Re: [quicwg/base-drafts] handling of coalesced packets with decryption errors creates DoS opportunity (#2308)

[Christian Huitema <[e-mail]>]

I am a bit skeptical of the idea of using coalesced packets for PMTU discovery. If I receive a handshake packet after the handshake is finished, I will most probably ignore it. I don't have the keys anymore, I cannot remove header protection or packet protection. Packet will be treated as a decryption error, and whatever is coalesced with it will most likely be ignored.

I understand the goal of coalescing is to find a CID in the ICMP payload, but that seems somewhat brittle. Parsing the ICMP based on the 5 tuple would be much more robust, and does not require playing games with packet coalescing.

I may be a bit biased because I would rather just completely ignore all ICMP packets, and rely only on end-to-end ACKs. It removes the need to find a portable way to obtain ICMP messages through the socket API, and it also prevents attacks by ICMP spoofing.

-- 
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/2308#issuecomment-490736116