// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
/* Copyright (c) 2016-2018 Mellanox Technologies. All rights reserved */

#include <linux/kernel.h>
#include <linux/types.h>
#include <linux/rhashtable.h>
#include <linux/bitops.h>
#include <linux/in6.h>
#include <linux/notifier.h>
#include <linux/inetdevice.h>
#include <linux/netdevice.h>
#include <linux/if_bridge.h>
#include <linux/socket.h>
#include <linux/route.h>
#include <linux/gcd.h>
#include <linux/if_macvlan.h>
#include <linux/refcount.h>
#include <linux/jhash.h>
#include <linux/net_namespace.h>
#include <linux/mutex.h>
#include <net/netevent.h>
#include <net/neighbour.h>
#include <net/arp.h>
#include <net/ip_fib.h>
#include <net/ip6_fib.h>
#include <net/nexthop.h>
#include <net/fib_rules.h>
#include <net/ip_tunnels.h>
#include <net/l3mdev.h>
#include <net/addrconf.h>
#include <net/ndisc.h>
#include <net/ipv6.h>
#include <net/fib_notifier.h>
#include <net/switchdev.h>

#include "spectrum.h"
#include "core.h"
#include "reg.h"
#include "spectrum_cnt.h"
#include "spectrum_dpipe.h"
#include "spectrum_ipip.h"
#include "spectrum_mr.h"
#include "spectrum_mr_tcam.h"
#include "spectrum_router.h"
#include "spectrum_span.h"

struct mlxsw_sp_fib;
struct mlxsw_sp_vr;
struct mlxsw_sp_lpm_tree;
struct mlxsw_sp_rif_ops;

struct mlxsw_sp_rif {
        struct list_head nexthop_list;
        struct list_head neigh_list;
        struct net_device *dev; /* NULL for underlay RIF */
        struct mlxsw_sp_fid *fid;
        unsigned char addr[ETH_ALEN];
        int mtu;
        u16 rif_index;
        u16 vr_id;
        const struct mlxsw_sp_rif_ops *ops;
        struct mlxsw_sp *mlxsw_sp;

        unsigned int counter_ingress;
        bool counter_ingress_valid;
        unsigned int counter_egress;
        bool counter_egress_valid;
};

struct mlxsw_sp_rif_params {
        struct net_device *dev;
        union {
                u16 system_port;
                u16 lag_id;
        };
        u16 vid;
        bool lag;
};

struct mlxsw_sp_rif_subport {
        struct mlxsw_sp_rif common;
        refcount_t ref_count;
        union {
                u16 system_port;
                u16 lag_id;
        };
        u16 vid;
        bool lag;
};

struct mlxsw_sp_rif_ipip_lb {
        struct mlxsw_sp_rif common;
        struct mlxsw_sp_rif_ipip_lb_config lb_config;
        u16 ul_vr_id; /* Reserved for Spectrum-2. */
        u16 ul_rif_id; /* Reserved for Spectrum. */
};

struct mlxsw_sp_rif_params_ipip_lb {
        struct mlxsw_sp_rif_params common;
        struct mlxsw_sp_rif_ipip_lb_config lb_config;
};

struct mlxsw_sp_rif_ops {
        enum mlxsw_sp_rif_type type;
        size_t rif_size;

        void (*setup)(struct mlxsw_sp_rif *rif,
                      const struct mlxsw_sp_rif_params *params);
        int (*configure)(struct mlxsw_sp_rif *rif);
        void (*deconfigure)(struct mlxsw_sp_rif *rif);
        struct mlxsw_sp_fid * (*fid_get)(struct mlxsw_sp_rif *rif,
                                         struct netlink_ext_ack *extack);
        void (*fdb_del)(struct mlxsw_sp_rif *rif, const char *mac);
};

struct mlxsw_sp_router_ops {
        int (*init)(struct mlxsw_sp *mlxsw_sp);
};

static struct mlxsw_sp_rif *
mlxsw_sp_rif_find_by_dev(const struct mlxsw_sp *mlxsw_sp,
                         const struct net_device *dev);
static void mlxsw_sp_rif_destroy(struct mlxsw_sp_rif *rif);
static void mlxsw_sp_lpm_tree_hold(struct mlxsw_sp_lpm_tree *lpm_tree);
static void mlxsw_sp_lpm_tree_put(struct mlxsw_sp *mlxsw_sp,
                                  struct mlxsw_sp_lpm_tree *lpm_tree);
static int mlxsw_sp_vr_lpm_tree_bind(struct mlxsw_sp *mlxsw_sp,
                                     const struct mlxsw_sp_fib *fib,
                                     u8 tree_id);
static int mlxsw_sp_vr_lpm_tree_unbind(struct mlxsw_sp *mlxsw_sp,
                                       const struct mlxsw_sp_fib *fib);

static unsigned int *
mlxsw_sp_rif_p_counter_get(struct mlxsw_sp_rif *rif,
                           enum mlxsw_sp_rif_counter_dir dir)
{
        switch (dir) {
        case MLXSW_SP_RIF_COUNTER_EGRESS:
                return &rif->counter_egress;
        case MLXSW_SP_RIF_COUNTER_INGRESS:
                return &rif->counter_ingress;
        }
        return NULL;
}

static bool
mlxsw_sp_rif_counter_valid_get(struct mlxsw_sp_rif *rif,
                               enum mlxsw_sp_rif_counter_dir dir)
{
        switch (dir) {
        case MLXSW_SP_RIF_COUNTER_EGRESS:
                return rif->counter_egress_valid;
        case MLXSW_SP_RIF_COUNTER_INGRESS:
                return rif->counter_ingress_valid;
        }
        return false;
}

static void
mlxsw_sp_rif_counter_valid_set(struct mlxsw_sp_rif *rif,
                               enum mlxsw_sp_rif_counter_dir dir,
                               bool valid)
{
        switch (dir) {
        case MLXSW_SP_RIF_COUNTER_EGRESS:
                rif->counter_egress_valid = valid;
                break;
        case MLXSW_SP_RIF_COUNTER_INGRESS:
                rif->counter_ingress_valid = valid;
                break;
        }
}

static int mlxsw_sp_rif_counter_edit(struct mlxsw_sp *mlxsw_sp, u16 rif_index,
                                     unsigned int counter_index, bool enable,
                                     enum mlxsw_sp_rif_counter_dir dir)
{
        char ritr_pl[MLXSW_REG_RITR_LEN];
        bool is_egress = false;
        int err;

        if (dir == MLXSW_SP_RIF_COUNTER_EGRESS)
                is_egress = true;
        mlxsw_reg_ritr_rif_pack(ritr_pl, rif_index);
        err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ritr), ritr_pl);
        if (err)
                return err;

        mlxsw_reg_ritr_counter_pack(ritr_pl, counter_index, enable,
                                    is_egress);
        return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ritr), ritr_pl);
}

int mlxsw_sp_rif_counter_value_get(struct mlxsw_sp *mlxsw_sp,
                                   struct mlxsw_sp_rif *rif,
                                   enum mlxsw_sp_rif_counter_dir dir, u64 *cnt)
{
        char ricnt_pl[MLXSW_REG_RICNT_LEN];
        unsigned int *p_counter_index;
        bool valid;
        int err;

        valid = mlxsw_sp_rif_counter_valid_get(rif, dir);
        if (!valid)
                return -EINVAL;

        p_counter_index = mlxsw_sp_rif_p_counter_get(rif, dir);
        if (!p_counter_index)
                return -EINVAL;
        mlxsw_reg_ricnt_pack(ricnt_pl, *p_counter_index,
                             MLXSW_REG_RICNT_OPCODE_NOP);
        err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ricnt), ricnt_pl);
        if (err)
                return err;
        *cnt = mlxsw_reg_ricnt_good_unicast_packets_get(ricnt_pl);
        return 0;
}

static int mlxsw_sp_rif_counter_clear(struct mlxsw_sp *mlxsw_sp,
                                      unsigned int counter_index)
{
        char ricnt_pl[MLXSW_REG_RICNT_LEN];

        mlxsw_reg_ricnt_pack(ricnt_pl, counter_index,
                             MLXSW_REG_RICNT_OPCODE_CLEAR);
        return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ricnt), ricnt_pl);
}

int mlxsw_sp_rif_counter_alloc(struct mlxsw_sp *mlxsw_sp,
                               struct mlxsw_sp_rif *rif,
                               enum mlxsw_sp_rif_counter_dir dir)
{
        unsigned int *p_counter_index;
        int err;

        p_counter_index = mlxsw_sp_rif_p_counter_get(rif, dir);
        if (!p_counter_index)
                return -EINVAL;
        err = mlxsw_sp_counter_alloc(mlxsw_sp, MLXSW_SP_COUNTER_SUB_POOL_RIF,
                                     p_counter_index);
        if (err)
                return err;

        err = mlxsw_sp_rif_counter_clear(mlxsw_sp, *p_counter_index);
        if (err)
                goto err_counter_clear;

        err = mlxsw_sp_rif_counter_edit(mlxsw_sp, rif->rif_index,
                                        *p_counter_index, true, dir);
        if (err)
                goto err_counter_edit;
        mlxsw_sp_rif_counter_valid_set(rif, dir, true);
        return 0;

err_counter_edit:
err_counter_clear:
        mlxsw_sp_counter_free(mlxsw_sp, MLXSW_SP_COUNTER_SUB_POOL_RIF,
                              *p_counter_index);
        return err;
}

void mlxsw_sp_rif_counter_free(struct mlxsw_sp *mlxsw_sp,
                               struct mlxsw_sp_rif *rif,
                               enum mlxsw_sp_rif_counter_dir dir)
{
        unsigned int *p_counter_index;

        if (!mlxsw_sp_rif_counter_valid_get(rif, dir))
                return;

        p_counter_index = mlxsw_sp_rif_p_counter_get(rif, dir);
        if (WARN_ON(!p_counter_index))
                return;
        mlxsw_sp_rif_counter_edit(mlxsw_sp, rif->rif_index,
                                  *p_counter_index, false, dir);
        mlxsw_sp_counter_free(mlxsw_sp, MLXSW_SP_COUNTER_SUB_POOL_RIF,
                              *p_counter_index);
        mlxsw_sp_rif_counter_valid_set(rif, dir, false);
}

static void mlxsw_sp_rif_counters_alloc(struct mlxsw_sp_rif *rif)
{
        struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
        struct devlink *devlink;

        devlink = priv_to_devlink(mlxsw_sp->core);
        if (!devlink_dpipe_table_counter_enabled(devlink,
                                                 MLXSW_SP_DPIPE_TABLE_NAME_ERIF))
                return;
        mlxsw_sp_rif_counter_alloc(mlxsw_sp, rif, MLXSW_SP_RIF_COUNTER_EGRESS);
}

static void mlxsw_sp_rif_counters_free(struct mlxsw_sp_rif *rif)
{
        struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;

        mlxsw_sp_rif_counter_free(mlxsw_sp, rif, MLXSW_SP_RIF_COUNTER_EGRESS);
}

#define MLXSW_SP_PREFIX_COUNT (sizeof(struct in6_addr) * BITS_PER_BYTE + 1)

struct mlxsw_sp_prefix_usage {
        DECLARE_BITMAP(b, MLXSW_SP_PREFIX_COUNT);
};

#define mlxsw_sp_prefix_usage_for_each(prefix, prefix_usage) \
        for_each_set_bit(prefix, (prefix_usage)->b, MLXSW_SP_PREFIX_COUNT)

static bool
mlxsw_sp_prefix_usage_eq(struct mlxsw_sp_prefix_usage *prefix_usage1,
                         struct mlxsw_sp_prefix_usage *prefix_usage2)
{
        return !memcmp(prefix_usage1, prefix_usage2, sizeof(*prefix_usage1));
}

static void
mlxsw_sp_prefix_usage_cpy(struct mlxsw_sp_prefix_usage *prefix_usage1,
                          struct mlxsw_sp_prefix_usage *prefix_usage2)
{
        memcpy(prefix_usage1, prefix_usage2, sizeof(*prefix_usage1));
}

static void
mlxsw_sp_prefix_usage_set(struct mlxsw_sp_prefix_usage *prefix_usage,
                          unsigned char prefix_len)
{
        set_bit(prefix_len, prefix_usage->b);
}

static void
mlxsw_sp_prefix_usage_clear(struct mlxsw_sp_prefix_usage *prefix_usage,
                            unsigned char prefix_len)
{
        clear_bit(prefix_len, prefix_usage->b);
}

struct mlxsw_sp_fib_key {
        unsigned char addr[sizeof(struct in6_addr)];
        unsigned char prefix_len;
};

enum mlxsw_sp_fib_entry_type {
        MLXSW_SP_FIB_ENTRY_TYPE_REMOTE,
        MLXSW_SP_FIB_ENTRY_TYPE_LOCAL,
        MLXSW_SP_FIB_ENTRY_TYPE_TRAP,
        MLXSW_SP_FIB_ENTRY_TYPE_BLACKHOLE,
        MLXSW_SP_FIB_ENTRY_TYPE_UNREACHABLE,

        /* This is a special case of local delivery, where a packet should be
         * decapsulated on reception. Note that there is no corresponding ENCAP,
         * because that's a type of next hop, not of FIB entry. (There can be
         * several next hops in a REMOTE entry, and some of them may be
         * encapsulating entries.)
         */
        MLXSW_SP_FIB_ENTRY_TYPE_IPIP_DECAP,
        MLXSW_SP_FIB_ENTRY_TYPE_NVE_DECAP,
};

struct mlxsw_sp_nexthop_group_info;
struct mlxsw_sp_nexthop_group;
struct mlxsw_sp_fib_entry;

struct mlxsw_sp_fib_node {
        struct mlxsw_sp_fib_entry *fib_entry;
        struct list_head list;
        struct rhash_head ht_node;
        struct mlxsw_sp_fib *fib;
        struct mlxsw_sp_fib_key key;
};

struct mlxsw_sp_fib_entry_decap {
        struct mlxsw_sp_ipip_entry *ipip_entry;
        u32 tunnel_index;
};

static struct mlxsw_sp_fib_entry_priv *
mlxsw_sp_fib_entry_priv_create(const struct mlxsw_sp_router_ll_ops *ll_ops)
{
        struct mlxsw_sp_fib_entry_priv *priv;

        if (!ll_ops->fib_entry_priv_size)
                /* No need to have priv */
                return NULL;

        priv = kzalloc(sizeof(*priv) + ll_ops->fib_entry_priv_size, GFP_KERNEL);
        if (!priv)
                return ERR_PTR(-ENOMEM);
        refcount_set(&priv->refcnt, 1);
        return priv;
}

static void
mlxsw_sp_fib_entry_priv_destroy(struct mlxsw_sp_fib_entry_priv *priv)
{
        kfree(priv);
}

static void mlxsw_sp_fib_entry_priv_hold(struct mlxsw_sp_fib_entry_priv *priv)
{
        refcount_inc(&priv->refcnt);
}

static void mlxsw_sp_fib_entry_priv_put(struct mlxsw_sp_fib_entry_priv *priv)
{
        if (!priv || !refcount_dec_and_test(&priv->refcnt))
                return;
        mlxsw_sp_fib_entry_priv_destroy(priv);
}

static void mlxsw_sp_fib_entry_op_ctx_priv_hold(struct mlxsw_sp_fib_entry_op_ctx *op_ctx,
                                                struct mlxsw_sp_fib_entry_priv *priv)
{
        if (!priv)
                return;
        mlxsw_sp_fib_entry_priv_hold(priv);
        list_add(&priv->list, &op_ctx->fib_entry_priv_list);
}

static void mlxsw_sp_fib_entry_op_ctx_priv_put_all(struct mlxsw_sp_fib_entry_op_ctx *op_ctx)
{
        struct mlxsw_sp_fib_entry_priv *priv, *tmp;

        list_for_each_entry_safe(priv, tmp, &op_ctx->fib_entry_priv_list, list)
                mlxsw_sp_fib_entry_priv_put(priv);
        INIT_LIST_HEAD(&op_ctx->fib_entry_priv_list);
}

struct mlxsw_sp_fib_entry {
        struct mlxsw_sp_fib_node *fib_node;
        enum mlxsw_sp_fib_entry_type type;
        struct list_head nexthop_group_node;
        struct mlxsw_sp_nexthop_group *nh_group;
        struct mlxsw_sp_fib_entry_decap decap; /* Valid for decap entries. */
        struct mlxsw_sp_fib_entry_priv *priv;
};

struct mlxsw_sp_fib4_entry {
        struct mlxsw_sp_fib_entry common;
        struct fib_info *fi;
        u32 tb_id;
        u8 tos;
        u8 type;
};

struct mlxsw_sp_fib6_entry {
        struct mlxsw_sp_fib_entry common;
        struct list_head rt6_list;
        unsigned int nrt6;
};

struct mlxsw_sp_rt6 {
        struct list_head list;
        struct fib6_info *rt;
};

struct mlxsw_sp_lpm_tree {
        u8 id; /* tree ID */
        unsigned int ref_count;
        enum mlxsw_sp_l3proto proto;
        unsigned long prefix_ref_count[MLXSW_SP_PREFIX_COUNT];
        struct mlxsw_sp_prefix_usage prefix_usage;
};

struct mlxsw_sp_fib {
        struct rhashtable ht;
        struct list_head node_list;
        struct mlxsw_sp_vr *vr;
        struct mlxsw_sp_lpm_tree *lpm_tree;
        enum mlxsw_sp_l3proto proto;
        const struct mlxsw_sp_router_ll_ops *ll_ops;
};

struct mlxsw_sp_vr {
        u16 id; /* virtual router ID */
        u32 tb_id; /* kernel fib table id */
        unsigned int rif_count;
        struct mlxsw_sp_fib *fib4;
        struct mlxsw_sp_fib *fib6;
        struct mlxsw_sp_mr_table *mr_table[MLXSW_SP_L3_PROTO_MAX];
        struct mlxsw_sp_rif *ul_rif;
        refcount_t ul_rif_refcnt;
};

static int mlxsw_sp_router_ll_basic_init(struct mlxsw_sp *mlxsw_sp, u16 vr_id,
                                         enum mlxsw_sp_l3proto proto)
{
        return 0;
}

static int mlxsw_sp_router_ll_basic_ralta_write(struct mlxsw_sp *mlxsw_sp, char *xralta_pl)
{
        return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ralta),
                               xralta_pl + MLXSW_REG_XRALTA_RALTA_OFFSET);
}

static int mlxsw_sp_router_ll_basic_ralst_write(struct mlxsw_sp *mlxsw_sp, char *xralst_pl)
{
        return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ralst),
                               xralst_pl + MLXSW_REG_XRALST_RALST_OFFSET);
}

static int mlxsw_sp_router_ll_basic_raltb_write(struct mlxsw_sp *mlxsw_sp, char *xraltb_pl)
{
        return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(raltb),
                               xraltb_pl + MLXSW_REG_XRALTB_RALTB_OFFSET);
}

static const struct rhashtable_params mlxsw_sp_fib_ht_params;

static struct mlxsw_sp_fib *mlxsw_sp_fib_create(struct mlxsw_sp *mlxsw_sp,
                                                struct mlxsw_sp_vr *vr,
                                                enum mlxsw_sp_l3proto proto)
{
        const struct mlxsw_sp_router_ll_ops *ll_ops = mlxsw_sp->router->proto_ll_ops[proto];
        struct mlxsw_sp_lpm_tree *lpm_tree;
        struct mlxsw_sp_fib *fib;
        int err;

        err = ll_ops->init(mlxsw_sp, vr->id, proto);
        if (err)
                return ERR_PTR(err);

        lpm_tree = mlxsw_sp->router->lpm.proto_trees[proto];
        fib = kzalloc(sizeof(*fib), GFP_KERNEL);
        if (!fib)
                return ERR_PTR(-ENOMEM);
        err = rhashtable_init(&fib->ht, &mlxsw_sp_fib_ht_params);
        if (err)
                goto err_rhashtable_init;
        INIT_LIST_HEAD(&fib->node_list);
        fib->proto = proto;
        fib->vr = vr;
        fib->lpm_tree = lpm_tree;
        fib->ll_ops = ll_ops;
        mlxsw_sp_lpm_tree_hold(lpm_tree);
        err = mlxsw_sp_vr_lpm_tree_bind(mlxsw_sp, fib, lpm_tree->id);
        if (err)
                goto err_lpm_tree_bind;
        return fib;

err_lpm_tree_bind:
        mlxsw_sp_lpm_tree_put(mlxsw_sp, lpm_tree);
err_rhashtable_init:
        kfree(fib);
        return ERR_PTR(err);
}

static void mlxsw_sp_fib_destroy(struct mlxsw_sp *mlxsw_sp,
                                 struct mlxsw_sp_fib *fib)
{
        mlxsw_sp_vr_lpm_tree_unbind(mlxsw_sp, fib);
        mlxsw_sp_lpm_tree_put(mlxsw_sp, fib->lpm_tree);
        WARN_ON(!list_empty(&fib->node_list));
        rhashtable_destroy(&fib->ht);
        kfree(fib);
}

static struct mlxsw_sp_lpm_tree *
mlxsw_sp_lpm_tree_find_unused(struct mlxsw_sp *mlxsw_sp)
{
        static struct mlxsw_sp_lpm_tree *lpm_tree;
        int i;

        for (i = 0; i < mlxsw_sp->router->lpm.tree_count; i++) {
                lpm_tree = &mlxsw_sp->router->lpm.trees[i];
                if (lpm_tree->ref_count == 0)
                        return lpm_tree;
        }
        return NULL;
}

static int mlxsw_sp_lpm_tree_alloc(struct mlxsw_sp *mlxsw_sp,
                                   const struct mlxsw_sp_router_ll_ops *ll_ops,
                                   struct mlxsw_sp_lpm_tree *lpm_tree)
{
        char xralta_pl[MLXSW_REG_XRALTA_LEN];

        mlxsw_reg_xralta_pack(xralta_pl, true,
                              (enum mlxsw_reg_ralxx_protocol) lpm_tree->proto,
                              lpm_tree->id);
        return ll_ops->ralta_write(mlxsw_sp, xralta_pl);
}

static void mlxsw_sp_lpm_tree_free(struct mlxsw_sp *mlxsw_sp,
                                   const struct mlxsw_sp_router_ll_ops *ll_ops,
                                   struct mlxsw_sp_lpm_tree *lpm_tree)
{
        char xralta_pl[MLXSW_REG_XRALTA_LEN];

        mlxsw_reg_xralta_pack(xralta_pl, false,
                              (enum mlxsw_reg_ralxx_protocol) lpm_tree->proto,
                              lpm_tree->id);
        ll_ops->ralta_write(mlxsw_sp, xralta_pl);
}

static int
mlxsw_sp_lpm_tree_left_struct_set(struct mlxsw_sp *mlxsw_sp,
                                  const struct mlxsw_sp_router_ll_ops *ll_ops,
                                  struct mlxsw_sp_prefix_usage *prefix_usage,
                                  struct mlxsw_sp_lpm_tree *lpm_tree)
{
        char xralst_pl[MLXSW_REG_XRALST_LEN];
        u8 root_bin = 0;
        u8 prefix;
        u8 last_prefix = MLXSW_REG_RALST_BIN_NO_CHILD;

        mlxsw_sp_prefix_usage_for_each(prefix, prefix_usage)
                root_bin = prefix;

        mlxsw_reg_xralst_pack(xralst_pl, root_bin, lpm_tree->id);
        mlxsw_sp_prefix_usage_for_each(prefix, prefix_usage) {
                if (prefix == 0)
                        continue;
                mlxsw_reg_xralst_bin_pack(xralst_pl, prefix, last_prefix,
                                          MLXSW_REG_RALST_BIN_NO_CHILD);
                last_prefix = prefix;
        }
        return ll_ops->ralst_write(mlxsw_sp, xralst_pl);
}

static struct mlxsw_sp_lpm_tree *
mlxsw_sp_lpm_tree_create(struct mlxsw_sp *mlxsw_sp,
                         const struct mlxsw_sp_router_ll_ops *ll_ops,
                         struct mlxsw_sp_prefix_usage *prefix_usage,
                         enum mlxsw_sp_l3proto proto)
{
        struct mlxsw_sp_lpm_tree *lpm_tree;
        int err;

        lpm_tree = mlxsw_sp_lpm_tree_find_unused(mlxsw_sp);
        if (!lpm_tree)
                return ERR_PTR(-EBUSY);
        lpm_tree->proto = proto;
        err = mlxsw_sp_lpm_tree_alloc(mlxsw_sp, ll_ops, lpm_tree);
        if (err)
                return ERR_PTR(err);

        err = mlxsw_sp_lpm_tree_left_struct_set(mlxsw_sp, ll_ops, prefix_usage, lpm_tree);
        if (err)
                goto err_left_struct_set;
        memcpy(&lpm_tree->prefix_usage, prefix_usage,
               sizeof(lpm_tree->prefix_usage));
        memset(&lpm_tree->prefix_ref_count, 0,
               sizeof(lpm_tree->prefix_ref_count));
        lpm_tree->ref_count = 1;
        return lpm_tree;

err_left_struct_set:
        mlxsw_sp_lpm_tree_free(mlxsw_sp, ll_ops, lpm_tree);
        return ERR_PTR(err);
}

static void mlxsw_sp_lpm_tree_destroy(struct mlxsw_sp *mlxsw_sp,
                                      const struct mlxsw_sp_router_ll_ops *ll_ops,
                                      struct mlxsw_sp_lpm_tree *lpm_tree)
{
        mlxsw_sp_lpm_tree_free(mlxsw_sp, ll_ops, lpm_tree);
}

static struct mlxsw_sp_lpm_tree *
mlxsw_sp_lpm_tree_get(struct mlxsw_sp *mlxsw_sp,
                      struct mlxsw_sp_prefix_usage *prefix_usage,
                      enum mlxsw_sp_l3proto proto)
{
        const struct mlxsw_sp_router_ll_ops *ll_ops = mlxsw_sp->router->proto_ll_ops[proto];
        struct mlxsw_sp_lpm_tree *lpm_tree;
        int i;

        for (i = 0; i < mlxsw_sp->router->lpm.tree_count; i++) {
                lpm_tree = &mlxsw_sp->router->lpm.trees[i];
                if (lpm_tree->ref_count != 0 &&
                    lpm_tree->proto == proto &&
                    mlxsw_sp_prefix_usage_eq(&lpm_tree->prefix_usage,
                                             prefix_usage)) {
                        mlxsw_sp_lpm_tree_hold(lpm_tree);
                        return lpm_tree;
                }
        }
        return mlxsw_sp_lpm_tree_create(mlxsw_sp, ll_ops, prefix_usage, proto);
}

static void mlxsw_sp_lpm_tree_hold(struct mlxsw_sp_lpm_tree *lpm_tree)
{
        lpm_tree->ref_count++;
}

static void mlxsw_sp_lpm_tree_put(struct mlxsw_sp *mlxsw_sp,
                                  struct mlxsw_sp_lpm_tree *lpm_tree)
{
        const struct mlxsw_sp_router_ll_ops *ll_ops =
                                mlxsw_sp->router->proto_ll_ops[lpm_tree->proto];

        if (--lpm_tree->ref_count == 0)
                mlxsw_sp_lpm_tree_destroy(mlxsw_sp, ll_ops, lpm_tree);
}

#define MLXSW_SP_LPM_TREE_MIN 1 /* tree 0 is reserved */

static int mlxsw_sp_lpm_init(struct mlxsw_sp *mlxsw_sp)
{
        struct mlxsw_sp_prefix_usage req_prefix_usage = {{ 0 } };
        struct mlxsw_sp_lpm_tree *lpm_tree;
        u64 max_trees;
        int err, i;

        if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, MAX_LPM_TREES))
                return -EIO;

        max_trees = MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_LPM_TREES);
        mlxsw_sp->router->lpm.tree_count = max_trees - MLXSW_SP_LPM_TREE_MIN;
        mlxsw_sp->router->lpm.trees = kcalloc(mlxsw_sp->router->lpm.tree_count,
                                             sizeof(struct mlxsw_sp_lpm_tree),
                                             GFP_KERNEL);
        if (!mlxsw_sp->router->lpm.trees)
                return -ENOMEM;

        for (i = 0; i < mlxsw_sp->router->lpm.tree_count; i++) {
                lpm_tree = &mlxsw_sp->router->lpm.trees[i];
                lpm_tree->id = i + MLXSW_SP_LPM_TREE_MIN;
        }

        lpm_tree = mlxsw_sp_lpm_tree_get(mlxsw_sp, &req_prefix_usage,
                                         MLXSW_SP_L3_PROTO_IPV4);
        if (IS_ERR(lpm_tree)) {
                err = PTR_ERR(lpm_tree);
                goto err_ipv4_tree_get;
        }
        mlxsw_sp->router->lpm.proto_trees[MLXSW_SP_L3_PROTO_IPV4] = lpm_tree;

        lpm_tree = mlxsw_sp_lpm_tree_get(mlxsw_sp, &req_prefix_usage,
                                         MLXSW_SP_L3_PROTO_IPV6);
        if (IS_ERR(lpm_tree)) {
                err = PTR_ERR(lpm_tree);
                goto err_ipv6_tree_get;
        }
        mlxsw_sp->router->lpm.proto_trees[MLXSW_SP_L3_PROTO_IPV6] = lpm_tree;

        return 0;

err_ipv6_tree_get:
        lpm_tree = mlxsw_sp->router->lpm.proto_trees[MLXSW_SP_L3_PROTO_IPV4];
        mlxsw_sp_lpm_tree_put(mlxsw_sp, lpm_tree);
err_ipv4_tree_get:
        kfree(mlxsw_sp->router->lpm.trees);
        return err;
}

static void mlxsw_sp_lpm_fini(struct mlxsw_sp *mlxsw_sp)
{
        struct mlxsw_sp_lpm_tree *lpm_tree;

        lpm_tree = mlxsw_sp->router->lpm.proto_trees[MLXSW_SP_L3_PROTO_IPV6];
        mlxsw_sp_lpm_tree_put(mlxsw_sp, lpm_tree);

        lpm_tree = mlxsw_sp->router->lpm.proto_trees[MLXSW_SP_L3_PROTO_IPV4];
        mlxsw_sp_lpm_tree_put(mlxsw_sp, lpm_tree);

        kfree(mlxsw_sp->router->lpm.trees);
}

static bool mlxsw_sp_vr_is_used(const struct mlxsw_sp_vr *vr)
{
        return !!vr->fib4 || !!vr->fib6 ||
               !!vr->mr_table[MLXSW_SP_L3_PROTO_IPV4] ||
               !!vr->mr_table[MLXSW_SP_L3_PROTO_IPV6];
}

static struct mlxsw_sp_vr *mlxsw_sp_vr_find_unused(struct mlxsw_sp *mlxsw_sp)
{
        struct mlxsw_sp_vr *vr;
        int i;

        for (i = 0; i < MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_VRS); i++) {
                vr = &mlxsw_sp->router->vrs[i];
                if (!mlxsw_sp_vr_is_used(vr))
                        return vr;
        }
        return NULL;
}

static int mlxsw_sp_vr_lpm_tree_bind(struct mlxsw_sp *mlxsw_sp,
                                     const struct mlxsw_sp_fib *fib, u8 tree_id)
{
        char xraltb_pl[MLXSW_REG_XRALTB_LEN];

        mlxsw_reg_xraltb_pack(xraltb_pl, fib->vr->id,
                              (enum mlxsw_reg_ralxx_protocol) fib->proto,
                              tree_id);
        return fib->ll_ops->raltb_write(mlxsw_sp, xraltb_pl);
}

static int mlxsw_sp_vr_lpm_tree_unbind(struct mlxsw_sp *mlxsw_sp,
                                       const struct mlxsw_sp_fib *fib)
{
        char xraltb_pl[MLXSW_REG_XRALTB_LEN];

        /* Bind to tree 0 which is default */
        mlxsw_reg_xraltb_pack(xraltb_pl, fib->vr->id,
                              (enum mlxsw_reg_ralxx_protocol) fib->proto, 0);
        return fib->ll_ops->raltb_write(mlxsw_sp, xraltb_pl);
}

static u32 mlxsw_sp_fix_tb_id(u32 tb_id)
{
        /* For our purpose, squash main, default and local tables into one */
        if (tb_id == RT_TABLE_LOCAL || tb_id == RT_TABLE_DEFAULT)
                tb_id = RT_TABLE_MAIN;
        return tb_id;
}

static struct mlxsw_sp_vr *mlxsw_sp_vr_find(struct mlxsw_sp *mlxsw_sp,
                                            u32 tb_id)
{
        struct mlxsw_sp_vr *vr;
        int i;

        tb_id = mlxsw_sp_fix_tb_id(tb_id);

        for (i = 0; i < MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_VRS); i++) {
                vr = &mlxsw_sp->router->vrs[i];
                if (mlxsw_sp_vr_is_used(vr) && vr->tb_id == tb_id)
                        return vr;
        }
        return NULL;
}

int mlxsw_sp_router_tb_id_vr_id(struct mlxsw_sp *mlxsw_sp, u32 tb_id,
                                u16 *vr_id)
{
        struct mlxsw_sp_vr *vr;
        int err = 0;

        mutex_lock(&mlxsw_sp->router->lock);
        vr = mlxsw_sp_vr_find(mlxsw_sp, tb_id);
        if (!vr) {
                err = -ESRCH;
                goto out;
        }
        *vr_id = vr->id;
out:
        mutex_unlock(&mlxsw_sp->router->lock);
        return err;
}

static struct mlxsw_sp_fib *mlxsw_sp_vr_fib(const struct mlxsw_sp_vr *vr,
                                            enum mlxsw_sp_l3proto proto)
{
        switch (proto) {
        case MLXSW_SP_L3_PROTO_IPV4:
                return vr->fib4;
        case MLXSW_SP_L3_PROTO_IPV6:
                return vr->fib6;
        }
        return NULL;
}

static struct mlxsw_sp_vr *mlxsw_sp_vr_create(struct mlxsw_sp *mlxsw_sp,
                                              u32 tb_id,
                                              struct netlink_ext_ack *extack)
{
        struct mlxsw_sp_mr_table *mr4_table, *mr6_table;
        struct mlxsw_sp_fib *fib4;
        struct mlxsw_sp_fib *fib6;
        struct mlxsw_sp_vr *vr;
        int err;

        vr = mlxsw_sp_vr_find_unused(mlxsw_sp);
        if (!vr) {
                NL_SET_ERR_MSG_MOD(extack, "Exceeded number of supported virtual routers");
                return ERR_PTR(-EBUSY);
        }
        fib4 = mlxsw_sp_fib_create(mlxsw_sp, vr, MLXSW_SP_L3_PROTO_IPV4);
        if (IS_ERR(fib4))
                return ERR_CAST(fib4);
        fib6 = mlxsw_sp_fib_create(mlxsw_sp, vr, MLXSW_SP_L3_PROTO_IPV6);
        if (IS_ERR(fib6)) {
                err = PTR_ERR(fib6);
                goto err_fib6_create;
        }
        mr4_table = mlxsw_sp_mr_table_create(mlxsw_sp, vr->id,
                                             MLXSW_SP_L3_PROTO_IPV4);
        if (IS_ERR(mr4_table)) {
                err = PTR_ERR(mr4_table);
                goto err_mr4_table_create;
        }
        mr6_table = mlxsw_sp_mr_table_create(mlxsw_sp, vr->id,
                                             MLXSW_SP_L3_PROTO_IPV6);
        if (IS_ERR(mr6_table)) {
                err = PTR_ERR(mr6_table);
                goto err_mr6_table_create;
        }

        vr->fib4 = fib4;
        vr->fib6 = fib6;
        vr->mr_table[MLXSW_SP_L3_PROTO_IPV4] = mr4_table;
        vr->mr_table[MLXSW_SP_L3_PROTO_IPV6] = mr6_table;
        vr->tb_id = tb_id;
        return vr;

err_mr6_table_create:
        mlxsw_sp_mr_table_destroy(mr4_table);
err_mr4_table_create:
        mlxsw_sp_fib_destroy(mlxsw_sp, fib6);
err_fib6_create:
        mlxsw_sp_fib_destroy(mlxsw_sp, fib4);
        return ERR_PTR(err);
}

static void mlxsw_sp_vr_destroy(struct mlxsw_sp *mlxsw_sp,
                                struct mlxsw_sp_vr *vr)
{
        mlxsw_sp_mr_table_destroy(vr->mr_table[MLXSW_SP_L3_PROTO_IPV6]);
        vr->mr_table[MLXSW_SP_L3_PROTO_IPV6] = NULL;
        mlxsw_sp_mr_table_destroy(vr->mr_table[MLXSW_SP_L3_PROTO_IPV4]);
        vr->mr_table[MLXSW_SP_L3_PROTO_IPV4] = NULL;
        mlxsw_sp_fib_destroy(mlxsw_sp, vr->fib6);
        vr->fib6 = NULL;
        mlxsw_sp_fib_destroy(mlxsw_sp, vr->fib4);
        vr->fib4 = NULL;
}

static struct mlxsw_sp_vr *mlxsw_sp_vr_get(struct mlxsw_sp *mlxsw_sp, u32 tb_id,
                                           struct netlink_ext_ack *extack)
{
        struct mlxsw_sp_vr *vr;

        tb_id = mlxsw_sp_fix_tb_id(tb_id);
        vr = mlxsw_sp_vr_find(mlxsw_sp, tb_id);
        if (!vr)
                vr = mlxsw_sp_vr_create(mlxsw_sp, tb_id, extack);
        return vr;
}

static void mlxsw_sp_vr_put(struct mlxsw_sp *mlxsw_sp, struct mlxsw_sp_vr *vr)
{
        if (!vr->rif_count && list_empty(&vr->fib4->node_list) &&
            list_empty(&vr->fib6->node_list) &&
            mlxsw_sp_mr_table_empty(vr->mr_table[MLXSW_SP_L3_PROTO_IPV4]) &&
            mlxsw_sp_mr_table_empty(vr->mr_table[MLXSW_SP_L3_PROTO_IPV6]))
                mlxsw_sp_vr_destroy(mlxsw_sp, vr);
}

static bool
mlxsw_sp_vr_lpm_tree_should_replace(struct mlxsw_sp_vr *vr,
                                    enum mlxsw_sp_l3proto proto, u8 tree_id)
{
        struct mlxsw_sp_fib *fib = mlxsw_sp_vr_fib(vr, proto);

        if (!mlxsw_sp_vr_is_used(vr))
                return false;
        if (fib->lpm_tree->id == tree_id)
                return true;
        return false;
}

static int mlxsw_sp_vr_lpm_tree_replace(struct mlxsw_sp *mlxsw_sp,
                                        struct mlxsw_sp_fib *fib,
                                        struct mlxsw_sp_lpm_tree *new_tree)
{
        struct mlxsw_sp_lpm_tree *old_tree = fib->lpm_tree;
        int err;

        fib->lpm_tree = new_tree;
        mlxsw_sp_lpm_tree_hold(new_tree);
        err = mlxsw_sp_vr_lpm_tree_bind(mlxsw_sp, fib, new_tree->id);
        if (err)
                goto err_tree_bind;
        mlxsw_sp_lpm_tree_put(mlxsw_sp, old_tree);
        return 0;

err_tree_bind:
        mlxsw_sp_lpm_tree_put(mlxsw_sp, new_tree);
        fib->lpm_tree = old_tree;
        return err;
}

static int mlxsw_sp_vrs_lpm_tree_replace(struct mlxsw_sp *mlxsw_sp,
                                         struct mlxsw_sp_fib *fib,
                                         struct mlxsw_sp_lpm_tree *new_tree)
{
        enum mlxsw_sp_l3proto proto = fib->proto;
        struct mlxsw_sp_lpm_tree *old_tree;
        u8 old_id, new_id = new_tree->id;
        struct mlxsw_sp_vr *vr;
        int i, err;

        old_tree = mlxsw_sp->router->lpm.proto_trees[proto];
        old_id = old_tree->id;

        for (i = 0; i < MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_VRS); i++) {
                vr = &mlxsw_sp->router->vrs[i];
                if (!mlxsw_sp_vr_lpm_tree_should_replace(vr, proto, old_id))
                        continue;
                err = mlxsw_sp_vr_lpm_tree_replace(mlxsw_sp,
                                                   mlxsw_sp_vr_fib(vr, proto),
                                                   new_tree);
                if (err)
                        goto err_tree_replace;
        }

        memcpy(new_tree->prefix_ref_count, old_tree->prefix_ref_count,
               sizeof(new_tree->prefix_ref_count));
        mlxsw_sp->router->lpm.proto_trees[proto] = new_tree;
        mlxsw_sp_lpm_tree_put(mlxsw_sp, old_tree);

        return 0;

err_tree_replace:
        for (i--; i >= 0; i--) {
                if (!mlxsw_sp_vr_lpm_tree_should_replace(vr, proto, new_id))
                        continue;
                mlxsw_sp_vr_lpm_tree_replace(mlxsw_sp,
                                             mlxsw_sp_vr_fib(vr, proto),
                                             old_tree);
        }
        return err;
}

static int mlxsw_sp_vrs_init(struct mlxsw_sp *mlxsw_sp)
{
        struct mlxsw_sp_vr *vr;
        u64 max_vrs;
        int i;

        if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, MAX_VRS))
                return -EIO;

        max_vrs = MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_VRS);
        mlxsw_sp->router->vrs = kcalloc(max_vrs, sizeof(struct mlxsw_sp_vr),
                                        GFP_KERNEL);
        if (!mlxsw_sp->router->vrs)
                return -ENOMEM;

        for (i = 0; i < max_vrs; i++) {
                vr = &mlxsw_sp->router->vrs[i];
                vr->id = i;
        }

        return 0;
}

static void mlxsw_sp_router_fib_flush(struct mlxsw_sp *mlxsw_sp);

static void mlxsw_sp_vrs_fini(struct mlxsw_sp *mlxsw_sp)
{
        /* At this stage we're guaranteed not to have new incoming
         * FIB notifications and the work queue is free from FIBs
         * sitting on top of mlxsw netdevs. However, we can still
         * have other FIBs queued. Flush the queue before flushing
         * the device's tables. No need for locks, as we're the only
         * writer.
         */
        mlxsw_core_flush_owq();
        mlxsw_sp_router_fib_flush(mlxsw_sp);
        kfree(mlxsw_sp->router->vrs);
}

static struct net_device *
__mlxsw_sp_ipip_netdev_ul_dev_get(const struct net_device *ol_dev)
{
        struct ip_tunnel *tun = netdev_priv(ol_dev);
        struct net *net = dev_net(ol_dev);

        return dev_get_by_index_rcu(net, tun->parms.link);
}

u32 mlxsw_sp_ipip_dev_ul_tb_id(const struct net_device *ol_dev)
{
        struct net_device *d;
        u32 tb_id;

        rcu_read_lock();
        d = __mlxsw_sp_ipip_netdev_ul_dev_get(ol_dev);
        if (d)
                tb_id = l3mdev_fib_table(d) ? : RT_TABLE_MAIN;
        else
                tb_id = RT_TABLE_MAIN;
        rcu_read_unlock();

        return tb_id;
}

static struct mlxsw_sp_rif *
mlxsw_sp_rif_create(struct mlxsw_sp *mlxsw_sp,
                    const struct mlxsw_sp_rif_params *params,
                    struct netlink_ext_ack *extack);

static struct mlxsw_sp_rif_ipip_lb *
mlxsw_sp_ipip_ol_ipip_lb_create(struct mlxsw_sp *mlxsw_sp,
                                enum mlxsw_sp_ipip_type ipipt,
                                struct net_device *ol_dev,
                                struct netlink_ext_ack *extack)
{
        struct mlxsw_sp_rif_params_ipip_lb lb_params;
        const struct mlxsw_sp_ipip_ops *ipip_ops;
        struct mlxsw_sp_rif *rif;

        ipip_ops = mlxsw_sp->router->ipip_ops_arr[ipipt];
        lb_params = (struct mlxsw_sp_rif_params_ipip_lb) {
                .common.dev = ol_dev,
                .common.lag = false,
                .lb_config = ipip_ops->ol_loopback_config(mlxsw_sp, ol_dev),
        };

        rif = mlxsw_sp_rif_create(mlxsw_sp, &lb_params.common, extack);
        if (IS_ERR(rif))
                return ERR_CAST(rif);
        return container_of(rif, struct mlxsw_sp_rif_ipip_lb, common);
}

static struct mlxsw_sp_ipip_entry *
mlxsw_sp_ipip_entry_alloc(struct mlxsw_sp *mlxsw_sp,
                          enum mlxsw_sp_ipip_type ipipt,
                          struct net_device *ol_dev)
{
        const struct mlxsw_sp_ipip_ops *ipip_ops;
        struct mlxsw_sp_ipip_entry *ipip_entry;
        struct mlxsw_sp_ipip_entry *ret = NULL;

        ipip_ops = mlxsw_sp->router->ipip_ops_arr[ipipt];
        ipip_entry = kzalloc(sizeof(*ipip_entry), GFP_KERNEL);
        if (!ipip_entry)
                return ERR_PTR(-ENOMEM);

        ipip_entry->ol_lb = mlxsw_sp_ipip_ol_ipip_lb_create(mlxsw_sp, ipipt,
                                                            ol_dev, NULL);
        if (IS_ERR(ipip_entry->ol_lb)) {
                ret = ERR_CAST(ipip_entry->ol_lb);
                goto err_ol_ipip_lb_create;
        }

        ipip_entry->ipipt = ipipt;
        ipip_entry->ol_dev = ol_dev;

        switch (ipip_ops->ul_proto) {
        case MLXSW_SP_L3_PROTO_IPV4:
                ipip_entry->parms4 = mlxsw_sp_ipip_netdev_parms4(ol_dev);
                break;
        case MLXSW_SP_L3_PROTO_IPV6:
                WARN_ON(1);
                break;
        }

        return ipip_entry;

err_ol_ipip_lb_create:
        kfree(ipip_entry);
        return ret;
}

static void
mlxsw_sp_ipip_entry_dealloc(struct mlxsw_sp_ipip_entry *ipip_entry)
{
        mlxsw_sp_rif_destroy(&ipip_entry->ol_lb->common);
        kfree(ipip_entry);
}

static bool
mlxsw_sp_ipip_entry_saddr_matches(struct mlxsw_sp *mlxsw_sp,
                                  const enum mlxsw_sp_l3proto ul_proto,
                                  union mlxsw_sp_l3addr saddr,
                                  u32 ul_tb_id,
                                  struct mlxsw_sp_ipip_entry *ipip_entry)
{
        u32 tun_ul_tb_id = mlxsw_sp_ipip_dev_ul_tb_id(ipip_entry->ol_dev);
        enum mlxsw_sp_ipip_type ipipt = ipip_entry->ipipt;
        union mlxsw_sp_l3addr tun_saddr;

        if (mlxsw_sp->router->ipip_ops_arr[ipipt]->ul_proto != ul_proto)
                return false;

        tun_saddr = mlxsw_sp_ipip_netdev_saddr(ul_proto, ipip_entry->ol_dev);
        return tun_ul_tb_id == ul_tb_id &&
               mlxsw_sp_l3addr_eq(&tun_saddr, &saddr);
}

static int
mlxsw_sp_fib_entry_decap_init(struct mlxsw_sp *mlxsw_sp,
                              struct mlxsw_sp_fib_entry *fib_entry,
                              struct mlxsw_sp_ipip_entry *ipip_entry)
{
        u32 tunnel_index;
        int err;

        err = mlxsw_sp_kvdl_alloc(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ADJ,
                                  1, &tunnel_index);
        if (err)
                return err;

        ipip_entry->decap_fib_entry = fib_entry;
        fib_entry->decap.ipip_entry = ipip_entry;
        fib_entry->decap.tunnel_index = tunnel_index;
        return 0;
}

static void mlxsw_sp_fib_entry_decap_fini(struct mlxsw_sp *mlxsw_sp,
                                          struct mlxsw_sp_fib_entry *fib_entry)
{
        /* Unlink this node from the IPIP entry that it's the decap entry of. */
        fib_entry->decap.ipip_entry->decap_fib_entry = NULL;
        fib_entry->decap.ipip_entry = NULL;
        mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ADJ,
                           1, fib_entry->decap.tunnel_index);
}

static struct mlxsw_sp_fib_node *
mlxsw_sp_fib_node_lookup(struct mlxsw_sp_fib *fib, const void *addr,
                         size_t addr_len, unsigned char prefix_len);
static int mlxsw_sp_fib_entry_update(struct mlxsw_sp *mlxsw_sp,
                                     struct mlxsw_sp_fib_entry *fib_entry);

static void
mlxsw_sp_ipip_entry_demote_decap(struct mlxsw_sp *mlxsw_sp,
                                 struct mlxsw_sp_ipip_entry *ipip_entry)
{
        struct mlxsw_sp_fib_entry *fib_entry = ipip_entry->decap_fib_entry;

        mlxsw_sp_fib_entry_decap_fini(mlxsw_sp, fib_entry);
        fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_TRAP;

        mlxsw_sp_fib_entry_update(mlxsw_sp, fib_entry);
}

static void
mlxsw_sp_ipip_entry_promote_decap(struct mlxsw_sp *mlxsw_sp,
                                  struct mlxsw_sp_ipip_entry *ipip_entry,
                                  struct mlxsw_sp_fib_entry *decap_fib_entry)
{
        if (mlxsw_sp_fib_entry_decap_init(mlxsw_sp, decap_fib_entry,
                                          ipip_entry))
                return;
        decap_fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_IPIP_DECAP;

        if (mlxsw_sp_fib_entry_update(mlxsw_sp, decap_fib_entry))
                mlxsw_sp_ipip_entry_demote_decap(mlxsw_sp, ipip_entry);
}

static struct mlxsw_sp_fib_entry *
mlxsw_sp_router_ip2me_fib_entry_find(struct mlxsw_sp *mlxsw_sp, u32 tb_id,
                                     enum mlxsw_sp_l3proto proto,
                                     const union mlxsw_sp_l3addr *addr,
                                     enum mlxsw_sp_fib_entry_type type)
{
        struct mlxsw_sp_fib_node *fib_node;
        unsigned char addr_prefix_len;
        struct mlxsw_sp_fib *fib;
        struct mlxsw_sp_vr *vr;
        const void *addrp;
        size_t addr_len;
        u32 addr4;

        vr = mlxsw_sp_vr_find(mlxsw_sp, tb_id);
        if (!vr)
                return NULL;
        fib = mlxsw_sp_vr_fib(vr, proto);

        switch (proto) {
        case MLXSW_SP_L3_PROTO_IPV4:
                addr4 = be32_to_cpu(addr->addr4);
                addrp = &addr4;
                addr_len = 4;
                addr_prefix_len = 32;
                break;
        case MLXSW_SP_L3_PROTO_IPV6:
        default:
                WARN_ON(1);
                return NULL;
        }

        fib_node = mlxsw_sp_fib_node_lookup(fib, addrp, addr_len,
                                            addr_prefix_len);
        if (!fib_node || fib_node->fib_entry->type != type)
                return NULL;

        return fib_node->fib_entry;
}

/* Given an IPIP entry, find the corresponding decap route. */
static struct mlxsw_sp_fib_entry *
mlxsw_sp_ipip_entry_find_decap(struct mlxsw_sp *mlxsw_sp,
                               struct mlxsw_sp_ipip_entry *ipip_entry)
{
        static struct mlxsw_sp_fib_node *fib_node;
        const struct mlxsw_sp_ipip_ops *ipip_ops;
        unsigned char saddr_prefix_len;
        union mlxsw_sp_l3addr saddr;
        struct mlxsw_sp_fib *ul_fib;
        struct mlxsw_sp_vr *ul_vr;
        const void *saddrp;
        size_t saddr_len;
        u32 ul_tb_id;
        u32 saddr4;

        ipip_ops = mlxsw_sp->router->ipip_ops_arr[ipip_entry->ipipt];

        ul_tb_id = mlxsw_sp_ipip_dev_ul_tb_id(ipip_entry->ol_dev);
        ul_vr = mlxsw_sp_vr_find(mlxsw_sp, ul_tb_id);
        if (!ul_vr)
                return NULL;

        ul_fib = mlxsw_sp_vr_fib(ul_vr, ipip_ops->ul_proto);
        saddr = mlxsw_sp_ipip_netdev_saddr(ipip_ops->ul_proto,
                                           ipip_entry->ol_dev);

        switch (ipip_ops->ul_proto) {
        case MLXSW_SP_L3_PROTO_IPV4:
                saddr4 = be32_to_cpu(saddr.addr4);
                saddrp = &saddr4;
                saddr_len = 4;
                saddr_prefix_len = 32;
                break;
        default:
                WARN_ON(1);
                return NULL;
        }

        fib_node = mlxsw_sp_fib_node_lookup(ul_fib, saddrp, saddr_len,
                                            saddr_prefix_len);
        if (!fib_node ||
            fib_node->fib_entry->type != MLXSW_SP_FIB_ENTRY_TYPE_TRAP)
                return NULL;

        return fib_node->fib_entry;
}

static struct mlxsw_sp_ipip_entry *
mlxsw_sp_ipip_entry_create(struct mlxsw_sp *mlxsw_sp,
                           enum mlxsw_sp_ipip_type ipipt,
                           struct net_device *ol_dev)
{
        struct mlxsw_sp_ipip_entry *ipip_entry;

        ipip_entry = mlxsw_sp_ipip_entry_alloc(mlxsw_sp, ipipt, ol_dev);
        if (IS_ERR(ipip_entry))
                return ipip_entry;

        list_add_tail(&ipip_entry->ipip_list_node,
                      &mlxsw_sp->router->ipip_list);

        return ipip_entry;
}

static void
mlxsw_sp_ipip_entry_destroy(struct mlxsw_sp *mlxsw_sp,
                            struct mlxsw_sp_ipip_entry *ipip_entry)
{
        list_del(&ipip_entry->ipip_list_node);
        mlxsw_sp_ipip_entry_dealloc(ipip_entry);
}

static bool
mlxsw_sp_ipip_entry_matches_decap(struct mlxsw_sp *mlxsw_sp,
                                  const struct net_device *ul_dev,
                                  enum mlxsw_sp_l3proto ul_proto,
                                  union mlxsw_sp_l3addr ul_dip,
                                  struct mlxsw_sp_ipip_entry *ipip_entry)
{
        u32 ul_tb_id = l3mdev_fib_table(ul_dev) ? : RT_TABLE_MAIN;
        enum mlxsw_sp_ipip_type ipipt = ipip_entry->ipipt;

        if (mlxsw_sp->router->ipip_ops_arr[ipipt]->ul_proto != ul_proto)
                return false;

        return mlxsw_sp_ipip_entry_saddr_matches(mlxsw_sp, ul_proto, ul_dip,
                                                 ul_tb_id, ipip_entry);
}

/* Given decap parameters, find the corresponding IPIP entry. */
static struct mlxsw_sp_ipip_entry *
mlxsw_sp_ipip_entry_find_by_decap(struct mlxsw_sp *mlxsw_sp, int ul_dev_ifindex,
                                  enum mlxsw_sp_l3proto ul_proto,
                                  union mlxsw_sp_l3addr ul_dip)
{
        struct mlxsw_sp_ipip_entry *ipip_entry = NULL;
        struct net_device *ul_dev;

        rcu_read_lock();

        ul_dev = dev_get_by_index_rcu(mlxsw_sp_net(mlxsw_sp), ul_dev_ifindex);
        if (!ul_dev)
                goto out_unlock;

        list_for_each_entry(ipip_entry, &mlxsw_sp->router->ipip_list,
                            ipip_list_node)
                if (mlxsw_sp_ipip_entry_matches_decap(mlxsw_sp, ul_dev,
                                                      ul_proto, ul_dip,
                                                      ipip_entry))
                        goto out_unlock;

        rcu_read_unlock();

        return NULL;

out_unlock:
        rcu_read_unlock();
        return ipip_entry;
}

static bool mlxsw_sp_netdev_ipip_type(const struct mlxsw_sp *mlxsw_sp,
                                      const struct net_device *dev,
                                      enum mlxsw_sp_ipip_type *p_type)
{
        struct mlxsw_sp_router *router = mlxsw_sp->router;
        const struct mlxsw_sp_ipip_ops *ipip_ops;
        enum mlxsw_sp_ipip_type ipipt;

        for (ipipt = 0; ipipt < MLXSW_SP_IPIP_TYPE_MAX; ++ipipt) {
                ipip_ops = router->ipip_ops_arr[ipipt];
                if (dev->type == ipip_ops->dev_type) {
                        if (p_type)
                                *p_type = ipipt;
                        return true;
                }
        }
        return false;
}

bool mlxsw_sp_netdev_is_ipip_ol(const struct mlxsw_sp *mlxsw_sp,
                                const struct net_device *dev)
{
        return mlxsw_sp_netdev_ipip_type(mlxsw_sp, dev, NULL);
}

static struct mlxsw_sp_ipip_entry *
mlxsw_sp_ipip_entry_find_by_ol_dev(struct mlxsw_sp *mlxsw_sp,
                                   const struct net_device *ol_dev)
{
        struct mlxsw_sp_ipip_entry *ipip_entry;

        list_for_each_entry(ipip_entry, &mlxsw_sp->router->ipip_list,
                            ipip_list_node)
                if (ipip_entry->ol_dev == ol_dev)
                        return ipip_entry;

        return NULL;
}

static struct mlxsw_sp_ipip_entry *
mlxsw_sp_ipip_entry_find_by_ul_dev(const struct mlxsw_sp *mlxsw_sp,
                                   const struct net_device *ul_dev,
                                   struct mlxsw_sp_ipip_entry *start)
{
        struct mlxsw_sp_ipip_entry *ipip_entry;

        ipip_entry = list_prepare_entry(start, &mlxsw_sp->router->ipip_list,
                                        ipip_list_node);
        list_for_each_entry_continue(ipip_entry, &mlxsw_sp->router->ipip_list,
                                     ipip_list_node) {
                struct net_device *ol_dev = ipip_entry->ol_dev;
                struct net_device *ipip_ul_dev;

                rcu_read_lock();
                ipip_ul_dev = __mlxsw_sp_ipip_netdev_ul_dev_get(ol_dev);
                rcu_read_unlock();

                if (ipip_ul_dev == ul_dev)
                        return ipip_entry;
        }

        return NULL;
}

bool mlxsw_sp_netdev_is_ipip_ul(struct mlxsw_sp *mlxsw_sp,
                                const struct net_device *dev)
{
        bool is_ipip_ul;

        mutex_lock(&mlxsw_sp->router->lock);
        is_ipip_ul = mlxsw_sp_ipip_entry_find_by_ul_dev(mlxsw_sp, dev, NULL);
        mutex_unlock(&mlxsw_sp->router->lock);

        return is_ipip_ul;
}

static bool mlxsw_sp_netdevice_ipip_can_offload(struct mlxsw_sp *mlxsw_sp,
                                                const struct net_device *ol_dev,
                                                enum mlxsw_sp_ipip_type ipipt)
{
        const struct mlxsw_sp_ipip_ops *ops
                = mlxsw_sp->router->ipip_ops_arr[ipipt];

        return ops->can_offload(mlxsw_sp, ol_dev);
}

static int mlxsw_sp_netdevice_ipip_ol_reg_event(struct mlxsw_sp *mlxsw_sp,
                                                struct net_device *ol_dev)
{
        enum mlxsw_sp_ipip_type ipipt = MLXSW_SP_IPIP_TYPE_MAX;
        struct mlxsw_sp_ipip_entry *ipip_entry;
        enum mlxsw_sp_l3proto ul_proto;
        union mlxsw_sp_l3addr saddr;
        u32 ul_tb_id;

        mlxsw_sp_netdev_ipip_type(mlxsw_sp, ol_dev, &ipipt);
        if (mlxsw_sp_netdevice_ipip_can_offload(mlxsw_sp, ol_dev, ipipt)) {
                ul_tb_id = mlxsw_sp_ipip_dev_ul_tb_id(ol_dev);
                ul_proto = mlxsw_sp->router->ipip_ops_arr[ipipt]->ul_proto;
                saddr = mlxsw_sp_ipip_netdev_saddr(ul_proto, ol_dev);
                if (!mlxsw_sp_ipip_demote_tunnel_by_saddr(mlxsw_sp, ul_proto,
                                                          saddr, ul_tb_id,
                                                          NULL)) {
                        ipip_entry = mlxsw_sp_ipip_entry_create(mlxsw_sp, ipipt,
                                                                ol_dev);
                        if (IS_ERR(ipip_entry))
                                return PTR_ERR(ipip_entry);
                }
        }

        return 0;
}

static void mlxsw_sp_netdevice_ipip_ol_unreg_event(struct mlxsw_sp *mlxsw_sp,
                                                   struct net_device *ol_dev)
{
        struct mlxsw_sp_ipip_entry *ipip_entry;

        ipip_entry = mlxsw_sp_ipip_entry_find_by_ol_dev(mlxsw_sp, ol_dev);
        if (ipip_entry)
                mlxsw_sp_ipip_entry_destroy(mlxsw_sp, ipip_entry);
}

static void
mlxsw_sp_ipip_entry_ol_up_event(struct mlxsw_sp *mlxsw_sp,
                                struct mlxsw_sp_ipip_entry *ipip_entry)
{
        struct mlxsw_sp_fib_entry *decap_fib_entry;

        decap_fib_entry = mlxsw_sp_ipip_entry_find_decap(mlxsw_sp, ipip_entry);
        if (decap_fib_entry)
                mlxsw_sp_ipip_entry_promote_decap(mlxsw_sp, ipip_entry,
                                                  decap_fib_entry);
}

static int
mlxsw_sp_rif_ipip_lb_op(struct mlxsw_sp_rif_ipip_lb *lb_rif, u16 ul_vr_id,
                        u16 ul_rif_id, bool enable)
{
        struct mlxsw_sp_rif_ipip_lb_config lb_cf = lb_rif->lb_config;
        struct mlxsw_sp_rif *rif = &lb_rif->common;
        struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
        char ritr_pl[MLXSW_REG_RITR_LEN];
        u32 saddr4;

        switch (lb_cf.ul_protocol) {
        case MLXSW_SP_L3_PROTO_IPV4:
                saddr4 = be32_to_cpu(lb_cf.saddr.addr4);
                mlxsw_reg_ritr_pack(ritr_pl, enable, MLXSW_REG_RITR_LOOPBACK_IF,
                                    rif->rif_index, rif->vr_id, rif->dev->mtu);
                mlxsw_reg_ritr_loopback_ipip4_pack(ritr_pl, lb_cf.lb_ipipt,
                            MLXSW_REG_RITR_LOOPBACK_IPIP_OPTIONS_GRE_KEY_PRESET,
                            ul_vr_id, ul_rif_id, saddr4, lb_cf.okey);
                break;

        case MLXSW_SP_L3_PROTO_IPV6:
                return -EAFNOSUPPORT;
        }

        return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ritr), ritr_pl);
}

static int mlxsw_sp_netdevice_ipip_ol_update_mtu(struct mlxsw_sp *mlxsw_sp,
                                                 struct net_device *ol_dev)
{
        struct mlxsw_sp_ipip_entry *ipip_entry;
        struct mlxsw_sp_rif_ipip_lb *lb_rif;
        int err = 0;

        ipip_entry = mlxsw_sp_ipip_entry_find_by_ol_dev(mlxsw_sp, ol_dev);
        if (ipip_entry) {
                lb_rif = ipip_entry->ol_lb;
                err = mlxsw_sp_rif_ipip_lb_op(lb_rif, lb_rif->ul_vr_id,
                                              lb_rif->ul_rif_id, true);
                if (err)
                        goto out;
                lb_rif->common.mtu = ol_dev->mtu;
        }

out:
        return err;
}

static void mlxsw_sp_netdevice_ipip_ol_up_event(struct mlxsw_sp *mlxsw_sp,
                                                struct net_device *ol_dev)
{
        struct mlxsw_sp_ipip_entry *ipip_entry;

        ipip_entry = mlxsw_sp_ipip_entry_find_by_ol_dev(mlxsw_sp, ol_dev);
        if (ipip_entry)
                mlxsw_sp_ipip_entry_ol_up_event(mlxsw_sp, ipip_entry);
}

static void
mlxsw_sp_ipip_entry_ol_down_event(struct mlxsw_sp *mlxsw_sp,
                                  struct mlxsw_sp_ipip_entry *ipip_entry)
{
        if (ipip_entry->decap_fib_entry)
                mlxsw_sp_ipip_entry_demote_decap(mlxsw_sp, ipip_entry);
}

static void mlxsw_sp_netdevice_ipip_ol_down_event(struct mlxsw_sp *mlxsw_sp,
                                                  struct net_device *ol_dev)
{
        struct mlxsw_sp_ipip_entry *ipip_entry;

        ipip_entry = mlxsw_sp_ipip_entry_find_by_ol_dev(mlxsw_sp, ol_dev);
        if (ipip_entry)
                mlxsw_sp_ipip_entry_ol_down_event(mlxsw_sp, ipip_entry);
}

static void mlxsw_sp_nexthop_rif_migrate(struct mlxsw_sp *mlxsw_sp,
                                         struct mlxsw_sp_rif *old_rif,
                                         struct mlxsw_sp_rif *new_rif);
static int
mlxsw_sp_ipip_entry_ol_lb_update(struct mlxsw_sp *mlxsw_sp,
                                 struct mlxsw_sp_ipip_entry *ipip_entry,
                                 bool keep_encap,
                                 struct netlink_ext_ack *extack)
{
        struct mlxsw_sp_rif_ipip_lb *old_lb_rif = ipip_entry->ol_lb;
        struct mlxsw_sp_rif_ipip_lb *new_lb_rif;

        new_lb_rif = mlxsw_sp_ipip_ol_ipip_lb_create(mlxsw_sp,
                                                     ipip_entry->ipipt,
                                                     ipip_entry->ol_dev,
                                                     extack);
        if (IS_ERR(new_lb_rif))
                return PTR_ERR(new_lb_rif);
        ipip_entry->ol_lb = new_lb_rif;

        if (keep_encap)
                mlxsw_sp_nexthop_rif_migrate(mlxsw_sp, &old_lb_rif->common,
                                             &new_lb_rif->common);

        mlxsw_sp_rif_destroy(&old_lb_rif->common);

        return 0;
}

static void mlxsw_sp_nexthop_rif_update(struct mlxsw_sp *mlxsw_sp,
                                        struct mlxsw_sp_rif *rif);

/**
 * __mlxsw_sp_ipip_entry_update_tunnel - Update offload related to IPIP entry.
 * @mlxsw_sp: mlxsw_sp.
 * @ipip_entry: IPIP entry.
 * @recreate_loopback: Recreates the associated loopback RIF.
 * @keep_encap: Updates next hops that use the tunnel netdevice. This is only
 *              relevant when recreate_loopback is true.
 * @update_nexthops: Updates next hops, keeping the current loopback RIF. This
 *                   is only relevant when recreate_loopback is false.
 * @extack: extack.
 *
 * Return: Non-zero value on failure.
 */
int __mlxsw_sp_ipip_entry_update_tunnel(struct mlxsw_sp *mlxsw_sp,
                                        struct mlxsw_sp_ipip_entry *ipip_entry,
                                        bool recreate_loopback,
                                        bool keep_encap,
                                        bool update_nexthops,
                                        struct netlink_ext_ack *extack)
{
        int err;

        /* RIFs can't be edited, so to update loopback, we need to destroy and
         * recreate it. That creates a window of opportunity where RALUE and
         * RATR registers end up referencing a RIF that's already gone. RATRs
         * are handled in mlxsw_sp_ipip_entry_ol_lb_update(), and to take care
         * of RALUE, demote the decap route back.
         */
        if (ipip_entry->decap_fib_entry)
                mlxsw_sp_ipip_entry_demote_decap(mlxsw_sp, ipip_entry);

        if (recreate_loopback) {
                err = mlxsw_sp_ipip_entry_ol_lb_update(mlxsw_sp, ipip_entry,
                                                       keep_encap, extack);
                if (err)
                        return err;
        } else if (update_nexthops) {
                mlxsw_sp_nexthop_rif_update(mlxsw_sp,
                                            &ipip_entry->ol_lb->common);
        }

        if (ipip_entry->ol_dev->flags & IFF_UP)
                mlxsw_sp_ipip_entry_ol_up_event(mlxsw_sp, ipip_entry);

        return 0;
}

static int mlxsw_sp_netdevice_ipip_ol_vrf_event(struct mlxsw_sp *mlxsw_sp,
                                                struct net_device *ol_dev,
                                                struct netlink_ext_ack *extack)
{
        struct mlxsw_sp_ipip_entry *ipip_entry =
                mlxsw_sp_ipip_entry_find_by_ol_dev(mlxsw_sp, ol_dev);

        if (!ipip_entry)
                return 0;

        return __mlxsw_sp_ipip_entry_update_tunnel(mlxsw_sp, ipip_entry,
                                                   true, false, false, extack);
}

static int
mlxsw_sp_netdevice_ipip_ul_vrf_event(struct mlxsw_sp *mlxsw_sp,
                                     struct mlxsw_sp_ipip_entry *ipip_entry,
                                     struct net_device *ul_dev,
                                     bool *demote_this,
                                     struct netlink_ext_ack *extack)
{
        u32 ul_tb_id = l3mdev_fib_table(ul_dev) ? : RT_TABLE_MAIN;
        enum mlxsw_sp_l3proto ul_proto;
        union mlxsw_sp_l3addr saddr;

        /* Moving underlay to a different VRF might cause local address
         * conflict, and the conflicting tunnels need to be demoted.
         */
        ul_proto = mlxsw_sp->router->ipip_ops_arr[ipip_entry->ipipt]->ul_proto;
        saddr = mlxsw_sp_ipip_netdev_saddr(ul_proto, ipip_entry->ol_dev);
        if (mlxsw_sp_ipip_demote_tunnel_by_saddr(mlxsw_sp, ul_proto,
                                                 saddr, ul_tb_id,
                                                 ipip_entry)) {
                *demote_this = true;
                return 0;
        }

        return __mlxsw_sp_ipip_entry_update_tunnel(mlxsw_sp, ipip_entry,
                                                   true, true, false, extack);
}

static int
mlxsw_sp_netdevice_ipip_ul_up_event(struct mlxsw_sp *mlxsw_sp,
                                    struct mlxsw_sp_ipip_entry *ipip_entry,
                                    struct net_device *ul_dev)
{
        return __mlxsw_sp_ipip_entry_update_tunnel(mlxsw_sp, ipip_entry,
                                                   false, false, true, NULL);
}

static int
mlxsw_sp_netdevice_ipip_ul_down_event(struct mlxsw_sp *mlxsw_sp,
                                      struct mlxsw_sp_ipip_entry *ipip_entry,
                                      struct net_device *ul_dev)
{
        /* A down underlay device causes encapsulated packets to not be
         * forwarded, but decap still works. So refresh next hops without
         * touching anything else.
         */
        return __mlxsw_sp_ipip_entry_update_tunnel(mlxsw_sp, ipip_entry,
                                                   false, false, true, NULL);
}

static int
mlxsw_sp_netdevice_ipip_ol_change_event(struct mlxsw_sp *mlxsw_sp,
                                        struct net_device *ol_dev,
                                        struct netlink_ext_ack *extack)
{
        const struct mlxsw_sp_ipip_ops *ipip_ops;
        struct mlxsw_sp_ipip_entry *ipip_entry;
        int err;

        ipip_entry = mlxsw_sp_ipip_entry_find_by_ol_dev(mlxsw_sp, ol_dev);
        if (!ipip_entry)
                /* A change might make a tunnel eligible for offloading, but
                 * that is currently not implemented. What falls to slow path
                 * stays there.
                 */
                return 0;

        /* A change might make a tunnel not eligible for offloading. */
        if (!mlxsw_sp_netdevice_ipip_can_offload(mlxsw_sp, ol_dev,
                                                 ipip_entry->ipipt)) {
                mlxsw_sp_ipip_entry_demote_tunnel(mlxsw_sp, ipip_entry);
                return 0;
        }

        ipip_ops = mlxsw_sp->router->ipip_ops_arr[ipip_entry->ipipt];
        err = ipip_ops->ol_netdev_change(mlxsw_sp, ipip_entry, extack);
        return err;
}

void mlxsw_sp_ipip_entry_demote_tunnel(struct mlxsw_sp *mlxsw_sp,
                                       struct mlxsw_sp_ipip_entry *ipip_entry)
{
        struct net_device *ol_dev = ipip_entry->ol_dev;

        if (ol_dev->flags & IFF_UP)
                mlxsw_sp_ipip_entry_ol_down_event(mlxsw_sp, ipip_entry);
        mlxsw_sp_ipip_entry_destroy(mlxsw_sp, ipip_entry);
}

/* The configuration where several tunnels have the same local address in the
 * same underlay table needs special treatment in the HW. That is currently not
 * implemented in the driver. This function finds and demotes the first tunnel
 * with a given source address, except the one passed in in the argument
 * `except'.
 */
bool
mlxsw_sp_ipip_demote_tunnel_by_saddr(struct mlxsw_sp *mlxsw_sp,
                                     enum mlxsw_sp_l3proto ul_proto,
                                     union mlxsw_sp_l3addr saddr,
                                     u32 ul_tb_id,
                                     const struct mlxsw_sp_ipip_entry *except)
{
        struct mlxsw_sp_ipip_entry *ipip_entry, *tmp;

        list_for_each_entry_safe(ipip_entry, tmp, &mlxsw_sp->router->ipip_list,
                                 ipip_list_node) {
                if (ipip_entry != except &&
                    mlxsw_sp_ipip_entry_saddr_matches(mlxsw_sp, ul_proto, saddr,
                                                      ul_tb_id, ipip_entry)) {
                        mlxsw_sp_ipip_entry_demote_tunnel(mlxsw_sp, ipip_entry);
                        return true;
                }
        }

        return false;
}

static void mlxsw_sp_ipip_demote_tunnel_by_ul_netdev(struct mlxsw_sp *mlxsw_sp,
                                                     struct net_device *ul_dev)
{
        struct mlxsw_sp_ipip_entry *ipip_entry, *tmp;

        list_for_each_entry_safe(ipip_entry, tmp, &mlxsw_sp->router->ipip_list,
                                 ipip_list_node) {
                struct net_device *ol_dev = ipip_entry->ol_dev;
                struct net_device *ipip_ul_dev;

                rcu_read_lock();
                ipip_ul_dev = __mlxsw_sp_ipip_netdev_ul_dev_get(ol_dev);
                rcu_read_unlock();
                if (ipip_ul_dev == ul_dev)
                        mlxsw_sp_ipip_entry_demote_tunnel(mlxsw_sp, ipip_entry);
        }
}

int mlxsw_sp_netdevice_ipip_ol_event(struct mlxsw_sp *mlxsw_sp,
                                     struct net_device *ol_dev,
                                     unsigned long event,
                                     struct netdev_notifier_info *info)
{
        struct netdev_notifier_changeupper_info *chup;
        struct netlink_ext_ack *extack;
        int err = 0;

        mutex_lock(&mlxsw_sp->router->lock);
        switch (event) {
        case NETDEV_REGISTER:
                err = mlxsw_sp_netdevice_ipip_ol_reg_event(mlxsw_sp, ol_dev);
                break;
        case NETDEV_UNREGISTER:
                mlxsw_sp_netdevice_ipip_ol_unreg_event(mlxsw_sp, ol_dev);
                break;
        case NETDEV_UP:
                mlxsw_sp_netdevice_ipip_ol_up_event(mlxsw_sp, ol_dev);
                break;
        case NETDEV_DOWN:
                mlxsw_sp_netdevice_ipip_ol_down_event(mlxsw_sp, ol_dev);
                break;
        case NETDEV_CHANGEUPPER:
                chup = container_of(info, typeof(*chup), info);
                extack = info->extack;
                if (netif_is_l3_master(chup->upper_dev))
                        err = mlxsw_sp_netdevice_ipip_ol_vrf_event(mlxsw_sp,
                                                                   ol_dev,
                                                                   extack);
                break;
        case NETDEV_CHANGE:
                extack = info->extack;
                err = mlxsw_sp_netdevice_ipip_ol_change_event(mlxsw_sp,
                                                              ol_dev, extack);
                break;
        case NETDEV_CHANGEMTU:
                err = mlxsw_sp_netdevice_ipip_ol_update_mtu(mlxsw_sp, ol_dev);
                break;
        }
        mutex_unlock(&mlxsw_sp->router->lock);
        return err;
}

static int
__mlxsw_sp_netdevice_ipip_ul_event(struct mlxsw_sp *mlxsw_sp,
                                   struct mlxsw_sp_ipip_entry *ipip_entry,
                                   struct net_device *ul_dev,
                                   bool *demote_this,
                                   unsigned long event,
                                   struct netdev_notifier_info *info)
{
        struct netdev_notifier_changeupper_info *chup;
        struct netlink_ext_ack *extack;

        switch (event) {
        case NETDEV_CHANGEUPPER:
                chup = container_of(info, typeof(*chup), info);
                extack = info->extack;
                if (netif_is_l3_master(chup->upper_dev))
                        return mlxsw_sp_netdevice_ipip_ul_vrf_event(mlxsw_sp,
                                                                    ipip_entry,
                                                                    ul_dev,
                                                                    demote_this,
                                                                    extack);
                break;

        case NETDEV_UP:
                return mlxsw_sp_netdevice_ipip_ul_up_event(mlxsw_sp, ipip_entry,
                                                           ul_dev);
        case NETDEV_DOWN:
                return mlxsw_sp_netdevice_ipip_ul_down_event(mlxsw_sp,
                                                             ipip_entry,
                                                             ul_dev);
        }
        return 0;
}

int
mlxsw_sp_netdevice_ipip_ul_event(struct mlxsw_sp *mlxsw_sp,
                                 struct net_device *ul_dev,
                                 unsigned long event,
                                 struct netdev_notifier_info *info)
{
        struct mlxsw_sp_ipip_entry *ipip_entry = NULL;
        int err = 0;

        mutex_lock(&mlxsw_sp->router->lock);
        while ((ipip_entry = mlxsw_sp_ipip_entry_find_by_ul_dev(mlxsw_sp,
                                                                ul_dev,
                                                                ipip_entry))) {
                struct mlxsw_sp_ipip_entry *prev;
                bool demote_this = false;

                err = __mlxsw_sp_netdevice_ipip_ul_event(mlxsw_sp, ipip_entry,
                                                         ul_dev, &demote_this,
                                                         event, info);
                if (err) {
                        mlxsw_sp_ipip_demote_tunnel_by_ul_netdev(mlxsw_sp,
                                                                 ul_dev);
                        break;
                }

                if (demote_this) {
                        if (list_is_first(&ipip_entry->ipip_list_node,
                                          &mlxsw_sp->router->ipip_list))
                                prev = NULL;
                        else
                                /* This can't be cached from previous iteration,
                                 * because that entry could be gone now.
                                 */
                                prev = list_prev_entry(ipip_entry,
                                                       ipip_list_node);
                        mlxsw_sp_ipip_entry_demote_tunnel(mlxsw_sp, ipip_entry);
                        ipip_entry = prev;
                }
        }
        mutex_unlock(&mlxsw_sp->router->lock);

        return err;
}

int mlxsw_sp_router_nve_promote_decap(struct mlxsw_sp *mlxsw_sp, u32 ul_tb_id,
                                      enum mlxsw_sp_l3proto ul_proto,
                                      const union mlxsw_sp_l3addr *ul_sip,
                                      u32 tunnel_index)
{
        enum mlxsw_sp_fib_entry_type type = MLXSW_SP_FIB_ENTRY_TYPE_TRAP;
        struct mlxsw_sp_router *router = mlxsw_sp->router;
        struct mlxsw_sp_fib_entry *fib_entry;
        int err = 0;

        mutex_lock(&mlxsw_sp->router->lock);

        if (WARN_ON_ONCE(router->nve_decap_config.valid)) {
                err = -EINVAL;
                goto out;
        }

        router->nve_decap_config.ul_tb_id = ul_tb_id;
        router->nve_decap_config.tunnel_index = tunnel_index;
        router->nve_decap_config.ul_proto = ul_proto;
        router->nve_decap_config.ul_sip = *ul_sip;
        router->nve_decap_config.valid = true;

        /* It is valid to create a tunnel with a local IP and only later
         * assign this IP address to a local interface
         */
        fib_entry = mlxsw_sp_router_ip2me_fib_entry_find(mlxsw_sp, ul_tb_id,
                                                         ul_proto, ul_sip,
                                                         type);
        if (!fib_entry)
                goto out;

        fib_entry->decap.tunnel_index = tunnel_index;
        fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_NVE_DECAP;

        err = mlxsw_sp_fib_entry_update(mlxsw_sp, fib_entry);
        if (err)
                goto err_fib_entry_update;

        goto out;

err_fib_entry_update:

        fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_TRAP;
        mlxsw_sp_fib_entry_update(mlxsw_sp, fib_entry);
out:
        mutex_unlock(&mlxsw_sp->router->lock);
        return err;
}

void mlxsw_sp_router_nve_demote_decap(struct mlxsw_sp *mlxsw_sp, u32 ul_tb_id,
                                      enum mlxsw_sp_l3proto ul_proto,
                                      const union mlxsw_sp_l3addr *ul_sip)
{
        enum mlxsw_sp_fib_entry_type type = MLXSW_SP_FIB_ENTRY_TYPE_NVE_DECAP;
        struct mlxsw_sp_router *router = mlxsw_sp->router;
        struct mlxsw_sp_fib_entry *fib_entry;

        mutex_lock(&mlxsw_sp->router->lock);

        if (WARN_ON_ONCE(!router->nve_decap_config.valid))
                goto out;

        router->nve_decap_config.valid = false;

        fib_entry = mlxsw_sp_router_ip2me_fib_entry_find(mlxsw_sp, ul_tb_id,
                                                         ul_proto, ul_sip,
                                                         type);
        if (!fib_entry)
                goto out;

        fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_TRAP;
        mlxsw_sp_fib_entry_update(mlxsw_sp, fib_entry);
out:
        mutex_unlock(&mlxsw_sp->router->lock);
}

static bool mlxsw_sp_router_nve_is_decap(struct mlxsw_sp *mlxsw_sp,
                                         u32 ul_tb_id,
                                         enum mlxsw_sp_l3proto ul_proto,
                                         const union mlxsw_sp_l3addr *ul_sip)
{
        struct mlxsw_sp_router *router = mlxsw_sp->router;

        return router->nve_decap_config.valid &&
               router->nve_decap_config.ul_tb_id == ul_tb_id &&
               router->nve_decap_config.ul_proto == ul_proto &&
               !memcmp(&router->nve_decap_config.ul_sip, ul_sip,
                       sizeof(*ul_sip));
}

struct mlxsw_sp_neigh_key {
        struct neighbour *n;
};

struct mlxsw_sp_neigh_entry {
        struct list_head rif_list_node;
        struct rhash_head ht_node;
        struct mlxsw_sp_neigh_key key;
        u16 rif;
        bool connected;
        unsigned char ha[ETH_ALEN];
        struct list_head nexthop_list; /* list of nexthops using
                                        * this neigh entry
                                        */
        struct list_head nexthop_neighs_list_node;
        unsigned int counter_index;
        bool counter_valid;
};

static const struct rhashtable_params mlxsw_sp_neigh_ht_params = {
        .key_offset = offsetof(struct mlxsw_sp_neigh_entry, key),
        .head_offset = offsetof(struct mlxsw_sp_neigh_entry, ht_node),
        .key_len = sizeof(struct mlxsw_sp_neigh_key),
};

struct mlxsw_sp_neigh_entry *
mlxsw_sp_rif_neigh_next(struct mlxsw_sp_rif *rif,
                        struct mlxsw_sp_neigh_entry *neigh_entry)
{
        if (!neigh_entry) {
                if (list_empty(&rif->neigh_list))
                        return NULL;
                else
                        return list_first_entry(&rif->neigh_list,
                                                typeof(*neigh_entry),
                                                rif_list_node);
        }
        if (list_is_last(&neigh_entry->rif_list_node, &rif->neigh_list))
                return NULL;
        return list_next_entry(neigh_entry, rif_list_node);
}

int mlxsw_sp_neigh_entry_type(struct mlxsw_sp_neigh_entry *neigh_entry)
{
        return neigh_entry->key.n->tbl->family;
}

unsigned char *
mlxsw_sp_neigh_entry_ha(struct mlxsw_sp_neigh_entry *neigh_entry)
{
        return neigh_entry->ha;
}

u32 mlxsw_sp_neigh4_entry_dip(struct mlxsw_sp_neigh_entry *neigh_entry)
{
        struct neighbour *n;

        n = neigh_entry->key.n;
        return ntohl(*((__be32 *) n->primary_key));
}

struct in6_addr *
mlxsw_sp_neigh6_entry_dip(struct mlxsw_sp_neigh_entry *neigh_entry)
{
        struct neighbour *n;

        n = neigh_entry->key.n;
        return (struct in6_addr *) &n->primary_key;
}

int mlxsw_sp_neigh_counter_get(struct mlxsw_sp *mlxsw_sp,
                               struct mlxsw_sp_neigh_entry *neigh_entry,
                               u64 *p_counter)
{
        if (!neigh_entry->counter_valid)
                return -EINVAL;

        return mlxsw_sp_flow_counter_get(mlxsw_sp, neigh_entry->counter_index,
                                         p_counter, NULL);
}

static struct mlxsw_sp_neigh_entry *
mlxsw_sp_neigh_entry_alloc(struct mlxsw_sp *mlxsw_sp, struct neighbour *n,
                           u16 rif)
{
        struct mlxsw_sp_neigh_entry *neigh_entry;

        neigh_entry = kzalloc(sizeof(*neigh_entry), GFP_KERNEL);
        if (!neigh_entry)
                return NULL;

        neigh_entry->key.n = n;
        neigh_entry->rif = rif;
        INIT_LIST_HEAD(&neigh_entry->nexthop_list);

        return neigh_entry;
}

static void mlxsw_sp_neigh_entry_free(struct mlxsw_sp_neigh_entry *neigh_entry)
{
        kfree(neigh_entry);
}

static int
mlxsw_sp_neigh_entry_insert(struct mlxsw_sp *mlxsw_sp,
                            struct mlxsw_sp_neigh_entry *neigh_entry)
{
        return rhashtable_insert_fast(&mlxsw_sp->router->neigh_ht,
                                      &neigh_entry->ht_node,
                                      mlxsw_sp_neigh_ht_params);
}

static void
mlxsw_sp_neigh_entry_remove(struct mlxsw_sp *mlxsw_sp,
                            struct mlxsw_sp_neigh_entry *neigh_entry)
{
        rhashtable_remove_fast(&mlxsw_sp->router->neigh_ht,
                               &neigh_entry->ht_node,
                               mlxsw_sp_neigh_ht_params);
}

static bool
mlxsw_sp_neigh_counter_should_alloc(struct mlxsw_sp *mlxsw_sp,
                                    struct mlxsw_sp_neigh_entry *neigh_entry)
{
        struct devlink *devlink;
        const char *table_name;

        switch (mlxsw_sp_neigh_entry_type(neigh_entry)) {
        case AF_INET:
                table_name = MLXSW_SP_DPIPE_TABLE_NAME_HOST4;
                break;
        case AF_INET6:
                table_name = MLXSW_SP_DPIPE_TABLE_NAME_HOST6;
                break;
        default:
                WARN_ON(1);
                return false;
        }

        devlink = priv_to_devlink(mlxsw_sp->core);
        return devlink_dpipe_table_counter_enabled(devlink, table_name);
}

static void
mlxsw_sp_neigh_counter_alloc(struct mlxsw_sp *mlxsw_sp,
                             struct mlxsw_sp_neigh_entry *neigh_entry)
{
        if (!mlxsw_sp_neigh_counter_should_alloc(mlxsw_sp, neigh_entry))
                return;

        if (mlxsw_sp_flow_counter_alloc(mlxsw_sp, &neigh_entry->counter_index))
                return;

        neigh_entry->counter_valid = true;
}

static void
mlxsw_sp_neigh_counter_free(struct mlxsw_sp *mlxsw_sp,
                            struct mlxsw_sp_neigh_entry *neigh_entry)
{
        if (!neigh_entry->counter_valid)
                return;
        mlxsw_sp_flow_counter_free(mlxsw_sp,
                                   neigh_entry->counter_index);
        neigh_entry->counter_valid = false;
}

static struct mlxsw_sp_neigh_entry *
mlxsw_sp_neigh_entry_create(struct mlxsw_sp *mlxsw_sp, struct neighbour *n)
{
        struct mlxsw_sp_neigh_entry *neigh_entry;
        struct mlxsw_sp_rif *rif;
        int err;

        rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, n->dev);
        if (!rif)
                return ERR_PTR(-EINVAL);

        neigh_entry = mlxsw_sp_neigh_entry_alloc(mlxsw_sp, n, rif->rif_index);
        if (!neigh_entry)
                return ERR_PTR(-ENOMEM);

        err = mlxsw_sp_neigh_entry_insert(mlxsw_sp, neigh_entry);
        if (err)
                goto err_neigh_entry_insert;

        mlxsw_sp_neigh_counter_alloc(mlxsw_sp, neigh_entry);
        list_add(&neigh_entry->rif_list_node, &rif->neigh_list);

        return neigh_entry;

err_neigh_entry_insert:
        mlxsw_sp_neigh_entry_free(neigh_entry);
        return ERR_PTR(err);
}

static void
mlxsw_sp_neigh_entry_destroy(struct mlxsw_sp *mlxsw_sp,
                             struct mlxsw_sp_neigh_entry *neigh_entry)
{
        list_del(&neigh_entry->rif_list_node);
        mlxsw_sp_neigh_counter_free(mlxsw_sp, neigh_entry);
        mlxsw_sp_neigh_entry_remove(mlxsw_sp, neigh_entry);
        mlxsw_sp_neigh_entry_free(neigh_entry);
}

static struct mlxsw_sp_neigh_entry *
mlxsw_sp_neigh_entry_lookup(struct mlxsw_sp *mlxsw_sp, struct neighbour *n)
{
        struct mlxsw_sp_neigh_key key;

        key.n = n;
        return rhashtable_lookup_fast(&mlxsw_sp->router->neigh_ht,
                                      &key, mlxsw_sp_neigh_ht_params);
}

static void
mlxsw_sp_router_neighs_update_interval_init(struct mlxsw_sp *mlxsw_sp)
{
        unsigned long interval;

#if IS_ENABLED(CONFIG_IPV6)
        interval = min_t(unsigned long,
                         NEIGH_VAR(&arp_tbl.parms, DELAY_PROBE_TIME),
                         NEIGH_VAR(&nd_tbl.parms, DELAY_PROBE_TIME));
#else
        interval = NEIGH_VAR(&arp_tbl.parms, DELAY_PROBE_TIME);
#endif
        mlxsw_sp->router->neighs_update.interval = jiffies_to_msecs(interval);
}

static void mlxsw_sp_router_neigh_ent_ipv4_process(struct mlxsw_sp *mlxsw_sp,
                                                   char *rauhtd_pl,
                                                   int ent_index)
{
        u64 max_rifs = MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_RIFS);
        struct net_device *dev;
        struct neighbour *n;
        __be32 dipn;
        u32 dip;
        u16 rif;

        mlxsw_reg_rauhtd_ent_ipv4_unpack(rauhtd_pl, ent_index, &rif, &dip);

        if (WARN_ON_ONCE(rif >= max_rifs))
                return;
        if (!mlxsw_sp->router->rifs[rif]) {
                dev_err_ratelimited(mlxsw_sp->bus_info->dev, "Incorrect RIF in neighbour entry\n");
                return;
        }

        dipn = htonl(dip);
        dev = mlxsw_sp->router->rifs[rif]->dev;
        n = neigh_lookup(&arp_tbl, &dipn, dev);
        if (!n)
                return;

        netdev_dbg(dev, "Updating neighbour with IP=%pI4h\n", &dip);
        neigh_event_send(n, NULL);
        neigh_release(n);
}

#if IS_ENABLED(CONFIG_IPV6)
static void mlxsw_sp_router_neigh_ent_ipv6_process(struct mlxsw_sp *mlxsw_sp,
                                                   char *rauhtd_pl,
                                                   int rec_index)
{
        struct net_device *dev;
        struct neighbour *n;
        struct in6_addr dip;
        u16 rif;

        mlxsw_reg_rauhtd_ent_ipv6_unpack(rauhtd_pl, rec_index, &rif,
                                         (char *) &dip);

        if (!mlxsw_sp->router->rifs[rif]) {
                dev_err_ratelimited(mlxsw_sp->bus_info->dev, "Incorrect RIF in neighbour entry\n");
                return;
        }

        dev = mlxsw_sp->router->rifs[rif]->dev;
        n = neigh_lookup(&nd_tbl, &dip, dev);
        if (!n)
                return;

        netdev_dbg(dev, "Updating neighbour with IP=%pI6c\n", &dip);
        neigh_event_send(n, NULL);
        neigh_release(n);
}
#else
static void mlxsw_sp_router_neigh_ent_ipv6_process(struct mlxsw_sp *mlxsw_sp,
                                                   char *rauhtd_pl,
                                                   int rec_index)
{
}
#endif

static void mlxsw_sp_router_neigh_rec_ipv4_process(struct mlxsw_sp *mlxsw_sp,
                                                   char *rauhtd_pl,
                                                   int rec_index)
{
        u8 num_entries;
        int i;

        num_entries = mlxsw_reg_rauhtd_ipv4_rec_num_entries_get(rauhtd_pl,
                                                                rec_index);
        /* Hardware starts counting at 0, so add 1. */
        num_entries++;

        /* Each record consists of several neighbour entries. */
        for (i = 0; i < num_entries; i++) {
                int ent_index;

                ent_index = rec_index * MLXSW_REG_RAUHTD_IPV4_ENT_PER_REC + i;
                mlxsw_sp_router_neigh_ent_ipv4_process(mlxsw_sp, rauhtd_pl,
                                                       ent_index);
        }

}

static void mlxsw_sp_router_neigh_rec_ipv6_process(struct mlxsw_sp *mlxsw_sp,
                                                   char *rauhtd_pl,
                                                   int rec_index)
{
        /* One record contains one entry. */
        mlxsw_sp_router_neigh_ent_ipv6_process(mlxsw_sp, rauhtd_pl,
                                               rec_index);
}

static void mlxsw_sp_router_neigh_rec_process(struct mlxsw_sp *mlxsw_sp,
                                              char *rauhtd_pl, int rec_index)
{
        switch (mlxsw_reg_rauhtd_rec_type_get(rauhtd_pl, rec_index)) {
        case MLXSW_REG_RAUHTD_TYPE_IPV4:
                mlxsw_sp_router_neigh_rec_ipv4_process(mlxsw_sp, rauhtd_pl,
                                                       rec_index);
                break;
        case MLXSW_REG_RAUHTD_TYPE_IPV6:
                mlxsw_sp_router_neigh_rec_ipv6_process(mlxsw_sp, rauhtd_pl,
                                                       rec_index);
                break;
        }
}

static bool mlxsw_sp_router_rauhtd_is_full(char *rauhtd_pl)
{
        u8 num_rec, last_rec_index, num_entries;

        num_rec = mlxsw_reg_rauhtd_num_rec_get(rauhtd_pl);
        last_rec_index = num_rec - 1;

        if (num_rec < MLXSW_REG_RAUHTD_REC_MAX_NUM)
                return false;
        if (mlxsw_reg_rauhtd_rec_type_get(rauhtd_pl, last_rec_index) ==
            MLXSW_REG_RAUHTD_TYPE_IPV6)
                return true;

        num_entries = mlxsw_reg_rauhtd_ipv4_rec_num_entries_get(rauhtd_pl,
                                                                last_rec_index);
        if (++num_entries == MLXSW_REG_RAUHTD_IPV4_ENT_PER_REC)
                return true;
        return false;
}

static int
__mlxsw_sp_router_neighs_update_rauhtd(struct mlxsw_sp *mlxsw_sp,
                                       char *rauhtd_pl,
                                       enum mlxsw_reg_rauhtd_type type)
{
        int i, num_rec;
        int err;

        /* Ensure the RIF we read from the device does not change mid-dump. */
        mutex_lock(&mlxsw_sp->router->lock);
        do {
                mlxsw_reg_rauhtd_pack(rauhtd_pl, type);
                err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(rauhtd),
                                      rauhtd_pl);
                if (err) {
                        dev_err_ratelimited(mlxsw_sp->bus_info->dev, "Failed to dump neighbour table\n");
                        break;
                }
                num_rec = mlxsw_reg_rauhtd_num_rec_get(rauhtd_pl);
                for (i = 0; i < num_rec; i++)
                        mlxsw_sp_router_neigh_rec_process(mlxsw_sp, rauhtd_pl,
                                                          i);
        } while (mlxsw_sp_router_rauhtd_is_full(rauhtd_pl));
        mutex_unlock(&mlxsw_sp->router->lock);

        return err;
}

static int mlxsw_sp_router_neighs_update_rauhtd(struct mlxsw_sp *mlxsw_sp)
{
        enum mlxsw_reg_rauhtd_type type;
        char *rauhtd_pl;
        int err;

        rauhtd_pl = kmalloc(MLXSW_REG_RAUHTD_LEN, GFP_KERNEL);
        if (!rauhtd_pl)
                return -ENOMEM;

        type = MLXSW_REG_RAUHTD_TYPE_IPV4;
        err = __mlxsw_sp_router_neighs_update_rauhtd(mlxsw_sp, rauhtd_pl, type);
        if (err)
                goto out;

        type = MLXSW_REG_RAUHTD_TYPE_IPV6;
        err = __mlxsw_sp_router_neighs_update_rauhtd(mlxsw_sp, rauhtd_pl, type);
out:
        kfree(rauhtd_pl);
        return err;
}

static void mlxsw_sp_router_neighs_update_nh(struct mlxsw_sp *mlxsw_sp)
{
        struct mlxsw_sp_neigh_entry *neigh_entry;

        mutex_lock(&mlxsw_sp->router->lock);
        list_for_each_entry(neigh_entry, &mlxsw_sp->router->nexthop_neighs_list,
                            nexthop_neighs_list_node)
                /* If this neigh have nexthops, make the kernel think this neigh
                 * is active regardless of the traffic.
                 */
                neigh_event_send(neigh_entry->key.n, NULL);
        mutex_unlock(&mlxsw_sp->router->lock);
}

static void
mlxsw_sp_router_neighs_update_work_schedule(struct mlxsw_sp *mlxsw_sp)
{
        unsigned long interval = mlxsw_sp->router->neighs_update.interval;

        mlxsw_core_schedule_dw(&mlxsw_sp->router->neighs_update.dw,
                               msecs_to_jiffies(interval));
}

static void mlxsw_sp_router_neighs_update_work(struct work_struct *work)
{
        struct mlxsw_sp_router *router;
        int err;

        router = container_of(work, struct mlxsw_sp_router,
                              neighs_update.dw.work);
        err = mlxsw_sp_router_neighs_update_rauhtd(router->mlxsw_sp);
        if (err)
                dev_err(router->mlxsw_sp->bus_info->dev, "Could not update kernel for neigh activity");

        mlxsw_sp_router_neighs_update_nh(router->mlxsw_sp);

        mlxsw_sp_router_neighs_update_work_schedule(router->mlxsw_sp);
}

static void mlxsw_sp_router_probe_unresolved_nexthops(struct work_struct *work)
{
        struct mlxsw_sp_neigh_entry *neigh_entry;
        struct mlxsw_sp_router *router;

        router = container_of(work, struct mlxsw_sp_router,
                              nexthop_probe_dw.work);
        /* Iterate over nexthop neighbours, find those who are unresolved and
         * send arp on them. This solves the chicken-egg problem when
         * the nexthop wouldn't get offloaded until the neighbor is resolved
         * but it wouldn't get resolved ever in case traffic is flowing in HW
         * using different nexthop.
         */
        mutex_lock(&router->lock);
        list_for_each_entry(neigh_entry, &router->nexthop_neighs_list,
                            nexthop_neighs_list_node)
                if (!neigh_entry->connected)
                        neigh_event_send(neigh_entry->key.n, NULL);
        mutex_unlock(&router->lock);

        mlxsw_core_schedule_dw(&router->nexthop_probe_dw,
                               MLXSW_SP_UNRESOLVED_NH_PROBE_INTERVAL);
}

static void
mlxsw_sp_nexthop_neigh_update(struct mlxsw_sp *mlxsw_sp,
                              struct mlxsw_sp_neigh_entry *neigh_entry,
                              bool removing, bool dead);

static enum mlxsw_reg_rauht_op mlxsw_sp_rauht_op(bool adding)
{
        return adding ? MLXSW_REG_RAUHT_OP_WRITE_ADD :
                        MLXSW_REG_RAUHT_OP_WRITE_DELETE;
}

static int
mlxsw_sp_router_neigh_entry_op4(struct mlxsw_sp *mlxsw_sp,
                                struct mlxsw_sp_neigh_entry *neigh_entry,
                                enum mlxsw_reg_rauht_op op)
{
        struct neighbour *n = neigh_entry->key.n;
        u32 dip = ntohl(*((__be32 *) n->primary_key));
        char rauht_pl[MLXSW_REG_RAUHT_LEN];

        mlxsw_reg_rauht_pack4(rauht_pl, op, neigh_entry->rif, neigh_entry->ha,
                              dip);
        if (neigh_entry->counter_valid)
                mlxsw_reg_rauht_pack_counter(rauht_pl,
                                             neigh_entry->counter_index);
        return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rauht), rauht_pl);
}

static int
mlxsw_sp_router_neigh_entry_op6(struct mlxsw_sp *mlxsw_sp,
                                struct mlxsw_sp_neigh_entry *neigh_entry,
                                enum mlxsw_reg_rauht_op op)
{
        struct neighbour *n = neigh_entry->key.n;
        char rauht_pl[MLXSW_REG_RAUHT_LEN];
        const char *dip = n->primary_key;

        mlxsw_reg_rauht_pack6(rauht_pl, op, neigh_entry->rif, neigh_entry->ha,
                              dip);
        if (neigh_entry->counter_valid)
                mlxsw_reg_rauht_pack_counter(rauht_pl,
                                             neigh_entry->counter_index);
        return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rauht), rauht_pl);
}

bool mlxsw_sp_neigh_ipv6_ignore(struct mlxsw_sp_neigh_entry *neigh_entry)
{
        struct neighbour *n = neigh_entry->key.n;

        /* Packets with a link-local destination address are trapped
         * after LPM lookup and never reach the neighbour table, so
         * there is no need to program such neighbours to the device.
         */
        if (ipv6_addr_type((struct in6_addr *) &n->primary_key) &
            IPV6_ADDR_LINKLOCAL)
                return true;
        return false;
}

static void
mlxsw_sp_neigh_entry_update(struct mlxsw_sp *mlxsw_sp,
                            struct mlxsw_sp_neigh_entry *neigh_entry,
                            bool adding)
{
        enum mlxsw_reg_rauht_op op = mlxsw_sp_rauht_op(adding);
        int err;

        if (!adding && !neigh_entry->connected)
                return;
        neigh_entry->connected = adding;
        if (neigh_entry->key.n->tbl->family == AF_INET) {
                err = mlxsw_sp_router_neigh_entry_op4(mlxsw_sp, neigh_entry,
                                                      op);
                if (err)
                        return;
        } else if (neigh_entry->key.n->tbl->family == AF_INET6) {
                if (mlxsw_sp_neigh_ipv6_ignore(neigh_entry))
                        return;
                err = mlxsw_sp_router_neigh_entry_op6(mlxsw_sp, neigh_entry,
                                                      op);
                if (err)
                        return;
        } else {
                WARN_ON_ONCE(1);
                return;
        }

        if (adding)
                neigh_entry->key.n->flags |= NTF_OFFLOADED;
        else
                neigh_entry->key.n->flags &= ~NTF_OFFLOADED;
}

void
mlxsw_sp_neigh_entry_counter_update(struct mlxsw_sp *mlxsw_sp,
                                    struct mlxsw_sp_neigh_entry *neigh_entry,
                                    bool adding)
{
        if (adding)
                mlxsw_sp_neigh_counter_alloc(mlxsw_sp, neigh_entry);
        else
                mlxsw_sp_neigh_counter_free(mlxsw_sp, neigh_entry);
        mlxsw_sp_neigh_entry_update(mlxsw_sp, neigh_entry, true);
}

struct mlxsw_sp_netevent_work {
        struct work_struct work;
        struct mlxsw_sp *mlxsw_sp;
        struct neighbour *n;
};

static void mlxsw_sp_router_neigh_event_work(struct work_struct *work)
{
        struct mlxsw_sp_netevent_work *net_work =
                container_of(work, struct mlxsw_sp_netevent_work, work);
        struct mlxsw_sp *mlxsw_sp = net_work->mlxsw_sp;
        struct mlxsw_sp_neigh_entry *neigh_entry;
        struct neighbour *n = net_work->n;
        unsigned char ha[ETH_ALEN];
        bool entry_connected;
        u8 nud_state, dead;

        /* If these parameters are changed after we release the lock,
         * then we are guaranteed to receive another event letting us
         * know about it.
         */
        read_lock_bh(&n->lock);
        memcpy(ha, n->ha, ETH_ALEN);
        nud_state = n->nud_state;
        dead = n->dead;
        read_unlock_bh(&n->lock);

        mutex_lock(&mlxsw_sp->router->lock);
        mlxsw_sp_span_respin(mlxsw_sp);

        entry_connected = nud_state & NUD_VALID && !dead;
        neigh_entry = mlxsw_sp_neigh_entry_lookup(mlxsw_sp, n);
        if (!entry_connected && !neigh_entry)
                goto out;
        if (!neigh_entry) {
                neigh_entry = mlxsw_sp_neigh_entry_create(mlxsw_sp, n);
                if (IS_ERR(neigh_entry))
                        goto out;
        }

        if (neigh_entry->connected && entry_connected &&
            !memcmp(neigh_entry->ha, ha, ETH_ALEN))
                goto out;

        memcpy(neigh_entry->ha, ha, ETH_ALEN);
        mlxsw_sp_neigh_entry_update(mlxsw_sp, neigh_entry, entry_connected);
        mlxsw_sp_nexthop_neigh_update(mlxsw_sp, neigh_entry, !entry_connected,
                                      dead);

        if (!neigh_entry->connected && list_empty(&neigh_entry->nexthop_list))
                mlxsw_sp_neigh_entry_destroy(mlxsw_sp, neigh_entry);

out:
        mutex_unlock(&mlxsw_sp->router->lock);
        neigh_release(n);
        kfree(net_work);
}

static int mlxsw_sp_mp_hash_init(struct mlxsw_sp *mlxsw_sp);

static void mlxsw_sp_router_mp_hash_event_work(struct work_struct *work)
{
        struct mlxsw_sp_netevent_work *net_work =
                container_of(work, struct mlxsw_sp_netevent_work, work);
        struct mlxsw_sp *mlxsw_sp = net_work->mlxsw_sp;

        mlxsw_sp_mp_hash_init(mlxsw_sp);
        kfree(net_work);
}

static int __mlxsw_sp_router_init(struct mlxsw_sp *mlxsw_sp);

static void mlxsw_sp_router_update_priority_work(struct work_struct *work)
{
        struct mlxsw_sp_netevent_work *net_work =
                container_of(work, struct mlxsw_sp_netevent_work, work);
        struct mlxsw_sp *mlxsw_sp = net_work->mlxsw_sp;

        __mlxsw_sp_router_init(mlxsw_sp);
        kfree(net_work);
}

static int mlxsw_sp_router_schedule_work(struct net *net,
                                         struct notifier_block *nb,
                                         void (*cb)(struct work_struct *))
{
        struct mlxsw_sp_netevent_work *net_work;
        struct mlxsw_sp_router *router;

        router = container_of(nb, struct mlxsw_sp_router, netevent_nb);
        if (!net_eq(net, mlxsw_sp_net(router->mlxsw_sp)))
                return NOTIFY_DONE;

        net_work = kzalloc(sizeof(*net_work), GFP_ATOMIC);
        if (!net_work)
                return NOTIFY_BAD;

        INIT_WORK(&net_work->work, cb);
        net_work->mlxsw_sp = router->mlxsw_sp;
        mlxsw_core_schedule_work(&net_work->work);
        return NOTIFY_DONE;
}

static int mlxsw_sp_router_netevent_event(struct notifier_block *nb,
                                          unsigned long event, void *ptr)
{
        struct mlxsw_sp_netevent_work *net_work;
        struct mlxsw_sp_port *mlxsw_sp_port;
        struct mlxsw_sp *mlxsw_sp;
        unsigned long interval;
        struct neigh_parms *p;
        struct neighbour *n;

        switch (event) {
        case NETEVENT_DELAY_PROBE_TIME_UPDATE:
                p = ptr;

                /* We don't care about changes in the default table. */
                if (!p->dev || (p->tbl->family != AF_INET &&
                                p->tbl->family != AF_INET6))
                        return NOTIFY_DONE;

                /* We are in atomic context and can't take RTNL mutex,
                 * so use RCU variant to walk the device chain.
                 */
                mlxsw_sp_port = mlxsw_sp_port_lower_dev_hold(p->dev);
                if (!mlxsw_sp_port)
                        return NOTIFY_DONE;

                mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
                interval = jiffies_to_msecs(NEIGH_VAR(p, DELAY_PROBE_TIME));
                mlxsw_sp->router->neighs_update.interval = interval;

                mlxsw_sp_port_dev_put(mlxsw_sp_port);
                break;
        case NETEVENT_NEIGH_UPDATE:
                n = ptr;

                if (n->tbl->family != AF_INET && n->tbl->family != AF_INET6)
                        return NOTIFY_DONE;

                mlxsw_sp_port = mlxsw_sp_port_lower_dev_hold(n->dev);
                if (!mlxsw_sp_port)
                        return NOTIFY_DONE;

                net_work = kzalloc(sizeof(*net_work), GFP_ATOMIC);
                if (!net_work) {
                        mlxsw_sp_port_dev_put(mlxsw_sp_port);
                        return NOTIFY_BAD;
                }

                INIT_WORK(&net_work->work, mlxsw_sp_router_neigh_event_work);
                net_work->mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
                net_work->n = n;

                /* Take a reference to ensure the neighbour won't be
                 * destructed until we drop the reference in delayed
                 * work.
                 */
                neigh_clone(n);
                mlxsw_core_schedule_work(&net_work->work);
                mlxsw_sp_port_dev_put(mlxsw_sp_port);
                break;
        case NETEVENT_IPV4_MPATH_HASH_UPDATE:
        case NETEVENT_IPV6_MPATH_HASH_UPDATE:
                return mlxsw_sp_router_schedule_work(ptr, nb,
                                mlxsw_sp_router_mp_hash_event_work);

        case NETEVENT_IPV4_FWD_UPDATE_PRIORITY_UPDATE:
                return mlxsw_sp_router_schedule_work(ptr, nb,
                                mlxsw_sp_router_update_priority_work);
        }

        return NOTIFY_DONE;
}

static int mlxsw_sp_neigh_init(struct mlxsw_sp *mlxsw_sp)
{
        int err;

        err = rhashtable_init(&mlxsw_sp->router->neigh_ht,
                              &mlxsw_sp_neigh_ht_params);
        if (err)
                return err;

        /* Initialize the polling interval according to the default
         * table.
         */
        mlxsw_sp_router_neighs_update_interval_init(mlxsw_sp);

        /* Create the delayed works for the activity_update */
        INIT_DELAYED_WORK(&mlxsw_sp->router->neighs_update.dw,
                          mlxsw_sp_router_neighs_update_work);
        INIT_DELAYED_WORK(&mlxsw_sp->router->nexthop_probe_dw,
                          mlxsw_sp_router_probe_unresolved_nexthops);
        mlxsw_core_schedule_dw(&mlxsw_sp->router->neighs_update.dw, 0);
        mlxsw_core_schedule_dw(&mlxsw_sp->router->nexthop_probe_dw, 0);
        return 0;
}

static void mlxsw_sp_neigh_fini(struct mlxsw_sp *mlxsw_sp)
{
        cancel_delayed_work_sync(&mlxsw_sp->router->neighs_update.dw);
        cancel_delayed_work_sync(&mlxsw_sp->router->nexthop_probe_dw);
        rhashtable_destroy(&mlxsw_sp->router->neigh_ht);
}

static void mlxsw_sp_neigh_rif_gone_sync(struct mlxsw_sp *mlxsw_sp,
                                         struct mlxsw_sp_rif *rif)
{
        struct mlxsw_sp_neigh_entry *neigh_entry, *tmp;

        list_for_each_entry_safe(neigh_entry, tmp, &rif->neigh_list,
                                 rif_list_node) {
                mlxsw_sp_neigh_entry_update(mlxsw_sp, neigh_entry, false);
                mlxsw_sp_neigh_entry_destroy(mlxsw_sp, neigh_entry);
        }
}

enum mlxsw_sp_nexthop_type {
        MLXSW_SP_NEXTHOP_TYPE_ETH,
        MLXSW_SP_NEXTHOP_TYPE_IPIP,
};

enum mlxsw_sp_nexthop_action {
        /* Nexthop forwards packets to an egress RIF */
        MLXSW_SP_NEXTHOP_ACTION_FORWARD,
        /* Nexthop discards packets */
        MLXSW_SP_NEXTHOP_ACTION_DISCARD,
        /* Nexthop traps packets */
        MLXSW_SP_NEXTHOP_ACTION_TRAP,
};

struct mlxsw_sp_nexthop_key {
        struct fib_nh *fib_nh;
};

struct mlxsw_sp_nexthop {
        struct list_head neigh_list_node; /* member of neigh entry list */
        struct list_head rif_list_node;
        struct list_head router_list_node;
        struct mlxsw_sp_nexthop_group_info *nhgi; /* pointer back to the group
                                                   * this nexthop belongs to
                                                   */
        struct rhash_head ht_node;
        struct neigh_table *neigh_tbl;
        struct mlxsw_sp_nexthop_key key;
        unsigned char gw_addr[sizeof(struct in6_addr)];
        int ifindex;
        int nh_weight;
        int norm_nh_weight;
        int num_adj_entries;
        struct mlxsw_sp_rif *rif;
        u8 should_offload:1, /* set indicates this nexthop should be written
                              * to the adjacency table.
                              */
           offloaded:1, /* set indicates this nexthop was written to the
                         * adjacency table.
                         */
           update:1; /* set indicates this nexthop should be updated in the
                      * adjacency table (f.e., its MAC changed).
                      */
        enum mlxsw_sp_nexthop_action action;
        enum mlxsw_sp_nexthop_type type;
        union {
                struct mlxsw_sp_neigh_entry *neigh_entry;
                struct mlxsw_sp_ipip_entry *ipip_entry;
        };
        unsigned int counter_index;
        bool counter_valid;
};

enum mlxsw_sp_nexthop_group_type {
        MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV4,
        MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV6,
        MLXSW_SP_NEXTHOP_GROUP_TYPE_OBJ,
};

struct mlxsw_sp_nexthop_group_info {
        struct mlxsw_sp_nexthop_group *nh_grp;
        u32 adj_index;
        u16 ecmp_size;
        u16 count;
        int sum_norm_weight;
        u8 adj_index_valid:1,
           gateway:1, /* routes using the group use a gateway */
           is_resilient:1;
        struct list_head list; /* member in nh_res_grp_list */
        struct mlxsw_sp_nexthop nexthops[0];
#define nh_rif  nexthops[0].rif
};

struct mlxsw_sp_nexthop_group_vr_key {
        u16 vr_id;
        enum mlxsw_sp_l3proto proto;
};

struct mlxsw_sp_nexthop_group_vr_entry {
        struct list_head list; /* member in vr_list */
        struct rhash_head ht_node; /* member in vr_ht */
        refcount_t ref_count;
        struct mlxsw_sp_nexthop_group_vr_key key;
};

struct mlxsw_sp_nexthop_group {
        struct rhash_head ht_node;
        struct list_head fib_list; /* list of fib entries that use this group */
        union {
                struct {
                        struct fib_info *fi;
                } ipv4;
                struct {
                        u32 id;
                } obj;
        };
        struct mlxsw_sp_nexthop_group_info *nhgi;
        struct list_head vr_list;
        struct rhashtable vr_ht;
        enum mlxsw_sp_nexthop_group_type type;
        bool can_destroy;
};

void mlxsw_sp_nexthop_counter_alloc(struct mlxsw_sp *mlxsw_sp,
                                    struct mlxsw_sp_nexthop *nh)
{
        struct devlink *devlink;

        devlink = priv_to_devlink(mlxsw_sp->core);
        if (!devlink_dpipe_table_counter_enabled(devlink,
                                                 MLXSW_SP_DPIPE_TABLE_NAME_ADJ))
                return;

        if (mlxsw_sp_flow_counter_alloc(mlxsw_sp, &nh->counter_index))
                return;

        nh->counter_valid = true;
}

void mlxsw_sp_nexthop_counter_free(struct mlxsw_sp *mlxsw_sp,
                                   struct mlxsw_sp_nexthop *nh)
{
        if (!nh->counter_valid)
                return;
        mlxsw_sp_flow_counter_free(mlxsw_sp, nh->counter_index);
        nh->counter_valid = false;
}

int mlxsw_sp_nexthop_counter_get(struct mlxsw_sp *mlxsw_sp,
                                 struct mlxsw_sp_nexthop *nh, u64 *p_counter)
{
        if (!nh->counter_valid)
                return -EINVAL;

        return mlxsw_sp_flow_counter_get(mlxsw_sp, nh->counter_index,
                                         p_counter, NULL);
}

struct mlxsw_sp_nexthop *mlxsw_sp_nexthop_next(struct mlxsw_sp_router *router,
                                               struct mlxsw_sp_nexthop *nh)
{
        if (!nh) {
                if (list_empty(&router->nexthop_list))
                        return NULL;
                else
                        return list_first_entry(&router->nexthop_list,
                                                typeof(*nh), router_list_node);
        }
        if (list_is_last(&nh->router_list_node, &router->nexthop_list))
                return NULL;