]> OpenPGP Aiven
paul.wouters@aiven.io
Proton AG
d.huigens@protonmail.com
Sequoia-PGP
justus@sequoia-pgp.org
FSIJ
gniibe@fsij.org
sec Internet-Draft This document specifies the message formats used in OpenPGP. OpenPGP provides encryption with public-key or symmetric cryptographic algorithms, digital signatures, compression and key management. This document is maintained in order to publish all necessary information needed to develop interoperable applications based on the OpenPGP format. It is not a step-by-step cookbook for writing an application. It describes only the format and methods needed to read, check, generate, and write conforming packets crossing any network. It does not deal with storage and implementation questions. It does, however, discuss implementation issues necessary to avoid security flaws. This document obsoletes: RFC 4880 (OpenPGP), RFC 5581 (Camellia in OpenPGP) and RFC 6637 (Elliptic Curves in OpenPGP). The latest revision of this draft can be found at . Status information for this document may be found at . Discussion of this document takes place on the OpenPGP Working Group mailing list (), which is archived at . Subscribe at . Source for this draft and an issue tracker can be found at .
Introduction This document provides information on the message-exchange packet formats used by OpenPGP to provide encryption, decryption, signing, and key management functions. It is a revision of RFC 4880, "OpenPGP Message Format", which is a revision of RFC 2440, which itself replaces RFC 1991, "PGP Message Exchange Formats" . This document obsoletes: (OpenPGP), (Camellia in OpenPGP) and (Elliptic Curves in OpenPGP). This document incorporates all - at the time of writing - outstanding verified errata which are listed in . Software that has already implemented those previous standards may want to review for pointers to what has changed.
Terms The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 when, and only when, they appear in all capitals, as shown here. The key words "PRIVATE USE", "SPECIFICATION REQUIRED", and "RFC REQUIRED" that appear in this document when used to describe namespace allocation are to be interpreted as described in . Some terminology used in this document has been improved from previous versions of the OpenPGP specification. See for more details.
General functions OpenPGP provides data confidentiality and integrity for messages and data files by using public-key and/or symmetric encryption, and digital signatures. It provides formats for encoding and transferring encrypted and/or signed messages. In addition, OpenPGP provides functionality for encoding and transferring keys and certificates, though key storage and management is beyond the scope of this document.
Confidentiality via Encryption OpenPGP combines symmetric-key encryption and (optionally) public-key encryption to provide confidentiality. When using public keys, first the object is encrypted using a symmetric encryption algorithm. Each symmetric key is used only once, for a single object. A new "session key" is generated as a random number for each object (sometimes referred to as a session). Since it is used only once, the session key is bound to the message and transmitted with it. To protect the key, it is encrypted with the receiver's public key. The sequence is as follows: The sender creates a message. The sending OpenPGP implementation generates a random session key for this message. The session key is encrypted using each recipient's public key. These "encrypted session keys" start the message. The sending OpenPGP implementation optionally compresses the message, and then encrypts it using a message key derived from the session key. The encrypted message forms the remainder of the OpenPGP message. The receiving OpenPGP implementation decrypts the session key using the recipient's private key. The receiving OpenPGP implementation decrypts the message using the message key derived from the session key. If the message was compressed, it will be decompressed. When using symmetric-key encryption, a similar process as described above is used, but the session key is encrypted with a symmetric algorithm derived from a shared secret. Both digital signature and confidentiality services may be applied to the same message. First, a signature is generated for the message and attached to the message. Then the message plus signature is encrypted using a symmetric message key derived from the session key. Finally, the session key is encrypted using public-key encryption and prefixed to the encrypted block.
Authentication via Digital Signature The digital signature uses a cryptographic hash function and a public-key signature algorithm. The sequence is as follows: The sender creates a message. The sending implementation generates a hash digest of the message. The sending implementation generates a signature from the hash digest using the sender's private key. The signature is attached to or transmitted alongside the message. The receiving implementation obtains a copy of the message and the message signature. The receiving implementation generates a new hash digest for the received message and verifies it using the message's signature. If the verification is successful, the message is accepted as authentic.
Compression An OpenPGP implementation MAY support the compression of data. Many existing OpenPGP messages are compressed. Implementers, such as those working on constrained implementations that do not want to support compression, might want to consider at least implementing decompression.
Conversion to Base64 OpenPGP's underlying native representation for encrypted messages, signatures, keys, and certificates is a stream of arbitrary octets. Some systems only permit the use of blocks consisting of seven-bit, printable text. For transporting OpenPGP's native raw binary octets through channels that are not safe to transport raw binary data, a printable encoding of these binary octets is defined. The raw 8-bit binary octet stream can be converted to a stream of printable ASCII characters using base64 encoding, in a format called ASCII Armor (see ). Implementations SHOULD support base64 conversions.
Signature-Only Applications OpenPGP is designed for applications that use both encryption and signatures, but there are a number of use cases that only require a signature-only implementation. Although this specification requires both encryption and signatures, it is reasonable for there to be subset implementations that are non-conformant only in that they omit encryption support.
Data Element Formats This section describes the data elements used by OpenPGP.
Scalar Numbers Scalar numbers are unsigned and are always stored in big-endian format. Using n[k] to refer to the kth octet being interpreted, the value of a two-octet scalar is ((n[0] << 8) + n[1]). The value of a four-octet scalar is ((n[0] << 24) + (n[1] << 16) + (n[2] << 8) + n[3]).
Multiprecision Integers Multiprecision integers (also called MPIs) are unsigned integers used to hold large integers such as the ones used in cryptographic calculations. An MPI consists of two pieces: a two-octet scalar that is the length of the MPI in bits followed by a string of octets that contain the actual integer. These octets form a big-endian number; a big-endian number can be made into an MPI by prefixing it with the appropriate length. Examples: (all numbers in the octet strings identified by square brackets are in hexadecimal) The string of octets [00 00] forms an MPI with the value 0. The string of octets [00 01 01] forms an MPI with the value 1. The string [00 09 01 FF] forms an MPI with the value of 511. Additional rules: The size of an MPI is ((MPI.length + 7) / 8) + 2 octets. The length field of an MPI describes the length starting from its most significant non-zero bit. Thus, the MPI [00 02 01] is not formed correctly. It should be [00 01 01]. When parsing an MPI in a v6 Key, Signature, or Public-Key Encrypted Session Key packet, the implementation MUST check that the encoded length matches the length starting from the most significant non-zero bit, and reject the packet as malformed if not. Unused bits of an MPI MUST be zero.
Using MPIs to encode other data Note that MPIs are in some places used to encode non-integer data, such as an elliptic curve point (see ), or an octet string of known, fixed length (see ). The wire representation is the same: two octets of length in bits counted from the first non-zero bit, followed by the smallest series of octets that can represent the value while stripping off any leading zero octets.
Key IDs and Fingerprints A Key ID is an eight-octet scalar that identifies a key. Implementations SHOULD NOT assume that Key IDs are unique. A fingerprint is more likely to be unique than a key ID. The fingerprint and key ID of a key are calculated differently according to the version of the key. describes how Key IDs and Fingerprints are formed.
Text Unless otherwise specified, the character set for text is the UTF-8 encoding of Unicode .
Time Fields A time field is an unsigned four-octet number containing the number of seconds elapsed since midnight, 1 January 1970 UTC.
Keyrings A keyring is a collection of one or more keys in a file or database. Traditionally, a keyring is simply a sequential list of keys, but may be any suitable database. It is beyond the scope of this standard to discuss the details of keyrings or other databases.
String-to-Key (S2K) Specifier A string-to-key (S2K) specifier type is used to convert a passphrase string into a symmetric-key encryption/decryption key. Passphrases requiring use of S2K conversion are currently used in two places: to encrypt the secret part of private keys, and for symmetrically encrypted messages.
String-to-Key (S2K) Specifier Types There are four types of S2K Specifier Types currently specified, and some reserved values: ID S2K Type S2K field size (octets) Reference Generate? 0 Simple S2K 2 No 1 Salted S2K 10 Only when string is high entropy 2 Reserved value - - No 3 Iterated and Salted S2K 11 Yes 4 Argon2 20 Yes 100 to 110 Private/Experimental S2K - - As appropriate These are described in the subsections below. If the "Generate?" column is not "Yes", the S2K entry is used only for reading in backwards compatibility mode and SHOULD NOT be used to generate new output.
Simple S2K This directly hashes the string to produce the key data. See below for how this hashing is done.
Simple S2K hashes the passphrase to produce the session key. The manner in which this is done depends on the size of the session key (which depends on the cipher the session key will be used with) and the size of the hash algorithm's output. If the hash size is greater than the session key size, the high-order (leftmost) octets of the hash are used as the key. If the hash size is less than the key size, multiple instances of the hash context are created --- enough to produce the required key data. These instances are preloaded with 0, 1, 2, ... octets of zeros (that is to say, the first instance has no preloading, the second gets preloaded with 1 octet of zero, the third is preloaded with two octets of zeros, and so forth). As the data is hashed, it is given independently to each hash context. Since the contexts have been initialized differently, they will each produce different hash output. Once the passphrase is hashed, the output data from the multiple hashes is concatenated, first hash leftmost, to produce the key data, with any excess octets on the right discarded.
Salted S2K This includes a "salt" value in the S2K specifier --- some arbitrary data --- that gets hashed along with the passphrase string, to help prevent dictionary attacks.
Salted S2K is exactly like Simple S2K, except that the input to the hash function(s) consists of the 8 octets of salt from the S2K specifier, followed by the passphrase.
Iterated and Salted S2K This includes both a salt and an octet count. The salt is combined with the passphrase and the resulting value is repeated and then hashed. This further increases the amount of work an attacker must do to try dictionary attacks.
The count is coded into a one-octet number using the following formula:
> 4) + EXPBIAS); ]]>
The above formula is in , where "Int32" is a type for a 32-bit integer, and the variable "c" is the coded count, Octet 10. Iterated-Salted S2K hashes the passphrase and salt data multiple times. The total number of octets to be hashed is specified in the encoded count in the S2K specifier. Note that the resulting count value is an octet count of how many octets will be hashed, not an iteration count. Initially, one or more hash contexts are set up as with the other S2K algorithms, depending on how many octets of key data are needed. Then the salt, followed by the passphrase data, is repeatedly processed as input to each hash context until the number of octets specified by the octet count has been hashed. The input is truncated to the octet count, except if the octet count is less than the initial isize of the salt plus passphrase. That is, at least one copy of the full salt plus passphrase will be provided as input to each hash context regardless of the octet count. After the hashing is done, the key data is produced from the hash digest(s) as with the other S2K algorithms.
Argon2 This S2K method hashes the passphrase using Argon2, specified in . This provides memory-hardness, further protecting the passphrase against brute-force attacks.
The salt SHOULD be unique for each passphrase. The number of passes t and the degree of parallelism p MUST be non-zero. The memory size m is 2**encoded_m kibibytes of RAM. The encoded memory size MUST be a value from 3+ceil(log_2(p)) to 31, such that the decoded memory size m is a value from 8*p to 2**31. Note that memory-hardness size is indicated in kibibytes (KiB), not octets. Argon2 is invoked with the passphrase as P, the salt as S, the values of t, p and m as described above, the required key size as the tag length T, 0x13 as the version v, and Argon2id as the type. For the recommended values of t, p and m, see . If the recommended value of m for a given application is not a power of 2, it is RECOMMENDED to round up to the next power of 2 if the resulting performance would be acceptable, and round down otherwise (keeping in mind that m must be at least 8*p). As an example, with the first recommended option (t=1, p=4, m=2**21), the full S2K specifier would be:
(where XX represents a random octet of salt).
String-to-Key Usage Simple S2K and Salted S2K specifiers can be brute-forced when used with a low-entropy string, such as those typically provided by users. In addition, the usage of Simple S2K can lead to key and IV reuse (see ). Therefore, when generating an S2K specifier, an implementation MUST NOT use Simple S2K. Furthermore, an implementation SHOULD NOT generate a Salted S2K unless the implementation knows that the input string is high-entropy (for example, it generated the string itself using a known-good source of randomness). It is RECOMMENDED that implementations use Argon2. If Argon2 is not available, Iterated and Salted S2K MAY be used if care is taken to use a high octet count and a strong passphrase. However, this method does not provide memory-hardness, unlike Argon2.
Secret-Key Encryption The first octet following the public key material in a secret key packet () indicates whether and how the secret key material is passphrase-protected. This first octet is known as the "S2K usage octet". If S2K usage octet is zero, the secret key data is unprotected. If it is non-zero, it describes how to use a passphrase to unlock the secret key. Implementations predating indicated a protected key by storing a symmetric cipher algorithm ID (see ) in the S2K usage octet. In this case, the MD5 hash function was always used to convert the passphrase to a key for the specified cipher algorithm. Later implementations indicate a protected secret key by storing a special value 253 (AEAD), 254 (CFB), or 255 (MalleableCFB) in the S2K usage octet. The S2K usage octet is then followed immediately by a set of fields that describe how to convert a passphrase to a symmetric key that can unlock the secret material, plus other parameters relevant to the type of encryption used. The wire format fields also differ based on the version of the enclosing OpenPGP packet. The table below, indexed by S2K usage octet, summarizes the specifics described in . In the table below, check(x) means the "2-octet checksum" meaning the sum of all octets in x mod 65536. The info and packetprefix parameters are described in detail in . S2K usage octet Shorthand Encryption parameter fields Encryption Generate? 0 Unprotected - v3 or v4 keys: [cleartext secrets || check(secrets)]
v6 keys: [cleartext secrets] Yes Known symmetric cipher algo ID (see ) LegacyCFB IV CFB(MD5(passphrase), secrets || check(secrets)) No 253 AEAD params-length (v6-only), cipher-algo, AEAD-mode, S2K-specifier-length (v6-only), S2K-specifier, nonce AEAD(HKDF(S2K(passphrase), info), secrets, packetprefix) Yes 254 CFB params-length (v6-only), cipher-algo, S2K-specifier-length (v6-only), S2K-specifier, IV CFB(S2K(passphrase), secrets || SHA1(secrets)) Yes 255 MalleableCFB cipher-algo, S2K-specifier, IV CFB(S2K(passphrase), secrets || check(secrets)) No When emitting a secret key (with or without passphrase-protection) an implementation MUST only produce data from a row with "Generate?" marked as "Yes". Each row with "Generate?" marked as "No" is described for backward compatibility (for reading v4 and earlier keys only), and MUST NOT be used to generate new output. Version 6 secret keys using these formats MUST be rejected. Note that compared to a version 4 secret key, the parameters of a passphrase-protected version 6 secret key are stored with an additional pair of length counts, each of which is one octet wide. Argon2 is only used with AEAD (S2K usage octet 253). An implementation MUST NOT create and MUST reject as malformed any secret key packet where the S2K usage octet is not AEAD (253) and the S2K specifier type is Argon2.
Symmetric-Key Message Encryption OpenPGP can create a Symmetric-key Encrypted Session Key (ESK) packet at the front of a message. This is used to allow S2K specifiers to be used for the passphrase conversion or to create messages with a mix of symmetric-key ESKs and public-key ESKs. This allows a message to be decrypted either with a passphrase or a public-key pair. Implementations predating always used IDEA with Simple string-to-key conversion when encrypting a message with a symmetric algorithm. See . This MUST NOT be generated, but MAY be consumed for backward-compatibility.
Packet Syntax This section describes the packets used by OpenPGP.
Overview An OpenPGP message is constructed from a number of records that are traditionally called packets. A packet is a chunk of data that has a type ID specifying its meaning. An OpenPGP message, keyring, certificate, detached signature, and so forth consists of a number of packets. Some of those packets may contain other OpenPGP packets (for example, a compressed data packet, when uncompressed, contains OpenPGP packets). Each packet consists of a packet header, followed by the packet body. The packet header is of variable length. When handling a stream of packets, the length information in each packet header is the canonical source of packet boundaries. An implementation handling a packet stream that wants to find the next packet MUST look for it at the precise offset indicated in the previous packet header. Additionally, some packets contain internal length indicators (for example, a subfield within the packet). In the event that a subfield length indicator within a packet implies inclusion of octets outside the range indicated in the packet header, a parser MUST abort without writing outside the indicated range and MUST treat the packet as malformed and unusable. An implementation MUST NOT interpret octets outside the range indicated in the packet header as part of the contents of the packet.
Packet Headers The first octet of the packet denotes the format of the rest of the header, and encodes the Packet Type ID, indicating the type of the packet (see ). The remainder of the packet header is the length of the packet. There are two packet formats, the (current) OpenPGP packet format specified by this document and its predecessors and , and the Legacy packet format as used by implementations predating any IETF specification of the protocol. Note that the most significant bit is the leftmost bit, called bit 7. A mask for this bit is 0x80 in hexadecimal.
Bit 6 of the first octet of the packet header indicates whether the packet is encoded in the OpenPGP or Legacy packet format. The Legacy packet format MAY be used when consuming packets to facilitate interoperability and accessing archived data. The Legacy packet format SHOULD NOT be used to generate new data, unless the recipient is known to only support the Legacy packet format. This latter case is extremely unlikely, as the Legacy packet format was obsoleted by in 1998. An implementation that consumes and re-distributes pre-existing OpenPGP data (such as Transferable Public Keys) may encounter packets framed with the Legacy packet format. Such an implementation MAY either re-distribute these packets in their Legacy format, or transform them to the current OpenPGP packet format before re-distribution. Note that Legacy format headers only have 4 bits for the packet type ID, and hence can only encode packet type IDs less than 16, whereas the OpenPGP format headers can encode IDs as great as 63.
OpenPGP Format Packet Lengths OpenPGP format packets have four possible ways of encoding length: A one-octet Body Length header encodes packet lengths of up to 191 octets. A two-octet Body Length header encodes packet lengths of 192 to 8383 octets. A five-octet Body Length header encodes packet lengths of up to 4,294,967,295 (0xFFFFFFFF) octets in length. (This actually encodes a four-octet scalar number.) When the length of the packet body is not known in advance by the issuer, Partial Body Length headers encode a packet of indeterminate length, effectively making it a stream.
One-Octet Lengths A one-octet Body Length header encodes a length of 0 to 191 octets. This type of length header is recognized because the one octet value is less than 192. The body length is equal to:
Two-Octet Lengths A two-octet Body Length header encodes a length of 192 to 8383 octets. It is recognized because its first octet is in the range 192 to 223. The body length is equal to:
Five-Octet Lengths A five-octet Body Length header consists of a single octet holding the value 255, followed by a four-octet scalar. The body length is equal to:
This basic set of one, two, and five-octet lengths is also used internally to some packets.
Partial Body Lengths A Partial Body Length header is one octet long and encodes the length of only part of the data packet. This length is a power of 2, from 1 to 1,073,741,824 (2 to the 30th power). It is recognized by its one octet value that is greater than or equal to 224, and less than 255. The Partial Body Length is equal to:
Each Partial Body Length header is followed by a portion of the packet body data. The Partial Body Length header specifies this portion's length. Another length header (one octet, two-octet, five-octet, or partial) follows that portion. The last length header in the packet MUST NOT be a Partial Body Length header. Partial Body Length headers may only be used for the non-final parts of the packet. Note also that the last Body Length header can be a zero-length header. An implementation MAY use Partial Body Lengths for data packets, be they literal, compressed, or encrypted. The first partial length MUST be at least 512 octets long. Partial Body Lengths MUST NOT be used for any other packet types.
Legacy Format Packet Lengths A zero in bit 6 of the first octet of the packet indicates a Legacy packet format. Bits 1 and 0 of the first octet of a Legacy packet are the "length-type" field. The meaning of the length-type in Legacy format packets is:
0
The packet has a one-octet length. The header is 2 octets long.
1
The packet has a two-octet length. The header is 3 octets long.
2
The packet has a four-octet length. The header is 5 octets long.
3
The packet is of indeterminate length. The header is 1 octet long, and the implementation must determine how long the packet is. If the packet is in a file, this means that the packet extends until the end of the file. The OpenPGP format headers have a mechanism for precisely encoding data of indeterminate length. An implementation MUST NOT generate a Legacy format packet with indeterminate length. An implementation MAY interpret an indeterminate length Legacy format packet in order to deal with historic data, or data generated by a legacy system that predates support for .
Packet Length Examples These examples show ways that OpenPGP format packets might encode the packet body lengths. A packet body with length 100 may have its length encoded in one octet: 0x64. This is followed by 100 octets of data. A packet body with length 1723 may have its length encoded in two octets: 0xC5, 0xFB. This header is followed by the 1723 octets of data. A packet body with length 100000 may have its length encoded in five octets: 0xFF, 0x00, 0x01, 0x86, 0xA0. It might also be encoded in the following octet stream: 0xEF, first 32768 octets of data; 0xE1, next two octets of data; 0xE0, next one octet of data; 0xF0, next 65536 octets of data; 0xC5, 0xDD, last 1693 octets of data. This is just one possible encoding, and many variations are possible on the size of the Partial Body Length headers, as long as a regular Body Length header encodes the last portion of the data. Please note that in all of these explanations, the total length of the packet is the length of the header(s) plus the length of the body.
Packet Criticality The Packet Type ID space is partitioned into critical packets and non-critical packets. If an implementation encounters a critical packet where the packet type is unknown in a packet sequence, it MUST reject the whole packet sequence (see ). On the other hand, an unknown non-critical packet MUST be ignored. Packets with Type IDs from 0 to 39 are critical. Packets with Type IDs from 40 to 63 are non-critical.
Packet Types The defined packet types are as follows: ID Critical Packet Type Description Reference Shorthand 0 yes Reserved - a packet MUST NOT have this packet type ID     1 yes Public-Key Encrypted Session Key Packet PKESK 2 yes Signature Packet SIG 3 yes Symmetric-Key Encrypted Session Key Packet SKESK 4 yes One-Pass Signature Packet OPS 5 yes Secret-Key Packet SECKEY 6 yes Public-Key Packet PUBKEY 7 yes Secret-Subkey Packet SECSUBKEY 8 yes Compressed Data Packet COMP 9 yes Symmetrically Encrypted Data Packet SED 10 yes Marker Packet MARKER 11 yes Literal Data Packet LIT 12 yes Trust Packet TRUST 13 yes User ID Packet UID 14 yes Public-Subkey Packet PUBSUBKEY 17 yes User Attribute Packet UAT 18 yes Symmetrically Encrypted and Integrity Protected Data Packet SEIPD 19 yes Reserved (formerly Modification Detection Code Packet) (see )   20 yes Reserved     21 yes Padding Packet PADDING 22 to 39 yes Unassigned Critical Packet     40 to 59 no Unassigned Non-Critical Packet     60 to 63 no Private or Experimental Values     The labels in the "Shorthand" column are used for compact reference elsewhere in this draft, and may also be used by implementations that provide debugging or inspection affordances for streams of OpenPGP packets.
Public-Key Encrypted Session Key Packet (Type ID 1) Zero or more Public-Key Encrypted Session Key (PKESK) packets and/or Symmetric-Key Encrypted Session Key packets () precede an encryption container (that is, a Symmetrically Encrypted Integrity Protected Data packet or --- for historic data --- a Symmetrically Encrypted Data packet), which holds an encrypted message. The message is encrypted with the session key, and the session key is itself encrypted and stored in the Encrypted Session Key packet(s). The encryption container is preceded by one Public-Key Encrypted Session Key packet for each OpenPGP key to which the message is encrypted. The recipient of the message finds a session key that is encrypted to their public key, decrypts the session key, and then uses the session key to decrypt the message. The body of this packet starts with a one-octet number giving the version number of the packet type. The currently defined versions are 3 and 6. The remainder of the packet depends on the version. The versions differ in how they identify the recipient key, and in what they encode. The version of the PKESK packet must align with the version of the SEIPD packet (see ). Any new version of the PKESK packet should be registered in the registry established in .
Version 3 Public-Key Encrypted Session Key Packet Format A version 3 Public-Key Encrypted Session Key (PKESK) packet precedes a version 1 Symmetrically Encrypted Integrity Protected Data (v1 SEIPD, see ) packet. In historic data, it is sometimes found preceding a deprecated Symmetrically Encrypted Data packet (SED, see ). A v3 PKESK packet MUST NOT precede a v2 SEIPD packet (see ). The v3 PKESK packet consists of: A one-octet version number with value 3. An eight-octet number that gives the Key ID of the public key to which the session key is encrypted. If the session key is encrypted to a subkey, then the Key ID of this subkey is used here instead of the Key ID of the primary key. The Key ID may also be all zeros, for an "anonymous recipient" (see ). A one-octet number giving the public-key algorithm used. A series of values comprising the encrypted session key. This is algorithm-specific and described below. The public-key encryption algorithm (described in subsequent sections) is passed two values: The session key. The one-octet algorithm identifier that specifies the symmetric encryption algorithm used to encrypt the following v1 SEIPD packet.
Version 6 Public-Key Encrypted Session Key Packet Format A version 6 Public-Key Encrypted Session Key (PKESK) packet precedes a version 2 Symmetrically Encrypted Integrity Protected Data (v2 SEIPD, see ) packet. A v6 PKESK packet MUST NOT precede a v1 SEIPD packet or a deprecated Symmetrically Encrypted Data packet (see ). The v6 PKESK packet consists of the following fields: A one-octet version number with value 6. A one-octet size of the following two fields. This size may be zero, if the key version number field and the fingerprint field are omitted for an "anonymous recipient" (see ). A one octet key version number. The fingerprint of the public key or subkey to which the session key is encrypted. Note that the length N of the fingerprint for a version 4 key is 20 octets; for a version 6 key N is 32. A one-octet number giving the public-key algorithm used. A series of values comprising the encrypted session key. This is algorithm-specific and described below. The session key is encrypted according to the public-key algorithm used, as described below. No symmetric encryption algorithm identifier is passed to the public-key algorithm for a v6 PKESK packet, as it is included in the v2 SEIPD packet.
Algorithm-Specific Fields for RSA encryption Multiprecision integer (MPI) of RSA-encrypted value m**e mod n. To produce the value "m" in the above formula, first concatenate the following values: The one-octet algorithm identifier, if it was passed (in the case of a v3 PKESK packet). The session key. A two-octet checksum of the session key, equal to the sum of the session key octets, modulo 65536. Then, the above values are encoded using the PKCS#1 block encoding EME-PKCS1-v1_5 described in step 2 of (see also ). When decoding "m" during decryption, an implementation should follow step 3 of (see also ). Note that when an implementation forms several PKESKs with one session key, forming a message that can be decrypted by several keys, the implementation MUST make a new PKCS#1 encoding for each key. This defends against attacks such as those discussed in .
Algorithm-Specific Fields for Elgamal encryption MPI of Elgamal (Diffie-Hellman) value g**k mod p. MPI of Elgamal (Diffie-Hellman) value m * y**k mod p. To produce the value "m" in the above formula, first concatenate the following values: The one-octet algorithm identifier, if it was passed (in the case of a v3 PKESK packet). The session key. A two-octet checksum of the session key, equal to the sum of the session key octets, modulo 65536. Then, the above values are encoded using the PKCS#1 block encoding EME-PKCS1-v1_5 described in step 2 of (see also ). When decoding "m" during decryption, an implementation should follow step 3 of (see also ). Note that when an implementation forms several PKESKs with one session key, forming a message that can be decrypted by several keys, the implementation MUST make a new PKCS#1 encoding for each key. This defends against attacks such as those discussed in . An implementation MUST NOT generate ElGamal v6 PKESKs.
Algorithm-Specific Fields for ECDH encryption MPI of an EC point representing an ephemeral public key, in the point format associated with the curve as specified in . A one-octet size, followed by a symmetric key encoded using the method described in .
Algorithm-Specific Fields for X25519 encryption 32 octets representing an ephemeral X25519 public key. A one-octet size of the following fields. The one-octet algorithm identifier, if it was passed (in the case of a v3 PKESK packet). The encrypted session key. See for more details on the computation of the ephemeral public key and the shared secret. HKDF () is then used with SHA256 and an info parameter of "OpenPGP X25519" and no salt. The input of HKDF is the concatenation of the following three values: 32 octets of the ephemeral X25519 public key from this packet. 32 octets of the recipient public key material. 32 octets of the shared secret. The key produced from HKDF is used to encrypt the session key with AES-128 key wrap, as defined in . Note that unlike ECDH, no checksum or padding are appended to the session key before key wrapping. Finally, note that unlike the other public-key algorithms, in the case of a v3 PKESK packet, the symmetric algorithm ID is not encrypted. Instead, it is prepended to the encrypted session key in plaintext. In this case, the symmetric algorithm used MUST be AES-128, AES-192 or AES-256 (algorithm ID 7, 8 or 9).
Algorithm-Specific Fields for X448 encryption 56 octets representing an ephemeral X448 public key. A one-octet size of the following fields. The one-octet algorithm identifier, if it was passed (in the case of a v3 PKESK packet). The encrypted session key. See for more details on the computation of the ephemeral public key and the shared secret. HKDF () is then used with SHA512 () and an info parameter of "OpenPGP X448" and no salt. The input of HKDF is the concatenation of the following three values: 56 octets of the ephemeral X448 public key from this packet. 56 octets of the recipient public key material. 56 octets of the shared secret. The key produced from HKDF is used to encrypt the session key with AES-256 key wrap, as defined in . Note that unlike ECDH, no checksum or padding are appended to the session key before key wrapping. Finally, note that unlike the other public-key algorithms, in the case of a v3 PKESK packet, the symmetric algorithm ID is not encrypted. Instead, it is prepended to the encrypted session key in plaintext. In this case, the symmetric algorithm used MUST be AES-128, AES-192 or AES-256 (algorithm ID 7, 8 or 9).
Notes on PKESK An implementation MAY accept or use a Key ID of all zeros, or an omitted key fingerprint, to hide the intended decryption key. In this case, the receiving implementation would try all available private keys, checking for a valid decrypted session key. This format helps reduce traffic analysis of messages.
Signature Packet (Type ID 2) A Signature packet describes a binding between some public key and some data. The most common signatures are a signature of a file or a block of text, and a signature that is a certification of a User ID. Three versions of Signature packets are defined. Version 3 provides basic signature information, while versions 4 and 6 provide an expandable format with subpackets that can specify more information about the signature. For historical reasons, versions 1, 2, and 5 of the Signature packet are unspecified. Any new Signature packet version should be registered in the registry established in . An implementation MUST generate a version 6 signature when signing with a version 6 key. An implementation MUST generate a version 4 signature when signing with a version 4 key. Implementations MUST NOT create version 3 signatures; they MAY accept version 3 signatures. See for more details about packet version correspondence between keys and signatures.
Signature Types There are a number of possible meanings for a signature, which are indicated by the signature type ID in any given signature. Please note that the vagueness of these meanings is not a flaw, but a feature of the system. Because OpenPGP places final authority for validity upon the receiver of a signature, it may be that one signer's casual act might be more rigorous than some other authority's positive act. See for detailed information on how to compute and verify signatures of each type. ID Name Reference 0x00 Binary Signature 0x01 Text Signature 0x02 Standalone Signature 0x10 Generic Certification 0x11 Persona Certification 0x12 Casual Certification 0x13 Positive Certification 0x18 Subkey Binding Signature 0x19 Primary Key Binding Signature 0x1F Direct Key Signature 0x20 Key Revocation 0x28 Subkey Revocation 0x30 Certification Revocation 0x40 Timestamp Signature 0x50 Third-Party Confirmation 0xFF Reserved These meanings of each signature type are described in the subsections below.
Signature of a binary document (type ID 0x00) This means the signer owns it, created it, or certifies that it has not been modified.
Signature of a canonical text document (type ID 0x01) This means the signer owns it, created it, or certifies that it has not been modified. The signature is calculated over the text data with its line endings converted to <CR><LF>.
Standalone signature (type ID 0x02) This signature is a signature of only its own subpacket contents. It is calculated identically to a signature over a zero-length binary document. V3 standalone signatures MUST NOT be generated and MUST be ignored.
Generic certification of a User ID and Public-Key packet (type ID 0x10) The issuer of this certification does not make any particular assertion as to how well the certifier has checked that the owner of the key is in fact the person described by the User ID.
Persona certification of a User ID and Public-Key packet (type ID 0x11) The issuer of this certification has not done any verification of the claim that the owner of this key is the User ID specified.
Casual certification of a User ID and Public-Key packet (type ID 0x12) The issuer of this certification has done some casual verification of the claim of identity.
Positive certification of a User ID and Public-Key packet (type ID 0x13) The issuer of this certification has done substantial verification of the claim of identity. Most OpenPGP implementations make their "key signatures" as generic (type ID 0x10) certifications. Some implementations can issue 0x11-0x13 certifications, but few differentiate between the types.
Subkey Binding Signature (type ID 0x18) This signature is a statement by the top-level signing key that indicates that it owns the subkey. This signature is calculated directly on the primary key and subkey, and not on any User ID or other packets. A signature that binds a signing subkey MUST have an Embedded Signature subpacket in this binding signature that contains a 0x19 signature made by the signing subkey on the primary key and subkey.
Primary Key Binding Signature (type ID 0x19) This signature is a statement by a signing subkey, indicating that it is owned by the primary key. This signature is calculated the same way as a subkey binding signature (0x18): directly on the primary key and subkey, and not on any User ID or other packets.
Direct Key Signature (type ID 0x1F) This signature is calculated directly on a key. It binds the information in the Signature subpackets to the key, and is appropriate to be used for subpackets that provide information about the key, such as the Key Flags subpacket or (deprecated) Revocation Key. It is also appropriate for statements that non-self certifiers want to make about the key itself, rather than the binding between a key and a name.
Key revocation signature (type ID 0x20) The signature is calculated directly on the key being revoked. A revoked key is not to be used. Only revocation signatures by the key being revoked, or by a (deprecated) Revocation Key, should be considered valid revocation signatures.
Subkey revocation signature (type ID 0x28) The signature is calculated directly on the primary key and the subkey being revoked. A revoked subkey is not to be used. Only revocation signatures by the top-level signature key that is bound to this subkey, or by a (deprecated) Revocation Key, should be considered valid revocation signatures.
Certification revocation signature (type ID 0x30) This signature revokes an earlier User ID certification signature (signature class 0x10 through 0x13) or direct key signature (0x1F). It should be issued by the same key that issued the revoked signature or by a (deprecated) Revocation Key. The signature is computed over the same data as the certification that it revokes, and should have a later creation date than that certification.
Timestamp signature (type ID 0x40) This signature is only meaningful for the timestamp contained in it.
Third-Party Confirmation signature (type ID 0x50) This signature is a signature over some other OpenPGP Signature packet(s). It is analogous to a notary seal on the signed data. A third-party signature SHOULD include Signature Target subpacket(s) to give easy identification. Note that we really do mean SHOULD. There are plausible uses for this (such as a blind party that only sees the signature, not the key or source document) that cannot include a target subpacket.
Reserved (type ID 0xFF) An implementation MUST NOT create any signature with this type, and MUST NOT validate any signature made with this type. See for more details.
Version 3 Signature Packet Format The body of a version 3 Signature packet contains: One-octet version number (3). One-octet length of following hashed material. MUST be 5. One-octet signature type ID. Four-octet creation time. Eight-octet Key ID of signer. One-octet public-key algorithm. One-octet hash algorithm. Two-octet field holding left 16 bits of signed hash value. One or more multiprecision integers comprising the signature. This portion is algorithm-specific, as described below. The concatenation of the data to be signed, the signature type, and creation time from the Signature packet (5 additional octets) is hashed. The resulting hash value is used in the signature algorithm. The high 16 bits (first two octets) of the hash are included in the Signature packet to provide a way to reject some invalid signatures without performing a signature verification. Algorithm-Specific Fields for RSA signatures: Multiprecision integer (MPI) of RSA signature value m**d mod n. Algorithm-Specific Fields for DSA signatures: MPI of DSA value r. MPI of DSA value s. The signature calculation is based on a hash of the signed data, as described above. The details of the calculation are different for DSA signatures than for RSA signatures, see and .
Version 4 and 6 Signature Packet Formats The body of a v4 or v6 Signature packet contains: One-octet version number. This is 4 for v4 signatures and 6 for v6 signatures. One-octet signature type ID. One-octet public-key algorithm. One-octet hash algorithm. A scalar octet count for the hashed subpacket data that follows this field. For a v4 signature, this is a two-octet field. For a v6 signature, this is a four-octet field. Note that this is the length in octets of all of the hashed subpackets; an implementation's pointer incremented by this number will skip over the hashed subpackets. Hashed subpacket data set (zero or more subpackets). A scalar octet count for the unhashed subpacket data that follows this field. For a v4 signature, this is a two-octet field. For a v6 signature, this is a four-octet field. Note that this is the length in octets of all of the unhashed subpackets; an implementation's pointer incremented by this number will skip over the unhashed subpackets. Unhashed subpacket data set (zero or more subpackets). Two-octet field holding the left 16 bits of the signed hash value. Only for v6 signatures, a variable-length field containing: A one-octet salt size. The value MUST match the value defined for the hash algorithm as specified in . The salt; a random value of the specified size. One or more multiprecision integers comprising the signature. This portion is algorithm-specific:
Algorithm-Specific Fields for RSA signatures Multiprecision integer (MPI) of RSA signature value m**d mod n. With RSA signatures, the hash value is encoded using PKCS#1 encoding type EMSA-PKCS1-v1_5 as described in (see also ). This requires inserting the hash value as an octet string into an ASN.1 structure. The object identifier (OID) for the hash algorithm itself is also included in the structure, see the OIDs in .
Algorithm-Specific Fields for DSA or ECDSA signatures MPI of DSA or ECDSA value r. MPI of DSA or ECDSA value s. A version 3 signature MUST NOT be created and MUST NOT be used with ECDSA. A DSA signature MUST use a hash algorithm with a digest size of at least the number of bits of q, the group generated by the DSA key's generator value. If the output size of the chosen hash is larger than the number of bits of q, the hash result is truncated to fit by taking the number of leftmost bits equal to the number of bits of q. This (possibly truncated) hash function result is treated as a number and used directly in the DSA signature algorithm. An ECDSA signature MUST use a hash algorithm with a digest size of at least the curve's "fsize" value (see ), except in the case of NIST P-521, for which at least a 512-bit hash algorithm MUST be used.
Algorithm-Specific Fields for EdDSALegacy signatures (deprecated) Two MPI-encoded values, whose contents and formatting depend on the choice of curve used (see ). A version 3 signature MUST NOT be created and MUST NOT be used with EdDSALegacy. An EdDSALegacy signature MUST use a hash algorithm with a digest size of at least the curve's "fsize" value (see ). A verifying implementation MUST reject any EdDSALegacy signature that uses a hash algorithm with a smaller digest size.
Algorithm-Specific Fields for Ed25519Legacy signatures (deprecated) The two MPIs for Ed25519Legacy use octet strings R and S as described in . Ed25519Legacy MUST NOT be used in signature packets version 6 or above. MPI of an EC point R, represented as a (non-prefixed) native (little-endian) octet string up to 32 octets. MPI of EdDSA value S, also in (non-prefixed) native (little-endian) format with a length up to 32 octets.
Algorithm-Specific Fields for Ed25519 signatures 64 octets of the native signature. For more details, see . A version 3 signature MUST NOT be created and MUST NOT be used with Ed25519. An Ed25519 signature MUST use a hash algorithm with a digest size of at least 256 bits. A verifying implementation MUST reject any Ed25519 signature that uses a hash algorithm with a smaller digest size.
Algorithm-Specific Fields for Ed448 signatures 114 octets of the native signature. For more details, see . A version 3 signature MUST NOT be created and MUST NOT be used with Ed448. An Ed448 signature MUST use a hash algorithm with a digest size of at least 512 bits. A verifying implementation MUST reject any Ed448 signature that uses a hash algorithm with a smaller digest size.
Notes on Signatures The concatenation of the data being signed, the signature data from the version number through the hashed subpacket data (inclusive), and (for signature versions later than 3) a six-octet trailer (see ) is hashed. The resulting hash value is what is signed. The high 16 bits (first two octets) of the hash are included in the Signature packet to provide a way to reject some invalid signatures without performing a signature verification. When verifying a v6 signature, an implementation MUST reject the signature if these octets don't match the first two octets of the computed hash. There are two fields consisting of Signature subpackets. The first field is hashed with the rest of the signature data, while the second is not hashed into the signature. The second set of subpackets (the "unhashed section") is not cryptographically protected by the signature and should include only advisory information. See for more information. The differences between a v4 and v6 signature are two-fold: first, a v6 signature increases the width of the fields that indicate the size of the hashed and unhashed subpackets, making it possible to include significantly more data in subpackets. Second, the hash is salted with random data (see ). The algorithms for converting the hash function result to a signature are described in .
Signature Subpacket Specification A subpacket data set consists of zero or more Signature subpackets. In Signature packets, the subpacket data set is preceded by a two-octet (for v4 signatures) or four-octet (for v6 signatures) scalar count of the length in octets of all the subpackets. A pointer incremented by this number will skip over the subpacket data set. Each subpacket consists of a subpacket header and a body. The header consists of: The subpacket length (1, 2, or 5 octets), The encoded subpacket type ID (1 octet), and is followed by the subpacket-specific data. The length includes the encoded subpacket type ID octet but not this length. Its format is similar to the OpenPGP format packet header lengths, but cannot have Partial Body Lengths. That is:
= 192 and < 255, then lengthOfLength = 2 subpacketLen = ((1st_octet - 192) << 8) + (2nd_octet) + 192 if the 1st octet = 255, then lengthOfLength = 5 subpacket length = [four-octet scalar starting at 2nd_octet] ]]>
Bit 7 of the encoded subpacket type ID is the "critical" bit. If set, it denotes that the subpacket is one that is critical for the evaluator of the signature to recognize. If a subpacket is encountered that is marked critical but is unknown to the evaluating implementation, the evaluator SHOULD consider the signature to be in error. An implementation SHOULD ignore any non-critical subpacket of a type that it does not recognize. An evaluator may "recognize" a subpacket, but not implement it. The purpose of the critical bit is to allow the signer to tell an evaluator that it would prefer a new, unknown feature to generate an error rather than being ignored. The other bits of the encoded subpacket type ID (i.e. bits 6-0) contain the subpacket type ID. The following signature subpackets are defined: ID Description Reference 0 Reserved   1 Reserved   2 Signature Creation Time 3 Signature Expiration Time 4 Exportable Certification 5 Trust Signature 6 Regular Expression 7 Revocable 8 Reserved   9 Key Expiration Time 10 Placeholder for backward compatibility   11 Preferred Symmetric Ciphers for v1 SEIPD 12 Revocation Key (deprecated) 13 to 15 Reserved   16 Issuer Key ID 17 to 19 Reserved   20 Notation Data 21 Preferred Hash Algorithms 22 Preferred Compression Algorithms 23 Key Server Preferences 24 Preferred Key Server 25 Primary User ID 26 Policy URI 27 Key Flags 28 Signer's User ID 29 Reason for Revocation 30 Features 31 Signature Target 32 Embedded Signature 33 Issuer Fingerprint 34 Reserved   35 Intended Recipient Fingerprint 37 Reserved (Attested Certifications)   38 Reserved (Key Block)   39 Preferred AEAD Ciphersuites 100 to 110 Private or experimental   Implementations SHOULD implement the four preferred algorithm subpackets (11, 21, 22, and 39), as well as the "Features" subpacket and the "Reason for Revocation" subpacket. To avoid surreptitious forwarding (see ), implementations SHOULD also implement the "Intended Recipients" subpacket. Note that if an implementation chooses not to implement some of the preferences subpackets, it MUST default to the mandatory-to-implement algorithms to ensure interoperability. An encrypting implementation that does not implement the "Features" subpacket SHOULD select the type of encrypted data format based instead on the versions of the recipient keys or external inference (see for more details).
Signature Subpacket Types A number of subpackets are currently defined for OpenPGP signatures. Some subpackets apply to the signature itself and some are attributes of the key. Subpackets that are found on a self-signature are placed on a certification made by the key itself. Note that a key may have more than one User ID, and thus may have more than one self-signature, and differing subpackets. A subpacket may be found either in the hashed or unhashed subpacket sections of a signature. If a subpacket is not hashed, then the information in it cannot be considered definitive because it is not part of the signature proper. See for more discussion about hashed and unhashed subpackets.
Notes on Subpackets It is certainly possible for a signature to contain conflicting information in subpackets. For example, a signature may contain multiple copies of a preference or multiple expiration times. In most cases, an implementation SHOULD use the last subpacket in the hashed section of the signature, but MAY use any conflict resolution scheme that makes more sense. Please note that we are intentionally leaving conflict resolution to the implementer; most conflicts are simply syntax errors, and the wishy-washy language here allows a receiver to be generous in what they accept, while putting pressure on a creator to be stingy in what they generate. Some apparent conflicts may actually make sense --- for example, suppose a keyholder has a v3 key and a v4 key that share the same RSA key material. Either of these keys can verify a signature created by the other, and it may be reasonable for a signature to contain an Issuer Key ID subpacket () for each key, as a way of explicitly tying those keys to the signature.
Notes on Self-Signatures A self-signature is a binding signature made by the key to which the signature refers. There are three types of self-signatures, the certification signatures (type IDs 0x10-0x13), the direct key signature (type ID 0x1F), and the subkey binding signature (type ID 0x18). A cryptographically-valid self-signature should be accepted from any primary key, regardless of what Key Flags () apply to the primary key. In particular, a primary key does not need to have 0x01 set in the first octet of Key Flags order to make a valid self-signature. For certification self-signatures, each User ID MAY have a self-signature, and thus different subpackets in those self-signatures. For subkey binding signatures, each subkey MUST have a self-signature. Subpackets that appear in a certification self-signature apply to the User ID, and subpackets that appear in the subkey self-signature apply to the subkey. Lastly, subpackets on the direct key signature apply to the entire key. An implementation should interpret a self-signature's preference subpackets as narrowly as possible. For example, suppose a key has two user names, Alice and Bob. Suppose that Alice prefers the AEAD ciphersuite AES-256 with OCB, and Bob prefers Camellia-256 with GCM. If the implementation locates this key via Alice's name, then the preferred AEAD ciphersuite is AES-256 with OCB; if the implementation locates the key via Bob's name, then the preferred algorithm is Camellia-256 with GCM. If the key is located by Key ID, the algorithm of the primary User ID of the key provides the preferred AEAD ciphersuite. Revoking a self-signature or allowing it to expire has a semantic meaning that varies with the signature type. Revoking the self-signature on a User ID effectively retires that user name. The self-signature is a statement, "My name X is tied to my signing key K" and is corroborated by other users' certifications. If another user revokes their certification, they are effectively saying that they no longer believe that name and that key are tied together. Similarly, if the users themselves revoke their self-signature, then the users no longer go by that name, no longer have that email address, etc. Revoking a binding signature effectively retires that subkey. Revoking a direct key signature cancels that signature. Please see for more relevant detail. Since a self-signature contains important information about the key's use, an implementation SHOULD allow the user to rewrite the self-signature, and important information in it, such as preferences and key expiration. When an implementation imports a secret key, it SHOULD verify that the key's internal self-signatures do not advertise features or algorithms that the implementation doesn't support. If an implementation observes such a mismatch, it SHOULD warn the user and offer to create new self-signatures that advertise the actual set of features and algorithms supported by the implementation. An implementation that encounters multiple self-signatures on the same object MUST select the most recent valid self-signature, and ignore all other self-signatures. By convention, a version 4 key stores information about the primary Public-Key (key flags, key expiration, etc.) and the Transferable Public Key as a whole (features, algorithm preferences, etc.) in a User ID self-signature of type 0x10 or 0x13. Some implementations require at least one User ID with a valid self-signature to be present to use a v4 key. For this reason, it is RECOMMENDED to include at least one User ID with a self-signature in v4 keys. For version 6 keys, it is RECOMMENDED to store information about the primary Public-Key as well as the Transferable Public Key as a whole (key flags, key expiration, features, algorithm preferences, etc.) in a direct key signature (type ID 0x1F) over the Public-Key instead of placing that information in a User ID self-signature. An implementation MUST ensure that a valid direct key signature is present before using a v6 key. This prevents certain attacks where an adversary strips a self-signature specifying a key expiration time or certain preferences. An implementation SHOULD NOT require a User ID self-signature to be present in order to consume or use a key, unless the particular use is contingent on the keyholder identifying themselves with the textual label in the User ID. For example, when refreshing a key to learn about changes in expiration, advertised features, algorithm preferences, revocation, subkey rotation, and so forth, there is no need to require a User ID self-signature. On the other hand, when verifying a signature over an e-mail message, an implementation MAY choose to only accept a signature from a key that has a valid self-signature over a User ID that matches the message's From: header, as a way to avoid a signature transplant attack.
Signature Creation Time (4-octet time field) The time the signature was made. This subpacket MUST be present in the hashed area. When generating this subpacket, it SHOULD be marked as critical.
Issuer Key ID (8-octet Key ID) The OpenPGP Key ID of the key issuing the signature. If the version of that key is greater than 4, this subpacket MUST NOT be included in the signature. For these keys, consider the Issuer Fingerprint subpacket () instead. Note: in previous versions of this specification, this subpacket was simply known as the "Issuer" subpacket.
Key Expiration Time (4-octet time field) The validity period of the key. This is the number of seconds after the key creation time that the key expires. For a direct or certification self-signature, the key creation time is that of the primary key. For a subkey binding signature, the key creation time is that of the subkey. If this is not present or has a value of zero, the key never expires. This is found only on a self-signature. When an implementation generates this subpacket, it SHOULD be marked as critical.
Preferred Symmetric Ciphers for v1 SEIPD (array of one-octet values) A series of symmetric cipher algorithm IDs indicating how the keyholder prefers to receive version 1 Symmetrically Encrypted Integrity Protected Data (). The subpacket body is an ordered list of octets with the most preferred listed first. It is assumed that only algorithms listed are supported by the recipient's implementation. Algorithm IDs are defined in . This is only found on a self-signature. When generating a v2 SEIPD packet, this preference list is not relevant. See instead.
Preferred AEAD Ciphersuites (array of pairs of octets indicating Symmetric Cipher and AEAD algorithms) A series of paired algorithm IDs indicating how the keyholder prefers to receive version 2 Symmetrically Encrypted Integrity Protected Data (). Each pair of octets indicates a combination of a symmetric cipher and an AEAD mode that the key holder prefers to use. The symmetric cipher algorithm ID precedes the AEAD algorithm ID in each pair. The subpacket body is an ordered list of pairs of octets with the most preferred algorithm combination listed first. It is assumed that only the combinations of algorithms listed are supported by the recipient's implementation, with the exception of the mandatory-to-implement combination of AES-128 and OCB. If AES-128 and OCB are not found in the subpacket, it is implicitly listed at the end. AEAD algorithm IDs are listed in . Symmetric cipher algorithm IDs are listed in . For example, a subpacket with content of these six octets:
Indicates that the keyholder prefers to receive v2 SEIPD using AES-256 with OCB, then AES-256 with GCM, then Camellia-256 with OCB, and finally the implicit AES-128 with OCB. Note that support for version 2 of the Symmetrically Encrypted Integrity Protected Data packet () in general is indicated by a Features Flag (). This subpacket is only found on a self-signature. When generating a v1 SEIPD packet, this preference list is not relevant. See instead.
Preferred Hash Algorithms (array of one-octet values) Message digest algorithm IDs that indicate which algorithms the key holder prefers to receive. Like the preferred AEAD ciphersuites, the list is ordered. Algorithm IDs are defined in . This is only found on a self-signature.
Preferred Compression Algorithms (array of one-octet values) Compression algorithm IDs that indicate which algorithms the key holder prefers to use. Like the preferred AEAD ciphersuites, the list is ordered. Algorithm IDs are defined in . A zero, or the absence of this subpacket, denotes that uncompressed data is preferred; the key holder's implementation might have no compression support available. This is only found on a self-signature.
Signature Expiration Time (4-octet time field) The validity period of the signature. This is the number of seconds after the signature creation time that the signature expires. If this is not present or has a value of zero, it never expires. When an implementation generates this subpacket, it SHOULD be marked as critical.
Exportable Certification (1 octet of exportability, 0 for not, 1 for exportable) This subpacket denotes whether a certification signature is "exportable", to be used by other users than the signature's issuer. The packet body contains a Boolean flag indicating whether the signature is exportable. If this packet is not present, the certification is exportable; it is equivalent to a flag containing a 1. Non-exportable, or "local", certifications are signatures made by a user to mark a key as valid within that user's implementation only. Thus, when an implementation prepares a user's copy of a key for transport to another user (this is the process of "exporting" the key), any local certification signatures are deleted from the key. The receiver of a transported key "imports" it, and likewise trims any local certifications. In normal operation, there won't be any, assuming the import is performed on an exported key. However, there are instances where this can reasonably happen. For example, if an implementation allows keys to be imported from a key database in addition to an exported key, then this situation can arise. Some implementations do not represent the interest of a single user (for example, a key server). Such implementations always trim local certifications from any key they handle. When an implementation generates this subpacket and denotes the signature as non-exportable, the subpacket MUST be marked as critical.
Revocable (1 octet of revocability, 0 for not, 1 for revocable) Signature's revocability status. The packet body contains a Boolean flag indicating whether the signature is revocable. Signatures that are not revocable have any later revocation signatures ignored. They represent a commitment by the signer that he cannot revoke his signature for the life of his key. If this packet is not present, the signature is revocable.
Trust Signature (1 octet "level" (depth), 1 octet of trust amount) Signer asserts that the key is not only valid but also trustworthy at the specified level. Level 0 has the same meaning as an ordinary validity signature. Level 1 means that the signed key is asserted to be a valid trusted introducer, with the 2nd octet of the body specifying the degree of trust. Level 2 means that the signed key is asserted to be trusted to issue level 1 trust signatures; that is, the signed key is a "meta introducer". Generally, a level n trust signature asserts that a key is trusted to issue level n-1 trust signatures. The trust amount is in a range from 0-255, interpreted such that values less than 120 indicate partial trust and values of 120 or greater indicate complete trust. Implementations SHOULD emit values of 60 for partial trust and 120 for complete trust.
Regular Expression (null-terminated UTF-8 encoded regular expression) Used in conjunction with trust Signature packets (of level > 0) to limit the scope of trust that is extended. Only signatures by the target key on User IDs that match the regular expression in the body of this packet have trust extended by the trust Signature subpacket. The regular expression uses the same syntax as Henry Spencer's "almost public domain" regular expression package. A description of the syntax is found in . The regular expression matches (or does not match) a sequence of UTF-8-encoded Unicode characters from User IDs. The expression itself is also written with UTF-8 characters. For historical reasons, this subpacket includes a null character (octet with value zero) after the regular expression. When an implementation parses a regular expression subpacket, it MUST remove this octet; if it is not present, it MUST reject the subpacket (i.e. ignore the subpacket if it's non-critical and reject the signature if it's critical). When an implementation generates a regular expression subpacket, it MUST include the null terminator. When generating this subpacket, it SHOULD be marked as critical.
Revocation Key (1 octet of class, 1 octet of public-key algorithm ID, 20 octets of v4 fingerprint) This mechanism is deprecated. Applications MUST NOT generate such a subpacket. An application that wants the functionality of delegating revocation can use an escrowed Revocation Signature. See for more details. The remainder of this section describes how some implementations attempt to interpret this deprecated subpacket. This packet was intended to authorize the specified key to issue revocation signatures for this key. Class octet must have bit 0x80 set. If the bit 0x40 is set, then this means that the revocation information is sensitive. Other bits are for future expansion to other kinds of authorizations. This is only found on a direct key self-signature (type ID 0x1F). The use on other types of self-signatures is unspecified. If the "sensitive" flag is set, the keyholder feels this subpacket contains private trust information that describes a real-world sensitive relationship. If this flag is set, implementations SHOULD NOT export this signature to other users except in cases where the data needs to be available: when the signature is being sent to the designated revoker, or when it is accompanied by a revocation signature from that revoker. Note that it may be appropriate to isolate this subpacket within a separate signature so that it is not combined with other subpackets that need to be exported.
Notation Data (4 octets of flags, 2 octets of name length (M), 2 octets of value length (N), M octets of name data, N octets of value data) This subpacket describes a "notation" on the signature that the issuer wishes to make. The notation has a name and a value, each of which are strings of octets. There may be more than one notation in a signature. Notations can be used for any extension the issuer of the signature cares to make. The "flags" field holds four octets of flags. All undefined flags MUST be zero. Defined flags are as follows: Flag Position Shorthand Description Reference 0x80000000 (first bit of first octet) human-readable Notation value is UTF-8 text. This document Notation names are arbitrary strings encoded in UTF-8. They reside in two namespaces: The IETF namespace and the user namespace. The IETF namespace is registered with IANA. These names MUST NOT contain the "@" character (0x40). This is a tag for the user namespace. Notation Name Data Type Allowed Values Reference         This registry is initially empty. Names in the user namespace consist of a UTF-8 string tag followed by "@" followed by a DNS domain name. Note that the tag MUST NOT contain an "@" character. For example, the "sample" tag used by Example Corporation could be "sample@example.com". Names in a user space are owned and controlled by the owners of that domain. Obviously, it's bad form to create a new name in a DNS space that you don't own. Since the user namespace is in the form of an email address, implementers MAY wish to arrange for that address to reach a person who can be consulted about the use of the named tag. Note that due to UTF-8 encoding, not all valid user space name tags are valid email addresses. If there is a critical notation, the criticality applies to that specific notation and not to notations in general.
Key Server Preferences (N octets of flags) This is a list of one-bit flags that indicate preferences that the key holder has about how the key is handled on a key server. All undefined flags MUST be zero. Flag Shorthand Definition 0x80... No-modify The key holder requests that this key only be modified or updated by the key holder or an administrator of the key server. This is found only on a self-signature.
Preferred Key Server (String) This is a URI of a key server that the key holder prefers be used for updates. Note that keys with multiple User IDs can have a preferred key server for each User ID. Note also that since this is a URI, the key server can actually be a copy of the key retrieved by https, ftp, http, etc.
Primary User ID (1 octet, Boolean) This is a flag in a User ID's self-signature that states whether this User ID is the main User ID for this key. It is reasonable for an implementation to resolve ambiguities in preferences, for example, by referring to the primary User ID. If this flag is absent, its value is zero. If more than one User ID in a key is marked as primary, the implementation may resolve the ambiguity in any way it sees fit, but it is RECOMMENDED that priority be given to the User ID with the most recent self-signature. When appearing on a self-signature on a User ID packet, this subpacket applies only to User ID packets. When appearing on a self-signature on a User Attribute packet, this subpacket applies only to User Attribute packets. That is to say, there are two different and independent "primaries" --- one for User IDs, and one for User Attributes.
Policy URI (String) This subpacket contains a URI of a document that describes the policy under which the signature was issued.
Key Flags (N octets of flags) This subpacket contains a list of binary flags that hold information about a key. It is a string of octets, and an implementation MUST NOT assume a fixed size. This is so it can grow over time. If a list is shorter than an implementation expects, the unstated flags are considered to be zero. The defined flags are as follows: Flag Definition 0x01... This key may be used to make User ID certifications (signature type IDs 0x10-0x13) or direct key signatures (signature type ID 0x1F) over other keys. 0x02... This key may be used to sign data. 0x04... This key may be used to encrypt communications. 0x08... This key may be used to encrypt storage. 0x10... The private component of this key may have been split by a secret-sharing mechanism. 0x20... This key may be used for authentication. 0x80... The private component of this key may be in the possession of more than one person. 0x0004... Reserved (ADSK). 0x0008... Reserved (timestamping). Usage notes: The flags in this packet may appear in self-signatures or in certification signatures. They mean different things depending on who is making the statement --- for example, a certification signature that has the "sign data" flag is stating that the certification is for that use. On the other hand, the "communications encryption" flag in a self-signature is stating a preference that a given key be used for communications. Note however, that it is a thorny issue to determine what is "communications" and what is "storage". This decision is left wholly up to the implementation; the authors of this document do not claim any special wisdom on the issue and realize that accepted opinion may change. The "split key" (0x10) and "group key" (0x80) flags are placed on a self-signature only; they are meaningless on a certification signature. They SHOULD be placed only on a direct key signature (type ID 0x1F) or a subkey signature (type ID 0x18), one that refers to the key the flag applies to. When an implementation generates this subpacket, it SHOULD be marked as critical.
Signer's User ID (String) This subpacket allows a keyholder to state which User ID is responsible for the signing. Many keyholders use a single key for different purposes, such as business communications as well as personal communications. This subpacket allows such a keyholder to state which of their roles is making a signature. This subpacket is not appropriate to use to refer to a User Attribute packet.
Reason for Revocation (1 octet of revocation code, N octets of reason string) This subpacket is used only in key revocation and certification revocation signatures. It describes the reason why the key or certification was revoked. The first octet contains a machine-readable code that denotes the reason for the revocation: Code Reason 0 No reason specified (key revocations or cert revocations) 1 Key is superseded (key revocations) 2 Key material has been compromised (key revocations) 3 Key is retired and no longer used (key revocations) 32 User ID information is no longer valid (cert revocations) 100-110 Private Use Following the revocation code is a string of octets that gives information about the Reason for Revocation in human-readable form (UTF-8). The string may be null (of zero length). The length of the subpacket is the length of the reason string plus one. An implementation SHOULD implement this subpacket, include it in all revocation signatures, and interpret revocations appropriately. There are important semantic differences between the reasons, and there are thus important reasons for revoking signatures. If a key has been revoked because of a compromise, all signatures created by that key are suspect. However, if it was merely superseded or retired, old signatures are still valid. If the revoked signature is the self-signature for certifying a User ID, a revocation denotes that that user name is no longer in use. Such a signature revocation SHOULD include a Reason for Revocation subpacket containing code 32. Note that any signature may be revoked, including a certification on some other person's key. There are many good reasons for revoking a certification signature, such as the case where the keyholder leaves the employ of a business with an email address. A revoked certification is no longer a part of validity calculations.
Features (N octets of flags) The Features subpacket denotes which advanced OpenPGP features a user's implementation supports. This is so that as features are added to OpenPGP that cannot be backwards-compatible, a user can state that they can use that feature. The flags are single bits that indicate that a given feature is supported. This subpacket is similar to a preferences subpacket, and only appears in a self-signature. An implementation SHOULD NOT use a feature listed when sending to a user who does not state that they can use it, unless the implementation can infer support for the feature from another implementation-dependent mechanism. Defined features are as follows: First octet: Feature Definition Reference 0x01... Symmetrically Encrypted Integrity Protected Data packet version 1 0x02... Reserved   0x04... Reserved   0x08... Symmetrically Encrypted Integrity Protected Data packet version 2 If an implementation implements any of the defined features, it SHOULD implement the Features subpacket, too. See for details about how to use the Features subpacket when generating encryption data.
Signature Target (1 octet public-key algorithm, 1 octet hash algorithm, N octets hash) This subpacket identifies a specific target signature to which a signature refers. For revocation signatures, this subpacket provides explicit designation of which signature is being revoked. For a third-party or timestamp signature, this designates what signature is signed. All arguments are an identifier of that target signature. The N octets of hash data MUST be the size of the hash of the signature. For example, a target signature with a SHA-1 hash MUST have 20 octets of hash data.
Embedded Signature (1 signature packet body) This subpacket contains a complete Signature packet body as specified in . It is useful when one signature needs to refer to, or be incorporated in, another signature.
Issuer Fingerprint (1 octet key version number, N octets of fingerprint) The OpenPGP Key fingerprint of the key issuing the signature. This subpacket SHOULD be included in all signatures. If the version of the issuing key is 4 and an Issuer Key ID subpacket () is also included in the signature, the key ID of the Issuer Key ID subpacket MUST match the low 64 bits of the fingerprint. Note that the length N of the fingerprint for a version 4 key is 20 octets; for a version 6 key N is 32. Since the version of the signature is bound to the version of the key, the version octet here MUST match the version of the signature. If the version octet does not match the signature version, the receiving implementation MUST treat it as a malformed signature (see ).
Intended Recipient Fingerprint (1 octet key version number, N octets of fingerprint) The OpenPGP Key fingerprint of the intended recipient primary key. If one or more subpackets of this type are included in a signature, it SHOULD be considered valid only in an encrypted context, where the key it was encrypted to is one of the indicated primary keys, or one of their subkeys. This can be used to prevent forwarding a signature outside of its intended, encrypted context (see ). Note that the length N of the fingerprint for a version 4 key is 20 octets; for a version 6 key N is 32. An implementation SHOULD generate this subpacket when creating a signed and encrypted message. When generating this subpacket in a v6 signature, it SHOULD be marked as critical.
Computing Signatures All signatures are formed by producing a hash over the signature data, and then using the resulting hash in the signature algorithm. When creating or verifying a v6 signature, the salt is fed into the hash context before any other data. For binary document signatures (type ID 0x00), the document data is hashed directly. For text document signatures (type ID 0x01), the implementation MUST first canonicalize the document by converting line endings to <CR><LF> and encoding it in UTF-8 (see ). The resulting UTF-8 bytestream is hashed. When a v4 signature is made over a key, the hash data starts with the octet 0x99, followed by a two-octet length of the key, and then the body of the key packet. When a v6 signature is made over a key, the hash data starts with the salt, then octet 0x9B, followed by a four-octet length of the key, and then the body of the key packet. A subkey binding signature (type ID 0x18) or primary key binding signature (type ID 0x19) then hashes the subkey using the same format as the main key (also using 0x99 or 0x9B as the first octet). Primary key revocation signatures (type ID 0x20) hash only the key being revoked. Subkey revocation signature (type ID 0x28) hash first the primary key and then the subkey being revoked. A certification signature (type ID 0x10 through 0x13) hashes the User ID being bound to the key into the hash context after the above data. A v3 certification hashes the contents of the User ID or User Attribute packet, without the packet header. A v4 or v6 certification hashes the constant 0xB4 for User ID certifications or the constant 0xD1 for User Attribute certifications, followed by a four-octet number giving the length of the User ID or User Attribute data, and then the User ID or User Attribute data. When a signature is made over a Signature packet (type ID 0x50, "Third-Party Confirmation signature"), the hash data starts with the salt (v6 signatures only), followed by the octet 0x88, followed by the four-octet length of the signature, and then the body of the Signature packet. (Note that this is a Legacy packet header for a Signature packet with the length-of-length field set to zero.) The unhashed subpacket data of the Signature packet being hashed is not included in the hash, and the unhashed subpacket data length value is set to zero. Once the data body is hashed, then a trailer is hashed. This trailer depends on the version of the signature. A v3 signature hashes five octets of the packet body, starting from the signature type field. This data is the signature type, followed by the four-octet signature creation time. A v4 or v6 signature hashes the packet body starting from its first field, the version number, through the end of the hashed subpacket data and a final extra trailer. Thus, the hashed fields are: An octet indicating the signature version (0x04 for v4, 0x06 for v6), The signature type, The public-key algorithm, The hash algorithm, The hashed subpacket length, The hashed subpacket body, A second version octet (0x04 for v4, 0x06 for v6) A single octet 0xFF, A number representing the length (in octets) of the hashed data from the Signature packet through the hashed subpacket body. This a four-octet big-endian unsigned integer of the length modulo 2**32. After all this has been hashed in a single hash context, the resulting hash field is used in the signature algorithm and its first two octets are placed in the Signature packet, as described in . For worked examples of the data hashed during a signature, see .
Notes About Signature Computation The data actually hashed by OpenPGP varies depending on signature version, in order to ensure that a signature made using one version cannot be repurposed as a signature with a different version over subtly different data. The hashed data streams differ based on their trailer, most critically in the fifth and sixth octets from the end of the stream. In particular: A v3 signature uses the fifth octet from the end to store its signature type ID. This MUST NOT be signature type ID 0xFF. All signature versions later than v3 always use a literal 0xFF in the fifth octet from the end. For these later signature versions, the sixth octet from the end (the octet before the 0xFF) stores the signature version number.
Malformed and Unknown Signatures In some cases, a signature packet (or its corresponding One-Pass Signature packet, see ) may be malformed or unknown. For example, it might encounter any of the following problems (this is not an exhaustive list): An unknown signature type An unknown signature version An unsupported signature version An unknown "critical" subpacket (see ) in the hashed area A subpacket with a length that diverges from the expected length A hashed subpacket area with length that exceeds the length of the signature packet itself A known-weak hash algorithm (e.g. MD5) A mismatch between the hash algorithm expected salt length and the actual salt length A mismatch between the One-Pass Signature version and the Signature version (see ) A signature with a version other than 6, made by a v6 key When an implementation encounters such a malformed or unknown signature, it MUST ignore the signature for validation purposes. It MUST NOT indicate a successful signature validation for such a signature. At the same time, it MUST NOT halt processing on the packet stream or reject other signatures in the same packet stream just because an unknown or invalid signature exists. This requirement is necessary for forward-compatibility. Producing an output that indicates that no successful signatures were found is preferable to aborting processing entirely.
Symmetric-Key Encrypted Session Key Packet (Type ID 3) The Symmetric-Key Encrypted Session Key (SKESK) packet holds the symmetric-key encryption of a session key used to encrypt a message. Zero or more Public-Key Encrypted Session Key packets () and/or Symmetric-Key Encrypted Session Key packets precede an encryption container (that is, a Symmetrically Encrypted Integrity Protected Data packet or --- for historic data --- a Symmetrically Encrypted Data packet) that holds an encrypted message. The message is encrypted with a session key, and the session key is itself encrypted and stored in the Encrypted Session Key packet(s). If the encryption container is preceded by one or more Symmetric-Key Encrypted Session Key packets, each specifies a passphrase that may be used to decrypt the message. This allows a message to be encrypted to a number of public keys, and also to one or more passphrases. The body of this packet starts with a one-octet number giving the version number of the packet type. The currently defined versions are 4 and 6. The remainder of the packet depends on the version. The versions differ in how they encrypt the session key with the passphrase, and in what they encode. The version of the SKESK packet must align with the version of the SEIPD packet (see ). Any new version of the SKESK packet should be registered in the registry established in .
Version 4 Symmetric-Key Encrypted Session Key Packet Format A version 4 Symmetric-Key Encrypted Session Key (SKESK) packet precedes a version 1 Symmetrically Encrypted Integrity Protected Data (v1 SEIPD, see ) packet. In historic data, it is sometimes found preceding a deprecated Symmetrically Encrypted Data packet (SED, see ). A v4 SKESK packet MUST NOT precede a v2 SEIPD packet (see ). A version 4 Symmetric-Key Encrypted Session Key packet consists of: A one-octet version number with value 4. A one-octet number describing the symmetric algorithm used. A string-to-key (S2K) specifier. The length of the string-to-key specifier depends on its type (see ). Optionally, the encrypted session key itself, which is decrypted with the string-to-key object. If the encrypted session key is not present (which can be detected on the basis of packet length and S2K specifier size), then the S2K algorithm applied to the passphrase produces the session key for decrypting the message, using the symmetric cipher algorithm from the Symmetric-Key Encrypted Session Key packet. If the encrypted session key is present, the result of applying the S2K algorithm to the passphrase is used to decrypt just that encrypted session key field, using CFB mode with an IV of all zeros. The decryption result consists of a one-octet algorithm identifier that specifies the symmetric-key encryption algorithm used to encrypt the following encryption container, followed by the session key octets themselves. Note: because an all-zero IV is used for this decryption, the S2K specifier MUST use a salt value, either a Salted S2K, an Iterated-Salted S2K, or Argon2. The salt value will ensure that the decryption key is not repeated even if the passphrase is reused.
Version 6 Symmetric-Key Encrypted Session Key Packet Format A version 6 Symmetric-Key Encrypted Session Key (SKESK) packet precedes a version 2 Symmetrically Encrypted Integrity Protected Data (v2 SEIPD, see ) packet. A v6 SKESK packet MUST NOT precede a v1 SEIPD packet or a deprecated Symmetrically Encrypted Data packet (see ). A version 6 Symmetric-Key Encrypted Session Key packet consists of: A one-octet version number with value 6. A one-octet scalar octet count for the 5 fields following this octet. A one-octet symmetric cipher algorithm ID from . A one-octet AEAD algorithm identifier from . A one-octet scalar octet count of the following field. A string-to-key (S2K) specifier. The length of the string-to-key specifier depends on its type (see ). A starting initialization vector of size specified by the AEAD algorithm. The encrypted session key itself. An authentication tag for the AEAD mode. A key-encryption key is derived using HKDF () with SHA256 () as the hash algorithm. The Initial Keying Material (IKM) for HKDF is the key derived from S2K. No salt is used. The info parameter is comprised of the Packet Type ID in OpenPGP format encoding (bits 7 and 6 set, bits 5-0 carry the packet type ID), the packet version, and the cipher-algo and AEAD-mode used to encrypt the key material. Then, the session key is encrypted using the resulting key, with the AEAD algorithm specified for version 2 of the Symmetrically Encrypted Integrity Protected Data packet. Note that no chunks are used and that there is only one authentication tag. The Packet Type ID encoded in OpenPGP format (bits 7 and 6 set, bits 5-0 carry the packet type ID), the packet version number, the cipher algorithm ID, and the AEAD algorithm ID are given as additional data. For example, the additional data used with AES-128 with OCB consists of the octets 0xC3, 0x06, 0x07, and 0x02.
One-Pass Signature Packet (Type ID 4) The One-Pass Signature packet precedes the signed data and contains enough information to allow the receiver to begin calculating any hashes needed to verify the signature. It allows the Signature packet to be placed at the end of the message, so that the signer can compute the entire signed message in one pass. The body of this packet consists of: A one-octet version number. The currently defined versions are 3 and 6. Any new One-Pass Signature packet version should be registered in the registry established in . A one-octet signature type ID. Signature types are described in . A one-octet number describing the hash algorithm used. A one-octet number describing the public-key algorithm used. Only for v6 packets, a variable-length field containing: A one-octet salt size. The value MUST match the value defined for the hash algorithm as specified in . The salt; a random value of the specified size. The value MUST match the salt field of the corresponding Signature packet. Only for v3 packets, an eight-octet number holding the Key ID of the signing key. Only for v6 packets, 32 octets of the fingerprint of the signing key. Since a v6 signature can only be made by a v6 key, the length of the fingerprint is fixed. A one-octet number holding a flag showing whether the signature is nested. A zero value indicates that the next packet is another One-Pass Signature packet that describes another signature to be applied to the same message data. When generating a one-pass signature, the OPS packet version MUST correspond to the version of the associated signature packet, except for the historical accident that v4 keys use a v3 one-pass signature packet (there is no v4 OPS). See for the full correspondence of versions between Keys, Signatures, and One-Pass Signatures. Note that if a message contains more than one one-pass signature, then the Signature packets bracket the message; that is, the first Signature packet after the message corresponds to the last one-pass packet and the final Signature packet corresponds to the first one-pass packet.
Key Material Packets A key material packet contains all the information about a public or private key. There are four variants of this packet type, two major versions (versions 4 and 6), and two strongly deprecated versions (versions 2 and 3). Consequently, this section is complex. For historical reasons, versions 1 and 5 of the key packets are unspecified.
Key Packet Variants
Public-Key Packet (Type ID 6) A Public-Key packet starts a series of packets that forms an OpenPGP key (sometimes called an OpenPGP certificate).
Public-Subkey Packet (Type ID 14) A Public-Subkey packet (type ID 14) has exactly the same format as a Public-Key packet, but denotes a subkey. One or more subkeys may be associated with a top-level key. By convention, the top-level key offers certification capability, but does not provide encryption services, while a dedicated subkey provides encryption (see ).
Secret-Key Packet (Type ID 5) A Secret-Key packet contains all the information that is found in a Public-Key packet, including the public-key material, but also includes the secret-key material after all the public-key fields.
Secret-Subkey Packet (Type ID 7) A Secret-Subkey packet (type ID 7) is the subkey analog of the Secret-Key packet and has exactly the same format.
Public-Key Packet Formats There are four versions of key-material packets. The V2 and V3 versions have been deprecated since 1998. The V4 version has been deprecated by this document in 2023. OpenPGP implementations SHOULD create keys with version 6 format. V4 keys are deprecated; an implementation SHOULD NOT generate a v4 key, but SHOULD accept it. V3 keys are deprecated; an implementation MUST NOT generate a v3 key, but MAY accept it. V2 keys are deprecated; an implementation MUST NOT generate a v2 key, but MAY accept it. Any new Key version must be registered in the registry established in .
Version 3 Public Keys V2 keys are identical to v3 keys except for the version number. A version 3 public key or public-subkey packet contains: A one-octet version number (3). A four-octet number denoting the time that the key was created. A two-octet number denoting the time in days that this key is valid. If this number is zero, then it does not expire. A one-octet number denoting the public-key algorithm of this key. A series of multiprecision integers comprising the key material: A multiprecision integer (MPI) of RSA public modulus n; An MPI of RSA public encryption exponent e. V3 keys are deprecated. They contain three weaknesses. First, it is relatively easy to construct a v3 key that has the same Key ID as any other key because the Key ID is simply the low 64 bits of the public modulus. Secondly, because the fingerprint of a v3 key hashes the key material, but not its length, there is an increased opportunity for fingerprint collisions. Third, there are weaknesses in the MD5 hash algorithm that make developers prefer other algorithms. See for a fuller discussion of Key IDs and fingerprints.
Version 4 Public Keys The version 4 format is similar to the version 3 format except for the absence of a validity period. This has been moved to the Signature packet. In addition, fingerprints of version 4 keys are calculated differently from version 3 keys, as described in . A version 4 packet contains: A one-octet version number (4). A four-octet number denoting the time that the key was created. A one-octet number denoting the public-key algorithm of this key. A series of values comprising the key material. This is algorithm-specific and described in .
Version 6 Public Keys The version 6 format is similar to the version 4 format except for the addition of a count for the key material. This count helps parsing secret key packets (which are an extension of the public key packet format) in the case of an unknown algorithm. In addition, fingerprints of version 6 keys are calculated differently from version 4 keys, as described in . A version 6 packet contains: A one-octet version number (6). A four-octet number denoting the time that the key was created. A one-octet number denoting the public-key algorithm of this key. A four-octet scalar octet count for the following public key material. A series of values comprising the public key material. This is algorithm-specific and described in .
Secret-Key Packet Formats The Secret-Key and Secret-Subkey packets contain all the data of the Public-Key and Public-Subkey packets, with additional algorithm-specific secret-key data appended, usually in encrypted form. The packet contains: The fields of a Public-Key or Public-Subkey packet, as described above. One octet (the "S2K usage octet") indicating whether and how the secret key material is protected by a passphrase. Zero indicates that the secret-key data is not encrypted. 255 (MalleableCFB), 254 (CFB), or 253 (AEAD) indicates that a string-to-key specifier and other parameters will follow. Any other value is a symmetric-key encryption algorithm identifier. A version 6 packet MUST NOT use the value 255 (MalleableCFB). Only for a version 6 packet where the secret key material is encrypted (that is, where the previous octet is not zero), a one-octet scalar octet count of the cumulative length of all the following conditionally included string-to-key parameter fields. Conditionally included string-to-key parameter fields: If string-to-key usage octet was 255, 254, or 253, a one-octet symmetric encryption algorithm. If string-to-key usage octet was 253 (AEAD), a one-octet AEAD algorithm. Only for a version 6 packet, and if string-to-key usage octet was 254, or 253, a one-octet count of the size of the one field following this octet. If string-to-key usage octet was 255, 254, or 253, a string-to-key (S2K) specifier. The length of the string-to-key specifier depends on its type (see ). If string-to-key usage octet was 253 (AEAD), an initialization vector (IV) of size specified by the AEAD algorithm (see ), which is used as the nonce for the AEAD algorithm. If string-to-key usage octet was 255, 254, or a cipher algorithm ID (that is, the secret data uses some form of CFB encryption), an initialization vector (IV) of the same length as the cipher's block size. Plain or encrypted multiprecision integers comprising the secret key data. This is algorithm-specific and described in . If the string-to-key usage octet is 253 (AEAD), then an AEAD authentication tag is at the end of that data. If the string-to-key usage octet is 254 (CFB), a 20-octet SHA-1 hash of the plaintext of the algorithm-specific portion is appended to plaintext and encrypted with it. If the string-to-key usage octet is 255 (MalleableCFB) or another nonzero value (that is, a symmetric-key encryption algorithm identifier), a two-octet checksum of the plaintext of the algorithm-specific portion (sum of all octets, mod 65536) is appended to plaintext and encrypted with it. (This is deprecated and SHOULD NOT be used, see below.) Only for a version 3 or 4 packet where the string-to-key usage octet is zero, a two-octet checksum of the algorithm-specific portion (sum of all octets, mod 65536). The details about storing algorithm-specific secrets above are summarized in . Note that the version 6 packet format adds two count values to help parsing packets with unknown S2K or public key algorithms. Secret MPI values can be encrypted using a passphrase. If a string-to-key specifier is given, that describes the algorithm for converting the passphrase to a key, else a simple MD5 hash of the passphrase is used. An implementation producing a passphrase-protected secret key packet MUST use a string-to-key specifier; the simple hash is for read-only backward compatibility, though implementations MAY continue to use existing private keys in the old format. The cipher for encrypting the MPIs is specified in the Secret-Key packet. Encryption/decryption of the secret data is done using the key created from the passphrase and the initialization vector from the packet. If the string-to-key usage octet is not 253, CFB mode is used. A different mode is used with v3 keys (which are only RSA) than with other key formats. With v3 keys, the MPI bit count prefix (that is, the first two octets) is not encrypted. Only the MPI non-prefix data is encrypted. Furthermore, the CFB state is resynchronized at the beginning of each new MPI value, so that the CFB block boundary is aligned with the start of the MPI data. With v4 and v6 keys, a simpler method is used. All secret MPI values are encrypted, including the MPI bitcount prefix. If the string-to-key usage octet is 253, the key encryption key is derived using HKDF () to provide key separation. SHA256 () is used as the hash algorithm for HKDF. The Initial Keying Material (IKM) for HKDF is the key derived from S2K. No salt is used. The info parameter is comprised of the Packet Type ID encoded in OpenPGP format (bits 7 and 6 set, bits 5-0 carry the packet type ID), the packet version, and the cipher-algo and AEAD-mode used to encrypt the key material. Then, the encrypted MPI values are encrypted as one combined plaintext using one of the AEAD algorithms specified for version 2 of the Symmetrically Encrypted Integrity Protected Data packet. Note that no chunks are used and that there is only one authentication tag. As additional data, the Packet Type ID in OpenPGP format encoding (bits 7 and 6 set, bits 5-0 carry the packet type ID), followed by the public key packet fields, starting with the packet version number, are passed to the AEAD algorithm. For example, the additional data used with a Secret-Key packet of version 4 consists of the octets 0xC5, 0x04, followed by four octets of creation time, one octet denoting the public-key algorithm, and the algorithm-specific public-key parameters. For a Secret-Subkey packet, the first octet would be 0xC7. For a version 6 key packet, the second octet would be 0x06, and the four-octet octet count of the public key material would be included as well (see ). The two-octet checksum that follows the algorithm-specific portion is the algebraic sum, mod 65536, of the plaintext of all the algorithm-specific octets (including MPI prefix and data). With v3 keys, the checksum is stored in the clear. With v4 keys, the checksum is encrypted like the algorithm-specific data. This value is used to check that the passphrase was correct. However, this checksum is deprecated; an implementation SHOULD NOT use it, but should rather use the SHA-1 hash denoted with a usage octet of 254. The reason for this is that there are some attacks that involve undetectably modifying the secret key. If the string-to-key usage octet is 253 no checksum or SHA-1 hash is used but the authentication tag of the AEAD algorithm follows. When decrypting the secret key material using any of these schemes (that is, where the usage octet is non-zero), the resulting cleartext octet stream must be well-formed. In particular, an implementation MUST NOT interpret octets beyond the unwrapped cleartext octet stream as part of any of the unwrapped MPI objects. Furthermore, an implementation MUST reject as unusable any secret key material whose cleartext length does not align with the lengths of the unwrapped MPI objects.
Key IDs and Fingerprints Every OpenPGP key has a fingerprint and a key ID. The computation of these values differs based on the key version. The fingerprint length varies with the key version, but the key ID (which is only used in v3 PKESK packets, see ) is always 64 bits. The following registry represents the subsections below: Key Version Fingerprint Fingerprint Length (bits) Key ID Reference 3 MD5(MPIs without length octets) 128 low 64 bits of RSA modulus 4 SHA1(normalized pubkey packet) 160 last 64 bits of fingerprint 6 SHA256(normalized pubkey packet) 256 first 64 bits of fingerprint
Version 3 Key ID and Fingerprint For a v3 key, the eight-octet Key ID consists of the low 64 bits of the public modulus of the RSA key. The fingerprint of a v3 key is formed by hashing the body (but not the two-octet length) of the MPIs that form the key material (public modulus n, followed by exponent e) with MD5. Note that both v3 keys and MD5 are deprecated.
Version 4 Key ID and Fingerprint A v4 fingerprint is the 160-bit SHA-1 hash of the octet 0x99, followed by the two-octet packet length, followed by the entire Public-Key packet starting with the version field. The Key ID is the low-order 64 bits of the fingerprint. Here are the fields of the hash material, with the example of an Ed25519 key: a.1) 0x99 (1 octet) a.2) two-octet, big-endian scalar octet count of (b)-(e) b) version number = 4 (1 octet); c) timestamp of key creation (4 octets); d) algorithm (1 octet): 27 = Ed25519 (example); e) Algorithm-specific fields. Algorithm-Specific Fields for Ed25519 keys (example): e.1) 32 octets representing the public key.
Version 6 Key ID and Fingerprint A v6 fingerprint is the 256-bit SHA2-256 hash of the octet 0x9B, followed by the four-octet packet length, followed by the entire Public-Key packet starting with the version field. The Key ID is the high-order 64 bits of the fingerprint. Here are the fields of the hash material, with the example of an Ed25519 key: a.1) 0x9B (1 octet) a.2) four-octet scalar octet count of (b)-(f) b) version number = 6 (1 octet); c) timestamp of key creation (4 octets); d) algorithm (1 octet): 27 = Ed25519 (example); e) four-octet scalar octet count for the following key material; f) algorithm-specific fields. Algorithm-Specific Fields for Ed25519 keys (example): e.1) 32 octets representing the public key. Note that it is possible for there to be collisions of Key IDs --- two different keys with the same Key ID. Note that there is a much smaller, but still non-zero, probability that two different keys have the same fingerprint. Also note that if v3, v4, and v6 format keys share the same RSA key material, they will have different Key IDs as well as different fingerprints. Finally, the Key ID and fingerprint of a subkey are calculated in the same way as for a primary key, including the 0x99 (v4 key) or 0x9B (v6 key) as the first octet (even though this is not a valid packet type ID for a public subkey).
Algorithm-specific Parts of Keys The public and secret key format specifies algorithm-specific parts of a key. The following sections describe them in detail.
Algorithm-Specific Part for RSA Keys The public key is this series of multiprecision integers: MPI of RSA public modulus n; MPI of RSA public encryption exponent e. The secret key is this series of multiprecision integers: MPI of RSA secret exponent d; MPI of RSA secret prime value p; MPI of RSA secret prime value q (p < q); MPI of u, the multiplicative inverse of p, mod q.
Algorithm-Specific Part for DSA Keys The public key is this series of multiprecision integers: MPI of DSA prime p; MPI of DSA group order q (q is a prime divisor of p-1); MPI of DSA group generator g; MPI of DSA public-key value y (= g**x mod p where x is secret). The secret key is this single multiprecision integer: MPI of DSA secret exponent x.
Algorithm-Specific Part for Elgamal Keys The public key is this series of multiprecision integers: MPI of Elgamal prime p; MPI of Elgamal group generator g; MPI of Elgamal public key value y (= g**x mod p where x is secret). The secret key is this single multiprecision integer: MPI of Elgamal secret exponent x.
Algorithm-Specific Part for ECDSA Keys The public key is this series of values: A variable-length field containing a curve OID, which is formatted as follows: A one-octet size of the following field; values 0 and 0xFF are reserved for future extensions, The octets representing a curve OID (defined in ); MPI of an EC point representing a public key. The secret key is this single multiprecision integer: MPI of an integer representing the secret key, which is a scalar of the public EC point.
Algorithm-Specific Part for EdDSALegacy Keys (deprecated) The public key is this series of values: A variable-length field containing a curve OID, formatted as follows: A one-octet size of the following field; values 0 and 0xFF are reserved for future extensions, The octets representing a curve OID, defined in ; An MPI of an EC point representing a public key Q in prefixed native form (see ). The secret key is this single multiprecision integer: An MPI-encoded octet string representing the native form of the secret key, in the curve-specific format described in . Note that the native form for an EdDSA secret key is a fixed-width sequence of unstructured random octets, with size corresponding to the specific curve. That sequence of random octets is used with a cryptographic digest to produce both a curve-specific secret scalar and a prefix used when making a signature. See for more details about how to use the native octet strings for Ed25519Legacy. The value stored in an OpenPGP EdDSALegacy secret key packet is the original sequence of random octets. Note that the only curve defined for use with EdDSALegacy is the Ed25519Legacy OID.
Algorithm-Specific Part for ECDH Keys The public key is this series of values: A variable-length field containing a curve OID, which is formatted as follows: A one-octet size of the following field; values 0 and 0xFF are reserved for future extensions, Octets representing a curve OID, defined in ; MPI of an EC point representing a public key, in the point format associated with the curve as specified in . A variable-length field containing KDF parameters, which is formatted as follows: A one-octet size of the following fields; values 0 and 0xFF are reserved for future extensions, A one-octet value 1, reserved for future extensions, A one-octet hash function ID used with a KDF, A one-octet algorithm ID for the symmetric algorithm used to wrap the symmetric key used for the message encryption; see for details. The secret key is this single multiprecision integer: An MPI representing the secret key, in the curve-specific format described in .
ECDH Secret Key Material When curve NIST P-256, NIST P-384, NIST P-521, brainpoolP256r1, brainpoolP384r1, or brainpoolP512r1 are used in ECDH, their secret keys are represented as a simple integer in standard MPI form. Other curves are presented on the wire differently (though still as a single MPI), as described below and in .
Curve25519Legacy ECDH Secret Key Material (deprecated) A Curve25519Legacy secret key is stored as a standard integer in big-endian MPI form. Curve25519Legacy MUST NOT be used in key packets version 6 or above. Note that this form is in reverse octet order from the little-endian "native" form found in . Note also that the integer for a Curve25519Legacy secret key for OpenPGP MUST have the appropriate form: that is, it MUST be divisible by 8, MUST be at least 2**254, and MUST be less than 2**255. The length of this MPI in bits is by definition always 255, so the two leading octets of the MPI will always be 00 FF and reversing the following 32 octets from the wire will produce the "native" form. When generating a new Curve25519Legacy secret key from 32 fully-random octets, the following pseudocode produces the MPI wire format (note the similarity to decodeScalar25519 from ):
Algorithm-Specific Part for X25519 Keys The public key is this single value: 32 octets of the native public key. The secret key is this single value: 32 octets of the native secret key. See for more details about how to use the native octet strings. The value stored in an OpenPGP X25519 secret key packet is the original sequence of random octets. The value stored in an OpenPGP X25519 public key packet is the value X25519(secretKey, 9).
Algorithm-Specific Part for X448 Keys The public key is this single value: 56 octets of the native public key. The secret key is this single value: 56 octets of the native secret key. See for more details about how to use the native octet strings. The value stored in an OpenPGP X448 secret key packet is the original sequence of random octets. The value stored in an OpenPGP X448 public key packet is the value X448(secretKey, 5).
Algorithm-Specific Part for Ed25519 Keys The public key is this single value: 32 octets of the native public key. The secret key is this single value: 32 octets of the native secret key. See for more details about how to use the native octet strings. The value stored in an OpenPGP Ed25519 secret key packet is the original sequence of random octets.
Algorithm-Specific Part for Ed448 Keys The public key is this single value: 57 octets of the native public key. The secret key is this single value: 57 octets of the native secret key. See for more details about how to use the native octet strings. The value stored in an OpenPGP Ed448 secret key packet is the original sequence of random octets.
Compressed Data Packet (Type ID 8) The Compressed Data packet contains compressed data. Typically, this packet is found as the contents of an encrypted packet, or following a Signature or One-Pass Signature packet, and contains a literal data packet. The body of this packet consists of: One octet that gives the algorithm used to compress the packet. Compressed data, which makes up the remainder of the packet. A Compressed Data packet's body contains data that is a compression of a series of OpenPGP packets. See for details on how messages are formed. A ZIP-compressed series of packets is compressed into raw DEFLATE blocks. A ZLIB-compressed series of packets is compressed with raw ZLIB-style blocks. A BZip2-compressed series of packets is compressed using the BZip2 algorithm. An implementation that generates a Compressed Data packet MUST use the non-legacy format for packet framing (see ). It MUST NOT generate a Compressed Data packet with Legacy format () An implementation that deals with either historic data or data generated by legacy implementations predating support for MAY interpret Compressed Data packets that use the Legacy format for packet framing.
Symmetrically Encrypted Data Packet (Type ID 9) The Symmetrically Encrypted Data packet contains data encrypted with a symmetric-key algorithm. When it has been decrypted, it contains other packets (usually a literal data packet or compressed data packet, but in theory other Symmetrically Encrypted Data packets or sequences of packets that form whole OpenPGP messages). This packet is obsolete. An implementation MUST NOT create this packet. An implementation SHOULD reject such a packet and stop processing the message. If an implementation chooses to process the packet anyway, it MUST return a clear warning that a non-integrity protected packet has been processed. This packet format is impossible to handle safely in general because the ciphertext it provides is malleable. See about selecting a better OpenPGP encryption container that does not have this flaw. The body of this packet consists of: A random prefix, containing block-size random octets (for example, 16 octets for a 128-bit block length) followed by a copy of the last two octets, encrypted together using Cipher Feedback (CFB) mode, with an Initial Vector (IV) of all zeros. Data encrypted using CFB mode, with the last block-size octets of the first ciphertext as the IV. The symmetric cipher used may be specified in a Public-Key or Symmetric-Key Encrypted Session Key packet that precedes the Symmetrically Encrypted Data packet. In that case, the cipher algorithm ID is prefixed to the session key before it is encrypted. If no packets of these types precede the encrypted data, the IDEA algorithm is used with the session key calculated as the MD5 hash of the passphrase, though this use is deprecated. The data is encrypted in CFB mode (see ). For the random prefix, the Initial Vector (IV) is specified as all zeros. Instead of achieving randomized encryption through an IV, a string of length equal to the block size of the cipher plus two is encrypted for this purpose. The first block-size octets (for example, 16 octets for a 128-bit block length) are random, and the following two octets are copies of the last two octets of the first block-size random octets. For example, for a 16-octet block length, octet 17 is a copy of octet 15 and octet 18 is a copy of octet 16. For a cipher of block length 8, octet 9 is a copy of octet 7, and octet 10 is a copy of octet 8. (In both these examples, we consider the first octet to be numbered 1.) After encrypting these block-size-plus-two octets, a new CFB context is created for the encryption of the data, with the last block-size octets of the first ciphertext as the IV. (Alternatively and equivalently, the CFB state is resynchronized: the last block-size octets of ciphertext are passed through the cipher and the block boundary is reset.) The repetition of two octets in the random prefix allows the receiver to immediately check whether the session key is incorrect. See for hints on the proper use of this "quick check".
Marker Packet (Type ID 10) The body of this packet consists of: The three octets 0x50, 0x47, 0x50 (which spell "PGP" in UTF-8). Such a packet MUST be ignored when received.
Literal Data Packet (Type ID 11) A Literal Data packet contains the body of a message; data that is not to be further interpreted. The body of this packet consists of: A one-octet field that describes how the data is formatted. If it is a b (0x62), then the Literal packet contains binary data. If it is a u (0x75), then the Literal packet contains UTF-8-encoded text data, and thus may need line ends converted to local form, or other text mode changes. Older versions of OpenPGP used t (0x74) to indicate textual data, but did not specify the character encoding. Implementations SHOULD NOT emit this value. An implementation that receives a literal data packet with this value in the format field SHOULD interpret the packet data as UTF-8 encoded text, unless reliable (not attacker-controlled) context indicates a specific alternate text encoding. This mode is deprecated due to its ambiguity. Some implementations predating also defined a value of l as a 'local' mode for machine-local conversions. incorrectly stated this local mode flag as 1 (ASCII numeral one). Both of these local modes are deprecated. File name as a string (one-octet length, followed by a file name). This may be a zero-length string. Commonly, if the source of the encrypted data is a file, this will be the name of the encrypted file. An implementation MAY consider the file name in the Literal packet to be a more authoritative name than the actual file name. A four-octet number that indicates a date associated with the literal data. Commonly, the date might be the modification date of a file, or the time the packet was created, or a zero that indicates no specific time. The remainder of the packet is literal data. Text data MUST be encoded with UTF-8 (see ) and stored with <CR><LF> text endings (that is, network-normal line endings). These should be converted to native line endings by the receiving implementation. Note that OpenPGP signatures do not include the formatting octet, the file name, and the date field of the literal packet in a signature hash and thus those fields are not protected against tampering in a signed document. A receiving implementation MUST NOT treat those fields as though they were cryptographically secured by the surrounding signature either when representing them to the user or acting on them. Due to their inherent malleability, an implementation that generates a literal data packet SHOULD avoid storing any significant data in these fields. If the implementation is certain that the data is textual and is encoded with UTF-8 (for example, if it will follow this literal data packet with a signature packet of type 0x01 (see ), it MAY set the format octet to u. Otherwise, it MUST set the format octet to b. It SHOULD set the filename to the empty string (encoded as a single zero octet), and the timestamp to zero (encoded as four zero octets). An application that wishes to include such filesystem metadata within a signature is advised to sign an encapsulated archive (for example, ). An implementation that generates a Literal Data packet MUST use the OpenPGP format for packet framing (see ). It MUST NOT generate a Literal Data packet with Legacy format () An implementation that deals with either historic data or data generated by an implementation that predates support for MAY interpret Literal Data packets that use the Legacy format for packet framing.
Special Filename _CONSOLE (Deprecated) The Literal Data packet's filename field has a historical special case for the special name _CONSOLE. When the filename field is _CONSOLE, the message is considered to be "for your eyes only". This advises that the message data is unusually sensitive, and the receiving program should process it more carefully, perhaps avoiding storing the received data to disk, for example. An OpenPGP deployment that generates literal data packets MUST NOT depend on this indicator being honored in any particular way. It cannot be enforced, and the field itself is not covered by any cryptographic signature. It is NOT RECOMMENDED to use this special filename in a newly-generated literal data packet.
Trust Packet (Type ID 12) The Trust packet is used only within keyrings and is not normally exported. Trust packets contain data that record the user's specifications of which key holders are trustworthy introducers, along with other information that implementation uses for trust information. The format of Trust packets is defined by a given implementation. Trust packets SHOULD NOT be emitted to output streams that are transferred to other users, and they SHOULD be ignored on any input other than local keyring files.
User ID Packet (Type ID 13) A User ID packet consists of UTF-8 text that is intended to represent the name and email address of the key holder. By convention, it includes an mail name-addr, but there are no restrictions on its content. The packet length in the header specifies the length of the User ID.
User Attribute Packet (Type ID 17) The User Attribute packet is a variation of the User ID packet. It is capable of storing more types of data than the User ID packet, which is limited to text. Like the User ID packet, a User Attribute packet may be certified by the key owner ("self-signed") or any other key owner who cares to certify it. Except as noted, a User Attribute packet may be used anywhere that a User ID packet may be used. While User Attribute packets are not a required part of the OpenPGP standard, implementations SHOULD provide at least enough compatibility to properly handle a certification signature on the User Attribute packet. A simple way to do this is by treating the User Attribute packet as a User ID packet with opaque contents, but an implementation may use any method desired. The User Attribute packet is made up of one or more attribute subpackets. Each subpacket consists of a subpacket header and a body. The header consists of: The subpacket length (1, 2, or 5 octets) The subpacket type ID (1 octet) and is followed by the subpacket specific data. The following table lists the currently known subpackets: ID Attribute Subpacket Reference 1 Image Attribute Subpacket 100-110 Private/Experimental Use   An implementation SHOULD ignore any subpacket of a type that it does not recognize.
The Image Attribute Subpacket The Image Attribute subpacket is used to encode an image, presumably (but not required to be) that of the key owner. The Image Attribute subpacket begins with an image header. The first two octets of the image header contain the length of the image header. Note that unlike other multi-octet numerical values in this document, due to a historical accident this value is encoded as a little-endian number. The image header length is followed by a single octet for the image header version. The only currently defined version of the image header is 1, which is a 16-octet image header. The first three octets of a version 1 image header are thus 0x10, 0x00, 0x01. Version Reference 1 The fourth octet of a version 1 image header designates the encoding format of the image. The only currently defined encoding format is the value 1 to indicate JPEG. Image format IDs 100 through 110 are reserved for private or experimental use. The rest of the version 1 image header is made up of 12 reserved octets, all of which MUST be set to 0. ID Encoding Reference 1 JPEG JPEG File Interchange Format () 100-110 Private/Experimental use   The rest of the image subpacket contains the image itself. As the only currently defined image type is JPEG, the image is encoded in the JPEG File Interchange Format (JFIF), a standard file format for JPEG images . An implementation MAY try to determine the type of an image by examination of the image data if it is unable to handle a particular version of the image header or if a specified encoding format value is not recognized.
Symmetrically Encrypted Integrity Protected Data Packet (Type ID 18) This packet (the "SEIPD" packet) contains integrity protected and encrypted data. When it has been decrypted, it will contain other packets forming an OpenPGP Message (see ). The first octet of this packet is always used to indicate the version number, but different versions contain differently-structured ciphertext. Version 1 of this packet contains data encrypted with a symmetric-key algorithm and protected against modification by the SHA-1 hash algorithm. This mechanism was introduced in and offers some protections against ciphertext malleability. Version 2 of this packet contains data encrypted with an authenticated encryption and additional data (AEAD) construction. This offers a more cryptographically rigorous defense against ciphertext malleability. See for more details on choosing between these formats. Any new version of the SEIPD packet should be registered in the registry established in .
Version 1 Symmetrically Encrypted Integrity Protected Data Packet Format A version 1 Symmetrically Encrypted Integrity Protected Data packet consists of: A one-octet version number with value 1. Encrypted data, the output of the selected symmetric-key cipher operating in Cipher Feedback (CFB) mode. The symmetric cipher used MUST be specified in a Public-Key or Symmetric-Key Encrypted Session Key packet that precedes the Symmetrically Encrypted Integrity Protected Data packet. In either case, the cipher algorithm ID is prefixed to the session key before it is encrypted. The data is encrypted in CFB mode (see ). The Initial Vector (IV) is specified as all zeros. Instead of achieving randomized encryption through an IV, OpenPGP prefixes an octet string to the data before it is encrypted for this purpose. The length of the octet string equals the block size of the cipher in octets, plus two. The first octets in the group, of length equal to the block size of the cipher, are random; the last two octets are each copies of their 2nd preceding octet. For example, with a cipher whose block size is 128 bits or 16 octets, the prefix data will contain 16 random octets, then two more octets, which are copies of the 15th and 16th octets, respectively. Unlike the deprecated Symmetrically Encrypted Data packet (), this prefix data is encrypted in the same CFB context, and no special CFB resynchronization is done. The repetition of 16 bits in the random data prefixed to the message allows the receiver to immediately check whether the session key is incorrect. See for hints on the proper use of this "quick check". Two constant octets with the values 0xD3 and 0x14 are appended to the plaintext. Then, the plaintext of the data to be encrypted is passed through the SHA-1 hash function. The input to the hash function includes the prefix data described above; it includes all of the plaintext, including the trailing constant octets 0xD3, 0x14. The 20 octets of the SHA-1 hash are then appended to the plaintext (after the constant octets 0xD3, 0x14) and encrypted along with the plaintext using the same CFB context. This trailing checksum is known as the Modification Detection Code (MDC). During decryption, the plaintext data should be hashed with SHA-1, including the prefix data as well as the trailing constant octets 0xD3, 0x14, but excluding the last 20 octets containing the SHA-1 hash. The computed SHA-1 hash is then compared with the last 20 octets of plaintext. A mismatch of the hash indicates that the message has been modified and MUST be treated as a security problem. Any failure SHOULD be reported to the user.
  • NON-NORMATIVE EXPLANATION The Modification Detection Code (MDC) system, as the integrity protection mechanism of version 1 of the Symmetrically Encrypted Integrity Protected Data packet is called, was created to provide an integrity mechanism that is less strong than a signature, yet stronger than bare CFB encryption. It is a limitation of CFB encryption that damage to the ciphertext will corrupt the affected cipher blocks and the block following. Additionally, if data is removed from the end of a CFB-encrypted block, that removal is undetectable. (Note also that CBC mode has a similar limitation, but data removed from the front of the block is undetectable.) The obvious way to protect or authenticate an encrypted block is to digitally sign it. However, many people do not wish to habitually sign data, for a large number of reasons beyond the scope of this document. Suffice it to say that many people consider properties such as deniability to be as valuable as integrity. OpenPGP addresses this desire to have more security than raw encryption and yet preserve deniability with the MDC system. An MDC is intentionally not a MAC. Its name was not selected by accident. It is analogous to a checksum. Despite the fact that it is a relatively modest system, it has proved itself in the real world. It is an effective defense to several attacks that have surfaced since it has been created. It has met its modest goals admirably. Consequently, because it is a modest security system, it has modest requirements on the hash function(s) it employs. It does not rely on a hash function being collision-free, it relies on a hash function being one-way. If a forger, Frank, wishes to send Alice a (digitally) unsigned message that says, "I've always secretly loved you, signed Bob", it is far easier for him to construct a new message than it is to modify anything intercepted from Bob. (Note also that if Bob wishes to communicate secretly with Alice, but without authentication or identification and with a threat model that includes forgers, he has a problem that transcends mere cryptography.) Note also that unlike nearly every other OpenPGP subsystem, there are no parameters in the MDC system. It hard-defines SHA-1 as its hash function. This is not an accident. It is an intentional choice to avoid downgrade and cross-grade attacks while making a simple, fast system. (A downgrade attack would be an attack that replaced SHA2-256 with SHA-1, for example. A cross-grade attack would replace SHA-1 with another 160-bit hash, such as RIPEMD-160, for example.) However, no update will be needed because the MDC has been replaced by the AEAD encryption described in this document.
Version 2 Symmetrically Encrypted Integrity Protected Data Packet Format A version 2 Symmetrically Encrypted Integrity Protected Data packet consists of: A one-octet version number with value 2. A one-octet cipher algorithm ID. A one-octet AEAD algorithm identifier. A one-octet chunk size. Thirty-two octets of salt. The salt is used to derive the message key and MUST be securely generated (See ). Encrypted data, the output of the selected symmetric-key cipher operating in the given AEAD mode. A final, summary authentication tag for the AEAD mode. The decrypted session key and the salt are used to derive an M-bit message key and N-64 bits used as initialization vector, where M is the key size of the symmetric algorithm and N is the nonce size of the AEAD algorithm. M + N - 64 bits are derived using HKDF (see ). The left-most M bits are used as symmetric algorithm key, the remaining N - 64 bits are used as initialization vector. HKDF is used with SHA256 () as hash algorithm, the session key as Initial Keying Material (IKM), the salt as salt, and the Packet Type ID in OpenPGP format encoding (bits 7 and 6 set, bits 5-0 carry the packet type ID), version number, cipher algorithm ID, AEAD algorithm ID, and chunk size octet as info parameter. The KDF mechanism provides key separation between cipher and AEAD algorithms. Furthermore, an implementation can securely reply to a message even if a recipient's certificate is unknown by reusing the encrypted session key packets and replying with a different salt yielding a new, unique message key. See for guidance on how applications can securely implement this feature. A v2 SEIPD packet consists of one or more chunks of data. The plaintext of each chunk is of a size specified using the chunk size octet using the method specified below. The encrypted data consists of the encryption of each chunk of plaintext, followed immediately by the relevant authentication tag. If the last chunk of plaintext is smaller than the chunk size, the ciphertext for that data may be shorter; it is nevertheless followed by a full authentication tag. For each chunk, the AEAD construction is given the Packet Type ID encoded in OpenPGP format (bits 7 and 6 set, bits 5-0 carry the packet type ID), version number, cipher algorithm ID, AEAD algorithm ID, and chunk size octet as additional data. For example, the additional data of the first chunk using EAX and AES-128 with a chunk size of 2**22 octets consists of the octets 0xD2, 0x02, 0x07, 0x01, and 0x10. After the final chunk, the AEAD algorithm is used to produce a final authentication tag encrypting the empty string. This AEAD instance is given the additional data specified above, plus an eight-octet, big-endian value specifying the total number of plaintext octets encrypted. This allows detection of a truncated ciphertext. The chunk size octet specifies the size of chunks using the following formula (in ), where c is the chunk size octet:
An implementation MUST accept chunk size octets with values from 0 to 16. An implementation MUST NOT create data with a chunk size octet value larger than 16 (4 MiB chunks). The nonce for AEAD mode consists of two parts. Let N be the size of the nonce. The left-most N - 64 bits are the initialization vector derived using HKDF. The right-most 64 bits are the chunk index as big-endian value. The index of the first chunk is zero.
EAX Mode The EAX AEAD Algorithm used in this document is defined in . The EAX algorithm can only use block ciphers with 16-octet blocks. The nonce is 16 octets long. EAX authentication tags are 16 octets long.
OCB Mode The OCB AEAD Algorithm used in this document is defined in . The OCB algorithm can only use block ciphers with 16-octet blocks. The nonce is 15 octets long. OCB authentication tags are 16 octets long.
GCM Mode The GCM AEAD Algorithm used in this document is defined in . The GCM algorithm can only use block ciphers with 16-octet blocks. The nonce is 12 octets long. GCM authentication tags are 16 octets long.
Padding Packet (Type ID 21) The Padding packet contains random data, and can be used to defend against traffic analysis (see ) on version 2 SEIPD messages (see ) and Transferable Public Keys (see ). Such a packet MUST be ignored when received. Its contents SHOULD be random octets to make the length obfuscation it provides more robust even when compressed. An implementation adding padding to an OpenPGP stream SHOULD place such a packet: At the end of a v6 Transferable Public Key that is transferred over an encrypted channel (see ). As the last packet of an Optionally Padded Message within a version 2 Symmetrically Encrypted Integrity Protected Data packet (see ). An implementation MUST be able to process padding packets anywhere else in an OpenPGP stream, so that future revisions of this document may specify further locations for padding. Policy about how large to make such a packet to defend against traffic analysis is beyond the scope of this document.
Base64 Conversions As stated in the introduction, OpenPGP's underlying native representation for objects is a stream of arbitrary octets, and some systems desire these objects to be immune to damage caused by character set translation, data conversions, etc. In principle, any printable encoding scheme that met the requirements of the unsafe channel would suffice, since it would not change the underlying binary bit streams of the native OpenPGP data structures. The OpenPGP standard specifies one such printable encoding scheme to ensure interoperability, . The encoding is composed of two parts: a base64 encoding of the binary data and an optional checksum. The base64 encoding used is described in , and it is wrapped into lines of no more than 76 characters each. When decoding base64, an OpenPGP implementation MUST ignore all white space.
Optional checksum The optional checksum is a 24-bit Cyclic Redundancy Check (CRC) converted to four characters of base64 encoding by the same MIME base64 transformation, preceded by an equal sign (=). The CRC is computed by using the generator 0x864CFB and an initialization of 0xB704CE. The accumulation is done on the data before it is converted to base64, rather than on the converted data. A sample implementation of this algorithm is in . If present, the checksum with its leading equal sign MUST appear on the next line after the base64 encoded data. An implementation MUST NOT reject an OpenPGP object when the CRC24 footer is present, missing, malformed, or disagrees with the computed CRC24 sum. When forming ASCII Armor, the CRC24 footer SHOULD NOT be generated, unless interoperability with implementations that require the CRC24 footer to be present is a concern. The CRC24 footer MUST NOT be generated if it can be determined by context or by the OpenPGP object being encoded that the consuming implementation accepts base64 encoded blocks without CRC24 footer. Notably: An ASCII-armored Encrypted Message packet sequence that ends in an v2 SEIPD packet MUST NOT contain a CRC24 footer. An ASCII-armored sequence of Signature packets that only includes v6 Signature packets MUST NOT contain a CRC24 footer. An ASCII-armored Transferable Public Key packet sequence of a v6 key MUST NOT contain a CRC24 footer. An ASCII-armored keyring consisting of only v6 keys MUST NOT contain a CRC24 footer. Rationale: Previous versions of this document state that the CRC24 footer is optional, but the text was ambiguous. In practice, very few implementations require the CRC24 footer to be present. Computing the CRC24 incurs a significant cost, while providing no meaningful integrity protection. Therefore, generating it is now discouraged.
An Implementation of the CRC-24 in "C" The following code is written in .
Forming ASCII Armor When OpenPGP encodes data into ASCII Armor, it puts specific headers around the base64 encoded data, so OpenPGP can reconstruct the data later. An OpenPGP implementation MAY use ASCII armor to protect raw binary data. OpenPGP informs the user what kind of data is encoded in the ASCII armor through the use of the headers. Concatenating the following data creates ASCII Armor: An Armor Header Line, appropriate for the type of data Armor Headers A blank (zero-length, or containing only whitespace) line The ASCII-Armored data An optional Armor Checksum (discouraged, see ) The Armor Tail, which depends on the Armor Header Line
Armor Header Line An Armor Header Line consists of the appropriate header line text surrounded by five (5) dashes (-, 0x2D) on either side of the header line text. The header line text is chosen based upon the type of data that is being encoded in Armor, and how it is being encoded. Header line texts include the following strings: Armor Header Use BEGIN PGP MESSAGE Used for signed, encrypted, or compressed files. BEGIN PGP PUBLIC KEY BLOCK Used for armoring public keys. BEGIN PGP PRIVATE KEY BLOCK Used for armoring private keys. BEGIN PGP SIGNATURE Used for detached signatures, OpenPGP/MIME signatures, and cleartext signatures. Note that all these Armor Header Lines are to consist of a complete line. The header lines, therefore, MUST start at the beginning of a line, and MUST NOT have text other than whitespace following them on the same line.
Armor Headers The Armor Headers are pairs of strings that can give the user or the receiving OpenPGP implementation some information about how to decode or use the message. The Armor Headers are a part of the armor, not a part of the message, and hence are not protected by any signatures applied to the message. The format of an Armor Header is that of a key-value pair. A colon (: 0x3A) and a single space (0x20) separate the key and value. An OpenPGP implementation may consider improperly formatted Armor Headers to be corruption of the ASCII Armor, but SHOULD make an effort to recover. Unknown keys should be silently ignored, and an OpenPGP implementation SHOULD continue to process the message. Note that some transport methods are sensitive to line length. While there is a limit of 76 characters for the base64 data (), there is no limit to the length of Armor Headers. Care should be taken that the Armor Headers are short enough to survive transport. One way to do this is to repeat an Armor Header Key multiple times with different values for each so that no one line is overly long. Currently defined Armor Header Keys are as follows: Key Summary Reference Version Implementation information Comment Arbitrary text Hash Hash algorithms used in some v4 cleartext signed messages Charset Character set
"Version" Armor Header The armor header key Version describes the OpenPGP implementation and version used to encode the message. To minimize metadata, implementations SHOULD NOT emit this key and its corresponding value except for debugging purposes with explicit user consent.
"Comment" Armor Header The armor header key Comment supplies a user-defined comment. OpenPGP defines all text to be in UTF-8. A comment may be any UTF-8 string. However, the whole point of armoring is to provide seven-bit-clean data. Consequently, if a comment has characters that are outside the US-ASCII range of UTF, they may very well not survive transport.
"Hash" Armor Header This header is deprecated, but some older implementations expect it in messages using the Cleartext Signature Framework (). When present, The armor header key Hash contains a comma-separated list of hash algorithms used in the signatures on message, with digest names as specified in "Text Name" column in . These headers SHOULD NOT be emitted unless: The cleartext signed message contains a v4 signature made using a SHA2-based digest (SHA224, SHA256, SHA384, or SHA512), and The cleartext signed message might be verified by a legacy OpenPGP implementation that requires this header. A verifying application MUST decline to validate any signature in a message with a non-conformant Hash header (that is, a Hash header that contains anything other than a comma-separated list of hash algorithms). When a conformant Hash header is present, a verifying application MUST ignore its contents, deferring to the hash algorithm indicated in the embedded signature.
"Charset" Armor Header The armor header key Charset contains a description of the character set that the plaintext is in (see ). Please note that OpenPGP defines text to be in UTF-8. An implementation will get best results by translating into and out of UTF-8. However, there are many instances where this is easier said than done. Also, there are communities of users who have no need for UTF-8 because they are all happy with a character set like ISO Latin-5 or a Japanese character set. In such instances, an implementation MAY override the UTF-8 default by using this header key. An implementation MAY implement this key and any translations it cares to; an implementation MAY ignore it and assume all text is UTF-8.
Armor Tail Line The Armor Tail Line is composed in the same manner as the Armor Header Line, except the string "BEGIN" is replaced by the string "END".
Cleartext Signature Framework It is desirable to be able to sign a textual octet stream without ASCII armoring the stream itself, so the signed text is still readable with any tool capable of rendering text. In order to bind a signature to such a cleartext, this framework is used, which follows the same basic format and restrictions as the ASCII armoring described in . (Note that this framework is not intended to be reversible. defines another way to sign cleartext messages for environments that support MIME.)
Cleartext Signed Message Structure An OpenPGP cleartext signed message consists of: The cleartext header -----BEGIN PGP SIGNED MESSAGE----- on a single line, Some implementations MAY include one or more legacy Hash Armor Headers, which MUST be ignored when well-formed (see ), An empty line (not included into the message digest), The dash-escaped cleartext, A line ending separating the cleartext and following armored signature (not included into the message digest), The ASCII armored signature(s) including the -----BEGIN PGP SIGNATURE----- Armor Header and Armor Tail Lines. As with binary signatures on text documents, a cleartext signature is calculated on the text using canonical <CR><LF> line endings. As described above, the line ending before the -----BEGIN PGP SIGNATURE----- Armor Header Line of the armored signature is not considered part of the signed text. Also, any trailing whitespace --- spaces (0x20) and tabs (0x09) --- at the end of any line is removed before signing or verifying a cleartext signed message. Between the -----BEGIN PGP SIGNED MESSAGE----- line and the first empty line, the only Armor Header permitted is a well-formed Hash Armor Header (see ). To reduce the risk of confusion about what has been signed, a verifying implementation MUST decline to validate any signature in a cleartext message if that message has any other Armor Header present in this location.
Dash-Escaped Text The cleartext content of the message must also be dash-escaped. Dash-escaped cleartext is the ordinary cleartext where every line starting with a - is prefixed by the sequence - and . This prevents the parser from recognizing armor headers of the cleartext itself. An implementation MAY dash-escape any line, SHOULD dash-escape lines commencing "From" followed by a space, and MUST dash-escape any line commencing in a dash. The message digest is computed using the cleartext itself, not the dash-escaped form. When reversing dash-escaping, an implementation MUST strip the string - if it occurs at the beginning of a line, and SHOULD warn on - and any character other than a space at the beginning of a line.
Issues with the Cleartext Signature Framework Since creating a cleartext signed message involves trimming trailing whitespace on every line, the Cleartext Signature Framework will fail to safely round-trip any textual stream that may include semantically meaningful whitespace. For example, the Unified Diff format contains semantically meaningful whitespace: an empty line of context will consist of a line with a single character, and any line that has trailing whitespace added or removed will represent such a change with semantically meaningful whitespace. Furthermore, a Cleartext Signature Framework message that is very large is unlikely to work well. In particular, it will be difficult for any human reading the message to know which part of the message is covered by the signature because they can't understand the whole message at once, in case an Armor Header line is placed somewhere in the body. And, very large Cleartext Signature Framework messages cannot be processed in a single pass, since the signature salt and digest algorithms are only discovered at the end. An implementation that knows it is working with a textual stream with any of the above characteristics SHOULD NOT use the Cleartext Signature Framework. Safe alternatives for a semantically meaningful OpenPGP signature over such a file format are: A Signed Message, as described in . A Detached Signature as described in . Either of these alternatives may be ASCII-armored (see ) if they need to be transmitted across a text-only (or 7-bit clean) channel. Finally, when a Cleartext Signature Framework message is presented to the user as-is, an attacker can include additional text in the Hash header, which may mislead the user into thinking it is part of the signed text. The signature validation constraints described in and help to mitigate the risk of arbitrary or misleading text in the Armor Headers.
Regular Expressions A regular expression is zero or more branches, separated by |. It matches anything that matches one of the branches. A branch is zero or more pieces, concatenated. It matches a match for the first, followed by a match for the second, etc. A piece is an atom possibly followed by *, +, or ?. An atom followed by * matches a sequence of 0 or more matches of the atom. An atom followed by + matches a sequence of 1 or more matches of the atom. An atom followed by ? matches a match of the atom, or the null string. An atom is a regular expression in parentheses (matching a match for the regular expression), a range (see below), . (matching any single Unicode character), ^ (matching the null string at the beginning of the input string), $ (matching the null string at the end of the input string), a \ followed by a single Unicode character (matching that character), or a single Unicode character with no other significance (matching that character). A range is a sequence of characters enclosed in []. It normally matches any single character from the sequence. If the sequence begins with ^, it matches any single Unicode character not from the rest of the sequence. If two characters in the sequence are separated by -, this is shorthand for the full list of Unicode characters between them in codepoint order (for example, [0-9] matches any decimal digit). To include a literal ] in the sequence, make it the first character (following a possible ^). To include a literal -, make it the first or last character.
Constants This section describes the constants used in OpenPGP. Note that these tables are not exhaustive lists; an implementation MAY implement an algorithm not on these lists, so long as the algorithm IDs are chosen from the private or experimental algorithm range. See for more discussion of the algorithms.
Public-Key Algorithms ID Algorithm Public Key Format Secret Key Format Signature Format PKESK Format 0 Reserved         1 RSA (Encrypt or Sign) MPI(n), MPI(e) [] MPI(d), MPI(p), MPI(q), MPI(u) MPI(m**d mod n) [] MPI(m**e mod n) [] 2 RSA Encrypt-Only MPI(n), MPI(e) [] MPI(d), MPI(p), MPI(q), MPI(u) N/A MPI(m**e mod n) [] 3 RSA Sign-Only MPI(n), MPI(e) [] MPI(d), MPI(p), MPI(q), MPI(u) MPI(m**d mod n) [] N/A 16 Elgamal (Encrypt-Only) MPI(p), MPI(g), MPI(y) [] MPI(x) N/A MPI(g**k mod p), MPI (m * y**k mod p) [] 17 DSA (Digital Signature Algorithm) MPI(p), MPI(q), MPI(g), MPI(y) [] MPI(x) MPI(r), MPI(s) [] N/A 18 ECDH public key algorithm OID, MPI(point in curve-specific point format), KDFParams [see , ] MPI(value in curve-specific format) [] N/A MPI(point in curve-specific point format), size octet, encoded key [, , ] 19 ECDSA public key algorithm OID, MPI(point in SEC1 format) [] MPI(value) MPI(r), MPI(s) [] N/A 20 Reserved (formerly Elgamal Encrypt or Sign)         21 Reserved for Diffie-Hellman (X9.42, as defined for IETF-S/MIME)         22 EdDSALegacy (deprecated) OID, MPI(point in prefixed native format) [see , ] MPI(value in curve-specific format) [see ] MPI, MPI [see , ] N/A 23 Reserved (AEDH)         24 Reserved (AEDSA)         25 X25519 32 octets [see ] 32 octets N/A 32 octets, size octet, encoded key [see ] 26 X448 56 octets [see ] 56 octets N/A 56 octets, size octet, encoded key [see ] 27 Ed25519 32 octets [see ] 32 octets 64 octets [see ]   28 Ed448 57 octets [see ] 57 octets 114 octets [see ]   100 to 110 Private/Experimental algorithm         Implementations MUST implement Ed25519 (27) for signatures, and X25519 (25) for encryption. Implementations SHOULD implement Ed448 (28) and X448 (26). RSA (1) keys are deprecated and SHOULD NOT be generated, but may be interpreted. RSA Encrypt-Only (2) and RSA Sign-Only (3) are deprecated and MUST NOT be generated. See . Elgamal (16) keys are deprecated and MUST NOT be generated (see ). DSA (17) keys are deprecated and MUST NOT be generated (see ). See for notes on Elgamal Encrypt or Sign (20), and X9.42 (21). Implementations MAY implement any other algorithm. Note that an implementation conforming to the previous version of this standard () has only DSA (17) and Elgamal (16) as its MUST-implement algorithms. A compatible specification of ECDSA is given in as "KT-I Signatures" and in ; ECDH is defined in of this document.
ECC Curves for OpenPGP The parameter curve OID is an array of octets that defines a named curve. The table below specifies the exact sequence of octets for each named curve referenced in this document. It also specifies which public key algorithms the curve can be used with, as well as the size of expected elements in octets: ASN.1 Object Identifier OID len Curve OID octets in hex Curve name Usage Field Size (fsize) 1.2.840.10045.3.1.7 8 2A 86 48 CE 3D 03 01 07 NIST P-256 ECDSA, ECDH 32 1.3.132.0.34 5 2B 81 04 00 22 NIST P-384 ECDSA, ECDH 48 1.3.132.0.35 5 2B 81 04 00 23 NIST P-521 ECDSA, ECDH 66 1.3.36.3.3.2.8.1.1.7 9 2B 24 03 03 02 08 01 01 07 brainpoolP256r1 ECDSA, ECDH 32 1.3.36.3.3.2.8.1.1.11 9 2B 24 03 03 02 08 01 01 0B brainpoolP384r1 ECDSA, ECDH 48 1.3.36.3.3.2.8.1.1.13 9 2B 24 03 03 02 08 01 01 0D brainpoolP512r1 ECDSA, ECDH 64 1.3.6.1.4.1.11591.15.1 9 2B 06 01 04 01 DA 47 0F 01 Ed25519Legacy EdDSALegacy 32 1.3.6.1.4.1.3029.1.5.1 10 2B 06 01 04 01 97 55 01 05 01 Curve25519Legacy ECDH 32 The "Field Size (fsize)" column represents the field size of the group in number of octets, rounded up, such that x or y coordinates for a point on the curve or native point representations for the curve can be represented in that many octets. For the curves specified here, also scalars such as the base point order and the private key can be represented in fsize octets. Note that, however, there exist curves outside this specification where the representation of scalars requires an additional octet. The sequence of octets in the third column is the result of applying the Distinguished Encoding Rules (DER) to the ASN.1 Object Identifier with subsequent truncation. The truncation removes the two fields of encoded Object Identifier. The first omitted field is one octet representing the Object Identifier tag, and the second omitted field is the length of the Object Identifier body. For example, the complete ASN.1 DER encoding for the NIST P-256 curve OID is "06 08 2A 86 48 CE 3D 03 01 07", from which the first entry in the table above is constructed by omitting the first two octets. Only the truncated sequence of octets is the valid representation of a curve OID. The deprecated OIDs for Ed25519Legacy and Curve25519Legacy are used only in version 4 keys and signatures. Implementations MAY implement these variants for compatibility with existing v4 key material and signatures. Implementations MUST NOT accept or generate v6 key material using the deprecated OIDs.
Curve-Specific Wire Formats Some Elliptic Curve Public Key Algorithms use different conventions for specific fields depending on the curve in use. Each field is always formatted as an MPI, but with a curve-specific framing. This table summarizes those distinctions. Curve ECDH Point Format ECDH Secret Key MPI EdDSA Secret Key MPI EdDSA Signature first MPI EdDSA Signature second MPI NIST P-256 SEC1 integer N/A N/A N/A NIST P-384 SEC1 integer N/A N/A N/A NIST P-521 SEC1 integer N/A N/A N/A brainpoolP256r1 SEC1 integer N/A N/A N/A brainpoolP384r1 SEC1 integer N/A N/A N/A brainpoolP512r1 SEC1 integer N/A N/A N/A Ed25519Legacy N/A N/A 32 octets of secret 32 octets of R 32 octets of S Curve25519Legacy prefixed native integer (see ) N/A N/A N/A For the native octet-string forms of Ed25519Legacy values, see . For the native octet-string forms of Curve25519Legacy secret scalars and points, see .
Symmetric-Key Algorithms ID Algorithm 0 Plaintext or unencrypted data 1 IDEA 2 TripleDES (DES-EDE, - 168 bit key derived from 192) 3 CAST5 (128 bit key, as per ) 4 Blowfish (128 bit key, 16 rounds) 5 Reserved 6 Reserved 7 AES with 128-bit key 8 AES with 192-bit key 9 AES with 256-bit key 10 Twofish with 256-bit key 11 Camellia with 128-bit key 12 Camellia with 192-bit key 13 Camellia with 256-bit key 100 to 110 Private/Experimental algorithm 253, 254 and 255 Reserved to avoid collision with Secret Key Encryption (see and ) Implementations MUST implement AES-128. Implementations SHOULD implement AES-256. Implementations MUST NOT encrypt data with IDEA, TripleDES, or CAST5. Implementations MAY decrypt data that uses IDEA, TripleDES, or CAST5 for the sake of reading older messages or new messages from implementations predating support for . An Implementation that decrypts data using IDEA, TripleDES, or CAST5 SHOULD generate a deprecation warning about the symmetric algorithm, indicating that message confidentiality is suspect. Implementations MAY implement any other algorithm.
Compression Algorithms ID Algorithm 0 Uncompressed 1 ZIP 2 ZLIB 3 BZip2 100 to 110 Private/Experimental algorithm Implementations MUST implement uncompressed data. Implementations SHOULD implement ZLIB. For interoperability reasons implementations SHOULD be able to decompress using ZIP. Implementations MAY implement any other algorithm.
Hash Algorithms ID Algorithm Text Name V6 signature salt size 1 MD5 "MD5" N/A 2 SHA-1 , "SHA1" N/A 3 RIPEMD-160 "RIPEMD160" N/A 4 Reserved     5 Reserved     6 Reserved     7 Reserved     8 SHA2-256 "SHA256" 16 9 SHA2-384 "SHA384" 24 10 SHA2-512 "SHA512" 32 11 SHA2-224 "SHA224" 16 12 SHA3-256 "SHA3-256" 16 13 Reserved     14 SHA3-512 "SHA3-512" 32 100 to 110 Private/Experimental algorithm     Hash Algorithm OID Full hash prefix MD5 1.2.840.113549.2.5 0x30, 0x20, 0x30, 0x0C, 0x06, 0x08, 0x2A, 0x86, 0x48, 0x86, 0xF7, 0x0D, 0x02, 0x05, 0x05, 0x00, 0x04, 0x10 SHA-1 1.3.14.3.2.26 0x30, 0x21, 0x30, 0x09, 0x06, 0x05, 0x2B, 0x0E, 0x03, 0x02, 0x1A, 0x05, 0x00, 0x04, 0x14 RIPEMD-160 1.3.36.3.2.1 0x30, 0x21, 0x30, 0x09, 0x06, 0x05, 0x2B, 0x24, 0x03, 0x02, 0x01, 0x05, 0x00, 0x04, 0x14 SHA2-256 2.16.840.1.101.3.4.2.1 0x30, 0x31, 0x30, 0x0D, 0x06, 0x09, 0x60, 0x86, 0x48, 0x01, 0x65, 0x03, 0x04, 0x02, 0x01, 0x05, 0x00, 0x04, 0x20 SHA2-384 2.16.840.1.101.3.4.2.2 0x30, 0x41, 0x30, 0x0D, 0x06, 0x09, 0x60, 0x86, 0x48, 0x01, 0x65, 0x03, 0x04, 0x02, 0x02, 0x05, 0x00, 0x04, 0x30 SHA2-512 2.16.840.1.101.3.4.2.3 0x30, 0x51, 0x30, 0x0D, 0x06, 0x09, 0x60, 0x86, 0x48, 0x01, 0x65, 0x03, 0x04, 0x02, 0x03, 0x05, 0x00, 0x04, 0x40 SHA2-224 2.16.840.1.101.3.4.2.4 0x30, 0x2D, 0x30, 0x0D, 0x06, 0x09, 0x60, 0x86, 0x48, 0x01, 0x65, 0x03, 0x04, 0x02, 0x04, 0x05, 0x00, 0x04, 0x1C SHA3-256 2.16.840.1.101.3.4.2.8 0x30, 0x31, 0x30, 0x0D, 0x06, 0x09, 0x60, 0x86, 0x48, 0x01, 0x65, 0x03, 0x04, 0x02, 0x08, 0x05, 0x00, 0x04, 0x20 SHA3-512 2.16.840.1.101.3.4.2.10 0x30, 0x51, 0x30, 0x0D, 0x06, 0x09, 0x60, 0x86, 0x48, 0x01, 0x65, 0x03, 0x04, 0x02, 0x0a, 0x05, 0x00, 0x04, 0x40 Implementations MUST implement SHA2-256. Implementations SHOULD implement SHA2-384 and SHA2-512. Implementations MAY implement other algorithms. Implementations SHOULD NOT create messages which require the use of SHA-1 with the exception of computing version 4 key fingerprints and for purposes of the Modification Detection Code (MDC) in version 1 Symmetrically Encrypted Integrity Protected Data packets. Implementations MUST NOT generate signatures with MD5, SHA-1, or RIPEMD-160. Implementations MUST NOT use MD5, SHA-1, or RIPEMD-160 as a hash function in an ECDH KDF. Implementations MUST NOT generate packets using MD5, SHA-1, or RIPEMD-160 as a hash function in an S2K KDF. Implementations MUST NOT decrypt a secret using MD5, SHA-1, or RIPEMD-160 as a hash function in an S2K KDF in a version 6 (or later) packet. Implementations MUST NOT validate any recent signature that depends on MD5, SHA-1, or RIPEMD-160. Implementations SHOULD NOT validate any old signature that depends on MD5, SHA-1, or RIPEMD-160 unless the signature's creation date predates known weakness of the algorithm used, and the implementation is confident that the message has been in the secure custody of the user the whole time.
AEAD Algorithms ID Name Reference Nonce length (octets) authentication tag length (octets) 1 EAX 16 16 2 OCB 15 16 3 GCM 12 16 100 to 110 Private/Experimental algorithm       Implementations MUST implement OCB. Implementations MAY implement EAX, GCM and other algorithms.
Packet Sequence Composition OpenPGP packets are assembled into sequences in order to create messages and to transfer keys. Not all possible packet sequences are meaningful and correct. This section describes the rules for how packets should be placed into sequences. There are three distinct sequences of packets: Transferable Public Keys () and their close counterpart, Transferable Secret Keys () OpenPGP Messages () Detached Signatures () Each sequence has an explicit grammar of what packet types () can appear in what place. The presence of an unknown critical packet, or a known but unexpected packet is a critical error, invalidating the entire sequence (see ). On the other hand, unknown non-critical packets can appear anywhere within any sequence. This provides a structured way to introduce new packets into the protocol, while making sure that certain packets will be handled strictly. An implementation may "recognize" a packet, but not implement it. The purpose of Packet Criticality is to allow the producer to tell the consumer whether it would prefer a new, unknown packet to generate an error or be ignored. Note that previous versions of this document did not have a concept of Packet Criticality, and did not give clear guidance on what to do when unknown packets are encountered. Therefore, implementations of older versions of this document may reject unknown non-critical packets, or accept unknown critical packets. When generating a sequence of OpenPGP packets according to one of the three grammars, an implementation MUST NOT inject a critical packet of a type that does not adhere to the grammar. When consuming a sequence of OpenPGP packets according to one of the three grammars, an implementation MUST reject the sequence with an error if it encounters a critical packet of inappropriate type according to the grammar.
Transferable Public Keys OpenPGP users may transfer public keys. This section describes the structure of public keys in transit to ensure interoperability. An OpenPGP Transferable Public Key is also known as an OpenPGP certificate, in order to distinguish it from both its constituent Public-Key packets ( and ) and the underlying cryptographic key material.
OpenPGP v6 Certificate Structure The format of an OpenPGP v6 certificate is as follows. Entries in square brackets are optional and ellipses indicate repetition.
In addition to these rules, a marker packet () can appear anywhere in the sequence. Note, that a v6 key uses a self-signed direct key signature to store algorithm preferences. Every subkey for a v6 primary key MUST be a v6 subkey. Every subkey MUST have at least one subkey binding signature. Every subkey binding signature MUST be a self-signature (that is, made by the v6 primary key). Like all other signatures, every self-signature made by a v6 key MUST be a v6 signature.
OpenPGP v6 Revocation Certificate When a primary v6 Public Key is revoked, it is sometimes distributed with only the revocation signature:
In this case, the direct key signature is no longer necessary, since the primary key itself has been marked as unusable.
OpenPGP v4 Certificate Structure The format of an OpenPGP v4 key is as follows.
In addition to these rules, a marker packet () can appear anywhere in the sequence. A subkey always has at least one subkey binding signature after it that is issued using the primary key to tie the two keys together. These binding signatures may be in either v3 or v4 format, but SHOULD be in v4 format. Subkeys that can issue signatures MUST have a v4 binding signature due to the REQUIRED embedded primary key binding signature. Every subkey for a v4 primary key MUST be a v4 subkey. When a primary v4 Public Key is revoked, the revocation signature is sometimes distributed by itself, without the primary key packet it applies to. This is referred to as a "revocation certificate". Instead, a v6 revocation certificate MUST include the primary key packet, as described in .
OpenPGP v3 Key Structure The format of an OpenPGP v3 key is as follows.
In addition to these rules, a marker packet () can appear anywhere in the sequence. Each signature certifies the RSA public key and the preceding User ID. The RSA public key can have many User IDs and each User ID can have many signatures. V3 keys are deprecated. Implementations MUST NOT generate new v3 keys, but MAY continue to use existing ones. V3 keys MUST NOT have subkeys.
Common requirements The Public-Key packet occurs first. The primary key MUST be an algorithm capable of making signatures (that is, not an encryption-only algorithm). This is because the primary key needs to be able to create self-signatures (see ). The subkeys may be keys of any type. For example, there may be a single-key RSA key, an Ed25519 primary key with an RSA encryption key, or an Ed25519 primary key with an X25519 subkey, etc. Each of the following User ID packets provides the identity of the owner of this public key. If there are multiple User ID packets, this corresponds to multiple means of identifying the same unique individual user; for example, a user may have more than one email address, and construct a User ID for each one. A transferable public key SHOULD include at least one User ID packet unless storage requirements prohibit this. Immediately following each User ID packet, there are zero or more Signature packets. Each Signature packet is calculated on the immediately preceding User ID packet and the initial Public-Key packet. The signature serves to certify the corresponding public key and User ID. In effect, the signer is testifying to his or her belief that this public key belongs to the user identified by this User ID. Within the same section as the User ID packets, there are zero or more User Attribute packets. Like the User ID packets, a User Attribute packet is followed by zero or more Signature packets calculated on the immediately preceding User Attribute packet and the initial Public-Key packet. User Attribute packets and User ID packets may be freely intermixed in this section, so long as the signatures that follow them are maintained on the proper User Attribute or User ID packet. After the sequence of User ID packets and Attribute packets and their associated signatures, zero or more Subkey packets follow, each with their own signatures. In general, subkeys are provided in cases where the top-level public key is a certification-only key. However, any v4 or v6 key may have subkeys, and the subkeys may be encryption keys, signing keys, authentication keys, etc. It is good practice to use separate subkeys for every operation (i.e. signature-only, encryption-only, authentication-only keys, etc.). Each Subkey packet MUST be followed by one Signature packet, which should be a subkey binding signature issued by the top-level key. For subkeys that can issue signatures, the subkey binding signature MUST contain an Embedded Signature subpacket with a primary key binding signature (0x19) issued by the subkey on the top-level key. Subkey and Key packets may each be followed by a revocation Signature packet to indicate that the key is revoked. Revocation signatures are only accepted if they are issued by the key itself, or by a key that is authorized to issue revocations via a Revocation Key subpacket in a self-signature by the top-level key. The optional trailing Padding packet is a mechanism to defend against traffic analysis (see ). For maximum interoperability, if the Public-Key packet is a v4 key, the optional Padding packet SHOULD NOT be present unless the recipient has indicated that they are capable of ignoring it successfully. An implementation that is capable of receiving a transferable public key with a v6 Public-Key primary key MUST be able to accept (and ignore) the trailing optional Padding packet. Transferable public-key packet sequences may be concatenated to allow transferring multiple public keys in one operation (see ).
Transferable Secret Keys OpenPGP users may transfer secret keys. The format of a transferable secret key is the same as a transferable public key except that Secret-Key and Secret-Subkey packets can be used in addition to the Public-Key and Public-Subkey packets. If a single Secret-Key or Secret-Subkey packet is included in a packet sequence, it is a transferable secret key and should be handled and marked as such (see ). An implementation SHOULD include self-signatures on any User IDs and subkeys, as this allows for a complete public key to be automatically extracted from the transferable secret key. An implementation MAY choose to omit the self-signatures, especially if a transferable public key accompanies the transferable secret key.
OpenPGP Messages An OpenPGP message is a packet or sequence of packets that corresponds to the following grammatical rules (comma (,) represents sequential composition, and vertical bar (|) separates alternatives):
OpenPGP Message :-
Encrypted Message | Signed Message | Compressed Message | Literal Message.
Compressed Message :-
Compressed Data Packet.
Literal Message :-
Literal Data Packet.
ESK :-
Public-Key Encrypted Session Key Packet | Symmetric-Key Encrypted Session Key Packet.
ESK Sequence :-
ESK | ESK Sequence, ESK.
Encrypted Data :-
Symmetrically Encrypted Data Packet | Symmetrically Encrypted Integrity Protected Data Packet
Encrypted Message :-
Encrypted Data | ESK Sequence, Encrypted Data.
One-Pass Signed Message :-
One-Pass Signature Packet, OpenPGP Message, Corresponding Signature Packet.
Signed Message :-
Signature Packet, OpenPGP Message | One-Pass Signed Message.
Optionally Padded Message :-
OpenPGP Message | OpenPGP Message, Padding Packet.
In addition to these rules, a marker packet () can appear anywhere in the sequence.
Unwrapping Encrypted and Compressed Messages In addition to the above grammar, certain messages can be "unwrapped" to yield new messages. In particular: Decrypting a version 2 Symmetrically Encrypted and Integrity Protected Data packet MUST yield a valid Optionally Padded Message. Decrypting a version 1 Symmetrically Encrypted and Integrity Protected Data packet or --- for historic data --- a Symmetrically Encrypted Data packet MUST yield a valid OpenPGP Message. Decompressing a Compressed Data packet MUST also yield a valid OpenPGP Message. When any unwrapping is performed, the resulting stream of octets is parsed into a series of OpenPGP packets like any other stream of octets. The packet boundaries found in the series of octets are expected to align with the length of the unwrapped octet stream. An implementation MUST NOT interpret octets beyond the boundaries of the unwrapped octet stream as part of any OpenPGP packet. If an implementation encounters a packet whose header length indicates that it would extend beyond the boundaries of the unwrapped octet stream, the implementation MUST reject that packet as malformed and unusable.
Additional Constraints on Packet Sequences Note that some subtle combinations that are formally acceptable by this grammar are nonetheless unacceptable.
Packet Versions in Encrypted Messages As noted above, an Encrypted Message is a sequence of zero or more PKESKs () and SKESKs (), followed by an SEIPD () payload. In some historic data, the payload may be a deprecated SED () packet instead of SEIPD, though implementations MUST NOT generate SED packets (see ). The versions of the preceding ESK packets within an Encrypted Message MUST align with the version of the payload SEIPD packet, as described in this section. v3 PKESK and v4 SKESK packets both contain in their cleartext the symmetric cipher algorithm ID in addition to the session key for the subsequent SEIPD packet. Since a v1 SEIPD does not contain a symmetric algorithm ID, all ESK packets preceding a v1 SEIPD payload MUST be either v3 PKESK or v4 SKESK. On the other hand, the cleartext of the v6 ESK packets (either PKESK or SKESK) do not contain a symmetric cipher algorithm ID, so they cannot be used in combination with a v1 SEIPD payload. The payload following any v6 PKESK or v6 SKESK packet MUST be a v2 SEIPD. Additionally, to avoid potentially conflicting cipher algorithm IDs, and for simplicity, implementations MUST NOT precede a v2 SEIPD payload with either v3 PKESK or v4 SKESK packets. The versions of packets found in an Encrypted Message are summarized in the following table. An implementation MUST only generate an Encrypted Message using packet versions that match a row with "Yes" in the "Generate?" column. Other rows are provided for the purpose of historic interoperability. A conforming implementation MUST only generate an Encrypted Message using packets whose versions correspond to a single row. Version of Encrypted Data payload Version of preceding Symmetric-Key ESK (if any) Version of preceding Public-Key ESK (if any) Generate? SED () - v2 PKESK () No SED () v4 SKESK () v3 PKESK () No v1 SEIPD () v4 SKESK () v3 PKESK () Yes v2 SEIPD () v6 SKESK () v6 PKESK () Yes An implementation processing an Encrypted Message MUST discard any preceding ESK packet with a version that does not align with the version of the payload.
Packet Versions in Signatures OpenPGP key packets and signature packets are also versioned. The version of a Signature typically matches the version of the signing key. When a v6 key produces a signature packet, it MUST produce a version 6 signature packet, regardless of the signature packet type. When a message is signed or verified using the one-pass construction, the version of the One-Pass Signature packet () should also be aligned to the other versions. Some legacy implementations have produced unaligned signature versions for older key material, which are also described in the table below for purpose of historic interoperability. A conforming implementation MUST only generate signature packets with version numbers matching rows with "Yes" in the "Generate?" column. Signing key version Signature packet version OPS packet version Generate? 3 () 3 () 3 No 4 () 3 () 3 No 4 () 4 () 3 Yes 6 () 6 () 6 Yes Note, however, that a version mismatch between these packets does not invalidate the packet sequence as a whole, it merely invalidates the signature, as a signature with an unknown version SHOULD be discarded (see ).
Detached Signatures Some OpenPGP applications use so-called "detached signatures". For example, a program bundle may contain a file, and with it a second file that is a detached signature of the first file. These detached signatures are simply one or more Signature packets stored separately from the data for which they are a signature. In addition, a marker packet () and a padding packet () can appear anywhere in the sequence.
Elliptic Curve Cryptography This section describes algorithms and parameters used with Elliptic Curve Cryptography (ECC) keys. A thorough introduction to ECC can be found in . None of the ECC methods described in this document are allowed with deprecated v3 keys. Refer to , B.4.1, for the method to generate a uniformly distributed ECC private key.
ECC Curves This document references three named prime field curves defined in as "Curve P-256", "Curve P-384", and "Curve P-521"; and three named prime field curves defined in as "brainpoolP256r1", "brainpoolP384r1", and "brainpoolP512r1". The three curves and the three curves can be used with ECDSA and ECDH public key algorithms. They are referenced using a sequence of octets, referred to as the curve OID. describes in detail how this sequence of octets is formed. Separate algorithms are also defined for the use of X25519 and X448, defined in ; and Ed25519 and Ed448, defined in . Additionally, legacy OIDs are defined for "Curve25519Legacy" (for encryption using the ECDH algorithm) and "Ed25519Legacy" (for signing using the EdDSALegacy algorithm).
EC Point Wire Formats A point on an elliptic curve will always be represented on the wire as an MPI. Each curve uses a specific point format for the data within the MPI itself. Each format uses a designated prefix octet to ensure that the high octet has at least one bit set to make the MPI a constant size. Name Wire Format Reference SEC1 0x04 || x || y Prefixed native 0x40 || native
SEC1 EC Point Wire Format For a SEC1-encoded (uncompressed) point the content of the MPI is:
where x and y are coordinates of the point P = (x, y), and each is encoded in the big-endian format and zero-padded to the adjusted underlying field size. The adjusted underlying field size is the underlying field size rounded up to the nearest 8-bit boundary, as noted in the "fsize" column in . This encoding is compatible with the definition given in .
Prefixed Native EC Point Wire Format For a custom compressed point the content of the MPI is:
where p is the public key of the point encoded using the rules defined for the specified curve. This format is used for ECDH keys based on curves expressed in Montgomery form, and for points when using EdDSA.
Notes on EC Point Wire Formats Given the above definitions, the exact size of the MPI payload for an encoded point is 515 bits for both NIST P-256 and brainpoolP256r1, 771 for both NIST P-384 and brainpoolP384r1, 1059 for NIST P-521, 1027 for brainpoolP512r1, and 263 for both Curve25519Legacy and Ed25519Legacy. For example, the length of a EdDSALegacy public key for the curve Ed25519Legacy is 263 bits: 7 bits to represent the 0x40 prefix octet and 32 octets for the native value of the public key. Even though the zero point, also called the point at infinity, may occur as a result of arithmetic operations on points of an elliptic curve, it SHALL NOT appear in data structures defined in this document. Each particular curve uses a designated wire format for the point found in its public key or ECDH data structure. An implementation MUST NOT use a different wire format for a point than the wire format associated with the curve.
EC Scalar Wire Formats Some non-curve values in elliptic curve cryptography (for example, secret keys and signature components) are not points on a curve, but are also encoded on the wire in OpenPGP as an MPI. Because of different patterns of deployment, some curves treat these values as opaque bit strings with the high bit set, while others are treated as actual integers, encoded in the standard OpenPGP big-endian form. The choice of encoding is specific to the public key algorithm in use. Type Description Reference integer An integer, big-endian encoded as a standard OpenPGP MPI octet string An octet string of fixed length, that may be shorter on the wire due to leading zeros being stripped by the MPI encoding, and may need to be zero-padded before use prefixed N octets An octet string of fixed length N, prefixed with octet 0x40 to ensure no leading zero octet
EC Octet String Wire Format Some opaque strings of octets are represented on the wire as an MPI by simply stripping the leading zeros and counting the remaining bits. These strings are of known, fixed length. They are represented in this document as MPI(N octets of X) where N is the expected length in octets of the octet string. For example, a five-octet opaque string (MPI(5 octets of X)) where X has the value 00 02 EE 19 00 would be represented on the wire as an MPI like so: 00 1A 02 EE 19 00. To encode X to the wire format, we set the MPI's two-octet bit counter to the value of the highest set bit (bit 26, or 0x001A), and do not transfer the leading all-zero octet to the wire. To reverse the process, an implementation that knows this value has an expected length of 5 octets can take the following steps: Ensure that the MPI's two-octet bitcount is less than or equal to 40 (5 octets of 8 bits) Allocate 5 octets, setting all to zero initially Copy the MPI data octets (without the two count octets) into the lower octets of the allocated space
Elliptic Curve Prefixed Octet String Wire Format Another way to ensure that a fixed-length bytestring is encoded simply to the wire while remaining in MPI format is to prefix the bytestring with a dedicated non-zero octet. This specification uses 0x40 as the prefix octet. This is represented in this standard as MPI(prefixed N octets of X), where N is the known bytestring length. For example, a five-octet opaque string using MPI(prefixed 5 octets of X) where X has the value 00 02 EE 19 00 would be written to the wire form as: 00 2F 40 00 02 EE 19 00. To encode the string, we prefix it with the octet 0x40 (whose 7th bit is set), then set the MPI's two-octet bit counter to 47 (0x002F, 7 bits for the prefix octet and 40 bits for the string). To decode the string from the wire, an implementation that knows that the variable is formed in this way can: Ensure that the first three octets of the MPI (the two bit-count octets plus the prefix octet) are 00 2F 40, and Use the remainder of the MPI directly off the wire. Note that this is a similar approach to that used in the EC point encodings found in .
Key Derivation Function A key derivation function (KDF) is necessary to implement EC encryption. The Concatenation Key Derivation Function (Approved Alternative 1) with the KDF hash function that is SHA2-256 or stronger is REQUIRED. For convenience, the synopsis of the encoding method is given below with significant simplifications attributable to the restricted choice of hash functions in this document. However, is the normative source of the definition.
Note that ZB in the KDF description above is the compact representation of X as defined in .
EC DH Algorithm (ECDH) The method is a combination of an ECC Diffie-Hellman method to establish a shared secret, a key derivation method to process the shared secret into a derived key, and a key wrapping method that uses the derived key to protect a session key used to encrypt a message. The One-Pass Diffie-Hellman method C(1, 1, ECC CDH) MUST be implemented with the following restrictions: the ECC CDH primitive employed by this method is modified to always assume the cofactor is 1, the KDF specified in is used, and the KDF parameters specified below are used. The KDF parameters are encoded as a concatenation of the following 5 variable-length and fixed-length fields, which are compatible with the definition of the OtherInfo bitstring : A variable-length field containing a curve OID, which is formatted as follows: A one-octet size of the following field, The octets representing a curve OID defined in ; A one-octet public key algorithm ID defined in ; A variable-length field containing KDF parameters, which are identical to the corresponding field in the ECDH public key, and are formatted as follows: A one-octet size of the following fields; values 0 and 0xFF are reserved for future extensions, A one-octet value 0x01, reserved for future extensions, A one-octet hash function ID used with the KDF, A one-octet algorithm ID for the symmetric algorithm used to wrap the symmetric key for message encryption; see for details; 20 octets representing the UTF-8 encoding of the string Anonymous Sender    , which is the octet sequence 41 6E 6F 6E 79 6D 6F 75 73 20 53 65 6E 64 65 72 20 20 20 20; A variable-length field containing the fingerprint of the recipient encryption subkey identifying the key material that is needed for decryption. For version 4 keys, this field is 20 octets. For version 6 keys, this field is 32 octets. The size in octets of the KDF parameters sequence, defined above, for encrypting to a v4 key is either 54 for curve NIST P-256, 51 for curves NIST P-384 and NIST P-521, 55 for curves brainpoolP256r1, brainpoolP384r1 and brainpoolP512r1, or 56 for Curve25519Legacy. For encrypting to a v6 key, the size of the sequence is either 66 for curve NIST P-256, 63 for curves NIST P-384 and NIST P-521, or 67 for curves brainpoolP256r1, brainpoolP384r1 and brainpoolP512r1. The key wrapping method is described in . The KDF produces a symmetric key that is used as a key-encryption key (KEK) as specified in . Refer to for the details regarding the choice of the KEK algorithm, which SHOULD be one of the three AES algorithms. Key wrapping and unwrapping is performed with the default initial value of . To produce the input to the key wrapping method, first concatenate the following values: The one-octet algorithm identifier, if it was passed (in the case of a v3 PKESK packet). The session key. A two-octet checksum of the session key, equal to the sum of the session key octets, modulo 65536. Then, the above values are padded using the method described in to an 8-octet granularity. For example, in a v3 Public-Key Encrypted Session Key packet, an AES-256 session key is encoded as follows, forming a 40 octet sequence:
The octets k0 to k31 above denote the session key, and the octets s0 and s1 denote the checksum of the session key octets. This encoding allows the sender to obfuscate the size of the symmetric encryption key used to encrypt the data. For example, assuming that an AES algorithm is used for the session key, the sender MAY use 21, 13, and 5 octets of padding for AES-128, AES-192, and AES-256, respectively, to provide the same number of octets, 40 total, as an input to the key wrapping method. In a v6 Public-Key Encrypted Session Key packet, the symmetric algorithm is not included, as described in . For example, an AES-256 session key would be composed as follows:
The octets k0 to k31 above again denote the session key, and the octets s0 and s1 denote the checksum. In this case, assuming that an AES algorithm is used for the session key, the sender MAY use 22, 14, and 6 octets of padding for AES-128, AES-192, and AES-256, respectively, to provide the same number of octets, 40 total, as an input to the key wrapping method. The output of the method consists of two fields. The first field is the MPI containing the ephemeral key used to establish the shared secret. The second field is composed of the following two subfields: One octet encoding the size in octets of the result of the key wrapping method; the value 255 is reserved for future extensions; Up to 254 octets representing the result of the key wrapping method, applied to the 8-octet padded session key, as described above. Note that for session key sizes 128, 192, and 256 bits, the size of the result of the key wrapping method is, respectively, 32, 40, and 48 octets, unless size obfuscation is used. For convenience, the synopsis of the encoding method is given below; however, this section, , and are the normative sources of the definition. Obtain the authenticated recipient public key R Generate an ephemeral, single-use key pair {v, V=vG} Compute the shared point S = vR; m = symm_alg_ID || session key || checksum || pkcs5_padding; curve_OID_len = (octet)len(curve_OID); Param = curve_OID_len || curve_OID || public_key_alg_ID || 03 || 01 || KDF_hash_ID || KEK_alg_ID for AESKeyWrap || Anonymous Sender     || recipient_fingerprint; Z_len = the key size for the KEK_alg_ID used with AESKeyWrap Compute Z = KDF( S, Z_len, Param ); Compute C = AESKeyWrap( Z, m ); (as per ) Wipe the memory that contained S, v, and Z to avoid leaking ephemeral secrets VB = convert point V to the octet string Output (MPI(VB) || len(C) || C). The decryption is the inverse of the method given. Note that the recipient with key pair (r,R) obtains the shared secret by calculating:
ECDH Parameters ECDH keys have a hash algorithm parameter for key derivation and a symmetric algorithm for key encapsulation. For v6 keys, the following algorithms MUST be used depending on the curve. An implementation MUST NOT generate a v6 ECDH key over any listed curve that uses different KDF or KEK parameters. An implementation MUST NOT encrypt any message to a v6 ECDH key over a listed curve that announces a different KDF or KEK parameter. For v4 keys, the following algorithms SHOULD be used depending on the curve. An implementation SHOULD only use an AES algorithm as a KEK algorithm. Curve Hash algorithm Symmetric algorithm NIST P-256 SHA2-256 AES-128 NIST P-384 SHA2-384 AES-192 NIST P-521 SHA2-512 AES-256 brainpoolP256r1 SHA2-256 AES-128 brainpoolP384r1 SHA2-384 AES-192 brainpoolP512r1 SHA2-512 AES-256 Curve25519Legacy SHA2-256 AES-128
Notes on Algorithms
PKCS#1 Encoding in OpenPGP This standard makes use of the PKCS#1 functions EME-PKCS1-v1_5 and EMSA-PKCS1-v1_5. However, the calling conventions of these functions has changed in the past. To avoid potential confusion and interoperability problems, we are including local copies in this document, adapted from those in PKCS#1 v2.1 . should be treated as the ultimate authority on PKCS#1 for OpenPGP. Nonetheless, we believe that there is value in having a self-contained document that avoids problems in the future with needed changes in the conventions.
EME-PKCS1-v1_5-ENCODE Input:
k =
the length in octets of the key modulus.
M =
message to be encoded, an octet string of length mLen, where mLen <= k - 11.
Output:
EM =
encoded message, an octet string of length k.
Error: "message too long". Length checking: If mLen > k - 11, output "message too long" and stop. Generate an octet string PS of length k - mLen - 3 consisting of pseudo-randomly generated nonzero octets. The length of PS will be at least eight octets. Concatenate PS, the message M, and other padding to form an encoded message EM of length k octets as
Output EM.
EME-PKCS1-v1_5-DECODE Input:
EM =
encoded message, an octet string
Output:
M =
message, an octet string.
Error: "decryption error". To decode an EME-PKCS1_v1_5 message, separate the encoded message EM into an octet string PS consisting of nonzero octets and a message M as follows
If the first octet of EM does not have hexadecimal value 0x00, if the second octet of EM does not have hexadecimal value 0x02, if there is no octet with hexadecimal value 0x00 to separate PS from M, or if the length of PS is less than 8 octets, output "decryption error" and stop. See also regarding differences in reporting between a decryption error and a padding error.
EMSA-PKCS1-v1_5 This encoding method is deterministic and only has an encoding operation. Option:
Hash -
a hash function in which hLen denotes the length in octets of the hash function output.
Input:
M =
message to be encoded.
emLen =
intended length in octets of the encoded message, at least tLen + 11, where tLen is the octet length of the DER encoding T of a certain value computed during the encoding operation.
Output:
EM =
encoded message, an octet string of length emLen.
Errors: "message too long"; "intended encoded message length too short". Steps: Apply the hash function to the message M to produce a hash value H: H = Hash(M). If the hash function outputs "message too long," output "message too long" and stop. Using the list in , produce an ASN.1 DER value for the hash function used. Let T be the full hash prefix from the list, and let tLen be the length in octets of T. If emLen < tLen + 11, output "intended encoded message length too short" and stop. Generate an octet string PS consisting of emLen - tLen - 3 octets with hexadecimal value 0xFF. The length of PS will be at least 8 octets. Concatenate PS, the hash prefix T, and other padding to form the encoded message EM as
Output EM.
Symmetric Algorithm Preferences The symmetric algorithm preference is an ordered list of algorithms that the keyholder accepts. Since it is found on a self-signature, it is possible that a keyholder may have multiple, different preferences. For example, Alice may have AES-128 only specified for "alice@work.com" but Camellia-256, Twofish, and AES-128 specified for "alice@home.org". Note that it is also possible for preferences to be in a subkey's binding signature. Since AES-128 is the MUST-implement algorithm, if it is not explicitly in the list, it is tacitly at the end. However, it is good form to place it there explicitly. Note also that if an implementation does not implement the preference, then it is implicitly an AES-128-only implementation. Note further that implementations conforming to previous versions of this standard have TripleDES as its only MUST-implement algorithm. An implementation MUST NOT use a symmetric algorithm that is not in the recipient's preference list. When encrypting to more than one recipient, the implementation finds a suitable algorithm by taking the intersection of the preferences of the recipients. Note that the MUST-implement algorithm, AES-128, ensures that the intersection is non-empty. The implementation may use any mechanism to pick an algorithm in the intersection. If an implementation can decrypt a message that a keyholder doesn't have in their preferences, the implementation SHOULD decrypt the message anyway, but MUST warn the keyholder that the protocol has been violated. For example, suppose that Alice, above, has an implementation that implements all algorithms in this specification. Nonetheless, she prefers subsets for work or home. If she is sent a message encrypted with IDEA, which is not in her preferences, the implementation warns her that someone sent her an IDEA-encrypted message, but it would ideally decrypt it anyway.
Plaintext Algorithm 0, "plaintext", may only be used to denote secret keys that are stored in the clear. An implementation MUST NOT use algorithm 0 as the indicated symmetric cipher for an encrypted data packet ( or ); it can use a Literal Data packet () to encode unencrypted literal data.
Other Algorithm Preferences Other algorithm preferences work similarly to the symmetric algorithm preference, in that they specify which algorithms the keyholder accepts. There are two interesting cases that other comments need to be made about, though, the compression preferences and the hash preferences.
Compression Preferences Like the algorithm preferences, an implementation MUST NOT use an algorithm that is not in the preference vector. If Uncompressed (0) is not explicitly in the list, it is tacitly at the end. That is, uncompressed messages may always be sent. Note that earlier implementations may assume that the absence of compression preferences means that [ZIP(1), Uncompressed(0)] are preferred, and default to ZIP compression. Therefore, an implementation that prefers uncompressed data SHOULD explicitly state this in the preferred compression algorithms.
Uncompressed Algorithm 0, "uncompressed", may only be used to denote a preference for uncompressed data. An implementation MUST NOT use algorithm 0 as the indicated compression algorithm in a Compressed Data packet (); it can use a Literal Data packet () to encode uncompressed literal data.
Hash Algorithm Preferences Typically, the choice of a hash algorithm is something the signer does, rather than the verifier, because a signer rarely knows who is going to be verifying the signature. This preference, though, allows a protocol based upon digital signatures ease in negotiation. Thus, if Alice is authenticating herself to Bob with a signature, it makes sense for her to use a hash algorithm that Bob's implementation uses. This preference allows Bob to state in his key which algorithms Alice may use. Since SHA2-256 is the MUST-implement hash algorithm, if it is not explicitly in the list, it is tacitly at the end. However, it is good form to place it there explicitly.
RSA The PKCS1-v1_5 padding scheme, used by the RSA algorithms defined in this document, is no longer recommended, and its use is deprecated by . Therefore, an implementation SHOULD NOT generate RSA keys. There are algorithm types for RSA Sign-Only, and RSA Encrypt-Only keys. These types are deprecated. The "key flags" subpacket in a signature is a much better way to express the same idea, and generalizes it to all algorithms. An implementation MUST NOT create such a key, but MAY interpret it. An implementation MUST NOT generate RSA keys of size less than 3072 bits. An implementation SHOULD NOT encrypt, sign or verify using RSA keys of size less than 3072 bits. An implementation MUST NOT encrypt, sign or verify using RSA keys of size less than 2048 bits. An implementation that decrypts a message using an RSA secret key of size less than 3072 bits SHOULD generate a deprecation warning that the key is too weak for modern use.
DSA DSA is no longer recommended. It has also been deprecated in . Therefore, an implementation MUST NOT generate DSA keys. An implementation MUST NOT sign or verify using DSA keys.
Elgamal The PKCS1-v1_5 padding scheme, used by the Elgamal algorithm defined in this document, is no longer recommended, and its use is deprecated by . Therefore, an implementation MUST NOT generate Elgamal keys. An implementation MUST NOT encrypt using Elgamal keys. An implementation that decrypts a message using an Elgamal secret key SHOULD generate a deprecation warning that the key is too weak for modern use.
EdDSA Although the EdDSA algorithm allows arbitrary data as input, its use with OpenPGP requires that a digest of the message is used as input (pre-hashed). See for details. Truncation of the resulting digest is never applied; the resulting digest value is used verbatim as input to the EdDSA algorithm. For clarity: while describes different variants of EdDSA, OpenPGP uses the "pure" variant (PureEdDSA). The hashing that happens with OpenPGP is done as part of the standard OpenPGP signature process, and that hash itself is fed as the input message to the PureEdDSA algorithm. As specified in , Ed448 also expects a "context string". In OpenPGP, Ed448 is used with the empty string as a context string.
Reserved Algorithm IDs A number of algorithm IDs have been reserved for algorithms that would be useful to use in an OpenPGP implementation, yet there are issues that prevent an implementer from actually implementing the algorithm. These are marked as reserved in . The reserved public-key algorithm X9.42 (21) does not have the necessary parameters, parameter order, or semantics defined. The same is currently true for reserved public-key algorithms AEDH (23) and AEDSA (24). Previous versions of OpenPGP permitted Elgamal signatures with a public-key algorithm ID of 20. These are no longer permitted. An implementation MUST NOT generate such keys. An implementation MUST NOT generate Elgamal signatures. See .
CFB Mode The Cipher Feedback (CFB) mode used in this document is defined in Section 6.3 of . The CFB segment size s is equal to the block size of the cipher (i.e., n-bit CFB mode where n is the block size is used).
Private or Experimental Parameters S2K specifiers, Signature subpacket type IDs, User Attribute subpacket type IDs, image format IDs, and the various algorithm IDs described in all reserve the range 100 to 110 for private and experimental use. Packet type IDs reserve the range 60 to 63 for private and experimental use. These are intentionally managed with the PRIVATE USE method, as described in . However, implementations need to be careful with these and promote them to full IANA-managed parameters when they grow beyond the original, limited system.
Meta-Considerations for Expansion If OpenPGP is extended in a way that is not backwards-compatible, meaning that old implementations will not gracefully handle their absence of a new feature, the extension proposal can be declared in the key holder's self-signature as part of the Features signature subpacket. We cannot state definitively what extensions will not be upwards-compatible, but typically new algorithms are upwards-compatible, whereas new packets are not. If an extension proposal does not update the Features system, it SHOULD include an explanation of why this is unnecessary. If the proposal contains neither an extension to the Features system nor an explanation of why such an extension is unnecessary, the proposal SHOULD be rejected.
Security Considerations As with any technology involving cryptography, implementors should check the current literature to determine if any algorithms used here have been found to be vulnerable to an attack. If so, implementors should consider disallowing such algorithms for new data and warn or prevent the enduser when they are trying to consume data protected by such now vulnerable algorithms. This specification uses Public-Key Cryptography technologies. It is assumed that the private key portion of a public-private key pair is controlled and secured by the proper party or parties. The MD5 and SHA-1 hash algorithms have been found to have weaknesses, with collisions found in a number of cases. MD5 and SHA-1 are deprecated for use in OpenPGP (See ). Many security protocol designers think that it is a bad idea to use a single key for both privacy (encryption) and integrity (signatures). In fact, this was one of the motivating forces behind the v4 key format with separate signature and encryption keys. Using a single key for encrypting and signing is discouraged. The DSA algorithm will work with any hash, but is sensitive to the quality of the hash algorithm. Verifiers should be aware that even if the signer used a strong hash, an attacker could have modified the signature to use a weak one. Only signatures using acceptably strong hash algorithms should be accepted as valid. As OpenPGP combines many different asymmetric, symmetric, and hash algorithms, each with different measures of strength, care should be taken that the weakest element of an OpenPGP message is still sufficiently strong for the purpose at hand. While consensus about the strength of a given algorithm may evolve, NIST Special Publication 800-57 contains recommendations current at the time of this publication about equivalent security levels of different algorithms. There is a somewhat-related potential security problem in signatures. If an attacker can find a message that hashes to the same hash with a different algorithm, a bogus signature structure can be constructed that evaluates correctly. For example, suppose Alice DSA signs message M using hash algorithm H. Suppose that Mallet finds a message M' that has the same hash value as M with H'. Mallet can then construct a signature block that verifies as Alice's signature of M' with H'. However, this would also constitute a weakness in either H or H' or both. Should this ever occur, a revision will have to be made to this document to revise the allowed hash algorithms. If you are building an authentication system, the recipient may specify a preferred signing algorithm. However, the signer would be foolish to use a weak algorithm simply because the recipient requests it. Some of the encryption algorithms mentioned in this document have been analyzed less than others. For example, although TWOFISH is presently considered reasonably strong, it has been analyzed much less than AES. Other algorithms may have other concerns surrounding them. In late summer 2002, Jallad, Katz, and Schneier published an interesting attack on older versions of the OpenPGP protocol and some of its implementations . In this attack, the attacker modifies a message and sends it to a user who then returns the erroneously decrypted message to the attacker. The attacker is thus using the user as a decryption oracle, and can often decrypt the message. This attack is a particular form of ciphertext malleability. See for information on how to defend against such an attack using more recent versions of OpenPGP.
SHA-1 Collision Detection As described in , the SHA-1 digest algorithm is not collision-resistant. However, an OpenPGP implementation cannot completely discard the SHA-1 algorithm, because it is required for implementing v4 public keys. In particular, the v4 fingerprint derivation uses SHA-1. So as long as an OpenPGP implementation supports v4 public keys, it will need to implement SHA-1 in at least some scenarios. To avoid the risk of uncertain breakage from a maliciously introduced SHA-1 collision, an OpenPGP implementation MAY attempt to detect when a hash input is likely from a known collision attack, and then either deliberately reject the hash input or modify the hash output. This should convert an uncertain breakage (where it is unclear what the effect of a collision will be) to an explicit breakage, which is more desirable for a robust implementation. describes a method for detecting indicators of well-known SHA-1 collision attacks. Some example C code implementing this technique can be found at .
Advantages of Salted Signatures V6 signatures include a salt that is hashed first, which size depends on the hashing algorithm. This makes v6 OpenPGP signatures non-deterministic and protects against a broad class of attacks that depend on creating a signature over a predictable message. By selecting a new random salt for each signature made, the signed hashes and the signatures are not predictable. While the material to be signed could be attacker-controlled, hashing the salt first means that there is no attacker controlled hashed prefix. An example of this kind of attack is described in the paper "SHA-1 Is A Shambles" , which leverages a chosen prefix collision attack against SHA-1. This means that an adversary carrying out a chosen-message attack will not be able to control the hash that is being signed, and will need to break second-preimage resistance instead of the simpler collision resistance to create two messages having the same signature. The size of the salt is bound to the hash function to match the expected collision resistance level, and at least 16 octets. In some cases, an attacker may be able to induce a signature to be made, even if they do not control the content of the message. In some scenarios, a repeated signature over the exact same message may risk leakage of part or all of the signing key, for example see discussion of hardware faults over EdDSA and deterministic ECDSA in . Choosing a new random salt for each signature ensures that no repeated signatures are produced, and mitigates this risk.
Elliptic Curve Side Channels Side channel attacks are a concern when a compliant application's use of the OpenPGP format can be modeled by a decryption or signing oracle, for example, when an application is a network service performing decryption to unauthenticated remote users. ECC scalar multiplication operations used in ECDSA and ECDH are vulnerable to side channel attacks. Countermeasures can often be taken at the higher protocol level, such as limiting the number of allowed failures or time-blinding of the operations associated with each network interface. Mitigations at the scalar multiplication level seek to eliminate any measurable distinction between the ECC point addition and doubling operations.
Risks of a Quick Check Oracle In winter 2005, Serge Mister and Robert Zuccherato from Entrust released a paper describing a way that the "quick check" in v1 SEIPD and SED packets can be used as an oracle to decrypt two octets of every cipher block . This check was intended for early detection of session key decryption errors, particularly to detect a wrong passphrase, since v4 SKESK packets do not include an integrity check. There is a danger to using the quick check if timing or error information about the check can be exposed to an attacker, particularly via an automated service that allows rapidly repeated queries. Disabling the quick check prevents the attack. For very large encrypted data whose session key is protected by a passphrase using a version 4 SKESK, the quick check may be convenient to the user, by informing them early that they typed the wrong passphrase. But the implementation should use the quick check with care. The recommended approach for secure and early detection of decryption failure is to encrypt data using v2 SEIPD. If the session key is public-key encrypted, the quick check is not useful as the public-key encryption of the session key should guarantee that it is the right session key. The quick check oracle attack is a particular type of attack that exploits ciphertext malleability. For information about other similar attacks, see .
Avoiding Leaks From PKCS#1 Errors The PKCS#1 padding (used in RSA-encrypted and ElGamal-encrypted PKESK) has been found to be vulnerable to attacks in which a system that allows distinguishing padding errors from other decryption errors can act as a decryption and/or signing oracle that can leak the session key or allow signing arbitrary data, respectively . The number of queries required to carry out an attack can range from thousands to millions, depending on how strict and careful an implementation is in processing the padding. To make the attack more difficult, an implementation SHOULD implement strict, robust, constant time padding checks. To prevent the attack, in settings where the attacker does not have access to timing information concerning message decryption, the simplest solution is to report a single error code for all variants of PKESK processing errors as well as SEIPD integrity errors (this includes also session key parsing errors, such as on invalid cipher algorithm for v3 PKESK, or session key size mismatch for v6 PKESK). If the attacker may have access to timing information, then a constant time solution is also needed. This requires careful design, especially for v3 PKESK, where session key size and cipher information is typically not known in advance, as it is part of the PKESK encrypted payload.
Fingerprint Usability This specification uses fingerprints in several places on the wire (e.g., , , and ), and in processing (e.g., in ECDH KDF ). An implementation may also use the fingerprint internally, for example as an index to a keystore. Additionally, some OpenPGP users have historically used manual fingerprint comparison to verify the public key of a peer. For a version 4 fingerprint, this has typically been done with the fingerprint represented as 40 hexadecimal digits, often broken into groups of four digits with whitespace between each group. When a human is actively involved, the result of such a verification is dubious. There is little evidence that most humans are good at precise comparison of high-entropy data, particularly when that data is represented in compact textual form like a hexadecimal fingerprint. The version 6 fingerprint makes the challenge for a human verifier even worse. At 256 bits (compared to v4's 160 bit fingerprint), a v6 fingerprint is even harder for a human to successfully compare. An OpenPGP implementation should prioritize mechanical fingerprint transfer and comparison where possible, and SHOULD NOT promote manual transfer or comparison of full fingerprints by a human unless there is no other way to achieve the desired result. While this subsection acknowledges existing practice for human-representable v4 fingerprints, this document does not attempt to standardize any specific human-readable form of v6 fingerprint for this discouraged use case. NOTE: the topic of interoperable human-in-the-loop key verification needs more work, to be done in a separate document.
Avoiding Ciphertext Malleability If ciphertext can be modified by an attacker but still subsequently decrypted to some new plaintext, it is considered "malleable". A number of attacks can arise in any cryptosystem that uses malleable encryption, so and later versions of OpenPGP offer mechanisms to defend against it. However, OpenPGP data may have been created before these defense mechanisms were available. Because OpenPGP implementations deal with historic stored data, they may encounter malleable ciphertexts. When an OpenPGP implementation discovers that it is decrypting data that appears to be malleable, it MUST indicate a clear error message that the integrity of the message is suspect, SHOULD NOT attempt to parse nor release decrypted data to the user, and SHOULD halt with an error. Parsing or releasing decrypted data before having confirmed its integrity can leak the decrypted data , . In the case of AEAD encrypted data, if the authentication tag fails to verify, the implementation MUST NOT attempt to parse nor release decrypted data to the user, and MUST halt with an error. An implementation that encounters malleable ciphertext MAY choose to release cleartext to the user if it is not encrypted using AEAD, and it is known to be dealing with historic archived legacy data, and the user is aware of the risks. In the case of AEAD encrypted messages, if the message is truncated, i.e. the final zero-octet chunk and possibly (part of) some chunks before it are missing, the implementation MAY choose to release cleartext from fully authenticated chunks before it to the user if it is operating in a streaming fashion, but it MUST indicate a clear error message as soon as the truncation is detected. Any of the following OpenPGP data elements indicate that malleable ciphertext is present: All Symmetrically Encrypted Data packets (). Within any encrypted container, any Compressed Data packet () where there is a decompression failure. Any version 1 Symmetrically Encrypted Integrity Protected Data packet () where the internal Modification Detection Code does not validate. Any version 2 Symmetrically Encrypted Integrity Protected Data packet () where the authentication tag of any chunk fails, or where there is no final zero-octet chunk. Any Secret-Key packet with encrypted secret key material () where there is an integrity failure, based on the value of the secret key protection octet: Value 255 (MalleableCFB) or raw cipher algorithm: where the trailing 2-octet checksum does not match. Value 254 (CFB): where the SHA1 checksum is mismatched. Value 253 (AEAD): where the AEAD authentication tag is invalid. To avoid these circumstances, an implementation that generates OpenPGP encrypted data SHOULD select the encrypted container format with the most robust protections that can be handled by the intended recipients. In particular: The SED packet is deprecated, and MUST NOT be generated. When encrypting to one or more public keys: If all recipient keys indicate support for version 2 of the Symmetrically Encrypted Integrity Protected Data packet in their Features subpacket (), or are v6 keys without a Features subpacket, or the implementation can otherwise infer that all recipients support v2 SEIPD packets, the implementation SHOULD encrypt using a v2 SEIPD packet. If one of the recipients does not support v2 SEIPD packets, then the message generator MAY use a v1 SEIPD packet instead. Passphrase-protected secret key material in a v6 Secret Key or v6 Secret Subkey packet SHOULD be protected with AEAD encryption (S2K usage octet 253) unless it will be transferred to an implementation that is known to not support AEAD. An implementation should be aware that, in scenarios where an attacker has write access to encrypted private keys, CFB-encrypted keys (S2K usage octet 254 or 255) are vulnerable to corruption attacks that can cause leakage of secret data when the secret key is used , . Implementers should implement AEAD (v2 SEIPD and S2K usage octet 253) promptly and encourage its spread. Users are RECOMMENDED to migrate to AEAD.
Secure Use of the v2 SEIPD Session-Key-Reuse Feature The salted key derivation of v2 SEIPD packets () allows the recipient of an encrypted message to reply to the sender and all other recipients without needing their public keys but by using the same v6 PKESK packets he received and a different random salt value. This ensures a secure mechanism on the cryptographic level that enables the use of message encryption in cases where a sender does not have a copy of an encryption-capable certificate for one or more participants in the conversation and thus can enhance the overall security of an application. However, care must be taken when using this mechanism not to create security vulnerabilities, such as the following. Replying to only a subset of the original recipients and the original sender by use of the session-key-reuse feature would mean that the remaining recipients (including the sender) of the original message could read the encrypted reply message, too. Adding a further recipient to the reply that is encrypted using the session-key-reuse feature gives that further recipient also cryptographic access to the original message that is being replied to (and potentially to a longer history of previous messages). A modification of the list of recipients addressed in the above points needs also to be safeguarded when a message is initially composed as a reply with session-key reuse but then first stored (e.g. as a draft) and later reopened for further editing and finally sent. There is the potential threat that an attacker with network or mailbox access, who is at the same time a recipient of the original message, silently removes themselves from the message before the victim's client receives it. The victim's client that then uses the mechanism for replying with session-key reuse would unknowingly compose an encrypted message that could be read by the attacker. Implementations are encouraged to use the Intended Recipient Fingerprint () subpacket when composing messages and to use it to check the consistency of the set of recipients of a message before replying to it with session-key reuse. When using the session-key-reuse feature in any higher-layer protocol, care should be taken that there is no other potentially interfering practice of session-key reuse established in that protocol. Such interfering session-key reuse could for instance be given if an initial message is already composed by reusing the session key of an existing encrypted file the access to which may be shared among a group of users already. Using the session-key-reuse feature to compose an encrypted reply to such a message would unknowingly give this whole group of users cryptographic access to the encrypted message. Generally, the use of the session-key-reuse feature should be under the control of the user. Specifically, care should be taken that this feature is not silently used when the user assumes that proper public-key encryption is used. This can be the case for instance when the public key of one of the recipients of the reply is known but has expired. Special care should be taken to ensure that users do not get caught in continued use of the session-key reuse unknowingly but instead receive the chance to switch to proper fresh public-key encryption as soon as possible. Whenever possible, a client should prefer a fresh public key encryption over the session-key reuse. Even though this not necessarily being a security aspect, note that initially composing an encrypted reply using the session-key-reuse feature on one client and storing it (e.g. as a draft) and later reopening the stored unfinished reply with another client that does not support the session-key-reuse feature may lead to interoperability problems. Avoiding the pitfalls described above requires context-specific expertise. An implementation should only make use of the session-key-reuse feature in any particular application layer when it can follow reasonable documentation about how to deploy the feature safely in the specific application. At the time of this writing, there is no known documentation about safe reuse of OpenPGP session keys for any specific context. An implementer that intends to make use of this feature should publish their own proposed guidance for others to review.
Escrowed Revocation Signatures A keyholder, Alice, may wish to designate a third party to be able to revoke Alice's own key. The preferred way for her to do this is to produce a specific Revocation Signature (signature type IDs 0x20, 0x28, or 0x30) and distribute it securely to her preferred revoker who can hold it in escrow. The preferred revoker can then publish the escrowed Revocation Signature at whatever time is deemed appropriate, rather than generating a revocation signature themselves. There are multiple advantages of using an escrowed Revocation Signature over the deprecated Revocation Key subpacket (): The keyholder can constrain what types of revocation the preferred revoker can issue, by only escrowing those specific signatures. There is no public/visible linkage between the keyholder and the preferred revoker. Third parties can verify the revocation without needing to find the key of the preferred revoker. The preferred revoker doesn't even need to have a public OpenPGP key if some other secure transport is possible between them and the keyholder. Implementation support for enforcing a revocation from an authorized Revocation Key subpacket is uneven and unreliable. If the fingerprint mechanism suffers a cryptanalytic flaw, the escrowed Revocation Signature is not affected. A Revocation Signature may also be split up into shares and distributed among multiple parties, requiring some subset of those parties to collaborate before the escrowed Revocation Signature is recreated.
Random Number Generation and Seeding OpenPGP requires a cryptographically secure pseudorandom number generator (CSPRNG). In most cases, the operating system provides an appropriate facility such as a getrandom() syscall on Linux or BSD, which should be used absent other (for example, performance) concerns. It is RECOMMENDED to use an existing CSPRNG implementation in preference to crafting a new one. Many adequate cryptographic libraries are already available under favorable license terms. Should those prove unsatisfactory, provides guidance on the generation of random values. OpenPGP uses random data with three different levels of visibility: In publicly-visible fields such as nonces, IVs, public padding material, or salts, In shared-secret values, such as session keys for encrypted data or padding material within an encrypted packet, and In entirely private data, such as asymmetric key generation. With a properly functioning CSPRNG, this range of visibility does not present a security problem, as it is not feasible to determine the CSPRNG state from its output. However, with a broken CSPRNG, it may be possible for an attacker to use visible output to determine the CSPRNG internal state and thereby predict less-visible data like keying material, as documented in . An implementation can provide extra security against this form of attack by using separate CSPRNGs to generate random data with different levels of visibility.
Traffic Analysis When sending OpenPGP data through the network, the size of the data may leak information to an attacker. There are circumstances where such a leak could be unacceptable from a security perspective. For example, if possible cleartext messages for a given protocol are known to be either yes (three octets) and no (two octets) and the messages are sent within a Symmetrically-Encrypted Integrity Protected Data packet, the length of the encrypted message will reveal the contents of the cleartext. In another example, sending an OpenPGP Transferable Public Key over an encrypted network connection might reveal the length of the certificate. Since the length of an OpenPGP certificate varies based on the content, an external observer interested in metadata (who is trying to contact whom) may be able to guess the identity of the certificate sent, if its length is unique. In both cases, an implementation can adjust the size of the compound structure by including a Padding packet (see ).
Surreptitious Forwarding When an attacker obtains a signature for some text, e.g. by receiving a signed message, they may be able to use that signature maliciously by sending a message purporting to come from the original sender, with the same body and signature, to a different recipient. To prevent this, an implementation SHOULD implement the Intended Recipient Fingerprint signature subpacket ().
Hashed vs. Unhashed Subpackets Each OpenPGP signature can have subpackets in two different sections. The first set of subpackets (the "hashed section") is covered by the signature itself. The second set has no cryptographic protections, and is used for advisory material only, including locally-stored annotations about the signature. For example, consider an implementation working with a specific signature that happens to know that the signature was made by a certain key, even though the signature contains no Issuer Fingerprint subpacket () in the hashed section. That implementation MAY synthesize an Issuer Fingerprint subpacket and store it in the unhashed section so that in the future it will be able to recall which key issued the signature. Some subpackets are only useful when they are in the hashed section, and an implementation SHOULD ignore them when they are found with unknown provenance in the unhashed section. For example, a Preferred AEAD Ciphersuites subpacket () in a direct key self-signature indicates the preferences of the key holder when encrypting SEIPD v2 data to the key. An implementation that observes such a subpacket found in the unhashed section would open itself to an attack where the recipient's certificate is tampered with to encourage the use of a specific cipher or mode of operation.
Malicious Compressed Data It is possible to form a compression quine that produces itself upon decompression, leading to infinite regress in any implementation willing to parse arbitrary numbers of layers of compression. This could cause resource exhaustion which itself could lead to it being terminated by the operating system. If the operating system would create a "crash report", that report could contain confidential information. An OpenPGP implementation SHOULD limit the number of layers of compression it is willing to decompress in a single message.
Implementation Considerations This section is a collection of comments to help an implementer, particularly with an eye to backward compatibility. Often the differences are small, but small differences are frequently more vexing than large differences. Thus, this is a non-comprehensive list of potential problems and gotchas for a developer who is trying to be backward-compatible. There are many ways possible for two keys to have the same key material, but different fingerprints (and thus Key IDs). For example, since a v4 fingerprint is constructed by hashing the key creation time along with other things, two v4 keys created at different times, yet with the same key material will have different fingerprints. OpenPGP does not put limits on the size of public keys. However, larger keys are not necessarily better keys. Larger keys take more computation time to use, and this can quickly become impractical. Different OpenPGP implementations may also use different upper bounds for public key sizes, and so care should be taken when choosing sizes to maintain interoperability. ASCII armor is an optional feature of OpenPGP. The OpenPGP working group strives for a minimal set of mandatory-to-implement features, and since there could be useful implementations that only use binary object formats, this is not a "MUST" feature for an implementation. For example, an implementation that is using OpenPGP as a mechanism for file signatures may find ASCII armor unnecessary. OpenPGP permits an implementation to declare what features it does and does not support, but ASCII armor is not one of these. Since most implementations allow binary and armored objects to be used indiscriminately, an implementation that does not implement ASCII armor may find itself with compatibility issues with general-purpose implementations. Moreover, implementations of OpenPGP-MIME already have a requirement for ASCII armor so those implementations will necessarily have support. What this document calls Legacy packet format is what older documents called the "old packet format". It is the packet format used by implementations predating . Older RFCs called the current OpenPGP packet format the "new packet format". This is the format introduced in and maintained through to this document.
Constrained Legacy Fingerprint Storage for v6 Keys Some OpenPGP implementations have fixed length constraints for key fingerprint storage that will not fit all 32 octets of a v6 fingerprint. For example, reserves 20 octets for each stored fingerprint. An OpenPGP implementation MUST NOT attempt to map any part of a v6 fingerprint to such a constrained field unless the relevant spec for the constrained environment has explicit guidance for storing a v6 fingerprint that distinguishes it from a v4 fingerprint. An implementation interacting with such a constrained field SHOULD directly calculate the v6 fingerprint from public key material and associated metadata instead of relying on the constrained field.
IANA Considerations This document obsoletes . IANA is requested to update all registration information that references to instead reference this RFC.
Rename "Pretty Good Privacy (PGP)" Protocol Group to "OpenPGP" IANA bundles a set of registries associated with a particular protocol into a "protocol group". This document requests IANA to update the name of the "Pretty Good Privacy (PGP)" protocol group (i.e., the group of registries described at https://www.iana.org/assignments/pgp-parameters/) to "OpenPGP". If renaming the protocol group results in new URLs for the registries in this protocol group, please arrange for a permanent redirection (e.g., HTTP 301) from the existing URLs to the new URLs. All further updates specified below are for registries within this same "OpenPGP" protocol group.
Registries to be Renamed and Updated IANA is requested to rename the "PGP String-to-Key (S2K)" registry to "String-to-Key (S2K) Types" and update its content to . IANA is requested to rename the "PGP Packet Types/Tags" registry to "Packet Types" and update its content to . IANA is requested to rename the "PGP User Attribute Types" registry to "User Attribute Subpacket Types" and update its content to . IANA is requested to rename the "Image Format Subpacket Types" registry to "Image Attribute Encoding Format" and update its content to . IANA is requested to rename the "Key Server Preference Extensions" registry to "Key Server Preference Flags" and update its contents to . IANA is requested to rename the "Reason for Revocation Extensions" registry to "Reason for Revocation Code" and update its contents to . IANA is requested to rename the "Key Flags Extensions" registry to "Key Flags" and update its contents to . IANA is requested to rename the "Implementation Features" registry to "Features Flags" and update its contents to .
Registries to be Updated IANA is requested to update the "Public Key Algorithms" registry with the contents of . IANA is requested to update the "Symmetric Key Algorithms" registry with the contents of . IANA is requested to update the "Compression Algorithms" registry with the contents of . IANA is requested to update the "Hash Algorithms" registry with the contents of . IANA is requested to update the "Signature Subpacket Types" registry with the contents of .
Registries to be Removed IANA is requested to remove the empty "New Packet Versions" registry. A tombstone note should be added to the OpenPGP protocol group with the following content: Those wishing to use the removed "New Packet Versions" registry should instead register new versions of the relevant packets in the "Key and Signature Versions", "Key ID and Fingerprint" and "Encrypted Message Packet Versions" registries.
Registries to be Added IANA is requested to add the following registries in the OpenPGP protocol group: Secret Key Encryption (S2K Usage Octet) containing . Signature Types containing . Signature Notation Data Subpacket Notation Flags containing . Signature Notation Data Subpacket Types containing . Key ID and Fingerprint containing . Image Attribute Version containing . Armor Header Line containing . Armor Header Key containing . ECC Curve OID and Usage containing . ECC Curve-specific Wire Formats containing . Hash Algorithm Identifiers for RSA Signatures use of EMSA-PKCS1-v1_5 Padding containing . AEAD Algorithms containing . Encrypted Message Packet Versions containing . Key and Signature Versions containing . Elliptic Curve Point Wire Formats containing . Elliptic Curve Scalar Encodings containing . ECDH KDF and KEK Parameters containing .
Registration Policies IANA is requested to set all registries within the OpenPGP protocol group to use the SPECIFICATION REQUIRED registration policy, see with the exception of the registries listed in , below. This policy means that review and approval by a designated expert is required, and that the IDs and their meanings must be documented in a permanent and readily available public specification, in sufficient detail so that interoperability between independent implementations is possible.
Registries that are RFC REQUIRED The following registries use the RFC REQUIRED registration policy, as described in : Packet Types registry () Key and Signature Versions registry () Key ID and Fingerprint registry () Encrypted Message Packet Versions registry ()
Designated Experts The designated experts will determine whether the new registrations retain the security properties that are expected by the base implementation and that these new registrations do not cause interoperability issues with existing implementations other than not producing or consuming the IDs associated with these new registrations. Registration proposals that fail to meet these criteria could instead be proposed as new work items for the OpenPGP working group or its successor. The subsections below describe specific guidance for classes of registry updates that a designated expert will consider. The designated experts should also consider when reviewing proposed additions to the OpenPGP registries.
Key and Signature Versions When defining a new version of OpenPGP keys or signatures, should be updated, When a new version of OpenPGP key is defined, should also be updated.
Encryption Versions When defining a new version of the Symmetrically Encrypted Integrity Protected Data Packet (), Public Key Encrypted Session Key Packet (), and/or Symmetric Key Encrypted Session Key Packet (), the registry from needs to be updated. When the SEIPD is updated, consider also adding a corresponding flag to .
Algorithms lists the cryptographic and compression algorithms that OpenPGP uses. Adding new algorithms is usually simple, in some cases as little as allocating an ID and pointing to a reference. But some algorithm registries require some subtle additional details when a new algorithm is introduced.
Elliptic Curve Algorithms To register a new elliptic curve for use with OpenPGP, its OID needs to be registered in , its wire format needs to be documented in , and if used for ECDH, its KDF and KEK parameters must be populated in . If the wire format(s) used are not already defined in or , they should be defined there as well.
Symmetric-Key Algorithms When registering a new symmetric cipher with a block size of 64 or 128 bits and a key size that is a multiple of 64 bits, no new considerations are needed. If the new cipher has a different block size, there needs to be additional documentation describing how to use the cipher in CFB mode. If the new cipher has an unusual key size, then padding needs to be considered for X25519 and X448 keywrap, which currently needs no padding.
Hash Algorithms When registering a new hash algorithm (in ), if the algorithm is also to be used with RSA signing schemes, it must also have an entry in .
Description of a New Variable-Length Key, 64-Bit Block Cipher (Blowfish) The Bzip2 and libbzip2 home page A Conventional Authenticated-Encryption Mode A Public-Key Cryptosystem and a Signature Scheme Based on Discrete Logarithms On the design and security of block ciphers Information Technology - Universal Multiple-octet coded Character Set (UCS) - Part 1: Architecture and Basic Multilingual Plane International Organization for Standardization Information technology – Digital compression and coding of continuous-tone still images: JPEG File Interchange Format (JFIF) International Telecommunication Union The MD5 Message-Digest Algorithm This document describes the MD5 message-digest algorithm. The algorithm takes as input a message of arbitrary length and produces as output a 128-bit "fingerprint" or "message digest" of the input. This memo provides information for the Internet community. It does not specify an Internet standard. ZLIB Compressed Data Format Specification version 3.3 This specification defines a lossless compressed data format. This memo provides information for the Internet community. This memo does not specify an Internet standard of any kind. DEFLATE Compressed Data Format Specification version 1.3 This specification defines a lossless compressed data format that compresses data using a combination of the LZ77 algorithm and Huffman coding, with efficiency comparable to the best currently available general-purpose compression methods. This memo provides information for the Internet community. This memo does not specify an Internet standard of any kind. The CAST-128 Encryption Algorithm There is a need in the Internet community for an unencumbered encryption algorithm with a range of key sizes that can provide security for a variety of cryptographic applications and protocols. This document describes an existing algorithm that can be used to satisfy this requirement. This memo provides information for the Internet community. This memo does not specify an Internet standard of any kind. Internet Message Format This document specifies a syntax for text messages that are sent between computer users, within the framework of "electronic mail" messages. [STANDARDS-TRACK] PKCS #5: Password-Based Cryptography Specification Version 2.0 This document provides recommendations for the implementation of password-based cryptography, covering key derivation functions, encryption schemes, message-authentication schemes, and ASN.1 syntax identifying the techniques. This memo provides information for the Internet community. MIME Security with OpenPGP This document describes how the OpenPGP Message Format can be used to provide privacy and authentication using the Multipurpose Internet Mail Extensions (MIME) security content types described in RFC 1847. [STANDARDS-TRACK] Advanced Encryption Standard (AES) Key Wrap Algorithm UTF-8, a transformation format of ISO 10646 ISO/IEC 10646-1 defines a large character set called the Universal Character Set (UCS) which encompasses most of the world's writing systems. The originally proposed encodings of the UCS, however, were not compatible with many current applications and protocols, and this has led to the development of UTF-8, the object of this memo. UTF-8 has the characteristic of preserving the full US-ASCII range, providing compatibility with file systems, parsers and other software that rely on US-ASCII values but are transparent to other values. This memo obsoletes and replaces RFC 2279. A Description of the Camellia Encryption Algorithm This document describes the Camellia encryption algorithm. Camellia is a block cipher with 128-bit block size and 128-, 192-, and 256-bit keys. The algorithm description is presented together with key scheduling part and data randomizing part. This memo provides information for the Internet community. Randomness Requirements for Security Security systems are built on strong cryptographic algorithms that foil pattern analysis attempts. However, the security of these systems is dependent on generating secret quantities for passwords, cryptographic keys, and similar quantities. The use of pseudo-random processes to generate secret quantities can result in pseudo-security. A sophisticated attacker may find it easier to reproduce the environment that produced the secret quantities and to search the resulting small set of possibilities than to locate the quantities in the whole of the potential number space. Choosing random quantities to foil a resourceful and motivated adversary is surprisingly difficult. This document points out many pitfalls in using poor entropy sources or traditional pseudo-random number generation techniques for generating such quantities. It recommends the use of truly random hardware techniques and shows that the existing hardware on many systems can be used for this purpose. It provides suggestions to ameliorate the problem when a hardware solution is not available, and it gives examples of how large such quantities need to be for some applications. This document specifies an Internet Best Current Practices for the Internet Community, and requests discussion and suggestions for improvements. The Base16, Base32, and Base64 Data Encodings This document describes the commonly used base 64, base 32, and base 16 encoding schemes. It also discusses the use of line-feeds in encoded data, use of padding in encoded data, use of non-alphabet characters in encoded data, use of different encoding alphabets, and canonical encodings. [STANDARDS-TRACK] US Secure Hash Algorithms (SHA and SHA-based HMAC and HKDF) Federal Information Processing Standard, FIPS The OCB Authenticated-Encryption Algorithm This document specifies OCB, a shared-key blockcipher-based encryption scheme that provides confidentiality and authenticity for plaintexts and authenticity for associated data. This document is a product of the Crypto Forum Research Group (CFRG). Elliptic Curves for Security This memo specifies two elliptic curves over prime fields that offer a high level of practical security in cryptographic applications, including Transport Layer Security (TLS). These curves are intended to operate at the ~128-bit and ~224-bit security level, respectively, and are generated deterministically based on a list of required properties. PKCS #1: RSA Cryptography Specifications Version 2.2 This document provides recommendations for the implementation of public-key cryptography based on the RSA algorithm, covering cryptographic primitives, encryption schemes, signature schemes with appendix, and ASN.1 syntax for representing keys and for identifying the schemes. This document represents a republication of PKCS #1 v2.2 from RSA Laboratories' Public-Key Cryptography Standards (PKCS) series. By publishing this RFC, change control is transferred to the IETF. This document also obsoletes RFC 3447. Edwards-Curve Digital Signature Algorithm (EdDSA) This document describes elliptic curve signature scheme Edwards-curve Digital Signature Algorithm (EdDSA). The algorithm is instantiated with recommended parameters for the edwards25519 and edwards448 curves. An example implementation and test vectors are provided. Guidelines for Writing an IANA Considerations Section in RFCs Many protocols make use of points of extensibility that use constants to identify various protocol parameters. To ensure that the values in these fields do not have conflicting uses and to promote interoperability, their allocations are often coordinated by a central record keeper. For IETF protocols, that role is filled by the Internet Assigned Numbers Authority (IANA). To make assignments in a given registry prudently, guidance describing the conditions under which new values should be assigned, as well as when and how modifications to existing values can be made, is needed. This document defines a framework for the documentation of these guidelines by specification authors, in order to assure that the provided guidance for the IANA Considerations is clear and addresses the various issues that are likely in the operation of a registry. This is the third edition of this document; it obsoletes RFC 5226. Argon2 Memory-Hard Function for Password Hashing and Proof-of-Work Applications This document describes the Argon2 memory-hard function for password hashing and proof-of-work applications. We provide an implementer-oriented description with test vectors. The purpose is to simplify adoption of Argon2 for Internet protocols. This document is a product of the Crypto Forum Research Group (CFRG) in the IRTF. Information technology - Security techniques - Hash-functions - Part 3: Dedicated hash-functions, International Organization for Standardization, Geneva, Switzerland Recommendation for Block Cipher Modes of Operation: Methods and Techniques Recommendation for Block Cipher Modes of Operation: Galois/Counter Mode (GCM) and GMAC Recommendation for Pair-Wise Key Establishment Schemes Using Discrete Logarithm Cryptography Recommendation for the Triple Data Encryption Algorithm (TDEA) Block Cipher NIST The Twofish Encryption Algorithm Key words for use in RFCs to Indicate Requirement Levels In many standards track documents several words are used to signify the requirements in the specification. These words are often capitalized. This document defines these words as they should be interpreted in IETF documents. This document specifies an Internet Best Current Practices for the Internet Community, and requests discussion and suggestions for improvements. Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words RFC 2119 specifies common key words that may be used in protocol specifications. This document aims to reduce the ambiguity by clarifying that only UPPERCASE usage of the key words have the defined special meanings. Digital Signature Standard (DSS) National Institute of Standards and Technology Advanced encryption standard (AES) Secure Hash Standard SHA-3 Standard: Permutation-Based Hash and Extendable-Output Functions Generating ElGamal Signatures Without Knowing the Secret Key Chosen Ciphertext Attacks Against Protocols Based on the RSA Encryption Standard PKCS \#1 Programming languages - C: C99, correction 3:2007, ISO/IEC 9899:1999/Cor 3:2007 Efail: Breaking S/MIME and OpenPGP Email Encryption using Exfiltration Channels Errata Report 2199 - S2K hash/cipher octet correction Errata Report 2200 - No implicit use of IDEA correction Errata Report 2206 - PKESK acronym expansion Errata Report 2208 - Signature key owner clarification Errata Report 2214 - Signature hashing clarification Errata Report 2216 - Self signature applies to user ID correction Errata Report 2219 - Session key encryption storage clarification Errata Report 2222 - Simple hash MUST/MAY clarification Errata Report 2226 - Native line endings SHOULD clarification Errata Report 2234 - Radix-64 / base64 clarification Errata Report 2235 - ASCII / UTF-8 collation sequence clarification Errata Report 2236 - Packet Composition is a sequence clarification Errata Report 2238 - Subkey packets come after all User ID packets clarification Errata Report 2240 - Subkey removal clarification Errata Report 2242 - mL / emLen variable correction Errata Report 2243 - CFB mode initialization vector (IV) clarification Errata Report 2270 - SHA-224 octet sequence correction Errata Report 2271 - Radix-64 correction Errata Report 3298 - Key revocation signatures correction Errata Report 5491 - C code fix for CRC24_POLY define Errata Report 7545 - Armor Header colon hex fix Solving Simultaneous Modular Equations of Low Degree Implementation of Chosen-Ciphertext Attacks against PGP and GnuPG A course in number theory and cryptography, Chapter VI. Elliptic Curves Victory by KO: Attacking OpenPGP Using Key Overwriting Attack on Private Signature Keys of the OpenPGP Format, PGP(TM) Programs and Other Applications Compatible with OpenPGP Format Oracles on OpenPGP An Attack on CFB Mode Encryption As Used By OpenPGP Functional Specification of the OpenPGP application on ISO Smart Card Operating Systems (version 3.4.1) IEEE Standard for Information Technology--Portable Operating System Interface (POSIX(R)) Base Specifications, Issue 7: pax - portable archive interchange The Open Group Attacking Deterministic Signature Schemes using Fault Attacks Mastering Regular Expressions O'Reilly PGP Message Exchange Formats This document describes the format of "PGP files", i.e., messages that have been encrypted and/or signed with PGP. This memo provides information for the Internet community. This memo does not specify an Internet standard of any kind. OpenPGP Message Format This document is maintained in order to publish all necessary information needed to develop interoperable applications based on the OpenPGP format. [STANDARDS-TRACK] OpenPGP Message Format This document is maintained in order to publish all necessary information needed to develop interoperable applications based on the OpenPGP format. It is not a step-by-step cookbook for writing an application. It describes only the format and methods needed to read, check, generate, and write conforming packets crossing any network. It does not deal with storage and implementation questions. It does, however, discuss implementation issues necessary to avoid security flaws. OpenPGP software uses a combination of strong public-key and symmetric cryptography to provide security services for electronic communications and data storage. These services include confidentiality, key management, authentication, and digital signatures. This document specifies the message formats used in OpenPGP. [STANDARDS-TRACK] The Camellia Cipher in OpenPGP This document presents the necessary information to use the Camellia symmetric block cipher in the OpenPGP protocol. This memo provides information for the Internet community. Elliptic Curve Cryptography (ECC) Brainpool Standard Curves and Curve Generation This memo proposes several elliptic curve domain parameters over finite prime fields for use in cryptographic applications. The domain parameters are consistent with the relevant international standards, and can be used in X.509 certificates and certificate revocation lists (CRLs), for Internet Key Exchange (IKE), Transport Layer Security (TLS), XML signatures, and all applications or protocols based on the cryptographic message syntax (CMS). This document is not an Internet Standards Track specification; it is published for informational purposes. HMAC-based Extract-and-Expand Key Derivation Function (HKDF) This document specifies a simple Hashed Message Authentication Code (HMAC)-based key derivation function (HKDF), which can be used as a building block in various protocols and applications. The key derivation function (KDF) is intended to support a wide range of applications and requirements, and is conservative in its use of cryptographic hash functions. This document is not an Internet Standards Track specification; it is published for informational purposes. Fundamental Elliptic Curve Cryptography Algorithms This note describes the fundamental algorithms of Elliptic Curve Cryptography (ECC) as they were defined in some seminal references from 1994 and earlier. These descriptions may be useful for implementing the fundamental algorithms without using any of the specialized methods that were developed in following years. Only elliptic curves defined over fields of characteristic greater than three are in scope; these curves are those used in Suite B. This document is not an Internet Standards Track specification; it is published for informational purposes. Elliptic Curve Cryptography (ECC) in OpenPGP This document defines an Elliptic Curve Cryptography extension to the OpenPGP public key format and specifies three Elliptic Curves that enjoy broad support by other standards, including standards published by the US National Institute of Standards and Technology. The document specifies the conventions for interoperability between compliant OpenPGP implementations that make use of this extension and these Elliptic Curves. [STANDARDS-TRACK] Standards for Efficient Cryptography 1 (SEC 1) Standards for Efficient Cryptography Group sha1collisiondetection Sha-1 is a shambles: First chosen-prefix collision on sha-1 and application to the PGP web of trust Recommendation on Key Management: Part 1 - General NIST Transitioning the Use of Cryptographic Algorithms and Key Lengths Counter-cryptanalysis Detailed Description of Unified Format Free Software Foundation IANA Charset Registration Procedures Multipurpose Internet Mail Extensions (MIME) and various other Internet protocols are capable of using many different charsets. This in turn means that the ability to label different charsets is essential. This document specifies an Internet Best Current Practices for the Internet Community, and requests discussion and suggestions for improvements. A Systematic Analysis of the Juniper Dual EC Incident University of Illinois at Chicago, Chicago, IL, USA UC San Diego, La Jolla, CA, USA Johns Hopkins University, Baltimore, MD, USA University of Pennsylvania, Philadelphia, PA, USA University of Pennsylvania, Philadelphia, PA, USA Johns Hopkins University, Baltimore, MD, USA University of Pennsylvania, Philadelphia, PA, USA Comsecuris, Duisburg, Germany UC San Diego, La Jolla, CA, USA UC San Diego, La Jolla, CA, USA
Test vectors To help implementing this specification a set of non-normative examples follow here.
Sample v4 Ed25519Legacy key The secret key used for this example is: D: 1a8b1ff05ded48e18bf50166c664ab023ea70003d78d9e41f5758a91d850f8d2 Note that this is the raw secret key used as input to the EdDSA signing operation. The key was created on 2014-08-19 14:28:27 and thus the fingerprint of the OpenPGP key is:
The algorithm-specific input parameters without the MPI length headers are:
The entire public key packet is thus:
The same packet, represented in ASCII-armored form is:
Sample v4 Ed25519Legacy signature The signature is created using the sample key over the input data "OpenPGP" on 2015-09-16 12:24:53 UTC and thus the input to the hash function is:
Using the SHA2-256 hash algorithm yields the digest:
Which is fed into the EdDSA signature function and yields this signature:
The entire signature packet is thus:
The same packet represented in ASCII-armored form is:
Sample v6 Certificate (Transferable Public Key) Here is a Transferable Public Key consisting of: A v6 Ed25519 Public-Key packet A v6 direct key self-signature A v6 X25519 Public-Subkey packet A v6 subkey binding signature The primary key has the fingerprint CB186C4F0609A697E4D52DFA6C722B0C1F1E27C18A56708F6525EC27BAD9ACC9. The subkey has the fingerprint 12C83F1E706F6308FE151A417743A1F033790E93E9978488D1DB378DA9930885.
The corresponding Transferable Secret Key can be found in .
Hashed Data Stream for Signature Verification The direct key self-signature in the certificate in is made over the following sequence of data:
The same data, broken out by octet and semantics, is:
The subkey binding signature in is made over the following sequence of data:
The same data, broken out by octet and semantics, is:
Sample v6 Secret Key (Transferable Secret Key) Here is a Transferable Secret Key consisting of: A v6 Ed25519 Secret-Key packet A v6 direct key self-signature A v6 X25519 Secret-Subkey packet A v6 subkey binding signature
The corresponding Transferable Public Key can be found in .
Sample locked v6 Secret Key (Transferable Secret Key) Here is the same secret key as in , but the secret key material is locked with a passphrase using AEAD and Argon2. The passphrase is the ASCII string:
Intermediate Data for Locked Primary Key The S2K-derived material for the primary key is:
After HKDF, the symmetric key used for AEAD encryption of the primary key is:
The additional data for AEAD for the primary key is:
Intermediate Data for Locked Subkey The S2K-derived key material for the subkey is:
After HKDF, the symmetric key used for AEAD encryption of the subkey is:
The additional data for AEAD for the subkey is:
Sample Cleartext Signed Message Here is a signed message that uses the cleartext signature framework (). It can be verified with the certificate from (). Note that this message makes use of dash-escaping () due to its contents.
The signature packet here is:
The signature is made over the following data:
The same data, broken out by octet and semantics, is:
The calculated SHA2-512 hash digest over this data is:
Sample inline-signed message This is the same message and signature as in , but as inline-signed message. The hashed data is exactly the same, and all intermediate values and annotated hex dumps are also applicable.
Sample X25519-AEAD-OCB encryption and decryption This example encrypts the cleartext string Hello, world! for the sample cert (see ), using AES-128 with AEAD-OCB encryption.
Sample public-key encrypted session key packet (v6) This packet contains the following series of octets:
The same data, broken out by octet and semantics:
X25519 encryption/decryption of the session key Ephemeral key:
This ephemeral key is derived from the following ephemeral secret key material, which is never placed on the wire:
Public key from target certificate (see ):
The corresponding long-lived X25519 private key material (see ):
Shared point:
HKDF output:
Decrypted session key:
Sample v2 SEIPD packet This packet contains the following series of octets:
The same data, broken out by octet and semantics:
Decryption of data Starting AEAD-OCB decryption of data, using the session key. HKDF info:
HKDF output:
Message key:
Initialization vector:
Chunk #0: Nonce:
Additional authenticated data:
Encrypted data chunk:
Decrypted chunk #0. Literal data packet with the string contents Hello, world!:
Padding packet:
Authenticating final tag: Final nonce:
Final additional authenticated data:
Complete X25519-AEAD-OCB encrypted packet sequence
Sample AEAD-EAX encryption and decryption This example encrypts the cleartext string Hello, world! with the passphrase password, using AES-128 with AEAD-EAX encryption.
Sample symmetric-key encrypted session key packet (v6) This packet contains the following series of octets:
The same data, broken out by octet and semantics:
Starting AEAD-EAX decryption of the session key The derived key is:
HKDF info:
HKDF output:
Authenticated Data:
Nonce:
Decrypted session key:
Sample v2 SEIPD packet This packet contains the following series of octets:
The same data, broken out by octet and semantics:
Decryption of data Starting AEAD-EAX decryption of data, using the session key. HKDF info:
HKDF output:
Message key:
Initialization vector:
Chunk #0: Nonce:
Additional authenticated data:
Decrypted chunk #0. Literal data packet with the string contents Hello, world!:
Padding packet:
Authenticating final tag: Final nonce:
Final additional authenticated data:
Complete AEAD-EAX encrypted packet sequence
Sample AEAD-OCB encryption and decryption This example encrypts the cleartext string Hello, world! with the passphrase password, using AES-128 with AEAD-OCB encryption.
Sample symmetric-key encrypted session key packet (v6) This packet contains the following series of octets:
The same data, broken out by octet and semantics:
Starting AEAD-OCB decryption of the session key The derived key is:
HKDF info:
HKDF output:
Authenticated Data:
Nonce:
Decrypted session key:
Sample v2 SEIPD packet This packet contains the following series of octets:
The same data, broken out by octet and semantics:
Decryption of data Starting AEAD-OCB decryption of data, using the session key. HKDF info:
HKDF output:
Message key:
Initialization vector:
Chunk #0: Nonce:
Additional authenticated data:
Decrypted chunk #0. Literal data packet with the string contents Hello, world!:
Padding packet:
Authenticating final tag: Final nonce:
Final additional authenticated data:
Complete AEAD-OCB encrypted packet sequence
Sample AEAD-GCM encryption and decryption This example encrypts the cleartext string Hello, world! with the passphrase password, using AES-128 with AEAD-GCM encryption.
Sample symmetric-key encrypted session key packet (v6) This packet contains the following series of octets:
The same data, broken out by octet and semantics:
Starting AEAD-GCM decryption of the session key The derived key is:
HKDF info:
HKDF output:
Authenticated Data:
Nonce:
Decrypted session key:
Sample v2 SEIPD packet This packet contains the following series of octets:
The same data, broken out by octet and semantics:
Decryption of data Starting AEAD-GCM decryption of data, using the session key. HKDF info:
HKDF output:
Message key:
Initialization vector:
Chunk #0: Nonce:
Additional authenticated data:
Decrypted chunk #0. Literal data packet with the string contents Hello, world!:
Padding packet:
Authenticating final tag: Final nonce:
Final additional authenticated data:
Complete AEAD-GCM encrypted packet sequence
Sample messages encrypted using Argon2 These messages are the literal data "Hello, world!" encrypted using v1 SEIPD, with Argon2 and the passphrase "password", using different session key sizes. In each example, the choice of symmetric cipher is the same in both the v4 SKESK packet and v1 SEIPD packet. In all cases, the Argon2 parameters are t = 1, p = 4, and m = 21.
Version 4 SKESK using Argon2 with AES-128
Version 4 SKESK using Argon2 with AES-192
Version 4 SKESK using Argon2 with AES-256
Upgrade Guidance (Adapting Implementations from RFC 4880 and RFC 6637) This subsection offers a concise, non-normative summary of the substantial additions to and departures from and . It is intended to help implementers who are augmenting an existing implementation from those standards to this standard. Cryptographic algorithms marked with "MTI" are mandatory to implement. Public Key signing algorithms: Ed25519 ( and ), MTI Ed448 ( and ) EdDSALegacy with Ed25519Legacy ( and ) ECDSA with Brainpool curves () Public Key encryption algorithms: X25519 ( and ), MTI X448 ( and ) ECDH with Curve25519Legacy () ECDH with Brainpool curves () AEAD Encryption: Version 2 SEIPD () AEAD modes: OCB mode (), MTI EAX mode () GCM mode () Version 6 PKESK () Version 6 SKESK () Features subpacket: add flag for SEIPDv2 () Subpacket: Preferred AEAD Ciphersuites () Secret key encryption: AEAD "S2K usage octet" ( and ) Version 6 Keys and Signatures: Version 6 Public keys () Version 6 Fingerprint and Key ID () Version 6 Secret keys () Version 6 Signatures () Version 6 One-Pass Signatures () Certificate (Transferable Public Key) Structure: Preferences subpackets in Direct Key Signatures () Self-verifying revocation certificate () User ID is explicitly optional () S2K: Argon2 () Subpacket: Intended Recipient Fingerprint () Digest algorithms: SHA3-256 and SHA3-512 () Packet: Padding () Message structure: Packet Criticality () Deprecations: Public Key Algorithms: Avoid RSA weak keys () Avoid DSA () Avoid ElGamal (, ) For Version 6 Keys: Avoid EdDSA25519Legacy, Curve25519Legacy () Digest Algorithms: Avoid MD5, SHA1, RIPEMD160 () Symmetric Key Algorithms: Avoid IDEA, TripleDES, CAST5 () S2K Specifier: Avoid Simple S2K () Secret Key protections (a.k.a. S2K Usage): Avoid MalleableCFB () Packet Types: Avoid Symmetrically-Encrypted Data (, ) Literal Data packet metadata: Avoid Filename and Date fields () Avoid Special _CONSOLE "filename" () Packet Versions: Avoid Version 3 Public Keys () Avoid Version 3 Signatures () Signature Types: Avoid Reserved Signature Type ID 0xFF (, ) Signature Subpackets: For Version 6 Signatures: Avoid Issuer Key ID () Avoid Revocation Key () ASCII Armor: Ignore, do not emit CRC () Do not emit "Version" armor header () Cleartext Signature Framework: Ignore, avoid emitting unnecessary Hash: headers () Reject CSF signatures with invalid Hash: headers () or any other Armor Header ()
Terminology Changes Note that some of the words used in previous revisions of the OpenPGP standard have been improved in this document. In previous revisions, the following terms were used: "Radix-64" was used to refer to OpenPGP's ASCII Armor base64 encoding (). "Old packet format" was used to refer to the Legacy packet format () predating . "New packet format" was used to refer to the OpenPGP packet format () introduced in . "Certificate" was used ambiguously to mean multiple things. In this document, it is used to mean "Transferable Public Key" exclusively. "Preferred Symmetric Algorithms" was the old name for the "Preferred Symmetric Ciphers for v1 SEIPD" subpacket () "Modification Detection Code" or "MDC" was originally described as a distinct packet (packet type ID 19), and its corresponding flag in the Features subpacket () was known as "Modification Detection". It is now described as an intrinsic part of v1 SEIPD (), and the same corresponding flag is known as "Symmetrically Encrypted Integrity Protected Data packet version 1". "Packet Tag" was used to refer to the Packet Type ID (), or sometimes to the encoded Packet Type ID ().
Acknowledgements Thanks to the openpgp design team for working on this document to prepare it for working group consumption: Stephen Farrell, Daniel Kahn Gillmor, Daniel Huigens, Jeffrey Lau, Yutaka Niibe, Justus Winter and Paul Wouters. Thanks to Werner Koch for the early work on rfc4880bis and Andrey Jivsov for . This document also draws on much previous work from a number of other authors, including: Derek Atkins, Charles Breed, Dave Del Torto, Marc Dyksterhouse, Gail Haspert, Gene Hoffman, Paul Hoffman, Ben Laurie, Raph Levien, Colin Plumb, Will Price, David Shaw, William Stallings, Mark Weaver, and Philip R. Zimmermann.
Errata addressed by this document The following verified errata have been incorporated or are otherwise resolved by this document: - S2K hash/cipher octet correction - No implicit use of IDEA correction - PKESK acronym expansion - Signature key owner clarification - Signature hashing clarification - Self signature applies to user ID correction - Session key encryption storage clarification - Simple hash MUST/MAY clarification - Native line endings SHOULD clarification - Radix-64 / base64 clarification - ASCII / UTF-8 collation sequence clarification - Packet Composition is a sequence clarification - Subkey packets come after all User ID packets clarification - Subkey removal clarification - mL / emLen variable correction - CFB mode initialization vector (IV) clarification - SHA-224 octet sequence correction - Radix-64 correction - Key revocation signatures correction - C code fix for CRC24_POLY define - Armor Header colon hex fix