X-Git-Url: http://ftp.safe.ca/?a=blobdiff_plain;f=net%2Fsched%2Fsch_generic.c;h=5f5efe4e6072e2959d6f6802a268cc7c5afca66b;hb=3480b25743cb7404928d57efeaa3d085708b04c2;hp=dda78ee314ec9bc79b03fbe03c81a0a799c5affa;hpb=68dfb42798e1eb2d42acbf872925cc75f1487d9b;p=safe%2Fjmp%2Flinux-2.6 diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c index dda78ee..5f5efe4 100644 --- a/net/sched/sch_generic.c +++ b/net/sched/sch_generic.c @@ -29,71 +29,56 @@ /* Main transmission queue. */ /* Modifications to data participating in scheduling must be protected with - * queue->lock spinlock. + * qdisc_lock(qdisc) spinlock. * * The idea is the following: - * - enqueue, dequeue are serialized via top level device - * spinlock queue->lock. - * - ingress filtering is serialized via top level device - * spinlock dev->rx_queue.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->tx_queue.lock) - __acquires(dev->rx_queue.lock) -{ - spin_lock_bh(&dev->tx_queue.lock); - spin_lock(&dev->rx_queue.lock); -} -EXPORT_SYMBOL(qdisc_lock_tree); - -void qdisc_unlock_tree(struct net_device *dev) - __releases(dev->rx_queue.lock) - __releases(dev->tx_queue.lock) -{ - spin_unlock(&dev->rx_queue.lock); - spin_unlock_bh(&dev->tx_queue.lock); -} -EXPORT_SYMBOL(qdisc_unlock_tree); - static inline int qdisc_qlen(struct Qdisc *q) { return q->q.qlen; } -static inline int dev_requeue_skb(struct sk_buff *skb, struct net_device *dev, - struct Qdisc *q) +static inline int dev_requeue_skb(struct sk_buff *skb, struct Qdisc *q) { - if (unlikely(skb->next)) - dev->gso_skb = skb; - else - q->ops->requeue(skb, q); + q->gso_skb = skb; + q->qstats.requeues++; + __netif_schedule(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 (unlikely(skb)) { + struct net_device *dev = qdisc_dev(q); + struct netdev_queue *txq; - if ((skb = dev->gso_skb)) - dev->gso_skb = NULL; - else + /* 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; + 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 +88,7 @@ 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 { /* @@ -111,22 +96,22 @@ 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 queue->lock with locally disabled BH. + * NOTE: Called under qdisc_lock(q) with locally disabled BH. * - * __LINK_STATE_QDISC_RUNNING guarantees only one CPU can process this - * device at a time. queue->lock serializes queue accesses for - * this device AND txq->qdisc pointer itself. + * __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. * - * queue->lock and netif_tx_lock are mutually exclusive, + * 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 @@ -136,28 +121,33 @@ static inline int handle_dev_cpu_collision(struct sk_buff *skb, * >0 - queue is not empty. * */ -static inline int qdisc_restart(struct net_device *dev) +static inline int qdisc_restart(struct Qdisc *q) { - struct netdev_queue *txq = &dev->tx_queue; - struct Qdisc *q = txq->qdisc; - struct sk_buff *skb; + struct netdev_queue *txq; int ret = NETDEV_TX_BUSY; + struct net_device *dev; + spinlock_t *root_lock; + struct sk_buff *skb; /* Dequeue packet */ - if (unlikely((skb = dev_dequeue_skb(dev, q)) == NULL)) + if (unlikely((skb = dequeue_skb(q)) == NULL)) return 0; + root_lock = qdisc_lock(q); - /* And release queue */ - spin_unlock(&txq->lock); + /* And release qdisc */ + spin_unlock(root_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); + dev = qdisc_dev(q); + txq = netdev_get_tx_queue(dev, skb_get_queue_mapping(skb)); - spin_lock(&txq->lock); - q = txq->qdisc; + 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); + + spin_lock(root_lock); switch (ret) { case NETDEV_TX_OK: @@ -167,7 +157,7 @@ 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: @@ -176,54 +166,69 @@ 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) +void __qdisc_run(struct Qdisc *q) { unsigned long start_time = jiffies; - while (qdisc_restart(dev)) { - if (netif_queue_stopped(dev)) - break; - + 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(dev); + __netif_schedule(q); break; } } - clear_bit(__LINK_STATE_QDISC_RUNNING, &dev->state); + clear_bit(__QDISC_STATE_RUNNING, &q->state); } static void dev_watchdog(unsigned long arg) { struct net_device *dev = (struct net_device *)arg; - struct netdev_queue *txq = &dev->tx_queue; netif_tx_lock(dev); - if (txq->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_stopped = 0; + unsigned int i; + + for (i = 0; i < dev->num_tx_queues; i++) { + struct netdev_queue *txq; + + txq = netdev_get_tx_queue(dev, i); + if (netif_tx_queue_stopped(txq)) { + some_queue_stopped = 1; + break; + } + } - printk(KERN_INFO "NETDEV WATCHDOG: %s: transmit timed out\n", - dev->name); - dev->tx_timeout(dev); - WARN_ON_ONCE(1); + if (some_queue_stopped && + time_after(jiffies, (dev->trans_start + + dev->watchdog_timeo))) { + char drivername[64]; + WARN_ONCE(1, KERN_INFO "NETDEV WATCHDOG: %s (%s): transmit timed out\n", + dev->name, netdev_drivername(dev, drivername, 64)); + 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); } } @@ -234,7 +239,7 @@ 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, @@ -265,6 +270,8 @@ 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); @@ -280,8 +287,11 @@ 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); @@ -301,30 +311,28 @@ 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); @@ -333,16 +341,24 @@ 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, }; @@ -389,10 +405,17 @@ 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) @@ -435,7 +458,7 @@ 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, @@ -495,7 +518,7 @@ errout: } EXPORT_SYMBOL(qdisc_create_dflt); -/* Under queue->lock and BH! */ +/* Under qdisc_lock(qdisc) and BH! */ void qdisc_reset(struct Qdisc *qdisc) { @@ -503,19 +526,11 @@ 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); + kfree_skb(qdisc->gso_skb); + qdisc->gso_skb = NULL; } - -/* Under queue->lock and BH! */ +EXPORT_SYMBOL(qdisc_reset); void qdisc_destroy(struct Qdisc *qdisc) { @@ -525,7 +540,11 @@ 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); @@ -534,13 +553,62 @@ void qdisc_destroy(struct Qdisc *qdisc) module_put(ops->owner); dev_put(qdisc_dev(qdisc)); - call_rcu(&qdisc->q_rcu, __qdisc_destroy); + + 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; + } + } 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) + *need_watchdog_p = 1; +} + void dev_activate(struct net_device *dev) { - struct netdev_queue *txq = &dev->tx_queue; + int need_watchdog; /* No queueing discipline is attached to device; create default one i.e. pfifo_fast for devices, @@ -548,61 +616,75 @@ void dev_activate(struct net_device *dev) virtual interfaces */ - if (txq->qdisc_sleeping == &noop_qdisc) { - struct Qdisc *qdisc; - if (dev->tx_queue_len) { - qdisc = qdisc_create_dflt(dev, txq, - &pfifo_fast_ops, - TC_H_ROOT); - if (qdisc == NULL) { - printk(KERN_INFO "%s: activation failed\n", dev->name); - return; - } - list_add_tail(&qdisc->list, &txq->qdisc_list); - } else { - qdisc = &noqueue_qdisc; - } - txq->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(&txq->lock); - rcu_assign_pointer(txq->qdisc, txq->qdisc_sleeping); - if (txq->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(&txq->lock); } static void dev_deactivate_queue(struct net_device *dev, struct netdev_queue *dev_queue, - struct Qdisc *qdisc_default) + void *_qdisc_default) { - struct Qdisc *qdisc = dev_queue->qdisc; + struct Qdisc *qdisc_default = _qdisc_default; + struct Qdisc *qdisc; + qdisc = dev_queue->qdisc; if (qdisc) { - dev_queue->qdisc = qdisc_default; + spin_lock_bh(qdisc_lock(qdisc)); + + if (!(qdisc->flags & TCQ_F_BUILTIN)) + set_bit(__QDISC_STATE_DEACTIVATED, &qdisc->state); + + rcu_assign_pointer(dev_queue->qdisc, qdisc_default); qdisc_reset(qdisc); + + spin_unlock_bh(qdisc_lock(qdisc)); } } -void dev_deactivate(struct net_device *dev) +static bool some_qdisc_is_busy(struct net_device *dev) { - struct sk_buff *skb; - int running; + unsigned int i; - spin_lock_bh(&dev->tx_queue.lock); - dev_deactivate_queue(dev, &dev->tx_queue, &noop_qdisc); + for (i = 0; i < dev->num_tx_queues; i++) { + struct netdev_queue *dev_queue; + spinlock_t *root_lock; + struct Qdisc *q; + int val; - skb = dev->gso_skb; - dev->gso_skb = NULL; - spin_unlock_bh(&dev->tx_queue.lock); + dev_queue = netdev_get_tx_queue(dev, i); + q = dev_queue->qdisc_sleeping; + root_lock = qdisc_lock(q); - kfree_skb(skb); + 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); @@ -610,56 +692,37 @@ 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(); - - /* - * Double-check inside queue lock to ensure that all effects - * of the queue run are visible when we return. - */ - spin_lock_bh(&dev->tx_queue.lock); - running = test_bit(__LINK_STATE_QDISC_RUNNING, &dev->state); - spin_unlock_bh(&dev->tx_queue.lock); - - /* - * 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)); + while (some_qdisc_is_busy(dev)) + yield(); } static void dev_init_scheduler_queue(struct net_device *dev, struct netdev_queue *dev_queue, - struct Qdisc *qdisc) + void *_qdisc) { + struct Qdisc *qdisc = _qdisc; + dev_queue->qdisc = qdisc; dev_queue->qdisc_sleeping = qdisc; - INIT_LIST_HEAD(&dev_queue->qdisc_list); } void dev_init_scheduler(struct net_device *dev) { - qdisc_lock_tree(dev); - dev_init_scheduler_queue(dev, &dev->tx_queue, &noop_qdisc); - dev_init_scheduler_queue(dev, &dev->rx_queue, NULL); - 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); } -static void dev_shutdown_scheduler_queue(struct net_device *dev, - struct netdev_queue *dev_queue, - struct Qdisc *qdisc_default) +static void shutdown_scheduler_queue(struct net_device *dev, + struct netdev_queue *dev_queue, + void *_qdisc_default) { struct Qdisc *qdisc = dev_queue->qdisc_sleeping; + struct Qdisc *qdisc_default = _qdisc_default; if (qdisc) { - dev_queue->qdisc = qdisc_default; + rcu_assign_pointer(dev_queue->qdisc, qdisc_default); dev_queue->qdisc_sleeping = qdisc_default; qdisc_destroy(qdisc); @@ -668,9 +731,7 @@ static void dev_shutdown_scheduler_queue(struct net_device *dev, void dev_shutdown(struct net_device *dev) { - qdisc_lock_tree(dev); - dev_shutdown_scheduler_queue(dev, &dev->tx_queue, &noop_qdisc); - dev_shutdown_scheduler_queue(dev, &dev->rx_queue, NULL); - BUG_TRAP(!timer_pending(&dev->watchdog_timer)); - qdisc_unlock_tree(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)); }