/*
- * Copyright (c) 2009 Nicira Networks.
+ * Copyright (c) 2009, 2010, 2011 Nicira Networks.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
#define DPIF_PROVIDER_H 1
/* Provider interface to dpifs, which provide an interface to an Open vSwitch
- * datapath. */
+ * datapath. A datapath is a collection of physical or virtual ports that are
+ * exposed over OpenFlow as a single switch. Datapaths and the collections of
+ * ports that they contain may be fixed or dynamic. */
#include <assert.h>
+#include "openflow/openflow.h"
#include "dpif.h"
+#include "util.h"
#ifdef __cplusplus
extern "C" {
* This structure should be treated as opaque by dpif implementations. */
struct dpif {
const struct dpif_class *dpif_class;
- char *name;
+ char *base_name;
+ char *full_name;
uint8_t netflow_engine_type;
uint8_t netflow_engine_id;
};
void dpif_init(struct dpif *, const struct dpif_class *, const char *name,
uint8_t netflow_engine_type, uint8_t netflow_engine_id);
+void dpif_uninit(struct dpif *dpif, bool close);
+
static inline void dpif_assert_class(const struct dpif *dpif,
const struct dpif_class *dpif_class)
{
* EWOULDBLOCK or EINPROGRESS. We may relax this requirement in the future if
* and when we encounter performance problems. */
struct dpif_class {
- /* Prefix for names of dpifs in this class, e.g. "netdev:".
+ /* Type of dpif in this class, e.g. "system", "netdev", etc.
*
- * One dpif class may have the empty string "" as its prefix, in which case
- * that dpif class is associated with dpif names that don't match any other
- * class name. */
- const char *prefix;
-
- /* Class name, for use in error messages. */
- const char *name;
+ * One of the providers should supply a "system" type, since this is
+ * the type assumed if no type is specified when opening a dpif. */
+ const char *type;
/* Performs periodic work needed by dpifs of this class, if any is
* necessary. */
* case this function may be a null pointer. */
int (*enumerate)(struct svec *all_dps);
- /* Attempts to open an existing dpif, if 'create' is false, or to open an
- * existing dpif or create a new one, if 'create' is true. 'name' is the
- * full dpif name provided by the user, e.g. "udatapath:/var/run/mypath".
- * This name is useful for error messages but must not be modified.
+ /* Attempts to open an existing dpif called 'name', if 'create' is false,
+ * or to open an existing dpif or create a new one, if 'create' is true.
*
- * 'suffix' is a copy of 'name' following the dpif's 'prefix'.
+ * 'dpif_class' is the class of dpif to open.
*
- * If successful, stores a pointer to the new dpif in '*dpifp'. On failure
- * there are no requirements on what is stored in '*dpifp'. */
- int (*open)(const char *name, char *suffix, bool create,
- struct dpif **dpifp);
+ * If successful, stores a pointer to the new dpif in '*dpifp', which must
+ * have class 'dpif_class'. On failure there are no requirements on what
+ * is stored in '*dpifp'. */
+ int (*open)(const struct dpif_class *dpif_class,
+ const char *name, bool create, struct dpif **dpifp);
/* Closes 'dpif' and frees associated memory. */
void (*close)(struct dpif *dpif);
* meaning is the same as for the get_drop_frags member function. */
int (*set_drop_frags)(struct dpif *dpif, bool drop_frags);
- /* Creates a new port in 'dpif' connected to network device 'devname'.
- * 'flags' is a set of ODP_PORT_* flags. If successful, sets '*port_no'
+ /* Adds 'netdev' as a new port in 'dpif'. If successful, sets '*port_no'
* to the new port's port number. */
- int (*port_add)(struct dpif *dpif, const char *devname, uint16_t flags,
+ int (*port_add)(struct dpif *dpif, struct netdev *netdev,
uint16_t *port_no);
/* Removes port numbered 'port_no' from 'dpif'. */
int (*port_del)(struct dpif *dpif, uint16_t port_no);
/* Queries 'dpif' for a port with the given 'port_no' or 'devname'. Stores
- * information about the port into '*port' if successful. */
+ * information about the port into '*port' if successful.
+ *
+ * The caller takes ownership of data in 'port' and must free it with
+ * dpif_port_destroy() when it is no longer needed. */
int (*port_query_by_number)(const struct dpif *dpif, uint16_t port_no,
- struct odp_port *port);
+ struct dpif_port *port);
int (*port_query_by_name)(const struct dpif *dpif, const char *devname,
- struct odp_port *port);
+ struct dpif_port *port);
+
+ /* Returns one greater than the largest port number accepted in flow
+ * actions. */
+ int (*get_max_ports)(const struct dpif *dpif);
+
+ /* Attempts to begin dumping the ports in a dpif. On success, returns 0
+ * and initializes '*statep' with any data needed for iteration. On
+ * failure, returns a positive errno value. */
+ int (*port_dump_start)(const struct dpif *dpif, void **statep);
+
+ /* Attempts to retrieve another port from 'dpif' for 'state', which was
+ * initialized by a successful call to the 'port_dump_start' function for
+ * 'dpif'. On success, stores a new dpif_port into 'port' and returns 0.
+ * Returns EOF if the end of the port table has been reached, or a positive
+ * errno value on error. This function will not be called again once it
+ * returns nonzero once for a given iteration (but the 'port_dump_done'
+ * function will be called afterward).
+ *
+ * The dpif provider retains ownership of the data stored in 'port'. It
+ * must remain valid until at least the next call to 'port_dump_next' or
+ * 'port_dump_done' for 'state'. */
+ int (*port_dump_next)(const struct dpif *dpif, void *state,
+ struct dpif_port *port);
- /* Stores in 'ports' information about up to 'n' ports attached to 'dpif',
- * in no particular order. Returns the number of ports attached to 'dpif'
- * (not the number stored), if successful, otherwise a negative errno
- * value. */
- int (*port_list)(const struct dpif *dpif, struct odp_port *ports, int n);
+ /* Releases resources from 'dpif' for 'state', which was initialized by a
+ * successful call to the 'port_dump_start' function for 'dpif'. */
+ int (*port_dump_done)(const struct dpif *dpif, void *state);
/* Polls for changes in the set of ports in 'dpif'. If the set of ports in
* 'dpif' has changed, then this function should do one of the
* value other than EAGAIN. */
void (*port_poll_wait)(const struct dpif *dpif);
- /* Stores in 'ports' the port numbers of up to 'n' ports that belong to
- * 'group' in 'dpif'. Returns the number of ports in 'group' (not the
- * number stored), if successful, otherwise a negative errno value. */
- int (*port_group_get)(const struct dpif *dpif, int group,
- uint16_t ports[], int n);
-
- /* Changes port group 'group' in 'dpif' to consist of the 'n' ports whose
- * numbers are given in 'ports'.
- *
- * Use the get_stats member function to obtain the number of supported port
- * groups. */
- int (*port_group_set)(struct dpif *dpif, int group,
- const uint16_t ports[], int n);
-
/* For each flow 'flow' in the 'n' flows in 'flows':
*
* - If a flow matching 'flow->key' exists in 'dpif':
* packets. */
int (*flow_flush)(struct dpif *dpif);
- /* Stores up to 'n' flows in 'dpif' into 'flows', updating their statistics
- * and actions as described under the flow_get member function. If
- * successful, returns the number of flows actually present in 'dpif',
- * which might be greater than the number stored (if 'dpif' has more than
- * 'n' flows). On failure, returns a negative errno value. */
- int (*flow_list)(const struct dpif *dpif, struct odp_flow flows[], int n);
-
- /* Performs the 'n_actions' actions in 'actions' on the Ethernet frame
- * specified in 'packet'.
+ /* Attempts to begin dumping the flows in a dpif. On success, returns 0
+ * and initializes '*statep' with any data needed for iteration. On
+ * failure, returns a positive errno value. */
+ int (*flow_dump_start)(const struct dpif *dpif, void **statep);
+
+ /* Attempts to retrieve another flow from 'dpif' for 'state', which was
+ * initialized by a successful call to the 'flow_dump_start' function for
+ * 'dpif'. On success, stores a new odp_flow into 'flow' and returns 0.
+ * Returns EOF if the end of the flow table has been reached, or a positive
+ * errno value on error. This function will not be called again once it
+ * returns nonzero once for a given iteration (but the 'flow_dump_done'
+ * function will be called afterward).
*
- * Pretends that the frame was originally received on the port numbered
- * 'in_port'. This affects only ODPAT_OUTPUT_GROUP actions, which will not
- * send a packet out their input port. Specify the number of an unused
- * port (e.g. UINT16_MAX is currently always unused) to avoid this
- * behavior. */
- int (*execute)(struct dpif *dpif, uint16_t in_port,
- const union odp_action actions[], int n_actions,
- const struct ofpbuf *packet);
+ * Dumping flow actions is optional. If the caller does not want to dump
+ * actions it will initialize 'flow->actions' to NULL and
+ * 'flow->actions_len' to 0. Otherwise, 'flow->actions' points to an array
+ * of struct nlattr and 'flow->actions_len' contains the number of bytes of
+ * Netlink attributes. The implemention should fill in as many actions as
+ * will fit into the provided array and update 'flow->actions_len' with the
+ * number of bytes required (regardless of whether they fit in the provided
+ * space). */
+ int (*flow_dump_next)(const struct dpif *dpif, void *state,
+ struct odp_flow *flow);
+
+ /* Releases resources from 'dpif' for 'state', which was initialized by a
+ * successful call to the 'flow_dump_start' function for 'dpif'. */
+ int (*flow_dump_done)(const struct dpif *dpif, void *state);
+
+ /* Performs the 'actions_len' bytes of actions in 'actions' on the Ethernet
+ * frame specified in 'packet'. */
+ int (*execute)(struct dpif *dpif, const struct nlattr *actions,
+ size_t actions_len, const struct ofpbuf *packet);
/* Retrieves 'dpif''s "listen mask" into '*listen_mask'. Each ODPL_* bit
* set in '*listen_mask' indicates the 'dpif' will receive messages of the
* corresponding type when it calls the recv member function. */
int (*recv_set_mask)(struct dpif *dpif, int listen_mask);
- /* Attempts to receive a message from 'dpif'. If successful, stores the
- * message into '*packetp'. The message, if one is received, must begin
- * with 'struct odp_msg' as a header. Only messages of the types selected
- * with the set_listen_mask member function should be received.
+ /* Retrieves 'dpif''s sFlow sampling probability into '*probability'.
+ * Return value is 0 or a positive errno value. EOPNOTSUPP indicates that
+ * the datapath does not support sFlow, as does a null pointer.
+ *
+ * '*probability' is expressed as the number of packets out of UINT_MAX to
+ * sample, e.g. probability/UINT_MAX is the probability of sampling a given
+ * packet. */
+ int (*get_sflow_probability)(const struct dpif *dpif,
+ uint32_t *probability);
+
+ /* Sets 'dpif''s sFlow sampling probability to 'probability'. Return value
+ * is 0 or a positive errno value. EOPNOTSUPP indicates that the datapath
+ * does not support sFlow, as does a null pointer.
+ *
+ * 'probability' is expressed as the number of packets out of UINT_MAX to
+ * sample, e.g. probability/UINT_MAX is the probability of sampling a given
+ * packet. */
+ int (*set_sflow_probability)(struct dpif *dpif, uint32_t probability);
+
+ /* Translates OpenFlow queue ID 'queue_id' (in host byte order) into a
+ * priority value for use in the ODPAT_SET_PRIORITY action in
+ * '*priority'. */
+ int (*queue_to_priority)(const struct dpif *dpif, uint32_t queue_id,
+ uint32_t *priority);
+
+ /* Polls for an upcall from 'dpif'. If successful, stores the upcall into
+ * '*upcall'. Only upcalls of the types selected with the set_listen_mask
+ * member function should be received.
*
- * This function must not block. If no message is ready to be received
- * when it is called, it should return EAGAIN without blocking. */
- int (*recv)(struct dpif *dpif, struct ofpbuf **packetp);
+ * The caller takes ownership of the data that 'upcall' points to.
+ * 'upcall->key' and 'upcall->actions' (if nonnull) point into data owned
+ * by 'upcall->packet', so their memory cannot be freed separately. (This
+ * is hardly a great way to do things but it works out OK for the dpif
+ * providers that exist so far.)
+ *
+ * For greatest efficiency, 'upcall->packet' should have at least
+ * offsetof(struct ofp_packet_in, data) bytes of headroom.
+ *
+ * This function must not block. If no upcall is pending when it is
+ * called, it should return EAGAIN without blocking. */
+ int (*recv)(struct dpif *dpif, struct dpif_upcall *upcall);
/* Arranges for the poll loop to wake up when 'dpif' has a message queued
* to be received with the recv member function. */
void (*recv_wait)(struct dpif *dpif);
+
+ /* Throws away any queued upcalls that 'dpif' currently has ready to
+ * return. */
+ void (*recv_purge)(struct dpif *dpif);
};
extern const struct dpif_class dpif_linux_class;