Re: [TLS] I-D Action: draft-ietf-tls-svcb-ech-01.txt

[Raghu Saxena <[e-mail]>]

On 3/28/24 04:37, [e-mail] wrote:
> Internet-Draft draft-ietf-tls-svcb-ech-01.txt is now available. It is a work
> item of the Transport Layer Security (TLS) WG of the IETF.
>
>     Title:   Bootstrapping TLS Encrypted ClientHello with DNS Service Bindings
>     Authors: Ben Schwartz
>              Mike Bishop
>              Erik Nygren
>     Name:    draft-ietf-tls-svcb-ech-01.txt
>     Pages:   6
>     Dates:   2024-03-27

Just wondering, do we want to explicitly mention looking up the SVCB 
record for ECHConfig using "Port Prefix Naming" e.g. with words like 
MUST? From the SVCB RFC (i.e. RFC 9460), it's mentioned under Section 
2.3 that non-standard ports MAY be specified.

However, for something security-related like ECH, I think a client MUST 
lookup the port-prefixed HTTPS record for determining which ECHConfig to 
use. As an example, for my personal ECH test website[0], different ports 
advertise different ECHConfigs. Chromium handles this correctly, but 
Firefox does not, which is considered to be a bug[1]. In my particular 
example, all ports have the same backend which can find a relevant ECH 
private key to use, but in some cases this may not be the case, so using 
the ECHConfig of port 443 (default) on another port could lead to problems.

I'm not sure if we need to explicitly mention it here, but since the 
draft seems to re-iterate some points of both SVCB & ECH, it may be useful.

Regards,

Raghu Saxena

[0] https://rfc5746.mywaifu.best:4443/

[1] https://bugzilla.mozilla.org/show_bug.cgi?id=1860038

P.S. The current draft links to SVCB as 
"https://datatracker.ietf.org/doc/html/draft-ietf-dnsop-svcb-https-12", 
but since it is now standardized as RFC 9460, I guess it should be updated.