Eric Dumazet edb09eb17e net: sched: do not acquire qdisc spinlock in qdisc/class stats dump ... Large tc dumps (tc -s {qdisc|class} sh dev ethX) done by Google BwE host agent [1] are problematic at scale : For each qdisc/class found in the dump, we currently lock the root qdisc spinlock in order to get stats. Sampling stats every 5 seconds from thousands of HTB classes is a challenge when the root qdisc spinlock is under high pressure. Not only the dumps take time, they also slow down the fast path (queue/dequeue packets) by 10 % to 20 % in some cases. An audit of existing qdiscs showed that sch_fq_codel is the only qdisc that might need the qdisc lock in fq_codel_dump_stats() and fq_codel_dump_class_stats() In v2 of this patch, I now use the Qdisc running seqcount to provide consistent reads of packets/bytes counters, regardless of 32/64 bit arches. I also changed rate estimators to use the same infrastructure so that they no longer need to lock root qdisc lock. [1] http://static.googleusercontent.com/media/research.google.com/en//pubs/archive/43838.pdf Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Cong Wang <xiyou.wangcong@gmail.com> Cc: Jamal Hadi Salim <jhs@mojatatu.com> Cc: John Fastabend <john.fastabend@gmail.com> Cc: Kevin Athey <kda@google.com> Cc: Xiaotian Pei <xiaotian@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>		2016-06-07 16:37:14 -07:00
..
act_api.c	net: sched: do not acquire qdisc spinlock in qdisc/class stats dump	2016-06-07 16:37:14 -07:00
act_bpf.c	net sched: indentation and other OCD stylistic fixes	2016-06-07 15:53:54 -07:00
act_connmark.c	net sched actions: aggregate dumping of actions timeinfo	2016-06-07 15:53:43 -07:00
act_csum.c	net sched actions: aggregate dumping of actions timeinfo	2016-06-07 15:53:43 -07:00
act_gact.c	net sched: indentation and other OCD stylistic fixes	2016-06-07 15:53:54 -07:00
act_ife.c	net sched actions: aggregate dumping of actions timeinfo	2016-06-07 15:53:43 -07:00
act_ipt.c	net sched: indentation and other OCD stylistic fixes	2016-06-07 15:53:54 -07:00
act_meta_mark.c	Support to encoding decoding skb mark on IFE action	2016-03-01 17:15:23 -05:00
act_meta_skbprio.c	Support to encoding decoding skb prio on IFE action	2016-03-01 17:15:23 -05:00
act_mirred.c	net sched actions: aggregate dumping of actions timeinfo	2016-06-07 15:53:43 -07:00
act_nat.c	net sched actions: aggregate dumping of actions timeinfo	2016-06-07 15:53:43 -07:00
act_pedit.c	net sched actions: aggregate dumping of actions timeinfo	2016-06-07 15:53:43 -07:00
act_police.c	net: sched: do not acquire qdisc spinlock in qdisc/class stats dump	2016-06-07 16:37:14 -07:00
act_simple.c	net sched actions: aggregate dumping of actions timeinfo	2016-06-07 15:53:43 -07:00
act_skbedit.c	net sched actions: aggregate dumping of actions timeinfo	2016-06-07 15:53:43 -07:00
act_vlan.c	net sched: indentation and other OCD stylistic fixes	2016-06-07 15:53:54 -07:00
cls_api.c	net sched: indentation and other OCD stylistic fixes	2016-06-07 15:53:54 -07:00
cls_basic.c
cls_bpf.c	bpf: wire in data and data_end for cls_act_bpf	2016-05-06 16:01:54 -04:00
cls_cgroup.c
cls_flow.c	sched: cls_flow: use skb_to_full_sk() helper	2015-11-08 20:56:39 -05:00
cls_flower.c	net/sched: cls_flower: Introduce support in SKIP SW flag	2016-06-07 15:49:53 -07:00
cls_fw.c	net: revert "net_sched: move tp->root allocation into fw_init()"	2015-09-24 14:33:30 -07:00
cls_route.c
cls_rsvp6.c
cls_rsvp.c
cls_rsvp.h	net_sched: convert rsvp to call tcf_exts_destroy from rcu callback	2015-08-26 11:01:45 -07:00
cls_tcindex.c	net_sched: convert tcindex to call tcf_exts_destroy from rcu callback	2015-08-26 11:01:44 -07:00
cls_u32.c	net: cls_u32: Add support for skip-sw flag to tc u32 classifier.	2016-05-16 13:30:57 -04:00
em_canid.c
em_cmp.c
em_ipset.c	netfilter: x_tables: Pass struct net in xt_action_param	2015-09-18 21:58:14 +02:00
em_meta.c	qdisc: constify meta_type_ops structures	2016-04-14 00:35:30 -04:00
em_nbyte.c
em_text.c
em_u32.c
ematch.c
Kconfig	Support to encoding decoding skb prio on IFE action	2016-03-01 17:15:23 -05:00
Makefile	Support to encoding decoding skb prio on IFE action	2016-03-01 17:15:23 -05:00
sch_api.c	net: sched: do not acquire qdisc spinlock in qdisc/class stats dump	2016-06-07 16:37:14 -07:00
sch_atm.c	net: sched: do not acquire qdisc spinlock in qdisc/class stats dump	2016-06-07 16:37:14 -07:00
sch_blackhole.c	net/sched: make sch_blackhole.c explicitly non-modular	2015-10-09 07:52:28 -07:00
sch_cbq.c	net: sched: do not acquire qdisc spinlock in qdisc/class stats dump	2016-06-07 16:37:14 -07:00
sch_choke.c	net_sched: update hierarchical backlog too	2016-02-29 17:02:33 -05:00
sch_codel.c	codel: split into multiple files	2016-04-25 16:44:27 -04:00
sch_drr.c	net: sched: do not acquire qdisc spinlock in qdisc/class stats dump	2016-06-07 16:37:14 -07:00
sch_dsmark.c	net_sched: dsmark: use qdisc_dequeue_peeked()	2016-03-08 14:35:13 -05:00
sch_fifo.c
sch_fq_codel.c	net: sched: do not acquire qdisc spinlock in qdisc/class stats dump	2016-06-07 16:37:14 -07:00
sch_fq.c	net_sched: update hierarchical backlog too	2016-02-29 17:02:33 -05:00
sch_generic.c	net_sched: transform qdisc running bit into a seqcount	2016-06-07 16:37:13 -07:00
sch_gred.c
sch_hfsc.c	net: sched: do not acquire qdisc spinlock in qdisc/class stats dump	2016-06-07 16:37:14 -07:00
sch_hhf.c	net_sched: update hierarchical backlog too	2016-02-29 17:02:33 -05:00
sch_htb.c	net: sched: do not acquire qdisc spinlock in qdisc/class stats dump	2016-06-07 16:37:14 -07:00
sch_ingress.c	net, sched: add clsact qdisc	2016-01-10 22:13:15 -05:00
sch_mq.c	net: sched: do not acquire qdisc spinlock in qdisc/class stats dump	2016-06-07 16:37:14 -07:00
sch_mqprio.c	net: sched: do not acquire qdisc spinlock in qdisc/class stats dump	2016-06-07 16:37:14 -07:00
sch_multiq.c	net: sched: do not acquire qdisc spinlock in qdisc/class stats dump	2016-06-07 16:37:14 -07:00
sch_netem.c	Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net	2016-05-04 00:52:29 -04:00
sch_pie.c	net_sched: update hierarchical backlog too	2016-02-29 17:02:33 -05:00
sch_plug.c
sch_prio.c	net: sched: do not acquire qdisc spinlock in qdisc/class stats dump	2016-06-07 16:37:14 -07:00
sch_qfq.c	net: sched: do not acquire qdisc spinlock in qdisc/class stats dump	2016-06-07 16:37:14 -07:00
sch_red.c	net_sched: update hierarchical backlog too	2016-02-29 17:02:33 -05:00
sch_sfb.c	net_sched: update hierarchical backlog too	2016-02-29 17:02:33 -05:00
sch_sfq.c	net_sched: update hierarchical backlog too	2016-02-29 17:02:33 -05:00
sch_tbf.c	sched: use nla_put_u64_64bit()	2016-04-25 15:09:09 -04:00
sch_teql.c