Merge branch 'linux-2.6' into for-2.6.22

9 files changed, 373 insertions, 65 deletions

diff --git a/drivers/net/cxgb3/common.h b/drivers/net/cxgb3/common.h
index e23deeb7d06..8d137963369 100644
--- a/drivers/net/cxgb3/common.h
+++ b/drivers/net/cxgb3/common.h
@@ -112,8 +112,7 @@ enum {
 };
 
 enum {
-	SUPPORTED_OFFLOAD = 1 << 24,
-	SUPPORTED_IRQ = 1 << 25
+	SUPPORTED_IRQ      = 1 << 24
 };
 
 enum {				/* adapter interrupt-maintained statistics */
@@ -260,6 +259,10 @@ struct mac_stats {
 	unsigned long serdes_signal_loss;
 	unsigned long xaui_pcs_ctc_err;
 	unsigned long xaui_pcs_align_change;
+
+	unsigned long num_toggled; /* # times toggled TxEn due to stuck TX */
+	unsigned long num_resets;  /* # times reset due to stuck TX */
+
 };
 
 struct tp_mib_stats {
@@ -354,6 +357,9 @@ enum {
 	MC5_MODE_72_BIT = 2
 };
 
+/* MC5 min active region size */
+enum { MC5_MIN_TIDS = 16 };
+
 struct vpd_params {
 	unsigned int cclk;
 	unsigned int mclk;
@@ -398,6 +404,13 @@ struct adapter_params {
 	unsigned int stats_update_period;	/* MAC stats accumulation period */
 	unsigned int linkpoll_period;	/* link poll period in 0.1s */
 	unsigned int rev;	/* chip revision */
+	unsigned int offload;
+};
+
+enum {					    /* chip revisions */
+	T3_REV_A  = 0,
+	T3_REV_B  = 2,
+	T3_REV_B2 = 3,
 };
 
 struct trace_params {
@@ -465,6 +478,13 @@ struct cmac {
 	struct adapter *adapter;
 	unsigned int offset;
 	unsigned int nucast;	/* # of address filters for unicast MACs */
+	unsigned int tx_tcnt;
+	unsigned int tx_xcnt;
+	u64 tx_mcnt;
+	unsigned int rx_xcnt;
+	u64 rx_mcnt;
+	unsigned int toggle_cnt;
+	unsigned int txen;
 	struct mac_stats stats;
 };
 
@@ -588,7 +608,7 @@ static inline int is_10G(const struct adapter *adap)
 
 static inline int is_offload(const struct adapter *adap)
 {
-	return adapter_info(adap)->caps & SUPPORTED_OFFLOAD;
+	return adap->params.offload;
 }
 
 static inline unsigned int core_ticks_per_usec(const struct adapter *adap)
@@ -666,6 +686,7 @@ int t3_mac_set_address(struct cmac *mac, unsigned int idx, u8 addr[6]);
 int t3_mac_set_num_ucast(struct cmac *mac, int n);
 const struct mac_stats *t3_mac_update_stats(struct cmac *mac);
 int t3_mac_set_speed_duplex_fc(struct cmac *mac, int speed, int duplex, int fc);
+int t3b2_mac_watchdog_task(struct cmac *mac);
 
 void t3_mc5_prep(struct adapter *adapter, struct mc5 *mc5, int mode);
 int t3_mc5_init(struct mc5 *mc5, unsigned int nservers, unsigned int nfilters,
diff --git a/drivers/net/cxgb3/cxgb3_main.c b/drivers/net/cxgb3/cxgb3_main.c
index 7ff834e45d6..67b4b219d92 100644
--- a/drivers/net/cxgb3/cxgb3_main.c
+++ b/drivers/net/cxgb3/cxgb3_main.c
@@ -42,6 +42,7 @@
 #include <linux/workqueue.h>
 #include <linux/proc_fs.h>
 #include <linux/rtnetlink.h>
+#include <linux/firmware.h>
 #include <asm/uaccess.h>
 
 #include "common.h"
@@ -184,16 +185,24 @@ void t3_os_link_changed(struct adapter *adapter, int port_id, int link_stat,
 			int speed, int duplex, int pause)
 {
 	struct net_device *dev = adapter->port[port_id];
+	struct port_info *pi = netdev_priv(dev);
+	struct cmac *mac = &pi->mac;
 
 	/* Skip changes from disabled ports. */
 	if (!netif_running(dev))
 		return;
 
 	if (link_stat != netif_carrier_ok(dev)) {
-		if (link_stat)
+		if (link_stat) {
+			t3_mac_enable(mac, MAC_DIRECTION_RX);
 			netif_carrier_on(dev);
-		else
+		} else {
 			netif_carrier_off(dev);
+			pi->phy.ops->power_down(&pi->phy, 1);
+			t3_mac_disable(mac, MAC_DIRECTION_RX);
+			t3_link_start(&pi->phy, mac, &pi->link_config);
+		}
+
 		link_report(dev);
 	}
 }
@@ -406,7 +415,7 @@ static void quiesce_rx(struct adapter *adap)
 static int setup_sge_qsets(struct adapter *adap)
 {
 	int i, j, err, irq_idx = 0, qset_idx = 0, dummy_dev_idx = 0;
-	unsigned int ntxq = is_offload(adap) ? SGE_TXQ_PER_SET : 1;
+	unsigned int ntxq = SGE_TXQ_PER_SET;
 
 	if (adap->params.rev > 0 && !(adap->flags & USING_MSI))
 		irq_idx = -1;
@@ -484,12 +493,14 @@ static ssize_t show_##name(struct device *d, struct device_attribute *attr, \
 static ssize_t set_nfilters(struct net_device *dev, unsigned int val)
 {
 	struct adapter *adap = dev->priv;
+	int min_tids = is_offload(adap) ? MC5_MIN_TIDS : 0;
 
 	if (adap->flags & FULL_INIT_DONE)
 		return -EBUSY;
 	if (val && adap->params.rev == 0)
 		return -EINVAL;
-	if (val > t3_mc5_size(&adap->mc5) - adap->params.mc5.nservers)
+	if (val > t3_mc5_size(&adap->mc5) - adap->params.mc5.nservers -
+	    min_tids)
 		return -EINVAL;
 	adap->params.mc5.nfilters = val;
 	return 0;
@@ -507,7 +518,8 @@ static ssize_t set_nservers(struct net_device *dev, unsigned int val)
 
 	if (adap->flags & FULL_INIT_DONE)
 		return -EBUSY;
-	if (val > t3_mc5_size(&adap->mc5) - adap->params.mc5.nfilters)
+	if (val > t3_mc5_size(&adap->mc5) - adap->params.mc5.nfilters -
+	    MC5_MIN_TIDS)
 		return -EINVAL;
 	adap->params.mc5.nservers = val;
 	return 0;
@@ -707,6 +719,28 @@ static void bind_qsets(struct adapter *adap)
 	}
 }
 
+#define FW_FNAME "t3fw-%d.%d.%d.bin"
+
+static int upgrade_fw(struct adapter *adap)
+{
+	int ret;
+	char buf[64];
+	const struct firmware *fw;
+	struct device *dev = &adap->pdev->dev;
+
+	snprintf(buf, sizeof(buf), FW_FNAME, FW_VERSION_MAJOR,
+		 FW_VERSION_MINOR, FW_VERSION_MICRO);
+	ret = request_firmware(&fw, buf, dev);
+	if (ret < 0) {
+		dev_err(dev, "could not upgrade firmware: unable to load %s\n",
+			buf);
+		return ret;
+	}
+	ret = t3_load_fw(adap, fw->data, fw->size);
+	release_firmware(fw);
+	return ret;
+}
+
 /**
  *	cxgb_up - enable the adapter
  *	@adapter: adapter being enabled
@@ -723,6 +757,8 @@ static int cxgb_up(struct adapter *adap)
 
 	if (!(adap->flags & FULL_INIT_DONE)) {
 		err = t3_check_fw_version(adap);
+		if (err == -EINVAL)
+			err = upgrade_fw(adap);
 		if (err)
 			goto out;
 
@@ -734,6 +770,8 @@ static int cxgb_up(struct adapter *adap)
 		if (err)
 			goto out;
 
+		t3_write_reg(adap, A_ULPRX_TDDP_PSZ, V_HPZ0(PAGE_SHIFT - 12));
+		
 		err = setup_sge_qsets(adap);
 		if (err)
 			goto out;
@@ -894,7 +932,7 @@ static int cxgb_open(struct net_device *dev)
 		return err;
 
 	set_bit(pi->port_id, &adapter->open_device_map);
-	if (!ofld_disable) {
+	if (is_offload(adapter) && !ofld_disable) {
 		err = offload_open(dev);
 		if (err)
 			printk(KERN_WARNING
@@ -1031,7 +1069,11 @@ static char stats_strings[][ETH_GSTRING_LEN] = {
 	"VLANinsertions     ",
 	"TxCsumOffload      ",
 	"RxCsumGood         ",
-	"RxDrops            "
+	"RxDrops            ",
+
+	"CheckTXEnToggled   ",
+	"CheckResets        ",
+
 };
 
 static int get_stats_count(struct net_device *dev)
@@ -1145,6 +1187,9 @@ static void get_stats(struct net_device *dev, struct ethtool_stats *stats,
 	*data++ = collect_sge_port_stats(adapter, pi, SGE_PSTAT_TX_CSUM);
 	*data++ = collect_sge_port_stats(adapter, pi, SGE_PSTAT_RX_CSUM_GOOD);
 	*data++ = s->rx_cong_drops;
+
+	*data++ = s->num_toggled;
+	*data++ = s->num_resets;
 }
 
 static inline void reg_block_dump(struct adapter *ap, void *buf,
@@ -1362,23 +1407,27 @@ static int set_rx_csum(struct net_device *dev, u32 data)
 
 static void get_sge_param(struct net_device *dev, struct ethtool_ringparam *e)
 {
-	struct adapter *adapter = dev->priv;
+	const struct adapter *adapter = dev->priv;
+	const struct port_info *pi = netdev_priv(dev);
+	const struct qset_params *q = &adapter->params.sge.qset[pi->first_qset];
 
 	e->rx_max_pending = MAX_RX_BUFFERS;
 	e->rx_mini_max_pending = 0;
 	e->rx_jumbo_max_pending = MAX_RX_JUMBO_BUFFERS;
 	e->tx_max_pending = MAX_TXQ_ENTRIES;
 
-	e->rx_pending = adapter->params.sge.qset[0].fl_size;
-	e->rx_mini_pending = adapter->params.sge.qset[0].rspq_size;
-	e->rx_jumbo_pending = adapter->params.sge.qset[0].jumbo_size;
-	e->tx_pending = adapter->params.sge.qset[0].txq_size[0];
+	e->rx_pending = q->fl_size;
+	e->rx_mini_pending = q->rspq_size;
+	e->rx_jumbo_pending = q->jumbo_size;
+	e->tx_pending = q->txq_size[0];
 }
 
 static int set_sge_param(struct net_device *dev, struct ethtool_ringparam *e)
 {
 	int i;
+	struct qset_params *q;
 	struct adapter *adapter = dev->priv;
+	const struct port_info *pi = netdev_priv(dev);
 
 	if (e->rx_pending > MAX_RX_BUFFERS ||
 	    e->rx_jumbo_pending > MAX_RX_JUMBO_BUFFERS ||
@@ -1393,9 +1442,8 @@ static int set_sge_param(struct net_device *dev, struct ethtool_ringparam *e)
 	if (adapter->flags & FULL_INIT_DONE)
 		return -EBUSY;
 
-	for (i = 0; i < SGE_QSETS; ++i) {
-		struct qset_params *q = &adapter->params.sge.qset[i];
-
+	q = &adapter->params.sge.qset[pi->first_qset];
+	for (i = 0; i < pi->nqsets; ++i, ++q) {
 		q->rspq_size = e->rx_mini_pending;
 		q->fl_size = e->rx_pending;
 		q->jumbo_size = e->rx_jumbo_pending;
@@ -2067,6 +2115,42 @@ static void check_link_status(struct adapter *adapter)
 	}
 }
 
+static void check_t3b2_mac(struct adapter *adapter)
+{
+	int i;
+
+	if (!rtnl_trylock())	/* synchronize with ifdown */
+		return;
+
+	for_each_port(adapter, i) {
+		struct net_device *dev = adapter->port[i];
+		struct port_info *p = netdev_priv(dev);
+		int status;
+
+		if (!netif_running(dev))
+			continue;
+
+		status = 0;
+		if (netif_running(dev) && netif_carrier_ok(dev))
+			status = t3b2_mac_watchdog_task(&p->mac);
+		if (status == 1)
+			p->mac.stats.num_toggled++;
+		else if (status == 2) {
+			struct cmac *mac = &p->mac;
+
+			t3_mac_set_mtu(mac, dev->mtu);
+			t3_mac_set_address(mac, 0, dev->dev_addr);
+			cxgb_set_rxmode(dev);
+			t3_link_start(&p->phy, mac, &p->link_config);
+			t3_mac_enable(mac, MAC_DIRECTION_RX | MAC_DIRECTION_TX);
+			t3_port_intr_enable(adapter, p->port_id);
+			p->mac.stats.num_resets++;
+		}
+	}
+	rtnl_unlock();
+}
+
+
 static void t3_adap_check_task(struct work_struct *work)
 {
 	struct adapter *adapter = container_of(work, struct adapter,
@@ -2087,6 +2171,9 @@ static void t3_adap_check_task(struct work_struct *work)
 		adapter->check_task_cnt = 0;
 	}
 
+	if (p->rev == T3_REV_B2)
+		check_t3b2_mac(adapter);
+
 	/* Schedule the next check update if any port is active. */
 	spin_lock(&adapter->work_lock);
 	if (adapter->open_device_map & PORT_MASK)
@@ -2195,9 +2282,9 @@ static void __devinit print_port_info(struct adapter *adap,
 
 		if (!test_bit(i, &adap->registered_device_map))
 			continue;
-		printk(KERN_INFO "%s: %s %s RNIC (rev %d) %s%s\n",
+		printk(KERN_INFO "%s: %s %s %sNIC (rev %d) %s%s\n",
 		       dev->name, ai->desc, pi->port_type->desc,
-		       adap->params.rev, buf,
+		       is_offload(adap) ? "R" : "", adap->params.rev, buf,
 		       (adap->flags & USING_MSIX) ? " MSI-X" :
 		       (adap->flags & USING_MSI) ? " MSI" : "");
 		if (adap->name == dev->name && adap->params.vpd.mclk)
diff --git a/drivers/net/cxgb3/cxgb3_offload.c b/drivers/net/cxgb3/cxgb3_offload.c
index f6ed033efb5..48649244673 100644
--- a/drivers/net/cxgb3/cxgb3_offload.c
+++ b/drivers/net/cxgb3/cxgb3_offload.c
@@ -553,7 +553,9 @@ int cxgb3_alloc_atid(struct t3cdev *tdev, struct cxgb3_client *client,
 	struct tid_info *t = &(T3C_DATA(tdev))->tid_maps;
 
 	spin_lock_bh(&t->atid_lock);
-	if (t->afree) {
+	if (t->afree &&
+	    t->atids_in_use + atomic_read(&t->tids_in_use) + MC5_MIN_TIDS <=
+	    t->ntids) {
 		union active_open_entry *p = t->afree;
 
 		atid = (p - t->atid_tab) + t->atid_base;
@@ -741,17 +743,6 @@ static int do_act_establish(struct t3cdev *dev, struct sk_buff *skb)
 	}
 }
 
-static int do_set_tcb_rpl(struct t3cdev *dev, struct sk_buff *skb)
-{
-	struct cpl_set_tcb_rpl *rpl = cplhdr(skb);
-
-	if (rpl->status != CPL_ERR_NONE)
-		printk(KERN_ERR
-		       "Unexpected SET_TCB_RPL status %u for tid %u\n",
-		       rpl->status, GET_TID(rpl));
-	return CPL_RET_BUF_DONE;
-}
-
 static int do_trace(struct t3cdev *dev, struct sk_buff *skb)
 {
 	struct cpl_trace_pkt *p = cplhdr(skb);
@@ -1213,7 +1204,8 @@ void __init cxgb3_offload_init(void)
 	t3_register_cpl_handler(CPL_CLOSE_CON_RPL, do_hwtid_rpl);
 	t3_register_cpl_handler(CPL_ABORT_REQ_RSS, do_abort_req_rss);
 	t3_register_cpl_handler(CPL_ACT_ESTABLISH, do_act_establish);
-	t3_register_cpl_handler(CPL_SET_TCB_RPL, do_set_tcb_rpl);
+	t3_register_cpl_handler(CPL_SET_TCB_RPL, do_hwtid_rpl);
+	t3_register_cpl_handler(CPL_GET_TCB_RPL, do_hwtid_rpl);
 	t3_register_cpl_handler(CPL_RDMA_TERMINATE, do_term);
 	t3_register_cpl_handler(CPL_RDMA_EC_STATUS, do_hwtid_rpl);
 	t3_register_cpl_handler(CPL_TRACE_PKT, do_trace);
diff --git a/drivers/net/cxgb3/mc5.c b/drivers/net/cxgb3/mc5.c
index 644d62ea86a..84c1ffa8e2d 100644
--- a/drivers/net/cxgb3/mc5.c
+++ b/drivers/net/cxgb3/mc5.c
@@ -328,6 +328,9 @@ int t3_mc5_init(struct mc5 *mc5, unsigned int nservers, unsigned int nfilters,
 	unsigned int tcam_size = mc5->tcam_size;
 	struct adapter *adap = mc5->adapter;
 
+	if (!tcam_size)
+		return 0;
+
 	if (nroutes > MAX_ROUTES || nroutes + nservers + nfilters > tcam_size)
 		return -EINVAL;
 
diff --git a/drivers/net/cxgb3/regs.h b/drivers/net/cxgb3/regs.h
index b56c5f52bcd..e5a553410e2 100644
--- a/drivers/net/cxgb3/regs.h
+++ b/drivers/net/cxgb3/regs.h
@@ -1206,6 +1206,14 @@
 
 #define A_TP_RX_TRC_KEY0 0x120
 
+#define A_TP_TX_DROP_CNT_CH0 0x12d
+
+#define S_TXDROPCNTCH0RCVD    0
+#define M_TXDROPCNTCH0RCVD    0xffff
+#define V_TXDROPCNTCH0RCVD(x) ((x) << S_TXDROPCNTCH0RCVD)
+#define G_TXDROPCNTCH0RCVD(x) (((x) >> S_TXDROPCNTCH0RCVD) & \
+			       M_TXDROPCNTCH0RCVD)
+
 #define A_ULPRX_CTL 0x500
 
 #define S_ROUND_ROBIN    4
@@ -1226,9 +1234,15 @@
 
 #define A_ULPRX_ISCSI_TAGMASK 0x514
 
+#define S_HPZ0    0
+#define M_HPZ0    0xf
+#define V_HPZ0(x) ((x) << S_HPZ0)
+#define G_HPZ0(x) (((x) >> S_HPZ0) & M_HPZ0)
+
 #define A_ULPRX_TDDP_LLIMIT 0x51c
 
 #define A_ULPRX_TDDP_ULIMIT 0x520
+#define A_ULPRX_TDDP_PSZ 0x528
 
 #define A_ULPRX_STAG_LLIMIT 0x52c
 
@@ -1834,6 +1848,8 @@
 #define V_TXPAUSEEN(x) ((x) << S_TXPAUSEEN)
 #define F_TXPAUSEEN    V_TXPAUSEEN(1U)
 
+#define A_XGM_TX_PAUSE_QUANTA 0x808
+
 #define A_XGM_RX_CTRL 0x80c
 
 #define S_RXEN    0
@@ -1920,11 +1936,20 @@
 
 #define A_XGM_TXFIFO_CFG 0x888
 
+#define S_TXIPG    13
+#define M_TXIPG    0xff
+#define V_TXIPG(x) ((x) << S_TXIPG)
+#define G_TXIPG(x) (((x) >> S_TXIPG) & M_TXIPG)
+
 #define S_TXFIFOTHRESH    4
 #define M_TXFIFOTHRESH    0x1ff
 
 #define V_TXFIFOTHRESH(x) ((x) << S_TXFIFOTHRESH)
 
+#define S_ENDROPPKT    21
+#define V_ENDROPPKT(x) ((x) << S_ENDROPPKT)
+#define F_ENDROPPKT    V_ENDROPPKT(1U)
+
 #define A_XGM_SERDES_CTRL 0x890
 #define A_XGM_SERDES_CTRL0 0x8e0
 
@@ -2190,6 +2215,13 @@
 
 #define A_XGM_RX_MAX_PKT_SIZE_ERR_CNT 0x9a4
 
+#define A_XGM_TX_SPI4_SOP_EOP_CNT 0x9a8
+
+#define S_TXSPI4SOPCNT    16
+#define M_TXSPI4SOPCNT    0xffff
+#define V_TXSPI4SOPCNT(x) ((x) << S_TXSPI4SOPCNT)
+#define G_TXSPI4SOPCNT(x) (((x) >> S_TXSPI4SOPCNT) & M_TXSPI4SOPCNT)
+
 #define A_XGM_RX_SPI4_SOP_EOP_CNT 0x9ac
 
 #define XGMAC0_1_BASE_ADDR 0xa00
diff --git a/drivers/net/cxgb3/sge.c b/drivers/net/cxgb3/sge.c
index c23783432e5..027ab2c3825 100644
--- a/drivers/net/cxgb3/sge.c
+++ b/drivers/net/cxgb3/sge.c
@@ -2631,7 +2631,7 @@ int t3_sge_alloc_qset(struct adapter *adapter, unsigned int id, int nports,
 	q->txq[TXQ_ETH].stop_thres = nports *
 	    flits_to_desc(sgl_len(MAX_SKB_FRAGS + 1) + 3);
 
-	if (ntxq == 1) {
+	if (!is_offload(adapter)) {
 #ifdef USE_RX_PAGE
 		q->fl[0].buf_size = RX_PAGE_SIZE;
 #else
diff --git a/drivers/net/cxgb3/t3_hw.c b/drivers/net/cxgb3/t3_hw.c
index eaa7a2e89a3..d83f075ef2d 100644
--- a/drivers/net/cxgb3/t3_hw.c
+++ b/drivers/net/cxgb3/t3_hw.c
@@ -438,23 +438,23 @@ static const struct adapter_info t3_adap_info[] = {
 	{2, 0, 0, 0,
 	 F_GPIO2_OEN | F_GPIO4_OEN |
 	 F_GPIO2_OUT_VAL | F_GPIO4_OUT_VAL, F_GPIO3 | F_GPIO5,
-	 SUPPORTED_OFFLOAD,
+	 0,
 	 &mi1_mdio_ops, "Chelsio PE9000"},
 	{2, 0, 0, 0,
 	 F_GPIO2_OEN | F_GPIO4_OEN |
 	 F_GPIO2_OUT_VAL | F_GPIO4_OUT_VAL, F_GPIO3 | F_GPIO5,
-	 SUPPORTED_OFFLOAD,
+	 0,
 	 &mi1_mdio_ops, "Chelsio T302"},
 	{1, 0, 0, 0,
 	 F_GPIO1_OEN | F_GPIO6_OEN | F_GPIO7_OEN | F_GPIO10_OEN |
 	 F_GPIO1_OUT_VAL | F_GPIO6_OUT_VAL | F_GPIO10_OUT_VAL, 0,
-	 SUPPORTED_10000baseT_Full | SUPPORTED_AUI | SUPPORTED_OFFLOAD,
+	 SUPPORTED_10000baseT_Full | SUPPORTED_AUI,
 	 &mi1_mdio_ext_ops, "Chelsio T310"},
 	{2, 0, 0, 0,
 	 F_GPIO1_OEN | F_GPIO2_OEN | F_GPIO4_OEN | F_GPIO5_OEN | F_GPIO6_OEN |
 	 F_GPIO7_OEN | F_GPIO10_OEN | F_GPIO11_OEN | F_GPIO1_OUT_VAL |
 	 F_GPIO5_OUT_VAL | F_GPIO6_OUT_VAL | F_GPIO10_OUT_VAL, 0,
-	 SUPPORTED_10000baseT_Full | SUPPORTED_AUI | SUPPORTED_OFFLOAD,
+	 SUPPORTED_10000baseT_Full | SUPPORTED_AUI,
 	 &mi1_mdio_ext_ops, "Chelsio T320"},
 };
 
@@ -681,7 +681,8 @@ enum {
 	SF_ERASE_SECTOR = 0xd8,	/* erase sector */
 
 	FW_FLASH_BOOT_ADDR = 0x70000,	/* start address of FW in flash */
-	FW_VERS_ADDR = 0x77ffc	/* flash address holding FW version */
+	FW_VERS_ADDR = 0x77ffc,    /* flash address holding FW version */
+	FW_MIN_SIZE = 8            /* at least version and csum */
 };
 
 /**
@@ -935,7 +936,7 @@ int t3_load_fw(struct adapter *adapter, const u8 *fw_data, unsigned int size)
 	const u32 *p = (const u32 *)fw_data;
 	int ret, addr, fw_sector = FW_FLASH_BOOT_ADDR >> 16;
 
-	if (size & 3)
+	if ((size & 3) || size < FW_MIN_SIZE)
 		return -EINVAL;
 	if (size > FW_VERS_ADDR + 8 - FW_FLASH_BOOT_ADDR)
 		return -EFBIG;
@@ -2899,6 +2900,9 @@ static int mc7_init(struct mc7 *mc7, unsigned int mc7_clock, int mem_type)
 	struct adapter *adapter = mc7->adapter;
 	const struct mc7_timing_params *p = &mc7_timings[mem_type];
 
+	if (!mc7->size)
+		return 0;
+
 	val = t3_read_reg(adapter, mc7->offset + A_MC7_CFG);
 	slow = val & F_SLOW;
 	width = G_WIDTH(val);
@@ -3099,8 +3103,10 @@ int t3_init_hw(struct adapter *adapter, u32 fw_params)
 	do {			/* wait for uP to initialize */
 		msleep(20);
 	} while (t3_read_reg(adapter, A_CIM_HOST_ACC_DATA) && --attempts);
-	if (!attempts)
+	if (!attempts) {
+		CH_ERR(adapter, "uP initialization timed out\n");
 		goto out_err;
+	}
 
 	err = 0;
 out_err:
@@ -3200,7 +3206,7 @@ static void __devinit mc7_prep(struct adapter *adapter, struct mc7 *mc7,
 	mc7->name = name;
 	mc7->offset = base_addr - MC7_PMRX_BASE_ADDR;
 	cfg = t3_read_reg(adapter, mc7->offset + A_MC7_CFG);
-	mc7->size = mc7_calc_size(cfg);
+	mc7->size = mc7->size = G_DEN(cfg) == M_DEN ? 0 : mc7_calc_size(cfg);
 	mc7->width = G_WIDTH(cfg);
 }
 
@@ -3227,6 +3233,7 @@ void early_hw_init(struct adapter *adapter, const struct adapter_info *ai)
 		     V_I2C_CLKDIV(adapter->params.vpd.cclk / 80 - 1));
 	t3_write_reg(adapter, A_T3DBG_GPIO_EN,
 		     ai->gpio_out | F_GPIO0_OEN | F_GPIO0_OUT_VAL);
+	t3_write_reg(adapter, A_MC5_DB_SERVER_INDEX, 0);
 
 	if (adapter->params.rev == 0 || !uses_xaui(adapter))
 		val |= F_ENRGMII;
@@ -3243,15 +3250,17 @@ void early_hw_init(struct adapter *adapter, const struct adapter_info *ai)
 }
 
 /*
- * Reset the adapter.  PCIe cards lose their config space during reset, PCI-X
+ * Reset the adapter. 
+ * Older PCIe cards lose their config space during reset, PCI-X
  * ones don't.
  */
 int t3_reset_adapter(struct adapter *adapter)
 {
-	int i;
+	int i, save_and_restore_pcie = 
+	    adapter->params.rev < T3_REV_B2 && is_pcie(adapter);
 	uint16_t devid = 0;
 
-	if (is_pcie(adapter))
+	if (save_and_restore_pcie)
 		pci_save_state(adapter->pdev);
 	t3_write_reg(adapter, A_PL_RST, F_CRSTWRM | F_CRSTWRMMODE);
 
@@ -3269,7 +3278,7 @@ int t3_reset_adapter(struct adapter *adapter)
 	if (devid != 0x1425)
 		return -1;
 
-	if (is_pcie(adapter))
+	if (save_and_restore_pcie)
 		pci_restore_state(adapter->pdev);
 	return 0;
 }
@@ -3323,7 +3332,13 @@ int __devinit t3_prep_adapter(struct adapter *adapter,
 		p->tx_num_pgs = pm_num_pages(p->chan_tx_size, p->tx_pg_size);
 		p->ntimer_qs = p->cm_size >= (128 << 20) ||
 		    adapter->params.rev > 0 ? 12 : 6;
+	}
+
+	adapter->params.offload = t3_mc7_size(&adapter->pmrx) &&
+				  t3_mc7_size(&adapter->pmtx) &&
+				  t3_mc7_size(&adapter->cm);
 
+	if (is_offload(adapter)) {
 		adapter->params.mc5.nservers = DEFAULT_NSERVERS;
 		adapter->params.mc5.nfilters = adapter->params.rev > 0 ?
 		    DEFAULT_NFILTERS : 0;
diff --git a/drivers/net/cxgb3/version.h b/drivers/net/cxgb3/version.h
index 82278f85025..042e27e291c 100644
--- a/drivers/net/cxgb3/version.h
+++ b/drivers/net/cxgb3/version.h
@@ -36,6 +36,9 @@
 #define DRV_NAME "cxgb3"
 /* Driver version */
 #define DRV_VERSION "1.0-ko"
+
+/* Firmware version */
 #define FW_VERSION_MAJOR 3
-#define FW_VERSION_MINOR 2
+#define FW_VERSION_MINOR 3
+#define FW_VERSION_MICRO 0
 #endif				/* __CHELSIO_VERSION_H */
diff --git a/drivers/net/cxgb3/xgmac.c b/drivers/net/cxgb3/xgmac.c
index 907a272ae32..a506792f957 100644
--- a/drivers/net/cxgb3/xgmac.c
+++ b/drivers/net/cxgb3/xgmac.c
@@ -124,9 +124,6 @@ int t3_mac_reset(struct cmac *mac)
 			xaui_serdes_reset(mac);
 	}
 
-	if (adap->params.rev > 0)
-		t3_write_reg(adap, A_XGM_PAUSE_TIMER + oft, 0xf000);
-
 	val = F_MAC_RESET_;
 	if (is_10G(adap))
 		val |= F_PCS_RESET_;
@@ -145,6 +142,58 @@ int t3_mac_reset(struct cmac *mac)
 	return 0;
 }
 
+int t3b2_mac_reset(struct cmac *mac)
+{
+	struct adapter *adap = mac->adapter;
+	unsigned int oft = mac->offset;
+	u32 val;
+
+	if (!macidx(mac)) 
+		t3_set_reg_field(adap, A_MPS_CFG, F_PORT0ACTIVE, 0);
+	else
+		t3_set_reg_field(adap, A_MPS_CFG, F_PORT1ACTIVE, 0);
+
+	t3_write_reg(adap, A_XGM_RESET_CTRL + oft, F_MAC_RESET_);
+	t3_read_reg(adap, A_XGM_RESET_CTRL + oft);    /* flush */
+
+	msleep(10);
+
+	/* Check for xgm Rx fifo empty */
+	if (t3_wait_op_done(adap, A_XGM_RX_MAX_PKT_SIZE_ERR_CNT + oft,
+			    0x80000000, 1, 5, 2)) {
+		CH_ERR(adap, "MAC %d Rx fifo drain failed\n",
+		       macidx(mac));
+		return -1;
+	}
+
+	t3_write_reg(adap, A_XGM_RESET_CTRL + oft, 0);
+	t3_read_reg(adap, A_XGM_RESET_CTRL + oft);    /* flush */
+
+	val = F_MAC_RESET_;
+	if (is_10G(adap))
+		val |= F_PCS_RESET_;
+	else if (uses_xaui(adap))
+		val |= F_PCS_RESET_ | F_XG2G_RESET_;
+	else
+		val |= F_RGMII_RESET_ | F_XG2G_RESET_;
+	t3_write_reg(adap, A_XGM_RESET_CTRL + oft, val);
+	t3_read_reg(adap, A_XGM_RESET_CTRL + oft);  /* flush */
+	if ((val & F_PCS_RESET_) && adap->params.rev) {
+		msleep(1);
+		t3b_pcs_reset(mac);
+	}
+	t3_write_reg(adap, A_XGM_RX_CFG + oft, 
+		     F_DISPAUSEFRAMES | F_EN1536BFRAMES |
+		     F_RMFCS | F_ENJUMBO | F_ENHASHMCAST);
+
+	if (!macidx(mac)) 
+		t3_set_reg_field(adap, A_MPS_CFG, 0, F_PORT0ACTIVE);
+	else
+		t3_set_reg_field(adap, A_MPS_CFG, 0, F_PORT1ACTIVE);
+
+	return 0;
+}
+
 /*
  * Set the exact match register 'idx' to recognize the given Ethernet address.
  */
@@ -251,9 +300,11 @@ int t3_mac_set_mtu(struct cmac *mac, unsigned int mtu)
 	 * Adjust the PAUSE frame watermarks.  We always set the LWM, and the
 	 * HWM only if flow-control is enabled.
 	 */
-	hwm = max(MAC_RXFIFO_SIZE - 3 * mtu, MAC_RXFIFO_SIZE / 2U);
-	hwm = min(hwm, 3 * MAC_RXFIFO_SIZE / 4 + 1024);
-	lwm = hwm - 1024;
+	hwm = max_t(unsigned int, MAC_RXFIFO_SIZE - 3 * mtu, 
+		    MAC_RXFIFO_SIZE * 38 / 100);
+	hwm = min(hwm, MAC_RXFIFO_SIZE - 8192);
+	lwm = min(3 * (int)mtu, MAC_RXFIFO_SIZE / 4);
+
 	v = t3_read_reg(adap, A_XGM_RXFIFO_CFG + mac->offset);
 	v &= ~V_RXFIFOPAUSELWM(M_RXFIFOPAUSELWM);
 	v |= V_RXFIFOPAUSELWM(lwm / 8);
@@ -270,7 +321,15 @@ int t3_mac_set_mtu(struct cmac *mac, unsigned int mtu)
 	thres = mtu > thres ? (mtu - thres + 7) / 8 : 0;
 	thres = max(thres, 8U);	/* need at least 8 */
 	t3_set_reg_field(adap, A_XGM_TXFIFO_CFG + mac->offset,
-			 V_TXFIFOTHRESH(M_TXFIFOTHRESH), V_TXFIFOTHRESH(thres));
+			 V_TXFIFOTHRESH(M_TXFIFOTHRESH) | V_TXIPG(M_TXIPG),
+			 V_TXFIFOTHRESH(thres) | V_TXIPG(1));
+
+	if (adap->params.rev > 0)
+		t3_write_reg(adap, A_XGM_PAUSE_TIMER + mac->offset,
+			     (hwm - lwm) * 4 / 8);
+	t3_write_reg(adap, A_XGM_TX_PAUSE_QUANTA + mac->offset,
+		     MAC_RXFIFO_SIZE * 4 * 8 / 512);
+
 	return 0;
 }
 
@@ -298,12 +357,6 @@ int t3_mac_set_speed_duplex_fc(struct cmac *mac, int speed, int duplex, int fc)
 				 V_PORTSPEED(M_PORTSPEED), val);
 	}
 
-	val = t3_read_reg(adap, A_XGM_RXFIFO_CFG + oft);
-	val &= ~V_RXFIFOPAUSEHWM(M_RXFIFOPAUSEHWM);
-	if (fc & PAUSE_TX)
-		val |= V_RXFIFOPAUSEHWM(G_RXFIFOPAUSELWM(val) + 128);	/* +1KB */
-	t3_write_reg(adap, A_XGM_RXFIFO_CFG + oft, val);
-
 	t3_set_reg_field(adap, A_XGM_TX_CFG + oft, F_TXPAUSEEN,
 			 (fc & PAUSE_RX) ? F_TXPAUSEEN : 0);
 	return 0;
@@ -314,13 +367,28 @@ int t3_mac_enable(struct cmac *mac, int which)
 	int idx = macidx(mac);
 	struct adapter *adap = mac->adapter;
 	unsigned int oft = mac->offset;
-
+	struct mac_stats *s = &mac->stats;
+	
 	if (which & MAC_DIRECTION_TX) {
 		t3_write_reg(adap, A_XGM_TX_CTRL + oft, F_TXEN);
 		t3_write_reg(adap, A_TP_PIO_ADDR, A_TP_TX_DROP_CFG_CH0 + idx);
-		t3_write_reg(adap, A_TP_PIO_DATA, 0xbf000001);
+		t3_write_reg(adap, A_TP_PIO_DATA, 0xc0ede401);
 		t3_write_reg(adap, A_TP_PIO_ADDR, A_TP_TX_DROP_MODE);
 		t3_set_reg_field(adap, A_TP_PIO_DATA, 1 << idx, 1 << idx);
+
+		t3_write_reg(adap, A_TP_PIO_ADDR, A_TP_TX_DROP_CNT_CH0 + idx);
+		mac->tx_mcnt = s->tx_frames;
+		mac->tx_tcnt = (G_TXDROPCNTCH0RCVD(t3_read_reg(adap,
+							A_TP_PIO_DATA)));
+		mac->tx_xcnt = (G_TXSPI4SOPCNT(t3_read_reg(adap,
+						A_XGM_TX_SPI4_SOP_EOP_CNT +
+						oft)));
+		mac->rx_mcnt = s->rx_frames;
+		mac->rx_xcnt = (G_TXSPI4SOPCNT(t3_read_reg(adap,
+						A_XGM_RX_SPI4_SOP_EOP_CNT +
+						oft)));
+		mac->txen = F_TXEN;
+		mac->toggle_cnt = 0;
 	}
 	if (which & MAC_DIRECTION_RX)
 		t3_write_reg(adap, A_XGM_RX_CTRL + oft, F_RXEN);
@@ -331,19 +399,102 @@ int t3_mac_disable(struct cmac *mac, int which)
 {
 	int idx = macidx(mac);
 	struct adapter *adap = mac->adapter;
+	int val;
 
 	if (which & MAC_DIRECTION_TX) {
 		t3_write_reg(adap, A_XGM_TX_CTRL + mac->offset, 0);
 		t3_write_reg(adap, A_TP_PIO_ADDR, A_TP_TX_DROP_CFG_CH0 + idx);
 		t3_write_reg(adap, A_TP_PIO_DATA, 0xc000001f);
 		t3_write_reg(adap, A_TP_PIO_ADDR, A_TP_TX_DROP_MODE);
-		t3_set_reg_field(adap, A_TP_PIO_DATA, 1 << idx, 0);
+		t3_set_reg_field(adap, A_TP_PIO_DATA, 1 << idx, 1 << idx);
+		mac->txen = 0;
 	}
-	if (which & MAC_DIRECTION_RX)
+	if (which & MAC_DIRECTION_RX) {
+		t3_set_reg_field(mac->adapter, A_XGM_RESET_CTRL + mac->offset,
+				 F_PCS_RESET_, 0);
+		msleep(100);
 		t3_write_reg(adap, A_XGM_RX_CTRL + mac->offset, 0);
+		val = F_MAC_RESET_;
+		if (is_10G(adap))
+			val |= F_PCS_RESET_;
+		else if (uses_xaui(adap))
+			val |= F_PCS_RESET_ | F_XG2G_RESET_;
+		else
+			val |= F_RGMII_RESET_ | F_XG2G_RESET_;
+		t3_write_reg(mac->adapter, A_XGM_RESET_CTRL + mac->offset, val);
+	}
 	return 0;
 }
 
+int t3b2_mac_watchdog_task(struct cmac *mac)
+{
+	struct adapter *adap = mac->adapter;
+	struct mac_stats *s = &mac->stats;
+	unsigned int tx_tcnt, tx_xcnt;
+	unsigned int tx_mcnt = s->tx_frames;
+	unsigned int rx_mcnt = s->rx_frames;
+	unsigned int rx_xcnt;
+	int status;
+
+	if (tx_mcnt == mac->tx_mcnt) {
+		tx_xcnt = (G_TXSPI4SOPCNT(t3_read_reg(adap,
+						A_XGM_TX_SPI4_SOP_EOP_CNT +
+					       	mac->offset)));
+		if (tx_xcnt == 0) {
+			t3_write_reg(adap, A_TP_PIO_ADDR,
+				     A_TP_TX_DROP_CNT_CH0 + macidx(mac));
+			tx_tcnt = (G_TXDROPCNTCH0RCVD(t3_read_reg(adap,
+						      A_TP_PIO_DATA)));
+		} else {
+			mac->toggle_cnt = 0;
+			return 0;
+		}
+	} else {
+		mac->toggle_cnt = 0;
+		return 0;
+	}
+
+	if (((tx_tcnt != mac->tx_tcnt) &&
+	     (tx_xcnt == 0) && (mac->tx_xcnt == 0)) ||
+	    ((mac->tx_mcnt == tx_mcnt) &&
+	     (tx_xcnt != 0) && (mac->tx_xcnt != 0))) {
+		if (mac->toggle_cnt > 4)
+			status = 2;
+		else 
+			status = 1;
+	} else {
+		mac->toggle_cnt = 0;
+		return 0;
+	}
+
+	if (rx_mcnt != mac->rx_mcnt)
+		rx_xcnt = (G_TXSPI4SOPCNT(t3_read_reg(adap,
+						A_XGM_RX_SPI4_SOP_EOP_CNT +
+						mac->offset)));
+	else 
+		return 0;
+
+	if (mac->rx_mcnt != s->rx_frames && rx_xcnt == 0 && mac->rx_xcnt == 0) 
+		status = 2;
+	
+	mac->tx_tcnt = tx_tcnt;
+	mac->tx_xcnt = tx_xcnt;
+	mac->tx_mcnt = s->tx_frames;
+	mac->rx_xcnt = rx_xcnt;
+	mac->rx_mcnt = s->rx_frames;
+	if (status == 1) {
+		t3_write_reg(adap, A_XGM_TX_CTRL + mac->offset, 0);
+		t3_read_reg(adap, A_XGM_TX_CTRL + mac->offset);  /* flush */
+		t3_write_reg(adap, A_XGM_TX_CTRL + mac->offset, mac->txen);
+		t3_read_reg(adap, A_XGM_TX_CTRL + mac->offset);  /* flush */
+		mac->toggle_cnt++;
+	} else if (status == 2) {
+		t3b2_mac_reset(mac);
+		mac->toggle_cnt = 0;
+	}
+	return status;
+}
+
 /*
  * This function is called periodically to accumulate the current values of the
  * RMON counters into the port statistics.  Since the packet counters are only
@@ -373,7 +524,11 @@ const struct mac_stats *t3_mac_update_stats(struct cmac *mac)
 	RMON_UPDATE(mac, rx_symbol_errs, RX_SYM_CODE_ERR_FRAMES);
 
 	RMON_UPDATE(mac, rx_too_long, RX_OVERSIZE_FRAMES);
-	mac->stats.rx_too_long += RMON_READ(mac, A_XGM_RX_MAX_PKT_SIZE_ERR_CNT);
+
+	v = RMON_READ(mac, A_XGM_RX_MAX_PKT_SIZE_ERR_CNT);
+	if (mac->adapter->params.rev == T3_REV_B2)
+		v &= 0x7fffffff;
+	mac->stats.rx_too_long += v;
 
 	RMON_UPDATE(mac, rx_frames_64, RX_64B_FRAMES);
 	RMON_UPDATE(mac, rx_frames_65_127, RX_65_127B_FRAMES);