Re: [quicwg/base-drafts] Stateless Reset packet sizes should not depend on the maximum connection ID length (#2869)

[ianswett <[e-mail]>]

The goal is not to avoid the attacker dropping the packet(I'm not sure what other type of interference is possible here?).  If they can do that, they can drop all packets and that's likely even worse.  The goal is to ensure the packet can't be reliably identified as a stateless reset, which is a much lower bar.

As I stated on #2770, if the client wants to ensure it's reset is always large enough to be indistinguishable, it needs to send a packet size that's one byte larger than the smallest packet it expects to receive.  Typically, this is a no-op for the client, so compliance is trivial.

No matter what happens, a server may send a stateless reset in response to any QUIC packet that's large enough to be valid, and it's not sensible to try to prevent a server from doing that.

Even if the packet is smaller than other packets, it's just odd, not invalid, as David said.

If we don't do this, I fear clients will increase their effective minimum packet size substantially, which is a cost everyone has to bear, when a large number of clients will use 0 byte CIDs.

-- 
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/issues/2869#issuecomment-507873146