Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
[linux-3.10.git] / drivers / net / igb / igb_main.c
index e253435..ffd7315 100644 (file)
@@ -942,6 +942,8 @@ int igb_up(struct igb_adapter *adapter)
        rd32(E1000_ICR);
        igb_irq_enable(adapter);
 
+       netif_tx_start_all_queues(adapter->netdev);
+
        /* Fire a link change interrupt to start the watchdog. */
        wr32(E1000_ICS, E1000_ICS_LSC);
        return 0;
@@ -1343,6 +1345,9 @@ static int __devinit igb_probe(struct pci_dev *pdev,
        if (pci_using_dac)
                netdev->features |= NETIF_F_HIGHDMA;
 
+       if (adapter->hw.mac.type == e1000_82576)
+               netdev->features |= NETIF_F_SCTP_CSUM;
+
        adapter->en_mng_pt = igb_enable_mng_pass_thru(&adapter->hw);
 
        /* before reading the NVM, reset the controller to put the device in a
@@ -1390,8 +1395,6 @@ static int __devinit igb_probe(struct pci_dev *pdev,
 
        igb_validate_mdi_setting(hw);
 
-       adapter->rx_csum = 1;
-
        /* Initial Wake on LAN setting If APM wake is enabled in the EEPROM,
         * enable the ACPI Magic Packet filter
         */
@@ -1442,15 +1445,14 @@ static int __devinit igb_probe(struct pci_dev *pdev,
         * driver. */
        igb_get_hw_control(adapter);
 
-       /* tell the stack to leave us alone until igb_open() is called */
-       netif_carrier_off(netdev);
-       netif_tx_stop_all_queues(netdev);
-
        strcpy(netdev->name, "eth%d");
        err = register_netdev(netdev);
        if (err)
                goto err_register;
 
+       /* carrier off reporting is important to ethtool even BEFORE open */
+       netif_carrier_off(netdev);
+
 #ifdef CONFIG_IGB_DCA
        if (dca_add_requester(&pdev->dev) == 0) {
                adapter->flags |= IGB_FLAG_DCA_ENABLED;
@@ -1699,6 +1701,8 @@ static int igb_open(struct net_device *netdev)
        if (test_bit(__IGB_TESTING, &adapter->state))
                return -EBUSY;
 
+       netif_carrier_off(netdev);
+
        /* allocate transmit descriptors */
        err = igb_setup_all_tx_resources(adapter);
        if (err)
@@ -2231,29 +2235,24 @@ static void igb_configure_rx(struct igb_adapter *adapter)
                mrqc |= (E1000_MRQC_RSS_FIELD_IPV6_UDP_EX |
                         E1000_MRQC_RSS_FIELD_IPV6_TCP_EX);
 
-
                wr32(E1000_MRQC, mrqc);
-
-               /* Multiqueue and raw packet checksumming are mutually
-                * exclusive.  Note that this not the same as TCP/IP
-                * checksumming, which works fine. */
-               rxcsum = rd32(E1000_RXCSUM);
-               rxcsum |= E1000_RXCSUM_PCSD;
-               wr32(E1000_RXCSUM, rxcsum);
-       } else {
+       } else if (adapter->vfs_allocated_count) {
                /* Enable multi-queue for sr-iov */
-               if (adapter->vfs_allocated_count)
-                       wr32(E1000_MRQC, E1000_MRQC_ENABLE_VMDQ);
-               /* Enable Receive Checksum Offload for TCP and UDP */
-               rxcsum = rd32(E1000_RXCSUM);
-               if (adapter->rx_csum)
-                       rxcsum |= E1000_RXCSUM_TUOFL | E1000_RXCSUM_IPPCSE;
-               else
-                       rxcsum &= ~(E1000_RXCSUM_TUOFL | E1000_RXCSUM_IPPCSE);
-
-               wr32(E1000_RXCSUM, rxcsum);
+               wr32(E1000_MRQC, E1000_MRQC_ENABLE_VMDQ);
        }
 
+       /* Enable Receive Checksum Offload for TCP and UDP */
+       rxcsum = rd32(E1000_RXCSUM);
+       /* Disable raw packet checksumming */
+       rxcsum |= E1000_RXCSUM_PCSD;
+
+       if (adapter->hw.mac.type == e1000_82576)
+               /* Enable Receive Checksum Offload for SCTP */
+               rxcsum |= E1000_RXCSUM_CRCOFL;
+
+       /* Don't need to set TUOFL or IPOFL, they default to 1 */
+       wr32(E1000_RXCSUM, rxcsum);
+
        /* Set the default pool for the PF's first queue */
        igb_configure_vt_default_pool(adapter);
 
@@ -2661,7 +2660,6 @@ static void igb_watchdog_task(struct work_struct *work)
                        }
 
                        netif_carrier_on(netdev);
-                       netif_tx_wake_all_queues(netdev);
 
                        igb_ping_all_vfs(adapter);
 
@@ -2678,7 +2676,6 @@ static void igb_watchdog_task(struct work_struct *work)
                        printk(KERN_INFO "igb: %s NIC Link is Down\n",
                               netdev->name);
                        netif_carrier_off(netdev);
-                       netif_tx_stop_all_queues(netdev);
 
                        igb_ping_all_vfs(adapter);
 
@@ -2712,6 +2709,8 @@ link_up:
                         * (Do the reset outside of interrupt context). */
                        adapter->tx_timeout_count++;
                        schedule_work(&adapter->reset_task);
+                       /* return immediately since reset is imminent */
+                       return;
                }
        }
 
@@ -2895,13 +2894,13 @@ static void igb_set_itr(struct igb_adapter *adapter)
        switch (current_itr) {
        /* counts and packets in update_itr are dependent on these numbers */
        case lowest_latency:
-               new_itr = 70000;
+               new_itr = 56;  /* aka 70,000 ints/sec */
                break;
        case low_latency:
-               new_itr = 20000; /* aka hwitr = ~200 */
+               new_itr = 196; /* aka 20,000 ints/sec */
                break;
        case bulk_latency:
-               new_itr = 4000;
+               new_itr = 980; /* aka 4,000 ints/sec */
                break;
        default:
                break;
@@ -2920,7 +2919,8 @@ set_itr_now:
                 * by adding intermediate steps when interrupt rate is
                 * increasing */
                new_itr = new_itr > adapter->itr ?
-                            min(adapter->itr + (new_itr >> 2), new_itr) :
+                            max((new_itr * adapter->itr) /
+                                (new_itr + (adapter->itr >> 2)), new_itr) :
                             new_itr;
                /* Don't write the value here; it resets the adapter's
                 * internal timer, and causes us to delay far longer than
@@ -2929,7 +2929,7 @@ set_itr_now:
                 * ends up being correct.
                 */
                adapter->itr = new_itr;
-               adapter->rx_ring->itr_val = 1000000000 / (new_itr * 256);
+               adapter->rx_ring->itr_val = new_itr;
                adapter->rx_ring->set_itr = 1;
        }
 
@@ -3068,11 +3068,15 @@ static inline bool igb_tx_csum_adv(struct igb_adapter *adapter,
                                tu_cmd |= E1000_ADVTXD_TUCMD_IPV4;
                                if (ip_hdr(skb)->protocol == IPPROTO_TCP)
                                        tu_cmd |= E1000_ADVTXD_TUCMD_L4T_TCP;
+                               else if (ip_hdr(skb)->protocol == IPPROTO_SCTP)
+                                       tu_cmd |= E1000_ADVTXD_TUCMD_L4T_SCTP;
                                break;
                        case cpu_to_be16(ETH_P_IPV6):
                                /* XXX what about other V6 headers?? */
                                if (ipv6_hdr(skb)->nexthdr == IPPROTO_TCP)
                                        tu_cmd |= E1000_ADVTXD_TUCMD_L4T_TCP;
+                               else if (ipv6_hdr(skb)->nexthdr == IPPROTO_SCTP)
+                                       tu_cmd |= E1000_ADVTXD_TUCMD_L4T_SCTP;
                                break;
                        default:
                                if (unlikely(net_ratelimit()))
@@ -4434,20 +4438,12 @@ static void igb_receive_skb(struct igb_ring *ring, u8 status,
        bool vlan_extracted = (adapter->vlgrp && (status & E1000_RXD_STAT_VP));
 
        skb_record_rx_queue(skb, ring->queue_index);
-       if (skb->ip_summed == CHECKSUM_UNNECESSARY) {
-               if (vlan_extracted)
-                       vlan_gro_receive(&ring->napi, adapter->vlgrp,
-                                        le16_to_cpu(rx_desc->wb.upper.vlan),
-                                        skb);
-               else
-                       napi_gro_receive(&ring->napi, skb);
-       } else {
-               if (vlan_extracted)
-                       vlan_hwaccel_receive_skb(skb, adapter->vlgrp,
-                                         le16_to_cpu(rx_desc->wb.upper.vlan));
-               else
-                       netif_receive_skb(skb);
-       }
+       if (vlan_extracted)
+               vlan_gro_receive(&ring->napi, adapter->vlgrp,
+                                le16_to_cpu(rx_desc->wb.upper.vlan),
+                                skb);
+       else
+               napi_gro_receive(&ring->napi, skb);
 }
 
 static inline void igb_rx_checksum_adv(struct igb_adapter *adapter,
@@ -4456,19 +4452,28 @@ static inline void igb_rx_checksum_adv(struct igb_adapter *adapter,
        skb->ip_summed = CHECKSUM_NONE;
 
        /* Ignore Checksum bit is set or checksum is disabled through ethtool */
-       if ((status_err & E1000_RXD_STAT_IXSM) || !adapter->rx_csum)
+       if ((status_err & E1000_RXD_STAT_IXSM) ||
+           (adapter->flags & IGB_FLAG_RX_CSUM_DISABLED))
                return;
        /* TCP/UDP checksum error bit is set */
        if (status_err &
            (E1000_RXDEXT_STATERR_TCPE | E1000_RXDEXT_STATERR_IPE)) {
+               /*
+                * work around errata with sctp packets where the TCPE aka
+                * L4E bit is set incorrectly on 64 byte (60 byte w/o crc)
+                * packets, (aka let the stack check the crc32c)
+                */
+               if (!((adapter->hw.mac.type == e1000_82576) &&
+                     (skb->len == 60)))
+                       adapter->hw_csum_err++;
                /* let the stack verify checksum errors */
-               adapter->hw_csum_err++;
                return;
        }
        /* It must be a TCP or UDP packet with a valid checksum */
        if (status_err & (E1000_RXD_STAT_TCPCS | E1000_RXD_STAT_UDPCS))
                skb->ip_summed = CHECKSUM_UNNECESSARY;
 
+       dev_dbg(&adapter->pdev->dev, "cksum success: bits %08X\n", status_err);
        adapter->hw_csum_good++;
 }