Merge gregkh@master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6

10 files changed, 72 insertions, 41 deletions

diff --git a/drivers/net/tg3.c b/drivers/net/tg3.c
index 6f97962..eafabb2 100644
--- a/drivers/net/tg3.c
+++ b/drivers/net/tg3.c
@@ -68,8 +68,8 @@
 
 #define DRV_MODULE_NAME		"tg3"
 #define PFX DRV_MODULE_NAME	": "
-#define DRV_MODULE_VERSION	"3.64"
-#define DRV_MODULE_RELDATE	"July 31, 2006"
+#define DRV_MODULE_VERSION	"3.65"
+#define DRV_MODULE_RELDATE	"August 07, 2006"
 
 #define TG3_DEF_MAC_MODE	0
 #define TG3_DEF_RX_MODE		0
@@ -123,9 +123,6 @@
 			           TG3_RX_RCB_RING_SIZE(tp))
 #define TG3_TX_RING_BYTES	(sizeof(struct tg3_tx_buffer_desc) * \
 				 TG3_TX_RING_SIZE)
-#define TX_BUFFS_AVAIL(TP)						\
-	((TP)->tx_pending -						\
-	 (((TP)->tx_prod - (TP)->tx_cons) & (TG3_TX_RING_SIZE - 1)))
 #define NEXT_TX(N)		(((N) + 1) & (TG3_TX_RING_SIZE - 1))
 
 #define RX_PKT_BUF_SZ		(1536 + tp->rx_offset + 64)
@@ -2987,6 +2984,13 @@ static void tg3_tx_recover(struct tg3 *tp)
 	spin_unlock(&tp->lock);
 }
 
+static inline u32 tg3_tx_avail(struct tg3 *tp)
+{
+	smp_mb();
+	return (tp->tx_pending -
+		((tp->tx_prod - tp->tx_cons) & (TG3_TX_RING_SIZE - 1)));
+}
+
 /* Tigon3 never reports partial packet sends.  So we do not
  * need special logic to handle SKBs that have not had all
  * of their frags sent yet, like SunGEM does.
@@ -3038,12 +3042,20 @@ static void tg3_tx(struct tg3 *tp)
 
 	tp->tx_cons = sw_idx;
 
-	if (unlikely(netif_queue_stopped(tp->dev))) {
-		spin_lock(&tp->tx_lock);
+	/* Need to make the tx_cons update visible to tg3_start_xmit()
+	 * before checking for netif_queue_stopped().  Without the
+	 * memory barrier, there is a small possibility that tg3_start_xmit()
+	 * will miss it and cause the queue to be stopped forever.
+	 */
+	smp_mb();
+
+	if (unlikely(netif_queue_stopped(tp->dev) &&
+		     (tg3_tx_avail(tp) > TG3_TX_WAKEUP_THRESH))) {
+		netif_tx_lock(tp->dev);
 		if (netif_queue_stopped(tp->dev) &&
-		    (TX_BUFFS_AVAIL(tp) > TG3_TX_WAKEUP_THRESH))
+		    (tg3_tx_avail(tp) > TG3_TX_WAKEUP_THRESH))
 			netif_wake_queue(tp->dev);
-		spin_unlock(&tp->tx_lock);
+		netif_tx_unlock(tp->dev);
 	}
 }
 
@@ -3101,7 +3113,6 @@ static int tg3_alloc_rx_skb(struct tg3 *tp, u32 opaque_key,
 	if (skb == NULL)
 		return -ENOMEM;
 
-	skb->dev = tp->dev;
 	skb_reserve(skb, tp->rx_offset);
 
 	mapping = pci_map_single(tp->pdev, skb->data,
@@ -3274,7 +3285,6 @@ static int tg3_rx(struct tg3 *tp, int budget)
 			if (copy_skb == NULL)
 				goto drop_it_no_recycle;
 
-			copy_skb->dev = tp->dev;
 			skb_reserve(copy_skb, 2);
 			skb_put(copy_skb, len);
 			pci_dma_sync_single_for_cpu(tp->pdev, dma_addr, len, PCI_DMA_FROMDEVICE);
@@ -3797,7 +3807,7 @@ static int tg3_start_xmit(struct sk_buff *skb, struct net_device *dev)
 	 * interrupt.  Furthermore, IRQ processing runs lockless so we have
 	 * no IRQ context deadlocks to worry about either.  Rejoice!
 	 */
-	if (unlikely(TX_BUFFS_AVAIL(tp) <= (skb_shinfo(skb)->nr_frags + 1))) {
+	if (unlikely(tg3_tx_avail(tp) <= (skb_shinfo(skb)->nr_frags + 1))) {
 		if (!netif_queue_stopped(dev)) {
 			netif_stop_queue(dev);
 
@@ -3893,12 +3903,10 @@ static int tg3_start_xmit(struct sk_buff *skb, struct net_device *dev)
 	tw32_tx_mbox((MAILBOX_SNDHOST_PROD_IDX_0 + TG3_64BIT_REG_LOW), entry);
 
 	tp->tx_prod = entry;
-	if (unlikely(TX_BUFFS_AVAIL(tp) <= (MAX_SKB_FRAGS + 1))) {
-		spin_lock(&tp->tx_lock);
+	if (unlikely(tg3_tx_avail(tp) <= (MAX_SKB_FRAGS + 1))) {
 		netif_stop_queue(dev);
-		if (TX_BUFFS_AVAIL(tp) > TG3_TX_WAKEUP_THRESH)
+		if (tg3_tx_avail(tp) > TG3_TX_WAKEUP_THRESH)
 			netif_wake_queue(tp->dev);
-		spin_unlock(&tp->tx_lock);
 	}
 
 out_unlock:
@@ -3920,7 +3928,7 @@ static int tg3_tso_bug(struct tg3 *tp, struct sk_buff *skb)
 	struct sk_buff *segs, *nskb;
 
 	/* Estimate the number of fragments in the worst case */
-	if (unlikely(TX_BUFFS_AVAIL(tp) <= (skb_shinfo(skb)->gso_segs * 3))) {
+	if (unlikely(tg3_tx_avail(tp) <= (skb_shinfo(skb)->gso_segs * 3))) {
 		netif_stop_queue(tp->dev);
 		return NETDEV_TX_BUSY;
 	}
@@ -3960,7 +3968,7 @@ static int tg3_start_xmit_dma_bug(struct sk_buff *skb, struct net_device *dev)
 	 * interrupt.  Furthermore, IRQ processing runs lockless so we have
 	 * no IRQ context deadlocks to worry about either.  Rejoice!
 	 */
-	if (unlikely(TX_BUFFS_AVAIL(tp) <= (skb_shinfo(skb)->nr_frags + 1))) {
+	if (unlikely(tg3_tx_avail(tp) <= (skb_shinfo(skb)->nr_frags + 1))) {
 		if (!netif_queue_stopped(dev)) {
 			netif_stop_queue(dev);
 
@@ -4110,12 +4118,10 @@ static int tg3_start_xmit_dma_bug(struct sk_buff *skb, struct net_device *dev)
 	tw32_tx_mbox((MAILBOX_SNDHOST_PROD_IDX_0 + TG3_64BIT_REG_LOW), entry);
 
 	tp->tx_prod = entry;
-	if (unlikely(TX_BUFFS_AVAIL(tp) <= (MAX_SKB_FRAGS + 1))) {
-		spin_lock(&tp->tx_lock);
+	if (unlikely(tg3_tx_avail(tp) <= (MAX_SKB_FRAGS + 1))) {
 		netif_stop_queue(dev);
-		if (TX_BUFFS_AVAIL(tp) > TG3_TX_WAKEUP_THRESH)
+		if (tg3_tx_avail(tp) > TG3_TX_WAKEUP_THRESH)
 			netif_wake_queue(tp->dev);
-		spin_unlock(&tp->tx_lock);
 	}
 
 out_unlock:
@@ -11474,7 +11480,6 @@ static int __devinit tg3_init_one(struct pci_dev *pdev,
 	tp->grc_mode |= GRC_MODE_BSWAP_NONFRM_DATA;
 #endif
 	spin_lock_init(&tp->lock);
-	spin_lock_init(&tp->tx_lock);
 	spin_lock_init(&tp->indirect_lock);
 	INIT_WORK(&tp->reset_task, tg3_reset_task, tp);
 
diff --git a/drivers/net/tg3.h b/drivers/net/tg3.h
index ba2c987..3ecf356 100644
--- a/drivers/net/tg3.h
+++ b/drivers/net/tg3.h
@@ -2079,9 +2079,9 @@ struct tg3 {
 	 * lock: Held during reset, PHY access, timer, and when
 	 *       updating tg3_flags and tg3_flags2.
 	 *
-	 * tx_lock: Held during tg3_start_xmit and tg3_tx only
-	 *          when calling netif_[start|stop]_queue.
-	 *          tg3_start_xmit is protected by netif_tx_lock.
+	 * netif_tx_lock: Held during tg3_start_xmit. tg3_tx holds
+	 *                netif_tx_lock when it needs to call
+	 *                netif_wake_queue.
 	 *
 	 * Both of these locks are to be held with BH safety.
 	 *
@@ -2118,8 +2118,6 @@ struct tg3 {
 	u32				tx_cons;
 	u32				tx_pending;
 
-	spinlock_t			tx_lock;
-
 	struct tg3_tx_buffer_desc	*tx_ring;
 	struct tx_ring_info		*tx_buffers;
 	dma_addr_t			tx_desc_mapping;
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 19c96d4..3573ba9 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -1081,7 +1081,7 @@ static inline void __skb_queue_purge(struct sk_buff_head *list)
  *	the headroom they think they need without accounting for the
  *	built in space. The built in space is used for optimisations.
  *
- *	%NULL is returned in there is no free memory.
+ *	%NULL is returned if there is no free memory.
  */
 static inline struct sk_buff *__dev_alloc_skb(unsigned int length,
 					      gfp_t gfp_mask)
@@ -1101,7 +1101,7 @@ static inline struct sk_buff *__dev_alloc_skb(unsigned int length,
  *	the headroom they think they need without accounting for the
  *	built in space. The built in space is used for optimisations.
  *
- *	%NULL is returned in there is no free memory. Although this function
+ *	%NULL is returned if there is no free memory. Although this function
  *	allocates memory it can be called from an interrupt.
  */
 static inline struct sk_buff *dev_alloc_skb(unsigned int length)
diff --git a/net/core/dst.c b/net/core/dst.c
index 470c05b..1a5e49d 100644
--- a/net/core/dst.c
+++ b/net/core/dst.c
@@ -95,12 +95,11 @@ static void dst_run_gc(unsigned long dummy)
 		dst_gc_timer_inc = DST_GC_INC;
 		dst_gc_timer_expires = DST_GC_MIN;
 	}
-	dst_gc_timer.expires = jiffies + dst_gc_timer_expires;
 #if RT_CACHE_DEBUG >= 2
 	printk("dst_total: %d/%d %ld\n",
 	       atomic_read(&dst_total), delayed,  dst_gc_timer_expires);
 #endif
-	add_timer(&dst_gc_timer);
+	mod_timer(&dst_gc_timer, jiffies + dst_gc_timer_expires);
 
 out:
 	spin_unlock(&dst_lock);
diff --git a/net/core/pktgen.c b/net/core/pktgen.c
index 67ed14d..6a7320b 100644
--- a/net/core/pktgen.c
+++ b/net/core/pktgen.c
@@ -2149,6 +2149,8 @@ static struct sk_buff *fill_packet_ipv4(struct net_device *odev,
 	skb->mac.raw = ((u8 *) iph) - 14 - pkt_dev->nr_labels*sizeof(u32);
 	skb->dev = odev;
 	skb->pkt_type = PACKET_HOST;
+	skb->nh.iph = iph;
+	skb->h.uh = udph;
 
 	if (pkt_dev->nfrags <= 0)
 		pgh = (struct pktgen_hdr *)skb_put(skb, datalen);
@@ -2460,6 +2462,8 @@ static struct sk_buff *fill_packet_ipv6(struct net_device *odev,
 	skb->protocol = protocol;
 	skb->dev = odev;
 	skb->pkt_type = PACKET_HOST;
+	skb->nh.ipv6h = iph;
+	skb->h.uh = udph;
 
 	if (pkt_dev->nfrags <= 0)
 		pgh = (struct pktgen_hdr *)skb_put(skb, datalen);
diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index 20e5bb7..30cc1ba 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -394,6 +394,9 @@ static int do_setlink(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 	}
 
 	if (ida[IFLA_ADDRESS - 1]) {
+		struct sockaddr *sa;
+		int len;
+
 		if (!dev->set_mac_address) {
 			err = -EOPNOTSUPP;
 			goto out;
@@ -405,7 +408,17 @@ static int do_setlink(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 		if (ida[IFLA_ADDRESS - 1]->rta_len != RTA_LENGTH(dev->addr_len))
 			goto out;
 
-		err = dev->set_mac_address(dev, RTA_DATA(ida[IFLA_ADDRESS - 1]));
+		len = sizeof(sa_family_t) + dev->addr_len;
+		sa = kmalloc(len, GFP_KERNEL);
+		if (!sa) {
+			err = -ENOMEM;
+			goto out;
+		}
+		sa->sa_family = dev->type;
+		memcpy(sa->sa_data, RTA_DATA(ida[IFLA_ADDRESS - 1]),
+		       dev->addr_len);
+		err = dev->set_mac_address(dev, sa);
+		kfree(sa);
 		if (err)
 			goto out;
 		send_addr_notify = 1;
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 022d889..c54f366 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -268,8 +268,10 @@ struct sk_buff *__netdev_alloc_skb(struct net_device *dev,
 	struct sk_buff *skb;
 
 	skb = alloc_skb(length + NET_SKB_PAD, gfp_mask);
-	if (likely(skb))
+	if (likely(skb)) {
 		skb_reserve(skb, NET_SKB_PAD);
+		skb->dev = dev;
+	}
 	return skb;
 }
 
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index 19bd49d..b873cbc 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -3157,7 +3157,7 @@ int __init ip_rt_init(void)
 					rhash_entries,
 					(num_physpages >= 128 * 1024) ?
 					15 : 17,
-					HASH_HIGHMEM,
+					0,
 					&rt_hash_log,
 					&rt_hash_mask,
 					0);
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 5c08ea2..507adef 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -466,7 +466,8 @@ static int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it,
 	if (skb->len != tcp_header_size)
 		tcp_event_data_sent(tp, skb, sk);
 
-	TCP_INC_STATS(TCP_MIB_OUTSEGS);
+	if (after(tcb->end_seq, tp->snd_nxt) || tcb->seq == tcb->end_seq)
+		TCP_INC_STATS(TCP_MIB_OUTSEGS);
 
 	err = icsk->icsk_af_ops->queue_xmit(skb, 0);
 	if (likely(err <= 0))
@@ -2157,10 +2158,9 @@ int tcp_connect(struct sock *sk)
 	skb_shinfo(buff)->gso_size = 0;
 	skb_shinfo(buff)->gso_type = 0;
 	buff->csum = 0;
+	tp->snd_nxt = tp->write_seq;
 	TCP_SKB_CB(buff)->seq = tp->write_seq++;
 	TCP_SKB_CB(buff)->end_seq = tp->write_seq;
-	tp->snd_nxt = tp->write_seq;
-	tp->pushed_seq = tp->write_seq;
 
 	/* Send it off. */
 	TCP_SKB_CB(buff)->when = tcp_time_stamp;
@@ -2170,6 +2170,12 @@ int tcp_connect(struct sock *sk)
 	sk_charge_skb(sk, buff);
 	tp->packets_out += tcp_skb_pcount(buff);
 	tcp_transmit_skb(sk, buff, 1, GFP_KERNEL);
+
+	/* We change tp->snd_nxt after the tcp_transmit_skb() call
+	 * in order to make this packet get counted in tcpOutSegs.
+	 */
+	tp->snd_nxt = tp->write_seq;
+	tp->pushed_seq = tp->write_seq;
 	TCP_INC_STATS(TCP_MIB_ACTIVEOPENS);
 
 	/* Timer for repeating the SYN until an answer. */
diff --git a/net/ipx/af_ipx.c b/net/ipx/af_ipx.c
index aa34ff4..40196420 100644
--- a/net/ipx/af_ipx.c
+++ b/net/ipx/af_ipx.c
@@ -1642,13 +1642,17 @@ static int ipx_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_ty
 	if ((skb = skb_share_check(skb, GFP_ATOMIC)) == NULL)
 		goto out;
 
-	ipx		= ipx_hdr(skb);
-	ipx_pktsize	= ntohs(ipx->ipx_pktsize);
+	if (!pskb_may_pull(skb, sizeof(struct ipxhdr)))
+		goto drop;
+
+	ipx_pktsize = ntohs(ipxhdr(skb)->ipx_pktsize);
 	
 	/* Too small or invalid header? */
-	if (ipx_pktsize < sizeof(struct ipxhdr) || ipx_pktsize > skb->len)
+	if (ipx_pktsize < sizeof(struct ipxhdr) ||
+	    !pskb_may_pull(skb, ipx_pktsize))
 		goto drop;
                         
+	ipx = ipx_hdr(skb);
 	if (ipx->ipx_checksum != IPX_NO_CHECKSUM &&
 	   ipx->ipx_checksum != ipx_cksum(ipx, ipx_pktsize))
 		goto drop;