kernel/linux.git/net/sched/Makefile, branch v6.18.22

sched: Add enqueue/dequeue of dualpi2 qdisc

2025-07-24T00:52:07+00:00

DualPI2 provides L4S-type low latency & loss to traffic that uses a scalable congestion controller (e.g. TCP-Prague, DCTCP) without degrading the performance of 'classic' traffic (e.g. Reno, Cubic etc.). It is to be the reference implementation of IETF RFC9332 DualQ Coupled AQM (https://datatracker.ietf.org/doc/html/rfc9332). Note that creating two independent queues cannot meet the goal of DualPI2 mentioned in RFC9332: "...to preserve fairness between ECN-capable and non-ECN-capable traffic." Further, it could even lead to starvation of Classic traffic, which is also inconsistent with the requirements in RFC9332: "...although priority MUST be bounded in order not to starve Classic traffic." DualPI2 is designed to maintain approximate per-flow fairness on L-queue and C-queue by forming a single qdisc using the coupling factor and scheduler between two queues. The qdisc provides two queues called low latency and classic. It classifies packets based on the ECN field in the IP headers. By default it directs non-ECN and ECT(0) into the classic queue and ECT(1) and CE into the low latency queue, as per the IETF spec. Each queue runs its own AQM: * The classic AQM is called PI2, which is similar to the PIE AQM but more responsive and simpler. Classic traffic requires a decent target queue (default 15ms for Internet deployment) to fully utilize the link and to avoid high drop rates. * The low latency AQM is, by default, a very shallow ECN marking threshold (1ms) similar to that used for DCTCP. The DualQ isolates the low queuing delay of the Low Latency queue from the larger delay of the 'Classic' queue. However, from a bandwidth perspective, flows in either queue will share out the link capacity as if there was just a single queue. This bandwidth pooling effect is achieved by coupling together the drop and ECN-marking probabilities of the two AQMs. The PI2 AQM has two main parameters in addition to its target delay. The integral gain factor alpha is used to slowly correct any persistent standing queue error from the target delay, while the proportional gain factor beta is used to quickly compensate for queue changes (growth or shrinkage). Either alpha and beta are given as a parameter, or they can be calculated by tc from alternative typical and maximum RTT parameters. Internally, the output of a linear Proportional Integral (PI) controller is used for both queues. This output is squared to calculate the drop or ECN-marking probability of the classic queue. This counterbalances the square-root rate equation of Reno/Cubic, which is the trick that balances flow rates across the queues. For the ECN-marking probability of the low latency queue, the output of the base AQM is multiplied by a coupling factor. This determines the balance between the flow rates in each queue. The default setting makes the flow rates roughly equal, which should be generally applicable. If DUALPI2 AQM has detected overload (due to excessive non-responsive traffic in either queue), it will switch to signaling congestion solely using drop, irrespective of the ECN field. Alternatively, it can be configured to limit the drop probability and let the queue grow and eventually overflow (like tail-drop). GSO splitting in DUALPI2 is configurable from userspace while the default behavior is to split gso. When running DUALPI2 at unshaped 10gigE with 4 download streams test, splitting gso apart results in halving the latency with no loss in throughput: Summary of tcp_4down run 'no_split_gso': avg median # data pts Ping (ms) ICMP : 0.53 0.30 ms 350 TCP download avg : 2326.86 N/A Mbits/s 350 TCP download sum : 9307.42 N/A Mbits/s 350 TCP download::1 : 2672.99 2568.73 Mbits/s 350 TCP download::2 : 2586.96 2570.51 Mbits/s 350 TCP download::3 : 1786.26 1798.82 Mbits/s 350 TCP download::4 : 2261.21 2309.49 Mbits/s 350 Summart of tcp_4down run 'split_gso': avg median # data pts Ping (ms) ICMP : 0.22 0.23 ms 350 TCP download avg : 2335.02 N/A Mbits/s 350 TCP download sum : 9340.09 N/A Mbits/s 350 TCP download::1 : 2335.30 2334.22 Mbits/s 350 TCP download::2 : 2334.72 2334.20 Mbits/s 350 TCP download::3 : 2335.28 2334.58 Mbits/s 350 TCP download::4 : 2334.79 2334.39 Mbits/s 350 A similar result is observed when running DUALPI2 at unshaped 1gigE with 1 download stream test: Summary of tcp_1down run 'no_split_gso': avg median # data pts Ping (ms) ICMP : 1.13 1.25 ms 350 TCP download : 941.41 941.46 Mbits/s 350 Summart of tcp_1down run 'split_gso': avg median # data pts Ping (ms) ICMP : 0.51 0.55 ms 350 TCP download : 941.41 941.45 Mbits/s 350 Additional details can be found in the draft: https://datatracker.ietf.org/doc/html/rfc9332 Signed-off-by: Koen De Schepper Co-developed-by: Olga Albisser Signed-off-by: Olga Albisser Co-developed-by: Olivier Tilmans Signed-off-by: Olivier Tilmans Co-developed-by: Henrik Steen Signed-off-by: Henrik Steen Co-developed-by: Chia-Yu Chang Signed-off-by: Chia-Yu Chang Signed-off-by: Bob Briscoe Signed-off-by: Ilpo Järvinen Acked-by: Dave Taht Link: https://patch.msgid.link/20250722095915.24485-4-chia-yu.chang@nokia-bell-labs.com Signed-off-by: Jakub Kicinski

bpf: net_sched: Support implementation of Qdisc_ops in bpf

2025-04-17T17:54:33+00:00

The recent advancement in bpf such as allocated objects, bpf list and bpf rbtree has provided powerful and flexible building blocks to realize sophisticated packet scheduling algorithms. As struct_ops now supports core operators in Qdisc_ops, start allowing qdisc to be implemented using bpf struct_ops with this patch. Users can implement Qdisc_ops.{enqueue, dequeue, init, reset, destroy} in bpf and register the qdisc dynamically into the kernel. Co-developed-by: Cong Wang Signed-off-by: Cong Wang Signed-off-by: Amery Hung Signed-off-by: Martin KaFai Lau Acked-by: Cong Wang Acked-by: Toke Høiland-Jørgensen Link: https://patch.msgid.link/20250409214606.2000194-3-ameryhung@gmail.com

net/sched: Retire ipt action

2024-01-02T12:41:16+00:00

The tc ipt action was intended to run all netfilter/iptables target. Unfortunately it has not benefitted over the years from proper updates when netfilter changes, and for that reason it has remained rudimentary. Pinging a bunch of people that i was aware were using this indicates that removing it wont affect them. Retire it to reduce maintenance efforts. Buh-bye. Reviewed-by: Victor Noguiera Reviewed-by: Pedro Tammela Signed-off-by: Jamal Hadi Salim Signed-off-by: David S. Miller

net/sched: Retire rsvp classifier

2023-02-16T08:27:07+00:00

The rsvp classifier has served us well for about a quarter of a century but has has not been getting much maintenance attention due to lack of known users. Signed-off-by: Jamal Hadi Salim Acked-by: Jiri Pirko Signed-off-by: Paolo Abeni

net/sched: Retire tcindex classifier

2023-02-16T08:27:07+00:00

The tcindex classifier has served us well for about a quarter of a century but has not been getting much TLC due to lack of known users. Most recently it has become easy prey to syzkaller. For this reason, we are retiring it. Signed-off-by: Jamal Hadi Salim Acked-by: Jiri Pirko Signed-off-by: Paolo Abeni

net/sched: Retire dsmark qdisc

2023-02-16T08:27:06+00:00

The dsmark qdisc has served us well over the years for diffserv but has not been getting much attention due to other more popular approaches to do diffserv services. Most recently it has become a shooting target for syzkaller. For this reason, we are retiring it. Signed-off-by: Jamal Hadi Salim Acked-by: Jiri Pirko Signed-off-by: Paolo Abeni

net/sched: Retire ATM qdisc

2023-02-16T08:27:06+00:00

The ATM qdisc has served us well over the years but has not been getting much TLC due to lack of known users. Most recently it has become a shooting target for syzkaller. For this reason, we are retiring it. Signed-off-by: Jamal Hadi Salim Acked-by: Jiri Pirko Signed-off-by: Paolo Abeni

net/sched: Retire CBQ qdisc

2023-02-16T08:27:06+00:00

While this amazing qdisc has served us well over the years it has not been getting any tender love and care and has bitrotted over time. It has become mostly a shooting target for syzkaller lately. For this reason, we are retiring it. Goodbye CBQ - we loved you. Signed-off-by: Jamal Hadi Salim Acked-by: Jiri Pirko Signed-off-by: Paolo Abeni

net/sched: taprio: centralize mqprio qopt validation

2023-02-06T10:06:44+00:00

There is a lot of code in taprio which is "borrowed" from mqprio. It makes sense to put a stop to the "borrowing" and start actually reusing code. Because taprio and mqprio are built as part of different kernel modules, code reuse can only take place either by writing it as static inline (limiting), putting it in sch_generic.o (not generic enough), or creating a third auto-selectable kernel module which only holds library code. I opted for the third variant. In a previous change, mqprio gained support for reverse TC:TXQ mappings, something which taprio still denies. Make taprio use the same validation logic so that it supports this configuration as well. The taprio code didn't enforce TXQ overlaps in txtime-assist mode and that looks intentional, even if I've no idea why that might be. Preserve that, but add a comment. There isn't any dedicated MAINTAINERS entry for mqprio, so nothing to update there. Signed-off-by: Vladimir Oltean Reviewed-by: Simon Horman Reviewed-by: Gerhard Engleder Signed-off-by: David S. Miller

net/sched: sch_frag: add generic packet fragment support.

2020-11-27T22:36:02+00:00

Currently kernel tc subsystem can do conntrack in cat_ct. But when several fragment packets go through the act_ct, function tcf_ct_handle_fragments will defrag the packets to a big one. But the last action will redirect mirred to a device which maybe lead the reassembly big packet over the mtu of target device. This patch add support for a xmit hook to mirred, that gets executed before xmiting the packet. Then, when act_ct gets loaded, it configs that hook. The frag xmit hook maybe reused by other modules. Signed-off-by: wenxu Acked-by: Cong Wang Acked-by: Jamal Hadi Salim Signed-off-by: Jakub Kicinski