Textual Encodings of PKIX, PKCS, and CMS Structures

Textual Encodings of PKIX, PKCS, and CMS Structures SJD AB

Johan Olof Wallins Vaeg 13 Solna 171 64 Sweden simon@josefsson.org http://josefsson.org/

Penango, Inc.

5900 Wilshire Boulevard 21st Floor Los Angeles CA 90036 United States dev+ietf@seantek.com http://www.penango.com/

textual encoding PKIX PKCS CMS base64 PEM encapsulation boundary X.509 certificate cert ASN.1 BER DER This document describes and discusses the textual encodings of the Public-Key Infrastructure X.509 (PKIX), Public-Key Cryptography Standards (PKCS), and Cryptographic Message Syntax (CMS). The textual encodings are well-known, are implemented by several applications and libraries, and are widely deployed. This document articulates the de facto rules by which existing implementations operate and defines them so that future implementations can interoperate.

Several security-related standards used on the Internet define ASN.1 data formats that are normally encoded using the Basic Encoding Rules (BER) or Distinguished Encoding Rules (DER), which are binary, octet-oriented encodings. This document is about the textual encodings of the following formats: Certificates, Certificate Revocation Lists (CRLs), and Subject Public Key Info structures in the Internet X.509 Public Key Infrastructure Certificate and Certificate Revocation List (CRL) Profile. PKCS #10: Certification Request Syntax. PKCS #7: Cryptographic Message Syntax. Cryptographic Message Syntax. PKCS #8: Private-Key Information Syntax, renamed to One Asymmetric Key in Asymmetric Key Package, and Encrypted Private-Key Information Syntax in the same documents. Attribute Certificates in An Internet Attribute Certificate Profile for Authorization. A disadvantage of a binary data format is that it cannot be interchanged in textual transports, such as email or text documents. One advantage with text-based encodings is that they are easy to modify using common text editors; for example, a user may concatenate several certificates to form a certificate chain with copy-and-paste operations. The tradition within the RFC series can be traced back to Privacy-Enhanced Mail (PEM), based on a proposal by Marshall Rose in Message Encapsulation. Originally called "PEM encapsulation mechanism", "encapsulated PEM message", or (arguably) "PEM printable encoding", today the format is sometimes referred to as "PEM encoding". Variations include OpenPGP ASCII armor and OpenSSH key file format. For reasons that basically boil down to non-coordination or inattention, many PKIX, PKCS, and CMS libraries implement a text-based encoding that is similar to -- but not identical with -- PEM encoding. This document specifies the _textual encoding_ format, articulates the de facto rules that most implementations operate by, and provides recommendations that will promote interoperability going forward. This document also provides common nomenclature for syntax elements, reflecting the evolution of this de facto standard format. Peter Gutmann's "X.509 Style Guide" contains a section "base64 Encoding" that describes the formats and contains suggestions similar to what is in this document. All figures are real, functional examples, with key lengths and inner contents chosen to be as small as practicable. 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 RFC 2119.

Textual encoding begins with a line comprising -----BEGIN , a label, and -----, and ends with a line comprising -----END , a label, and -----. Between these lines, or "encapsulation boundaries", are base64-encoded data according to Section 4 of . (PEM referred to this data as the "encapsulated text portion".) Data before the encapsulation boundaries are permitted, and parsers MUST NOT malfunction when processing such data. Furthermore, parsers SHOULD ignore whitespace and other non-base64 characters and MUST handle different newline conventions. The type of data encoded is labeled depending on the type label in the -----BEGIN line (pre-encapsulation boundary). For example, the line may be -----BEGIN CERTIFICATE----- to indicate that the content is a PKIX certificate (see further below). Generators MUST put the same label on the -----END line (post-encapsulation boundary) as the corresponding -----BEGIN line. Labels are formally case-sensitive, uppercase, and comprised of zero or more characters; they do not contain consecutive spaces or hyphen-minuses, nor do they contain spaces or hyphen-minuses at either end. Parsers MAY disregard the label in the post-encapsulation boundary instead of signaling an error if there is a label mismatch: some extant implementations require the labels to match; others do not. There is exactly one space character (SP) separating the BEGIN or END from the label. There are exactly five hyphen-minus (also known as dash) characters (-) on both ends of the encapsulation boundaries, no more, no less. The label type implies that the encoded data follows the specified syntax. Parsers MUST handle non-conforming data gracefully. However, not all parsers or generators prior to this document behave consistently. A conforming parser MAY interpret the contents as another label type but ought to be aware of the security implications discussed in the section. The labels described in this document identify container formats that are not specific to any particular cryptographic algorithm, a property consistent with algorithm agility. These formats use the ASN.1 AlgorithmIdentifier structure as described in Section 4.1.1.2 of . Unlike legacy PEM encoding, OpenPGP ASCII armor, and the OpenSSH key file format, textual encoding does *not* define or permit headers to be encoded alongside the data. Empty space can appear between the pre-encapsulation boundary and the base64, but generators SHOULD NOT emit such any such spacing. (The provision for this empty area is a throwback to PEM, which defined an "encapsulated header portion".) Implementers need to be aware that extant parsers diverge considerably on the handling of whitespace. In this document, "whitespace" means any character or series of characters that represent horizontal or vertical space in typography. In US-ASCII, whitespace means HT (0x09), VT (0x0B), FF (0x0C), SP (0x20), CR (0x0D), and LF (0x0A); "blank" means HT and SP; lines are divided with CRLF, CR, or LF. The common ABNF production WSP is congruent with "blank"; a new production W is used for "whitespace". The ABNF in is specific to US-ASCII. As these textual encodings can be used on many different systems as well as on long-term archival storage media such as paper or engravings, an implementer ought to use the spirit rather than the letter of the rules when generating or parsing these formats in environments that are not strictly limited to US-ASCII. Most extant parsers ignore blanks at the ends of lines; blanks at the beginnings of lines or in the middle of the base64-encoded data are far less compatible. These observations are codified in . The most lax parser implementations are not line-oriented at all and will accept any mixture of whitespace outside of the encapsulation boundaries (see ). Such lax parsing may run the risk of accepting text that was not intended to be accepted in the first place (e.g., because the text was a snippet or sample). Generators MUST wrap the base64-encoded lines so that each line consists of exactly 64 characters except for the final line, which will encode the remainder of the data (within the 64-character line boundary), and they MUST NOT emit extraneous whitespace. Parsers MAY handle other line sizes. These requirements are consistent with PEM. Files MAY contain multiple textual encoding instances. This is used, for example, when a file contains several certificates. Whether the instances are ordered or unordered depends on the context.

The ABNF of the textual encoding is:

= is not good, but is valid base64text = *base64line base64finl ; we could also use from RFC 1421, which requires ; 16 groups of 4 chars, which means exactly 64 chars per ; line, except the final line, but this is more accurate labelchar = %x21-2C / %x2E-7E ; any printable character, ; except hyphen-minus label = [ labelchar *( ["-" / SP] labelchar ) ] ; empty ok eol = CRLF / CR / LF eolWSP = WSP / CR / LF ; compare with LWSP]]>

New implementations SHOULD emit the strict format specified above. The choice of parsing strategy depends on the context of use.

For convenience, these figures summarize the structures, encodings, and references in the following sections:

* This OID does not actually appear in PKCS #7 v1.5. It was defined in the ASN.1 module to PKCS #7 v1.6, and has been carried forward through PKCS #12.

Public-key certificates are encoded using the CERTIFICATE label. The encoded data MUST be a BER (DER strongly preferred; see ) encoded ASN.1 Certificate structure as described in Section 4 of .

-----BEGIN CERTIFICATE----- MIICLDCCAdKgAwIBAgIBADAKBggqhkjOPQQDAjB9MQswCQYDVQQGEwJCRTEPMA0G A1UEChMGR251VExTMSUwIwYDVQQLExxHbnVUTFMgY2VydGlmaWNhdGUgYXV0aG9y aXR5MQ8wDQYDVQQIEwZMZXV2ZW4xJTAjBgNVBAMTHEdudVRMUyBjZXJ0aWZpY2F0 ZSBhdXRob3JpdHkwHhcNMTEwNTIzMjAzODIxWhcNMTIxMjIyMDc0MTUxWjB9MQsw CQYDVQQGEwJCRTEPMA0GA1UEChMGR251VExTMSUwIwYDVQQLExxHbnVUTFMgY2Vy dGlmaWNhdGUgYXV0aG9yaXR5MQ8wDQYDVQQIEwZMZXV2ZW4xJTAjBgNVBAMTHEdu dVRMUyBjZXJ0aWZpY2F0ZSBhdXRob3JpdHkwWTATBgcqhkjOPQIBBggqhkjOPQMB BwNCAARS2I0jiuNn14Y2sSALCX3IybqiIJUvxUpj+oNfzngvj/Niyv2394BWnW4X uQ4RTEiywK87WRcWMGgJB5kX/t2no0MwQTAPBgNVHRMBAf8EBTADAQH/MA8GA1Ud DwEB/wQFAwMHBgAwHQYDVR0OBBYEFPC0gf6YEr+1KLlkQAPLzB9mTigDMAoGCCqG SM49BAMCA0gAMEUCIDGuwD1KPyG+hRf88MeyMQcqOFZD0TbVleF+UsAGQ4enAiEA l4wOuDwKQa+upc8GftXE2C//4mKANBC6It01gUaTIpo= -----END CERTIFICATE----- Historically, the label X509 CERTIFICATE and also less commonly X.509 CERTIFICATE have been used. Generators conforming to this document MUST generate CERTIFICATE labels and MUST NOT generate X509 CERTIFICATE or X.509 CERTIFICATE labels. Parsers SHOULD NOT treat X509 CERTIFICATE or X.509 CERTIFICATE as equivalent to CERTIFICATE, but a valid exception may be for backwards compatibility (potentially together with a warning).

Many tools are known to emit explanatory text before the BEGIN and after the END lines for PKIX certificates, more than any other type. If emitted, such text SHOULD be related to the certificate, such as providing a textual representation of key data elements in the certificate.

Although textual encodings of PKIX structures can occur anywhere, many tools are known to offer an option to output this encoding when serializing PKIX structures. To promote interoperability and to separate DER encodings from textual encodings, the extension .crt SHOULD be used for the textual encoding of a certificate. Implementations should be aware that in spite of this recommendation, many tools still default to encode certificates in this textual encoding with the extension .cer. This section does not disturb the official application/pkix-cert registration in any way (which states that "each '.cer' file contains exactly one certificate, encoded in DER format"), but merely articulates a widespread, de facto alternative.

Certificate Revocation Lists (CRLs) are encoded using the X509 CRL label. The encoded data MUST be a BER (DER strongly preferred; see ) encoded ASN.1 CertificateList structure as described in Section 5 of .

-----BEGIN X509 CRL----- MIIB9DCCAV8CAQEwCwYJKoZIhvcNAQEFMIIBCDEXMBUGA1UEChMOVmVyaVNpZ24s IEluYy4xHzAdBgNVBAsTFlZlcmlTaWduIFRydXN0IE5ldHdvcmsxRjBEBgNVBAsT PXd3dy52ZXJpc2lnbi5jb20vcmVwb3NpdG9yeS9SUEEgSW5jb3JwLiBieSBSZWYu LExJQUIuTFREKGMpOTgxHjAcBgNVBAsTFVBlcnNvbmEgTm90IFZhbGlkYXRlZDEm MCQGA1UECxMdRGlnaXRhbCBJRCBDbGFzcyAxIC0gTmV0c2NhcGUxGDAWBgNVBAMU D1NpbW9uIEpvc2Vmc3NvbjEiMCAGCSqGSIb3DQEJARYTc2ltb25Aam9zZWZzc29u Lm9yZxcNMDYxMjI3MDgwMjM0WhcNMDcwMjA3MDgwMjM1WjAjMCECEC4QNwPfRoWd elUNpllhhTgXDTA2MTIyNzA4MDIzNFowCwYJKoZIhvcNAQEFA4GBAD0zX+J2hkcc Nbrq1Dn5IKL8nXLgPGcHv1I/le1MNo9t1ohGQxB5HnFUkRPAY82fR6Epor4aHgVy b+5y+neKN9Kn2mPF4iiun+a4o26CjJ0pArojCL1p8T0yyi9Xxvyc/ezaZ98HiIyP c3DGMNR+oUmSjKZ0jIhAYmeLxaPHfQwR -----END X509 CRL----- Historically, the label CRL has rarely been used. Today, it is not common and many popular tools do not understand the label. Therefore, this document standardizes X509 CRL in order to promote interoperability and backwards-compatibility. Generators conforming to this document MUST generate X509 CRL labels and MUST NOT generate CRL labels. Parsers SHOULD NOT treat CRL as equivalent to X509 CRL.

PKCS #10 Certification Requests are encoded using the CERTIFICATE REQUEST label. The encoded data MUST be a BER (DER strongly preferred; see ) encoded ASN.1 CertificationRequest structure as described in .

The label NEW CERTIFICATE REQUEST is also in wide use. Generators conforming to this document MUST generate CERTIFICATE REQUEST labels. Parsers MAY treat NEW CERTIFICATE REQUEST as equivalent to CERTIFICATE REQUEST.

PKCS #7 Cryptographic Message Syntax structures are encoded using the PKCS7 label. The encoded data MUST be a BER-encoded ASN.1 ContentInfo structure as described in .

The label CERTIFICATE CHAIN has been in use to denote a degenerate PKCS #7 structure that contains only a list of certificates (see Section 9 of ). Several modern tools do not support this label. Generators MUST NOT generate the CERTIFICATE CHAIN label. Parsers SHOULD NOT treat CERTIFICATE CHAIN as equivalent to PKCS7. PKCS #7 is an old specification that has long been superseded by CMS. Implementations SHOULD NOT generate PKCS #7 when CMS is an alternative.

Cryptographic Message Syntax structures are encoded using the CMS label. The encoded data MUST be a BER-encoded ASN.1 ContentInfo structure as described in .

CMS is the IETF successor to PKCS #7. Section 1.1.1 of describes the changes since PKCS #7 v1.5. Implementations SHOULD generate CMS when it is an alternative, promoting interoperability and forwards-compatibility.

Unencrypted PKCS #8 Private Key Information Syntax structures (PrivateKeyInfo), renamed to Asymmetric Key Packages (OneAsymmetricKey), are encoded using the PRIVATE KEY label. The encoded data MUST be a BER (DER preferred; see ) encoded ASN.1 PrivateKeyInfo structure as described in PKCS #8, or a OneAsymmetricKey structure as described in . The two are semantically identical and can be distinguished by version number.

Encrypted PKCS #8 Private Key Information Syntax structures (EncryptedPrivateKeyInfo), called the same in , are encoded using the ENCRYPTED PRIVATE KEY label. The encoded data MUST be a BER (DER preferred; see ) encoded ASN.1 EncryptedPrivateKeyInfo structure as described in PKCS #8 and .

Attribute certificates are encoded using the ATTRIBUTE CERTIFICATE label. The encoded data MUST be a BER (DER strongly preferred; see ) encoded ASN.1 AttributeCertificate structure as described in .

Public keys are encoded using the PUBLIC KEY label. The encoded data MUST be a BER (DER preferred; see ) encoded ASN.1 SubjectPublicKeyInfo structure as described in Section 4.1.2.7 of .

Data in this format often originates from untrusted sources, thus parsers must be prepared to handle unexpected data without causing security vulnerabilities. Implementers building implementations that rely on canonical representation or the ability to fingerprint a particular data object need to understand that this document does not define canonical encodings. The first ambiguity is introduced by permitting the text-encoded representation instead of the binary BER or DER encodings, but further ambiguities arise when multiple labels are treated as similar. Variations of whitespace and non-base64 alphabetic characters can create further ambiguities. Data encoding ambiguities also create opportunities for side channels. If canonical encodings are desired, the encoded structure must be decoded and processed into a canonical form (namely, DER encoding).

&rfc2119; &rfc2315; &rfc2986; &rfc4648; &rfc5280; &RFC5234; &RFC5652; &rfc5755; &RFC5958; Information Technology - ASN.1 encoding rules: Specification of Basic Encoding Rules (BER), Canonical Encoding Rules (CER) and Distinguished Encoding Rules (DER) International Telecommunications Union &RFC0934; &rfc1421; &RFC2585; &rfc4716; &rfc4880; &RFC5208; &RFC7292; Extensions and Revisions to PKCS #7 (Version 1.6 Bulletin) X.509 Style Guide

This appendix contains examples for the non-recommended label variants described earlier in this document. As discussed earlier, supporting these is not required and is sometimes discouraged. Still, they can be useful for interoperability testing and for easy reference.

-----BEGIN X.509 CERTIFICATE----- MIIBHDCBxaADAgECAgIcxzAJBgcqhkjOPQQBMBAxDjAMBgNVBAMUBVBLSVghMB4X DTE0MDkxNDA2MTU1MFoXDTI0MDkxNDA2MTU1MFowEDEOMAwGA1UEAxQFUEtJWCEw WTATBgcqhkjOPQIBBggqhkjOPQMBBwNCAATwoQSr863QrR0PoRIYQ96H7WykDePH Wa0eVAE24bth43wCNc+U5aZ761dhGhSSJkVWRgVH5+prLIr+nzfIq+X4oxAwDjAM BgNVHRMBAf8EAjAAMAkGByqGSM49BAEDRwAwRAIfMdKS5F63lMnWVhi7uaKJzKCs NnY/OKgBex6MIEAv2AIhAI2GdvfL+mGvhyPZE+JxRxWChmggb5/9eHdUcmW/jkOH -----END X.509 CERTIFICATE-----

-----BEGIN NEW CERTIFICATE REQUEST----- MIIBWDCCAQcCAQAwTjELMAkGA1UEBhMCU0UxJzAlBgNVBAoTHlNpbW9uIEpvc2Vm c3NvbiBEYXRha29uc3VsdCBBQjEWMBQGA1UEAxMNam9zZWZzc29uLm9yZzBOMBAG ByqGSM49AgEGBSuBBAAhAzoABLLPSkuXY0l66MbxVJ3Mot5FCFuqQfn6dTs+9/CM EOlSwVej77tj56kj9R/j9Q+LfysX8FO9I5p3oGIwYAYJKoZIhvcNAQkOMVMwUTAY BgNVHREEETAPgg1qb3NlZnNzb24ub3JnMAwGA1UdEwEB/wQCMAAwDwYDVR0PAQH/ BAUDAwegADAWBgNVHSUBAf8EDDAKBggrBgEFBQcDATAKBggqhkjOPQQDAgM/ADA8 AhxBvfhxPFfbBbsE1NoFmCUczOFApEuQVUw3ZP69AhwWXk3dgSUsKnuwL5g/ftAY dEQc8B8jAcnuOrfU -----END NEW CERTIFICATE REQUEST-----

This appendix is informative. Consult the respective standards for the normative rules. DER is a restricted profile of BER ; thus, all DER encodings of data values are BER encodings, but just one of the BER encodings is the DER encoding for a data value. Canonical encoding matters when performing cryptographic operations; additionally, canonical encoding has certain efficiency advantages for parsers. There are three principal reasons to encode with DER: A digital signature is (supposed to be) computed over the DER encoding of the semantic content, so providing anything other than the DER encoding is senseless. (In practice, an implementer might choose to have an implementation parse and digest the data as is, but this practice amounts to guesswork.) In practice, cryptographic hashes are computed over the DER encoding for identification. In practice, the content is small. DER always encodes data values in definite-length form (where the length is stated at the beginning of the encoding); thus, a parser can anticipate memory or resource usage up front. matches the structures in this document with the particular reasons for DER encoding:

Peter Gutmann suggested to document labels for Attribute Certificates and PKCS #7 messages, and to add examples for the non-standard variants. Dr. Stephen Henson suggested distinguishing when BER versus DER is appropriate or necessary.