net: adopt u64_stats_t in struct pcpu_sw_netstats

As explained in commit 316580b6 ("u64_stats: provide u64_stats_t type")
we should use u64_stats_t and related accessors to avoid load/store tearing.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

drivers/net/macsec.c

@@ -523,8 +523,8 @@ static void count_tx(struct net_device *dev, int ret, int len)
 		struct pcpu_sw_netstats *stats = this_cpu_ptr(dev->tstats);
 
 		u64_stats_update_begin(&stats->syncp);
-		stats->tx_packets++;
-		stats->tx_bytes += len;
+		u64_stats_inc(&stats->tx_packets);
+		u64_stats_add(&stats->tx_bytes, len);
 		u64_stats_update_end(&stats->syncp);
 	}
 }
@@ -825,8 +825,8 @@ static void count_rx(struct net_device *dev, int len)
 	struct pcpu_sw_netstats *stats = this_cpu_ptr(dev->tstats);
 
 	u64_stats_update_begin(&stats->syncp);
-	stats->rx_packets++;
-	stats->rx_bytes += len;
+	u64_stats_inc(&stats->rx_packets);
+	u64_stats_add(&stats->rx_bytes, len);
 	u64_stats_update_end(&stats->syncp);
 }
 

drivers/net/usb/usbnet.c

@@ -337,8 +337,8 @@ void usbnet_skb_return (struct usbnet *dev, struct sk_buff *skb)
 		skb->protocol = eth_type_trans (skb, dev->net);
 
 	flags = u64_stats_update_begin_irqsave(&stats64->syncp);
-	stats64->rx_packets++;
-	stats64->rx_bytes += skb->len;
+	u64_stats_inc(&stats64->rx_packets);
+	u64_stats_add(&stats64->rx_bytes, skb->len);
 	u64_stats_update_end_irqrestore(&stats64->syncp, flags);
 
 	netif_dbg(dev, rx_status, dev->net, "< rx, len %zu, type 0x%x\n",
@@ -1258,8 +1258,8 @@ static void tx_complete (struct urb *urb)
 		unsigned long flags;
 
 		flags = u64_stats_update_begin_irqsave(&stats64->syncp);
-		stats64->tx_packets += entry->packets;
-		stats64->tx_bytes += entry->length;
+		u64_stats_add(&stats64->tx_packets, entry->packets);
+		u64_stats_add(&stats64->tx_bytes, entry->length);
 		u64_stats_update_end_irqrestore(&stats64->syncp, flags);
 	} else {
 		dev->net->stats.tx_errors++;

drivers/net/vxlan/vxlan_core.c

@@ -2385,15 +2385,15 @@ static void vxlan_encap_bypass(struct sk_buff *skb, struct vxlan_dev *src_vxlan,
 		vxlan_snoop(dev, &loopback, eth_hdr(skb)->h_source, 0, vni);
 
 	u64_stats_update_begin(&tx_stats->syncp);
-	tx_stats->tx_packets++;
-	tx_stats->tx_bytes += len;
+	u64_stats_inc(&tx_stats->tx_packets);
+	u64_stats_add(&tx_stats->tx_bytes, len);
 	u64_stats_update_end(&tx_stats->syncp);
 	vxlan_vnifilter_count(src_vxlan, vni, NULL, VXLAN_VNI_STATS_TX, len);
 
 	if (__netif_rx(skb) == NET_RX_SUCCESS) {
 		u64_stats_update_begin(&rx_stats->syncp);
-		rx_stats->rx_packets++;
-		rx_stats->rx_bytes += len;
+		u64_stats_inc(&rx_stats->rx_packets);
+		u64_stats_add(&rx_stats->rx_bytes, len);
 		u64_stats_update_end(&rx_stats->syncp);
 		vxlan_vnifilter_count(dst_vxlan, vni, NULL, VXLAN_VNI_STATS_RX,
 				      len);

include/linux/netdevice.h

@@ -2636,10 +2636,10 @@ struct packet_offload {
 
 /* often modified stats are per-CPU, other are shared (netdev->stats) */
 struct pcpu_sw_netstats {
-	u64     rx_packets;
-	u64     rx_bytes;
-	u64     tx_packets;
-	u64     tx_bytes;
+	u64_stats_t		rx_packets;
+	u64_stats_t		rx_bytes;
+	u64_stats_t		tx_packets;
+	u64_stats_t		tx_bytes;
 	struct u64_stats_sync   syncp;
 } __aligned(4 * sizeof(u64));
 
@@ -2656,8 +2656,8 @@ static inline void dev_sw_netstats_rx_add(struct net_device *dev, unsigned int l
 	struct pcpu_sw_netstats *tstats = this_cpu_ptr(dev->tstats);
 
 	u64_stats_update_begin(&tstats->syncp);
-	tstats->rx_bytes += len;
-	tstats->rx_packets++;
+	u64_stats_add(&tstats->rx_bytes, len);
+	u64_stats_inc(&tstats->rx_packets);
 	u64_stats_update_end(&tstats->syncp);
 }
 
@@ -2668,8 +2668,8 @@ static inline void dev_sw_netstats_tx_add(struct net_device *dev,
 	struct pcpu_sw_netstats *tstats = this_cpu_ptr(dev->tstats);
 
 	u64_stats_update_begin(&tstats->syncp);
-	tstats->tx_bytes += len;
-	tstats->tx_packets += packets;
+	u64_stats_add(&tstats->tx_bytes, len);
+	u64_stats_add(&tstats->tx_packets, packets);
 	u64_stats_update_end(&tstats->syncp);
 }
 

include/net/ip_tunnels.h

@@ -456,8 +456,8 @@ static inline void iptunnel_xmit_stats(struct net_device *dev, int pkt_len)
 		struct pcpu_sw_netstats *tstats = get_cpu_ptr(dev->tstats);
 
 		u64_stats_update_begin(&tstats->syncp);
-		tstats->tx_bytes += pkt_len;
-		tstats->tx_packets++;
+		u64_stats_add(&tstats->tx_bytes, pkt_len);
+		u64_stats_inc(&tstats->tx_packets);
 		u64_stats_update_end(&tstats->syncp);
 		put_cpu_ptr(tstats);
 	} else {

net/bridge/br_netlink.c

@@ -1770,10 +1770,10 @@ static int br_fill_linkxstats(struct sk_buff *skb,
 			if (v->vid == pvid)
 				vxi.flags |= BRIDGE_VLAN_INFO_PVID;
 			br_vlan_get_stats(v, &stats);
-			vxi.rx_bytes = stats.rx_bytes;
-			vxi.rx_packets = stats.rx_packets;
-			vxi.tx_bytes = stats.tx_bytes;
-			vxi.tx_packets = stats.tx_packets;
+			vxi.rx_bytes = u64_stats_read(&stats.rx_bytes);
+			vxi.rx_packets = u64_stats_read(&stats.rx_packets);
+			vxi.tx_bytes = u64_stats_read(&stats.tx_bytes);
+			vxi.tx_packets = u64_stats_read(&stats.tx_packets);
 
 			if (nla_put(skb, BRIDGE_XSTATS_VLAN, sizeof(vxi), &vxi))
 				goto nla_put_failure;

net/bridge/br_vlan.c

@@ -505,8 +505,8 @@ struct sk_buff *br_handle_vlan(struct net_bridge *br,
 	if (br_opt_get(br, BROPT_VLAN_STATS_ENABLED)) {
 		stats = this_cpu_ptr(v->stats);
 		u64_stats_update_begin(&stats->syncp);
-		stats->tx_bytes += skb->len;
-		stats->tx_packets++;
+		u64_stats_add(&stats->tx_bytes, skb->len);
+		u64_stats_inc(&stats->tx_packets);
 		u64_stats_update_end(&stats->syncp);
 	}
 
@@ -624,8 +624,8 @@ static bool __allowed_ingress(const struct net_bridge *br,
 	if (br_opt_get(br, BROPT_VLAN_STATS_ENABLED)) {
 		stats = this_cpu_ptr(v->stats);
 		u64_stats_update_begin(&stats->syncp);
-		stats->rx_bytes += skb->len;
-		stats->rx_packets++;
+		u64_stats_add(&stats->rx_bytes, skb->len);
+		u64_stats_inc(&stats->rx_packets);
 		u64_stats_update_end(&stats->syncp);
 	}
 
@@ -1379,16 +1379,16 @@ void br_vlan_get_stats(const struct net_bridge_vlan *v,
 		cpu_stats = per_cpu_ptr(v->stats, i);
 		do {
 			start = u64_stats_fetch_begin_irq(&cpu_stats->syncp);
-			rxpackets = cpu_stats->rx_packets;
-			rxbytes = cpu_stats->rx_bytes;
-			txbytes = cpu_stats->tx_bytes;
-			txpackets = cpu_stats->tx_packets;
+			rxpackets = u64_stats_read(&cpu_stats->rx_packets);
+			rxbytes = u64_stats_read(&cpu_stats->rx_bytes);
+			txbytes = u64_stats_read(&cpu_stats->tx_bytes);
+			txpackets = u64_stats_read(&cpu_stats->tx_packets);
 		} while (u64_stats_fetch_retry_irq(&cpu_stats->syncp, start));
 
-		stats->rx_packets += rxpackets;
-		stats->rx_bytes += rxbytes;
-		stats->tx_bytes += txbytes;
-		stats->tx_packets += txpackets;
+		u64_stats_add(&stats->rx_packets, rxpackets);
+		u64_stats_add(&stats->rx_bytes, rxbytes);
+		u64_stats_add(&stats->tx_bytes, txbytes);
+		u64_stats_add(&stats->tx_packets, txpackets);
 	}
 }
 
@@ -1779,14 +1779,18 @@ static bool br_vlan_stats_fill(struct sk_buff *skb,
 		return false;
 
 	br_vlan_get_stats(v, &stats);
-	if (nla_put_u64_64bit(skb, BRIDGE_VLANDB_STATS_RX_BYTES, stats.rx_bytes,
+	if (nla_put_u64_64bit(skb, BRIDGE_VLANDB_STATS_RX_BYTES,
+			      u64_stats_read(&stats.rx_bytes),
 			      BRIDGE_VLANDB_STATS_PAD) ||
 	    nla_put_u64_64bit(skb, BRIDGE_VLANDB_STATS_RX_PACKETS,
-			      stats.rx_packets, BRIDGE_VLANDB_STATS_PAD) ||
-	    nla_put_u64_64bit(skb, BRIDGE_VLANDB_STATS_TX_BYTES, stats.tx_bytes,
+			      u64_stats_read(&stats.rx_packets),
+			      BRIDGE_VLANDB_STATS_PAD) ||
+	    nla_put_u64_64bit(skb, BRIDGE_VLANDB_STATS_TX_BYTES,
+			      u64_stats_read(&stats.tx_bytes),
 			      BRIDGE_VLANDB_STATS_PAD) ||
 	    nla_put_u64_64bit(skb, BRIDGE_VLANDB_STATS_TX_PACKETS,
-			      stats.tx_packets, BRIDGE_VLANDB_STATS_PAD))
+			      u64_stats_read(&stats.tx_packets),
+			      BRIDGE_VLANDB_STATS_PAD))
 		goto out_err;
 
 	nla_nest_end(skb, nest);

net/core/dev.c

@@ -10459,23 +10459,23 @@ void dev_fetch_sw_netstats(struct rtnl_link_stats64 *s,
 	int cpu;
 
 	for_each_possible_cpu(cpu) {
+		u64 rx_packets, rx_bytes, tx_packets, tx_bytes;
 		const struct pcpu_sw_netstats *stats;
-		struct pcpu_sw_netstats tmp;
 		unsigned int start;
 
 		stats = per_cpu_ptr(netstats, cpu);
 		do {
 			start = u64_stats_fetch_begin_irq(&stats->syncp);
-			tmp.rx_packets = stats->rx_packets;
-			tmp.rx_bytes   = stats->rx_bytes;
-			tmp.tx_packets = stats->tx_packets;
-			tmp.tx_bytes   = stats->tx_bytes;
+			rx_packets = u64_stats_read(&stats->rx_packets);
+			rx_bytes   = u64_stats_read(&stats->rx_bytes);
+			tx_packets = u64_stats_read(&stats->tx_packets);
+			tx_bytes   = u64_stats_read(&stats->tx_bytes);
 		} while (u64_stats_fetch_retry_irq(&stats->syncp, start));
 
-		s->rx_packets += tmp.rx_packets;
-		s->rx_bytes   += tmp.rx_bytes;
-		s->tx_packets += tmp.tx_packets;
-		s->tx_bytes   += tmp.tx_bytes;
+		s->rx_packets += rx_packets;
+		s->rx_bytes   += rx_bytes;
+		s->tx_packets += tx_packets;
+		s->tx_bytes   += tx_bytes;
 	}
 }
 EXPORT_SYMBOL_GPL(dev_fetch_sw_netstats);

net/dsa/slave.c

@@ -935,10 +935,10 @@ static void dsa_slave_get_ethtool_stats(struct net_device *dev,
 		s = per_cpu_ptr(dev->tstats, i);
 		do {
 			start = u64_stats_fetch_begin_irq(&s->syncp);
-			tx_packets = s->tx_packets;
-			tx_bytes = s->tx_bytes;
-			rx_packets = s->rx_packets;
-			rx_bytes = s->rx_bytes;
+			tx_packets = u64_stats_read(&s->tx_packets);
+			tx_bytes = u64_stats_read(&s->tx_bytes);
+			rx_packets = u64_stats_read(&s->rx_packets);
+			rx_bytes = u64_stats_read(&s->rx_bytes);
 		} while (u64_stats_fetch_retry_irq(&s->syncp, start));
 		data[0] += tx_packets;
 		data[1] += tx_bytes;