netlink: constify nlmsghdr arguments
[safe/jmp/linux-2.6] / net / sched / sch_generic.c
index 51e64ac..693df7a 100644 (file)
 /* Main transmission queue. */
 
 /* Modifications to data participating in scheduling must be protected with
- * dev->queue_lock spinlock.
+ * qdisc_lock(qdisc) spinlock.
  *
  * The idea is the following:
- * - enqueue, dequeue are serialized via top level device
- *   spinlock dev->queue_lock.
- * - ingress filtering is serialized via top level device
- *   spinlock dev->ingress_lock.
+ * - enqueue, dequeue are serialized via qdisc root lock
+ * - ingress filtering is also serialized via qdisc root lock
  * - updates to tree and tree walking are only done under the rtnl mutex.
  */
 
-void qdisc_lock_tree(struct net_device *dev)
-       __acquires(dev->queue_lock)
-       __acquires(dev->ingress_lock)
+static inline int dev_requeue_skb(struct sk_buff *skb, struct Qdisc *q)
 {
-       spin_lock_bh(&dev->queue_lock);
-       spin_lock(&dev->ingress_lock);
-}
-EXPORT_SYMBOL(qdisc_lock_tree);
-
-void qdisc_unlock_tree(struct net_device *dev)
-       __releases(dev->ingress_lock)
-       __releases(dev->queue_lock)
-{
-       spin_unlock(&dev->ingress_lock);
-       spin_unlock_bh(&dev->queue_lock);
-}
-EXPORT_SYMBOL(qdisc_unlock_tree);
-
-static inline int qdisc_qlen(struct Qdisc *q)
-{
-       return q->q.qlen;
-}
+       q->gso_skb = skb;
+       q->qstats.requeues++;
+       q->q.qlen++;    /* it's still part of the queue */
+       __netif_schedule(q);
 
-static inline int dev_requeue_skb(struct sk_buff *skb, struct net_device *dev,
-                                 struct Qdisc *q)
-{
-       if (unlikely(skb->next))
-               dev->gso_skb = skb;
-       else
-               q->ops->requeue(skb, q);
-
-       netif_schedule(dev);
        return 0;
 }
 
-static inline struct sk_buff *dev_dequeue_skb(struct net_device *dev,
-                                             struct Qdisc *q)
+static inline struct sk_buff *dequeue_skb(struct Qdisc *q)
 {
-       struct sk_buff *skb;
+       struct sk_buff *skb = q->gso_skb;
 
-       if ((skb = dev->gso_skb))
-               dev->gso_skb = NULL;
-       else
+       if (unlikely(skb)) {
+               struct net_device *dev = qdisc_dev(q);
+               struct netdev_queue *txq;
+
+               /* check the reason of requeuing without tx lock first */
+               txq = netdev_get_tx_queue(dev, skb_get_queue_mapping(skb));
+               if (!netif_tx_queue_stopped(txq) &&
+                   !netif_tx_queue_frozen(txq)) {
+                       q->gso_skb = NULL;
+                       q->q.qlen--;
+               } else
+                       skb = NULL;
+       } else {
                skb = q->dequeue(q);
+       }
 
        return skb;
 }
 
 static inline int handle_dev_cpu_collision(struct sk_buff *skb,
-                                          struct net_device *dev,
+                                          struct netdev_queue *dev_queue,
                                           struct Qdisc *q)
 {
        int ret;
 
-       if (unlikely(dev->xmit_lock_owner == smp_processor_id())) {
+       if (unlikely(dev_queue->xmit_lock_owner == smp_processor_id())) {
                /*
                 * Same CPU holding the lock. It may be a transient
                 * configuration error, when hard_start_xmit() recurses. We
@@ -103,7 +86,7 @@ static inline int handle_dev_cpu_collision(struct sk_buff *skb,
                kfree_skb(skb);
                if (net_ratelimit())
                        printk(KERN_WARNING "Dead loop on netdevice %s, "
-                              "fix it urgently!\n", dev->name);
+                              "fix it urgently!\n", dev_queue->dev->name);
                ret = qdisc_qlen(q);
        } else {
                /*
@@ -111,52 +94,37 @@ static inline int handle_dev_cpu_collision(struct sk_buff *skb,
                 * some time.
                 */
                __get_cpu_var(netdev_rx_stat).cpu_collision++;
-               ret = dev_requeue_skb(skb, dev, q);
+               ret = dev_requeue_skb(skb, q);
        }
 
        return ret;
 }
 
 /*
- * NOTE: Called under dev->queue_lock with locally disabled BH.
- *
- * __LINK_STATE_QDISC_RUNNING guarantees only one CPU can process this
- * device at a time. dev->queue_lock serializes queue accesses for
- * this device AND dev->qdisc pointer itself.
- *
- *  netif_tx_lock serializes accesses to device driver.
- *
- *  dev->queue_lock and netif_tx_lock are mutually exclusive,
- *  if one is grabbed, another must be free.
- *
- * Note, that this procedure can be called by a watchdog timer
+ * Transmit one skb, and handle the return status as required. Holding the
+ * __QDISC_STATE_RUNNING bit guarantees that only one CPU can execute this
+ * function.
  *
  * Returns to the caller:
  *                             0  - queue is empty or throttled.
  *                             >0 - queue is not empty.
- *
  */
-static inline int qdisc_restart(struct net_device *dev)
+int sch_direct_xmit(struct sk_buff *skb, struct Qdisc *q,
+                   struct net_device *dev, struct netdev_queue *txq,
+                   spinlock_t *root_lock)
 {
-       struct Qdisc *q = dev->qdisc;
-       struct sk_buff *skb;
        int ret = NETDEV_TX_BUSY;
 
-       /* Dequeue packet */
-       if (unlikely((skb = dev_dequeue_skb(dev, q)) == NULL))
-               return 0;
+       /* And release qdisc */
+       spin_unlock(root_lock);
 
+       HARD_TX_LOCK(dev, txq, smp_processor_id());
+       if (!netif_tx_queue_stopped(txq) &&
+           !netif_tx_queue_frozen(txq))
+               ret = dev_hard_start_xmit(skb, dev, txq);
+       HARD_TX_UNLOCK(dev, txq);
 
-       /* And release queue */
-       spin_unlock(&dev->queue_lock);
-
-       HARD_TX_LOCK(dev, smp_processor_id());
-       if (!netif_subqueue_stopped(dev, skb))
-               ret = dev_hard_start_xmit(skb, dev);
-       HARD_TX_UNLOCK(dev);
-
-       spin_lock(&dev->queue_lock);
-       q = dev->qdisc;
+       spin_lock(root_lock);
 
        switch (ret) {
        case NETDEV_TX_OK:
@@ -166,7 +134,7 @@ static inline int qdisc_restart(struct net_device *dev)
 
        case NETDEV_TX_LOCKED:
                /* Driver try lock failed */
-               ret = handle_dev_cpu_collision(skb, dev, q);
+               ret = handle_dev_cpu_collision(skb, txq, q);
                break;
 
        default:
@@ -175,40 +143,127 @@ static inline int qdisc_restart(struct net_device *dev)
                        printk(KERN_WARNING "BUG %s code %d qlen %d\n",
                               dev->name, ret, q->q.qlen);
 
-               ret = dev_requeue_skb(skb, dev, q);
+               ret = dev_requeue_skb(skb, q);
                break;
        }
 
+       if (ret && (netif_tx_queue_stopped(txq) ||
+                   netif_tx_queue_frozen(txq)))
+               ret = 0;
+
        return ret;
 }
 
-void __qdisc_run(struct net_device *dev)
+/*
+ * NOTE: Called under qdisc_lock(q) with locally disabled BH.
+ *
+ * __QDISC_STATE_RUNNING guarantees only one CPU can process
+ * this qdisc at a time. qdisc_lock(q) serializes queue accesses for
+ * this queue.
+ *
+ *  netif_tx_lock serializes accesses to device driver.
+ *
+ *  qdisc_lock(q) and netif_tx_lock are mutually exclusive,
+ *  if one is grabbed, another must be free.
+ *
+ * Note, that this procedure can be called by a watchdog timer
+ *
+ * Returns to the caller:
+ *                             0  - queue is empty or throttled.
+ *                             >0 - queue is not empty.
+ *
+ */
+static inline int qdisc_restart(struct Qdisc *q)
+{
+       struct netdev_queue *txq;
+       struct net_device *dev;
+       spinlock_t *root_lock;
+       struct sk_buff *skb;
+
+       /* Dequeue packet */
+       skb = dequeue_skb(q);
+       if (unlikely(!skb))
+               return 0;
+
+       root_lock = qdisc_lock(q);
+       dev = qdisc_dev(q);
+       txq = netdev_get_tx_queue(dev, skb_get_queue_mapping(skb));
+
+       return sch_direct_xmit(skb, q, dev, txq, root_lock);
+}
+
+void __qdisc_run(struct Qdisc *q)
 {
-       do {
-               if (!qdisc_restart(dev))
+       unsigned long start_time = jiffies;
+
+       while (qdisc_restart(q)) {
+               /*
+                * Postpone processing if
+                * 1. another process needs the CPU;
+                * 2. we've been doing it for too long.
+                */
+               if (need_resched() || jiffies != start_time) {
+                       __netif_schedule(q);
                        break;
-       } while (!netif_queue_stopped(dev));
+               }
+       }
 
-       clear_bit(__LINK_STATE_QDISC_RUNNING, &dev->state);
+       clear_bit(__QDISC_STATE_RUNNING, &q->state);
 }
 
+unsigned long dev_trans_start(struct net_device *dev)
+{
+       unsigned long val, res = dev->trans_start;
+       unsigned int i;
+
+       for (i = 0; i < dev->num_tx_queues; i++) {
+               val = netdev_get_tx_queue(dev, i)->trans_start;
+               if (val && time_after(val, res))
+                       res = val;
+       }
+       dev->trans_start = res;
+       return res;
+}
+EXPORT_SYMBOL(dev_trans_start);
+
 static void dev_watchdog(unsigned long arg)
 {
        struct net_device *dev = (struct net_device *)arg;
 
        netif_tx_lock(dev);
-       if (dev->qdisc != &noop_qdisc) {
+       if (!qdisc_tx_is_noop(dev)) {
                if (netif_device_present(dev) &&
                    netif_running(dev) &&
                    netif_carrier_ok(dev)) {
-                       if (netif_queue_stopped(dev) &&
-                           time_after(jiffies, dev->trans_start + dev->watchdog_timeo)) {
+                       int some_queue_timedout = 0;
+                       unsigned int i;
+                       unsigned long trans_start;
+
+                       for (i = 0; i < dev->num_tx_queues; i++) {
+                               struct netdev_queue *txq;
+
+                               txq = netdev_get_tx_queue(dev, i);
+                               /*
+                                * old device drivers set dev->trans_start
+                                */
+                               trans_start = txq->trans_start ? : dev->trans_start;
+                               if (netif_tx_queue_stopped(txq) &&
+                                   time_after(jiffies, (trans_start +
+                                                        dev->watchdog_timeo))) {
+                                       some_queue_timedout = 1;
+                                       break;
+                               }
+                       }
 
-                               printk(KERN_INFO "NETDEV WATCHDOG: %s: transmit timed out\n",
-                                      dev->name);
-                               dev->tx_timeout(dev);
+                       if (some_queue_timedout) {
+                               char drivername[64];
+                               WARN_ONCE(1, KERN_INFO "NETDEV WATCHDOG: %s (%s): transmit queue %u timed out\n",
+                                      dev->name, netdev_drivername(dev, drivername, 64), i);
+                               dev->netdev_ops->ndo_tx_timeout(dev);
                        }
-                       if (!mod_timer(&dev->watchdog_timer, round_jiffies(jiffies + dev->watchdog_timeo)))
+                       if (!mod_timer(&dev->watchdog_timer,
+                                      round_jiffies(jiffies +
+                                                    dev->watchdog_timeo)))
                                dev_hold(dev);
                }
        }
@@ -219,7 +274,7 @@ static void dev_watchdog(unsigned long arg)
 
 void __netdev_watchdog_up(struct net_device *dev)
 {
-       if (dev->tx_timeout) {
+       if (dev->netdev_ops->ndo_tx_timeout) {
                if (dev->watchdog_timeo <= 0)
                        dev->watchdog_timeo = 5*HZ;
                if (!mod_timer(&dev->watchdog_timer,
@@ -250,6 +305,8 @@ static void dev_watchdog_down(struct net_device *dev)
 void netif_carrier_on(struct net_device *dev)
 {
        if (test_and_clear_bit(__LINK_STATE_NOCARRIER, &dev->state)) {
+               if (dev->reg_state == NETREG_UNINITIALIZED)
+                       return;
                linkwatch_fire_event(dev);
                if (netif_running(dev))
                        __netdev_watchdog_up(dev);
@@ -265,8 +322,11 @@ EXPORT_SYMBOL(netif_carrier_on);
  */
 void netif_carrier_off(struct net_device *dev)
 {
-       if (!test_and_set_bit(__LINK_STATE_NOCARRIER, &dev->state))
+       if (!test_and_set_bit(__LINK_STATE_NOCARRIER, &dev->state)) {
+               if (dev->reg_state == NETREG_UNINITIALIZED)
+                       return;
                linkwatch_fire_event(dev);
+       }
 }
 EXPORT_SYMBOL(netif_carrier_off);
 
@@ -286,30 +346,28 @@ static struct sk_buff *noop_dequeue(struct Qdisc * qdisc)
        return NULL;
 }
 
-static int noop_requeue(struct sk_buff *skb, struct Qdisc* qdisc)
-{
-       if (net_ratelimit())
-               printk(KERN_DEBUG "%s deferred output. It is buggy.\n",
-                      skb->dev->name);
-       kfree_skb(skb);
-       return NET_XMIT_CN;
-}
-
 struct Qdisc_ops noop_qdisc_ops __read_mostly = {
        .id             =       "noop",
        .priv_size      =       0,
        .enqueue        =       noop_enqueue,
        .dequeue        =       noop_dequeue,
-       .requeue        =       noop_requeue,
+       .peek           =       noop_dequeue,
        .owner          =       THIS_MODULE,
 };
 
+static struct netdev_queue noop_netdev_queue = {
+       .qdisc          =       &noop_qdisc,
+       .qdisc_sleeping =       &noop_qdisc,
+};
+
 struct Qdisc noop_qdisc = {
        .enqueue        =       noop_enqueue,
        .dequeue        =       noop_dequeue,
        .flags          =       TCQ_F_BUILTIN,
        .ops            =       &noop_qdisc_ops,
        .list           =       LIST_HEAD_INIT(noop_qdisc.list),
+       .q.lock         =       __SPIN_LOCK_UNLOCKED(noop_qdisc.q.lock),
+       .dev_queue      =       &noop_netdev_queue,
 };
 EXPORT_SYMBOL(noop_qdisc);
 
@@ -318,16 +376,24 @@ static struct Qdisc_ops noqueue_qdisc_ops __read_mostly = {
        .priv_size      =       0,
        .enqueue        =       noop_enqueue,
        .dequeue        =       noop_dequeue,
-       .requeue        =       noop_requeue,
+       .peek           =       noop_dequeue,
        .owner          =       THIS_MODULE,
 };
 
+static struct Qdisc noqueue_qdisc;
+static struct netdev_queue noqueue_netdev_queue = {
+       .qdisc          =       &noqueue_qdisc,
+       .qdisc_sleeping =       &noqueue_qdisc,
+};
+
 static struct Qdisc noqueue_qdisc = {
        .enqueue        =       NULL,
        .dequeue        =       noop_dequeue,
        .flags          =       TCQ_F_BUILTIN,
        .ops            =       &noqueue_qdisc_ops,
        .list           =       LIST_HEAD_INIT(noqueue_qdisc.list),
+       .q.lock         =       __SPIN_LOCK_UNLOCKED(noqueue_qdisc.q.lock),
+       .dev_queue      =       &noqueue_netdev_queue,
 };
 
 
@@ -351,7 +417,7 @@ static int pfifo_fast_enqueue(struct sk_buff *skb, struct Qdisc* qdisc)
 {
        struct sk_buff_head *list = prio2list(skb, qdisc);
 
-       if (skb_queue_len(list) < qdisc->dev->tx_queue_len) {
+       if (skb_queue_len(list) < qdisc_dev(qdisc)->tx_queue_len) {
                qdisc->q.qlen++;
                return __qdisc_enqueue_tail(skb, qdisc, list);
        }
@@ -374,10 +440,17 @@ static struct sk_buff *pfifo_fast_dequeue(struct Qdisc* qdisc)
        return NULL;
 }
 
-static int pfifo_fast_requeue(struct sk_buff *skb, struct Qdisc* qdisc)
+static struct sk_buff *pfifo_fast_peek(struct Qdisc* qdisc)
 {
-       qdisc->q.qlen++;
-       return __qdisc_requeue(skb, qdisc, prio2list(skb, qdisc));
+       int prio;
+       struct sk_buff_head *list = qdisc_priv(qdisc);
+
+       for (prio = 0; prio < PFIFO_FAST_BANDS; prio++) {
+               if (!skb_queue_empty(list + prio))
+                       return skb_peek(list + prio);
+       }
+
+       return NULL;
 }
 
 static void pfifo_fast_reset(struct Qdisc* qdisc)
@@ -397,14 +470,14 @@ static int pfifo_fast_dump(struct Qdisc *qdisc, struct sk_buff *skb)
        struct tc_prio_qopt opt = { .bands = PFIFO_FAST_BANDS };
 
        memcpy(&opt.priomap, prio2band, TC_PRIO_MAX+1);
-       RTA_PUT(skb, TCA_OPTIONS, sizeof(opt), &opt);
+       NLA_PUT(skb, TCA_OPTIONS, sizeof(opt), &opt);
        return skb->len;
 
-rtattr_failure:
+nla_put_failure:
        return -1;
 }
 
-static int pfifo_fast_init(struct Qdisc *qdisc, struct rtattr *opt)
+static int pfifo_fast_init(struct Qdisc *qdisc, struct nlattr *opt)
 {
        int prio;
        struct sk_buff_head *list = qdisc_priv(qdisc);
@@ -420,14 +493,15 @@ static struct Qdisc_ops pfifo_fast_ops __read_mostly = {
        .priv_size      =       PFIFO_FAST_BANDS * sizeof(struct sk_buff_head),
        .enqueue        =       pfifo_fast_enqueue,
        .dequeue        =       pfifo_fast_dequeue,
-       .requeue        =       pfifo_fast_requeue,
+       .peek           =       pfifo_fast_peek,
        .init           =       pfifo_fast_init,
        .reset          =       pfifo_fast_reset,
        .dump           =       pfifo_fast_dump,
        .owner          =       THIS_MODULE,
 };
 
-struct Qdisc *qdisc_alloc(struct net_device *dev, struct Qdisc_ops *ops)
+struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
+                         struct Qdisc_ops *ops)
 {
        void *p;
        struct Qdisc *sch;
@@ -449,24 +523,25 @@ struct Qdisc *qdisc_alloc(struct net_device *dev, struct Qdisc_ops *ops)
        sch->ops = ops;
        sch->enqueue = ops->enqueue;
        sch->dequeue = ops->dequeue;
-       sch->dev = dev;
-       dev_hold(dev);
+       sch->dev_queue = dev_queue;
+       dev_hold(qdisc_dev(sch));
        atomic_set(&sch->refcnt, 1);
 
        return sch;
 errout:
-       return ERR_PTR(-err);
+       return ERR_PTR(err);
 }
 
-struct Qdisc * qdisc_create_dflt(struct net_device *dev, struct Qdisc_ops *ops,
+struct Qdisc * qdisc_create_dflt(struct net_device *dev,
+                                struct netdev_queue *dev_queue,
+                                struct Qdisc_ops *ops,
                                 unsigned int parentid)
 {
        struct Qdisc *sch;
 
-       sch = qdisc_alloc(dev, ops);
+       sch = qdisc_alloc(dev_queue, ops);
        if (IS_ERR(sch))
                goto errout;
-       sch->stats_lock = &dev->queue_lock;
        sch->parent = parentid;
 
        if (!ops->init || ops->init(sch, NULL) == 0)
@@ -478,7 +553,7 @@ errout:
 }
 EXPORT_SYMBOL(qdisc_create_dflt);
 
-/* Under dev->queue_lock and BH! */
+/* Under qdisc_lock(qdisc) and BH! */
 
 void qdisc_reset(struct Qdisc *qdisc)
 {
@@ -486,19 +561,14 @@ void qdisc_reset(struct Qdisc *qdisc)
 
        if (ops->reset)
                ops->reset(qdisc);
-}
-EXPORT_SYMBOL(qdisc_reset);
-
-/* this is the rcu callback function to clean up a qdisc when there
- * are no further references to it */
 
-static void __qdisc_destroy(struct rcu_head *head)
-{
-       struct Qdisc *qdisc = container_of(head, struct Qdisc, q_rcu);
-       kfree((char *) qdisc - qdisc->padded);
+       if (qdisc->gso_skb) {
+               kfree_skb(qdisc->gso_skb);
+               qdisc->gso_skb = NULL;
+               qdisc->q.qlen = 0;
+       }
 }
-
-/* Under dev->queue_lock and BH! */
+EXPORT_SYMBOL(qdisc_reset);
 
 void qdisc_destroy(struct Qdisc *qdisc)
 {
@@ -508,7 +578,11 @@ void qdisc_destroy(struct Qdisc *qdisc)
            !atomic_dec_and_test(&qdisc->refcnt))
                return;
 
-       list_del(&qdisc->list);
+#ifdef CONFIG_NET_SCHED
+       qdisc_list_del(qdisc);
+
+       qdisc_put_stab(qdisc->stab);
+#endif
        gen_kill_estimator(&qdisc->bstats, &qdisc->rate_est);
        if (ops->reset)
                ops->reset(qdisc);
@@ -516,65 +590,144 @@ void qdisc_destroy(struct Qdisc *qdisc)
                ops->destroy(qdisc);
 
        module_put(ops->owner);
-       dev_put(qdisc->dev);
-       call_rcu(&qdisc->q_rcu, __qdisc_destroy);
+       dev_put(qdisc_dev(qdisc));
+
+       kfree_skb(qdisc->gso_skb);
+       kfree((char *) qdisc - qdisc->padded);
 }
 EXPORT_SYMBOL(qdisc_destroy);
 
+static bool dev_all_qdisc_sleeping_noop(struct net_device *dev)
+{
+       unsigned int i;
+
+       for (i = 0; i < dev->num_tx_queues; i++) {
+               struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
+
+               if (txq->qdisc_sleeping != &noop_qdisc)
+                       return false;
+       }
+       return true;
+}
+
+static void attach_one_default_qdisc(struct net_device *dev,
+                                    struct netdev_queue *dev_queue,
+                                    void *_unused)
+{
+       struct Qdisc *qdisc;
+
+       if (dev->tx_queue_len) {
+               qdisc = qdisc_create_dflt(dev, dev_queue,
+                                         &pfifo_fast_ops, TC_H_ROOT);
+               if (!qdisc) {
+                       printk(KERN_INFO "%s: activation failed\n", dev->name);
+                       return;
+               }
+
+               /* Can by-pass the queue discipline for default qdisc */
+               qdisc->flags |= TCQ_F_CAN_BYPASS;
+       } else {
+               qdisc =  &noqueue_qdisc;
+       }
+       dev_queue->qdisc_sleeping = qdisc;
+}
+
+static void transition_one_qdisc(struct net_device *dev,
+                                struct netdev_queue *dev_queue,
+                                void *_need_watchdog)
+{
+       struct Qdisc *new_qdisc = dev_queue->qdisc_sleeping;
+       int *need_watchdog_p = _need_watchdog;
+
+       if (!(new_qdisc->flags & TCQ_F_BUILTIN))
+               clear_bit(__QDISC_STATE_DEACTIVATED, &new_qdisc->state);
+
+       rcu_assign_pointer(dev_queue->qdisc, new_qdisc);
+       if (need_watchdog_p && new_qdisc != &noqueue_qdisc) {
+               dev_queue->trans_start = 0;
+               *need_watchdog_p = 1;
+       }
+}
+
 void dev_activate(struct net_device *dev)
 {
+       int need_watchdog;
+
        /* No queueing discipline is attached to device;
           create default one i.e. pfifo_fast for devices,
           which need queueing and noqueue_qdisc for
           virtual interfaces
         */
 
-       if (dev->qdisc_sleeping == &noop_qdisc) {
-               struct Qdisc *qdisc;
-               if (dev->tx_queue_len) {
-                       qdisc = qdisc_create_dflt(dev, &pfifo_fast_ops,
-                                                 TC_H_ROOT);
-                       if (qdisc == NULL) {
-                               printk(KERN_INFO "%s: activation failed\n", dev->name);
-                               return;
-                       }
-                       list_add_tail(&qdisc->list, &dev->qdisc_list);
-               } else {
-                       qdisc =  &noqueue_qdisc;
-               }
-               dev->qdisc_sleeping = qdisc;
-       }
+       if (dev_all_qdisc_sleeping_noop(dev))
+               netdev_for_each_tx_queue(dev, attach_one_default_qdisc, NULL);
 
        if (!netif_carrier_ok(dev))
                /* Delay activation until next carrier-on event */
                return;
 
-       spin_lock_bh(&dev->queue_lock);
-       rcu_assign_pointer(dev->qdisc, dev->qdisc_sleeping);
-       if (dev->qdisc != &noqueue_qdisc) {
+       need_watchdog = 0;
+       netdev_for_each_tx_queue(dev, transition_one_qdisc, &need_watchdog);
+       transition_one_qdisc(dev, &dev->rx_queue, NULL);
+
+       if (need_watchdog) {
                dev->trans_start = jiffies;
                dev_watchdog_up(dev);
        }
-       spin_unlock_bh(&dev->queue_lock);
 }
 
-void dev_deactivate(struct net_device *dev)
+static void dev_deactivate_queue(struct net_device *dev,
+                                struct netdev_queue *dev_queue,
+                                void *_qdisc_default)
 {
+       struct Qdisc *qdisc_default = _qdisc_default;
        struct Qdisc *qdisc;
-       struct sk_buff *skb;
-       int running;
 
-       spin_lock_bh(&dev->queue_lock);
-       qdisc = dev->qdisc;
-       dev->qdisc = &noop_qdisc;
+       qdisc = dev_queue->qdisc;
+       if (qdisc) {
+               spin_lock_bh(qdisc_lock(qdisc));
 
-       qdisc_reset(qdisc);
+               if (!(qdisc->flags & TCQ_F_BUILTIN))
+                       set_bit(__QDISC_STATE_DEACTIVATED, &qdisc->state);
 
-       skb = dev->gso_skb;
-       dev->gso_skb = NULL;
-       spin_unlock_bh(&dev->queue_lock);
+               rcu_assign_pointer(dev_queue->qdisc, qdisc_default);
+               qdisc_reset(qdisc);
 
-       kfree_skb(skb);
+               spin_unlock_bh(qdisc_lock(qdisc));
+       }
+}
+
+static bool some_qdisc_is_busy(struct net_device *dev)
+{
+       unsigned int i;
+
+       for (i = 0; i < dev->num_tx_queues; i++) {
+               struct netdev_queue *dev_queue;
+               spinlock_t *root_lock;
+               struct Qdisc *q;
+               int val;
+
+               dev_queue = netdev_get_tx_queue(dev, i);
+               q = dev_queue->qdisc_sleeping;
+               root_lock = qdisc_lock(q);
+
+               spin_lock_bh(root_lock);
+
+               val = (test_bit(__QDISC_STATE_RUNNING, &q->state) ||
+                      test_bit(__QDISC_STATE_SCHED, &q->state));
+
+               spin_unlock_bh(root_lock);
+
+               if (val)
+                       return true;
+       }
+       return false;
+}
+
+void dev_deactivate(struct net_device *dev)
+{
+       netdev_for_each_tx_queue(dev, dev_deactivate_queue, &noop_qdisc);
+       dev_deactivate_queue(dev, &dev->rx_queue, &noop_qdisc);
 
        dev_watchdog_down(dev);
 
@@ -582,55 +735,46 @@ void dev_deactivate(struct net_device *dev)
        synchronize_rcu();
 
        /* Wait for outstanding qdisc_run calls. */
-       do {
-               while (test_bit(__LINK_STATE_QDISC_RUNNING, &dev->state))
-                       yield();
+       while (some_qdisc_is_busy(dev))
+               yield();
+}
 
-               /*
-                * Double-check inside queue lock to ensure that all effects
-                * of the queue run are visible when we return.
-                */
-               spin_lock_bh(&dev->queue_lock);
-               running = test_bit(__LINK_STATE_QDISC_RUNNING, &dev->state);
-               spin_unlock_bh(&dev->queue_lock);
+static void dev_init_scheduler_queue(struct net_device *dev,
+                                    struct netdev_queue *dev_queue,
+                                    void *_qdisc)
+{
+       struct Qdisc *qdisc = _qdisc;
 
-               /*
-                * The running flag should never be set at this point because
-                * we've already set dev->qdisc to noop_qdisc *inside* the same
-                * pair of spin locks.  That is, if any qdisc_run starts after
-                * our initial test it should see the noop_qdisc and then
-                * clear the RUNNING bit before dropping the queue lock.  So
-                * if it is set here then we've found a bug.
-                */
-       } while (WARN_ON_ONCE(running));
+       dev_queue->qdisc = qdisc;
+       dev_queue->qdisc_sleeping = qdisc;
 }
 
 void dev_init_scheduler(struct net_device *dev)
 {
-       qdisc_lock_tree(dev);
-       dev->qdisc = &noop_qdisc;
-       dev->qdisc_sleeping = &noop_qdisc;
-       INIT_LIST_HEAD(&dev->qdisc_list);
-       qdisc_unlock_tree(dev);
+       netdev_for_each_tx_queue(dev, dev_init_scheduler_queue, &noop_qdisc);
+       dev_init_scheduler_queue(dev, &dev->rx_queue, &noop_qdisc);
 
        setup_timer(&dev->watchdog_timer, dev_watchdog, (unsigned long)dev);
 }
 
-void dev_shutdown(struct net_device *dev)
+static void shutdown_scheduler_queue(struct net_device *dev,
+                                    struct netdev_queue *dev_queue,
+                                    void *_qdisc_default)
 {
-       struct Qdisc *qdisc;
+       struct Qdisc *qdisc = dev_queue->qdisc_sleeping;
+       struct Qdisc *qdisc_default = _qdisc_default;
+
+       if (qdisc) {
+               rcu_assign_pointer(dev_queue->qdisc, qdisc_default);
+               dev_queue->qdisc_sleeping = qdisc_default;
 
-       qdisc_lock_tree(dev);
-       qdisc = dev->qdisc_sleeping;
-       dev->qdisc = &noop_qdisc;
-       dev->qdisc_sleeping = &noop_qdisc;
-       qdisc_destroy(qdisc);
-#if defined(CONFIG_NET_SCH_INGRESS) || defined(CONFIG_NET_SCH_INGRESS_MODULE)
-       if ((qdisc = dev->qdisc_ingress) != NULL) {
-               dev->qdisc_ingress = NULL;
                qdisc_destroy(qdisc);
        }
-#endif
-       BUG_TRAP(!timer_pending(&dev->watchdog_timer));
-       qdisc_unlock_tree(dev);
+}
+
+void dev_shutdown(struct net_device *dev)
+{
+       netdev_for_each_tx_queue(dev, shutdown_scheduler_queue, &noop_qdisc);
+       shutdown_scheduler_queue(dev, &dev->rx_queue, &noop_qdisc);
+       WARN_ON(timer_pending(&dev->watchdog_timer));
 }