Re: [quicwg/base-drafts] Allow Smaller Stateless Resets (#2927)

[Martin Thomson <[e-mail]>]

It's pretty much "it's OK to go smaller if you are willing to risk stateless reset messages being identifiable", but a little stronger than that.  The sender of a stateless reset is now no longer bound by the upper limit on connection ID size.  That allows for a smaller packet to be sent and still have it get a stateless reset in response.

However, the onus for ensuring that the resulting stateless reset is indistinguishable from regular packets moves to the endpoint that would receive the stateless reset.  It has to ensure that it never sends a packet small enough that a stateless reset packet would be too small to appear as valid.

For instance, take the case where a client needs at least 8 bytes of connection ID, but the server had no connection ID.  Though the client has no real need to send larger packets toward the server, it has to pad outbound packets so that the server, should it be forced to send a stateless reset, wouldn't end up sending a packet that appears to have a 7 byte connection ID (which, assuming an observer knows about the client's constraints, would be easy to identify as a stateless reset).

-- 
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/2927#issuecomment-527692685