Re: [PATCH v10 net-next 00/15] AccECN protocol case handling series
From: Neal Cardwell <ncardwell@google.com>
Date: 2026-01-21 21:07:49
Also in:
bpf, linux-kselftest, netdev
On Wed, Jan 21, 2026 at 3:55 PM Jakub Kicinski [off-list ref] wrote:
On Wed, 21 Jan 2026 15:32:00 +0100 chia-yu.chang@nokia-bell-labs.com wrote:quoted
From: Chia-Yu Chang <redacted> Hello, Plesae find the v10 AccECN case handling patch series, which covers several excpetional case handling of Accurate ECN spec (RFC9768), adds new identifiers to be used by CC modules, adds ecn_delta into rate_sample, and keeps the ACE counter for computation, etc.Looks like we went in the wrong directions, multiple tests are failing now: https://netdev.bots.linux.dev/contest.html?branch=net-next-2026-01-21--18-00&executor=vmksft-packetdrill&pw-n=0&pass=0
Looks like the first two are failing because the v10 series has a different layout for tcp_info than what packetdrill currently expects: https://netdev-ctrl.bots.linux.dev/logs/vmksft/packetdrill/results/483641/30-tcp-accecn-delivered-updates-pkt/stdout https://netdev-ctrl.bots.linux.dev/logs/vmksft/packetdrill/results/483641/58-tcp-accecn-delivered-falseoverflow-detect-pkt/stdout Chia-Yu, can you please post a patch for packetdrill to make it expect the new layout from the v10 series? For the third one: https://netdev-ctrl.bots.linux.dev/logs/vmksft/packetdrill/results/483641/69-tcp-accecn-client-accecn-options-lost-pkt/stdout ... it looks like the test line that says: +.002 ... 0.052 connect(4, ..., ...) = 0 needs to be changed to: +.002 ... 0.102 connect(4, ..., ...) = 0 It seems like running the tests before emailing the patches should have caught these 3 issues? Chia-Yu, can you please make sure you run all the AccECN packetdrill tests with the latest kernel and latest packetdrill binary you are proposing, before posting a v11 kernel patch series and packetdrill patch? Thanks, neal