net_sched: gen_estimator: complete rewrite of rate estimators
1) Old code was hard to maintain, due to complex lock chains.
(We probably will be able to remove some kfree_rcu() in callers)
2) Using a single timer to update all estimators does not scale.
3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity
is not supposed to work well)
In this rewrite :
- I removed the RB tree that had to be scanned in
gen_estimator_active(). qdisc dumps should be much faster.
- Each estimator has its own timer.
- Estimations are maintained in net_rate_estimator structure,
instead of dirtying the qdisc. Minor, but part of the simplification.
- Reading the estimator uses RCU and a seqcount to provide proper
support for 32bit kernels.
- We reduce memory need when estimators are not used, since
we store a pointer, instead of the bytes/packets counters.
- xt_rateest_mt() no longer has to grab a spinlock.
(In the future, xt_rateest_tg() could be switched to per cpu counters)
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
diff --git a/net/sched/act_api.c b/net/sched/act_api.c
index f893d18..2095c83 100644
--- a/net/sched/act_api.c
+++ b/net/sched/act_api.c
@@ -41,8 +41,7 @@
spin_lock_bh(&hinfo->lock);
hlist_del(&p->tcfa_head);
spin_unlock_bh(&hinfo->lock);
- gen_kill_estimator(&p->tcfa_bstats,
- &p->tcfa_rate_est);
+ gen_kill_estimator(&p->tcfa_rate_est);
/*
* gen_estimator est_timer() might access p->tcfa_lock
* or bstats, wait a RCU grace period before freeing p
@@ -237,8 +236,7 @@
void tcf_hash_cleanup(struct tc_action *a, struct nlattr *est)
{
if (est)
- gen_kill_estimator(&a->tcfa_bstats,
- &a->tcfa_rate_est);
+ gen_kill_estimator(&a->tcfa_rate_est);
call_rcu(&a->tcfa_rcu, free_tcf);
}
EXPORT_SYMBOL(tcf_hash_cleanup);
@@ -670,8 +668,7 @@
goto errout;
if (gnet_stats_copy_basic(NULL, &d, p->cpu_bstats, &p->tcfa_bstats) < 0 ||
- gnet_stats_copy_rate_est(&d, &p->tcfa_bstats,
- &p->tcfa_rate_est) < 0 ||
+ gnet_stats_copy_rate_est(&d, &p->tcfa_rate_est) < 0 ||
gnet_stats_copy_queue(&d, p->cpu_qstats,
&p->tcfa_qstats,
p->tcfa_qstats.qlen) < 0)
diff --git a/net/sched/act_police.c b/net/sched/act_police.c
index c990b73..0ba91d1 100644
--- a/net/sched/act_police.c
+++ b/net/sched/act_police.c
@@ -142,8 +142,7 @@
goto failure_unlock;
} else if (tb[TCA_POLICE_AVRATE] &&
(ret == ACT_P_CREATED ||
- !gen_estimator_active(&police->tcf_bstats,
- &police->tcf_rate_est))) {
+ !gen_estimator_active(&police->tcf_rate_est))) {
err = -EINVAL;
goto failure_unlock;
}
@@ -216,13 +215,17 @@
bstats_update(&police->tcf_bstats, skb);
tcf_lastuse_update(&police->tcf_tm);
- if (police->tcfp_ewma_rate &&
- police->tcf_rate_est.bps >= police->tcfp_ewma_rate) {
- police->tcf_qstats.overlimits++;
- if (police->tcf_action == TC_ACT_SHOT)
- police->tcf_qstats.drops++;
- spin_unlock(&police->tcf_lock);
- return police->tcf_action;
+ if (police->tcfp_ewma_rate) {
+ struct gnet_stats_rate_est64 sample;
+
+ if (!gen_estimator_read(&police->tcf_rate_est, &sample) ||
+ sample.bps >= police->tcfp_ewma_rate) {
+ police->tcf_qstats.overlimits++;
+ if (police->tcf_action == TC_ACT_SHOT)
+ police->tcf_qstats.drops++;
+ spin_unlock(&police->tcf_lock);
+ return police->tcf_action;
+ }
}
if (qdisc_pkt_len(skb) <= police->tcfp_mtu) {
diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index f337f1b..d7b9342 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c
@@ -1395,7 +1395,7 @@
if (gnet_stats_copy_basic(qdisc_root_sleeping_running(q),
&d, cpu_bstats, &q->bstats) < 0 ||
- gnet_stats_copy_rate_est(&d, &q->bstats, &q->rate_est) < 0 ||
+ gnet_stats_copy_rate_est(&d, &q->rate_est) < 0 ||
gnet_stats_copy_queue(&d, cpu_qstats, &q->qstats, qlen) < 0)
goto nla_put_failure;
diff --git a/net/sched/sch_cbq.c b/net/sched/sch_cbq.c
index beb554a..9ffe1c2 100644
--- a/net/sched/sch_cbq.c
+++ b/net/sched/sch_cbq.c
@@ -122,7 +122,7 @@
psched_time_t penalized;
struct gnet_stats_basic_packed bstats;
struct gnet_stats_queue qstats;
- struct gnet_stats_rate_est64 rate_est;
+ struct net_rate_estimator __rcu *rate_est;
struct tc_cbq_xstats xstats;
struct tcf_proto __rcu *filter_list;
@@ -1346,7 +1346,7 @@
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_rate_est(d, &cl->rate_est) < 0 ||
gnet_stats_copy_queue(d, NULL, &cl->qstats, cl->q->q.qlen) < 0)
return -1;
@@ -1405,7 +1405,7 @@
tcf_destroy_chain(&cl->filter_list);
qdisc_destroy(cl->q);
qdisc_put_rtab(cl->R_tab);
- gen_kill_estimator(&cl->bstats, &cl->rate_est);
+ gen_kill_estimator(&cl->rate_est);
if (cl != &q->link)
kfree(cl);
}
diff --git a/net/sched/sch_drr.c b/net/sched/sch_drr.c
index 8af5c59..bb4cbdf 100644
--- a/net/sched/sch_drr.c
+++ b/net/sched/sch_drr.c
@@ -25,7 +25,7 @@
struct gnet_stats_basic_packed bstats;
struct gnet_stats_queue qstats;
- struct gnet_stats_rate_est64 rate_est;
+ struct net_rate_estimator __rcu *rate_est;
struct list_head alist;
struct Qdisc *qdisc;
@@ -142,7 +142,7 @@
static void drr_destroy_class(struct Qdisc *sch, struct drr_class *cl)
{
- gen_kill_estimator(&cl->bstats, &cl->rate_est);
+ gen_kill_estimator(&cl->rate_est);
qdisc_destroy(cl->qdisc);
kfree(cl);
}
@@ -283,7 +283,7 @@
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_rate_est(d, &cl->rate_est) < 0 ||
gnet_stats_copy_queue(d, NULL, &cl->qdisc->qstats, qlen) < 0)
return -1;
diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
index 6cfb6e9..6eb9c8e 100644
--- a/net/sched/sch_generic.c
+++ b/net/sched/sch_generic.c
@@ -709,7 +709,7 @@
qdisc_put_stab(rtnl_dereference(qdisc->stab));
#endif
- gen_kill_estimator(&qdisc->bstats, &qdisc->rate_est);
+ gen_kill_estimator(&qdisc->rate_est);
if (ops->reset)
ops->reset(qdisc);
if (ops->destroy)
diff --git a/net/sched/sch_hfsc.c b/net/sched/sch_hfsc.c
index 000f1d3..3ffaa6f 100644
--- a/net/sched/sch_hfsc.c
+++ b/net/sched/sch_hfsc.c
@@ -114,7 +114,7 @@
struct gnet_stats_basic_packed bstats;
struct gnet_stats_queue qstats;
- struct gnet_stats_rate_est64 rate_est;
+ struct net_rate_estimator __rcu *rate_est;
struct tcf_proto __rcu *filter_list; /* filter list */
unsigned int filter_cnt; /* filter count */
unsigned int level; /* class level in hierarchy */
@@ -1091,7 +1091,7 @@
tcf_destroy_chain(&cl->filter_list);
qdisc_destroy(cl->qdisc);
- gen_kill_estimator(&cl->bstats, &cl->rate_est);
+ gen_kill_estimator(&cl->rate_est);
if (cl != &q->root)
kfree(cl);
}
@@ -1348,7 +1348,7 @@
xstats.rtwork = cl->cl_cumul;
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_rate_est(d, &cl->rate_est) < 0 ||
gnet_stats_copy_queue(d, NULL, &cl->qstats, cl->qdisc->q.qlen) < 0)
return -1;
diff --git a/net/sched/sch_htb.c b/net/sched/sch_htb.c
index 9926fe4..760f39e 100644
--- a/net/sched/sch_htb.c
+++ b/net/sched/sch_htb.c
@@ -111,7 +111,7 @@
unsigned int children;
struct htb_class *parent; /* parent class */
- struct gnet_stats_rate_est64 rate_est;
+ struct net_rate_estimator __rcu *rate_est;
/*
* Written often fields
@@ -1145,7 +1145,7 @@
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_rate_est(d, &cl->rate_est) < 0 ||
gnet_stats_copy_queue(d, NULL, &qs, qlen) < 0)
return -1;
@@ -1228,7 +1228,7 @@
WARN_ON(!cl->un.leaf.q);
qdisc_destroy(cl->un.leaf.q);
}
- gen_kill_estimator(&cl->bstats, &cl->rate_est);
+ gen_kill_estimator(&cl->rate_est);
tcf_destroy_chain(&cl->filter_list);
kfree(cl);
}
diff --git a/net/sched/sch_qfq.c b/net/sched/sch_qfq.c
index ca0516e..f9e712c 100644
--- a/net/sched/sch_qfq.c
+++ b/net/sched/sch_qfq.c
@@ -137,7 +137,7 @@
struct gnet_stats_basic_packed bstats;
struct gnet_stats_queue qstats;
- struct gnet_stats_rate_est64 rate_est;
+ struct net_rate_estimator __rcu *rate_est;
struct Qdisc *qdisc;
struct list_head alist; /* Link for active-classes list. */
struct qfq_aggregate *agg; /* Parent aggregate. */
@@ -508,7 +508,7 @@
new_agg = kzalloc(sizeof(*new_agg), GFP_KERNEL);
if (new_agg == NULL) {
err = -ENOBUFS;
- gen_kill_estimator(&cl->bstats, &cl->rate_est);
+ gen_kill_estimator(&cl->rate_est);
goto destroy_class;
}
sch_tree_lock(sch);
@@ -533,7 +533,7 @@
struct qfq_sched *q = qdisc_priv(sch);
qfq_rm_from_agg(q, cl);
- gen_kill_estimator(&cl->bstats, &cl->rate_est);
+ gen_kill_estimator(&cl->rate_est);
qdisc_destroy(cl->qdisc);
kfree(cl);
}
@@ -667,7 +667,7 @@
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_rate_est(d, &cl->rate_est) < 0 ||
gnet_stats_copy_queue(d, NULL,
&cl->qdisc->qstats, cl->qdisc->q.qlen) < 0)
return -1;