summaryrefslogtreecommitdiffstats
path: root/net/sched/sch_mqprio.c
diff options
context:
space:
mode:
authorEric Dumazet <edumazet@google.com>2016-06-06 09:37:16 -0700
committerDavid S. Miller <davem@davemloft.net>2016-06-07 16:37:14 -0700
commitedb09eb17ed89eaa82a52dd306beac93e292b485 (patch)
tree1f241506d6b781b65d1033925f1c1ce6a39c3394 /net/sched/sch_mqprio.c
parentf9eb8aea2a1e12fc2f584d1627deeb957435a801 (diff)
downloadop-kernel-dev-edb09eb17ed89eaa82a52dd306beac93e292b485.zip
op-kernel-dev-edb09eb17ed89eaa82a52dd306beac93e292b485.tar.gz
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_mqprio.c')
-rw-r--r--net/sched/sch_mqprio.c11
1 files changed, 7 insertions, 4 deletions
diff --git a/net/sched/sch_mqprio.c b/net/sched/sch_mqprio.c
index b8002ce..549c663 100644
--- a/net/sched/sch_mqprio.c
+++ b/net/sched/sch_mqprio.c
@@ -342,7 +342,8 @@ static int mqprio_dump_class_stats(struct Qdisc *sch, unsigned long cl,
* hold here is the look on dev_queue->qdisc_sleeping
* also acquired below.
*/
- spin_unlock_bh(d->lock);
+ if (d->lock)
+ spin_unlock_bh(d->lock);
for (i = tc.offset; i < tc.offset + tc.count; i++) {
struct netdev_queue *q = netdev_get_tx_queue(dev, i);
@@ -359,15 +360,17 @@ static int mqprio_dump_class_stats(struct Qdisc *sch, unsigned long cl,
spin_unlock_bh(qdisc_lock(qdisc));
}
/* Reclaim root sleeping lock before completing stats */
- spin_lock_bh(d->lock);
- if (gnet_stats_copy_basic(d, NULL, &bstats) < 0 ||
+ if (d->lock)
+ spin_lock_bh(d->lock);
+ if (gnet_stats_copy_basic(NULL, d, NULL, &bstats) < 0 ||
gnet_stats_copy_queue(d, NULL, &qstats, qlen) < 0)
return -1;
} else {
struct netdev_queue *dev_queue = mqprio_queue_get(sch, cl);
sch = dev_queue->qdisc_sleeping;
- if (gnet_stats_copy_basic(d, NULL, &sch->bstats) < 0 ||
+ if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch),
+ d, NULL, &sch->bstats) < 0 ||
gnet_stats_copy_queue(d, NULL,
&sch->qstats, sch->q.qlen) < 0)
return -1;
OpenPOWER on IntegriCloud