SASL - Implementation of SASL mechanisms for Tcl
The Simple Authentication and Security Layer (SASL) is a framework for providing authentication and authorization to comunications protocols. The SASL framework is structured to permit negotiation among a number of authentication mechanisms. SASL may be used in SMTP, IMAP and HTTP authentication. It is also in use in XMPP, LDAP and BEEP. See MECHANISMS for the set of available SASL mechanisms provided with tcllib.
The SASL framework operates using a simple multi-step challenge response mechanism. All the mechanisms work the same way although the number of steps may vary. In this implementation a callback procedure must be provided from which the SASL framework will obtain users details. See CALLBACK PROCEDURE for details of this procedure.
Contruct a new SASL context. See OPTIONS for details of the possible options to this command. A context token is required for most of the SASL procedures.
Modify and inspect the SASL context option. See OPTIONS for further details.
This is the core procedure for using the SASL framework. The step procedure should be called until it returns 0. Each step takes a server challenge string and the response is calculated and stored in the context. Each mechanism may require one or more steps. For some steps there may be no server challenge required in which case an empty string should be provided for this parameter. All mechanisms should accept an initial empty challenge.
Returns the next response string that should be sent to the server.
Re-initialize the SASL context. Discards any internal state and permits the token to be reused.
Release all resources associated with the SASL context. The context token may not be used again after this procedure has been called.
Returns a list of all the available SASL mechanisms. The list is sorted by the mechanism preference value (see register) with the preferred mechanisms and the head of the list. Any mechanism with a preference value less than theminimum (which defaults to 0) is removed from the returned list. This permits a security threshold to be set. Mechanisms with a preference less that 25 transmit authentication are particularly susceptible to eavesdropping and should not be provided unless a secure channel is in use (eg: tls).
The type parameter may be one of client or server and defaults to client. Only mechanisms that have an implementation matching the type are returned (this permits servers to correctly declare support only for mechanisms that actually provide a server implementation).
New mechanisms can be added to the package by registering the mechanism name and the implementing procedures. The server procedure is optional. The preference value is an integer that is used to order the list returned by the mechanisms command. Higher values indicate a preferred mechanism. If the mechanism is already registered then the recorded values are updated.
Specify a command to be evaluated when the SASL mechanism requires information about the user. The command is called with the current SASL context and a name specifying the information desired. See EXAMPLES.
Set the SASL mechanism to be used. See mechanisms for a list of supported authentication mechanisms.
Set the service type for this context. Some mechanisms may make use of this parameter (eg DIGEST-MD5, GSSAPI and Kerberos). If not set it defaults to an empty string. If the -type is set to 'server' then this option should be set to a valid service identity. Some examples of valid service names are smtp, ldap, beep and xmpp.
This option is used to set the server name used in SASL challenges when operating as a SASL server.
The context type may be one of 'client' or 'server'. The default is to operate as a client application and respond to server challenges. Mechanisms may be written to support server-side SASL and setting this option will cause each step to issue the next challenge. A new context must be created for each incoming client connection when in server mode.
When the SASL framework requires any user details it will call the procedure provided when the context was created with an argument that specfies the item of information required.
In all cases a single response string should be returned.
The callback procedure should return the users authorization identity. Return an empty string unless this is to be different to the authentication identity. Read [1] for a discussion about the specific meaning of authorization and authentication identities within SASL.
The callback procedure should return the users authentication identity. Read [1] for a discussion about the specific meaning of authorization and authentication identities within SASL.
The callback procedure should return the password that matches the authentication identity as used within the current realm.
For server mechanisms the password callback should always be called with the authentication identity and the realm as the first two parameters.
Some SASL mechanisms use realms to partition authentication identities. The realm string is protocol dependent and is often the current DNS domain or in the case of the NTLM mechanism it is the Windows NT domain name.
Returns the client host name - typically [info host].
As used in FTP this mechanism only passes an email address for authentication. The ANONYMOUS mechanism is specified in [2].
This is the simplest mechanism. The users authentication details are transmitted in plain text. This mechanism should not be provided unless an encrypted link is in use - typically after SSL or TLS has been negotiated.
The LOGIN [1] mechanism transmits the users details with base64 encoding. This is no more secure than PLAIN and likewise should not be used without a secure link.
This mechanism avoids sending the users password over the network in plain text by hashing the password with a server provided random value (known as a nonce). A disadvantage of this mechanism is that the server must maintain a database of plaintext passwords for comparison. CRAM-MD5 was defined in [4].
This mechanism improves upon the CRAM-MD5 mechanism by avoiding the need for the server to store plaintext passwords. With digest authentication the server needs to store the MD5 digest of the users password which helps to make the system more secure. As in CRAM-MD5 the password is hashed with a server nonce and other data before being transmitted across the network. Specified in [3].
OTP is the One-Time Password system described in RFC 2289 [6]. This mechanism is secure against replay attacks and also avoids storing password or password equivalents on the server. Only a digest of a seed and a passphrase is ever transmitted across the network. Requires the otp package from tcllib and one or more of the cryptographic digest packages (md5 or sha-1 are the most commonly used).
This is a proprietary protocol developed by Microsoft [5] and is in common use for authenticating users in a Windows network environment. NTLM uses DES encryption and MD4 digests of the users password to authenticate a connection. Certain weaknesses have been found in NTLM and thus there are a number of versions of the protocol. As this mechanism has additional dependencies it is made available as a separate sub-package. To enable this mechanism your application must load the SASL::NTLM package.
This is a proprietary protocol developed by Google and used for authenticating users for the Google Talk service. This mechanism makes a pair of HTTP requests over an SSL channel and so this mechanism depends upon the availability of the tls and http packages. To enable this mechanism your application must load the SASL::XGoogleToken package. In addition you are recommended to make use of the autoproxy package to handle HTTP proxies reasonably transparently.
This is a protocol specified in RFC 5802 [7]. To enable this mechanism your application must load the SASL::SCRAM package.
See the examples subdirectory for more complete samples using SASL with network protocols. The following should give an idea how the SASL commands are to be used. In reality this should be event driven. Each time the step command is called, the last server response should be provided as the command argument so that the SASL mechanism can take appropriate action.
proc ClientCallback {context command args} {
switch -exact -- $command {
login { return "" }
username { return $::tcl_platform(user) }
password { return "SecRet" }
realm { return "" }
hostname { return [info host] }
default { return -code error unxpected }
}
}
proc Demo {{mech PLAIN}} {
set ctx [SASL::new -mechanism $mech -callback ClientCallback]
set challenge ""
while {1} {
set more_steps [SASL::step $ctx challenge]
puts "Send '[SASL::response $ctx]'"
puts "Read server response into challenge var"
if {!$more_steps} {break}
}
SASL::cleanup $ctx
}
Myers, J. "Simple Authentication and Security Layer (SASL)", RFC 2222, October 1997. (http://www.ietf.org/rfc/rfc2222.txt)
Newman, C. "Anonymous SASL Mechanism", RFC 2245, November 1997. (http://www.ietf.org/rfc/rfc2245.txt)
Leach, P., Newman, C. "Using Digest Authentication as a SASL Mechanism", RFC 2831, May 2000, (http://www.ietf.org/rfc/rfc2831.txt)
Klensin, J., Catoe, R. and Krumviede, P., "IMAP/POP AUTHorize Extension for Simple Challenge/Response" RFC 2195, September 1997. (http://www.ietf.org/rfc/rfc2195.txt)
No official specification is available. However, http://davenport.sourceforge.net/ntlm.html provides a good description.
Haller, N. et al., "A One-Time Password System", RFC 2289, February 1998, (http://www.ieft.org/rfc/rfc2289.txt)
Newman, C. et al., "Salted Challenge Response Authentication Mechanism (SCRAM) SASL and GSS-API Mechanisms", RFC 5802, July 2010, (http://www.ieft.org/rfc/rfc5802.txt)
Pat Thoyts
This document, and the package it describes, will undoubtedly contain bugs and other problems. Please report such in the category sasl of the Tcllib Trackers. Please also report any ideas for enhancements you may have for either package and/or documentation.
Networking
Copyright © 2005-2006, Pat Thoyts <patthoyts@users.sourceforge.net>