P6R's Secure Shell Public Key Subsystem
Organize keys and certificates in namespaces over SSH.
Joseph, M. and J. Susoy, "P6R's Secure Shell Public Key Subsystem", RFC 7076, DOI 10.17487/RFC7076, November 2013, <https://www.rfc-editor.org/info/rfc7076>.
@misc{rfc7076,
series = {Request for Comments},
number = 7076,
howpublished = {RFC 7076},
publisher = {RFC Editor},
doi = {10.17487/RFC7076},
url = {https://www.rfc-editor.org/info/rfc7076},
author = {M. Joseph and J. Susoy},
title = {{P6R's Secure Shell Public Key Subsystem}},
pagetotal = 11,
year = 2013,
month = nov,
} RFC text
Contents
Independent Submission M. Joseph
Request for Comments: 7076 J. Susoy
Category: Informational P6R, Inc
ISSN: 2070-1721 November 2013
P6R's Secure Shell Public Key Subsystem
Abstract
The Secure Shell (SSH) Public Key Subsystem protocol defines a key
distribution protocol that is limited to provisioning an SSH server
with a user's public keys. This document describes a new protocol
that builds on the protocol defined in RFC 4819 to allow the
provisioning of keys and certificates to a server using the SSH
transport.
The new protocol allows the calling client to organize keys and
certificates in different namespaces on a server. These namespaces
can be used by the server to allow a client to configure any
application running on the server (e.g., SSH, Key Management
Interoperability Protocol (KMIP), Simple Network Management Protocol
(SNMP)).
The new protocol provides a server-independent mechanism for clients
to add public keys, remove public keys, add certificates, remove
certificates, and list the current set of keys and certificates known
by the server by namespace (e.g., list all public keys in the SSH
namespace).
Rights to manage keys and certificates in a particular namespace are
specific and limited to the authorized user and are defined as part
of the server's implementation. The described protocol is backward
compatible to version 2 defined by RFC 4819.
Status of This Memo
This document is not an Internet Standards Track specification; it is
published for informational purposes.
This is a contribution to the RFC Series, independently of any other
RFC stream. The RFC Editor has chosen to publish this document at
its discretion and makes no statement about its value for
implementation or deployment. Documents approved for publication by
the RFC Editor are not a candidate for any level of Internet
Standard; see Section 2 of RFC 5741.
Information about the current status of this document, any errata,
and how to provide feedback on it may be obtained at
http://www.rfc-editor.org/info/rfc7076.
Copyright Notice
Copyright (c) 2013 IETF Trust and the persons identified as the
document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents
(http://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents
carefully, as they describe your rights and restrictions with respect
to this document.
Table of Contents
1. Introduction ....................................................3
2. Terminology .....................................................3
3. Overview of Extensions to the Public Key Subsystem ..............3
3.1. Extended Status Codes ......................................4
3.2. The Version Packet .........................................4
3.3. The Namespace Attribute ....................................4
4. New Operations ..................................................5
4.1. Adding a Certificate .......................................5
4.2. Removing a Certificate .....................................6
4.3. Listing Certificates .......................................6
4.4. Listing Namespaces .........................................7
5. Extending Public Key Operations .................................8
5.1. Adding a Public Key ........................................8
5.2. Removing a Public Key ......................................8
5.3. Listing Public Keys ........................................9
6. Security Considerations .........................................9
7. IANA Considerations ............................................10
8. References .....................................................10
8.1. Normative References ......................................10
8.2. Informative References ....................................10
1. Introduction
This document describes a new protocol that builds on the protocol
defined in RFC 4819 that can be used to configure public keys and
certificates in an implementation-independent fashion. The concept
of a namespace is added to the protocol's operations; it allows the
client to organize keys and certificates by application or
organizational structure.
P6R's Secure Shell Public Key Subsystem has been designed to run on
top of the Secure Shell transport layer [3] and user authentication
protocols [4]. It provides a simple mechanism for the client to
manage the public keys and certificates on the server related to that
client. These keys and certificates are normally used for
authentication of the client to a service, but they can be used for
encrypting results back to the client as well. Uploaded keys and
certificates are meant to be able to configure all protocols running
on a server (e.g., SSH, SSL, KMIP [8]) that use keys and
certificates, as well as the applications that run on a server.
This document should be read only after reading the Secure Shell
Public Key Subsystem [1] document. The new protocol described in
this document builds on and is meant to be backwards compatible with
the protocol described in [1].
2. Terminology
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in RFC 2119 [2].
3. Overview of Extensions to the Public Key Subsystem
The Public Key Subsystem provides a server-independent mechanism for
clients to add public keys, remove public keys, list the current
public keys known by the server, add certificates, remove
certificates, and list the current set of certificates known by the
server. This secure key distribution mechanism is implemented by a
new SSH subsystem with the name of "publickey@p6r.com".
3.1. Extended Status Codes
The status code gives the status in a more machine-readable format
(suitable for localization) and can have the following values:
SSH_PUBLICKEY_CERTIFICATE_NOT_FOUND 192
SSH_PUBLICKEY_CERTIFICATE_NOT_SUPPORTED 193
SSH_PUBLICKEY_CERTIFICATE_ALREADY_PRESENT 194
SSH_PUBLICKEY_ACTION_NOT_AUTHORIZED 195
SSH_PUBLICKEY_CANNOT_CREATE_NAMESPACE 196
The meaning of the failure codes is as implied by their names. See
Security Considerations for the use of the failure code:
SSH_PUBLICKEY_ACTION_NOT_AUTHORIZED.
3.2. The Version Packet
Both sides MUST start a connection by sending a version packet that
indicates the version of the protocol they are using.
string "version"
uint32 protocol-version-number
This document defines version 3 of the new protocol. We are using
version 3 so that it can be backward compatible with the protocol
defined by RFC 4819 [1].
3.3. The Namespace Attribute
The "namespace" attribute is added as an extension to what was
described in RFC 4819. The purpose of this attribute is to be able
to organize the uploaded keys and certificates into groups where each
group represents an application or organization structure. This
attribute is a string that should not be longer than 300 characters
and MUST be specified in UTF-8 format [5].
This new protocol uses the "ssh" namespace for the manipulation of
public keys in an SSH server and should be considered as the default
namespace when none is provided.
As a convention, namespaces used for protocols are lowercase strings
of the protocol's standard abbreviation. For example, "ssl" should
be the namespace used for the Secure Sockets Layer protocol.
Namespaces for applications should contain the product and vendor's
name. To help determine what namespaces already exist on a server, a
new operation "list-namespaces" is defined in Section 4.
4. New Operations
P6R's Public Key Subsystem extends the functionality defined in RFC
4819 with the following operations: add-certificate,
remove-certificate, list-certificates, and list-namespaces.
4.1. Adding a Certificate
If the client wishes to add a certificate, the client sends:
string "add-certificate"
string certificate format name
string certificate blob
boolean overwrite
uint32 attribute-count
string attrib-name
string attrib-value
bool critical
repeated attribute-count times
This request MUST include at least the "namespace" attribute so that
the server knows where to save the certificate. Only one namespace
attribute can be used per an add-certificate request. It is possible
for the same user to save the same certificate into multiple
namespaces, but this must be done with several separate
add-certificate requests.
If the namespace appearing in an add-certificate request does not
already exist on a server, then it is created by this operation.
However, if the user is not authorized to create a namespace, the
server MUST return SSH_PUBLICKEY_CANNOT_CREATE_NAMESPACE.
If the overwrite field is false and the specified certificate already
exists in the given namespace, the server MUST return
SSH_PUBLICKEY_CERTIFICATE_ALREADY_PRESENT. If the server returns
this, the client SHOULD provide an option to the user to overwrite
the certificate. If the overwrite field is true and the specified
key already exists in the given namespace but cannot be overwritten,
the server MUST return SSH_PUBLICKEY_ACCESS_DENIED.
However, a user may not be authorized to add a certificate to the
specified namespace. If the user does not have permission to add a
certificate, then the server MUST return
SSH_PUBLICKEY_ACTION_NOT_AUTHORIZED.
Examples of possible "certificate format names" are: "X509",
"pgp-sign-rsa", and "pgp-sign-dss". The format of the public key and
certificate blobs are detailed in Section 6.6, "Public Key
Algorithms", of the SSH Transport Protocol document [3], where X.509
certificates are to be encoded using a DER format [6] [7] in a
certificate blob.
4.2. Removing a Certificate
If the client wishes to remove a certificate, the client sends:
string "remove-certificate"
string certificate format name
string certificate blob
uint32 attribute-count
string attrib-name
string attrib-value
repeated attribute-count times
This request MUST include at least the "namespace" attribute so that
the server knows from where to delete the certificate. Only one
namespace attribute can be used per remove-certificate request. The
server MUST attempt to remove the certificate from the appropriate
location.
However, a user may not be authorized to remove a certificate from
the specified namespace. If the user does not have permission to
remove the certificate, then the server MUST return
SSH_PUBLICKEY_ACTION_NOT_AUTHORIZED.
Examples of possible "certificate format names" are: "X509",
"pgp-sign-rsa", and "pgp-sign-dss".
4.3. Listing Certificates
If the client wishes to list the known certificates, the client
sends:
string "list-certificates"
The server will respond with zero or more of the following responses:
string "certificate"
string certificate format name
string certificate blob
uint32 attribute-count
string attrib-name
string attrib-value
repeated attribute-count times
There is no requirement that the responses be in any particular
order. Whilst some server implementations may send the responses in
some order, client implementations should not rely on responses being
in any order.
This response MUST include at least the "namespace" attribute so that
a client can tell in which namespace the certificate resides. Only
one namespace attribute can be used per list-certificate request.
Following the last "certificate" response, a status packet MUST be
sent.
4.4. Listing Namespaces
If the client wishes to know existing namespaces on the server, it
sends:
string "list-namespaces"
The server will respond with zero or more of the following responses:
string "namespace"
string namespace name
It is possible that not all namespaces will be visible to every
authenticated user. In this case, the responding server will return
a subset of existing namespaces. See Security Considerations below.
Following the last "namespace" response, a status packet MUST be
sent.
5. Extending Public Key Operations
In addition to adding new operations, this document describes
extensions to the operations defined in RFC 4819.
5.1. Adding a Public Key
If the client wishes to add a public key, the client sends:
string "add"
string public key algorithm name
string public key blob
boolean overwrite
uint32 attribute-count
string attrib-name
string attrib-value
bool critical
repeated attribute-count times
This request MAY include one "namespace" attribute so that a client
can save the public key into a specific namespace. It is possible
for the same user to save the same key into multiple namespaces, but
this requires multiple add requests.
If the namespace appearing in an add public key request does not
already exist on a server, then it is created by this operation.
However, if the user is not authorized to create a namespace the
server MUST return SSH_PUBLICKEY_CANNOT_CREATE_NAMESPACE,
5.2. Removing a Public Key
If the client wishes to remove a public key, the client sends:
string "remove"
string public key algorithm name
string public key blob
uint32 attribute-count
string attrib-name
string attrib-value
bool critical
repeated attribute-count times
This extension allows attributes to be added to a remove request.
This request MAY include one "namespace" attribute so that a client
can remove the public key from a specific namespace.
5.3. Listing Public Keys
If the client wishes to list the known public keys, the client sends:
string "list"
uint32 attribute-count
string attrib-name
string attrib-value
bool critical
repeated attribute-count times
This extension allows attributes to be added to a list request. This
request MAY include one "namespace" attribute so that a client can
list the public keys from a specific namespace.
The server will respond with zero or more of the following responses:
string "publickey"
string public key algorithm name
string public key blob
uint32 attribute-count
string attrib-name
string attrib-value
repeated attribute-count times
This response MAY include the "namespace" attribute so that a client
can tell in which namespace the key resides.
6. Security Considerations
This protocol assumes that it is run over a secure channel and that
the endpoints of the channel have been authenticated. Thus, this
protocol assumes that it is externally protected from network-level
attacks.
This protocol provides a mechanism that allows key and certificate
material to be uploaded and manipulated into a server application.
It is the responsibility of the server implementation to enforce
access controls that may be required to limit any particular user's
access to the data in a namespace. For example, one user may be
allowed to list only the contents of a namespace but not add or
remove keys or certificates to/from it. The server MUST return
SSH_PUBLICKEY_ACTION_NOT_AUTHORIZED when a user's action goes against
its defined access controls.
This protocol requires the client to assume that the server will
correctly implement and observe attributes applied to keys.
Implementation errors in the server could cause clients to authorize
keys and certificates for access they were not intended to have, or
to apply fewer restrictions than were intended.
7. IANA Considerations
Although Section 3.1 defines four new status codes, these are in the
'Private Use' range of IANA's Publickey Subsystem Status Codes
registry as defined by Section 6.6.1 ("Conventions") in [1]. No IANA
actions are required for this document.
8. References
8.1. Normative References
[1] Galbraith, J., Van Dyke, J., and J. Bright, "Secure Shell Public
Key Subsystem", RFC 4819, March 2007.
[2] Bradner, S., "Key words for use in RFCs to Indicate Requirement
Levels", BCP 14, RFC 2119, March 1997.
[3] Ylonen, T. and C. Lonvick, Ed., "The Secure Shell (SSH) Transport
Layer Protocol", RFC 4253, January 2006.
[4] Ylonen, T. and C. Lonvick, Ed., "The Secure Shell (SSH)
Authentication Protocol", RFC 4252, January 2006.
[5] Yergeau, F., "UTF-8, a transformation format of ISO 10646", STD
63, RFC 3629, November 2003.
[6] Cooper, D., Santesson, S., Farrell, S., Boeyen, S., Housley, R.,
and W. Polk, "Internet X.509 Public Key Infrastructure
Certificate and Certificate Revocation List (CRL) Profile", RFC
5280, May 2008.
[7] ITU-T Recommendation X.690 (2002) | ISO/IEC 8825-1:2002,
Information technology -- ASN.1 encoding rules: Specification of
Basic Encoding Rules (BER), Canonical Encoding Rules (CER) and
Distinguished Encoding Rules (DER).
8.2. Informative References
[8] OASIS, "Key Management Interoperability Protocol (KMIP) 1.1",
January 2013, <http://docs.oasis-open.org/kmip/spec/v1.1/os/
kmip-spec-v1.1-os.html>.
Authors' Addresses
Mark Joseph, PhD
P6R, Inc
1840 41st Ave
Suite 102-139
Capitola, CA 95010
US
Phone: +1 888 452 2580 (x702)
EMail: mark@p6r.com
Jim Susoy
P6R, Inc
1840 41st Ave
Suite 102-139
Capitola, CA 95010
US
Phone: +1 888 452 2580 (x701)
EMail: jim@p6r.com
AI ELI5
On this page
§1Introduction to the Extension
The Secure Shell (SSH) Public Key Subsystem, defined in RFC 4819, provides a way to upload and manage a user's public keys on an SSH server. This document describes an extension to that protocol, adding a namespace concept that lets clients organize keys and certificates by the application or service that will use them.
P6R's extension runs as an SSH subsystem named publickey@p6r.com and operates on top of the SSH transport layer. Keys and certificates can be used for client authentication or to encrypt responses back to the client. The namespace system allows a single server to use the same key distribution protocol for different applications like SSH, KMIP, SNMP, or SSL, each with its own namespace.
The protocol is designed to be backward compatible with RFC 4819 version 2 while introducing version 3 with the new capabilities.
SSH lets you send keys to a server. This adds labels (namespaces) so a server can keep different types of keys separate.
§2Core Extensions
Three main extensions are added to the original protocol. First, a new set of status codes are defined: SSH_PUBLICKEY_CERTIFICATE_NOT_FOUND (192), SSH_PUBLICKEY_CERTIFICATE_NOT_SUPPORTED (193), SSH_PUBLICKEY_CERTIFICATE_ALREADY_PRESENT (194), SSH_PUBLICKEY_ACTION_NOT_AUTHORIZED (195), and SSH_PUBLICKEY_CANNOT_CREATE_NAMESPACE (196). These allow the server to report specific failure reasons in machine-readable form.
Second, a version packet starts every connection, letting both sides declare which protocol version they support. This document defines version 3.
Third, a namespace attribute is added to all operations. A namespace is a UTF-8 string of up to 300 characters that groups keys and certificates. The default namespace is ssh when none is provided. By convention, protocol namespaces use lowercase abbreviations (e.g., ssl, kmip, snmp), while application namespaces include the product and vendor name.
The extension adds version numbers so client and server agree on what they support, error codes so the server can explain what went wrong, and namespace names so everything knows which box to use.
§3Certificate Operations
The protocol adds four new operations for working with certificates. Add-certificate uploads a certificate in formats such as X.509, pgp-sign-rsa, or pgp-sign-dss, with a namespace attribute telling the server which namespace to store it in. If the namespace does not exist, the server creates it (unless the user lacks permission). The operation can overwrite an existing certificate or reject if one already exists.
Remove-certificate deletes a certificate from a namespace by its format and blob. List-certificates queries the server and returns all certificates in a namespace, zero or more at a time, ending with a status packet. List-namespaces shows all namespaces visible to the authenticated user; the server may hide namespaces the user is not authorized to access.
All certificate operations require the namespace attribute. Only one namespace can be specified per request. A user can upload the same certificate to multiple namespaces by sending separate requests.
You can add a certificate to a box, remove one, see all certificates in a box, or ask which boxes exist. Each action needs to say which box to use.
§4Extended Public Key Operations
The protocol also extends the original RFC 4819 public key operations. Add now optionally accepts a namespace attribute, allowing a public key to be saved to a specific namespace instead of just the default. If the namespace does not exist, the server creates it. An attribute system allows metadata to be attached to keys.
Remove now accepts attributes and can specify a namespace from which to delete a public key. List also accepts attributes and an optional namespace parameter to query public keys from a specific namespace. Server responses may include the namespace attribute so the client knows which namespace each key is in.
Unlike certificate operations, the namespace attribute is optional for public key operations, allowing backward compatibility with servers that do not understand namespaces.
The original way to add and remove keys gets updated to work with boxes too, but without requiring them so that old servers still work.
§5Security Considerations and Deployment
The protocol assumes it runs over a secure, authenticated SSH channel, protecting it from network-level attacks. The server implementation must enforce access controls to restrict what each user can do in each namespace. For example, one user might be able to list a namespace's keys but not add or remove them. If a user's action violates the server's access policy, the server returns SSH_PUBLICKEY_ACTION_NOT_AUTHORIZED.
Implementation correctness is critical. If a server incorrectly handles attributes attached to keys and certificates, it could authorize access that was not intended or apply fewer restrictions than desired. The protocol relies on correct server behavior and places the responsibility for access control on the server implementation, not the protocol itself.
Because the protocol is informational and vendor-specific, IANA registration is not required for the new status codes, which fall in the Private Use range of the SSH Public Key Subsystem registry.
The server has to be trusted to keep its boxes locked properly, and if it makes a mistake, keys could end up in the wrong hands.