net_sched: RCU conversion of stab
This patch converts stab qdisc management to RCU, so that we can perform
the qdisc_calculate_pkt_len() call before getting qdisc lock.
This shortens the lock's held time in __dev_xmit_skb().
This permits more qdiscs to get TCQ_F_CAN_BYPASS status, avoiding lot of
cache misses and so reducing latencies.
Signed-off-by: Eric Dumazet <[email protected]>
CC: Patrick McHardy <[email protected]>
CC: Jesper Dangaard Brouer <[email protected]>
CC: Jarek Poplawski <[email protected]>
CC: Jamal Hadi Salim <[email protected]>
CC: Stephen Hemminger <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index 374fcbe..1507415 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c
@@ -398,6 +398,11 @@
return stab;
}
+static void stab_kfree_rcu(struct rcu_head *head)
+{
+ kfree(container_of(head, struct qdisc_size_table, rcu));
+}
+
void qdisc_put_stab(struct qdisc_size_table *tab)
{
if (!tab)
@@ -407,7 +412,7 @@
if (--tab->refcnt == 0) {
list_del(&tab->list);
- kfree(tab);
+ call_rcu_bh(&tab->rcu, stab_kfree_rcu);
}
spin_unlock(&qdisc_stab_lock);
@@ -430,7 +435,7 @@
return -1;
}
-void qdisc_calculate_pkt_len(struct sk_buff *skb, struct qdisc_size_table *stab)
+void __qdisc_calculate_pkt_len(struct sk_buff *skb, const struct qdisc_size_table *stab)
{
int pkt_len, slot;
@@ -456,7 +461,7 @@
pkt_len = 1;
qdisc_skb_cb(skb)->pkt_len = pkt_len;
}
-EXPORT_SYMBOL(qdisc_calculate_pkt_len);
+EXPORT_SYMBOL(__qdisc_calculate_pkt_len);
void qdisc_warn_nonwc(char *txt, struct Qdisc *qdisc)
{
@@ -835,7 +840,7 @@
err = PTR_ERR(stab);
goto err_out4;
}
- sch->stab = stab;
+ rcu_assign_pointer(sch->stab, stab);
}
if (tca[TCA_RATE]) {
spinlock_t *root_lock;
@@ -875,7 +880,7 @@
* Any broken qdiscs that would require a ops->reset() here?
* The qdisc was never in action so it shouldn't be necessary.
*/
- qdisc_put_stab(sch->stab);
+ qdisc_put_stab(rtnl_dereference(sch->stab));
if (ops->destroy)
ops->destroy(sch);
goto err_out3;
@@ -883,7 +888,7 @@
static int qdisc_change(struct Qdisc *sch, struct nlattr **tca)
{
- struct qdisc_size_table *stab = NULL;
+ struct qdisc_size_table *ostab, *stab = NULL;
int err = 0;
if (tca[TCA_OPTIONS]) {
@@ -900,8 +905,9 @@
return PTR_ERR(stab);
}
- qdisc_put_stab(sch->stab);
- sch->stab = stab;
+ ostab = rtnl_dereference(sch->stab);
+ rcu_assign_pointer(sch->stab, stab);
+ qdisc_put_stab(ostab);
if (tca[TCA_RATE]) {
/* NB: ignores errors from replace_estimator
@@ -1180,6 +1186,7 @@
struct nlmsghdr *nlh;
unsigned char *b = skb_tail_pointer(skb);
struct gnet_dump d;
+ struct qdisc_size_table *stab;
nlh = NLMSG_NEW(skb, pid, seq, event, sizeof(*tcm), flags);
tcm = NLMSG_DATA(nlh);
@@ -1195,7 +1202,8 @@
goto nla_put_failure;
q->qstats.qlen = q->q.qlen;
- if (q->stab && qdisc_dump_stab(skb, q->stab) < 0)
+ stab = rtnl_dereference(q->stab);
+ if (stab && qdisc_dump_stab(skb, stab) < 0)
goto nla_put_failure;
if (gnet_stats_start_copy_compat(skb, TCA_STATS2, TCA_STATS, TCA_XSTATS,
