W3C home > Mailing lists > Public > public-xmlsec@w3.org > January 2009

Re: Summary of problems with the definition of ECDSA public key in RFC 4050

From: Magnus Nyström <magnus@rsa.com>
Date: Mon, 19 Jan 2009 00:03:50 -0500 (Eastern Standard Time)
To: public-xmlsec@w3.org
Message-ID: <Pine.WNT.4.64.0901182317390.744@W-JNISBETTEST-1.tablus.com>

(This is a long overdue review of Kelvin's analysis of issues with RFC 
4050.)

Having looked at this in more detail, I agree with Kevin's analysis. The 
important areas as I see it are the incompatibility with X9.62 wrt 
ECPoints and the potential issues arising from the use of 
nonNegativeInteger.

As for the current proposal, I have the following comments/questions:

- You don't want to include the "implicitCA" option for ECKeyValueType?
   This allows for re-using an algorithm specified by a CA policy. (For
   clarity and, perhaps future flexibility, I'd also suggest having a
   separate ECDomainParameterType type as in X9.62 and then the
   ECKeyValueType etc. would become:

   <complexType name="ECKeyValueType">
     <sequence>
       <element ref="ds:ECDomainParameters"/>
       <element name="PublicKey" type="ds:ECPointType"/>
     </sequence>
   </complexType>

   <element name="ECDomainParameters" type="ds:ECDomainParametersType"/>
   <complexType name="ECDomainParametersType">
     <sequence>
       <choice>
         <element ref="ds:ExplicitECParams"/>
         <element name="NamedCurve..."/>
         <element name="ImplicitCA"/>
       </choice>
     </sequence>
   </complexType>

   <element name="ExplicitECParams" type="ds:ExplicitECParamsType"/>
   <complexType name="ExplicitECParamsType">
     <sequence ... -> same as current ECParametersType>

- I'm missing a <Hash> element in the ds:ECParametersType type definition.
   In X9.62, this component is used to verifiably generate the curve. Do
   you feel we do not need it, Kelvin?

- Also in the ECParameterType type definition, note that the <Order>
   element may also be a large number and hence the use of the
   "integer" type may again be problematic.

- For the PrimeFieldParameterType, note that primes will also be large
   (larger than 18 decimal digits).

- The reference for conversion of an EC point to an Octet String would
   perhaps better be to Appendix A.5.7 ("Point to Octet String").

Best,
-- Magnus
Received on Monday, 19 January 2009 05:04:27 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:43:57 GMT