]> pilppa.org Git - linux-2.6-omap-h63xx.git/blobdiff - drivers/net/s2io.c
Merge git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq
[linux-2.6-omap-h63xx.git] / drivers / net / s2io.c
index fa57c49c0c51fad86de064e98e96a36249efb829..157fd932e95140ce54936e3d0a79bdb30cd25048 100644 (file)
@@ -50,6 +50,8 @@
  *                 Possible values '1' for enable , '0' for disable.
  *                 Default is '2' - which means disable in promisc mode
  *                 and enable in non-promiscuous mode.
+ * multiq: This parameter used to enable/disable MULTIQUEUE support.
+ *      Possible values '1' for enable and '0' for disable. Default is '0'
  ************************************************************************/
 
 #include <linux/module.h>
@@ -84,7 +86,7 @@
 #include "s2io.h"
 #include "s2io-regs.h"
 
-#define DRV_VERSION "2.0.26.10"
+#define DRV_VERSION "2.0.26.22"
 
 /* S2io Driver name & version. */
 static char s2io_driver_name[] = "Neterion";
@@ -115,20 +117,6 @@ static inline int RXD_IS_UP2DT(struct RxD_t *rxdp)
 
 #define LINK_IS_UP(val64) (!(val64 & (ADAPTER_STATUS_RMAC_REMOTE_FAULT | \
                                      ADAPTER_STATUS_RMAC_LOCAL_FAULT)))
-#define TASKLET_IN_USE test_and_set_bit(0, (&sp->tasklet_status))
-#define PANIC  1
-#define LOW    2
-static inline int rx_buffer_level(struct s2io_nic * sp, int rxb_size, int ring)
-{
-       struct mac_info *mac_control;
-
-       mac_control = &sp->mac_control;
-       if (rxb_size <= rxd_count[sp->rxd_mode])
-               return PANIC;
-       else if ((mac_control->rings[ring].pkt_cnt - rxb_size) > 16)
-               return  LOW;
-       return 0;
-}
 
 static inline int is_s2io_card_up(const struct s2io_nic * sp)
 {
@@ -335,10 +323,9 @@ static char ethtool_driver_stats_keys[][ETH_GSTRING_LEN] = {
        {"mc_err_cnt"}
 };
 
-#define S2IO_XENA_STAT_LEN sizeof(ethtool_xena_stats_keys)/ ETH_GSTRING_LEN
-#define S2IO_ENHANCED_STAT_LEN sizeof(ethtool_enhanced_stats_keys)/ \
-                                       ETH_GSTRING_LEN
-#define S2IO_DRIVER_STAT_LEN sizeof(ethtool_driver_stats_keys)/ ETH_GSTRING_LEN
+#define S2IO_XENA_STAT_LEN     ARRAY_SIZE(ethtool_xena_stats_keys)
+#define S2IO_ENHANCED_STAT_LEN ARRAY_SIZE(ethtool_enhanced_stats_keys)
+#define S2IO_DRIVER_STAT_LEN   ARRAY_SIZE(ethtool_driver_stats_keys)
 
 #define XFRAME_I_STAT_LEN (S2IO_XENA_STAT_LEN + S2IO_DRIVER_STAT_LEN )
 #define XFRAME_II_STAT_LEN (XFRAME_I_STAT_LEN + S2IO_ENHANCED_STAT_LEN )
@@ -346,7 +333,7 @@ static char ethtool_driver_stats_keys[][ETH_GSTRING_LEN] = {
 #define XFRAME_I_STAT_STRINGS_LEN ( XFRAME_I_STAT_LEN * ETH_GSTRING_LEN )
 #define XFRAME_II_STAT_STRINGS_LEN ( XFRAME_II_STAT_LEN * ETH_GSTRING_LEN )
 
-#define S2IO_TEST_LEN  sizeof(s2io_gstrings) / ETH_GSTRING_LEN
+#define S2IO_TEST_LEN  ARRAY_SIZE(s2io_gstrings)
 #define S2IO_STRINGS_LEN       S2IO_TEST_LEN * ETH_GSTRING_LEN
 
 #define S2IO_TIMER_CONF(timer, handle, arg, exp)               \
@@ -369,17 +356,44 @@ static void do_s2io_copy_mac_addr(struct s2io_nic *sp, int offset, u64 mac_addr)
 static void s2io_vlan_rx_register(struct net_device *dev,
                                        struct vlan_group *grp)
 {
+       int i;
        struct s2io_nic *nic = dev->priv;
-       unsigned long flags;
+       unsigned long flags[MAX_TX_FIFOS];
+       struct mac_info *mac_control = &nic->mac_control;
+       struct config_param *config = &nic->config;
+
+       for (i = 0; i < config->tx_fifo_num; i++)
+               spin_lock_irqsave(&mac_control->fifos[i].tx_lock, flags[i]);
 
-       spin_lock_irqsave(&nic->tx_lock, flags);
        nic->vlgrp = grp;
-       spin_unlock_irqrestore(&nic->tx_lock, flags);
+       for (i = config->tx_fifo_num - 1; i >= 0; i--)
+               spin_unlock_irqrestore(&mac_control->fifos[i].tx_lock,
+                               flags[i]);
 }
 
 /* A flag indicating whether 'RX_PA_CFG_STRIP_VLAN_TAG' bit is set or not */
 static int vlan_strip_flag;
 
+/* Unregister the vlan */
+static void s2io_vlan_rx_kill_vid(struct net_device *dev, unsigned long vid)
+{
+       int i;
+       struct s2io_nic *nic = dev->priv;
+       unsigned long flags[MAX_TX_FIFOS];
+       struct mac_info *mac_control = &nic->mac_control;
+       struct config_param *config = &nic->config;
+
+       for (i = 0; i < config->tx_fifo_num; i++)
+               spin_lock_irqsave(&mac_control->fifos[i].tx_lock, flags[i]);
+
+       if (nic->vlgrp)
+               vlan_group_set_device(nic->vlgrp, vid, NULL);
+
+       for (i = config->tx_fifo_num - 1; i >= 0; i--)
+               spin_unlock_irqrestore(&mac_control->fifos[i].tx_lock,
+                       flags[i]);
+}
+
 /*
  * Constants to be programmed into the Xena's registers, to configure
  * the XAUI.
@@ -450,10 +464,9 @@ MODULE_VERSION(DRV_VERSION);
 
 
 /* Module Loadable parameters. */
-S2IO_PARM_INT(tx_fifo_num, 1);
+S2IO_PARM_INT(tx_fifo_num, FIFO_DEFAULT_NUM);
 S2IO_PARM_INT(rx_ring_num, 1);
-
-
+S2IO_PARM_INT(multiq, 0);
 S2IO_PARM_INT(rx_ring_mode, 1);
 S2IO_PARM_INT(use_continuous_tx_intrs, 1);
 S2IO_PARM_INT(rmac_pause_time, 0x100);
@@ -463,6 +476,8 @@ S2IO_PARM_INT(shared_splits, 0);
 S2IO_PARM_INT(tmac_util_period, 5);
 S2IO_PARM_INT(rmac_util_period, 5);
 S2IO_PARM_INT(l3l4hdr_size, 128);
+/* 0 is no steering, 1 is Priority steering, 2 is Default steering */
+S2IO_PARM_INT(tx_steering_type, TX_DEFAULT_STEERING);
 /* Frequency of Rx desc syncs expressed as power of 2 */
 S2IO_PARM_INT(rxsync_frequency, 3);
 /* Interrupt type. Values can be 0(INTA), 2(MSI_X) */
@@ -527,6 +542,101 @@ static struct pci_driver s2io_driver = {
 /* A simplifier macro used both by init and free shared_mem Fns(). */
 #define TXD_MEM_PAGE_CNT(len, per_each) ((len+per_each - 1) / per_each)
 
+/* netqueue manipulation helper functions */
+static inline void s2io_stop_all_tx_queue(struct s2io_nic *sp)
+{
+       int i;
+#ifdef CONFIG_NETDEVICES_MULTIQUEUE
+       if (sp->config.multiq) {
+               for (i = 0; i < sp->config.tx_fifo_num; i++)
+                       netif_stop_subqueue(sp->dev, i);
+       } else
+#endif
+       {
+               for (i = 0; i < sp->config.tx_fifo_num; i++)
+                       sp->mac_control.fifos[i].queue_state = FIFO_QUEUE_STOP;
+               netif_stop_queue(sp->dev);
+       }
+}
+
+static inline void s2io_stop_tx_queue(struct s2io_nic *sp, int fifo_no)
+{
+#ifdef CONFIG_NETDEVICES_MULTIQUEUE
+       if (sp->config.multiq)
+               netif_stop_subqueue(sp->dev, fifo_no);
+       else
+#endif
+       {
+               sp->mac_control.fifos[fifo_no].queue_state =
+                       FIFO_QUEUE_STOP;
+               netif_stop_queue(sp->dev);
+       }
+}
+
+static inline void s2io_start_all_tx_queue(struct s2io_nic *sp)
+{
+       int i;
+#ifdef CONFIG_NETDEVICES_MULTIQUEUE
+       if (sp->config.multiq) {
+               for (i = 0; i < sp->config.tx_fifo_num; i++)
+                       netif_start_subqueue(sp->dev, i);
+       } else
+#endif
+       {
+               for (i = 0; i < sp->config.tx_fifo_num; i++)
+                       sp->mac_control.fifos[i].queue_state = FIFO_QUEUE_START;
+               netif_start_queue(sp->dev);
+       }
+}
+
+static inline void s2io_start_tx_queue(struct s2io_nic *sp, int fifo_no)
+{
+#ifdef CONFIG_NETDEVICES_MULTIQUEUE
+       if (sp->config.multiq)
+               netif_start_subqueue(sp->dev, fifo_no);
+       else
+#endif
+       {
+               sp->mac_control.fifos[fifo_no].queue_state =
+                       FIFO_QUEUE_START;
+               netif_start_queue(sp->dev);
+       }
+}
+
+static inline void s2io_wake_all_tx_queue(struct s2io_nic *sp)
+{
+       int i;
+#ifdef CONFIG_NETDEVICES_MULTIQUEUE
+       if (sp->config.multiq) {
+               for (i = 0; i < sp->config.tx_fifo_num; i++)
+                       netif_wake_subqueue(sp->dev, i);
+       } else
+#endif
+       {
+               for (i = 0; i < sp->config.tx_fifo_num; i++)
+                       sp->mac_control.fifos[i].queue_state = FIFO_QUEUE_START;
+               netif_wake_queue(sp->dev);
+       }
+}
+
+static inline void s2io_wake_tx_queue(
+       struct fifo_info *fifo, int cnt, u8 multiq)
+{
+
+#ifdef CONFIG_NETDEVICES_MULTIQUEUE
+       if (multiq) {
+               if (cnt && __netif_subqueue_stopped(fifo->dev, fifo->fifo_no))
+                       netif_wake_subqueue(fifo->dev, fifo->fifo_no);
+       } else
+#endif
+       if (cnt && (fifo->queue_state == FIFO_QUEUE_STOP)) {
+               if (netif_queue_stopped(fifo->dev)) {
+                       fifo->queue_state = FIFO_QUEUE_START;
+                       netif_wake_queue(fifo->dev);
+               }
+       }
+}
+
 /**
  * init_shared_mem - Allocation and Initialization of Memory
  * @nic: Device private variable.
@@ -566,6 +676,21 @@ static int init_shared_mem(struct s2io_nic *nic)
                return -EINVAL;
        }
 
+       size = 0;
+       for (i = 0; i < config->tx_fifo_num; i++) {
+               size = config->tx_cfg[i].fifo_len;
+               /*
+                * Legal values are from 2 to 8192
+                */
+               if (size < 2) {
+                       DBG_PRINT(ERR_DBG, "s2io: Invalid fifo len (%d)", size);
+                       DBG_PRINT(ERR_DBG, "for fifo %d\n", i);
+                       DBG_PRINT(ERR_DBG, "s2io: Legal values for fifo len"
+                               "are 2 to 8192\n");
+                       return -EINVAL;
+               }
+       }
+
        lst_size = (sizeof(struct TxD) * config->max_txds);
        lst_per_page = PAGE_SIZE / lst_size;
 
@@ -593,6 +718,7 @@ static int init_shared_mem(struct s2io_nic *nic)
                mac_control->fifos[i].fifo_no = i;
                mac_control->fifos[i].nic = nic;
                mac_control->fifos[i].max_txds = MAX_SKB_FRAGS + 2;
+               mac_control->fifos[i].dev = dev;
 
                for (j = 0; j < page_num; j++) {
                        int k = 0;
@@ -640,10 +766,14 @@ static int init_shared_mem(struct s2io_nic *nic)
                }
        }
 
-       nic->ufo_in_band_v = kcalloc(size, sizeof(u64), GFP_KERNEL);
-       if (!nic->ufo_in_band_v)
-               return -ENOMEM;
-        mem_allocated += (size * sizeof(u64));
+       for (i = 0; i < config->tx_fifo_num; i++) {
+               size = config->tx_cfg[i].fifo_len;
+               mac_control->fifos[i].ufo_in_band_v
+                       = kcalloc(size, sizeof(u64), GFP_KERNEL);
+               if (!mac_control->fifos[i].ufo_in_band_v)
+                       return -ENOMEM;
+               mem_allocated += (size * sizeof(u64));
+       }
 
        /* Allocation and initialization of RXDs in Rings */
        size = 0;
@@ -830,7 +960,6 @@ static int init_shared_mem(struct s2io_nic *nic)
 static void free_shared_mem(struct s2io_nic *nic)
 {
        int i, j, blk_cnt, size;
-       u32 ufo_size = 0;
        void *tmp_v_addr;
        dma_addr_t tmp_p_addr;
        struct mac_info *mac_control;
@@ -851,7 +980,6 @@ static void free_shared_mem(struct s2io_nic *nic)
        lst_per_page = PAGE_SIZE / lst_size;
 
        for (i = 0; i < config->tx_fifo_num; i++) {
-               ufo_size += config->tx_cfg[i].fifo_len;
                page_num = TXD_MEM_PAGE_CNT(config->tx_cfg[i].fifo_len,
                                                        lst_per_page);
                for (j = 0; j < page_num; j++) {
@@ -941,18 +1069,21 @@ static void free_shared_mem(struct s2io_nic *nic)
                }
        }
 
+       for (i = 0; i < nic->config.tx_fifo_num; i++) {
+               if (mac_control->fifos[i].ufo_in_band_v) {
+                       nic->mac_control.stats_info->sw_stat.mem_freed
+                               += (config->tx_cfg[i].fifo_len * sizeof(u64));
+                       kfree(mac_control->fifos[i].ufo_in_band_v);
+               }
+       }
+
        if (mac_control->stats_mem) {
+               nic->mac_control.stats_info->sw_stat.mem_freed +=
+                       mac_control->stats_mem_sz;
                pci_free_consistent(nic->pdev,
                                    mac_control->stats_mem_sz,
                                    mac_control->stats_mem,
                                    mac_control->stats_mem_phy);
-               nic->mac_control.stats_info->sw_stat.mem_freed +=
-                       mac_control->stats_mem_sz;
-       }
-       if (nic->ufo_in_band_v) {
-               kfree(nic->ufo_in_band_v);
-               nic->mac_control.stats_info->sw_stat.mem_freed
-                       += (ufo_size * sizeof(u64));
        }
 }
 
@@ -1052,9 +1183,68 @@ static int s2io_print_pci_mode(struct s2io_nic *nic)
        return mode;
 }
 
+/**
+ *  init_tti - Initialization transmit traffic interrupt scheme
+ *  @nic: device private variable
+ *  @link: link status (UP/DOWN) used to enable/disable continuous
+ *  transmit interrupts
+ *  Description: The function configures transmit traffic interrupts
+ *  Return Value:  SUCCESS on success and
+ *  '-1' on failure
+ */
+
+static int init_tti(struct s2io_nic *nic, int link)
+{
+       struct XENA_dev_config __iomem *bar0 = nic->bar0;
+       register u64 val64 = 0;
+       int i;
+       struct config_param *config;
+
+       config = &nic->config;
+
+       for (i = 0; i < config->tx_fifo_num; i++) {
+               /*
+                * TTI Initialization. Default Tx timer gets us about
+                * 250 interrupts per sec. Continuous interrupts are enabled
+                * by default.
+                */
+               if (nic->device_type == XFRAME_II_DEVICE) {
+                       int count = (nic->config.bus_speed * 125)/2;
+                       val64 = TTI_DATA1_MEM_TX_TIMER_VAL(count);
+               } else
+                       val64 = TTI_DATA1_MEM_TX_TIMER_VAL(0x2078);
+
+               val64 |= TTI_DATA1_MEM_TX_URNG_A(0xA) |
+                               TTI_DATA1_MEM_TX_URNG_B(0x10) |
+                               TTI_DATA1_MEM_TX_URNG_C(0x30) |
+                               TTI_DATA1_MEM_TX_TIMER_AC_EN;
+
+               if (use_continuous_tx_intrs && (link == LINK_UP))
+                       val64 |= TTI_DATA1_MEM_TX_TIMER_CI_EN;
+               writeq(val64, &bar0->tti_data1_mem);
+
+               val64 = TTI_DATA2_MEM_TX_UFC_A(0x10) |
+                               TTI_DATA2_MEM_TX_UFC_B(0x20) |
+                               TTI_DATA2_MEM_TX_UFC_C(0x40) |
+                               TTI_DATA2_MEM_TX_UFC_D(0x80);
+
+               writeq(val64, &bar0->tti_data2_mem);
+
+               val64 = TTI_CMD_MEM_WE | TTI_CMD_MEM_STROBE_NEW_CMD |
+                               TTI_CMD_MEM_OFFSET(i);
+               writeq(val64, &bar0->tti_command_mem);
+
+               if (wait_for_cmd_complete(&bar0->tti_command_mem,
+                       TTI_CMD_MEM_STROBE_NEW_CMD, S2IO_BIT_RESET) != SUCCESS)
+                       return FAILURE;
+       }
+
+       return SUCCESS;
+}
+
 /**
  *  init_nic - Initialization of hardware
- *  @nic: device peivate variable
+ *  @nic: device private variable
  *  Description: The function sequentially configures every block
  *  of the H/W from their reset values.
  *  Return Value:  SUCCESS on success and
@@ -1159,9 +1349,9 @@ static int init_nic(struct s2io_nic *nic)
 
        for (i = 0, j = 0; i < config->tx_fifo_num; i++) {
                val64 |=
-                   vBIT(config->tx_cfg[i].fifo_len - 1, ((i * 32) + 19),
+                   vBIT(config->tx_cfg[i].fifo_len - 1, ((j * 32) + 19),
                         13) | vBIT(config->tx_cfg[i].fifo_priority,
-                                   ((i * 32) + 5), 3);
+                                   ((j * 32) + 5), 3);
 
                if (i == (config->tx_fifo_num - 1)) {
                        if (i % 2 == 0)
@@ -1172,17 +1362,25 @@ static int init_nic(struct s2io_nic *nic)
                case 1:
                        writeq(val64, &bar0->tx_fifo_partition_0);
                        val64 = 0;
+                       j = 0;
                        break;
                case 3:
                        writeq(val64, &bar0->tx_fifo_partition_1);
                        val64 = 0;
+                       j = 0;
                        break;
                case 5:
                        writeq(val64, &bar0->tx_fifo_partition_2);
                        val64 = 0;
+                       j = 0;
                        break;
                case 7:
                        writeq(val64, &bar0->tx_fifo_partition_3);
+                       val64 = 0;
+                       j = 0;
+                       break;
+               default:
+                       j++;
                        break;
                }
        }
@@ -1268,11 +1466,11 @@ static int init_nic(struct s2io_nic *nic)
 
        /*
         * Filling Tx round robin registers
-        * as per the number of FIFOs
+        * as per the number of FIFOs for equal scheduling priority
         */
        switch (config->tx_fifo_num) {
        case 1:
-               val64 = 0x0000000000000000ULL;
+               val64 = 0x0;
                writeq(val64, &bar0->tx_w_round_robin_0);
                writeq(val64, &bar0->tx_w_round_robin_1);
                writeq(val64, &bar0->tx_w_round_robin_2);
@@ -1280,87 +1478,78 @@ static int init_nic(struct s2io_nic *nic)
                writeq(val64, &bar0->tx_w_round_robin_4);
                break;
        case 2:
-               val64 = 0x0000010000010000ULL;
+               val64 = 0x0001000100010001ULL;
                writeq(val64, &bar0->tx_w_round_robin_0);
-               val64 = 0x0100000100000100ULL;
                writeq(val64, &bar0->tx_w_round_robin_1);
-               val64 = 0x0001000001000001ULL;
                writeq(val64, &bar0->tx_w_round_robin_2);
-               val64 = 0x0000010000010000ULL;
                writeq(val64, &bar0->tx_w_round_robin_3);
-               val64 = 0x0100000000000000ULL;
+               val64 = 0x0001000100000000ULL;
                writeq(val64, &bar0->tx_w_round_robin_4);
                break;
        case 3:
-               val64 = 0x0001000102000001ULL;
+               val64 = 0x0001020001020001ULL;
                writeq(val64, &bar0->tx_w_round_robin_0);
-               val64 = 0x0001020000010001ULL;
+               val64 = 0x0200010200010200ULL;
                writeq(val64, &bar0->tx_w_round_robin_1);
-               val64 = 0x0200000100010200ULL;
+               val64 = 0x0102000102000102ULL;
                writeq(val64, &bar0->tx_w_round_robin_2);
-               val64 = 0x0001000102000001ULL;
+               val64 = 0x0001020001020001ULL;
                writeq(val64, &bar0->tx_w_round_robin_3);
-               val64 = 0x0001020000000000ULL;
+               val64 = 0x0200010200000000ULL;
                writeq(val64, &bar0->tx_w_round_robin_4);
                break;
        case 4:
-               val64 = 0x0001020300010200ULL;
+               val64 = 0x0001020300010203ULL;
                writeq(val64, &bar0->tx_w_round_robin_0);
-               val64 = 0x0100000102030001ULL;
                writeq(val64, &bar0->tx_w_round_robin_1);
-               val64 = 0x0200010000010203ULL;
                writeq(val64, &bar0->tx_w_round_robin_2);
-               val64 = 0x0001020001000001ULL;
                writeq(val64, &bar0->tx_w_round_robin_3);
-               val64 = 0x0203000100000000ULL;
+               val64 = 0x0001020300000000ULL;
                writeq(val64, &bar0->tx_w_round_robin_4);
                break;
        case 5:
-               val64 = 0x0001000203000102ULL;
+               val64 = 0x0001020304000102ULL;
                writeq(val64, &bar0->tx_w_round_robin_0);
-               val64 = 0x0001020001030004ULL;
+               val64 = 0x0304000102030400ULL;
                writeq(val64, &bar0->tx_w_round_robin_1);
-               val64 = 0x0001000203000102ULL;
+               val64 = 0x0102030400010203ULL;
                writeq(val64, &bar0->tx_w_round_robin_2);
-               val64 = 0x0001020001030004ULL;
+               val64 = 0x0400010203040001ULL;
                writeq(val64, &bar0->tx_w_round_robin_3);
-               val64 = 0x0001000000000000ULL;
+               val64 = 0x0203040000000000ULL;
                writeq(val64, &bar0->tx_w_round_robin_4);
                break;
        case 6:
-               val64 = 0x0001020304000102ULL;
+               val64 = 0x0001020304050001ULL;
                writeq(val64, &bar0->tx_w_round_robin_0);
-               val64 = 0x0304050001020001ULL;
+               val64 = 0x0203040500010203ULL;
                writeq(val64, &bar0->tx_w_round_robin_1);
-               val64 = 0x0203000100000102ULL;
+               val64 = 0x0405000102030405ULL;
                writeq(val64, &bar0->tx_w_round_robin_2);
-               val64 = 0x0304000102030405ULL;
+               val64 = 0x0001020304050001ULL;
                writeq(val64, &bar0->tx_w_round_robin_3);
-               val64 = 0x0001000200000000ULL;
+               val64 = 0x0203040500000000ULL;
                writeq(val64, &bar0->tx_w_round_robin_4);
                break;
        case 7:
-               val64 = 0x0001020001020300ULL;
+               val64 = 0x0001020304050600ULL;
                writeq(val64, &bar0->tx_w_round_robin_0);
-               val64 = 0x0102030400010203ULL;
+               val64 = 0x0102030405060001ULL;
                writeq(val64, &bar0->tx_w_round_robin_1);
-               val64 = 0x0405060001020001ULL;
+               val64 = 0x0203040506000102ULL;
                writeq(val64, &bar0->tx_w_round_robin_2);
-               val64 = 0x0304050000010200ULL;
+               val64 = 0x0304050600010203ULL;
                writeq(val64, &bar0->tx_w_round_robin_3);
-               val64 = 0x0102030000000000ULL;
+               val64 = 0x0405060000000000ULL;
                writeq(val64, &bar0->tx_w_round_robin_4);
                break;
        case 8:
-               val64 = 0x0001020300040105ULL;
+               val64 = 0x0001020304050607ULL;
                writeq(val64, &bar0->tx_w_round_robin_0);
-               val64 = 0x0200030106000204ULL;
                writeq(val64, &bar0->tx_w_round_robin_1);
-               val64 = 0x0103000502010007ULL;
                writeq(val64, &bar0->tx_w_round_robin_2);
-               val64 = 0x0304010002060500ULL;
                writeq(val64, &bar0->tx_w_round_robin_3);
-               val64 = 0x0103020400000000ULL;
+               val64 = 0x0001020300000000ULL;
                writeq(val64, &bar0->tx_w_round_robin_4);
                break;
        }
@@ -1537,58 +1726,14 @@ static int init_nic(struct s2io_nic *nic)
            MAC_RX_LINK_UTIL_VAL(rmac_util_period);
        writeq(val64, &bar0->mac_link_util);
 
-
        /*
         * Initializing the Transmit and Receive Traffic Interrupt
         * Scheme.
         */
-       /*
-        * TTI Initialization. Default Tx timer gets us about
-        * 250 interrupts per sec. Continuous interrupts are enabled
-        * by default.
-        */
-       if (nic->device_type == XFRAME_II_DEVICE) {
-               int count = (nic->config.bus_speed * 125)/2;
-               val64 = TTI_DATA1_MEM_TX_TIMER_VAL(count);
-       } else {
-
-               val64 = TTI_DATA1_MEM_TX_TIMER_VAL(0x2078);
-       }
-       val64 |= TTI_DATA1_MEM_TX_URNG_A(0xA) |
-           TTI_DATA1_MEM_TX_URNG_B(0x10) |
-           TTI_DATA1_MEM_TX_URNG_C(0x30) | TTI_DATA1_MEM_TX_TIMER_AC_EN;
-               if (use_continuous_tx_intrs)
-                       val64 |= TTI_DATA1_MEM_TX_TIMER_CI_EN;
-       writeq(val64, &bar0->tti_data1_mem);
-
-       val64 = TTI_DATA2_MEM_TX_UFC_A(0x10) |
-           TTI_DATA2_MEM_TX_UFC_B(0x20) |
-           TTI_DATA2_MEM_TX_UFC_C(0x40) | TTI_DATA2_MEM_TX_UFC_D(0x80);
-       writeq(val64, &bar0->tti_data2_mem);
-
-       val64 = TTI_CMD_MEM_WE | TTI_CMD_MEM_STROBE_NEW_CMD;
-       writeq(val64, &bar0->tti_command_mem);
 
-       /*
-        * Once the operation completes, the Strobe bit of the command
-        * register will be reset. We poll for this particular condition
-        * We wait for a maximum of 500ms for the operation to complete,
-        * if it's not complete by then we return error.
-        */
-       time = 0;
-       while (TRUE) {
-               val64 = readq(&bar0->tti_command_mem);
-               if (!(val64 & TTI_CMD_MEM_STROBE_NEW_CMD)) {
-                       break;
-               }
-               if (time > 10) {
-                       DBG_PRINT(ERR_DBG, "%s: TTI init Failed\n",
-                                 dev->name);
-                       return -ENODEV;
-               }
-               msleep(50);
-               time++;
-       }
+       /* Initialize TTI */
+       if (SUCCESS != init_tti(nic, nic->last_link_state))
+               return -ENODEV;
 
        /* RTI Initialization */
        if (nic->device_type == XFRAME_II_DEVICE) {
@@ -2242,7 +2387,7 @@ static struct sk_buff *s2io_txdl_getskb(struct fifo_info *fifo_data, struct \
        u16 j, frg_cnt;
 
        txds = txdlp;
-       if (txds->Host_Control == (u64)(long)nic->ufo_in_band_v) {
+       if (txds->Host_Control == (u64)(long)fifo_data->ufo_in_band_v) {
                pci_unmap_single(nic->pdev, (dma_addr_t)
                        txds->Buffer_Pointer, sizeof(u64),
                        PCI_DMA_TODEVICE);
@@ -2297,7 +2442,9 @@ static void free_tx_buffers(struct s2io_nic *nic)
        config = &nic->config;
 
        for (i = 0; i < config->tx_fifo_num; i++) {
-               for (j = 0; j < config->tx_cfg[i].fifo_len - 1; j++) {
+               unsigned long flags;
+               spin_lock_irqsave(&mac_control->fifos[i].tx_lock, flags);
+               for (j = 0; j < config->tx_cfg[i].fifo_len; j++) {
                        txdp = (struct TxD *) \
                        mac_control->fifos[i].list_info[j].list_virt_addr;
                        skb = s2io_txdl_getskb(&mac_control->fifos[i], txdp, j);
@@ -2313,6 +2460,7 @@ static void free_tx_buffers(struct s2io_nic *nic)
                          dev->name, cnt, i);
                mac_control->fifos[i].tx_curr_get_info.offset = 0;
                mac_control->fifos[i].tx_curr_put_info.offset = 0;
+               spin_unlock_irqrestore(&mac_control->fifos[i].tx_lock, flags);
        }
 }
 
@@ -2382,7 +2530,6 @@ static int fill_rx_buffers(struct s2io_nic *nic, int ring_no)
        struct config_param *config;
        u64 tmp;
        struct buffAdd *ba;
-       unsigned long flags;
        struct RxD_t *first_rxdp = NULL;
        u64 Buffer0_ptr = 0, Buffer1_ptr = 0;
        struct RxD1 *rxdp1;
@@ -2430,15 +2577,7 @@ static int fill_rx_buffers(struct s2io_nic *nic, int ring_no)
                        DBG_PRINT(INTR_DBG, "%s: Next block at: %p\n",
                                  dev->name, rxdp);
                }
-               if(!napi) {
-                       spin_lock_irqsave(&nic->put_lock, flags);
-                       mac_control->rings[ring_no].put_pos =
-                       (block_no * (rxd_count[nic->rxd_mode] + 1)) + off;
-                       spin_unlock_irqrestore(&nic->put_lock, flags);
-               } else {
-                       mac_control->rings[ring_no].put_pos =
-                       (block_no * (rxd_count[nic->rxd_mode] + 1)) + off;
-               }
+
                if ((rxdp->Control_1 & RXD_OWN_XENA) &&
                        ((nic->rxd_mode == RXD_MODE_3B) &&
                                (rxdp->Control_2 & s2BIT(0)))) {
@@ -2816,7 +2955,7 @@ static void rx_intr_handler(struct ring_info *ring_data)
 {
        struct s2io_nic *nic = ring_data->nic;
        struct net_device *dev = (struct net_device *) nic->dev;
-       int get_block, put_block, put_offset;
+       int get_block, put_block;
        struct rx_curr_get_info get_info, put_info;
        struct RxD_t *rxdp;
        struct sk_buff *skb;
@@ -2825,19 +2964,11 @@ static void rx_intr_handler(struct ring_info *ring_data)
        struct RxD1* rxdp1;
        struct RxD3* rxdp3;
 
-       spin_lock(&nic->rx_lock);
-
        get_info = ring_data->rx_curr_get_info;
        get_block = get_info.block_index;
        memcpy(&put_info, &ring_data->rx_curr_put_info, sizeof(put_info));
        put_block = put_info.block_index;
        rxdp = ring_data->rx_blocks[get_block].rxds[get_info.offset].virt_addr;
-       if (!napi) {
-               spin_lock(&nic->put_lock);
-               put_offset = ring_data->put_pos;
-               spin_unlock(&nic->put_lock);
-       } else
-               put_offset = ring_data->put_pos;
 
        while (RXD_IS_UP2DT(rxdp)) {
                /*
@@ -2854,7 +2985,6 @@ static void rx_intr_handler(struct ring_info *ring_data)
                        DBG_PRINT(ERR_DBG, "%s: The skb is ",
                                  dev->name);
                        DBG_PRINT(ERR_DBG, "Null in Rx Intr\n");
-                       spin_unlock(&nic->rx_lock);
                        return;
                }
                if (nic->rxd_mode == RXD_MODE_1) {
@@ -2905,13 +3035,11 @@ static void rx_intr_handler(struct ring_info *ring_data)
                        struct lro *lro = &nic->lro0_n[i];
                        if (lro->in_use) {
                                update_L3L4_header(nic, lro);
-                               queue_rx_frame(lro->parent);
+                               queue_rx_frame(lro->parent, lro->vlan_tag);
                                clear_lro_session(lro);
                        }
                }
        }
-
-       spin_unlock(&nic->rx_lock);
 }
 
 /**
@@ -2929,12 +3057,16 @@ static void rx_intr_handler(struct ring_info *ring_data)
 static void tx_intr_handler(struct fifo_info *fifo_data)
 {
        struct s2io_nic *nic = fifo_data->nic;
-       struct net_device *dev = (struct net_device *) nic->dev;
        struct tx_curr_get_info get_info, put_info;
-       struct sk_buff *skb;
+       struct sk_buff *skb = NULL;
        struct TxD *txdlp;
+       int pkt_cnt = 0;
+       unsigned long flags = 0;
        u8 err_mask;
 
+       if (!spin_trylock_irqsave(&fifo_data->tx_lock, flags))
+                       return;
+
        get_info = fifo_data->tx_curr_get_info;
        memcpy(&put_info, &fifo_data->tx_curr_put_info, sizeof(put_info));
        txdlp = (struct TxD *) fifo_data->list_info[get_info.offset].
@@ -2983,11 +3115,13 @@ static void tx_intr_handler(struct fifo_info *fifo_data)
 
                skb = s2io_txdl_getskb(fifo_data, txdlp, get_info.offset);
                if (skb == NULL) {
+                       spin_unlock_irqrestore(&fifo_data->tx_lock, flags);
                        DBG_PRINT(ERR_DBG, "%s: Null skb ",
                        __FUNCTION__);
                        DBG_PRINT(ERR_DBG, "in Tx Free Intr\n");
                        return;
                }
+               pkt_cnt++;
 
                /* Updating the statistics block */
                nic->stats.tx_bytes += skb->len;
@@ -3003,10 +3137,9 @@ static void tx_intr_handler(struct fifo_info *fifo_data)
                    get_info.offset;
        }
 
-       spin_lock(&nic->tx_lock);
-       if (netif_queue_stopped(dev))
-               netif_wake_queue(dev);
-       spin_unlock(&nic->tx_lock);
+       s2io_wake_tx_queue(fifo_data, pkt_cnt, nic->config.multiq);
+
+       spin_unlock_irqrestore(&fifo_data->tx_lock, flags);
 }
 
 /**
@@ -3376,6 +3509,9 @@ static void s2io_reset(struct s2io_nic * sp)
        /* Set swapper to enable I/O register access */
        s2io_set_swapper(sp);
 
+       /* restore mac_addr entries */
+       do_s2io_restore_unicast_mc(sp);
+
        /* Restore the MSIX table entries from local variables */
        restore_xmsi_data(sp);
 
@@ -3434,9 +3570,6 @@ static void s2io_reset(struct s2io_nic * sp)
                writeq(val64, &bar0->pcc_err_reg);
        }
 
-       /* restore the previously assigned mac address */
-       do_s2io_prog_unicast(sp->dev, (u8 *)&sp->def_mac_addr[0].mac_addr);
-
        sp->device_enabled_once = FALSE;
 }
 
@@ -3773,7 +3906,7 @@ static int s2io_test_msi(struct s2io_nic *sp)
 
        if (!sp->msi_detected) {
                /* MSI(X) test failed, go back to INTx mode */
-               DBG_PRINT(ERR_DBG, "%s: PCI %s: No interrupt was generated"
+               DBG_PRINT(ERR_DBG, "%s: PCI %s: No interrupt was generated "
                        "using MSI(X) during test\n", sp->dev->name,
                        pci_name(pdev));
 
@@ -3848,8 +3981,6 @@ static int s2io_open(struct net_device *dev)
        netif_carrier_off(dev);
        sp->last_link_state = 0;
 
-       napi_enable(&sp->napi);
-
        if (sp->config.intr_type == MSI_X) {
                int ret = s2io_enable_msi_x(sp);
 
@@ -3887,12 +4018,10 @@ static int s2io_open(struct net_device *dev)
                err = -ENODEV;
                goto hw_init_failed;
        }
-
-       netif_start_queue(dev);
+       s2io_start_all_tx_queue(sp);
        return 0;
 
 hw_init_failed:
-       napi_disable(&sp->napi);
        if (sp->config.intr_type == MSI_X) {
                if (sp->entries) {
                        kfree(sp->entries);
@@ -3924,6 +4053,9 @@ hw_init_failed:
 static int s2io_close(struct net_device *dev)
 {
        struct s2io_nic *sp = dev->priv;
+       struct config_param *config = &sp->config;
+       u64 tmp64;
+       int offset;
 
        /* Return if the device is already closed               *
        *  Can happen when s2io_card_up failed in change_mtu    *
@@ -3931,9 +4063,14 @@ static int s2io_close(struct net_device *dev)
        if (!is_s2io_card_up(sp))
                return 0;
 
-       netif_stop_queue(dev);
-       napi_disable(&sp->napi);
-       /* Reset card, kill tasklet and free Tx and Rx buffers. */
+       s2io_stop_all_tx_queue(sp);
+       /* delete all populated mac entries */
+       for (offset = 1; offset < config->max_mc_addr; offset++) {
+               tmp64 = do_s2io_read_unicast_mc(sp, offset);
+               if (tmp64 != S2IO_DISABLE_MAC_ENTRY)
+                       do_s2io_delete_unicast_mc(sp, tmp64);
+       }
+
        s2io_card_down(sp);
 
        return 0;
@@ -3959,12 +4096,14 @@ static int s2io_xmit(struct sk_buff *skb, struct net_device *dev)
        register u64 val64;
        struct TxD *txdp;
        struct TxFIFO_element __iomem *tx_fifo;
-       unsigned long flags;
+       unsigned long flags = 0;
        u16 vlan_tag = 0;
-       int vlan_priority = 0;
+       struct fifo_info *fifo = NULL;
        struct mac_info *mac_control;
        struct config_param *config;
+       int do_spin_lock = 1;
        int offload_type;
+       int enable_per_list_interrupt = 0;
        struct swStat *stats = &sp->mac_control.stats_info->sw_stat;
 
        mac_control = &sp->mac_control;
@@ -3976,38 +4115,89 @@ static int s2io_xmit(struct sk_buff *skb, struct net_device *dev)
                DBG_PRINT(TX_DBG, "%s:Buffer has no data..\n", dev->name);
                dev_kfree_skb_any(skb);
                return 0;
-}
+       }
 
-       spin_lock_irqsave(&sp->tx_lock, flags);
        if (!is_s2io_card_up(sp)) {
                DBG_PRINT(TX_DBG, "%s: Card going down for reset\n",
                          dev->name);
-               spin_unlock_irqrestore(&sp->tx_lock, flags);
                dev_kfree_skb(skb);
                return 0;
        }
 
        queue = 0;
-       /* Get Fifo number to Transmit based on vlan priority */
-       if (sp->vlgrp && vlan_tx_tag_present(skb)) {
+       if (sp->vlgrp && vlan_tx_tag_present(skb))
                vlan_tag = vlan_tx_tag_get(skb);
-               vlan_priority = vlan_tag >> 13;
-               queue = config->fifo_mapping[vlan_priority];
+       if (sp->config.tx_steering_type == TX_DEFAULT_STEERING) {
+               if (skb->protocol == htons(ETH_P_IP)) {
+                       struct iphdr *ip;
+                       struct tcphdr *th;
+                       ip = ip_hdr(skb);
+
+                       if ((ip->frag_off & htons(IP_OFFSET|IP_MF)) == 0) {
+                               th = (struct tcphdr *)(((unsigned char *)ip) +
+                                               ip->ihl*4);
+
+                               if (ip->protocol == IPPROTO_TCP) {
+                                       queue_len = sp->total_tcp_fifos;
+                                       queue = (ntohs(th->source) +
+                                                       ntohs(th->dest)) &
+                                           sp->fifo_selector[queue_len - 1];
+                                       if (queue >= queue_len)
+                                               queue = queue_len - 1;
+                               } else if (ip->protocol == IPPROTO_UDP) {
+                                       queue_len = sp->total_udp_fifos;
+                                       queue = (ntohs(th->source) +
+                                                       ntohs(th->dest)) &
+                                           sp->fifo_selector[queue_len - 1];
+                                       if (queue >= queue_len)
+                                               queue = queue_len - 1;
+                                       queue += sp->udp_fifo_idx;
+                                       if (skb->len > 1024)
+                                               enable_per_list_interrupt = 1;
+                                       do_spin_lock = 0;
+                               }
+                       }
+               }
+       } else if (sp->config.tx_steering_type == TX_PRIORITY_STEERING)
+               /* get fifo number based on skb->priority value */
+               queue = config->fifo_mapping
+                                       [skb->priority & (MAX_TX_FIFOS - 1)];
+       fifo = &mac_control->fifos[queue];
+
+       if (do_spin_lock)
+               spin_lock_irqsave(&fifo->tx_lock, flags);
+       else {
+               if (unlikely(!spin_trylock_irqsave(&fifo->tx_lock, flags)))
+                       return NETDEV_TX_LOCKED;
        }
 
-       put_off = (u16) mac_control->fifos[queue].tx_curr_put_info.offset;
-       get_off = (u16) mac_control->fifos[queue].tx_curr_get_info.offset;
-       txdp = (struct TxD *) mac_control->fifos[queue].list_info[put_off].
-               list_virt_addr;
+#ifdef CONFIG_NETDEVICES_MULTIQUEUE
+       if (sp->config.multiq) {
+               if (__netif_subqueue_stopped(dev, fifo->fifo_no)) {
+                       spin_unlock_irqrestore(&fifo->tx_lock, flags);
+                       return NETDEV_TX_BUSY;
+               }
+       } else
+#endif
+       if (unlikely(fifo->queue_state == FIFO_QUEUE_STOP)) {
+               if (netif_queue_stopped(dev)) {
+                       spin_unlock_irqrestore(&fifo->tx_lock, flags);
+                       return NETDEV_TX_BUSY;
+               }
+       }
+
+       put_off = (u16) fifo->tx_curr_put_info.offset;
+       get_off = (u16) fifo->tx_curr_get_info.offset;
+       txdp = (struct TxD *) fifo->list_info[put_off].list_virt_addr;
 
-       queue_len = mac_control->fifos[queue].tx_curr_put_info.fifo_len + 1;
+       queue_len = fifo->tx_curr_put_info.fifo_len + 1;
        /* Avoid "put" pointer going beyond "get" pointer */
        if (txdp->Host_Control ||
                   ((put_off+1) == queue_len ? 0 : (put_off+1)) == get_off) {
                DBG_PRINT(TX_DBG, "Error in xmit, No free TXDs.\n");
-               netif_stop_queue(dev);
+               s2io_stop_tx_queue(sp, fifo->fifo_no);
                dev_kfree_skb(skb);
-               spin_unlock_irqrestore(&sp->tx_lock, flags);
+               spin_unlock_irqrestore(&fifo->tx_lock, flags);
                return 0;
        }
 
@@ -4023,9 +4213,11 @@ static int s2io_xmit(struct sk_buff *skb, struct net_device *dev)
        }
        txdp->Control_1 |= TXD_GATHER_CODE_FIRST;
        txdp->Control_1 |= TXD_LIST_OWN_XENA;
-       txdp->Control_2 |= config->tx_intr_type;
-
-       if (sp->vlgrp && vlan_tx_tag_present(skb)) {
+       txdp->Control_2 |= TXD_INT_NUMBER(fifo->fifo_no);
+       if (enable_per_list_interrupt)
+               if (put_off & (queue_len >> 5))
+                       txdp->Control_2 |= TXD_INT_TYPE_PER_LIST;
+       if (vlan_tag) {
                txdp->Control_2 |= TXD_VLAN_ENABLE;
                txdp->Control_2 |= TXD_VLAN_TAG(vlan_tag);
        }
@@ -4040,15 +4232,16 @@ static int s2io_xmit(struct sk_buff *skb, struct net_device *dev)
                txdp->Control_1 |= TXD_UFO_MSS(ufo_size);
                txdp->Control_1 |= TXD_BUFFER0_SIZE(8);
 #ifdef __BIG_ENDIAN
-               sp->ufo_in_band_v[put_off] =
-                               (u64)skb_shinfo(skb)->ip6_frag_id;
+               /* both variants do cpu_to_be64(be32_to_cpu(...)) */
+               fifo->ufo_in_band_v[put_off] =
+                               (__force u64)skb_shinfo(skb)->ip6_frag_id;
 #else
-               sp->ufo_in_band_v[put_off] =
-                               (u64)skb_shinfo(skb)->ip6_frag_id << 32;
+               fifo->ufo_in_band_v[put_off] =
+                               (__force u64)skb_shinfo(skb)->ip6_frag_id << 32;
 #endif
-               txdp->Host_Control = (unsigned long)sp->ufo_in_band_v;
+               txdp->Host_Control = (unsigned long)fifo->ufo_in_band_v;
                txdp->Buffer_Pointer = pci_map_single(sp->pdev,
-                                       sp->ufo_in_band_v,
+                                       fifo->ufo_in_band_v,
                                        sizeof(u64), PCI_DMA_TODEVICE);
                if((txdp->Buffer_Pointer == 0) ||
                        (txdp->Buffer_Pointer == DMA_ERROR_CODE))
@@ -4088,7 +4281,7 @@ static int s2io_xmit(struct sk_buff *skb, struct net_device *dev)
                frg_cnt++; /* as Txd0 was used for inband header */
 
        tx_fifo = mac_control->tx_FIFO_start[queue];
-       val64 = mac_control->fifos[queue].list_info[put_off].list_phy_addr;
+       val64 = fifo->list_info[put_off].list_phy_addr;
        writeq(val64, &tx_fifo->TxDL_Pointer);
 
        val64 = (TX_FIFO_LAST_TXD_NUM(frg_cnt) | TX_FIFO_FIRST_LIST |
@@ -4101,9 +4294,9 @@ static int s2io_xmit(struct sk_buff *skb, struct net_device *dev)
        mmiowb();
 
        put_off++;
-       if (put_off == mac_control->fifos[queue].tx_curr_put_info.fifo_len + 1)
+       if (put_off == fifo->tx_curr_put_info.fifo_len + 1)
                put_off = 0;
-       mac_control->fifos[queue].tx_curr_put_info.offset = put_off;
+       fifo->tx_curr_put_info.offset = put_off;
 
        /* Avoid "put" pointer going beyond "get" pointer */
        if (((put_off+1) == queue_len ? 0 : (put_off+1)) == get_off) {
@@ -4111,19 +4304,22 @@ static int s2io_xmit(struct sk_buff *skb, struct net_device *dev)
                DBG_PRINT(TX_DBG,
                          "No free TxDs for xmit, Put: 0x%x Get:0x%x\n",
                          put_off, get_off);
-               netif_stop_queue(dev);
+               s2io_stop_tx_queue(sp, fifo->fifo_no);
        }
        mac_control->stats_info->sw_stat.mem_allocated += skb->truesize;
        dev->trans_start = jiffies;
-       spin_unlock_irqrestore(&sp->tx_lock, flags);
+       spin_unlock_irqrestore(&fifo->tx_lock, flags);
+
+       if (sp->config.intr_type == MSI_X)
+               tx_intr_handler(fifo);
 
        return 0;
 pci_map_failed:
        stats->pci_map_fail_cnt++;
-       netif_stop_queue(dev);
+       s2io_stop_tx_queue(sp, fifo->fifo_no);
        stats->mem_freed += skb->truesize;
        dev_kfree_skb(skb);
-       spin_unlock_irqrestore(&sp->tx_lock, flags);
+       spin_unlock_irqrestore(&fifo->tx_lock, flags);
        return 0;
 }
 
@@ -4139,29 +4335,9 @@ s2io_alarm_handle(unsigned long data)
 
 static int s2io_chk_rx_buffers(struct s2io_nic *sp, int rng_n)
 {
-       int rxb_size, level;
-
-       if (!sp->lro) {
-               rxb_size = atomic_read(&sp->rx_bufs_left[rng_n]);
-               level = rx_buffer_level(sp, rxb_size, rng_n);
-
-               if ((level == PANIC) && (!TASKLET_IN_USE)) {
-                       int ret;
-                       DBG_PRINT(INTR_DBG, "%s: Rx BD hit ", __FUNCTION__);
-                       DBG_PRINT(INTR_DBG, "PANIC levels\n");
-                       if ((ret = fill_rx_buffers(sp, rng_n)) == -ENOMEM) {
-                               DBG_PRINT(INFO_DBG, "Out of memory in %s",
-                                         __FUNCTION__);
-                               clear_bit(0, (&sp->tasklet_status));
-                               return -1;
-                       }
-                       clear_bit(0, (&sp->tasklet_status));
-               } else if (level == LOW)
-                       tasklet_schedule(&sp->task);
-
-       } else if (fill_rx_buffers(sp, rng_n) == -ENOMEM) {
-                       DBG_PRINT(INFO_DBG, "%s:Out of memory", sp->dev->name);
-                       DBG_PRINT(INFO_DBG, " in Rx Intr!!\n");
+       if (fill_rx_buffers(sp, rng_n) == -ENOMEM) {
+               DBG_PRINT(INFO_DBG, "%s:Out of memory", sp->dev->name);
+               DBG_PRINT(INFO_DBG, " in Rx Intr!!\n");
        }
        return 0;
 }
@@ -4532,7 +4708,7 @@ static void s2io_handle_errors(void * dev_id)
        return;
 
 reset:
-       netif_stop_queue(dev);
+       s2io_stop_all_tx_queue(sp);
        schedule_work(&sp->rst_timer_task);
        sw_stat->soft_reset_cnt++;
        return;
@@ -4733,8 +4909,9 @@ static void s2io_set_multicast(struct net_device *dev)
        struct XENA_dev_config __iomem *bar0 = sp->bar0;
        u64 val64 = 0, multi_mac = 0x010203040506ULL, mask =
            0xfeffffffffffULL;
-       u64 dis_addr = 0xffffffffffffULL, mac_addr = 0;
+       u64 dis_addr = S2IO_DISABLE_MAC_ENTRY, mac_addr = 0;
        void __iomem *add;
+       struct config_param *config = &sp->config;
 
        if ((dev->flags & IFF_ALLMULTI) && (!sp->m_cast_flg)) {
                /*  Enable all Multicast addresses */
@@ -4744,7 +4921,7 @@ static void s2io_set_multicast(struct net_device *dev)
                       &bar0->rmac_addr_data1_mem);
                val64 = RMAC_ADDR_CMD_MEM_WE |
                    RMAC_ADDR_CMD_MEM_STROBE_NEW_CMD |
-                   RMAC_ADDR_CMD_MEM_OFFSET(MAC_MC_ALL_MC_ADDR_OFFSET);
+                   RMAC_ADDR_CMD_MEM_OFFSET(config->max_mc_addr - 1);
                writeq(val64, &bar0->rmac_addr_cmd_mem);
                /* Wait till command completes */
                wait_for_cmd_complete(&bar0->rmac_addr_cmd_mem,
@@ -4752,7 +4929,7 @@ static void s2io_set_multicast(struct net_device *dev)
                                        S2IO_BIT_RESET);
 
                sp->m_cast_flg = 1;
-               sp->all_multi_pos = MAC_MC_ALL_MC_ADDR_OFFSET;
+               sp->all_multi_pos = config->max_mc_addr - 1;
        } else if ((dev->flags & IFF_ALLMULTI) && (sp->m_cast_flg)) {
                /*  Disable all Multicast addresses */
                writeq(RMAC_ADDR_DATA0_MEM_ADDR(dis_addr),
@@ -4821,7 +4998,7 @@ static void s2io_set_multicast(struct net_device *dev)
        /*  Update individual M_CAST address list */
        if ((!sp->m_cast_flg) && dev->mc_count) {
                if (dev->mc_count >
-                   (MAX_ADDRS_SUPPORTED - MAC_MC_ADDR_START_OFFSET - 1)) {
+                   (config->max_mc_addr - config->max_mac_addr)) {
                        DBG_PRINT(ERR_DBG, "%s: No more Rx filters ",
                                  dev->name);
                        DBG_PRINT(ERR_DBG, "can be added, please enable ");
@@ -4841,7 +5018,7 @@ static void s2io_set_multicast(struct net_device *dev)
                        val64 = RMAC_ADDR_CMD_MEM_WE |
                            RMAC_ADDR_CMD_MEM_STROBE_NEW_CMD |
                            RMAC_ADDR_CMD_MEM_OFFSET
-                           (MAC_MC_ADDR_START_OFFSET + i);
+                           (config->mc_start_offset + i);
                        writeq(val64, &bar0->rmac_addr_cmd_mem);
 
                        /* Wait for command completes */
@@ -4873,7 +5050,7 @@ static void s2io_set_multicast(struct net_device *dev)
                        val64 = RMAC_ADDR_CMD_MEM_WE |
                            RMAC_ADDR_CMD_MEM_STROBE_NEW_CMD |
                            RMAC_ADDR_CMD_MEM_OFFSET
-                           (i + MAC_MC_ADDR_START_OFFSET);
+                           (i + config->mc_start_offset);
                        writeq(val64, &bar0->rmac_addr_cmd_mem);
 
                        /* Wait for command completes */
@@ -4889,8 +5066,78 @@ static void s2io_set_multicast(struct net_device *dev)
        }
 }
 
-/* add unicast MAC address to CAM */
-static int do_s2io_add_unicast(struct s2io_nic *sp, u64 addr, int off)
+/* read from CAM unicast & multicast addresses and store it in
+ * def_mac_addr structure
+ */
+void do_s2io_store_unicast_mc(struct s2io_nic *sp)
+{
+       int offset;
+       u64 mac_addr = 0x0;
+       struct config_param *config = &sp->config;
+
+       /* store unicast & multicast mac addresses */
+       for (offset = 0; offset < config->max_mc_addr; offset++) {
+               mac_addr = do_s2io_read_unicast_mc(sp, offset);
+               /* if read fails disable the entry */
+               if (mac_addr == FAILURE)
+                       mac_addr = S2IO_DISABLE_MAC_ENTRY;
+               do_s2io_copy_mac_addr(sp, offset, mac_addr);
+       }
+}
+
+/* restore unicast & multicast MAC to CAM from def_mac_addr structure */
+static void do_s2io_restore_unicast_mc(struct s2io_nic *sp)
+{
+       int offset;
+       struct config_param *config = &sp->config;
+       /* restore unicast mac address */
+       for (offset = 0; offset < config->max_mac_addr; offset++)
+               do_s2io_prog_unicast(sp->dev,
+                       sp->def_mac_addr[offset].mac_addr);
+
+       /* restore multicast mac address */
+       for (offset = config->mc_start_offset;
+               offset < config->max_mc_addr; offset++)
+               do_s2io_add_mc(sp, sp->def_mac_addr[offset].mac_addr);
+}
+
+/* add a multicast MAC address to CAM */
+static int do_s2io_add_mc(struct s2io_nic *sp, u8 *addr)
+{
+       int i;
+       u64 mac_addr = 0;
+       struct config_param *config = &sp->config;
+
+       for (i = 0; i < ETH_ALEN; i++) {
+               mac_addr <<= 8;
+               mac_addr |= addr[i];
+       }
+       if ((0ULL == mac_addr) || (mac_addr == S2IO_DISABLE_MAC_ENTRY))
+               return SUCCESS;
+
+       /* check if the multicast mac already preset in CAM */
+       for (i = config->mc_start_offset; i < config->max_mc_addr; i++) {
+               u64 tmp64;
+               tmp64 = do_s2io_read_unicast_mc(sp, i);
+               if (tmp64 == S2IO_DISABLE_MAC_ENTRY) /* CAM entry is empty */
+                       break;
+
+               if (tmp64 == mac_addr)
+                       return SUCCESS;
+       }
+       if (i == config->max_mc_addr) {
+               DBG_PRINT(ERR_DBG,
+                       "CAM full no space left for multicast MAC\n");
+               return FAILURE;
+       }
+       /* Update the internal structure with this new mac address */
+       do_s2io_copy_mac_addr(sp, i, mac_addr);
+
+       return (do_s2io_add_mac(sp, mac_addr, i));
+}
+
+/* add MAC address to CAM */
+static int do_s2io_add_mac(struct s2io_nic *sp, u64 addr, int off)
 {
        u64 val64;
        struct XENA_dev_config __iomem *bar0 = sp->bar0;
@@ -4907,15 +5154,62 @@ static int do_s2io_add_unicast(struct s2io_nic *sp, u64 addr, int off)
        if (wait_for_cmd_complete(&bar0->rmac_addr_cmd_mem,
                RMAC_ADDR_CMD_MEM_STROBE_CMD_EXECUTING,
                S2IO_BIT_RESET)) {
-               DBG_PRINT(INFO_DBG, "add_mac_addr failed\n");
+               DBG_PRINT(INFO_DBG, "do_s2io_add_mac failed\n");
                return FAILURE;
        }
        return SUCCESS;
 }
+/* deletes a specified unicast/multicast mac entry from CAM */
+static int do_s2io_delete_unicast_mc(struct s2io_nic *sp, u64 addr)
+{
+       int offset;
+       u64 dis_addr = S2IO_DISABLE_MAC_ENTRY, tmp64;
+       struct config_param *config = &sp->config;
+
+       for (offset = 1;
+               offset < config->max_mc_addr; offset++) {
+               tmp64 = do_s2io_read_unicast_mc(sp, offset);
+               if (tmp64 == addr) {
+                       /* disable the entry by writing  0xffffffffffffULL */
+                       if (do_s2io_add_mac(sp, dis_addr, offset) ==  FAILURE)
+                               return FAILURE;
+                       /* store the new mac list from CAM */
+                       do_s2io_store_unicast_mc(sp);
+                       return SUCCESS;
+               }
+       }
+       DBG_PRINT(ERR_DBG, "MAC address 0x%llx not found in CAM\n",
+                       (unsigned long long)addr);
+       return FAILURE;
+}
+
+/* read mac entries from CAM */
+static u64 do_s2io_read_unicast_mc(struct s2io_nic *sp, int offset)
+{
+       u64 tmp64 = 0xffffffffffff0000ULL, val64;
+       struct XENA_dev_config __iomem *bar0 = sp->bar0;
+
+       /* read mac addr */
+       val64 =
+               RMAC_ADDR_CMD_MEM_RD | RMAC_ADDR_CMD_MEM_STROBE_NEW_CMD |
+               RMAC_ADDR_CMD_MEM_OFFSET(offset);
+       writeq(val64, &bar0->rmac_addr_cmd_mem);
+
+       /* Wait till command completes */
+       if (wait_for_cmd_complete(&bar0->rmac_addr_cmd_mem,
+               RMAC_ADDR_CMD_MEM_STROBE_CMD_EXECUTING,
+               S2IO_BIT_RESET)) {
+               DBG_PRINT(INFO_DBG, "do_s2io_read_unicast_mc failed\n");
+               return FAILURE;
+       }
+       tmp64 = readq(&bar0->rmac_addr_data0_mem);
+       return (tmp64 >> 16);
+}
 
 /**
  * s2io_set_mac_addr driver entry point
  */
+
 static int s2io_set_mac_addr(struct net_device *dev, void *p)
 {
        struct sockaddr *addr = p;
@@ -4928,7 +5222,6 @@ static int s2io_set_mac_addr(struct net_device *dev, void *p)
        /* store the MAC address in CAM */
        return (do_s2io_prog_unicast(dev, dev->dev_addr));
 }
-
 /**
  *  do_s2io_prog_unicast - Programs the Xframe mac address
  *  @dev : pointer to the device structure.
@@ -4938,11 +5231,14 @@ static int s2io_set_mac_addr(struct net_device *dev, void *p)
  *  Return value: SUCCESS on success and an appropriate (-)ve integer
  *  as defined in errno.h file on failure.
  */
+
 static int do_s2io_prog_unicast(struct net_device *dev, u8 *addr)
 {
        struct s2io_nic *sp = dev->priv;
        register u64 mac_addr = 0, perm_addr = 0;
        int i;
+       u64 tmp64;
+       struct config_param *config = &sp->config;
 
        /*
        * Set the new MAC address as the new unicast filter and reflect this
@@ -4960,9 +5256,26 @@ static int do_s2io_prog_unicast(struct net_device *dev, u8 *addr)
        if (mac_addr == perm_addr)
                return SUCCESS;
 
+       /* check if the mac already preset in CAM */
+       for (i = 1; i < config->max_mac_addr; i++) {
+               tmp64 = do_s2io_read_unicast_mc(sp, i);
+               if (tmp64 == S2IO_DISABLE_MAC_ENTRY) /* CAM entry is empty */
+                       break;
+
+               if (tmp64 == mac_addr) {
+                       DBG_PRINT(INFO_DBG,
+                       "MAC addr:0x%llx already present in CAM\n",
+                       (unsigned long long)mac_addr);
+                       return SUCCESS;
+               }
+       }
+       if (i == config->max_mac_addr) {
+               DBG_PRINT(ERR_DBG, "CAM full no space left for Unicast MAC\n");
+               return FAILURE;
+       }
        /* Update the internal structure with this new mac address */
-       do_s2io_copy_mac_addr(sp, 0, mac_addr);
-       return (do_s2io_add_unicast(sp, mac_addr, 0));
+       do_s2io_copy_mac_addr(sp, i, mac_addr);
+       return (do_s2io_add_mac(sp, mac_addr, i));
 }
 
 /**
@@ -6382,16 +6695,15 @@ static int s2io_change_mtu(struct net_device *dev, int new_mtu)
 
        dev->mtu = new_mtu;
        if (netif_running(dev)) {
+               s2io_stop_all_tx_queue(sp);
                s2io_card_down(sp);
-               netif_stop_queue(dev);
                ret = s2io_card_up(sp);
                if (ret) {
                        DBG_PRINT(ERR_DBG, "%s: Device bring up failed\n",
                                  __FUNCTION__);
                        return ret;
                }
-               if (netif_queue_stopped(dev))
-                       netif_wake_queue(dev);
+               s2io_wake_all_tx_queue(sp);
        } else { /* Device is down */
                struct XENA_dev_config __iomem *bar0 = sp->bar0;
                u64 val64 = new_mtu;
@@ -6402,49 +6714,6 @@ static int s2io_change_mtu(struct net_device *dev, int new_mtu)
        return ret;
 }
 
-/**
- *  s2io_tasklet - Bottom half of the ISR.
- *  @dev_adr : address of the device structure in dma_addr_t format.
- *  Description:
- *  This is the tasklet or the bottom half of the ISR. This is
- *  an extension of the ISR which is scheduled by the scheduler to be run
- *  when the load on the CPU is low. All low priority tasks of the ISR can
- *  be pushed into the tasklet. For now the tasklet is used only to
- *  replenish the Rx buffers in the Rx buffer descriptors.
- *  Return value:
- *  void.
- */
-
-static void s2io_tasklet(unsigned long dev_addr)
-{
-       struct net_device *dev = (struct net_device *) dev_addr;
-       struct s2io_nic *sp = dev->priv;
-       int i, ret;
-       struct mac_info *mac_control;
-       struct config_param *config;
-
-       mac_control = &sp->mac_control;
-       config = &sp->config;
-
-       if (!TASKLET_IN_USE) {
-               for (i = 0; i < config->rx_ring_num; i++) {
-                       ret = fill_rx_buffers(sp, i);
-                       if (ret == -ENOMEM) {
-                               DBG_PRINT(INFO_DBG, "%s: Out of ",
-                                         dev->name);
-                               DBG_PRINT(INFO_DBG, "memory in tasklet\n");
-                               break;
-                       } else if (ret == -EFILL) {
-                               DBG_PRINT(INFO_DBG,
-                                         "%s: Rx Ring %d is full\n",
-                                         dev->name, i);
-                               break;
-                       }
-               }
-               clear_bit(0, (&sp->tasklet_status));
-       }
-}
-
 /**
  * s2io_set_link - Set the LInk status
  * @data: long pointer to device private structue
@@ -6499,7 +6768,7 @@ static void s2io_set_link(struct work_struct *work)
                        } else {
                                DBG_PRINT(ERR_DBG, "%s: Error: ", dev->name);
                                DBG_PRINT(ERR_DBG, "device is not Quiescent\n");
-                               netif_stop_queue(dev);
+                               s2io_stop_all_tx_queue(nic);
                        }
                }
                val64 = readq(&bar0->adapter_control);
@@ -6725,12 +6994,12 @@ static int s2io_add_isr(struct s2io_nic * sp)
                                /* If either data or addr is zero print it */
                                if(!(sp->msix_info[i].addr &&
                                        sp->msix_info[i].data)) {
-                                       DBG_PRINT(ERR_DBG, "%s @ Addr:0x%llx"
-                                               "Data:0x%lx\n",sp->desc[i],
+                                       DBG_PRINT(ERR_DBG, "%s @ Addr:0x%llx "
+                                               "Data:0x%llx\n",sp->desc[i],
                                                (unsigned long long)
                                                sp->msix_info[i].addr,
-                                               (unsigned long)
-                                               ntohl(sp->msix_info[i].data));
+                                               (unsigned long long)
+                                               sp->msix_info[i].data);
                                } else {
                                        msix_tx_cnt++;
                                }
@@ -6743,12 +7012,12 @@ static int s2io_add_isr(struct s2io_nic * sp)
                                /* If either data or addr is zero print it */
                                if(!(sp->msix_info[i].addr &&
                                        sp->msix_info[i].data)) {
-                                       DBG_PRINT(ERR_DBG, "%s @ Addr:0x%llx"
-                                               "Data:0x%lx\n",sp->desc[i],
+                                       DBG_PRINT(ERR_DBG, "%s @ Addr:0x%llx "
+                                               "Data:0x%llx\n",sp->desc[i],
                                                (unsigned long long)
                                                sp->msix_info[i].addr,
-                                               (unsigned long)
-                                               ntohl(sp->msix_info[i].data));
+                                               (unsigned long long)
+                                               sp->msix_info[i].data);
                                } else {
                                        msix_rx_cnt++;
                                }
@@ -6794,8 +7063,9 @@ static void do_s2io_card_down(struct s2io_nic * sp, int do_io)
 {
        int cnt = 0;
        struct XENA_dev_config __iomem *bar0 = sp->bar0;
-       unsigned long flags;
        register u64 val64 = 0;
+       struct config_param *config;
+       config = &sp->config;
 
        if (!is_s2io_card_up(sp))
                return;
@@ -6807,15 +7077,16 @@ static void do_s2io_card_down(struct s2io_nic * sp, int do_io)
        }
        clear_bit(__S2IO_STATE_CARD_UP, &sp->state);
 
+       /* Disable napi */
+       if (config->napi)
+               napi_disable(&sp->napi);
+
        /* disable Tx and Rx traffic on the NIC */
        if (do_io)
                stop_nic(sp);
 
        s2io_rem_isr(sp);
 
-       /* Kill tasklet. */
-       tasklet_kill(&sp->task);
-
        /* Check if the device is Quiescent and then Reset the NIC */
        while(do_io) {
                /* As per the HW requirement we need to replenish the
@@ -6846,15 +7117,11 @@ static void do_s2io_card_down(struct s2io_nic * sp, int do_io)
        if (do_io)
                s2io_reset(sp);
 
-       spin_lock_irqsave(&sp->tx_lock, flags);
        /* Free all Tx buffers */
        free_tx_buffers(sp);
-       spin_unlock_irqrestore(&sp->tx_lock, flags);
 
        /* Free all Rx buffers */
-       spin_lock_irqsave(&sp->rx_lock, flags);
        free_rx_buffers(sp);
-       spin_unlock_irqrestore(&sp->rx_lock, flags);
 
        clear_bit(__S2IO_STATE_LINK_TASK, &(sp->state));
 }
@@ -6900,6 +7167,11 @@ static int s2io_card_up(struct s2io_nic * sp)
                DBG_PRINT(INFO_DBG, "Buf in ring:%d is %d:\n", i,
                          atomic_read(&sp->rx_bufs_left[i]));
        }
+
+       /* Initialise napi */
+       if (config->napi)
+               napi_enable(&sp->napi);
+
        /* Maintain the state prior to the open */
        if (sp->promisc_flg)
                sp->promisc_flg = 0;
@@ -6938,9 +7210,6 @@ static int s2io_card_up(struct s2io_nic * sp)
 
        S2IO_TIMER_CONF(sp->alarm_timer, s2io_alarm_handle, sp, (HZ/2));
 
-       /* Enable tasklet for the device */
-       tasklet_init(&sp->task, s2io_tasklet, (unsigned long) dev);
-
        /*  Enable select interrupts */
        en_dis_err_alarms(sp, ENA_ALL_INTRS, ENABLE_INTRS);
        if (sp->config.intr_type != INTA)
@@ -6980,7 +7249,7 @@ static void s2io_restart_nic(struct work_struct *work)
                DBG_PRINT(ERR_DBG, "%s: Device bring up failed\n",
                          dev->name);
        }
-       netif_wake_queue(dev);
+       s2io_wake_all_tx_queue(sp);
        DBG_PRINT(ERR_DBG, "%s: was reset by Tx watchdog timer\n",
                  dev->name);
 out_unlock:
@@ -7170,7 +7439,8 @@ static int rx_osm_handler(struct ring_info *ring_data, struct RxD_t * rxdp)
                                        {
                                                lro_append_pkt(sp, lro,
                                                        skb, tcp_len);
-                                               queue_rx_frame(lro->parent);
+                                               queue_rx_frame(lro->parent,
+                                                       lro->vlan_tag);
                                                clear_lro_session(lro);
                                                sp->mac_control.stats_info->
                                                    sw_stat.flush_max_pkts++;
@@ -7181,7 +7451,8 @@ static int rx_osm_handler(struct ring_info *ring_data, struct RxD_t * rxdp)
                                                        lro->frags_len;
                                                sp->mac_control.stats_info->
                                                     sw_stat.sending_both++;
-                                               queue_rx_frame(lro->parent);
+                                               queue_rx_frame(lro->parent,
+                                                       lro->vlan_tag);
                                                clear_lro_session(lro);
                                                goto send_up;
                                        case 0: /* sessions exceeded */
@@ -7207,31 +7478,12 @@ static int rx_osm_handler(struct ring_info *ring_data, struct RxD_t * rxdp)
                         */
                        skb->ip_summed = CHECKSUM_NONE;
                }
-       } else {
+       } else
                skb->ip_summed = CHECKSUM_NONE;
-       }
+
        sp->mac_control.stats_info->sw_stat.mem_freed += skb->truesize;
-       if (!sp->lro) {
-               skb->protocol = eth_type_trans(skb, dev);
-               if ((sp->vlgrp && RXD_GET_VLAN_TAG(rxdp->Control_2) &&
-                       vlan_strip_flag)) {
-                       /* Queueing the vlan frame to the upper layer */
-                       if (napi)
-                               vlan_hwaccel_receive_skb(skb, sp->vlgrp,
-                                       RXD_GET_VLAN_TAG(rxdp->Control_2));
-                       else
-                               vlan_hwaccel_rx(skb, sp->vlgrp,
-                                       RXD_GET_VLAN_TAG(rxdp->Control_2));
-               } else {
-                       if (napi)
-                               netif_receive_skb(skb);
-                       else
-                               netif_rx(skb);
-               }
-       } else {
 send_up:
-               queue_rx_frame(skb);
-       }
+       queue_rx_frame(skb, RXD_GET_VLAN_TAG(rxdp->Control_2));
        dev->last_rx = jiffies;
 aggregate:
        atomic_dec(&sp->rx_bufs_left[ring_no]);
@@ -7256,8 +7508,10 @@ static void s2io_link(struct s2io_nic * sp, int link)
        struct net_device *dev = (struct net_device *) sp->dev;
 
        if (link != sp->last_link_state) {
+               init_tti(sp, link);
                if (link == LINK_DOWN) {
                        DBG_PRINT(ERR_DBG, "%s: Link down\n", dev->name);
+                       s2io_stop_all_tx_queue(sp);
                        netif_carrier_off(dev);
                        if(sp->mac_control.stats_info->sw_stat.link_up_cnt)
                        sp->mac_control.stats_info->sw_stat.link_up_time =
@@ -7270,6 +7524,7 @@ static void s2io_link(struct s2io_nic * sp, int link)
                                jiffies - sp->start_time;
                        sp->mac_control.stats_info->sw_stat.link_up_cnt++;
                        netif_carrier_on(dev);
+                       s2io_wake_all_tx_queue(sp);
                }
        }
        sp->last_link_state = link;
@@ -7306,14 +7561,48 @@ static void s2io_init_pci(struct s2io_nic * sp)
        pci_read_config_word(sp->pdev, PCI_COMMAND, &pci_cmd);
 }
 
-static int s2io_verify_parm(struct pci_dev *pdev, u8 *dev_intr_type)
+static int s2io_verify_parm(struct pci_dev *pdev, u8 *dev_intr_type,
+       u8 *dev_multiq)
 {
-       if ( tx_fifo_num > 8) {
-               DBG_PRINT(ERR_DBG, "s2io: Requested number of Tx fifos not "
+       if ((tx_fifo_num > MAX_TX_FIFOS) ||
+               (tx_fifo_num < 1)) {
+               DBG_PRINT(ERR_DBG, "s2io: Requested number of tx fifos "
+                       "(%d) not supported\n", tx_fifo_num);
+
+               if (tx_fifo_num < 1)
+                       tx_fifo_num = 1;
+               else
+                       tx_fifo_num = MAX_TX_FIFOS;
+
+               DBG_PRINT(ERR_DBG, "s2io: Default to %d ", tx_fifo_num);
+               DBG_PRINT(ERR_DBG, "tx fifos\n");
+       }
+
+#ifndef CONFIG_NETDEVICES_MULTIQUEUE
+       if (multiq) {
+               DBG_PRINT(ERR_DBG, "s2io: Multiqueue support not enabled\n");
+               multiq = 0;
+       }
+#endif
+       if (multiq)
+               *dev_multiq = multiq;
+
+       if (tx_steering_type && (1 == tx_fifo_num)) {
+               if (tx_steering_type != TX_DEFAULT_STEERING)
+                       DBG_PRINT(ERR_DBG,
+                               "s2io: Tx steering is not supported with "
+                               "one fifo. Disabling Tx steering.\n");
+               tx_steering_type = NO_STEERING;
+       }
+
+       if ((tx_steering_type < NO_STEERING) ||
+               (tx_steering_type > TX_DEFAULT_STEERING)) {
+               DBG_PRINT(ERR_DBG, "s2io: Requested transmit steering not "
                         "supported\n");
-               DBG_PRINT(ERR_DBG, "s2io: Default to 8 Tx fifos\n");
-               tx_fifo_num = 8;
+               DBG_PRINT(ERR_DBG, "s2io: Disabling transmit steering\n");
+               tx_steering_type = NO_STEERING;
        }
+
        if ( rx_ring_num > 8) {
                DBG_PRINT(ERR_DBG, "s2io: Requested number of Rx rings not "
                         "supported\n");
@@ -7348,7 +7637,7 @@ static int s2io_verify_parm(struct pci_dev *pdev, u8 *dev_intr_type)
 /**
  * rts_ds_steer - Receive traffic steering based on IPv4 or IPv6 TOS
  * or Traffic class respectively.
- * @nic: device peivate variable
+ * @nic: device private variable
  * Description: The function configures the receive steering to
  * desired receive ring.
  * Return Value:  SUCCESS on success and
@@ -7405,9 +7694,11 @@ s2io_init_nic(struct pci_dev *pdev, const struct pci_device_id *pre)
        struct config_param *config;
        int mode;
        u8 dev_intr_type = intr_type;
+       u8 dev_multiq = 0;
        DECLARE_MAC_BUF(mac);
 
-       if ((ret = s2io_verify_parm(pdev, &dev_intr_type)))
+       ret = s2io_verify_parm(pdev, &dev_intr_type, &dev_multiq);
+       if (ret)
                return ret;
 
        if ((ret = pci_enable_device(pdev))) {
@@ -7438,7 +7729,11 @@ s2io_init_nic(struct pci_dev *pdev, const struct pci_device_id *pre)
                pci_disable_device(pdev);
                return -ENODEV;
        }
-
+#ifdef CONFIG_NETDEVICES_MULTIQUEUE
+       if (dev_multiq)
+               dev = alloc_etherdev_mq(sizeof(struct s2io_nic), tx_fifo_num);
+       else
+#endif
        dev = alloc_etherdev(sizeof(struct s2io_nic));
        if (dev == NULL) {
                DBG_PRINT(ERR_DBG, "Device allocation failed\n");
@@ -7487,17 +7782,45 @@ s2io_init_nic(struct pci_dev *pdev, const struct pci_device_id *pre)
        config = &sp->config;
 
        config->napi = napi;
+       config->tx_steering_type = tx_steering_type;
 
        /* Tx side parameters. */
-       config->tx_fifo_num = tx_fifo_num;
-       for (i = 0; i < MAX_TX_FIFOS; i++) {
+       if (config->tx_steering_type == TX_PRIORITY_STEERING)
+               config->tx_fifo_num = MAX_TX_FIFOS;
+       else
+               config->tx_fifo_num = tx_fifo_num;
+
+       /* Initialize the fifos used for tx steering */
+       if (config->tx_fifo_num < 5) {
+                       if (config->tx_fifo_num  == 1)
+                               sp->total_tcp_fifos = 1;
+                       else
+                               sp->total_tcp_fifos = config->tx_fifo_num - 1;
+                       sp->udp_fifo_idx = config->tx_fifo_num - 1;
+                       sp->total_udp_fifos = 1;
+                       sp->other_fifo_idx = sp->total_tcp_fifos - 1;
+       } else {
+               sp->total_tcp_fifos = (tx_fifo_num - FIFO_UDP_MAX_NUM -
+                                               FIFO_OTHER_MAX_NUM);
+               sp->udp_fifo_idx = sp->total_tcp_fifos;
+               sp->total_udp_fifos = FIFO_UDP_MAX_NUM;
+               sp->other_fifo_idx = sp->udp_fifo_idx + FIFO_UDP_MAX_NUM;
+       }
+
+       config->multiq = dev_multiq;
+       for (i = 0; i < config->tx_fifo_num; i++) {
                config->tx_cfg[i].fifo_len = tx_fifo_len[i];
                config->tx_cfg[i].fifo_priority = i;
        }
 
        /* mapping the QoS priority to the configured fifos */
        for (i = 0; i < MAX_TX_FIFOS; i++)
-               config->fifo_mapping[i] = fifo_map[config->tx_fifo_num][i];
+               config->fifo_mapping[i] = fifo_map[config->tx_fifo_num - 1][i];
+
+       /* map the hashing selector table to the configured fifos */
+       for (i = 0; i < config->tx_fifo_num; i++)
+               sp->fifo_selector[i] = fifo_selector[i];
+
 
        config->tx_intr_type = TXD_INT_TYPE_UTILZ;
        for (i = 0; i < config->tx_fifo_num; i++) {
@@ -7582,6 +7905,7 @@ s2io_init_nic(struct pci_dev *pdev, const struct pci_device_id *pre)
        SET_ETHTOOL_OPS(dev, &netdev_ethtool_ops);
        dev->features |= NETIF_F_HW_VLAN_TX | NETIF_F_HW_VLAN_RX;
        dev->vlan_rx_register = s2io_vlan_rx_register;
+       dev->vlan_rx_kill_vid = (void *)s2io_vlan_rx_kill_vid;
 
        /*
         * will use eth_mac_addr() for  dev->set_mac_address
@@ -7602,7 +7926,10 @@ s2io_init_nic(struct pci_dev *pdev, const struct pci_device_id *pre)
                dev->features |= NETIF_F_UFO;
                dev->features |= NETIF_F_HW_CSUM;
        }
-
+#ifdef CONFIG_NETDEVICES_MULTIQUEUE
+       if (config->multiq)
+               dev->features |= NETIF_F_MULTI_QUEUE;
+#endif
        dev->tx_timeout = &s2io_tx_watchdog;
        dev->watchdog_timeo = WATCH_DOG_TIMEOUT;
        INIT_WORK(&sp->rst_timer_task, s2io_restart_nic);
@@ -7645,7 +7972,7 @@ s2io_init_nic(struct pci_dev *pdev, const struct pci_device_id *pre)
         */
        bar0 = sp->bar0;
        val64 = RMAC_ADDR_CMD_MEM_RD | RMAC_ADDR_CMD_MEM_STROBE_NEW_CMD |
-           RMAC_ADDR_CMD_MEM_OFFSET(0 + MAC_MAC_ADDR_START_OFFSET);
+           RMAC_ADDR_CMD_MEM_OFFSET(0 + S2IO_MAC_ADDR_START_OFFSET);
        writeq(val64, &bar0->rmac_addr_cmd_mem);
        wait_for_cmd_complete(&bar0->rmac_addr_cmd_mem,
                      RMAC_ADDR_CMD_MEM_STROBE_CMD_EXECUTING, S2IO_BIT_RESET);
@@ -7665,24 +7992,34 @@ s2io_init_nic(struct pci_dev *pdev, const struct pci_device_id *pre)
        memcpy(dev->dev_addr, sp->def_mac_addr, ETH_ALEN);
        memcpy(dev->perm_addr, dev->dev_addr, ETH_ALEN);
 
+       /* initialize number of multicast & unicast MAC entries variables */
+       if (sp->device_type == XFRAME_I_DEVICE) {
+               config->max_mc_addr = S2IO_XENA_MAX_MC_ADDRESSES;
+               config->max_mac_addr = S2IO_XENA_MAX_MAC_ADDRESSES;
+               config->mc_start_offset = S2IO_XENA_MC_ADDR_START_OFFSET;
+       } else if (sp->device_type == XFRAME_II_DEVICE) {
+               config->max_mc_addr = S2IO_HERC_MAX_MC_ADDRESSES;
+               config->max_mac_addr = S2IO_HERC_MAX_MAC_ADDRESSES;
+               config->mc_start_offset = S2IO_HERC_MC_ADDR_START_OFFSET;
+       }
+
+       /* store mac addresses from CAM to s2io_nic structure */
+       do_s2io_store_unicast_mc(sp);
+
         /* Store the values of the MSIX table in the s2io_nic structure */
        store_xmsi_data(sp);
        /* reset Nic and bring it to known state */
        s2io_reset(sp);
 
        /*
-        * Initialize the tasklet status and link state flags
+        * Initialize link state flags
         * and the card state parameter
         */
-       sp->tasklet_status = 0;
        sp->state = 0;
 
        /* Initialize spinlocks */
-       spin_lock_init(&sp->tx_lock);
-
-       if (!napi)
-               spin_lock_init(&sp->put_lock);
-       spin_lock_init(&sp->rx_lock);
+       for (i = 0; i < sp->config.tx_fifo_num; i++)
+               spin_lock_init(&mac_control->fifos[i].tx_lock);
 
        /*
         * SXE-002: Configure link and activity LED to init state
@@ -7736,6 +8073,10 @@ s2io_init_nic(struct pci_dev *pdev, const struct pci_device_id *pre)
 
        if (napi)
                DBG_PRINT(ERR_DBG, "%s: NAPI enabled\n", dev->name);
+
+       DBG_PRINT(ERR_DBG, "%s: Using %d Tx fifo(s)\n", dev->name,
+               sp->config.tx_fifo_num);
+
        switch(sp->config.intr_type) {
                case INTA:
                    DBG_PRINT(ERR_DBG, "%s: Interrupt type INTA\n", dev->name);
@@ -7744,6 +8085,29 @@ s2io_init_nic(struct pci_dev *pdev, const struct pci_device_id *pre)
                    DBG_PRINT(ERR_DBG, "%s: Interrupt type MSI-X\n", dev->name);
                    break;
        }
+       if (sp->config.multiq) {
+       for (i = 0; i < sp->config.tx_fifo_num; i++)
+               mac_control->fifos[i].multiq = config->multiq;
+               DBG_PRINT(ERR_DBG, "%s: Multiqueue support enabled\n",
+                       dev->name);
+       } else
+               DBG_PRINT(ERR_DBG, "%s: Multiqueue support disabled\n",
+                       dev->name);
+
+       switch (sp->config.tx_steering_type) {
+       case NO_STEERING:
+               DBG_PRINT(ERR_DBG, "%s: No steering enabled for"
+                       " transmit\n", dev->name);
+                       break;
+       case TX_PRIORITY_STEERING:
+               DBG_PRINT(ERR_DBG, "%s: Priority steering enabled for"
+                       " transmit\n", dev->name);
+               break;
+       case TX_DEFAULT_STEERING:
+               DBG_PRINT(ERR_DBG, "%s: Default steering enabled for"
+                       " transmit\n", dev->name);
+       }
+
        if (sp->lro)
                DBG_PRINT(ERR_DBG, "%s: Large receive offload enabled\n",
                          dev->name);
@@ -7838,7 +8202,8 @@ module_init(s2io_starter);
 module_exit(s2io_closer);
 
 static int check_L2_lro_capable(u8 *buffer, struct iphdr **ip,
-               struct tcphdr **tcp, struct RxD_t *rxdp)
+               struct tcphdr **tcp, struct RxD_t *rxdp,
+               struct s2io_nic *sp)
 {
        int ip_off;
        u8 l2_type = (u8)((rxdp->Control_1 >> 37) & 0x7), ip_len;
@@ -7849,19 +8214,20 @@ static int check_L2_lro_capable(u8 *buffer, struct iphdr **ip,
                return -1;
        }
 
-       /* TODO:
-        * By default the VLAN field in the MAC is stripped by the card, if this
-        * feature is turned off in rx_pa_cfg register, then the ip_off field
-        * has to be shifted by a further 2 bytes
-        */
-       switch (l2_type) {
-               case 0: /* DIX type */
-               case 4: /* DIX type with VLAN */
-                       ip_off = HEADER_ETHERNET_II_802_3_SIZE;
-                       break;
+       /* Checking for DIX type or DIX type with VLAN */
+       if ((l2_type == 0)
+               || (l2_type == 4)) {
+               ip_off = HEADER_ETHERNET_II_802_3_SIZE;
+               /*
+                * If vlan stripping is disabled and the frame is VLAN tagged,
+                * shift the offset by the VLAN header size bytes.
+                */
+               if ((!vlan_strip_flag) &&
+                       (rxdp->Control_1 & RXD_FRAME_VLAN_TAG))
+                       ip_off += HEADER_VLAN_SIZE;
+       } else {
                /* LLC, SNAP etc are considered non-mergeable */
-               default:
-                       return -1;
+               return -1;
        }
 
        *ip = (struct iphdr *)((u8 *)buffer + ip_off);
@@ -7888,26 +8254,27 @@ static inline int get_l4_pyld_length(struct iphdr *ip, struct tcphdr *tcp)
 }
 
 static void initiate_new_session(struct lro *lro, u8 *l2h,
-                    struct iphdr *ip, struct tcphdr *tcp, u32 tcp_pyld_len)
+       struct iphdr *ip, struct tcphdr *tcp, u32 tcp_pyld_len, u16 vlan_tag)
 {
        DBG_PRINT(INFO_DBG,"%s: Been here...\n", __FUNCTION__);
        lro->l2h = l2h;
        lro->iph = ip;
        lro->tcph = tcp;
        lro->tcp_next_seq = tcp_pyld_len + ntohl(tcp->seq);
-       lro->tcp_ack = ntohl(tcp->ack_seq);
+       lro->tcp_ack = tcp->ack_seq;
        lro->sg_num = 1;
        lro->total_len = ntohs(ip->tot_len);
        lro->frags_len = 0;
+       lro->vlan_tag = vlan_tag;
        /*
         * check if we saw TCP timestamp. Other consistency checks have
         * already been done.
         */
        if (tcp->doff == 8) {
-               u32 *ptr;
-               ptr = (u32 *)(tcp+1);
+               __be32 *ptr;
+               ptr = (__be32 *)(tcp+1);
                lro->saw_ts = 1;
-               lro->cur_tsval = *(ptr+1);
+               lro->cur_tsval = ntohl(*(ptr+1));
                lro->cur_tsecr = *(ptr+2);
        }
        lro->in_use = 1;
@@ -7933,7 +8300,7 @@ static void update_L3L4_header(struct s2io_nic *sp, struct lro *lro)
 
        /* Update tsecr field if this session has timestamps enabled */
        if (lro->saw_ts) {
-               u32 *ptr = (u32 *)(tcp + 1);
+               __be32 *ptr = (__be32 *)(tcp + 1);
                *(ptr+2) = lro->cur_tsecr;
        }
 
@@ -7958,10 +8325,10 @@ static void aggregate_new_rx(struct lro *lro, struct iphdr *ip,
        lro->window = tcp->window;
 
        if (lro->saw_ts) {
-               u32 *ptr;
+               __be32 *ptr;
                /* Update tsecr and tsval from this packet */
-               ptr = (u32 *) (tcp + 1);
-               lro->cur_tsval = *(ptr + 1);
+               ptr = (__be32 *)(tcp+1);
+               lro->cur_tsval = ntohl(*(ptr+1));
                lro->cur_tsecr = *(ptr + 2);
        }
 }
@@ -8012,11 +8379,11 @@ static int verify_l3_l4_lro_capable(struct lro *l_lro, struct iphdr *ip,
 
                /* Ensure timestamp value increases monotonically */
                if (l_lro)
-                       if (l_lro->cur_tsval > *((u32 *)(ptr+2)))
+                       if (l_lro->cur_tsval > ntohl(*((__be32 *)(ptr+2))))
                                return -1;
 
                /* timestamp echo reply should be non-zero */
-               if (*((u32 *)(ptr+6)) == 0)
+               if (*((__be32 *)(ptr+6)) == 0)
                        return -1;
        }
 
@@ -8030,15 +8397,16 @@ s2io_club_tcp_session(u8 *buffer, u8 **tcp, u32 *tcp_len, struct lro **lro,
        struct iphdr *ip;
        struct tcphdr *tcph;
        int ret = 0, i;
+       u16 vlan_tag = 0;
 
        if (!(ret = check_L2_lro_capable(buffer, &ip, (struct tcphdr **)tcp,
-                                        rxdp))) {
+                                        rxdp, sp))) {
                DBG_PRINT(INFO_DBG,"IP Saddr: %x Daddr: %x\n",
                          ip->saddr, ip->daddr);
-       } else {
+       } else
                return ret;
-       }
 
+       vlan_tag = RXD_GET_VLAN_TAG(rxdp->Control_2);
        tcph = (struct tcphdr *)*tcp;
        *tcp_len = get_l4_pyld_length(ip, tcph);
        for (i=0; i<MAX_LRO_SESSIONS; i++) {
@@ -8098,7 +8466,8 @@ s2io_club_tcp_session(u8 *buffer, u8 **tcp, u32 *tcp_len, struct lro **lro,
 
        switch (ret) {
                case 3:
-                       initiate_new_session(*lro, buffer, ip, tcph, *tcp_len);
+                       initiate_new_session(*lro, buffer, ip, tcph, *tcp_len,
+                                                               vlan_tag);
                        break;
                case 2:
                        update_L3L4_header(sp, *lro);
@@ -8126,15 +8495,25 @@ static void clear_lro_session(struct lro *lro)
        memset(lro, 0, lro_struct_size);
 }
 
-static void queue_rx_frame(struct sk_buff *skb)
+static void queue_rx_frame(struct sk_buff *skb, u16 vlan_tag)
 {
        struct net_device *dev = skb->dev;
+       struct s2io_nic *sp = dev->priv;
 
        skb->protocol = eth_type_trans(skb, dev);
-       if (napi)
-               netif_receive_skb(skb);
-       else
-               netif_rx(skb);
+       if (sp->vlgrp && vlan_tag
+               && (vlan_strip_flag)) {
+               /* Queueing the vlan frame to the upper layer */
+               if (sp->config.napi)
+                       vlan_hwaccel_receive_skb(skb, sp->vlgrp, vlan_tag);
+               else
+                       vlan_hwaccel_rx(skb, sp->vlgrp, vlan_tag);
+       } else {
+               if (sp->config.napi)
+                       netif_receive_skb(skb);
+               else
+                       netif_rx(skb);
+       }
 }
 
 static void lro_append_pkt(struct s2io_nic *sp, struct lro *lro,