net/mlx5_core: Introduce flow steering API

Introducing the following objects:

mlx5_flow_root_namespace: represent the root of specific flow table
type tree(e.g NIC receive, FDB, etc..)

mlx5_flow_group: define the mask of the flow specification.

fs_fte(flow steering flow table entry): defines the value of the
flow specification.

The following describes the relationships between the tree objects:
root_namespace --> priorities -->namespaces -->
priorities -->flow-tables --> flow-groups -->
flow-entries --> destinations

When we create new object(flow table/flow group/flow table entry), we
call to the FW command and then we add the related sw object to the tree.

When we destroy object, e.g. call to mlx5_destroy_flow_table, we use
the tree node destructor for destroying the FW object and remove the
node from the tree.
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Moni Shoua <monis@mellanox.com>
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/mellanox/mlx5/core/fs_core.c

@@ -35,6 +35,12 @@
 
 #include "mlx5_core.h"
 #include "fs_core.h"
+#include "fs_cmd.h"
+
+static void del_rule(struct fs_node *node);
+static void del_flow_table(struct fs_node *node);
+static void del_flow_group(struct fs_node *node);
+static void del_fte(struct fs_node *node);
 
 static void tree_init_node(struct fs_node *node,
 			   unsigned int refcount,
@@ -207,3 +213,461 @@ static bool compare_match_criteria(u8 match_criteria_enable1,
 	return match_criteria_enable1 == match_criteria_enable2 &&
 		!memcmp(mask1, mask2, MLX5_ST_SZ_BYTES(fte_match_param));
 }
+
+static struct mlx5_flow_root_namespace *find_root(struct fs_node *node)
+{
+	struct fs_node *root;
+	struct mlx5_flow_namespace *ns;
+
+	root = node->root;
+
+	if (WARN_ON(root->type != FS_TYPE_NAMESPACE)) {
+		pr_warn("mlx5: flow steering node is not in tree or garbaged\n");
+		return NULL;
+	}
+
+	ns = container_of(root, struct mlx5_flow_namespace, node);
+	return container_of(ns, struct mlx5_flow_root_namespace, ns);
+}
+
+static inline struct mlx5_core_dev *get_dev(struct fs_node *node)
+{
+	struct mlx5_flow_root_namespace *root = find_root(node);
+
+	if (root)
+		return root->dev;
+	return NULL;
+}
+
+static void del_flow_table(struct fs_node *node)
+{
+	struct mlx5_flow_table *ft;
+	struct mlx5_core_dev *dev;
+	struct fs_prio *prio;
+	int err;
+
+	fs_get_obj(ft, node);
+	dev = get_dev(&ft->node);
+
+	err = mlx5_cmd_destroy_flow_table(dev, ft);
+	if (err)
+		pr_warn("flow steering can't destroy ft\n");
+	fs_get_obj(prio, ft->node.parent);
+	prio->num_ft--;
+}
+
+static void del_rule(struct fs_node *node)
+{
+	struct mlx5_flow_rule *rule;
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_group *fg;
+	struct fs_fte *fte;
+	u32	*match_value;
+	struct mlx5_core_dev *dev = get_dev(node);
+	int match_len = MLX5_ST_SZ_BYTES(fte_match_param);
+	int err;
+
+	match_value = mlx5_vzalloc(match_len);
+	if (!match_value) {
+		pr_warn("failed to allocate inbox\n");
+		return;
+	}
+
+	fs_get_obj(rule, node);
+	fs_get_obj(fte, rule->node.parent);
+	fs_get_obj(fg, fte->node.parent);
+	memcpy(match_value, fte->val, sizeof(fte->val));
+	fs_get_obj(ft, fg->node.parent);
+	list_del(&rule->node.list);
+	fte->dests_size--;
+	if (fte->dests_size) {
+		err = mlx5_cmd_update_fte(dev, ft,
+					  fg->id, fte);
+		if (err)
+			pr_warn("%s can't del rule fg id=%d fte_index=%d\n",
+				__func__, fg->id, fte->index);
+	}
+	kvfree(match_value);
+}
+
+static void del_fte(struct fs_node *node)
+{
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_group *fg;
+	struct mlx5_core_dev *dev;
+	struct fs_fte *fte;
+	int err;
+
+	fs_get_obj(fte, node);
+	fs_get_obj(fg, fte->node.parent);
+	fs_get_obj(ft, fg->node.parent);
+
+	dev = get_dev(&ft->node);
+	err = mlx5_cmd_delete_fte(dev, ft,
+				  fte->index);
+	if (err)
+		pr_warn("flow steering can't delete fte in index %d of flow group id %d\n",
+			fte->index, fg->id);
+
+	fte->status = 0;
+	fg->num_ftes--;
+}
+
+static void del_flow_group(struct fs_node *node)
+{
+	struct mlx5_flow_group *fg;
+	struct mlx5_flow_table *ft;
+	struct mlx5_core_dev *dev;
+
+	fs_get_obj(fg, node);
+	fs_get_obj(ft, fg->node.parent);
+	dev = get_dev(&ft->node);
+
+	if (mlx5_cmd_destroy_flow_group(dev, ft, fg->id))
+		pr_warn("flow steering can't destroy fg %d of ft %d\n",
+			fg->id, ft->id);
+}
+
+static struct fs_fte *alloc_fte(u8 action,
+				u32 flow_tag,
+				u32 *match_value,
+				unsigned int index)
+{
+	struct fs_fte *fte;
+
+	fte = kzalloc(sizeof(*fte), GFP_KERNEL);
+	if (!fte)
+		return ERR_PTR(-ENOMEM);
+
+	memcpy(fte->val, match_value, sizeof(fte->val));
+	fte->node.type =  FS_TYPE_FLOW_ENTRY;
+	fte->flow_tag = flow_tag;
+	fte->index = index;
+	fte->action = action;
+
+	return fte;
+}
+
+static struct mlx5_flow_group *alloc_flow_group(u32 *create_fg_in)
+{
+	struct mlx5_flow_group *fg;
+	void *match_criteria = MLX5_ADDR_OF(create_flow_group_in,
+					    create_fg_in, match_criteria);
+	u8 match_criteria_enable = MLX5_GET(create_flow_group_in,
+					    create_fg_in,
+					    match_criteria_enable);
+	fg = kzalloc(sizeof(*fg), GFP_KERNEL);
+	if (!fg)
+		return ERR_PTR(-ENOMEM);
+
+	fg->mask.match_criteria_enable = match_criteria_enable;
+	memcpy(&fg->mask.match_criteria, match_criteria,
+	       sizeof(fg->mask.match_criteria));
+	fg->node.type =  FS_TYPE_FLOW_GROUP;
+	fg->start_index = MLX5_GET(create_flow_group_in, create_fg_in,
+				   start_flow_index);
+	fg->max_ftes = MLX5_GET(create_flow_group_in, create_fg_in,
+				end_flow_index) - fg->start_index + 1;
+	return fg;
+}
+
+static struct mlx5_flow_table *alloc_flow_table(int level, int max_fte,
+						enum fs_flow_table_type table_type)
+{
+	struct mlx5_flow_table *ft;
+
+	ft  = kzalloc(sizeof(*ft), GFP_KERNEL);
+	if (!ft)
+		return NULL;
+
+	ft->level = level;
+	ft->node.type = FS_TYPE_FLOW_TABLE;
+	ft->type = table_type;
+	ft->max_fte = max_fte;
+
+	return ft;
+}
+
+static struct mlx5_flow_table *mlx5_create_flow_table(struct mlx5_flow_namespace *ns,
+						      int prio,
+						      int max_fte)
+{
+	struct mlx5_flow_table *ft;
+	int err;
+	int log_table_sz;
+	struct mlx5_flow_root_namespace *root =
+		find_root(&ns->node);
+	struct fs_prio *fs_prio = NULL;
+
+	if (!root) {
+		pr_err("mlx5: flow steering failed to find root of namespace\n");
+		return ERR_PTR(-ENODEV);
+	}
+
+	fs_prio = find_prio(ns, prio);
+	if (!fs_prio)
+		return ERR_PTR(-EINVAL);
+
+	lock_ref_node(&fs_prio->node);
+	if (fs_prio->num_ft == fs_prio->max_ft) {
+		err = -ENOSPC;
+		goto unlock_prio;
+	}
+
+	ft = alloc_flow_table(find_next_free_level(fs_prio),
+			      roundup_pow_of_two(max_fte),
+			      root->table_type);
+	if (!ft) {
+		err = -ENOMEM;
+		goto unlock_prio;
+	}
+
+	tree_init_node(&ft->node, 1, del_flow_table);
+	log_table_sz = ilog2(ft->max_fte);
+	err = mlx5_cmd_create_flow_table(root->dev, ft->type, ft->level,
+					 log_table_sz, &ft->id);
+	if (err)
+		goto free_ft;
+
+	tree_add_node(&ft->node, &fs_prio->node);
+	list_add_tail(&ft->node.list, &fs_prio->node.children);
+	fs_prio->num_ft++;
+	unlock_ref_node(&fs_prio->node);
+
+	return ft;
+
+free_ft:
+	kfree(ft);
+unlock_prio:
+	unlock_ref_node(&fs_prio->node);
+	return ERR_PTR(err);
+}
+
+static struct mlx5_flow_group *mlx5_create_flow_group(struct mlx5_flow_table *ft,
+						      u32 *fg_in)
+{
+	struct mlx5_flow_group *fg;
+	struct mlx5_core_dev *dev = get_dev(&ft->node);
+	int err;
+
+	if (!dev)
+		return ERR_PTR(-ENODEV);
+
+	fg = alloc_flow_group(fg_in);
+	if (IS_ERR(fg))
+		return fg;
+
+	lock_ref_node(&ft->node);
+	err = mlx5_cmd_create_flow_group(dev, ft, fg_in, &fg->id);
+	if (err) {
+		kfree(fg);
+		unlock_ref_node(&ft->node);
+		return ERR_PTR(err);
+	}
+	/* Add node to tree */
+	tree_init_node(&fg->node, 1, del_flow_group);
+	tree_add_node(&fg->node, &ft->node);
+	/* Add node to group list */
+	list_add(&fg->node.list, ft->node.children.prev);
+	unlock_ref_node(&ft->node);
+
+	return fg;
+}
+
+static struct mlx5_flow_rule *alloc_rule(struct mlx5_flow_destination *dest)
+{
+	struct mlx5_flow_rule *rule;
+
+	rule = kzalloc(sizeof(*rule), GFP_KERNEL);
+	if (!rule)
+		return NULL;
+
+	rule->node.type = FS_TYPE_FLOW_DEST;
+	memcpy(&rule->dest_attr, dest, sizeof(*dest));
+
+	return rule;
+}
+
+/* fte should not be deleted while calling this function */
+static struct mlx5_flow_rule *add_rule_fte(struct fs_fte *fte,
+					   struct mlx5_flow_group *fg,
+					   struct mlx5_flow_destination *dest)
+{
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_rule *rule;
+	int err;
+
+	rule = alloc_rule(dest);
+	if (!rule)
+		return ERR_PTR(-ENOMEM);
+
+	fs_get_obj(ft, fg->node.parent);
+	/* Add dest to dests list- added as first element after the head */
+	tree_init_node(&rule->node, 1, del_rule);
+	list_add_tail(&rule->node.list, &fte->node.children);
+	fte->dests_size++;
+	if (fte->dests_size == 1)
+		err = mlx5_cmd_create_fte(get_dev(&ft->node),
+					  ft, fg->id, fte);
+	else
+		err = mlx5_cmd_update_fte(get_dev(&ft->node),
+					  ft, fg->id, fte);
+	if (err)
+		goto free_rule;
+
+	fte->status |= FS_FTE_STATUS_EXISTING;
+
+	return rule;
+
+free_rule:
+	list_del(&rule->node.list);
+	kfree(rule);
+	fte->dests_size--;
+	return ERR_PTR(err);
+}
+
+/* Assumed fg is locked */
+static unsigned int get_free_fte_index(struct mlx5_flow_group *fg,
+				       struct list_head **prev)
+{
+	struct fs_fte *fte;
+	unsigned int start = fg->start_index;
+
+	if (prev)
+		*prev = &fg->node.children;
+
+	/* assumed list is sorted by index */
+	fs_for_each_fte(fte, fg) {
+		if (fte->index != start)
+			return start;
+		start++;
+		if (prev)
+			*prev = &fte->node.list;
+	}
+
+	return start;
+}
+
+/* prev is output, prev->next = new_fte */
+static struct fs_fte *create_fte(struct mlx5_flow_group *fg,
+				 u32 *match_value,
+				 u8 action,
+				 u32 flow_tag,
+				 struct list_head **prev)
+{
+	struct fs_fte *fte;
+	int index;
+
+	index = get_free_fte_index(fg, prev);
+	fte = alloc_fte(action, flow_tag, match_value, index);
+	if (IS_ERR(fte))
+		return fte;
+
+	return fte;
+}
+
+/* Assuming parent fg(flow table) is locked */
+static struct mlx5_flow_rule *add_rule_fg(struct mlx5_flow_group *fg,
+					  u32 *match_value,
+					  u8 action,
+					  u32 flow_tag,
+					  struct mlx5_flow_destination *dest)
+{
+	struct fs_fte *fte;
+	struct mlx5_flow_rule *rule;
+	struct mlx5_flow_table *ft;
+	struct list_head *prev;
+
+	lock_ref_node(&fg->node);
+	fs_for_each_fte(fte, fg) {
+		nested_lock_ref_node(&fte->node);
+		if (compare_match_value(&fg->mask, match_value, &fte->val) &&
+		    action == fte->action && flow_tag == fte->flow_tag) {
+			rule = add_rule_fte(fte, fg, dest);
+			unlock_ref_node(&fte->node);
+			if (IS_ERR(rule))
+				goto unlock_fg;
+			else
+				goto add_rule;
+		}
+		unlock_ref_node(&fte->node);
+	}
+	fs_get_obj(ft, fg->node.parent);
+	if (fg->num_ftes >= fg->max_ftes) {
+		rule = ERR_PTR(-ENOSPC);
+		goto unlock_fg;
+	}
+
+	fte = create_fte(fg, match_value, action, flow_tag, &prev);
+	if (IS_ERR(fte)) {
+		rule = (void *)fte;
+		goto unlock_fg;
+	}
+	tree_init_node(&fte->node, 0, del_fte);
+	rule = add_rule_fte(fte, fg, dest);
+	if (IS_ERR(rule)) {
+		kfree(fte);
+		goto unlock_fg;
+	}
+
+	fg->num_ftes++;
+
+	tree_add_node(&fte->node, &fg->node);
+	list_add(&fte->node.list, prev);
+add_rule:
+	tree_add_node(&rule->node, &fte->node);
+unlock_fg:
+	unlock_ref_node(&fg->node);
+	return rule;
+}
+
+static struct mlx5_flow_rule *
+mlx5_add_flow_rule(struct mlx5_flow_table *ft,
+		   u8 match_criteria_enable,
+		   u32 *match_criteria,
+		   u32 *match_value,
+		   u32 action,
+		   u32 flow_tag,
+		   struct mlx5_flow_destination *dest)
+{
+	struct mlx5_flow_group *g;
+	struct mlx5_flow_rule *rule = ERR_PTR(-EINVAL);
+
+	tree_get_node(&ft->node);
+	lock_ref_node(&ft->node);
+	fs_for_each_fg(g, ft)
+		if (compare_match_criteria(g->mask.match_criteria_enable,
+					   match_criteria_enable,
+					   g->mask.match_criteria,
+					   match_criteria)) {
+			unlock_ref_node(&ft->node);
+			rule = add_rule_fg(g, match_value,
+					   action, flow_tag, dest);
+			goto put;
+		}
+	unlock_ref_node(&ft->node);
+put:
+	tree_put_node(&ft->node);
+	return rule;
+}
+
+static void mlx5_del_flow_rule(struct mlx5_flow_rule *rule)
+{
+	tree_remove_node(&rule->node);
+}
+
+static int mlx5_destroy_flow_table(struct mlx5_flow_table *ft)
+{
+	if (tree_remove_node(&ft->node))
+		mlx5_core_warn(get_dev(&ft->node), "Flow table %d wasn't destroyed, refcount > 1\n",
+			       ft->id);
+
+	return 0;
+}
+
+static void mlx5_destroy_flow_group(struct mlx5_flow_group *fg)
+{
+	if (tree_remove_node(&fg->node))
+		mlx5_core_warn(get_dev(&fg->node), "Flow group %d wasn't destroyed, refcount > 1\n",
+			       fg->id);
+}

drivers/net/ethernet/mellanox/mlx5/core/fs_core.h

@@ -69,13 +69,16 @@ struct mlx5_flow_rule {
 	struct mlx5_flow_destination		dest_attr;
 };
 
+/* Type of children is mlx5_flow_group */
 struct mlx5_flow_table {
 	struct fs_node			node;
 	u32				id;
+	unsigned int			max_fte;
 	unsigned int			level;
 	enum fs_flow_table_type		type;
 };
 
+/* Type of children is mlx5_flow_rule */
 struct fs_fte {
 	struct fs_node			node;
 	u32				val[MLX5_ST_SZ_DW(fte_match_param)];
@@ -83,15 +86,19 @@ struct fs_fte {
 	u32				flow_tag;
 	u32				index;
 	u32				action;
+	enum fs_fte_status		status;
 };
 
+/* Type of children is mlx5_flow_table/namespace */
 struct fs_prio {
 	struct fs_node			node;
 	unsigned int			max_ft;
 	unsigned int			start_level;
 	unsigned int			prio;
+	unsigned int			num_ft;
 };
 
+/* Type of children is fs_prio */
 struct mlx5_flow_namespace {
 	/* parent == NULL => root ns */
 	struct	fs_node			node;
@@ -102,6 +109,22 @@ struct mlx5_flow_group_mask {
 	u32	match_criteria[MLX5_ST_SZ_DW(fte_match_param)];
 };
 
+/* Type of children is fs_fte */
+struct mlx5_flow_group {
+	struct fs_node			node;
+	struct mlx5_flow_group_mask	mask;
+	u32				start_index;
+	u32				max_ftes;
+	u32				num_ftes;
+	u32				id;
+};
+
+struct mlx5_flow_root_namespace {
+	struct mlx5_flow_namespace	ns;
+	enum   fs_flow_table_type	table_type;
+	struct mlx5_core_dev		*dev;
+};
+
 #define fs_get_obj(v, _node)  {v = container_of((_node), typeof(*v), node); }
 
 #define fs_list_for_each_entry(pos, root)		\