[Docs] [txt|pdf] [Tracker] [WG] [Email] [Nits]
Versions: 00 01 02 03 04 05 06 07 08 09 10 11
12 13 14 15 16 17 18 19 20 RFC 6712
Internet Draft
PKIX Working Group Amit Kapoor (Certicom)
Expires in 6 months Ronald Tschalär (Certicom)
June 22 2000
Transport Protocols for CMP
<draft-ietf-pkix-cmp-transport-protocols-00.txt>
Status of this Memo
This document is an Internet-Draft and is in full conformance with
all provisions of Section 10 of RFC2026.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF), its areas, and its working groups. Note that
other groups may also distribute working documents as
Internet-Drafts.
Internet-Drafts are draft documents valid for a maximum of six
months and may be updated, replaced, or obsoleted by other documents
at any time. It is inappropriate to use Internet- Drafts as
reference material or to cite them other than as "work in progress."
The list of current Internet-Drafts can be accessed at
http://www.ietf.org/ietf/1id-abstracts.txt
The list of Internet-Draft Shadow Directories can be accessed at
http://www.ietf.org/shadow.html.
This Internet-Draft will expire on December 22, 2000
Copyright Notice
Copyright (C) The Internet Society (2000). All Rights Reserved.
Abstract
This document describes how to layer Certificate Management
Protocols [CMP] over various transport protocols.
The key words "MUST", "MUST NOT", "REQUIRED", "SHOULD", "SHOULD
NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document (in
uppercase, as shown) are to be interpreted as described in
[RFC2119].
This document version corresponds to the draft-ietf-pkix-cmp-transports-03.txt
published and discussed in the PKI Forum interoperability mailing list.
1. Motivation
Section 5 of the [RFC2510] spec specifies sending the DER-encoded
CMP message directly over various protocols. However, implementors,
during various interoperability workshops, found the protocol
lacking in the following respects:
1. No clear definition on when the connection is to be closed
and by whom.
2. No version number specified to allow for extensions.
3. Error messages cannot be processed by applications.
Realizing that this could not be achieved in a backward compatible
way, and acknowledging the changes being made to [RFC2510], the
decision was made to enhance the protocol now to avoid
interoperability conflicts later and to pull the transport section
out in a separate draft. This enhancement tries to keep as much of
the older protocol as possible, while ensuring that implementations
using the old protocol will not mistake a new message for a valid
message in the [RFC2510] format.
2. TCP-Based Management Protocol
While this section is called TCP-Based and the messages are called
TCP-message's, the same protocol can be used over any reliable,
connection oriented transport protocol (e.g. SNA, DECnet, etc.).
This protocol is suitable for cases where an end entity (or an RA)
initiates a transaction and can poll to pick up the results.
The client sends a TCP-message to the server, and the server
responds with another TCP-message.
The protocol basically assumes a listener process on an RA or CA
which can accept TCP-messages on a well-defined port (default port
number is 829). Typically a client initiates connection to the
server and submits a PKI message. The server replies with a PKI
message or with a reference number to be used later when polling
for the actual PKI message response.
If a polling-reference was supplied then the client will send a
polling request using this polling-reference after waiting for at
least the specified time. The server may again reply with a
polling-reference or with the actual PKI message response.
When the final PKI response message has been picked up by the
client then no new polling reference is supplied.
If a transaction is initiated by a PKI entity (RA or CA) then an
end entity must either supply a listener process or be supplied
with a polling reference (see below) in order to allow it to pick
up the PKI message from the PKI management component.
2.1 General Form
A TCP-message consists of:
length (32-bits)
version (8-bits)
flags (variable length)
message-type (8-bits),
value (defined below)
The length field contains the number of octets of the remainder of
the TCP-message (i.e., number of octets of <value> plus <flags-length>
plus 2). All bit values in this protocol are specified to be in
network byte order.
The version field indicates the version of the TCP-message. It MUST
be incremented for each specification which changes the flags field
in a way that is not fully backwards compatible with the previous
version (e.g. when the length of the flags field is changed).
The flags field is for transporting TCP-message specific data. The
length of this field is version dependent and is fixed for a given
version.
The message-type field is used to indicate the type of TCP-message.
The value field contains message-type dependent data.
2.2 Version Negotiation
If a client knows the protocol version(s) supported by the
server (e.g. from a previous TCP-message exchange or via some
out-of-band means) then it SHOULD send a TCP-message with the highest
version supported both by it and the server. If a client does
not know what version(s) the server supports then it SHOULD send
a TCP-message using the highest version it supports.
If a server receives a TCP-message version that it supports, then it
MUST reply with a TCP-message of the same version. If the version
received is higher than what the server supports, it MUST send
back a VersionNotSupported errorMsgRep (defined below) containing
the highest version it supports.
2.3 TCP-message Version 10
The TCP-message version will be 10 for this document. The number
has deliberately been chosen to prevent [RFC2510] compliant
applications from treating it as a valid message type. Applications
receiving a version less than 10 SHOULD interpret the message as
being an [RFC2510] style message.
The length of the flags field for this version is 1 octet. The LSB
is used to indicate a connection close; all other bits in the flags
octet MUST be ignored by receivers, and MUST be set to zero by
senders.
By default connections are kept open after the receipt of a
response. Either party (client or server) MAY set the connection
close bit at any time. If the connection close bit is set on a
request, then the server MUST set the bit in the response and
close the connection after sending the response. If the bit is set
on a response from the server, the client MUST NOT send any further
requests on that connection. Applications MAY decide to close an
idle connection (one on which no response is outstanding) after
some time-out. Because of the problem where a client sends a
request and the server closes the connection while the request is
still in flight, clients SHOULD automatically retry a request for
which no part of the response could be read due to a connection
close or reset.
If the connection is kept open, it MUST only be used for subsequent
request/response transactions started by the client - the server
MUST NOT use it to send requests to the client. Different
transactions may be freely interwoven on the same connection. E.g.
a CR/CP need not immediately be followed by the Confirm, but may be
followed by any other request from a different transaction.
2.4 Detecting and Interoperating with RFC-2510 Conformant Implementations
Servers wishing to interoperate with clients conforming to
[RFC2510] can do so by treating any received message with a version
less than 10 as an [RFC2510] message and responding in that format.
Servers not wishing to support [RFC2510] messages MUST respond with
a [RFC2510] errorMsgRep.
Clients wishing to interoperate with [RFC2510] compliant servers
SHOULD treat a response with a version less than 10 as an [RFC2510]
style message. If this message is an errorMsgRep (message-type 06)
then the client MAY automatically retry the request using the
[RFC2510] format; if the message is not an errorMsgRep or the
implementation does not wish to support [RFC2510] then it MUST
abort the corresponding CMP transaction.
2.5 Message Types
message-types 0-127 are reserved and will be issued under IANA
auspices. message-types 128-255 are reserved for application use.
The message-type's currently defined are:
Message name Message-type
pkiReq '00'H
pollRep '01'H
pollReq '02'H
finRep '03'H
pkiRep '05'H
errorMsgRep '06'H
If server receives an unknown message-type then it MUST reply with
an InvalidMessageType errorMsgRep. If a client receives an unknown
message-type then it MUST abort the CMP transaction.
The different TCP-messages are discussed in the following sections:
2.5.1 pkiReq
The pkiReq is to be used to carry a PKIMessage from the client to the
server. The <value> portion of this TCP-message will contain:
DER-encoded PKIMessage.
The type of PKIMessages that can be carried by this TCP-message are:
CRL Announcement
Certificate Confirmation
Poll Request
Subscription Request
CA Key Update Announcement
Certificate Announcement
Certification Request
Cross-Certification Request
Error Message
General Message
Initialization Request
Key Recovery Request
Key Update Request
Nested Message
PKCS-10 Request
POP Response
Revocation Request
2.5.2 pkiRep
This TCP-message is to be used to send back the response to the request.
The <value> portion of the pkiRep will contain:
DER encoded PKI message
The type of PKIMessages that can be carried by this TCP-message are:
Confirmation
Poll Response
Subscription Response
Certification Response
Error Message
General Response
Initialization Response
Key Recovery Response
Key Update Response
POP Challenge
Revocation Response
2.5.3 pollReq
The pollReq will be the used by the client to check the status of a
pending TCP-message. The <value> portion of the pollReq will
contain:
polling-reference (32 bits)
The <polling-reference> MUST be the one returned via the pollRep
TCP-message.
2.5.4 pollRep
The pollRep will be the response sent by the server to the client
when there are no TCP-message response ready. The <value> portion of
the pollRep will contain:
polling-reference (32 bits)
time-to-check-back (32 bits)
The <polling-reference> is a unique 32-bit number sent by the server.
The <time-to-check-back> is the time in seconds indicating the minimum
interval after which the client SHOULD check the status again.
The duration for which the server keeps the <polling-reference>
unique is left to the implementation.
2.5.5 finRep
finRep will be the response from the server indicating end of
transaction, i.e., there are no further messages to be delivered
from the server. The <value> portion of the finRep will
contain:
'00'H (8 bits)
2.5.6 errorMsgRep
This TCP-message is sent when a TCP-message level protocol error is
detected. Please note that PKIError messages MUST NOT be sent
using this. Examples of TCP-message level errors are:
1. Invalid protocol version
2. Invalid TCP message-type
3. Invalid polling reference number
The <value> field of the TCP-message SHALL contain:
error-type (16-bits)
data-length (16-bits)
data (<data-length> octets)
UTF8 String (SHOULD include a RFC 1766 language tag)
The <error-type> is of the form MMNN where M and N are hex digits
(0-F) and MM represents the major category and NN the minor. The
major categories defined by this specification are:
'01'H TCP-message version negotiation
'02'H client errors
'03'H server errors
The major categories '80'H-'FF'H are reserved for application use.
The <data-length> and <data> are additional information about the
error to be used by programs for further processing and recovery.
<data-length> contains the length of the <data> field in number of
octets. Error messages not needing additional information to be
conveyed MUST set the <data-length> to 0.
The UTF8 text string is for user readable error messages.
2.5.6.1 VersionNotSupported errorMsgRep
The VersionNotSupported errorMsgRep is defined as follows:
error-type: '0101'H
data-length: 1
data: <version>
UTF8-text String: implementation defined
where <version> is the highest version the server supports.
2.5.6.2 GeneralClientError errorMsgRep
The GeneralClientError errorMsgRep is defined as follows:
error-type: '0200'H
data-length: 0
data: <empty>
UTF8-text String: implementation defined
2.5.6.3 InvalidMessageType errorMsgRep
The InvalidMessageType errorMsgRep is defined as follows:
error-type: '0201'H
data-length: 1
data: <message-type>
UTF8-text String: implementation defined
where <message-type> is the message-type received by the
server.
2.5.6.4 InvalidPollID errorMsgRep
The InvalidPollID errorMsgRep is defined as follows:
error-type: '0202'H
data-length: 4
data: <polling-reference>
UTF8-text String: implementation defined
where <polling-reference> is the polling-reference received by
the server.
2.5.6.5 GeneralServerError errorMsgRep
The GeneralServerError errorMsgRep is defined as follows:
error-type: '0300'H
data-length: 0
data: <empty>
UTF8-text String: implementation defined
3. HTTP-Based Management Protocol
A client creates a TCP-message, as specified in section 2.0. The
message is then sent as the entity-body of an HTTP POST request. If
the HTTP request is successful then the server returns a similar
message in the body of the response. The response status code in
this case MUST be 200; other 2xx codes MUST NOT be used. The content
type of the request and response MUST be "application/pkixcmp".
Applications MAY wish to also recognized and use the
"application/x-pkixcmp" MIME type (specified in earlier versions of
this document) in order to support backward compatibility wherever
applicable. Content codings may be applied.
Note that a server may return any 1xx, 3xx, 4xx, or 5xx code if the
HTTP request needs further handling or is otherwise not acceptable.
Because in general CMP messages are not cacheable, requests and
responses should include a "Cache-Control: no-cache" (and, if either
side uses HTTP/1.0, a "Pragma: no-cache") to prevent the client from
getting cached responses. This is especially important for polling
requests and responses.
Connection management SHOULD be based on the HTTP provided mechanisms
(Connection and Proxy-Connection headers) and not on the connection
flag carried in the TCP-message.
4. File based protocol
A file containing a PKI message MUST contain only the DER encoding of
one PKI message, i.e., there MUST be no extraneous header or trailer
information in the file.
Such files can be used to transport PKI messages using, e.g., FTP.
5. Mail based protocol
This subsection specifies a means for conveying ASN.1-encoded
messages for the protocol exchanges via Internet mail.
A simple MIME object is specified as follows.
Content-Type: application/pkixcmp
Content-Transfer-Encoding: base64
<<the ASN.1 DER-encoded PKIX-CMP message, base64-encoded>>
This MIME object can be sent and received using common MIME
processing engines and provides a simple Internet mail transport for
PKIX-CMP messages. Implementations MAY wish to also recognize and
use the "application/x-pkixcmp" MIME type (specified in earlier
versions of this document) in order to support backward compatibility
wherever applicable.
6. Security Considerations
Three aspects need to be considered by server side implementors:
1. There is no security at the TCP and HTTP protocol level (unless
tunneled via SSL/TLS) and thus TCP-message should not be used
to change state of the transaction. Change of state should be
done on the signed PKIMessage being carried within the
TCP-message.
2. If the server is going to be sending messages with sensitive
information (not meant for public consumption) in the clear, it
is RECOMMENDED that the server send back the message directly
and not use the pollRep.
3. The polling request/response mechanism can be used for all kinds
of denial of service attacks. It is RECOMMENDED that the server
not change the polling-reference between polling requests.
7. Acknowledgments
The authors gratefully acknowledge the contributions of various
members of the IETF PKIX Working Group and the ICSA CA-talk mailing
list (a list solely devoted to discussing CMP interoperability
efforts).
8. References
[RFC2510] Adams, C., Farrell, S., "Internet X.509 Public Key
Infrastructure, Certificate Management Protocols", RFC 2510,
March 1999.
[HTTP] Fielding, R.T., et. al, "Hypertext Transfer Protocol --
HTTP/1.1", RFC 2616, June 1999.
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", RFC 2119, March 1997.
[RFC821] Postel, J., "Simple Mail Transfer Protocol", RFC 821,
August 1982.
Authors' Addresses
Amit Kapoor
Certicom
25801 Industrial Blvd
Hayward, CA 94545
US
E-Mail: amit@trustpoint.com
Ronald Tschalär
Certicom
25801 Industrial Blvd
Hayward, CA 94545
US
E-Mail: ronald@trustpoint.com
Full Copyright Statement
Copyright (C) The Internet Society (2000). All Rights Reserved.
This document and translations of it may be copied and furnished to
others, and derivative works that comment on or otherwise explain
it or assist in its implementation may be prepared, copied,
published and distributed, in whole or in part, without restriction
of any kind, provided that the above copyright notice and this
paragraph are included on all such copies and derivative works.
However, this document itself may not be modified in any way, such
as by removing the copyright notice or references to the Internet
Society or other Internet organizations, except as needed for the
purpose of developing Internet standards in which case the
procedures for copyrights defined in the Internet Standards process
must be followed, or as required to translate it into languages
other than English.
The limited permissions granted above are perpetual and will not be
revoked by the Internet Society or its successors or assigns.
This document and the information contained herein is provided on
an "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET
ENGINEERING TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF
THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED
WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
Html markup produced by rfcmarkup 1.129d, available from
https://tools.ietf.org/tools/rfcmarkup/