net: sched: when clearing NOLOCK, clear TCQ_F_CPUSTATS, too

Since stats updating is always consistent with TCQ_F_CPUSTATS flag,
we can disable it at qdisc creation time flipping such bit.

In my experiments, if the NOLOCK flag is cleared, per CPU stats
accounting does not give any measurable performance gain, but it
waste some memory.

Let's clear TCQ_F_CPUSTATS together with NOLOCK, when enslaving
a NOLOCK qdisc to 'lock' one.

Use stats update helper inside pfifo_fast, to cope correctly with
TCQ_F_CPUSTATS flag change.

As a side effect, q.qlen value for any child qdiscs is always
consistent for all lock classfull qdiscs.

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 2 insertions, 8 deletions

diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
index ddff2952be87..12a6e1a39fa0 100644
--- a/net/sched/sch_generic.c
+++ b/net/sched/sch_generic.c
@@ -629,11 +629,7 @@ static int pfifo_fast_enqueue(struct sk_buff *skb, struct Qdisc *qdisc,
 	if (unlikely(err))
 		return qdisc_drop_cpu(skb, qdisc, to_free);
 
-	qdisc_qstats_atomic_qlen_inc(qdisc);
-	/* Note: skb can not be used after skb_array_produce(),
-	 * so we better not use qdisc_qstats_cpu_backlog_inc()
-	 */
-	this_cpu_add(qdisc->cpu_qstats->backlog, pkt_len);
+	qdisc_update_stats_at_enqueue(qdisc, pkt_len);
 	return NET_XMIT_SUCCESS;
 }
 
@@ -652,9 +648,7 @@ static struct sk_buff *pfifo_fast_dequeue(struct Qdisc *qdisc)
 		skb = __skb_array_consume(q);
 	}
 	if (likely(skb)) {
-		qdisc_qstats_cpu_backlog_dec(qdisc, skb);
-		qdisc_bstats_cpu_update(qdisc, skb);
-		qdisc_qstats_atomic_qlen_dec(qdisc);
+		qdisc_update_stats_at_dequeue(qdisc, skb);
 	} else {
 		qdisc->empty = true;
 	}