diff options
Diffstat (limited to 'drivers/net/ethernet/mellanox/mlx4')
23 files changed, 1855 insertions, 287 deletions
diff --git a/drivers/net/ethernet/mellanox/mlx4/Makefile b/drivers/net/ethernet/mellanox/mlx4/Makefile index 3e9c70f15b42..c82217e0d22d 100644 --- a/drivers/net/ethernet/mellanox/mlx4/Makefile +++ b/drivers/net/ethernet/mellanox/mlx4/Makefile @@ -1,7 +1,8 @@ obj-$(CONFIG_MLX4_CORE) += mlx4_core.o -mlx4_core-y := alloc.o catas.o cmd.o cq.o eq.o fw.o icm.o intf.o main.o mcg.o \ - mr.o pd.o port.o profile.o qp.o reset.o sense.o srq.o resource_tracker.o +mlx4_core-y := alloc.o catas.o cmd.o cq.o eq.o fw.o fw_qos.o icm.o intf.o \ + main.o mcg.o mr.o pd.o port.o profile.o qp.o reset.o sense.o \ + srq.o resource_tracker.o obj-$(CONFIG_MLX4_EN) += mlx4_en.o diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c index 546ca4226916..f0fbb4ade85d 100644 --- a/drivers/net/ethernet/mellanox/mlx4/cmd.c +++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c @@ -48,6 +48,7 @@ #include "mlx4.h" #include "fw.h" +#include "fw_qos.h" #define CMD_POLL_TOKEN 0xffff #define INBOX_MASK 0xffffffffffffff00ULL @@ -726,7 +727,8 @@ static int mlx4_cmd_wait(struct mlx4_dev *dev, u64 in_param, u64 *out_param, */ if (op == MLX4_CMD_SET_PORT && (in_modifier == 1 || in_modifier == 2) && - op_modifier == 0 && context->fw_status == CMD_STAT_BAD_SIZE) + op_modifier == MLX4_SET_PORT_IB_OPCODE && + context->fw_status == CMD_STAT_BAD_SIZE) mlx4_dbg(dev, "command 0x%x failed: fw status = 0x%x\n", op, context->fw_status); else @@ -1456,6 +1458,24 @@ static struct mlx4_cmd_info cmd_info[] = { .wrapper = mlx4_CMD_EPERM_wrapper, }, { + .opcode = MLX4_CMD_ALLOCATE_VPP, + .has_inbox = false, + .has_outbox = true, + .out_is_imm = false, + .encode_slave_id = false, + .verify = NULL, + .wrapper = mlx4_CMD_EPERM_wrapper, + }, + { + .opcode = MLX4_CMD_SET_VPORT_QOS, + .has_inbox = false, + .has_outbox = true, + .out_is_imm = false, + .encode_slave_id = false, + .verify = NULL, + .wrapper = mlx4_CMD_EPERM_wrapper, + }, + { .opcode = MLX4_CMD_CONF_SPECIAL_QP, .has_inbox = false, .has_outbox = false, @@ -1500,6 +1520,15 @@ static struct mlx4_cmd_info cmd_info[] = { .verify = NULL, .wrapper = mlx4_ACCESS_REG_wrapper, }, + { + .opcode = MLX4_CMD_CONGESTION_CTRL_OPCODE, + .has_inbox = false, + .has_outbox = false, + .out_is_imm = false, + .encode_slave_id = false, + .verify = NULL, + .wrapper = mlx4_CMD_EPERM_wrapper, + }, /* Native multicast commands are not available for guests */ { .opcode = MLX4_CMD_QP_ATTACH, @@ -1782,7 +1811,8 @@ static int mlx4_master_immediate_activate_vlan_qos(struct mlx4_priv *priv, if (vp_oper->state.default_vlan == vp_admin->default_vlan && vp_oper->state.default_qos == vp_admin->default_qos && - vp_oper->state.link_state == vp_admin->link_state) + vp_oper->state.link_state == vp_admin->link_state && + vp_oper->state.qos_vport == vp_admin->qos_vport) return 0; if (!(priv->mfunc.master.slave_state[slave].active && @@ -1840,6 +1870,7 @@ static int mlx4_master_immediate_activate_vlan_qos(struct mlx4_priv *priv, vp_oper->state.default_vlan = vp_admin->default_vlan; vp_oper->state.default_qos = vp_admin->default_qos; vp_oper->state.link_state = vp_admin->link_state; + vp_oper->state.qos_vport = vp_admin->qos_vport; if (vp_admin->link_state == IFLA_VF_LINK_STATE_DISABLE) work->flags |= MLX4_VF_IMMED_VLAN_FLAG_LINK_DISABLE; @@ -1848,6 +1879,7 @@ static int mlx4_master_immediate_activate_vlan_qos(struct mlx4_priv *priv, work->port = port; work->slave = slave; work->qos = vp_oper->state.default_qos; + work->qos_vport = vp_oper->state.qos_vport; work->vlan_id = vp_oper->state.default_vlan; work->vlan_ix = vp_oper->vlan_idx; work->priv = priv; @@ -1857,6 +1889,63 @@ static int mlx4_master_immediate_activate_vlan_qos(struct mlx4_priv *priv, return 0; } +static void mlx4_set_default_port_qos(struct mlx4_dev *dev, int port) +{ + struct mlx4_qos_manager *port_qos_ctl; + struct mlx4_priv *priv = mlx4_priv(dev); + + port_qos_ctl = &priv->mfunc.master.qos_ctl[port]; + bitmap_zero(port_qos_ctl->priority_bm, MLX4_NUM_UP); + + /* Enable only default prio at PF init routine */ + set_bit(MLX4_DEFAULT_QOS_PRIO, port_qos_ctl->priority_bm); +} + +static void mlx4_allocate_port_vpps(struct mlx4_dev *dev, int port) +{ + int i; + int err; + int num_vfs; + u16 availible_vpp; + u8 vpp_param[MLX4_NUM_UP]; + struct mlx4_qos_manager *port_qos; + struct mlx4_priv *priv = mlx4_priv(dev); + + err = mlx4_ALLOCATE_VPP_get(dev, port, &availible_vpp, vpp_param); + if (err) { + mlx4_info(dev, "Failed query availible VPPs\n"); + return; + } + + port_qos = &priv->mfunc.master.qos_ctl[port]; + num_vfs = (availible_vpp / + bitmap_weight(port_qos->priority_bm, MLX4_NUM_UP)); + + for (i = 0; i < MLX4_NUM_UP; i++) { + if (test_bit(i, port_qos->priority_bm)) + vpp_param[i] = num_vfs; + } + + err = mlx4_ALLOCATE_VPP_set(dev, port, vpp_param); + if (err) { + mlx4_info(dev, "Failed allocating VPPs\n"); + return; + } + + /* Query actual allocated VPP, just to make sure */ + err = mlx4_ALLOCATE_VPP_get(dev, port, &availible_vpp, vpp_param); + if (err) { + mlx4_info(dev, "Failed query availible VPPs\n"); + return; + } + + port_qos->num_of_qos_vfs = num_vfs; + mlx4_dbg(dev, "Port %d Availible VPPs %d\n", port, availible_vpp); + + for (i = 0; i < MLX4_NUM_UP; i++) + mlx4_dbg(dev, "Port %d UP %d Allocated %d VPPs\n", port, i, + vpp_param[i]); +} static int mlx4_master_activate_admin_state(struct mlx4_priv *priv, int slave) { @@ -2204,6 +2293,9 @@ int mlx4_multi_func_init(struct mlx4_dev *dev) } if (mlx4_is_master(dev)) { + struct mlx4_vf_oper_state *vf_oper; + struct mlx4_vf_admin_state *vf_admin; + priv->mfunc.master.slave_state = kzalloc(dev->num_slaves * sizeof(struct mlx4_slave_state), GFP_KERNEL); @@ -2223,6 +2315,8 @@ int mlx4_multi_func_init(struct mlx4_dev *dev) goto err_comm_oper; for (i = 0; i < dev->num_slaves; ++i) { + vf_admin = &priv->mfunc.master.vf_admin[i]; + vf_oper = &priv->mfunc.master.vf_oper[i]; s_state = &priv->mfunc.master.slave_state[i]; s_state->last_cmd = MLX4_COMM_CMD_RESET; mutex_init(&priv->mfunc.master.gen_eqe_mutex[i]); @@ -2234,6 +2328,9 @@ int mlx4_multi_func_init(struct mlx4_dev *dev) &priv->mfunc.comm[i].slave_read); mmiowb(); for (port = 1; port <= MLX4_MAX_PORTS; port++) { + struct mlx4_vport_state *admin_vport; + struct mlx4_vport_state *oper_vport; + s_state->vlan_filter[port] = kzalloc(sizeof(struct mlx4_vlan_fltr), GFP_KERNEL); @@ -2242,15 +2339,30 @@ int mlx4_multi_func_init(struct mlx4_dev *dev) kfree(s_state->vlan_filter[port]); goto err_slaves; } + + admin_vport = &vf_admin->vport[port]; + oper_vport = &vf_oper->vport[port].state; INIT_LIST_HEAD(&s_state->mcast_filters[port]); - priv->mfunc.master.vf_admin[i].vport[port].default_vlan = MLX4_VGT; - priv->mfunc.master.vf_oper[i].vport[port].state.default_vlan = MLX4_VGT; - priv->mfunc.master.vf_oper[i].vport[port].vlan_idx = NO_INDX; - priv->mfunc.master.vf_oper[i].vport[port].mac_idx = NO_INDX; + admin_vport->default_vlan = MLX4_VGT; + oper_vport->default_vlan = MLX4_VGT; + admin_vport->qos_vport = + MLX4_VPP_DEFAULT_VPORT; + oper_vport->qos_vport = MLX4_VPP_DEFAULT_VPORT; + vf_oper->vport[port].vlan_idx = NO_INDX; + vf_oper->vport[port].mac_idx = NO_INDX; } spin_lock_init(&s_state->lock); } + if (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_QOS_VPP) { + for (port = 1; port <= dev->caps.num_ports; port++) { + if (mlx4_is_eth(dev, port)) { + mlx4_set_default_port_qos(dev, port); + mlx4_allocate_port_vpps(dev, port); + } + } + } + memset(&priv->mfunc.master.cmd_eqe, 0, dev->caps.eqe_size); priv->mfunc.master.cmd_eqe.type = MLX4_EVENT_TYPE_CMD; INIT_WORK(&priv->mfunc.master.comm_work, @@ -2671,6 +2783,103 @@ static int mlx4_slaves_closest_port(struct mlx4_dev *dev, int slave, int port) return port; } +static int mlx4_set_vport_qos(struct mlx4_priv *priv, int slave, int port, + int max_tx_rate) +{ + int i; + int err; + struct mlx4_qos_manager *port_qos; + struct mlx4_dev *dev = &priv->dev; + struct mlx4_vport_qos_param vpp_qos[MLX4_NUM_UP]; + + port_qos = &priv->mfunc.master.qos_ctl[port]; + memset(vpp_qos, 0, sizeof(struct mlx4_vport_qos_param) * MLX4_NUM_UP); + + if (slave > port_qos->num_of_qos_vfs) { + mlx4_info(dev, "No availible VPP resources for this VF\n"); + return -EINVAL; + } + + /* Query for default QoS values from Vport 0 is needed */ + err = mlx4_SET_VPORT_QOS_get(dev, port, 0, vpp_qos); + if (err) { + mlx4_info(dev, "Failed to query Vport 0 QoS values\n"); + return err; + } + + for (i = 0; i < MLX4_NUM_UP; i++) { + if (test_bit(i, port_qos->priority_bm) && max_tx_rate) { + vpp_qos[i].max_avg_bw = max_tx_rate; + vpp_qos[i].enable = 1; + } else { + /* if user supplied tx_rate == 0, meaning no rate limit + * configuration is required. so we are leaving the + * value of max_avg_bw as queried from Vport 0. + */ + vpp_qos[i].enable = 0; + } + } + + err = mlx4_SET_VPORT_QOS_set(dev, port, slave, vpp_qos); + if (err) { + mlx4_info(dev, "Failed to set Vport %d QoS values\n", slave); + return err; + } + + return 0; +} + +static bool mlx4_is_vf_vst_and_prio_qos(struct mlx4_dev *dev, int port, + struct mlx4_vport_state *vf_admin) +{ + struct mlx4_qos_manager *info; + struct mlx4_priv *priv = mlx4_priv(dev); + + if (!mlx4_is_master(dev) || + !(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_QOS_VPP)) + return false; + + info = &priv->mfunc.master.qos_ctl[port]; + + if (vf_admin->default_vlan != MLX4_VGT && + test_bit(vf_admin->default_qos, info->priority_bm)) + return true; + + return false; +} + +static bool mlx4_valid_vf_state_change(struct mlx4_dev *dev, int port, + struct mlx4_vport_state *vf_admin, + int vlan, int qos) +{ + struct mlx4_vport_state dummy_admin = {0}; + + if (!mlx4_is_vf_vst_and_prio_qos(dev, port, vf_admin) || + !vf_admin->tx_rate) + return true; + + dummy_admin.default_qos = qos; + dummy_admin.default_vlan = vlan; + + /* VF wants to move to other VST state which is valid with current + * rate limit. Either differnt default vlan in VST or other + * supported QoS priority. Otherwise we don't allow this change when + * the TX rate is still configured. + */ + if (mlx4_is_vf_vst_and_prio_qos(dev, port, &dummy_admin)) + return true; + + mlx4_info(dev, "Cannot change VF state to %s while rate is set\n", + (vlan == MLX4_VGT) ? "VGT" : "VST"); + + if (vlan != MLX4_VGT) + mlx4_info(dev, "VST priority %d not supported for QoS\n", qos); + + mlx4_info(dev, "Please set rate to 0 prior to this VF state change\n"); + + return false; +} + int mlx4_set_vf_mac(struct mlx4_dev *dev, int port, int vf, u64 mac) { struct mlx4_priv *priv = mlx4_priv(dev); @@ -2714,12 +2923,22 @@ int mlx4_set_vf_vlan(struct mlx4_dev *dev, int port, int vf, u16 vlan, u8 qos) port = mlx4_slaves_closest_port(dev, slave, port); vf_admin = &priv->mfunc.master.vf_admin[slave].vport[port]; + if (!mlx4_valid_vf_state_change(dev, port, vf_admin, vlan, qos)) + return -EPERM; + if ((0 == vlan) && (0 == qos)) vf_admin->default_vlan = MLX4_VGT; else vf_admin->default_vlan = vlan; vf_admin->default_qos = qos; + /* If rate was configured prior to VST, we saved the configured rate + * in vf_admin->rate and now, if priority supported we enforce the QoS + */ + if (mlx4_is_vf_vst_and_prio_qos(dev, port, vf_admin) && + vf_admin->tx_rate) + vf_admin->qos_vport = slave; + if (mlx4_master_immediate_activate_vlan_qos(priv, slave, port)) mlx4_info(dev, "updating vf %d port %d config will take effect on next VF restart\n", @@ -2728,6 +2947,69 @@ int mlx4_set_vf_vlan(struct mlx4_dev *dev, int port, int vf, u16 vlan, u8 qos) } EXPORT_SYMBOL_GPL(mlx4_set_vf_vlan); +int mlx4_set_vf_rate(struct mlx4_dev *dev, int port, int vf, int min_tx_rate, + int max_tx_rate) +{ + int err; + int slave; + struct mlx4_vport_state *vf_admin; + struct mlx4_priv *priv = mlx4_priv(dev); + + if (!mlx4_is_master(dev) || + !(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_QOS_VPP)) + return -EPROTONOSUPPORT; + + if (min_tx_rate) { + mlx4_info(dev, "Minimum BW share not supported\n"); + return -EPROTONOSUPPORT; + } + + slave = mlx4_get_slave_indx(dev, vf); + if (slave < 0) + return -EINVAL; + + port = mlx4_slaves_closest_port(dev, slave, port); + vf_admin = &priv->mfunc.master.vf_admin[slave].vport[port]; + + err = mlx4_set_vport_qos(priv, slave, port, max_tx_rate); + if (err) { + mlx4_info(dev, "vf %d failed to set rate %d\n", vf, + max_tx_rate); + return err; + } + + vf_admin->tx_rate = max_tx_rate; + /* if VF is not in supported mode (VST with supported prio), + * we do not change vport configuration for its QPs, but save + * the rate, so it will be enforced when it moves to supported + * mode next time. + */ + if (!mlx4_is_vf_vst_and_prio_qos(dev, port, vf_admin)) { + mlx4_info(dev, + "rate set for VF %d when not in valid state\n", vf); + + if (vf_admin->default_vlan != MLX4_VGT) + mlx4_info(dev, "VST priority not supported by QoS\n"); + else + mlx4_info(dev, "VF in VGT mode (needed VST)\n"); + + mlx4_info(dev, + "rate %d take affect when VF moves to valid state\n", + max_tx_rate); + return 0; + } + + /* If user sets rate 0 assigning default vport for its QPs */ + vf_admin->qos_vport = max_tx_rate ? slave : MLX4_VPP_DEFAULT_VPORT; + + if (priv->mfunc.master.slave_state[slave].active && + dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_UPDATE_QP) + mlx4_master_immediate_activate_vlan_qos(priv, slave, port); + + return 0; +} +EXPORT_SYMBOL_GPL(mlx4_set_vf_rate); + /* mlx4_get_slave_default_vlan - * return true if VST ( default vlan) * if VST, will return vlan & qos (if not NULL) @@ -2801,7 +3083,12 @@ int mlx4_get_vf_config(struct mlx4_dev *dev, int port, int vf, struct ifla_vf_in ivf->vlan = s_info->default_vlan; ivf->qos = s_info->default_qos; - ivf->max_tx_rate = s_info->tx_rate; + + if (mlx4_is_vf_vst_and_prio_qos(dev, port, s_info)) + ivf->max_tx_rate = s_info->tx_rate; + else + ivf->max_tx_rate = 0; + ivf->min_tx_rate = 0; ivf->spoofchk = s_info->spoofchk; ivf->linkstate = s_info->link_state; diff --git a/drivers/net/ethernet/mellanox/mlx4/en_clock.c b/drivers/net/ethernet/mellanox/mlx4/en_clock.c index 90b5309cdb5c..8a083d73efdb 100644 --- a/drivers/net/ethernet/mellanox/mlx4/en_clock.c +++ b/drivers/net/ethernet/mellanox/mlx4/en_clock.c @@ -164,20 +164,19 @@ static int mlx4_en_phc_adjtime(struct ptp_clock_info *ptp, s64 delta) * Read the timecounter and return the correct value in ns after converting * it into a struct timespec. **/ -static int mlx4_en_phc_gettime(struct ptp_clock_info *ptp, struct timespec *ts) +static int mlx4_en_phc_gettime(struct ptp_clock_info *ptp, + struct timespec64 *ts) { struct mlx4_en_dev *mdev = container_of(ptp, struct mlx4_en_dev, ptp_clock_info); unsigned long flags; - u32 remainder; u64 ns; write_lock_irqsave(&mdev->clock_lock, flags); ns = timecounter_read(&mdev->clock); write_unlock_irqrestore(&mdev->clock_lock, flags); - ts->tv_sec = div_u64_rem(ns, NSEC_PER_SEC, &remainder); - ts->tv_nsec = remainder; + *ts = ns_to_timespec64(ns); return 0; } @@ -191,11 +190,11 @@ static int mlx4_en_phc_gettime(struct ptp_clock_info *ptp, struct timespec *ts) * wall timer value. **/ static int mlx4_en_phc_settime(struct ptp_clock_info *ptp, - const struct timespec *ts) + const struct timespec64 *ts) { struct mlx4_en_dev *mdev = container_of(ptp, struct mlx4_en_dev, ptp_clock_info); - u64 ns = timespec_to_ns(ts); + u64 ns = timespec64_to_ns(ts); unsigned long flags; /* reset the timecounter */ @@ -232,8 +231,8 @@ static const struct ptp_clock_info mlx4_en_ptp_clock_info = { .pps = 0, .adjfreq = mlx4_en_phc_adjfreq, .adjtime = mlx4_en_phc_adjtime, - .gettime = mlx4_en_phc_gettime, - .settime = mlx4_en_phc_settime, + .gettime64 = mlx4_en_phc_gettime, + .settime64 = mlx4_en_phc_settime, .enable = mlx4_en_phc_enable, }; diff --git a/drivers/net/ethernet/mellanox/mlx4/en_dcb_nl.c b/drivers/net/ethernet/mellanox/mlx4/en_dcb_nl.c index c95ca252187c..f01918c63f28 100644 --- a/drivers/net/ethernet/mellanox/mlx4/en_dcb_nl.c +++ b/drivers/net/ethernet/mellanox/mlx4/en_dcb_nl.c @@ -35,6 +35,50 @@ #include <linux/math64.h> #include "mlx4_en.h" +#include "fw_qos.h" + +/* Definitions for QCN + */ + +struct mlx4_congestion_control_mb_prio_802_1_qau_params { + __be32 modify_enable_high; + __be32 modify_enable_low; + __be32 reserved1; + __be32 extended_enable; + __be32 rppp_max_rps; + __be32 rpg_time_reset; + __be32 rpg_byte_reset; + __be32 rpg_threshold; + __be32 rpg_max_rate; + __be32 rpg_ai_rate; + __be32 rpg_hai_rate; + __be32 rpg_gd; + __be32 rpg_min_dec_fac; + __be32 rpg_min_rate; + __be32 max_time_rise; + __be32 max_byte_rise; + __be32 max_qdelta; + __be32 min_qoffset; + __be32 gd_coefficient; + __be32 reserved2[5]; + __be32 cp_sample_base; + __be32 reserved3[39]; +}; + +struct mlx4_congestion_control_mb_prio_802_1_qau_statistics { + __be64 rppp_rp_centiseconds; + __be32 reserved1; + __be32 ignored_cnm; + __be32 rppp_created_rps; + __be32 estimated_total_rate; + __be32 max_active_rate_limiter_index; + __be32 dropped_cnms_busy_fw; + __be32 reserved2; + __be32 cnms_handled_successfully; + __be32 min_total_limiters_rate; + __be32 max_total_limiters_rate; + __be32 reserved3[4]; +}; static int mlx4_en_dcbnl_ieee_getets(struct net_device *dev, struct ieee_ets *ets) @@ -183,6 +227,10 @@ static int mlx4_en_dcbnl_ieee_setpfc(struct net_device *dev, prof->rx_ppp); if (err) en_err(priv, "Failed setting pause params\n"); + else + mlx4_en_update_pfc_stats_bitmap(mdev->dev, &priv->stats_bitmap, + prof->rx_ppp, prof->rx_pause, + prof->tx_ppp, prof->tx_pause); return err; } @@ -242,6 +290,178 @@ static int mlx4_en_dcbnl_ieee_setmaxrate(struct net_device *dev, return 0; } +#define RPG_ENABLE_BIT 31 +#define CN_TAG_BIT 30 + +static int mlx4_en_dcbnl_ieee_getqcn(struct net_device *dev, + struct ieee_qcn *qcn) +{ + struct mlx4_en_priv *priv = netdev_priv(dev); + struct mlx4_congestion_control_mb_prio_802_1_qau_params *hw_qcn; + struct mlx4_cmd_mailbox *mailbox_out = NULL; + u64 mailbox_in_dma = 0; + u32 inmod = 0; + int i, err; + + if (!(priv->mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_QCN)) + return -EOPNOTSUPP; + + mailbox_out = mlx4_alloc_cmd_mailbox(priv->mdev->dev); + if (IS_ERR(mailbox_out)) + return -ENOMEM; + hw_qcn = + (struct mlx4_congestion_control_mb_prio_802_1_qau_params *) + mailbox_out->buf; + + for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) { + inmod = priv->port | ((1 << i) << 8) | + (MLX4_CTRL_ALGO_802_1_QAU_REACTION_POINT << 16); + err = mlx4_cmd_box(priv->mdev->dev, mailbox_in_dma, + mailbox_out->dma, + inmod, MLX4_CONGESTION_CONTROL_GET_PARAMS, + MLX4_CMD_CONGESTION_CTRL_OPCODE, + MLX4_CMD_TIME_CLASS_C, + MLX4_CMD_NATIVE); + if (err) { + mlx4_free_cmd_mailbox(priv->mdev->dev, mailbox_out); + return err; + } + + qcn->rpg_enable[i] = + be32_to_cpu(hw_qcn->extended_enable) >> RPG_ENABLE_BIT; + qcn->rppp_max_rps[i] = + be32_to_cpu(hw_qcn->rppp_max_rps); + qcn->rpg_time_reset[i] = + be32_to_cpu(hw_qcn->rpg_time_reset); + qcn->rpg_byte_reset[i] = + be32_to_cpu(hw_qcn->rpg_byte_reset); + qcn->rpg_threshold[i] = + be32_to_cpu(hw_qcn->rpg_threshold); + qcn->rpg_max_rate[i] = + be32_to_cpu(hw_qcn->rpg_max_rate); + qcn->rpg_ai_rate[i] = + be32_to_cpu(hw_qcn->rpg_ai_rate); + qcn->rpg_hai_rate[i] = + be32_to_cpu(hw_qcn->rpg_hai_rate); + qcn->rpg_gd[i] = + be32_to_cpu(hw_qcn->rpg_gd); + qcn->rpg_min_dec_fac[i] = + be32_to_cpu(hw_qcn->rpg_min_dec_fac); + qcn->rpg_min_rate[i] = + be32_to_cpu(hw_qcn->rpg_min_rate); + qcn->cndd_state_machine[i] = + priv->cndd_state[i]; + } + mlx4_free_cmd_mailbox(priv->mdev->dev, mailbox_out); + return 0; +} + +static int mlx4_en_dcbnl_ieee_setqcn(struct net_device *dev, + struct ieee_qcn *qcn) +{ + struct mlx4_en_priv *priv = netdev_priv(dev); + struct mlx4_congestion_control_mb_prio_802_1_qau_params *hw_qcn; + struct mlx4_cmd_mailbox *mailbox_in = NULL; + u64 mailbox_in_dma = 0; + u32 inmod = 0; + int i, err; +#define MODIFY_ENABLE_HIGH_MASK 0xc0000000 +#define MODIFY_ENABLE_LOW_MASK 0xffc00000 + + if (!(priv->mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_QCN)) + return -EOPNOTSUPP; + + mailbox_in = mlx4_alloc_cmd_mailbox(priv->mdev->dev); + if (IS_ERR(mailbox_in)) + return -ENOMEM; + + mailbox_in_dma = mailbox_in->dma; + hw_qcn = + (struct mlx4_congestion_control_mb_prio_802_1_qau_params *)mailbox_in->buf; + for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) { + inmod = priv->port | ((1 << i) << 8) | + (MLX4_CTRL_ALGO_802_1_QAU_REACTION_POINT << 16); + + /* Before updating QCN parameter, + * need to set it's modify enable bit to 1 + */ + + hw_qcn->modify_enable_high = cpu_to_be32( + MODIFY_ENABLE_HIGH_MASK); + hw_qcn->modify_enable_low = cpu_to_be32(MODIFY_ENABLE_LOW_MASK); + + hw_qcn->extended_enable = cpu_to_be32(qcn->rpg_enable[i] << RPG_ENABLE_BIT); + hw_qcn->rppp_max_rps = cpu_to_be32(qcn->rppp_max_rps[i]); + hw_qcn->rpg_time_reset = cpu_to_be32(qcn->rpg_time_reset[i]); + hw_qcn->rpg_byte_reset = cpu_to_be32(qcn->rpg_byte_reset[i]); + hw_qcn->rpg_threshold = cpu_to_be32(qcn->rpg_threshold[i]); + hw_qcn->rpg_max_rate = cpu_to_be32(qcn->rpg_max_rate[i]); + hw_qcn->rpg_ai_rate = cpu_to_be32(qcn->rpg_ai_rate[i]); + hw_qcn->rpg_hai_rate = cpu_to_be32(qcn->rpg_hai_rate[i]); + hw_qcn->rpg_gd = cpu_to_be32(qcn->rpg_gd[i]); + hw_qcn->rpg_min_dec_fac = cpu_to_be32(qcn->rpg_min_dec_fac[i]); + hw_qcn->rpg_min_rate = cpu_to_be32(qcn->rpg_min_rate[i]); + priv->cndd_state[i] = qcn->cndd_state_machine[i]; + if (qcn->cndd_state_machine[i] == DCB_CNDD_INTERIOR_READY) + hw_qcn->extended_enable |= cpu_to_be32(1 << CN_TAG_BIT); + + err = mlx4_cmd(priv->mdev->dev, mailbox_in_dma, inmod, + MLX4_CONGESTION_CONTROL_SET_PARAMS, + MLX4_CMD_CONGESTION_CTRL_OPCODE, + MLX4_CMD_TIME_CLASS_C, + MLX4_CMD_NATIVE); + if (err) { + mlx4_free_cmd_mailbox(priv->mdev->dev, mailbox_in); + return err; + } + } + mlx4_free_cmd_mailbox(priv->mdev->dev, mailbox_in); + return 0; +} + +static int mlx4_en_dcbnl_ieee_getqcnstats(struct net_device *dev, + struct ieee_qcn_stats *qcn_stats) +{ + struct mlx4_en_priv *priv = netdev_priv(dev); + struct mlx4_congestion_control_mb_prio_802_1_qau_statistics *hw_qcn_stats; + struct mlx4_cmd_mailbox *mailbox_out = NULL; + u64 mailbox_in_dma = 0; + u32 inmod = 0; + int i, err; + + if (!(priv->mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_QCN)) + return -EOPNOTSUPP; + + mailbox_out = mlx4_alloc_cmd_mailbox(priv->mdev->dev); + if (IS_ERR(mailbox_out)) + return -ENOMEM; + + hw_qcn_stats = + (struct mlx4_congestion_control_mb_prio_802_1_qau_statistics *) + mailbox_out->buf; + + for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) { + inmod = priv->port | ((1 << i) << 8) | + (MLX4_CTRL_ALGO_802_1_QAU_REACTION_POINT << 16); + err = mlx4_cmd_box(priv->mdev->dev, mailbox_in_dma, + mailbox_out->dma, inmod, + MLX4_CONGESTION_CONTROL_GET_STATISTICS, + MLX4_CMD_CONGESTION_CTRL_OPCODE, + MLX4_CMD_TIME_CLASS_C, + MLX4_CMD_NATIVE); + if (err) { + mlx4_free_cmd_mailbox(priv->mdev->dev, mailbox_out); + return err; + } + qcn_stats->rppp_rp_centiseconds[i] = + be64_to_cpu(hw_qcn_stats->rppp_rp_centiseconds); + qcn_stats->rppp_created_rps[i] = + be32_to_cpu(hw_qcn_stats->rppp_created_rps); + } + mlx4_free_cmd_mailbox(priv->mdev->dev, mailbox_out); + return 0; +} + const struct dcbnl_rtnl_ops mlx4_en_dcbnl_ops = { .ieee_getets = mlx4_en_dcbnl_ieee_getets, .ieee_setets = mlx4_en_dcbnl_ieee_setets, @@ -252,6 +472,9 @@ const struct dcbnl_rtnl_ops mlx4_en_dcbnl_ops = { .getdcbx = mlx4_en_dcbnl_getdcbx, .setdcbx = mlx4_en_dcbnl_setdcbx, + .ieee_getqcn = mlx4_en_dcbnl_ieee_getqcn, + .ieee_setqcn = mlx4_en_dcbnl_ieee_setqcn, + .ieee_getqcnstats = mlx4_en_dcbnl_ieee_getqcnstats, }; const struct dcbnl_rtnl_ops mlx4_en_dcbnl_pfc_ops = { diff --git a/drivers/net/ethernet/mellanox/mlx4/en_ethtool.c b/drivers/net/ethernet/mellanox/mlx4/en_ethtool.c index a7b58ba8492b..3f44e2bbb982 100644 --- a/drivers/net/ethernet/mellanox/mlx4/en_ethtool.c +++ b/drivers/net/ethernet/mellanox/mlx4/en_ethtool.c @@ -38,6 +38,7 @@ #include <linux/mlx4/device.h> #include <linux/in.h> #include <net/ip.h> +#include <linux/bitmap.h> #include "mlx4_en.h" #include "en_port.h" @@ -104,6 +105,7 @@ static const char mlx4_en_priv_flags[][ETH_GSTRING_LEN] = { }; static const char main_strings[][ETH_GSTRING_LEN] = { + /* main statistics */ "rx_packets", "tx_packets", "rx_bytes", "tx_bytes", "rx_errors", "tx_errors", "rx_dropped", "tx_dropped", "multicast", "collisions", "rx_length_errors", "rx_over_errors", "rx_crc_errors", @@ -117,14 +119,76 @@ static const char main_strings[][ETH_GSTRING_LEN] = { "queue_stopped", "wake_queue", "tx_timeout", "rx_alloc_failed", "rx_csum_good", "rx_csum_none", "rx_csum_complete", "tx_chksum_offload", + /* priority flow control statistics rx */ + "rx_pause_prio_0", "rx_pause_duration_prio_0", + "rx_pause_transition_prio_0", + "rx_pause_prio_1", "rx_pause_duration_prio_1", + "rx_pause_transition_prio_1", + "rx_pause_prio_2", "rx_pause_duration_prio_2", + "rx_pause_transition_prio_2", + "rx_pause_prio_3", "rx_pause_duration_prio_3", + "rx_pause_transition_prio_3", + "rx_pause_prio_4", "rx_pause_duration_prio_4", + "rx_pause_transition_prio_4", + "rx_pause_prio_5", "rx_pause_duration_prio_5", + "rx_pause_transition_prio_5", + "rx_pause_prio_6", "rx_pause_duration_prio_6", + "rx_pause_transition_prio_6", + "rx_pause_prio_7", "rx_pause_duration_prio_7", + "rx_pause_transition_prio_7", + + /* flow control statistics rx */ + "rx_pause", "rx_pause_duration", "rx_pause_transition", + + /* priority flow control statistics tx */ + "tx_pause_prio_0", "tx_pause_duration_prio_0", + "tx_pause_transition_prio_0", + "tx_pause_prio_1", "tx_pause_duration_prio_1", + "tx_pause_transition_prio_1", + "tx_pause_prio_2", "tx_pause_duration_prio_2", + "tx_pause_transition_prio_2", + "tx_pause_prio_3", "tx_pause_duration_prio_3", + "tx_pause_transition_prio_3", + "tx_pause_prio_4", "tx_pause_duration_prio_4", + "tx_pause_transition_prio_4", + "tx_pause_prio_5", "tx_pause_duration_prio_5", + "tx_pause_transition_prio_5", + "tx_pause_prio_6", "tx_pause_duration_prio_6", + "tx_pause_transition_prio_6", + "tx_pause_prio_7", "tx_pause_duration_prio_7", + "tx_pause_transition_prio_7", + + /* flow control statistics tx */ + "tx_pause", "tx_pause_duration", "tx_pause_transition", + /* packet statistics */ - "broadcast", "rx_prio_0", "rx_prio_1", "rx_prio_2", "rx_prio_3", - "rx_prio_4", "rx_prio_5", "rx_prio_6", "rx_prio_7", "tx_prio_0", - "tx_prio_1", "tx_prio_2", "tx_prio_3", "tx_prio_4", "tx_prio_5", - "tx_prio_6", "tx_prio_7", + "rx_multicast_packets", + "rx_broadcast_packets", + "rx_jabbers", + "rx_in_range_length_error", + "rx_out_range_length_error", + "tx_multicast_packets", + "tx_broadcast_packets", + "rx_prio_0_packets", "rx_prio_0_bytes", + "rx_prio_1_packets", "rx_prio_1_bytes", + "rx_prio_2_packets", "rx_prio_2_bytes", + "rx_prio_3_packets", "rx_prio_3_bytes", + "rx_prio_4_packets", "rx_prio_4_bytes", + "rx_prio_5_packets", "rx_prio_5_bytes", + "rx_prio_6_packets", "rx_prio_6_bytes", + "rx_prio_7_packets", "rx_prio_7_bytes", + "rx_novlan_packets", "rx_novlan_bytes", + "tx_prio_0_packets", "tx_prio_0_bytes", + "tx_prio_1_packets", "tx_prio_1_bytes", + "tx_prio_2_packets", "tx_prio_2_bytes", + "tx_prio_3_packets", "tx_prio_3_bytes", + "tx_prio_4_packets", "tx_prio_4_bytes", + "tx_prio_5_packets", "tx_prio_5_bytes", + "tx_prio_6_packets", "tx_prio_6_bytes", + "tx_prio_7_packets", "tx_prio_7_bytes", + "tx_novlan_packets", "tx_novlan_bytes", + }; -#define NUM_MAIN_STATS 21 -#define NUM_ALL_STATS (NUM_MAIN_STATS + NUM_PORT_STATS + NUM_PKT_STATS + NUM_PERF_STATS) static const char mlx4_en_test_names[][ETH_GSTRING_LEN]= { "Interrupt Test", @@ -224,14 +288,50 @@ static int mlx4_en_set_wol(struct net_device *netdev, return err; } +struct bitmap_iterator { + unsigned long *stats_bitmap; + unsigned int count; + unsigned int iterator; + bool advance_array; /* if set, force no increments */ +}; + +static inline void bitmap_iterator_init(struct bitmap_iterator *h, + unsigned long *stats_bitmap, + int count) +{ + h->iterator = 0; + h->advance_array = !bitmap_empty(stats_bitmap, count); + h->count = h->advance_array ? bitmap_weight(stats_bitmap, count) + : count; + h->stats_bitmap = stats_bitmap; +} + +static inline int bitmap_iterator_test(struct bitmap_iterator *h) +{ + return !h->advance_array ? 1 : test_bit(h->iterator, h->stats_bitmap); +} + +static inline int bitmap_iterator_inc(struct bitmap_iterator *h) +{ + return h->iterator++; +} + +static inline unsigned int +bitmap_iterator_count(struct bitmap_iterator *h) +{ + return h->count; +} + static int mlx4_en_get_sset_count(struct net_device *dev, int sset) { struct mlx4_en_priv *priv = netdev_priv(dev); - int bit_count = hweight64(priv->stats_bitmap); + struct bitmap_iterator it; + + bitmap_iterator_init(&it, priv->stats_bitmap.bitmap, NUM_ALL_STATS); switch (sset) { case ETH_SS_STATS: - return (priv->stats_bitmap ? bit_count : NUM_ALL_STATS) + + return bitmap_iterator_count(&it) + (priv->tx_ring_num * 2) + #ifdef CONFIG_NET_RX_BUSY_POLL (priv->rx_ring_num * 5); @@ -253,34 +353,45 @@ static void mlx4_en_get_ethtool_stats(struct net_device *dev, { struct mlx4_en_priv *priv = netdev_priv(dev); int index = 0; - int i, j = 0; + int i; + struct bitmap_iterator it; + + bitmap_iterator_init(&it, priv->stats_bitmap.bitmap, NUM_ALL_STATS); spin_lock_bh(&priv->stats_lock); - if (!(priv->stats_bitmap)) { - for (i = 0; i < NUM_MAIN_STATS; i++) - data[index++] = - ((unsigned long *) &priv->stats)[i]; - for (i = 0; i < NUM_PORT_STATS; i++) + for (i = 0; i < NUM_MAIN_STATS; i++, bitmap_iterator_inc(&it)) + if (bitmap_iterator_test(&it)) + data[index++] = ((unsigned long *)&priv->stats)[i]; + + for (i = 0; i < NUM_PORT_STATS; i++, bitmap_iterator_inc(&it)) + if (bitmap_iterator_test(&it)) + data[index++] = ((unsigned long *)&priv->port_stats)[i]; + + for (i = 0; i < NUM_FLOW_PRIORITY_STATS_RX; + i++, bitmap_iterator_inc(&it)) + if (bitmap_iterator_test(&it)) data[index++] = - ((unsigned long *) &priv->port_stats)[i]; - for (i = 0; i < NUM_PKT_STATS; i++) + ((u64 *)&priv->rx_priority_flowstats)[i]; + + for (i = 0; i < NUM_FLOW_STATS_RX; i++, bitmap_iterator_inc(&it)) + if (bitmap_iterator_test(&it)) + data[index++] = ((u64 *)&priv->rx_flowstats)[i]; + + for (i = 0; i < NUM_FLOW_PRIORITY_STATS_TX; + i++, bitmap_iterator_inc(&it)) + if (bitmap_iterator_test(&it)) data[index++] = - ((unsigned long *) &priv->pkstats)[i]; - } else { - for (i = 0; i < NUM_MAIN_STATS; i++) { - if ((priv->stats_bitmap >> j) & 1) - data[index++] = - ((unsigned long *) &priv->stats)[i]; - j++; - } - for (i = 0; i < NUM_PORT_STATS; i++) { - if ((priv->stats_bitmap >> j) & 1) - data[index++] = - ((unsigned long *) &priv->port_stats)[i]; - j++; - } - } + ((u64 *)&priv->tx_priority_flowstats)[i]; + + for (i = 0; i < NUM_FLOW_STATS_TX; i++, bitmap_iterator_inc(&it)) + if (bitmap_iterator_test(&it)) + data[index++] = ((u64 *)&priv->tx_flowstats)[i]; + + for (i = 0; i < NUM_PKT_STATS; i++, bitmap_iterator_inc(&it)) + if (bitmap_iterator_test(&it)) + data[index++] = ((unsigned long *)&priv->pkstats)[i]; + for (i = 0; i < priv->tx_ring_num; i++) { data[index++] = priv->tx_ring[i]->packets; data[index++] = priv->tx_ring[i]->bytes; @@ -309,7 +420,10 @@ static void mlx4_en_get_strings(struct net_device *dev, { struct mlx4_en_priv *priv = netdev_priv(dev); int index = 0; - int i; + int i, strings = 0; + struct bitmap_iterator it; + + bitmap_iterator_init(&it, priv->stats_bitmap.bitmap, NUM_ALL_STATS); switch (stringset) { case ETH_SS_TEST: @@ -322,29 +436,30 @@ static void mlx4_en_get_strings(struct net_device *dev, case ETH_SS_STATS: /* Add main counters */ - if (!priv->stats_bitmap) { - for (i = 0; i < NUM_MAIN_STATS; i++) + for (i = 0; i < NUM_MAIN_STATS; i++, strings++, + bitmap_iterator_inc(&it)) + if (bitmap_iterator_test(&it)) + strcpy(data + (index++) * ETH_GSTRING_LEN, + main_strings[strings]); + + for (i = 0; i < NUM_PORT_STATS; i++, strings++, + bitmap_iterator_inc(&it)) + if (bitmap_iterator_test(&it)) strcpy(data + (index++) * ETH_GSTRING_LEN, - main_strings[i]); - for (i = 0; i < NUM_PORT_STATS; i++) + main_strings[strings]); + + for (i = 0; i < NUM_FLOW_STATS; i++, strings++, + bitmap_iterator_inc(&it)) + if (bitmap_iterator_test(&it)) strcpy(data + (index++) * ETH_GSTRING_LEN, - main_strings[i + - NUM_MAIN_STATS]); - for (i = 0; i < NUM_PKT_STATS; i++) + main_strings[strings]); + + for (i = 0; i < NUM_PKT_STATS; i++, strings++, + bitmap_iterator_inc(&it)) + if (bitmap_iterator_test(&it)) strcpy(data + (index++) * ETH_GSTRING_LEN, - main_strings[i + - NUM_MAIN_STATS + - NUM_PORT_STATS]); - } else - for (i = 0; i < NUM_MAIN_STATS + NUM_PORT_STATS; i++) { - if ((priv->stats_bitmap >> i) & 1) { - strcpy(data + - (index++) * ETH_GSTRING_LEN, - main_strings[i]); - } - if (!(priv->stats_bitmap >> i)) - break; - } + main_strings[strings]); + for (i = 0; i < priv->tx_ring_num; i++) { sprintf(data + (index++) * ETH_GSTRING_LEN, "tx%d_packets", i); @@ -885,6 +1000,12 @@ static int mlx4_en_set_pauseparam(struct net_device *dev, priv->prof->rx_ppp); if (err) en_err(priv, "Failed setting pause params\n"); + else + mlx4_en_update_pfc_stats_bitmap(mdev->dev, &priv->stats_bitmap, + priv->prof->rx_ppp, + priv->prof->rx_pause, + priv->prof->tx_ppp, + priv->prof->tx_pause); return err; } @@ -1818,6 +1939,32 @@ static int mlx4_en_get_module_eeprom(struct net_device *dev, return 0; } +static int mlx4_en_set_phys_id(struct net_device *dev, + enum ethtool_phys_id_state state) +{ + int err; + u16 beacon_duration; + struct mlx4_en_priv *priv = netdev_priv(dev); + struct mlx4_en_dev *mdev = priv->mdev; + + if (!(mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_PORT_BEACON)) + return -EOPNOTSUPP; + + switch (state) { + case ETHTOOL_ID_ACTIVE: + beacon_duration = PORT_BEACON_MAX_LIMIT; + break; + case ETHTOOL_ID_INACTIVE: + beacon_duration = 0; + break; + default: + return -EOPNOTSUPP; + } + + err = mlx4_SET_PORT_BEACON(mdev->dev, priv->port, beacon_duration); + return err; +} + const struct ethtool_ops mlx4_en_ethtool_ops = { .get_drvinfo = mlx4_en_get_drvinfo, .get_settings = mlx4_en_get_settings, @@ -1827,6 +1974,7 @@ const struct ethtool_ops mlx4_en_ethtool_ops = { .get_sset_count = mlx4_en_get_sset_count, .get_ethtool_stats = mlx4_en_get_ethtool_stats, .self_test = mlx4_en_self_test, + .set_phys_id = mlx4_en_set_phys_id, .get_wol = mlx4_en_get_wol, .set_wol = mlx4_en_set_wol, .get_msglevel = mlx4_en_get_msglevel, diff --git a/drivers/net/ethernet/mellanox/mlx4/en_main.c b/drivers/net/ethernet/mellanox/mlx4/en_main.c index 58d5a07d0ff4..913b716ed2e1 100644 --- a/drivers/net/ethernet/mellanox/mlx4/en_main.c +++ b/drivers/net/ethernet/mellanox/mlx4/en_main.c @@ -103,6 +103,11 @@ void mlx4_en_update_loopback_state(struct net_device *dev, { struct mlx4_en_priv *priv = netdev_priv(dev); + if (features & NETIF_F_LOOPBACK) + priv->ctrl_flags |= cpu_to_be32(MLX4_WQE_CTRL_FORCE_LOOPBACK); + else + priv->ctrl_flags &= cpu_to_be32(~MLX4_WQE_CTRL_FORCE_LOOPBACK); + priv->flags &= ~(MLX4_EN_FLAG_RX_FILTER_NEEDED| MLX4_EN_FLAG_ENABLE_HW_LOOPBACK); diff --git a/drivers/net/ethernet/mellanox/mlx4/en_netdev.c b/drivers/net/ethernet/mellanox/mlx4/en_netdev.c index 3485acf03014..0f1afc085d58 100644 --- a/drivers/net/ethernet/mellanox/mlx4/en_netdev.c +++ b/drivers/net/ethernet/mellanox/mlx4/en_netdev.c @@ -1685,7 +1685,7 @@ int mlx4_en_start_port(struct net_device *dev) } /* Attach rx QP to bradcast address */ - memset(&mc_list[10], 0xff, ETH_ALEN); + eth_broadcast_addr(&mc_list[10]); mc_list[5] = priv->port; /* needed for B0 steering support */ if (mlx4_multicast_attach(mdev->dev, &priv->rss_map.indir_qp, mc_list, priv->port, 0, MLX4_PROT_ETH, @@ -1786,7 +1786,7 @@ void mlx4_en_stop_port(struct net_device *dev, int detach) } /* Detach All multicasts */ - memset(&mc_list[10], 0xff, ETH_ALEN); + eth_broadcast_addr(&mc_list[10]); mc_list[5] = priv->port; /* needed for B0 steering support */ mlx4_multicast_detach(mdev->dev, &priv->rss_map.indir_qp, mc_list, MLX4_PROT_ETH, priv->broadcast_id); @@ -1888,6 +1888,12 @@ static void mlx4_en_clear_stats(struct net_device *dev) memset(&priv->pstats, 0, sizeof(priv->pstats)); memset(&priv->pkstats, 0, sizeof(priv->pkstats)); memset(&priv->port_stats, 0, sizeof(priv->port_stats)); + memset(&priv->rx_flowstats, 0, sizeof(priv->rx_flowstats)); + memset(&priv->tx_flowstats, 0, sizeof(priv->tx_flowstats)); + memset(&priv->rx_priority_flowstats, 0, + sizeof(priv->rx_priority_flowstats)); + memset(&priv->tx_priority_flowstats, 0, + sizeof(priv->tx_priority_flowstats)); for (i = 0; i < priv->tx_ring_num; i++) { priv->tx_ring[i]->bytes = 0; @@ -2189,31 +2195,50 @@ static int mlx4_en_set_features(struct net_device *netdev, netdev_features_t features) { struct mlx4_en_priv *priv = netdev_priv(netdev); + bool reset = false; int ret = 0; + if (DEV_FEATURE_CHANGED(netdev, features, NETIF_F_RXFCS)) { + en_info(priv, "Turn %s RX-FCS\n", + (features & NETIF_F_RXFCS) ? "ON" : "OFF"); + reset = true; + } + + if (DEV_FEATURE_CHANGED(netdev, features, NETIF_F_RXALL)) { + u8 ignore_fcs_value = (features & NETIF_F_RXALL) ? 1 : 0; + + en_info(priv, "Turn %s RX-ALL\n", + ignore_fcs_value ? "ON" : "OFF"); + ret = mlx4_SET_PORT_fcs_check(priv->mdev->dev, + priv->port, ignore_fcs_value); + if (ret) + return ret; + } + if (DEV_FEATURE_CHANGED(netdev, features, NETIF_F_HW_VLAN_CTAG_RX)) { en_info(priv, "Turn %s RX vlan strip offload\n", (features & NETIF_F_HW_VLAN_CTAG_RX) ? "ON" : "OFF"); - ret = mlx4_en_reset_config(netdev, priv->hwtstamp_config, - features); - if (ret) - return ret; + reset = true; } if (DEV_FEATURE_CHANGED(netdev, features, NETIF_F_HW_VLAN_CTAG_TX)) en_info(priv, "Turn %s TX vlan strip offload\n", (features & NETIF_F_HW_VLAN_CTAG_TX) ? "ON" : "OFF"); - if (features & NETIF_F_LOOPBACK) - priv->ctrl_flags |= cpu_to_be32(MLX4_WQE_CTRL_FORCE_LOOPBACK); - else - priv->ctrl_flags &= - cpu_to_be32(~MLX4_WQE_CTRL_FORCE_LOOPBACK); + if (DEV_FEATURE_CHANGED(netdev, features, NETIF_F_LOOPBACK)) { + en_info(priv, "Turn %s loopback\n", + (features & NETIF_F_LOOPBACK) ? "ON" : "OFF"); + mlx4_en_update_loopback_state(netdev, features); + } - mlx4_en_update_loopback_state(netdev, features); + if (reset) { + ret = mlx4_en_reset_config(netdev, priv->hwtstamp_config, + features); + if (ret) + return ret; + } return 0; - } static int mlx4_en_set_vf_mac(struct net_device *dev, int queue, u8 *mac) @@ -2236,6 +2261,16 @@ static int mlx4_en_set_vf_vlan(struct net_device *dev, int vf, u16 vlan, u8 qos) return mlx4_set_vf_vlan(mdev->dev, en_priv->port, vf, vlan, qos); } +static int mlx4_en_set_vf_rate(struct net_device *dev, int vf, int min_tx_rate, + int max_tx_rate) +{ + struct mlx4_en_priv *en_priv = netdev_priv(dev); + struct mlx4_en_dev *mdev = en_priv->mdev; + + return mlx4_set_vf_rate(mdev->dev, en_priv->port, vf, min_tx_rate, + max_tx_rate); +} + static int mlx4_en_set_vf_spoofchk(struct net_device *dev, int vf, bool setting) { struct mlx4_en_priv *en_priv = netdev_priv(dev); @@ -2373,10 +2408,38 @@ static netdev_features_t mlx4_en_features_check(struct sk_buff *skb, struct net_device *dev, netdev_features_t features) { + features = vlan_features_check(skb, features); return vxlan_features_check(skb, features); } #endif +static int mlx4_en_set_tx_maxrate(struct net_device *dev, int queue_index, u32 maxrate) +{ + struct mlx4_en_priv *priv = netdev_priv(dev); + struct mlx4_en_tx_ring *tx_ring = priv->tx_ring[queue_index]; + struct mlx4_update_qp_params params; + int err; + + if (!(priv->mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_QP_RATE_LIMIT)) + return -EOPNOTSUPP; + + /* rate provided to us in Mbs, check if it fits into 12 bits, if not use Gbs */ + if (maxrate >> 12) { + params.rate_unit = MLX4_QP_RATE_LIMIT_GBS; + params.rate_val = maxrate / 1000; + } else if (maxrate) { + params.rate_unit = MLX4_QP_RATE_LIMIT_MBS; + params.rate_val = maxrate; + } else { /* zero serves to revoke the QP rate-limitation */ + params.rate_unit = 0; + params.rate_val = 0; + } + + err = mlx4_update_qp(priv->mdev->dev, tx_ring->qpn, MLX4_UPDATE_QP_RATE_LIMIT, + ¶ms); + return err; +} + static const struct net_device_ops mlx4_netdev_ops = { .ndo_open = mlx4_en_open, .ndo_stop = mlx4_en_close, @@ -2408,6 +2471,7 @@ static const struct net_device_ops mlx4_netdev_ops = { .ndo_del_vxlan_port = mlx4_en_del_vxlan_port, .ndo_features_check = mlx4_en_features_check, #endif + .ndo_set_tx_maxrate = mlx4_en_set_tx_maxrate, }; static const struct net_device_ops mlx4_netdev_ops_master = { @@ -2425,6 +2489,7 @@ static const struct net_device_ops mlx4_netdev_ops_master = { .ndo_vlan_rx_kill_vid = mlx4_en_vlan_rx_kill_vid, .ndo_set_vf_mac = mlx4_en_set_vf_mac, .ndo_set_vf_vlan = mlx4_en_set_vf_vlan, + .ndo_set_vf_rate = mlx4_en_set_vf_rate, .ndo_set_vf_spoofchk = mlx4_en_set_vf_spoofchk, .ndo_set_vf_link_state = mlx4_en_set_vf_link_state, .ndo_get_vf_config = mlx4_en_get_vf_config, @@ -2442,6 +2507,7 @@ static const struct net_device_ops mlx4_netdev_ops_master = { .ndo_del_vxlan_port = mlx4_en_del_vxlan_port, .ndo_features_check = mlx4_en_features_check, #endif + .ndo_set_tx_maxrate = mlx4_en_set_tx_maxrate, }; struct mlx4_en_bond { @@ -2618,6 +2684,82 @@ int mlx4_en_netdev_event(struct notifier_block *this, return NOTIFY_DONE; } +void mlx4_en_update_pfc_stats_bitmap(struct mlx4_dev *dev, + struct mlx4_en_stats_bitmap *stats_bitmap, + u8 rx_ppp, u8 rx_pause, + u8 tx_ppp, u8 tx_pause) +{ + int last_i = NUM_MAIN_STATS + NUM_PORT_STATS; + + if (!mlx4_is_slave(dev) && + (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_FLOWSTATS_EN)) { + mutex_lock(&stats_bitmap->mutex); + bitmap_clear(stats_bitmap->bitmap, last_i, NUM_FLOW_STATS); + + if (rx_ppp) + bitmap_set(stats_bitmap->bitmap, last_i, + NUM_FLOW_PRIORITY_STATS_RX); + last_i += NUM_FLOW_PRIORITY_STATS_RX; + + if (rx_pause && !(rx_ppp)) + bitmap_set(stats_bitmap->bitmap, last_i, + NUM_FLOW_STATS_RX); + last_i += NUM_FLOW_STATS_RX; + + if (tx_ppp) + bitmap_set(stats_bitmap->bitmap, last_i, + NUM_FLOW_PRIORITY_STATS_TX); + last_i += NUM_FLOW_PRIORITY_STATS_TX; + + if (tx_pause && !(tx_ppp)) + bitmap_set(stats_bitmap->bitmap, last_i, + NUM_FLOW_STATS_TX); + last_i += NUM_FLOW_STATS_TX; + + mutex_unlock(&stats_bitmap->mutex); + } +} + +void mlx4_en_set_stats_bitmap(struct mlx4_dev *dev, + struct mlx4_en_stats_bitmap *stats_bitmap, + u8 rx_ppp, u8 rx_pause, + u8 tx_ppp, u8 tx_pause) +{ + int last_i = 0; + + mutex_init(&stats_bitmap->mutex); + bitmap_zero(stats_bitmap->bitmap, NUM_ALL_STATS); + + if (mlx4_is_slave(dev)) { + bitmap_set(stats_bitmap->bitmap, last_i + + MLX4_FIND_NETDEV_STAT(rx_packets), 1); + bitmap_set(stats_bitmap->bitmap, last_i + + MLX4_FIND_NETDEV_STAT(tx_packets), 1); + bitmap_set(stats_bitmap->bitmap, last_i + + MLX4_FIND_NETDEV_STAT(rx_bytes), 1); + bitmap_set(stats_bitmap->bitmap, last_i + + MLX4_FIND_NETDEV_STAT(tx_bytes), 1); + bitmap_set(stats_bitmap->bitmap, last_i + + MLX4_FIND_NETDEV_STAT(rx_dropped), 1); + bitmap_set(stats_bitmap->bitmap, last_i + + MLX4_FIND_NETDEV_STAT(tx_dropped), 1); + } else { + bitmap_set(stats_bitmap->bitmap, last_i, NUM_MAIN_STATS); + } + last_i += NUM_MAIN_STATS; + + bitmap_set(stats_bitmap->bitmap, last_i, NUM_PORT_STATS); + last_i += NUM_PORT_STATS; + + mlx4_en_update_pfc_stats_bitmap(dev, stats_bitmap, + rx_ppp, rx_pause, + tx_ppp, tx_pause); + last_i += NUM_FLOW_STATS; + + if (!mlx4_is_slave(dev)) + bitmap_set(stats_bitmap->bitmap, last_i, NUM_PKT_STATS); +} + int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port, struct mlx4_en_port_profile *prof) { @@ -2693,7 +2835,7 @@ int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port, priv->msg_enable = MLX4_EN_MSG_LEVEL; #ifdef CONFIG_MLX4_EN_DCB if (!mlx4_is_slave(priv->mdev->dev)) { - if (mdev->dev->caps.flags & MLX4_DEV_CAP_FLAG_SET_ETH_SCHED) { + if (mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_ETS_CFG) { dev->dcbnl_ops = &mlx4_en_dcbnl_ops; } else { en_info(priv, "enabling only PFC DCB ops\n"); @@ -2780,6 +2922,12 @@ int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port, dev->hw_features |= NETIF_F_LOOPBACK | NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_HW_VLAN_CTAG_RX; + if (mdev->dev->caps.flags & MLX4_DEV_CAP_FLAG_FCS_KEEP) + dev->hw_features |= NETIF_F_RXFCS; + + if (mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_IGNORE_FCS) + dev->hw_features |= NETIF_F_RXALL; + if (mdev->dev->caps.steering_mode == MLX4_STEERING_MODE_DEVICE_MANAGED && mdev->dev->caps.dmfs_high_steer_mode != MLX4_STEERING_DMFS_A0_STATIC) @@ -2844,7 +2992,11 @@ int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port, queue_delayed_work(mdev->workqueue, &priv->service_task, SERVICE_TASK_DELAY); - mlx4_set_stats_bitmap(mdev->dev, &priv->stats_bitmap); + mlx4_en_set_stats_bitmap(mdev->dev, &priv->stats_bitmap, + mdev->profile.prof[priv->port].rx_ppp, + mdev->profile.prof[priv->port].rx_pause, + mdev->profile.prof[priv->port].tx_ppp, + mdev->profile.prof[priv->port].tx_pause); err = register_netdev(dev); if (err) { @@ -2872,7 +3024,8 @@ int mlx4_en_reset_config(struct net_device *dev, if (priv->hwtstamp_config.tx_type == ts_config.tx_type && priv->hwtstamp_config.rx_filter == ts_config.rx_filter && - !DEV_FEATURE_CHANGED(dev, features, NETIF_F_HW_VLAN_CTAG_RX)) + !DEV_FEATURE_CHANGED(dev, features, NETIF_F_HW_VLAN_CTAG_RX) && + !DEV_FEATURE_CHANGED(dev, features, NETIF_F_RXFCS)) return 0; /* Nothing to change */ if (DEV_FEATURE_CHANGED(dev, features, NETIF_F_HW_VLAN_CTAG_RX) && @@ -2911,6 +3064,13 @@ int mlx4_en_reset_config(struct net_device *dev, dev->features &= ~NETIF_F_HW_VLAN_CTAG_RX; } + if (DEV_FEATURE_CHANGED(dev, features, NETIF_F_RXFCS)) { + if (features & NETIF_F_RXFCS) + dev->features |= NETIF_F_RXFCS; + else + dev->features &= ~NETIF_F_RXFCS; + } + /* RX vlan offload and RX time-stamping can't co-exist ! * Regardless of the caller's choice, * Turn Off RX vlan offload in case of time-stamping is ON diff --git a/drivers/net/ethernet/mellanox/mlx4/en_port.c b/drivers/net/ethernet/mellanox/mlx4/en_port.c index 6cb80072af6c..54f0e5ab2e55 100644 --- a/drivers/net/ethernet/mellanox/mlx4/en_port.c +++ b/drivers/net/ethernet/mellanox/mlx4/en_port.c @@ -128,9 +128,29 @@ out: return err; } +/* Each counter set is located in struct mlx4_en_stat_out_mbox + * with a const offset between its prio components. + * This function runs over a counter set and sum all of it's prio components. + */ +static unsigned long en_stats_adder(__be64 *start, __be64 *next, int num) +{ + __be64 *curr = start; + unsigned long ret = 0; + int i; + int offset = next - start; + + for (i = 0; i <= num; i++) { + ret += be64_to_cpu(*curr); + curr += offset; + } + + return ret; +} + int mlx4_en_DUMP_ETH_STATS(struct mlx4_en_dev *mdev, u8 port, u8 reset) { struct mlx4_en_stat_out_mbox *mlx4_en_stats; + struct mlx4_en_stat_out_flow_control_mbox *flowstats; struct mlx4_en_priv *priv = netdev_priv(mdev->pndev[port]); struct net_device_stats *stats = &priv->stats; struct mlx4_cmd_mailbox *mailbox; @@ -183,22 +203,25 @@ int mlx4_en_DUMP_ETH_STATS(struct mlx4_en_dev *mdev, u8 port, u8 reset) priv->port_stats.xmit_more += ring->xmit_more; } + /* net device stats */ stats->rx_errors = be64_to_cpu(mlx4_en_stats->PCS) + - be32_to_cpu(mlx4_en_stats->RdropLength) + be32_to_cpu(mlx4_en_stats->RJBBR) + be32_to_cpu(mlx4_en_stats->RCRC) + - be32_to_cpu(mlx4_en_stats->RRUNT); - stats->tx_errors = be32_to_cpu(mlx4_en_stats->TDROP); - stats->multicast = be64_to_cpu(mlx4_en_stats->MCAST_prio_0) + - be64_to_cpu(mlx4_en_stats->MCAST_prio_1) + - be64_to_cpu(mlx4_en_stats->MCAST_prio_2) + - be64_to_cpu(mlx4_en_stats->MCAST_prio_3) + - be64_to_cpu(mlx4_en_stats->MCAST_prio_4) + - be64_to_cpu(mlx4_en_stats->MCAST_prio_5) + - be64_to_cpu(mlx4_en_stats->MCAST_prio_6) + - be64_to_cpu(mlx4_en_stats->MCAST_prio_7) + - be64_to_cpu(mlx4_en_stats->MCAST_novlan); + be32_to_cpu(mlx4_en_stats->RRUNT) + + be64_to_cpu(mlx4_en_stats->RInRangeLengthErr) + + be64_to_cpu(mlx4_en_stats->ROutRangeLengthErr) + + be32_to_cpu(mlx4_en_stats->RSHORT) + + en_stats_adder(&mlx4_en_stats->RGIANT_prio_0, + &mlx4_en_stats->RGIANT_prio_1, + NUM_PRIORITIES); + stats->tx_errors = en_stats_adder(&mlx4_en_stats->TGIANT_prio_0, + &mlx4_en_stats->TGIANT_prio_1, + NUM_PRIORITIES); + stats->multicast = en_stats_adder(&mlx4_en_stats->MCAST_prio_0, + &mlx4_en_stats->MCAST_prio_1, + NUM_PRIORITIES); stats->collisions = 0; + stats->rx_dropped = be32_to_cpu(mlx4_en_stats->RDROP); stats->rx_length_errors = be32_to_cpu(mlx4_en_stats->RdropLength); stats->rx_over_errors = be32_to_cpu(mlx4_en_stats->RdropOvflw); stats->rx_crc_errors = be32_to_cpu(mlx4_en_stats->RCRC); @@ -210,33 +233,116 @@ int mlx4_en_DUMP_ETH_STATS(struct mlx4_en_dev *mdev, u8 port, u8 reset) stats->tx_fifo_errors = 0; stats->tx_heartbeat_errors = 0; stats->tx_window_errors = 0; + stats->tx_dropped = be32_to_cpu(mlx4_en_stats->TDROP); + + /* RX stats */ + priv->pkstats.rx_multicast_packets = stats->multicast; + priv->pkstats.rx_broadcast_packets = + en_stats_adder(&mlx4_en_stats->RBCAST_prio_0, + &mlx4_en_stats->RBCAST_prio_1, + NUM_PRIORITIES); + priv->pkstats.rx_jabbers = be32_to_cpu(mlx4_en_stats->RJBBR); + priv->pkstats.rx_in_range_length_error = + be64_to_cpu(mlx4_en_stats->RInRangeLengthErr); + priv->pkstats.rx_out_range_length_error = + be64_to_cpu(mlx4_en_stats->ROutRangeLengthErr); + + /* Tx stats */ + priv->pkstats.tx_multicast_packets = + en_stats_adder(&mlx4_en_stats->TMCAST_prio_0, + &mlx4_en_stats->TMCAST_prio_1, + NUM_PRIORITIES); + priv->pkstats.tx_broadcast_packets = + en_stats_adder(&mlx4_en_stats->TBCAST_prio_0, + &mlx4_en_stats->TBCAST_prio_1, + NUM_PRIORITIES); + + priv->pkstats.rx_prio[0][0] = be64_to_cpu(mlx4_en_stats->RTOT_prio_0); + priv->pkstats.rx_prio[0][1] = be64_to_cpu(mlx4_en_stats->ROCT_prio_0); + priv->pkstats.rx_prio[1][0] = be64_to_cpu(mlx4_en_stats->RTOT_prio_1); + priv->pkstats.rx_prio[1][1] = be64_to_cpu(mlx4_en_stats->ROCT_prio_1); + priv->pkstats.rx_prio[2][0] = be64_to_cpu(mlx4_en_stats->RTOT_prio_2); + priv->pkstats.rx_prio[2][1] = be64_to_cpu(mlx4_en_stats->ROCT_prio_2); + priv->pkstats.rx_prio[3][0] = be64_to_cpu(mlx4_en_stats->RTOT_prio_3); + priv->pkstats.rx_prio[3][1] = be64_to_cpu(mlx4_en_stats->ROCT_prio_3); + priv->pkstats.rx_prio[4][0] = be64_to_cpu(mlx4_en_stats->RTOT_prio_4); + priv->pkstats.rx_prio[4][1] = be64_to_cpu(mlx4_en_stats->ROCT_prio_4); + priv->pkstats.rx_prio[5][0] = be64_to_cpu(mlx4_en_stats->RTOT_prio_5); + priv->pkstats.rx_prio[5][1] = be64_to_cpu(mlx4_en_stats->ROCT_prio_5); + priv->pkstats.rx_prio[6][0] = be64_to_cpu(mlx4_en_stats->RTOT_prio_6); + priv->pkstats.rx_prio[6][1] = be64_to_cpu(mlx4_en_stats->ROCT_prio_6); + priv->pkstats.rx_prio[7][0] = be64_to_cpu(mlx4_en_stats->RTOT_prio_7); + priv->pkstats.rx_prio[7][1] = be64_to_cpu(mlx4_en_stats->ROCT_prio_7); + priv->pkstats.rx_prio[8][0] = be64_to_cpu(mlx4_en_stats->RTOT_novlan); + priv->pkstats.rx_prio[8][1] = be64_to_cpu(mlx4_en_stats->ROCT_novlan); + priv->pkstats.tx_prio[0][0] = be64_to_cpu(mlx4_en_stats->TTOT_prio_0); + priv->pkstats.tx_prio[0][1] = be64_to_cpu(mlx4_en_stats->TOCT_prio_0); + priv->pkstats.tx_prio[1][0] = be64_to_cpu(mlx4_en_stats->TTOT_prio_1); + priv->pkstats.tx_prio[1][1] = be64_to_cpu(mlx4_en_stats->TOCT_prio_1); + priv->pkstats.tx_prio[2][0] = be64_to_cpu(mlx4_en_stats->TTOT_prio_2); + priv->pkstats.tx_prio[2][1] = be64_to_cpu(mlx4_en_stats->TOCT_prio_2); + priv->pkstats.tx_prio[3][0] = be64_to_cpu(mlx4_en_stats->TTOT_prio_3); + priv->pkstats.tx_prio[3][1] = be64_to_cpu(mlx4_en_stats->TOCT_prio_3); + priv->pkstats.tx_prio[4][0] = be64_to_cpu(mlx4_en_stats->TTOT_prio_4); + priv->pkstats.tx_prio[4][1] = be64_to_cpu(mlx4_en_stats->TOCT_prio_4); + priv->pkstats.tx_prio[5][0] = be64_to_cpu(mlx4_en_stats->TTOT_prio_5); + priv->pkstats.tx_prio[5][1] = be64_to_cpu(mlx4_en_stats->TOCT_prio_5); + priv->pkstats.tx_prio[6][0] = be64_to_cpu(mlx4_en_stats->TTOT_prio_6); + priv->pkstats.tx_prio[6][1] = be64_to_cpu(mlx4_en_stats->TOCT_prio_6); + priv->pkstats.tx_prio[7][0] = be64_to_cpu(mlx4_en_stats->TTOT_prio_7); + priv->pkstats.tx_prio[7][1] = be64_to_cpu(mlx4_en_stats->TOCT_prio_7); + priv->pkstats.tx_prio[8][0] = be64_to_cpu(mlx4_en_stats->TTOT_novlan); + priv->pkstats.tx_prio[8][1] = be64_to_cpu(mlx4_en_stats->TOCT_novlan); + + spin_unlock_bh(&priv->stats_lock); + + /* 0xffs indicates invalid value */ + memset(mailbox->buf, 0xff, sizeof(*flowstats) * MLX4_NUM_PRIORITIES); + + if (mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_FLOWSTATS_EN) { + memset(mailbox->buf, 0, + sizeof(*flowstats) * MLX4_NUM_PRIORITIES); + err = mlx4_cmd_box(mdev->dev, 0, mailbox->dma, + in_mod | MLX4_DUMP_ETH_STATS_FLOW_CONTROL, + 0, MLX4_CMD_DUMP_ETH_STATS, + MLX4_CMD_TIME_CLASS_B, MLX4_CMD_WRAPPED); + if (err) + goto out; + } + + flowstats = mailbox->buf; + + spin_lock_bh(&priv->stats_lock); + + for (i = 0; i < MLX4_NUM_PRIORITIES; i++) { + priv->rx_priority_flowstats[i].rx_pause = + be64_to_cpu(flowstats[i].rx_pause); + priv->rx_priority_flowstats[i].rx_pause_duration = + be64_to_cpu(flowstats[i].rx_pause_duration); + priv->rx_priority_flowstats[i].rx_pause_transition = + be64_to_cpu(flowstats[i].rx_pause_transition); + priv->tx_priority_flowstats[i].tx_pause = + be64_to_cpu(flowstats[i].tx_pause); + priv->tx_priority_flowstats[i].tx_pause_duration = + be64_to_cpu(flowstats[i].tx_pause_duration); + priv->tx_priority_flowstats[i].tx_pause_transition = + be64_to_cpu(flowstats[i].tx_pause_transition); + } + + /* if pfc is not in use, all priorities counters have the same value */ + priv->rx_flowstats.rx_pause = + be64_to_cpu(flowstats[0].rx_pause); + priv->rx_flowstats.rx_pause_duration = + be64_to_cpu(flowstats[0].rx_pause_duration); + priv->rx_flowstats.rx_pause_transition = + be64_to_cpu(flowstats[0].rx_pause_transition); + priv->tx_flowstats.tx_pause = + be64_to_cpu(flowstats[0].tx_pause); + priv->tx_flowstats.tx_pause_duration = + be64_to_cpu(flowstats[0].tx_pause_duration); + priv->tx_flowstats.tx_pause_transition = + be64_to_cpu(flowstats[0].tx_pause_transition); - priv->pkstats.broadcast = - be64_to_cpu(mlx4_en_stats->RBCAST_prio_0) + - be64_to_cpu(mlx4_en_stats->RBCAST_prio_1) + - be64_to_cpu(mlx4_en_stats->RBCAST_prio_2) + - be64_to_cpu(mlx4_en_stats->RBCAST_prio_3) + - be64_to_cpu(mlx4_en_stats->RBCAST_prio_4) + - be64_to_cpu(mlx4_en_stats->RBCAST_prio_5) + - be64_to_cpu(mlx4_en_stats->RBCAST_prio_6) + - be64_to_cpu(mlx4_en_stats->RBCAST_prio_7) + - be64_to_cpu(mlx4_en_stats->RBCAST_novlan); - priv->pkstats.rx_prio[0] = be64_to_cpu(mlx4_en_stats->RTOT_prio_0); - priv->pkstats.rx_prio[1] = be64_to_cpu(mlx4_en_stats->RTOT_prio_1); - priv->pkstats.rx_prio[2] = be64_to_cpu(mlx4_en_stats->RTOT_prio_2); - priv->pkstats.rx_prio[3] = be64_to_cpu(mlx4_en_stats->RTOT_prio_3); - priv->pkstats.rx_prio[4] = be64_to_cpu(mlx4_en_stats->RTOT_prio_4); - priv->pkstats.rx_prio[5] = be64_to_cpu(mlx4_en_stats->RTOT_prio_5); - priv->pkstats.rx_prio[6] = be64_to_cpu(mlx4_en_stats->RTOT_prio_6); - priv->pkstats.rx_prio[7] = be64_to_cpu(mlx4_en_stats->RTOT_prio_7); - priv->pkstats.tx_prio[0] = be64_to_cpu(mlx4_en_stats->TTOT_prio_0); - priv->pkstats.tx_prio[1] = be64_to_cpu(mlx4_en_stats->TTOT_prio_1); - priv->pkstats.tx_prio[2] = be64_to_cpu(mlx4_en_stats->TTOT_prio_2); - priv->pkstats.tx_prio[3] = be64_to_cpu(mlx4_en_stats->TTOT_prio_3); - priv->pkstats.tx_prio[4] = be64_to_cpu(mlx4_en_stats->TTOT_prio_4); - priv->pkstats.tx_prio[5] = be64_to_cpu(mlx4_en_stats->TTOT_prio_5); - priv->pkstats.tx_prio[6] = be64_to_cpu(mlx4_en_stats->TTOT_prio_6); - priv->pkstats.tx_prio[7] = be64_to_cpu(mlx4_en_stats->TTOT_prio_7); spin_unlock_bh(&priv->stats_lock); out: diff --git a/drivers/net/ethernet/mellanox/mlx4/en_rx.c b/drivers/net/ethernet/mellanox/mlx4/en_rx.c index 698d60de1255..4fdd3c37e47b 100644 --- a/drivers/net/ethernet/mellanox/mlx4/en_rx.c +++ b/drivers/net/ethernet/mellanox/mlx4/en_rx.c @@ -771,7 +771,7 @@ int mlx4_en_process_rx_cq(struct net_device *dev, struct mlx4_en_cq *cq, int bud /* * make sure we read the CQE after we read the ownership bit */ - rmb(); + dma_rmb(); /* Drop packet on bad receive or bad checksum */ if (unlikely((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) == @@ -1116,7 +1116,10 @@ static int mlx4_en_config_rss_qp(struct mlx4_en_priv *priv, int qpn, /* Cancel FCS removal if FW allows */ if (mdev->dev->caps.flags & MLX4_DEV_CAP_FLAG_FCS_KEEP) { context->param3 |= cpu_to_be32(1 << 29); - ring->fcs_del = ETH_FCS_LEN; + if (priv->dev->features & NETIF_F_RXFCS) + ring->fcs_del = 0; + else + ring->fcs_del = ETH_FCS_LEN; } else ring->fcs_del = 0; diff --git a/drivers/net/ethernet/mellanox/mlx4/en_selftest.c b/drivers/net/ethernet/mellanox/mlx4/en_selftest.c index a61009f4b2df..b66e03d9711f 100644 --- a/drivers/net/ethernet/mellanox/mlx4/en_selftest.c +++ b/drivers/net/ethernet/mellanox/mlx4/en_selftest.c @@ -66,7 +66,7 @@ static int mlx4_en_test_loopback_xmit(struct mlx4_en_priv *priv) ethh = (struct ethhdr *)skb_put(skb, sizeof(struct ethhdr)); packet = (unsigned char *)skb_put(skb, packet_size); memcpy(ethh->h_dest, priv->dev->dev_addr, ETH_ALEN); - memset(ethh->h_source, 0, ETH_ALEN); + eth_zero_addr(ethh->h_source); ethh->h_proto = htons(ETH_P_ARP); skb_set_mac_header(skb, 0); for (i = 0; i < packet_size; ++i) /* fill our packet */ diff --git a/drivers/net/ethernet/mellanox/mlx4/en_tx.c b/drivers/net/ethernet/mellanox/mlx4/en_tx.c index 55f9f5c5344e..1783705273d8 100644 --- a/drivers/net/ethernet/mellanox/mlx4/en_tx.c +++ b/drivers/net/ethernet/mellanox/mlx4/en_tx.c @@ -416,7 +416,7 @@ static bool mlx4_en_process_tx_cq(struct net_device *dev, * make sure we read the CQE after we read the * ownership bit */ - rmb(); + dma_rmb(); if (unlikely((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) == MLX4_CQE_OPCODE_ERROR)) { @@ -667,7 +667,7 @@ static void build_inline_wqe(struct mlx4_en_tx_desc *tx_desc, skb_frag_size(&shinfo->frags[0])); } - wmb(); + dma_wmb(); inl->byte_count = cpu_to_be32(1 << 31 | (skb->len - spc)); } } @@ -804,7 +804,7 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev) data->addr = cpu_to_be64(dma); data->lkey = ring->mr_key; - wmb(); + dma_wmb(); data->byte_count = cpu_to_be32(byte_count); --data; } @@ -821,7 +821,7 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev) data->addr = cpu_to_be64(dma); data->lkey = ring->mr_key; - wmb(); + dma_wmb(); data->byte_count = cpu_to_be32(byte_count); } /* tx completion can avoid cache line miss for common cases */ @@ -938,7 +938,7 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev) /* Ensure new descriptor hits memory * before setting ownership of this descriptor to HW */ - wmb(); + dma_wmb(); tx_desc->ctrl.owner_opcode = op_own; wmb(); @@ -958,7 +958,7 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev) /* Ensure new descriptor hits memory * before setting ownership of this descriptor to HW */ - wmb(); + dma_wmb(); tx_desc->ctrl.owner_opcode = op_own; if (send_doorbell) { wmb(); diff --git a/drivers/net/ethernet/mellanox/mlx4/eq.c b/drivers/net/ethernet/mellanox/mlx4/eq.c index 6e70ffee8e87..190fd624bdfe 100644 --- a/drivers/net/ethernet/mellanox/mlx4/eq.c +++ b/drivers/net/ethernet/mellanox/mlx4/eq.c @@ -188,7 +188,7 @@ static void slave_event(struct mlx4_dev *dev, u8 slave, struct mlx4_eqe *eqe) memcpy(s_eqe, eqe, dev->caps.eqe_size - 1); s_eqe->slave_id = slave; /* ensure all information is written before setting the ownersip bit */ - wmb(); + dma_wmb(); s_eqe->owner = !!(slave_eq->prod & SLAVE_EVENT_EQ_SIZE) ? 0x0 : 0x80; ++slave_eq->prod; @@ -473,7 +473,7 @@ static int mlx4_eq_int(struct mlx4_dev *dev, struct mlx4_eq *eq) * Make sure we read EQ entry contents after we've * checked the ownership bit. */ - rmb(); + dma_rmb(); switch (eqe->type) { case MLX4_EVENT_TYPE_COMP: diff --git a/drivers/net/ethernet/mellanox/mlx4/fw.c b/drivers/net/ethernet/mellanox/mlx4/fw.c index 5a21e5dc94cb..b9881fc1252f 100644 --- a/drivers/net/ethernet/mellanox/mlx4/fw.c +++ b/drivers/net/ethernet/mellanox/mlx4/fw.c @@ -49,9 +49,9 @@ enum { extern void __buggy_use_of_MLX4_GET(void); extern void __buggy_use_of_MLX4_PUT(void); -static bool enable_qos; +static bool enable_qos = true; module_param(enable_qos, bool, 0444); -MODULE_PARM_DESC(enable_qos, "Enable Quality of Service support in the HCA (default: off)"); +MODULE_PARM_DESC(enable_qos, "Enable Enhanced QoS support (default: on)"); #define MLX4_GET(dest, source, offset) \ do { \ @@ -105,6 +105,7 @@ static void dump_dev_cap_flags(struct mlx4_dev *dev, u64 flags) [41] = "Unicast VEP steering support", [42] = "Multicast VEP steering support", [48] = "Counters support", + [52] = "RSS IP fragments support", [53] = "Port ETS Scheduler support", [55] = "Port link type sensing support", [59] = "Port management change event support", @@ -143,7 +144,14 @@ static void dump_dev_cap_flags2(struct mlx4_dev *dev, u64 flags) [18] = "More than 80 VFs support", [19] = "Performance optimized for limited rule configuration flow steering support", [20] = "Recoverable error events support", - [21] = "Port Remap support" + [21] = "Port Remap support", + [22] = "QCN support", + [23] = "QP rate limiting support", + [24] = "Ethernet Flow control statistics support", + [25] = "Granular QoS per VF support", + [26] = "Port ETS Scheduler support", + [27] = "Port beacon support", + [28] = "RX-ALL support", }; int i; @@ -641,6 +649,7 @@ int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap) #define QUERY_DEV_CAP_RSS_OFFSET 0x2e #define QUERY_DEV_CAP_MAX_RDMA_OFFSET 0x2f #define QUERY_DEV_CAP_RSZ_SRQ_OFFSET 0x33 +#define QUERY_DEV_CAP_PORT_BEACON_OFFSET 0x34 #define QUERY_DEV_CAP_ACK_DELAY_OFFSET 0x35 #define QUERY_DEV_CAP_MTU_WIDTH_OFFSET 0x36 #define QUERY_DEV_CAP_VL_PORT_OFFSET 0x37 @@ -670,12 +679,13 @@ int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap) #define QUERY_DEV_CAP_RSVD_XRC_OFFSET 0x66 #define QUERY_DEV_CAP_MAX_XRC_OFFSET 0x67 #define QUERY_DEV_CAP_MAX_COUNTERS_OFFSET 0x68 +#define QUERY_DEV_CAP_PORT_FLOWSTATS_COUNTERS_OFFSET 0x70 #define QUERY_DEV_CAP_EXT_2_FLAGS_OFFSET 0x70 #define QUERY_DEV_CAP_FLOW_STEERING_IPOIB_OFFSET 0x74 #define QUERY_DEV_CAP_FLOW_STEERING_RANGE_EN_OFFSET 0x76 #define QUERY_DEV_CAP_FLOW_STEERING_MAX_QP_OFFSET 0x77 #define QUERY_DEV_CAP_CQ_EQ_CACHE_LINE_STRIDE 0x7a -#define QUERY_DEV_CAP_ETH_PROT_CTRL_OFFSET 0x7a +#define QUERY_DEV_CAP_ECN_QCN_VER_OFFSET 0x7b #define QUERY_DEV_CAP_RDMARC_ENTRY_SZ_OFFSET 0x80 #define QUERY_DEV_CAP_QPC_ENTRY_SZ_OFFSET 0x82 #define QUERY_DEV_CAP_AUX_ENTRY_SZ_OFFSET 0x84 @@ -696,6 +706,10 @@ int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap) #define QUERY_DEV_CAP_MAD_DEMUX_OFFSET 0xb0 #define QUERY_DEV_CAP_DMFS_HIGH_RATE_QPN_BASE_OFFSET 0xa8 #define QUERY_DEV_CAP_DMFS_HIGH_RATE_QPN_RANGE_OFFSET 0xac +#define QUERY_DEV_CAP_QP_RATE_LIMIT_NUM_OFFSET 0xcc +#define QUERY_DEV_CAP_QP_RATE_LIMIT_MAX_OFFSET 0xd0 +#define QUERY_DEV_CAP_QP_RATE_LIMIT_MIN_OFFSET 0xd2 + dev_cap->flags2 = 0; mailbox = mlx4_alloc_cmd_mailbox(dev); @@ -767,16 +781,25 @@ int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap) MLX4_GET(field, outbox, QUERY_DEV_CAP_VL_PORT_OFFSET); dev_cap->num_ports = field & 0xf; MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_MSG_SZ_OFFSET); + MLX4_GET(field, outbox, QUERY_DEV_CAP_PORT_FLOWSTATS_COUNTERS_OFFSET); + if (field & 0x10) + dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_FLOWSTATS_EN; dev_cap->max_msg_sz = 1 << (field & 0x1f); MLX4_GET(field, outbox, QUERY_DEV_CAP_FLOW_STEERING_RANGE_EN_OFFSET); if (field & 0x80) dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_FS_EN; dev_cap->fs_log_max_ucast_qp_range_size = field & 0x1f; + MLX4_GET(field, outbox, QUERY_DEV_CAP_PORT_BEACON_OFFSET); + if (field & 0x80) + dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_PORT_BEACON; MLX4_GET(field, outbox, QUERY_DEV_CAP_FLOW_STEERING_IPOIB_OFFSET); if (field & 0x80) dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_DMFS_IPOIB; MLX4_GET(field, outbox, QUERY_DEV_CAP_FLOW_STEERING_MAX_QP_OFFSET); dev_cap->fs_max_num_qp_per_entry = field; + MLX4_GET(field, outbox, QUERY_DEV_CAP_ECN_QCN_VER_OFFSET); + if (field & 0x1) + dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_QCN; MLX4_GET(stat_rate, outbox, QUERY_DEV_CAP_RATE_SUPPORT_OFFSET); dev_cap->stat_rate_support = stat_rate; MLX4_GET(field, outbox, QUERY_DEV_CAP_CQ_TS_SUPPORT_OFFSET); @@ -856,6 +879,8 @@ int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap) MLX4_GET(size, outbox, QUERY_DEV_CAP_MAX_DESC_SZ_RQ_OFFSET); dev_cap->max_rq_desc_sz = size; MLX4_GET(field, outbox, QUERY_DEV_CAP_CQ_EQ_CACHE_LINE_STRIDE); + if (field & (1 << 4)) + dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_QOS_VPP; if (field & (1 << 5)) dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_ETH_PROT_CTRL; if (field & (1 << 6)) @@ -869,6 +894,8 @@ int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap) MLX4_GET(field, outbox, QUERY_DEV_CAP_CONFIG_DEV_OFFSET); if (field & 0x20) dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_CONFIG_DEV; + if (field & (1 << 2)) + dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_IGNORE_FCS; MLX4_GET(dev_cap->reserved_lkey, outbox, QUERY_DEV_CAP_RSVD_LKEY_OFFSET); MLX4_GET(field32, outbox, QUERY_DEV_CAP_ETH_BACKPL_OFFSET); @@ -882,6 +909,8 @@ int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap) MLX4_GET(field, outbox, QUERY_DEV_CAP_VXLAN); if (field & 1<<3) dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_VXLAN_OFFLOADS; + if (field & (1 << 5)) + dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_ETS_CFG; MLX4_GET(dev_cap->max_icm_sz, outbox, QUERY_DEV_CAP_MAX_ICM_SZ_OFFSET); if (dev_cap->flags & MLX4_DEV_CAP_FLAG_COUNTERS) @@ -900,6 +929,18 @@ int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap) QUERY_DEV_CAP_DMFS_HIGH_RATE_QPN_RANGE_OFFSET); dev_cap->dmfs_high_rate_qpn_range &= MGM_QPN_MASK; + MLX4_GET(size, outbox, QUERY_DEV_CAP_QP_RATE_LIMIT_NUM_OFFSET); + dev_cap->rl_caps.num_rates = size; + if (dev_cap->rl_caps.num_rates) { + dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_QP_RATE_LIMIT; + MLX4_GET(size, outbox, QUERY_DEV_CAP_QP_RATE_LIMIT_MAX_OFFSET); + dev_cap->rl_caps.max_val = size & 0xfff; + dev_cap->rl_caps.max_unit = size >> 14; + MLX4_GET(size, outbox, QUERY_DEV_CAP_QP_RATE_LIMIT_MIN_OFFSET); + dev_cap->rl_caps.min_val = size & 0xfff; + dev_cap->rl_caps.min_unit = size >> 14; + } + MLX4_GET(field32, outbox, QUERY_DEV_CAP_EXT_2_FLAGS_OFFSET); if (field32 & (1 << 16)) dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_UPDATE_QP; @@ -975,6 +1016,15 @@ void mlx4_dev_cap_dump(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap) dev_cap->dmfs_high_rate_qpn_base); mlx4_dbg(dev, "DMFS high rate steer QPn range: %d\n", dev_cap->dmfs_high_rate_qpn_range); + + if (dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_QP_RATE_LIMIT) { + struct mlx4_rate_limit_caps *rl_caps = &dev_cap->rl_caps; + + mlx4_dbg(dev, "QP Rate-Limit: #rates %d, unit/val max %d/%d, min %d/%d\n", + rl_caps->num_rates, rl_caps->max_unit, rl_caps->max_val, + rl_caps->min_unit, rl_caps->min_val); + } + dump_dev_cap_flags(dev, dev_cap->flags); dump_dev_cap_flags2(dev, dev_cap->flags2); } @@ -1058,6 +1108,7 @@ out: return err; } +#define DEV_CAP_EXT_2_FLAG_PFC_COUNTERS (1 << 28) #define DEV_CAP_EXT_2_FLAG_VLAN_CONTROL (1 << 26) #define DEV_CAP_EXT_2_FLAG_80_VFS (1 << 21) #define DEV_CAP_EXT_2_FLAG_FSM (1 << 20) @@ -1071,6 +1122,7 @@ int mlx4_QUERY_DEV_CAP_wrapper(struct mlx4_dev *dev, int slave, u64 flags; int err = 0; u8 field; + u16 field16; u32 bmme_flags, field32; int real_port; int slave_port; @@ -1101,6 +1153,9 @@ int mlx4_QUERY_DEV_CAP_wrapper(struct mlx4_dev *dev, int slave, } for (; slave_port < dev->caps.num_ports; ++slave_port) flags &= ~(MLX4_DEV_CAP_FLAG_WOL_PORT1 << slave_port); + + /* Not exposing RSS IP fragments to guests */ + flags &= ~MLX4_DEV_CAP_FLAG_RSS_IP_FRAG; MLX4_PUT(outbox->buf, flags, QUERY_DEV_CAP_EXT_FLAGS_OFFSET); MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_VL_PORT_OFFSET); @@ -1113,11 +1168,16 @@ int mlx4_QUERY_DEV_CAP_wrapper(struct mlx4_dev *dev, int slave, field &= 0x7f; MLX4_PUT(outbox->buf, field, QUERY_DEV_CAP_CQ_TS_SUPPORT_OFFSET); - /* For guests, disable vxlan tunneling */ + /* For guests, disable vxlan tunneling and QoS support */ MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_VXLAN); - field &= 0xf7; + field &= 0xd7; MLX4_PUT(outbox->buf, field, QUERY_DEV_CAP_VXLAN); + /* For guests, disable port BEACON */ + MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_PORT_BEACON_OFFSET); + field &= 0x7f; + MLX4_PUT(outbox->buf, field, QUERY_DEV_CAP_PORT_BEACON_OFFSET); + /* For guests, report Blueflame disabled */ MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_BF_OFFSET); field &= 0x7f; @@ -1146,9 +1206,28 @@ int mlx4_QUERY_DEV_CAP_wrapper(struct mlx4_dev *dev, int slave, /* turn off host side virt features (VST, FSM, etc) for guests */ MLX4_GET(field32, outbox->buf, QUERY_DEV_CAP_EXT_2_FLAGS_OFFSET); field32 &= ~(DEV_CAP_EXT_2_FLAG_VLAN_CONTROL | DEV_CAP_EXT_2_FLAG_80_VFS | - DEV_CAP_EXT_2_FLAG_FSM); + DEV_CAP_EXT_2_FLAG_FSM | DEV_CAP_EXT_2_FLAG_PFC_COUNTERS); MLX4_PUT(outbox->buf, field32, QUERY_DEV_CAP_EXT_2_FLAGS_OFFSET); + /* turn off QCN for guests */ + MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_ECN_QCN_VER_OFFSET); + field &= 0xfe; + MLX4_PUT(outbox->buf, field, QUERY_DEV_CAP_ECN_QCN_VER_OFFSET); + + /* turn off QP max-rate limiting for guests */ + field16 = 0; + MLX4_PUT(outbox->buf, field16, QUERY_DEV_CAP_QP_RATE_LIMIT_NUM_OFFSET); + + /* turn off QoS per VF support for guests */ + MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_CQ_EQ_CACHE_LINE_STRIDE); + field &= 0xef; + MLX4_PUT(outbox->buf, field, QUERY_DEV_CAP_CQ_EQ_CACHE_LINE_STRIDE); + + /* turn off ignore FCS feature for guests */ + MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_CONFIG_DEV_OFFSET); + field &= 0xfb; + MLX4_PUT(outbox->buf, field, QUERY_DEV_CAP_CONFIG_DEV_OFFSET); + return 0; } @@ -1648,13 +1727,17 @@ int mlx4_INIT_HCA(struct mlx4_dev *dev, struct mlx4_init_hca_param *param) *(inbox + INIT_HCA_FLAGS_OFFSET / 4) |= cpu_to_be32(1 << 3); /* Enable QoS support if module parameter set */ - if (enable_qos) + if (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_ETS_CFG && enable_qos) *(inbox + INIT_HCA_FLAGS_OFFSET / 4) |= cpu_to_be32(1 << 2); /* enable counters */ if (dev->caps.flags & MLX4_DEV_CAP_FLAG_COUNTERS) *(inbox + INIT_HCA_FLAGS_OFFSET / 4) |= cpu_to_be32(1 << 4); + /* Enable RSS spread to fragmented IP packets when supported */ + if (dev->caps.flags & MLX4_DEV_CAP_FLAG_RSS_IP_FRAG) + *(inbox + INIT_HCA_FLAGS_OFFSET / 4) |= cpu_to_be32(1 << 13); + /* CX3 is capable of extending CQEs/EQEs from 32 to 64 bytes */ if (dev->caps.flags & MLX4_DEV_CAP_FLAG_64B_EQE) { *(inbox + INIT_HCA_EQE_CQE_OFFSETS / 4) |= cpu_to_be32(1 << 29); @@ -1843,6 +1926,10 @@ int mlx4_QUERY_HCA(struct mlx4_dev *dev, else param->steering_mode = MLX4_STEERING_MODE_A0; } + + if (dword_field & (1 << 13)) + param->rss_ip_frags = 1; + /* steering attributes */ if (param->steering_mode == MLX4_STEERING_MODE_DEVICE_MANAGED) { MLX4_GET(param->mc_base, outbox, INIT_HCA_FS_BASE_OFFSET); diff --git a/drivers/net/ethernet/mellanox/mlx4/fw.h b/drivers/net/ethernet/mellanox/mlx4/fw.h index f44f7f6017ed..07cb7c2461ad 100644 --- a/drivers/net/ethernet/mellanox/mlx4/fw.h +++ b/drivers/net/ethernet/mellanox/mlx4/fw.h @@ -127,6 +127,7 @@ struct mlx4_dev_cap { u32 max_counters; u32 dmfs_high_rate_qpn_base; u32 dmfs_high_rate_qpn_range; + struct mlx4_rate_limit_caps rl_caps; struct mlx4_port_cap port_cap[MLX4_MAX_PORTS + 1]; }; @@ -202,6 +203,7 @@ struct mlx4_init_hca_param { u64 dev_cap_enabled; u16 cqe_size; /* For use only when CQE stride feature enabled */ u16 eqe_size; /* For use only when EQE stride feature enabled */ + u8 rss_ip_frags; }; struct mlx4_init_ib_param { diff --git a/drivers/net/ethernet/mellanox/mlx4/fw_qos.c b/drivers/net/ethernet/mellanox/mlx4/fw_qos.c new file mode 100644 index 000000000000..8f2fde0487c4 --- /dev/null +++ b/drivers/net/ethernet/mellanox/mlx4/fw_qos.c @@ -0,0 +1,289 @@ +/* + * Copyright (c) 2004, 2005 Topspin Communications. All rights reserved. + * Copyright (c) 2005, 2006, 2007, 2008 Mellanox Technologies. + * 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/export.h> +#include "fw_qos.h" +#include "fw.h" + +enum { + /* allocate vpp opcode modifiers */ + MLX4_ALLOCATE_VPP_ALLOCATE = 0x0, + MLX4_ALLOCATE_VPP_QUERY = 0x1 +}; + +enum { + /* set vport qos opcode modifiers */ + MLX4_SET_VPORT_QOS_SET = 0x0, + MLX4_SET_VPORT_QOS_QUERY = 0x1 +}; + +struct mlx4_set_port_prio2tc_context { + u8 prio2tc[4]; +}; + +struct mlx4_port_scheduler_tc_cfg_be { + __be16 pg; + __be16 bw_precentage; + __be16 max_bw_units; /* 3-100Mbps, 4-1Gbps, other values - reserved */ + __be16 max_bw_value; +}; + +struct mlx4_set_port_scheduler_context { + struct mlx4_port_scheduler_tc_cfg_be tc[MLX4_NUM_TC]; +}; + +/* Granular Qos (per VF) section */ +struct mlx4_alloc_vpp_param { + __be32 availible_vpp; + __be32 vpp_p_up[MLX4_NUM_UP]; +}; + +struct mlx4_prio_qos_param { + __be32 bw_share; + __be32 max_avg_bw; + __be32 reserved; + __be32 enable; + __be32 reserved1[4]; +}; + +struct mlx4_set_vport_context { + __be32 reserved[8]; + struct mlx4_prio_qos_param qos_p_up[MLX4_NUM_UP]; +}; + +int mlx4_SET_PORT_PRIO2TC(struct mlx4_dev *dev, u8 port, u8 *prio2tc) +{ + struct mlx4_cmd_mailbox *mailbox; + struct mlx4_set_port_prio2tc_context *context; + int err; + u32 in_mod; + int i; + + mailbox = mlx4_alloc_cmd_mailbox(dev); + if (IS_ERR(mailbox)) + return PTR_ERR(mailbox); + + context = mailbox->buf; + + for (i = 0; i < MLX4_NUM_UP; i += 2) + context->prio2tc[i >> 1] = prio2tc[i] << 4 | prio2tc[i + 1]; + + in_mod = MLX4_SET_PORT_PRIO2TC << 8 | port; + err = mlx4_cmd(dev, mailbox->dma, in_mod, 1, MLX4_CMD_SET_PORT, + MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE); + + mlx4_free_cmd_mailbox(dev, mailbox); + return err; +} +EXPORT_SYMBOL(mlx4_SET_PORT_PRIO2TC); + +int mlx4_SET_PORT_SCHEDULER(struct mlx4_dev *dev, u8 port, u8 *tc_tx_bw, + u8 *pg, u16 *ratelimit) +{ + struct mlx4_cmd_mailbox *mailbox; + struct mlx4_set_port_scheduler_context *context; + int err; + u32 in_mod; + int i; + + mailbox = mlx4_alloc_cmd_mailbox(dev); + if (IS_ERR(mailbox)) + return PTR_ERR(mailbox); + + context = mailbox->buf; + + for (i = 0; i < MLX4_NUM_TC; i++) { + struct mlx4_port_scheduler_tc_cfg_be *tc = &context->tc[i]; + u16 r; + + if (ratelimit && ratelimit[i]) { + if (ratelimit[i] <= MLX4_MAX_100M_UNITS_VAL) { + r = ratelimit[i]; + tc->max_bw_units = + htons(MLX4_RATELIMIT_100M_UNITS); + } else { + r = ratelimit[i] / 10; + tc->max_bw_units = + htons(MLX4_RATELIMIT_1G_UNITS); + } + tc->max_bw_value = htons(r); + } else { + tc->max_bw_value = htons(MLX4_RATELIMIT_DEFAULT); + tc->max_bw_units = htons(MLX4_RATELIMIT_1G_UNITS); + } + + tc->pg = htons(pg[i]); + tc->bw_precentage = htons(tc_tx_bw[i]); + } + + in_mod = MLX4_SET_PORT_SCHEDULER << 8 | port; + err = mlx4_cmd(dev, mailbox->dma, in_mod, 1, MLX4_CMD_SET_PORT, + MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE); + + mlx4_free_cmd_mailbox(dev, mailbox); + return err; +} +EXPORT_SYMBOL(mlx4_SET_PORT_SCHEDULER); + +int mlx4_ALLOCATE_VPP_get(struct mlx4_dev *dev, u8 port, + u16 *availible_vpp, u8 *vpp_p_up) +{ + int i; + int err; + struct mlx4_cmd_mailbox *mailbox; + struct mlx4_alloc_vpp_param *out_param; + + mailbox = mlx4_alloc_cmd_mailbox(dev); + if (IS_ERR(mailbox)) + return PTR_ERR(mailbox); + + out_param = mailbox->buf; + + err = mlx4_cmd_box(dev, 0, mailbox->dma, port, + MLX4_ALLOCATE_VPP_QUERY, + MLX4_CMD_ALLOCATE_VPP, + MLX4_CMD_TIME_CLASS_A, + MLX4_CMD_NATIVE); + if (err) + goto out; + + /* Total number of supported VPPs */ + *availible_vpp = (u16)be32_to_cpu(out_param->availible_vpp); + + for (i = 0; i < MLX4_NUM_UP; i++) + vpp_p_up[i] = (u8)be32_to_cpu(out_param->vpp_p_up[i]); + +out: + mlx4_free_cmd_mailbox(dev, mailbox); + + return err; +} +EXPORT_SYMBOL(mlx4_ALLOCATE_VPP_get); + +int mlx4_ALLOCATE_VPP_set(struct mlx4_dev *dev, u8 port, u8 *vpp_p_up) +{ + int i; + int err; + struct mlx4_cmd_mailbox *mailbox; + struct mlx4_alloc_vpp_param *in_param; + + mailbox = mlx4_alloc_cmd_mailbox(dev); + if (IS_ERR(mailbox)) + return PTR_ERR(mailbox); + + in_param = mailbox->buf; + + for (i = 0; i < MLX4_NUM_UP; i++) + in_param->vpp_p_up[i] = cpu_to_be32(vpp_p_up[i]); + + err = mlx4_cmd(dev, mailbox->dma, port, + MLX4_ALLOCATE_VPP_ALLOCATE, + MLX4_CMD_ALLOCATE_VPP, + MLX4_CMD_TIME_CLASS_A, + MLX4_CMD_NATIVE); + + mlx4_free_cmd_mailbox(dev, mailbox); + return err; +} +EXPORT_SYMBOL(mlx4_ALLOCATE_VPP_set); + +int mlx4_SET_VPORT_QOS_get(struct mlx4_dev *dev, u8 port, u8 vport, + struct mlx4_vport_qos_param *out_param) +{ + int i; + int err; + struct mlx4_cmd_mailbox *mailbox; + struct mlx4_set_vport_context *ctx; + + mailbox = mlx4_alloc_cmd_mailbox(dev); + if (IS_ERR(mailbox)) + return PTR_ERR(mailbox); + + ctx = mailbox->buf; + + err = mlx4_cmd_box(dev, 0, mailbox->dma, (vport << 8) | port, + MLX4_SET_VPORT_QOS_QUERY, + MLX4_CMD_SET_VPORT_QOS, + MLX4_CMD_TIME_CLASS_A, + MLX4_CMD_NATIVE); + if (err) + goto out; + + for (i = 0; i < MLX4_NUM_UP; i++) { + out_param[i].bw_share = be32_to_cpu(ctx->qos_p_up[i].bw_share); + out_param[i].max_avg_bw = + be32_to_cpu(ctx->qos_p_up[i].max_avg_bw); + out_param[i].enable = + !!(be32_to_cpu(ctx->qos_p_up[i].enable) & 31); + } + +out: + mlx4_free_cmd_mailbox(dev, mailbox); + + return err; +} +EXPORT_SYMBOL(mlx4_SET_VPORT_QOS_get); + +int mlx4_SET_VPORT_QOS_set(struct mlx4_dev *dev, u8 port, u8 vport, + struct mlx4_vport_qos_param *in_param) +{ + int i; + int err; + struct mlx4_cmd_mailbox *mailbox; + struct mlx4_set_vport_context *ctx; + + mailbox = mlx4_alloc_cmd_mailbox(dev); + if (IS_ERR(mailbox)) + return PTR_ERR(mailbox); + + ctx = mailbox->buf; + + for (i = 0; i < MLX4_NUM_UP; i++) { + ctx->qos_p_up[i].bw_share = cpu_to_be32(in_param[i].bw_share); + ctx->qos_p_up[i].max_avg_bw = + cpu_to_be32(in_param[i].max_avg_bw); + ctx->qos_p_up[i].enable = + cpu_to_be32(in_param[i].enable << 31); + } + + err = mlx4_cmd(dev, mailbox->dma, (vport << 8) | port, + MLX4_SET_VPORT_QOS_SET, + MLX4_CMD_SET_VPORT_QOS, + MLX4_CMD_TIME_CLASS_A, + MLX4_CMD_NATIVE); + + mlx4_free_cmd_mailbox(dev, mailbox); + return err; +} +EXPORT_SYMBOL(mlx4_SET_VPORT_QOS_set); diff --git a/drivers/net/ethernet/mellanox/mlx4/fw_qos.h b/drivers/net/ethernet/mellanox/mlx4/fw_qos.h new file mode 100644 index 000000000000..ac1f331878e6 --- /dev/null +++ b/drivers/net/ethernet/mellanox/mlx4/fw_qos.h @@ -0,0 +1,145 @@ +/* + * Copyright (c) 2004, 2005 Topspin Communications. All rights reserved. + * Copyright (c) 2005, 2006, 2007, 2008 Mellanox Technologies. + * 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. + */ + +#ifndef MLX4_FW_QOS_H +#define MLX4_FW_QOS_H + +#include <linux/mlx4/cmd.h> +#include <linux/mlx4/device.h> + +#define MLX4_NUM_UP 8 +#define MLX4_NUM_TC 8 + +/* Default supported priorities for VPP allocation */ +#define MLX4_DEFAULT_QOS_PRIO (0) + +/* Derived from FW feature definition, 0 is the default vport fo all QPs */ +#define MLX4_VPP_DEFAULT_VPORT (0) + +struct mlx4_vport_qos_param { + u32 bw_share; + u32 max_avg_bw; + u8 enable; +}; + +/** + * mlx4_SET_PORT_PRIO2TC - This routine maps user priorities to traffic + * classes of a given port and device. + * + * @dev: mlx4_dev. + * @port: Physical port number. + * @prio2tc: Array of TC associated with each priorities. + * + * Returns 0 on success or a negative mlx4_core errno code. + **/ +int mlx4_SET_PORT_PRIO2TC(struct mlx4_dev *dev, u8 port, u8 *prio2tc); + +/** + * mlx4_SET_PORT_SCHEDULER - This routine configures the arbitration between + * traffic classes (ETS) and configured rate limit for traffic classes. + * tc_tx_bw, pg and ratelimit are arrays where each index represents a TC. + * The description for those parameters below refers to a single TC. + * + * @dev: mlx4_dev. + * @port: Physical port number. + * @tc_tx_bw: The percentage of the bandwidth allocated for traffic class + * within a TC group. The sum of the bw_percentage of all the traffic + * classes within a TC group must equal 100% for correct operation. + * @pg: The TC group the traffic class is associated with. + * @ratelimit: The maximal bandwidth allowed for the use by this traffic class. + * + * Returns 0 on success or a negative mlx4_core errno code. + **/ +int mlx4_SET_PORT_SCHEDULER(struct mlx4_dev *dev, u8 port, u8 *tc_tx_bw, + u8 *pg, u16 *ratelimit); +/** + * mlx4_ALLOCATE_VPP_get - Query port VPP availible resources and allocation. + * Before distribution of VPPs to priorities, only availible_vpp is returned. + * After initialization it returns the distribution of VPPs among priorities. + * + * @dev: mlx4_dev. + * @port: Physical port number. + * @availible_vpp: Pointer to variable where number of availible VPPs is stored + * @vpp_p_up: Distribution of VPPs to priorities is stored in this array + * + * Returns 0 on success or a negative mlx4_core errno code. + **/ +int mlx4_ALLOCATE_VPP_get(struct mlx4_dev *dev, u8 port, + u16 *availible_vpp, u8 *vpp_p_up); +/** + * mlx4_ALLOCATE_VPP_set - Distribution of VPPs among differnt priorities. + * The total number of VPPs assigned to all for a port must not exceed + * the value reported by availible_vpp in mlx4_ALLOCATE_VPP_get. + * VPP allocation is allowed only after the port type has been set, + * and while no QPs are open for this port. + * + * @dev: mlx4_dev. + * @port: Physical port number. + * @vpp_p_up: Allocation of VPPs to different priorities. + * + * Returns 0 on success or a negative mlx4_core errno code. + **/ +int mlx4_ALLOCATE_VPP_set(struct mlx4_dev *dev, u8 port, u8 *vpp_p_up); + +/** + * mlx4_SET_VPORT_QOS_get - Query QoS proporties of a Vport. + * Each priority allowed for the Vport is assigned with a share of the BW, + * and a BW limitation. This commands query the current QoS values. + * + * @dev: mlx4_dev. + * @port: Physical port number. + * @vport: Vport id. + * @out_param: Array of mlx4_vport_qos_param that will contain the values. + * + * Returns 0 on success or a negative mlx4_core errno code. + **/ +int mlx4_SET_VPORT_QOS_get(struct mlx4_dev *dev, u8 port, u8 vport, + struct mlx4_vport_qos_param *out_param); + +/** + * mlx4_SET_VPORT_QOS_set - Set QoS proporties of a Vport. + * QoS parameters can be modified at any time, but must be initialized + * before any QP is associated with the VPort. + * + * @dev: mlx4_dev. + * @port: Physical port number. + * @vport: Vport id. + * @out_param: Array of mlx4_vport_qos_param which holds the requested values. + * + * Returns 0 on success or a negative mlx4_core errno code. + **/ +int mlx4_SET_VPORT_QOS_set(struct mlx4_dev *dev, u8 port, u8 vport, + struct mlx4_vport_qos_param *in_param); + +#endif /* MLX4_FW_QOS_H */ diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c index 7e487223489a..acceb75e8c44 100644 --- a/drivers/net/ethernet/mellanox/mlx4/main.c +++ b/drivers/net/ethernet/mellanox/mlx4/main.c @@ -297,6 +297,25 @@ static int mlx4_dev_port(struct mlx4_dev *dev, int port, return err; } +static inline void mlx4_enable_ignore_fcs(struct mlx4_dev *dev) +{ + if (!(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_IGNORE_FCS)) + return; + + if (mlx4_is_mfunc(dev)) { + mlx4_dbg(dev, "SRIOV mode - Disabling Ignore FCS"); + dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_IGNORE_FCS; + return; + } + + if (!(dev->caps.flags & MLX4_DEV_CAP_FLAG_FCS_KEEP)) { + mlx4_dbg(dev, + "Keep FCS is not supported - Disabling Ignore FCS"); + dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_IGNORE_FCS; + return; + } +} + #define MLX4_A0_STEERING_TABLE_SIZE 256 static int mlx4_dev_cap(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap) { @@ -489,6 +508,8 @@ static int mlx4_dev_cap(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap) dev->caps.dmfs_high_rate_qpn_range = MLX4_A0_STEERING_TABLE_SIZE; } + dev->caps.rl_caps = dev_cap->rl_caps; + dev->caps.reserved_qps_cnt[MLX4_QP_REGION_RSS_RAW_ETH] = dev->caps.dmfs_high_rate_qpn_range; @@ -526,10 +547,20 @@ static int mlx4_dev_cap(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap) dev->caps.alloc_res_qp_mask = (dev->caps.bf_reg_size ? MLX4_RESERVE_ETH_BF_QP : 0) | MLX4_RESERVE_A0_QP; + + if (!(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_ETS_CFG) && + dev->caps.flags & MLX4_DEV_CAP_FLAG_SET_ETH_SCHED) { + mlx4_warn(dev, "Old device ETS support detected\n"); + mlx4_warn(dev, "Consider upgrading device FW.\n"); + dev->caps.flags2 |= MLX4_DEV_CAP_FLAG2_ETS_CFG; + } + } else { dev->caps.alloc_res_qp_mask = 0; } + mlx4_enable_ignore_fcs(dev); + return 0; } @@ -883,6 +914,8 @@ static int mlx4_slave_cap(struct mlx4_dev *dev) mlx4_warn(dev, "Timestamping is not supported in slave mode\n"); slave_adjust_steering_mode(dev, &dev_cap, &hca_param); + mlx4_dbg(dev, "RSS support for IP fragments is %s\n", + hca_param.rss_ip_frags ? "on" : "off"); if (func_cap.extra_flags & MLX4_QUERY_FUNC_FLAGS_BF_RES_QP && dev->caps.bf_reg_size) diff --git a/drivers/net/ethernet/mellanox/mlx4/mlx4.h b/drivers/net/ethernet/mellanox/mlx4/mlx4.h index 1409d0cd6143..f30eeb730a86 100644 --- a/drivers/net/ethernet/mellanox/mlx4/mlx4.h +++ b/drivers/net/ethernet/mellanox/mlx4/mlx4.h @@ -50,6 +50,7 @@ #include <linux/mlx4/driver.h> #include <linux/mlx4/doorbell.h> #include <linux/mlx4/cmd.h> +#include "fw_qos.h" #define DRV_NAME "mlx4_core" #define PFX DRV_NAME ": " @@ -64,21 +65,6 @@ #define INIT_HCA_TPT_MW_ENABLE (1 << 7) -struct mlx4_set_port_prio2tc_context { - u8 prio2tc[4]; -}; - -struct mlx4_port_scheduler_tc_cfg_be { - __be16 pg; - __be16 bw_precentage; - __be16 max_bw_units; /* 3-100Mbps, 4-1Gbps, other values - reserved */ - __be16 max_bw_value; -}; - -struct mlx4_set_port_scheduler_context { - struct mlx4_port_scheduler_tc_cfg_be tc[MLX4_NUM_TC]; -}; - enum { MLX4_HCR_BASE = 0x80680, MLX4_HCR_SIZE = 0x0001c, @@ -175,7 +161,7 @@ enum mlx4_res_tracker_free_type { /* *Virtual HCR structures. - * mlx4_vhcr is the sw representation, in machine endianess + * mlx4_vhcr is the sw representation, in machine endianness * * mlx4_vhcr_cmd is the formalized structure, the one that is passed * to FW to go through communication channel. @@ -512,6 +498,7 @@ struct mlx4_vport_state { u32 tx_rate; bool spoofchk; u32 link_state; + u8 qos_vport; }; struct mlx4_vf_admin_state { @@ -568,6 +555,11 @@ struct mlx4_slave_event_eq { struct mlx4_eqe event_eqe[SLAVE_EVENT_EQ_SIZE]; }; +struct mlx4_qos_manager { + int num_of_qos_vfs; + DECLARE_BITMAP(priority_bm, MLX4_NUM_UP); +}; + struct mlx4_master_qp0_state { int proxy_qp0_active; int qp0_active; @@ -592,6 +584,7 @@ struct mlx4_mfunc_master_ctx { struct mlx4_eqe cmd_eqe; struct mlx4_slave_event_eq slave_eq; struct mutex gen_eqe_mutex[MLX4_MFUNC_MAX]; + struct mlx4_qos_manager qos_ctl[MLX4_MAX_PORTS + 1]; }; struct mlx4_mfunc { @@ -644,6 +637,7 @@ struct mlx4_vf_immed_vlan_work { int orig_vlan_ix; u8 port; u8 qos; + u8 qos_vport; u16 vlan_id; u16 orig_vlan_id; }; @@ -769,9 +763,11 @@ enum { struct mlx4_set_port_general_context { - u8 reserved[3]; + u16 reserved1; + u8 v_ignore_fcs; u8 flags; - u16 reserved2; + u8 ignore_fcs; + u8 reserved2; __be16 mtu; u8 pptx; u8 pfctx; diff --git a/drivers/net/ethernet/mellanox/mlx4/mlx4_en.h b/drivers/net/ethernet/mellanox/mlx4/mlx4_en.h index ebbe244e80dd..9de30216b146 100644 --- a/drivers/net/ethernet/mellanox/mlx4/mlx4_en.h +++ b/drivers/net/ethernet/mellanox/mlx4/mlx4_en.h @@ -55,6 +55,7 @@ #include <linux/mlx4/cmd.h> #include "en_port.h" +#include "mlx4_stats.h" #define DRV_NAME "mlx4_en" #define DRV_VERSION "2.2-1" @@ -171,7 +172,6 @@ enum { /* Number of samples to 'average' */ #define AVG_SIZE 128 #define AVG_FACTOR 1024 -#define NUM_PERF_STATS NUM_PERF_COUNTERS #define INC_PERF_COUNTER(cnt) (++(cnt)) #define ADD_PERF_COUNTER(cnt, add) ((cnt) += (add)) @@ -182,7 +182,6 @@ enum { #else -#define NUM_PERF_STATS 0 #define INC_PERF_COUNTER(cnt) do {} while (0) #define ADD_PERF_COUNTER(cnt, add) do {} while (0) #define AVG_PERF_COUNTER(cnt, sample) do {} while (0) @@ -435,37 +434,6 @@ struct mlx4_en_port_state { u32 flags; }; -struct mlx4_en_pkt_stats { - unsigned long broadcast; - unsigned long rx_prio[8]; - unsigned long tx_prio[8]; -#define NUM_PKT_STATS 17 -}; - -struct mlx4_en_port_stats { - unsigned long tso_packets; - unsigned long xmit_more; - unsigned long queue_stopped; - unsigned long wake_queue; - unsigned long tx_timeout; - unsigned long rx_alloc_failed; - unsigned long rx_chksum_good; - unsigned long rx_chksum_none; - unsigned long rx_chksum_complete; - unsigned long tx_chksum_offload; -#define NUM_PORT_STATS 10 -}; - -struct mlx4_en_perf_stats { - u32 tx_poll; - u64 tx_pktsz_avg; - u32 inflight_avg; - u16 tx_coal_avg; - u16 rx_coal_avg; - u32 napi_quota; -#define NUM_PERF_COUNTERS 6 -}; - enum mlx4_en_mclist_act { MCLIST_NONE, MCLIST_REM, @@ -514,9 +482,15 @@ enum { MLX4_EN_FLAG_RX_CSUM_NON_TCP_UDP = (1 << 5), }; +#define PORT_BEACON_MAX_LIMIT (65535) #define MLX4_EN_MAC_HASH_SIZE (1 << BITS_PER_BYTE) #define MLX4_EN_MAC_HASH_IDX 5 +struct mlx4_en_stats_bitmap { + DECLARE_BITMAP(bitmap, NUM_ALL_STATS); + struct mutex mutex; /* for mutual access to stats bitmap */ +}; + struct mlx4_en_priv { struct mlx4_en_dev *mdev; struct mlx4_en_port_profile *prof; @@ -592,8 +566,12 @@ struct mlx4_en_priv { #endif struct mlx4_en_perf_stats pstats; struct mlx4_en_pkt_stats pkstats; + struct mlx4_en_flow_stats_rx rx_priority_flowstats[MLX4_NUM_PRIORITIES]; + struct mlx4_en_flow_stats_tx tx_priority_flowstats[MLX4_NUM_PRIORITIES]; + struct mlx4_en_flow_stats_rx rx_flowstats; + struct mlx4_en_flow_stats_tx tx_flowstats; struct mlx4_en_port_stats port_stats; - u64 stats_bitmap; + struct mlx4_en_stats_bitmap stats_bitmap; struct list_head mc_list; struct list_head curr_list; u64 broadcast_id; @@ -608,6 +586,7 @@ struct mlx4_en_priv { #ifdef CONFIG_MLX4_EN_DCB struct ieee_ets ets; u16 maxrate[IEEE_8021QAZ_MAX_TCS]; + enum dcbnl_cndd_states cndd_state[IEEE_8021QAZ_MAX_TCS]; #endif #ifdef CONFIG_RFS_ACCEL spinlock_t filters_lock; @@ -761,6 +740,11 @@ int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port, int mlx4_en_start_port(struct net_device *dev); void mlx4_en_stop_port(struct net_device *dev, int detach); +void mlx4_en_set_stats_bitmap(struct mlx4_dev *dev, + struct mlx4_en_stats_bitmap *stats_bitmap, + u8 rx_ppp, u8 rx_pause, + u8 tx_ppp, u8 tx_pause); + void mlx4_en_free_resources(struct mlx4_en_priv *priv); int mlx4_en_alloc_resources(struct mlx4_en_priv *priv); @@ -846,7 +830,10 @@ void mlx4_en_ptp_overflow_check(struct mlx4_en_dev *mdev); int mlx4_en_reset_config(struct net_device *dev, struct hwtstamp_config ts_config, netdev_features_t new_features); - +void mlx4_en_update_pfc_stats_bitmap(struct mlx4_dev *dev, + struct mlx4_en_stats_bitmap *stats_bitmap, + u8 rx_ppp, u8 rx_pause, + u8 tx_ppp, u8 tx_pause); int mlx4_en_netdev_event(struct notifier_block *this, unsigned long event, void *ptr); diff --git a/drivers/net/ethernet/mellanox/mlx4/mlx4_stats.h b/drivers/net/ethernet/mellanox/mlx4/mlx4_stats.h new file mode 100644 index 000000000000..00555832a4ae --- /dev/null +++ b/drivers/net/ethernet/mellanox/mlx4/mlx4_stats.h @@ -0,0 +1,107 @@ +#ifndef _MLX4_STATS_ +#define _MLX4_STATS_ + +#ifdef MLX4_EN_PERF_STAT +#define NUM_PERF_STATS NUM_PERF_COUNTERS +#else +#define NUM_PERF_STATS 0 +#endif + +#define NUM_PRIORITIES 9 +#define NUM_PRIORITY_STATS 2 + +struct mlx4_en_pkt_stats { + unsigned long rx_multicast_packets; + unsigned long rx_broadcast_packets; + unsigned long rx_jabbers; + unsigned long rx_in_range_length_error; + unsigned long rx_out_range_length_error; + unsigned long tx_multicast_packets; + unsigned long tx_broadcast_packets; + unsigned long rx_prio[NUM_PRIORITIES][NUM_PRIORITY_STATS]; + unsigned long tx_prio[NUM_PRIORITIES][NUM_PRIORITY_STATS]; +#define NUM_PKT_STATS 43 +}; + +struct mlx4_en_port_stats { + unsigned long tso_packets; + unsigned long xmit_more; + unsigned long queue_stopped; + unsigned long wake_queue; + unsigned long tx_timeout; + unsigned long rx_alloc_failed; + unsigned long rx_chksum_good; + unsigned long rx_chksum_none; + unsigned long rx_chksum_complete; + unsigned long tx_chksum_offload; +#define NUM_PORT_STATS 10 +}; + +struct mlx4_en_perf_stats { + u32 tx_poll; + u64 tx_pktsz_avg; + u32 inflight_avg; + u16 tx_coal_avg; + u16 rx_coal_avg; + u32 napi_quota; +#define NUM_PERF_COUNTERS 6 +}; + +#define NUM_MAIN_STATS 21 + +#define MLX4_NUM_PRIORITIES 8 + +struct mlx4_en_flow_stats_rx { + u64 rx_pause; + u64 rx_pause_duration; + u64 rx_pause_transition; +#define NUM_FLOW_STATS_RX 3 +#define NUM_FLOW_PRIORITY_STATS_RX (NUM_FLOW_STATS_RX * \ + MLX4_NUM_PRIORITIES) +}; + +struct mlx4_en_flow_stats_tx { + u64 tx_pause; + u64 tx_pause_duration; + u64 tx_pause_transition; +#define NUM_FLOW_STATS_TX 3 +#define NUM_FLOW_PRIORITY_STATS_TX (NUM_FLOW_STATS_TX * \ + MLX4_NUM_PRIORITIES) +}; + +#define NUM_FLOW_STATS (NUM_FLOW_STATS_RX + NUM_FLOW_STATS_TX + \ + NUM_FLOW_PRIORITY_STATS_TX + \ + NUM_FLOW_PRIORITY_STATS_RX) + +struct mlx4_en_stat_out_flow_control_mbox { + /* Total number of PAUSE frames received from the far-end port */ + __be64 rx_pause; + /* Total number of microseconds that far-end port requested to pause + * transmission of packets + */ + __be64 rx_pause_duration; + /* Number of received transmission from XOFF state to XON state */ + __be64 rx_pause_transition; + /* Total number of PAUSE frames sent from the far-end port */ + __be64 tx_pause; + /* Total time in microseconds that transmission of packets has been + * paused + */ + __be64 tx_pause_duration; + /* Number of transmitter transitions from XOFF state to XON state */ + __be64 tx_pause_transition; + /* Reserverd */ + __be64 reserved[2]; +}; + +enum { + MLX4_DUMP_ETH_STATS_FLOW_CONTROL = 1 << 12 +}; + +#define NUM_ALL_STATS (NUM_MAIN_STATS + NUM_PORT_STATS + NUM_PKT_STATS + \ + NUM_FLOW_STATS + NUM_PERF_STATS) + +#define MLX4_FIND_NETDEV_STAT(n) (offsetof(struct net_device_stats, n) / \ + sizeof(((struct net_device_stats *)0)->n)) + +#endif diff --git a/drivers/net/ethernet/mellanox/mlx4/port.c b/drivers/net/ethernet/mellanox/mlx4/port.c index 9f268f05290a..c2b21313dba7 100644 --- a/drivers/net/ethernet/mellanox/mlx4/port.c +++ b/drivers/net/ethernet/mellanox/mlx4/port.c @@ -38,6 +38,7 @@ #include <linux/mlx4/cmd.h> #include "mlx4.h" +#include "mlx4_stats.h" #define MLX4_MAC_VALID (1ull << 63) @@ -49,6 +50,9 @@ #define MLX4_STATS_ERROR_COUNTERS_MASK 0x1ffc30ULL #define MLX4_STATS_PORT_COUNTERS_MASK 0x1fe00000ULL +#define MLX4_FLAG_V_IGNORE_FCS_MASK 0x2 +#define MLX4_IGNORE_FCS_MASK 0x1 + void mlx4_init_mac_table(struct mlx4_dev *dev, struct mlx4_mac_table *table) { int i; @@ -127,8 +131,9 @@ static int mlx4_set_port_mac_table(struct mlx4_dev *dev, u8 port, in_mod = MLX4_SET_PORT_MAC_TABLE << 8 | port; - err = mlx4_cmd(dev, mailbox->dma, in_mod, 1, MLX4_CMD_SET_PORT, - MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE); + err = mlx4_cmd(dev, mailbox->dma, in_mod, MLX4_SET_PORT_ETH_OPCODE, + MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B, + MLX4_CMD_NATIVE); mlx4_free_cmd_mailbox(dev, mailbox); return err; @@ -341,8 +346,9 @@ static int mlx4_set_port_vlan_table(struct mlx4_dev *dev, u8 port, memcpy(mailbox->buf, entries, MLX4_VLAN_TABLE_SIZE); in_mod = MLX4_SET_PORT_VLAN_TABLE << 8 | port; - err = mlx4_cmd(dev, mailbox->dma, in_mod, 1, MLX4_CMD_SET_PORT, - MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE); + err = mlx4_cmd(dev, mailbox->dma, in_mod, MLX4_SET_PORT_ETH_OPCODE, + MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B, + MLX4_CMD_NATIVE); mlx4_free_cmd_mailbox(dev, mailbox); @@ -629,9 +635,9 @@ static int mlx4_reset_roce_port_gids(struct mlx4_dev *dev, int slave, MLX4_ROCE_GID_ENTRY_SIZE); err = mlx4_cmd(dev, mailbox->dma, - ((u32)port) | (MLX4_SET_PORT_GID_TABLE << 8), 1, - MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B, - MLX4_CMD_NATIVE); + ((u32)port) | (MLX4_SET_PORT_GID_TABLE << 8), + MLX4_SET_PORT_ETH_OPCODE, MLX4_CMD_SET_PORT, + MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE); mutex_unlock(&(priv->port[port].gid_table.mutex)); return err; } @@ -837,6 +843,12 @@ static int mlx4_common_set_port(struct mlx4_dev *dev, int slave, u32 in_mod, MLX4_CMD_NATIVE); } + /* Slaves are not allowed to SET_PORT beacon (LED) blink */ + if (op_mod == MLX4_SET_PORT_BEACON_OPCODE) { + mlx4_warn(dev, "denying SET_PORT Beacon slave:%d\n", slave); + return -EPERM; + } + /* For IB, we only consider: * - The capability mask, which is set to the aggregate of all * slave function capabilities @@ -945,8 +957,9 @@ int mlx4_SET_PORT(struct mlx4_dev *dev, u8 port, int pkey_tbl_sz) (pkey_tbl_flag << MLX4_CHANGE_PORT_PKEY_TBL_SZ) | (dev->caps.port_ib_mtu[port] << MLX4_SET_PORT_MTU_CAP) | (vl_cap << MLX4_SET_PORT_VL_CAP)); - err = mlx4_cmd(dev, mailbox->dma, port, 0, MLX4_CMD_SET_PORT, - MLX4_CMD_TIME_CLASS_B, MLX4_CMD_WRAPPED); + err = mlx4_cmd(dev, mailbox->dma, port, + MLX4_SET_PORT_IB_OPCODE, MLX4_CMD_SET_PORT, + MLX4_CMD_TIME_CLASS_B, MLX4_CMD_WRAPPED); if (err != -ENOMEM) break; } @@ -975,8 +988,9 @@ int mlx4_SET_PORT_general(struct mlx4_dev *dev, u8 port, int mtu, context->pfcrx = pfcrx; in_mod = MLX4_SET_PORT_GENERAL << 8 | port; - err = mlx4_cmd(dev, mailbox->dma, in_mod, 1, MLX4_CMD_SET_PORT, - MLX4_CMD_TIME_CLASS_B, MLX4_CMD_WRAPPED); + err = mlx4_cmd(dev, mailbox->dma, in_mod, MLX4_SET_PORT_ETH_OPCODE, + MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B, + MLX4_CMD_WRAPPED); mlx4_free_cmd_mailbox(dev, mailbox); return err; @@ -1012,84 +1026,40 @@ int mlx4_SET_PORT_qpn_calc(struct mlx4_dev *dev, u8 port, u32 base_qpn, context->vlan_miss = MLX4_VLAN_MISS_IDX; in_mod = MLX4_SET_PORT_RQP_CALC << 8 | port; - err = mlx4_cmd(dev, mailbox->dma, in_mod, 1, MLX4_CMD_SET_PORT, - MLX4_CMD_TIME_CLASS_B, MLX4_CMD_WRAPPED); + err = mlx4_cmd(dev, mailbox->dma, in_mod, MLX4_SET_PORT_ETH_OPCODE, + MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B, + MLX4_CMD_WRAPPED); mlx4_free_cmd_mailbox(dev, mailbox); return err; } EXPORT_SYMBOL(mlx4_SET_PORT_qpn_calc); -int mlx4_SET_PORT_PRIO2TC(struct mlx4_dev *dev, u8 port, u8 *prio2tc) +int mlx4_SET_PORT_fcs_check(struct mlx4_dev *dev, u8 port, u8 ignore_fcs_value) { struct mlx4_cmd_mailbox *mailbox; - struct mlx4_set_port_prio2tc_context *context; - int err; + struct mlx4_set_port_general_context *context; u32 in_mod; - int i; - - mailbox = mlx4_alloc_cmd_mailbox(dev); - if (IS_ERR(mailbox)) - return PTR_ERR(mailbox); - context = mailbox->buf; - for (i = 0; i < MLX4_NUM_UP; i += 2) - context->prio2tc[i >> 1] = prio2tc[i] << 4 | prio2tc[i + 1]; - - in_mod = MLX4_SET_PORT_PRIO2TC << 8 | port; - err = mlx4_cmd(dev, mailbox->dma, in_mod, 1, MLX4_CMD_SET_PORT, - MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE); - - mlx4_free_cmd_mailbox(dev, mailbox); - return err; -} -EXPORT_SYMBOL(mlx4_SET_PORT_PRIO2TC); - -int mlx4_SET_PORT_SCHEDULER(struct mlx4_dev *dev, u8 port, u8 *tc_tx_bw, - u8 *pg, u16 *ratelimit) -{ - struct mlx4_cmd_mailbox *mailbox; - struct mlx4_set_port_scheduler_context *context; int err; - u32 in_mod; - int i; mailbox = mlx4_alloc_cmd_mailbox(dev); if (IS_ERR(mailbox)) return PTR_ERR(mailbox); context = mailbox->buf; + context->v_ignore_fcs |= MLX4_FLAG_V_IGNORE_FCS_MASK; + if (ignore_fcs_value) + context->ignore_fcs |= MLX4_IGNORE_FCS_MASK; + else + context->ignore_fcs &= ~MLX4_IGNORE_FCS_MASK; - for (i = 0; i < MLX4_NUM_TC; i++) { - struct mlx4_port_scheduler_tc_cfg_be *tc = &context->tc[i]; - u16 r; - - if (ratelimit && ratelimit[i]) { - if (ratelimit[i] <= MLX4_MAX_100M_UNITS_VAL) { - r = ratelimit[i]; - tc->max_bw_units = - htons(MLX4_RATELIMIT_100M_UNITS); - } else { - r = ratelimit[i]/10; - tc->max_bw_units = - htons(MLX4_RATELIMIT_1G_UNITS); - } - tc->max_bw_value = htons(r); - } else { - tc->max_bw_value = htons(MLX4_RATELIMIT_DEFAULT); - tc->max_bw_units = htons(MLX4_RATELIMIT_1G_UNITS); - } - - tc->pg = htons(pg[i]); - tc->bw_precentage = htons(tc_tx_bw[i]); - } - - in_mod = MLX4_SET_PORT_SCHEDULER << 8 | port; + in_mod = MLX4_SET_PORT_GENERAL << 8 | port; err = mlx4_cmd(dev, mailbox->dma, in_mod, 1, MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE); mlx4_free_cmd_mailbox(dev, mailbox); return err; } -EXPORT_SYMBOL(mlx4_SET_PORT_SCHEDULER); +EXPORT_SYMBOL(mlx4_SET_PORT_fcs_check); enum { VXLAN_ENABLE_MODIFY = 1 << 7, @@ -1125,14 +1095,35 @@ int mlx4_SET_PORT_VXLAN(struct mlx4_dev *dev, u8 port, u8 steering, int enable) context->steering = steering; in_mod = MLX4_SET_PORT_VXLAN << 8 | port; - err = mlx4_cmd(dev, mailbox->dma, in_mod, 1, MLX4_CMD_SET_PORT, - MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE); + err = mlx4_cmd(dev, mailbox->dma, in_mod, MLX4_SET_PORT_ETH_OPCODE, + MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B, + MLX4_CMD_NATIVE); mlx4_free_cmd_mailbox(dev, mailbox); return err; } EXPORT_SYMBOL(mlx4_SET_PORT_VXLAN); +int mlx4_SET_PORT_BEACON(struct mlx4_dev *dev, u8 port, u16 time) +{ + int err; + struct mlx4_cmd_mailbox *mailbox; + + mailbox = mlx4_alloc_cmd_mailbox(dev); + if (IS_ERR(mailbox)) + return PTR_ERR(mailbox); + + *((__be32 *)mailbox->buf) = cpu_to_be32(time); + + err = mlx4_cmd(dev, mailbox->dma, port, MLX4_SET_PORT_BEACON_OPCODE, + MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B, + MLX4_CMD_NATIVE); + + mlx4_free_cmd_mailbox(dev, mailbox); + return err; +} +EXPORT_SYMBOL(mlx4_SET_PORT_BEACON); + int mlx4_SET_MCAST_FLTR_wrapper(struct mlx4_dev *dev, int slave, struct mlx4_vhcr *vhcr, struct mlx4_cmd_mailbox *inbox, @@ -1184,22 +1175,6 @@ int mlx4_DUMP_ETH_STATS_wrapper(struct mlx4_dev *dev, int slave, vhcr->in_modifier, outbox); } -void mlx4_set_stats_bitmap(struct mlx4_dev *dev, u64 *stats_bitmap) -{ - if (!mlx4_is_mfunc(dev)) { - *stats_bitmap = 0; - return; - } - - *stats_bitmap = (MLX4_STATS_TRAFFIC_COUNTERS_MASK | - MLX4_STATS_TRAFFIC_DROPS_MASK | - MLX4_STATS_PORT_COUNTERS_MASK); - - if (mlx4_is_master(dev)) - *stats_bitmap |= MLX4_STATS_ERROR_COUNTERS_MASK; -} -EXPORT_SYMBOL(mlx4_set_stats_bitmap); - int mlx4_get_slave_from_roce_gid(struct mlx4_dev *dev, int port, u8 *gid, int *slave_id) { diff --git a/drivers/net/ethernet/mellanox/mlx4/qp.c b/drivers/net/ethernet/mellanox/mlx4/qp.c index eda29dbbfcd2..b75214a80d0e 100644 --- a/drivers/net/ethernet/mellanox/mlx4/qp.c +++ b/drivers/net/ethernet/mellanox/mlx4/qp.c @@ -442,6 +442,16 @@ int mlx4_update_qp(struct mlx4_dev *dev, u32 qpn, cmd->qp_context.param3 |= cpu_to_be32(MLX4_STRIP_VLAN); } + if (attr & MLX4_UPDATE_QP_RATE_LIMIT) { + qp_mask |= 1ULL << MLX4_UPD_QP_MASK_RATE_LIMIT; + cmd->qp_context.rate_limit_params = cpu_to_be16((params->rate_unit << 14) | params->rate_val); + } + + if (attr & MLX4_UPDATE_QP_QOS_VPORT) { + qp_mask |= 1ULL << MLX4_UPD_QP_MASK_QOS_VPP; + cmd->qp_context.qos_vport = params->qos_vport; + } + cmd->primary_addr_path_mask = cpu_to_be64(pri_addr_path_mask); cmd->qp_mask = cpu_to_be64(qp_mask); diff --git a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c index 6e413ac4e940..c7f28bf4b8e2 100644 --- a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c +++ b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c @@ -221,11 +221,6 @@ struct res_fs_rule { int qpn; }; -static int mlx4_is_eth(struct mlx4_dev *dev, int port) -{ - return dev->caps.port_mask[port] == MLX4_PORT_TYPE_IB ? 0 : 1; -} - static void *res_tracker_lookup(struct rb_root *root, u64 res_id) { struct rb_node *node = root->rb_node; @@ -770,6 +765,7 @@ static int update_vport_qp_param(struct mlx4_dev *dev, qpc->pri_path.feup |= MLX4_FEUP_FORCE_ETH_UP | MLX4_FVL_FORCE_ETH_VLAN; qpc->pri_path.sched_queue &= 0xC7; qpc->pri_path.sched_queue |= (vp_oper->state.default_qos) << 3; + qpc->qos_vport = vp_oper->state.qos_vport; } if (vp_oper->state.spoofchk) { qpc->pri_path.feup |= MLX4_FSM_FORCE_ETH_SRC_MAC; @@ -2947,8 +2943,12 @@ static int verify_qp_parameters(struct mlx4_dev *dev, qp_type = (be32_to_cpu(qp_ctx->flags) >> 16) & 0xff; optpar = be32_to_cpu(*(__be32 *) inbox->buf); - if (slave != mlx4_master_func_num(dev)) + if (slave != mlx4_master_func_num(dev)) { qp_ctx->params2 &= ~MLX4_QP_BIT_FPP; + /* setting QP rate-limit is disallowed for VFs */ + if (qp_ctx->rate_limit_params) + return -EPERM; + } switch (qp_type) { case MLX4_QP_ST_RC: @@ -3027,7 +3027,7 @@ int mlx4_WRITE_MTT_wrapper(struct mlx4_dev *dev, int slave, /* Call the SW implementation of write_mtt: * - Prepare a dummy mtt struct - * - Translate inbox contents to simple addresses in host endianess */ + * - Translate inbox contents to simple addresses in host endianness */ mtt.offset = 0; /* TBD this is broken but I don't handle it since we don't really use it */ mtt.order = 0; @@ -4918,6 +4918,11 @@ void mlx4_vf_immed_vlan_work_handler(struct work_struct *_work) qp->sched_queue & 0xC7; upd_context->qp_context.pri_path.sched_queue |= ((work->qos & 0x7) << 3); + upd_context->qp_mask |= + cpu_to_be64(1ULL << + MLX4_UPD_QP_MASK_QOS_VPP); + upd_context->qp_context.qos_vport = + work->qos_vport; } err = mlx4_cmd(dev, mailbox->dma, |