ipsec September 2009 archive
Main Archive Page > Month Archives  > ipsec archives
ipsec: [IPsec] Issue #26: Missing treatment of error cases

[IPsec] Issue #26: Missing treatment of error cases

From: Yoav Nir <ynir_at_nospam>
Date: Tue Sep 01 2009 - 06:43:26 GMT
To: "ipsec@ietf.org WG" <ipsec@ietf.org>

Hello all.

Issue #26 was submitted by Tero Kivinen. It concerns section 2.21 ("error handling") and states that several things are missing:

  • handling of errors before authentication
  • listing what error conditions cause the IKE SA to be deleted entirely
  • listing how errors are handled in the piggybacked exchanges.

Following is our suggested new text. Please let us know what you think. Also, please take a look at the description of "AUTHENTICATION_FAILED" in section 3.10.1. "response to an IKE_AUTH message" means either an IKE_AUTH response to an IKE_AUTH request, or an INFORMATIONAL request that describes an error in the IKE_AUTH response. Do you think this phrasing is clear enough?

2.21. Error Handling

    There are many kinds of errors that can occur during IKE processing.     If a request is received that is badly formatted, or unacceptable for

    reasons of policy (e.g., no matching cryptographic algorithms), the     response MUST contain a Notify payload indicating the error. If an     error occurs in the processing of a response, then the initiator     SHOULD initiate an INFORMATIONAL exchange with a Notify payload     describing the problem. If an error occurs outside the context of an

    IKE request (e.g., the node is getting ESP messages on a nonexistent     SPI), the node SHOULD initiate an INFORMATIONAL exchange with a     Notify payload describing the problem.

    Errors that occur before a cryptographically protected IKE SA is     established must be handled very carefully. There is a trade-off     between wanting to be helpful in diagnosing a problem and responding     to it and wanting to avoid being a dupe in a denial of service attack

    based on forged messages.

    If a node receives a message on UDP port 500 or 4500 outside the     context of an IKE SA known to it (and not a request to start one), it

    may be the result of a recent crash of the node. If the message is     marked as a response, the node MAY audit the suspicious event but     MUST NOT respond. If the message is marked as a request, the node     MAY audit the suspicious event and MAY send a response. If a     response is sent, the response MUST be sent to the IP address and     port from whence it came with the same IKE SPIs and the Message ID     copied. The response MUST NOT be cryptographically protected and     MUST contain a Notify payload indicating INVALID_IKE_SPI. The     INVALID_IKE_SPI notification indicates an IKE message was received     with an unrecognized destination SPI; this usually indicates that the

    recipient has rebooted and forgotten the existence of an IKE SA.

    A node receiving such an unprotected Notify payload MUST NOT respond     and MUST NOT change the state of any existing SAs. The message might

    be a forgery or might be a response, the genuine correspondent was     tricked into sending. A node should treat such a message (and also a

    network message like ICMP destination unreachable) as a hint that     there might be problems with SAs to that IP address and should     initiate a liveness check for any such IKE SA. An implementation     SHOULD limit the frequency of such tests to avoid being tricked into     participating in a denial of service attack.

    A node receiving a suspicious message from an IP address (and port,     if NAT traversal is used) with which it has an IKE SA MAY send an IKE
    Notify payload in an IKE INFORMATIONAL exchange over that SA. The     recipient MUST NOT change the state of any SAs as a result, but may     wish to audit the event to aid in diagnosing malfunctions. A node     MUST limit the rate at which it will send messages in response to     unprotected messages.

    All errors that occur in an IKE_AUTH exchange, causing the     authentication to fail for whatever reason (invalid shared secret,     unrecognized ID, untrusted certificate issuer, revoked or expired     certificate, etc.) MUST result in an AUTHENTICATION_FAILED     notification. If the error occurred on the responder, the     notification MUST be returned in the protected response, and MUST be     the only payload in that response. If the error occurs on the     initiator, the notification MUST be returned in a separate     INFORMATIONAL exchange, with no other payloads. Note, however, that     messages that contain an unsupported critical payload, or that are     otherwise malformed, MUST be rejected in their entirety, and only     lead to an UNSUPPORTED_CRITICAL_PAYLOAD or INVALID_SYNTAX     Notification. The receiver MUST NOT verify the payloads related to     authentication in this case.

    If authentication has succeeded in the IKE_AUTH exchange, the IKE SA     is established, provided that there are no unsupported critical     payloads. Establishing the child SA, or requesting configuration     information may still fail, but they do not automatically cause the     IKE SA to be deleted. Specifically, a responder may include all the     payloads associated with authentication (IDr, Cert and AUTH) while     sending error notifications for the piggybacked exchanges     (FAILED_CP_REQUIRED, INVALID_SELECTORS, NO_PROPOSAL_CHOSEN, etc.),     and the initiator MUST NOT fail the authentication because of this.     The initiator MAY, of course, for reasons of policy later delete such

    an IKE SA.

    Only authentication failures and malformed packets lead to a deletion

    of the IKE SA without requiring an explicit DELETE payload. Other     error conditions require such a payload. In an IKE_SA_INIT exchange,

    any error notification causes the exchange to fail. In an IKE_AUTH     exchange, or in the subsequent INFORMATIONAL exchnage, only the     following notifications cause the IKE SA to be deleted or not     created, without a DELETE payload:

    Extension documents may define new error notifications with these     semantics, but MUST NOT use them unless the peer is known to     understand them.

IPsec mailing list