Hello, using 5.15.0-1019-bluefield, we see the tuples were offloaded and we also see the offload entries.
Thanks ** Tags removed: verification-needed-jammy ** Tags added: verification-done-jammy -- You received this bug notification because you are a member of Kernel Packages, which is subscribed to linux-bluefield in Ubuntu. https://bugs.launchpad.net/bugs/2012571 Title: net/sched: cls_api: Support hardware miss to tc action Status in linux-bluefield package in Ubuntu: Invalid Status in linux-bluefield source package in Focal: Fix Released Status in linux-bluefield source package in Jammy: Fix Committed Bug description: * Explain the bug(s) Currently tc miss interface only supports resuming from a specific tc chain. If a packet modification is done before a missable action such as CT and there is a miss in CT after it, this may cause a miss match when resuming re-executing the same chain in software, and wrong packet count. This use case for example is a stateless (static) nat. * brief explanation of fixes Add support for missing to a specific action instance, and support of per action hardware stats to update what was actually done in hardware. * How to test Create OVS bridge with 2 devices mlx5 rep devices. Enable HW offload and configure regular connection tracking OpenFlow rules with packet modification before the CT action (such as statless nat): e.g: ovs-ofctl del-flows br-ovs ovs-ofctl add-flow br-ovs arp,actions=normal ovs-ofctl add-flow br-ovs "in_port=1,table=0, ip,ct_state=-trk actions=mod_nw_dst=1.1.1.2,ct(table=1)" ovs-ofctl add-flow br-ovs "in_port=1,table=1, ip,ct_state=+trk+new actions=ct(commit),output:2" ovs-ofctl add-flow br-ovs "in_port=1,table=1, ip,ct_state=+trk+est, actions=output:2" ovs-ofctl add-flow br-ovs "in_port=2,table=0, ip,ct_state=-trk actions=ct(table=1)" ovs-ofctl add-flow br-ovs "in_port=2,table=1, ip,ct_state=+trk+est, actions=mod_nw_src=1.1.1.2,output:1" Config VF1 ip 1.1.1.1, VF2 ip 1.1.1.2 For VF2, add route and static neighbour to floating (router) ip 5.5.5.5 Then run a TCP connection, e.g: on mlx5 VF1 iperf -s #(which will listen on 1.1.1.2) on mlx5 VF2 iperf -c 5.5.5.5 -t 10 #(this creates a packet from 1.1.1.1 -> 5.5.5.5, and nat will change this to 1.1.1.1->1.1.1.2) Optional: In different terminal, while traffic is running, check for offload: tcpdump -nnepi <RELEVANT_MLX5_REP> tcp and see no iperf tcp packets. Dump conntrack with relevant ip: cat /proc/net/nf_conntrack | grep -i 1.1.1.1 See tuples were offloaded: ipv4 2 tcp 6 src= 1.1.1.1 dst=1.1.1.2 sport=56394 dport=5001 packets=2 bytes=112 src=1.1.1.2 dst=1.1.1.1 sport=5001 dport=56394 packets=1777 bytes=665340 [HW_OFFLOAD] mark=0 zone=0 use=3 * What it could break. offload for modifications + ct and tc packet count. To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux-bluefield/+bug/2012571/+subscriptions -- Mailing list: https://launchpad.net/~kernel-packages Post to : kernel-packages@lists.launchpad.net Unsubscribe : https://launchpad.net/~kernel-packages More help : https://help.launchpad.net/ListHelp
