Re: [TLS] AIA cert fetching seen as harmful

[Nelson B Bolyard <nelson at bolyard.com>]

Eric Rescorla wrote, On 2008-04-10 18:08:
> At Thu, 10 Apr 2008 17:45:06 -0700,
> Nelson B Bolyard wrote:
>> Mike wrote, On 2008-04-10 09:01:
>>
>>> This could be made safe with some help from PKIX (if X.509 doesn't
>>> already support it -- I haven't read RFC 3280 or -bis in a while).
>>> If root certificates listed constraints on what constitutes a valid
>>> URL for retrieving issued certificates, then a server could scan
>>> the combined list from each trusted root to determine if it is safe
>>> to fetch a client certificate.
>> Are you all aware of this paper, now making a stir?
>>
>>     https://www.cynops.de/techzone/http_over_x509.html
> 
> Yes, Martin cited this paper a few weeks ago.
> 
>  
>> It claims that fetching CA certs from URLs found in AIA extensions in certs
>> that have not yet been validated is a vulnerability.  At least one browser
>> organization known to me agrees.
> 
> How does that organization feel about inline images in HTML pages?

The problem isn't so much when browsers initiate fetches for certs from
servers.  The major concerns are:
a) servers fetching URLs from unvetted client auth certs, and
b) mail clients fetching certs to verify signatures in emails from strangers.

Some email clients, in particular, are good at not fetching remote content
from html emails, which confirms email addresses to spammers.  AIA cert
fetching weakens their ability to defend against such attempts to validate
email addresses.

Servers see them selves as similarly weakened.

I'm receiving inquiries about white listing CA URLs for AIA fetching. :(
_______________________________________________
TLS mailing list
TLS at ietf.org
https://www.ietf.org/mailman/listinfo/tls