Re: [Sks-devel] multiple subkey binding

[Jason Harris]

On Wed, Dec 03, 2003 at 01:44:58PM -0500, David Shaw wrote:
> On Wed, Dec 03, 2003 at 12:19:29PM -0500, Jason Harris wrote:

> > FWIW, the keys Peter reported all have "duplicate" subkey binding
> > signatures with the same date and top two hash bytes, where one has
> > a type 101(?) subpacket (an GPG-ism from earlier versions) and the

[Yaron, please default the list setting "avoid duplicate copies of messages"
to NO.  Thanks.]

("An" GPG-ism?  Gah!)

> > other doesn't.  Good keyservers don't modify packets...
> 
> Interesting.  Is the 101 subpacket hashed or not?

All the ones I've ever seen have been unhashed.
 
> If it's hashed, it's more likely a PGP x.509 subpacket.
> 
> GnuPG never exported the local subpackets, so someone would have to
> make an extreme effort to get them onto the server.

While I routinely merge the updated Debian keyrings from
rsync://keyring.debian.org/keyrings/ because that data doesn't make it onto
the (synchronized) keyservers by itself, my last copy of debian-keyring.gpg
only has this subkey signature for the first key Peter mentioned:

Old: Signature Packet(tag 2)(70 bytes)
        Ver 4 - new
        Sig type - Subkey Binding Signature(0x18).
        Pub alg - DSA Digital Signature Standard(pub 17)
        Hash alg - SHA1(hash 2)
        Hashed Sub: signature creation time(sub 2)(4 bytes)
                Time - Wed Jan 30 15:25:21 EST 2002
        Sub: issuer key ID(sub 16)(8 bytes)
                Key ID - 0xC04974CFDF6807BE
        Hash left 2 bytes - 06 2c 
        DSA r(160 bits) - ...
        DSA s(160 bits) - ...
                -> hash(160 bits)

for example, so I can't confirm this as the mechanism for this key.
Also, I think I've seen the private unhashed subpackets on enough
(non-Debian) keys to classify this as a general phenomenon.

Hmmm, getting my key from keyserver.kjsl.com, gpgsplitting it, removing
the 1st subkey signature, importing it into a temporary keyring, and
--exporting it still leaves the (GPG-generated) unhashed subpacket:

Old: Signature Packet(tag 2)(78 bytes)
        Ver 4 - new
        Sig type - Subkey Binding Signature(0x18).
        Pub alg - DSA Digital Signature Standard(pub 17)
        Hash alg - SHA1(hash 2)
        Hashed Sub: signature creation time(sub 2)(4 bytes)
                Time - Sat Aug 19 14:29:36 EDT 2000
        Sub: issuer key ID(sub 16)(8 bytes)
                Key ID - 0x4B2A4897D39DA0E3
        Sub: unknown(sub 101)(6 bytes)
        Hash left 2 bytes - 3b 9a 
        DSA r(159 bits) - ...
        DSA s(156 bits) - ...
                -> hash(160 bits)

So, such legacy subpackets seem to be able to find the keyservers with
newer versions of GPG, at least.  How the versions of the signatures
without the type 101 subpackets are getting generated is still unclear
to me, however.

-- 
Jason Harris          | NIC:  JH329, PGP:  This _is_ PGP-signed, isn't it?
address@hidden | web:  http://keyserver.kjsl.com/~jharris/

pgpJCQiM2P1RV.pgp
Description: PGP signature