/*
- * Copyright (c) 2009, 2010, 2011 Nicira Networks.
+ * Copyright (c) 2009, 2010, 2011, 2012, 2013 Nicira, Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
/* Definitions for use within ofproto. */
#include "ofproto/ofproto.h"
+#include "cfm.h"
#include "classifier.h"
+#include "heap.h"
#include "list.h"
+#include "ofp-errors.h"
+#include "ofp-util.h"
#include "shash.h"
+#include "simap.h"
#include "timeval.h"
+struct match;
+struct ofpact;
+struct ofputil_flow_mod;
+struct bfd_cfg;
+
/* An OpenFlow switch.
*
* With few exceptions, ofproto implementations may look at these fields but
* should not modify them. */
struct ofproto {
+ struct hmap_node hmap_node; /* In global 'all_ofprotos' hmap. */
const struct ofproto_class *ofproto_class;
char *type; /* Datapath type. */
char *name; /* Datapath name. */
- struct hmap_node hmap_node; /* In global 'all_ofprotos' hmap. */
/* Settings. */
uint64_t fallback_dpid; /* Datapath ID if no better choice found. */
unsigned flow_eviction_threshold; /* Threshold at which to begin flow
* table eviction. Only affects the
* ofproto-dpif implementation */
- char *mfr_desc; /* Manufacturer. */
- char *hw_desc; /* Hardware. */
- char *sw_desc; /* Software version. */
- char *serial_desc; /* Serial number. */
- char *dp_desc; /* Datapath description. */
+ bool forward_bpdu; /* Option to allow forwarding of BPDU frames
+ * when NORMAL action is invoked. */
+ char *mfr_desc; /* Manufacturer (NULL for default)b. */
+ char *hw_desc; /* Hardware (NULL for default). */
+ char *sw_desc; /* Software version (NULL for default). */
+ char *serial_desc; /* Serial number (NULL for default). */
+ char *dp_desc; /* Datapath description (NULL for default). */
+ enum ofp_config_flags frag_handling; /* One of OFPC_*. */
/* Datapath. */
struct hmap ports; /* Contains "struct ofport"s. */
struct shash port_by_name;
+ unsigned long *ofp_port_ids;/* Bitmap of used OpenFlow port numbers. */
+ struct simap ofp_requests; /* OpenFlow port number requests. */
+ uint16_t alloc_port_no; /* Last allocated OpenFlow port number. */
+ uint16_t max_ports; /* Max possible OpenFlow port num, plus one. */
/* Flow tables. */
- struct classifier *tables; /* Each classifier contains "struct rule"s. */
+ struct oftable *tables;
int n_tables;
+ /* Optimisation for flow expiry.
+ * These flows should all be present in tables. */
+ struct list expirable; /* Expirable 'struct rule"s in all tables. */
+
/* OpenFlow connections. */
struct connmgr *connmgr;
/* Flow table operation tracking. */
int state; /* Internal state. */
struct list pending; /* List of "struct ofopgroup"s. */
+ unsigned int n_pending; /* list_size(&pending). */
struct hmap deletions; /* All OFOPERATION_DELETE "ofoperation"s. */
+
+ /* Flow table operation logging. */
+ int n_add, n_delete, n_modify; /* Number of unreported ops of each kind. */
+ long long int first_op, last_op; /* Range of times for unreported ops. */
+ long long int next_op_report; /* Time to report ops, or LLONG_MAX. */
+ long long int op_backoff; /* Earliest time to report ops again. */
+
+ /* Linux VLAN device support (e.g. "eth0.10" for VLAN 10.)
+ *
+ * This is deprecated. It is only for compatibility with broken device
+ * drivers in old versions of Linux that do not properly support VLANs when
+ * VLAN devices are not used. When broken device drivers are no longer in
+ * widespread use, we will delete these interfaces. */
+ unsigned long int *vlan_bitmap; /* 4096-bit bitmap of in-use VLANs. */
+ bool vlans_changed; /* True if new VLANs are in use. */
+ int min_mtu; /* Current MTU of non-internal ports. */
};
+void ofproto_init_tables(struct ofproto *, int n_tables);
+void ofproto_init_max_ports(struct ofproto *, uint16_t max_ports);
+
struct ofproto *ofproto_lookup(const char *name);
struct ofport *ofproto_get_port(const struct ofproto *, uint16_t ofp_port);
* With few exceptions, ofproto implementations may look at these fields but
* should not modify them. */
struct ofport {
- struct ofproto *ofproto; /* The ofproto that contains this port. */
struct hmap_node hmap_node; /* In struct ofproto's "ports" hmap. */
+ struct ofproto *ofproto; /* The ofproto that contains this port. */
struct netdev *netdev;
- struct ofp_phy_port opp;
+ struct ofputil_phy_port pp;
uint16_t ofp_port; /* OpenFlow port number. */
unsigned int change_seq;
+ int mtu;
+};
+
+void ofproto_port_set_state(struct ofport *, enum ofputil_port_state);
+
+/* OpenFlow table flags:
+ *
+ * - "Hidden" tables are not included in OpenFlow operations that operate on
+ * "all tables". For example, a request for flow stats on all tables will
+ * omit flows in hidden tables, table stats requests will omit the table
+ * entirely, and the switch features reply will not count the hidden table.
+ *
+ * However, operations that specifically name the particular table still
+ * operate on it. For example, flow_mods and flow stats requests on a
+ * hidden table work.
+ *
+ * To avoid gaps in table IDs (which have unclear validity in OpenFlow),
+ * hidden tables must be the highest-numbered tables that a provider
+ * implements.
+ *
+ * - "Read-only" tables can't be changed through OpenFlow operations. (At
+ * the moment all flow table operations go effectively through OpenFlow, so
+ * this means that read-only tables can't be changed at all after the
+ * read-only flag is set.)
+ *
+ * The generic ofproto layer never sets these flags. An ofproto provider can
+ * set them if it is appropriate.
+ */
+enum oftable_flags {
+ OFTABLE_HIDDEN = 1 << 0, /* Hide from most OpenFlow operations. */
+ OFTABLE_READONLY = 1 << 1 /* Don't allow OpenFlow to change this table. */
};
+/* A flow table within a "struct ofproto". */
+struct oftable {
+ enum oftable_flags flags;
+ struct classifier cls; /* Contains "struct rule"s. */
+ char *name; /* Table name exposed via OpenFlow, or NULL. */
+
+ /* Maximum number of flows or UINT_MAX if there is no limit besides any
+ * limit imposed by resource limitations. */
+ unsigned int max_flows;
+
+ /* These members determine the handling of an attempt to add a flow that
+ * would cause the table to have more than 'max_flows' flows.
+ *
+ * If 'eviction_fields' is NULL, overflows will be rejected with an error.
+ *
+ * If 'eviction_fields' is nonnull (regardless of whether n_eviction_fields
+ * is nonzero), an overflow will cause a flow to be removed. The flow to
+ * be removed is chosen to give fairness among groups distinguished by
+ * different values for the subfields within 'groups'. */
+ struct mf_subfield *eviction_fields;
+ size_t n_eviction_fields;
+
+ /* Eviction groups.
+ *
+ * When a flow is added that would cause the table to have more than
+ * 'max_flows' flows, and 'eviction_fields' is nonnull, these groups are
+ * used to decide which rule to evict: the rule is chosen from the eviction
+ * group that contains the greatest number of rules.*/
+ uint32_t eviction_group_id_basis;
+ struct hmap eviction_groups_by_id;
+ struct heap eviction_groups_by_size;
+};
+
+/* Assigns TABLE to each oftable, in turn, in OFPROTO.
+ *
+ * All parameters are evaluated multiple times. */
+#define OFPROTO_FOR_EACH_TABLE(TABLE, OFPROTO) \
+ for ((TABLE) = (OFPROTO)->tables; \
+ (TABLE) < &(OFPROTO)->tables[(OFPROTO)->n_tables]; \
+ (TABLE)++)
+
/* An OpenFlow flow within a "struct ofproto".
*
* With few exceptions, ofproto implementations may look at these fields but
* should not modify them. */
struct rule {
- struct ofproto *ofproto; /* The ofproto that contains this rule. */
struct list ofproto_node; /* Owned by ofproto base code. */
+ struct ofproto *ofproto; /* The ofproto that contains this rule. */
struct cls_rule cr; /* In owning ofproto's classifier. */
struct ofoperation *pending; /* Operation now in progress, if nonnull. */
ovs_be64 flow_cookie; /* Controller-issued identifier. */
long long int created; /* Creation time. */
- uint16_t idle_timeout; /* In seconds from time of last use. */
- uint16_t hard_timeout; /* In seconds from time of creation. */
+ long long int modified; /* Time of last modification. */
+ long long int used; /* Last use; time created if never used. */
+ uint16_t hard_timeout; /* In seconds from ->modified. */
+ uint16_t idle_timeout; /* In seconds from ->used. */
uint8_t table_id; /* Index in ofproto's 'tables' array. */
bool send_flow_removed; /* Send a flow removed message? */
- union ofp_action *actions; /* OpenFlow actions. */
- int n_actions; /* Number of elements in actions[]. */
+ /* Eviction groups. */
+ bool evictable; /* If false, prevents eviction. */
+ struct heap_node evg_node; /* In eviction_group's "rules" heap. */
+ struct eviction_group *eviction_group; /* NULL if not in any group. */
+
+ struct ofpact *ofpacts; /* Sequence of "struct ofpacts". */
+ unsigned int ofpacts_len; /* Size of 'ofpacts', in bytes. */
+
+ /* Flow monitors. */
+ enum nx_flow_monitor_flags monitor_flags;
+ uint64_t add_seqno; /* Sequence number when added. */
+ uint64_t modify_seqno; /* Sequence number when changed. */
+
+ /* Optimisation for flow expiry. */
+ struct list expirable; /* In ofproto's 'expirable' list if this rule
+ * is expirable, otherwise empty. */
};
static inline struct rule *
return cls_rule ? CONTAINER_OF(cls_rule, struct rule, cr) : NULL;
}
+void ofproto_rule_update_used(struct rule *, long long int used);
void ofproto_rule_expire(struct rule *, uint8_t reason);
void ofproto_rule_destroy(struct rule *);
-void ofoperation_complete(struct ofoperation *, int status);
+bool ofproto_rule_has_out_port(const struct rule *, uint16_t out_port);
+
+void ofoperation_complete(struct ofoperation *, enum ofperr);
struct rule *ofoperation_get_victim(struct ofoperation *);
+bool ofoperation_has_out_port(const struct ofoperation *, uint16_t out_port);
+
+bool ofproto_rule_is_hidden(const struct rule *);
+
/* ofproto class structure, to be defined by each ofproto implementation.
*
*
*
* Most of these functions return 0 if they are successful or a positive error
* code on failure. Depending on the function, valid error codes are either
- * errno values or OpenFlow error codes constructed with ofp_mkerr().
+ * errno values or OFPERR_* OpenFlow error codes.
*
* Most of these functions are expected to execute synchronously, that is, to
* block as necessary to obtain a result. Thus, these functions may return
/* ## Factory Functions ## */
/* ## ----------------- ## */
+ /* Initializes provider. The caller may pass in 'iface_hints',
+ * which contains an shash of "struct iface_hint" elements indexed
+ * by the interface's name. The provider may use these hints to
+ * describe the startup configuration in order to reinitialize its
+ * state. The caller owns the provided data, so a provider must
+ * make copies of anything required. An ofproto provider must
+ * remove any existing state that is not described by the hint, and
+ * may choose to remove it all. */
+ void (*init)(const struct shash *iface_hints);
+
/* Enumerates the types of all support ofproto types into 'types'. The
* caller has already initialized 'types' and other ofproto classes might
* already have added names to it. */
*/
int (*del)(const char *type, const char *name);
+ /* Returns the type to pass to netdev_open() when a datapath of type
+ * 'datapath_type' has a port of type 'port_type', for a few special
+ * cases when a netdev type differs from a port type. For example,
+ * when using the userspace datapath, a port of type "internal"
+ * needs to be opened as "tap".
+ *
+ * Returns either 'type' itself or a string literal, which must not
+ * be freed. */
+ const char *(*port_open_type)(const char *datapath_type,
+ const char *port_type);
+
+/* ## ------------------------ ## */
+/* ## Top-Level type Functions ## */
+/* ## ------------------------ ## */
+
+ /* Performs any periodic activity required on ofprotos of type
+ * 'type'.
+ *
+ * An ofproto provider may implement it or not, depending on whether
+ * it needs type-level maintenance.
+ *
+ * Returns 0 if successful, otherwise a positive errno value. */
+ int (*type_run)(const char *type);
+
+ /* Performs periodic activity required on ofprotos of type 'type'
+ * that needs to be done with the least possible latency.
+ *
+ * This is run multiple times per main loop. An ofproto provider may
+ * implement it or not, according to whether it provides a performance
+ * boost for that ofproto implementation.
+ *
+ * Returns 0 if successful, otherwise a positive errno value. */
+ int (*type_run_fast)(const char *type);
+
+ /* Causes the poll loop to wake up when a type 'type''s 'run'
+ * function needs to be called, e.g. by calling the timer or fd
+ * waiting functions in poll-loop.h.
+ *
+ * An ofproto provider may implement it or not, depending on whether
+ * it needs type-level maintenance. */
+ void (*type_wait)(const char *type);
+
/* ## --------------------------- ## */
/* ## Top-Level ofproto Functions ## */
/* ## --------------------------- ## */
*
* When ->construct() is called, the client does not yet know how many flow
* tables the datapath supports, so ofproto->n_tables will be 0 and
- * ofproto->tables will be NULL. ->construct() should store the number of
- * flow tables supported by the datapath (between 1 and 255, inclusive)
- * into '*n_tables'. After a successful return, the client will initialize
- * the base 'n_tables' member to '*n_tables' and allocate and initialize
- * the base 'tables' member as the specified number of empty flow tables.
- * Each flow table will be initially empty, so ->construct() should delete
- * flows from the underlying datapath, if necessary, rather than populating
- * the tables.
+ * ofproto->tables will be NULL. ->construct() should call
+ * ofproto_init_tables() to allocate and initialize ofproto->n_tables and
+ * ofproto->tables. Each flow table will be initially empty, so
+ * ->construct() should delete flows from the underlying datapath, if
+ * necessary, rather than populating the tables.
+ *
+ * If the ofproto knows the maximum port number that the datapath can have,
+ * then it can call ofproto_init_max_ports(). If it does so, then the
+ * client will ensure that the actions it allows to be used through
+ * OpenFlow do not refer to ports above that maximum number.
*
* Only one ofproto instance needs to be supported for any given datapath.
* If a datapath is already open as part of one "ofproto", then another
* returns.
*/
struct ofproto *(*alloc)(void);
- int (*construct)(struct ofproto *ofproto, int *n_tables);
+ int (*construct)(struct ofproto *ofproto);
void (*destruct)(struct ofproto *ofproto);
void (*dealloc)(struct ofproto *ofproto);
* - Call ofproto_rule_expire() for each OpenFlow flow that has reached
* its hard_timeout or idle_timeout, to expire the flow.
*
- * Returns 0 if successful, otherwise a positive errno value. The ENODEV
- * return value specifically means that the datapath underlying 'ofproto'
- * has been destroyed (externally, e.g. by an admin running ovs-dpctl).
- */
+ * (But rules that are part of a pending operation, e.g. rules for
+ * which ->pending is true, may not expire.)
+ *
+ * Returns 0 if successful, otherwise a positive errno value. */
int (*run)(struct ofproto *ofproto);
+ /* Performs periodic activity required by 'ofproto' that needs to be done
+ * with the least possible latency.
+ *
+ * This is run multiple times per main loop. An ofproto provider may
+ * implement it or not, according to whether it provides a performance
+ * boost for that ofproto implementation. */
+ int (*run_fast)(struct ofproto *ofproto);
+
/* Causes the poll loop to wake up when 'ofproto''s 'run' function needs to
* be called, e.g. by calling the timer or fd waiting functions in
* poll-loop.h. */
void (*wait)(struct ofproto *ofproto);
+ /* Adds some memory usage statistics for the implementation of 'ofproto'
+ * into 'usage', for use with memory_report().
+ *
+ * This function is optional. */
+ void (*get_memory_usage)(const struct ofproto *ofproto,
+ struct simap *usage);
+
/* Every "struct rule" in 'ofproto' is about to be deleted, one by one.
* This function may prepare for that, for example by clearing state in
* advance. It should *not* actually delete any "struct rule"s from
* otherwise.
*
* The implementation should store in '*actions' a bitmap of the supported
- * OpenFlow actions: the bit with value (1 << n) should be set to 1 if the
- * implementation supports the action with value 'n', and to 0 otherwise.
- * For example, if the implementation supports the OFPAT_OUTPUT and
- * OFPAT_ENQUEUE actions, but no others, it would set '*actions' to (1 <<
- * OFPAT_OUTPUT) | (1 << OFPAT_ENQUEUE). Vendor actions are not included
- * in '*actions'. */
+ * OpenFlow actions. Vendor actions are not included in '*actions'. */
void (*get_features)(struct ofproto *ofproto,
- bool *arp_match_ip, uint32_t *actions);
+ bool *arp_match_ip,
+ enum ofputil_action_bitmap *actions);
/* Helper for the OpenFlow OFPST_TABLE statistics request.
*
*
* - 'name' to "table#" where # is the table ID.
*
- * - 'wildcards' to OFPFW_ALL.
+ * - 'match' and 'wildcards' to OFPXMT12_MASK.
+ *
+ * - 'write_actions' and 'apply_actions' to OFPAT12_OUTPUT.
+ *
+ * - 'write_setfields' and 'apply_setfields' to OFPXMT12_MASK.
+ *
+ * - 'metadata_match' and 'metadata_write' to UINT64_MAX.
+ *
+ * - 'instructions' to OFPIT11_ALL.
+ *
+ * - 'config' to OFPTC11_TABLE_MISS_MASK.
*
* - 'max_entries' to 1,000,000.
*
* - 'wildcards' to the set of wildcards actually supported by the table
* (if it doesn't support all OpenFlow wildcards).
*
+ * - 'instructions' to set the instructions actually supported by
+ * the table.
+ *
+ * - 'write_actions' to set the write actions actually supported by
+ * the table (if it doesn't support all OpenFlow actions).
+ *
+ * - 'apply_actions' to set the apply actions actually supported by
+ * the table (if it doesn't support all OpenFlow actions).
+ *
+ * - 'write_setfields' to set the write setfields actually supported by
+ * the table.
+ *
+ * - 'apply_setfields' to set the apply setfields actually supported by
+ * the table.
+ *
* - 'max_entries' to the maximum number of flows actually supported by
* the hardware.
*
* - 'matched_count' to the number of packets looked up in this flow
* table so far that matched one of the flow entries.
*
- * Keep in mind that all of the members of struct ofp_table_stats are in
- * network byte order.
+ * All of the members of struct ofp12_table_stats are in network byte
+ * order.
*/
- void (*get_tables)(struct ofproto *ofproto, struct ofp_table_stats *ots);
+ void (*get_tables)(struct ofproto *ofproto, struct ofp12_table_stats *ots);
/* ## ---------------- ## */
/* ## ofport Functions ## */
/* Life-cycle functions for a "struct ofport" (see "Life Cycle" above).
*
* ->port_construct() should not modify any base members of the ofport.
+ * An ofproto implementation should use the 'ofp_port' member of
+ * "struct ofport" as the OpenFlow port number.
*
* ofports are managed by the base ofproto code. The ofproto
* implementation should only create and destroy them in response to calls
* function may use a null pointer. */
void (*port_modified)(struct ofport *ofport);
- /* Called after an OpenFlow OFPT_PORT_MOD request changes a port's
- * configuration. 'ofport->opp.config' contains the new configuration.
- * 'old_config' contains the previous configuration.
+ /* Called after an OpenFlow request changes a port's configuration.
+ * 'ofport->pp.config' contains the new configuration. 'old_config'
+ * contains the previous configuration.
*
- * The caller implements OFPPC_PORT_DOWN using netdev functions to turn
- * NETDEV_UP on and off, so this function doesn't have to do anything for
- * that bit (and it won't be called if that is the only bit that
+ * The caller implements OFPUTIL_PC_PORT_DOWN using netdev functions to
+ * turn NETDEV_UP on and off, so this function doesn't have to do anything
+ * for that bit (and it won't be called if that is the only bit that
* changes). */
- void (*port_reconfigured)(struct ofport *ofport, ovs_be32 old_config);
+ void (*port_reconfigured)(struct ofport *ofport,
+ enum ofputil_port_config old_config);
/* Looks up a port named 'devname' in 'ofproto'. On success, initializes
* '*port' appropriately.
const char *devname, struct ofproto_port *port);
/* Attempts to add 'netdev' as a port on 'ofproto'. Returns 0 if
- * successful, otherwise a positive errno value. If successful, sets
- * '*ofp_portp' to the new port's port number.
+ * successful, otherwise a positive errno value. The caller should
+ * inform the implementation of the OpenFlow port through the
+ * ->port_construct() method.
*
* It doesn't matter whether the new port will be returned by a later call
* to ->port_poll(); the implementation may do whatever is more
* convenient. */
- int (*port_add)(struct ofproto *ofproto, struct netdev *netdev,
- uint16_t *ofp_portp);
+ int (*port_add)(struct ofproto *ofproto, struct netdev *netdev);
/* Deletes port number 'ofp_port' from the datapath for 'ofproto'. Returns
* 0 if successful, otherwise a positive errno value.
* convenient. */
int (*port_del)(struct ofproto *ofproto, uint16_t ofp_port);
+ /* Get port stats */
+ int (*port_get_stats)(const struct ofport *port,
+ struct netdev_stats *stats);
+
/* Port iteration functions.
*
* The client might not be entirely in control of the ports within an
* ofproto. Some hardware implementations, for example, might have a fixed
- * set of ports in a datapath, and the Linux datapath allows the system
- * administrator to externally add and remove ports with ovs-dpctl. For
- * this reason, the client needs a way to iterate through all the ports
- * that are actually in a datapath. These functions provide that
- * functionality.
+ * set of ports in a datapath. For this reason, the client needs a way to
+ * iterate through all the ports that are actually in a datapath. These
+ * functions provide that functionality.
*
* The 'state' pointer provides the implementation a place to
* keep track of its position. Its format is opaque to the caller.
/* ## OpenFlow Rule Functions ## */
/* ## ----------------------- ## */
-
-
- /* Chooses an appropriate table for 'cls_rule' within 'ofproto'. On
+ /* Chooses an appropriate table for 'match' within 'ofproto'. On
* success, stores the table ID into '*table_idp' and returns 0. On
- * failure, returns an OpenFlow error code (as returned by ofp_mkerr()).
+ * failure, returns an OpenFlow error code.
*
- * The choice of table should be a function of 'cls_rule' and 'ofproto''s
+ * The choice of table should be a function of 'match' and 'ofproto''s
* datapath capabilities. It should not depend on the flows already in
* 'ofproto''s flow tables. Failure implies that an OpenFlow rule with
- * 'cls_rule' as its matching condition can never be inserted into
- * 'ofproto', even starting from an empty flow table.
+ * 'match' as its matching condition can never be inserted into 'ofproto',
+ * even starting from an empty flow table.
*
* If multiple tables are candidates for inserting the flow, the function
* should choose one arbitrarily (but deterministically).
*
* If this function is NULL then table 0 is always chosen. */
- int (*rule_choose_table)(const struct ofproto *ofproto,
- const struct cls_rule *cls_rule,
- uint8_t *table_idp);
+ enum ofperr (*rule_choose_table)(const struct ofproto *ofproto,
+ const struct match *match,
+ uint8_t *table_idp);
/* Life-cycle functions for a "struct rule" (see "Life Cycle" above).
*
*
* - 'rule' is replacing an existing rule in its flow table that had the
* same matching criteria and priority. In this case,
- * ofoperation_get_victim(rule) returns the rule being replaced.
+ * ofoperation_get_victim(rule) returns the rule being replaced (the
+ * "victim" rule).
*
* ->rule_construct() should set the following in motion:
*
* registers, then it is an error if 'rule->cr' does not wildcard all
* registers.
*
- * - Validate that 'rule->actions' and 'rule->n_actions' are well-formed
- * OpenFlow actions that the datapath can correctly implement. The
- * validate_actions() function (in ofp-util.c) can be useful as a model
- * for action validation, but it accepts all of the OpenFlow actions
- * that OVS understands. If your ofproto implementation only
- * implements a subset of those, then you should implement your own
- * action validation.
+ * - Validate that the datapath can correctly implement 'rule->ofpacts'.
*
* - If the rule is valid, update the datapath flow table, adding the new
* rule or replacing the existing one.
*
+ * - If 'rule' is replacing an existing rule, uninitialize any derived
+ * state for the victim rule, as in step 5 in the "Life Cycle"
+ * described above.
+ *
* (On failure, the ofproto code will roll back the insertion from the flow
- * table, either removing 'rule' or replacing it by the flow that was
- * originally in its place.)
+ * table, either removing 'rule' or replacing it by the victim rule if
+ * there is one.)
*
* ->rule_construct() must act in one of the following ways:
*
*
* * Call ofoperation_complete() and return 0.
*
- * * Return an OpenFlow error code (as returned by ofp_mkerr()). (Do
- * not call ofoperation_complete() in this case.)
+ * * Return an OpenFlow error code. (Do not call
+ * ofoperation_complete() in this case.)
*
- * In the former case, ->rule_destruct() will be called; in the latter
- * case, it will not. ->rule_dealloc() will be called in either case.
+ * Either way, ->rule_destruct() will not be called for 'rule', but
+ * ->rule_dealloc() will be.
*
* - If the operation is only partially complete, then it must return 0.
* Later, when the operation is complete, the ->run() or ->destruct()
*
* Rule destruction must not fail. */
struct rule *(*rule_alloc)(void);
- int (*rule_construct)(struct rule *rule);
+ enum ofperr (*rule_construct)(struct rule *rule);
void (*rule_destruct)(struct rule *rule);
void (*rule_dealloc)(struct rule *rule);
*
* 'flow' reflects the flow information for 'packet'. All of the
* information in 'flow' is extracted from 'packet', except for
- * flow->tun_id and flow->in_port, which are assigned the correct values
- * for the incoming packet. The register values are zeroed.
+ * flow->tunnel and flow->in_port, which are assigned the correct values
+ * for the incoming packet. The register values are zeroed. 'packet''s
+ * header pointers (e.g. packet->l3) are appropriately initialized.
*
- * The statistics for 'packet' should be included in 'rule'.
+ * The implementation should add the statistics for 'packet' into 'rule'.
*
- * Returns 0 if successful, otherwise an OpenFlow error code (as returned
- * by ofp_mkerr()). */
- int (*rule_execute)(struct rule *rule, struct flow *flow,
- struct ofpbuf *packet);
+ * Returns 0 if successful, otherwise an OpenFlow error code. */
+ enum ofperr (*rule_execute)(struct rule *rule, const struct flow *flow,
+ struct ofpbuf *packet);
/* When ->rule_modify_actions() is called, the caller has already replaced
* the OpenFlow actions in 'rule' by a new set. (The original actions are
*
* ->rule_modify_actions() should set the following in motion:
*
- * - Validate that the actions now in 'rule' are well-formed OpenFlow
- * actions that the datapath can correctly implement.
+ * - Validate that the datapath can correctly implement the actions now
+ * in 'rule'.
*
* - Update the datapath flow table with the new actions.
*
* rule. */
void (*rule_modify_actions)(struct rule *rule);
- /* These functions implement the OpenFlow IP fragment handling policy. By
- * default ('drop_frags' == false), an OpenFlow switch should treat IP
- * fragments the same way as other packets (although TCP and UDP port
- * numbers cannot be determined). With 'drop_frags' == true, the switch
- * should drop all IP fragments without passing them through the flow
- * table. */
- bool (*get_drop_frags)(struct ofproto *ofproto);
- void (*set_drop_frags)(struct ofproto *ofproto, bool drop_frags);
+ /* Changes the OpenFlow IP fragment handling policy to 'frag_handling',
+ * which takes one of the following values, with the corresponding
+ * meanings:
+ *
+ * - OFPC_FRAG_NORMAL: The switch should treat IP fragments the same way
+ * as other packets, omitting TCP and UDP port numbers (always setting
+ * them to 0).
+ *
+ * - OFPC_FRAG_DROP: The switch should drop all IP fragments without
+ * passing them through the flow table.
+ *
+ * - OFPC_FRAG_REASM: The switch should reassemble IP fragments before
+ * passing packets through the flow table.
+ *
+ * - OFPC_FRAG_NX_MATCH (a Nicira extension): Similar to OFPC_FRAG_NORMAL,
+ * except that TCP and UDP port numbers should be included in fragments
+ * with offset 0.
+ *
+ * Implementations are not required to support every mode.
+ * OFPC_FRAG_NORMAL is the default mode when an ofproto is created.
+ *
+ * At the time of the call to ->set_frag_handling(), the current mode is
+ * available in 'ofproto->frag_handling'. ->set_frag_handling() returns
+ * true if the requested mode was set, false if it is not supported.
+ *
+ * Upon successful return, the caller changes 'ofproto->frag_handling' to
+ * reflect the new mode.
+ */
+ bool (*set_frag_handling)(struct ofproto *ofproto,
+ enum ofp_config_flags frag_handling);
/* Implements the OpenFlow OFPT_PACKET_OUT command. The datapath should
- * execute the 'n_actions' in the 'actions' array on 'packet'.
+ * execute the 'ofpacts_len' bytes of "struct ofpacts" in 'ofpacts'.
*
- * The caller retains ownership of 'packet', so ->packet_out() should not
- * modify or free it.
+ * The caller retains ownership of 'packet' and of 'ofpacts', so
+ * ->packet_out() should not modify or free them.
*
- * This function must validate that the 'n_actions' elements in 'actions'
- * are well-formed OpenFlow actions that can be correctly implemented by
- * the datapath. If not, then it should return an OpenFlow error code (as
- * returned by ofp_mkerr()).
+ * This function must validate that it can correctly implement 'ofpacts'.
+ * If not, then it should return an OpenFlow error code.
*
* 'flow' reflects the flow information for 'packet'. All of the
* information in 'flow' is extracted from 'packet', except for
- * flow->in_port, which is taken from the OFPT_PACKET_OUT message.
- * flow->tun_id and its register values are zeroed.
+ * flow->in_port (see below). flow->tunnel and its register values are
+ * zeroed.
+ *
+ * flow->in_port comes from the OpenFlow OFPT_PACKET_OUT message. The
+ * implementation should reject invalid flow->in_port values by returning
+ * OFPERR_OFPBRC_BAD_PORT. (If the implementation called
+ * ofproto_init_max_ports(), then the client will reject these ports
+ * itself.) For consistency, the implementation should consider valid for
+ * flow->in_port any value that could possibly be seen in a packet that it
+ * passes to connmgr_send_packet_in(). Ideally, even an implementation
+ * that never generates packet-ins (e.g. due to hardware limitations)
+ * should still allow flow->in_port values for every possible physical port
+ * and OFPP_LOCAL. The only virtual ports (those above OFPP_MAX) that the
+ * caller will ever pass in as flow->in_port, other than OFPP_LOCAL, are
+ * OFPP_NONE and OFPP_CONTROLLER. The implementation should allow both of
+ * these, treating each of them as packets generated by the controller as
+ * opposed to packets originating from some switch port.
+ *
+ * (Ordinarily the only effect of flow->in_port is on output actions that
+ * involve the input port, such as actions that output to OFPP_IN_PORT,
+ * OFPP_FLOOD, or OFPP_ALL. flow->in_port can also affect Nicira extension
+ * "resubmit" actions.)
*
* 'packet' is not matched against the OpenFlow flow table, so its
* statistics should not be included in OpenFlow flow statistics.
*
- * Returns 0 if successful, otherwise an OpenFlow error code (as returned
- * by ofp_mkerr()). */
- int (*packet_out)(struct ofproto *ofproto, struct ofpbuf *packet,
- const struct flow *flow,
- const union ofp_action *actions,
- size_t n_actions);
+ * Returns 0 if successful, otherwise an OpenFlow error code. */
+ enum ofperr (*packet_out)(struct ofproto *ofproto, struct ofpbuf *packet,
+ const struct flow *flow,
+ const struct ofpact *ofpacts,
+ size_t ofpacts_len);
/* ## ------------------------- ## */
/* ## OFPP_NORMAL configuration ## */
int (*set_sflow)(struct ofproto *ofproto,
const struct ofproto_sflow_options *sflow_options);
+ /* Configures IPFIX on 'ofproto' according to the options in
+ * 'bridge_exporter_options' and the 'flow_exporters_options'
+ * array, or turns off IPFIX if 'bridge_exporter_options' and
+ * 'flow_exporters_options' is NULL.
+ *
+ * EOPNOTSUPP as a return value indicates that 'ofproto' does not support
+ * IPFIX, as does a null pointer. */
+ int (*set_ipfix)(
+ struct ofproto *ofproto,
+ const struct ofproto_ipfix_bridge_exporter_options
+ *bridge_exporter_options,
+ const struct ofproto_ipfix_flow_exporter_options
+ *flow_exporters_options, size_t n_flow_exporters_options);
+
/* Configures connectivity fault management on 'ofport'.
*
* If 'cfm_settings' is nonnull, configures CFM according to its members.
* support CFM, as does a null pointer. */
int (*set_cfm)(struct ofport *ofport, const struct cfm_settings *s);
- /* Checks the fault status of CFM configured on 'ofport'. Returns 1 if CFM
- * is faulted (generally indicating a connectivity problem), 0 if CFM is
- * not faulted, or -1 if CFM is not enabled on 'port'
+ /* Checks the status of CFM configured on 'ofport'. Returns true if the
+ * port's CFM status was successfully stored into '*status'. Returns false
+ * if the port did not have CFM configured, in which case '*status' is
+ * indeterminate.
*
- * This function may be a null pointer if the ofproto implementation does
- * not support CFM. */
- int (*get_cfm_fault)(const struct ofport *ofport);
+ * The caller must provide and owns '*status', but it does not own and must
+ * not modify or free the array returned in 'status->rmps'. */
+ bool (*get_cfm_status)(const struct ofport *ofport,
+ struct ofproto_cfm_status *status);
+
+ /* Configures BFD on 'ofport'.
+ *
+ * If 'cfg' is NULL, or 'cfg' does not contain the key value pair
+ * "enable=true", removes BFD from 'ofport'. Otherwise, configures BFD
+ * according to 'cfg'.
+ *
+ * EOPNOTSUPP as a return value indicates that this ofproto_class does not
+ * support BFD, as does a null pointer. */
+ int (*set_bfd)(struct ofport *ofport, const struct smap *cfg);
+
+ /* Populates 'smap' with the status of BFD on 'ofport'. Returns 0 on
+ * success, or a positive errno. EOPNOTSUPP as a return value indicates
+ * that this ofproto_class does not support BFD, as does a null pointer. */
+ int (*get_bfd_status)(struct ofport *ofport, struct smap *smap);
+
+ /* Configures spanning tree protocol (STP) on 'ofproto' using the
+ * settings defined in 's'.
+ *
+ * If 's' is nonnull, configures STP according to its members.
+ *
+ * If 's' is null, removes any STP configuration from 'ofproto'.
+ *
+ * EOPNOTSUPP as a return value indicates that this ofproto_class does not
+ * support STP, as does a null pointer. */
+ int (*set_stp)(struct ofproto *ofproto,
+ const struct ofproto_stp_settings *s);
+
+ /* Retrieves state of spanning tree protocol (STP) on 'ofproto'.
+ *
+ * Stores STP state for 'ofproto' in 's'. If the 'enabled' member
+ * is false, the other member values are not meaningful.
+ *
+ * EOPNOTSUPP as a return value indicates that this ofproto_class does not
+ * support STP, as does a null pointer. */
+ int (*get_stp_status)(struct ofproto *ofproto,
+ struct ofproto_stp_status *s);
+
+ /* Configures spanning tree protocol (STP) on 'ofport' using the
+ * settings defined in 's'.
+ *
+ * If 's' is nonnull, configures STP according to its members. The
+ * caller is responsible for assigning STP port numbers (using the
+ * 'port_num' member in the range of 1 through 255, inclusive) and
+ * ensuring there are no duplicates.
+ *
+ * If 's' is null, removes any STP configuration from 'ofport'.
+ *
+ * EOPNOTSUPP as a return value indicates that this ofproto_class does not
+ * support STP, as does a null pointer. */
+ int (*set_stp_port)(struct ofport *ofport,
+ const struct ofproto_port_stp_settings *s);
+
+ /* Retrieves spanning tree protocol (STP) port status of 'ofport'.
+ *
+ * Stores STP state for 'ofport' in 's'. If the 'enabled' member is
+ * false, the other member values are not meaningful.
+ *
+ * EOPNOTSUPP as a return value indicates that this ofproto_class does not
+ * support STP, as does a null pointer. */
+ int (*get_stp_port_status)(struct ofport *ofport,
+ struct ofproto_port_stp_status *s);
+
+ /* Registers meta-data associated with the 'n_qdscp' Qualities of Service
+ * 'queues' attached to 'ofport'. This data is not intended to be
+ * sufficient to implement QoS. Instead, providers may use this
+ * information to implement features which require knowledge of what queues
+ * exist on a port, and some basic information about them.
+ *
+ * EOPNOTSUPP as a return value indicates that this ofproto_class does not
+ * support QoS, as does a null pointer. */
+ int (*set_queues)(struct ofport *ofport,
+ const struct ofproto_port_queue *queues, size_t n_qdscp);
/* If 's' is nonnull, this function registers a "bundle" associated with
* client data pointer 'aux' in 'ofproto'. A bundle is the same concept as
* 'ofproto' associated with client data pointer 'aux'. If no such mirror
* has been registered, this has no effect.
*
- * This function affects only the behavior of the OFPP_NORMAL action. An
- * implementation that does not support it at all may set it to NULL or
- * return EOPNOTSUPP. An implementation that supports only a subset of the
- * functionality should implement what it can and return 0. */
+ * An implementation that does not support mirroring at all may set
+ * it to NULL or return EOPNOTSUPP. An implementation that supports
+ * only a subset of the functionality should implement what it can
+ * and return 0. */
int (*mirror_set)(struct ofproto *ofproto, void *aux,
const struct ofproto_mirror_settings *s);
+ /* Retrieves statistics from mirror associated with client data
+ * pointer 'aux' in 'ofproto'. Stores packet and byte counts in
+ * 'packets' and 'bytes', respectively. If a particular counter is
+ * not supported, the appropriate argument is set to UINT64_MAX.
+ *
+ * EOPNOTSUPP as a return value indicates that this ofproto_class does not
+ * support retrieving mirror statistics. */
+ int (*mirror_get_stats)(struct ofproto *ofproto, void *aux,
+ uint64_t *packets, uint64_t *bytes);
+
/* Configures the VLANs whose bits are set to 1 in 'flood_vlans' as VLANs
* on which all packets are flooded, instead of using MAC learning. If
* 'flood_vlans' is NULL, then MAC learning applies to all VLANs.
/* Returns true if 'aux' is a registered bundle that is currently in use as
* the output for a mirror. */
- bool (*is_mirror_output_bundle)(struct ofproto *ofproto, void *aux);
+ bool (*is_mirror_output_bundle)(const struct ofproto *ofproto, void *aux);
+
+ /* When the configuration option of forward_bpdu changes, this function
+ * will be invoked. */
+ void (*forward_bpdu_changed)(struct ofproto *ofproto);
+
+ /* Sets the MAC aging timeout for the OFPP_NORMAL action to 'idle_time', in
+ * seconds, and the maximum number of MAC table entries to
+ * 'max_entries'.
+ *
+ * An implementation that doesn't support configuring these features may
+ * set this function to NULL or implement it as a no-op. */
+ void (*set_mac_table_config)(struct ofproto *ofproto,
+ unsigned int idle_time, size_t max_entries);
+
+/* Linux VLAN device support (e.g. "eth0.10" for VLAN 10.)
+ *
+ * This is deprecated. It is only for compatibility with broken device drivers
+ * in old versions of Linux that do not properly support VLANs when VLAN
+ * devices are not used. When broken device drivers are no longer in
+ * widespread use, we will delete these interfaces. */
+
+ /* If 'realdev_ofp_port' is nonzero, then this function configures 'ofport'
+ * as a VLAN splinter port for VLAN 'vid', associated with the real device
+ * that has OpenFlow port number 'realdev_ofp_port'.
+ *
+ * If 'realdev_ofp_port' is zero, then this function deconfigures 'ofport'
+ * as a VLAN splinter port.
+ *
+ * This function should be NULL if a an implementation does not support
+ * it. */
+ int (*set_realdev)(struct ofport *ofport,
+ uint16_t realdev_ofp_port, int vid);
};
extern const struct ofproto_class ofproto_dpif_class;
int ofproto_class_register(const struct ofproto_class *);
int ofproto_class_unregister(const struct ofproto_class *);
-void ofproto_add_flow(struct ofproto *, const struct cls_rule *,
- const union ofp_action *, size_t n_actions);
-bool ofproto_delete_flow(struct ofproto *, const struct cls_rule *);
+/* ofproto_flow_mod() returns this value if the flow_mod could not be processed
+ * because it overlaps with an ongoing flow table operation that has not yet
+ * completed. The caller should retry the operation later.
+ *
+ * ofproto.c also uses this value internally for additional (similar) purposes.
+ *
+ * This particular value is a good choice because it is large, so that it does
+ * not collide with any errno value, but not large enough to collide with an
+ * OFPERR_* value. */
+enum { OFPROTO_POSTPONE = 1 << 16 };
+BUILD_ASSERT_DECL(OFPROTO_POSTPONE < OFPERR_OFS);
+
+int ofproto_flow_mod(struct ofproto *, const struct ofputil_flow_mod *);
+void ofproto_add_flow(struct ofproto *, const struct match *,
+ unsigned int priority,
+ const struct ofpact *ofpacts, size_t ofpacts_len);
+bool ofproto_delete_flow(struct ofproto *,
+ const struct match *, unsigned int priority);
void ofproto_flush_flows(struct ofproto *);
#endif /* ofproto/ofproto-provider.h */