int mlx4_check_port_params(struct mlx4_dev *dev,
enum mlx4_port_type *port_type)
{
int i;
for (i = 0; i < dev->caps.num_ports - 1; i++) {
if (port_type[i] != port_type[i + 1]) {
if (!(dev->caps.flags & MLX4_DEV_CAP_FLAG_DPDP)) {
mlx4_err(dev, "Only same port types supported "
"on this HCA, aborting.\n");
return -EINVAL;
}
if (port_type[i] == MLX4_PORT_TYPE_ETH &&
port_type[i + 1] == MLX4_PORT_TYPE_IB)
return -EINVAL;
}
}
for (i = 0; i < dev->caps.num_ports; i++) {
if (!(port_type[i] & dev->caps.supported_type[i+1])) {
mlx4_err(dev, "Requested port type for port %d is not "
"supported on this HCA\n", i + 1);
return -EINVAL;
}
}
return 0;
}
static int mlx4_load_fw(struct mlx4_dev *dev)
{
struct mlx4_priv *priv = mlx4_priv(dev);
int err;
priv->fw.fw_icm = mlx4_alloc_icm(dev, priv->fw.fw_pages,
GFP_HIGHUSER | __GFP_NOWARN, 0);
if (!priv->fw.fw_icm) {
mlx4_err(dev, "Couldn't allocate FW area, aborting.\n");
return -ENOMEM;
}
err = mlx4_MAP_FA(dev, priv->fw.fw_icm);
if (err) {
mlx4_err(dev, "MAP_FA command failed, aborting.\n");
goto err_free;
}
err = mlx4_RUN_FW(dev);
if (err) {
mlx4_err(dev, "RUN_FW command failed, aborting.\n");
goto err_unmap_fa;
}
return 0;
err_unmap_fa:
mlx4_UNMAP_FA(dev);
err_free:
mlx4_free_icm(dev, priv->fw.fw_icm, 0);
return err;
}
static int mlx4_en_config_rss_qp(struct mlx4_en_priv *priv,
int qpn, int srqn, int cqn,
enum mlx4_qp_state *state,
struct mlx4_qp *qp)
{
struct mlx4_en_dev *mdev = priv->mdev;
struct mlx4_qp_context *context;
int err = 0;
context = kmalloc(sizeof *context , GFP_KERNEL);
if (!context) {
mlx4_err(mdev, "Failed to allocate qp context\n");
return -ENOMEM;
}
err = mlx4_qp_alloc(mdev->dev, qpn, qp);
if (err) {
mlx4_err(mdev, "Failed to allocate qp #%d\n", qpn);
goto out;
return err;
}
qp->event = mlx4_en_sqp_event;
memset(context, 0, sizeof *context);
mlx4_en_fill_qp_context(priv, 0, 0, 0, 0, qpn, cqn, srqn, context);
err = mlx4_qp_to_ready(mdev->dev, &priv->res.mtt, context, qp, state);
if (err) {
mlx4_qp_remove(mdev->dev, qp);
mlx4_qp_free(mdev->dev, qp);
}
out:
kfree(context);
return err;
}
int mlx4_en_create_rx_ring(struct mlx4_en_priv *priv,
struct mlx4_en_rx_ring *ring, u32 size, u16 stride)
{
struct mlx4_en_dev *mdev = priv->mdev;
int err;
int tmp;
/* Sanity check SRQ size before proceeding */
if (size >= mdev->dev->caps.max_srq_wqes)
return -EINVAL;
ring->prod = 0;
ring->cons = 0;
ring->size = size;
ring->size_mask = size - 1;
ring->stride = stride;
ring->log_stride = ffs(ring->stride) - 1;
ring->buf_size = ring->size * ring->stride;
tmp = size * roundup_pow_of_two(MLX4_EN_MAX_RX_FRAGS *
sizeof(struct skb_frag_struct));
ring->rx_info = vmalloc(tmp);
if (!ring->rx_info) {
mlx4_err(mdev, "Failed allocating rx_info ring\n");
return -ENOMEM;
}
mlx4_dbg(DRV, priv, "Allocated rx_info ring at addr:%p size:%d\n",
ring->rx_info, tmp);
err = mlx4_alloc_hwq_res(mdev->dev, &ring->wqres,
ring->buf_size, 2 * PAGE_SIZE);
if (err)
goto err_ring;
err = mlx4_en_map_buffer(&ring->wqres.buf);
if (err) {
mlx4_err(mdev, "Failed to map RX buffer\n");
goto err_hwq;
}
ring->buf = ring->wqres.buf.direct.buf;
/* Allocate LRO sessions */
if (mlx4_en_lro_init(ring, mdev->profile.num_lro)) {
mlx4_err(mdev, "Failed allocating lro sessions\n");
goto err_map;
}
return 0;
err_map:
mlx4_en_unmap_buffer(&ring->wqres.buf);
err_hwq:
mlx4_free_hwq_res(mdev->dev, &ring->wqres, ring->buf_size);
err_ring:
vfree(ring->rx_info);
ring->rx_info = NULL;
return err;
}
static void dump_err_buf(struct mlx4_dev *dev)
{
struct mlx4_priv *priv = mlx4_priv(dev);
int i;
mlx4_err(dev, "Internal error detected:\n");
for (i = 0; i < priv->fw.catas_size; ++i)
mlx4_err(dev, " buf[%02x]: %08x\n",
i, swab32(readl(priv->catas_err.map + i)));
}
int mlx4_init_uar_table(struct mlx4_dev *dev)
{
if (dev->caps.num_uars <= 128) {
mlx4_err(dev, "Only %d UAR pages (need more than 128)\n",
dev->caps.num_uars);
mlx4_err(dev, "Increase firmware log2_uar_bar_megabytes?\n");
return -ENODEV;
}
return mlx4_bitmap_init(&mlx4_priv(dev)->uar_table.bitmap,
dev->caps.num_uars, dev->caps.num_uars - 1,
max(128, dev->caps.reserved_uars), 0);
}
static void mlx4_en_event(struct mlx4_dev *dev, void *endev_ptr,
enum mlx4_dev_event event, unsigned long port)
{
struct mlx4_en_dev *mdev = (struct mlx4_en_dev *) endev_ptr;
struct mlx4_en_priv *priv;
switch (event) {
case MLX4_DEV_EVENT_PORT_UP:
case MLX4_DEV_EVENT_PORT_DOWN:
if (!mdev->pndev[port])
return;
priv = netdev_priv(mdev->pndev[port]);
/* To prevent races, we poll the link state in a separate
task rather than changing it here */
priv->link_state = event;
queue_work(mdev->workqueue, &priv->linkstate_task);
break;
case MLX4_DEV_EVENT_CATASTROPHIC_ERROR:
mlx4_err(mdev, "Internal error detected, restarting device\n");
break;
case MLX4_DEV_EVENT_SLAVE_INIT:
case MLX4_DEV_EVENT_SLAVE_SHUTDOWN:
break;
default:
if (port < 1 || port > dev->caps.num_ports ||
!mdev->pndev[port])
return;
mlx4_warn(mdev, "Unhandled event %d for port %d\n", event,
(int) port);
}
}
static int mlx4_en_fill_rx_buffers(struct mlx4_en_priv *priv)
{
struct mlx4_en_dev *mdev = priv->mdev;
struct mlx4_en_rx_ring *ring;
int ring_ind;
int buf_ind;
for (buf_ind = 0; buf_ind < priv->prof->rx_ring_size; buf_ind++) {
for (ring_ind = 0; ring_ind < priv->rx_ring_num; ring_ind++) {
ring = &priv->rx_ring[ring_ind];
if (mlx4_en_prepare_rx_desc(priv, ring,
ring->actual_size)) {
if (ring->actual_size < MLX4_EN_MIN_RX_SIZE) {
mlx4_err(mdev, "Failed to allocate "
"enough rx buffers\n");
return -ENOMEM;
} else {
if (netif_msg_rx_err(priv))
mlx4_warn(mdev,
"Only %d buffers allocated\n",
ring->actual_size);
goto out;
}
}
ring->actual_size++;
ring->prod++;
}
}
out:
return 0;
}
/*
* Change the port configuration of the device.
* Every user of this function must hold the port mutex.
*/
int mlx4_change_port_types(struct mlx4_dev *dev,
enum mlx4_port_type *port_types)
{
int err = 0;
int change = 0;
int port;
for (port = 0; port < dev->caps.num_ports; port++) {
/* Change the port type only if the new type is different
* from the current, and not set to Auto */
if (port_types[port] != dev->caps.port_type[port + 1]) {
change = 1;
dev->caps.port_type[port + 1] = port_types[port];
}
}
if (change) {
mlx4_unregister_device(dev);
for (port = 1; port <= dev->caps.num_ports; port++) {
mlx4_CLOSE_PORT(dev, port);
err = mlx4_SET_PORT(dev, port);
if (err) {
mlx4_err(dev, "Failed to set port %d, "
"aborting\n", port);
goto out;
}
}
mlx4_set_port_mask(dev);
err = mlx4_register_device(dev);
}
out:
return err;
}
static int mlx4_uc_steer_add(struct mlx4_dev *dev, u8 port,
u64 mac, int *qpn, u8 reserve)
{
struct mlx4_qp qp;
u8 gid[16] = {0};
int err;
if (reserve) {
err = mlx4_qp_reserve_range(dev, 1, 1, qpn, 0xC0);
if (err) {
mlx4_err(dev, "Failed to reserve qp for mac registration\n");
return err;
}
}
qp.qpn = *qpn;
mac &= 0xffffffffffffULL;
mac = cpu_to_be64(mac << 16);
memcpy(&gid[10], &mac, ETH_ALEN);
gid[5] = port;
gid[7] = MLX4_UC_STEER << 1;
err = mlx4_qp_attach_common(dev, &qp, gid, 0,
MLX4_PROT_ETH, MLX4_UC_STEER, 0);
if (err && reserve)
mlx4_qp_release_range(dev, *qpn, 1);
return err;
}
static void mlx4_en_event(struct mlx4_dev *dev, void *endev_ptr,
enum mlx4_dev_event event, int port)
{
struct mlx4_en_dev *mdev = (struct mlx4_en_dev *) endev_ptr;
struct mlx4_en_priv *priv;
if (!mdev->pndev[port])
return;
priv = netdev_priv(mdev->pndev[port]);
switch (event) {
case MLX4_DEV_EVENT_PORT_UP:
case MLX4_DEV_EVENT_PORT_DOWN:
/* To prevent races, we poll the link state in a separate
task rather than changing it here */
priv->link_state = event;
queue_work(mdev->workqueue, &priv->linkstate_task);
break;
case MLX4_DEV_EVENT_CATASTROPHIC_ERROR:
mlx4_err(mdev, "Internal error detected, restarting device\n");
break;
default:
mlx4_warn(mdev, "Unhandled event: %d\n", event);
}
}
int mlx4_change_port_types(struct mlx4_dev *dev,
enum mlx4_port_type *port_types)
{
int err = 0;
int change = 0;
int port;
for (port = 0; port < dev->caps.num_ports; port++) {
if (port_types[port] != dev->caps.port_type[port + 1]) {
change = 1;
dev->caps.port_type[port + 1] = port_types[port];
}
}
if (change) {
mlx4_unregister_device(dev);
for (port = 1; port <= dev->caps.num_ports; port++) {
mlx4_CLOSE_PORT(dev, port);
err = mlx4_SET_PORT(dev, port);
if (err) {
mlx4_err(dev, "Failed to set port %d, "
"aborting\n", port);
goto out;
}
}
mlx4_set_port_mask(dev);
err = mlx4_register_device(dev);
}
out:
return err;
}
static void mlx4_en_event(struct mlx4_dev *dev, void *endev_ptr,
enum mlx4_dev_event event, unsigned long port)
{
struct mlx4_en_dev *mdev = (struct mlx4_en_dev *) endev_ptr;
struct mlx4_en_priv *priv;
/* check that port param is not a pointer */
if (port != (port & (unsigned long)0x0FFFF))
return;
switch (event) {
case MLX4_DEV_EVENT_PORT_UP:
/* To prevent races, we poll the link state in a separate
task rather than changing it here */
if (!mdev->pndev[port])
return;
priv = netdev_priv(mdev->pndev[port]);
priv->link_state = 1;
queue_work(mdev->workqueue, &priv->linkstate_task);
break;
case MLX4_DEV_EVENT_PORT_DOWN:
/* To prevent races, we poll the link state in a separate
task rather than changing it here */
if (!mdev->pndev[port])
return;
priv = netdev_priv(mdev->pndev[port]);
priv->link_state = 0;
queue_work(mdev->workqueue, &priv->linkstate_task);
break;
case MLX4_DEV_EVENT_CATASTROPHIC_ERROR:
#ifndef __VMKERNEL_MODULE__
mlx4_err(mdev, "Internal error detected, restarting device\n");
#else /* __VMKERNEL_MODULE__ */
mlx4_err(mdev, "Internal error detected, please reload the driver manually\n");
#endif /* __VMKERNEL_MODULE__ */
break;
default:
mlx4_warn(mdev, "Unhandled event: %d\n", event);
}
}
static int read_vendor_id(struct mlx4_dev *dev)
{
u16 vendor_id = 0;
int ret;
ret = pci_read_config_word(dev->persist->pdev, 0, &vendor_id);
if (ret) {
mlx4_err(dev, "Failed to read vendor ID, ret=%d\n", ret);
return ret;
}
if (vendor_id == 0xffff) {
mlx4_err(dev, "PCI can't be accessed to read vendor id\n");
return -EINVAL;
}
return 0;
}
void mlx4_enter_error_state(struct mlx4_dev_persistent *persist)
{
int err;
struct mlx4_dev *dev;
if (!mlx4_internal_err_reset)
return;
mutex_lock(&persist->device_state_mutex);
if (persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR)
goto out;
dev = persist->dev;
mlx4_err(dev, "device is going to be reset\n");
if (mlx4_is_slave(dev)) {
err = mlx4_reset_slave(dev);
} else {
mlx4_crdump_collect(dev);
err = mlx4_reset_master(dev);
}
if (!err) {
mlx4_err(dev, "device was reset successfully\n");
} else {
/* EEH could have disabled the PCI channel during reset. That's
* recoverable and the PCI error flow will handle it.
*/
if (!pci_channel_offline(dev->persist->pdev))
BUG_ON(1);
}
dev->persist->state |= MLX4_DEVICE_STATE_INTERNAL_ERROR;
mutex_unlock(&persist->device_state_mutex);
/* At that step HW was already reset, now notify clients */
mlx4_dispatch_event(dev, MLX4_DEV_EVENT_CATASTROPHIC_ERROR, 0);
mlx4_cmd_wake_completions(dev);
return;
out:
mutex_unlock(&persist->device_state_mutex);
}
int mlx4_register_mac(struct mlx4_dev *dev, u8 port, u64 mac, int *index)
{
struct mlx4_mac_table *table = &mlx4_priv(dev)->port[port].mac_table;
int i, err = 0;
int free = -1;
mlx4_dbg(dev, "Registering MAC: 0x%llx\n", (unsigned long long) mac);
mutex_lock(&table->mutex);
for (i = 0; i < MLX4_MAX_MAC_NUM - 1; i++) {
if (free < 0 && !table->refs[i]) {
free = i;
continue;
}
if (mac == (MLX4_MAC_MASK & be64_to_cpu(table->entries[i]))) {
/* MAC already registered, increase refernce count */
*index = i;
++table->refs[i];
goto out;
}
}
if (free < 0) {
err = -ENOMEM;
goto out;
}
mlx4_dbg(dev, "Free MAC index is %d\n", free);
if (table->total == table->max) {
/* No free mac entries */
err = -ENOSPC;
goto out;
}
/* Register new MAC */
table->refs[free] = 1;
table->entries[free] = cpu_to_be64(mac | MLX4_MAC_VALID);
err = mlx4_set_port_mac_table(dev, port, table->entries);
if (unlikely(err)) {
mlx4_err(dev, "Failed adding MAC: 0x%llx\n", (unsigned long long) mac);
table->refs[free] = 0;
table->entries[free] = 0;
goto out;
}
*index = free;
++table->total;
out:
mutex_unlock(&table->mutex);
return err;
}
static int mlx4_SENSE_PORT(struct mlx4_dev *dev, int port,
enum mlx4_port_type *type)
{
u64 out_param;
int err = 0;
err = mlx4_cmd_imm(dev, 0, &out_param, port, 0,
MLX4_CMD_SENSE_PORT, MLX4_CMD_TIME_CLASS_B);
if (err) {
mlx4_err(dev, "Sense command failed for port: %d\n", port);
return err;
}
if (out_param > 2) {
mlx4_err(dev, "Sense returned illegal value: 0x%llx\n", out_param);
return EINVAL;
}
*type = out_param;
return 0;
}
void mlx4_en_init_timestamp(struct mlx4_en_dev *mdev)
{
panic("Disabled");
#if 0 // AKAROS_PORT
struct mlx4_dev *dev = mdev->dev;
unsigned long flags;
uint64_t ns, zero = 0;
rwlock_init(&mdev->clock_lock);
memset(&mdev->cycles, 0, sizeof(mdev->cycles));
mdev->cycles.read = mlx4_en_read_clock;
mdev->cycles.mask = CLOCKSOURCE_MASK(48);
/* Using shift to make calculation more accurate. Since current HW
* clock frequency is 427 MHz, and cycles are given using a 48 bits
* register, the biggest shift when calculating using u64, is 14
* (max_cycles * multiplier < 2^64)
*/
mdev->cycles.shift = 14;
mdev->cycles.mult =
clocksource_khz2mult(1000 * dev->caps.hca_core_clock, mdev->cycles.shift);
mdev->nominal_c_mult = mdev->cycles.mult;
write_lock_irqsave(&mdev->clock_lock, flags);
timecounter_init(&mdev->clock, &mdev->cycles,
epoch_nsec());
write_unlock_irqrestore(&mdev->clock_lock, flags);
/* Calculate period in seconds to call the overflow watchdog - to make
* sure counter is checked at least once every wrap around.
*/
ns = cyclecounter_cyc2ns(&mdev->cycles, mdev->cycles.mask, zero, &zero);
do_div(ns, NSEC_PER_SEC / 2 / HZ);
mdev->overflow_period = ns;
/* Configure the PHC */
mdev->ptp_clock_info = mlx4_en_ptp_clock_info;
snprintf(mdev->ptp_clock_info.name, 16, "mlx4 ptp");
mdev->ptp_clock = ptp_clock_register(&mdev->ptp_clock_info,
&mdev->pdev->dev);
if (IS_ERR(mdev->ptp_clock)) {
mdev->ptp_clock = NULL;
mlx4_err(mdev, "ptp_clock_register failed\n");
} else {
mlx4_info(mdev, "registered PHC clock\n");
}
#endif
}
void mlx4_en_init_timestamp(struct mlx4_en_dev *mdev)
{
struct mlx4_dev *dev = mdev->dev;
unsigned long flags;
u64 ns;
/* mlx4_en_init_timestamp is called for each netdev.
* mdev->ptp_clock is common for all ports, skip initialization if
* was done for other port.
*/
if (mdev->ptp_clock)
return;
rwlock_init(&mdev->clock_lock);
memset(&mdev->cycles, 0, sizeof(mdev->cycles));
mdev->cycles.read = mlx4_en_read_clock;
mdev->cycles.mask = CLOCKSOURCE_MASK(48);
mdev->cycles.shift = freq_to_shift(dev->caps.hca_core_clock);
mdev->cycles.mult =
clocksource_khz2mult(1000 * dev->caps.hca_core_clock, mdev->cycles.shift);
mdev->nominal_c_mult = mdev->cycles.mult;
write_lock_irqsave(&mdev->clock_lock, flags);
timecounter_init(&mdev->clock, &mdev->cycles,
ktime_to_ns(ktime_get_real()));
write_unlock_irqrestore(&mdev->clock_lock, flags);
/* Calculate period in seconds to call the overflow watchdog - to make
* sure counter is checked at least once every wrap around.
*/
ns = cyclecounter_cyc2ns(&mdev->cycles, mdev->cycles.mask);
do_div(ns, NSEC_PER_SEC / 2 / HZ);
mdev->overflow_period = ns;
/* Configure the PHC */
mdev->ptp_clock_info = mlx4_en_ptp_clock_info;
snprintf(mdev->ptp_clock_info.name, 16, "mlx4 ptp");
mdev->ptp_clock = ptp_clock_register(&mdev->ptp_clock_info,
&mdev->pdev->dev);
if (IS_ERR(mdev->ptp_clock)) {
mdev->ptp_clock = NULL;
mlx4_err(mdev, "ptp_clock_register failed\n");
} else {
mlx4_info(mdev, "registered PHC clock\n");
}
}
static int mlx4_reset_master(struct mlx4_dev *dev)
{
int err = 0;
if (mlx4_is_master(dev))
mlx4_report_internal_err_comm_event(dev);
if (!pci_channel_offline(dev->persist->pdev)) {
err = read_vendor_id(dev);
/* If PCI can't be accessed to read vendor ID we assume that its
* link was disabled and chip was already reset.
*/
if (err)
return 0;
err = mlx4_reset(dev);
if (err)
mlx4_err(dev, "Fail to reset HCA\n");
}
return err;
}
static void mlx4_sense_port(struct work_struct *work)
{
struct delayed_work *delay = to_delayed_work(work);
struct mlx4_sense *sense = container_of(delay, struct mlx4_sense,
sense_poll);
struct mlx4_dev *dev = sense->dev;
struct mlx4_priv *priv = mlx4_priv(dev);
enum mlx4_port_type stype[MLX4_MAX_PORTS];
mutex_lock(&priv->port_mutex);
mlx4_do_sense_ports(dev, stype, &dev->caps.port_type[1]);
if (mlx4_check_port_params(dev, stype))
goto sense_again;
if (mlx4_change_port_types(dev, stype))
mlx4_err(dev, "Failed to change port_types\n");
sense_again:
mutex_unlock(&priv->port_mutex);
queue_delayed_work(mlx4_wq , &sense->sense_poll,
round_jiffies_relative(MLX4_SENSE_RANGE));
}
static void mlx4_en_remove(struct mlx4_dev *dev, void *endev_ptr)
{
struct mlx4_en_dev *mdev = endev_ptr;
int i, ret;
mutex_lock(&mdev->state_lock);
mdev->device_up = false;
mutex_unlock(&mdev->state_lock);
mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_ETH)
if (mdev->pndev[i])
mlx4_en_destroy_netdev(mdev->pndev[i]);
flush_workqueue(mdev->workqueue);
destroy_workqueue(mdev->workqueue);
ret = mlx4_mr_free(dev, &mdev->mr);
if (ret)
mlx4_err(mdev, "Error deregistering MR. The system may have become unstable.");
iounmap(mdev->uar_map);
mlx4_uar_free(dev, &mdev->priv_uar);
mlx4_pd_free(dev, mdev->priv_pdn);
kfree(mdev);
}
int mlx4_replace_mac(struct mlx4_dev *dev, u8 port, int qpn, u64 new_mac, u8 wrap)
{
struct mlx4_port_info *info = &mlx4_priv(dev)->port[port];
struct mlx4_mac_table *table = &info->mac_table;
int index = qpn - info->base_qpn;
struct mlx4_mac_entry *entry;
int err;
if (dev->caps.vep_uc_steering) {
entry = radix_tree_lookup(&info->mac_tree, qpn);
if (!entry)
return -EINVAL;
index = find_index(dev, table, entry->mac);
mlx4_uc_steer_release(dev, port, entry->mac, qpn, 0);
entry->mac = new_mac;
err = mlx4_uc_steer_add(dev, port, entry->mac, &qpn, 0);
if (err || index < 0)
return err;
}
mutex_lock(&table->mutex);
err = validate_index(dev, table, index);
if (err)
goto out;
table->entries[index] = cpu_to_be64(new_mac | MLX4_MAC_VALID);
err = mlx4_set_port_mac_table(dev, port, table->entries);
if (unlikely(err)) {
mlx4_err(dev, "Failed adding MAC: 0x%llx\n", (unsigned long long) new_mac);
table->entries[index] = 0;
}
out:
mutex_unlock(&table->mutex);
return err;
}
static int mlx4_init_port_info(struct mlx4_dev *dev, int port)
{
struct mlx4_port_info *info = &mlx4_priv(dev)->port[port];
int err = 0;
info->dev = dev;
info->port = port;
mlx4_init_mac_table(dev, &info->mac_table);
mlx4_init_vlan_table(dev, &info->vlan_table);
sprintf(info->dev_name, "mlx4_port%d", port);
info->port_attr.attr.name = info->dev_name;
info->port_attr.attr.mode = S_IRUGO | S_IWUSR;
info->port_attr.show = show_port_type;
info->port_attr.store = set_port_type;
err = device_create_file(&dev->pdev->dev, &info->port_attr);
if (err) {
mlx4_err(dev, "Failed to create file for port %d\n", port);
info->port = -1;
}
return err;
}
int mlx4_register_mac(struct mlx4_dev *dev, u8 port, u64 mac, int *qpn, u8 wrap)
{
struct mlx4_port_info *info = &mlx4_priv(dev)->port[port];
struct mlx4_mac_table *table = &info->mac_table;
struct mlx4_mac_entry *entry;
int i, err = 0;
int free = -1;
if (dev->caps.vep_uc_steering) {
err = mlx4_uc_steer_add(dev, port, mac, qpn, 1);
if (!err) {
entry = kmalloc(sizeof *entry, GFP_KERNEL);
if (!entry) {
mlx4_uc_steer_release(dev, port, mac, *qpn, 1);
return -ENOMEM;
}
entry->mac = mac;
err = radix_tree_insert(&info->mac_tree, *qpn, entry);
if (err) {
mlx4_uc_steer_release(dev, port, mac, *qpn, 1);
return err;
}
} else
return err;
}
mlx4_dbg(dev, "Registering MAC: 0x%llx\n", (unsigned long long) mac);
mutex_lock(&table->mutex);
for (i = 0; i < MLX4_MAX_MAC_NUM; i++) {
if (free < 0 && !table->entries[i]) {
free = i;
continue;
}
if (mac == (MLX4_MAC_MASK & be64_to_cpu(table->entries[i]))) {
/* MAC already registered, Must not have duplicates */
err = -EEXIST;
goto out;
}
}
mlx4_dbg(dev, "Free MAC index is %d\n", free);
if (table->total == table->max) {
/* No free mac entries */
err = -ENOSPC;
goto out;
}
/* Register new MAC */
table->entries[free] = cpu_to_be64(mac | MLX4_MAC_VALID);
err = mlx4_set_port_mac_table(dev, port, table->entries);
if (unlikely(err)) {
mlx4_err(dev, "Failed adding MAC: 0x%llx\n", (unsigned long long) mac);
table->entries[free] = 0;
goto out;
}
if (!dev->caps.vep_uc_steering)
*qpn = info->base_qpn + free;
++table->total;
out:
mutex_unlock(&table->mutex);
return err;
}
static int __mlx4_init_one(struct pci_dev *pdev, const struct pci_device_id *id)
{
struct mlx4_priv *priv;
struct mlx4_dev *dev;
int err;
int port;
printk(KERN_INFO PFX "Initializing %s\n",
pci_name(pdev));
err = pci_enable_device(pdev);
if (err) {
dev_err(&pdev->dev, "Cannot enable PCI device, "
"aborting.\n");
return err;
}
if (!(pci_resource_flags(pdev, 0) & IORESOURCE_MEM) ||
pci_resource_len(pdev, 0) != 1 << 20) {
dev_err(&pdev->dev, "Missing DCS, aborting.\n");
err = -ENODEV;
goto err_disable_pdev;
}
if (!(pci_resource_flags(pdev, 2) & IORESOURCE_MEM)) {
dev_err(&pdev->dev, "Missing UAR, aborting.\n");
err = -ENODEV;
goto err_disable_pdev;
}
err = pci_request_regions(pdev, DRV_NAME);
if (err) {
dev_err(&pdev->dev, "Couldn't get PCI resources, aborting\n");
goto err_disable_pdev;
}
pci_set_master(pdev);
err = pci_set_dma_mask(pdev, DMA_BIT_MASK(64));
if (err) {
dev_warn(&pdev->dev, "Warning: couldn't set 64-bit PCI DMA mask.\n");
err = pci_set_dma_mask(pdev, DMA_BIT_MASK(32));
if (err) {
dev_err(&pdev->dev, "Can't set PCI DMA mask, aborting.\n");
goto err_release_regions;
}
}
err = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(64));
if (err) {
dev_warn(&pdev->dev, "Warning: couldn't set 64-bit "
"consistent PCI DMA mask.\n");
err = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(32));
if (err) {
dev_err(&pdev->dev, "Can't set consistent PCI DMA mask, "
"aborting.\n");
goto err_release_regions;
}
}
priv = kzalloc(sizeof *priv, GFP_KERNEL);
if (!priv) {
dev_err(&pdev->dev, "Device struct alloc failed, "
"aborting.\n");
err = -ENOMEM;
goto err_release_regions;
}
dev = &priv->dev;
dev->pdev = pdev;
INIT_LIST_HEAD(&priv->ctx_list);
spin_lock_init(&priv->ctx_lock);
mutex_init(&priv->port_mutex);
INIT_LIST_HEAD(&priv->pgdir_list);
mutex_init(&priv->pgdir_mutex);
err = mlx4_reset(dev);
if (err) {
mlx4_err(dev, "Failed to reset HCA, aborting.\n");
goto err_free_dev;
}
if (mlx4_cmd_init(dev)) {
mlx4_err(dev, "Failed to init command interface, aborting.\n");
goto err_free_dev;
}
err = mlx4_init_hca(dev);
if (err)
goto err_cmd;
err = mlx4_alloc_eq_table(dev);
if (err)
goto err_close;
mlx4_enable_msi_x(dev);
err = mlx4_setup_hca(dev);
if (err == -EBUSY && (dev->flags & MLX4_FLAG_MSI_X)) {
dev->flags &= ~MLX4_FLAG_MSI_X;
pci_disable_msix(pdev);
err = mlx4_setup_hca(dev);
}
if (err)
goto err_free_eq;
for (port = 1; port <= dev->caps.num_ports; port++) {
err = mlx4_init_port_info(dev, port);
if (err)
goto err_port;
}
err = mlx4_register_device(dev);
if (err)
goto err_port;
mlx4_sense_init(dev);
mlx4_start_sense(dev);
pci_set_drvdata(pdev, dev);
return 0;
err_port:
for (port = 1; port <= dev->caps.num_ports; port++)
mlx4_cleanup_port_info(&priv->port[port]);
mlx4_cleanup_mcg_table(dev);
mlx4_cleanup_qp_table(dev);
mlx4_cleanup_srq_table(dev);
mlx4_cleanup_cq_table(dev);
mlx4_cmd_use_polling(dev);
mlx4_cleanup_eq_table(dev);
mlx4_cleanup_mr_table(dev);
mlx4_cleanup_pd_table(dev);
mlx4_cleanup_uar_table(dev);
err_free_eq:
mlx4_free_eq_table(dev);
err_close:
if (dev->flags & MLX4_FLAG_MSI_X)
pci_disable_msix(pdev);
mlx4_close_hca(dev);
err_cmd:
mlx4_cmd_cleanup(dev);
err_free_dev:
kfree(priv);
err_release_regions:
pci_release_regions(pdev);
err_disable_pdev:
pci_disable_device(pdev);
pci_set_drvdata(pdev, NULL);
return err;
}
static int mlx4_setup_hca(struct mlx4_dev *dev)
{
struct mlx4_priv *priv = mlx4_priv(dev);
int err;
int port;
__be32 ib_port_default_caps;
err = mlx4_init_uar_table(dev);
if (err) {
mlx4_err(dev, "Failed to initialize "
"user access region table, aborting.\n");
return err;
}
err = mlx4_uar_alloc(dev, &priv->driver_uar);
if (err) {
mlx4_err(dev, "Failed to allocate driver access region, "
"aborting.\n");
goto err_uar_table_free;
}
priv->kar = ioremap(priv->driver_uar.pfn << PAGE_SHIFT, PAGE_SIZE);
if (!priv->kar) {
mlx4_err(dev, "Couldn't map kernel access region, "
"aborting.\n");
err = -ENOMEM;
goto err_uar_free;
}
err = mlx4_init_pd_table(dev);
if (err) {
mlx4_err(dev, "Failed to initialize "
"protection domain table, aborting.\n");
goto err_kar_unmap;
}
err = mlx4_init_mr_table(dev);
if (err) {
mlx4_err(dev, "Failed to initialize "
"memory region table, aborting.\n");
goto err_pd_table_free;
}
err = mlx4_init_eq_table(dev);
if (err) {
mlx4_err(dev, "Failed to initialize "
"event queue table, aborting.\n");
goto err_mr_table_free;
}
err = mlx4_cmd_use_events(dev);
if (err) {
mlx4_err(dev, "Failed to switch to event-driven "
"firmware commands, aborting.\n");
goto err_eq_table_free;
}
err = mlx4_NOP(dev);
if (err) {
if (dev->flags & MLX4_FLAG_MSI_X) {
mlx4_warn(dev, "NOP command failed to generate MSI-X "
"interrupt IRQ %d).\n",
priv->eq_table.eq[dev->caps.num_comp_vectors].irq);
mlx4_warn(dev, "Trying again without MSI-X.\n");
} else {
mlx4_err(dev, "NOP command failed to generate interrupt "
"(IRQ %d), aborting.\n",
priv->eq_table.eq[dev->caps.num_comp_vectors].irq);
mlx4_err(dev, "BIOS or ACPI interrupt routing problem?\n");
}
goto err_cmd_poll;
}
mlx4_dbg(dev, "NOP command IRQ test passed\n");
err = mlx4_init_cq_table(dev);
if (err) {
mlx4_err(dev, "Failed to initialize "
"completion queue table, aborting.\n");
goto err_cmd_poll;
}
err = mlx4_init_srq_table(dev);
if (err) {
mlx4_err(dev, "Failed to initialize "
"shared receive queue table, aborting.\n");
goto err_cq_table_free;
}
err = mlx4_init_qp_table(dev);
if (err) {
mlx4_err(dev, "Failed to initialize "
"queue pair table, aborting.\n");
goto err_srq_table_free;
}
err = mlx4_init_mcg_table(dev);
if (err) {
mlx4_err(dev, "Failed to initialize "
"multicast group table, aborting.\n");
goto err_qp_table_free;
}
for (port = 1; port <= dev->caps.num_ports; port++) {
ib_port_default_caps = 0;
err = mlx4_get_port_ib_caps(dev, port, &ib_port_default_caps);
if (err)
mlx4_warn(dev, "failed to get port %d default "
"ib capabilities (%d). Continuing with "
"caps = 0\n", port, err);
dev->caps.ib_port_def_cap[port] = ib_port_default_caps;
err = mlx4_SET_PORT(dev, port);
if (err) {
mlx4_err(dev, "Failed to set port %d, aborting\n",
port);
goto err_mcg_table_free;
}
}
return 0;
err_mcg_table_free:
mlx4_cleanup_mcg_table(dev);
err_qp_table_free:
mlx4_cleanup_qp_table(dev);
err_srq_table_free:
mlx4_cleanup_srq_table(dev);
err_cq_table_free:
mlx4_cleanup_cq_table(dev);
err_cmd_poll:
mlx4_cmd_use_polling(dev);
err_eq_table_free:
mlx4_cleanup_eq_table(dev);
err_mr_table_free:
mlx4_cleanup_mr_table(dev);
err_pd_table_free:
mlx4_cleanup_pd_table(dev);
err_kar_unmap:
iounmap(priv->kar);
err_uar_free:
mlx4_uar_free(dev, &priv->driver_uar);
err_uar_table_free:
mlx4_cleanup_uar_table(dev);
return err;
}
static int mlx4_init_hca(struct mlx4_dev *dev)
{
struct mlx4_priv *priv = mlx4_priv(dev);
struct mlx4_adapter adapter;
struct mlx4_dev_cap dev_cap;
struct mlx4_mod_stat_cfg mlx4_cfg;
struct mlx4_profile profile;
struct mlx4_init_hca_param init_hca;
u64 icm_size;
int err;
err = mlx4_QUERY_FW(dev);
if (err) {
if (err == -EACCES)
mlx4_info(dev, "non-primary physical function, skipping.\n");
else
mlx4_err(dev, "QUERY_FW command failed, aborting.\n");
return err;
}
err = mlx4_load_fw(dev);
if (err) {
mlx4_err(dev, "Failed to start FW, aborting.\n");
return err;
}
mlx4_cfg.log_pg_sz_m = 1;
mlx4_cfg.log_pg_sz = 0;
err = mlx4_MOD_STAT_CFG(dev, &mlx4_cfg);
if (err)
mlx4_warn(dev, "Failed to override log_pg_sz parameter\n");
err = mlx4_dev_cap(dev, &dev_cap);
if (err) {
mlx4_err(dev, "QUERY_DEV_CAP command failed, aborting.\n");
goto err_stop_fw;
}
profile = default_profile;
icm_size = mlx4_make_profile(dev, &profile, &dev_cap, &init_hca);
if ((long long) icm_size < 0) {
err = icm_size;
goto err_stop_fw;
}
init_hca.log_uar_sz = ilog2(dev->caps.num_uars);
err = mlx4_init_icm(dev, &dev_cap, &init_hca, icm_size);
if (err)
goto err_stop_fw;
err = mlx4_INIT_HCA(dev, &init_hca);
if (err) {
mlx4_err(dev, "INIT_HCA command failed, aborting.\n");
goto err_free_icm;
}
err = mlx4_QUERY_ADAPTER(dev, &adapter);
if (err) {
mlx4_err(dev, "QUERY_ADAPTER command failed, aborting.\n");
goto err_close;
}
priv->eq_table.inta_pin = adapter.inta_pin;
memcpy(dev->board_id, adapter.board_id, sizeof dev->board_id);
return 0;
err_close:
mlx4_CLOSE_HCA(dev, 0);
err_free_icm:
mlx4_free_icms(dev);
err_stop_fw:
mlx4_UNMAP_FA(dev);
mlx4_free_icm(dev, priv->fw.fw_icm, 0);
return err;
}
static int mlx4_init_icm(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap,
struct mlx4_init_hca_param *init_hca, u64 icm_size)
{
struct mlx4_priv *priv = mlx4_priv(dev);
u64 aux_pages;
int err;
err = mlx4_SET_ICM_SIZE(dev, icm_size, &aux_pages);
if (err) {
mlx4_err(dev, "SET_ICM_SIZE command failed, aborting.\n");
return err;
}
mlx4_dbg(dev, "%lld KB of HCA context requires %lld KB aux memory.\n",
(unsigned long long) icm_size >> 10,
(unsigned long long) aux_pages << 2);
priv->fw.aux_icm = mlx4_alloc_icm(dev, aux_pages,
GFP_HIGHUSER | __GFP_NOWARN, 0);
if (!priv->fw.aux_icm) {
mlx4_err(dev, "Couldn't allocate aux memory, aborting.\n");
return -ENOMEM;
}
err = mlx4_MAP_ICM_AUX(dev, priv->fw.aux_icm);
if (err) {
mlx4_err(dev, "MAP_ICM_AUX command failed, aborting.\n");
goto err_free_aux;
}
err = mlx4_init_cmpt_table(dev, init_hca->cmpt_base, dev_cap->cmpt_entry_sz);
if (err) {
mlx4_err(dev, "Failed to map cMPT context memory, aborting.\n");
goto err_unmap_aux;
}
err = mlx4_init_icm_table(dev, &priv->eq_table.table,
init_hca->eqc_base, dev_cap->eqc_entry_sz,
dev->caps.num_eqs, dev->caps.num_eqs,
0, 0);
if (err) {
mlx4_err(dev, "Failed to map EQ context memory, aborting.\n");
goto err_unmap_cmpt;
}
dev->caps.reserved_mtts =
ALIGN(dev->caps.reserved_mtts * dev->caps.mtt_entry_sz,
dma_get_cache_alignment()) / dev->caps.mtt_entry_sz;
err = mlx4_init_icm_table(dev, &priv->mr_table.mtt_table,
init_hca->mtt_base,
dev->caps.mtt_entry_sz,
dev->caps.num_mtt_segs,
dev->caps.reserved_mtts, 1, 0);
if (err) {
mlx4_err(dev, "Failed to map MTT context memory, aborting.\n");
goto err_unmap_eq;
}
err = mlx4_init_icm_table(dev, &priv->mr_table.dmpt_table,
init_hca->dmpt_base,
dev_cap->dmpt_entry_sz,
dev->caps.num_mpts,
dev->caps.reserved_mrws, 1, 1);
if (err) {
mlx4_err(dev, "Failed to map dMPT context memory, aborting.\n");
goto err_unmap_mtt;
}
err = mlx4_init_icm_table(dev, &priv->qp_table.qp_table,
init_hca->qpc_base,
dev_cap->qpc_entry_sz,
dev->caps.num_qps,
dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW],
0, 0);
if (err) {
mlx4_err(dev, "Failed to map QP context memory, aborting.\n");
goto err_unmap_dmpt;
}
err = mlx4_init_icm_table(dev, &priv->qp_table.auxc_table,
init_hca->auxc_base,
dev_cap->aux_entry_sz,
dev->caps.num_qps,
dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW],
0, 0);
if (err) {
mlx4_err(dev, "Failed to map AUXC context memory, aborting.\n");
goto err_unmap_qp;
}
err = mlx4_init_icm_table(dev, &priv->qp_table.altc_table,
init_hca->altc_base,
dev_cap->altc_entry_sz,
dev->caps.num_qps,
dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW],
0, 0);
if (err) {
mlx4_err(dev, "Failed to map ALTC context memory, aborting.\n");
goto err_unmap_auxc;
}
err = mlx4_init_icm_table(dev, &priv->qp_table.rdmarc_table,
init_hca->rdmarc_base,
dev_cap->rdmarc_entry_sz << priv->qp_table.rdmarc_shift,
dev->caps.num_qps,
dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW],
0, 0);
if (err) {
mlx4_err(dev, "Failed to map RDMARC context memory, aborting\n");
goto err_unmap_altc;
}
err = mlx4_init_icm_table(dev, &priv->cq_table.table,
init_hca->cqc_base,
dev_cap->cqc_entry_sz,
dev->caps.num_cqs,
dev->caps.reserved_cqs, 0, 0);
if (err) {
mlx4_err(dev, "Failed to map CQ context memory, aborting.\n");
goto err_unmap_rdmarc;
}
err = mlx4_init_icm_table(dev, &priv->srq_table.table,
init_hca->srqc_base,
dev_cap->srq_entry_sz,
dev->caps.num_srqs,
dev->caps.reserved_srqs, 0, 0);
if (err) {
mlx4_err(dev, "Failed to map SRQ context memory, aborting.\n");
goto err_unmap_cq;
}
err = mlx4_init_icm_table(dev, &priv->mcg_table.table,
init_hca->mc_base, MLX4_MGM_ENTRY_SIZE,
dev->caps.num_mgms + dev->caps.num_amgms,
dev->caps.num_mgms + dev->caps.num_amgms,
0, 0);
if (err) {
mlx4_err(dev, "Failed to map MCG context memory, aborting.\n");
goto err_unmap_srq;
}
return 0;
err_unmap_srq:
mlx4_cleanup_icm_table(dev, &priv->srq_table.table);
err_unmap_cq:
mlx4_cleanup_icm_table(dev, &priv->cq_table.table);
err_unmap_rdmarc:
mlx4_cleanup_icm_table(dev, &priv->qp_table.rdmarc_table);
err_unmap_altc:
mlx4_cleanup_icm_table(dev, &priv->qp_table.altc_table);
err_unmap_auxc:
mlx4_cleanup_icm_table(dev, &priv->qp_table.auxc_table);
err_unmap_qp:
mlx4_cleanup_icm_table(dev, &priv->qp_table.qp_table);
err_unmap_dmpt:
mlx4_cleanup_icm_table(dev, &priv->mr_table.dmpt_table);
err_unmap_mtt:
mlx4_cleanup_icm_table(dev, &priv->mr_table.mtt_table);
err_unmap_eq:
mlx4_cleanup_icm_table(dev, &priv->eq_table.table);
err_unmap_cmpt:
mlx4_cleanup_icm_table(dev, &priv->eq_table.cmpt_table);
mlx4_cleanup_icm_table(dev, &priv->cq_table.cmpt_table);
mlx4_cleanup_icm_table(dev, &priv->srq_table.cmpt_table);
mlx4_cleanup_icm_table(dev, &priv->qp_table.cmpt_table);
err_unmap_aux:
mlx4_UNMAP_ICM_AUX(dev);
err_free_aux:
mlx4_free_icm(dev, priv->fw.aux_icm, 0);
return err;
}
static ssize_t set_port_type(struct device *dev,
struct device_attribute *attr,
const char *buf, size_t count)
{
struct mlx4_port_info *info = container_of(attr, struct mlx4_port_info,
port_attr);
struct mlx4_dev *mdev = info->dev;
struct mlx4_priv *priv = mlx4_priv(mdev);
enum mlx4_port_type types[MLX4_MAX_PORTS];
enum mlx4_port_type new_types[MLX4_MAX_PORTS];
int i;
int err = 0;
if (!strcmp(buf, "ib\n"))
info->tmp_type = MLX4_PORT_TYPE_IB;
else if (!strcmp(buf, "eth\n"))
info->tmp_type = MLX4_PORT_TYPE_ETH;
else if (!strcmp(buf, "auto\n"))
info->tmp_type = MLX4_PORT_TYPE_AUTO;
else {
mlx4_err(mdev, "%s is not supported port type\n", buf);
return -EINVAL;
}
mlx4_stop_sense(mdev);
mutex_lock(&priv->port_mutex);
mdev->caps.possible_type[info->port] = info->tmp_type;
for (i = 0; i < mdev->caps.num_ports; i++) {
types[i] = priv->port[i+1].tmp_type ? priv->port[i+1].tmp_type :
mdev->caps.possible_type[i+1];
if (types[i] == MLX4_PORT_TYPE_AUTO)
types[i] = mdev->caps.port_type[i+1];
}
if (!(mdev->caps.flags & MLX4_DEV_CAP_FLAG_DPDP)) {
for (i = 1; i <= mdev->caps.num_ports; i++) {
if (mdev->caps.possible_type[i] == MLX4_PORT_TYPE_AUTO) {
mdev->caps.possible_type[i] = mdev->caps.port_type[i];
err = -EINVAL;
}
}
}
if (err) {
mlx4_err(mdev, "Auto sensing is not supported on this HCA. "
"Set only 'eth' or 'ib' for both ports "
"(should be the same)\n");
goto out;
}
mlx4_do_sense_ports(mdev, new_types, types);
err = mlx4_check_port_params(mdev, new_types);
if (err)
goto out;
for (i = 0; i < mdev->caps.num_ports; i++)
priv->port[i + 1].tmp_type = 0;
err = mlx4_change_port_types(mdev, new_types);
out:
mlx4_start_sense(mdev);
mutex_unlock(&priv->port_mutex);
return err ? err : count;
}