On Tue, Jun 10, 2025, at 16:14, Dave Crocker wrote:
> On 6/10/2025 11:39 AM, Bron Gondwana wrote:
>> Because I don't see it.
> 
> 
> This means you have done the affirmative work of trying to consider various 
> legitimate vs. illegitimate scenarios.  Please document those analyses.
> 
> 
> 
> It is typical for those making assertions of what is true to carry the 
> initial burden of demonstrating the basis for their claims.  Simple 
> assertions of faith in the truth does not typically carry any force.  
> 
> The burden on those who challenge the claims typically comes /after/ there is 
> a significant basis for believing the claims.
> 
> 
> 
> Here's the simplest case you should analyze:
> 
>> Processing through a legitimate alias.
>> 
> How can a receiver distinguish between this, versus receiving a illegitimate 
> DKIM Replay?
> 

Legitmate alias: [email protected] -> [email protected] -> 
[email protected]
Source email from: [email protected]

DKIM2: i=1; [email protected]; [email protected]; d=bbiw.net
DKIM2: i=2; [email protected] [email protected]; 
d=fastmailteam.com
DKIM2; i=3; [email protected] [email protected]; 
d=fastmailteam.com

Now - since both those aliases are under fastmailteam.com and the system is 
under Fastmail's control, we could elide the second header and just create the 
third one, but if you had a more complex case where I was forwarding my mail to 
my gmail afterwards, then obviously it would need another signature on that 
alias too.

DKIM2: i=1; [email protected]; [email protected]; d=bbiw.net
DKIM2: i=2; [email protected] [email protected]; 
d=fastmailteam.com
DKIM2; i=3; [email protected] [email protected]; 
d=fastmailteam.com
DKIM2; i=4; [email protected] [email protected]; 
d=fastmail.fm

And all those signatures could be created by the same system if it had access 
to keys for multiple domains.  Obviously, every domain in the chain would need 
to have its own key.  And we couldn't elide both 2 and 3, because then 
[email protected] would NOT be aligned with [email protected] and 
that would make it indistinguishable from replay because there would be no 
contiguous chain of "signature on hop `i+1` is created by the same domain as 
`rt` of hop `i`".

Bron.



--
  Bron Gondwana, CEO, Fastmail Pty Ltd
  [email protected]

_______________________________________________
Ietf-dkim mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to