Interesting key merging bug with searching

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

Interesting key merging bug with searching

David Shaw
This was originally reported as a GnuPG bug, but it seems that SKS is
actually making the odd response here.

Basically there are two keys, both with key ID A56E15A3 (a natural
collision).  One has various user IDs indicating ownership by "Hubert
Figuiere".  The other one has a user ID for "Pedro R. Fernandez".  If
you search for "Figuiere" in SKS, you get both keys, and the user IDs
are joined so each key has IDs from the other.

$ curl 'http://keyserver.ubuntu.com:11371/pks/lookup?op=index&options=mr&search=Figuiere'
info:1:2
pub:A56E15A3:17:1024:1012389324::
uid:Hubert Figuiere <[hidden email]>:1119055887::
uid:Hubert Figuiere <[hidden email]>:1130648607::
uid:Hubert Figuiere <[hidden email]>:1028015454::
uid:Hubert Figuiere <[hidden email]>:1130648605::
uid:Pedro R. Fernandez (GNU/Linux User #000.204.555 http%3A//counter.li.org/) <[hidden email]>:::
pub:A56E15A3:17:1024:1002062559::
uid:Hubert Figuiere <[hidden email]>:::
uid:Hubert Figuiere <[hidden email]>:::
uid:Pedro R. Fernandez (GNU/Linux User #000.204.555 http%3A//counter.li.org/) <[hidden email]>:1002062559::

David


_______________________________________________
Sks-devel mailing list
[hidden email]
http://lists.nongnu.org/mailman/listinfo/sks-devel
Reply | Threaded
Open this post in threaded view
|

Re: Interesting key merging bug with searching

Yaron Minsky-2
I can't see offhand how this would show up as a result of a bug in the search mechanism:  the word searches are indexed by full key hashes, and so the word search should pull up the correct answer.  But what it actually looks like is that the keys of these people have been inadvetantly merged.  I have no reason to think offhand that the merging was done by SKS.  SKS only merges keys when the public key packet is identical, not merely when the keyids match.   But some other piece of software could have done this and fed it to SKS, which would then propagate it happily.  Does this show up only in the SKS servers, or do other servers have the same bad key?

y

On 11/6/05, David Shaw <[hidden email]> wrote:
This was originally reported as a GnuPG bug, but it seems that SKS is
actually making the odd response here.

Basically there are two keys, both with key ID A56E15A3 (a natural
collision).  One has various user IDs indicating ownership by "Hubert
Figuiere".  The other one has a user ID for "Pedro R. Fernandez".  If
you search for "Figuiere" in SKS, you get both keys, and the user IDs
are joined so each key has IDs from the other.

$ curl 'http://keyserver.ubuntu.com:11371/pks/lookup?op=index&options=mr&search=Figuiere'
info:1:2
pub:A56E15A3:17:1024:1012389324::
uid:Hubert Figuiere <[hidden email]>:1119055887::
uid:Hubert Figuiere <[hidden email]>:1130648607::
uid:Hubert Figuiere <[hidden email]>:1028015454::
uid:Hubert Figuiere <[hidden email]>:1130648605::
uid:Pedro R. Fernandez (GNU/Linux User #000.204.555 http%3A//counter.li.org/) < [hidden email]>:::
pub:A56E15A3:17:1024:1002062559::
uid:Hubert Figuiere <[hidden email]>:::
uid:Hubert Figuiere < [hidden email]>:::
uid:Pedro R. Fernandez (GNU/Linux User #000.204.555 http%3A//counter.li.org/) <[hidden email]>:1002062559::

David


_______________________________________________
Sks-devel mailing list
[hidden email]
http://lists.nongnu.org/mailman/listinfo/sks-devel


_______________________________________________
Sks-devel mailing list
[hidden email]
http://lists.nongnu.org/mailman/listinfo/sks-devel
Reply | Threaded
Open this post in threaded view
|

Re: Interesting key merging bug with searching

David Shaw
Good point.  I see this on the PKS servers as well.  Here's an example
of two different keys with colliding short key IDs sharing user IDs:

http://keyserver.kjsl.com:11371/pks/lookup?op=vindex&fingerprint=on&search=0x1EFFBA41

Been a while since I looked at the pksd code, but I would certainly
believe it could do this.  It seems that this only happens if the key
parameters are in common: I can't find a case where, for example, a
DSA and RSA key were merged.

David

On Sun, Nov 06, 2005 at 10:49:57PM -0500, Yaron Minsky wrote:

> I can't see offhand how this would show up as a result of a bug in the
> search mechanism: the word searches are indexed by full key hashes, and so
> the word search should pull up the correct answer. But what it actually
> looks like is that the keys of these people have been inadvetantly merged. I
> have no reason to think offhand that the merging was done by SKS. SKS only
> merges keys when the public key packet is identical, not merely when the
> keyids match. But some other piece of software could have done this and fed
> it to SKS, which would then propagate it happily. Does this show up only in
> the SKS servers, or do other servers have the same bad key?
>
> y
>
> On 11/6/05, David Shaw <[hidden email]> wrote:
> >
> > This was originally reported as a GnuPG bug, but it seems that SKS is
> > actually making the odd response here.
> >
> > Basically there are two keys, both with key ID A56E15A3 (a natural
> > collision). One has various user IDs indicating ownership by "Hubert
> > Figuiere". The other one has a user ID for "Pedro R. Fernandez". If
> > you search for "Figuiere" in SKS, you get both keys, and the user IDs
> > are joined so each key has IDs from the other.
> >
> > $ curl '
> > http://keyserver.ubuntu.com:11371/pks/lookup?op=index&options=mr&search=Figuiere
> > '
> > info:1:2
> > pub:A56E15A3:17:1024:1012389324::
> > uid:Hubert Figuiere <[hidden email]>:1119055887::
> > uid:Hubert Figuiere <[hidden email]>:1130648607::
> > uid:Hubert Figuiere <[hidden email]>:1028015454::
> > uid:Hubert Figuiere <[hidden email]>:1130648605::
> > uid:Pedro R. Fernandez (GNU/Linux User #000.204.555
> > http%3A//counter.li.org/) <[hidden email]>:::
> > pub:A56E15A3:17:1024:1002062559::
> > uid:Hubert Figuiere <[hidden email]>:::
> > uid:Hubert Figuiere <[hidden email]>:::
> > uid:Pedro R. Fernandez (GNU/Linux User #000.204.555
> > http%3A//counter.li.org/) <[hidden email]>:1002062559::
> >
> > David
> >
> >
> > _______________________________________________
> > Sks-devel mailing list
> > [hidden email]
> > http://lists.nongnu.org/mailman/listinfo/sks-devel
> >

> _______________________________________________
> Sks-devel mailing list
> [hidden email]
> http://lists.nongnu.org/mailman/listinfo/sks-devel



_______________________________________________
Sks-devel mailing list
[hidden email]
http://lists.nongnu.org/mailman/listinfo/sks-devel
Reply | Threaded
Open this post in threaded view
|

Re: Interesting key merging bug with searching

Yaron Minsky-2
On 11/6/05, David Shaw <[hidden email]> wrote:
Good point.  I see this on the PKS servers as well.  Here's an example
of two different keys with colliding short key IDs sharing user IDs:

http://keyserver.kjsl.com:11371/pks/lookup?op=vindex&fingerprint=on&search=0x1EFFBA41

Been a while since I looked at the pksd code, but I would certainly
believe it could do this.  It seems that this only happens if the key
parameters are in common: I can't find a case where, for example, a
DSA and RSA key were merged.

If that's true, it's hard to imagine SKS creating that issue, since things like RSA/DSA aren't even looked at by SKS.

y
 

David

On Sun, Nov 06, 2005 at 10:49:57PM -0500, Yaron Minsky wrote:

> I can't see offhand how this would show up as a result of a bug in the
> search mechanism: the word searches are indexed by full key hashes, and so
> the word search should pull up the correct answer. But what it actually
> looks like is that the keys of these people have been inadvetantly merged. I
> have no reason to think offhand that the merging was done by SKS. SKS only
> merges keys when the public key packet is identical, not merely when the
> keyids match. But some other piece of software could have done this and fed
> it to SKS, which would then propagate it happily. Does this show up only in
> the SKS servers, or do other servers have the same bad key?
>
> y
>
> On 11/6/05, David Shaw <[hidden email]> wrote:
> >
> > This was originally reported as a GnuPG bug, but it seems that SKS is
> > actually making the odd response here.
> >
> > Basically there are two keys, both with key ID A56E15A3 (a natural
> > collision). One has various user IDs indicating ownership by "Hubert
> > Figuiere". The other one has a user ID for "Pedro R. Fernandez". If
> > you search for "Figuiere" in SKS, you get both keys, and the user IDs
> > are joined so each key has IDs from the other.
> >
> > $ curl '
> > http://keyserver.ubuntu.com:11371/pks/lookup?op=index&options=mr&search=Figuiere
> > '
> > info:1:2
> > pub:A56E15A3:17:1024:1012389324::
> > uid:Hubert Figuiere <[hidden email]>:1119055887::
> > uid:Hubert Figuiere < [hidden email]>:1130648607::
> > uid:Hubert Figuiere <[hidden email]>:1028015454::
> > uid:Hubert Figuiere <[hidden email]>:1130648605::
> > uid:Pedro R. Fernandez (GNU/Linux User #000.204.555
> > http%3A//counter.li.org/) <[hidden email]>:::
> > pub:A56E15A3:17:1024:1002062559::
> > uid:Hubert Figuiere <[hidden email]>:::
> > uid:Hubert Figuiere <[hidden email]>:::
> > uid:Pedro R. Fernandez (GNU/Linux User #000.204.555
> > http%3A//counter.li.org/) <[hidden email]>:1002062559::
> >
> > David
> >
> >
> > _______________________________________________
> > Sks-devel mailing list
> > [hidden email]
> > http://lists.nongnu.org/mailman/listinfo/sks-devel
> >

> _______________________________________________
> Sks-devel mailing list
> [hidden email]
> http://lists.nongnu.org/mailman/listinfo/sks-devel



_______________________________________________
Sks-devel mailing list
[hidden email]
http://lists.nongnu.org/mailman/listinfo/sks-devel


_______________________________________________
Sks-devel mailing list
[hidden email]
http://lists.nongnu.org/mailman/listinfo/sks-devel