net: Use queue aware tests throughout.

This effectively "flips the switch" by making the core networking
and multiqueue-aware drivers use the new TX multiqueue structures.

Non-multiqueue drivers need no changes.  The interfaces they use such
as netif_stop_queue() degenerate into an operation on TX queue zero.
So everything "just works" for them.

Code that really wants to do "X" to all TX queues now invokes a
routine that does so, such as netif_tx_wake_all_queues(),
netif_tx_stop_all_queues(), etc.

pktgen and netpoll required a little bit more surgery than the others.

In particular the pktgen changes, whilst functional, could be largely
improved.  The initial check in pktgen_xmit() will sometimes check the
wrong queue, which is mostly harmless.  The thing to do is probably to
invoke fill_packet() earlier.

The bulk of the netpoll changes is to make the code operate solely on
the TX queue indicated by by the SKB queue mapping.

Setting of the SKB queue mapping is entirely confined inside of
net/core/dev.c:dev_pick_tx().  If we end up needing any kind of
special semantics (drops, for example) it will be implemented here.

Finally, we now have a "real_num_tx_queues" which is where the driver
indicates how many TX queues are actually active.

With IGB changes from Jeff Kirsher.

Signed-off-by: David S. Miller <davem@davemloft.net>

5 files changed, 74 insertions, 58 deletions

diff --git a/net/core/dev.c b/net/core/dev.c
index 69378f250695..f027a1ac4fbb 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -1598,7 +1598,8 @@ static int dev_gso_segment(struct sk_buff *skb)
 	return 0;
 }
 
-int dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev)
+int dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev,
+			struct netdev_queue *txq)
 {
 	if (likely(!skb->next)) {
 		if (!list_empty(&ptype_all))
@@ -1627,9 +1628,7 @@ gso:
 			skb->next = nskb;
 			return rc;
 		}
-		if (unlikely((netif_queue_stopped(dev) ||
-			     netif_subqueue_stopped(dev, skb)) &&
-			     skb->next))
+		if (unlikely(netif_tx_queue_stopped(txq) && skb->next))
 			return NETDEV_TX_BUSY;
 	} while (skb->next);
 
@@ -1669,7 +1668,10 @@ out_kfree_skb:
 static struct netdev_queue *dev_pick_tx(struct net_device *dev,
 					struct sk_buff *skb)
 {
-	return netdev_get_tx_queue(dev, 0);
+	u16 queue_index = 0;
+
+	skb_set_queue_mapping(skb, queue_index);
+	return netdev_get_tx_queue(dev, queue_index);
 }
 
 int dev_queue_xmit(struct sk_buff *skb)
@@ -1737,8 +1739,6 @@ gso:
 		spin_lock(&txq->lock);
 		q = txq->qdisc;
 		if (q->enqueue) {
-			/* reset queue_mapping to zero */
-			skb_set_queue_mapping(skb, 0);
 			rc = q->enqueue(skb, q);
 			qdisc_run(txq);
 			spin_unlock(&txq->lock);
@@ -1768,10 +1768,9 @@ gso:
 
 			HARD_TX_LOCK(dev, txq, cpu);
 
-			if (!netif_queue_stopped(dev) &&
-			    !netif_subqueue_stopped(dev, skb)) {
+			if (!netif_tx_queue_stopped(txq)) {
 				rc = 0;
-				if (!dev_hard_start_xmit(skb, dev)) {
+				if (!dev_hard_start_xmit(skb, dev, txq)) {
 					HARD_TX_UNLOCK(dev, txq);
 					goto out;
 				}
@@ -4160,8 +4159,7 @@ struct net_device *alloc_netdev_mq(int sizeof_priv, const char *name,
 
 	BUG_ON(strlen(name) >= sizeof(dev->name));
 
-	alloc_size = sizeof(struct net_device) +
-		     sizeof(struct net_device_subqueue) * (queue_count - 1);
+	alloc_size = sizeof(struct net_device);
 	if (sizeof_priv) {
 		/* ensure 32-byte alignment of private area */
 		alloc_size = (alloc_size + NETDEV_ALIGN_CONST) & ~NETDEV_ALIGN_CONST;
@@ -4191,16 +4189,14 @@ struct net_device *alloc_netdev_mq(int sizeof_priv, const char *name,
 
 	dev->_tx = tx;
 	dev->num_tx_queues = queue_count;
+	dev->real_num_tx_queues = queue_count;
 
 	if (sizeof_priv) {
 		dev->priv = ((char *)dev +
-			     ((sizeof(struct net_device) +
-			       (sizeof(struct net_device_subqueue) *
-				(queue_count - 1)) + NETDEV_ALIGN_CONST)
+			     ((sizeof(struct net_device) + NETDEV_ALIGN_CONST)
 			      & ~NETDEV_ALIGN_CONST));
 	}
 
-	dev->egress_subqueue_count = queue_count;
 	dev->gso_max_size = GSO_MAX_SIZE;
 
 	netdev_init_queues(dev);
diff --git a/net/core/netpoll.c b/net/core/netpoll.c
index 8fb134da0346..c12720895ecf 100644
--- a/net/core/netpoll.c
+++ b/net/core/netpoll.c
@@ -58,25 +58,27 @@ static void queue_process(struct work_struct *work)
 
 	while ((skb = skb_dequeue(&npinfo->txq))) {
 		struct net_device *dev = skb->dev;
+		struct netdev_queue *txq;
 
 		if (!netif_device_present(dev) || !netif_running(dev)) {
 			__kfree_skb(skb);
 			continue;
 		}
 
+		txq = netdev_get_tx_queue(dev, skb_get_queue_mapping(skb));
+
 		local_irq_save(flags);
-		netif_tx_lock(dev);
-		if ((netif_queue_stopped(dev) ||
-		     netif_subqueue_stopped(dev, skb)) ||
-		     dev->hard_start_xmit(skb, dev) != NETDEV_TX_OK) {
+		__netif_tx_lock(txq, smp_processor_id());
+		if (netif_tx_queue_stopped(txq) ||
+		    dev->hard_start_xmit(skb, dev) != NETDEV_TX_OK) {
 			skb_queue_head(&npinfo->txq, skb);
-			netif_tx_unlock(dev);
+			__netif_tx_unlock(txq);
 			local_irq_restore(flags);
 
 			schedule_delayed_work(&npinfo->tx_work, HZ/10);
 			return;
 		}
-		netif_tx_unlock(dev);
+		__netif_tx_unlock(txq);
 		local_irq_restore(flags);
 	}
 }
@@ -278,17 +280,19 @@ static void netpoll_send_skb(struct netpoll *np, struct sk_buff *skb)
 
 	/* don't get messages out of order, and no recursion */
 	if (skb_queue_len(&npinfo->txq) == 0 && !netpoll_owner_active(dev)) {
+		struct netdev_queue *txq;
 		unsigned long flags;
 
+		txq = netdev_get_tx_queue(dev, skb_get_queue_mapping(skb));
+
 		local_irq_save(flags);
 		/* try until next clock tick */
 		for (tries = jiffies_to_usecs(1)/USEC_PER_POLL;
 		     tries > 0; --tries) {
-			if (netif_tx_trylock(dev)) {
-				if (!netif_queue_stopped(dev) &&
-				    !netif_subqueue_stopped(dev, skb))
+			if (__netif_tx_trylock(txq)) {
+				if (!netif_tx_queue_stopped(txq))
 					status = dev->hard_start_xmit(skb, dev);
-				netif_tx_unlock(dev);
+				__netif_tx_unlock(txq);
 
 				if (status == NETDEV_TX_OK)
 					break;
diff --git a/net/core/pktgen.c b/net/core/pktgen.c
index fdf537707e51..906802db4ed4 100644
--- a/net/core/pktgen.c
+++ b/net/core/pktgen.c
@@ -2123,6 +2123,24 @@ static void get_ipsec_sa(struct pktgen_dev *pkt_dev, int flow)
 	}
 }
 #endif
+static void set_cur_queue_map(struct pktgen_dev *pkt_dev)
+{
+	if (pkt_dev->queue_map_min < pkt_dev->queue_map_max) {
+		__u16 t;
+		if (pkt_dev->flags & F_QUEUE_MAP_RND) {
+			t = random32() %
+				(pkt_dev->queue_map_max -
+				 pkt_dev->queue_map_min + 1)
+				+ pkt_dev->queue_map_min;
+		} else {
+			t = pkt_dev->cur_queue_map + 1;
+			if (t > pkt_dev->queue_map_max)
+				t = pkt_dev->queue_map_min;
+		}
+		pkt_dev->cur_queue_map = t;
+	}
+}
+
 /* Increment/randomize headers according to flags and current values
  * for IP src/dest, UDP src/dst port, MAC-Addr src/dst
  */
@@ -2325,19 +2343,7 @@ static void mod_cur_headers(struct pktgen_dev *pkt_dev)
 		pkt_dev->cur_pkt_size = t;
 	}
 
-	if (pkt_dev->queue_map_min < pkt_dev->queue_map_max) {
-		__u16 t;
-		if (pkt_dev->flags & F_QUEUE_MAP_RND) {
-			t = random32() %
-				(pkt_dev->queue_map_max - pkt_dev->queue_map_min + 1)
-				+ pkt_dev->queue_map_min;
-		} else {
-			t = pkt_dev->cur_queue_map + 1;
-			if (t > pkt_dev->queue_map_max)
-				t = pkt_dev->queue_map_min;
-		}
-		pkt_dev->cur_queue_map = t;
-	}
+	set_cur_queue_map(pkt_dev);
 
 	pkt_dev->flows[flow].count++;
 }
@@ -2458,7 +2464,7 @@ static struct sk_buff *fill_packet_ipv4(struct net_device *odev,
 	__be16 *vlan_encapsulated_proto = NULL;  /* packet type ID field (or len) for VLAN tag */
 	__be16 *svlan_tci = NULL;                /* Encapsulates priority and SVLAN ID */
 	__be16 *svlan_encapsulated_proto = NULL; /* packet type ID field (or len) for SVLAN tag */
-
+	u16 queue_map;
 
 	if (pkt_dev->nr_labels)
 		protocol = htons(ETH_P_MPLS_UC);
@@ -2469,6 +2475,7 @@ static struct sk_buff *fill_packet_ipv4(struct net_device *odev,
 	/* Update any of the values, used when we're incrementing various
 	 * fields.
 	 */
+	queue_map = pkt_dev->cur_queue_map;
 	mod_cur_headers(pkt_dev);
 
 	datalen = (odev->hard_header_len + 16) & ~0xf;
@@ -2507,7 +2514,7 @@ static struct sk_buff *fill_packet_ipv4(struct net_device *odev,
 	skb->network_header = skb->tail;
 	skb->transport_header = skb->network_header + sizeof(struct iphdr);
 	skb_put(skb, sizeof(struct iphdr) + sizeof(struct udphdr));
-	skb_set_queue_mapping(skb, pkt_dev->cur_queue_map);
+	skb_set_queue_mapping(skb, queue_map);
 	iph = ip_hdr(skb);
 	udph = udp_hdr(skb);
 
@@ -2797,6 +2804,7 @@ static struct sk_buff *fill_packet_ipv6(struct net_device *odev,
 	__be16 *vlan_encapsulated_proto = NULL;  /* packet type ID field (or len) for VLAN tag */
 	__be16 *svlan_tci = NULL;                /* Encapsulates priority and SVLAN ID */
 	__be16 *svlan_encapsulated_proto = NULL; /* packet type ID field (or len) for SVLAN tag */
+	u16 queue_map;
 
 	if (pkt_dev->nr_labels)
 		protocol = htons(ETH_P_MPLS_UC);
@@ -2807,6 +2815,7 @@ static struct sk_buff *fill_packet_ipv6(struct net_device *odev,
 	/* Update any of the values, used when we're incrementing various
 	 * fields.
 	 */
+	queue_map = pkt_dev->cur_queue_map;
 	mod_cur_headers(pkt_dev);
 
 	skb = alloc_skb(pkt_dev->cur_pkt_size + 64 + 16 +
@@ -2844,7 +2853,7 @@ static struct sk_buff *fill_packet_ipv6(struct net_device *odev,
 	skb->network_header = skb->tail;
 	skb->transport_header = skb->network_header + sizeof(struct ipv6hdr);
 	skb_put(skb, sizeof(struct ipv6hdr) + sizeof(struct udphdr));
-	skb_set_queue_mapping(skb, pkt_dev->cur_queue_map);
+	skb_set_queue_mapping(skb, queue_map);
 	iph = ipv6_hdr(skb);
 	udph = udp_hdr(skb);
 
@@ -3263,7 +3272,9 @@ static void pktgen_rem_thread(struct pktgen_thread *t)
 static __inline__ void pktgen_xmit(struct pktgen_dev *pkt_dev)
 {
 	struct net_device *odev = NULL;
+	struct netdev_queue *txq;
 	__u64 idle_start = 0;
+	u16 queue_map;
 	int ret;
 
 	odev = pkt_dev->odev;
@@ -3285,9 +3296,15 @@ static __inline__ void pktgen_xmit(struct pktgen_dev *pkt_dev)
 		}
 	}
 
-	if ((netif_queue_stopped(odev) ||
-	     (pkt_dev->skb &&
-	      netif_subqueue_stopped(odev, pkt_dev->skb))) ||
+	if (!pkt_dev->skb) {
+		set_cur_queue_map(pkt_dev);
+		queue_map = pkt_dev->cur_queue_map;
+	} else {
+		queue_map = skb_get_queue_mapping(pkt_dev->skb);
+	}
+
+	txq = netdev_get_tx_queue(odev, queue_map);
+	if (netif_tx_queue_stopped(txq) ||
 	    need_resched()) {
 		idle_start = getCurUs();
 
@@ -3303,8 +3320,7 @@ static __inline__ void pktgen_xmit(struct pktgen_dev *pkt_dev)
 
 		pkt_dev->idle_acc += getCurUs() - idle_start;
 
-		if (netif_queue_stopped(odev) ||
-		    netif_subqueue_stopped(odev, pkt_dev->skb)) {
+		if (netif_tx_queue_stopped(txq)) {
 			pkt_dev->next_tx_us = getCurUs();	/* TODO */
 			pkt_dev->next_tx_ns = 0;
 			goto out;	/* Try the next interface */
@@ -3331,9 +3347,12 @@ static __inline__ void pktgen_xmit(struct pktgen_dev *pkt_dev)
 		}
 	}
 
-	netif_tx_lock_bh(odev);
-	if (!netif_queue_stopped(odev) &&
-	    !netif_subqueue_stopped(odev, pkt_dev->skb)) {
+	/* fill_packet() might have changed the queue */
+	queue_map = skb_get_queue_mapping(pkt_dev->skb);
+	txq = netdev_get_tx_queue(odev, queue_map);
+
+	__netif_tx_lock_bh(txq);
+	if (!netif_tx_queue_stopped(txq)) {
 
 		atomic_inc(&(pkt_dev->skb->users));
 	      retry_now:
@@ -3377,7 +3396,7 @@ static __inline__ void pktgen_xmit(struct pktgen_dev *pkt_dev)
 		pkt_dev->next_tx_ns = 0;
 	}
 
-	netif_tx_unlock_bh(odev);
+	__netif_tx_unlock_bh(txq);
 
 	/* If pkt_dev->count is zero, then run forever */
 	if ((pkt_dev->count != 0) && (pkt_dev->sofar >= pkt_dev->count)) {
diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
index 4e2b865cbba0..2f575b9017d1 100644
--- a/net/sched/sch_generic.c
+++ b/net/sched/sch_generic.c
@@ -166,7 +166,7 @@ static inline int qdisc_restart(struct netdev_queue *txq)
 
 	HARD_TX_LOCK(dev, txq, smp_processor_id());
 	if (!netif_subqueue_stopped(dev, skb))
-		ret = dev_hard_start_xmit(skb, dev);
+		ret = dev_hard_start_xmit(skb, dev, txq);
 	HARD_TX_UNLOCK(dev, txq);
 
 	spin_lock(&txq->lock);
@@ -198,11 +198,10 @@ static inline int qdisc_restart(struct netdev_queue *txq)
 
 void __qdisc_run(struct netdev_queue *txq)
 {
-	struct net_device *dev = txq->dev;
 	unsigned long start_time = jiffies;
 
 	while (qdisc_restart(txq)) {
-		if (netif_queue_stopped(dev))
+		if (netif_tx_queue_stopped(txq))
 			break;
 
 		/*
diff --git a/net/sched/sch_teql.c b/net/sched/sch_teql.c
index 44a2c3451f4d..ade3372221c7 100644
--- a/net/sched/sch_teql.c
+++ b/net/sched/sch_teql.c
@@ -295,8 +295,7 @@ restart:
 		slave_txq = netdev_get_tx_queue(slave, 0);
 		if (slave_txq->qdisc_sleeping != q)
 			continue;
-		if (netif_queue_stopped(slave) ||
-		    __netif_subqueue_stopped(slave, subq) ||
+		if (__netif_subqueue_stopped(slave, subq) ||
 		    !netif_running(slave)) {
 			busy = 1;
 			continue;
@@ -305,8 +304,7 @@ restart:
 		switch (teql_resolve(skb, skb_res, slave)) {
 		case 0:
 			if (netif_tx_trylock(slave)) {
-				if (!netif_queue_stopped(slave) &&
-				    !__netif_subqueue_stopped(slave, subq) &&
+				if (!__netif_subqueue_stopped(slave, subq) &&
 				    slave->hard_start_xmit(skb, slave) == 0) {
 					netif_tx_unlock(slave);
 					master->slaves = NEXT_SLAVE(q);