Merge branch 'remove-qdisc-running-counter'

Sebastian Andrzej Siewior says:

====================
Try to simplify the gnet_stats and remove qdisc->running sequence counter.

The first few patches is a follow up to
    https://lore.kernel.org/all/20211007175000.2334713-1-bigeasy@linutronix.de/

The remaining patches (#5+) remove the seqcount_t (Qdisc::running) from
the Qdisc. The statistics (Qdisc::bstats and Qdisc::cpu_bstats) use
u64_stats_t and the "running state" is now represented by a bit in
Qdisc::state.

By removing the seqcount_t from Qdisc and decoupling the bstats
statistics from the seqcount_t it is possible to query the statistics
even if the Qdisc is running instead of waiting until it is idle again.

The try-lock like usage of the seqcount_t in qdisc_run_begin() is
problematic on PREEMPT_RT. Inside the qdisc_run_begin/end() qdisc->running
sequence counter write sections, at sch_direct_xmit(), the seqcount write
serialization lock is released then re-acquired. This is fine for !RT, because
the writer is in a BH disabled region and there is a no in-IRQ reader. For RT
though, BH sections are preemptible. The earlier introduced seqcount_LOCKNAME_t
mechanism, which for RT the reader acquires then relesaes the write
serailization lock to avoid infinite spinning if it preempts a seqcount write
section, cannot work: the qdisc->running write serialization lock is already
intermittingly released inside the seqcount write section.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/netronome/nfp/abm/qdisc.c

@@ -458,7 +458,7 @@ nfp_abm_qdisc_graft(struct nfp_abm_link *alink, u32 handle, u32 child_handle,
 static void
 nfp_abm_stats_calculate(struct nfp_alink_stats *new,
 			struct nfp_alink_stats *old,
-			struct gnet_stats_basic_packed *bstats,
+			struct gnet_stats_basic_sync *bstats,
 			struct gnet_stats_queue *qstats)
 {
 	_bstats_update(bstats, new->tx_bytes - old->tx_bytes,

include/linux/netdevice.h

@@ -1916,7 +1916,6 @@ enum netdev_ml_priv_type {
  *	@sfp_bus:	attached &struct sfp_bus structure.
  *
  *	@qdisc_tx_busylock: lockdep class annotating Qdisc->busylock spinlock
- *	@qdisc_running_key: lockdep class annotating Qdisc->running seqcount
  *
  *	@proto_down:	protocol port state information can be sent to the
  *			switch driver and used to set the phys state of the
@@ -2250,7 +2249,6 @@ struct net_device {
 	struct phy_device	*phydev;
 	struct sfp_bus		*sfp_bus;
 	struct lock_class_key	*qdisc_tx_busylock;
-	struct lock_class_key	*qdisc_running_key;
 	bool			proto_down;
 	unsigned		wol_enabled:1;
 	unsigned		threaded:1;
@@ -2360,13 +2358,11 @@ static inline void netdev_for_each_tx_queue(struct net_device *dev,
 #define netdev_lockdep_set_classes(dev)				\
 {								\
 	static struct lock_class_key qdisc_tx_busylock_key;	\
-	static struct lock_class_key qdisc_running_key;		\
 	static struct lock_class_key qdisc_xmit_lock_key;	\
 	static struct lock_class_key dev_addr_list_lock_key;	\
 	unsigned int i;						\
 								\
 	(dev)->qdisc_tx_busylock = &qdisc_tx_busylock_key;	\
-	(dev)->qdisc_running_key = &qdisc_running_key;		\
 	lockdep_set_class(&(dev)->addr_list_lock,		\
 			  &dev_addr_list_lock_key);		\
 	for (i = 0; i < (dev)->num_tx_queues; i++)		\

include/linux/u64_stats_sync.h

@@ -83,6 +83,11 @@ static inline u64 u64_stats_read(const u64_stats_t *p)
 	return local64_read(&p->v);
 }
 
+static inline void u64_stats_set(u64_stats_t *p, u64 val)
+{
+	local64_set(&p->v, val);
+}
+
 static inline void u64_stats_add(u64_stats_t *p, unsigned long val)
 {
 	local64_add(val, &p->v);
@@ -104,6 +109,11 @@ static inline u64 u64_stats_read(const u64_stats_t *p)
 	return p->v;
 }
 
+static inline void u64_stats_set(u64_stats_t *p, u64 val)
+{
+	p->v = val;
+}
+
 static inline void u64_stats_add(u64_stats_t *p, unsigned long val)
 {
 	p->v += val;

include/net/act_api.h

@@ -30,13 +30,13 @@ struct tc_action {
 	atomic_t			tcfa_bindcnt;
 	int				tcfa_action;
 	struct tcf_t			tcfa_tm;
-	struct gnet_stats_basic_packed	tcfa_bstats;
-	struct gnet_stats_basic_packed	tcfa_bstats_hw;
+	struct gnet_stats_basic_sync	tcfa_bstats;
+	struct gnet_stats_basic_sync	tcfa_bstats_hw;
 	struct gnet_stats_queue		tcfa_qstats;
 	struct net_rate_estimator __rcu *tcfa_rate_est;
 	spinlock_t			tcfa_lock;
-	struct gnet_stats_basic_cpu __percpu *cpu_bstats;
-	struct gnet_stats_basic_cpu __percpu *cpu_bstats_hw;
+	struct gnet_stats_basic_sync __percpu *cpu_bstats;
+	struct gnet_stats_basic_sync __percpu *cpu_bstats_hw;
 	struct gnet_stats_queue __percpu *cpu_qstats;
 	struct tc_cookie	__rcu *act_cookie;
 	struct tcf_chain	__rcu *goto_chain;
@@ -206,7 +206,7 @@ static inline void tcf_action_update_bstats(struct tc_action *a,
 					    struct sk_buff *skb)
 {
 	if (likely(a->cpu_bstats)) {
-		bstats_cpu_update(this_cpu_ptr(a->cpu_bstats), skb);
+		bstats_update(this_cpu_ptr(a->cpu_bstats), skb);
 		return;
 	}
 	spin_lock(&a->tcfa_lock);

include/net/gen_stats.h

@@ -7,14 +7,17 @@
 #include <linux/rtnetlink.h>
 #include <linux/pkt_sched.h>
 
-/* Note: this used to be in include/uapi/linux/gen_stats.h */
-struct gnet_stats_basic_packed {
-	__u64	bytes;
-	__u64	packets;
-};
-
-struct gnet_stats_basic_cpu {
-	struct gnet_stats_basic_packed bstats;
+/* Throughput stats.
+ * Must be initialized beforehand with gnet_stats_basic_sync_init().
+ *
+ * If no reads can ever occur parallel to writes (e.g. stack-allocated
+ * bstats), then the internal stat values can be written to and read
+ * from directly. Otherwise, use _bstats_set/update() for writes and
+ * gnet_stats_add_basic() for reads.
+ */
+struct gnet_stats_basic_sync {
+	u64_stats_t bytes;
+	u64_stats_t packets;
 	struct u64_stats_sync syncp;
 } __aligned(2 * sizeof(u64));
 
@@ -34,6 +37,7 @@ struct gnet_dump {
 	struct tc_stats   tc_stats;
 };
 
+void gnet_stats_basic_sync_init(struct gnet_stats_basic_sync *b);
 int gnet_stats_start_copy(struct sk_buff *skb, int type, spinlock_t *lock,
 			  struct gnet_dump *d, int padattr);
 
@@ -42,41 +46,38 @@ int gnet_stats_start_copy_compat(struct sk_buff *skb, int type,
 				 spinlock_t *lock, struct gnet_dump *d,
 				 int padattr);
 
-int gnet_stats_copy_basic(const seqcount_t *running,
-			  struct gnet_dump *d,
-			  struct gnet_stats_basic_cpu __percpu *cpu,
-			  struct gnet_stats_basic_packed *b);
-void __gnet_stats_copy_basic(const seqcount_t *running,
-			     struct gnet_stats_basic_packed *bstats,
-			     struct gnet_stats_basic_cpu __percpu *cpu,
-			     struct gnet_stats_basic_packed *b);
-int gnet_stats_copy_basic_hw(const seqcount_t *running,
-			     struct gnet_dump *d,
-			     struct gnet_stats_basic_cpu __percpu *cpu,
-			     struct gnet_stats_basic_packed *b);
+int gnet_stats_copy_basic(struct gnet_dump *d,
+			  struct gnet_stats_basic_sync __percpu *cpu,
+			  struct gnet_stats_basic_sync *b, bool running);
+void gnet_stats_add_basic(struct gnet_stats_basic_sync *bstats,
+			  struct gnet_stats_basic_sync __percpu *cpu,
+			  struct gnet_stats_basic_sync *b, bool running);
+int gnet_stats_copy_basic_hw(struct gnet_dump *d,
+			     struct gnet_stats_basic_sync __percpu *cpu,
+			     struct gnet_stats_basic_sync *b, bool running);
 int gnet_stats_copy_rate_est(struct gnet_dump *d,
 			     struct net_rate_estimator __rcu **ptr);
 int gnet_stats_copy_queue(struct gnet_dump *d,
 			  struct gnet_stats_queue __percpu *cpu_q,
 			  struct gnet_stats_queue *q, __u32 qlen);
-void __gnet_stats_copy_queue(struct gnet_stats_queue *qstats,
-			     const struct gnet_stats_queue __percpu *cpu_q,
-			     const struct gnet_stats_queue *q, __u32 qlen);
+void gnet_stats_add_queue(struct gnet_stats_queue *qstats,
+			  const struct gnet_stats_queue __percpu *cpu_q,
+			  const struct gnet_stats_queue *q);
 int gnet_stats_copy_app(struct gnet_dump *d, void *st, int len);
 
 int gnet_stats_finish_copy(struct gnet_dump *d);
 
-int gen_new_estimator(struct gnet_stats_basic_packed *bstats,
-		      struct gnet_stats_basic_cpu __percpu *cpu_bstats,
+int gen_new_estimator(struct gnet_stats_basic_sync *bstats,
+		      struct gnet_stats_basic_sync __percpu *cpu_bstats,
 		      struct net_rate_estimator __rcu **rate_est,
 		      spinlock_t *lock,
-		      seqcount_t *running, struct nlattr *opt);
+		      bool running, struct nlattr *opt);
 void gen_kill_estimator(struct net_rate_estimator __rcu **ptr);
-int gen_replace_estimator(struct gnet_stats_basic_packed *bstats,
-			  struct gnet_stats_basic_cpu __percpu *cpu_bstats,
+int gen_replace_estimator(struct gnet_stats_basic_sync *bstats,
+			  struct gnet_stats_basic_sync __percpu *cpu_bstats,
 			  struct net_rate_estimator __rcu **ptr,
 			  spinlock_t *lock,
-			  seqcount_t *running, struct nlattr *opt);
+			  bool running, struct nlattr *opt);
 bool gen_estimator_active(struct net_rate_estimator __rcu **ptr);
 bool gen_estimator_read(struct net_rate_estimator __rcu **ptr,
 			struct gnet_stats_rate_est64 *sample);

include/net/netfilter/xt_rateest.h

@@ -6,7 +6,7 @@
 
 struct xt_rateest {
 	/* keep lock and bstats on same cache line to speedup xt_rateest_tg() */
-	struct gnet_stats_basic_packed	bstats;
+	struct gnet_stats_basic_sync	bstats;
 	spinlock_t			lock;
 
 

include/net/pkt_cls.h

@@ -765,7 +765,7 @@ struct tc_cookie {
 };
 
 struct tc_qopt_offload_stats {
-	struct gnet_stats_basic_packed *bstats;
+	struct gnet_stats_basic_sync *bstats;
 	struct gnet_stats_queue *qstats;
 };
 
@@ -885,7 +885,7 @@ struct tc_gred_qopt_offload_params {
 };
 
 struct tc_gred_qopt_offload_stats {
-	struct gnet_stats_basic_packed bstats[MAX_DPs];
+	struct gnet_stats_basic_sync bstats[MAX_DPs];
 	struct gnet_stats_queue qstats[MAX_DPs];
 	struct red_stats *xstats[MAX_DPs];
 };

include/net/sch_generic.h

@@ -38,6 +38,10 @@ enum qdisc_state_t {
 	__QDISC_STATE_DEACTIVATED,
 	__QDISC_STATE_MISSED,
 	__QDISC_STATE_DRAINING,
+	/* Only for !TCQ_F_NOLOCK qdisc. Never access it directly.
+	 * Use qdisc_run_begin/end() or qdisc_is_running() instead.
+	 */
+	__QDISC_STATE_RUNNING,
 };
 
 #define QDISC_STATE_MISSED	BIT(__QDISC_STATE_MISSED)
@@ -97,7 +101,7 @@ struct Qdisc {
 	struct netdev_queue	*dev_queue;
 
 	struct net_rate_estimator __rcu *rate_est;
-	struct gnet_stats_basic_cpu __percpu *cpu_bstats;
+	struct gnet_stats_basic_sync __percpu *cpu_bstats;
 	struct gnet_stats_queue	__percpu *cpu_qstats;
 	int			pad;
 	refcount_t		refcnt;
@@ -107,8 +111,7 @@ struct Qdisc {
 	 */
 	struct sk_buff_head	gso_skb ____cacheline_aligned_in_smp;
 	struct qdisc_skb_head	q;
-	struct gnet_stats_basic_packed bstats;
-	seqcount_t		running;
+	struct gnet_stats_basic_sync bstats;
 	struct gnet_stats_queue	qstats;
 	unsigned long		state;
 	struct Qdisc            *next_sched;
@@ -143,11 +146,15 @@ static inline struct Qdisc *qdisc_refcount_inc_nz(struct Qdisc *qdisc)
 	return NULL;
 }
 
+/* For !TCQ_F_NOLOCK qdisc: callers must either call this within a qdisc
+ * root_lock section, or provide their own memory barriers -- ordering
+ * against qdisc_run_begin/end() atomic bit operations.
+ */
 static inline bool qdisc_is_running(struct Qdisc *qdisc)
 {
 	if (qdisc->flags & TCQ_F_NOLOCK)
 		return spin_is_locked(&qdisc->seqlock);
-	return (raw_read_seqcount(&qdisc->running) & 1) ? true : false;
+	return test_bit(__QDISC_STATE_RUNNING, &qdisc->state);
 }
 
 static inline bool nolock_qdisc_is_empty(const struct Qdisc *qdisc)
@@ -167,6 +174,9 @@ static inline bool qdisc_is_empty(const struct Qdisc *qdisc)
 	return !READ_ONCE(qdisc->q.qlen);
 }
 
+/* For !TCQ_F_NOLOCK qdisc, qdisc_run_begin/end() must be invoked with
+ * the qdisc root lock acquired.
+ */
 static inline bool qdisc_run_begin(struct Qdisc *qdisc)
 {
 	if (qdisc->flags & TCQ_F_NOLOCK) {
@@ -206,15 +216,8 @@ static inline bool qdisc_run_begin(struct Qdisc *qdisc)
 		 * after it releases the lock at the end of qdisc_run_end().
 		 */
 		return spin_trylock(&qdisc->seqlock);
-	} else if (qdisc_is_running(qdisc)) {
-		return false;
 	}
-	/* Variant of write_seqcount_begin() telling lockdep a trylock
-	 * was attempted.
-	 */
-	raw_write_seqcount_begin(&qdisc->running);
-	seqcount_acquire(&qdisc->running.dep_map, 0, 1, _RET_IP_);
-	return true;
+	return test_and_set_bit(__QDISC_STATE_RUNNING, &qdisc->state);
 }
 
 static inline void qdisc_run_end(struct Qdisc *qdisc)
@@ -226,7 +229,7 @@ static inline void qdisc_run_end(struct Qdisc *qdisc)
 				      &qdisc->state)))
 			__netif_schedule(qdisc);
 	} else {
-		write_seqcount_end(&qdisc->running);
+		clear_bit(__QDISC_STATE_RUNNING, &qdisc->state);
 	}
 }
 
@@ -592,14 +595,6 @@ static inline spinlock_t *qdisc_root_sleeping_lock(const struct Qdisc *qdisc)
 	return qdisc_lock(root);
 }
 
-static inline seqcount_t *qdisc_root_sleeping_running(const struct Qdisc *qdisc)
-{
-	struct Qdisc *root = qdisc_root_sleeping(qdisc);
-
-	ASSERT_RTNL();
-	return &root->running;
-}
-
 static inline struct net_device *qdisc_dev(const struct Qdisc *qdisc)
 {
 	return qdisc->dev_queue->dev;
@@ -849,14 +844,16 @@ static inline int qdisc_enqueue(struct sk_buff *skb, struct Qdisc *sch,
 	return sch->enqueue(skb, sch, to_free);
 }
 
-static inline void _bstats_update(struct gnet_stats_basic_packed *bstats,
+static inline void _bstats_update(struct gnet_stats_basic_sync *bstats,
 				  __u64 bytes, __u32 packets)
 {
-	bstats->bytes += bytes;
-	bstats->packets += packets;
+	u64_stats_update_begin(&bstats->syncp);
+	u64_stats_add(&bstats->bytes, bytes);
+	u64_stats_add(&bstats->packets, packets);
+	u64_stats_update_end(&bstats->syncp);
 }
 
-static inline void bstats_update(struct gnet_stats_basic_packed *bstats,
+static inline void bstats_update(struct gnet_stats_basic_sync *bstats,
 				 const struct sk_buff *skb)
 {
 	_bstats_update(bstats,
@@ -864,26 +861,10 @@ static inline void bstats_update(struct gnet_stats_basic_packed *bstats,
 		       skb_is_gso(skb) ? skb_shinfo(skb)->gso_segs : 1);
 }
 
-static inline void _bstats_cpu_update(struct gnet_stats_basic_cpu *bstats,
-				      __u64 bytes, __u32 packets)
-{
-	u64_stats_update_begin(&bstats->syncp);
-	_bstats_update(&bstats->bstats, bytes, packets);
-	u64_stats_update_end(&bstats->syncp);
-}
-
-static inline void bstats_cpu_update(struct gnet_stats_basic_cpu *bstats,
-				     const struct sk_buff *skb)
-{
-	u64_stats_update_begin(&bstats->syncp);
-	bstats_update(&bstats->bstats, skb);
-	u64_stats_update_end(&bstats->syncp);
-}
-
 static inline void qdisc_bstats_cpu_update(struct Qdisc *sch,
 					   const struct sk_buff *skb)
 {
-	bstats_cpu_update(this_cpu_ptr(sch->cpu_bstats), skb);
+	bstats_update(this_cpu_ptr(sch->cpu_bstats), skb);
 }
 
 static inline void qdisc_bstats_update(struct Qdisc *sch,
@@ -972,10 +953,9 @@ 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;
+	gnet_stats_add_queue(&qstats, sch->cpu_qstats, &sch->qstats);
+	*qlen = qstats.qlen + qdisc_qlen(sch);
 	*backlog = qstats.backlog;
 }
 
@@ -1316,7 +1296,7 @@ void psched_ppscfg_precompute(struct psched_pktrate *r, u64 pktrate64);
 struct mini_Qdisc {
 	struct tcf_proto *filter_list;
 	struct tcf_block *block;
-	struct gnet_stats_basic_cpu __percpu *cpu_bstats;
+	struct gnet_stats_basic_sync __percpu *cpu_bstats;
 	struct gnet_stats_queue	__percpu *cpu_qstats;
 	struct rcu_head rcu;
 };
@@ -1324,7 +1304,7 @@ struct mini_Qdisc {
 static inline void mini_qdisc_bstats_cpu_update(struct mini_Qdisc *miniq,
 						const struct sk_buff *skb)
 {
-	bstats_cpu_update(this_cpu_ptr(miniq->cpu_bstats), skb);
+	bstats_update(this_cpu_ptr(miniq->cpu_bstats), skb);
 }
 
 static inline void mini_qdisc_qstats_cpu_drop(struct mini_Qdisc *miniq)

net/core/gen_estimator.c

@@ -40,10 +40,10 @@
  */
 
 struct net_rate_estimator {
-	struct gnet_stats_basic_packed	*bstats;
+	struct gnet_stats_basic_sync	*bstats;
 	spinlock_t		*stats_lock;
-	seqcount_t		*running;
-	struct gnet_stats_basic_cpu __percpu *cpu_bstats;
+	bool			running;
+	struct gnet_stats_basic_sync __percpu *cpu_bstats;
 	u8			ewma_log;
 	u8			intvl_log; /* period : (250ms << intvl_log) */
 
@@ -60,13 +60,13 @@ struct net_rate_estimator {
 };
 
 static void est_fetch_counters(struct net_rate_estimator *e,
-			       struct gnet_stats_basic_packed *b)
+			       struct gnet_stats_basic_sync *b)
 {
-	memset(b, 0, sizeof(*b));
+	gnet_stats_basic_sync_init(b);
 	if (e->stats_lock)
 		spin_lock(e->stats_lock);
 
-	__gnet_stats_copy_basic(e->running, b, e->cpu_bstats, e->bstats);
+	gnet_stats_add_basic(b, e->cpu_bstats, e->bstats, e->running);
 
 	if (e->stats_lock)
 		spin_unlock(e->stats_lock);
@@ -76,14 +76,18 @@ static void est_fetch_counters(struct net_rate_estimator *e,
 static void est_timer(struct timer_list *t)
 {
 	struct net_rate_estimator *est = from_timer(est, t, timer);
-	struct gnet_stats_basic_packed b;
+	struct gnet_stats_basic_sync b;
+	u64 b_bytes, b_packets;
 	u64 rate, brate;
 
 	est_fetch_counters(est, &b);
-	brate = (b.bytes - est->last_bytes) << (10 - est->intvl_log);
+	b_bytes = u64_stats_read(&b.bytes);
+	b_packets = u64_stats_read(&b.packets);
+
+	brate = (b_bytes - est->last_bytes) << (10 - est->intvl_log);
 	brate = (brate >> est->ewma_log) - (est->avbps >> est->ewma_log);
 
-	rate = (b.packets - est->last_packets) << (10 - est->intvl_log);
+	rate = (b_packets - est->last_packets) << (10 - est->intvl_log);
 	rate = (rate >> est->ewma_log) - (est->avpps >> est->ewma_log);
 
 	write_seqcount_begin(&est->seq);
@@ -91,8 +95,8 @@ static void est_timer(struct timer_list *t)
 	est->avpps += rate;
 	write_seqcount_end(&est->seq);
 
-	est->last_bytes = b.bytes;
-	est->last_packets = b.packets;
+	est->last_bytes = b_bytes;
+	est->last_packets = b_packets;
 
 	est->next_jiffies += ((HZ/4) << est->intvl_log);
 
@@ -109,7 +113,9 @@ static void est_timer(struct timer_list *t)
  * @cpu_bstats: bstats per cpu
  * @rate_est: rate estimator statistics
  * @lock: lock for statistics and control path
- * @running: qdisc running seqcount
+ * @running: true if @bstats represents a running qdisc, thus @bstats'
+ *           internal values might change during basic reads. Only used
+ *           if @bstats_cpu is NULL
  * @opt: rate estimator configuration TLV
  *
  * Creates a new rate estimator with &bstats as source and &rate_est
@@ -121,16 +127,16 @@ static void est_timer(struct timer_list *t)
  * Returns 0 on success or a negative error code.
  *
  */
-int gen_new_estimator(struct gnet_stats_basic_packed *bstats,
-		      struct gnet_stats_basic_cpu __percpu *cpu_bstats,
+int gen_new_estimator(struct gnet_stats_basic_sync *bstats,
+		      struct gnet_stats_basic_sync __percpu *cpu_bstats,
 		      struct net_rate_estimator __rcu **rate_est,
 		      spinlock_t *lock,
-		      seqcount_t *running,
+		      bool running,
 		      struct nlattr *opt)
 {
 	struct gnet_estimator *parm = nla_data(opt);
 	struct net_rate_estimator *old, *est;
-	struct gnet_stats_basic_packed b;
+	struct gnet_stats_basic_sync b;
 	int intvl_log;
 
 	if (nla_len(opt) < sizeof(*parm))
@@ -164,8 +170,8 @@ int gen_new_estimator(struct gnet_stats_basic_packed *bstats,
 	est_fetch_counters(est, &b);
 	if (lock)
 		local_bh_enable();
-	est->last_bytes = b.bytes;
-	est->last_packets = b.packets;
+	est->last_bytes = u64_stats_read(&b.bytes);
+	est->last_packets = u64_stats_read(&b.packets);
 
 	if (lock)
 		spin_lock_bh(lock);
@@ -214,7 +220,9 @@ EXPORT_SYMBOL(gen_kill_estimator);
  * @cpu_bstats: bstats per cpu
  * @rate_est: rate estimator statistics
  * @lock: lock for statistics and control path
- * @running: qdisc running seqcount (might be NULL)
+ * @running: true if @bstats represents a running qdisc, thus @bstats'
+ *           internal values might change during basic reads. Only used
+ *           if @cpu_bstats is NULL
  * @opt: rate estimator configuration TLV
  *
  * Replaces the configuration of a rate estimator by calling
@@ -222,11 +230,11 @@ EXPORT_SYMBOL(gen_kill_estimator);
  *
  * Returns 0 on success or a negative error code.
  */
-int gen_replace_estimator(struct gnet_stats_basic_packed *bstats,
-			  struct gnet_stats_basic_cpu __percpu *cpu_bstats,
+int gen_replace_estimator(struct gnet_stats_basic_sync *bstats,
+			  struct gnet_stats_basic_sync __percpu *cpu_bstats,
 			  struct net_rate_estimator __rcu **rate_est,
 			  spinlock_t *lock,
-			  seqcount_t *running, struct nlattr *opt)
+			  bool running, struct nlattr *opt)
 {
 	return gen_new_estimator(bstats, cpu_bstats, rate_est,
 				 lock, running, opt);

net/core/gen_stats.c

@@ -18,7 +18,7 @@
 #include <linux/gen_stats.h>
 #include <net/netlink.h>
 #include <net/gen_stats.h>
-
+#include <net/sch_generic.h>
 
 static inline int
 gnet_stats_copy(struct gnet_dump *d, int type, void *buf, int size, int padattr)
@@ -114,63 +114,81 @@ gnet_stats_start_copy(struct sk_buff *skb, int type, spinlock_t *lock,
 }
 EXPORT_SYMBOL(gnet_stats_start_copy);
 
-static void
-__gnet_stats_copy_basic_cpu(struct gnet_stats_basic_packed *bstats,
-			    struct gnet_stats_basic_cpu __percpu *cpu)
+/* Must not be inlined, due to u64_stats seqcount_t lockdep key */
+void gnet_stats_basic_sync_init(struct gnet_stats_basic_sync *b)
 {
+	u64_stats_set(&b->bytes, 0);
+	u64_stats_set(&b->packets, 0);
+	u64_stats_init(&b->syncp);
+}
+EXPORT_SYMBOL(gnet_stats_basic_sync_init);
+
+static void gnet_stats_add_basic_cpu(struct gnet_stats_basic_sync *bstats,
+				     struct gnet_stats_basic_sync __percpu *cpu)
+{
+	u64 t_bytes = 0, t_packets = 0;
 	int i;
 
 	for_each_possible_cpu(i) {
-		struct gnet_stats_basic_cpu *bcpu = per_cpu_ptr(cpu, i);
+		struct gnet_stats_basic_sync *bcpu = per_cpu_ptr(cpu, i);
 		unsigned int start;
 		u64 bytes, packets;
 
 		do {
 			start = u64_stats_fetch_begin_irq(&bcpu->syncp);
-			bytes = bcpu->bstats.bytes;
-			packets = bcpu->bstats.packets;
+			bytes = u64_stats_read(&bcpu->bytes);
+			packets = u64_stats_read(&bcpu->packets);
 		} while (u64_stats_fetch_retry_irq(&bcpu->syncp, start));
 
-		bstats->bytes += bytes;
-		bstats->packets += packets;
+		t_bytes += bytes;
+		t_packets += packets;
 	}
+	_bstats_update(bstats, t_bytes, t_packets);
 }
 
-void
-__gnet_stats_copy_basic(const seqcount_t *running,
-			struct gnet_stats_basic_packed *bstats,
-			struct gnet_stats_basic_cpu __percpu *cpu,
-			struct gnet_stats_basic_packed *b)
+void gnet_stats_add_basic(struct gnet_stats_basic_sync *bstats,
+			  struct gnet_stats_basic_sync __percpu *cpu,
+			  struct gnet_stats_basic_sync *b, bool running)
 {
-	unsigned int seq;
+	unsigned int start;
+	u64 bytes = 0;
+	u64 packets = 0;
+
+	WARN_ON_ONCE((cpu || running) && !in_task());
 
 	if (cpu) {
-		__gnet_stats_copy_basic_cpu(bstats, cpu);
+		gnet_stats_add_basic_cpu(bstats, cpu);
 		return;
 	}
 	do {
 		if (running)
-			seq = read_seqcount_begin(running);
-		bstats->bytes = b->bytes;
-		bstats->packets = b->packets;
-	} while (running && read_seqcount_retry(running, seq));
+			start = u64_stats_fetch_begin_irq(&b->syncp);
+		bytes = u64_stats_read(&b->bytes);
+		packets = u64_stats_read(&b->packets);
+	} while (running && u64_stats_fetch_retry_irq(&b->syncp, start));
+
+	_bstats_update(bstats, bytes, packets);
 }
-EXPORT_SYMBOL(__gnet_stats_copy_basic);
+EXPORT_SYMBOL(gnet_stats_add_basic);
 
 static int
-___gnet_stats_copy_basic(const seqcount_t *running,
-			 struct gnet_dump *d,
-			 struct gnet_stats_basic_cpu __percpu *cpu,
-			 struct gnet_stats_basic_packed *b,
-			 int type)
+___gnet_stats_copy_basic(struct gnet_dump *d,
+			 struct gnet_stats_basic_sync __percpu *cpu,
+			 struct gnet_stats_basic_sync *b,
+			 int type, bool running)
 {
-	struct gnet_stats_basic_packed bstats = {0};
+	struct gnet_stats_basic_sync bstats;
+	u64 bstats_bytes, bstats_packets;
+
+	gnet_stats_basic_sync_init(&bstats);
+	gnet_stats_add_basic(&bstats, cpu, b, running);
 
-	__gnet_stats_copy_basic(running, &bstats, cpu, b);
+	bstats_bytes = u64_stats_read(&bstats.bytes);
+	bstats_packets = u64_stats_read(&bstats.packets);
 
 	if (d->compat_tc_stats && type == TCA_STATS_BASIC) {
-		d->tc_stats.bytes = bstats.bytes;
-		d->tc_stats.packets = bstats.packets;
+		d->tc_stats.bytes = bstats_bytes;
+		d->tc_stats.packets = bstats_packets;
 	}
 
 	if (d->tail) {
@@ -178,24 +196,28 @@ ___gnet_stats_copy_basic(const seqcount_t *running,
 		int res;
 
 		memset(&sb, 0, sizeof(sb));
-		sb.bytes = bstats.bytes;
-		sb.packets = bstats.packets;
+		sb.bytes = bstats_bytes;
+		sb.packets = bstats_packets;
 		res = gnet_stats_copy(d, type, &sb, sizeof(sb), TCA_STATS_PAD);
-		if (res < 0 || sb.packets == bstats.packets)
+		if (res < 0 || sb.packets == bstats_packets)
 			return res;
 		/* emit 64bit stats only if needed */
-		return gnet_stats_copy(d, TCA_STATS_PKT64, &bstats.packets,
-				       sizeof(bstats.packets), TCA_STATS_PAD);
+		return gnet_stats_copy(d, TCA_STATS_PKT64, &bstats_packets,
+				       sizeof(bstats_packets), TCA_STATS_PAD);
 	}
 	return 0;
 }
 
 /**
  * gnet_stats_copy_basic - copy basic statistics into statistic TLV
- * @running: seqcount_t pointer
  * @d: dumping handle
  * @cpu: copy statistic per cpu
  * @b: basic statistics
+ * @running: true if @b represents a running qdisc, thus @b's
+ *           internal values might change during basic reads.
+ *           Only used if @cpu is NULL
+ *
+ * Context: task; must not be run from IRQ or BH contexts
  *
  * Appends the basic statistics to the top level TLV created by
  * gnet_stats_start_copy().
@@ -204,22 +226,25 @@ ___gnet_stats_copy_basic(const seqcount_t *running,
  * if the room in the socket buffer was not sufficient.
  */
 int
-gnet_stats_copy_basic(const seqcount_t *running,
-		      struct gnet_dump *d,
-		      struct gnet_stats_basic_cpu __percpu *cpu,
-		      struct gnet_stats_basic_packed *b)
+gnet_stats_copy_basic(struct gnet_dump *d,
+		      struct gnet_stats_basic_sync __percpu *cpu,
+		      struct gnet_stats_basic_sync *b,
+		      bool running)
 {
-	return ___gnet_stats_copy_basic(running, d, cpu, b,
-					TCA_STATS_BASIC);
+	return ___gnet_stats_copy_basic(d, cpu, b, TCA_STATS_BASIC, running);
 }
 EXPORT_SYMBOL(gnet_stats_copy_basic);
 
 /**
  * gnet_stats_copy_basic_hw - copy basic hw statistics into statistic TLV
- * @running: seqcount_t pointer
  * @d: dumping handle
  * @cpu: copy statistic per cpu
  * @b: basic statistics
+ * @running: true if @b represents a running qdisc, thus @b's
+ *           internal values might change during basic reads.
+ *           Only used if @cpu is NULL
+ *
+ * Context: task; must not be run from IRQ or BH contexts
  *
  * Appends the basic statistics to the top level TLV created by
  * gnet_stats_start_copy().
@@ -228,13 +253,12 @@ EXPORT_SYMBOL(gnet_stats_copy_basic);
  * if the room in the socket buffer was not sufficient.
  */
 int
-gnet_stats_copy_basic_hw(const seqcount_t *running,
-			 struct gnet_dump *d,
-			 struct gnet_stats_basic_cpu __percpu *cpu,
-			 struct gnet_stats_basic_packed *b)
+gnet_stats_copy_basic_hw(struct gnet_dump *d,
+			 struct gnet_stats_basic_sync __percpu *cpu,
+			 struct gnet_stats_basic_sync *b,
+			 bool running)
 {
-	return ___gnet_stats_copy_basic(running, d, cpu, b,
-					TCA_STATS_BASIC_HW);
+	return ___gnet_stats_copy_basic(d, cpu, b, TCA_STATS_BASIC_HW, running);
 }
 EXPORT_SYMBOL(gnet_stats_copy_basic_hw);
 
@@ -282,16 +306,15 @@ gnet_stats_copy_rate_est(struct gnet_dump *d,
 }
 EXPORT_SYMBOL(gnet_stats_copy_rate_est);
 
-static void
-__gnet_stats_copy_queue_cpu(struct gnet_stats_queue *qstats,
-			    const struct gnet_stats_queue __percpu *q)
+static void gnet_stats_add_queue_cpu(struct gnet_stats_queue *qstats,
+				     const struct gnet_stats_queue __percpu *q)
 {
 	int i;
 
 	for_each_possible_cpu(i) {
 		const struct gnet_stats_queue *qcpu = per_cpu_ptr(q, i);
 
-		qstats->qlen = 0;
+		qstats->qlen += qcpu->backlog;
 		qstats->backlog += qcpu->backlog;
 		qstats->drops += qcpu->drops;
 		qstats->requeues += qcpu->requeues;
@@ -299,24 +322,21 @@ __gnet_stats_copy_queue_cpu(struct gnet_stats_queue *qstats,
 	}
 }
 
-void __gnet_stats_copy_queue(struct gnet_stats_queue *qstats,
-			     const struct gnet_stats_queue __percpu *cpu,
-			     const struct gnet_stats_queue *q,
-			     __u32 qlen)
+void gnet_stats_add_queue(struct gnet_stats_queue *qstats,
+			  const struct gnet_stats_queue __percpu *cpu,
+			  const struct gnet_stats_queue *q)
 {
 	if (cpu) {
-		__gnet_stats_copy_queue_cpu(qstats, cpu);
+		gnet_stats_add_queue_cpu(qstats, cpu);
 	} else {
-		qstats->qlen = q->qlen;
-		qstats->backlog = q->backlog;
-		qstats->drops = q->drops;
-		qstats->requeues = q->requeues;
-		qstats->overlimits = q->overlimits;
+		qstats->qlen += q->qlen;
+		qstats->backlog += q->backlog;
+		qstats->drops += q->drops;
+		qstats->requeues += q->requeues;
+		qstats->overlimits += q->overlimits;
 	}
-
-	qstats->qlen = qlen;
 }
-EXPORT_SYMBOL(__gnet_stats_copy_queue);
+EXPORT_SYMBOL(gnet_stats_add_queue);
 
 /**
  * gnet_stats_copy_queue - copy queue statistics into statistics TLV
@@ -339,7 +359,8 @@ gnet_stats_copy_queue(struct gnet_dump *d,
 {
 	struct gnet_stats_queue qstats = {0};
 
-	__gnet_stats_copy_queue(&qstats, cpu_q, q, qlen);
+	gnet_stats_add_queue(&qstats, cpu_q, q);
+	qstats.qlen = qlen;
 
 	if (d->compat_tc_stats) {
 		d->tc_stats.drops = qstats.drops;

net/netfilter/xt_RATEEST.c

@@ -94,11 +94,11 @@ static unsigned int
 xt_rateest_tg(struct sk_buff *skb, const struct xt_action_param *par)
 {
 	const struct xt_rateest_target_info *info = par->targinfo;
-	struct gnet_stats_basic_packed *stats = &info->est->bstats;
+	struct gnet_stats_basic_sync *stats = &info->est->bstats;
 
 	spin_lock_bh(&info->est->lock);
-	stats->bytes += skb->len;
-	stats->packets++;
+	u64_stats_add(&stats->bytes, skb->len);
+	u64_stats_inc(&stats->packets);
 	spin_unlock_bh(&info->est->lock);
 
 	return XT_CONTINUE;
@@ -143,6 +143,7 @@ static int xt_rateest_tg_checkentry(const struct xt_tgchk_param *par)
 	if (!est)
 		goto err1;
 
+	gnet_stats_basic_sync_init(&est->bstats);
 	strlcpy(est->name, info->name, sizeof(est->name));
 	spin_lock_init(&est->lock);
 	est->refcnt		= 1;

net/sched/act_api.c

@@ -480,16 +480,18 @@ int tcf_idr_create(struct tc_action_net *tn, u32 index, struct nlattr *est,
 		atomic_set(&p->tcfa_bindcnt, 1);
 
 	if (cpustats) {
-		p->cpu_bstats = netdev_alloc_pcpu_stats(struct gnet_stats_basic_cpu);
+		p->cpu_bstats = netdev_alloc_pcpu_stats(struct gnet_stats_basic_sync);
 		if (!p->cpu_bstats)
 			goto err1;
-		p->cpu_bstats_hw = netdev_alloc_pcpu_stats(struct gnet_stats_basic_cpu);
+		p->cpu_bstats_hw = netdev_alloc_pcpu_stats(struct gnet_stats_basic_sync);
 		if (!p->cpu_bstats_hw)
 			goto err2;
 		p->cpu_qstats = alloc_percpu(struct gnet_stats_queue);
 		if (!p->cpu_qstats)
 			goto err3;
 	}
+	gnet_stats_basic_sync_init(&p->tcfa_bstats);
+	gnet_stats_basic_sync_init(&p->tcfa_bstats_hw);
 	spin_lock_init(&p->tcfa_lock);
 	p->tcfa_index = index;
 	p->tcfa_tm.install = jiffies;
@@ -499,7 +501,7 @@ int tcf_idr_create(struct tc_action_net *tn, u32 index, struct nlattr *est,
 	if (est) {
 		err = gen_new_estimator(&p->tcfa_bstats, p->cpu_bstats,
 					&p->tcfa_rate_est,
-					&p->tcfa_lock, NULL, est);
+					&p->tcfa_lock, false, est);
 		if (err)
 			goto err4;
 	}
@@ -1126,13 +1128,13 @@ void tcf_action_update_stats(struct tc_action *a, u64 bytes, u64 packets,
 			     u64 drops, bool hw)
 {
 	if (a->cpu_bstats) {
-		_bstats_cpu_update(this_cpu_ptr(a->cpu_bstats), bytes, packets);
+		_bstats_update(this_cpu_ptr(a->cpu_bstats), bytes, packets);
 
 		this_cpu_ptr(a->cpu_qstats)->drops += drops;
 
 		if (hw)
-			_bstats_cpu_update(this_cpu_ptr(a->cpu_bstats_hw),
-					   bytes, packets);
+			_bstats_update(this_cpu_ptr(a->cpu_bstats_hw),
+				       bytes, packets);
 		return;
 	}
 
@@ -1171,9 +1173,10 @@ int tcf_action_copy_stats(struct sk_buff *skb, struct tc_action *p,
 	if (err < 0)
 		goto errout;
 
-	if (gnet_stats_copy_basic(NULL, &d, p->cpu_bstats, &p->tcfa_bstats) < 0 ||
-	    gnet_stats_copy_basic_hw(NULL, &d, p->cpu_bstats_hw,
-				     &p->tcfa_bstats_hw) < 0 ||
+	if (gnet_stats_copy_basic(&d, p->cpu_bstats,
+				  &p->tcfa_bstats, false) < 0 ||
+	    gnet_stats_copy_basic_hw(&d, p->cpu_bstats_hw,
+				     &p->tcfa_bstats_hw, false) < 0 ||
 	    gnet_stats_copy_rate_est(&d, &p->tcfa_rate_est) < 0 ||
 	    gnet_stats_copy_queue(&d, p->cpu_qstats,
 				  &p->tcfa_qstats,

net/sched/act_bpf.c

@@ -41,7 +41,7 @@ static int tcf_bpf_act(struct sk_buff *skb, const struct tc_action *act,
 	int action, filter_res;
 
 	tcf_lastuse_update(&prog->tcf_tm);
-	bstats_cpu_update(this_cpu_ptr(prog->common.cpu_bstats), skb);
+	bstats_update(this_cpu_ptr(prog->common.cpu_bstats), skb);
 
 	filter = rcu_dereference(prog->filter);
 	if (at_ingress) {

net/sched/act_ife.c

@@ -718,7 +718,7 @@ static int tcf_ife_decode(struct sk_buff *skb, const struct tc_action *a,
 	u8 *tlv_data;
 	u16 metalen;
 
-	bstats_cpu_update(this_cpu_ptr(ife->common.cpu_bstats), skb);
+	bstats_update(this_cpu_ptr(ife->common.cpu_bstats), skb);
 	tcf_lastuse_update(&ife->tcf_tm);
 
 	if (skb_at_tc_ingress(skb))
@@ -806,7 +806,7 @@ static int tcf_ife_encode(struct sk_buff *skb, const struct tc_action *a,
 			exceed_mtu = true;
 	}
 
-	bstats_cpu_update(this_cpu_ptr(ife->common.cpu_bstats), skb);
+	bstats_update(this_cpu_ptr(ife->common.cpu_bstats), skb);
 	tcf_lastuse_update(&ife->tcf_tm);
 
 	if (!metalen) {		/* no metadata to send */

net/sched/act_mpls.c

@@ -59,7 +59,7 @@ static int tcf_mpls_act(struct sk_buff *skb, const struct tc_action *a,
 	int ret, mac_len;
 
 	tcf_lastuse_update(&m->tcf_tm);
-	bstats_cpu_update(this_cpu_ptr(m->common.cpu_bstats), skb);
+	bstats_update(this_cpu_ptr(m->common.cpu_bstats), skb);
 
 	/* Ensure 'data' points at mac_header prior calling mpls manipulating
 	 * functions.

net/sched/act_police.c

@@ -125,7 +125,7 @@ static int tcf_police_init(struct net *net, struct nlattr *nla,
 					    police->common.cpu_bstats,
 					    &police->tcf_rate_est,
 					    &police->tcf_lock,
-					    NULL, est);
+					    false, est);
 		if (err)
 			goto failure;
 	} else if (tb[TCA_POLICE_AVRATE] &&
@@ -248,7 +248,7 @@ static int tcf_police_act(struct sk_buff *skb, const struct tc_action *a,
 	int ret;
 
 	tcf_lastuse_update(&police->tcf_tm);
-	bstats_cpu_update(this_cpu_ptr(police->common.cpu_bstats), skb);
+	bstats_update(this_cpu_ptr(police->common.cpu_bstats), skb);
 
 	ret = READ_ONCE(police->tcf_action);
 	p = rcu_dereference_bh(police->params);

net/sched/act_sample.c

@@ -163,7 +163,7 @@ static int tcf_sample_act(struct sk_buff *skb, const struct tc_action *a,
 	int retval;
 
 	tcf_lastuse_update(&s->tcf_tm);
-	bstats_cpu_update(this_cpu_ptr(s->common.cpu_bstats), skb);
+	bstats_update(this_cpu_ptr(s->common.cpu_bstats), skb);
 	retval = READ_ONCE(s->tcf_action);
 
 	psample_group = rcu_dereference_bh(s->psample_group);

net/sched/act_simple.c

@@ -36,7 +36,8 @@ static int tcf_simp_act(struct sk_buff *skb, const struct tc_action *a,
 	 * then it would look like "hello_3" (without quotes)
 	 */
 	pr_info("simple: %s_%llu\n",
-	       (char *)d->tcfd_defdata, d->tcf_bstats.packets);
+		(char *)d->tcfd_defdata,
+		u64_stats_read(&d->tcf_bstats.packets));
 	spin_unlock(&d->tcf_lock);
 	return d->tcf_action;
 }

net/sched/act_skbedit.c

@@ -31,7 +31,7 @@ static int tcf_skbedit_act(struct sk_buff *skb, const struct tc_action *a,
 	int action;
 
 	tcf_lastuse_update(&d->tcf_tm);
-	bstats_cpu_update(this_cpu_ptr(d->common.cpu_bstats), skb);
+	bstats_update(this_cpu_ptr(d->common.cpu_bstats), skb);
 
 	params = rcu_dereference_bh(d->params);
 	action = READ_ONCE(d->tcf_action);

net/sched/act_skbmod.c

@@ -31,7 +31,7 @@ static int tcf_skbmod_act(struct sk_buff *skb, const struct tc_action *a,
 	u64 flags;
 
 	tcf_lastuse_update(&d->tcf_tm);
-	bstats_cpu_update(this_cpu_ptr(d->common.cpu_bstats), skb);
+	bstats_update(this_cpu_ptr(d->common.cpu_bstats), skb);
 
 	action = READ_ONCE(d->tcf_action);
 	if (unlikely(action == TC_ACT_SHOT))

net/sched/sch_api.c

@@ -885,7 +885,7 @@ static void qdisc_offload_graft_root(struct net_device *dev,
 static int tc_fill_qdisc(struct sk_buff *skb, struct Qdisc *q, u32 clid,
 			 u32 portid, u32 seq, u16 flags, int event)
 {
-	struct gnet_stats_basic_cpu __percpu *cpu_bstats = NULL;
+	struct gnet_stats_basic_sync __percpu *cpu_bstats = NULL;
 	struct gnet_stats_queue __percpu *cpu_qstats = NULL;
 	struct tcmsg *tcm;
 	struct nlmsghdr  *nlh;
@@ -943,8 +943,7 @@ static int tc_fill_qdisc(struct sk_buff *skb, struct Qdisc *q, u32 clid,
 		cpu_qstats = q->cpu_qstats;
 	}
 
-	if (gnet_stats_copy_basic(qdisc_root_sleeping_running(q),
-				  &d, cpu_bstats, &q->bstats) < 0 ||
+	if (gnet_stats_copy_basic(&d, cpu_bstats, &q->bstats, true) < 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;
@@ -1265,26 +1264,17 @@ static struct Qdisc *qdisc_create(struct net_device *dev,
 		rcu_assign_pointer(sch->stab, stab);
 	}
 	if (tca[TCA_RATE]) {
-		seqcount_t *running;
-
 		err = -EOPNOTSUPP;
 		if (sch->flags & TCQ_F_MQROOT) {
 			NL_SET_ERR_MSG(extack, "Cannot attach rate estimator to a multi-queue root qdisc");
 			goto err_out4;
 		}
 
-		if (sch->parent != TC_H_ROOT &&
-		    !(sch->flags & TCQ_F_INGRESS) &&
-		    (!p || !(p->flags & TCQ_F_MQROOT)))
-			running = qdisc_root_sleeping_running(sch);
-		else
-			running = &sch->running;
-
 		err = gen_new_estimator(&sch->bstats,
 					sch->cpu_bstats,
 					&sch->rate_est,
 					NULL,
-					running,
+					true,
 					tca[TCA_RATE]);
 		if (err) {
 			NL_SET_ERR_MSG(extack, "Failed to generate new estimator");
@@ -1360,7 +1350,7 @@ static int qdisc_change(struct Qdisc *sch, struct nlattr **tca,
 				      sch->cpu_bstats,
 				      &sch->rate_est,
 				      NULL,
-				      qdisc_root_sleeping_running(sch),
+				      true,
 				      tca[TCA_RATE]);
 	}
 out:

net/sched/sch_atm.c

@@ -52,7 +52,7 @@ struct atm_flow_data {
 	struct atm_qdisc_data	*parent;	/* parent qdisc */
 	struct socket		*sock;		/* for closing */
 	int			ref;		/* reference count */
-	struct gnet_stats_basic_packed	bstats;
+	struct gnet_stats_basic_sync	bstats;
 	struct gnet_stats_queue	qstats;
 	struct list_head	list;
 	struct atm_flow_data	*excess;	/* flow for excess traffic;
@@ -548,6 +548,7 @@ static int atm_tc_init(struct Qdisc *sch, struct nlattr *opt,
 	pr_debug("atm_tc_init(sch %p,[qdisc %p],opt %p)\n", sch, p, opt);
 	INIT_LIST_HEAD(&p->flows);
 	INIT_LIST_HEAD(&p->link.list);
+	gnet_stats_basic_sync_init(&p->link.bstats);
 	list_add(&p->link.list, &p->flows);
 	p->link.q = qdisc_create_dflt(sch->dev_queue,
 				      &pfifo_qdisc_ops, sch->handle, extack);
@@ -652,8 +653,7 @@ atm_tc_dump_class_stats(struct Qdisc *sch, unsigned long arg,
 {
 	struct atm_flow_data *flow = (struct atm_flow_data *)arg;
 
-	if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch),
-				  d, NULL, &flow->bstats) < 0 ||
+	if (gnet_stats_copy_basic(d, NULL, &flow->bstats, true) < 0 ||
 	    gnet_stats_copy_queue(d, NULL, &flow->qstats, flow->q->q.qlen) < 0)
 		return -1;
 

net/sched/sch_cbq.c

@@ -116,7 +116,7 @@ struct cbq_class {
 	long			avgidle;
 	long			deficit;	/* Saved deficit for WRR */
 	psched_time_t		penalized;
-	struct gnet_stats_basic_packed bstats;
+	struct gnet_stats_basic_sync bstats;
 	struct gnet_stats_queue qstats;
 	struct net_rate_estimator __rcu *rate_est;
 	struct tc_cbq_xstats	xstats;
@@ -565,8 +565,7 @@ cbq_update(struct cbq_sched_data *q)
 		long avgidle = cl->avgidle;
 		long idle;
 
-		cl->bstats.packets++;
-		cl->bstats.bytes += len;
+		_bstats_update(&cl->bstats, len, 1);
 
 		/*
 		 * (now - last) is total time between packet right edges.
@@ -1384,8 +1383,7 @@ cbq_dump_class_stats(struct Qdisc *sch, unsigned long arg,
 	if (cl->undertime != PSCHED_PASTPERFECT)
 		cl->xstats.undertime = cl->undertime - q->now;
 
-	if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch),
-				  d, NULL, &cl->bstats) < 0 ||
+	if (gnet_stats_copy_basic(d, NULL, &cl->bstats, true) < 0 ||
 	    gnet_stats_copy_rate_est(d, &cl->rate_est) < 0 ||
 	    gnet_stats_copy_queue(d, NULL, &cl->qstats, qlen) < 0)
 		return -1;
@@ -1519,7 +1517,7 @@ cbq_change_class(struct Qdisc *sch, u32 classid, u32 parentid, struct nlattr **t
 			err = gen_replace_estimator(&cl->bstats, NULL,
 						    &cl->rate_est,
 						    NULL,
-						    qdisc_root_sleeping_running(sch),
+						    true,
 						    tca[TCA_RATE]);
 			if (err) {
 				NL_SET_ERR_MSG(extack, "Failed to replace specified rate estimator");
@@ -1611,6 +1609,7 @@ cbq_change_class(struct Qdisc *sch, u32 classid, u32 parentid, struct nlattr **t
 	if (cl == NULL)
 		goto failure;
 
+	gnet_stats_basic_sync_init(&cl->bstats);
 	err = tcf_block_get(&cl->block, &cl->filter_list, sch, extack);
 	if (err) {
 		kfree(cl);
@@ -1619,9 +1618,7 @@ cbq_change_class(struct Qdisc *sch, u32 classid, u32 parentid, struct nlattr **t
 
 	if (tca[TCA_RATE]) {
 		err = gen_new_estimator(&cl->bstats, NULL, &cl->rate_est,
-					NULL,
-					qdisc_root_sleeping_running(sch),
-					tca[TCA_RATE]);
+					NULL, true, tca[TCA_RATE]);
 		if (err) {
 			NL_SET_ERR_MSG(extack, "Couldn't create new estimator");
 			tcf_block_put(cl->block);

net/sched/sch_drr.c

@@ -19,7 +19,7 @@ struct drr_class {
 	struct Qdisc_class_common	common;
 	unsigned int			filter_cnt;
 
-	struct gnet_stats_basic_packed		bstats;
+	struct gnet_stats_basic_sync		bstats;
 	struct gnet_stats_queue		qstats;
 	struct net_rate_estimator __rcu *rate_est;
 	struct list_head		alist;
@@ -85,8 +85,7 @@ static int drr_change_class(struct Qdisc *sch, u32 classid, u32 parentid,
 		if (tca[TCA_RATE]) {
 			err = gen_replace_estimator(&cl->bstats, NULL,
 						    &cl->rate_est,
-						    NULL,
-						    qdisc_root_sleeping_running(sch),
+						    NULL, true,
 						    tca[TCA_RATE]);
 			if (err) {
 				NL_SET_ERR_MSG(extack, "Failed to replace estimator");
@@ -106,6 +105,7 @@ static int drr_change_class(struct Qdisc *sch, u32 classid, u32 parentid,
 	if (cl == NULL)
 		return -ENOBUFS;
 
+	gnet_stats_basic_sync_init(&cl->bstats);
 	cl->common.classid = classid;
 	cl->quantum	   = quantum;
 	cl->qdisc	   = qdisc_create_dflt(sch->dev_queue,
@@ -118,9 +118,7 @@ static int drr_change_class(struct Qdisc *sch, u32 classid, u32 parentid,
 
 	if (tca[TCA_RATE]) {
 		err = gen_replace_estimator(&cl->bstats, NULL, &cl->rate_est,
-					    NULL,
-					    qdisc_root_sleeping_running(sch),
-					    tca[TCA_RATE]);
+					    NULL, true, tca[TCA_RATE]);
 		if (err) {
 			NL_SET_ERR_MSG(extack, "Failed to replace estimator");
 			qdisc_put(cl->qdisc);
@@ -267,8 +265,7 @@ static int drr_dump_class_stats(struct Qdisc *sch, unsigned long arg,
 	if (qlen)
 		xstats.deficit = cl->deficit;
 
-	if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch),
-				  d, NULL, &cl->bstats) < 0 ||
+	if (gnet_stats_copy_basic(d, NULL, &cl->bstats, true) < 0 ||
 	    gnet_stats_copy_rate_est(d, &cl->rate_est) < 0 ||
 	    gnet_stats_copy_queue(d, cl_q->cpu_qstats, &cl_q->qstats, qlen) < 0)
 		return -1;

net/sched/sch_ets.c

@@ -41,7 +41,7 @@ struct ets_class {
 	struct Qdisc *qdisc;
 	u32 quantum;
 	u32 deficit;
-	struct gnet_stats_basic_packed bstats;
+	struct gnet_stats_basic_sync bstats;
 	struct gnet_stats_queue qstats;
 };
 
@@ -325,8 +325,7 @@ static int ets_class_dump_stats(struct Qdisc *sch, unsigned long arg,
 	struct ets_class *cl = ets_class_from_arg(sch, arg);
 	struct Qdisc *cl_q = cl->qdisc;
 
-	if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch),
-				  d, NULL, &cl_q->bstats) < 0 ||
+	if (gnet_stats_copy_basic(d, NULL, &cl_q->bstats, true) < 0 ||
 	    qdisc_qstats_copy(d, cl_q) < 0)
 		return -1;
 
@@ -689,7 +688,7 @@ static int ets_qdisc_change(struct Qdisc *sch, struct nlattr *opt,
 		q->classes[i].qdisc = NULL;
 		q->classes[i].quantum = 0;
 		q->classes[i].deficit = 0;
-		memset(&q->classes[i].bstats, 0, sizeof(q->classes[i].bstats));
+		gnet_stats_basic_sync_init(&q->classes[i].bstats);
 		memset(&q->classes[i].qstats, 0, sizeof(q->classes[i].qstats));
 	}
 	return 0;

net/sched/sch_generic.c

@@ -304,8 +304,8 @@ static struct sk_buff *dequeue_skb(struct Qdisc *q, bool *validate,
 
 /*
  * Transmit possibly several skbs, and handle the return status as
- * required. Owning running seqcount bit guarantees that
- * only one CPU can execute this function.
+ * required. Owning qdisc running bit guarantees that only one CPU
+ * can execute this function.
  *
  * Returns to the caller:
  *				false  - hardware queue frozen backoff
@@ -606,7 +606,6 @@ struct Qdisc noop_qdisc = {
 	.ops		=	&noop_qdisc_ops,
 	.q.lock		=	__SPIN_LOCK_UNLOCKED(noop_qdisc.q.lock),
 	.dev_queue	=	&noop_netdev_queue,
-	.running	=	SEQCNT_ZERO(noop_qdisc.running),
 	.busylock	=	__SPIN_LOCK_UNLOCKED(noop_qdisc.busylock),
 	.gso_skb = {
 		.next = (struct sk_buff *)&noop_qdisc.gso_skb,
@@ -867,7 +866,6 @@ struct Qdisc_ops pfifo_fast_ops __read_mostly = {
 EXPORT_SYMBOL(pfifo_fast_ops);
 
 static struct lock_class_key qdisc_tx_busylock;
-static struct lock_class_key qdisc_running_key;
 
 struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
 			  const struct Qdisc_ops *ops,
@@ -892,11 +890,12 @@ struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
 	__skb_queue_head_init(&sch->gso_skb);
 	__skb_queue_head_init(&sch->skb_bad_txq);
 	qdisc_skb_head_init(&sch->q);
+	gnet_stats_basic_sync_init(&sch->bstats);
 	spin_lock_init(&sch->q.lock);
 
 	if (ops->static_flags & TCQ_F_CPUSTATS) {
 		sch->cpu_bstats =
-			netdev_alloc_pcpu_stats(struct gnet_stats_basic_cpu);
+			netdev_alloc_pcpu_stats(struct gnet_stats_basic_sync);
 		if (!sch->cpu_bstats)
 			goto errout1;
 
@@ -916,10 +915,6 @@ struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
 	lockdep_set_class(&sch->seqlock,
 			  dev->qdisc_tx_busylock ?: &qdisc_tx_busylock);
 
-	seqcount_init(&sch->running);
-	lockdep_set_class(&sch->running,
-			  dev->qdisc_running_key ?: &qdisc_running_key);
-
 	sch->ops = ops;
 	sch->flags = ops->static_flags;
 	sch->enqueue = ops->enqueue;

net/sched/sch_gred.c

@@ -353,6 +353,7 @@ static int gred_offload_dump_stats(struct Qdisc *sch)
 {
 	struct gred_sched *table = qdisc_priv(sch);
 	struct tc_gred_qopt_offload *hw_stats;
+	u64 bytes = 0, packets = 0;
 	unsigned int i;
 	int ret;
 
@@ -364,9 +365,11 @@ static int gred_offload_dump_stats(struct Qdisc *sch)
 	hw_stats->handle = sch->handle;
 	hw_stats->parent = sch->parent;
 
-	for (i = 0; i < MAX_DPs; i++)
+	for (i = 0; i < MAX_DPs; i++) {
+		gnet_stats_basic_sync_init(&hw_stats->stats.bstats[i]);
 		if (table->tab[i])
 			hw_stats->stats.xstats[i] = &table->tab[i]->stats;
+	}
 
 	ret = qdisc_offload_dump_helper(sch, TC_SETUP_QDISC_GRED, hw_stats);
 	/* Even if driver returns failure adjust the stats - in case offload
@@ -375,19 +378,19 @@ static int gred_offload_dump_stats(struct Qdisc *sch)
 	for (i = 0; i < MAX_DPs; i++) {
 		if (!table->tab[i])
 			continue;
-		table->tab[i]->packetsin += hw_stats->stats.bstats[i].packets;
-		table->tab[i]->bytesin += hw_stats->stats.bstats[i].bytes;
+		table->tab[i]->packetsin += u64_stats_read(&hw_stats->stats.bstats[i].packets);
+		table->tab[i]->bytesin += u64_stats_read(&hw_stats->stats.bstats[i].bytes);
 		table->tab[i]->backlog += hw_stats->stats.qstats[i].backlog;
 
-		_bstats_update(&sch->bstats,
-			       hw_stats->stats.bstats[i].bytes,
-			       hw_stats->stats.bstats[i].packets);
+		bytes += u64_stats_read(&hw_stats->stats.bstats[i].bytes);
+		packets += u64_stats_read(&hw_stats->stats.bstats[i].packets);
 		sch->qstats.qlen += hw_stats->stats.qstats[i].qlen;
 		sch->qstats.backlog += hw_stats->stats.qstats[i].backlog;
 		sch->qstats.drops += hw_stats->stats.qstats[i].drops;
 		sch->qstats.requeues += hw_stats->stats.qstats[i].requeues;
 		sch->qstats.overlimits += hw_stats->stats.qstats[i].overlimits;
 	}
+	_bstats_update(&sch->bstats, bytes, packets);
 
 	kfree(hw_stats);
 	return ret;

net/sched/sch_hfsc.c

@@ -111,7 +111,7 @@ enum hfsc_class_flags {
 struct hfsc_class {
 	struct Qdisc_class_common cl_common;
 
-	struct gnet_stats_basic_packed bstats;
+	struct gnet_stats_basic_sync bstats;
 	struct gnet_stats_queue qstats;
 	struct net_rate_estimator __rcu *rate_est;
 	struct tcf_proto __rcu *filter_list; /* filter list */
@@ -965,7 +965,7 @@ hfsc_change_class(struct Qdisc *sch, u32 classid, u32 parentid,
 			err = gen_replace_estimator(&cl->bstats, NULL,
 						    &cl->rate_est,
 						    NULL,
-						    qdisc_root_sleeping_running(sch),
+						    true,
 						    tca[TCA_RATE]);
 			if (err)
 				return err;
@@ -1033,9 +1033,7 @@ hfsc_change_class(struct Qdisc *sch, u32 classid, u32 parentid,
 
 	if (tca[TCA_RATE]) {
 		err = gen_new_estimator(&cl->bstats, NULL, &cl->rate_est,
-					NULL,
-					qdisc_root_sleeping_running(sch),
-					tca[TCA_RATE]);
+					NULL, true, tca[TCA_RATE]);
 		if (err) {
 			tcf_block_put(cl->block);
 			kfree(cl);
@@ -1328,7 +1326,7 @@ hfsc_dump_class_stats(struct Qdisc *sch, unsigned long arg,
 	xstats.work    = cl->cl_total;
 	xstats.rtwork  = cl->cl_cumul;
 
-	if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch), d, NULL, &cl->bstats) < 0 ||
+	if (gnet_stats_copy_basic(d, NULL, &cl->bstats, true) < 0 ||
 	    gnet_stats_copy_rate_est(d, &cl->rate_est) < 0 ||
 	    gnet_stats_copy_queue(d, NULL, &cl->qstats, qlen) < 0)
 		return -1;
@@ -1406,6 +1404,7 @@ hfsc_init_qdisc(struct Qdisc *sch, struct nlattr *opt,
 	if (err)
 		return err;
 
+	gnet_stats_basic_sync_init(&q->root.bstats);
 	q->root.cl_common.classid = sch->handle;
 	q->root.sched   = q;
 	q->root.qdisc = qdisc_create_dflt(sch->dev_queue, &pfifo_qdisc_ops,

net/sched/sch_htb.c

@@ -113,8 +113,8 @@ struct htb_class {
 	/*
 	 * Written often fields
 	 */
-	struct gnet_stats_basic_packed bstats;
-	struct gnet_stats_basic_packed bstats_bias;
+	struct gnet_stats_basic_sync bstats;
+	struct gnet_stats_basic_sync bstats_bias;
 	struct tc_htb_xstats	xstats;	/* our special stats */
 
 	/* token bucket parameters */
@@ -1308,10 +1308,11 @@ static int htb_dump_class(struct Qdisc *sch, unsigned long arg,
 static void htb_offload_aggregate_stats(struct htb_sched *q,
 					struct htb_class *cl)
 {
+	u64 bytes = 0, packets = 0;
 	struct htb_class *c;
 	unsigned int i;
 
-	memset(&cl->bstats, 0, sizeof(cl->bstats));
+	gnet_stats_basic_sync_init(&cl->bstats);
 
 	for (i = 0; i < q->clhash.hashsize; i++) {
 		hlist_for_each_entry(c, &q->clhash.hash[i], common.hnode) {
@@ -1323,14 +1324,15 @@ static void htb_offload_aggregate_stats(struct htb_sched *q,
 			if (p != cl)
 				continue;
 
-			cl->bstats.bytes += c->bstats_bias.bytes;
-			cl->bstats.packets += c->bstats_bias.packets;
+			bytes += u64_stats_read(&c->bstats_bias.bytes);
+			packets += u64_stats_read(&c->bstats_bias.packets);
 			if (c->level == 0) {
-				cl->bstats.bytes += c->leaf.q->bstats.bytes;
-				cl->bstats.packets += c->leaf.q->bstats.packets;
+				bytes += u64_stats_read(&c->leaf.q->bstats.bytes);
+				packets += u64_stats_read(&c->leaf.q->bstats.packets);
 			}
 		}
 	}
+	_bstats_update(&cl->bstats, bytes, packets);
 }
 
 static int
@@ -1357,16 +1359,16 @@ htb_dump_class_stats(struct Qdisc *sch, unsigned long arg, struct gnet_dump *d)
 			if (cl->leaf.q)
 				cl->bstats = cl->leaf.q->bstats;
 			else
-				memset(&cl->bstats, 0, sizeof(cl->bstats));
-			cl->bstats.bytes += cl->bstats_bias.bytes;
-			cl->bstats.packets += cl->bstats_bias.packets;
+				gnet_stats_basic_sync_init(&cl->bstats);
+			_bstats_update(&cl->bstats,
+				       u64_stats_read(&cl->bstats_bias.bytes),
+				       u64_stats_read(&cl->bstats_bias.packets));
 		} else {
 			htb_offload_aggregate_stats(q, cl);
 		}
 	}
 
-	if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch),
-				  d, NULL, &cl->bstats) < 0 ||
+	if (gnet_stats_copy_basic(d, NULL, &cl->bstats, true) < 0 ||
 	    gnet_stats_copy_rate_est(d, &cl->rate_est) < 0 ||
 	    gnet_stats_copy_queue(d, NULL, &qs, qlen) < 0)
 		return -1;
@@ -1578,8 +1580,9 @@ static int htb_destroy_class_offload(struct Qdisc *sch, struct htb_class *cl,
 		WARN_ON(old != q);
 
 	if (cl->parent) {
-		cl->parent->bstats_bias.bytes += q->bstats.bytes;
-		cl->parent->bstats_bias.packets += q->bstats.packets;
+		_bstats_update(&cl->parent->bstats_bias,
+			       u64_stats_read(&q->bstats.bytes),
+			       u64_stats_read(&q->bstats.packets));
 	}
 
 	offload_opt = (struct tc_htb_qopt_offload) {
@@ -1849,6 +1852,9 @@ static int htb_change_class(struct Qdisc *sch, u32 classid,
 		if (!cl)
 			goto failure;
 
+		gnet_stats_basic_sync_init(&cl->bstats);
+		gnet_stats_basic_sync_init(&cl->bstats_bias);
+
 		err = tcf_block_get(&cl->block, &cl->filter_list, sch, extack);
 		if (err) {
 			kfree(cl);
@@ -1858,7 +1864,7 @@ static int htb_change_class(struct Qdisc *sch, u32 classid,
 			err = gen_new_estimator(&cl->bstats, NULL,
 						&cl->rate_est,
 						NULL,
-						qdisc_root_sleeping_running(sch),
+						true,
 						tca[TCA_RATE] ? : &est.nla);
 			if (err)
 				goto err_block_put;
@@ -1922,8 +1928,9 @@ static int htb_change_class(struct Qdisc *sch, u32 classid,
 				htb_graft_helper(dev_queue, old_q);
 				goto err_kill_estimator;
 			}
-			parent->bstats_bias.bytes += old_q->bstats.bytes;
-			parent->bstats_bias.packets += old_q->bstats.packets;
+			_bstats_update(&parent->bstats_bias,
+				       u64_stats_read(&old_q->bstats.bytes),
+				       u64_stats_read(&old_q->bstats.packets));
 			qdisc_put(old_q);
 		}
 		new_q = qdisc_create_dflt(dev_queue, &pfifo_qdisc_ops,
@@ -1983,7 +1990,7 @@ static int htb_change_class(struct Qdisc *sch, u32 classid,
 			err = gen_replace_estimator(&cl->bstats, NULL,
 						    &cl->rate_est,
 						    NULL,
-						    qdisc_root_sleeping_running(sch),
+						    true,
 						    tca[TCA_RATE]);
 			if (err)
 				return err;

net/sched/sch_mq.c

@@ -130,10 +130,9 @@ static int mq_dump(struct Qdisc *sch, struct sk_buff *skb)
 	struct net_device *dev = qdisc_dev(sch);
 	struct Qdisc *qdisc;
 	unsigned int ntx;
-	__u32 qlen = 0;
 
 	sch->q.qlen = 0;
-	memset(&sch->bstats, 0, sizeof(sch->bstats));
+	gnet_stats_basic_sync_init(&sch->bstats);
 	memset(&sch->qstats, 0, sizeof(sch->qstats));
 
 	/* MQ supports lockless qdiscs. However, statistics accounting needs
@@ -145,25 +144,11 @@ static int mq_dump(struct Qdisc *sch, struct sk_buff *skb)
 		qdisc = netdev_get_tx_queue(dev, ntx)->qdisc_sleeping;
 		spin_lock_bh(qdisc_lock(qdisc));
 
-		if (qdisc_is_percpu_stats(qdisc)) {
-			qlen = qdisc_qlen_sum(qdisc);
-			__gnet_stats_copy_basic(NULL, &sch->bstats,
-						qdisc->cpu_bstats,
-						&qdisc->bstats);
-			__gnet_stats_copy_queue(&sch->qstats,
-						qdisc->cpu_qstats,
-						&qdisc->qstats, qlen);
-			sch->q.qlen		+= qlen;
-		} else {
-			sch->q.qlen		+= qdisc->q.qlen;
-			sch->bstats.bytes	+= qdisc->bstats.bytes;
-			sch->bstats.packets	+= qdisc->bstats.packets;
-			sch->qstats.qlen	+= qdisc->qstats.qlen;
-			sch->qstats.backlog	+= qdisc->qstats.backlog;
-			sch->qstats.drops	+= qdisc->qstats.drops;
-			sch->qstats.requeues	+= qdisc->qstats.requeues;
-			sch->qstats.overlimits	+= qdisc->qstats.overlimits;
-		}
+		gnet_stats_add_basic(&sch->bstats, qdisc->cpu_bstats,
+				     &qdisc->bstats, false);
+		gnet_stats_add_queue(&sch->qstats, qdisc->cpu_qstats,
+				     &qdisc->qstats);
+		sch->q.qlen += qdisc_qlen(qdisc);
 
 		spin_unlock_bh(qdisc_lock(qdisc));
 	}
@@ -246,8 +231,7 @@ static int mq_dump_class_stats(struct Qdisc *sch, unsigned long cl,
 	struct netdev_queue *dev_queue = mq_queue_get(sch, cl);
 
 	sch = dev_queue->qdisc_sleeping;
-	if (gnet_stats_copy_basic(&sch->running, d, sch->cpu_bstats,
-				  &sch->bstats) < 0 ||
+	if (gnet_stats_copy_basic(d, sch->cpu_bstats, &sch->bstats, true) < 0 ||
 	    qdisc_qstats_copy(d, sch) < 0)
 		return -1;
 	return 0;

net/sched/sch_mqprio.c

@@ -390,7 +390,7 @@ static int mqprio_dump(struct Qdisc *sch, struct sk_buff *skb)
 	unsigned int ntx, tc;
 
 	sch->q.qlen = 0;
-	memset(&sch->bstats, 0, sizeof(sch->bstats));
+	gnet_stats_basic_sync_init(&sch->bstats);
 	memset(&sch->qstats, 0, sizeof(sch->qstats));
 
 	/* MQ supports lockless qdiscs. However, statistics accounting needs
@@ -402,25 +402,11 @@ static int mqprio_dump(struct Qdisc *sch, struct sk_buff *skb)
 		qdisc = netdev_get_tx_queue(dev, ntx)->qdisc_sleeping;
 		spin_lock_bh(qdisc_lock(qdisc));
 
-		if (qdisc_is_percpu_stats(qdisc)) {
-			__u32 qlen = qdisc_qlen_sum(qdisc);
-
-			__gnet_stats_copy_basic(NULL, &sch->bstats,
-						qdisc->cpu_bstats,
-						&qdisc->bstats);
-			__gnet_stats_copy_queue(&sch->qstats,
-						qdisc->cpu_qstats,
-						&qdisc->qstats, qlen);
-			sch->q.qlen		+= qlen;
-		} else {
-			sch->q.qlen		+= qdisc->q.qlen;
-			sch->bstats.bytes	+= qdisc->bstats.bytes;
-			sch->bstats.packets	+= qdisc->bstats.packets;
-			sch->qstats.backlog	+= qdisc->qstats.backlog;
-			sch->qstats.drops	+= qdisc->qstats.drops;
-			sch->qstats.requeues	+= qdisc->qstats.requeues;
-			sch->qstats.overlimits	+= qdisc->qstats.overlimits;
-		}
+		gnet_stats_add_basic(&sch->bstats, qdisc->cpu_bstats,
+				     &qdisc->bstats, false);
+		gnet_stats_add_queue(&sch->qstats, qdisc->cpu_qstats,
+				     &qdisc->qstats);
+		sch->q.qlen += qdisc_qlen(qdisc);
 
 		spin_unlock_bh(qdisc_lock(qdisc));
 	}
@@ -512,12 +498,13 @@ static int mqprio_dump_class_stats(struct Qdisc *sch, unsigned long cl,
 {
 	if (cl >= TC_H_MIN_PRIORITY) {
 		int i;
-		__u32 qlen = 0;
+		__u32 qlen;
 		struct gnet_stats_queue qstats = {0};
-		struct gnet_stats_basic_packed bstats = {0};
+		struct gnet_stats_basic_sync bstats;
 		struct net_device *dev = qdisc_dev(sch);
 		struct netdev_tc_txq tc = dev->tc_to_txq[cl & TC_BITMASK];
 
+		gnet_stats_basic_sync_init(&bstats);
 		/* Drop lock here it will be reclaimed before touching
 		 * statistics this is required because the d->lock we
 		 * hold here is the look on dev_queue->qdisc_sleeping
@@ -532,40 +519,28 @@ static int mqprio_dump_class_stats(struct Qdisc *sch, unsigned long cl,
 
 			spin_lock_bh(qdisc_lock(qdisc));
 
-			if (qdisc_is_percpu_stats(qdisc)) {
-				qlen = qdisc_qlen_sum(qdisc);
-
-				__gnet_stats_copy_basic(NULL, &bstats,
-							qdisc->cpu_bstats,
-							&qdisc->bstats);
-				__gnet_stats_copy_queue(&qstats,
-							qdisc->cpu_qstats,
-							&qdisc->qstats,
-							qlen);
-			} else {
-				qlen		+= qdisc->q.qlen;
-				bstats.bytes	+= qdisc->bstats.bytes;
-				bstats.packets	+= qdisc->bstats.packets;
-				qstats.backlog	+= qdisc->qstats.backlog;
-				qstats.drops	+= qdisc->qstats.drops;
-				qstats.requeues	+= qdisc->qstats.requeues;
-				qstats.overlimits += qdisc->qstats.overlimits;
-			}
+			gnet_stats_add_basic(&bstats, qdisc->cpu_bstats,
+					     &qdisc->bstats, false);
+			gnet_stats_add_queue(&qstats, qdisc->cpu_qstats,
+					     &qdisc->qstats);
+			sch->q.qlen += qdisc_qlen(qdisc);
+
 			spin_unlock_bh(qdisc_lock(qdisc));
 		}
+		qlen = qdisc_qlen(sch) + qstats.qlen;
 
 		/* Reclaim root sleeping lock before completing stats */
 		if (d->lock)
 			spin_lock_bh(d->lock);
-		if (gnet_stats_copy_basic(NULL, d, NULL, &bstats) < 0 ||
+		if (gnet_stats_copy_basic(d, NULL, &bstats, false) < 0 ||
 		    gnet_stats_copy_queue(d, NULL, &qstats, qlen) < 0)
 			return -1;
 	} else {
 		struct netdev_queue *dev_queue = mqprio_queue_get(sch, cl);
 
 		sch = dev_queue->qdisc_sleeping;
-		if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch), d,
-					  sch->cpu_bstats, &sch->bstats) < 0 ||
+		if (gnet_stats_copy_basic(d, sch->cpu_bstats,
+					  &sch->bstats, true) < 0 ||
 		    qdisc_qstats_copy(d, sch) < 0)
 			return -1;
 	}

net/sched/sch_multiq.c

@@ -338,8 +338,7 @@ static int multiq_dump_class_stats(struct Qdisc *sch, unsigned long cl,
 	struct Qdisc *cl_q;
 
 	cl_q = q->queues[cl - 1];
-	if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch),
-				  d, cl_q->cpu_bstats, &cl_q->bstats) < 0 ||
+	if (gnet_stats_copy_basic(d, cl_q->cpu_bstats, &cl_q->bstats, true) < 0 ||
 	    qdisc_qstats_copy(d, cl_q) < 0)
 		return -1;
 

net/sched/sch_prio.c

@@ -361,8 +361,8 @@ static int prio_dump_class_stats(struct Qdisc *sch, unsigned long cl,
 	struct Qdisc *cl_q;
 
 	cl_q = q->queues[cl - 1];
-	if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch),
-				  d, cl_q->cpu_bstats, &cl_q->bstats) < 0 ||
+	if (gnet_stats_copy_basic(d, cl_q->cpu_bstats,
+				  &cl_q->bstats, true) < 0 ||
 	    qdisc_qstats_copy(d, cl_q) < 0)
 		return -1;
 

net/sched/sch_qfq.c

@@ -131,7 +131,7 @@ struct qfq_class {
 
 	unsigned int filter_cnt;
 
-	struct gnet_stats_basic_packed bstats;
+	struct gnet_stats_basic_sync bstats;
 	struct gnet_stats_queue qstats;
 	struct net_rate_estimator __rcu *rate_est;
 	struct Qdisc *qdisc;
@@ -451,7 +451,7 @@ static int qfq_change_class(struct Qdisc *sch, u32 classid, u32 parentid,
 			err = gen_replace_estimator(&cl->bstats, NULL,
 						    &cl->rate_est,
 						    NULL,
-						    qdisc_root_sleeping_running(sch),
+						    true,
 						    tca[TCA_RATE]);
 			if (err)
 				return err;
@@ -465,6 +465,7 @@ static int qfq_change_class(struct Qdisc *sch, u32 classid, u32 parentid,
 	if (cl == NULL)
 		return -ENOBUFS;
 
+	gnet_stats_basic_sync_init(&cl->bstats);
 	cl->common.classid = classid;
 	cl->deficit = lmax;
 
@@ -477,7 +478,7 @@ static int qfq_change_class(struct Qdisc *sch, u32 classid, u32 parentid,
 		err = gen_new_estimator(&cl->bstats, NULL,
 					&cl->rate_est,
 					NULL,
-					qdisc_root_sleeping_running(sch),
+					true,
 					tca[TCA_RATE]);
 		if (err)
 			goto destroy_class;
@@ -639,8 +640,7 @@ static int qfq_dump_class_stats(struct Qdisc *sch, unsigned long arg,
 	xstats.weight = cl->agg->class_weight;
 	xstats.lmax = cl->agg->lmax;
 
-	if (gnet_stats_copy_basic(qdisc_root_sleeping_running(sch),
-				  d, NULL, &cl->bstats) < 0 ||
+	if (gnet_stats_copy_basic(d, NULL, &cl->bstats, true) < 0 ||
 	    gnet_stats_copy_rate_est(d, &cl->rate_est) < 0 ||
 	    qdisc_qstats_copy(d, cl->qdisc) < 0)
 		return -1;
@@ -1234,8 +1234,7 @@ static int qfq_enqueue(struct sk_buff *skb, struct Qdisc *sch,
 		return err;
 	}
 
-	cl->bstats.bytes += len;
-	cl->bstats.packets += gso_segs;
+	_bstats_update(&cl->bstats, len, gso_segs);
 	sch->qstats.backlog += len;
 	++sch->q.qlen;
 

net/sched/sch_taprio.c

@@ -1977,7 +1977,7 @@ static int taprio_dump_class_stats(struct Qdisc *sch, unsigned long cl,
 	struct netdev_queue *dev_queue = taprio_queue_get(sch, cl);
 
 	sch = dev_queue->qdisc_sleeping;
-	if (gnet_stats_copy_basic(&sch->running, d, NULL, &sch->bstats) < 0 ||
+	if (gnet_stats_copy_basic(d, NULL, &sch->bstats, true) < 0 ||
 	    qdisc_qstats_copy(d, sch) < 0)
 		return -1;
 	return 0;