diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h
index 7d1a0483a17b..43e4e17aa938 100644
--- a/include/net/sch_generic.h
+++ b/include/net/sch_generic.h
@@ -923,6 +923,24 @@ static inline void qdisc_qstats_overlimit(struct Qdisc *sch)
 	sch->qstats.overlimits++;
 }
 
+static inline int qdisc_qstats_copy(struct gnet_dump *d, struct Qdisc *sch)
+{
+	__u32 qlen = qdisc_qlen_sum(sch);
+
+	return gnet_stats_copy_queue(d, sch->cpu_qstats, &sch->qstats, qlen);
+}
+
+static inline void qdisc_qstats_qlen_backlog(struct Qdisc *sch,  __u32 *qlen,
+					     __u32 *backlog)
+{
+	struct gnet_stats_queue qstats = { 0 };
+	__u32 len = qdisc_qlen_sum(sch);
+
+	__gnet_stats_copy_queue(&qstats, sch->cpu_qstats, &sch->qstats, len);
+	*qlen = qstats.qlen;
+	*backlog = qstats.backlog;
+}
+
 static inline void qdisc_skb_head_init(struct qdisc_skb_head *qh)
 {
 	qh->head = NULL;
diff --git a/net/sched/sch_cbq.c b/net/sched/sch_cbq.c
index 4dc05409e3fb..651879c1b655 100644
--- a/net/sched/sch_cbq.c
+++ b/net/sched/sch_cbq.c
@@ -1358,9 +1358,11 @@ cbq_dump_class_stats(struct Qdisc *sch, unsigned long arg,
 {
 	struct cbq_sched_data *q = qdisc_priv(sch);
 	struct cbq_class *cl = (struct cbq_class *)arg;
+	__u32 qlen;
 
 	cl->xstats.avgidle = cl->avgidle;
 	cl->xstats.undertime = 0;
+	qdisc_qstats_qlen_backlog(cl->q, &qlen, &cl->qstats.backlog);
 
 	if (cl->undertime != PSCHED_PASTPERFECT)
 		cl->xstats.undertime = cl->undertime - q->now;
@@ -1368,7 +1370,7 @@ cbq_dump_class_stats(struct Qdisc *sch, unsigned long arg,
 	if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch),
 				  d, NULL, &cl->bstats) < 0 ||
 	    gnet_stats_copy_rate_est(d, &cl->rate_est) < 0 ||
-	    gnet_stats_copy_queue(d, NULL, &cl->qstats, cl->q->q.qlen) < 0)
+	    gnet_stats_copy_queue(d, NULL, &cl->qstats, qlen) < 0)
 		return -1;
 
 	return gnet_stats_copy_app(d, &cl->xstats, sizeof(cl->xstats));
diff --git a/net/sched/sch_drr.c b/net/sched/sch_drr.c
index 09b800991065..8a181591b0ea 100644
--- a/net/sched/sch_drr.c
+++ b/net/sched/sch_drr.c
@@ -269,7 +269,8 @@ static int drr_dump_class_stats(struct Qdisc *sch, unsigned long arg,
 				struct gnet_dump *d)
 {
 	struct drr_class *cl = (struct drr_class *)arg;
-	__u32 qlen = cl->qdisc->q.qlen;
+	__u32 qlen = qdisc_qlen_sum(cl->qdisc);
+	struct Qdisc *cl_q = cl->qdisc;
 	struct tc_drr_stats xstats;
 
 	memset(&xstats, 0, sizeof(xstats));
@@ -279,7 +280,7 @@ static int drr_dump_class_stats(struct Qdisc *sch, unsigned long arg,
 	if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch),
 				  d, NULL, &cl->bstats) < 0 ||
 	    gnet_stats_copy_rate_est(d, &cl->rate_est) < 0 ||
-	    gnet_stats_copy_queue(d, NULL, &cl->qdisc->qstats, qlen) < 0)
+	    gnet_stats_copy_queue(d, cl_q->cpu_qstats, &cl_q->qstats, qlen) < 0)
 		return -1;
 
 	return gnet_stats_copy_app(d, &xstats, sizeof(xstats));
diff --git a/net/sched/sch_hfsc.c b/net/sched/sch_hfsc.c
index 24cc220a3218..a946a419d717 100644
--- a/net/sched/sch_hfsc.c
+++ b/net/sched/sch_hfsc.c
@@ -1328,8 +1328,9 @@ hfsc_dump_class_stats(struct Qdisc *sch, unsigned long arg,
 {
 	struct hfsc_class *cl = (struct hfsc_class *)arg;
 	struct tc_hfsc_stats xstats;
+	__u32 qlen;
 
-	cl->qstats.backlog = cl->qdisc->qstats.backlog;
+	qdisc_qstats_qlen_backlog(cl->qdisc, &qlen, &cl->qstats.backlog);
 	xstats.level   = cl->level;
 	xstats.period  = cl->cl_vtperiod;
 	xstats.work    = cl->cl_total;
@@ -1337,7 +1338,7 @@ hfsc_dump_class_stats(struct Qdisc *sch, unsigned long arg,
 
 	if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch), d, NULL, &cl->bstats) < 0 ||
 	    gnet_stats_copy_rate_est(d, &cl->rate_est) < 0 ||
-	    gnet_stats_copy_queue(d, NULL, &cl->qstats, cl->qdisc->q.qlen) < 0)
+	    gnet_stats_copy_queue(d, NULL, &cl->qstats, qlen) < 0)
 		return -1;
 
 	return gnet_stats_copy_app(d, &xstats, sizeof(xstats));
diff --git a/net/sched/sch_htb.c b/net/sched/sch_htb.c
index 30f9da7e1076..ed92836f528a 100644
--- a/net/sched/sch_htb.c
+++ b/net/sched/sch_htb.c
@@ -1127,10 +1127,9 @@ htb_dump_class_stats(struct Qdisc *sch, unsigned long arg, struct gnet_dump *d)
 	};
 	__u32 qlen = 0;
 
-	if (!cl->level && cl->leaf.q) {
-		qlen = cl->leaf.q->q.qlen;
-		qs.backlog = cl->leaf.q->qstats.backlog;
-	}
+	if (!cl->level && cl->leaf.q)
+		qdisc_qstats_qlen_backlog(cl->leaf.q, &qlen, &qs.backlog);
+
 	cl->xstats.tokens = clamp_t(s64, PSCHED_NS2TICKS(cl->tokens),
 				    INT_MIN, INT_MAX);
 	cl->xstats.ctokens = clamp_t(s64, PSCHED_NS2TICKS(cl->ctokens),
diff --git a/net/sched/sch_mq.c b/net/sched/sch_mq.c
index 203659bc3906..3a3312467692 100644
--- a/net/sched/sch_mq.c
+++ b/net/sched/sch_mq.c
@@ -249,7 +249,7 @@ static int mq_dump_class_stats(struct Qdisc *sch, unsigned long cl,
 
 	sch = dev_queue->qdisc_sleeping;
 	if (gnet_stats_copy_basic(&sch->running, d, NULL, &sch->bstats) < 0 ||
-	    gnet_stats_copy_queue(d, NULL, &sch->qstats, sch->q.qlen) < 0)
+	    qdisc_qstats_copy(d, sch) < 0)
 		return -1;
 	return 0;
 }
diff --git a/net/sched/sch_mqprio.c b/net/sched/sch_mqprio.c
index d364e63c396d..ea0dc112b38d 100644
--- a/net/sched/sch_mqprio.c
+++ b/net/sched/sch_mqprio.c
@@ -561,8 +561,7 @@ static int mqprio_dump_class_stats(struct Qdisc *sch, unsigned long cl,
 		sch = dev_queue->qdisc_sleeping;
 		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)
+		    qdisc_qstats_copy(d, sch) < 0)
 			return -1;
 	}
 	return 0;
diff --git a/net/sched/sch_multiq.c b/net/sched/sch_multiq.c
index 7410ce4d0321..53c918a11378 100644
--- a/net/sched/sch_multiq.c
+++ b/net/sched/sch_multiq.c
@@ -344,7 +344,7 @@ static int multiq_dump_class_stats(struct Qdisc *sch, unsigned long cl,
 	cl_q = q->queues[cl - 1];
 	if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch),
 				  d, NULL, &cl_q->bstats) < 0 ||
-	    gnet_stats_copy_queue(d, NULL, &cl_q->qstats, cl_q->q.qlen) < 0)
+	    qdisc_qstats_copy(d, cl_q) < 0)
 		return -1;
 
 	return 0;
diff --git a/net/sched/sch_prio.c b/net/sched/sch_prio.c
index 847141cd900f..dfb06d5bfacc 100644
--- a/net/sched/sch_prio.c
+++ b/net/sched/sch_prio.c
@@ -365,7 +365,7 @@ static int prio_dump_class_stats(struct Qdisc *sch, unsigned long cl,
 	cl_q = q->queues[cl - 1];
 	if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch),
 				  d, NULL, &cl_q->bstats) < 0 ||
-	    gnet_stats_copy_queue(d, NULL, &cl_q->qstats, cl_q->q.qlen) < 0)
+	    qdisc_qstats_copy(d, cl_q) < 0)
 		return -1;
 
 	return 0;
diff --git a/net/sched/sch_qfq.c b/net/sched/sch_qfq.c
index 29f5c4a24688..9fbda3ec5861 100644
--- a/net/sched/sch_qfq.c
+++ b/net/sched/sch_qfq.c
@@ -655,8 +655,7 @@ static int qfq_dump_class_stats(struct Qdisc *sch, unsigned long arg,
 	if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch),
 				  d, NULL, &cl->bstats) < 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)
+	    qdisc_qstats_copy(d, cl->qdisc) < 0)
 		return -1;
 
 	return gnet_stats_copy_app(d, &xstats, sizeof(xstats));
diff --git a/net/sched/sch_taprio.c b/net/sched/sch_taprio.c
index 206e4dbed12f..c7041999eb5d 100644
--- a/net/sched/sch_taprio.c
+++ b/net/sched/sch_taprio.c
@@ -895,7 +895,7 @@ static int taprio_dump_class_stats(struct Qdisc *sch, unsigned long cl,
 
 	sch = dev_queue->qdisc_sleeping;
 	if (gnet_stats_copy_basic(&sch->running, d, NULL, &sch->bstats) < 0 ||
-	    gnet_stats_copy_queue(d, NULL, &sch->qstats, sch->q.qlen) < 0)
+	    qdisc_qstats_copy(d, sch) < 0)
 		return -1;
 	return 0;
 }