implementation representations for a selected subset of the product. Content and presentation of evidence elements: (Req. 244) ADV_IMP.1.1C: The implementation representations shall unambiguously define the product to a level of detail such that it can be generated without further design decisions. Evaluator action elements: (Req. 245) ADV_IMP.1.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. (Req. 246) ADV_IMP.1.2E: The evaluator shall determine if the KRS functional requirements are addressed by the representation of the product. [NOTE: DOES THIS MEAN THAT THE FUNCTIONAL REQUIREMENTS NEED TO BE IDENTIFIED AS SUCH SO THAT THERE IS NO CONFUSION WITH SECURITY OR OPERATIONAL REQUIREMENTS?] 4.3.4 Development ADV_LLD - Low-Level Design Objectives The low-level design of a product provides a description of the internal workings of the product in terms of modules and their interrelationships and dependencies. The low-level design provides assurance that the modules have been correctly and effectively refined. For each module of the product, the low-level design describes its purpose, function, interfaces, dependencies, and the implementation of any security policy enforcing functions. Application notes In addition to the content indicated in the following requirements, the low- level design shall also include any additional specific detail specified by the documentation notes in the related functional components. The evaluator of the product is expected to make determinations regarding the functional requirements relevant to the low-level design. In the course of the low-level design evaluation, there are essentially three types of evaluator determination: specific functional requirements are met and no further work (e.g., with a less abstract representation of the product) is necessary; specific functional requirements are violated and the product fails to meet its requirements; and specific functional requirements have not been addressed and further analysis (of another product representation) is necessary. Whenever more analysis is necessary, the evaluator is expected to carry that information forward to the analysis of other product representations. If requirements are not addressed after the analysis of the last provided product representation, this also represents a failure of the product evaluation. Note that this more comprehensive failure determination requirement is realized in the Representation correspondence (ADV_RCR) family. In all cases, it is important that the evaluator evaluates the product as a unit since, in many cases, the security functions must cooperate to meet specific functional requirements, and each security function must not interfere with the operation of any other security function. 4.3.4.1 ADV_LLD.1 Descriptive Low-Level Design Application notes Only representations for modules in the product need to be provided. Developer action elements: (Req. 247) ADV_LLD.1.1D: The developer shall provide the low-level design of the product. Content and presentation of evidence elements: (Req. 248) ADV_LLD.1.1C: The presentation of the low-level design shall be informal. (Req. 249) ADV_LLD.1.2C: The low-level design shall describe the product in terms of modules. (Req. 250) ADV_LLD.1.3C: The low-level design shall describe the purpose of each module. (Req. 251) ADV_LLD.1.4C: The low-level design shall define the interrelationships between the modules in terms of provided functionality and dependencies on other modules. (Req. 252) ADV_LLD.1.5C: The low-level design shall describe the implementation of all security policy enforcing functions. (Req. 253) ADV_LLD.1.6C: The low-level design shall describe the interfaces of each module in terms of their syntax and semantics. (Req. 254) ADV_LLD.1.7C: The low-level design shall provide a demonstration that the product is completely represented. (Req. 255) ADV_LLD.1.8C: The low-level design shall identify the interfaces of the modules of the product which are visible at the external interface of the product. Evaluator action elements: (Req. 256) ADV_LLD.1.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. (Req. 257) ADV_LLD.1.2E: The evaluator shall determine if the functional requirements in the KRS are addressed by the representation of the product. 4.3.5 Development ADV_RCR - Representation Correspondence Objectives The correspondence between the various representations (i.e. functional requirements expressed in the KRS, functional specification, high-level design, low-level design, implementation) addresses the correct and complete instantiation of the requirements to the least abstract representation provided. This conclusion is achieved by step-wise refinement and the cumulative results of correspondence determinations between all adjacent abstractions of representation. Application notes The developer must demonstrate to the evaluator that the most detailed, or least abstract, representation of the product is an accurate, consistent, and complete instantiation of the functions expressed as functional requirements in this standard. This is accomplished by showing correspondence between adjacent representations at a commensurate level of rigor. The evaluator must analyze each demonstration of correspondence between abstractions, as well as the results of the analysis of each product representation, and then make a determination as to whether the functional requirements in this standard have been satisfied. This family of requirements is not intended to address correspondence relating to the security policy model. Rather, it is intended to address correspondence between the requirements in this standard as well as the product, functional specification, high-level design, low-level design, and implementation representation. 4.3.5.1 ADV_RCR.1 Informal Correspondence Demonstration Developer action elements: (Req. 258) ADV_RCR.1.1D: The developer shall provide evidence that the least abstract product representation provided is an accurate, consistent, and complete instantiation of the functional requirements expressed in this standard. Content and presentation of evidence elements: (Req. 259) ADV_RCR.1.1C: For each adjacent pair of product representations, the evidence shall demonstrate that all parts of the more abstract representation are refined in the less abstract representation. (Req. 260) ADV_RCR.1.2C: For each adjacent pair of product representations, the demonstration of correspondence between the representations may be informal. Evaluator action elements: (Req. 261) ADV_RCR.1.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. (Req. 262) ADV_RCR.1.2E: The evaluator shall analyze the correspondence between the functional requirements expressed in this standard and the least abstract representation provided by the developer in order to ensure accuracy, consistency, and completeness. 4.4 Guidance Documents 4.4.1 Guidance Documents AGD_ADM Administrator Guidance Objectives Administrator guidance refers to written material that is intended to be used by those persons responsible for configuring, maintaining, and administering the product in a correct manner for maximum security. Because the secure operation of the product is dependent upon the correct performance of the product, persons responsible for performing these functions are trusted by the product. Administrator guidance is intended to help administrators understand the security functions provided by the product, including both those functions that require the administrator to perform security-critical actions and those functions that provide security-critical information. Application notes The requirements AGD_ADM.1.2C and AGD_ADM.1.11C encompass the aspect that any warnings to the users of a product with regard to the product security environment and the security objectives described in this standard are appropriately covered in the administrator guidance. Those topics that are relevant to administrator guidance for the understanding and proper application of the security functions should be considered for inclusion in the administrator guidance requirements. An example of an administrator guidance document is a reference manual. 4.4.1.1 AGD_ADM.1 Administrator Guidance Developer action elements: (Req. 263) AGD_ADM.1.1D: The developer shall provide administrator guidance addressed to system administrative personnel. Content and presentation of evidence elements: (Req. 264) AGD_ADM.1.1C: The administrator guidance shall describe how to administer the product in a secure manner. (Req. 265) AGD_ADM.1.2C: The administrator guidance shall contain warnings about functions and privileges that should be controlled in a secure processing environment. (Req. 266) AGD_ADM.1.3C: The administrator guidance shall contain guidelines on the consistent and effective use of the security functions within the product. (Req. 267) AGD_ADM.1.4C: The administrator guidance shall describe the difference between two types of functions: those which allow an administrator to control security parameters, and those which allow the administrator to obtain information only. (Req. 268) AGD_ADM.1.5C: The administrator guidance shall describe all security parameters under the administrator's control. (Req. 269) AGD_ADM.1.6C: The administrator guidance shall describe each type of security-relevant event relative to the administrative functions that need to be performed, including changing the security characteristics of entities under the control of the product. (Req. 270) AGD_ADM.1.7C: The administrator guidance shall contain guidelines on how the security functions interact. (Req. 271) AGD_ADM.1.8C: The administrator guidance shall contain instructions regarding how to configure the product. (Req. 272) AGD_ADM.1.9C: The administrator guidance shall describe all configuration options that may be used during the secure installation of the product. (Req. 273) AGD_ADM.1.10C: The administrator guidance shall describe details, sufficient for the use of procedures relevant to the administration of security. (Req. 274) AGD_ADM.1.11C: The administrator guidance shall be consistent with all other documents supplied for evaluation. Evaluator action elements: (Req. 275) AGD_ADM.1.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. (Req. 276) AGD_ADM.1.2E: The evaluator shall confirm that the installation procedures result in a secure configuration. 4.4.2 Guidance Documents AGD_USR - User Guidance Objectives User guidance refers to written material that is intended to be used by non- administrative (human) users of the product. User guidance describes the security functions provided by the product and provides instructions and guidelines, including warnings, for its secure use. The user guidance provides a basis for assumptions about the use of the product and a measure of confidence that non-malicious users and application providers will understand the secure operation of the product and will use it as intended. Application notes The requirement AGD_USR.1.3.C and AGD_USR.1.5C encompass the aspect that any warnings to the users of a product with regard to the product security environment and the security objectives described in this standard are appropriately covered in the user guidance. Those topics in this standard that are relevant to user guidance aimed at the understanding and proper use of the security functions should be considered for inclusion in the user guidance requirements. Examples of user guidance are reference manuals, user guides, and on-line help. 4.4.2.1 AGD_USR.1 User Guidance Developer action elements: (Req. 277) AGD_USR.1.1D: The developer shall provide user guidance. Content and presentation of evidence elements: (Req. 278) AGD_USR.1.1C: The user guidance shall describe the product and interfaces available to the user. (Req. 279) AGD_USR.1.2C: The user guidance shall contain guidelines on the use of security functions provided by the product. (Req. 280) AGD_USR.1.3C: The user guidance shall contain warnings about functions and privileges that should be controlled in a secure processing environment. (Req. 281) AGD_USR.1.4C: The user guidance shall describe the interaction between user-visible security functions. (Req. 282) AGD_USR.1.5C: The user guidance shall be consistent with all other documentation delivered for evaluation. Evaluator action elements: (Req. 283) AGD_USR.1.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. 4.5 Life Cycle Support 4.5.1 Life Cycle Support ALC_FLR - Flaw Remediation Objectives Flaw remediation requires that discovered flaws be tracked and corrected by the developer. Although future compliance with flaw remediation procedures cannot be determined at the time of the product evaluation, it is possible to evaluate the policies and procedures that a developer has in place to track and correct flaws, and to distribute the flaw information and corrections. Application notes None 4.5.1.1 ALC_FLR.1 Basic Flaw Remediation Developer action elements: (Req. 284) ALC_FLR.1.1D: The developer shall document the flaw remediation procedures. Content and presentation of evidence elements: (Req. 285) ALC_FLR.1.1C: The flaw remediation procedures documentation shall describe the procedures used to track all reported security flaws in each release of the product. (Req. 286) ALC_FLR.1.2C: The flaw remediation procedures shall require that a description of the nature and effect of each security flaw be provided, as well as the status of finding a correction to that flaw. (Req. 287) ALC_FLR.1.3C: The flaw remediation procedures shall require that corrective actions be identified for each of the security flaws. (Req. 288) ALC_FLR.1.4C: The flaw remediation procedures documentation shall describe the methods used to provide flaw information and corrections to product users. Evaluator action elements: (Req. 289) ALC_FLR.1.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. 4.5.1.2 ALC_FLR.2 Flaw Reporting Procedures Developer action elements: (Req. 290) ALC_FLR.2.1D: The developer shall document the flaw remediation procedures. (Req. 291) ALC_FLR.2.2D: The developer shall establish a procedure for accepting and acting upon user reports of security flaws and requests for corrections to those flaws. Content and presentation of evidence elements: (Req. 292) ALC_FLR.2.1C: The flaw remediation procedures documentation shall describe the procedures used to track all reported security flaws in each release of the product. (Req. 293) ALC_FLR.2.2C: The flaw remediation procedures shall require that a description of the nature and effect of each security flaw be provided, as well as the status of finding a correction to that flaw. (Req. 294) ALC_FLR.2.3C: The flaw remediation procedures shall require that corrective actions be identified for each of the security flaws. (Req. 295) ALC_FLR.2.4C: The flaw remediation procedures documentation shall describe the methods used to provide flaw information and corrections to product users. (Req. 296) ALC_FLR.2.5C: The procedures for processing reported security flaws shall ensure that any reported flaws are corrected and the correction issued to product users. Evaluator action elements: (Req. 297) ALC_FLR.2.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. 4.6 Tests 4.6.1 Tests ATE_COV - Coverage Objectives This family addresses those aspects of testing that deal with completeness of testing. That is, it addresses the extent to which the product security functions are tested, whether or not the testing is sufficiently extensive to demonstrate that the product operates as specified, and whether or not the order in which testing proceeds correctly accounts for functional dependencies between the portions of the product being tested. Application notes The specific documentation required by the coverage components will be determined, in most cases, by the documentation stipulated in the level of ATE_FUN that is specified. 4.6.1.1 ATE_COV.1 Complete Coverage - Informal Objectives In this component, the objective is that testing completely address the security functions. Application notes While the testing objective is to completely cover the product, there is no more than an informal explanation to support this assertion. Developer action elements: (Req. 298) ATE_COV.1.1D: The developer shall provide an analysis of the test coverage. Content and presentation of evidence elements: (Req. 299) ATE_COV.1.1C: The analysis of the test coverage shall demonstrate that the tests identified in the test documentation cover the product. Evaluator action elements: (Req. 300) ATE_COV.1.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. 4.6.2 Tests ATE_DPT - Depth Objectives The components in this family deal with the level of detail to which the product is tested. The testing of security functions is based upon an increasing depth of information derived from the analysis of the representations. The objective is to counter the risk of missing an error in the development of the product. Additionally, the components of this family, especially as testing is more concerned with the internals of the product, are more likely to discover any malicious code that has been inserted. Application notes The specific amount and type of documentation and evidence will, in general, be determined by that required by the level of ATE_FUN selected. 4.6.2.1 ATE_DPT.1 Testing - Functional Specification Objectives The functional specification of a product provides a high level description of the external workings of the product. Testing at the level of the functional specification, in order to demonstrate the presence of any flaws, provides assurance that the product functional specification has been correctly realized. Application notes The functional specification representation is used to express the notion of the most abstract representation of the product. Developer action elements: (Req. 301) ATE_DPT.1.1D: The developer shall provide the analysis of the depth of testing. Content and presentation of evidence elements: (Req. 302) ATE_DPT.1.1C: The depth analysis shall demonstrate that the tests identified in the test documentation are sufficient to demonstrate that the product operates in accordance with the functional specification of the product. Evaluator action elements: (Req. 303) ATE_DPT.1.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. 4.6.3 Tests ATE_FUN - Functional Tests Objectives Functional testing establishes that the product exhibits the properties necessary to satisfy the functional requirements of this standard. Functional testing provides assurance that the product satisfies at least the security functional requirements, although it cannot establish that the product does no more than what was specified. The ``Functional tests'' family is focused on the type and amount of documentation or support tools required, and what is to be demonstrated through testing. This family contributes to providing assurance that the likelihood of undiscovered flaws is relatively small. Application notes Procedures for performing tests are expected to provide instructions for using test programs and test suites, including the test environment, test conditions, test data parameters and values. The test procedures should also show how the test results are derived from the test inputs. The developer shall eliminate all security relevant flaws discovered during testing. The developer shall test the product to determine that no new security relevant flaws have been introduced as a result of eliminating discovered security relevant flaws. Tests shall include an examination of procedures and documents that assist in implementing the product security policy. 4.6.3.1 ATE_FUN.1 Functional Testing Objectives The objective is for the developer to demonstrate that all security functions perform as specified. The developer is required to perform testing and to provide test documentation. Developer action elements: (Req. 304) ATE_FUN.1.1D: The developer shall test the product and document the results. (Req. 305) ATE_FUN.1.2D: The developer shall provide test documentation. Content and presentation of evidence elements: (Req. 306) ATE_FUN.1.1C: The test documentation shall consist of test plans, test procedure descriptions, and test results. (Req. 307) ATE_FUN.1.2C: The test plans shall identify the security functions to be tested and describe the goal of the tests to be performed. (Req. 308) ATE_FUN.1.3C: The test procedure descriptions shall identify the tests to be performed and describe the scenarios for testing each security function. (Req. 309) ATE_FUN.1.4C: The test results in the test documentation shall show the expected results of each test. (Req. 310) ATE_FUN.1.5C: The test results from the execution of the tests by the developer shall demonstrate that each security function operates as specified. Evaluator action elements: (Req. 311) ATE_FUN.1.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. 4.6.4 Tests ATE_IND - Independent Testing Objectives The objective is to demonstrate that the security functions perform as specified. An additional objective is to counter the risk of an incorrect assessment of the test outcomes on the part of the developer which results in the incorrect implementation of the specifications, or overlooks code that is non-compliant with the specifications. Application notes The testing specified in this family can be performed by a party other than the evaluator (e.g., an independent laboratory, an objective consumer organization). This family deals with the degree to which there is independent functional testing of the product. Independent functional testing may take the form of repeating the developer's functional tests in whole or in part. It may also take the form of the augmentation of the developer's functional tests, either to extend the scope or the depth of the developer's tests. Independent testing shall be performed by an independent third party certified and accredited by the Government. The Government will supply some tests to validate compliance and conformance. Examples include: cryptographic algorithms and cryptographic protocols. The evaluator (which happens to be the independent third party) shall execute these government supplied tests in addition to the tests provided by the developer, and tests developed by the evaluator. 4.6.4.1 ATE_IND.2 Independent Testing - Sample Objectives The objective is to demonstrate that the security functions perform as specified. In this component, the objective is to select and repeat a sample of the developer testing. Application notes The suitability of the product for testing is based on access to the product, and the supporting documentation and information required to run tests. The need for documentation is supported by other assurance families (e.g., ATE_FUN) Additionally, the suitability of the product for testing may be based on other considerations (e.g., the version of the product submitted by the developer is not the final version). The developer is required to perform testing and to provide test documentation and test results. This is addressed by the ATE_FUN family. Testing may be selective and is based upon all available documentation. Developer action elements: (Req. 312) ATE_IND.2.1D: The developer shall provide the product for testing. Content and presentation of evidence elements: (Req. 313) ATE_IND.2.1C: The product shall be suitable for testing. Evaluator action elements: (Req. 314) ATE_IND.2.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. (Req. 315) ATE_IND.2.2E: The evaluator shall test the product to confirm that the product operates as specified. (Req. 316) ATE_IND.2.3E: The evaluator shall execute a sample of tests in the test documentation to verify the developer test results. 4.6.4.2 ATE_IND.3 Independent Testing - Complete Objectives The objective is to demonstrate that the security functions perform as specified. In this component, the objective is to repeat the developer testing. Application notes The suitability of the product for testing is based on access to the product, and the supporting documentation and information required to run tests. The need for documentation is supported by other assurance families (e.g., ATE_FUN) Additionally, the suitability of the product for testing may be based on other considerations (e.g., the version of the product submitted by the developer is not the final version). The developer is required to perform testing and to provide test documentation and test results. This is addressed by the ATE_FUN family. Developer action elements: (Req. 317) ATE_IND.3.1D: The developer shall provide the product for testing. Content and presentation of evidence elements: (Req. 318) ATE_IND.3.1C: The product shall be suitable for testing. Evaluator action elements: (Req. 319) ATE_IND.3.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. (Req. 320) ATE_IND.3.2E: The evaluator shall test the product to confirm that the product operates as specified. (Req. 321) ATE_IND.3.3E: The evaluator shall execute all tests in the test documentation to verify the developer test results. 4.7 Vulnerability Assessment 4.7.1 Vulnerability Assessment AVA_VLA - Vulnerability Analysis Objectives Vulnerability analysis is an assessment to determine whether vulnerabilities could allow malicious users to violate the security policy. These vulnerabilities will be identified during the evaluation by flaw hypotheses. Vulnerability analysis deals with the threats that a malicious user will be able to discover flaws that will allow access to resources (e.g., data), allow the ability to interfere with or alter the product, or interfere with the authorized capabilities of other users. Application notes The vulnerability analysis should consider the contents of all the product deliverables for the targeted evaluation assurance level. Obvious vulnerabilities are those that allow common attacks or those that might be suggested by the product interface description. Obvious vulnerabilities are those in the public domain, details of which should be known to a developer, publicly available, or available from NIST. The evidence identifies all the product documentation upon which the search for flaws was based. 4.7.1.1 AVA_VLA.1 Developer Vulnerability Analysis Objectives A vulnerability analysis is performed by the developer to ascertain the presence of ``obvious'' security vulnerabilities. The objective is to confirm that no identified security vulnerabilities can be exploited in the intended environment for the product. Application notes Obvious vulnerabilities are those which are open to exploitations which require a minimum of understanding of the product, skill, technical sophistication, and resources. Developer action elements: (Req. 322) AVA_VLA.1.1D: The developer shall perform and document an analysis of the product deliverables searching for obvious ways in which a user can violate the security policy. (Req. 323) AVA_VLA.1.2D: The developer shall document the disposition of identified vulnerabilities. Content and presentation of evidence elements: (Req. 324) AVA_VLA.1.1C: The evidence shall show, for each vulnerability, that the vulnerability cannot be exploited in the intended environment for the product. Evaluator action elements: (Req. 325) AVA_VLA.1.1E: The evaluator shall confirm that the information provided meets all requirements for content and presentation of evidence. (Req. 326) AVA_VLA.1.2E: The evaluator shall conduct penetration testing, based on the developer vulnerability analysis, to ensure that obvious vulnerabilities have been addressed. 4.8 Excluded Assurance Requirements This section contains the Common Criteria assurance requirements that are recommended for exclusion. The ADV_INT family relates to modularity, layering, information hiding, etc. For economic reasons, this family has not been included. ALC_DVS (Developmental Security), ALC_LCD (Life Cycle Definition), and ALC_TAT (Tools and Techniques) are excluded in order to provide engineering independence for the vendors, spur commercial product development, and align assurance requirements with the commercial practices. AVA_CCA (Covert Channel Analysis), AVA_SOF (Strength of Function, e.g., work factor for cryptographic operation) are excluded since they are not particularly relevant here. AVA_CCA in non-discretionary policy environments can be implemented using procedural controls such as executing trusted software only. Cryptanalysis work factors will be provided or implied by the FIPS cryptographic algorithms. AVA_MSU (Misuse Analysis) is excluded since obvious flaws and known flaws will come under AVA_VLA (Vulnerability Analysis). Given this is a standard for SBU data, vulnerability analysis may be an overkill. Appendix A: Examples A.1 Key Recovery Function Distribution The functions of a KRS may be integrated into products in a variety of configurations in order to accommodate different user environments. In Figure 9, the KRI Generation, Delivery, Acquisition and Validation functions are provided in a single cryptographic end system product. The Requestor and KRA functions are each available as independent products. The separate Requestor System might be appropriate in an organization which prefers to centralize the key recovery process. In Figure 10, the KRI Generation and Delivery Functions are provided in one product, while the Requestor Function and KRA Function are in a separate product. This configuration may be appropriate for a storage application, where files are encrypted by a user, KRI is attached to the file and thereafter ignored unless the decryption key becomes unavailable and recovery is required. The user could then go to a special recovery system in order to recover the appropriate key. In Figure 11, the KRA function is bundled with the KRI Generation and Delivery Functions. This might be appropriate for an environment in which the KRA generates the encryption key pair, sends it off to the user and/or a CA for certification, and caches a copy of the private key for potential recovery at a later time. In Figure12, the KRI Generation, Delivery, Acquisition, Validation and Requestor Functions are provided in a single cryptographic end system. The KRA Function is a separate product. There may be an electronic connection between the end user system and the KRA in order to effect the recovery process. A.2 Multiple KRI Generation Functions Figure 13 provides an example of multiple KRI Generation Functions which are required to provide the aggregate of KRI needed to recover a target key. Suppose that System B or a trusted generation service generates an encryption key pair for System B and provides the public key to a Certificate Authority (CA) along with other information which will be useful in providing key recovery. The CA generates a certificate containing this information. System A uses this certificate along with other internally generated information to create KRI for messages to be sent to System B. In this case, System A, the CA and whoever generates System B's key pair participate in the generation of the KRI that will allow System B to recover. A.3 KRI Generation Scenarios Assume that each system has an encryption public key certificate (hereafter called an encryption certificate) that identifies the key recovery method and the identity of the KRA(s). Encryption certificates are also available for the KRAs. A.3.1 Interactive Communications A.3.1.1 Between Two Encapsulation Techniques In Figure 5, cryptographic end systems A and B are two systems that employ two different encapsulation methods for key recovery, but use a common key recovery block (KRB). A key transport method of key exchange is used (e.g., the data key is encrypted using the receiver's encryption public key). System A has a key recovery policy stating that key recovery information is not created for interactive communications. System B has a key recovery policy that states: (1) key recovery information must be created for itself for all communications when that information is not present, and (2) key recovery information must also be created for the other party whenever possible. System A creates a data key to be used for the communication session and encrypts the data key using the public encryption key of System B (obtained from System B's encryption certificate). System A sends the encrypted key as part of the normal key exchange process. System A then encrypts a message for System B, and sends the encrypted message on the communications path. When System B determines that no key recovery information is available for the message received from System A (i.e., no KRB is present), System B decrypts the encrypted data key (received as part of the key exchange process), and uses the resulting plaintext data key to create key recovery information for itself and/or its Key Recovery Agent. The KRI is placed in a KRB in accordance with its key recovery scheme. By examining System A's certificate, the identity of System A's KRA(s) can be determined, and the KRA encryption certificate(s) can be acquired. If System B can create a KRB for System A's key recovery technique and all information is available, key recovery information is created for System A and/or its Key Recovery Agent(s). System B then uses the data key to decrypt the received message. The newly created key recovery information is then attached to the next message in the communication session and sent back to System A. In subsequent messages received by System A within this interactive session, System A can recognize the presence of the KRI (perhaps perform some processing of the KRI in the KRB) and decrypt the received messages. A.3.1.2 Between Encapsulated and Key Escrow Techniques Figure 8 includes cryptographic end systems A and B that use key escrow and KRI encapsulation methods of key recovery, respectively. System B uses a KRB. A key agreement method of key exchange is used (e.g., the encryption public and private keys pairs of both parties to a communication are used along with randomly generated values to generate a shared data key at the cryptographic end systems). System A has a key recovery policy that requires that all incoming communications must have KRI available for the sender. System B has a policy stating that communications will only be conducted with other parties that employ key recovery techniques, and that KRI is always created for itself in outgoing communications. System B wants to initiate a communication session with System A. By obtaining System A's encryption certificate, System B obtains System A's public encryption key as well as determining that System A uses a key escrow method of key recovery. System B initiates a key exchange with System A to agree upon a data key, then encapsulates the data key and other KRI in a KRB for itself and its KRA. The data key is then used to encrypt the data, and the encrypted data and the KRB are sent to system A. System A (probably during the key exchange process) determines that System B uses an encapsulated method of key recovery by examining System B's encryption certificate. When the initial message is received from System B, System A is able to recognize that there is a KRB for System B. System A then proceeds to decrypt the received message. A.3.2 Store and Forward Communications A.3.2.1 Between Two Key Escrow Key Recovery Schemes In Figure 6, cryptographic end Systems A and B employ key escrow methods of key recovery. A key transport method of key exchange is used. System B has a policy stating that all outgoing email messages will be archived and recoverable (i.e., KRI must be available to recover encrypted email messages that have been archived). System A is able to recover incoming encrypted email messages if key transport is used for key exchange. System B generates a data key and encrypts the key using the encryption public key of the receiver SA) for use in the key exchange (key transport process). Even though System B uses key escrow, there is nothing yet which allows System B to recover after the outgoing message is archived. System B encrypts the data key using his own encryption public key, and places it in a KRB. System B then encrypts the message with the data key, and sends the encrypted message and System A's copy of the encrypted data key to System A. The encrypted message and the KRB are archived. System A decrypts the data key received via the key transport mechanism and decrypts the received message using that key. A.3.2.2 Between an Encapsulated Scheme and an End User System with No Key Recovery Capability In Figure 5, cryptographic end System A uses an encapsulated method of key recovery. System B has no key recovery capability. A key transport method of key exchange is in use (e.g., the data key is encrypted by the receiver's encryption public key). System A has a key recovery policy that states: (1) key recovery information must always be created for itself and/or its Key Recovery Agent, and (2) Key recovery Information is not created for anyone else. System A retains a copy of all outgoing email messages. System A sends the KRB along an alternate path from that of the encrypted messages; this allows system B to ignore key recovery information so that interoperability is possible. System A creates a data key, then creates key recovery information for itself and/or its Key Recovery Agent, and places the KRI in a KRB. The KRB is sent along the alternate communication path. The data key is encrypted by system B's encryption public key (obtained from System B's encryption certificate) and then used to encrypt an e-mail message. The encrypted key is placed in the message header (the method of key transport that is employed in this example) and sent with the encrypted message to System B. Upon receipt of the encrypted message and key exchange information , System B decrypts the data encryption key in the message header, and uses the decrypted data encryption key to decrypt the message. A.3.3 Data Storage A.3.3.1 Creation by an End User with an Encapsulated Scheme; Read Access by Anyone For data storage applications, the Encryptor and Decryptor may not be the same entity (e.g., shared files). In Figure 5, end user system A uses an encapsulated method for key recovery. System A's organization has a policy stating that key recovery information must exist for all stored data. Read only access can be granted to a list of other systems in the organization, whether or not those systems have a key recovery capability. System A creates a data key and uses the encryption public key of each system on the access list to encrypt a copy of the data key for that system (including itself). System A also encrypts the data key using the encryption public key of the organization's KRA. The data key is then used to encrypt the data. All copies of the encrypted key are placed in a file along with the encrypted data. When accessing the encrypted file, the acquiring system decrypts the appropriate copy of the encrypted data key, and uses the decrypted data key to decrypt the file. A.4 Key Recovery Scenarios A.4.1 Interactive Session Referring back to scenario A.3.1.2, when System A initially tries to participate in the key exchange process, it is discovered that the private key of the encryption public key pair is lost. System A immediately requests the recovery of its private key from the KRA using its automated ability to request key recovery. When the private key is provided, system A can continue with the key exchange process and participate in the determination of the data key to be used for the communication session. A.4.2 Store-and-Forward Communications In scenario A.3.2.1, the email message received by System A is stored in the in- box until read. Suppose that the user receives a large number of email messages before reading them. When attempting to read the encrypted messages, it is discovered that the private key of the encryption public key pair is corrupted. The user requests a recovery of the private key from the key recovery function, uses the recovered private key to decrypt the data key for each message, and then uses the data key to decrypt the associated message. A.4.3 Data Storage In scenario A.3.3.1, System A could create a file for himself (i.e., no one else is on the access list, so the data key is not encrypted for anyone else). At some later time, the user needs to retrieve the file, but has lost access to his decryption key. The data key can be recovered by sending the copy of the key which was encrypted using the KRA's encryption public key to the KRA for decryption. Appendix B: Key Recovery Block B.1 Introduction When different key recovery products that employ KRI encapsulation need to interoperate with one another, one of the major obstacles is the inability of the receiver product to recognize and validate the key recovery information received from the sender product. In order to allow the interoperability of various key recovery techniques which require the use of KRI encapsulaton, a common structure -- a Key Recovery Block (KRB) -- may be required. The KRB serves as a container2 for technique-specific key recovery information, and supports generic mechanisms to identify and validate the contained key recovery information. Various levels of validation may be performed depending on the key recovery techniques used by the sending and receiving parties, including: ( Verification of the presence of the KRB, ( Validation of the integrity of the KRB, ( Authentication of the source and validation of the integrity of the KRB [WILL THIS BE THE CASE? INFO MAY NEED TO BE ADDED], and ( Verification that the KRI can be used to recover the data key. The KRB is independent of the encryption algorithm used to protect the confidentiality of the data, and independent of the communication or storage protocol used to carry the encrypted data. B.2 KRB Fields The KRB should include the following fields of information: * The KRB version number, * The KRB length - beginning at the version number and ending at the last word/byte of the Integrity Field, * Object Identifier (OID) for the key recovery technique used to generate the KRI field. * Encrypted Data Sensitivity (EDS) Field Type: Type = 0: NONE (no EDS field is specified) * Encrypted Data Sensitivity (EDS) Field Length: Number of {words/bytes} in the EDS field. * Encrypted Data Sensitivity (EDS) Field- the sensitivity of the data recoverable by this KRB [THIS NEED WAS IDENTIFIED IN THE BUSINESS REQUIREMENTS PAPER PRODUCED BY THE KEY RECOVERY ALLIANCE - SEE SCENARIO 13, 2ND COLUMN, 4TH ITEM] . * KRI Field length - in {words,bytes}. * KRI Field (KRIF) - the KRI as specified by the indicated key recovery technique using the format employed by that technique, * Encrypted Data Locator (EDL) Field Type - identifies the method used to generate the EDL Field. Defined methods include: type = 0: NONE (no EDL field was calculated) * Encrypted Data Locator (EDL) Field Length: Number of {words/bytes} in the EDL field. * Encrypted Data Locator (EDL) Field: The value of the Encrypted Data Locator Field. This is reserved for possible future use in locating the encrypted data that may be recovered using this KRB. * Integrity Field Type: Identifies the method used to generate the Integrity Field. Defined methods include: type = 0: NONE (no integrity field was calculated) type = 1: SEMANTIC (no integrity field was calculated) type = 2: PROTOCOL (no integrity field was calculated) type = 3: CONF-HMAC-SHA-1-96 (integrity field calculated using HMAC and SHA-1 and the confidentiality key associated with the KRF - described in RFC 2104 and draft-ietf-ipsec-hmac-sha196-00.txt) type = 4: CONF-HMAC-MD5-96 (integrity field calculated using HMAC and MD5 and the confidentiality key associated with the KRF type = 5: INTEG-HMAC-SHA-1-96 (integrity field calculated using HMAC and SHA-1 and the integrity key associated with the session - described in RFC 2104 and draft-ietf-ipsec-hmac-sha196-00.txt) type = 6: INTEG-HMAC-MD5-96 (integrity field calculated using HMAC and MD5 and the integrity key associated with the session - described in RFC 2104 and draft-ietf-ipsec-hmac-md5-96-00.txt) type = 7: SIGNATURE-PKCS7 (integrity field calculated as a PKCS #7 envelope with ContentType = "signed data" - described in the PKCS #7 specification. The data content that is carried within the PKCS#7 envelope is the hash of the KRF. The hash algorithm used is the same one that is specified within the PKCS#7 Content. * Integrity Field Length: Number of {words/bytes} in the Integrity Field. The Integrity Field Length must be consistent with the Integrity Field Type: Integrity Field Type Integrity Field Length 0 0 1 0 2 0 3 5 4 4 5 5 6 4 7 Varies * Integrity Field Value: The value of the Integrity Field that is calculated over all fields of the KRB except for the Integrity Field Value itself. For Integrity Field Types 0 through 2, the Integrity Field value does not exist. For Integrity Field Types 3 and 5, the Integrity Field Value is a 20 byte hash of the KRF using HMAC and SHA-1. For Integrity Field Types 4 and 6, it is a 16 byte hash of the KRF using HMAC and MD5. For Integrity Field Type 7, the Integrity Field Value is a PKCS#7 envelope [SEE ENVELOPE STRUCTURE BELOW] whose content is a hash of the relevant fields of the KRB using the digestAlgorithmIdentifier specified within the PKCS#7 Content. [NOTE: THE FOLLOWING MAY NEED TO BE REMOVED OR EXTENSIVELY REVISED BASED ON THE THE FIPS VALIDATION REQUIREMENTS.] Further Notes on the Integrity Field: Certain key recovery products do not require any verification of the KRIF to be done at the receiving side. These products should use Integrity Field Type "NONE", indicating that KRIF verification is unnecessary. Certain other products use technique-specific validation methods for the KRIF since these may be potentially stronger than the KRIF integrity checking techniques that are supported by the KRB. Products of this class should construct KRBs with Integrity Field Type "SEMANTIC", implying that the KRIF should be validated semantically using the technique-specific algorithm. A major drawback of using semantic validation techniques is that interoperability between products using dissimilar key recovery techniques may not be supportable. Some key recovery products are based on secure communication protocols which provide integrity protection for the KRB when it is carried as an integral part of the secure association. This class of products should use Integrity Field Type "PROTOCOL", implying that the KRIF need not be checked for integrity since the carrier protocol provides integrity protection for the entire KRB. Finally, there are a class of key recovery products which require KRIF validation by the receiver who cannot rely on the carrier protocol to provide integrity protection to the KRB, and require interoperability between heterogeneous key recovery systems. This class of products should use the supported integrity checking mechanisms of the KRB by using Integrity Field Types 3 to 7. The Integrity Field should contain the value corresponding to the specified type. Certain products may like to use keyed-hash based integrity checks for the KRB. These products will generate KRBs with Integrity Field Types 3 to 6. The keyed- hash Integrity Field Types are defined for systems that use a single key for confidentiality and integrity protection, as well as systems using separate confidentiality and integrity keys. Types 3 and 4 use the confidentiality key associated with the session in generating the HMAC value, while types 5 and 6 use the integrity key associated with a session for the HMAC. A careful analysis of the cryptographic system is required when the same key does double duty as the encryption key and the HMAC key for the key recovery block. Certain products may like to use digital signature techniques to validate the integrity of the KRB. These products will generate KRBs with Integrity Field Type 7, which denotes that the Integrity Field Value is a PKCS#7 envelope that carries a digital signature over the relevant fields of the KRB. The PKCS#7 format was chosen as a vehicle for carrying the signature value since it allows the pertinent certificates (needed for signature verification) to be conveniently packaged in a well-known format. It may be noted that the Content within the PKCS#7 envelope is a hash of the relevant fields of the KRB. Thus, the actual signature carried within the PKCS#7 envelope will be calculated on the hash of the KRB, rather than the KRB itself. It may be noted that a product that generates a KRB specifies the Integrity Field Type based on its assumptions about its operating environment and its policy related to KRIF verification. Similarly, the types of KRBs that may be accepted by a receiver product are based on the receiver's assumptions about its operating environment and its policy related to KRIF verification. This proposal in no way mandates that a receiver product accept a KRB with all possible integrity Field Types; it leaves the usage and acceptability of specific Integrity Field Types to the discretion of the sending and receiving products. The KRB format can also be used very conveniently to identify the KRIF carried within it. Certain vendors may like to use the KRB format for KRIF identification purposes only, but may not want to incur the overhead of generating and verifying the integrity field. It is recommended that these vendors use Integrity Field Type "NONE". The KRB integrity field is a "robust" mechanism for verifying the integrity of the enclosed KRIF. The integrity field is not susceptible to typical man-in-the- middle attacks (MITM). Modification of the Integrity Field Type is not useful to an attacker, since the communicating peers have a security association that demands specific Integrity Field Types. Substitution of the KRIF and the corresponding Integrity Field value (for types 3 to 6) does not succeed, since the MITM does not have the session key necessary to generate a valid Integrity Field value for the bogus KRIF that was substituted. B.3 KRB Format [NEED TO DECIDE HOW DEEPLY TO SPECIFY THE KRB. SHOULD PROBABLY SPECIFY ENOUGH THAT ANY PROTOCOL USING THE KRB INFORMATION WOULD USE THE SAME STRUCTURE, ALLOWING DEVELOPERS TO DESIGN TO THE SAME FORMAT.] 31 23 15 7 Version KRB Length Key Recovery Technique OID EDS Field Type EDS Field Length Encrypted Data Sensitivity (EDS) Reserved KRIF Length Key Recovery Information Field (KRIF) EDL Field Type EDL Field Length Encrypted Data Locator (EDL) Field Integrity Field Type Integrity field length Integrity Field Value B.4 Implementation Guidance Vendors that are compliant with the common KRB format, would design their products so that their KRIF (in its proprietary format) is placed within the common KRB defined above. The appropriate information should be provided in the other fields of the KRB. When a compliant product receives a KRB with an integrity field, the product can validate the KRIF embedded within the KRB, either by using the KRB integrity field, or the technique-specific validation algorithm (if known). ---------------- [THE PRESENCE OF THE PKCS#7 STRUCTURE IS FOR INFORAMTION PURPOSES ONLY.] PKCS #7 Structure PKCS #7 contains the following fields: * Version * Digest Algorithm ID * Content * Certificates (opt.) * CRLs (Opt.) * Signer Info version, issuer & serial ID, digest algorithm ID, authenticated attribute (opt.), encrypted digest, unauthenticated attributes (opt.) Note: There may be multiple Signer Info fields Appendix C: Certificate Extensions C.1 Introduction In order to facilitate the recovery of a key in a Public Key Infrastructure (PKI), the appropriate certificates should be extended to include key recovery information. Modifications may include: ( The encryption certificate for a KRA should include: (1) a key usage bit which indicates that the public key is to be used for key recovery purposes. ASN.1 for this has two alternatives: a. Recommend adding a bit to the X.509 key usage extension: keyRecoveryAgent (9) and mark the key usage extension critical, or b. Register the following object as the key purpose object identifier and mark it critical. This seems to be the more appropriate approach rather than expecting X.509 to add a bit to the key usage extension. {joint-iso-ccitt(2) country(16) us(840) organization(1) gov(101) csor(3) pki(2) keypurpose(2) krakey(1)} (2) an identification of the key recovery technique(s) with which the public key may be used. It is not clear that this extension is required at all. Santosh suggests not having this extension at all. Note that if this extension is used, the first extension is not required since this extension implies that the public key belongs to the KRA. Thus, it is recommended that if this extension is used, the first extension (extended key usage) should be dropped. The following is the syntax for this private, critical3 extension to the certificate: kRTechnique EXTENSION ::= { SYNTAX KRTechnique IDENTIFIED BY id-extensions-kRTechnique } KRTechnique ::= SEQUENCE { technique technique.&id, parameters OPTIONAL } -- technique is an object identifier. The parameters syntax is registered when the technique OID is registered * A certificate for the entity using key recovery should include: (1) an indication that the entity has a key recovery capability, This is done by using the following private, non-critical extension keyRecoveryCapable EXTENSION ::= { SYNTAX SubjectKeyIdentifier IDENTIFIED BY id-extensions-KeyRecoveryCapable } KeyRecoveryCapable ::= BOOLEAN DEFAULT FALSE (2) identify the KRA(s), This is done using the following private, non- critical extension. Please note that if this extension is included, the first extension (key recovery capable) is not required. kR EXTENSION ::= { SYNTAX KR IDENTIFIED BY id-extensions-KR } KR ::= SEQUENCE SIZE (1...MAX) OF KRS KRS ::= SEQUENCE { technique KRTechnique SEQUENCE SIZE (1...MAX) OF AGENT } AGENT ::= SEQUENCE { agentName directoryName agentkey KeyIdentifier - OPTIONAL agentpol KRAPolicy - OPTIONAL} KRAPolicy ::= OBJECT IDENTIFIER (3) indicate the KRA certificate(s) containing the appropriate KRA public key(s). Please note that this is in the KR extension. (4) identify the key recovery technique(s) supported by the entity. Please note that this is in the extension for the key recovery (item 2 above) (5) include any key recovery technique information required. Please note that this is in optional parameters extension of the key recovery technique. CSOR REGISTERED TECHNICAL OBJECTS Prefix for CSOR-unique technical objects: {2.16.840.1.101.3} {joint-iso-ccitt(2) country(16) us(840) organization(1) gov(101) csor(3)} -- Technical Object Identifiers -- Types of information security objects id-slabel ID ::= {id-csor 1} id-pki ID ::= {id-csor 2} id-arpa ID ::= {id-csor 3} -- Certificate Policies -- {joint-iso-ccitt(2) country(16) us(840) organization(1) gov(101) csor(3) pki(2) certpolicies(1)} -- Key Purpose -- {joint-iso-ccitt(2) country(16) us(840) organization(1) gov(101) csor(3) pki(2) keypurpose(2)} id-kRAKey ID ::= {id-keypurpose 1} -- Extensions -- {joint-iso-ccitt(2) country(16) us(840) organization(1) gov(101) csor(3) pki(2) extensions(3)} id-kRTechnique ID ::= {id-extensions 1} id-kRecoveryCapable ID ::= {id-extensions 2} id-kR ID ::= {id-extensions 3} -- Key Recovery Schemes -- {joint-iso-ccitt(2) country(16) us(840) organization(1) gov(101) csor(3) pki(2) keyrecoveryschemes(4)} -- Key Recovery Policy -- {joint-iso-ccitt(2) country(16) us(840) organization(1) gov(101) csor(3) pki(2) krapol(5)} Appendix D: Interoperability Examples D.1 Introduction D.2 S/MIME The Secure MIME (S/MIME) protocol provides encryption for Internet electronic mail that uses the MIME encoding format. S/MIME defines two security wrappers: one for digital signatures and one for encryption. To encrypt and sign a message, both wrappers are applied. Both of these wrappers build on the formats defined in PKCS#7 version 1.5. For encryption, the EnvelopedData wrapper is used. The EnvelopedData wrapper requires RSA key management, and the RSA public keys must be carried in certificates. S/MIME does not include a location that can be used to carry a key recovery field. However, the key recovery center could be a recipient on every message, even if the message is not delivered to the key recovery center. In this way, the key recovery center private key can be used to recover the message plaintext content. Key recovery may also be done as part of certificate management. This technique only works if the originator is a recipient of the message. That is, a RecipientInfo field for the originator must be included to ensure that the key used to encrypt the message content is available to the key recovery center who holds a copy of the originator's RSA private key. D.3 MSP The Message Security Protocol (MSP) provides encryption for Internet and X.400 electronic mail. MSP is used in the Defense Message System, and MSP is specified in SDN.701. Like S/MIME, MSP supports both digital signatures and encryption; however, MSP defines one wrapper to provide both services. MSP is algorithm independent. MSP includes two locations that could be used to carry a key recovery field: the token and the extensions. To carry a key recovery field in the token, a separate object identifier for a new key management technique must be assigned. This approach would destroy interoperability with existing implementations. To carry a key recovery field in the extensions, a non-critical extension is added to the end of the message. MSP does not encrypt the extensions; therefore a key recovery field carried in an extension would be accessible. Alternatively, the key recovery center could be a recipient on every message, even if the message is not delivered to the key recovery center. In this way, the key recovery center private key can be used to recover the message plaintext content. Key recovery may also be done as part of certificate management. MSP includes a token for the originator. If the mail transfer system is unable to deliver the MSP protected message and returns the message to the originator as part of non- delivery notification, this token allows the originator to decrypt the message to determine which one was returned. If the key recovery center holds a copy of the originator's private key, then the key recovery center can also use the originator token to decrypt the message content. D.4 PEM The Privacy Enhanced Mail (PEM) protocol provides encryption for Internet electronic mail. PEM defines one encapsulation mechanism for digital signatures and encryption. PEM is defined in Internet RFCs 1421 through 1424. Two forms of key management are supported for encryption: RSA key management using certificates, and out-of-band distribution of symmetric key encryption keys. PEM includes one location that could be used to carry a key recovery field: the Key-Info header line. This header line is used for both forms of key management. To carry a key recovery field in the Key-Info line, a separate Date Encryption Key protection algorithm identifier must be assigned. This approach would destroy interoperability with existing implementations. Key recovery may also be done as part of certificate management. RFC 1421 recommends that a Key-Info header line be included for the originator as well as each recipient. This technique only works if the originator Key-Info header line is included. That is, a Key-Info header line for the originator must be included to ensure that the key used to encrypt the message content is available to the key recovery center who holds a copy of the originator's RSA private key. RFC 1424 specifies the certificate management for PEM, and a single RSA key is used for key management and digital signature. Thus, this form of key recovery permits a malicious key recovery center to masquerade as the originator by generating signed PEM messages. These unauthorized messages could also be encrypted. D.5 ISAKM 1 The C2, B1 and B2 ratings are in accordance with the TCSEC (see the cross index in the Announcement section. 2 Note that the KRB is not itself KRI - the KRB contains KRI plus other information, including an integrity checking value. 3 The extension must be critical since only those who understand it in the key recovery context should use this public key. ADVISORY COMMITTEE DRAFT May 29, 1998 ADVISORY COMMITTEE DRAFT May 29, 1998 4 1 17 1