diff options
Diffstat (limited to 'drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c')
-rw-r--r-- | drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c | 289 |
1 files changed, 208 insertions, 81 deletions
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c index 6bc6b48a56dc..f3f6eb081948 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c @@ -12,6 +12,7 @@ #include <net/flow_offload.h> #include <net/netfilter/nf_flow_table.h> #include <linux/workqueue.h> +#include <linux/refcount.h> #include <linux/xarray.h> #include "lib/fs_chains.h" @@ -27,6 +28,7 @@ #define MLX5_CT_STATE_ESTABLISHED_BIT BIT(1) #define MLX5_CT_STATE_TRK_BIT BIT(2) #define MLX5_CT_STATE_NAT_BIT BIT(3) +#define MLX5_CT_STATE_REPLY_BIT BIT(4) #define MLX5_FTE_ID_BITS (mlx5e_tc_attr_to_reg_mappings[FTEID_TO_REG].mlen * 8) #define MLX5_FTE_ID_MAX GENMASK(MLX5_FTE_ID_BITS - 1, 0) @@ -51,11 +53,11 @@ struct mlx5_tc_ct_priv { struct mlx5_flow_table *ct_nat; struct mlx5_flow_table *post_ct; struct mutex control_lock; /* guards parallel adds/dels */ - struct mutex shared_counter_lock; struct mapping_ctx *zone_mapping; struct mapping_ctx *labels_mapping; enum mlx5_flow_namespace_type ns_type; struct mlx5_fs_chains *chains; + spinlock_t ht_lock; /* protects ft entries */ }; struct mlx5_ct_flow { @@ -124,6 +126,10 @@ struct mlx5_ct_counter { bool is_shared; }; +enum { + MLX5_CT_ENTRY_FLAG_VALID, +}; + struct mlx5_ct_entry { struct rhash_head node; struct rhash_head tuple_node; @@ -134,6 +140,12 @@ struct mlx5_ct_entry { struct mlx5_ct_tuple tuple; struct mlx5_ct_tuple tuple_nat; struct mlx5_ct_zone_rule zone_rules[2]; + + struct mlx5_tc_ct_priv *ct_priv; + struct work_struct work; + + refcount_t refcnt; + unsigned long flags; }; static const struct rhashtable_params cts_ht_params = { @@ -641,6 +653,7 @@ mlx5_tc_ct_entry_create_mod_hdr(struct mlx5_tc_ct_priv *ct_priv, } ct_state |= MLX5_CT_STATE_ESTABLISHED_BIT | MLX5_CT_STATE_TRK_BIT; + ct_state |= meta->ct_metadata.orig_dir ? 0 : MLX5_CT_STATE_REPLY_BIT; err = mlx5_tc_ct_entry_set_registers(ct_priv, &mod_acts, ct_state, meta->ct_metadata.mark, @@ -709,11 +722,11 @@ mlx5_tc_ct_entry_add_rule(struct mlx5_tc_ct_priv *ct_priv, attr->outer_match_level = MLX5_MATCH_L4; attr->counter = entry->counter->counter; attr->flags |= MLX5_ESW_ATTR_FLAG_NO_IN_PORT; + if (ct_priv->ns_type == MLX5_FLOW_NAMESPACE_FDB) + attr->esw_attr->in_mdev = priv->mdev; mlx5_tc_ct_set_tuple_match(netdev_priv(ct_priv->netdev), spec, flow_rule); - mlx5e_tc_match_to_reg_match(spec, ZONE_TO_REG, - entry->tuple.zone & MLX5_CT_ZONE_MASK, - MLX5_CT_ZONE_MASK); + mlx5e_tc_match_to_reg_match(spec, ZONE_TO_REG, entry->tuple.zone, MLX5_CT_ZONE_MASK); zone_rule->rule = mlx5_tc_rule_insert(priv, spec, attr); if (IS_ERR(zone_rule->rule)) { @@ -740,6 +753,87 @@ err_attr: return err; } +static bool +mlx5_tc_ct_entry_valid(struct mlx5_ct_entry *entry) +{ + return test_bit(MLX5_CT_ENTRY_FLAG_VALID, &entry->flags); +} + +static struct mlx5_ct_entry * +mlx5_tc_ct_entry_get(struct mlx5_tc_ct_priv *ct_priv, struct mlx5_ct_tuple *tuple) +{ + struct mlx5_ct_entry *entry; + + entry = rhashtable_lookup_fast(&ct_priv->ct_tuples_ht, tuple, + tuples_ht_params); + if (entry && mlx5_tc_ct_entry_valid(entry) && + refcount_inc_not_zero(&entry->refcnt)) { + return entry; + } else if (!entry) { + entry = rhashtable_lookup_fast(&ct_priv->ct_tuples_nat_ht, + tuple, tuples_nat_ht_params); + if (entry && mlx5_tc_ct_entry_valid(entry) && + refcount_inc_not_zero(&entry->refcnt)) + return entry; + } + + return entry ? ERR_PTR(-EINVAL) : NULL; +} + +static void mlx5_tc_ct_entry_remove_from_tuples(struct mlx5_ct_entry *entry) +{ + struct mlx5_tc_ct_priv *ct_priv = entry->ct_priv; + + rhashtable_remove_fast(&ct_priv->ct_tuples_nat_ht, + &entry->tuple_nat_node, + tuples_nat_ht_params); + rhashtable_remove_fast(&ct_priv->ct_tuples_ht, &entry->tuple_node, + tuples_ht_params); +} + +static void mlx5_tc_ct_entry_del(struct mlx5_ct_entry *entry) +{ + struct mlx5_tc_ct_priv *ct_priv = entry->ct_priv; + + mlx5_tc_ct_entry_del_rules(ct_priv, entry); + + spin_lock_bh(&ct_priv->ht_lock); + mlx5_tc_ct_entry_remove_from_tuples(entry); + spin_unlock_bh(&ct_priv->ht_lock); + + mlx5_tc_ct_counter_put(ct_priv, entry); + kfree(entry); +} + +static void +mlx5_tc_ct_entry_put(struct mlx5_ct_entry *entry) +{ + if (!refcount_dec_and_test(&entry->refcnt)) + return; + + mlx5_tc_ct_entry_del(entry); +} + +static void mlx5_tc_ct_entry_del_work(struct work_struct *work) +{ + struct mlx5_ct_entry *entry = container_of(work, struct mlx5_ct_entry, work); + + mlx5_tc_ct_entry_del(entry); +} + +static void +__mlx5_tc_ct_entry_put(struct mlx5_ct_entry *entry) +{ + struct mlx5e_priv *priv; + + if (!refcount_dec_and_test(&entry->refcnt)) + return; + + priv = netdev_priv(entry->ct_priv->netdev); + INIT_WORK(&entry->work, mlx5_tc_ct_entry_del_work); + queue_work(priv->wq, &entry->work); +} + static struct mlx5_ct_counter * mlx5_tc_ct_counter_create(struct mlx5_tc_ct_priv *ct_priv) { @@ -770,7 +864,6 @@ mlx5_tc_ct_shared_counter_get(struct mlx5_tc_ct_priv *ct_priv, struct mlx5_ct_counter *shared_counter; struct mlx5_ct_entry *rev_entry; __be16 tmp_port; - int ret; /* get the reversed tuple */ tmp_port = rev_tuple.port.src; @@ -792,23 +885,31 @@ mlx5_tc_ct_shared_counter_get(struct mlx5_tc_ct_priv *ct_priv, } /* Use the same counter as the reverse direction */ - mutex_lock(&ct_priv->shared_counter_lock); - rev_entry = rhashtable_lookup_fast(&ct_priv->ct_tuples_ht, &rev_tuple, - tuples_ht_params); - if (rev_entry) { - if (refcount_inc_not_zero(&rev_entry->counter->refcount)) { - mutex_unlock(&ct_priv->shared_counter_lock); - return rev_entry->counter; - } + spin_lock_bh(&ct_priv->ht_lock); + rev_entry = mlx5_tc_ct_entry_get(ct_priv, &rev_tuple); + + if (IS_ERR(rev_entry)) { + spin_unlock_bh(&ct_priv->ht_lock); + goto create_counter; } - mutex_unlock(&ct_priv->shared_counter_lock); - shared_counter = mlx5_tc_ct_counter_create(ct_priv); - if (IS_ERR(shared_counter)) { - ret = PTR_ERR(shared_counter); - return ERR_PTR(ret); + if (rev_entry && refcount_inc_not_zero(&rev_entry->counter->refcount)) { + ct_dbg("Using shared counter entry=0x%p rev=0x%p\n", entry, rev_entry); + shared_counter = rev_entry->counter; + spin_unlock_bh(&ct_priv->ht_lock); + + mlx5_tc_ct_entry_put(rev_entry); + return shared_counter; } + spin_unlock_bh(&ct_priv->ht_lock); + +create_counter: + + shared_counter = mlx5_tc_ct_counter_create(ct_priv); + if (IS_ERR(shared_counter)) + return shared_counter; + shared_counter->is_shared = true; refcount_set(&shared_counter->refcount, 1); return shared_counter; @@ -866,10 +967,14 @@ mlx5_tc_ct_block_flow_offload_add(struct mlx5_ct_ft *ft, if (!meta_action) return -EOPNOTSUPP; - entry = rhashtable_lookup_fast(&ft->ct_entries_ht, &cookie, - cts_ht_params); - if (entry) - return 0; + spin_lock_bh(&ct_priv->ht_lock); + entry = rhashtable_lookup_fast(&ft->ct_entries_ht, &cookie, cts_ht_params); + if (entry && refcount_inc_not_zero(&entry->refcnt)) { + spin_unlock_bh(&ct_priv->ht_lock); + mlx5_tc_ct_entry_put(entry); + return -EEXIST; + } + spin_unlock_bh(&ct_priv->ht_lock); entry = kzalloc(sizeof(*entry), GFP_KERNEL); if (!entry) @@ -878,6 +983,8 @@ mlx5_tc_ct_block_flow_offload_add(struct mlx5_ct_ft *ft, entry->tuple.zone = ft->zone; entry->cookie = flow->cookie; entry->restore_cookie = meta_action->ct_metadata.cookie; + refcount_set(&entry->refcnt, 2); + entry->ct_priv = ct_priv; err = mlx5_tc_ct_rule_to_tuple(&entry->tuple, flow_rule); if (err) @@ -888,35 +995,40 @@ mlx5_tc_ct_block_flow_offload_add(struct mlx5_ct_ft *ft, if (err) goto err_set; - err = rhashtable_insert_fast(&ct_priv->ct_tuples_ht, - &entry->tuple_node, - tuples_ht_params); + spin_lock_bh(&ct_priv->ht_lock); + + err = rhashtable_lookup_insert_fast(&ft->ct_entries_ht, &entry->node, + cts_ht_params); + if (err) + goto err_entries; + + err = rhashtable_lookup_insert_fast(&ct_priv->ct_tuples_ht, + &entry->tuple_node, + tuples_ht_params); if (err) goto err_tuple; if (memcmp(&entry->tuple, &entry->tuple_nat, sizeof(entry->tuple))) { - err = rhashtable_insert_fast(&ct_priv->ct_tuples_nat_ht, - &entry->tuple_nat_node, - tuples_nat_ht_params); + err = rhashtable_lookup_insert_fast(&ct_priv->ct_tuples_nat_ht, + &entry->tuple_nat_node, + tuples_nat_ht_params); if (err) goto err_tuple_nat; } + spin_unlock_bh(&ct_priv->ht_lock); err = mlx5_tc_ct_entry_add_rules(ct_priv, flow_rule, entry, ft->zone_restore_id); if (err) goto err_rules; - err = rhashtable_insert_fast(&ft->ct_entries_ht, &entry->node, - cts_ht_params); - if (err) - goto err_insert; + set_bit(MLX5_CT_ENTRY_FLAG_VALID, &entry->flags); + mlx5_tc_ct_entry_put(entry); /* this function reference */ return 0; -err_insert: - mlx5_tc_ct_entry_del_rules(ct_priv, entry); err_rules: + spin_lock_bh(&ct_priv->ht_lock); if (mlx5_tc_ct_entry_has_nat(entry)) rhashtable_remove_fast(&ct_priv->ct_tuples_nat_ht, &entry->tuple_nat_node, tuples_nat_ht_params); @@ -925,47 +1037,43 @@ err_tuple_nat: &entry->tuple_node, tuples_ht_params); err_tuple: + rhashtable_remove_fast(&ft->ct_entries_ht, + &entry->node, + cts_ht_params); +err_entries: + spin_unlock_bh(&ct_priv->ht_lock); err_set: kfree(entry); - netdev_warn(ct_priv->netdev, - "Failed to offload ct entry, err: %d\n", err); + if (err != -EEXIST) + netdev_warn(ct_priv->netdev, "Failed to offload ct entry, err: %d\n", err); return err; } -static void -mlx5_tc_ct_del_ft_entry(struct mlx5_tc_ct_priv *ct_priv, - struct mlx5_ct_entry *entry) -{ - mlx5_tc_ct_entry_del_rules(ct_priv, entry); - mutex_lock(&ct_priv->shared_counter_lock); - if (mlx5_tc_ct_entry_has_nat(entry)) - rhashtable_remove_fast(&ct_priv->ct_tuples_nat_ht, - &entry->tuple_nat_node, - tuples_nat_ht_params); - rhashtable_remove_fast(&ct_priv->ct_tuples_ht, &entry->tuple_node, - tuples_ht_params); - mutex_unlock(&ct_priv->shared_counter_lock); - mlx5_tc_ct_counter_put(ct_priv, entry); - -} - static int mlx5_tc_ct_block_flow_offload_del(struct mlx5_ct_ft *ft, struct flow_cls_offload *flow) { + struct mlx5_tc_ct_priv *ct_priv = ft->ct_priv; unsigned long cookie = flow->cookie; struct mlx5_ct_entry *entry; - entry = rhashtable_lookup_fast(&ft->ct_entries_ht, &cookie, - cts_ht_params); - if (!entry) + spin_lock_bh(&ct_priv->ht_lock); + entry = rhashtable_lookup_fast(&ft->ct_entries_ht, &cookie, cts_ht_params); + if (!entry) { + spin_unlock_bh(&ct_priv->ht_lock); return -ENOENT; + } - mlx5_tc_ct_del_ft_entry(ft->ct_priv, entry); - WARN_ON(rhashtable_remove_fast(&ft->ct_entries_ht, - &entry->node, - cts_ht_params)); - kfree(entry); + if (!mlx5_tc_ct_entry_valid(entry)) { + spin_unlock_bh(&ct_priv->ht_lock); + return -EINVAL; + } + + rhashtable_remove_fast(&ft->ct_entries_ht, &entry->node, cts_ht_params); + mlx5_tc_ct_entry_remove_from_tuples(entry); + spin_unlock_bh(&ct_priv->ht_lock); + + mlx5_tc_ct_entry_put(entry); return 0; } @@ -974,19 +1082,30 @@ static int mlx5_tc_ct_block_flow_offload_stats(struct mlx5_ct_ft *ft, struct flow_cls_offload *f) { + struct mlx5_tc_ct_priv *ct_priv = ft->ct_priv; unsigned long cookie = f->cookie; struct mlx5_ct_entry *entry; u64 lastuse, packets, bytes; - entry = rhashtable_lookup_fast(&ft->ct_entries_ht, &cookie, - cts_ht_params); - if (!entry) + spin_lock_bh(&ct_priv->ht_lock); + entry = rhashtable_lookup_fast(&ft->ct_entries_ht, &cookie, cts_ht_params); + if (!entry) { + spin_unlock_bh(&ct_priv->ht_lock); return -ENOENT; + } + + if (!mlx5_tc_ct_entry_valid(entry) || !refcount_inc_not_zero(&entry->refcnt)) { + spin_unlock_bh(&ct_priv->ht_lock); + return -EINVAL; + } + + spin_unlock_bh(&ct_priv->ht_lock); mlx5_fc_query_cached(entry->counter->counter, &bytes, &packets, &lastuse); flow_stats_update(&f->stats, bytes, packets, 0, lastuse, FLOW_ACTION_HW_STATS_DELAYED); + mlx5_tc_ct_entry_put(entry); return 0; } @@ -1088,8 +1207,8 @@ mlx5_tc_ct_match_add(struct mlx5_tc_ct_priv *priv, struct netlink_ext_ack *extack) { struct flow_rule *rule = flow_cls_offload_flow_rule(f); + bool trk, est, untrk, unest, new, rpl, unrpl; struct flow_dissector_key_ct *mask, *key; - bool trk, est, untrk, unest, new; u32 ctstate = 0, ctstate_mask = 0; u16 ct_state_on, ct_state_off; u16 ct_state, ct_state_mask; @@ -1115,9 +1234,10 @@ mlx5_tc_ct_match_add(struct mlx5_tc_ct_priv *priv, if (ct_state_mask & ~(TCA_FLOWER_KEY_CT_FLAGS_TRACKED | TCA_FLOWER_KEY_CT_FLAGS_ESTABLISHED | - TCA_FLOWER_KEY_CT_FLAGS_NEW)) { + TCA_FLOWER_KEY_CT_FLAGS_NEW | + TCA_FLOWER_KEY_CT_FLAGS_REPLY)) { NL_SET_ERR_MSG_MOD(extack, - "only ct_state trk, est and new are supported for offload"); + "only ct_state trk, est, new and rpl are supported for offload"); return -EOPNOTSUPP; } @@ -1126,13 +1246,17 @@ mlx5_tc_ct_match_add(struct mlx5_tc_ct_priv *priv, trk = ct_state_on & TCA_FLOWER_KEY_CT_FLAGS_TRACKED; new = ct_state_on & TCA_FLOWER_KEY_CT_FLAGS_NEW; est = ct_state_on & TCA_FLOWER_KEY_CT_FLAGS_ESTABLISHED; + rpl = ct_state_on & TCA_FLOWER_KEY_CT_FLAGS_REPLY; untrk = ct_state_off & TCA_FLOWER_KEY_CT_FLAGS_TRACKED; unest = ct_state_off & TCA_FLOWER_KEY_CT_FLAGS_ESTABLISHED; + unrpl = ct_state_off & TCA_FLOWER_KEY_CT_FLAGS_REPLY; ctstate |= trk ? MLX5_CT_STATE_TRK_BIT : 0; ctstate |= est ? MLX5_CT_STATE_ESTABLISHED_BIT : 0; + ctstate |= rpl ? MLX5_CT_STATE_REPLY_BIT : 0; ctstate_mask |= (untrk || trk) ? MLX5_CT_STATE_TRK_BIT : 0; ctstate_mask |= (unest || est) ? MLX5_CT_STATE_ESTABLISHED_BIT : 0; + ctstate_mask |= (unrpl || rpl) ? MLX5_CT_STATE_REPLY_BIT : 0; if (new) { NL_SET_ERR_MSG_MOD(extack, @@ -1247,9 +1371,8 @@ static int tc_ct_pre_ct_add_rules(struct mlx5_ct_ft *ct_ft, pre_ct->flow_rule = rule; /* add miss rule */ - memset(spec, 0, sizeof(*spec)); dest.ft = nat ? ct_priv->ct_nat : ct_priv->ct; - rule = mlx5_add_flow_rules(ft, spec, &flow_act, &dest, 1); + rule = mlx5_add_flow_rules(ft, NULL, &flow_act, &dest, 1); if (IS_ERR(rule)) { err = PTR_ERR(rule); ct_dbg("Failed to add pre ct miss rule zone %d", zone); @@ -1478,11 +1601,9 @@ err_mapping: static void mlx5_tc_ct_flush_ft_entry(void *ptr, void *arg) { - struct mlx5_tc_ct_priv *ct_priv = arg; struct mlx5_ct_entry *entry = ptr; - mlx5_tc_ct_del_ft_entry(ct_priv, entry); - kfree(entry); + mlx5_tc_ct_entry_put(entry); } static void @@ -1760,7 +1881,6 @@ __mlx5_tc_ct_flow_offload_clear(struct mlx5_tc_ct_priv *ct_priv, goto err_set_registers; } - dealloc_mod_hdr_actions(mod_acts); pre_ct_attr->modify_hdr = mod_hdr; pre_ct_attr->action |= MLX5_FLOW_CONTEXT_ACTION_MOD_HDR; @@ -1960,6 +2080,7 @@ mlx5_tc_ct_init(struct mlx5e_priv *priv, struct mlx5_fs_chains *chains, goto err_mapping_labels; } + spin_lock_init(&ct_priv->ht_lock); ct_priv->ns_type = ns_type; ct_priv->chains = chains; ct_priv->netdev = priv->netdev; @@ -1994,7 +2115,6 @@ mlx5_tc_ct_init(struct mlx5e_priv *priv, struct mlx5_fs_chains *chains, idr_init(&ct_priv->fte_ids); mutex_init(&ct_priv->control_lock); - mutex_init(&ct_priv->shared_counter_lock); rhashtable_init(&ct_priv->zone_ht, &zone_params); rhashtable_init(&ct_priv->ct_tuples_ht, &tuples_ht_params); rhashtable_init(&ct_priv->ct_tuples_nat_ht, &tuples_nat_ht_params); @@ -2037,7 +2157,6 @@ mlx5_tc_ct_clean(struct mlx5_tc_ct_priv *ct_priv) rhashtable_destroy(&ct_priv->ct_tuples_nat_ht); rhashtable_destroy(&ct_priv->zone_ht); mutex_destroy(&ct_priv->control_lock); - mutex_destroy(&ct_priv->shared_counter_lock); idr_destroy(&ct_priv->fte_ids); kfree(ct_priv); } @@ -2059,14 +2178,22 @@ mlx5e_tc_ct_restore_flow(struct mlx5_tc_ct_priv *ct_priv, if (!mlx5_tc_ct_skb_to_tuple(skb, &tuple, zone)) return false; - entry = rhashtable_lookup_fast(&ct_priv->ct_tuples_ht, &tuple, - tuples_ht_params); - if (!entry) - entry = rhashtable_lookup_fast(&ct_priv->ct_tuples_nat_ht, - &tuple, tuples_nat_ht_params); - if (!entry) + spin_lock(&ct_priv->ht_lock); + + entry = mlx5_tc_ct_entry_get(ct_priv, &tuple); + if (!entry) { + spin_unlock(&ct_priv->ht_lock); + return false; + } + + if (IS_ERR(entry)) { + spin_unlock(&ct_priv->ht_lock); return false; + } + spin_unlock(&ct_priv->ht_lock); tcf_ct_flow_table_restore_skb(skb, entry->restore_cookie); + __mlx5_tc_ct_entry_put(entry); + return true; } |