i40e: support DCB mode
authorJingjing Wu <jingjing.wu@intel.com>
Sat, 31 Oct 2015 15:57:25 +0000 (23:57 +0800)
committerThomas Monjalon <thomas.monjalon@6wind.com>
Sun, 1 Nov 2015 13:52:02 +0000 (14:52 +0100)
This patch enables DCB feature on Intel XL710/X710 NICs. It includes:
  Receive queue classification based on traffic class
  Round Robin ETS schedule (rx and tx)
  Priority flow control

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Acked-by: Jijiang Liu <jijiang.liu@intel.com>
Acked-by: Helin Zhang <helin.zhang@intel.com>
doc/guides/rel_notes/release_2_2.rst
drivers/net/i40e/i40e_ethdev.c
drivers/net/i40e/i40e_ethdev.h
drivers/net/i40e/i40e_rxtx.c
drivers/net/i40e/i40e_rxtx.h

index 9a6d15e..ad16691 100644 (file)
@@ -22,6 +22,8 @@ New Features
 
 * **Added i40e flow control support.**
 
+* **Added DCB support to i40e PF driver.**
+
 * **Added fm10k TSO support for both PF and VF.**
 
 * **New NIC Boulder Rapid support.**
index 016838a..ce4efb2 100644 (file)
@@ -56,6 +56,7 @@
 #include "base/i40e_adminq_cmd.h"
 #include "base/i40e_type.h"
 #include "base/i40e_register.h"
+#include "base/i40e_dcb.h"
 #include "i40e_ethdev.h"
 #include "i40e_rxtx.h"
 #include "i40e_pf.h"
 #define I40E_PRTTSYN_TSYNENA  0x80000000
 #define I40E_PRTTSYN_TSYNTYPE 0x0e000000
 
+#define I40E_MAX_PERCENT            100
+#define I40E_DEFAULT_DCB_APP_NUM    1
+#define I40E_DEFAULT_DCB_APP_PRIO   3
+
 static int eth_i40e_dev_init(struct rte_eth_dev *eth_dev);
 static int eth_i40e_dev_uninit(struct rte_eth_dev *eth_dev);
 static int i40e_dev_configure(struct rte_eth_dev *dev);
@@ -189,6 +194,8 @@ static int i40e_pf_parameter_init(struct rte_eth_dev *dev);
 static int i40e_pf_setup(struct i40e_pf *pf);
 static int i40e_dev_rxtx_init(struct i40e_pf *pf);
 static int i40e_vmdq_setup(struct rte_eth_dev *dev);
+static int i40e_dcb_init_configure(struct rte_eth_dev *dev, bool sw_dcb);
+static int i40e_dcb_setup(struct rte_eth_dev *dev);
 static void i40e_stat_update_32(struct i40e_hw *hw, uint32_t reg,
                bool offset_loaded, uint64_t *offset, uint64_t *stat);
 static void i40e_stat_update_48(struct i40e_hw *hw,
@@ -517,11 +524,6 @@ eth_i40e_dev_init(struct rte_eth_dev *dev)
                     ((hw->nvm.version >> 4) & 0xff),
                     (hw->nvm.version & 0xf), hw->nvm.eetrack);
 
-       /* Disable LLDP */
-       ret = i40e_aq_stop_lldp(hw, true, NULL);
-       if (ret != I40E_SUCCESS) /* Its failure can be ignored */
-               PMD_INIT_LOG(INFO, "Failed to stop lldp");
-
        /* Clear PXE mode */
        i40e_clear_pxe_mode(hw);
 
@@ -642,6 +644,13 @@ eth_i40e_dev_init(struct rte_eth_dev *dev)
        /* initialize mirror rule list */
        TAILQ_INIT(&pf->mirror_list);
 
+       /* Init dcb to sw mode by default */
+       ret = i40e_dcb_init_configure(dev, TRUE);
+       if (ret != I40E_SUCCESS) {
+               PMD_INIT_LOG(INFO, "Failed to init dcb.");
+               pf->flags &= ~I40E_FLAG_DCB;
+       }
+
        return 0;
 
 err_mac_alloc:
@@ -728,7 +737,7 @@ i40e_dev_configure(struct rte_eth_dev *dev)
                I40E_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
        struct i40e_pf *pf = I40E_DEV_PRIVATE_TO_PF(dev->data->dev_private);
        enum rte_eth_rx_mq_mode mq_mode = dev->data->dev_conf.rxmode.mq_mode;
-       int ret;
+       int i, ret;
 
        /* Initialize to TRUE. If any of Rx queues doesn't meet the
         * bulk allocation or vector Rx preconditions we will reset it.
@@ -773,8 +782,27 @@ i40e_dev_configure(struct rte_eth_dev *dev)
                if (ret)
                        goto err;
        }
+
+       if (mq_mode & ETH_MQ_RX_DCB_FLAG) {
+               ret = i40e_dcb_setup(dev);
+               if (ret) {
+                       PMD_DRV_LOG(ERR, "failed to configure DCB.");
+                       goto err_dcb;
+               }
+       }
+
        return 0;
+
+err_dcb:
+       /* need to release vmdq resource if exists */
+       for (i = 0; i < pf->nb_cfg_vmdq_vsi; i++) {
+               i40e_vsi_release(pf->vmdq[i].vsi);
+               pf->vmdq[i].vsi = NULL;
+       }
+       rte_free(pf->vmdq);
+       pf->vmdq = NULL;
 err:
+       /* need to release fdir resource if exists */
        i40e_fdir_teardown(pf);
        return ret;
 }
@@ -2517,6 +2545,9 @@ i40e_pf_parameter_init(struct rte_eth_dev *dev)
                 */
        }
 
+       if (hw->func_caps.dcb)
+               pf->flags |= I40E_FLAG_DCB;
+
        if (sum_vsis > pf->max_num_vsi ||
                sum_queues > hw->func_caps.num_rx_qp) {
                PMD_INIT_LOG(ERR, "VSI/QUEUE setting can't be satisfied");
@@ -2922,7 +2953,7 @@ i40e_vsi_config_tc_queue_mapping(struct i40e_vsi *vsi,
                                 struct i40e_aqc_vsi_properties_data *info,
                                 uint8_t enabled_tcmap)
 {
-       int ret, total_tc = 0, i;
+       int ret, i, total_tc = 0;
        uint16_t qpnum_per_tc, bsf, qp_idx;
 
        ret = validate_tcmap_parameter(vsi, enabled_tcmap);
@@ -5479,11 +5510,6 @@ i40e_pf_config_mq_rx(struct i40e_pf *pf)
        int ret = 0;
        enum rte_eth_rx_mq_mode mq_mode = pf->dev_data->dev_conf.rxmode.mq_mode;
 
-       if (mq_mode & ETH_MQ_RX_DCB_FLAG) {
-               PMD_INIT_LOG(ERR, "i40e doesn't support DCB yet");
-               return -ENOTSUP;
-       }
-
        /* RSS setup */
        if (mq_mode & ETH_MQ_RX_RSS_FLAG)
                ret = i40e_pf_config_rss(pf);
@@ -6508,3 +6534,485 @@ i40e_timesync_read_tx_timestamp(struct rte_eth_dev *dev,
 
        return  0;
 }
+
+/*
+ * i40e_parse_dcb_configure - parse dcb configure from user
+ * @dev: the device being configured
+ * @dcb_cfg: pointer of the result of parse
+ * @*tc_map: bit map of enabled traffic classes
+ *
+ * Returns 0 on success, negative value on failure
+ */
+static int
+i40e_parse_dcb_configure(struct rte_eth_dev *dev,
+                        struct i40e_dcbx_config *dcb_cfg,
+                        uint8_t *tc_map)
+{
+       struct rte_eth_dcb_rx_conf *dcb_rx_conf;
+       uint8_t i, tc_bw, bw_lf;
+
+       memset(dcb_cfg, 0, sizeof(struct i40e_dcbx_config));
+
+       dcb_rx_conf = &dev->data->dev_conf.rx_adv_conf.dcb_rx_conf;
+       if (dcb_rx_conf->nb_tcs > I40E_MAX_TRAFFIC_CLASS) {
+               PMD_INIT_LOG(ERR, "number of tc exceeds max.");
+               return -EINVAL;
+       }
+
+       /* assume each tc has the same bw */
+       tc_bw = I40E_MAX_PERCENT / dcb_rx_conf->nb_tcs;
+       for (i = 0; i < dcb_rx_conf->nb_tcs; i++)
+               dcb_cfg->etscfg.tcbwtable[i] = tc_bw;
+       /* to ensure the sum of tcbw is equal to 100 */
+       bw_lf = I40E_MAX_PERCENT % dcb_rx_conf->nb_tcs;
+       for (i = 0; i < bw_lf; i++)
+               dcb_cfg->etscfg.tcbwtable[i]++;
+
+       /* assume each tc has the same Transmission Selection Algorithm */
+       for (i = 0; i < dcb_rx_conf->nb_tcs; i++)
+               dcb_cfg->etscfg.tsatable[i] = I40E_IEEE_TSA_ETS;
+
+       for (i = 0; i < I40E_MAX_USER_PRIORITY; i++)
+               dcb_cfg->etscfg.prioritytable[i] =
+                               dcb_rx_conf->dcb_tc[i];
+
+       /* FW needs one App to configure HW */
+       dcb_cfg->numapps = I40E_DEFAULT_DCB_APP_NUM;
+       dcb_cfg->app[0].selector = I40E_APP_SEL_ETHTYPE;
+       dcb_cfg->app[0].priority = I40E_DEFAULT_DCB_APP_PRIO;
+       dcb_cfg->app[0].protocolid = I40E_APP_PROTOID_FCOE;
+
+       if (dcb_rx_conf->nb_tcs == 0)
+               *tc_map = 1; /* tc0 only */
+       else
+               *tc_map = RTE_LEN2MASK(dcb_rx_conf->nb_tcs, uint8_t);
+
+       if (dev->data->dev_conf.dcb_capability_en & ETH_DCB_PFC_SUPPORT) {
+               dcb_cfg->pfc.willing = 0;
+               dcb_cfg->pfc.pfccap = I40E_MAX_TRAFFIC_CLASS;
+               dcb_cfg->pfc.pfcenable = *tc_map;
+       }
+       return 0;
+}
+
+/*
+ * i40e_vsi_get_bw_info - Query VSI BW Information
+ * @vsi: the VSI being queried
+ *
+ * Returns 0 on success, negative value on failure
+ */
+static int
+i40e_vsi_get_bw_info(struct i40e_vsi *vsi)
+{
+       struct i40e_aqc_query_vsi_ets_sla_config_resp bw_ets_config = {0};
+       struct i40e_aqc_query_vsi_bw_config_resp bw_config = {0};
+       struct i40e_hw *hw = I40E_VSI_TO_HW(vsi);
+       int i, ret;
+       uint32_t tc_bw_max;
+
+       /* Get the VSI level BW configuration */
+       ret = i40e_aq_query_vsi_bw_config(hw, vsi->seid, &bw_config, NULL);
+       if (ret) {
+               PMD_INIT_LOG(ERR,
+                        "couldn't get PF vsi bw config, err %s aq_err %s\n",
+                        i40e_stat_str(hw, ret),
+                        i40e_aq_str(hw, hw->aq.asq_last_status));
+               return -EINVAL;
+       }
+
+       /* Get the VSI level BW configuration per TC */
+       ret = i40e_aq_query_vsi_ets_sla_config(hw, vsi->seid, &bw_ets_config,
+                                                 NULL);
+       if (ret) {
+               PMD_INIT_LOG(ERR,
+                        "couldn't get PF vsi ets bw config, err %s aq_err %s\n",
+                        i40e_stat_str(hw, ret),
+                        i40e_aq_str(hw, hw->aq.asq_last_status));
+               return -EINVAL;
+       }
+
+       if (bw_config.tc_valid_bits != bw_ets_config.tc_valid_bits) {
+               PMD_INIT_LOG(WARNING,
+                        "Enabled TCs mismatch from querying VSI BW info"
+                        " 0x%08x 0x%08x\n", bw_config.tc_valid_bits,
+                        bw_ets_config.tc_valid_bits);
+               /* Still continuing */
+       }
+
+       vsi->bw_info.bw_limit = rte_le_to_cpu_16(bw_config.port_bw_limit);
+       vsi->bw_info.bw_max_quanta = bw_config.max_bw;
+       tc_bw_max = rte_le_to_cpu_16(bw_ets_config.tc_bw_max[0]) |
+                   (rte_le_to_cpu_16(bw_ets_config.tc_bw_max[1]) << 16);
+       for (i = 0; i < I40E_MAX_TRAFFIC_CLASS; i++) {
+               vsi->bw_info.bw_ets_share_credits[i] =
+                               bw_ets_config.share_credits[i];
+               vsi->bw_info.bw_ets_limit_credits[i] =
+                               rte_le_to_cpu_16(bw_ets_config.credits[i]);
+               /* 3 bits out of 4 for each TC */
+               vsi->bw_info.bw_ets_max_quanta[i] =
+                       (uint8_t)((tc_bw_max >> (i * 4)) & 0x7);
+               PMD_INIT_LOG(DEBUG,
+                        "%s: vsi seid = %d, TC = %d, qset = 0x%x\n",
+                        __func__, vsi->seid, i, bw_config.qs_handles[i]);
+       }
+
+       return 0;
+}
+
+static int
+i40e_vsi_update_queue_mapping(struct i40e_vsi *vsi,
+                             struct i40e_aqc_vsi_properties_data *info,
+                             uint8_t enabled_tcmap)
+{
+       int ret, i, total_tc = 0;
+       uint16_t qpnum_per_tc, bsf, qp_idx;
+       struct rte_eth_dev_data *dev_data = I40E_VSI_TO_DEV_DATA(vsi);
+
+       ret = validate_tcmap_parameter(vsi, enabled_tcmap);
+       if (ret != I40E_SUCCESS)
+               return ret;
+
+       for (i = 0; i < I40E_MAX_TRAFFIC_CLASS; i++) {
+               if (enabled_tcmap & (1 << i))
+                       total_tc++;
+       }
+       if (total_tc == 0)
+               total_tc = 1;
+       vsi->enabled_tc = enabled_tcmap;
+
+       qpnum_per_tc = dev_data->nb_rx_queues / total_tc;
+       /* Number of queues per enabled TC */
+       if (qpnum_per_tc == 0) {
+               PMD_INIT_LOG(ERR, " number of queues is less that tcs.");
+               return I40E_ERR_INVALID_QP_ID;
+       }
+       qpnum_per_tc = RTE_MIN(i40e_align_floor(qpnum_per_tc),
+                               I40E_MAX_Q_PER_TC);
+       bsf = rte_bsf32(qpnum_per_tc);
+
+       /**
+        * Configure TC and queue mapping parameters, for enabled TC,
+        * allocate qpnum_per_tc queues to this traffic. For disabled TC,
+        * default queue will serve it.
+        */
+       qp_idx = 0;
+       for (i = 0; i < I40E_MAX_TRAFFIC_CLASS; i++) {
+               if (vsi->enabled_tc & (1 << i)) {
+                       info->tc_mapping[i] = rte_cpu_to_le_16((qp_idx <<
+                                       I40E_AQ_VSI_TC_QUE_OFFSET_SHIFT) |
+                               (bsf << I40E_AQ_VSI_TC_QUE_NUMBER_SHIFT));
+                       qp_idx += qpnum_per_tc;
+               } else
+                       info->tc_mapping[i] = 0;
+       }
+
+       /* Associate queue number with VSI, Keep vsi->nb_qps unchanged */
+       if (vsi->type == I40E_VSI_SRIOV) {
+               info->mapping_flags |=
+                       rte_cpu_to_le_16(I40E_AQ_VSI_QUE_MAP_NONCONTIG);
+               for (i = 0; i < vsi->nb_qps; i++)
+                       info->queue_mapping[i] =
+                               rte_cpu_to_le_16(vsi->base_queue + i);
+       } else {
+               info->mapping_flags |=
+                       rte_cpu_to_le_16(I40E_AQ_VSI_QUE_MAP_CONTIG);
+               info->queue_mapping[0] = rte_cpu_to_le_16(vsi->base_queue);
+       }
+       info->valid_sections |=
+               rte_cpu_to_le_16(I40E_AQ_VSI_PROP_QUEUE_MAP_VALID);
+
+       return I40E_SUCCESS;
+}
+
+/*
+ * i40e_vsi_config_tc - Configure VSI tc setting for given TC map
+ * @vsi: VSI to be configured
+ * @tc_map: enabled TC bitmap
+ *
+ * Returns 0 on success, negative value on failure
+ */
+static int
+i40e_vsi_config_tc(struct i40e_vsi *vsi, u8 tc_map)
+{
+       struct i40e_aqc_configure_vsi_tc_bw_data bw_data;
+       struct i40e_vsi_context ctxt;
+       struct i40e_hw *hw = I40E_VSI_TO_HW(vsi);
+       int ret = 0;
+       int i;
+
+       /* Check if enabled_tc is same as existing or new TCs */
+       if (vsi->enabled_tc == tc_map)
+               return ret;
+
+       /* configure tc bandwidth */
+       memset(&bw_data, 0, sizeof(bw_data));
+       bw_data.tc_valid_bits = tc_map;
+       /* Enable ETS TCs with equal BW Share for now across all VSIs */
+       for (i = 0; i < I40E_MAX_TRAFFIC_CLASS; i++) {
+               if (tc_map & BIT_ULL(i))
+                       bw_data.tc_bw_credits[i] = 1;
+       }
+       ret = i40e_aq_config_vsi_tc_bw(hw, vsi->seid, &bw_data, NULL);
+       if (ret) {
+               PMD_INIT_LOG(ERR, "AQ command Config VSI BW allocation"
+                       " per TC failed = %d",
+                       hw->aq.asq_last_status);
+               goto out;
+       }
+       for (i = 0; i < I40E_MAX_TRAFFIC_CLASS; i++)
+               vsi->info.qs_handle[i] = bw_data.qs_handles[i];
+
+       /* Update Queue Pairs Mapping for currently enabled UPs */
+       ctxt.seid = vsi->seid;
+       ctxt.pf_num = hw->pf_id;
+       ctxt.vf_num = 0;
+       ctxt.uplink_seid = vsi->uplink_seid;
+       ctxt.info = vsi->info;
+       i40e_get_cap(hw);
+       ret = i40e_vsi_update_queue_mapping(vsi, &ctxt.info, tc_map);
+       if (ret)
+               goto out;
+
+       /* Update the VSI after updating the VSI queue-mapping information */
+       ret = i40e_aq_update_vsi_params(hw, &ctxt, NULL);
+       if (ret) {
+               PMD_INIT_LOG(ERR, "Failed to configure "
+                           "TC queue mapping = %d",
+                           hw->aq.asq_last_status);
+               goto out;
+       }
+       /* update the local VSI info with updated queue map */
+       (void)rte_memcpy(&vsi->info.tc_mapping, &ctxt.info.tc_mapping,
+                                       sizeof(vsi->info.tc_mapping));
+       (void)rte_memcpy(&vsi->info.queue_mapping,
+                       &ctxt.info.queue_mapping,
+               sizeof(vsi->info.queue_mapping));
+       vsi->info.mapping_flags = ctxt.info.mapping_flags;
+       vsi->info.valid_sections = 0;
+
+       /* Update current VSI BW information */
+       ret = i40e_vsi_get_bw_info(vsi);
+       if (ret) {
+               PMD_INIT_LOG(ERR,
+                        "Failed updating vsi bw info, err %s aq_err %s",
+                        i40e_stat_str(hw, ret),
+                        i40e_aq_str(hw, hw->aq.asq_last_status));
+               goto out;
+       }
+
+       vsi->enabled_tc = tc_map;
+
+out:
+       return ret;
+}
+
+/*
+ * i40e_dcb_hw_configure - program the dcb setting to hw
+ * @pf: pf the configuration is taken on
+ * @new_cfg: new configuration
+ * @tc_map: enabled TC bitmap
+ *
+ * Returns 0 on success, negative value on failure
+ */
+static enum i40e_status_code
+i40e_dcb_hw_configure(struct i40e_pf *pf,
+                     struct i40e_dcbx_config *new_cfg,
+                     uint8_t tc_map)
+{
+       struct i40e_hw *hw = I40E_PF_TO_HW(pf);
+       struct i40e_dcbx_config *old_cfg = &hw->local_dcbx_config;
+       struct i40e_vsi *main_vsi = pf->main_vsi;
+       struct i40e_vsi_list *vsi_list;
+       int i, ret;
+       uint32_t val;
+
+       /* Use the FW API if FW > v4.4*/
+       if (!((hw->aq.fw_maj_ver == 4) && (hw->aq.fw_min_ver >= 4))) {
+               PMD_INIT_LOG(ERR, "FW < v4.4, can not use FW LLDP API"
+                                 " to configure DCB");
+               return I40E_ERR_FIRMWARE_API_VERSION;
+       }
+
+       /* Check if need reconfiguration */
+       if (!memcmp(new_cfg, old_cfg, sizeof(struct i40e_dcbx_config))) {
+               PMD_INIT_LOG(ERR, "No Change in DCB Config required.");
+               return I40E_SUCCESS;
+       }
+
+       /* Copy the new config to the current config */
+       *old_cfg = *new_cfg;
+       old_cfg->etsrec = old_cfg->etscfg;
+       ret = i40e_set_dcb_config(hw);
+       if (ret) {
+               PMD_INIT_LOG(ERR,
+                        "Set DCB Config failed, err %s aq_err %s\n",
+                        i40e_stat_str(hw, ret),
+                        i40e_aq_str(hw, hw->aq.asq_last_status));
+               return ret;
+       }
+       /* set receive Arbiter to RR mode and ETS scheme by default */
+       for (i = 0; i <= I40E_PRTDCB_RETSTCC_MAX_INDEX; i++) {
+               val = I40E_READ_REG(hw, I40E_PRTDCB_RETSTCC(i));
+               val &= ~(I40E_PRTDCB_RETSTCC_BWSHARE_MASK     |
+                        I40E_PRTDCB_RETSTCC_UPINTC_MODE_MASK |
+                        I40E_PRTDCB_RETSTCC_ETSTC_SHIFT);
+               val |= ((uint32_t)old_cfg->etscfg.tcbwtable[i] <<
+                       I40E_PRTDCB_RETSTCC_BWSHARE_SHIFT) &
+                        I40E_PRTDCB_RETSTCC_BWSHARE_MASK;
+               val |= ((uint32_t)1 << I40E_PRTDCB_RETSTCC_UPINTC_MODE_SHIFT) &
+                        I40E_PRTDCB_RETSTCC_UPINTC_MODE_MASK;
+               val |= ((uint32_t)1 << I40E_PRTDCB_RETSTCC_ETSTC_SHIFT) &
+                        I40E_PRTDCB_RETSTCC_ETSTC_MASK;
+               I40E_WRITE_REG(hw, I40E_PRTDCB_RETSTCC(i), val);
+       }
+       /* get local mib to check whether it is configured correctly */
+       /* IEEE mode */
+       hw->local_dcbx_config.dcbx_mode = I40E_DCBX_MODE_IEEE;
+       /* Get Local DCB Config */
+       i40e_aq_get_dcb_config(hw, I40E_AQ_LLDP_MIB_LOCAL, 0,
+                                    &hw->local_dcbx_config);
+
+       /* Update each VSI */
+       i40e_vsi_config_tc(main_vsi, tc_map);
+       if (main_vsi->veb) {
+               TAILQ_FOREACH(vsi_list, &main_vsi->veb->head, list) {
+                       /* Beside main VSI, only enable default
+                        * TC for other VSIs
+                        */
+                       ret = i40e_vsi_config_tc(vsi_list->vsi,
+                                               I40E_DEFAULT_TCMAP);
+                       if (ret)
+                               PMD_INIT_LOG(WARNING,
+                                        "Failed configuring TC for VSI seid=%d\n",
+                                        vsi_list->vsi->seid);
+                       /* continue */
+               }
+       }
+       return I40E_SUCCESS;
+}
+
+/*
+ * i40e_dcb_init_configure - initial dcb config
+ * @dev: device being configured
+ * @sw_dcb: indicate whether dcb is sw configured or hw offload
+ *
+ * Returns 0 on success, negative value on failure
+ */
+static int
+i40e_dcb_init_configure(struct rte_eth_dev *dev, bool sw_dcb)
+{
+       struct i40e_pf *pf = I40E_DEV_PRIVATE_TO_PF(dev->data->dev_private);
+       struct i40e_hw *hw = I40E_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+       int ret = 0;
+
+       if ((pf->flags & I40E_FLAG_DCB) == 0) {
+               PMD_INIT_LOG(ERR, "HW doesn't support DCB");
+               return -ENOTSUP;
+       }
+
+       /* DCB initialization:
+        * Update DCB configuration from the Firmware and configure
+        * LLDP MIB change event.
+        */
+       if (sw_dcb == TRUE) {
+               ret = i40e_aq_stop_lldp(hw, TRUE, NULL);
+               if (ret != I40E_SUCCESS)
+                       PMD_INIT_LOG(DEBUG, "Failed to stop lldp");
+
+               ret = i40e_init_dcb(hw);
+               /* if sw_dcb, lldp agent is stopped, the return from
+                * i40e_init_dcb we expect is failure with I40E_AQ_RC_EPERM
+                * adminq status.
+                */
+               if (ret != I40E_SUCCESS &&
+                   hw->aq.asq_last_status == I40E_AQ_RC_EPERM) {
+                       memset(&hw->local_dcbx_config, 0,
+                               sizeof(struct i40e_dcbx_config));
+                       /* set dcb default configuration */
+                       hw->local_dcbx_config.etscfg.willing = 0;
+                       hw->local_dcbx_config.etscfg.maxtcs = 0;
+                       hw->local_dcbx_config.etscfg.tcbwtable[0] = 100;
+                       hw->local_dcbx_config.etscfg.tsatable[0] =
+                                               I40E_IEEE_TSA_ETS;
+                       hw->local_dcbx_config.etsrec =
+                               hw->local_dcbx_config.etscfg;
+                       hw->local_dcbx_config.pfc.willing = 0;
+                       hw->local_dcbx_config.pfc.pfccap =
+                                               I40E_MAX_TRAFFIC_CLASS;
+                       /* FW needs one App to configure HW */
+                       hw->local_dcbx_config.numapps = 1;
+                       hw->local_dcbx_config.app[0].selector =
+                                               I40E_APP_SEL_ETHTYPE;
+                       hw->local_dcbx_config.app[0].priority = 3;
+                       hw->local_dcbx_config.app[0].protocolid =
+                                               I40E_APP_PROTOID_FCOE;
+                       ret = i40e_set_dcb_config(hw);
+                       if (ret) {
+                               PMD_INIT_LOG(ERR, "default dcb config fails."
+                                       " err = %d, aq_err = %d.", ret,
+                                         hw->aq.asq_last_status);
+                               return -ENOSYS;
+                       }
+               } else {
+                       PMD_INIT_LOG(ERR, "DCBX configuration failed, err = %d,"
+                                         " aq_err = %d.", ret,
+                                         hw->aq.asq_last_status);
+                       return -ENOTSUP;
+               }
+       } else {
+               ret = i40e_aq_start_lldp(hw, NULL);
+               if (ret != I40E_SUCCESS)
+                       PMD_INIT_LOG(DEBUG, "Failed to start lldp");
+
+               ret = i40e_init_dcb(hw);
+               if (!ret) {
+                       if (hw->dcbx_status == I40E_DCBX_STATUS_DISABLED) {
+                               PMD_INIT_LOG(ERR, "HW doesn't support"
+                                                 " DCBX offload.");
+                               return -ENOTSUP;
+                       }
+               } else {
+                       PMD_INIT_LOG(ERR, "DCBX configuration failed, err = %d,"
+                                         " aq_err = %d.", ret,
+                                         hw->aq.asq_last_status);
+                       return -ENOTSUP;
+               }
+       }
+       return 0;
+}
+
+/*
+ * i40e_dcb_setup - setup dcb related config
+ * @dev: device being configured
+ *
+ * Returns 0 on success, negative value on failure
+ */
+static int
+i40e_dcb_setup(struct rte_eth_dev *dev)
+{
+       struct i40e_pf *pf = I40E_DEV_PRIVATE_TO_PF(dev->data->dev_private);
+       struct i40e_dcbx_config dcb_cfg;
+       uint8_t tc_map = 0;
+       int ret = 0;
+
+       if ((pf->flags & I40E_FLAG_DCB) == 0) {
+               PMD_INIT_LOG(ERR, "HW doesn't support DCB");
+               return -ENOTSUP;
+       }
+
+       if (pf->vf_num != 0 ||
+           (dev->data->dev_conf.rxmode.mq_mode & ETH_MQ_RX_VMDQ_FLAG))
+               PMD_INIT_LOG(DEBUG, " DCB only works on main vsi.");
+
+       ret = i40e_parse_dcb_configure(dev, &dcb_cfg, &tc_map);
+       if (ret) {
+               PMD_INIT_LOG(ERR, "invalid dcb config");
+               return -EINVAL;
+       }
+       ret = i40e_dcb_hw_configure(pf, &dcb_cfg, tc_map);
+       if (ret) {
+               PMD_INIT_LOG(ERR, "dcb sw configure fails");
+               return -ENOSYS;
+       }
+       return 0;
+}
index 832c036..0c9c78c 100644 (file)
@@ -204,6 +204,19 @@ struct i40e_macvlan_filter {
        uint16_t vlan_id;
 };
 
+/* Bandwidth limit information */
+struct i40e_bw_info {
+       uint16_t bw_limit;      /* BW Limit (0 = disabled) */
+       uint8_t  bw_max_quanta; /* Max Quanta when BW limit is enabled */
+
+       /* Relative TC credits across VSIs */
+       uint8_t  bw_ets_share_credits[I40E_MAX_TRAFFIC_CLASS];
+       /* TC BW limit credits within VSI */
+       uint8_t  bw_ets_limit_credits[I40E_MAX_TRAFFIC_CLASS];
+       /* TC BW limit max quanta within VSI */
+       uint8_t  bw_ets_max_quanta[I40E_MAX_TRAFFIC_CLASS];
+};
+
 /*
  * Structure that defines a VSI, associated with a adapter.
  */
@@ -249,6 +262,7 @@ struct i40e_vsi {
        uint16_t vsi_id;
        uint16_t msix_intr; /* The MSIX interrupt binds to VSI */
        uint8_t enabled_tc; /* The traffic class enabled */
+       struct i40e_bw_info bw_info; /* VSI bandwidth information */
 };
 
 struct pool_entry {
index 8731712..e7af655 100644 (file)
@@ -2110,7 +2110,8 @@ i40e_dev_rx_queue_setup(struct rte_eth_dev *dev,
        struct i40e_rx_queue *rxq;
        const struct rte_memzone *rz;
        uint32_t ring_size;
-       uint16_t len;
+       uint16_t len, i;
+       uint16_t base, bsf, tc_mapping;
        int use_def_burst_func = 1;
 
        if (hw->mac.type == I40E_MAC_VF) {
@@ -2231,6 +2232,19 @@ i40e_dev_rx_queue_setup(struct rte_eth_dev *dev,
                ad->rx_bulk_alloc_allowed = false;
        }
 
+       for (i = 0; i < I40E_MAX_TRAFFIC_CLASS; i++) {
+               if (!(vsi->enabled_tc & (1 << i)))
+                       continue;
+               tc_mapping = rte_le_to_cpu_16(vsi->info.tc_mapping[i]);
+               base = (tc_mapping & I40E_AQ_VSI_TC_QUE_OFFSET_MASK) >>
+                       I40E_AQ_VSI_TC_QUE_OFFSET_SHIFT;
+               bsf = (tc_mapping & I40E_AQ_VSI_TC_QUE_NUMBER_MASK) >>
+                       I40E_AQ_VSI_TC_QUE_NUMBER_SHIFT;
+
+               if (queue_idx >= base && queue_idx < (base + BIT(bsf)))
+                       rxq->dcb_tc = i;
+       }
+
        return 0;
 }
 
@@ -2323,6 +2337,7 @@ i40e_dev_tx_queue_setup(struct rte_eth_dev *dev,
        const struct rte_memzone *tz;
        uint32_t ring_size;
        uint16_t tx_rs_thresh, tx_free_thresh;
+       uint16_t i, base, bsf, tc_mapping;
 
        if (hw->mac.type == I40E_MAC_VF) {
                struct i40e_vf *vf =
@@ -2492,6 +2507,19 @@ i40e_dev_tx_queue_setup(struct rte_eth_dev *dev,
        /* Use a simple TX queue without offloads or multi segs if possible */
        i40e_set_tx_function_flag(dev, txq);
 
+       for (i = 0; i < I40E_MAX_TRAFFIC_CLASS; i++) {
+               if (!(vsi->enabled_tc & (1 << i)))
+                       continue;
+               tc_mapping = rte_le_to_cpu_16(vsi->info.tc_mapping[i]);
+               base = (tc_mapping & I40E_AQ_VSI_TC_QUE_OFFSET_MASK) >>
+                       I40E_AQ_VSI_TC_QUE_OFFSET_SHIFT;
+               bsf = (tc_mapping & I40E_AQ_VSI_TC_QUE_NUMBER_MASK) >>
+                       I40E_AQ_VSI_TC_QUE_NUMBER_SHIFT;
+
+               if (queue_idx >= base && queue_idx < (base + BIT(bsf)))
+                       txq->dcb_tc = i;
+       }
+
        return 0;
 }
 
@@ -2704,7 +2732,7 @@ i40e_tx_queue_init(struct i40e_tx_queue *txq)
 #ifdef RTE_LIBRTE_IEEE1588
        tx_ctx.timesync_ena = 1;
 #endif
-       tx_ctx.rdylist = rte_le_to_cpu_16(vsi->info.qs_handle[0]);
+       tx_ctx.rdylist = rte_le_to_cpu_16(vsi->info.qs_handle[txq->dcb_tc]);
        if (vsi->type == I40E_VSI_FDIR)
                tx_ctx.fd_ena = TRUE;
 
index 39cb95a..c35828c 100644 (file)
@@ -133,6 +133,7 @@ struct i40e_rx_queue {
        bool q_set; /**< indicate if rx queue has been configured */
        bool rx_deferred_start; /**< don't start this queue in dev start */
        uint16_t rx_using_sse; /**<flag indicate the usage of vPMD for rx */
+       uint8_t dcb_tc;         /**< Traffic class of rx queue */
 };
 
 struct i40e_tx_entry {
@@ -173,6 +174,7 @@ struct i40e_tx_queue {
        uint16_t tx_next_rs;
        bool q_set; /**< indicate if tx queue has been configured */
        bool tx_deferred_start; /**< don't start this queue in dev start */
+       uint8_t dcb_tc;         /**< Traffic class of tx queue */
 };
 
 /** Offload features */