oom handling added into winbond-840.c
data members in netdev_private reordered for better readability (and
perhaps better cache line independace of TX and RX codepaths).
Tested on i386 SMP.
-- Manfred --------------F81AE2453B592209FBC7FA27 Content-Type: text/plain; charset=us-ascii; name="patch-winbond-oom" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="patch-winbond-oom"--- 2.5/drivers/net/winbond-840.c Thu Oct 11 15:19:59 2001 +++ build-2.5/drivers/net/winbond-840.c Sun Jan 13 18:57:26 2002 @@ -36,6 +36,8 @@ power management. support for big endian descriptors Copyright (C) 2001 Manfred Spraul + * OOM handling + Copyright (C) 2002 Manfred Spraul TODO: * enable pci_power_off @@ -338,39 +340,52 @@ DescIntr=0x80000000, }; -#define PRIV_ALIGN 15 /* Required alignment mask */ #define MII_CNT 1 /* winbond only supports one MII */ struct netdev_private { + /* 1) RX data - accessed only by the hw irq handler, + * by the timer after disable_irq(), + * by the resume & shutdown code, after del_timer_sync() + * and disable_irq(). + */ struct w840_rx_desc *rx_ring; - dma_addr_t rx_addr[RX_RING_SIZE]; - struct w840_tx_desc *tx_ring; - dma_addr_t tx_addr[RX_RING_SIZE]; - dma_addr_t ring_dma_addr; + struct w840_rx_desc *rx_head_desc; + unsigned int rx_buf_sz; /* Based on MTU+slack. */ + unsigned int cur_rx, dirty_rx; /* Producer/consumer ring indices */ + int oom; /* The addresses of receive-in-place skbuffs. */ struct sk_buff* rx_skbuff[RX_RING_SIZE]; - /* The saved address of a sent-in-place packet/buffer, for later free(). */ - struct sk_buff* tx_skbuff[TX_RING_SIZE]; + dma_addr_t rx_addr[RX_RING_SIZE]; + + /* 2) misc members, most for media setup */ + dma_addr_t ring_dma_addr; struct net_device_stats stats; struct timer_list timer; /* Media monitoring timer. */ - /* Frequently used values: keep some adjacent for cache effect. */ - spinlock_t lock; int chip_id, drv_flags; struct pci_dev *pci_dev; int csr6; - struct w840_rx_desc *rx_head_desc; - unsigned int cur_rx, dirty_rx; /* Producer/consumer ring indices */ - unsigned int rx_buf_sz; /* Based on MTU+slack. */ - unsigned int cur_tx, dirty_tx; - unsigned int tx_q_bytes; - unsigned int tx_full; /* The Tx queue is full. */ /* These values are keep track of the transceiver/media in use. */ unsigned int full_duplex:1; /* Full-duplex operation requested. */ unsigned int duplex_lock:1; /* MII transceiver section. */ - int mii_cnt; /* MII device addresses. */ - u16 advertising; /* NWay media advertisement */ - unsigned char phys[MII_CNT]; /* MII device addresses, but only the first is used */ + int mii_cnt; /* MII device addresses. */ + u16 advertising; /* NWay media advertisement */ + unsigned char phys[MII_CNT]; u32 mii; + + /* 3) TX data - accessed under spin_lock_irq(np->lock);. + * Exception: start_tx (protected by dev_xmit_lock) reads + * cur_tx + * The spinlock is part of TX data, since packet TX + * is the only common codepath that needs the lock. + */ + spinlock_t lock; + struct w840_tx_desc *tx_ring; + unsigned int cur_tx, dirty_tx; + /* The saved address of a sent-in-place packet/buffer, for later free(). */ + unsigned int tx_q_bytes; + unsigned int tx_full; /* The Tx queue is full. */ + struct sk_buff* tx_skbuff[TX_RING_SIZE]; + dma_addr_t tx_addr[TX_RING_SIZE]; }; static int eeprom_read(long ioaddr, int location); @@ -378,6 +393,7 @@ static void mdio_write(struct net_device *dev, int phy_id, int location, int value); static int netdev_open(struct net_device *dev); static int update_link(struct net_device *dev); +static void refill_rx(struct net_device *dev); static void netdev_timer(unsigned long data); static void init_rxtx_rings(struct net_device *dev); static void free_rxtx_rings(struct netdev_private *np); @@ -825,11 +841,40 @@ np->full_duplex = 1; } +static void refill_rx(struct net_device *dev) +{ + struct netdev_private *np = dev->priv; + + /* Refill the Rx ring buffers. */ + np->oom = 0; + for (; np->cur_rx - np->dirty_rx > 0; np->dirty_rx++) { + struct sk_buff *skb; + int entry; + entry = np->dirty_rx % RX_RING_SIZE; + if (np->rx_skbuff[entry] == NULL) { + skb = dev_alloc_skb(np->rx_buf_sz); + np->rx_skbuff[entry] = skb; + if (skb == NULL) + break; /* Better luck next round. */ + skb->dev = dev; /* Mark as being used by this device. */ + np->rx_addr[entry] = pci_map_single(np->pci_dev, + skb->tail, + skb->len, PCI_DMA_FROMDEVICE); + np->rx_ring[entry].buffer1 = np->rx_addr[entry]; + } + wmb(); + np->rx_ring[entry].status = DescOwn; + } + if (np->cur_rx-np->dirty_rx == RX_RING_SIZE) + np->oom = 1; +} + static void netdev_timer(unsigned long data) { struct net_device *dev = (struct net_device *)data; struct netdev_private *np = dev->priv; long ioaddr = dev->base_addr; + long next = 10*HZ; if (debug > 2) printk(KERN_DEBUG "%s: Media selection timer tick, status %8.8x " @@ -839,8 +884,16 @@ spin_lock_irq(&np->lock); update_csr6(dev, update_link(dev)); spin_unlock_irq(&np->lock); - np->timer.expires = jiffies + 10*HZ; - add_timer(&np->timer); + if (np->oom) { + disable_irq(dev->irq); + refill_rx(dev); + if (np->oom) + next = 1; + else + writel(0, ioaddr + RxStartDemand); + enable_irq(dev->irq); + } + mod_timer(&np->timer, jiffies + next); } static void init_rxtx_rings(struct net_device *dev) @@ -860,22 +913,9 @@ /* Mark the last entry as wrapping the ring. */ np->rx_ring[i-1].length |= DescEndRing; - /* Fill in the Rx buffers. Handle allocation failure gracefully. */ - for (i = 0; i < RX_RING_SIZE; i++) { - struct sk_buff *skb = dev_alloc_skb(np->rx_buf_sz); - np->rx_skbuff[i] = skb; - if (skb == NULL) - break; - skb->dev = dev; /* Mark as being used by this device. */ - np->rx_addr[i] = pci_map_single(np->pci_dev,skb->tail, - skb->len,PCI_DMA_FROMDEVICE); - - np->rx_ring[i].buffer1 = np->rx_addr[i]; - np->rx_ring[i].status = DescOwn; - } - - np->cur_rx = 0; - np->dirty_rx = (unsigned int)(i - RX_RING_SIZE); + np->cur_rx = RX_RING_SIZE; + np->dirty_rx = 0; + refill_rx(dev); /* Initialize the Tx descriptors */ for (i = 0; i < TX_RING_SIZE; i++) { @@ -1193,8 +1233,9 @@ if (intr_status & (IntrRxDone | RxNoBuf)) netdev_rx(dev); - if (intr_status & RxNoBuf) - writel(0, ioaddr + RxStartDemand); + /* RxNoBuf is an error interrupt, we kick + * RxStartDemand in netdev_error() + */ if (intr_status & (TxIdle | IntrTxDone) && np->cur_tx != np->dirty_tx) { @@ -1326,25 +1367,7 @@ entry = (++np->cur_rx) % RX_RING_SIZE; np->rx_head_desc = &np->rx_ring[entry]; } - - /* Refill the Rx ring buffers. */ - for (; np->cur_rx - np->dirty_rx > 0; np->dirty_rx++) { - struct sk_buff *skb; - entry = np->dirty_rx % RX_RING_SIZE; - if (np->rx_skbuff[entry] == NULL) { - skb = dev_alloc_skb(np->rx_buf_sz); - np->rx_skbuff[entry] = skb; - if (skb == NULL) - break; /* Better luck next round. */ - skb->dev = dev; /* Mark as being used by this device. */ - np->rx_addr[entry] = pci_map_single(np->pci_dev, - skb->tail, - skb->len, PCI_DMA_FROMDEVICE); - np->rx_ring[entry].buffer1 = np->rx_addr[entry]; - } - wmb(); - np->rx_ring[entry].status = DescOwn; - } + refill_rx(dev); return 0; } @@ -1388,8 +1411,14 @@ if (netif_device_present(dev)) writel(0x1A0F5, ioaddr + IntrEnable); } + /* strictly only needed with RxNoBuf, but + * kicking too often probably doesn't hurt. + */ + if (np->oom) + mod_timer(&np->timer, jiffies+1); + else + writel(0, ioaddr + RxStartDemand); np->stats.rx_missed_errors += readl(ioaddr + RxMissed) & 0xffff; - writel(0, ioaddr + RxStartDemand); spin_unlock(&np->lock); } @@ -1537,6 +1566,7 @@ dev->name, np->cur_tx, np->dirty_tx, np->cur_rx, np->dirty_rx); } + del_timer_sync(&np->timer); /* Stop the chip's Tx and Rx processes. */ spin_lock_irq(&np->lock); netif_device_detach(dev); @@ -1555,10 +1585,10 @@ if (debug > 2) { int i; - printk("\n"KERN_DEBUG" Tx ring at %8.8x:\n", + printk(KERN_DEBUG" Tx ring at %8.8x:\n", (int)np->tx_ring); for (i = 0; i < TX_RING_SIZE; i++) - printk(" #%d desc. %4.4x %4.4x %8.8x.\n", + printk(KERN_DEBUG " #%d desc. %4.4x %4.4x %8.8x.\n", i, np->tx_ring[i].length, np->tx_ring[i].status, np->tx_ring[i].buffer1); printk("\n"KERN_DEBUG " Rx ring %8.8x:\n", @@ -1571,8 +1601,6 @@ } #endif /* __i386__ debugging only */ - del_timer_sync(&np->timer); - free_rxtx_rings(np); free_ringdesc(np); @@ -1658,7 +1686,6 @@ return 0; } - static int w840_resume (struct pci_dev *pdev) { struct net_device *dev = pci_get_drvdata (pdev); @@ -1682,8 +1709,7 @@ netif_wake_queue(dev); - np->timer.expires = jiffies + 1*HZ; - add_timer(&np->timer); + mod_timer(&np->timer, jiffies + 1*HZ); } else { netif_device_attach(dev); }
--------------F81AE2453B592209FBC7FA27--
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/