ietf-dkim
[Top] [All Lists]

[ietf-dkim] NEW ISSUE: Security Threats are well defined

2008-02-11 18:11:54
Section 5 is pretty clear about security threat potentials about removed all considerations in SSP-02 to deal with them.

| 5.  Security Considerations
|
|   Security considerations in the Sender Signing Practices are mostly
|   related to attempts on the part of malicious senders to represent
|   themselves as other authors, often in an attempt to defraud either
|   the recipient or an Alleged Author.
|
|   Additional security considerations regarding Sender Signing
|   Practices may be found in the DKIM threat analysis [RFC4686].
|
|      *<<<THIS SECTION IS NOT COMPLETE.>>>*


I suggest the following be added in part or in whole to the section:

  There are clear and present danger with high potential frauds which
  SHOULD NOT be ignored if we are to gain any payoff or add-valued
  secured benefit for signers, verifiers and users:

  The fraud are exhibit when transactions contradict the following
  domain expectations:

    o NO MAIL EXPECTED
    o OPTIONAL 1ST PARTY SIGNING ONLY
    o STRICT 1ST PARTY SIGNING ONLY
    o 1ST OR 3rd PARTY SIGNING REQUIRED

  These represent the key propositions in the protocol consistency of
  the DKIM model and they can offer immediate and high potential
  security protection against current legacy domain forge abuses and
  expected DKIM-adapting mail abuses.

  Please keep in mind that these are not signature or domain reputation
  evaluators but fundamental DKIM protocol consistency checks.  It may
  be viewed a "DKIM Policy Declaration" Checking.  In addition, there
  is no verifier mandated handling requirement. All handling language
  recommendations cited below are defined by RFC2119.

  o NO MAIL EXPECTED

    One of the largest abuse of domains are forged mail transactions
    which are using RFC2822 From: domains which are not valid Email
    hosting domains and/or were not expected to be used for email
    transaction by the principal owner of the domain.

    Protection is offered in SSP-02 as follows:

    Domains can implicitly state mail is not expected by creating a NULL
    public key to force invalidation of a signature. Augmenting this
    with a DKIM=DISCARDABLE enforces the requirement of a valid
    signature, therefore with the lack of valid signature, the message
    SHOULD be rejected without prejudice.

  o OPTIONAL 1ST PARTY SIGNING ONLY

    Conventional wisdom suggest that domain explorers of DKIM
    will test the waters using an optional 1st party signing
    policy.  Some messages will be signed while others are not,
    but they don't expect 3rd party signers to "test" the waters
    for domain.

    Unexpected 3rd party signing protection is not offered via SSP-02.

    A new explicit policy DKIM=OPTIONAL is recommended that
    distinguishes itself from the current DKIM=UNKNOWN which allows
    anyone to sign and misrepresent the domain.

    DKIM=OPTIONAL (NEW)

    This policy declares the domain signing practices are optional, but
    only expect its own domain to sign the mail.  No other 3rd party
    representation is expected.  A message containing a 3rd party
    signature SHOULD be rejected without prejudice.

  o STRICT 1ST PARTY SIGNING ONLY

    This protection is offered using DKIM=DISCARDABLE. A message with no
    signature or containing a 3rd party signature SHOULD be rejected
    without prejudice.

  o 1ST OR 3rd PARTY SIGNING REQUIRED

    A weak protection is offered using DKIM=ALL. It is weak because it
    allows for an inconsistent state to exist without prejudice and
    lacks handling semantics for the detectable inconsistent DKIM=ALL
    policy transaction. A message with no signature or containing a 3rd
    party signature MAY be rejected without prejudice.


--
Sincerely

Hector Santos, CTO
http://www.santronics.com
http://santronics.blogspot.com

_______________________________________________
NOTE WELL: This list operates according to http://mipassoc.org/dkim/ietf-list-rules.html

<Prev in Thread] Current Thread [Next in Thread>
  • [ietf-dkim] NEW ISSUE: Security Threats are well defined, Hector Santos <=