net/mlx5e: CT: Use own workqueue instead of mlx5e priv
authorRoi Dayan <roid@nvidia.com>
Tue, 21 Jun 2022 07:43:55 +0000 (10:43 +0300)
committerSaeed Mahameed <saeedm@nvidia.com>
Wed, 6 Jul 2022 23:11:55 +0000 (16:11 -0700)
Allocate a ct priv workqueue instead of using mlx5e priv one
so flushing will only be of related CT entries.
Also move flushing of the workqueue before rhashtable destroy
otherwise entries won't be valid.

Fixes: b069e14fff46 ("net/mlx5e: CT: Fix queued up restore put() executing after relevant ft release")
Signed-off-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Oz Shlomo <ozsh@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c

index 25f51f8..ba171c7 100644 (file)
@@ -76,6 +76,7 @@ struct mlx5_tc_ct_priv {
        struct mlx5_ct_fs *fs;
        struct mlx5_ct_fs_ops *fs_ops;
        spinlock_t ht_lock; /* protects ft entries */
+       struct workqueue_struct *wq;
 
        struct mlx5_tc_ct_debugfs debugfs;
 };
@@ -941,14 +942,11 @@ static void mlx5_tc_ct_entry_del_work(struct work_struct *work)
 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);
+       queue_work(entry->ct_priv->wq, &entry->work);
 }
 
 static struct mlx5_ct_counter *
@@ -1759,19 +1757,16 @@ mlx5_tc_ct_flush_ft_entry(void *ptr, void *arg)
 static void
 mlx5_tc_ct_del_ft_cb(struct mlx5_tc_ct_priv *ct_priv, struct mlx5_ct_ft *ft)
 {
-       struct mlx5e_priv *priv;
-
        if (!refcount_dec_and_test(&ft->refcount))
                return;
 
+       flush_workqueue(ct_priv->wq);
        nf_flow_table_offload_del_cb(ft->nf_ft,
                                     mlx5_tc_ct_block_flow_offload, ft);
        rhashtable_remove_fast(&ct_priv->zone_ht, &ft->node, zone_params);
        rhashtable_free_and_destroy(&ft->ct_entries_ht,
                                    mlx5_tc_ct_flush_ft_entry,
                                    ct_priv);
-       priv = netdev_priv(ct_priv->netdev);
-       flush_workqueue(priv->wq);
        mlx5_tc_ct_free_pre_ct_tables(ft);
        mapping_remove(ct_priv->zone_mapping, ft->zone_restore_id);
        kfree(ft);
@@ -2176,6 +2171,12 @@ mlx5_tc_ct_init(struct mlx5e_priv *priv, struct mlx5_fs_chains *chains,
        if (rhashtable_init(&ct_priv->ct_tuples_nat_ht, &tuples_nat_ht_params))
                goto err_ct_tuples_nat_ht;
 
+       ct_priv->wq = alloc_ordered_workqueue("mlx5e_ct_priv_wq", 0);
+       if (!ct_priv->wq) {
+               err = -ENOMEM;
+               goto err_wq;
+       }
+
        err = mlx5_tc_ct_fs_init(ct_priv);
        if (err)
                goto err_init_fs;
@@ -2184,6 +2185,8 @@ mlx5_tc_ct_init(struct mlx5e_priv *priv, struct mlx5_fs_chains *chains,
        return ct_priv;
 
 err_init_fs:
+       destroy_workqueue(ct_priv->wq);
+err_wq:
        rhashtable_destroy(&ct_priv->ct_tuples_nat_ht);
 err_ct_tuples_nat_ht:
        rhashtable_destroy(&ct_priv->ct_tuples_ht);
@@ -2213,6 +2216,7 @@ mlx5_tc_ct_clean(struct mlx5_tc_ct_priv *ct_priv)
        if (!ct_priv)
                return;
 
+       destroy_workqueue(ct_priv->wq);
        mlx5_ct_tc_remove_dbgfs(ct_priv);
        chains = ct_priv->chains;