[netdrvr tg3] fix NAPI deadlock

* do not hold driver spinlock during RX processing in tg3_poll
  (this is the deadlock fix)
* create netif_poll_{en,dis}able to synchronize against dev->poll()
* create __netif_rx_complete to avoid a third irq-save in tg3_poll
* create tg3_netif_{start,stop} as driver-specific helper functions
  which disable and enable NAPI polling and TX queueing.  Note that
  the TX queueing enable/disable is purely advisory, and is not
  intended to prevent any races.
* remove tg3_halt call from tg3_set_power_state, as all callers
  have already called tg3_halt, making it redundant.  Removing this
  function call also eliminates some locking complications.
* use new helper __netif_rx_complete in tg3_poll
* create tg3_reset_task, as a function that runs in process context
  which resets the NIC.  This is needed because tg3_netif_stop()
  calls schedule() in the process of disabling dev->poll.
* schedule tg3_reset_task from tg3_tx_timeout
* schedule tg3_reset_task from tg3_timer
* wrap several tg3_halt...tg3_init_hw sequences with
  tg3_netif_stop...tg3_netif_start.  In addition to synchronizing
  with dev->poll, this additionally fixes bugs where we were not
  calling netif_wake_queue, when we should have been.
* move netif_start_queue call to very bottom of tg3_open
* add missing tg3_netif_{start,stop} to tg3_{suspend,resume},
  further fixing obvious bugs.

drivers/net/tg3.c

@@ -25,6 +25,7 @@
 #include <linux/if_vlan.h>
 #include <linux/ip.h>
 #include <linux/tcp.h>
+#include <linux/workqueue.h>
 
 #include <asm/system.h>
 #include <asm/io.h>
@@ -228,6 +229,46 @@ static void tg3_enable_ints(struct tg3 *tp)
 		     tp->grc_local_ctrl | GRC_LCLCTRL_SETINT);
 }
 
+/* these three netif_xxx funcs should be moved into generic net layer */
+static void netif_poll_disable(struct net_device *dev)
+{
+	while (test_and_set_bit(__LINK_STATE_RX_SCHED, &dev->state)) {
+		current->state = TASK_INTERRUPTIBLE;
+		schedule_timeout(1);
+	}
+}
+
+static inline void netif_poll_enable(struct net_device *dev)
+{
+	clear_bit(__LINK_STATE_RX_SCHED, &dev->state);
+}
+
+/* same as netif_rx_complete, except that local_irq_save(flags)
+ * has already been issued
+ */
+static inline void __netif_rx_complete(struct net_device *dev)
+{
+	if (!test_bit(__LINK_STATE_RX_SCHED, &dev->state)) BUG();
+	list_del(&dev->poll_list);
+	clear_bit(__LINK_STATE_RX_SCHED, &dev->state);
+}
+
+static inline void tg3_netif_stop(struct tg3 *tp)
+{
+	netif_stop_queue(tp->dev);
+	netif_poll_disable(tp->dev);
+}
+
+static inline void tg3_netif_start(struct tg3 *tp)
+{
+	netif_poll_enable(tp->dev);
+	netif_wake_queue(tp->dev);
+	/* NOTE: unconditional netif_wake_queue is only appropriate
+	 * so long as all callers are assured to have free tx slots
+	 * (such as after tg3_init_hw)
+	 */
+}
+
 static void tg3_switch_clocks(struct tg3 *tp)
 {
 	if (tr32(TG3PCI_CLOCK_CTRL) & CLOCK_CTRL_44MHZ_CORE) {
@@ -387,7 +428,6 @@ static int tg3_phy_reset(struct tg3 *tp, int force)
 }
 
 static int tg3_setup_phy(struct tg3 *);
-static int tg3_halt(struct tg3 *);
 
 static int tg3_set_power_state(struct tg3 *tp, int state)
 {
@@ -458,8 +498,6 @@ static int tg3_set_power_state(struct tg3 *tp, int state)
 		tg3_setup_phy(tp);
 	}
 
-	tg3_halt(tp);
-
 	pci_read_config_word(tp->pdev, pm + PCI_PM_PMC, &power_caps);
 
 	if (tp->tg3_flags & TG3_FLAG_WOL_ENABLE) {
@@ -2044,7 +2082,12 @@ static int tg3_poll(struct net_device *netdev, int *budget)
 		spin_unlock(&tp->tx_lock);
 	}
 
-	/* run RX thread, within the bounds set by NAPI */
+	spin_unlock_irqrestore(&tp->lock, flags);
+
+	/* run RX thread, within the bounds set by NAPI.
+	 * All RX "locking" is done by ensuring outside
+	 * code synchronizes with dev->poll()
+	 */
 	done = 1;
 	if (sblk->idx[0].rx_producer != tp->rx_rcb_ptr) {
 		int orig_budget = *budget;
@@ -2064,11 +2107,11 @@ static int tg3_poll(struct net_device *netdev, int *budget)
 
 	/* if no more work, tell net stack and NIC we're done */
 	if (done) {
-		netif_rx_complete(netdev);
+		spin_lock_irqsave(&tp->lock, flags);
+		__netif_rx_complete(netdev);
 		tg3_enable_ints(tp);
-	}
-
 		spin_unlock_irqrestore(&tp->lock, flags);
+	}
 
 	return (done ? 0 : 1);
 }
@@ -2136,17 +2179,21 @@ static void tg3_interrupt(int irq, void *dev_id, struct pt_regs *regs)
 
 static void tg3_init_rings(struct tg3 *);
 static int tg3_init_hw(struct tg3 *);
+static int tg3_halt(struct tg3 *);
 
-static void tg3_tx_timeout(struct net_device *dev)
+static void tg3_reset_task(void *_data)
 {
-	struct tg3 *tp = dev->priv;
+	struct tg3 *tp = _data;
+	unsigned int restart_timer;
 
-	printk(KERN_ERR PFX "%s: transmit timed out, resetting\n",
-	       dev->name);
+	tg3_netif_stop(tp);
 
 	spin_lock_irq(&tp->lock);
 	spin_lock(&tp->tx_lock);
 
+	restart_timer = tp->tg3_flags2 & TG3_FLG2_RESTART_TIMER;
+	tp->tg3_flags2 &= ~TG3_FLG2_RESTART_TIMER;
+
 	tg3_halt(tp);
 	tg3_init_rings(tp);
 	tg3_init_hw(tp);
@@ -2154,7 +2201,20 @@ static void tg3_tx_timeout(struct net_device *dev)
 	spin_unlock(&tp->tx_lock);
 	spin_unlock_irq(&tp->lock);
 
-	netif_wake_queue(dev);
+	tg3_netif_start(tp);
+
+	if (restart_timer)
+		mod_timer(&tp->timer, jiffies + 1);
+}
+
+static void tg3_tx_timeout(struct net_device *dev)
+{
+	struct tg3 *tp = dev->priv;
+
+	printk(KERN_ERR PFX "%s: transmit timed out, resetting\n",
+	       dev->name);
+
+	schedule_work(&tp->reset_task);
 }
 
 #if !PCI_DMA_BUS_IS_PHYS
@@ -2686,6 +2746,7 @@ static int tg3_change_mtu(struct net_device *dev, int new_mtu)
 		return 0;
 	}
 
+	tg3_netif_stop(tp);
 	spin_lock_irq(&tp->lock);
 	spin_lock(&tp->tx_lock);
 
@@ -2698,6 +2759,7 @@ static int tg3_change_mtu(struct net_device *dev, int new_mtu)
 
 	spin_unlock(&tp->tx_lock);
 	spin_unlock_irq(&tp->lock);
+	tg3_netif_start(tp);
 
 	return 0;
 }
@@ -4408,9 +4470,11 @@ static void tg3_timer(unsigned long __opaque)
 	}
 
 	if (!(tr32(WDMAC_MODE) & WDMAC_MODE_ENABLE)) {
-		tg3_halt(tp);
-		tg3_init_rings(tp);
-		tg3_init_hw(tp);
+		tp->tg3_flags2 |= TG3_FLG2_RESTART_TIMER;
+		spin_unlock(&tp->tx_lock);
+		spin_unlock_irqrestore(&tp->lock, flags);
+		schedule_work(&tp->reset_task);
+		return;
 	}
 
 	/* This part only runs once per second. */
@@ -4541,8 +4605,6 @@ static int tg3_open(struct net_device *dev)
 		return err;
 	}
 
-	netif_start_queue(dev);
-
 	spin_lock_irq(&tp->lock);
 	spin_lock(&tp->tx_lock);
 
@@ -4551,6 +4613,8 @@ static int tg3_open(struct net_device *dev)
 	spin_unlock(&tp->tx_lock);
 	spin_unlock_irq(&tp->lock);
 
+	netif_start_queue(dev);
+
 	return 0;
 }
 
@@ -5316,6 +5380,7 @@ static int tg3_ethtool_ioctl (struct net_device *dev, void *useraddr)
 		    (ering.tx_pending > TG3_TX_RING_SIZE - 1))
 			return -EINVAL;
 
+		tg3_netif_stop(tp);
 		spin_lock_irq(&tp->lock);
 		spin_lock(&tp->tx_lock);
 
@@ -5329,6 +5394,7 @@ static int tg3_ethtool_ioctl (struct net_device *dev, void *useraddr)
 		netif_wake_queue(tp->dev);
 		spin_unlock(&tp->tx_lock);
 		spin_unlock_irq(&tp->lock);
+		tg3_netif_start(tp);
 
 		return 0;
 	}
@@ -5351,6 +5417,7 @@ static int tg3_ethtool_ioctl (struct net_device *dev, void *useraddr)
 		if (copy_from_user(&epause, useraddr, sizeof(epause)))
 			return -EFAULT;
 
+		tg3_netif_stop(tp);
 		spin_lock_irq(&tp->lock);
 		spin_lock(&tp->tx_lock);
 		if (epause.autoneg)
@@ -5370,6 +5437,7 @@ static int tg3_ethtool_ioctl (struct net_device *dev, void *useraddr)
 		tg3_init_hw(tp);
 		spin_unlock(&tp->tx_lock);
 		spin_unlock_irq(&tp->lock);
+		tg3_netif_start(tp);
 
 		return 0;
 	}
@@ -6724,6 +6792,7 @@ static int __devinit tg3_init_one(struct pci_dev *pdev,
 	spin_lock_init(&tp->lock);
 	spin_lock_init(&tp->tx_lock);
 	spin_lock_init(&tp->indirect_lock);
+	PREPARE_WORK(&tp->reset_task, tg3_reset_task, tp);
 
 	tp->regs = (unsigned long) ioremap(tg3reg_base, tg3reg_len);
 	if (tp->regs == 0UL) {
@@ -6865,6 +6934,8 @@ static int tg3_suspend(struct pci_dev *pdev, u32 state)
 	if (!netif_running(dev))
 		return 0;
 
+	tg3_netif_stop(tp);
+
 	spin_lock_irq(&tp->lock);
 	spin_lock(&tp->tx_lock);
 	tg3_disable_ints(tp);
@@ -6891,6 +6962,7 @@ static int tg3_suspend(struct pci_dev *pdev, u32 state)
 		spin_unlock_irq(&tp->lock);
 
 		netif_device_attach(dev);
+		tg3_netif_start(tp);
 	}
 
 	return err;
@@ -6921,6 +6993,8 @@ static int tg3_resume(struct pci_dev *pdev)
 	spin_unlock(&tp->tx_lock);
 	spin_unlock_irq(&tp->lock);
 
+	tg3_netif_start(tp);
+
 	return 0;
 }
 

drivers/net/tg3.h

@@ -1821,6 +1821,8 @@ struct tg3 {
 #define TG3_FLAG_GOT_SERDES_FLOWCTL	0x20000000
 #define TG3_FLAG_SPLIT_MODE		0x40000000
 #define TG3_FLAG_INIT_COMPLETE		0x80000000
+	u32				tg3_flags2;
+#define TG3_FLG2_RESTART_TIMER		0x00000001
 
 	u32				split_mode_max_reqs;
 #define SPLIT_MODE_5704_MAX_REQ		3
@@ -1889,6 +1891,7 @@ struct tg3 {
 
 	struct tg3_hw_stats		*hw_stats;
 	dma_addr_t			stats_mapping;
+	struct work_struct		reset_task;
 };
 
 #endif /* !(_T3_H) */