net/ice: support ACL filter in DCF
authorSimei Su <simei.su@intel.com>
Tue, 20 Oct 2020 11:32:42 +0000 (19:32 +0800)
committerFerruh Yigit <ferruh.yigit@intel.com>
Tue, 3 Nov 2020 22:24:24 +0000 (23:24 +0100)
Add ice_acl_create_filter to create a rule and ice_acl_destroy_filter
to destroy a rule. If a flow is matched by ACL filter, filter rule
will be set to HW. Currently IPV4/IPV4_UDP/IPV4_TCP/IPV4_SCTP pattern
and drop action are supported.

Signed-off-by: Simei Su <simei.su@intel.com>
Signed-off-by: Xuan Ding <xuan.ding@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
doc/guides/rel_notes/release_20_11.rst
drivers/net/ice/base/ice_flow.c
drivers/net/ice/base/ice_flow.h
drivers/net/ice/ice_acl_filter.c [new file with mode: 0644]
drivers/net/ice/ice_ethdev.h
drivers/net/ice/ice_generic_flow.c
drivers/net/ice/meson.build

index 062b90a..462a8ca 100644 (file)
@@ -169,6 +169,11 @@ New Features
   Added the FEC PMD which provides functions for query FEC capabilities and
   current FEC mode from device. Also, PMD for configuring FEC mode is also provided.
 
+* **Updated Intel ice driver.**
+
+  * Used write combining stores.
+  * Added ACL filter support for Intel DCF.
+
 * **Updated Mellanox mlx5 driver.**
 
   Updated Mellanox mlx5 driver with new features and improvements, including:
@@ -199,10 +204,6 @@ New Features
 
   Updated the Intel ixgbe driver to use write combining stores.
 
-* **Updated Intel ice driver.**
-
-  Updated the Intel ice driver to use write combining stores.
-
 * **Updated Intel qat driver.**
 
   Updated the Intel qat driver to use write combining stores.
index de5dfb2..80ac0b6 100644 (file)
@@ -2125,7 +2125,7 @@ ice_flow_assoc_vsig_vsi(struct ice_hw *hw, enum ice_block blk, u16 vsi_handle,
  * Assumption: the caller has acquired the lock to the profile list
  * and the software VSI handle has been validated
  */
-static enum ice_status
+enum ice_status
 ice_flow_assoc_prof(struct ice_hw *hw, enum ice_block blk,
                    struct ice_flow_prof *prof, u16 vsi_handle)
 {
index 0a52409..698a230 100644 (file)
@@ -499,6 +499,9 @@ ice_flow_add_prof(struct ice_hw *hw, enum ice_block blk, enum ice_flow_dir dir,
 enum ice_status
 ice_flow_rem_prof(struct ice_hw *hw, enum ice_block blk, u64 prof_id);
 enum ice_status
+ice_flow_assoc_prof(struct ice_hw *hw, enum ice_block blk,
+                   struct ice_flow_prof *prof, u16 vsi_handle);
+enum ice_status
 ice_flow_assoc_vsig_vsi(struct ice_hw *hw, enum ice_block blk, u16 vsi_handle,
                        u16 vsig);
 enum ice_status
diff --git a/drivers/net/ice/ice_acl_filter.c b/drivers/net/ice/ice_acl_filter.c
new file mode 100644 (file)
index 0000000..ca483f0
--- /dev/null
@@ -0,0 +1,1011 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2020 Intel Corporation
+ */
+
+#include <sys/queue.h>
+#include <stdio.h>
+#include <errno.h>
+#include <stdint.h>
+#include <string.h>
+#include <unistd.h>
+#include <stdarg.h>
+#include <rte_debug.h>
+#include <rte_ether.h>
+#include <rte_ethdev_driver.h>
+#include <rte_log.h>
+#include <rte_malloc.h>
+#include <rte_eth_ctrl.h>
+#include <rte_tailq.h>
+#include <rte_flow_driver.h>
+#include <rte_flow.h>
+#include <rte_bitmap.h>
+#include "base/ice_type.h"
+#include "base/ice_acl.h"
+#include "ice_logs.h"
+#include "ice_ethdev.h"
+#include "ice_generic_flow.h"
+#include "base/ice_flow.h"
+
+#define MAX_ACL_SLOTS_ID 2048
+
+#define ICE_ACL_INSET_ETH_IPV4 ( \
+       ICE_INSET_SMAC | ICE_INSET_DMAC | \
+       ICE_INSET_IPV4_SRC | ICE_INSET_IPV4_DST)
+#define ICE_ACL_INSET_ETH_IPV4_UDP ( \
+       ICE_ACL_INSET_ETH_IPV4 | \
+       ICE_INSET_UDP_SRC_PORT | ICE_INSET_UDP_DST_PORT)
+#define ICE_ACL_INSET_ETH_IPV4_TCP ( \
+       ICE_ACL_INSET_ETH_IPV4 | \
+       ICE_INSET_TCP_SRC_PORT | ICE_INSET_TCP_DST_PORT)
+#define ICE_ACL_INSET_ETH_IPV4_SCTP ( \
+       ICE_ACL_INSET_ETH_IPV4 | \
+       ICE_INSET_SCTP_SRC_PORT | ICE_INSET_SCTP_DST_PORT)
+
+static struct ice_flow_parser ice_acl_parser;
+
+struct acl_rule {
+       enum ice_fltr_ptype flow_type;
+       uint32_t entry_id[4];
+};
+
+static struct
+ice_pattern_match_item ice_acl_pattern[] = {
+       {pattern_eth_ipv4,      ICE_ACL_INSET_ETH_IPV4,      ICE_INSET_NONE},
+       {pattern_eth_ipv4_udp,  ICE_ACL_INSET_ETH_IPV4_UDP,  ICE_INSET_NONE},
+       {pattern_eth_ipv4_tcp,  ICE_ACL_INSET_ETH_IPV4_TCP,  ICE_INSET_NONE},
+       {pattern_eth_ipv4_sctp, ICE_ACL_INSET_ETH_IPV4_SCTP, ICE_INSET_NONE},
+};
+
+static int
+ice_acl_prof_alloc(struct ice_hw *hw)
+{
+       enum ice_fltr_ptype ptype, fltr_ptype;
+
+       if (!hw->acl_prof) {
+               hw->acl_prof = (struct ice_fd_hw_prof **)
+                       ice_malloc(hw, ICE_FLTR_PTYPE_MAX *
+                                  sizeof(*hw->acl_prof));
+               if (!hw->acl_prof)
+                       return -ENOMEM;
+       }
+
+       for (ptype = ICE_FLTR_PTYPE_NONF_NONE + 1;
+            ptype < ICE_FLTR_PTYPE_MAX; ptype++) {
+               if (!hw->acl_prof[ptype]) {
+                       hw->acl_prof[ptype] = (struct ice_fd_hw_prof *)
+                               ice_malloc(hw, sizeof(**hw->acl_prof));
+                       if (!hw->acl_prof[ptype])
+                               goto fail_mem;
+               }
+       }
+
+       return 0;
+
+fail_mem:
+       for (fltr_ptype = ICE_FLTR_PTYPE_NONF_NONE + 1;
+            fltr_ptype < ptype; fltr_ptype++) {
+               rte_free(hw->acl_prof[fltr_ptype]);
+               hw->acl_prof[fltr_ptype] = NULL;
+       }
+
+       rte_free(hw->acl_prof);
+       hw->acl_prof = NULL;
+
+       return -ENOMEM;
+}
+
+/**
+ * ice_acl_setup - Reserve and initialize the ACL resources
+ * @pf: board private structure
+ */
+static int
+ice_acl_setup(struct ice_pf *pf)
+{
+       struct ice_hw *hw = ICE_PF_TO_HW(pf);
+       uint32_t pf_num = hw->dev_caps.num_funcs;
+       struct ice_acl_tbl_params params;
+       uint16_t scen_id;
+       int err = 0;
+
+       memset(&params, 0, sizeof(params));
+
+       /* create for IPV4 table */
+       if (pf_num < 4)
+               params.width = ICE_AQC_ACL_KEY_WIDTH_BYTES * 6;
+       else
+               params.width = ICE_AQC_ACL_KEY_WIDTH_BYTES * 3;
+
+       params.depth = ICE_AQC_ACL_TCAM_DEPTH;
+       params.entry_act_pairs = 1;
+       params.concurr = false;
+
+       err = ice_acl_create_tbl(hw, &params);
+       if (err)
+               return err;
+
+       err = ice_acl_create_scen(hw, params.width, params.depth,
+                                 &scen_id);
+       if (err)
+               return err;
+
+       return 0;
+}
+
+/**
+ * ice_deinit_acl - Unroll the initialization of the ACL block
+ * @pf: ptr to PF device
+ *
+ * returns 0 on success, negative on error
+ */
+static void ice_deinit_acl(struct ice_pf *pf)
+{
+       struct ice_hw *hw = ICE_PF_TO_HW(pf);
+
+       ice_acl_destroy_tbl(hw);
+
+       rte_free(hw->acl_tbl);
+       hw->acl_tbl = NULL;
+
+       if (pf->acl.slots) {
+               rte_free(pf->acl.slots);
+               pf->acl.slots = NULL;
+       }
+}
+
+static void
+acl_add_prof_prepare(struct ice_hw *hw, struct ice_flow_seg_info *seg,
+                    bool is_l4, uint16_t src_port, uint16_t dst_port)
+{
+       uint16_t val_loc, mask_loc;
+
+       if (hw->dev_caps.num_funcs < 4) {
+               /* mac source address */
+               val_loc = offsetof(struct ice_fdir_fltr,
+                                  ext_data.src_mac);
+               mask_loc = offsetof(struct ice_fdir_fltr,
+                                   ext_mask.src_mac);
+               ice_flow_set_fld(seg, ICE_FLOW_FIELD_IDX_ETH_SA,
+                                val_loc, mask_loc,
+                                ICE_FLOW_FLD_OFF_INVAL, false);
+
+               /* mac destination address */
+               val_loc = offsetof(struct ice_fdir_fltr,
+                                  ext_data.dst_mac);
+               mask_loc = offsetof(struct ice_fdir_fltr,
+                                   ext_mask.dst_mac);
+               ice_flow_set_fld(seg, ICE_FLOW_FIELD_IDX_ETH_DA,
+                                val_loc, mask_loc,
+                                ICE_FLOW_FLD_OFF_INVAL, false);
+       }
+
+       /* IP source address */
+       val_loc = offsetof(struct ice_fdir_fltr, ip.v4.src_ip);
+       mask_loc = offsetof(struct ice_fdir_fltr, mask.v4.src_ip);
+       ice_flow_set_fld(seg, ICE_FLOW_FIELD_IDX_IPV4_SA, val_loc,
+                        mask_loc, ICE_FLOW_FLD_OFF_INVAL, false);
+
+       /* IP destination address */
+       val_loc = offsetof(struct ice_fdir_fltr, ip.v4.dst_ip);
+       mask_loc = offsetof(struct ice_fdir_fltr, mask.v4.dst_ip);
+       ice_flow_set_fld(seg, ICE_FLOW_FIELD_IDX_IPV4_DA, val_loc,
+                        mask_loc, ICE_FLOW_FLD_OFF_INVAL, false);
+
+       if (is_l4) {
+               /* Layer 4 source port */
+               val_loc = offsetof(struct ice_fdir_fltr, ip.v4.src_port);
+               mask_loc = offsetof(struct ice_fdir_fltr, mask.v4.src_port);
+               ice_flow_set_fld(seg, src_port, val_loc,
+                                mask_loc, ICE_FLOW_FLD_OFF_INVAL, false);
+
+               /* Layer 4 destination port */
+               val_loc = offsetof(struct ice_fdir_fltr, ip.v4.dst_port);
+               mask_loc = offsetof(struct ice_fdir_fltr, mask.v4.dst_port);
+               ice_flow_set_fld(seg, dst_port, val_loc,
+                                mask_loc, ICE_FLOW_FLD_OFF_INVAL, false);
+       }
+}
+
+/**
+ * ice_acl_prof_init - Initialize ACL profile
+ * @pf: ice PF structure
+ *
+ * Returns 0 on success.
+ */
+static int
+ice_acl_prof_init(struct ice_pf *pf)
+{
+       struct ice_hw *hw = ICE_PF_TO_HW(pf);
+       struct ice_flow_prof *prof_ipv4 = NULL;
+       struct ice_flow_prof *prof_ipv4_udp = NULL;
+       struct ice_flow_prof *prof_ipv4_tcp = NULL;
+       struct ice_flow_prof *prof_ipv4_sctp = NULL;
+       struct ice_flow_seg_info *seg;
+       int i;
+       int ret;
+
+       seg = (struct ice_flow_seg_info *)
+                ice_malloc(hw, sizeof(*seg));
+       if (!seg)
+               return -ENOMEM;
+
+       ICE_FLOW_SET_HDRS(seg, ICE_FLOW_SEG_HDR_IPV4);
+       acl_add_prof_prepare(hw, seg, false, 0, 0);
+       ret = ice_flow_add_prof(hw, ICE_BLK_ACL, ICE_FLOW_RX,
+                               ICE_FLTR_PTYPE_NONF_IPV4_OTHER,
+                               seg, 1, NULL, 0, &prof_ipv4);
+       if (ret)
+               goto err_add_prof;
+
+       ice_memset(seg, 0, sizeof(*seg), ICE_NONDMA_MEM);
+       ICE_FLOW_SET_HDRS(seg, ICE_FLOW_SEG_HDR_UDP | ICE_FLOW_SEG_HDR_IPV4);
+       acl_add_prof_prepare(hw, seg, true,
+                            ICE_FLOW_FIELD_IDX_UDP_SRC_PORT,
+                            ICE_FLOW_FIELD_IDX_UDP_DST_PORT);
+       ret = ice_flow_add_prof(hw, ICE_BLK_ACL, ICE_FLOW_RX,
+                               ICE_FLTR_PTYPE_NONF_IPV4_UDP,
+                               seg, 1, NULL, 0, &prof_ipv4_udp);
+       if (ret)
+               goto err_add_prof_ipv4_udp;
+
+       ice_memset(seg, 0, sizeof(*seg), ICE_NONDMA_MEM);
+       ICE_FLOW_SET_HDRS(seg, ICE_FLOW_SEG_HDR_TCP | ICE_FLOW_SEG_HDR_IPV4);
+       acl_add_prof_prepare(hw, seg, true,
+                            ICE_FLOW_FIELD_IDX_TCP_SRC_PORT,
+                            ICE_FLOW_FIELD_IDX_TCP_DST_PORT);
+       ret = ice_flow_add_prof(hw, ICE_BLK_ACL, ICE_FLOW_RX,
+                               ICE_FLTR_PTYPE_NONF_IPV4_TCP,
+                               seg, 1, NULL, 0, &prof_ipv4_tcp);
+       if (ret)
+               goto err_add_prof_ipv4_tcp;
+
+       ice_memset(seg, 0, sizeof(*seg), ICE_NONDMA_MEM);
+       ICE_FLOW_SET_HDRS(seg, ICE_FLOW_SEG_HDR_SCTP | ICE_FLOW_SEG_HDR_IPV4);
+       acl_add_prof_prepare(hw, seg, true,
+                            ICE_FLOW_FIELD_IDX_SCTP_SRC_PORT,
+                            ICE_FLOW_FIELD_IDX_SCTP_DST_PORT);
+       ret = ice_flow_add_prof(hw, ICE_BLK_ACL, ICE_FLOW_RX,
+                               ICE_FLTR_PTYPE_NONF_IPV4_SCTP,
+                               seg, 1, NULL, 0, &prof_ipv4_sctp);
+       if (ret)
+               goto err_add_prof_ipv4_sctp;
+
+       for (i = 0; i < pf->main_vsi->idx; i++) {
+               ret = ice_flow_assoc_prof(hw, ICE_BLK_ACL, prof_ipv4, i);
+               if (ret)
+                       goto err_assoc_prof;
+
+               ret = ice_flow_assoc_prof(hw, ICE_BLK_ACL, prof_ipv4_udp, i);
+               if (ret)
+                       goto err_assoc_prof;
+
+               ret = ice_flow_assoc_prof(hw, ICE_BLK_ACL, prof_ipv4_tcp, i);
+               if (ret)
+                       goto err_assoc_prof;
+
+               ret = ice_flow_assoc_prof(hw, ICE_BLK_ACL, prof_ipv4_sctp, i);
+               if (ret)
+                       goto err_assoc_prof;
+       }
+       return 0;
+
+err_assoc_prof:
+       ice_flow_rem_prof(hw, ICE_BLK_ACL, ICE_FLTR_PTYPE_NONF_IPV4_SCTP);
+err_add_prof_ipv4_sctp:
+       ice_flow_rem_prof(hw, ICE_BLK_ACL, ICE_FLTR_PTYPE_NONF_IPV4_TCP);
+err_add_prof_ipv4_tcp:
+       ice_flow_rem_prof(hw, ICE_BLK_ACL, ICE_FLTR_PTYPE_NONF_IPV4_UDP);
+err_add_prof_ipv4_udp:
+       ice_flow_rem_prof(hw, ICE_BLK_ACL, ICE_FLTR_PTYPE_NONF_IPV4_OTHER);
+err_add_prof:
+       ice_free(hw, seg);
+       return ret;
+}
+
+/**
+ * ice_acl_set_input_set - Helper function to set the input set for ACL
+ * @hw: pointer to HW instance
+ * @filter: pointer to ACL info
+ * @input: filter structure
+ *
+ * Return error value or 0 on success.
+ */
+static int
+ice_acl_set_input_set(struct ice_acl_conf *filter, struct ice_fdir_fltr *input)
+{
+       if (!input)
+               return ICE_ERR_BAD_PTR;
+
+       input->q_index = filter->input.q_index;
+       input->dest_vsi = filter->input.dest_vsi;
+       input->dest_ctl = filter->input.dest_ctl;
+       input->fltr_status = ICE_FLTR_PRGM_DESC_FD_STATUS_FD_ID;
+       input->flow_type = filter->input.flow_type;
+
+       switch (input->flow_type) {
+       case ICE_FLTR_PTYPE_NONF_IPV4_TCP:
+       case ICE_FLTR_PTYPE_NONF_IPV4_UDP:
+       case ICE_FLTR_PTYPE_NONF_IPV4_SCTP:
+               input->ip.v4.dst_port = filter->input.ip.v4.dst_port;
+               input->ip.v4.src_port = filter->input.ip.v4.src_port;
+               input->ip.v4.dst_ip = filter->input.ip.v4.dst_ip;
+               input->ip.v4.src_ip = filter->input.ip.v4.src_ip;
+
+               input->mask.v4.dst_port = filter->input.mask.v4.dst_port;
+               input->mask.v4.src_port = filter->input.mask.v4.src_port;
+               input->mask.v4.dst_ip = filter->input.mask.v4.dst_ip;
+               input->mask.v4.src_ip = filter->input.mask.v4.src_ip;
+
+               ice_memcpy(&input->ext_data.src_mac,
+                          &filter->input.ext_data.src_mac,
+                          RTE_ETHER_ADDR_LEN,
+                          ICE_NONDMA_TO_NONDMA);
+
+               ice_memcpy(&input->ext_mask.src_mac,
+                          &filter->input.ext_mask.src_mac,
+                          RTE_ETHER_ADDR_LEN,
+                          ICE_NONDMA_TO_NONDMA);
+
+               ice_memcpy(&input->ext_data.dst_mac,
+                          &filter->input.ext_data.dst_mac,
+                          RTE_ETHER_ADDR_LEN,
+                          ICE_NONDMA_TO_NONDMA);
+               ice_memcpy(&input->ext_mask.dst_mac,
+                          &filter->input.ext_mask.dst_mac,
+                          RTE_ETHER_ADDR_LEN,
+                          ICE_NONDMA_TO_NONDMA);
+
+               break;
+       case ICE_FLTR_PTYPE_NONF_IPV4_OTHER:
+               ice_memcpy(&input->ip.v4, &filter->input.ip.v4,
+                          sizeof(struct ice_fdir_v4),
+                          ICE_NONDMA_TO_NONDMA);
+               ice_memcpy(&input->mask.v4, &filter->input.mask.v4,
+                          sizeof(struct ice_fdir_v4),
+                          ICE_NONDMA_TO_NONDMA);
+
+               ice_memcpy(&input->ext_data.src_mac,
+                          &filter->input.ext_data.src_mac,
+                          RTE_ETHER_ADDR_LEN,
+                          ICE_NONDMA_TO_NONDMA);
+               ice_memcpy(&input->ext_mask.src_mac,
+                          &filter->input.ext_mask.src_mac,
+                          RTE_ETHER_ADDR_LEN,
+                          ICE_NONDMA_TO_NONDMA);
+
+               ice_memcpy(&input->ext_data.dst_mac,
+                          &filter->input.ext_data.dst_mac,
+                          RTE_ETHER_ADDR_LEN,
+                          ICE_NONDMA_TO_NONDMA);
+               ice_memcpy(&input->ext_mask.dst_mac,
+                          &filter->input.ext_mask.dst_mac,
+                          RTE_ETHER_ADDR_LEN,
+                          ICE_NONDMA_TO_NONDMA);
+
+               break;
+       default:
+               return -EINVAL;
+       }
+
+       return 0;
+}
+
+static inline int
+ice_acl_alloc_slot_id(struct rte_bitmap *slots, uint32_t *slot_id)
+{
+       uint32_t pos = 0;
+       uint64_t slab = 0;
+       uint32_t i = 0;
+
+       __rte_bitmap_scan_init(slots);
+       if (!rte_bitmap_scan(slots, &pos, &slab))
+               return -rte_errno;
+
+       i = rte_bsf64(slab);
+       pos += i;
+       rte_bitmap_clear(slots, pos);
+
+       *slot_id = pos;
+       return 0;
+}
+
+static inline int
+ice_acl_hw_set_conf(struct ice_pf *pf, struct ice_fdir_fltr *input,
+                   struct ice_flow_action *acts, struct acl_rule *rule,
+                   enum ice_fltr_ptype flow_type, int32_t entry_idx)
+{
+       struct ice_hw *hw = ICE_PF_TO_HW(pf);
+       enum ice_block blk = ICE_BLK_ACL;
+       uint64_t entry_id, hw_entry;
+       uint32_t slot_id = 0;
+       int act_cnt = 1;
+       int ret = 0;
+
+       /* Allocate slot_id from bitmap table. */
+       ret = ice_acl_alloc_slot_id(pf->acl.slots, &slot_id);
+       if (ret) {
+               PMD_DRV_LOG(ERR, "fail to alloc slot id.");
+               return ret;
+       }
+
+       /* For IPV4_OTHER type, should add entry for all types.
+        * For IPV4_UDP/TCP/SCTP type, only add entry for each.
+        */
+       if (slot_id < MAX_ACL_ENTRIES) {
+               entry_id = ((uint64_t)flow_type << 32) | slot_id;
+               ret = ice_flow_add_entry(hw, blk, flow_type,
+                                        entry_id, pf->main_vsi->idx,
+                                        ICE_FLOW_PRIO_NORMAL, input,
+                                        acts, act_cnt, &hw_entry);
+               if (ret) {
+                       PMD_DRV_LOG(ERR, "Fail to add entry.");
+                       return ret;
+               }
+               rule->entry_id[entry_idx] = slot_id;
+               pf->acl.hw_entry_id[slot_id] = hw_entry;
+       } else {
+               PMD_DRV_LOG(ERR, "Exceed the maximum entry number(%d)"
+                           " HW supported!", MAX_ACL_ENTRIES);
+               return -1;
+       }
+
+       return 0;
+}
+
+static inline void
+ice_acl_hw_rem_conf(struct ice_pf *pf, struct acl_rule *rule, int32_t entry_idx)
+{
+       uint32_t slot_id;
+       int32_t i;
+       struct ice_hw *hw = ICE_PF_TO_HW(pf);
+
+       for (i = 0; i < entry_idx; i++) {
+               slot_id = rule->entry_id[i];
+               rte_bitmap_set(pf->acl.slots, slot_id);
+               ice_flow_rem_entry(hw, ICE_BLK_ACL,
+                                  pf->acl.hw_entry_id[slot_id]);
+       }
+}
+
+static int
+ice_acl_create_filter(struct ice_adapter *ad,
+                     struct rte_flow *flow,
+                     void *meta,
+                     struct rte_flow_error *error)
+{
+       struct ice_acl_conf *filter = meta;
+       enum ice_fltr_ptype flow_type = filter->input.flow_type;
+       struct ice_flow_action acts[1];
+       struct ice_pf *pf = &ad->pf;
+       struct ice_fdir_fltr *input;
+       struct acl_rule *rule;
+       int ret;
+
+       rule = rte_zmalloc("acl_rule", sizeof(*rule), 0);
+       if (!rule) {
+               rte_flow_error_set(error, ENOMEM,
+                                  RTE_FLOW_ERROR_TYPE_HANDLE, NULL,
+                                  "Failed to allocate memory for acl rule");
+               return -rte_errno;
+       }
+
+       input = rte_zmalloc("acl_entry", sizeof(*input), 0);
+       if (!input) {
+               rte_flow_error_set(error, ENOMEM,
+                                  RTE_FLOW_ERROR_TYPE_HANDLE, NULL,
+                                  "Failed to allocate memory for acl input");
+               ret = -rte_errno;
+               goto err_acl_input_alloc;
+       }
+
+       ret = ice_acl_set_input_set(filter, input);
+       if (ret) {
+               rte_flow_error_set(error, -ret,
+                                  RTE_FLOW_ERROR_TYPE_HANDLE, NULL,
+                                  "failed to set input set.");
+               ret = -rte_errno;
+               goto err_acl_set_input;
+       }
+
+       if (filter->input.dest_ctl == ICE_FLTR_PRGM_DESC_DEST_DROP_PKT) {
+               acts[0].type = ICE_FLOW_ACT_DROP;
+               acts[0].data.acl_act.mdid = ICE_MDID_RX_PKT_DROP;
+               acts[0].data.acl_act.prio = 0x3;
+               acts[0].data.acl_act.value = CPU_TO_LE16(0x1);
+       }
+
+       input->acl_fltr = true;
+       ret = ice_acl_hw_set_conf(pf, input, acts, rule, flow_type, 0);
+       if (ret) {
+               rte_flow_error_set(error, -ret,
+                                  RTE_FLOW_ERROR_TYPE_HANDLE, NULL,
+                                  "failed to set hw configure.");
+               ret = -rte_errno;
+               return ret;
+       }
+
+       if (flow_type == ICE_FLTR_PTYPE_NONF_IPV4_OTHER) {
+               ret = ice_acl_hw_set_conf(pf, input, acts, rule,
+                                         ICE_FLTR_PTYPE_NONF_IPV4_UDP, 1);
+               if (ret)
+                       goto err_acl_hw_set_conf_udp;
+               ret = ice_acl_hw_set_conf(pf, input, acts, rule,
+                                         ICE_FLTR_PTYPE_NONF_IPV4_TCP, 2);
+               if (ret)
+                       goto err_acl_hw_set_conf_tcp;
+               ret = ice_acl_hw_set_conf(pf, input, acts, rule,
+                                         ICE_FLTR_PTYPE_NONF_IPV4_SCTP, 3);
+               if (ret)
+                       goto err_acl_hw_set_conf_sctp;
+       }
+
+       rule->flow_type = flow_type;
+       flow->rule = rule;
+       return 0;
+
+err_acl_hw_set_conf_sctp:
+       ice_acl_hw_rem_conf(pf, rule, 3);
+err_acl_hw_set_conf_tcp:
+       ice_acl_hw_rem_conf(pf, rule, 2);
+err_acl_hw_set_conf_udp:
+       ice_acl_hw_rem_conf(pf, rule, 1);
+err_acl_set_input:
+       rte_free(input);
+err_acl_input_alloc:
+       rte_free(rule);
+       return ret;
+}
+
+static int
+ice_acl_destroy_filter(struct ice_adapter *ad,
+                      struct rte_flow *flow,
+                      struct rte_flow_error *error __rte_unused)
+{
+       struct acl_rule *rule = (struct acl_rule *)flow->rule;
+       uint32_t slot_id, i;
+       struct ice_pf *pf = &ad->pf;
+       struct ice_hw *hw = ICE_PF_TO_HW(pf);
+       int ret = 0;
+
+       switch (rule->flow_type) {
+       case ICE_FLTR_PTYPE_NONF_IPV4_OTHER:
+               for (i = 0; i < 4; i++) {
+                       slot_id = rule->entry_id[i];
+                       rte_bitmap_set(pf->acl.slots, slot_id);
+                       ice_flow_rem_entry(hw, ICE_BLK_ACL,
+                                          pf->acl.hw_entry_id[slot_id]);
+               }
+               break;
+       case ICE_FLTR_PTYPE_NONF_IPV4_UDP:
+       case ICE_FLTR_PTYPE_NONF_IPV4_TCP:
+       case ICE_FLTR_PTYPE_NONF_IPV4_SCTP:
+               slot_id = rule->entry_id[0];
+               rte_bitmap_set(pf->acl.slots, slot_id);
+               ice_flow_rem_entry(hw, ICE_BLK_ACL,
+                                  pf->acl.hw_entry_id[slot_id]);
+               break;
+       default:
+               rte_flow_error_set(error, EINVAL,
+                                  RTE_FLOW_ERROR_TYPE_ITEM,
+                                  NULL, "Unsupported flow type.");
+               break;
+       }
+
+       flow->rule = NULL;
+       rte_free(rule);
+       return ret;
+}
+
+static void
+ice_acl_filter_free(struct rte_flow *flow)
+{
+       rte_free(flow->rule);
+       flow->rule = NULL;
+}
+
+static int
+ice_acl_parse_action(__rte_unused struct ice_adapter *ad,
+                    const struct rte_flow_action actions[],
+                    struct rte_flow_error *error,
+                    struct ice_acl_conf *filter)
+{
+       uint32_t dest_num = 0;
+
+       for (; actions->type != RTE_FLOW_ACTION_TYPE_END; actions++) {
+               switch (actions->type) {
+               case RTE_FLOW_ACTION_TYPE_VOID:
+                       break;
+               case RTE_FLOW_ACTION_TYPE_DROP:
+                       dest_num++;
+
+                       filter->input.dest_ctl =
+                               ICE_FLTR_PRGM_DESC_DEST_DROP_PKT;
+                       break;
+               default:
+                       rte_flow_error_set(error, EINVAL,
+                                  RTE_FLOW_ERROR_TYPE_ACTION, actions,
+                                  "Invalid action.");
+                       return -rte_errno;
+               }
+       }
+
+       if (dest_num == 0 || dest_num >= 2) {
+               rte_flow_error_set(error, EINVAL,
+                          RTE_FLOW_ERROR_TYPE_ACTION, actions,
+                          "Unsupported action combination");
+               return -rte_errno;
+       }
+
+       return 0;
+}
+
+static int
+ice_acl_parse_pattern(__rte_unused struct ice_adapter *ad,
+                      const struct rte_flow_item pattern[],
+                      struct rte_flow_error *error,
+                      struct ice_acl_conf *filter)
+{
+       const struct rte_flow_item *item = pattern;
+       enum rte_flow_item_type item_type;
+       enum rte_flow_item_type l3 = RTE_FLOW_ITEM_TYPE_END;
+       const struct rte_flow_item_eth *eth_spec, *eth_mask;
+       const struct rte_flow_item_ipv4 *ipv4_spec, *ipv4_mask;
+       const struct rte_flow_item_tcp *tcp_spec, *tcp_mask;
+       const struct rte_flow_item_udp *udp_spec, *udp_mask;
+       const struct rte_flow_item_sctp *sctp_spec, *sctp_mask;
+       uint64_t input_set = ICE_INSET_NONE;
+       uint8_t flow_type = ICE_FLTR_PTYPE_NONF_NONE;
+
+       for (item = pattern; item->type != RTE_FLOW_ITEM_TYPE_END; item++) {
+               item_type = item->type;
+
+               switch (item_type) {
+               case RTE_FLOW_ITEM_TYPE_ETH:
+                       eth_spec = item->spec;
+                       eth_mask = item->mask;
+
+                       if (eth_spec && eth_mask) {
+                               if (!rte_is_zero_ether_addr(&eth_spec->src) &&
+                                   !rte_is_zero_ether_addr(&eth_mask->src)) {
+                                       input_set |= ICE_INSET_SMAC;
+                                       ice_memcpy(&filter->input.ext_data.src_mac,
+                                                  &eth_spec->src,
+                                                  RTE_ETHER_ADDR_LEN,
+                                                  ICE_NONDMA_TO_NONDMA);
+                                       ice_memcpy(&filter->input.ext_mask.src_mac,
+                                                  &eth_mask->src,
+                                                  RTE_ETHER_ADDR_LEN,
+                                                  ICE_NONDMA_TO_NONDMA);
+                               }
+
+                               if (!rte_is_zero_ether_addr(&eth_spec->dst) &&
+                                   !rte_is_zero_ether_addr(&eth_mask->dst)) {
+                                       input_set |= ICE_INSET_DMAC;
+                                       ice_memcpy(&filter->input.ext_data.dst_mac,
+                                                  &eth_spec->dst,
+                                                  RTE_ETHER_ADDR_LEN,
+                                                  ICE_NONDMA_TO_NONDMA);
+                                       ice_memcpy(&filter->input.ext_mask.dst_mac,
+                                                  &eth_mask->dst,
+                                                  RTE_ETHER_ADDR_LEN,
+                                                  ICE_NONDMA_TO_NONDMA);
+                               }
+                       }
+                       break;
+               case RTE_FLOW_ITEM_TYPE_IPV4:
+                       l3 = RTE_FLOW_ITEM_TYPE_IPV4;
+                       ipv4_spec = item->spec;
+                       ipv4_mask = item->mask;
+
+                       if (ipv4_spec && ipv4_mask) {
+                               /* Check IPv4 mask and update input set */
+                               if (ipv4_mask->hdr.version_ihl ||
+                                   ipv4_mask->hdr.total_length ||
+                                   ipv4_mask->hdr.packet_id ||
+                                   ipv4_mask->hdr.fragment_offset ||
+                                   ipv4_mask->hdr.hdr_checksum) {
+                                       rte_flow_error_set(error, EINVAL,
+                                               RTE_FLOW_ERROR_TYPE_ITEM,
+                                               item,
+                                               "Invalid IPv4 mask.");
+                                       return -rte_errno;
+                               }
+
+                               if (ipv4_mask->hdr.src_addr) {
+                                       filter->input.ip.v4.src_ip =
+                                               ipv4_spec->hdr.src_addr;
+                                       filter->input.mask.v4.src_ip =
+                                               ipv4_mask->hdr.src_addr;
+
+                                       input_set |= ICE_INSET_IPV4_SRC;
+                               }
+
+                               if (ipv4_mask->hdr.dst_addr) {
+                                       filter->input.ip.v4.dst_ip =
+                                               ipv4_spec->hdr.dst_addr;
+                                       filter->input.mask.v4.dst_ip =
+                                               ipv4_mask->hdr.dst_addr;
+
+                                       input_set |= ICE_INSET_IPV4_DST;
+                               }
+                       }
+
+                       flow_type = ICE_FLTR_PTYPE_NONF_IPV4_OTHER;
+                       break;
+               case RTE_FLOW_ITEM_TYPE_TCP:
+                       tcp_spec = item->spec;
+                       tcp_mask = item->mask;
+
+                       if (l3 == RTE_FLOW_ITEM_TYPE_IPV4)
+                               flow_type = ICE_FLTR_PTYPE_NONF_IPV4_TCP;
+
+                       if (tcp_spec && tcp_mask) {
+                               /* Check TCP mask and update input set */
+                               if (tcp_mask->hdr.sent_seq ||
+                                   tcp_mask->hdr.recv_ack ||
+                                   tcp_mask->hdr.data_off ||
+                                   tcp_mask->hdr.tcp_flags ||
+                                   tcp_mask->hdr.rx_win ||
+                                   tcp_mask->hdr.cksum ||
+                                   tcp_mask->hdr.tcp_urp) {
+                                       rte_flow_error_set(error, EINVAL,
+                                               RTE_FLOW_ERROR_TYPE_ITEM,
+                                               item,
+                                               "Invalid TCP mask");
+                                       return -rte_errno;
+                               }
+
+                               if (l3 == RTE_FLOW_ITEM_TYPE_IPV4 &&
+                                   tcp_mask->hdr.src_port) {
+                                       input_set |= ICE_INSET_TCP_SRC_PORT;
+                                       filter->input.ip.v4.src_port =
+                                               tcp_spec->hdr.src_port;
+                                       filter->input.mask.v4.src_port =
+                                               tcp_mask->hdr.src_port;
+                               }
+
+                               if (l3 == RTE_FLOW_ITEM_TYPE_IPV4 &&
+                                   tcp_mask->hdr.dst_port) {
+                                       input_set |= ICE_INSET_TCP_DST_PORT;
+                                       filter->input.ip.v4.dst_port =
+                                               tcp_spec->hdr.dst_port;
+                                       filter->input.mask.v4.dst_port =
+                                               tcp_mask->hdr.dst_port;
+                               }
+                       }
+                       break;
+               case RTE_FLOW_ITEM_TYPE_UDP:
+                       udp_spec = item->spec;
+                       udp_mask = item->mask;
+
+                       if (l3 == RTE_FLOW_ITEM_TYPE_IPV4)
+                               flow_type = ICE_FLTR_PTYPE_NONF_IPV4_UDP;
+
+                       if (udp_spec && udp_mask) {
+                               /* Check UDP mask and update input set*/
+                               if (udp_mask->hdr.dgram_len ||
+                                   udp_mask->hdr.dgram_cksum) {
+                                       rte_flow_error_set(error, EINVAL,
+                                               RTE_FLOW_ERROR_TYPE_ITEM,
+                                               item,
+                                               "Invalid UDP mask");
+                                       return -rte_errno;
+                               }
+
+                               if (l3 == RTE_FLOW_ITEM_TYPE_IPV4 &&
+                                   udp_mask->hdr.src_port) {
+                                       input_set |= ICE_INSET_UDP_SRC_PORT;
+                                       filter->input.ip.v4.src_port =
+                                               udp_spec->hdr.src_port;
+                                       filter->input.mask.v4.src_port =
+                                               udp_mask->hdr.src_port;
+                               }
+
+                               if (l3 == RTE_FLOW_ITEM_TYPE_IPV4 &&
+                                   udp_mask->hdr.dst_port) {
+                                       input_set |= ICE_INSET_UDP_DST_PORT;
+                                       filter->input.ip.v4.dst_port =
+                                               udp_spec->hdr.dst_port;
+                                       filter->input.mask.v4.dst_port =
+                                               udp_mask->hdr.dst_port;
+                               }
+                       }
+                       break;
+               case RTE_FLOW_ITEM_TYPE_SCTP:
+                       sctp_spec = item->spec;
+                       sctp_mask = item->mask;
+
+                       if (l3 == RTE_FLOW_ITEM_TYPE_IPV4)
+                               flow_type = ICE_FLTR_PTYPE_NONF_IPV4_SCTP;
+
+                       if (sctp_spec && sctp_mask) {
+                               if (l3 == RTE_FLOW_ITEM_TYPE_IPV4 &&
+                                   sctp_mask->hdr.src_port) {
+                                       input_set |= ICE_INSET_SCTP_SRC_PORT;
+                                       filter->input.ip.v4.src_port =
+                                               sctp_spec->hdr.src_port;
+                                       filter->input.mask.v4.src_port =
+                                               sctp_mask->hdr.src_port;
+                               }
+
+                               if (l3 == RTE_FLOW_ITEM_TYPE_IPV4 &&
+                                   sctp_mask->hdr.dst_port) {
+                                       input_set |= ICE_INSET_SCTP_DST_PORT;
+                                       filter->input.ip.v4.dst_port =
+                                               sctp_spec->hdr.dst_port;
+                                       filter->input.mask.v4.dst_port =
+                                               sctp_mask->hdr.dst_port;
+                               }
+                       }
+                       break;
+               case RTE_FLOW_ITEM_TYPE_VOID:
+                       break;
+               default:
+                       rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_ITEM,
+                               item,
+                               "Invalid pattern item.");
+                       return -rte_errno;
+               }
+       }
+
+       filter->input.flow_type = flow_type;
+       filter->input_set = input_set;
+
+       return 0;
+}
+
+static int
+ice_acl_parse(struct ice_adapter *ad,
+              struct ice_pattern_match_item *array,
+              uint32_t array_len,
+              const struct rte_flow_item pattern[],
+              const struct rte_flow_action actions[],
+              void **meta,
+              struct rte_flow_error *error)
+{
+       struct ice_pf *pf = &ad->pf;
+       struct ice_acl_conf *filter = &pf->acl.conf;
+       struct ice_pattern_match_item *item = NULL;
+       uint64_t input_set;
+       int ret;
+
+       memset(filter, 0, sizeof(*filter));
+       item = ice_search_pattern_match_item(pattern, array, array_len, error);
+       if (!item)
+               return -rte_errno;
+
+       ret = ice_acl_parse_pattern(ad, pattern, error, filter);
+       if (ret)
+               goto error;
+       input_set = filter->input_set;
+       if (!input_set || input_set & ~item->input_set_mask) {
+               rte_flow_error_set(error, EINVAL,
+                                  RTE_FLOW_ERROR_TYPE_ITEM_SPEC,
+                                  pattern,
+                                  "Invalid input set");
+               ret = -rte_errno;
+               goto error;
+       }
+
+       ret = ice_acl_parse_action(ad, actions, error, filter);
+       if (ret)
+               goto error;
+
+       if (meta)
+               *meta = filter;
+
+error:
+       rte_free(item);
+       return ret;
+}
+
+static int
+ice_acl_bitmap_init(struct ice_pf *pf)
+{
+       uint32_t bmp_size;
+       void *mem = NULL;
+       struct rte_bitmap *slots;
+       int ret = 0;
+       bmp_size = rte_bitmap_get_memory_footprint(MAX_ACL_SLOTS_ID);
+       mem = rte_zmalloc("create_acl_bmap", bmp_size, RTE_CACHE_LINE_SIZE);
+       if (mem == NULL) {
+               PMD_DRV_LOG(ERR, "Failed to allocate memory for acl bitmap.");
+               return -rte_errno;
+       }
+
+       slots = rte_bitmap_init_with_all_set(MAX_ACL_SLOTS_ID, mem, bmp_size);
+       if (slots == NULL) {
+               PMD_DRV_LOG(ERR, "Failed to initialize acl bitmap.");
+               ret = -rte_errno;
+               goto err_acl_mem_alloc;
+       }
+       pf->acl.slots = slots;
+       return 0;
+
+err_acl_mem_alloc:
+       rte_free(mem);
+       return ret;
+}
+
+static int
+ice_acl_init(struct ice_adapter *ad)
+{
+       int ret = 0;
+       struct ice_pf *pf = &ad->pf;
+       struct ice_hw *hw = ICE_PF_TO_HW(pf);
+       struct ice_flow_parser *parser = &ice_acl_parser;
+
+       ret = ice_acl_prof_alloc(hw);
+       if (ret) {
+               PMD_DRV_LOG(ERR, "Cannot allocate memory for "
+                           "ACL profile.");
+               return -ENOMEM;
+       }
+
+       ret = ice_acl_setup(pf);
+       if (ret)
+               return ret;
+
+       ret = ice_acl_bitmap_init(pf);
+       if (ret)
+               return ret;
+
+       ret = ice_acl_prof_init(pf);
+       if (ret)
+               return ret;
+
+       return ice_register_parser(parser, ad);
+}
+
+static void
+ice_acl_prof_free(struct ice_hw *hw)
+{
+       enum ice_fltr_ptype ptype;
+
+       for (ptype = ICE_FLTR_PTYPE_NONF_NONE + 1;
+            ptype < ICE_FLTR_PTYPE_MAX; ptype++) {
+               rte_free(hw->acl_prof[ptype]);
+               hw->acl_prof[ptype] = NULL;
+       }
+
+       rte_free(hw->acl_prof);
+       hw->acl_prof = NULL;
+}
+
+static void
+ice_acl_uninit(struct ice_adapter *ad)
+{
+       struct ice_pf *pf = &ad->pf;
+       struct ice_hw *hw = ICE_PF_TO_HW(pf);
+       struct ice_flow_parser *parser = &ice_acl_parser;
+
+       ice_unregister_parser(parser, ad);
+
+       ice_deinit_acl(pf);
+       ice_acl_prof_free(hw);
+}
+
+static struct
+ice_flow_engine ice_acl_engine = {
+       .init = ice_acl_init,
+       .uninit = ice_acl_uninit,
+       .create = ice_acl_create_filter,
+       .destroy = ice_acl_destroy_filter,
+       .free = ice_acl_filter_free,
+       .type = ICE_FLOW_ENGINE_ACL,
+};
+
+static struct
+ice_flow_parser ice_acl_parser = {
+       .engine = &ice_acl_engine,
+       .array = ice_acl_pattern,
+       .array_len = RTE_DIM(ice_acl_pattern),
+       .parse_pattern_action = ice_acl_parse,
+       .stage = ICE_FLOW_STAGE_DISTRIBUTOR,
+};
+
+RTE_INIT(ice_acl_engine_init)
+{
+       struct ice_flow_engine *engine = &ice_acl_engine;
+       ice_register_flow_engine(engine);
+}
index 9789096..05218af 100644 (file)
@@ -49,6 +49,8 @@
 #define ICE_PKG_FILE_SEARCH_PATH_UPDATES "/lib/firmware/updates/intel/ice/ddp/"
 #define ICE_MAX_PKG_FILENAME_SIZE   256
 
+#define MAX_ACL_ENTRIES    512
+
 /**
  * vlan_id is a 12 bit number.
  * The VFTA array is actually a 4096 bit array, 128 of 32bit elements.
@@ -398,6 +400,20 @@ struct ice_hash_ctx {
        struct ice_hash_gtpu_ctx gtpu6;
 };
 
+struct ice_acl_conf {
+       struct ice_fdir_fltr input;
+       uint64_t input_set;
+};
+
+/**
+ * A structure used to define fields of ACL related info.
+ */
+struct ice_acl_info {
+       struct ice_acl_conf conf;
+       struct rte_bitmap *slots;
+       uint64_t hw_entry_id[MAX_ACL_ENTRIES];
+};
+
 struct ice_pf {
        struct ice_adapter *adapter; /* The adapter this PF associate to */
        struct ice_vsi *main_vsi; /* pointer to main VSI structure */
@@ -421,6 +437,7 @@ struct ice_pf {
        uint16_t fdir_nb_qps; /* The number of queue pairs of Flow Director */
        uint16_t fdir_qp_offset;
        struct ice_fdir_info fdir; /* flow director info */
+       struct ice_acl_info acl; /* ACL info */
        struct ice_hash_ctx hash_ctx;
        uint16_t hw_prof_cnt[ICE_FLTR_PTYPE_MAX][ICE_FD_HW_SEG_MAX];
        uint16_t fdir_fltr_cnt[ICE_FLTR_PTYPE_MAX][ICE_FD_HW_SEG_MAX];
index 54b0316..1429cbc 100644 (file)
@@ -1896,6 +1896,8 @@ ice_register_parser(struct ice_flow_parser *parser,
                        TAILQ_INSERT_TAIL(list, parser_node, node);
                else if (parser->engine->type == ICE_FLOW_ENGINE_FDIR)
                        TAILQ_INSERT_HEAD(list, parser_node, node);
+               else if (parser->engine->type == ICE_FLOW_ENGINE_ACL)
+                       TAILQ_INSERT_HEAD(list, parser_node, node);
                else
                        return -EINVAL;
        }
index f6a64d9..40cae3c 100644 (file)
@@ -10,7 +10,8 @@ sources = files(
        'ice_switch_filter.c',
        'ice_generic_flow.c',
        'ice_fdir_filter.c',
-       'ice_hash.c'
+       'ice_hash.c',
+       'ice_acl_filter.c'
        )
 
 deps += ['hash', 'net', 'common_iavf']