net_sched: sch_fq: defer skb freeing
authorEric Dumazet <edumazet@google.com>
Tue, 14 Jun 2016 03:21:53 +0000 (20:21 -0700)
committerDavid S. Miller <davem@davemloft.net>
Wed, 15 Jun 2016 21:08:35 +0000 (14:08 -0700)
Both fq_change() and fq_reset() can use rtnl_kfree_skbs()

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/sched/sch_fq.c

index f49c81e..6eb0667 100644 (file)
@@ -514,17 +514,25 @@ out:
        return skb;
 }
 
+static void fq_flow_purge(struct fq_flow *flow)
+{
+       rtnl_kfree_skbs(flow->head, flow->tail);
+       flow->head = NULL;
+       flow->qlen = 0;
+}
+
 static void fq_reset(struct Qdisc *sch)
 {
        struct fq_sched_data *q = qdisc_priv(sch);
        struct rb_root *root;
-       struct sk_buff *skb;
        struct rb_node *p;
        struct fq_flow *f;
        unsigned int idx;
 
-       while ((skb = fq_dequeue_head(sch, &q->internal)) != NULL)
-               kfree_skb(skb);
+       sch->q.qlen = 0;
+       sch->qstats.backlog = 0;
+
+       fq_flow_purge(&q->internal);
 
        if (!q->fq_root)
                return;
@@ -535,8 +543,7 @@ static void fq_reset(struct Qdisc *sch)
                        f = container_of(p, struct fq_flow, fq_node);
                        rb_erase(p, root);
 
-                       while ((skb = fq_dequeue_head(sch, f)) != NULL)
-                               kfree_skb(skb);
+                       fq_flow_purge(f);
 
                        kmem_cache_free(fq_flow_cachep, f);
                }
@@ -737,7 +744,7 @@ static int fq_change(struct Qdisc *sch, struct nlattr *opt)
                if (!skb)
                        break;
                drop_len += qdisc_pkt_len(skb);
-               kfree_skb(skb);
+               rtnl_kfree_skbs(skb, skb);
                drop_count++;
        }
        qdisc_tree_reduce_backlog(sch, drop_count, drop_len);