From: Alan Barrett <barrett@daisy.ee.und.ac.za>
Date: Fri, 27 May 94 00:02:19 PDT
To: Paul Ferguson <paul@hawksbill.sprintmrn.com>
Subject: Re: Malformed Signatures?
In-Reply-To: <9405270230.AA05721@hawksbill.sprintmrn.com>
Message-ID: <Pine.3.89.9405270820.r180-0100000@newdaisy.ee.und.ac.za>
MIME-Version: 1.0
Content-Type: text/plain


> a "pgp -kc paul" PGP 2.6 declares my signatories: 
>             (Malformed or obsolete signature format)
> and asks if I would like it to:
>             Remove bad signatures (Y/n)? 
> Does this mean what I think it means? 

A PGP signature is made like this (I think):

  1.  Decide what to sign.  In the case of a signature on a key, it
      includes the key and the user-ID to which the signature will
      attest, the date, and some other stuff.

  2.  Generate a digest of the message to be signed.  This uses MD5.

  3.  Encapsulate the digest in some way.  The old way used some simple
      padding, and the new way uses slightly more complex padding.
      PGP versions up to 2.2 always generated the old format, but 2.2
      could understand both formats.  PGP 2.3 could understand both
      formats, and could generate either format under control of the
      "pkcs_compat" option.  PGP 2.5 and 2.6 don't like the old format,
      and I am not sure how easy it would be to teach them to understand
      it.

  4.  RSA encrypt the encapsulated digest, using the secret key of the
      signer.  The results obviously depend on the type of encapsulation
      chosen in step 3.

  5.  Encapsulate the encrypted digest and some other stuff in a PGP
      "packet".

If you want PGP 2.5 and 2.6 to understand your signatures, you have to
use the PKCS encapsulation at step 3 above.  This means that you have to
ask people who have already signed your key using the old method to sign
it again using the new method.

--apb (Alan Barrett)