Re: [quicwg/base-drafts] token-based greasing / initial packet protection (#3166)

[Mike Bishop <[e-mail]>]

MikeBishop commented on this pull request.

I can see both sides with regard to a fallback path.  A MitM that wanted to force you back to v1 could manufacture a VN packet in response to versions it doesn't understand, swallowing the Initial or the server's actual reply.  Prohibiting fallback means that such a MitM can only prevent connections, not inspect the ClientHello.

That said, I'm dubious that most clients will not fall back at all.  The alternative is falling back to TCP, which is even more inspectable, or total connection failure.  History shows that total connection failure is not an outcome most clients or servers are comfortable with.

I think what we really want is a fallback path where an intermediary's attempt at interference becomes visible; endpoints can decide how to proceed in that situation without the spec telling them.  But then we're back to designing QUIC's version negotiation and downgrade protection mechanism, something we said we weren't going to do in v1.

> @@ -4970,6 +5080,30 @@ The NEW_TOKEN frame is as follows:
 
 NEW_TOKEN frames contain the following fields:
 
+Lifetime:
+
+: Indicates the lifetime of the values contained in this frame in milliseconds.
+  An endpoint MUST NOT use the values provided by this frame, once the time that

```suggestion
  An endpoint MUST NOT use the values provided by this frame once the time that
```

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/quicwg/base-drafts/pull/3166#pullrequestreview-309480916