diff options
Diffstat (limited to 'drivers/net/cxgb3/cxgb3_main.c')
| -rw-r--r-- | drivers/net/cxgb3/cxgb3_main.c | 2515 |
1 files changed, 2515 insertions, 0 deletions
diff --git a/drivers/net/cxgb3/cxgb3_main.c b/drivers/net/cxgb3/cxgb3_main.c new file mode 100644 index 00000000000..dfa035a1ad4 --- /dev/null +++ b/drivers/net/cxgb3/cxgb3_main.c @@ -0,0 +1,2515 @@ +/* + * Copyright (c) 2003-2007 Chelsio, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ +#include <linux/module.h> +#include <linux/moduleparam.h> +#include <linux/init.h> +#include <linux/pci.h> +#include <linux/dma-mapping.h> +#include <linux/netdevice.h> +#include <linux/etherdevice.h> +#include <linux/if_vlan.h> +#include <linux/mii.h> +#include <linux/sockios.h> +#include <linux/workqueue.h> +#include <linux/proc_fs.h> +#include <linux/rtnetlink.h> +#include <asm/uaccess.h> + +#include "common.h" +#include "cxgb3_ioctl.h" +#include "regs.h" +#include "cxgb3_offload.h" +#include "version.h" + +#include "cxgb3_ctl_defs.h" +#include "t3_cpl.h" +#include "firmware_exports.h" + +enum { + MAX_TXQ_ENTRIES = 16384, + MAX_CTRL_TXQ_ENTRIES = 1024, + MAX_RSPQ_ENTRIES = 16384, + MAX_RX_BUFFERS = 16384, + MAX_RX_JUMBO_BUFFERS = 16384, + MIN_TXQ_ENTRIES = 4, + MIN_CTRL_TXQ_ENTRIES = 4, + MIN_RSPQ_ENTRIES = 32, + MIN_FL_ENTRIES = 32 +}; + +#define PORT_MASK ((1 << MAX_NPORTS) - 1) + +#define DFLT_MSG_ENABLE (NETIF_MSG_DRV | NETIF_MSG_PROBE | NETIF_MSG_LINK | \ + NETIF_MSG_TIMER | NETIF_MSG_IFDOWN | NETIF_MSG_IFUP |\ + NETIF_MSG_RX_ERR | NETIF_MSG_TX_ERR) + +#define EEPROM_MAGIC 0x38E2F10C + +#define to_net_dev(class) container_of(class, struct net_device, class_dev) + +#define CH_DEVICE(devid, ssid, idx) \ + { PCI_VENDOR_ID_CHELSIO, devid, PCI_ANY_ID, ssid, 0, 0, idx } + +static const struct pci_device_id cxgb3_pci_tbl[] = { + CH_DEVICE(0x20, 1, 0), /* PE9000 */ + CH_DEVICE(0x21, 1, 1), /* T302E */ + CH_DEVICE(0x22, 1, 2), /* T310E */ + CH_DEVICE(0x23, 1, 3), /* T320X */ + CH_DEVICE(0x24, 1, 1), /* T302X */ + CH_DEVICE(0x25, 1, 3), /* T320E */ + CH_DEVICE(0x26, 1, 2), /* T310X */ + CH_DEVICE(0x30, 1, 2), /* T3B10 */ + CH_DEVICE(0x31, 1, 3), /* T3B20 */ + CH_DEVICE(0x32, 1, 1), /* T3B02 */ + {0,} +}; + +MODULE_DESCRIPTION(DRV_DESC); +MODULE_AUTHOR("Chelsio Communications"); +MODULE_LICENSE("Dual BSD/GPL"); +MODULE_VERSION(DRV_VERSION); +MODULE_DEVICE_TABLE(pci, cxgb3_pci_tbl); + +static int dflt_msg_enable = DFLT_MSG_ENABLE; + +module_param(dflt_msg_enable, int, 0644); +MODULE_PARM_DESC(dflt_msg_enable, "Chelsio T3 default message enable bitmap"); + +/* + * The driver uses the best interrupt scheme available on a platform in the + * order MSI-X, MSI, legacy pin interrupts. This parameter determines which + * of these schemes the driver may consider as follows: + * + * msi = 2: choose from among all three options + * msi = 1: only consider MSI and pin interrupts + * msi = 0: force pin interrupts + */ +static int msi = 2; + +module_param(msi, int, 0644); +MODULE_PARM_DESC(msi, "whether to use MSI or MSI-X"); + +/* + * The driver enables offload as a default. + * To disable it, use ofld_disable = 1. + */ + +static int ofld_disable = 0; + +module_param(ofld_disable, int, 0644); +MODULE_PARM_DESC(ofld_disable, "whether to enable offload at init time or not"); + +/* + * We have work elements that we need to cancel when an interface is taken + * down. Normally the work elements would be executed by keventd but that + * can deadlock because of linkwatch. If our close method takes the rtnl + * lock and linkwatch is ahead of our work elements in keventd, linkwatch + * will block keventd as it needs the rtnl lock, and we'll deadlock waiting + * for our work to complete. Get our own work queue to solve this. + */ +static struct workqueue_struct *cxgb3_wq; + +/** + * link_report - show link status and link speed/duplex + * @p: the port whose settings are to be reported + * + * Shows the link status, speed, and duplex of a port. + */ +static void link_report(struct net_device *dev) +{ + if (!netif_carrier_ok(dev)) + printk(KERN_INFO "%s: link down\n", dev->name); + else { + const char *s = "10Mbps"; + const struct port_info *p = netdev_priv(dev); + + switch (p->link_config.speed) { + case SPEED_10000: + s = "10Gbps"; + break; + case SPEED_1000: + s = "1000Mbps"; + break; + case SPEED_100: + s = "100Mbps"; + break; + } + + printk(KERN_INFO "%s: link up, %s, %s-duplex\n", dev->name, s, + p->link_config.duplex == DUPLEX_FULL ? "full" : "half"); + } +} + +/** + * t3_os_link_changed - handle link status changes + * @adapter: the adapter associated with the link change + * @port_id: the port index whose limk status has changed + * @link_stat: the new status of the link + * @speed: the new speed setting + * @duplex: the new duplex setting + * @pause: the new flow-control setting + * + * This is the OS-dependent handler for link status changes. The OS + * neutral handler takes care of most of the processing for these events, + * then calls this handler for any OS-specific processing. + */ +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]; + + /* Skip changes from disabled ports. */ + if (!netif_running(dev)) + return; + + if (link_stat != netif_carrier_ok(dev)) { + if (link_stat) + netif_carrier_on(dev); + else + netif_carrier_off(dev); + link_report(dev); + } +} + +static void cxgb_set_rxmode(struct net_device *dev) +{ + struct t3_rx_mode rm; + struct port_info *pi = netdev_priv(dev); + + init_rx_mode(&rm, dev, dev->mc_list); + t3_mac_set_rx_mode(&pi->mac, &rm); +} + +/** + * link_start - enable a port + * @dev: the device to enable + * + * Performs the MAC and PHY actions needed to enable a port. + */ +static void link_start(struct net_device *dev) +{ + struct t3_rx_mode rm; + struct port_info *pi = netdev_priv(dev); + struct cmac *mac = &pi->mac; + + init_rx_mode(&rm, dev, dev->mc_list); + t3_mac_reset(mac); + t3_mac_set_mtu(mac, dev->mtu); + t3_mac_set_address(mac, 0, dev->dev_addr); + t3_mac_set_rx_mode(mac, &rm); + t3_link_start(&pi->phy, mac, &pi->link_config); + t3_mac_enable(mac, MAC_DIRECTION_RX | MAC_DIRECTION_TX); +} + +static inline void cxgb_disable_msi(struct adapter *adapter) +{ + if (adapter->flags & USING_MSIX) { + pci_disable_msix(adapter->pdev); + adapter->flags &= ~USING_MSIX; + } else if (adapter->flags & USING_MSI) { + pci_disable_msi(adapter->pdev); + adapter->flags &= ~USING_MSI; + } +} + +/* + * Interrupt handler for asynchronous events used with MSI-X. + */ +static irqreturn_t t3_async_intr_handler(int irq, void *cookie) +{ + t3_slow_intr_handler(cookie); + return IRQ_HANDLED; +} + +/* + * Name the MSI-X interrupts. + */ +static void name_msix_vecs(struct adapter *adap) +{ + int i, j, msi_idx = 1, n = sizeof(adap->msix_info[0].desc) - 1; + + snprintf(adap->msix_info[0].desc, n, "%s", adap->name); + adap->msix_info[0].desc[n] = 0; + + for_each_port(adap, j) { + struct net_device *d = adap->port[j]; + const struct port_info *pi = netdev_priv(d); + + for (i = 0; i < pi->nqsets; i++, msi_idx++) { + snprintf(adap->msix_info[msi_idx].desc, n, + "%s (queue %d)", d->name, i); + adap->msix_info[msi_idx].desc[n] = 0; + } + } +} + +static int request_msix_data_irqs(struct adapter *adap) +{ + int i, j, err, qidx = 0; + + for_each_port(adap, i) { + int nqsets = adap2pinfo(adap, i)->nqsets; + + for (j = 0; j < nqsets; ++j) { + err = request_irq(adap->msix_info[qidx + 1].vec, + t3_intr_handler(adap, + adap->sge.qs[qidx]. + rspq.polling), 0, + adap->msix_info[qidx + 1].desc, + &adap->sge.qs[qidx]); + if (err) { + while (--qidx >= 0) + free_irq(adap->msix_info[qidx + 1].vec, + &adap->sge.qs[qidx]); + return err; + } + qidx++; + } + } + return 0; +} + +/** + * setup_rss - configure RSS + * @adap: the adapter + * + * Sets up RSS to distribute packets to multiple receive queues. We + * configure the RSS CPU lookup table to distribute to the number of HW + * receive queues, and the response queue lookup table to narrow that + * down to the response queues actually configured for each port. + * We always configure the RSS mapping for two ports since the mapping + * table has plenty of entries. + */ +static void setup_rss(struct adapter *adap) +{ + int i; + unsigned int nq0 = adap2pinfo(adap, 0)->nqsets; + unsigned int nq1 = adap->port[1] ? adap2pinfo(adap, 1)->nqsets : 1; + u8 cpus[SGE_QSETS + 1]; + u16 rspq_map[RSS_TABLE_SIZE]; + + for (i = 0; i < SGE_QSETS; ++i) + cpus[i] = i; + cpus[SGE_QSETS] = 0xff; /* terminator */ + + for (i = 0; i < RSS_TABLE_SIZE / 2; ++i) { + rspq_map[i] = i % nq0; + rspq_map[i + RSS_TABLE_SIZE / 2] = (i % nq1) + nq0; + } + + t3_config_rss(adap, F_RQFEEDBACKENABLE | F_TNLLKPEN | F_TNLMAPEN | + F_TNLPRTEN | F_TNL2TUPEN | F_TNL4TUPEN | + V_RRCPLCPUSIZE(6), cpus, rspq_map); +} + +/* + * If we have multiple receive queues per port serviced by NAPI we need one + * netdevice per queue as NAPI operates on netdevices. We already have one + * netdevice, namely the one associated with the interface, so we use dummy + * ones for any additional queues. Note that these netdevices exist purely + * so that NAPI has something to work with, they do not represent network + * ports and are not registered. + */ +static int init_dummy_netdevs(struct adapter *adap) +{ + int i, j, dummy_idx = 0; + struct net_device *nd; + + for_each_port(adap, i) { + struct net_device *dev = adap->port[i]; + const struct port_info *pi = netdev_priv(dev); + + for (j = 0; j < pi->nqsets - 1; j++) { + if (!adap->dummy_netdev[dummy_idx]) { + nd = alloc_netdev(0, "", ether_setup); + if (!nd) + goto free_all; + + nd->priv = adap; + nd->weight = 64; + set_bit(__LINK_STATE_START, &nd->state); + adap->dummy_netdev[dummy_idx] = nd; + } + strcpy(adap->dummy_netdev[dummy_idx]->name, dev->name); + dummy_idx++; + } + } + return 0; + +free_all: + while (--dummy_idx >= 0) { + free_netdev(adap->dummy_netdev[dummy_idx]); + adap->dummy_netdev[dummy_idx] = NULL; + } + return -ENOMEM; +} + +/* + * Wait until all NAPI handlers are descheduled. This includes the handlers of + * both netdevices representing interfaces and the dummy ones for the extra + * queues. + */ +static void quiesce_rx(struct adapter *adap) +{ + int i; + struct net_device *dev; + + for_each_port(adap, i) { + dev = adap->port[i]; + while (test_bit(__LINK_STATE_RX_SCHED, &dev->state)) + msleep(1); + } + + for (i = 0; i < ARRAY_SIZE(adap->dummy_netdev); i++) { + dev = adap->dummy_netdev[i]; + if (dev) + while (test_bit(__LINK_STATE_RX_SCHED, &dev->state)) + msleep(1); + } +} + +/** + * setup_sge_qsets - configure SGE Tx/Rx/response queues + * @adap: the adapter + * + * Determines how many sets of SGE queues to use and initializes them. + * We support multiple queue sets per port if we have MSI-X, otherwise + * just one queue set per port. + */ +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; + + if (adap->params.rev > 0 && !(adap->flags & USING_MSI)) + irq_idx = -1; + + for_each_port(adap, i) { + struct net_device *dev = adap->port[i]; + const struct port_info *pi = netdev_priv(dev); + + for (j = 0; j < pi->nqsets; ++j, ++qset_idx) { + err = t3_sge_alloc_qset(adap, qset_idx, 1, + (adap->flags & USING_MSIX) ? qset_idx + 1 : + irq_idx, + &adap->params.sge.qset[qset_idx], ntxq, + j == 0 ? dev : + adap-> dummy_netdev[dummy_dev_idx++]); + if (err) { + t3_free_sge_resources(adap); + return err; + } + } + } + + return 0; +} + +static ssize_t attr_show(struct class_device *cd, char *buf, + ssize_t(*format) (struct adapter *, char *)) +{ + ssize_t len; + struct adapter *adap = to_net_dev(cd)->priv; + + /* Synchronize with ioctls that may shut down the device */ + rtnl_lock(); + len = (*format) (adap, buf); + rtnl_unlock(); + return len; +} + +static ssize_t attr_store(struct class_device *cd, const char *buf, size_t len, + ssize_t(*set) (struct adapter *, unsigned int), + unsigned int min_val, unsigned int max_val) +{ + char *endp; + ssize_t ret; + unsigned int val; + struct adapter *adap = to_net_dev(cd)->priv; + + if (!capable(CAP_NET_ADMIN)) + return -EPERM; + + val = simple_strtoul(buf, &endp, 0); + if (endp == buf || val < min_val || val > max_val) + return -EINVAL; + + rtnl_lock(); + ret = (*set) (adap, val); + if (!ret) + ret = len; + rtnl_unlock(); + return ret; +} + +#define CXGB3_SHOW(name, val_expr) \ +static ssize_t format_##name(struct adapter *adap, char *buf) \ +{ \ + return sprintf(buf, "%u\n", val_expr); \ +} \ +static ssize_t show_##name(struct class_device *cd, char *buf) \ +{ \ + return attr_show(cd, buf, format_##name); \ +} + +static ssize_t set_nfilters(struct adapter *adap, unsigned int val) +{ + 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) + return -EINVAL; + adap->params.mc5.nfilters = val; + return 0; +} + +static ssize_t store_nfilters(struct class_device *cd, const char *buf, + size_t len) +{ + return attr_store(cd, buf, len, set_nfilters, 0, ~0); +} + +static ssize_t set_nservers(struct adapter *adap, unsigned int val) +{ + if (adap->flags & FULL_INIT_DONE) + return -EBUSY; + if (val > t3_mc5_size(&adap->mc5) - adap->params.mc5.nfilters) + return -EINVAL; + adap->params.mc5.nservers = val; + return 0; +} + +static ssize_t store_nservers(struct class_device *cd, const char *buf, + size_t len) +{ + return attr_store(cd, buf, len, set_nservers, 0, ~0); +} + +#define CXGB3_ATTR_R(name, val_expr) \ +CXGB3_SHOW(name, val_expr) \ +static CLASS_DEVICE_ATTR(name, S_IRUGO, show_##name, NULL) + +#define CXGB3_ATTR_RW(name, val_expr, store_method) \ +CXGB3_SHOW(name, val_expr) \ +static CLASS_DEVICE_ATTR(name, S_IRUGO | S_IWUSR, show_##name, store_method) + +CXGB3_ATTR_R(cam_size, t3_mc5_size(&adap->mc5)); +CXGB3_ATTR_RW(nfilters, adap->params.mc5.nfilters, store_nfilters); +CXGB3_ATTR_RW(nservers, adap->params.mc5.nservers, store_nservers); + +static struct attribute *cxgb3_attrs[] = { + &class_device_attr_cam_size.attr, + &class_device_attr_nfilters.attr, + &class_device_attr_nservers.attr, + NULL +}; + +static struct attribute_group cxgb3_attr_group = {.attrs = cxgb3_attrs }; + +static ssize_t tm_attr_show(struct class_device *cd, char *buf, int sched) +{ + ssize_t len; + unsigned int v, addr, bpt, cpt; + struct adapter *adap = to_net_dev(cd)->priv; + + addr = A_TP_TX_MOD_Q1_Q0_RATE_LIMIT - sched / 2; + rtnl_lock(); + t3_write_reg(adap, A_TP_TM_PIO_ADDR, addr); + v = t3_read_reg(adap, A_TP_TM_PIO_DATA); + if (sched & 1) + v >>= 16; + bpt = (v >> 8) & 0xff; + cpt = v & 0xff; + if (!cpt) + len = sprintf(buf, "disabled\n"); + else { + v = (adap->params.vpd.cclk * 1000) / cpt; + len = sprintf(buf, "%u Kbps\n", (v * bpt) / 125); + } + rtnl_unlock(); + return len; +} + +static ssize_t tm_attr_store(struct class_device *cd, const char *buf, + size_t len, int sched) +{ + char *endp; + ssize_t ret; + unsigned int val; + struct adapter *adap = to_net_dev(cd)->priv; + + if (!capable(CAP_NET_ADMIN)) + return -EPERM; + + val = simple_strtoul(buf, &endp, 0); + if (endp == buf || val > 10000000) + return -EINVAL; + + rtnl_lock(); + ret = t3_config_sched(adap, val, sched); + if (!ret) + ret = len; + rtnl_unlock(); + return ret; +} + +#define TM_ATTR(name, sched) \ +static ssize_t show_##name(struct class_device *cd, char *buf) \ +{ \ + return tm_attr_show(cd, buf, sched); \ +} \ +static ssize_t store_##name(struct class_device *cd, const char *buf, size_t len) \ +{ \ + return tm_attr_store(cd, buf, len, sched); \ +} \ +static CLASS_DEVICE_ATTR(name, S_IRUGO | S_IWUSR, show_##name, store_##name) + +TM_ATTR(sched0, 0); +TM_ATTR(sched1, 1); +TM_ATTR(sched2, 2); +TM_ATTR(sched3, 3); +TM_ATTR(sched4, 4); +TM_ATTR(sched5, 5); +TM_ATTR(sched6, 6); +TM_ATTR(sched7, 7); + +static struct attribute *offload_attrs[] = { + &class_device_attr_sched0.attr, + &class_device_attr_sched1.attr, + &class_device_attr_sched2.attr, + &class_device_attr_sched3.attr, + &class_device_attr_sched4.attr, + &class_device_attr_sched5.attr, + &class_device_attr_sched6.attr, + &class_device_attr_sched7.attr, + NULL +}; + +static struct attribute_group offload_attr_group = {.attrs = offload_attrs }; + +/* + * Sends an sk_buff to an offload queue driver + * after dealing with any active network taps. + */ +static inline int offload_tx(struct t3cdev *tdev, struct sk_buff *skb) +{ + int ret; + + local_bh_disable(); + ret = t3_offload_tx(tdev, skb); + local_bh_enable(); + return ret; +} + +static int write_smt_entry(struct adapter *adapter, int idx) +{ + struct cpl_smt_write_req *req; + struct sk_buff *skb = alloc_skb(sizeof(*req), GFP_KERNEL); + + if (!skb) + return -ENOMEM; + + req = (struct cpl_smt_write_req *)__skb_put(skb, sizeof(*req)); + req->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_FORWARD)); + OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_SMT_WRITE_REQ, idx)); + req->mtu_idx = NMTUS - 1; /* should be 0 but there's a T3 bug */ + req->iff = idx; + memset(req->src_mac1, 0, sizeof(req->src_mac1)); + memcpy(req->src_mac0, adapter->port[idx]->dev_addr, ETH_ALEN); + skb->priority = 1; + offload_tx(&adapter->tdev, skb); + return 0; +} + +static int init_smt(struct adapter *adapter) +{ + int i; + + for_each_port(adapter, i) + write_smt_entry(adapter, i); + return 0; +} + +static void init_port_mtus(struct adapter *adapter) +{ + unsigned int mtus = adapter->port[0]->mtu; + + if (adapter->port[1]) + mtus |= adapter->port[1]->mtu << 16; + t3_write_reg(adapter, A_TP_MTU_PORT_TABLE, mtus); +} + +static void send_pktsched_cmd(struct adapter *adap, int sched, int qidx, int lo, + int hi, int port) +{ + struct sk_buff *skb; + struct mngt_pktsched_wr *req; + + skb = alloc_skb(sizeof(*req), GFP_KERNEL | __GFP_NOFAIL); + req = (struct mngt_pktsched_wr *)skb_put(skb, sizeof(*req)); + req->wr_hi = htonl(V_WR_OP(FW_WROPCODE_MNGT)); + req->mngt_opcode = FW_MNGTOPCODE_PKTSCHED_SET; + req->sched = sched; + req->idx = qidx; + req->min = lo; + req->max = hi; + req->binding = port; + t3_mgmt_tx(adap, skb); +} + +static void bind_qsets(struct adapter *adap) +{ + int i, j; + + for_each_port(adap, i) { + const struct port_info *pi = adap2pinfo(adap, i); + + for (j = 0; j < pi->nqsets; ++j) + send_pktsched_cmd(adap, 1, pi->first_qset + j, -1, + -1, i); + } +} + +/** + * cxgb_up - enable the adapter + * @adapter: adapter being enabled + * + * Called when the first port is enabled, this function performs the + * actions necessary to make an adapter operational, such as completing + * the initialization of HW modules, and enabling interrupts. + * + * Must be called with the rtnl lock held. + */ +static int cxgb_up(struct adapter *adap) +{ + int err = 0; + + if (!(adap->flags & FULL_INIT_DONE)) { + err = t3_check_fw_version(adap); + if (err) + goto out; + + err = init_dummy_netdevs(adap); + if (err) + goto out; + + err = t3_init_hw(adap, 0); + if (err) + goto out; + + err = setup_sge_qsets(adap); + if (err) + goto out; + + setup_rss(adap); + adap->flags |= FULL_INIT_DONE; + } + + t3_intr_clear(adap); + + if (adap->flags & USING_MSIX) { + name_msix_vecs(adap); + err = request_irq(adap->msix_info[0].vec, + t3_async_intr_handler, 0, + adap->msix_info[0].desc, adap); + if (err) + goto irq_err; + + if (request_msix_data_irqs(adap)) { + free_irq(adap->msix_info[0].vec, adap); + goto irq_err; + } + } else if ((err = request_irq(adap->pdev->irq, + t3_intr_handler(adap, + adap->sge.qs[0].rspq. + polling), + (adap->flags & USING_MSI) ? 0 : SA_SHIRQ, + adap->name, adap))) + goto irq_err; + + t3_sge_start(adap); + t3_intr_enable(adap); + + if ((adap->flags & (USING_MSIX | QUEUES_BOUND)) == USING_MSIX) + bind_qsets(adap); + adap->flags |= QUEUES_BOUND; + +out: + return err; +irq_err: + CH_ERR(adap, "request_irq failed, err %d\n", err); + goto out; +} + +/* + * Release resources when all the ports and offloading have been stopped. + */ +static void cxgb_down(struct adapter *adapter) +{ + t3_sge_stop(adapter); + spin_lock_irq(&adapter->work_lock); /* sync with PHY intr task */ + t3_intr_disable(adapter); + spin_unlock_irq(&adapter->work_lock); + + if (adapter->flags & USING_MSIX) { + int i, n = 0; + + free_irq(adapter->msix_info[0].vec, adapter); + for_each_port(adapter, i) + n += adap2pinfo(adapter, i)->nqsets; + + for (i = 0; i < n; ++i) + free_irq(adapter->msix_info[i + 1].vec, + &adapter->sge.qs[i]); + } else + free_irq(adapter->pdev->irq, adapter); + + flush_workqueue(cxgb3_wq); /* wait for external IRQ handler */ + quiesce_rx(adapter); +} + +static void schedule_chk_task(struct adapter *adap) +{ + unsigned int timeo; + + timeo = adap->params.linkpoll_period ? + (HZ * adap->params.linkpoll_period) / 10 : + adap->params.stats_update_period * HZ; + if (timeo) + queue_delayed_work(cxgb3_wq, &adap->adap_check_task, timeo); +} + +static int offload_open(struct net_device *dev) +{ + struct adapter *adapter = dev->priv; + struct t3cdev *tdev = T3CDEV(dev); + int adap_up = adapter->open_device_map & PORT_MASK; + int err = 0; + + if (test_and_set_bit(OFFLOAD_DEVMAP_BIT, &adapter->open_device_map)) + return 0; + + if (!adap_up && (err = cxgb_up(adapter)) < 0) + return err; + + t3_tp_set_offload_mode(adapter, 1); + tdev->lldev = adapter->port[0]; + err = cxgb3_offload_activate(adapter); + if (err) + goto out; + + init_port_mtus(adapter); + t3_load_mtus(adapter, adapter->params.mtus, adapter->params.a_wnd, + adapter->params.b_wnd, + adapter->params.rev == 0 ? + adapter->port[0]->mtu : 0xffff); + init_smt(adapter); + + /* Never mind if the next step fails */ + sysfs_create_group(&tdev->lldev->class_dev.kobj, &offload_attr_group); + + /* Call back all registered clients */ + cxgb3_add_clients(tdev); + +out: + /* restore them in case the offload module has changed them */ + if (err) { + t3_tp_set_offload_mode(adapter, 0); + clear_bit(OFFLOAD_DEVMAP_BIT, &adapter->open_device_map); + cxgb3_set_dummy_ops(tdev); + } + return err; +} + +static int offload_close(struct t3cdev *tdev) +{ + struct adapter *adapter = tdev2adap(tdev); + + if (!test_bit(OFFLOAD_DEVMAP_BIT, &adapter->open_device_map)) + return 0; + + /* Call back all registered clients */ + cxgb3_remove_clients(tdev); + + sysfs_remove_group(&tdev->lldev->class_dev.kobj, &offload_attr_group); + + tdev->lldev = NULL; + cxgb3_set_dummy_ops(tdev); + t3_tp_set_offload_mode(adapter, 0); + clear_bit(OFFLOAD_DEVMAP_BIT, &adapter->open_device_map); + + if (!adapter->open_device_map) + cxgb_down(adapter); + + cxgb3_offload_deactivate(adapter); + return 0; +} + +static int cxgb_open(struct net_device *dev) +{ + int err; + struct adapter *adapter = dev->priv; + struct port_info *pi = netdev_priv(dev); + int other_ports = adapter->open_device_map & PORT_MASK; + + if (!adapter->open_device_map && (err = cxgb_up(adapter)) < 0) + return err; + + set_bit(pi->port_id, &adapter->open_device_map); + if (!ofld_disable) { + err = offload_open(dev); + if (err) + printk(KERN_WARNING + "Could not initialize offload capabilities\n"); + } + + link_start(dev); + t3_port_intr_enable(adapter, pi->port_id); + netif_start_queue(dev); + if (!other_ports) + schedule_chk_task(adapter); + + return 0; +} + +static int cxgb_close(struct net_device *dev) +{ + struct adapter *adapter = dev->priv; + struct port_info *p = netdev_priv(dev); + + t3_port_intr_disable(adapter, p->port_id); + netif_stop_queue(dev); + p->phy.ops->power_down(&p->phy, 1); + netif_carrier_off(dev); + t3_mac_disable(&p->mac, MAC_DIRECTION_TX | MAC_DIRECTION_RX); + + spin_lock(&adapter->work_lock); /* sync with update task */ + clear_bit(p->port_id, &adapter->open_device_map); + spin_unlock(&adapter->work_lock); + + if (!(adapter->open_device_map & PORT_MASK)) + cancel_rearming_delayed_workqueue(cxgb3_wq, + &adapter->adap_check_task); + + if (!adapter->open_device_map) + cxgb_down(adapter); + + return 0; +} + +static struct net_device_stats *cxgb_get_stats(struct net_device *dev) +{ + struct adapter *adapter = dev->priv; + struct port_info *p = netdev_priv(dev); + struct net_device_stats *ns = &p->netstats; + const struct mac_stats *pstats; + + spin_lock(&adapter->stats_lock); + pstats = t3_mac_update_stats(&p->mac); + spin_unlock(&adapter->stats_lock); + + ns->tx_bytes = pstats->tx_octets; + ns->tx_packets = pstats->tx_frames; + ns->rx_bytes = pstats->rx_octets; + ns->rx_packets = pstats->rx_frames; + ns->multicast = pstats->rx_mcast_frames; + + ns->tx_errors = pstats->tx_underrun; + ns->rx_errors = pstats->rx_symbol_errs + pstats->rx_fcs_errs + + pstats->rx_too_long + pstats->rx_jabber + pstats->rx_short + + pstats->rx_fifo_ovfl; + + /* detailed rx_errors */ + ns->rx_length_errors = pstats->rx_jabber + pstats->rx_too_long; + ns->rx_over_errors = 0; + ns->rx_crc_errors = pstats->rx_fcs_errs; + ns->rx_frame_errors = pstats->rx_symbol_errs; + ns->rx_fifo_errors = pstats->rx_fifo_ovfl; + ns->rx_missed_errors = pstats->rx_cong_drops; + + /* detailed tx_errors */ + ns->tx_aborted_errors = 0; + ns->tx_carrier_errors = 0; + ns->tx_fifo_errors = pstats->tx_underrun; + ns->tx_heartbeat_errors = 0; + ns->tx_window_errors = 0; + return ns; +} + +static u32 get_msglevel(struct net_device *dev) +{ + struct adapter *adapter = dev->priv; + + return adapter->msg_enable; +} + +static void set_msglevel(struct net_device *dev, u32 val) +{ + struct adapter *adapter = dev->priv; + + adapter->msg_enable = val; +} + +static char stats_strings[][ETH_GSTRING_LEN] = { + "TxOctetsOK ", + "TxFramesOK ", + "TxMulticastFramesOK", + "TxBroadcastFramesOK", + "TxPauseFrames ", + "TxUnderrun ", + "TxExtUnderrun ", + + "TxFrames64 ", + "TxFrames65To127 ", + "TxFrames128To255 ", + "TxFrames256To511 ", + "TxFrames512To1023 ", + "TxFrames1024To1518 ", + "TxFrames1519ToMax ", + + "RxOctetsOK ", + "RxFramesOK ", + "RxMulticastFramesOK", + "RxBroadcastFramesOK", + "RxPauseFrames ", + "RxFCSErrors ", + "RxSymbolErrors ", + "RxShortErrors ", + "RxJabberErrors ", + "RxLengthErrors ", + "RxFIFOoverflow ", + + "RxFrames64 ", + "RxFrames65To127 ", + "RxFrames128To255 ", + "RxFrames256To511 ", + "RxFrames512To1023 ", + "RxFrames1024To1518 ", + "RxFrames1519ToMax ", + + "PhyFIFOErrors ", + "TSO ", + "VLANextractions ", + "VLANinsertions ", + "TxCsumOffload ", + "RxCsumGood ", + "RxDrops " +}; + +static int get_stats_count(struct net_device *dev) +{ + return ARRAY_SIZE(stats_strings); +} + +#define T3_REGMAP_SIZE (3 * 1024) + +static int get_regs_len(struct net_device *dev) +{ + return T3_REGMAP_SIZE; +} + +static int get_eeprom_len(struct net_device *dev) +{ + return EEPROMSIZE; +} + +static void get_drvinfo(struct net_device *dev, struct ethtool_drvinfo *info) +{ + u32 fw_vers = 0; + struct adapter *adapter = dev->priv; + + t3_get_fw_version(adapter, &fw_vers); + + strcpy(info->driver, DRV_NAME); + strcpy(info->version, DRV_VERSION); + strcpy(info->bus_info, pci_name(adapter->pdev)); + if (!fw_vers) + strcpy(info->fw_version, "N/A"); + else { + snprintf(info->fw_version, sizeof(info->fw_version), + "%s %u.%u.%u", + G_FW_VERSION_TYPE(fw_vers) ? "T" : "N", + G_FW_VERSION_MAJOR(fw_vers), + G_FW_VERSION_MINOR(fw_vers), + G_FW_VERSION_MICRO(fw_vers)); + } +} + +static void get_strings(struct net_device *dev, u32 stringset, u8 * data) +{ + if (stringset == ETH_SS_STATS) + memcpy(data, stats_strings, sizeof(stats_strings)); +} + +static unsigned long collect_sge_port_stats(struct adapter *adapter, + struct port_info *p, int idx) +{ + int i; + unsigned long tot = 0; + + for (i = 0; i < p->nqsets; ++i) + tot += adapter->sge.qs[i + p->first_qset].port_stats[idx]; + return tot; +} + +static void get_stats(struct net_device *dev, struct ethtool_stats *stats, + u64 *data) +{ + struct adapter *adapter = dev->priv; + struct port_info *pi = netdev_priv(dev); + const struct mac_stats *s; + + spin_lock(&adapter->stats_lock); + s = t3_mac_update_stats(&pi->mac); + spin_unlock(&adapter->stats_lock); + + *data++ = s->tx_octets; + *data++ = s->tx_frames; + *data++ = s->tx_mcast_frames; + *data++ = s->tx_bcast_frames; + *data++ = s->tx_pause; + *data++ = s->tx_underrun; + *data++ = s->tx_fifo_urun; + + *data++ = s->tx_frames_64; + *data++ = s->tx_frames_65_127; + *data++ = s->tx_frames_128_255; + *data++ = s->tx_frames_256_511; + *data++ = s->tx_frames_512_1023; + *data++ = s->tx_frames_1024_1518; + *data++ = s->tx_frames_1519_max; + + *data++ = s->rx_octets; + *data++ = s->rx_frames; + *data++ = s->rx_mcast_frames; + *data++ = s->rx_bcast_frames; + *data++ = s->rx_pause; + *data++ = s->rx_fcs_errs; + *data++ = s->rx_symbol_errs; + *data++ = s->rx_short; + *data++ = s->rx_jabber; + *data++ = s->rx_too_long; + *data++ = s->rx_fifo_ovfl; + + *data++ = s->rx_frames_64; + *data++ = s->rx_frames_65_127; + *data++ = s->rx_frames_128_255; + *data++ = s->rx_frames_256_511; + *data++ = s->rx_frames_512_1023; + *data++ = s->rx_frames_1024_1518; + *data++ = s->rx_frames_1519_max; + + *data++ = pi->phy.fifo_errors; + + *data++ = collect_sge_port_stats(adapter, pi, SGE_PSTAT_TSO); + *data++ = collect_sge_port_stats(adapter, pi, SGE_PSTAT_VLANEX); + *data++ = collect_sge_port_stats(adapter, pi, SGE_PSTAT_VLANINS); + *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; +} + +static inline void reg_block_dump(struct adapter *ap, void *buf, + unsigned int start, unsigned int end) +{ + u32 *p = buf + start; + + for (; start <= end; start += sizeof(u32)) + *p++ = t3_read_reg(ap, start); +} + +static void get_regs(struct net_device *dev, struct ethtool_regs *regs, + void *buf) +{ + struct adapter *ap = dev->priv; + + /* + * Version scheme: + * bits 0..9: chip version + * bits 10..15: chip revision + * bit 31: set for PCIe cards + */ + regs->version = 3 | (ap->params.rev << 10) | (is_pcie(ap) << 31); + + /* + * We skip the MAC statistics registers because they are clear-on-read. + * Also reading multi-register stats would need to synchronize with the + * periodic mac stats accumulation. Hard to justify the complexity. + */ + memset(buf, 0, T3_REGMAP_SIZE); + reg_block_dump(ap, buf, 0, A_SG_RSPQ_CREDIT_RETURN); + reg_block_dump(ap, buf, A_SG_HI_DRB_HI_THRSH, A_ULPRX_PBL_ULIMIT); + reg_block_dump(ap, buf, A_ULPTX_CONFIG, A_MPS_INT_CAUSE); + reg_block_dump(ap, buf, A_CPL_SWITCH_CNTRL, A_CPL_MAP_TBL_DATA); + reg_block_dump(ap, buf, A_SMB_GLOBAL_TIME_CFG, A_XGM_SERDES_STAT3); + reg_block_dump(ap, buf, A_XGM_SERDES_STATUS0, + XGM_REG(A_XGM_SERDES_STAT3, 1)); + reg_block_dump(ap, buf, XGM_REG(A_XGM_SERDES_STATUS0, 1), + XGM_REG(A_XGM_RX_SPI4_SOP_EOP_CNT, 1)); +} + +static int restart_autoneg(struct net_device *dev) +{ + |