diff options
| author |   Eric Dumazet <edumazet@google.com> | 2016-06-06 09:37:16 -0700 |
|---|---|---|
| committer |   David S. Miller <davem@davemloft.net> | 2016-06-07 16:37:14 -0700 |
| commit | edb09eb17ed89eaa82a52dd306beac93e292b485 (patch) | |
| tree | 1f241506d6b781b65d1033925f1c1ce6a39c3394 /net/sched/sch_cbq.c | |
| parent | net_sched: transform qdisc running bit into a seqcount (diff) | |
| download | linux-dev-edb09eb17ed89eaa82a52dd306beac93e292b485.tar.xz linux-dev-edb09eb17ed89eaa82a52dd306beac93e292b485.zip | |
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>
Diffstat (limited to 'net/sched/sch_cbq.c')
| -rw-r--r-- | net/sched/sch_cbq.c | 9 |
1 files changed, 6 insertions, 3 deletions
diff --git a/net/sched/sch_cbq.c b/net/sched/sch_cbq.c index baafddf229ce..1b8128fb845d 100644 --- a/net/sched/sch_cbq.c +++ b/net/sched/sch_cbq.c @@ -1600,7 +1600,8 @@ cbq_dump_class_stats(struct Qdisc *sch, unsigned long arg, if (cl->undertime != PSCHED_PASTPERFECT) cl->xstats.undertime = cl->undertime - q->now; - if (gnet_stats_copy_basic(d, NULL, &cl->bstats) < 0 || + if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch), + d, NULL, &cl->bstats) < 0 || gnet_stats_copy_rate_est(d, &cl->bstats, &cl->rate_est) < 0 || gnet_stats_copy_queue(d, NULL, &cl->qstats, cl->q->q.qlen) < 0) return -1; @@ -1755,7 +1756,8 @@ cbq_change_class(struct Qdisc *sch, u32 classid, u32 parentid, struct nlattr **t if (tca[TCA_RATE]) { err = gen_replace_estimator(&cl->bstats, NULL, &cl->rate_est, - qdisc_root_sleeping_lock(sch), + NULL, + qdisc_root_sleeping_running(sch), tca[TCA_RATE]); if (err) { qdisc_put_rtab(rtab); @@ -1848,7 +1850,8 @@ cbq_change_class(struct Qdisc *sch, u32 classid, u32 parentid, struct nlattr **t if (tca[TCA_RATE]) { err = gen_new_estimator(&cl->bstats, NULL, &cl->rate_est, - qdisc_root_sleeping_lock(sch), + NULL, + qdisc_root_sleeping_running(sch), tca[TCA_RATE]); if (err) { kfree(cl); |