diff options
| author | Kiran Patil <kiran.patil@intel.com> | 2021-12-29 21:54:33 +0300 |
|---|---|---|
| committer | David S. Miller <davem@davemloft.net> | 2021-12-30 16:16:07 +0300 |
| commit | 40319796b7321e25d23671e525e01a1f4d85f6b6 (patch) | |
| tree | 1cdbc64e8a4c073dece11383ac0bcfb3f82ffcea /drivers/net/ethernet/intel/ice/ice_ethtool_fdir.c | |
| parent | a1f18c5fe554750eb14f6b41d2526a13a697c185 (diff) | |
| download | linux-40319796b7321e25d23671e525e01a1f4d85f6b6.tar.xz | |
ice: Add flow director support for channel mode
Add support to enable flow-director filter when multiple TCs are
configured. Flow director filter can be configured using ethtool
(--config-ntuple option). When multiple TCs are configured, each
TC is mapped to an unique HW VSI. So VSI corresponding to queue
used in filter is identified and flow director context is updated
with correct VSI while configuring ntuple filter in HW.
Signed-off-by: Kiran Patil <kiran.patil@intel.com>
Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
Signed-off-by: Sudheer Mogilappagari <sudheer.mogilappagari@intel.com>
Tested-by: Bharathi Sreenivas <bharathi.sreenivas@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'drivers/net/ethernet/intel/ice/ice_ethtool_fdir.c')
| -rw-r--r-- | drivers/net/ethernet/intel/ice/ice_ethtool_fdir.c | 260 |
1 files changed, 251 insertions, 9 deletions
diff --git a/drivers/net/ethernet/intel/ice/ice_ethtool_fdir.c b/drivers/net/ethernet/intel/ice/ice_ethtool_fdir.c index bbc64d6ce4cd..5d10c4f84a36 100644 --- a/drivers/net/ethernet/intel/ice/ice_ethtool_fdir.c +++ b/drivers/net/ethernet/intel/ice/ice_ethtool_fdir.c @@ -5,6 +5,7 @@ #include "ice.h" #include "ice_lib.h" +#include "ice_fdir.h" #include "ice_flow.h" static struct in6_addr full_ipv6_addr_mask = { @@ -205,7 +206,7 @@ int ice_get_ethtool_fdir_entry(struct ice_hw *hw, struct ethtool_rxnfc *cmd) if (rule->dest_ctl == ICE_FLTR_PRGM_DESC_DEST_DROP_PKT) fsp->ring_cookie = RX_CLS_FLOW_DISC; else - fsp->ring_cookie = rule->q_index; + fsp->ring_cookie = rule->orig_q_index; idx = ice_ethtool_flow_to_fltr(fsp->flow_type); if (idx == ICE_FLTR_PTYPE_NONF_NONE) { @@ -257,6 +258,80 @@ release_lock: } /** + * ice_fdir_remap_entries - update the FDir entries in profile + * @prof: FDir structure pointer + * @tun: tunneled or non-tunneled packet + * @idx: FDir entry index + */ +static void +ice_fdir_remap_entries(struct ice_fd_hw_prof *prof, int tun, int idx) +{ + if (idx != prof->cnt && tun < ICE_FD_HW_SEG_MAX) { + int i; + + for (i = idx; i < (prof->cnt - 1); i++) { + u64 old_entry_h; + + old_entry_h = prof->entry_h[i + 1][tun]; + prof->entry_h[i][tun] = old_entry_h; + prof->vsi_h[i] = prof->vsi_h[i + 1]; + } + + prof->entry_h[i][tun] = 0; + prof->vsi_h[i] = 0; + } +} + +/** + * ice_fdir_rem_adq_chnl - remove an ADQ channel from HW filter rules + * @hw: hardware structure containing filter list + * @vsi_idx: VSI handle + */ +void ice_fdir_rem_adq_chnl(struct ice_hw *hw, u16 vsi_idx) +{ + int status, flow; + + if (!hw->fdir_prof) + return; + + for (flow = 0; flow < ICE_FLTR_PTYPE_MAX; flow++) { + struct ice_fd_hw_prof *prof = hw->fdir_prof[flow]; + int tun, i; + + if (!prof || !prof->cnt) + continue; + + for (tun = 0; tun < ICE_FD_HW_SEG_MAX; tun++) { + u64 prof_id; + + prof_id = flow + tun * ICE_FLTR_PTYPE_MAX; + + for (i = 0; i < prof->cnt; i++) { + if (prof->vsi_h[i] != vsi_idx) + continue; + + prof->entry_h[i][tun] = 0; + prof->vsi_h[i] = 0; + break; + } + + /* after clearing FDir entries update the remaining */ + ice_fdir_remap_entries(prof, tun, i); + + /* find flow profile corresponding to prof_id and clear + * vsi_idx from bitmap. + */ + status = ice_flow_rem_vsi_prof(hw, vsi_idx, prof_id); + if (status) { + dev_err(ice_hw_to_dev(hw), "ice_flow_rem_vsi_prof() failed status=%d\n", + status); + } + } + prof->cnt--; + } +} + +/** * ice_fdir_get_hw_prof - return the ice_fd_hw_proc associated with a flow * @hw: hardware structure containing the filter list * @blk: hardware block @@ -514,6 +589,28 @@ ice_fdir_alloc_flow_prof(struct ice_hw *hw, enum ice_fltr_ptype flow) } /** + * ice_fdir_prof_vsi_idx - find or insert a vsi_idx in structure + * @prof: pointer to flow director HW profile + * @vsi_idx: vsi_idx to locate + * + * return the index of the vsi_idx. if vsi_idx is not found insert it + * into the vsi_h table. + */ +static u16 +ice_fdir_prof_vsi_idx(struct ice_fd_hw_prof *prof, int vsi_idx) +{ + u16 idx = 0; + + for (idx = 0; idx < prof->cnt; idx++) + if (prof->vsi_h[idx] == vsi_idx) + return idx; + + if (idx == prof->cnt) + prof->vsi_h[prof->cnt++] = vsi_idx; + return idx; +} + +/** * ice_fdir_set_hw_fltr_rule - Configure HW tables to generate a FDir rule * @pf: pointer to the PF structure * @seg: protocol header description pointer @@ -532,8 +629,10 @@ ice_fdir_set_hw_fltr_rule(struct ice_pf *pf, struct ice_flow_seg_info *seg, struct ice_hw *hw = &pf->hw; u64 entry1_h = 0; u64 entry2_h = 0; + bool del_last; u64 prof_id; int err; + int idx; main_vsi = ice_get_main_vsi(pf); if (!main_vsi) @@ -603,8 +702,60 @@ ice_fdir_set_hw_fltr_rule(struct ice_pf *pf, struct ice_flow_seg_info *seg, if (!hw_prof->cnt) hw_prof->cnt = 2; + for (idx = 1; idx < ICE_CHNL_MAX_TC; idx++) { + u16 vsi_idx; + u16 vsi_h; + + if (!ice_is_adq_active(pf) || !main_vsi->tc_map_vsi[idx]) + continue; + + entry1_h = 0; + vsi_h = main_vsi->tc_map_vsi[idx]->idx; + err = ice_flow_add_entry(hw, ICE_BLK_FD, prof_id, + main_vsi->idx, vsi_h, + ICE_FLOW_PRIO_NORMAL, seg, + &entry1_h); + if (err) { + dev_err(dev, "Could not add Channel VSI %d to flow group\n", + idx); + goto err_unroll; + } + + vsi_idx = ice_fdir_prof_vsi_idx(hw_prof, + main_vsi->tc_map_vsi[idx]->idx); + hw_prof->entry_h[vsi_idx][tun] = entry1_h; + } + return 0; +err_unroll: + entry1_h = 0; + hw_prof->fdir_seg[tun] = NULL; + + /* The variable del_last will be used to determine when to clean up + * the VSI group data. The VSI data is not needed if there are no + * segments. + */ + del_last = true; + for (idx = 0; idx < ICE_FD_HW_SEG_MAX; idx++) + if (hw_prof->fdir_seg[idx]) { + del_last = false; + break; + } + + for (idx = 0; idx < hw_prof->cnt; idx++) { + u16 vsi_num = ice_get_hw_vsi_num(hw, hw_prof->vsi_h[idx]); + + if (!hw_prof->entry_h[idx][tun]) + continue; + ice_rem_prof_id_flow(hw, ICE_BLK_FD, vsi_num, prof_id); + ice_flow_rem_entry(hw, ICE_BLK_FD, hw_prof->entry_h[idx][tun]); + hw_prof->entry_h[idx][tun] = 0; + if (del_last) + hw_prof->vsi_h[idx] = 0; + } + if (del_last) + hw_prof->cnt = 0; err_entry: ice_rem_prof_id_flow(hw, ICE_BLK_FD, ice_get_hw_vsi_num(hw, main_vsi->idx), prof_id); @@ -1169,6 +1320,31 @@ err_exit: } /** + * ice_update_per_q_fltr + * @vsi: ptr to VSI + * @q_index: queue index + * @inc: true to increment or false to decrement per queue filter count + * + * This function is used to keep track of per queue sideband filters + */ +static void ice_update_per_q_fltr(struct ice_vsi *vsi, u32 q_index, bool inc) +{ + struct ice_rx_ring *rx_ring; + + if (!vsi->num_rxq || q_index >= vsi->num_rxq) + return; + + rx_ring = vsi->rx_rings[q_index]; + if (!rx_ring || !rx_ring->ch) + return; + + if (inc) + atomic_inc(&rx_ring->ch->num_sb_fltr); + else + atomic_dec_if_positive(&rx_ring->ch->num_sb_fltr); +} + +/** * ice_fdir_write_fltr - send a flow director filter to the hardware * @pf: PF data structure * @input: filter structure @@ -1314,13 +1490,32 @@ int ice_fdir_create_dflt_rules(struct ice_pf *pf) } /** + * ice_fdir_del_all_fltrs - Delete all flow director filters + * @vsi: the VSI being changed + * + * This function needs to be called while holding hw->fdir_fltr_lock + */ +void ice_fdir_del_all_fltrs(struct ice_vsi *vsi) +{ + struct ice_fdir_fltr *f_rule, *tmp; + struct ice_pf *pf = vsi->back; + struct ice_hw *hw = &pf->hw; + + list_for_each_entry_safe(f_rule, tmp, &hw->fdir_list_head, fltr_node) { + ice_fdir_write_all_fltr(pf, f_rule, false); + ice_fdir_update_cntrs(hw, f_rule->flow_type, false); + list_del(&f_rule->fltr_node); + devm_kfree(ice_pf_to_dev(pf), f_rule); + } +} + +/** * ice_vsi_manage_fdir - turn on/off flow director * @vsi: the VSI being changed * @ena: boolean value indicating if this is an enable or disable request */ void ice_vsi_manage_fdir(struct ice_vsi *vsi, bool ena) { - struct ice_fdir_fltr *f_rule, *tmp; struct ice_pf *pf = vsi->back; struct ice_hw *hw = &pf->hw; enum ice_fltr_ptype flow; @@ -1334,13 +1529,8 @@ void ice_vsi_manage_fdir(struct ice_vsi *vsi, bool ena) mutex_lock(&hw->fdir_fltr_lock); if (!test_and_clear_bit(ICE_FLAG_FD_ENA, pf->flags)) goto release_lock; - list_for_each_entry_safe(f_rule, tmp, &hw->fdir_list_head, fltr_node) { - /* ignore return value */ - ice_fdir_write_all_fltr(pf, f_rule, false); - ice_fdir_update_cntrs(hw, f_rule->flow_type, false); - list_del(&f_rule->fltr_node); - devm_kfree(ice_hw_to_dev(hw), f_rule); - } + + ice_fdir_del_all_fltrs(vsi); if (hw->fdir_prof) for (flow = ICE_FLTR_PTYPE_NONF_NONE; flow < ICE_FLTR_PTYPE_MAX; @@ -1391,18 +1581,25 @@ ice_fdir_update_list_entry(struct ice_pf *pf, struct ice_fdir_fltr *input, { struct ice_fdir_fltr *old_fltr; struct ice_hw *hw = &pf->hw; + struct ice_vsi *vsi; int err = -ENOENT; /* Do not update filters during reset */ if (ice_is_reset_in_progress(pf->state)) return -EBUSY; + vsi = ice_get_main_vsi(pf); + if (!vsi) + return -EINVAL; + old_fltr = ice_fdir_find_fltr_by_idx(hw, fltr_idx); if (old_fltr) { err = ice_fdir_write_all_fltr(pf, old_fltr, false); if (err) return err; ice_fdir_update_cntrs(hw, old_fltr->flow_type, false); + /* update sb-filters count, specific to ring->channel */ + ice_update_per_q_fltr(vsi, old_fltr->orig_q_index, false); if (!input && !hw->fdir_fltr_cnt[old_fltr->flow_type]) /* we just deleted the last filter of flow_type so we * should also delete the HW filter info. @@ -1414,6 +1611,8 @@ ice_fdir_update_list_entry(struct ice_pf *pf, struct ice_fdir_fltr *input, if (!input) return err; ice_fdir_list_add_fltr(hw, input); + /* update sb-filters count, specific to ring->channel */ + ice_update_per_q_fltr(vsi, input->orig_q_index, true); ice_fdir_update_cntrs(hw, input->flow_type, true); return 0; } @@ -1453,6 +1652,39 @@ int ice_del_fdir_ethtool(struct ice_vsi *vsi, struct ethtool_rxnfc *cmd) } /** + * ice_update_ring_dest_vsi - update dest ring and dest VSI + * @vsi: pointer to target VSI + * @dest_vsi: ptr to dest VSI index + * @ring: ptr to dest ring + * + * This function updates destination VSI and queue if user specifies + * target queue which falls in channel's (aka ADQ) queue region + */ +static void +ice_update_ring_dest_vsi(struct ice_vsi *vsi, u16 *dest_vsi, u32 *ring) +{ + struct ice_channel *ch; + + list_for_each_entry(ch, &vsi->ch_list, list) { + if (!ch->ch_vsi) + continue; + + /* make sure to locate corresponding channel based on "queue" + * specified + */ + if ((*ring < ch->base_q) || + (*ring >= (ch->base_q + ch->num_rxq))) + continue; + + /* update the dest_vsi based on channel */ + *dest_vsi = ch->ch_vsi->idx; + + /* update the "ring" to be correct based on channel */ + *ring -= ch->base_q; + } +} + +/** * ice_set_fdir_input_set - Set the input set for Flow Director * @vsi: pointer to target VSI * @fsp: pointer to ethtool Rx flow specification @@ -1463,6 +1695,7 @@ ice_set_fdir_input_set(struct ice_vsi *vsi, struct ethtool_rx_flow_spec *fsp, struct ice_fdir_fltr *input) { u16 dest_vsi, q_index = 0; + u16 orig_q_index = 0; struct ice_pf *pf; struct ice_hw *hw; int flow_type; @@ -1489,6 +1722,8 @@ ice_set_fdir_input_set(struct ice_vsi *vsi, struct ethtool_rx_flow_spec *fsp, if (ring >= vsi->num_rxq) return -EINVAL; + orig_q_index = ring; + ice_update_ring_dest_vsi(vsi, &dest_vsi, &ring); dest_ctl = ICE_FLTR_PRGM_DESC_DEST_DIRECT_PKT_QINDEX; q_index = ring; } @@ -1497,6 +1732,11 @@ ice_set_fdir_input_set(struct ice_vsi *vsi, struct ethtool_rx_flow_spec *fsp, input->q_index = q_index; flow_type = fsp->flow_type & ~FLOW_EXT; + /* Record the original queue index as specified by user. + * with channel configuration 'q_index' becomes relative + * to TC (channel). + */ + input->orig_q_index = orig_q_index; input->dest_vsi = dest_vsi; input->dest_ctl = dest_ctl; input->fltr_status = ICE_FLTR_PRGM_DESC_FD_STATUS_FD_ID; @@ -1684,6 +1924,8 @@ int ice_add_fdir_ethtool(struct ice_vsi *vsi, struct ethtool_rxnfc *cmd) remove_sw_rule: ice_fdir_update_cntrs(hw, input->flow_type, false); + /* update sb-filters count, specific to ring->channel */ + ice_update_per_q_fltr(vsi, input->orig_q_index, false); list_del(&input->fltr_node); release_lock: mutex_unlock(&hw->fdir_fltr_lock); |
