Copyright ©2001 W3C® (MIT, INRIA, Keio), All Rights Reserved. W3C liability, trademark, document use and software licensing rules apply.
This document specifies a process for encrypting data and representing the result in XML. The data may be arbitrary data (including an XML document), an XML element, or XML element content. The result of encrypting data is an XML Encryption element which contains or references the cipher data.
This is an Editors' draft with no official standing.
This is Last Call for the "XML Encryption Syntax and Processing" specification from the XML Encryption Working Group (Activity). This version specifies the Working Group's approach to satisfying the requirements stemming from the March 2001 face-to-face meeting and subsequent discussion on the xml-encryption@w3.org mailing list. The Last Call period is 3 weeks, ending on 9 November 2001.
The Working Group will try to use a new namespace when changes in its syntax or processing are substantive. However, this namespace might be reused (prior to reaching Candidate Recommendation) by subsequent drafts in such a way as to cause instances using the namespace to become invalid or to change in meaning or affect the operation of existing software. Requests for a more stringent level of namespace stability should be made to the Working Group.
Publication of this document does not imply endorsement by the W3C membership. This is a draft document and may be updated, replaced or obsoleted by other documents at any time. It is inappropriate to cite a W3C Working Draft as anything other than a "work in progress." A list of current W3C working drafts can be found at http://www.w3.org/TR/.
Please send comments to the editors (<reagle@w3.org>, <dee3@torque.pothole.com>) and cc: the list xml-encryption@w3.org (archives)
Patent disclosures relevant to this specification may be found on the Working Group's patent disclosure page in conformance with W3C policy.
This section discusses algorithms used with the XML Encryption
specification. Entries contain the identifier to be used as the
value of the Algorithm
attribute of the
EncryptionMethod
element or other element
representing the role of the algorithm, a reference to the formal
specification, definitions for the representation of keys and the
results of cryptographic operations where applicable, and general
applicability comments.
All algorithms listed below have implicit parameters depending on their role. For example, the data to be encrypted or decrypted, keying material, and direction of operation (encrypting or decrypting) for encryption algorithms. Any explicit additional parameters to an algorithm appear as content elements within the element. Such parameter child elements have descriptive element names, which are frequently algorithm specific, and SHOULD be in the same namespace as this XML Encryption specification, the XML Signature specification, or in an algorithm specific namespace. An example of such an explicit parameter could be a nonce (unique quantity) provided to a key agreement algorithm.
This specification defines a set of algorithms, their URIs, and requirements for implementation. Requirements are specified over implementation, not over requirements for encryption use. Furthermore, the mechanism is extensible, and alternative algorithms may be used.
The table below lists the categories of algorithms. Within each category, a brief name, the level of implementation requirement, and an identifying URI are given for each algorithm.
The schema for EncryptionMethod
is as follows:
Schema Definition: <complexType name='EncryptionMethodType' mixed='true'> <sequence> <element name='KeySize' minOccurs='0' type='KeySizeType'/> <element ref'ds:DigestMethod' minOccurs='0'/> <element name='OAEPparams' minOccurs='0' type='base64Binary'/> <any namespace='##other' minOccurs='0' maxOccurs='unbounded'/> <!-- (0,unbounded) elements from (1,1) external namespace --> </sequence> <attribute name='Algorithm' type='anyURI' use='required'/> </complexType>
NOTE: Which child elements to the EncryptionMethod
algorithm role are allowed or required depends on the specific
value of the Algorithm
attribute URI; however, the
KeySize
child element is always permitted. (Schema
does not provide a facility for expressing conditionality of child
element occurrence based on attribute value.) The presence of any
child element under EncryptionMethod
which is not
permitted by the algorithm or the presence of a
KeySize
child inconsistent with the algorithm MUST be
treated as an error. (All algorithm URIs specified in this document
imply a key size but this is not true in general. Most popular
stream cipher algorithms take variable size keys.)
Block encryption algorithms are designed for encrypting and
decrypting data in multiple octet chunks. Their identifiers appear
as the value of the Algorithm
attributes of
EncryptionMethod
elements that are children of
EncryptedData
.
Block encryption algorithms take, as implicit arguments, the data to be encrypted or decrypted, the keying material, and their direction of operation. For all of these algorithms specified below, an initialization vector (IV) is required that is encoded with the cipher text. For user specified block encryption algorithms, the IV, if any, could be specified as being with the cipher data, as an algorithm content element, or elsewhere.
The IV is encoded with and before the cipher text for the algorithms below for ease of availability to the decryption code and to emphasize its association with the cipher text. Good cryptographic practice requires that a different IV be used for every encryption.
Since the data being encrypted is an arbitrary number of octets, it may not be a multiple of the block size. This is solved by padding the plain text up to the block size before encryption and unpadding after decrytion. (This us done after prepending the nonce for encryption.) The padding algorithm is to calculate the smallest non-zero number of octets, say N, that must be suffixed to the plain text to bring it up to a multiple of the block size. We will assume the block size is B octets os N is in the range of 1 to B. Pad by suffixing the plain text with N-1 arbitrary pad bytes and a final byte whose value is N. On decryption, just take the last byte and, after sanity checking it, strip that many bytes from the end of the decrypted cipher text.
For example, assume an 8 byte block size and plain text of 0x616263. The padded plain text would then be 0x616263????????05 where the "??" bytes can be any value. Similarly, plain text of 0x2122232425262728 would be padded to 0x2122232425262728??????????????08.
ANSI X9.52 [TRIPLEDES] specifies three sequential FIPS 46-3 [DES] operations. The XML Encryption TRIPLEDES consists of a DES encrypt, a DES decrypt, and a DES encrypt used in the Cipher Block Chaining (CBC) mode with 192 bits of key and a 64 bit Initialization Vector (IV). Of the key bits, the first 64 are used in the first DES operation, the second 64 bits in the middle DES operation, and the third 64 bits in the last DES operation.
Note: Each of these 64 bits of key contain 56 effective bits and 8 parity bits. Thus there are only 168 operational bits out of the 192 being transported for a TRIPLEDES key. (Depending on the criterion used for analysis, the effective strength of the key may thought to be 112 bits (due to meet in the middle attacks) or even less.)
The resulting cipher text is prefixed by the IV. If included in XML output, it is then base64 encoded. An example TRIPLEDES EncryptionMethod is as follows:
<EncryptionMethod
Algorithm="http://www.w3.org/2001/04/xmlenc#tripledes-cbc"/>
[AES] is used in the Cipher Block Chaining (CBC) mode with a 128 bit initialization vector (IV). The resulting cipher text is prefixed by the IV. If included in XML output, it is then base64 encoded. An example AES EncryptionMethod is as follows:
<EncryptionMethod
Algorithm="http://www.w3.org/2001/04/xmlenc#aes128-cbc"/>
Simple stream encryption algorithms generate, based on the key,
a stream of bytes which are XORed with the plain text data bytes to
produce the cipher text on encryption and with the cipher text
bytes to produce plain text on decryption. They are normally used
for the encryption of data and are specified by the value of the
Algorithm
attribute of the
EncryptionMethod
child of an
EncryptedData
element.
NOTE: It is critical that each simple stream encryption key (or key and initialization vector (IV) if an IV is also used) be used once only. If the same key (or key and IV) is ever used on two messages then, by XORing the two cipher texts, you can obtain the XOR of the two plain texts. This is usually very compromising.
No specific steam encryption algorithms are specified herein but this section is included to provide general guidelines.
Stream algorithms typically use the optional
KeySize
explicit parameter. In cases where the key
size is not apparent from the algorithm URI or key source, as in
the use of key agreement methods, this parameter sets the key size.
If the size of the key to be used is apparent and disagrees with
the KeySize
parameter, an error MUST be returned.
Implementation of any stream algorithms is optional. The schema for
the KeySize parameter is as follows:
Schema Definition: <simpleType name='KeySizeType'> <restriction base="integer"/> </simpleType>
Key Transport algorithms are public key encryption algorithms
especially specified for encrypting and decrypting keys. Their
identifiers appear as Algorithm
attributes to
EncryptionMethod
elements that are children of
EncryptedKey
. EncryptedKey
is in turn the
child of a KeyInfo
element. The type of key being
transported, that is to say the algorithm in which it is planned
to use the transported key, is given by the Algorithm
attribute of the
EncryptionMethod
child of the EncryptedData
or
EncryptedKey
parent of this KeyInfo
element.
(Key Transport algorithms may optionally be used to encrypt data
in which case they appear directly as the Algorithm
attriubte of an EncryptionMethod
child of an
EncryptedData
element.
Because they use public key algorithms directly, Key Transport
algorithms are not efficient for the transport of any amounts of
data significantly larger than symmetric keys.)
The Key Transport algorithms given below are those used in conjunction with the Cryptographic Message Syntax (CMS) of S/MIME [CMS-Algorithms, CMS-AES].
The RSAES-PKCS1-v1_5 algorithm, specified in RFC 2437 [PKCS1], takes no explicit parameters. An
example of an RSA Version 1.5 EncryptionMethod
element
is:
<EncryptionMethod
Algorithm="http://www.w3.org/2001/04/xmlenc#rsa-1_5"/>
The CipherValue
for such an encrypted key is the
base64 [MIME] encoding of the octet string
computed as per RFC 2437 [PKCS1, section
7.2.1: Encryption operation]. As specified in the EME-PKCS1-v1_5
function RFC 2437 [PKCS1, section
9.1.2.1], the value input to the key transport function is as
follows:
CRYPT ( PAD ( KEY ))
where the padding is of the following special form:
02 | PS* | 00 | key
where "|" is concatenation, "02" and "00" are fixed octets of the corresponding hexadecimal value, PS is a string of strong pseudo-random octets [RANDOM] at least eight octets long, containing no zero octets, and long enough that the value of the quantity being CRYPTed is one octet shorter than the RSA modulus, and "key" is the key being transported. The key is 192 bits for TRIPLEDES and 128, 192, or 256 bits for AES. Support of this key transport algorithm for transporting 192 bit keys is mandatory to implement. Support of this algorithm for transporting other keys is optional. RAS-OAEP is recommended for the transport of AES keys.
The resulting base64 [MIME] string is
the value of the child text node of the CipherData
element, e.g.
<CipherData> IWijxQjUrcXBYoCei4QxjWo9Kg8D3p9tlWoT4 t0/gyTE96639In0FZFY2/rvP+/bMJ01EArmKZsR5VW3rwoPxw= </CipherData>
THE RSAES-OAEP-ENCRYPT, as specified in RFC 2437 [PKCS1], algorithm takes as explicit
parameters a message digest function and an optional octet string
OAEPparams
. The message digest function is indicated
by the Algorithm
attribute of a child
DigestMethod
element and the octet string is the
base64 decoding of the content of an optional
OAEPparams
child element. An example of an RSA-OAEP
element is:
<EncryptionMethod Algorithm="http://www.w3.org/2001/04/xmlenc#rsa-oaep-mgf1p"> <ds:DigestMethod Algorithm="http://www.w3.org/2000/09/xmldsig#sha1"/> <OAEPparams> 9lWu3Q== </OAEPparams> <EncryptionMethod>
The CipherData
for an RSA-OAEP encrypted key is the
base64 [MIME] encoding of the octet string
computed as per RFC 2437 [PKCS1, section
7.1.1: Encryption operation]. As described in the EME-OAEP-ENCODE
function RFC 2437 [PKCS1, section
9.1.1.1], the value input to the key transport function is
calculated use the message digest function and string specified in
the DigestMethod
and OAEPparams
elements
and using the mask generator function MGF1 specified in RFC 2437.
The desired output length for EME-OAEP-ENCODE is one byte shorter
than the RSA modulus.
The transported key size is 192 bits for TRIPLEDES and 128, 192, or 256 bits for AES. Implementations MUST implement RSA-OAEP for the transport of 128 and 256 bit keys. They MAY implement RSA-OAEP for the transport of other keys.
A Key Agreement algorithm provides for the agreement to a shared
secret quantity based on certain types of compatible public keys
from both the sender and the recipient. Information to determine
the key associated with the originator is indicated by an optional
OriginatorKeyInfo
parameter child of an
AgreementMethod
element while that associated with the
recipient is indicated by an optional
RecipientKeyInfo
. A shared key is derived from this
shared secret by a method determined by the Key Agreement
algorithm.
Note: XML Encryption does NOT provide an on-line key agreement negotiation
protocol. The key agreement structure defined here is only suitable
for off-line agreement. Agreement
based, for example, on trusted recipient key information
obtained previously from some public key infrastructure by the originator.
If a key has been agreed to via some previous on-line protocol, it would be
more natural to use a KeyName
or the like to refer to an
agreed name.
The AgreementMethod
element appears as the content
of a ds:KeyInfo
since, like other
ds:KeyInfo
children, it yields a key. This
ds:KeyInfo
is in turn a child of an
EncryptedData
or EncryptedKey
element.
The Algorithm
attribute and KeySize
child
of the EncryptionMethod
element under this
EncryptedData
or EncryptedKey
element are
implicit parameters to the key agreement computation. In cases
where this EncryptionMethod
algorithm URI is
insufficient to determine the key length, a KeySize
MUST have been included. In addition, the sender may place a
Nonce
element under AgreementMethod
to
assure that different keying material is generated even for
repeated agreements using the same sender and recipient public
keys. For example:
<EncryptedData> <EncryptionMethod Algorithm="Example:Block/Alg" <KeySize>80</KeySize> </EncryptionMethod> <ds:KeyInfo xmlns:ds="http://www.w3.org/2000/09/xmldsig#"> <AgreementMethod Algorithm="Example:Agreement/Algorithm"> <Nonce> Zm9v </Nonce> <DigestMethod Algorithm="http://www.w3.org/2001/04/xmlenc#sha256"> <OriginatorKeyInfo> <KeyValue>...originator...</KeyValue> </OriginatorKeyInfo> <RecipientKeyInfo> <KeyValue>...recipient...</KeyValue> </RecipientKeyInfo> </AgreementMethod> </ds:KeyInfo> <CipherData>...</CipherData> </EncryptedData>
If the agreed key is being used to wrap a key, rather than data
as above, then AgreementMethod
would appear inside a
ds:KeyInfo
inside an EncryptedKey
element.
The Schema for AgreementMethod
is as follows:
Schema Definition: <element name="AgreementMethod" type="xenc:AgreementMethodType"/> <complexType name="AgreementMethodType" mixed="true"> <sequence> <element name="Nonce" minOccurs="0" type="base64Binary"/> <element ref="ds:DigestMethod" minOccurs="0"/> <element name="OriginatorKeyInfo" minOccurs="0" type="ds:KeyInfoType"/> <element name="RecipientKeyInfo" minOccurs="0" type="ds:KeyInfoType"/> <any namespace="##other" minOccurs="0" maxOccurs="unbounded"/> <!-- (0,unbounded) elements from (1,1) external namespace --> </sequence> <attribute name="Algorithm" type="anyURI" use="required"/> </complexType>
Diffie-Hellman keys can appear directly within
KeyValue
elements or be obtained by
ds:RetrievalMethod
fetches as well as appearing in
certificates and the like. The above identifier can be used as the
value of the Type
attribute of Reference
or ds:RetrievalMethod
elements.
A DH public key consists of three quantities, a large prime p, a
"generator" g, and "public" such that public = g**x mod p. The
corresponding private key is x. Because a prime and generator can
be safely shared over many DH keys, they may be known from the
application environment and are optional. The schema for a
DHKeyValue
is as follows:
Schema:
<element name="DHKeyValue" type="xenc:DHKeyValueType"/>
<complexType name="DHKeyValueType">
<sequence>
<element name="Prime"
type="ds:CryptoBinary" minOccurs="0"/>
<element name="Generator"
type="ds:CryptoBinary" minOccurs="0"/>
<element name="Public"
type="ds:CryptoBinary"/>
</sequence>
</complexType>
The Diffie-Hellman (DH) key agreement protocol [ESDH] involves the derivation of shared secret information based on compatible DH keys from the sender and recipient. Two DH public keys are compatible if they have the same prime and generator. If, for the second one, Y = g**y mod p, then the two parties can calculate the shared secret ZZ = ( g**(x*y) mod p ) even though each knows only their own private key and the other party's public key. Leading zero bytes MUST be maintained in ZZ so it will be the same length, in bytes, as p. We require that p be at least 512 bits and g at least 160 bits. There are numerous other complex security considerations in the selection of g, p, and a random x as described in [ESDH].
Diffie-Hellman key agreement is optional to implement. An
example of a DH AgreementMethod
element is as
follows:
<xenc:AgreementMethod xenc:Algorithm="http://www.w3.org/2001/04/xmlenc#dh" xmlns="http://www.w3.org/2000/09/xmldsig#"> <xenc:Nonce>Zm9v</Nonce> <DigestMethod Algorithm="http://www.w3.org/2000/09/xmldsig#sha1"/> <xenc:OriginatorKeyInfo> <X509Data><X509Certificate> ... </X509Certificate></X509Data> </OriginatorKeyInfo> <xenc:RecipientKeyInfo><KeyValue> ... </KeyValue></RecipientKeyInfo> </xenc:AgreementMethod>
(Add example with real "bits"...)
Assume the Diffie-Hellman shared secret is the octet sequence ZZ. The shared keying material needed will then be calculated as follows:
Keying Material = KM(1) | KM(2) | ...
where "|" is byte stream concatenation and
KM(counter) = DigestAlg ( ZZ | counter | EncryptionAlg | Nonce | KeySize ).
DigestMethod
child of
AgreementMethod
.Algorithm
attribute of the
EncryptionMethod
child of the
EncryptedData
or EncryptedKey
grandparent
of AgreementMethod
.Nonce
child
of AgreementMethod
, if present. If the
Nonce
element is absent, it is null.For example, the initial (KM(1)) calculation for the example above where the binary 1 counter byte is represented as the two character UTF-8 sequence "01", would be calculated as follows:
SHA-256(Example:Block/AlgZZ01foo40 )
where ZZ is the binary shared secret octet sequence and "foo" is the base64 decoding of "Zm9v".
(Actual digest value and agreed key to be added...)
Each application of DigestAlg for successive values of Counter will produce some additional number of bytes of keying material. From the concatenated string of one or more KM's, enough leading bytes are taken to meet the need for an actual key and the remainder discarded. For example, if DigestAlg is SHA1 which produces 20 octets of hash, then for 128 bit AES the first 16 bytes from KM(1) would be taken and the remaining 4 bytes discarded. For 256 bit AES, all of KM(1) suffixed with the first 12 bytes of KM(2) would be taken and the remaining 8 bytes of KM(2) discarded.
Symmetric Key Wrap algorithms are shared secret key encryption
algorithms especially specified for encrypting and decrypting
symmetric keys. Their identifiers appear as Algorithm
attribute values to EncryptionMethod
elements that are
children of EncryptedKey
which is in turn a child of
KeyInfo
which is in turn a child of
EncryptedData
or another EncryptedKey
.
The type of the key being wrapped is indicated by the
Algorithm
attribute of EncryptionMethod
child of the parent of the KeyInfo
grandparent of the
EncryptionMethod
specifying the symmetric key wrap
algorithm.
Some key wrap algorithms make use of the key checksum defined in CMS [CMS-Wrap]. This is used to provide an integrity check value for the key being wrapped. The algorithm is
XML Encryption implementations MUST support TRIPLEDES wrapping of 168 bit keys and may optionally support TRIPLEDES wrapping of other keys.
An example of a TRIPLEDES Key Wrap EncryptionMethod
element is a as follows:
<EncryptionMethod
Algorithm="http://www.w3.org/2001/04/xmlenc#kw-tripledes"/>
The following algorithm wraps (encrypts) a key (the wrapped key, WK) under a TRIPLEDES key-encryption-key (KEK) as specified in [CMS-Algorithms]:
The following algorithm unwraps (decrypts) a key as specified in [CMS-Algorithms]:
Implementation of AES key wrap as specified by NIST/NSA/CMS will be mandatory for AES 128 and AES 256 and optional for AES 192 -- when it has been completely specified by them.
Message digest algorithms can be used in
AgreementMethod
as part of the
key derivation, within RSA-OAEP encryption as a hash function, and
in connection with the HMAC message authentication code method as
described in [XML-DSIG].)
The SHA-1 algorithm [SHA] takes no
explicit parameters. XML encryption implementations MUST implement
SHA-1. An example of an SHA-1 DigestMethod
element
is:
<DigestMethod
Algorithm="http://www.w3.org/2000/09/xmldsig#sha1"/>
A SHA-1 digest is a 160-bit string. The content of the
DigestValue
element shall be the base64 encoding of
this bit string viewed as a 20-octet octet stream. For example, the
DigestValue
element for the message digest:
A9993E36 4706816A BA3E2571 7850C26C 9CD0D89D
from Appendix A of the SHA-1 standard would be:
<DigestValue>qZk+NkcGgWq6PiVxeFDCbJzQ2J0=</DigestValue>
The SHA-256 algorithm [SHA] takes no
explicit parameters. It is RECOMMENDED that XML encryption
implementations implement SHA-256. An example of an SHA-256
DigestMethod
element is:
<DigestMethod
Algorithm="http://www.w3.org/2001/04/xmlenc#sha256"/>
A SHA-256 digest is a 256-bit string. The content of the
DigestValue
element shall be the base64 encoding of
this bit string viewed as a 32-octet octet stream.
The SHA-512 algorithm [SHA] takes no
explicit parameters. An example of an SHA-512
DigestMethod
element is:
<DigestMethod
Algorithm="http://www.w3.org/2001/04/xmlenc#sha512"/>
A SHA-512 digest is a 512-bit string. The content of the
DigestValue
element shall be the base64 encoding of
this bit string viewed as a 64-octet octet stream.
The RIPEMD-160 algorithm [RIPEMD-160] takes no explicit
parameters. An example of an RIPEMD-160 DigestMethod
element is:
<DigestMethod
Algorithm="http://www.w3.org/2001/04/xmlenc#ripemd160"/>
A RIPEMD-160 digest is a 160-bit string. The content of the
DigestValue
element shall be the base64 encoding of
this bit string viewed as a 20-octet octet stream.
XML Signature [XML-DSIG] is optional to implement for XML encryption applications. It is the recommended way to provide key based authentication.
A Canonicalization of XML is a method of consistently serializing XML into an octet stream as is necessary prior to encrypting XML.
Canonical XML [Canon] is a method of serializing XML which includes the in scope namespace and xml namespace attribute context from ancestors of the XML being serialized.
If XML is to be encrypted and then later decrypted into a different environment and it is desired to preserve namespace prefix bindings and the value of attributes in the "xml" namespace of its original environment, then the canonical XML with comments version of the XML should be the serialization that is encrypted
Exclusive XML Canonicalization [Exclusive] serializes XML in such a way as to include to the minimum extent practical the namespace prefix binding and xml namespace attribute context inherited from ancestor elements.
It is the recommended method where the outer context of a fragment which was signed and then encrypted may be changed. Otherwise the validation of the signature over the fragment may fail because the canonicalization by signature validation may include unnecessary namespaces into the fragment.