skb_queue_head_init(&dp->queues[i]);
init_waitqueue_head(&dp->waitqueue);
- kobject_set_name(&dp->ifobj, SYSFS_BRIDGE_PORT_SUBDIR); /* "bridge" */
+ /* Initialize kobject for bridge. This will be added as
+ * /sys/class/net/<devname>/bridge later, if sysfs is enabled. */
- dp->ifobj.parent = NULL;
+ dp->ifobj.kset = NULL;
+ kobject_init(&dp->ifobj, &dp_ktype);
+
/* Allocate table. */
err = -ENOMEM;
rcu_assign_pointer(dp->table, dp_table_create(DP_L1_SIZE));
mutex_unlock(&dp_mutex);
rtnl_unlock();
- #ifdef SUPPORT_SYSFS
- if (dp_add_dp_hook)
- dp_add_dp_hook(dp);
+ dp_sysfs_add_dp(dp);
- #endif
return 0;
if (p->port_no != ODPP_LOCAL)
dp_del_port(p);
- #ifdef SUPPORT_SYSFS
- if (dp_del_dp_hook)
- dp_del_dp_hook(dp);
+ dp_sysfs_del_dp(dp);
- #endif
rcu_assign_pointer(dps[dp->dp_idx], NULL);
return err;
}
- #ifdef SUPPORT_SYSFS
+static void release_nbp(struct kobject *kobj)
+{
+ struct net_bridge_port *p = container_of(kobj, struct net_bridge_port, kobj);
+ kfree(p);
+}
+
+struct kobj_type brport_ktype = {
++#ifdef CONFIG_SYSFS
+ .sysfs_ops = &brport_sysfs_ops,
+#endif
+ .release = release_nbp
+};
+
/* Called with RTNL lock and dp_mutex. */
static int new_nbp(struct datapath *dp, struct net_device *dev, int port_no)
{
list_add_rcu(&p->node, &dp->port_list);
dp->n_ports++;
- kobject_set_name(&p->kobj, SYSFS_BRIDGE_PORT_ATTR); /* "brport" */
+ /* Initialize kobject for bridge. This will be added as
+ * /sys/class/net/<devname>/brport later, if sysfs is enabled. */
- p->kobj.parent = &p->dev->NETDEV_DEV_MEMBER.kobj;
+ p->kobj.kset = NULL;
+ kobject_init(&p->kobj, &brport_ktype);
+
dp_ifinfo_notify(RTM_NEWLINK, p);
return 0;
if (err)
goto out_put;
- #ifdef SUPPORT_SYSFS
- if (dp_add_if_hook)
- dp_add_if_hook(dp->ports[port_no]);
+ dp_sysfs_add_if(dp->ports[port_no]);
- #endif
+
+ err = __put_user(port_no, &port.port);
out_put:
dev_put(dev);
{
ASSERT_RTNL();
- #ifdef SUPPORT_SYSFS
- if (p->port_no != ODPP_LOCAL && dp_del_if_hook) {
-#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,25)
- sysfs_remove_link(&p->dp->ifobj, p->dev->name);
-#else
- sysfs_remove_link(p->dp->ifobj, p->dev->name);
-#endif
- }
+ if (p->port_no != ODPP_LOCAL)
+ dp_sysfs_del_if(p);
- #endif
dp_ifinfo_notify(RTM_DELLINK, p);
p->dp->n_ports--;
#include <linux/netdevice.h>
#include <linux/workqueue.h>
#include <linux/skbuff.h>
+ #include <linux/version.h>
#include "flow.h"
-#include "brc_sysfs.h"
+#include "dp_sysfs.h"
/* Mask for the priority bits in a vlan header. If we ever merge upstream
* then this should go into include/linux/if_vlan.h. */
--- /dev/null
- #include <linux/version.h>
- #if LINUX_VERSION_CODE == KERNEL_VERSION(2,6,18)
- #define SUPPORT_SYSFS 1
- #else
- /* We only support sysfs on Linux 2.6.18 because that's the only place we
- * really need it (on Xen, for brcompat) and it's a big pain to try to support
- * multiple versions. */
- #endif
-
- #ifdef SUPPORT_SYSFS
+/*
+ * Copyright (c) 2009 Nicira Networks.
+ * Distributed under the terms of the GNU GPL version 2.
+ *
+ * Significant portions of this file may be copied from parts of the Linux
+ * kernel, by Linus Torvalds and others.
+ */
+
+#ifndef DP_SYSFS_H
+#define DP_SYSFS_H 1
+
+struct datapath;
+struct net_bridge_port;
+
+/* dp_sysfs_dp.c */
+int dp_sysfs_add_dp(struct datapath *dp);
+int dp_sysfs_del_dp(struct datapath *dp);
+
+/* dp_sysfs_if.c */
+int dp_sysfs_add_if(struct net_bridge_port *p);
+int dp_sysfs_del_if(struct net_bridge_port *p);
+
++#ifdef CONFIG_SYSFS
+extern struct sysfs_ops brport_sysfs_ops;
+#endif
+
+#endif /* dp_sysfs.h */
+
--- /dev/null
- #ifdef SUPPORT_SYSFS
+/*
+ * Copyright (c) 2009 Nicira Networks.
+ * Distributed under the terms of the GNU GPL version 2.
+ *
+ * Significant portions of this file may be copied from parts of the Linux
+ * kernel, by Linus Torvalds and others.
+ */
+
+#include <linux/version.h>
+
+/*
+ * Sysfs attributes of bridge for Open vSwitch
+ *
+ * This has been shamelessly copied from the kernel sources.
+ */
+
+#include <linux/capability.h>
+#include <linux/device.h>
+#include <linux/kernel.h>
+#include <linux/netdevice.h>
+#include <linux/if_bridge.h>
+#include <linux/rtnetlink.h>
+#include <linux/spinlock.h>
+#include <linux/times.h>
+#include <linux/version.h>
+
+#include "dp_sysfs.h"
+#include "datapath.h"
+#include "dp_dev.h"
+
- #define to_kobj(d) &(d)->class_dev.kobj
++#ifdef CONFIG_SYSFS
+#define to_dev(obj) container_of(obj, struct device, kobj)
+
+/* Hack to attempt to build on more platforms. */
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,21)
- #define to_kobj(d) &(d)->dev.kobj
+#define DP_DEVICE_ATTR CLASS_DEVICE_ATTR
++#define DEVICE_PARAMS struct class_device *d
++#define DEVICE_ARGS d
++#define DEV_ATTR(NAME) class_device_attr_##NAME
+#else
- static ssize_t store_bridge_parm(struct class_device *d,
+#define DP_DEVICE_ATTR DEVICE_ATTR
++#define DEVICE_PARAMS struct device *d, struct device_attribute *attr
++#define DEVICE_ARGS d, attr
++#define DEV_ATTR(NAME) dev_attr_##NAME
+#endif
+
+/*
+ * Common code for storing bridge parameters.
+ */
- static ssize_t show_forward_delay(struct class_device *d,
- char *buf)
++static ssize_t store_bridge_parm(DEVICE_PARAMS,
+ const char *buf, size_t len,
+ void (*set)(struct datapath *, unsigned long))
+{
+ struct datapath *dp = dp_dev_get_dp(to_net_dev(d));
+ char *endp;
+ unsigned long val;
+
+ if (!capable(CAP_NET_ADMIN))
+ return -EPERM;
+
+ val = simple_strtoul(buf, &endp, 0);
+ if (endp == buf)
+ return -EINVAL;
+
+#if 0
+ spin_lock_bh(&br->lock);
+ (*set)(br, val);
+ spin_unlock_bh(&br->lock);
+#else
+ /* xxx We use a default value of 0 for all fields. If the caller is
+ * xxx attempting to set the value to our default, just silently
+ * xxx ignore the request.
+ */
+ if (val != 0) {
+ printk("%s: xxx writing dp parms not supported yet!\n",
+ dp_name(dp));
+ }
+#endif
+ return len;
+}
+
+
- static ssize_t store_forward_delay(struct class_device *d,
++static ssize_t show_forward_delay(DEVICE_PARAMS, char *buf)
+{
+#if 0
+ struct datapath *dp = dp_dev_get_dp(to_net_dev(d));
+ return sprintf(buf, "%lu\n", jiffies_to_clock_t(br->forward_delay));
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+
+static void set_forward_delay(struct datapath *dp, unsigned long val)
+{
+#if 0
+ unsigned long delay = clock_t_to_jiffies(val);
+ br->forward_delay = delay;
+ if (br_is_root_bridge(br))
+ br->bridge_forward_delay = delay;
+#else
+ printk("%s: xxx attempt to set_forward_delay()\n", dp_name(dp));
+#endif
+}
+
- return store_bridge_parm(d, buf, len, set_forward_delay);
++static ssize_t store_forward_delay(DEVICE_PARAMS,
+ const char *buf, size_t len)
+{
- static ssize_t show_hello_time(struct class_device *d, char *buf)
++ return store_bridge_parm(DEVICE_ARGS, buf, len, set_forward_delay);
+}
+static DP_DEVICE_ATTR(forward_delay, S_IRUGO | S_IWUSR,
+ show_forward_delay, store_forward_delay);
+
- static ssize_t store_hello_time(struct class_device *d,
++static ssize_t show_hello_time(DEVICE_PARAMS, char *buf)
+{
+#if 0
+ return sprintf(buf, "%lu\n",
+ jiffies_to_clock_t(to_bridge(d)->hello_time));
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+
+static void set_hello_time(struct datapath *dp, unsigned long val)
+{
+#if 0
+ unsigned long t = clock_t_to_jiffies(val);
+ br->hello_time = t;
+ if (br_is_root_bridge(br))
+ br->bridge_hello_time = t;
+#else
+ printk("%s: xxx attempt to set_hello_time()\n", dp_name(dp));
+#endif
+}
+
- return store_bridge_parm(d, buf, len, set_hello_time);
++static ssize_t store_hello_time(DEVICE_PARAMS,
+ const char *buf,
+ size_t len)
+{
- static ssize_t show_max_age(struct class_device *d,
- char *buf)
++ return store_bridge_parm(DEVICE_ARGS, buf, len, set_hello_time);
+}
+static DP_DEVICE_ATTR(hello_time, S_IRUGO | S_IWUSR, show_hello_time,
+ store_hello_time);
+
- static ssize_t store_max_age(struct class_device *d,
++static ssize_t show_max_age(DEVICE_PARAMS, char *buf)
+{
+#if 0
+ return sprintf(buf, "%lu\n",
+ jiffies_to_clock_t(to_bridge(d)->max_age));
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+
+static void set_max_age(struct datapath *dp, unsigned long val)
+{
+#if 0
+ unsigned long t = clock_t_to_jiffies(val);
+ br->max_age = t;
+ if (br_is_root_bridge(br))
+ br->bridge_max_age = t;
+#else
+ printk("%s: xxx attempt to set_max_age()\n", dp_name(dp));
+#endif
+}
+
- return store_bridge_parm(d, buf, len, set_max_age);
++static ssize_t store_max_age(DEVICE_PARAMS,
+ const char *buf, size_t len)
+{
- static ssize_t show_ageing_time(struct class_device *d,
- char *buf)
++ return store_bridge_parm(DEVICE_ARGS, buf, len, set_max_age);
+}
+static DP_DEVICE_ATTR(max_age, S_IRUGO | S_IWUSR, show_max_age, store_max_age);
+
- static ssize_t store_ageing_time(struct class_device *d,
++static ssize_t show_ageing_time(DEVICE_PARAMS, char *buf)
+{
+#if 0
+ struct datapath *dp = dp_dev_get_dp(to_net_dev(d));
+ return sprintf(buf, "%lu\n", jiffies_to_clock_t(br->ageing_time));
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+
+static void set_ageing_time(struct datapath *dp, unsigned long val)
+{
+#if 0
+ br->ageing_time = clock_t_to_jiffies(val);
+#else
+ printk("%s: xxx attempt to set_ageing_time()\n", dp_name(dp));
+#endif
+}
+
- return store_bridge_parm(d, buf, len, set_ageing_time);
++static ssize_t store_ageing_time(DEVICE_PARAMS,
+ const char *buf, size_t len)
+{
- static ssize_t show_stp_state(struct class_device *d,
- char *buf)
++ return store_bridge_parm(DEVICE_ARGS, buf, len, set_ageing_time);
+}
+static DP_DEVICE_ATTR(ageing_time, S_IRUGO | S_IWUSR, show_ageing_time,
+ store_ageing_time);
+
- static ssize_t store_stp_state(struct class_device *d,
++static ssize_t show_stp_state(DEVICE_PARAMS, char *buf)
+{
+#if 0
+ struct datapath *dp = dp_dev_get_dp(to_net_dev(d));
+ return sprintf(buf, "%d\n", br->stp_enabled);
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+
+
- static ssize_t show_priority(struct class_device *d,
- char *buf)
++static ssize_t store_stp_state(DEVICE_PARAMS,
+ const char *buf,
+ size_t len)
+{
+ struct datapath *dp = dp_dev_get_dp(to_net_dev(d));
+#if 0
+ char *endp;
+ unsigned long val;
+
+ if (!capable(CAP_NET_ADMIN))
+ return -EPERM;
+
+ val = simple_strtoul(buf, &endp, 0);
+ if (endp == buf)
+ return -EINVAL;
+
+ rtnl_lock();
+ br_stp_set_enabled(br, val);
+ rtnl_unlock();
+#else
+ printk("%s: xxx attempt to set_stp_state()\n", dp_name(dp));
+#endif
+
+ return len;
+}
+static DP_DEVICE_ATTR(stp_state, S_IRUGO | S_IWUSR, show_stp_state,
+ store_stp_state);
+
- static ssize_t store_priority(struct class_device *d,
++static ssize_t show_priority(DEVICE_PARAMS, char *buf)
+{
+#if 0
+ struct datapath *dp = dp_dev_get_dp(to_net_dev(d));
+ return sprintf(buf, "%d\n",
+ (br->bridge_id.prio[0] << 8) | br->bridge_id.prio[1]);
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+
+static void set_priority(struct datapath *dp, unsigned long val)
+{
+#if 0
+ br_stp_set_bridge_priority(br, (u16) val);
+#else
+ printk("%s: xxx attempt to set_priority()\n", dp_name(dp));
+#endif
+}
+
- return store_bridge_parm(d, buf, len, set_priority);
++static ssize_t store_priority(DEVICE_PARAMS,
+ const char *buf, size_t len)
+{
- static ssize_t show_root_id(struct class_device *d,
- char *buf)
++ return store_bridge_parm(DEVICE_ARGS, buf, len, set_priority);
+}
+static DP_DEVICE_ATTR(priority, S_IRUGO | S_IWUSR, show_priority, store_priority);
+
- static ssize_t show_bridge_id(struct class_device *d,
- char *buf)
++static ssize_t show_root_id(DEVICE_PARAMS, char *buf)
+{
+#if 0
+ return br_show_bridge_id(buf, &to_bridge(d)->designated_root);
+#else
+ return sprintf(buf, "0000.010203040506\n");
+#endif
+}
+static DP_DEVICE_ATTR(root_id, S_IRUGO, show_root_id, NULL);
+
- static ssize_t show_root_port(struct class_device *d,
- char *buf)
++static ssize_t show_bridge_id(DEVICE_PARAMS, char *buf)
+{
+ struct datapath *dp = dp_dev_get_dp(to_net_dev(d));
+ const unsigned char *addr = dp->ports[ODPP_LOCAL]->dev->dev_addr;
+
+ /* xxx Do we need a lock of some sort? */
+ return sprintf(buf, "%.2x%.2x.%.2x%.2x%.2x%.2x%.2x%.2x\n",
+ 0, 0, addr[0], addr[1], addr[2], addr[3], addr[4], addr[5]);
+}
+static DP_DEVICE_ATTR(bridge_id, S_IRUGO, show_bridge_id, NULL);
+
- static ssize_t show_root_path_cost(struct class_device *d,
- char *buf)
++static ssize_t show_root_port(DEVICE_PARAMS, char *buf)
+{
+#if 0
+ return sprintf(buf, "%d\n", to_bridge(d)->root_port);
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static DP_DEVICE_ATTR(root_port, S_IRUGO, show_root_port, NULL);
+
- static ssize_t show_topology_change(struct class_device *d,
- char *buf)
++static ssize_t show_root_path_cost(DEVICE_PARAMS, char *buf)
+{
+#if 0
+ return sprintf(buf, "%d\n", to_bridge(d)->root_path_cost);
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static DP_DEVICE_ATTR(root_path_cost, S_IRUGO, show_root_path_cost, NULL);
+
- static ssize_t show_topology_change_detected(struct class_device *d,
- char *buf)
++static ssize_t show_topology_change(DEVICE_PARAMS, char *buf)
+{
+#if 0
+ return sprintf(buf, "%d\n", to_bridge(d)->topology_change);
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static DP_DEVICE_ATTR(topology_change, S_IRUGO, show_topology_change, NULL);
+
- static ssize_t show_hello_timer(struct class_device *d,
- char *buf)
++static ssize_t show_topology_change_detected(DEVICE_PARAMS, char *buf)
+{
+#if 0
+ struct datapath *dp = dp_dev_get_dp(to_net_dev(d));
+ return sprintf(buf, "%d\n", br->topology_change_detected);
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static DP_DEVICE_ATTR(topology_change_detected, S_IRUGO,
+ show_topology_change_detected, NULL);
+
- static ssize_t show_tcn_timer(struct class_device *d,
- char *buf)
++static ssize_t show_hello_timer(DEVICE_PARAMS, char *buf)
+{
+#if 0
+ struct datapath *dp = dp_dev_get_dp(to_net_dev(d));
+ return sprintf(buf, "%ld\n", br_timer_value(&br->hello_timer));
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static DP_DEVICE_ATTR(hello_timer, S_IRUGO, show_hello_timer, NULL);
+
- static ssize_t show_topology_change_timer(struct class_device *d,
- char *buf)
++static ssize_t show_tcn_timer(DEVICE_PARAMS, char *buf)
+{
+#if 0
+ struct datapath *dp = dp_dev_get_dp(to_net_dev(d));
+ return sprintf(buf, "%ld\n", br_timer_value(&br->tcn_timer));
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static DP_DEVICE_ATTR(tcn_timer, S_IRUGO, show_tcn_timer, NULL);
+
- static ssize_t show_gc_timer(struct class_device *d,
- char *buf)
++static ssize_t show_topology_change_timer(DEVICE_PARAMS, char *buf)
+{
+#if 0
+ struct datapath *dp = dp_dev_get_dp(to_net_dev(d));
+ return sprintf(buf, "%ld\n", br_timer_value(&br->topology_change_timer));
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static DP_DEVICE_ATTR(topology_change_timer, S_IRUGO, show_topology_change_timer,
+ NULL);
+
- static ssize_t show_group_addr(struct class_device *d,
- char *buf)
++static ssize_t show_gc_timer(DEVICE_PARAMS, char *buf)
+{
+#if 0
+ struct datapath *dp = dp_dev_get_dp(to_net_dev(d));
+ return sprintf(buf, "%ld\n", br_timer_value(&br->gc_timer));
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static DP_DEVICE_ATTR(gc_timer, S_IRUGO, show_gc_timer, NULL);
+
- static ssize_t store_group_addr(struct class_device *d,
++static ssize_t show_group_addr(DEVICE_PARAMS, char *buf)
+{
+#if 0
+ struct datapath *dp = dp_dev_get_dp(to_net_dev(d));
+ return sprintf(buf, "%x:%x:%x:%x:%x:%x\n",
+ br->group_addr[0], br->group_addr[1],
+ br->group_addr[2], br->group_addr[3],
+ br->group_addr[4], br->group_addr[5]);
+#else
+ return sprintf(buf, "00:01:02:03:04:05\n");
+#endif
+}
+
- &class_device_attr_forward_delay.attr,
- &class_device_attr_hello_time.attr,
- &class_device_attr_max_age.attr,
- &class_device_attr_ageing_time.attr,
- &class_device_attr_stp_state.attr,
- &class_device_attr_priority.attr,
- &class_device_attr_bridge_id.attr,
- &class_device_attr_root_id.attr,
- &class_device_attr_root_path_cost.attr,
- &class_device_attr_root_port.attr,
- &class_device_attr_topology_change.attr,
- &class_device_attr_topology_change_detected.attr,
- &class_device_attr_hello_timer.attr,
- &class_device_attr_tcn_timer.attr,
- &class_device_attr_topology_change_timer.attr,
- &class_device_attr_gc_timer.attr,
- &class_device_attr_group_addr.attr,
++static ssize_t store_group_addr(DEVICE_PARAMS,
+ const char *buf, size_t len)
+{
+ struct datapath *dp = dp_dev_get_dp(to_net_dev(d));
+#if 0
+ unsigned new_addr[6];
+ int i;
+
+ if (!capable(CAP_NET_ADMIN))
+ return -EPERM;
+
+ if (sscanf(buf, "%x:%x:%x:%x:%x:%x",
+ &new_addr[0], &new_addr[1], &new_addr[2],
+ &new_addr[3], &new_addr[4], &new_addr[5]) != 6)
+ return -EINVAL;
+
+ /* Must be 01:80:c2:00:00:0X */
+ for (i = 0; i < 5; i++)
+ if (new_addr[i] != br_group_address[i])
+ return -EINVAL;
+
+ if (new_addr[5] & ~0xf)
+ return -EINVAL;
+
+ if (new_addr[5] == 1 /* 802.3x Pause address */
+ || new_addr[5] == 2 /* 802.3ad Slow protocols */
+ || new_addr[5] == 3) /* 802.1X PAE address */
+ return -EINVAL;
+
+ spin_lock_bh(&br->lock);
+ for (i = 0; i < 6; i++)
+ br->group_addr[i] = new_addr[i];
+ spin_unlock_bh(&br->lock);
+#else
+ printk("%s: xxx attempt to store_group_addr()\n", dp_name(dp));
+#endif
+ return len;
+}
+
+static DP_DEVICE_ATTR(group_addr, S_IRUGO | S_IWUSR,
+ show_group_addr, store_group_addr);
+
+static struct attribute *bridge_attrs[] = {
- struct kobject *kobj = to_kobj(dp->ports[ODPP_LOCAL]->dev);
++ &DEV_ATTR(forward_delay).attr,
++ &DEV_ATTR(hello_time).attr,
++ &DEV_ATTR(max_age).attr,
++ &DEV_ATTR(ageing_time).attr,
++ &DEV_ATTR(stp_state).attr,
++ &DEV_ATTR(priority).attr,
++ &DEV_ATTR(bridge_id).attr,
++ &DEV_ATTR(root_id).attr,
++ &DEV_ATTR(root_path_cost).attr,
++ &DEV_ATTR(root_port).attr,
++ &DEV_ATTR(topology_change).attr,
++ &DEV_ATTR(topology_change_detected).attr,
++ &DEV_ATTR(hello_timer).attr,
++ &DEV_ATTR(tcn_timer).attr,
++ &DEV_ATTR(topology_change_timer).attr,
++ &DEV_ATTR(gc_timer).attr,
++ &DEV_ATTR(group_addr).attr,
+ NULL
+};
+
+static struct attribute_group bridge_group = {
+ .name = SYSFS_BRIDGE_ATTR,
+ .attrs = bridge_attrs,
+};
+
+/*
+ * Add entries in sysfs onto the existing network class device
+ * for the bridge.
+ * Adds a attribute group "bridge" containing tuning parameters.
+ * Sub directory to hold links to interfaces.
+ *
+ * Note: the ifobj exists only to be a subdirectory
+ * to hold links. The ifobj exists in the same data structure
+ * as its parent the bridge so reference counting works.
+ */
+int dp_sysfs_add_dp(struct datapath *dp)
+{
- /* Create /sys/class/net/<devname>/bridge directory. */
- dp->ifobj.parent = kobj;
- err = kobject_add(&dp->ifobj);
++ struct kobject *kobj = &dp->ports[ODPP_LOCAL]->dev->NETDEV_DEV_MEMBER.kobj;
+ int err;
+
+ err = sysfs_create_group(kobj, &bridge_group);
+ if (err) {
+ pr_info("%s: can't create group %s/%s\n",
+ __func__, dp_name(dp), bridge_group.name);
+ goto out1;
+ }
+
- __FUNCTION__, dp_name(dp), dp->ifobj.name);
++ /* Create /sys/class/net/<devname>/brif directory. */
++ err = kobject_add(&dp->ifobj, kobj, SYSFS_BRIDGE_PORT_SUBDIR);
+ if (err) {
+ pr_info("%s: can't add kobject (directory) %s/%s\n",
- struct kobject *kobj = to_kobj(dp->ports[ODPP_LOCAL]->dev);
++ __FUNCTION__, dp_name(dp), kobject_name(&dp->ifobj));
+ goto out2;
+ }
+ kobject_uevent(&dp->ifobj, KOBJ_ADD);
+ return 0;
+
+ out2:
+ sysfs_remove_group(kobj, &bridge_group);
+ out1:
+ return err;
+}
+
+int dp_sysfs_del_dp(struct datapath *dp)
+{
- #else /* !SUPPORT_SYSFS */
++ struct kobject *kobj = &dp->ports[ODPP_LOCAL]->dev->NETDEV_DEV_MEMBER.kobj;
+
+ kobject_del(&dp->ifobj);
+ sysfs_remove_group(kobj, &bridge_group);
+
+ return 0;
+}
- int dp_sysfs_del_if(struct net_bridge_port *p)
- {
- dev_put(p->dev);
- kfree(p);
- return 0;
- }
- #endif /* !SUPPORT_SYSFS */
++#else /* !CONFIG_SYSFS */
+int dp_sysfs_add_dp(struct datapath *dp) { return 0; }
+int dp_sysfs_del_dp(struct datapath *dp) { return 0; }
+int dp_sysfs_add_if(struct net_bridge_port *p) { return 0; }
++int dp_sysfs_del_if(struct net_bridge_port *p) { return 0; }
++#endif /* !CONFIG_SYSFS */
--- /dev/null
- #ifdef SUPPORT_SYSFS
+/*
+ * Copyright (c) 2009 Nicira Networks.
+ * Distributed under the terms of the GNU GPL version 2.
+ *
+ * Significant portions of this file may be copied from parts of the Linux
+ * kernel, by Linus Torvalds and others.
+ */
+
+/*
+ * Sysfs attributes of bridge ports for Open vSwitch
+ *
+ * This has been shamelessly copied from the kernel sources.
+ */
+
+#include <linux/capability.h>
+#include <linux/kernel.h>
+#include <linux/netdevice.h>
+#include <linux/if_bridge.h>
+#include <linux/rtnetlink.h>
+#include <linux/spinlock.h>
+#include "dp_sysfs.h"
+#include "datapath.h"
+
- err = kobject_add(&p->kobj);
++#ifdef CONFIG_SYSFS
+
+struct brport_attribute {
+ struct attribute attr;
+ ssize_t (*show)(struct net_bridge_port *, char *);
+ ssize_t (*store)(struct net_bridge_port *, unsigned long);
+};
+
+#define BRPORT_ATTR(_name,_mode,_show,_store) \
+struct brport_attribute brport_attr_##_name = { \
+ .attr = {.name = __stringify(_name), \
+ .mode = _mode, \
+ .owner = THIS_MODULE, }, \
+ .show = _show, \
+ .store = _store, \
+};
+
+static ssize_t show_path_cost(struct net_bridge_port *p, char *buf)
+{
+#if 0
+ return sprintf(buf, "%d\n", p->path_cost);
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static ssize_t store_path_cost(struct net_bridge_port *p, unsigned long v)
+{
+#if 0
+ br_stp_set_path_cost(p, v);
+#endif
+ return 0;
+}
+static BRPORT_ATTR(path_cost, S_IRUGO | S_IWUSR,
+ show_path_cost, store_path_cost);
+
+static ssize_t show_priority(struct net_bridge_port *p, char *buf)
+{
+#if 0
+ return sprintf(buf, "%d\n", p->priority);
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static ssize_t store_priority(struct net_bridge_port *p, unsigned long v)
+{
+#if 0
+ if (v >= (1<<(16-BR_PORT_BITS)))
+ return -ERANGE;
+ br_stp_set_port_priority(p, v);
+#endif
+ return 0;
+}
+static BRPORT_ATTR(priority, S_IRUGO | S_IWUSR,
+ show_priority, store_priority);
+
+static ssize_t show_designated_root(struct net_bridge_port *p, char *buf)
+{
+#if 0
+ return br_show_bridge_id(buf, &p->designated_root);
+#else
+ return sprintf(buf, "0000.010203040506\n");
+#endif
+}
+static BRPORT_ATTR(designated_root, S_IRUGO, show_designated_root, NULL);
+
+static ssize_t show_designated_bridge(struct net_bridge_port *p, char *buf)
+{
+#if 0
+ return br_show_bridge_id(buf, &p->designated_bridge);
+#else
+ return sprintf(buf, "0000.060504030201\n");
+#endif
+}
+static BRPORT_ATTR(designated_bridge, S_IRUGO, show_designated_bridge, NULL);
+
+static ssize_t show_designated_port(struct net_bridge_port *p, char *buf)
+{
+#if 0
+ return sprintf(buf, "%d\n", p->designated_port);
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static BRPORT_ATTR(designated_port, S_IRUGO, show_designated_port, NULL);
+
+static ssize_t show_designated_cost(struct net_bridge_port *p, char *buf)
+{
+#if 0
+ return sprintf(buf, "%d\n", p->designated_cost);
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static BRPORT_ATTR(designated_cost, S_IRUGO, show_designated_cost, NULL);
+
+static ssize_t show_port_id(struct net_bridge_port *p, char *buf)
+{
+#if 0
+ return sprintf(buf, "0x%x\n", p->port_id);
+#else
+ return sprintf(buf, "0x%x\n", 0);
+#endif
+}
+static BRPORT_ATTR(port_id, S_IRUGO, show_port_id, NULL);
+
+static ssize_t show_port_no(struct net_bridge_port *p, char *buf)
+{
+ return sprintf(buf, "0x%x\n", p->port_no);
+}
+
+static BRPORT_ATTR(port_no, S_IRUGO, show_port_no, NULL);
+
+static ssize_t show_change_ack(struct net_bridge_port *p, char *buf)
+{
+#if 0
+ return sprintf(buf, "%d\n", p->topology_change_ack);
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static BRPORT_ATTR(change_ack, S_IRUGO, show_change_ack, NULL);
+
+static ssize_t show_config_pending(struct net_bridge_port *p, char *buf)
+{
+#if 0
+ return sprintf(buf, "%d\n", p->config_pending);
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static BRPORT_ATTR(config_pending, S_IRUGO, show_config_pending, NULL);
+
+static ssize_t show_port_state(struct net_bridge_port *p, char *buf)
+{
+#if 0
+ return sprintf(buf, "%d\n", p->state);
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static BRPORT_ATTR(state, S_IRUGO, show_port_state, NULL);
+
+static ssize_t show_message_age_timer(struct net_bridge_port *p,
+ char *buf)
+{
+#if 0
+ return sprintf(buf, "%ld\n", br_timer_value(&p->message_age_timer));
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static BRPORT_ATTR(message_age_timer, S_IRUGO, show_message_age_timer, NULL);
+
+static ssize_t show_forward_delay_timer(struct net_bridge_port *p,
+ char *buf)
+{
+#if 0
+ return sprintf(buf, "%ld\n", br_timer_value(&p->forward_delay_timer));
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static BRPORT_ATTR(forward_delay_timer, S_IRUGO, show_forward_delay_timer, NULL);
+
+static ssize_t show_hold_timer(struct net_bridge_port *p,
+ char *buf)
+{
+#if 0
+ return sprintf(buf, "%ld\n", br_timer_value(&p->hold_timer));
+#else
+ return sprintf(buf, "%d\n", 0);
+#endif
+}
+static BRPORT_ATTR(hold_timer, S_IRUGO, show_hold_timer, NULL);
+
+static struct brport_attribute *brport_attrs[] = {
+ &brport_attr_path_cost,
+ &brport_attr_priority,
+ &brport_attr_port_id,
+ &brport_attr_port_no,
+ &brport_attr_designated_root,
+ &brport_attr_designated_bridge,
+ &brport_attr_designated_port,
+ &brport_attr_designated_cost,
+ &brport_attr_state,
+ &brport_attr_change_ack,
+ &brport_attr_config_pending,
+ &brport_attr_message_age_timer,
+ &brport_attr_forward_delay_timer,
+ &brport_attr_hold_timer,
+ NULL
+};
+
+#define to_brport_attr(_at) container_of(_at, struct brport_attribute, attr)
+#define to_brport(obj) container_of(obj, struct net_bridge_port, kobj)
+
+static ssize_t brport_show(struct kobject * kobj,
+ struct attribute * attr, char * buf)
+{
+ struct brport_attribute * brport_attr = to_brport_attr(attr);
+ struct net_bridge_port * p = to_brport(kobj);
+
+ return brport_attr->show(p, buf);
+}
+
+static ssize_t brport_store(struct kobject * kobj,
+ struct attribute * attr,
+ const char * buf, size_t count)
+{
+ struct net_bridge_port * p = to_brport(kobj);
+#if 0
+ struct brport_attribute * brport_attr = to_brport_attr(attr);
+ char *endp;
+ unsigned long val;
+#endif
+ ssize_t ret = -EINVAL;
+
+ if (!capable(CAP_NET_ADMIN))
+ return -EPERM;
+
+#if 0
+ val = simple_strtoul(buf, &endp, 0);
+ if (endp != buf) {
+ rtnl_lock();
+ if (p->dev && p->br && brport_attr->store) {
+ spin_lock_bh(&p->br->lock);
+ ret = brport_attr->store(p, val);
+ spin_unlock_bh(&p->br->lock);
+ if (ret == 0)
+ ret = count;
+ }
+ rtnl_unlock();
+ }
+#else
+ printk("%s: xxx writing port parms not supported yet!\n",
+ dp_name(p->dp));
+#endif
+ return ret;
+}
+
+struct sysfs_ops brport_sysfs_ops = {
+ .show = brport_show,
+ .store = brport_store,
+};
+
+/*
+ * Add sysfs entries to ethernet device added to a bridge.
+ * Creates a brport subdirectory with bridge attributes.
+ * Puts symlink in bridge's brport subdirectory
+ */
+int dp_sysfs_add_if(struct net_bridge_port *p)
+{
+ struct datapath *dp = p->dp;
+ struct brport_attribute **a;
+ int err;
+
+ /* Create /sys/class/net/<devname>/brport directory. */
- goto err_put;
++ err = kobject_add(&p->kobj, &p->dev->NETDEV_DEV_MEMBER.kobj,
++ SYSFS_BRIDGE_PORT_ATTR);
+ if (err)
- return err;
++ goto err;
+
+ /* Create symlink from /sys/class/net/<devname>/brport/bridge to
+ * /sys/class/net/<bridgename>. */
+ err = sysfs_create_link(&p->kobj,
+ &dp->ports[ODPP_LOCAL]->dev->NETDEV_DEV_MEMBER.kobj,
+ SYSFS_BRIDGE_PORT_LINK); /* "bridge" */
+ if (err)
+ goto err_del;
+
+ /* Populate /sys/class/net/<devname>/brport directory with files. */
+ for (a = brport_attrs; *a; ++a) {
+ err = sysfs_create_file(&p->kobj, &((*a)->attr));
+ if (err)
+ goto err_del;
+ }
+
+ /* Create symlink from /sys/class/net/<bridgename>/brif/<devname> to
+ * /sys/class/net/<devname>/brport. */
+ err = sysfs_create_link(&dp->ifobj, &p->kobj, p->dev->name);
+ if (err)
+ goto err_del;
+ strcpy(p->linkname, p->dev->name);
+
+ kobject_uevent(&p->kobj, KOBJ_ADD);
+
- err_put:
- kobject_put(&p->kobj);
-
- /* Ensure that dp_sysfs_del_if becomes a no-op. */
- p->kobj.dentry = NULL;
++ return 0;
+
+err_del:
+ kobject_del(&p->kobj);
- p->linkname[0] = '\0';
- }
- if (p->kobj.dentry) {
++err:
++ p->linkname[0] = 0;
+ return err;
+}
+
+int dp_sysfs_del_if(struct net_bridge_port *p)
+{
+ if (p->linkname[0]) {
+ sysfs_remove_link(&p->dp->ifobj, p->linkname);
- #endif /* SUPPORT_SYSFS */
+ kobject_uevent(&p->kobj, KOBJ_REMOVE);
+ kobject_del(&p->kobj);
++ p->linkname[0] = '\0';
+ }
+ return 0;
+}
++#endif /* CONFIG_SYSFS */
#include_next <linux/kobject.h>
#include <linux/version.h>
++
#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,25)
-static inline int kobject_init_and_add(struct kobject *kobj,
- struct kobj_type *ktype,
- struct kobject *parent,
- const char *name)
+#define kobject_init(kobj, ktype) rpl_kobject_init(kobj, ktype)
+static inline void rpl_kobject_init(struct kobject *kobj, struct kobj_type *ktype)
{
- kobject_init(kobj);
- kobject_set_name(kobj, "%s", name);
kobj->ktype = ktype;
- kobj->kset = NULL;
- kobj->parent = parent;
+ (kobject_init)(kobj);
+}
+
- return kobject_add(kobj);
++#define kobject_add(kobj, parent, name) rpl_kobject_add(kobj, parent, name)
++static inline int rpl_kobject_add(struct kobject *kobj,
++ struct kobject *parent,
++ const char *name)
++{
++ int err = kobject_set_name(kobj, "%s", name);
++ if (err)
++ return err;
++ kobj->parent = parent;
++ return (kobject_add)(kobj);
+ }
#endif
++
#endif /* linux/kobject.h wrapper */
#RATE_LIMIT=1000
# INACTIVITY_PROBE: The maximum number of seconds of inactivity on the
- # controller connection before secchan sends an inactivity probe
+ # controller connection before ovs-openflowd sends an inactivity probe
# message to the controller. The valid range is 5 and up. If unset,
- # secchan defaults to 5 seconds.
-# ovs-openflowd defaults to 15 seconds.
++# ovs-openflowd defaults to 5 seconds.
#INACTIVITY_PROBE=5
- # MAX_BACKOFF: The maximum time that secchan will wait between
+ # MAX_BACKOFF: The maximum time that ovs-openflowd will wait between
# attempts to connect to the controller. The valid range is 1 and up.
- # If unset, secchan defaults to 8 seconds.
-# If unset, ovs-openflowd defaults to 15 seconds.
-#MAX_BACKOFF=15
++# If unset, ovs-openflowd defaults to 8 seconds.
+#MAX_BACKOFF=8
- # DAEMON_OPTS: Additional options to pass to secchan, e.g. "--fail=open"
+ # DAEMON_OPTS: Additional options to pass to ovs-openflowd, e.g. "--fail=open"
DAEMON_OPTS=""
# CORE_LIMIT: Maximum size for core dumps.
flow_extract(&b, 0, &flow);
if (flow.dl_type != htons(ETH_TYPE_IP)
|| flow.nw_proto != IP_TYPE_UDP
- || flow.tp_dst != htons(68)
+ || flow.tp_dst != htons(DHCP_CLIENT_PORT)
|| !(eth_addr_is_broadcast(flow.dl_dst)
- || eth_addr_equals(flow.dl_dst,
- netdev_get_etheraddr(cli->netdev)))) {
+ || eth_addr_equals(flow.dl_dst, cli_mac))) {
continue;
}
--- /dev/null
- VLOG_WARN("Could not connect to controller for %d seconds, "
- "failing open", disconn_secs);
+ /*
+ * Copyright (c) 2008, 2009 Nicira Networks.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+ #include <config.h>
+ #include "fail-open.h"
+ #include <inttypes.h>
+ #include <stdlib.h>
+ #include "flow.h"
+ #include "mac-learning.h"
+ #include "odp-util.h"
+ #include "ofproto.h"
+ #include "rconn.h"
+ #include "status.h"
+ #include "timeval.h"
+
+ #define THIS_MODULE VLM_fail_open
+ #include "vlog.h"
+
+ struct fail_open {
+ struct ofproto *ofproto;
+ struct rconn *controller;
+ int trigger_duration;
+ int last_disconn_secs;
+ struct status_category *ss_cat;
+ };
+
+ /* Causes the switch to enter or leave fail-open mode, if appropriate. */
+ void
+ fail_open_run(struct fail_open *fo)
+ {
+ int disconn_secs = rconn_failure_duration(fo->controller);
+ bool open = disconn_secs >= fo->trigger_duration;
+ if (open != (fo->last_disconn_secs != 0)) {
+ if (!open) {
+ flow_t flow;
+
+ VLOG_WARN("No longer in fail-open mode");
+ fo->last_disconn_secs = 0;
+
+ memset(&flow, 0, sizeof flow);
+ ofproto_delete_flow(fo->ofproto, &flow, OFPFW_ALL, 70000);
+ } else {
++ VLOG_WARN("Could not connect to controller (or switch failed "
++ "controller's post-connection admission control "
++ "policy) for %d seconds, failing open", disconn_secs);
+ fo->last_disconn_secs = disconn_secs;
+
+ /* Flush all OpenFlow and datapath flows. We will set up our
+ * fail-open rule from fail_open_flushed() when
+ * ofproto_flush_flows() calls back to us. */
+ ofproto_flush_flows(fo->ofproto);
+ }
+ } else if (open && disconn_secs > fo->last_disconn_secs + 60) {
+ VLOG_INFO("Still in fail-open mode after %d seconds disconnected "
+ "from controller", disconn_secs);
+ fo->last_disconn_secs = disconn_secs;
+ }
+ }
+
+ void
+ fail_open_wait(struct fail_open *fo UNUSED)
+ {
+ /* Nothing to do. */
+ }
+
+ void
+ fail_open_flushed(struct fail_open *fo)
+ {
+ int disconn_secs = rconn_failure_duration(fo->controller);
+ bool open = disconn_secs >= fo->trigger_duration;
+ if (open) {
+ union ofp_action action;
+ flow_t flow;
+
+ /* Set up a flow that matches every packet and directs them to
+ * OFPP_NORMAL. */
+ memset(&action, 0, sizeof action);
+ action.type = htons(OFPAT_OUTPUT);
+ action.output.len = htons(sizeof action);
+ action.output.port = htons(OFPP_NORMAL);
+ memset(&flow, 0, sizeof flow);
+ ofproto_add_flow(fo->ofproto, &flow, OFPFW_ALL, 70000,
+ &action, 1, 0);
+ }
+ }
+
+ static void
+ fail_open_status_cb(struct status_reply *sr, void *fo_)
+ {
+ struct fail_open *fo = fo_;
+ int cur_duration = rconn_failure_duration(fo->controller);
+
+ status_reply_put(sr, "trigger-duration=%d", fo->trigger_duration);
+ status_reply_put(sr, "current-duration=%d", cur_duration);
+ status_reply_put(sr, "triggered=%s",
+ cur_duration >= fo->trigger_duration ? "true" : "false");
+ }
+
+ struct fail_open *
+ fail_open_create(struct ofproto *ofproto,
+ int trigger_duration, struct switch_status *switch_status,
+ struct rconn *controller)
+ {
+ struct fail_open *fo = xmalloc(sizeof *fo);
+ fo->ofproto = ofproto;
+ fo->controller = controller;
+ fo->trigger_duration = trigger_duration;
+ fo->last_disconn_secs = 0;
+ fo->ss_cat = switch_status_register(switch_status, "fail-open",
+ fail_open_status_cb, fo);
+ return fo;
+ }
+
+ void
+ fail_open_set_trigger_duration(struct fail_open *fo, int trigger_duration)
+ {
+ fo->trigger_duration = trigger_duration;
+ }
+
+ void
+ fail_open_destroy(struct fail_open *fo)
+ {
+ if (fo) {
+ /* We don't own fo->controller. */
+ switch_status_unregister(fo->ss_cat);
+ free(fo);
+ }
+ }
--- /dev/null
- char *save_ptr;
+ /*
+ * Copyright (c) 2008, 2009 Nicira Networks.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+ #include <config.h>
+ #include "netflow.h"
+ #include <arpa/inet.h>
+ #include <errno.h>
+ #include <stdlib.h>
+ #include <unistd.h>
+ #include "cfg.h"
+ #include "flow.h"
+ #include "netflow.h"
+ #include "ofpbuf.h"
+ #include "ofproto.h"
+ #include "packets.h"
+ #include "socket-util.h"
+ #include "svec.h"
+ #include "timeval.h"
+ #include "util.h"
+ #include "xtoxll.h"
+
+ #define THIS_MODULE VLM_netflow
+ #include "vlog.h"
+
+ #define NETFLOW_V5_VERSION 5
+
+ /* Every NetFlow v5 message contains the header that follows. This is
+ * followed by up to thirty records that describe a terminating flow.
+ * We only send a single record per NetFlow message.
+ */
+ struct netflow_v5_header {
+ uint16_t version; /* NetFlow version is 5. */
+ uint16_t count; /* Number of records in this message. */
+ uint32_t sysuptime; /* System uptime in milliseconds. */
+ uint32_t unix_secs; /* Number of seconds since Unix epoch. */
+ uint32_t unix_nsecs; /* Number of residual nanoseconds
+ after epoch seconds. */
+ uint32_t flow_seq; /* Number of flows since sending
+ messages began. */
+ uint8_t engine_type; /* Engine type. */
+ uint8_t engine_id; /* Engine id. */
+ uint16_t sampling_interval; /* Set to zero. */
+ };
+ BUILD_ASSERT_DECL(sizeof(struct netflow_v5_header) == 24);
+
+ /* A NetFlow v5 description of a terminating flow. It is preceded by a
+ * NetFlow v5 header.
+ */
+ struct netflow_v5_record {
+ uint32_t src_addr; /* Source IP address. */
+ uint32_t dst_addr; /* Destination IP address. */
+ uint32_t nexthop; /* IP address of next hop. Set to 0. */
+ uint16_t input; /* Input interface index. */
+ uint16_t output; /* Output interface index. */
+ uint32_t packet_count; /* Number of packets. */
+ uint32_t byte_count; /* Number of bytes. */
+ uint32_t init_time; /* Value of sysuptime on first packet. */
+ uint32_t used_time; /* Value of sysuptime on last packet. */
+
+ /* The 'src_port' and 'dst_port' identify the source and destination
+ * port, respectively, for TCP and UDP. For ICMP, the high-order
+ * byte identifies the type and low-order byte identifies the code
+ * in the 'dst_port' field. */
+ uint16_t src_port;
+ uint16_t dst_port;
+
+ uint8_t pad1;
+ uint8_t tcp_flags; /* Union of seen TCP flags. */
+ uint8_t ip_proto; /* IP protocol. */
+ uint8_t ip_tos; /* IP TOS value. */
+ uint16_t src_as; /* Source AS ID. Set to 0. */
+ uint16_t dst_as; /* Destination AS ID. Set to 0. */
+ uint8_t src_mask; /* Source mask bits. Set to 0. */
+ uint8_t dst_mask; /* Destination mask bits. Set to 0. */
+ uint8_t pad[2];
+ };
+ BUILD_ASSERT_DECL(sizeof(struct netflow_v5_record) == 48);
+
+ struct netflow {
+ uint8_t engine_type; /* Value of engine_type to use. */
+ uint8_t engine_id; /* Value of engine_id to use. */
+ long long int boot_time; /* Time when netflow_create() was called. */
+ int *fds; /* Sockets for NetFlow collectors. */
+ size_t n_fds; /* Number of Netflow collectors. */
+ bool add_id_to_iface; /* Put the 7 least signficiant bits of
+ * 'engine_id' into the most signficant
+ * bits of the interface fields. */
+ uint32_t netflow_cnt; /* Flow sequence number for NetFlow. */
+ struct ofpbuf packet; /* NetFlow packet being accumulated. */
+ };
+
+ static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 5);
+
+ static int
+ open_collector(char *dst)
+ {
++ char *save_ptr = NULL;
+ const char *host_name;
+ const char *port_string;
+ struct sockaddr_in sin;
+ int retval;
+ int fd;
+
+ /* Glibc 2.7 has a bug in strtok_r when compiling with optimization that
+ * can cause segfaults here:
+ * http://sources.redhat.com/bugzilla/show_bug.cgi?id=5614.
+ * Using "::" instead of the obvious ":" works around it. */
+ host_name = strtok_r(dst, ":", &save_ptr);
+ port_string = strtok_r(NULL, ":", &save_ptr);
+ if (!host_name) {
+ ovs_error(0, "%s: bad peer name format", dst);
+ return -EAFNOSUPPORT;
+ }
+ if (!port_string) {
+ ovs_error(0, "%s: bad port format", dst);
+ return -EAFNOSUPPORT;
+ }
+
+ memset(&sin, 0, sizeof sin);
+ sin.sin_family = AF_INET;
+ if (lookup_ip(host_name, &sin.sin_addr)) {
+ return -ENOENT;
+ }
+ sin.sin_port = htons(atoi(port_string));
+
+ fd = socket(AF_INET, SOCK_DGRAM, 0);
+ if (fd < 0) {
+ VLOG_ERR("%s: socket: %s", dst, strerror(errno));
+ return -errno;
+ }
+
+ retval = set_nonblocking(fd);
+ if (retval) {
+ close(fd);
+ return -retval;
+ }
+
+ retval = connect(fd, (struct sockaddr *) &sin, sizeof sin);
+ if (retval < 0) {
+ int error = errno;
+ VLOG_ERR("%s: connect: %s", dst, strerror(error));
+ close(fd);
+ return -error;
+ }
+
+ return fd;
+ }
+
+ void
+ netflow_expire(struct netflow *nf, const struct ofexpired *expired)
+ {
+ struct netflow_v5_header *nf_hdr;
+ struct netflow_v5_record *nf_rec;
+ struct timeval now;
+
+ /* NetFlow only reports on IP packets. */
+ if (expired->flow.dl_type != htons(ETH_TYPE_IP)) {
+ return;
+ }
+
+ time_timeval(&now);
+
+ if (!nf->packet.size) {
+ nf_hdr = ofpbuf_put_zeros(&nf->packet, sizeof *nf_hdr);
+ nf_hdr->version = htons(NETFLOW_V5_VERSION);
+ nf_hdr->count = htons(0);
+ nf_hdr->sysuptime = htonl(time_msec() - nf->boot_time);
+ nf_hdr->unix_secs = htonl(now.tv_sec);
+ nf_hdr->unix_nsecs = htonl(now.tv_usec * 1000);
+ nf_hdr->flow_seq = htonl(nf->netflow_cnt++);
+ nf_hdr->engine_type = nf->engine_type;
+ nf_hdr->engine_id = nf->engine_id;
+ nf_hdr->sampling_interval = htons(0);
+ }
+
+ nf_hdr = nf->packet.data;
+ nf_hdr->count = htons(ntohs(nf_hdr->count) + 1);
+
+ nf_rec = ofpbuf_put_zeros(&nf->packet, sizeof *nf_rec);
+ nf_rec->src_addr = expired->flow.nw_src;
+ nf_rec->dst_addr = expired->flow.nw_dst;
+ nf_rec->nexthop = htons(0);
+ if (nf->add_id_to_iface) {
+ uint16_t iface = (nf->engine_id & 0x7f) << 9;
+ nf_rec->input = htons(iface | (expired->flow.in_port & 0x1ff));
+ nf_rec->output = htons(iface);
+ printf("input: %x\n", ntohs(nf_rec->input));
+ } else {
+ nf_rec->input = htons(expired->flow.in_port);
+ nf_rec->output = htons(0);
+ }
+ nf_rec->packet_count = htonl(MIN(expired->packet_count, UINT32_MAX));
+ nf_rec->byte_count = htonl(MIN(expired->byte_count, UINT32_MAX));
+ nf_rec->init_time = htonl(expired->created - nf->boot_time);
+ nf_rec->used_time = htonl(MAX(expired->created, expired->used)
+ - nf->boot_time);
+ if (expired->flow.nw_proto == IP_TYPE_ICMP) {
+ /* In NetFlow, the ICMP type and code are concatenated and
+ * placed in the 'dst_port' field. */
+ uint8_t type = ntohs(expired->flow.tp_src);
+ uint8_t code = ntohs(expired->flow.tp_dst);
+ nf_rec->src_port = htons(0);
+ nf_rec->dst_port = htons((type << 8) | code);
+ } else {
+ nf_rec->src_port = expired->flow.tp_src;
+ nf_rec->dst_port = expired->flow.tp_dst;
+ }
+ nf_rec->tcp_flags = expired->tcp_flags;
+ nf_rec->ip_proto = expired->flow.nw_proto;
+ nf_rec->ip_tos = expired->ip_tos;
+
+ /* NetFlow messages are limited to 30 records. A length of 1400
+ * bytes guarantees that the limit is not exceeded. */
+ if (nf->packet.size >= 1400) {
+ netflow_run(nf);
+ }
+ }
+
+ void
+ netflow_run(struct netflow *nf)
+ {
+ size_t i;
+
+ if (!nf->packet.size) {
+ return;
+ }
+
+ for (i = 0; i < nf->n_fds; i++) {
+ if (send(nf->fds[i], nf->packet.data, nf->packet.size, 0) == -1) {
+ VLOG_WARN_RL(&rl, "netflow message send failed: %s",
+ strerror(errno));
+ }
+ }
+ nf->packet.size = 0;
+ }
+
+ static void
+ clear_collectors(struct netflow *nf)
+ {
+ size_t i;
+
+ for (i = 0; i < nf->n_fds; i++) {
+ close(nf->fds[i]);
+ }
+ free(nf->fds);
+ nf->fds = NULL;
+ nf->n_fds = 0;
+ }
+
+ int
+ netflow_set_collectors(struct netflow *nf, const struct svec *collectors_)
+ {
+ struct svec collectors;
+ int error = 0;
+ size_t i;
+
+ clear_collectors(nf);
+
+ svec_clone(&collectors, collectors_);
+ svec_sort_unique(&collectors);
+
+ nf->fds = xmalloc(sizeof *nf->fds * collectors.n);
+ for (i = 0; i < collectors.n; i++) {
+ const char *name = collectors.names[i];
+ char *tmpname = xstrdup(name);
+ int fd = open_collector(tmpname);
+ free(tmpname);
+ if (fd >= 0) {
+ nf->fds[nf->n_fds++] = fd;
+ } else {
+ VLOG_WARN("couldn't open connection to collector (%s), "
+ "ignoring %s\n", strerror(-fd), name);
+ if (!error) {
+ error = -fd;
+ }
+ }
+ }
+
+ svec_destroy(&collectors);
+ return error;
+ }
+
+ void
+ netflow_set_engine(struct netflow *nf, uint8_t engine_type,
+ uint8_t engine_id, bool add_id_to_iface)
+ {
+ nf->engine_type = engine_type;
+ nf->engine_id = engine_id;
+ nf->add_id_to_iface = add_id_to_iface;
+ }
+
+ struct netflow *
+ netflow_create(void)
+ {
+ struct netflow *nf = xmalloc(sizeof *nf);
+ nf->engine_type = 0;
+ nf->engine_id = 0;
+ nf->boot_time = time_msec();
+ nf->fds = NULL;
+ nf->n_fds = 0;
+ nf->add_id_to_iface = false;
+ nf->netflow_cnt = 0;
+ ofpbuf_init(&nf->packet, 1500);
+ return nf;
+ }
+
+ void
+ netflow_destroy(struct netflow *nf)
+ {
+ if (nf) {
+ ofpbuf_uninit(&nf->packet);
+ clear_collectors(nf);
+ free(nf);
+ }
+ }
--- /dev/null
- p->controller = ofconn_create(p, rconn_create(15, 15));
+ /*
+ * Copyright (c) 2009 Nicira Networks.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+ #include <config.h>
+ #include "ofproto.h"
+ #include <errno.h>
+ #include <inttypes.h>
+ #include <net/if.h>
+ #include <netinet/in.h>
+ #include <stdbool.h>
+ #include <stdlib.h>
+ #include "classifier.h"
+ #include "coverage.h"
+ #include "discovery.h"
+ #include "dpif.h"
++#include "dynamic-string.h"
+ #include "executer.h"
+ #include "fail-open.h"
+ #include "in-band.h"
+ #include "mac-learning.h"
+ #include "netdev.h"
+ #include "netflow.h"
+ #include "odp-util.h"
+ #include "ofp-print.h"
+ #include "ofpbuf.h"
+ #include "openflow/nicira-ext.h"
+ #include "openflow/openflow.h"
+ #include "openflow/openflow-mgmt.h"
+ #include "openvswitch/datapath-protocol.h"
+ #include "packets.h"
+ #include "pinsched.h"
+ #include "pktbuf.h"
+ #include "poll-loop.h"
+ #include "port-array.h"
+ #include "rconn.h"
+ #include "shash.h"
+ #include "status.h"
+ #include "stp.h"
+ #include "svec.h"
+ #include "tag.h"
+ #include "timeval.h"
++#include "unixctl.h"
+ #include "vconn.h"
+ #include "vconn-ssl.h"
+ #include "xtoxll.h"
+
+ #define THIS_MODULE VLM_ofproto
+ #include "vlog.h"
+
+ enum {
+ DP_GROUP_FLOOD = 0,
+ DP_GROUP_ALL = 1
+ };
+
+ enum {
+ TABLEID_HASH = 0,
+ TABLEID_CLASSIFIER = 1
+ };
+
+ struct ofport {
+ struct netdev *netdev;
+ struct ofp_phy_port opp; /* In host byte order. */
+ };
+
+ static void ofport_free(struct ofport *);
+ static void hton_ofp_phy_port(struct ofp_phy_port *);
+
+ static int xlate_actions(const union ofp_action *in, size_t n_in,
+ const flow_t *flow, struct ofproto *ofproto,
+ const struct ofpbuf *packet,
+ struct odp_actions *out, tag_type *tags,
+ bool *may_setup_flow);
+
+ struct rule {
+ struct cls_rule cr;
+
+ uint16_t idle_timeout; /* In seconds from time of last use. */
+ uint16_t hard_timeout; /* In seconds from time of creation. */
+ long long int used; /* Last-used time (0 if never used). */
+ long long int created; /* Creation time. */
+ uint64_t packet_count; /* Number of packets received. */
+ uint64_t byte_count; /* Number of bytes received. */
+ uint64_t accounted_bytes; /* Number of bytes passed to account_cb. */
+ uint8_t tcp_flags; /* Bitwise-OR of all TCP flags seen. */
+ uint8_t ip_tos; /* Last-seen IP type-of-service. */
+ tag_type tags; /* Tags (set only by hooks). */
+
+ /* If 'super' is non-NULL, this rule is a subrule, that is, it is an
+ * exact-match rule (having cr.wc.wildcards of 0) generated from the
+ * wildcard rule 'super'. In this case, 'list' is an element of the
+ * super-rule's list.
+ *
+ * If 'super' is NULL, this rule is a super-rule, and 'list' is the head of
+ * a list of subrules. A super-rule with no wildcards (where
+ * cr.wc.wildcards is 0) will never have any subrules. */
+ struct rule *super;
+ struct list list;
+
+ /* OpenFlow actions.
+ *
+ * A subrule has no actions (it uses the super-rule's actions). */
+ int n_actions;
+ union ofp_action *actions;
+
+ /* Datapath actions.
+ *
+ * A super-rule with wildcard fields never has ODP actions (since the
+ * datapath only supports exact-match flows). */
+ bool installed; /* Installed in datapath? */
+ bool may_install; /* True ordinarily; false if actions must
+ * be reassessed for every packet. */
+ int n_odp_actions;
+ union odp_action *odp_actions;
+ };
+
+ static inline bool
+ rule_is_hidden(const struct rule *rule)
+ {
+ /* Subrules are merely an implementation detail, so hide them from the
+ * controller. */
+ if (rule->super != NULL) {
+ return true;
+ }
+
+ /* Rules with priority higher than UINT16_MAX are set up by ofproto itself
+ * (e.g. by in-band control) and are intentionally hidden from the
+ * controller. */
+ if (rule->cr.priority > UINT16_MAX) {
+ return true;
+ }
+
+ return false;
+ }
+
+ static struct rule *rule_create(struct rule *super, const union ofp_action *,
+ size_t n_actions, uint16_t idle_timeout,
+ uint16_t hard_timeout);
+ static void rule_free(struct rule *);
+ static void rule_destroy(struct ofproto *, struct rule *);
+ static struct rule *rule_from_cls_rule(const struct cls_rule *);
+ static void rule_insert(struct ofproto *, struct rule *,
+ struct ofpbuf *packet, uint16_t in_port);
+ static void rule_remove(struct ofproto *, struct rule *);
+ static bool rule_make_actions(struct ofproto *, struct rule *,
+ const struct ofpbuf *packet);
+ static void rule_install(struct ofproto *, struct rule *,
+ struct rule *displaced_rule);
+ static void rule_uninstall(struct ofproto *, struct rule *);
+ static void rule_post_uninstall(struct ofproto *, struct rule *);
+
+ struct ofconn {
+ struct list node;
+ struct rconn *rconn;
+ struct pktbuf *pktbuf;
+ bool send_flow_exp;
+ int miss_send_len;
+
+ struct rconn_packet_counter *packet_in_counter;
+
+ /* Number of OpenFlow messages queued as replies to OpenFlow requests, and
+ * the maximum number before we stop reading OpenFlow requests. */
+ #define OFCONN_REPLY_MAX 100
+ struct rconn_packet_counter *reply_counter;
+ };
+
+ static struct ofconn *ofconn_create(struct ofproto *, struct rconn *);
+ static void ofconn_destroy(struct ofconn *, struct ofproto *);
+ static void ofconn_run(struct ofconn *, struct ofproto *);
+ static void ofconn_wait(struct ofconn *);
+ static void queue_tx(struct ofpbuf *msg, const struct ofconn *ofconn,
+ struct rconn_packet_counter *counter);
+
+ struct ofproto {
+ /* Settings. */
+ uint64_t datapath_id; /* Datapath ID. */
+ uint64_t fallback_dpid; /* Datapath ID if no better choice found. */
+ uint64_t mgmt_id; /* Management channel identifier. */
+ char *manufacturer; /* Manufacturer. */
+ char *hardware; /* Hardware. */
+ char *software; /* Software version. */
+ char *serial; /* Serial number. */
+
+ /* Datapath. */
+ struct dpif *dpif;
+ struct netdev_monitor *netdev_monitor;
+ struct port_array ports; /* Index is ODP port nr; ofport->opp.port_no is
+ * OFP port nr. */
+ struct shash port_by_name;
+ uint32_t max_ports;
+
+ /* Configuration. */
+ struct switch_status *switch_status;
+ struct status_category *ss_cat;
+ struct in_band *in_band;
+ struct discovery *discovery;
+ struct fail_open *fail_open;
+ struct pinsched *miss_sched, *action_sched;
+ struct executer *executer;
+ struct netflow *netflow;
+
+ /* Flow table. */
+ struct classifier cls;
+ bool need_revalidate;
+ long long int next_expiration;
+ struct tag_set revalidate_set;
+
+ /* OpenFlow connections. */
+ struct list all_conns;
+ struct ofconn *controller;
+ struct pvconn **listeners;
+ size_t n_listeners;
+ struct pvconn **snoops;
+ size_t n_snoops;
+
+ /* Hooks for ovs-vswitchd. */
+ const struct ofhooks *ofhooks;
+ void *aux;
+
+ /* Used by default ofhooks. */
+ struct mac_learning *ml;
+ };
+
+ static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 5);
+
+ static const struct ofhooks default_ofhooks;
+
+ static uint64_t pick_datapath_id(const struct ofproto *);
+ static uint64_t pick_fallback_dpid(void);
+ static void send_packet_in_miss(struct ofpbuf *, void *ofproto);
+ static void send_packet_in_action(struct ofpbuf *, void *ofproto);
+ static void update_used(struct ofproto *);
+ static void update_stats(struct rule *, const struct odp_flow_stats *);
+ static void expire_rule(struct cls_rule *, void *ofproto);
+ static bool revalidate_rule(struct ofproto *p, struct rule *rule);
+ static void revalidate_cb(struct cls_rule *rule_, void *p_);
+
+ static void handle_odp_msg(struct ofproto *, struct ofpbuf *);
+
+ static void handle_openflow(struct ofconn *, struct ofproto *,
+ struct ofpbuf *);
+
+ static void refresh_port_group(struct ofproto *, unsigned int group);
+ static void update_port(struct ofproto *, const char *devname);
+ static int init_ports(struct ofproto *);
+ static void reinit_ports(struct ofproto *);
+
+ int
+ ofproto_create(const char *datapath, const struct ofhooks *ofhooks, void *aux,
+ struct ofproto **ofprotop)
+ {
+ struct odp_stats stats;
+ struct ofproto *p;
+ struct dpif *dpif;
+ int error;
+
+ *ofprotop = NULL;
+
+ /* Connect to datapath and start listening for messages. */
+ error = dpif_open(datapath, &dpif);
+ if (error) {
+ VLOG_ERR("failed to open datapath %s: %s", datapath, strerror(error));
+ return error;
+ }
+ error = dpif_get_dp_stats(dpif, &stats);
+ if (error) {
+ VLOG_ERR("failed to obtain stats for datapath %s: %s",
+ datapath, strerror(error));
+ dpif_close(dpif);
+ return error;
+ }
+ error = dpif_recv_set_mask(dpif, ODPL_MISS | ODPL_ACTION);
+ if (error) {
+ VLOG_ERR("failed to listen on datapath %s: %s",
+ datapath, strerror(error));
+ dpif_close(dpif);
+ return error;
+ }
+ dpif_flow_flush(dpif);
+ dpif_recv_purge(dpif);
+
+ /* Initialize settings. */
+ p = xcalloc(1, sizeof *p);
+ p->fallback_dpid = pick_fallback_dpid();
+ p->datapath_id = p->fallback_dpid;
+ p->manufacturer = xstrdup("Nicira Networks, Inc.");
+ p->hardware = xstrdup("Reference Implementation");
+ p->software = xstrdup(VERSION BUILDNR);
+ p->serial = xstrdup("None");
+
+ /* Initialize datapath. */
+ p->dpif = dpif;
+ p->netdev_monitor = netdev_monitor_create();
+ port_array_init(&p->ports);
+ shash_init(&p->port_by_name);
+ p->max_ports = stats.max_ports;
+
+ /* Initialize submodules. */
+ p->switch_status = switch_status_create(p);
+ p->in_band = NULL;
+ p->discovery = NULL;
+ p->fail_open = NULL;
+ p->miss_sched = p->action_sched = NULL;
+ p->executer = NULL;
+ p->netflow = NULL;
+
+ /* Initialize flow table. */
+ classifier_init(&p->cls);
+ p->need_revalidate = false;
+ p->next_expiration = time_msec() + 1000;
+ tag_set_init(&p->revalidate_set);
+
+ /* Initialize OpenFlow connections. */
+ list_init(&p->all_conns);
- struct ofport *ofport;
++ p->controller = ofconn_create(p, rconn_create(5, 8));
+ p->controller->pktbuf = pktbuf_create();
+ p->controller->miss_send_len = OFP_DEFAULT_MISS_SEND_LEN;
+ p->listeners = NULL;
+ p->n_listeners = 0;
+ p->snoops = NULL;
+ p->n_snoops = 0;
+
+ /* Initialize hooks. */
+ if (ofhooks) {
+ p->ofhooks = ofhooks;
+ p->aux = aux;
+ p->ml = NULL;
+ } else {
+ p->ofhooks = &default_ofhooks;
+ p->aux = p;
+ p->ml = mac_learning_create();
+ }
+
+ /* Register switch status category. */
+ p->ss_cat = switch_status_register(p->switch_status, "remote",
+ rconn_status_cb, p->controller->rconn);
+
+ /* Almost done... */
+ error = init_ports(p);
+ if (error) {
+ ofproto_destroy(p);
+ return error;
+ }
+
+ /* Pick final datapath ID. */
+ p->datapath_id = pick_datapath_id(p);
+ VLOG_INFO("using datapath ID %012"PRIx64, p->datapath_id);
+
+ *ofprotop = p;
+ return 0;
+ }
+
+ void
+ ofproto_set_datapath_id(struct ofproto *p, uint64_t datapath_id)
+ {
+ uint64_t old_dpid = p->datapath_id;
+ p->datapath_id = datapath_id ? datapath_id : pick_datapath_id(p);
+ if (p->datapath_id != old_dpid) {
+ VLOG_INFO("datapath ID changed to %012"PRIx64, p->datapath_id);
+ rconn_reconnect(p->controller->rconn);
+ }
+ }
+
+ void
+ ofproto_set_mgmt_id(struct ofproto *p, uint64_t mgmt_id)
+ {
+ p->mgmt_id = mgmt_id;
+ }
+
+ void
+ ofproto_set_probe_interval(struct ofproto *p, int probe_interval)
+ {
+ probe_interval = probe_interval ? MAX(probe_interval, 5) : 0;
+ rconn_set_probe_interval(p->controller->rconn, probe_interval);
+ if (p->fail_open) {
+ int trigger_duration = probe_interval ? probe_interval * 3 : 15;
+ fail_open_set_trigger_duration(p->fail_open, trigger_duration);
+ }
+ }
+
+ void
+ ofproto_set_max_backoff(struct ofproto *p, int max_backoff)
+ {
+ rconn_set_max_backoff(p->controller->rconn, max_backoff);
+ }
+
+ void
+ ofproto_set_desc(struct ofproto *p,
+ const char *manufacturer, const char *hardware,
+ const char *software, const char *serial)
+ {
+ if (manufacturer) {
+ free(p->manufacturer);
+ p->manufacturer = xstrdup(manufacturer);
+ }
+ if (hardware) {
+ free(p->hardware);
+ p->hardware = xstrdup(hardware);
+ }
+ if (software) {
+ free(p->software);
+ p->software = xstrdup(software);
+ }
+ if (serial) {
+ free(p->serial);
+ p->serial = xstrdup(serial);
+ }
+ }
+
+ int
+ ofproto_set_in_band(struct ofproto *p, bool in_band)
+ {
+ if (in_band != (p->in_band != NULL)) {
+ if (in_band) {
+ in_band_create(p, p->switch_status, p->controller->rconn,
+ &p->in_band);
+ return 0;
+ } else {
+ ofproto_set_discovery(p, false, NULL, true);
+ in_band_destroy(p->in_band);
+ p->in_band = NULL;
+ }
+ rconn_reconnect(p->controller->rconn);
+ }
+ return 0;
+ }
+
+ int
+ ofproto_set_discovery(struct ofproto *p, bool discovery,
+ const char *re, bool update_resolv_conf)
+ {
+ if (discovery != (p->discovery != NULL)) {
+ if (discovery) {
+ int error = ofproto_set_in_band(p, true);
+ if (error) {
+ return error;
+ }
+ error = discovery_create(re, update_resolv_conf,
+ p->dpif, p->switch_status,
+ &p->discovery);
+ if (error) {
+ return error;
+ }
+ } else {
+ discovery_destroy(p->discovery);
+ p->discovery = NULL;
+ }
+ rconn_disconnect(p->controller->rconn);
+ } else if (discovery) {
+ discovery_set_update_resolv_conf(p->discovery, update_resolv_conf);
+ return discovery_set_accept_controller_re(p->discovery, re);
+ }
+ return 0;
+ }
+
+ int
+ ofproto_set_controller(struct ofproto *ofproto, const char *controller)
+ {
+ if (ofproto->discovery) {
+ return EINVAL;
+ } else if (controller) {
+ if (strcmp(rconn_get_name(ofproto->controller->rconn), controller)) {
+ return rconn_connect(ofproto->controller->rconn, controller);
+ } else {
+ return 0;
+ }
+ } else {
+ rconn_disconnect(ofproto->controller->rconn);
+ return 0;
+ }
+ }
+
+ static int
+ set_pvconns(struct pvconn ***pvconnsp, size_t *n_pvconnsp,
+ const struct svec *svec)
+ {
+ struct pvconn **pvconns = *pvconnsp;
+ size_t n_pvconns = *n_pvconnsp;
+ int retval = 0;
+ size_t i;
+
+ for (i = 0; i < n_pvconns; i++) {
+ pvconn_close(pvconns[i]);
+ }
+ free(pvconns);
+
+ pvconns = xmalloc(svec->n * sizeof *pvconns);
+ n_pvconns = 0;
+ for (i = 0; i < svec->n; i++) {
+ const char *name = svec->names[i];
+ struct pvconn *pvconn;
+ int error;
+
+ error = pvconn_open(name, &pvconn);
+ if (!error) {
+ pvconns[n_pvconns++] = pvconn;
+ } else {
+ VLOG_ERR("failed to listen on %s: %s", name, strerror(error));
+ if (!retval) {
+ retval = error;
+ }
+ }
+ }
+
+ *pvconnsp = pvconns;
+ *n_pvconnsp = n_pvconns;
+
+ return retval;
+ }
+
+ int
+ ofproto_set_listeners(struct ofproto *ofproto, const struct svec *listeners)
+ {
+ return set_pvconns(&ofproto->listeners, &ofproto->n_listeners, listeners);
+ }
+
+ int
+ ofproto_set_snoops(struct ofproto *ofproto, const struct svec *snoops)
+ {
+ return set_pvconns(&ofproto->snoops, &ofproto->n_snoops, snoops);
+ }
+
+ int
+ ofproto_set_netflow(struct ofproto *ofproto, const struct svec *collectors,
+ uint8_t engine_type, uint8_t engine_id, bool add_id_to_iface)
+ {
+ if (collectors && collectors->n) {
+ if (!ofproto->netflow) {
+ ofproto->netflow = netflow_create();
+ }
+ netflow_set_engine(ofproto->netflow, engine_type, engine_id,
+ add_id_to_iface);
+ return netflow_set_collectors(ofproto->netflow, collectors);
+ } else {
+ netflow_destroy(ofproto->netflow);
+ ofproto->netflow = NULL;
+ return 0;
+ }
+ }
+
+ void
+ ofproto_set_failure(struct ofproto *ofproto, bool fail_open)
+ {
+ if (fail_open) {
+ struct rconn *rconn = ofproto->controller->rconn;
+ int trigger_duration = rconn_get_probe_interval(rconn) * 3;
+ if (!ofproto->fail_open) {
+ ofproto->fail_open = fail_open_create(ofproto, trigger_duration,
+ ofproto->switch_status,
+ rconn);
+ } else {
+ fail_open_set_trigger_duration(ofproto->fail_open,
+ trigger_duration);
+ }
+ } else {
+ fail_open_destroy(ofproto->fail_open);
+ ofproto->fail_open = NULL;
+ }
+ }
+
+ void
+ ofproto_set_rate_limit(struct ofproto *ofproto,
+ int rate_limit, int burst_limit)
+ {
+ if (rate_limit > 0) {
+ if (!ofproto->miss_sched) {
+ ofproto->miss_sched = pinsched_create(rate_limit, burst_limit,
+ ofproto->switch_status);
+ ofproto->action_sched = pinsched_create(rate_limit, burst_limit,
+ NULL);
+ } else {
+ pinsched_set_limits(ofproto->miss_sched, rate_limit, burst_limit);
+ pinsched_set_limits(ofproto->action_sched,
+ rate_limit, burst_limit);
+ }
+ } else {
+ pinsched_destroy(ofproto->miss_sched);
+ ofproto->miss_sched = NULL;
+ pinsched_destroy(ofproto->action_sched);
+ ofproto->action_sched = NULL;
+ }
+ }
+
+ int
+ ofproto_set_stp(struct ofproto *ofproto UNUSED, bool enable_stp)
+ {
+ /* XXX */
+ if (enable_stp) {
+ VLOG_WARN("STP is not yet implemented");
+ return EINVAL;
+ } else {
+ return 0;
+ }
+ }
+
+ int
+ ofproto_set_remote_execution(struct ofproto *ofproto, const char *command_acl,
+ const char *command_dir)
+ {
+ if (command_acl) {
+ if (!ofproto->executer) {
+ return executer_create(command_acl, command_dir,
+ &ofproto->executer);
+ } else {
+ executer_set_acl(ofproto->executer, command_acl, command_dir);
+ }
+ } else {
+ executer_destroy(ofproto->executer);
+ ofproto->executer = NULL;
+ }
+ return 0;
+ }
+
+ uint64_t
+ ofproto_get_datapath_id(const struct ofproto *ofproto)
+ {
+ return ofproto->datapath_id;
+ }
+
+ uint64_t
+ ofproto_get_mgmt_id(const struct ofproto *ofproto)
+ {
+ return ofproto->mgmt_id;
+ }
+
+ int
+ ofproto_get_probe_interval(const struct ofproto *ofproto)
+ {
+ return rconn_get_probe_interval(ofproto->controller->rconn);
+ }
+
+ int
+ ofproto_get_max_backoff(const struct ofproto *ofproto)
+ {
+ return rconn_get_max_backoff(ofproto->controller->rconn);
+ }
+
+ bool
+ ofproto_get_in_band(const struct ofproto *ofproto)
+ {
+ return ofproto->in_band != NULL;
+ }
+
+ bool
+ ofproto_get_discovery(const struct ofproto *ofproto)
+ {
+ return ofproto->discovery != NULL;
+ }
+
+ const char *
+ ofproto_get_controller(const struct ofproto *ofproto)
+ {
+ return rconn_get_name(ofproto->controller->rconn);
+ }
+
+ void
+ ofproto_get_listeners(const struct ofproto *ofproto, struct svec *listeners)
+ {
+ size_t i;
+
+ for (i = 0; i < ofproto->n_listeners; i++) {
+ svec_add(listeners, pvconn_get_name(ofproto->listeners[i]));
+ }
+ }
+
+ void
+ ofproto_get_snoops(const struct ofproto *ofproto, struct svec *snoops)
+ {
+ size_t i;
+
+ for (i = 0; i < ofproto->n_snoops; i++) {
+ svec_add(snoops, pvconn_get_name(ofproto->snoops[i]));
+ }
+ }
+
+ void
+ ofproto_destroy(struct ofproto *p)
+ {
+ struct ofconn *ofconn, *next_ofconn;
+ struct ofport *ofport;
+ unsigned int port_no;
+ size_t i;
+
+ if (!p) {
+ return;
+ }
+
+ ofproto_flush_flows(p);
+ classifier_destroy(&p->cls);
+
+ LIST_FOR_EACH_SAFE (ofconn, next_ofconn, struct ofconn, node,
+ &p->all_conns) {
+ ofconn_destroy(ofconn, p);
+ }
+
+ dpif_close(p->dpif);
+ netdev_monitor_destroy(p->netdev_monitor);
+ PORT_ARRAY_FOR_EACH (ofport, &p->ports, port_no) {
+ ofport_free(ofport);
+ }
+ shash_destroy(&p->port_by_name);
+
+ switch_status_destroy(p->switch_status);
+ in_band_destroy(p->in_band);
+ discovery_destroy(p->discovery);
+ fail_open_destroy(p->fail_open);
+ pinsched_destroy(p->miss_sched);
+ pinsched_destroy(p->action_sched);
+ executer_destroy(p->executer);
+ netflow_destroy(p->netflow);
+
+ switch_status_unregister(p->ss_cat);
+
+ for (i = 0; i < p->n_listeners; i++) {
+ pvconn_close(p->listeners[i]);
+ }
+ free(p->listeners);
+
+ for (i = 0; i < p->n_snoops; i++) {
+ pvconn_close(p->snoops[i]);
+ }
+ free(p->snoops);
+
+ mac_learning_destroy(p->ml);
+
+ free(p);
+ }
+
+ int
+ ofproto_run(struct ofproto *p)
+ {
+ int error = ofproto_run1(p);
+ if (!error) {
+ error = ofproto_run2(p, false);
+ }
+ return error;
+ }
+
+ static void
+ process_port_change(struct ofproto *ofproto, int error, char *devname)
+ {
+ if (error == ENOBUFS) {
+ reinit_ports(ofproto);
+ } else if (!error) {
+ update_port(ofproto, devname);
+ free(devname);
+ }
+ }
+
+ int
+ ofproto_run1(struct ofproto *p)
+ {
+ struct ofconn *ofconn, *next_ofconn;
+ char *devname;
+ int error;
+ int i;
+
+ for (i = 0; i < 50; i++) {
+ struct ofpbuf *buf;
+ int error;
+
+ error = dpif_recv(p->dpif, &buf);
+ if (error) {
+ if (error == ENODEV) {
+ /* Someone destroyed the datapath behind our back. The caller
+ * better destroy us and give up, because we're just going to
+ * spin from here on out. */
+ static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 5);
+ VLOG_ERR_RL(&rl, "%s: datapath was destroyed externally",
+ dpif_name(p->dpif));
+ return ENODEV;
+ }
+ break;
+ }
+
+ handle_odp_msg(p, buf);
+ }
+
+ while ((error = dpif_port_poll(p->dpif, &devname)) != EAGAIN) {
+ process_port_change(p, error, devname);
+ }
+ while ((error = netdev_monitor_poll(p->netdev_monitor,
+ &devname)) != EAGAIN) {
+ process_port_change(p, error, devname);
+ }
+
+ if (p->in_band) {
+ in_band_run(p->in_band);
+ }
+ if (p->discovery) {
+ char *controller_name;
+ if (rconn_is_connectivity_questionable(p->controller->rconn)) {
+ discovery_question_connectivity(p->discovery);
+ }
+ if (discovery_run(p->discovery, &controller_name)) {
+ if (controller_name) {
+ rconn_connect(p->controller->rconn, controller_name);
+ } else {
+ rconn_disconnect(p->controller->rconn);
+ }
+ }
+ }
+ if (p->fail_open) {
+ fail_open_run(p->fail_open);
+ }
+ pinsched_run(p->miss_sched, send_packet_in_miss, p);
+ pinsched_run(p->action_sched, send_packet_in_action, p);
+ if (p->executer) {
+ executer_run(p->executer);
+ }
+
+ LIST_FOR_EACH_SAFE (ofconn, next_ofconn, struct ofconn, node,
+ &p->all_conns) {
+ ofconn_run(ofconn, p);
+ }
+
+ for (i = 0; i < p->n_listeners; i++) {
+ struct vconn *vconn;
+ int retval;
+
+ retval = pvconn_accept(p->listeners[i], OFP_VERSION, &vconn);
+ if (!retval) {
+ ofconn_create(p, rconn_new_from_vconn("passive", vconn));
+ } else if (retval != EAGAIN) {
+ VLOG_WARN_RL(&rl, "accept failed (%s)", strerror(retval));
+ }
+ }
+
+ for (i = 0; i < p->n_snoops; i++) {
+ struct vconn *vconn;
+ int retval;
+
+ retval = pvconn_accept(p->snoops[i], OFP_VERSION, &vconn);
+ if (!retval) {
+ rconn_add_monitor(p->controller->rconn, vconn);
+ } else if (retval != EAGAIN) {
+ VLOG_WARN_RL(&rl, "accept failed (%s)", strerror(retval));
+ }
+ }
+
+ if (time_msec() >= p->next_expiration) {
+ COVERAGE_INC(ofproto_expiration);
+ p->next_expiration = time_msec() + 1000;
+ update_used(p);
+
+ classifier_for_each(&p->cls, CLS_INC_ALL, expire_rule, p);
+
+ /* Let the hook know that we're at a stable point: all outstanding data
+ * in existing flows has been accounted to the account_cb. Thus, the
+ * hook can now reasonably do operations that depend on having accurate
+ * flow volume accounting (currently, that's just bond rebalancing). */
+ if (p->ofhooks->account_checkpoint_cb) {
+ p->ofhooks->account_checkpoint_cb(p->aux);
+ }
+ }
+
+ if (p->netflow) {
+ netflow_run(p->netflow);
+ }
+
+ return 0;
+ }
+
+ struct revalidate_cbdata {
+ struct ofproto *ofproto;
+ bool revalidate_all; /* Revalidate all exact-match rules? */
+ bool revalidate_subrules; /* Revalidate all exact-match subrules? */
+ struct tag_set revalidate_set; /* Set of tags to revalidate. */
+ };
+
+ int
+ ofproto_run2(struct ofproto *p, bool revalidate_all)
+ {
+ if (p->need_revalidate || revalidate_all
+ || !tag_set_is_empty(&p->revalidate_set)) {
+ struct revalidate_cbdata cbdata;
+ cbdata.ofproto = p;
+ cbdata.revalidate_all = revalidate_all;
+ cbdata.revalidate_subrules = p->need_revalidate;
+ cbdata.revalidate_set = p->revalidate_set;
+ tag_set_init(&p->revalidate_set);
+ COVERAGE_INC(ofproto_revalidate);
+ classifier_for_each(&p->cls, CLS_INC_EXACT, revalidate_cb, &cbdata);
+ p->need_revalidate = false;
+ }
+
+ return 0;
+ }
+
+ void
+ ofproto_wait(struct ofproto *p)
+ {
+ struct ofconn *ofconn;
+ size_t i;
+
+ dpif_recv_wait(p->dpif);
+ dpif_port_poll_wait(p->dpif);
+ netdev_monitor_poll_wait(p->netdev_monitor);
+ LIST_FOR_EACH (ofconn, struct ofconn, node, &p->all_conns) {
+ ofconn_wait(ofconn);
+ }
+ if (p->in_band) {
+ in_band_wait(p->in_band);
+ }
+ if (p->discovery) {
+ discovery_wait(p->discovery);
+ }
+ if (p->fail_open) {
+ fail_open_wait(p->fail_open);
+ }
+ pinsched_wait(p->miss_sched);
+ pinsched_wait(p->action_sched);
+ if (p->executer) {
+ executer_wait(p->executer);
+ }
+ if (!tag_set_is_empty(&p->revalidate_set)) {
+ poll_immediate_wake();
+ }
+ if (p->need_revalidate) {
+ /* Shouldn't happen, but if it does just go around again. */
+ VLOG_DBG_RL(&rl, "need revalidate in ofproto_wait_cb()");
+ poll_immediate_wake();
+ } else if (p->next_expiration != LLONG_MAX) {
+ poll_timer_wait(p->next_expiration - time_msec());
+ }
+ for (i = 0; i < p->n_listeners; i++) {
+ pvconn_wait(p->listeners[i]);
+ }
+ for (i = 0; i < p->n_snoops; i++) {
+ pvconn_wait(p->snoops[i]);
+ }
+ }
+
+ void
+ ofproto_revalidate(struct ofproto *ofproto, tag_type tag)
+ {
+ tag_set_add(&ofproto->revalidate_set, tag);
+ }
+
+ struct tag_set *
+ ofproto_get_revalidate_set(struct ofproto *ofproto)
+ {
+ return &ofproto->revalidate_set;
+ }
+
+ bool
+ ofproto_is_alive(const struct ofproto *p)
+ {
+ return p->discovery || rconn_is_alive(p->controller->rconn);
+ }
+
+ int
+ ofproto_send_packet(struct ofproto *p, const flow_t *flow,
+ const union ofp_action *actions, size_t n_actions,
+ const struct ofpbuf *packet)
+ {
+ struct odp_actions odp_actions;
+ int error;
+
+ error = xlate_actions(actions, n_actions, flow, p, packet, &odp_actions,
+ NULL, NULL);
+ if (error) {
+ return error;
+ }
+
+ /* XXX Should we translate the dpif_execute() errno value into an OpenFlow
+ * error code? */
+ dpif_execute(p->dpif, flow->in_port, odp_actions.actions,
+ odp_actions.n_actions, packet);
+ return 0;
+ }
+
+ void
+ ofproto_add_flow(struct ofproto *p,
+ const flow_t *flow, uint32_t wildcards, unsigned int priority,
+ const union ofp_action *actions, size_t n_actions,
+ int idle_timeout)
+ {
+ struct rule *rule;
+ rule = rule_create(NULL, actions, n_actions,
+ idle_timeout >= 0 ? idle_timeout : 5 /* XXX */, 0);
+ cls_rule_from_flow(&rule->cr, flow, wildcards, priority);
+ rule_insert(p, rule, NULL, 0);
+ }
+
+ void
+ ofproto_delete_flow(struct ofproto *ofproto, const flow_t *flow,
+ uint32_t wildcards, unsigned int priority)
+ {
+ struct rule *rule;
+
+ rule = rule_from_cls_rule(classifier_find_rule_exactly(&ofproto->cls,
+ flow, wildcards,
+ priority));
+ if (rule) {
+ rule_remove(ofproto, rule);
+ }
+ }
+
+ static void
+ destroy_rule(struct cls_rule *rule_, void *ofproto_)
+ {
+ struct rule *rule = rule_from_cls_rule(rule_);
+ struct ofproto *ofproto = ofproto_;
+
+ /* Mark the flow as not installed, even though it might really be
+ * installed, so that rule_remove() doesn't bother trying to uninstall it.
+ * There is no point in uninstalling it individually since we are about to
+ * blow away all the flows with dpif_flow_flush(). */
+ rule->installed = false;
+
+ rule_remove(ofproto, rule);
+ }
+
+ void
+ ofproto_flush_flows(struct ofproto *ofproto)
+ {
+ COVERAGE_INC(ofproto_flush);
+ classifier_for_each(&ofproto->cls, CLS_INC_ALL, destroy_rule, ofproto);
+ dpif_flow_flush(ofproto->dpif);
+ if (ofproto->in_band) {
+ in_band_flushed(ofproto->in_band);
+ }
+ if (ofproto->fail_open) {
+ fail_open_flushed(ofproto->fail_open);
+ }
+ }
+ \f
+ static void
+ reinit_ports(struct ofproto *p)
+ {
+ struct svec devnames;
+ struct ofport *ofport;
+ unsigned int port_no;
+ struct odp_port *odp_ports;
+ size_t n_odp_ports;
+ size_t i;
+
+ svec_init(&devnames);
+ PORT_ARRAY_FOR_EACH (ofport, &p->ports, port_no) {
+ svec_add (&devnames, (char *) ofport->opp.name);
+ }
+ dpif_port_list(p->dpif, &odp_ports, &n_odp_ports);
+ for (i = 0; i < n_odp_ports; i++) {
+ svec_add (&devnames, odp_ports[i].devname);
+ }
+ free(odp_ports);
+
+ svec_sort_unique(&devnames);
+ for (i = 0; i < devnames.n; i++) {
+ update_port(p, devnames.names[i]);
+ }
+ svec_destroy(&devnames);
+ }
+
+ static void
+ refresh_port_group(struct ofproto *p, unsigned int group)
+ {
+ uint16_t *ports;
+ size_t n_ports;
+ struct ofport *port;
+ unsigned int port_no;
+
+ assert(group == DP_GROUP_ALL || group == DP_GROUP_FLOOD);
+
+ ports = xmalloc(port_array_count(&p->ports) * sizeof *ports);
+ n_ports = 0;
+ PORT_ARRAY_FOR_EACH (port, &p->ports, port_no) {
+ if (group == DP_GROUP_ALL || !(port->opp.config & OFPPC_NO_FLOOD)) {
+ ports[n_ports++] = port_no;
+ }
+ }
+ dpif_port_group_set(p->dpif, group, ports, n_ports);
+ free(ports);
+ }
+
+ static void
+ refresh_port_groups(struct ofproto *p)
+ {
+ refresh_port_group(p, DP_GROUP_FLOOD);
+ refresh_port_group(p, DP_GROUP_ALL);
+ }
+
+ static struct ofport *
+ make_ofport(const struct odp_port *odp_port)
+ {
+ enum netdev_flags flags;
+ struct ofport *ofport;
+ struct netdev *netdev;
+ bool carrier;
+ int error;
+
+ error = netdev_open(odp_port->devname, NETDEV_ETH_TYPE_NONE, &netdev);
+ if (error) {
+ VLOG_WARN_RL(&rl, "ignoring port %s (%"PRIu16") because netdev %s "
+ "cannot be opened (%s)",
+ odp_port->devname, odp_port->port,
+ odp_port->devname, strerror(error));
+ return NULL;
+ }
+
+ ofport = xmalloc(sizeof *ofport);
+ ofport->netdev = netdev;
+ ofport->opp.port_no = odp_port_to_ofp_port(odp_port->port);
+ netdev_get_etheraddr(netdev, ofport->opp.hw_addr);
+ memcpy(ofport->opp.name, odp_port->devname,
+ MIN(sizeof ofport->opp.name, sizeof odp_port->devname));
+ ofport->opp.name[sizeof ofport->opp.name - 1] = '\0';
+
+ netdev_get_flags(netdev, &flags);
+ ofport->opp.config = flags & NETDEV_UP ? 0 : OFPPC_PORT_DOWN;
+
+ netdev_get_carrier(netdev, &carrier);
+ ofport->opp.state = carrier ? 0 : OFPPS_LINK_DOWN;
+
+ netdev_get_features(netdev,
+ &ofport->opp.curr, &ofport->opp.advertised,
+ &ofport->opp.supported, &ofport->opp.peer);
+ return ofport;
+ }
+
+ static bool
+ ofport_conflicts(const struct ofproto *p, const struct odp_port *odp_port)
+ {
+ if (port_array_get(&p->ports, odp_port->port)) {
+ VLOG_WARN_RL(&rl, "ignoring duplicate port %"PRIu16" in datapath",
+ odp_port->port);
+ return true;
+ } else if (shash_find(&p->port_by_name, odp_port->devname)) {
+ VLOG_WARN_RL(&rl, "ignoring duplicate device %s in datapath",
+ odp_port->devname);
+ return true;
+ } else {
+ return false;
+ }
+ }
+
+ static int
+ ofport_equal(const struct ofport *a_, const struct ofport *b_)
+ {
+ const struct ofp_phy_port *a = &a_->opp;
+ const struct ofp_phy_port *b = &b_->opp;
+
+ BUILD_ASSERT_DECL(sizeof *a == 48); /* Detect ofp_phy_port changes. */
+ return (a->port_no == b->port_no
+ && !memcmp(a->hw_addr, b->hw_addr, sizeof a->hw_addr)
+ && !strcmp((char *) a->name, (char *) b->name)
+ && a->state == b->state
+ && a->config == b->config
+ && a->curr == b->curr
+ && a->advertised == b->advertised
+ && a->supported == b->supported
+ && a->peer == b->peer);
+ }
+
+ static void
+ send_port_status(struct ofproto *p, const struct ofport *ofport,
+ uint8_t reason)
+ {
+ /* XXX Should limit the number of queued port status change messages. */
+ struct ofconn *ofconn;
+ LIST_FOR_EACH (ofconn, struct ofconn, node, &p->all_conns) {
+ struct ofp_port_status *ops;
+ struct ofpbuf *b;
+
+ ops = make_openflow_xid(sizeof *ops, OFPT_PORT_STATUS, 0, &b);
+ ops->reason = reason;
+ ops->desc = ofport->opp;
+ hton_ofp_phy_port(&ops->desc);
+ queue_tx(b, ofconn, NULL);
+ }
+ if (p->ofhooks->port_changed_cb) {
+ p->ofhooks->port_changed_cb(reason, &ofport->opp, p->aux);
+ }
+ }
+
+ static void
+ ofport_install(struct ofproto *p, struct ofport *ofport)
+ {
+ netdev_monitor_add(p->netdev_monitor, ofport->netdev);
+ port_array_set(&p->ports, ofp_port_to_odp_port(ofport->opp.port_no),
+ ofport);
+ shash_add(&p->port_by_name, (char *) ofport->opp.name, ofport);
+ }
+
+ static void
+ ofport_remove(struct ofproto *p, struct ofport *ofport)
+ {
+ netdev_monitor_remove(p->netdev_monitor, ofport->netdev);
+ port_array_set(&p->ports, ofp_port_to_odp_port(ofport->opp.port_no), NULL);
+ shash_delete(&p->port_by_name,
+ shash_find(&p->port_by_name, (char *) ofport->opp.name));
+ }
+
+ static void
+ ofport_free(struct ofport *ofport)
+ {
+ if (ofport) {
+ netdev_close(ofport->netdev);
+ free(ofport);
+ }
+ }
+
+ static void
+ update_port(struct ofproto *p, const char *devname)
+ {
+ struct odp_port odp_port;
- ofport = shash_find_data(&p->port_by_name, devname);
++ struct ofport *old_ofport;
++ struct ofport *new_ofport;
+ int error;
+
+ COVERAGE_INC(ofproto_update_port);
- if (!error) {
- if (!ofport) {
- /* New port. */
- if (!ofport_conflicts(p, &odp_port)) {
- ofport = make_ofport(&odp_port);
- if (ofport) {
- ofport_install(p, ofport);
- send_port_status(p, ofport, OFPPR_ADD);
- }
- }
- } else {
- /* Modified port. */
- struct ofport *new_ofport = make_ofport(&odp_port);
- if (!new_ofport) {
- return;
- }
++
++ /* Query the datapath for port information. */
+ error = dpif_port_query_by_name(p->dpif, devname, &odp_port);
- new_ofport->opp.config &= OFPPC_PORT_DOWN;
- new_ofport->opp.config |= ofport->opp.config & ~OFPPC_PORT_DOWN;
- if (ofport_equal(ofport, new_ofport)) {
- /* False alarm--no change. */
- ofport_free(new_ofport);
- } else {
- ofport_remove(p, ofport);
- ofport_install(p, new_ofport);
- ofport_free(ofport);
- send_port_status(p, new_ofport, OFPPR_MODIFY);
- }
- }
- } else if (error == ENOENT || error == ENODEV) {
- /* Deleted port. */
- if (ofport) {
- send_port_status(p, ofport, OFPPR_DELETE);
- ofport_remove(p, ofport);
- ofport_free(ofport);
+
- } else {
++ /* Find the old ofport. */
++ old_ofport = shash_find_data(&p->port_by_name, devname);
++ if (!error) {
++ if (!old_ofport) {
++ /* There's no port named 'devname' but there might be a port with
++ * the same port number. This could happen if a port is deleted
++ * and then a new one added in its place very quickly, or if a port
++ * is renamed. In the former case we want to send an OFPPR_DELETE
++ * and an OFPPR_ADD, and in the latter case we want to send a
++ * single OFPPR_MODIFY. We can distinguish the cases by comparing
++ * the old port's ifindex against the new port, or perhaps less
++ * reliably but more portably by comparing the old port's MAC
++ * against the new port's MAC. However, this code isn't that smart
++ * and always sends an OFPPR_MODIFY (XXX). */
++ old_ofport = port_array_get(&p->ports, odp_port.port);
+ }
++ } else if (error != ENOENT && error != ENODEV) {
+ VLOG_WARN_RL(&rl, "dpif_port_query_by_name returned unexpected error "
+ "%s", strerror(error));
+ return;
+ }
++
++ /* Create a new ofport. */
++ new_ofport = !error ? make_ofport(&odp_port) : NULL;
++
++ /* Eliminate a few pathological cases. */
++ if (!old_ofport && !new_ofport) {
++ return;
++ } else if (old_ofport && new_ofport) {
++ /* Most of the 'config' bits are OpenFlow soft state, but
++ * OFPPC_PORT_DOWN is maintained the kernel. So transfer the OpenFlow
++ * bits from old_ofport. (make_ofport() only sets OFPPC_PORT_DOWN and
++ * leaves the other bits 0.) */
++ new_ofport->opp.config |= old_ofport->opp.config & ~OFPPC_PORT_DOWN;
++
++ if (ofport_equal(old_ofport, new_ofport)) {
++ /* False alarm--no change. */
++ ofport_free(new_ofport);
++ return;
++ }
++ }
++
++ /* Now deal with the normal cases. */
++ if (old_ofport) {
++ ofport_remove(p, old_ofport);
++ }
++ if (new_ofport) {
++ ofport_install(p, new_ofport);
++ }
++ send_port_status(p, new_ofport ? new_ofport : old_ofport,
++ (!old_ofport ? OFPPR_ADD
++ : !new_ofport ? OFPPR_DELETE
++ : OFPPR_MODIFY));
++ ofport_free(old_ofport);
++
++ /* Update port groups. */
+ refresh_port_groups(p);
+ }
+
+ static int
+ init_ports(struct ofproto *p)
+ {
+ struct odp_port *ports;
+ size_t n_ports;
+ size_t i;
+ int error;
+
+ error = dpif_port_list(p->dpif, &ports, &n_ports);
+ if (error) {
+ return error;
+ }
+
+ for (i = 0; i < n_ports; i++) {
+ const struct odp_port *odp_port = &ports[i];
+ if (!ofport_conflicts(p, odp_port)) {
+ struct ofport *ofport = make_ofport(odp_port);
+ if (ofport) {
+ ofport_install(p, ofport);
+ }
+ }
+ }
+ free(ports);
+ refresh_port_groups(p);
+ return 0;
+ }
+ \f
+ static struct ofconn *
+ ofconn_create(struct ofproto *p, struct rconn *rconn)
+ {
+ struct ofconn *ofconn = xmalloc(sizeof *ofconn);
+ list_push_back(&p->all_conns, &ofconn->node);
+ ofconn->rconn = rconn;
+ ofconn->pktbuf = NULL;
+ ofconn->send_flow_exp = false;
+ ofconn->miss_send_len = 0;
+ ofconn->packet_in_counter = rconn_packet_counter_create ();
+ ofconn->reply_counter = rconn_packet_counter_create ();
+ return ofconn;
+ }
+
+ static void
+ ofconn_destroy(struct ofconn *ofconn, struct ofproto *p)
+ {
+ if (p->executer) {
+ executer_rconn_closing(p->executer, ofconn->rconn);
+ }
+
+ list_remove(&ofconn->node);
+ rconn_destroy(ofconn->rconn);
+ rconn_packet_counter_destroy(ofconn->packet_in_counter);
+ rconn_packet_counter_destroy(ofconn->reply_counter);
+ pktbuf_destroy(ofconn->pktbuf);
+ free(ofconn);
+ }
+
+ static void
+ ofconn_run(struct ofconn *ofconn, struct ofproto *p)
+ {
+ int iteration;
+
+ rconn_run(ofconn->rconn);
+
+ if (rconn_packet_counter_read (ofconn->reply_counter) < OFCONN_REPLY_MAX) {
+ /* Limit the number of iterations to prevent other tasks from
+ * starving. */
+ for (iteration = 0; iteration < 50; iteration++) {
+ struct ofpbuf *of_msg = rconn_recv(ofconn->rconn);
+ if (!of_msg) {
+ break;
+ }
+ handle_openflow(ofconn, p, of_msg);
+ ofpbuf_delete(of_msg);
+ }
+ }
+
+ if (ofconn != p->controller && !rconn_is_alive(ofconn->rconn)) {
+ ofconn_destroy(ofconn, p);
+ }
+ }
+
+ static void
+ ofconn_wait(struct ofconn *ofconn)
+ {
+ rconn_run_wait(ofconn->rconn);
+ if (rconn_packet_counter_read (ofconn->reply_counter) < OFCONN_REPLY_MAX) {
+ rconn_recv_wait(ofconn->rconn);
+ } else {
+ COVERAGE_INC(ofproto_ofconn_stuck);
+ }
+ }
+ \f
+ /* Caller is responsible for initializing the 'cr' member of the returned
+ * rule. */
+ static struct rule *
+ rule_create(struct rule *super,
+ const union ofp_action *actions, size_t n_actions,
+ uint16_t idle_timeout, uint16_t hard_timeout)
+ {
+ struct rule *rule = xcalloc(1, sizeof *rule);
+ rule->idle_timeout = idle_timeout;
+ rule->hard_timeout = hard_timeout;
+ rule->used = rule->created = time_msec();
+ rule->super = super;
+ if (super) {
+ list_push_back(&super->list, &rule->list);
+ } else {
+ list_init(&rule->list);
+ }
+ rule->n_actions = n_actions;
+ rule->actions = xmemdup(actions, n_actions * sizeof *actions);
+ return rule;
+ }
+
+ static struct rule *
+ rule_from_cls_rule(const struct cls_rule *cls_rule)
+ {
+ return cls_rule ? CONTAINER_OF(cls_rule, struct rule, cr) : NULL;
+ }
+
+ static void
+ rule_free(struct rule *rule)
+ {
+ free(rule->actions);
+ free(rule->odp_actions);
+ free(rule);
+ }
+
+ /* Destroys 'rule'. If 'rule' is a subrule, also removes it from its
+ * super-rule's list of subrules. If 'rule' is a super-rule, also iterates
+ * through all of its subrules and revalidates them, destroying any that no
+ * longer has a super-rule (which is probably all of them).
+ *
+ * Before calling this function, the caller must make have removed 'rule' from
+ * the classifier. If 'rule' is an exact-match rule, the caller is also
+ * responsible for ensuring that it has been uninstalled from the datapath. */
+ static void
+ rule_destroy(struct ofproto *ofproto, struct rule *rule)
+ {
+ if (!rule->super) {
+ struct rule *subrule, *next;
+ LIST_FOR_EACH_SAFE (subrule, next, struct rule, list, &rule->list) {
+ revalidate_rule(ofproto, subrule);
+ }
+ } else {
+ list_remove(&rule->list);
+ }
+ rule_free(rule);
+ }
+
+ static bool
+ rule_has_out_port(const struct rule *rule, uint16_t out_port)
+ {
+ const union ofp_action *oa;
+ struct actions_iterator i;
+
+ if (out_port == htons(OFPP_NONE)) {
+ return true;
+ }
+ for (oa = actions_first(&i, rule->actions, rule->n_actions); oa;
+ oa = actions_next(&i)) {
+ if (oa->type == htons(OFPAT_OUTPUT) && oa->output.port == out_port) {
+ return true;
+ }
+ }
+ return false;
+ }
+
+ /* Executes the actions indicated by 'rule' on 'packet', which is in flow
+ * 'flow' and is considered to have arrived on ODP port 'in_port'.
+ *
+ * The flow that 'packet' actually contains does not need to actually match
+ * 'rule'; the actions in 'rule' will be applied to it either way. Likewise,
+ * the packet and byte counters for 'rule' will be credited for the packet sent
+ * out whether or not the packet actually matches 'rule'.
+ *
+ * If 'rule' is an exact-match rule and 'flow' actually equals the rule's flow,
+ * the caller must already have accurately composed ODP actions for it given
+ * 'packet' using rule_make_actions(). If 'rule' is a wildcard rule, or if
+ * 'rule' is an exact-match rule but 'flow' is not the rule's flow, then this
+ * function will compose a set of ODP actions based on 'rule''s OpenFlow
+ * actions and apply them to 'packet'. */
+ static void
+ rule_execute(struct ofproto *ofproto, struct rule *rule,
+ struct ofpbuf *packet, const flow_t *flow)
+ {
+ const union odp_action *actions;
+ size_t n_actions;
+ struct odp_actions a;
+
+ /* Grab or compose the ODP actions.
+ *
+ * The special case for an exact-match 'rule' where 'flow' is not the
+ * rule's flow is important to avoid, e.g., sending a packet out its input
+ * port simply because the ODP actions were composed for the wrong
+ * scenario. */
+ if (rule->cr.wc.wildcards || !flow_equal(flow, &rule->cr.flow)) {
+ struct rule *super = rule->super ? rule->super : rule;
+ if (xlate_actions(super->actions, super->n_actions, flow, ofproto,
+ packet, &a, NULL, 0)) {
+ return;
+ }
+ actions = a.actions;
+ n_actions = a.n_actions;
+ } else {
+ actions = rule->odp_actions;
+ n_actions = rule->n_odp_actions;
+ }
+
+ /* Execute the ODP actions. */
+ if (!dpif_execute(ofproto->dpif, flow->in_port,
+ actions, n_actions, packet)) {
+ struct odp_flow_stats stats;
+ flow_extract_stats(flow, packet, &stats);
+ update_stats(rule, &stats);
+ rule->used = time_msec();
+ }
+ }
+
+ static void
+ rule_insert(struct ofproto *p, struct rule *rule, struct ofpbuf *packet,
+ uint16_t in_port)
+ {
+ struct rule *displaced_rule;
+
+ /* Insert the rule in the classifier. */
+ displaced_rule = rule_from_cls_rule(classifier_insert(&p->cls, &rule->cr));
+ if (!rule->cr.wc.wildcards) {
+ rule_make_actions(p, rule, packet);
+ }
+
+ /* Send the packet and credit it to the rule. */
+ if (packet) {
+ flow_t flow;
+ flow_extract(packet, in_port, &flow);
+ rule_execute(p, rule, packet, &flow);
+ }
+
+ /* Install the rule in the datapath only after sending the packet, to
+ * avoid packet reordering. */
+ if (rule->cr.wc.wildcards) {
+ COVERAGE_INC(ofproto_add_wc_flow);
+ p->need_revalidate = true;
+ } else {
+ rule_install(p, rule, displaced_rule);
+ }
+
+ /* Free the rule that was displaced, if any. */
+ if (displaced_rule) {
+ rule_destroy(p, displaced_rule);
+ }
+ }
+
+ static struct rule *
+ rule_create_subrule(struct ofproto *ofproto, struct rule *rule,
+ const flow_t *flow)
+ {
+ struct rule *subrule = rule_create(rule, NULL, 0,
+ rule->idle_timeout, rule->hard_timeout);
+ COVERAGE_INC(ofproto_subrule_create);
+ cls_rule_from_flow(&subrule->cr, flow, 0,
+ (rule->cr.priority <= UINT16_MAX ? UINT16_MAX
+ : rule->cr.priority));
+ classifier_insert_exact(&ofproto->cls, &subrule->cr);
+
+ return subrule;
+ }
+
+ static void
+ rule_remove(struct ofproto *ofproto, struct rule *rule)
+ {
+ if (rule->cr.wc.wildcards) {
+ COVERAGE_INC(ofproto_del_wc_flow);
+ ofproto->need_revalidate = true;
+ } else {
+ rule_uninstall(ofproto, rule);
+ }
+ classifier_remove(&ofproto->cls, &rule->cr);
+ rule_destroy(ofproto, rule);
+ }
+
+ /* Returns true if the actions changed, false otherwise. */
+ static bool
+ rule_make_actions(struct ofproto *p, struct rule *rule,
+ const struct ofpbuf *packet)
+ {
+ const struct rule *super;
+ struct odp_actions a;
+ size_t actions_len;
+
+ assert(!rule->cr.wc.wildcards);
+
+ super = rule->super ? rule->super : rule;
+ rule->tags = 0;
+ xlate_actions(super->actions, super->n_actions, &rule->cr.flow, p,
+ packet, &a, &rule->tags, &rule->may_install);
+
+ actions_len = a.n_actions * sizeof *a.actions;
+ if (rule->n_odp_actions != a.n_actions
+ || memcmp(rule->odp_actions, a.actions, actions_len)) {
+ COVERAGE_INC(ofproto_odp_unchanged);
+ free(rule->odp_actions);
+ rule->n_odp_actions = a.n_actions;
+ rule->odp_actions = xmemdup(a.actions, actions_len);
+ return true;
+ } else {
+ return false;
+ }
+ }
+
+ static int
+ do_put_flow(struct ofproto *ofproto, struct rule *rule, int flags,
+ struct odp_flow_put *put)
+ {
+ memset(&put->flow.stats, 0, sizeof put->flow.stats);
+ put->flow.key = rule->cr.flow;
+ put->flow.actions = rule->odp_actions;
+ put->flow.n_actions = rule->n_odp_actions;
+ put->flags = flags;
+ return dpif_flow_put(ofproto->dpif, put);
+ }
+
+ static void
+ rule_install(struct ofproto *p, struct rule *rule, struct rule *displaced_rule)
+ {
+ assert(!rule->cr.wc.wildcards);
+
+ if (rule->may_install) {
+ struct odp_flow_put put;
+ if (!do_put_flow(p, rule,
+ ODPPF_CREATE | ODPPF_MODIFY | ODPPF_ZERO_STATS,
+ &put)) {
+ rule->installed = true;
+ if (displaced_rule) {
+ update_stats(rule, &put.flow.stats);
+ rule_post_uninstall(p, displaced_rule);
+ }
+ }
+ } else if (displaced_rule) {
+ rule_uninstall(p, displaced_rule);
+ }
+ }
+
+ static void
+ rule_reinstall(struct ofproto *ofproto, struct rule *rule)
+ {
+ if (rule->installed) {
+ struct odp_flow_put put;
+ COVERAGE_INC(ofproto_dp_missed);
+ do_put_flow(ofproto, rule, ODPPF_CREATE | ODPPF_MODIFY, &put);
+ } else {
+ rule_install(ofproto, rule, NULL);
+ }
+ }
+
+ static void
+ rule_update_actions(struct ofproto *ofproto, struct rule *rule)
+ {
+ bool actions_changed = rule_make_actions(ofproto, rule, NULL);
+ if (rule->may_install) {
+ if (rule->installed) {
+ if (actions_changed) {
+ /* XXX should really do rule_post_uninstall() for the *old* set
+ * of actions, and distinguish the old stats from the new. */
+ struct odp_flow_put put;
+ do_put_flow(ofproto, rule, ODPPF_CREATE | ODPPF_MODIFY, &put);
+ }
+ } else {
+ rule_install(ofproto, rule, NULL);
+ }
+ } else {
+ rule_uninstall(ofproto, rule);
+ }
+ }
+
+ static void
+ rule_account(struct ofproto *ofproto, struct rule *rule, uint64_t extra_bytes)
+ {
+ uint64_t total_bytes = rule->byte_count + extra_bytes;
+
+ if (ofproto->ofhooks->account_flow_cb
+ && total_bytes > rule->accounted_bytes)
+ {
+ ofproto->ofhooks->account_flow_cb(
+ &rule->cr.flow, rule->odp_actions, rule->n_odp_actions,
+ total_bytes - rule->accounted_bytes, ofproto->aux);
+ rule->accounted_bytes = total_bytes;
+ }
+ }
+
+ static void
+ rule_uninstall(struct ofproto *p, struct rule *rule)
+ {
+ assert(!rule->cr.wc.wildcards);
+ if (rule->installed) {
+ struct odp_flow odp_flow;
+
+ odp_flow.key = rule->cr.flow;
+ odp_flow.actions = NULL;
+ odp_flow.n_actions = 0;
+ if (!dpif_flow_del(p->dpif, &odp_flow)) {
+ update_stats(rule, &odp_flow.stats);
+ }
+ rule->installed = false;
+
+ rule_post_uninstall(p, rule);
+ }
+ }
+
+ static void
+ rule_post_uninstall(struct ofproto *ofproto, struct rule *rule)
+ {
+ struct rule *super = rule->super;
+
+ rule_account(ofproto, rule, 0);
+ if (ofproto->netflow) {
+ struct ofexpired expired;
+ expired.flow = rule->cr.flow;
+ expired.packet_count = rule->packet_count;
+ expired.byte_count = rule->byte_count;
+ expired.used = rule->used;
+ expired.created = rule->created;
+ expired.tcp_flags = rule->tcp_flags;
+ expired.ip_tos = rule->ip_tos;
+ netflow_expire(ofproto->netflow, &expired);
+ }
+ if (super) {
+ super->packet_count += rule->packet_count;
+ super->byte_count += rule->byte_count;
+ super->tcp_flags |= rule->tcp_flags;
+ if (rule->packet_count) {
+ super->ip_tos = rule->ip_tos;
+ }
+ }
+
+ /* Reset counters to prevent double counting if the rule ever gets
+ * reinstalled. */
+ rule->packet_count = 0;
+ rule->byte_count = 0;
+ rule->accounted_bytes = 0;
+ rule->tcp_flags = 0;
+ rule->ip_tos = 0;
+ }
+ \f
+ static void
+ queue_tx(struct ofpbuf *msg, const struct ofconn *ofconn,
+ struct rconn_packet_counter *counter)
+ {
+ update_openflow_length(msg);
+ if (rconn_send(ofconn->rconn, msg, counter)) {
+ ofpbuf_delete(msg);
+ }
+ }
+
+ static void
+ send_error(const struct ofconn *ofconn, const struct ofp_header *oh,
+ int error, const void *data, size_t len)
+ {
+ struct ofpbuf *buf;
+ struct ofp_error_msg *oem;
+
+ if (!(error >> 16)) {
+ VLOG_WARN_RL(&rl, "not sending bad error code %d to controller",
+ error);
+ return;
+ }
+
+ COVERAGE_INC(ofproto_error);
+ oem = make_openflow_xid(len + sizeof *oem, OFPT_ERROR,
+ oh ? oh->xid : 0, &buf);
+ oem->type = htons((unsigned int) error >> 16);
+ oem->code = htons(error & 0xffff);
+ memcpy(oem->data, data, len);
+ queue_tx(buf, ofconn, ofconn->reply_counter);
+ }
+
+ static void
+ send_error_oh(const struct ofconn *ofconn, const struct ofp_header *oh,
+ int error)
+ {
+ size_t oh_length = ntohs(oh->length);
+ send_error(ofconn, oh, error, oh, MIN(oh_length, 64));
+ }
+
+ static void
+ hton_ofp_phy_port(struct ofp_phy_port *opp)
+ {
+ opp->port_no = htons(opp->port_no);
+ opp->config = htonl(opp->config);
+ opp->state = htonl(opp->state);
+ opp->curr = htonl(opp->curr);
+ opp->advertised = htonl(opp->advertised);
+ opp->supported = htonl(opp->supported);
+ opp->peer = htonl(opp->peer);
+ }
+
+ static int
+ handle_echo_request(struct ofconn *ofconn, struct ofp_header *oh)
+ {
+ struct ofp_header *rq = oh;
+ queue_tx(make_echo_reply(rq), ofconn, ofconn->reply_counter);
+ return 0;
+ }
+
+ static int
+ handle_features_request(struct ofproto *p, struct ofconn *ofconn,
+ struct ofp_header *oh)
+ {
+ struct ofp_switch_features *osf;
+ struct ofpbuf *buf;
+ unsigned int port_no;
+ struct ofport *port;
+
+ osf = make_openflow_xid(sizeof *osf, OFPT_FEATURES_REPLY, oh->xid, &buf);
+ osf->datapath_id = htonll(p->datapath_id);
+ osf->n_buffers = htonl(pktbuf_capacity());
+ osf->n_tables = 2;
+ osf->capabilities = htonl(OFPC_FLOW_STATS | OFPC_TABLE_STATS |
+ OFPC_PORT_STATS | OFPC_MULTI_PHY_TX);
+ osf->actions = htonl((1u << OFPAT_OUTPUT) |
+ (1u << OFPAT_SET_VLAN_VID) |
+ (1u << OFPAT_SET_VLAN_PCP) |
+ (1u << OFPAT_STRIP_VLAN) |
+ (1u << OFPAT_SET_DL_SRC) |
+ (1u << OFPAT_SET_DL_DST) |
+ (1u << OFPAT_SET_NW_SRC) |
+ (1u << OFPAT_SET_NW_DST) |
+ (1u << OFPAT_SET_TP_SRC) |
+ (1u << OFPAT_SET_TP_DST));
+
+ PORT_ARRAY_FOR_EACH (port, &p->ports, port_no) {
+ hton_ofp_phy_port(ofpbuf_put(buf, &port->opp, sizeof port->opp));
+ }
+
+ queue_tx(buf, ofconn, ofconn->reply_counter);
+ return 0;
+ }
+
+ static int
+ handle_get_config_request(struct ofproto *p, struct ofconn *ofconn,
+ struct ofp_header *oh)
+ {
+ struct ofpbuf *buf;
+ struct ofp_switch_config *osc;
+ uint16_t flags;
+ bool drop_frags;
+
+ /* Figure out flags. */
+ dpif_get_drop_frags(p->dpif, &drop_frags);
+ flags = drop_frags ? OFPC_FRAG_DROP : OFPC_FRAG_NORMAL;
+ if (ofconn->send_flow_exp) {
+ flags |= OFPC_SEND_FLOW_EXP;
+ }
+
+ /* Send reply. */
+ osc = make_openflow_xid(sizeof *osc, OFPT_GET_CONFIG_REPLY, oh->xid, &buf);
+ osc->flags = htons(flags);
+ osc->miss_send_len = htons(ofconn->miss_send_len);
+ queue_tx(buf, ofconn, ofconn->reply_counter);
+
+ return 0;
+ }
+
+ static int
+ handle_set_config(struct ofproto *p, struct ofconn *ofconn,
+ struct ofp_switch_config *osc)
+ {
+ uint16_t flags;
+ int error;
+
+ error = check_ofp_message(&osc->header, OFPT_SET_CONFIG, sizeof *osc);
+ if (error) {
+ return error;
+ }
+ flags = ntohs(osc->flags);
+
+ ofconn->send_flow_exp = (flags & OFPC_SEND_FLOW_EXP) != 0;
+
+ if (ofconn == p->controller) {
+ switch (flags & OFPC_FRAG_MASK) {
+ case OFPC_FRAG_NORMAL:
+ dpif_set_drop_frags(p->dpif, false);
+ break;
+ case OFPC_FRAG_DROP:
+ dpif_set_drop_frags(p->dpif, true);
+ break;
+ default:
+ VLOG_WARN_RL(&rl, "requested bad fragment mode (flags=%"PRIx16")",
+ osc->flags);
+ break;
+ }
+ }
+
+ if ((ntohs(osc->miss_send_len) != 0) != (ofconn->miss_send_len != 0)) {
+ if (ntohs(osc->miss_send_len) != 0) {
+ ofconn->pktbuf = pktbuf_create();
+ } else {
+ pktbuf_destroy(ofconn->pktbuf);
+ }
+ }
+
+ ofconn->miss_send_len = ntohs(osc->miss_send_len);
+
+ return 0;
+ }
+
+ static void
+ add_output_group_action(struct odp_actions *actions, uint16_t group)
+ {
+ odp_actions_add(actions, ODPAT_OUTPUT_GROUP)->output_group.group = group;
+ }
+
+ static void
+ add_controller_action(struct odp_actions *actions,
+ const struct ofp_action_output *oao)
+ {
+ union odp_action *a = odp_actions_add(actions, ODPAT_CONTROLLER);
+ a->controller.arg = oao->max_len ? ntohs(oao->max_len) : UINT32_MAX;
+ }
+
+ struct action_xlate_ctx {
+ /* Input. */
+ const flow_t *flow; /* Flow to which these actions correspond. */
+ int recurse; /* Recursion level, via xlate_table_action. */
+ struct ofproto *ofproto;
+ const struct ofpbuf *packet; /* The packet corresponding to 'flow', or a
+ * null pointer if we are revalidating
+ * without a packet to refer to. */
+
+ /* Output. */
+ struct odp_actions *out; /* Datapath actions. */
+ tag_type *tags; /* Tags associated with OFPP_NORMAL actions. */
+ bool may_setup_flow; /* True ordinarily; false if the actions must
+ * be reassessed for every packet. */
+ };
+
+ static void do_xlate_actions(const union ofp_action *in, size_t n_in,
+ struct action_xlate_ctx *ctx);
+
+ static void
+ add_output_action(struct action_xlate_ctx *ctx, uint16_t port)
+ {
+ const struct ofport *ofport = port_array_get(&ctx->ofproto->ports, port);
+ if (!ofport || !(ofport->opp.config & OFPPC_NO_FWD)) {
+ odp_actions_add(ctx->out, ODPAT_OUTPUT)->output.port = port;
+ }
+ }
+
+ static struct rule *
+ lookup_valid_rule(struct ofproto *ofproto, const flow_t *flow)
+ {
+ struct rule *rule;
+ rule = rule_from_cls_rule(classifier_lookup(&ofproto->cls, flow));
+
+ /* The rule we found might not be valid, since we could be in need of
+ * revalidation. If it is not valid, don't return it. */
+ if (rule
+ && rule->super
+ && ofproto->need_revalidate
+ && !revalidate_rule(ofproto, rule)) {
+ COVERAGE_INC(ofproto_invalidated);
+ return NULL;
+ }
+
+ return rule;
+ }
+
+ static void
+ xlate_table_action(struct action_xlate_ctx *ctx, uint16_t in_port)
+ {
+ if (!ctx->recurse) {
+ struct rule *rule;
+ flow_t flow;
+
+ flow = *ctx->flow;
+ flow.in_port = in_port;
+
+ rule = lookup_valid_rule(ctx->ofproto, &flow);
+ if (rule) {
+ if (rule->super) {
+ rule = rule->super;
+ }
+
+ ctx->recurse++;
+ do_xlate_actions(rule->actions, rule->n_actions, ctx);
+ ctx->recurse--;
+ }
+ }
+ }
+
+ static void
+ xlate_output_action(struct action_xlate_ctx *ctx,
+ const struct ofp_action_output *oao)
+ {
+ uint16_t odp_port;
+
+ switch (ntohs(oao->port)) {
+ case OFPP_IN_PORT:
+ add_output_action(ctx, ctx->flow->in_port);
+ break;
+ case OFPP_TABLE:
+ xlate_table_action(ctx, ctx->flow->in_port);
+ break;
+ case OFPP_NORMAL:
+ if (!ctx->ofproto->ofhooks->normal_cb(ctx->flow, ctx->packet,
+ ctx->out, ctx->tags,
+ ctx->ofproto->aux)) {
+ COVERAGE_INC(ofproto_uninstallable);
+ ctx->may_setup_flow = false;
+ }
+ break;
+ case OFPP_FLOOD:
+ add_output_group_action(ctx->out, DP_GROUP_FLOOD);
+ break;
+ case OFPP_ALL:
+ add_output_group_action(ctx->out, DP_GROUP_ALL);
+ break;
+ case OFPP_CONTROLLER:
+ add_controller_action(ctx->out, oao);
+ break;
+ case OFPP_LOCAL:
+ add_output_action(ctx, ODPP_LOCAL);
+ break;
+ default:
+ odp_port = ofp_port_to_odp_port(ntohs(oao->port));
+ if (odp_port != ctx->flow->in_port) {
+ add_output_action(ctx, odp_port);
+ }
+ break;
+ }
+ }
+
+ static void
+ xlate_nicira_action(struct action_xlate_ctx *ctx,
+ const struct nx_action_header *nah)
+ {
+ const struct nx_action_resubmit *nar;
+ int subtype = ntohs(nah->subtype);
+
+ assert(nah->vendor == htonl(NX_VENDOR_ID));
+ switch (subtype) {
+ case NXAST_RESUBMIT:
+ nar = (const struct nx_action_resubmit *) nah;
+ xlate_table_action(ctx, ofp_port_to_odp_port(ntohs(nar->in_port)));
+ break;
+
+ default:
+ VLOG_DBG_RL(&rl, "unknown Nicira action type %"PRIu16, subtype);
+ break;
+ }
+ }
+
+ static void
+ do_xlate_actions(const union ofp_action *in, size_t n_in,
+ struct action_xlate_ctx *ctx)
+ {
+ struct actions_iterator iter;
+ const union ofp_action *ia;
+ const struct ofport *port;
+
+ port = port_array_get(&ctx->ofproto->ports, ctx->flow->in_port);
+ if (port && port->opp.config & (OFPPC_NO_RECV | OFPPC_NO_RECV_STP) &&
+ port->opp.config & (eth_addr_equals(ctx->flow->dl_dst, stp_eth_addr)
+ ? OFPPC_NO_RECV_STP : OFPPC_NO_RECV)) {
+ /* Drop this flow. */
+ return;
+ }
+
+ for (ia = actions_first(&iter, in, n_in); ia; ia = actions_next(&iter)) {
+ uint16_t type = ntohs(ia->type);
+ union odp_action *oa;
+
+ switch (type) {
+ case OFPAT_OUTPUT:
+ xlate_output_action(ctx, &ia->output);
+ break;
+
+ case OFPAT_SET_VLAN_VID:
+ oa = odp_actions_add(ctx->out, ODPAT_SET_VLAN_VID);
+ oa->vlan_vid.vlan_vid = ia->vlan_vid.vlan_vid;
+ break;
+
+ case OFPAT_SET_VLAN_PCP:
+ oa = odp_actions_add(ctx->out, ODPAT_SET_VLAN_PCP);
+ oa->vlan_pcp.vlan_pcp = ia->vlan_pcp.vlan_pcp;
+ break;
+
+ case OFPAT_STRIP_VLAN:
+ odp_actions_add(ctx->out, ODPAT_STRIP_VLAN);
+ break;
+
+ case OFPAT_SET_DL_SRC:
+ oa = odp_actions_add(ctx->out, ODPAT_SET_DL_SRC);
+ memcpy(oa->dl_addr.dl_addr,
+ ((struct ofp_action_dl_addr *) ia)->dl_addr, ETH_ADDR_LEN);
+ break;
+
+ case OFPAT_SET_DL_DST:
+ oa = odp_actions_add(ctx->out, ODPAT_SET_DL_DST);
+ memcpy(oa->dl_addr.dl_addr,
+ ((struct ofp_action_dl_addr *) ia)->dl_addr, ETH_ADDR_LEN);
+ break;
+
+ case OFPAT_SET_NW_SRC:
+ oa = odp_actions_add(ctx->out, ODPAT_SET_NW_SRC);
+ oa->nw_addr.nw_addr = ia->nw_addr.nw_addr;
+ break;
+
+ case OFPAT_SET_TP_SRC:
+ oa = odp_actions_add(ctx->out, ODPAT_SET_TP_SRC);
+ oa->tp_port.tp_port = ia->tp_port.tp_port;
+ break;
+
+ case OFPAT_VENDOR:
+ xlate_nicira_action(ctx, (const struct nx_action_header *) ia);
+ break;
+
+ default:
+ VLOG_DBG_RL(&rl, "unknown action type %"PRIu16, type);
+ break;
+ }
+ }
+ }
+
+ static int
+ xlate_actions(const union ofp_action *in, size_t n_in,
+ const flow_t *flow, struct ofproto *ofproto,
+ const struct ofpbuf *packet,
+ struct odp_actions *out, tag_type *tags, bool *may_setup_flow)
+ {
+ tag_type no_tags = 0;
+ struct action_xlate_ctx ctx;
+ COVERAGE_INC(ofproto_ofp2odp);
+ odp_actions_init(out);
+ ctx.flow = flow;
+ ctx.recurse = 0;
+ ctx.ofproto = ofproto;
+ ctx.packet = packet;
+ ctx.out = out;
+ ctx.tags = tags ? tags : &no_tags;
+ ctx.may_setup_flow = true;
+ do_xlate_actions(in, n_in, &ctx);
+ if (may_setup_flow) {
+ *may_setup_flow = ctx.may_setup_flow;
+ }
+ if (odp_actions_overflow(out)) {
+ odp_actions_init(out);
+ return ofp_mkerr(OFPET_BAD_ACTION, OFPBAC_TOO_MANY);
+ }
+ return 0;
+ }
+
+ static int
+ handle_packet_out(struct ofproto *p, struct ofconn *ofconn,
+ struct ofp_header *oh)
+ {
+ struct ofp_packet_out *opo;
+ struct ofpbuf payload, *buffer;
+ struct odp_actions actions;
+ int n_actions;
+ uint16_t in_port;
+ flow_t flow;
+ int error;
+
+ error = check_ofp_packet_out(oh, &payload, &n_actions, p->max_ports);
+ if (error) {
+ return error;
+ }
+ opo = (struct ofp_packet_out *) oh;
+
+ COVERAGE_INC(ofproto_packet_out);
+ if (opo->buffer_id != htonl(UINT32_MAX)) {
+ error = pktbuf_retrieve(ofconn->pktbuf, ntohl(opo->buffer_id),
+ &buffer, &in_port);
+ if (error) {
+ return error;
+ }
+ payload = *buffer;
+ } else {
+ buffer = NULL;
+ }
+
+ flow_extract(&payload, ofp_port_to_odp_port(ntohs(opo->in_port)), &flow);
+ error = xlate_actions((const union ofp_action *) opo->actions, n_actions,
+ &flow, p, &payload, &actions, NULL, NULL);
+ if (error) {
+ return error;
+ }
+
+ dpif_execute(p->dpif, flow.in_port, actions.actions, actions.n_actions,
+ &payload);
+ ofpbuf_delete(buffer);
+
+ return 0;
+ }
+
+ static void
+ update_port_config(struct ofproto *p, struct ofport *port,
+ uint32_t config, uint32_t mask)
+ {
+ mask &= config ^ port->opp.config;
+ if (mask & OFPPC_PORT_DOWN) {
+ if (config & OFPPC_PORT_DOWN) {
+ netdev_turn_flags_off(port->netdev, NETDEV_UP, true);
+ } else {
+ netdev_turn_flags_on(port->netdev, NETDEV_UP, true);
+ }
+ }
+ #define REVALIDATE_BITS (OFPPC_NO_RECV | OFPPC_NO_RECV_STP | OFPPC_NO_FWD)
+ if (mask & REVALIDATE_BITS) {
+ COVERAGE_INC(ofproto_costly_flags);
+ port->opp.config ^= mask & REVALIDATE_BITS;
+ p->need_revalidate = true;
+ }
+ #undef REVALIDATE_BITS
+ if (mask & OFPPC_NO_FLOOD) {
+ port->opp.config ^= OFPPC_NO_FLOOD;
+ refresh_port_group(p, DP_GROUP_FLOOD);
+ }
+ if (mask & OFPPC_NO_PACKET_IN) {
+ port->opp.config ^= OFPPC_NO_PACKET_IN;
+ }
+ }
+
+ static int
+ handle_port_mod(struct ofproto *p, struct ofp_header *oh)
+ {
+ const struct ofp_port_mod *opm;
+ struct ofport *port;
+ int error;
+
+ error = check_ofp_message(oh, OFPT_PORT_MOD, sizeof *opm);
+ if (error) {
+ return error;
+ }
+ opm = (struct ofp_port_mod *) oh;
+
+ port = port_array_get(&p->ports,
+ ofp_port_to_odp_port(ntohs(opm->port_no)));
+ if (!port) {
+ return ofp_mkerr(OFPET_PORT_MOD_FAILED, OFPPMFC_BAD_PORT);
+ } else if (memcmp(port->opp.hw_addr, opm->hw_addr, OFP_ETH_ALEN)) {
+ return ofp_mkerr(OFPET_PORT_MOD_FAILED, OFPPMFC_BAD_HW_ADDR);
+ } else {
+ update_port_config(p, port, ntohl(opm->config), ntohl(opm->mask));
+ if (opm->advertise) {
+ netdev_set_advertisements(port->netdev, ntohl(opm->advertise));
+ }
+ }
+ return 0;
+ }
+
+ static struct ofpbuf *
+ make_stats_reply(uint32_t xid, uint16_t type, size_t body_len)
+ {
+ struct ofp_stats_reply *osr;
+ struct ofpbuf *msg;
+
+ msg = ofpbuf_new(MIN(sizeof *osr + body_len, UINT16_MAX));
+ osr = put_openflow_xid(sizeof *osr, OFPT_STATS_REPLY, xid, msg);
+ osr->type = type;
+ osr->flags = htons(0);
+ return msg;
+ }
+
+ static struct ofpbuf *
+ start_stats_reply(const struct ofp_stats_request *request, size_t body_len)
+ {
+ return make_stats_reply(request->header.xid, request->type, body_len);
+ }
+
+ static void *
+ append_stats_reply(size_t nbytes, struct ofconn *ofconn, struct ofpbuf **msgp)
+ {
+ struct ofpbuf *msg = *msgp;
+ assert(nbytes <= UINT16_MAX - sizeof(struct ofp_stats_reply));
+ if (nbytes + msg->size > UINT16_MAX) {
+ struct ofp_stats_reply *reply = msg->data;
+ reply->flags = htons(OFPSF_REPLY_MORE);
+ *msgp = make_stats_reply(reply->header.xid, reply->type, nbytes);
+ queue_tx(msg, ofconn, ofconn->reply_counter);
+ }
+ return ofpbuf_put_uninit(*msgp, nbytes);
+ }
+
+ static int
+ handle_desc_stats_request(struct ofproto *p, struct ofconn *ofconn,
+ struct ofp_stats_request *request)
+ {
+ struct ofp_desc_stats *ods;
+ struct ofpbuf *msg;
+
+ msg = start_stats_reply(request, sizeof *ods);
+ ods = append_stats_reply(sizeof *ods, ofconn, &msg);
+ strncpy(ods->mfr_desc, p->manufacturer, sizeof ods->mfr_desc);
+ strncpy(ods->hw_desc, p->hardware, sizeof ods->hw_desc);
+ strncpy(ods->sw_desc, p->software, sizeof ods->sw_desc);
+ strncpy(ods->serial_num, p->serial, sizeof ods->serial_num);
+ queue_tx(msg, ofconn, ofconn->reply_counter);
+
+ return 0;
+ }
+
+ static void
+ count_subrules(struct cls_rule *cls_rule, void *n_subrules_)
+ {
+ struct rule *rule = rule_from_cls_rule(cls_rule);
+ int *n_subrules = n_subrules_;
+
+ if (rule->super) {
+ (*n_subrules)++;
+ }
+ }
+
+ static int
+ handle_table_stats_request(struct ofproto *p, struct ofconn *ofconn,
+ struct ofp_stats_request *request)
+ {
+ struct ofp_table_stats *ots;
+ struct ofpbuf *msg;
+ struct odp_stats dpstats;
+ int n_exact, n_subrules, n_wild;
+
+ msg = start_stats_reply(request, sizeof *ots * 2);
+
+ /* Count rules of various kinds. */
+ n_subrules = 0;
+ classifier_for_each(&p->cls, CLS_INC_EXACT, count_subrules, &n_subrules);
+ n_exact = classifier_count_exact(&p->cls) - n_subrules;
+ n_wild = classifier_count(&p->cls) - classifier_count_exact(&p->cls);
+
+ /* Hash table. */
+ dpif_get_dp_stats(p->dpif, &dpstats);
+ ots = append_stats_reply(sizeof *ots, ofconn, &msg);
+ memset(ots, 0, sizeof *ots);
+ ots->table_id = TABLEID_HASH;
+ strcpy(ots->name, "hash");
+ ots->wildcards = htonl(0);
+ ots->max_entries = htonl(dpstats.max_capacity);
+ ots->active_count = htonl(n_exact);
+ ots->lookup_count = htonll(dpstats.n_frags + dpstats.n_hit +
+ dpstats.n_missed);
+ ots->matched_count = htonll(dpstats.n_hit); /* XXX */
+
+ /* Classifier table. */
+ ots = append_stats_reply(sizeof *ots, ofconn, &msg);
+ memset(ots, 0, sizeof *ots);
+ ots->table_id = TABLEID_CLASSIFIER;
+ strcpy(ots->name, "classifier");
+ ots->wildcards = htonl(OFPFW_ALL);
+ ots->max_entries = htonl(65536);
+ ots->active_count = htonl(n_wild);
+ ots->lookup_count = htonll(0); /* XXX */
+ ots->matched_count = htonll(0); /* XXX */
+
+ queue_tx(msg, ofconn, ofconn->reply_counter);
+ return 0;
+ }
+
+ static int
+ handle_port_stats_request(struct ofproto *p, struct ofconn *ofconn,
+ struct ofp_stats_request *request)
+ {
+ struct ofp_port_stats *ops;
+ struct ofpbuf *msg;
+ struct ofport *port;
+ unsigned int port_no;
+
+ msg = start_stats_reply(request, sizeof *ops * 16);
+ PORT_ARRAY_FOR_EACH (port, &p->ports, port_no) {
+ struct netdev_stats stats;
+
+ /* Intentionally ignore return value, since errors will set 'stats' to
+ * all-1s, which is correct for OpenFlow, and netdev_get_stats() will
+ * log errors. */
+ netdev_get_stats(port->netdev, &stats);
+
+ ops = append_stats_reply(sizeof *ops, ofconn, &msg);
+ ops->port_no = htons(odp_port_to_ofp_port(port_no));
+ memset(ops->pad, 0, sizeof ops->pad);
+ ops->rx_packets = htonll(stats.rx_packets);
+ ops->tx_packets = htonll(stats.tx_packets);
+ ops->rx_bytes = htonll(stats.rx_bytes);
+ ops->tx_bytes = htonll(stats.tx_bytes);
+ ops->rx_dropped = htonll(stats.rx_dropped);
+ ops->tx_dropped = htonll(stats.tx_dropped);
+ ops->rx_errors = htonll(stats.rx_errors);
+ ops->tx_errors = htonll(stats.tx_errors);
+ ops->rx_frame_err = htonll(stats.rx_frame_errors);
+ ops->rx_over_err = htonll(stats.rx_over_errors);
+ ops->rx_crc_err = htonll(stats.rx_crc_errors);
+ ops->collisions = htonll(stats.collisions);
+ }
+
+ queue_tx(msg, ofconn, ofconn->reply_counter);
+ return 0;
+ }
+
+ struct flow_stats_cbdata {
+ struct ofproto *ofproto;
+ struct ofconn *ofconn;
+ uint16_t out_port;
+ struct ofpbuf *msg;
+ };
+
+ static void
+ query_stats(struct ofproto *p, struct rule *rule,
+ uint64_t *packet_countp, uint64_t *byte_countp)
+ {
+ uint64_t packet_count, byte_count;
+ struct rule *subrule;
+ struct odp_flow *odp_flows;
+ size_t n_odp_flows;
+
+ n_odp_flows = rule->cr.wc.wildcards ? list_size(&rule->list) : 1;
+ odp_flows = xcalloc(1, n_odp_flows * sizeof *odp_flows);
+ if (rule->cr.wc.wildcards) {
+ size_t i = 0;
+ LIST_FOR_EACH (subrule, struct rule, list, &rule->list) {
+ odp_flows[i++].key = subrule->cr.flow;
+ }
+ } else {
+ odp_flows[0].key = rule->cr.flow;
+ }
+
+ packet_count = rule->packet_count;
+ byte_count = rule->byte_count;
+ if (!dpif_flow_get_multiple(p->dpif, odp_flows, n_odp_flows)) {
+ size_t i;
+ for (i = 0; i < n_odp_flows; i++) {
+ struct odp_flow *odp_flow = &odp_flows[i];
+ packet_count += odp_flow->stats.n_packets;
+ byte_count += odp_flow->stats.n_bytes;
+ }
+ }
+ free(odp_flows);
+
+ *packet_countp = packet_count;
+ *byte_countp = byte_count;
+ }
+
+ static void
+ flow_stats_cb(struct cls_rule *rule_, void *cbdata_)
+ {
+ struct rule *rule = rule_from_cls_rule(rule_);
+ struct flow_stats_cbdata *cbdata = cbdata_;
+ struct ofp_flow_stats *ofs;
+ uint64_t packet_count, byte_count;
+ size_t act_len, len;
+
+ if (rule_is_hidden(rule) || !rule_has_out_port(rule, cbdata->out_port)) {
+ return;
+ }
+
+ act_len = sizeof *rule->actions * rule->n_actions;
+ len = offsetof(struct ofp_flow_stats, actions) + act_len;
+
+ query_stats(cbdata->ofproto, rule, &packet_count, &byte_count);
+
+ ofs = append_stats_reply(len, cbdata->ofconn, &cbdata->msg);
+ ofs->length = htons(len);
+ ofs->table_id = rule->cr.wc.wildcards ? TABLEID_CLASSIFIER : TABLEID_HASH;
+ ofs->pad = 0;
+ flow_to_match(&rule->cr.flow, rule->cr.wc.wildcards, &ofs->match);
+ ofs->duration = htonl((time_msec() - rule->created) / 1000);
+ ofs->priority = htons(rule->cr.priority);
+ ofs->idle_timeout = htons(rule->idle_timeout);
+ ofs->hard_timeout = htons(rule->hard_timeout);
+ memset(ofs->pad2, 0, sizeof ofs->pad2);
+ ofs->packet_count = htonll(packet_count);
+ ofs->byte_count = htonll(byte_count);
+ memcpy(ofs->actions, rule->actions, act_len);
+ }
+
+ static int
+ table_id_to_include(uint8_t table_id)
+ {
+ return (table_id == TABLEID_HASH ? CLS_INC_EXACT
+ : table_id == TABLEID_CLASSIFIER ? CLS_INC_WILD
+ : table_id == 0xff ? CLS_INC_ALL
+ : 0);
+ }
+
+ static int
+ handle_flow_stats_request(struct ofproto *p, struct ofconn *ofconn,
+ const struct ofp_stats_request *osr,
+ size_t arg_size)
+ {
+ struct ofp_flow_stats_request *fsr;
+ struct flow_stats_cbdata cbdata;
+ struct cls_rule target;
+
+ if (arg_size != sizeof *fsr) {
+ return ofp_mkerr(OFPET_BAD_REQUEST, OFPBRC_BAD_LENGTH);
+ }
+ fsr = (struct ofp_flow_stats_request *) osr->body;
+
+ COVERAGE_INC(ofproto_flows_req);
+ cbdata.ofproto = p;
+ cbdata.ofconn = ofconn;
+ cbdata.out_port = fsr->out_port;
+ cbdata.msg = start_stats_reply(osr, 1024);
+ cls_rule_from_match(&target, &fsr->match, 0);
+ classifier_for_each_match(&p->cls, &target,
+ table_id_to_include(fsr->table_id),
+ flow_stats_cb, &cbdata);
+ queue_tx(cbdata.msg, ofconn, ofconn->reply_counter);
+ return 0;
+ }
+
++struct flow_stats_ds_cbdata {
++ struct ofproto *ofproto;
++ struct ds *results;
++};
++
++static void
++flow_stats_ds_cb(struct cls_rule *rule_, void *cbdata_)
++{
++ struct rule *rule = rule_from_cls_rule(rule_);
++ struct flow_stats_ds_cbdata *cbdata = cbdata_;
++ struct ds *results = cbdata->results;
++ struct ofp_match match;
++ uint64_t packet_count, byte_count;
++ size_t act_len = sizeof *rule->actions * rule->n_actions;
++
++ /* Don't report on subrules. */
++ if (rule->super != NULL) {
++ return;
++ }
++
++ query_stats(cbdata->ofproto, rule, &packet_count, &byte_count);
++ flow_to_match(&rule->cr.flow, rule->cr.wc.wildcards, &match);
++
++ ds_put_format(results, "duration=%llds, ",
++ (time_msec() - rule->created) / 1000);
++ ds_put_format(results, "priority=%u", rule->cr.priority);
++ ds_put_format(results, "n_packets=%"PRIu64", ", packet_count);
++ ds_put_format(results, "n_bytes=%"PRIu64", ", byte_count);
++ ofp_print_match(results, &match, true);
++ ofp_print_actions(results, &rule->actions->header, act_len);
++ ds_put_cstr(results, "\n");
++}
++
++/* Adds a pretty-printed description of all flows to 'results', including
++ * those marked hidden by secchan (e.g., by in-band control). */
++void
++ofproto_get_all_flows(struct ofproto *p, struct ds *results)
++{
++ struct ofp_match match;
++ struct cls_rule target;
++ struct flow_stats_ds_cbdata cbdata;
++
++ memset(&match, 0, sizeof match);
++ match.wildcards = htonl(OFPFW_ALL);
++
++ cbdata.ofproto = p;
++ cbdata.results = results;
++
++ cls_rule_from_match(&target, &match, 0);
++ classifier_for_each_match(&p->cls, &target, CLS_INC_ALL,
++ flow_stats_ds_cb, &cbdata);
++}
++
+ struct aggregate_stats_cbdata {
+ struct ofproto *ofproto;
+ uint16_t out_port;
+ uint64_t packet_count;
+ uint64_t byte_count;
+ uint32_t n_flows;
+ };
+
+ static void
+ aggregate_stats_cb(struct cls_rule *rule_, void *cbdata_)
+ {
+ struct rule *rule = rule_from_cls_rule(rule_);
+ struct aggregate_stats_cbdata *cbdata = cbdata_;
+ uint64_t packet_count, byte_count;
+
+ if (rule_is_hidden(rule) || !rule_has_out_port(rule, cbdata->out_port)) {
+ return;
+ }
+
+ query_stats(cbdata->ofproto, rule, &packet_count, &byte_count);
+
+ cbdata->packet_count += packet_count;
+ cbdata->byte_count += byte_count;
+ cbdata->n_flows++;
+ }
+
+ static int
+ handle_aggregate_stats_request(struct ofproto *p, struct ofconn *ofconn,
+ const struct ofp_stats_request *osr,
+ size_t arg_size)
+ {
+ struct ofp_aggregate_stats_request *asr;
+ struct ofp_aggregate_stats_reply *reply;
+ struct aggregate_stats_cbdata cbdata;
+ struct cls_rule target;
+ struct ofpbuf *msg;
+
+ if (arg_size != sizeof *asr) {
+ return ofp_mkerr(OFPET_BAD_REQUEST, OFPBRC_BAD_LENGTH);
+ }
+ asr = (struct ofp_aggregate_stats_request *) osr->body;
+
+ COVERAGE_INC(ofproto_agg_request);
+ cbdata.ofproto = p;
+ cbdata.out_port = asr->out_port;
+ cbdata.packet_count = 0;
+ cbdata.byte_count = 0;
+ cbdata.n_flows = 0;
+ cls_rule_from_match(&target, &asr->match, 0);
+ classifier_for_each_match(&p->cls, &target,
+ table_id_to_include(asr->table_id),
+ aggregate_stats_cb, &cbdata);
+
+ msg = start_stats_reply(osr, sizeof *reply);
+ reply = append_stats_reply(sizeof *reply, ofconn, &msg);
+ reply->flow_count = htonl(cbdata.n_flows);
+ reply->packet_count = htonll(cbdata.packet_count);
+ reply->byte_count = htonll(cbdata.byte_count);
+ queue_tx(msg, ofconn, ofconn->reply_counter);
+ return 0;
+ }
+
+ static int
+ handle_stats_request(struct ofproto *p, struct ofconn *ofconn,
+ struct ofp_header *oh)
+ {
+ struct ofp_stats_request *osr;
+ size_t arg_size;
+ int error;
+
+ error = check_ofp_message_array(oh, OFPT_STATS_REQUEST, sizeof *osr,
+ 1, &arg_size);
+ if (error) {
+ return error;
+ }
+ osr = (struct ofp_stats_request *) oh;
+
+ switch (ntohs(osr->type)) {
+ case OFPST_DESC:
+ return handle_desc_stats_request(p, ofconn, osr);
+
+ case OFPST_FLOW:
+ return handle_flow_stats_request(p, ofconn, osr, arg_size);
+
+ case OFPST_AGGREGATE:
+ return handle_aggregate_stats_request(p, ofconn, osr, arg_size);
+
+ case OFPST_TABLE:
+ return handle_table_stats_request(p, ofconn, osr);
+
+ case OFPST_PORT:
+ return handle_port_stats_request(p, ofconn, osr);
+
+ case OFPST_VENDOR:
+ return ofp_mkerr(OFPET_BAD_REQUEST, OFPBRC_BAD_VENDOR);
+
+ default:
+ return ofp_mkerr(OFPET_BAD_REQUEST, OFPBRC_BAD_STAT);
+ }
+ }
+
+ static long long int
+ msec_from_nsec(uint64_t sec, uint32_t nsec)
+ {
+ return !sec ? 0 : sec * 1000 + nsec / 1000000;
+ }
+
+ static void
+ update_time(struct rule *rule, const struct odp_flow_stats *stats)
+ {
+ long long int used = msec_from_nsec(stats->used_sec, stats->used_nsec);
+ if (used > rule->used) {
+ rule->used = used;
+ }
+ }
+
+ static void
+ update_stats(struct rule *rule, const struct odp_flow_stats *stats)
+ {
+ update_time(rule, stats);
+ rule->packet_count += stats->n_packets;
+ rule->byte_count += stats->n_bytes;
+ rule->tcp_flags |= stats->tcp_flags;
+ if (stats->n_packets) {
+ rule->ip_tos = stats->ip_tos;
+ }
+ }
+
+ static int
+ add_flow(struct ofproto *p, struct ofconn *ofconn,
+ struct ofp_flow_mod *ofm, size_t n_actions)
+ {
+ struct ofpbuf *packet;
+ struct rule *rule;
+ uint16_t in_port;
+ int error;
+
+ rule = rule_create(NULL, (const union ofp_action *) ofm->actions,
+ n_actions, ntohs(ofm->idle_timeout),
+ ntohs(ofm->hard_timeout));
+ cls_rule_from_match(&rule->cr, &ofm->match, ntohs(ofm->priority));
+
+ packet = NULL;
+ error = 0;
+ if (ofm->buffer_id != htonl(UINT32_MAX)) {
+ error = pktbuf_retrieve(ofconn->pktbuf, ntohl(ofm->buffer_id),
+ &packet, &in_port);
+ }
+
+ rule_insert(p, rule, packet, in_port);
+ ofpbuf_delete(packet);
+ return error;
+ }
+
+ static int
+ modify_flow(struct ofproto *p, const struct ofp_flow_mod *ofm,
+ size_t n_actions, uint16_t command, struct rule *rule)
+ {
+ if (rule_is_hidden(rule)) {
+ return 0;
+ }
+
+ if (command == OFPFC_DELETE) {
+ rule_remove(p, rule);
+ } else {
+ size_t actions_len = n_actions * sizeof *rule->actions;
+
+ if (n_actions == rule->n_actions
+ && !memcmp(ofm->actions, rule->actions, actions_len))
+ {
+ return 0;
+ }
+
+ free(rule->actions);
+ rule->actions = xmemdup(ofm->actions, actions_len);
+ rule->n_actions = n_actions;
+
+ if (rule->cr.wc.wildcards) {
+ COVERAGE_INC(ofproto_mod_wc_flow);
+ p->need_revalidate = true;
+ } else {
+ rule_update_actions(p, rule);
+ }
+ }
+
+ return 0;
+ }
+
+ static int
+ modify_flows_strict(struct ofproto *p, const struct ofp_flow_mod *ofm,
+ size_t n_actions, uint16_t command)
+ {
+ struct rule *rule;
+ uint32_t wildcards;
+ flow_t flow;
+
+ flow_from_match(&flow, &wildcards, &ofm->match);
+ rule = rule_from_cls_rule(classifier_find_rule_exactly(
+ &p->cls, &flow, wildcards,
+ ntohs(ofm->priority)));
+
+ if (rule) {
+ if (command == OFPFC_DELETE
+ && ofm->out_port != htons(OFPP_NONE)
+ && !rule_has_out_port(rule, ofm->out_port)) {
+ return 0;
+ }
+
+ modify_flow(p, ofm, n_actions, command, rule);
+ }
+ return 0;
+ }
+
+ struct modify_flows_cbdata {
+ struct ofproto *ofproto;
+ const struct ofp_flow_mod *ofm;
+ uint16_t out_port;
+ size_t n_actions;
+ uint16_t command;
+ };
+
+ static void
+ modify_flows_cb(struct cls_rule *rule_, void *cbdata_)
+ {
+ struct rule *rule = rule_from_cls_rule(rule_);
+ struct modify_flows_cbdata *cbdata = cbdata_;
+
+ if (cbdata->out_port != htons(OFPP_NONE)
+ && !rule_has_out_port(rule, cbdata->out_port)) {
+ return;
+ }
+
+ modify_flow(cbdata->ofproto, cbdata->ofm, cbdata->n_actions,
+ cbdata->command, rule);
+ }
+
+ static int
+ modify_flows_loose(struct ofproto *p, const struct ofp_flow_mod *ofm,
+ size_t n_actions, uint16_t command)
+ {
+ struct modify_flows_cbdata cbdata;
+ struct cls_rule target;
+
+ cbdata.ofproto = p;
+ cbdata.ofm = ofm;
+ cbdata.out_port = (command == OFPFC_DELETE ? ofm->out_port
+ : htons(OFPP_NONE));
+ cbdata.n_actions = n_actions;
+ cbdata.command = command;
+
+ cls_rule_from_match(&target, &ofm->match, 0);
+
+ classifier_for_each_match(&p->cls, &target, CLS_INC_ALL,
+ modify_flows_cb, &cbdata);
+ return 0;
+ }
+
+ static int
+ handle_flow_mod(struct ofproto *p, struct ofconn *ofconn,
+ struct ofp_flow_mod *ofm)
+ {
+ size_t n_actions;
+ int error;
+
+ error = check_ofp_message_array(&ofm->header, OFPT_FLOW_MOD, sizeof *ofm,
+ sizeof *ofm->actions, &n_actions);
+ if (error) {
+ return error;
+ }
+
+ normalize_match(&ofm->match);
+ if (!ofm->match.wildcards) {
+ ofm->priority = htons(UINT16_MAX);
+ }
+
+ error = validate_actions((const union ofp_action *) ofm->actions,
+ n_actions, p->max_ports);
+ if (error) {
+ return error;
+ }
+
+ switch (ntohs(ofm->command)) {
+ case OFPFC_ADD:
+ return add_flow(p, ofconn, ofm, n_actions);
+
+ case OFPFC_MODIFY:
+ return modify_flows_loose(p, ofm, n_actions, OFPFC_MODIFY);
+
+ case OFPFC_MODIFY_STRICT:
+ return modify_flows_strict(p, ofm, n_actions, OFPFC_MODIFY);
+
+ case OFPFC_DELETE:
+ return modify_flows_loose(p, ofm, n_actions, OFPFC_DELETE);
+
+ case OFPFC_DELETE_STRICT:
+ return modify_flows_strict(p, ofm, n_actions, OFPFC_DELETE);
+
+ default:
+ return ofp_mkerr(OFPET_FLOW_MOD_FAILED, OFPFMFC_BAD_COMMAND);
+ }
+ }
+
+ static void
+ send_capability_reply(struct ofproto *p, struct ofconn *ofconn, uint32_t xid)
+ {
+ struct ofmp_capability_reply *ocr;
+ struct ofpbuf *b;
+ char capabilities[] = "com.nicira.mgmt.manager=false\n";
+
+ ocr = make_openflow_xid(sizeof(*ocr), OFPT_VENDOR, xid, &b);
+ ocr->header.header.vendor = htonl(NX_VENDOR_ID);
+ ocr->header.header.subtype = htonl(NXT_MGMT);
+ ocr->header.type = htons(OFMPT_CAPABILITY_REPLY);
+
+ ocr->format = htonl(OFMPCOF_SIMPLE);
+ ocr->mgmt_id = htonll(p->mgmt_id);
+
+ ofpbuf_put(b, capabilities, strlen(capabilities));
+
+ queue_tx(b, ofconn, ofconn->reply_counter);
+ }
+
+ static int
+ handle_ofmp(struct ofproto *p, struct ofconn *ofconn,
+ struct ofmp_header *ofmph)
+ {
+ size_t msg_len = ntohs(ofmph->header.header.length);
+ if (msg_len < sizeof(*ofmph)) {
+ VLOG_WARN_RL(&rl, "dropping short managment message: %d\n", msg_len);
+ return ofp_mkerr(OFPET_BAD_REQUEST, OFPBRC_BAD_LENGTH);
+ }
+
+ if (ofmph->type == htons(OFMPT_CAPABILITY_REQUEST)) {
+ struct ofmp_capability_request *ofmpcr;
+
+ if (msg_len < sizeof(struct ofmp_capability_request)) {
+ VLOG_WARN_RL(&rl, "dropping short capability request: %d\n",
+ msg_len);
+ return ofp_mkerr(OFPET_BAD_REQUEST, OFPBRC_BAD_LENGTH);
+ }
+
+ ofmpcr = (struct ofmp_capability_request *)ofmph;
+ if (ofmpcr->format != htonl(OFMPCAF_SIMPLE)) {
+ /* xxx Find a better type than bad subtype */
+ return ofp_mkerr(OFPET_BAD_REQUEST, OFPBRC_BAD_SUBTYPE);
+ }
+
+ send_capability_reply(p, ofconn, ofmph->header.header.xid);
+ return 0;
+ } else {
+ return ofp_mkerr(OFPET_BAD_REQUEST, OFPBRC_BAD_SUBTYPE);
+ }
+ }
+
+ static int
+ handle_vendor(struct ofproto *p, struct ofconn *ofconn, void *msg)
+ {
+ struct ofp_vendor_header *ovh = msg;
+ struct nicira_header *nh;
+
+ if (ntohs(ovh->header.length) < sizeof(struct ofp_vendor_header)) {
+ return ofp_mkerr(OFPET_BAD_REQUEST, OFPBRC_BAD_LENGTH);
+ }
+ if (ovh->vendor != htonl(NX_VENDOR_ID)) {
+ return ofp_mkerr(OFPET_BAD_REQUEST, OFPBRC_BAD_VENDOR);
+ }
+ if (ntohs(ovh->header.length) < sizeof(struct nicira_header)) {
+ return ofp_mkerr(OFPET_BAD_REQUEST, OFPBRC_BAD_LENGTH);
+ }
+
+ nh = msg;
+ switch (ntohl(nh->subtype)) {
+ case NXT_STATUS_REQUEST:
+ return switch_status_handle_request(p->switch_status, ofconn->rconn,
+ msg);
+
+ case NXT_ACT_SET_CONFIG:
+ return ofp_mkerr(OFPET_BAD_REQUEST, OFPBRC_BAD_SUBTYPE); /* XXX */
+
+ case NXT_ACT_GET_CONFIG:
+ return ofp_mkerr(OFPET_BAD_REQUEST, OFPBRC_BAD_SUBTYPE); /* XXX */
+
+ case NXT_COMMAND_REQUEST:
+ if (p->executer) {
+ return executer_handle_request(p->executer, ofconn->rconn, msg);
+ }
+ break;
+
+ case NXT_MGMT:
+ return handle_ofmp(p, ofconn, msg);
+ }
+
+ return ofp_mkerr(OFPET_BAD_REQUEST, OFPBRC_BAD_SUBTYPE);
+ }
+
+ static void
+ handle_openflow(struct ofconn *ofconn, struct ofproto *p,
+ struct ofpbuf *ofp_msg)
+ {
+ struct ofp_header *oh = ofp_msg->data;
+ int error;
+
+ COVERAGE_INC(ofproto_recv_openflow);
+ switch (oh->type) {
+ case OFPT_ECHO_REQUEST:
+ error = handle_echo_request(ofconn, oh);
+ break;
+
+ case OFPT_ECHO_REPLY:
+ error = 0;
+ break;
+
+ case OFPT_FEATURES_REQUEST:
+ error = handle_features_request(p, ofconn, oh);
+ break;
+
+ case OFPT_GET_CONFIG_REQUEST:
+ error = handle_get_config_request(p, ofconn, oh);
+ break;
+
+ case OFPT_SET_CONFIG:
+ error = handle_set_config(p, ofconn, ofp_msg->data);
+ break;
+
+ case OFPT_PACKET_OUT:
+ error = handle_packet_out(p, ofconn, ofp_msg->data);
+ break;
+
+ case OFPT_PORT_MOD:
+ error = handle_port_mod(p, oh);
+ break;
+
+ case OFPT_FLOW_MOD:
+ error = handle_flow_mod(p, ofconn, ofp_msg->data);
+ break;
+
+ case OFPT_STATS_REQUEST:
+ error = handle_stats_request(p, ofconn, oh);
+ break;
+
+ case OFPT_VENDOR:
+ error = handle_vendor(p, ofconn, ofp_msg->data);
+ break;
+
+ default:
+ if (VLOG_IS_WARN_ENABLED()) {
+ char *s = ofp_to_string(oh, ntohs(oh->length), 2);
+ VLOG_DBG_RL(&rl, "OpenFlow message ignored: %s", s);
+ free(s);
+ }
+ error = ofp_mkerr(OFPET_BAD_REQUEST, OFPBRC_BAD_TYPE);
+ break;
+ }
+
+ if (error) {
+ send_error_oh(ofconn, ofp_msg->data, error);
+ }
+ }
+ \f
+ static void
+ handle_odp_msg(struct ofproto *p, struct ofpbuf *packet)
+ {
+ struct odp_msg *msg = packet->data;
+ uint16_t in_port = odp_port_to_ofp_port(msg->port);
+ struct rule *rule;
+ struct ofpbuf payload;
+ flow_t flow;
+
+ /* Handle controller actions. */
+ if (msg->type == _ODPL_ACTION_NR) {
+ COVERAGE_INC(ofproto_ctlr_action);
+ pinsched_send(p->action_sched, in_port, packet,
+ send_packet_in_action, p);
+ return;
+ }
+
+ payload.data = msg + 1;
+ payload.size = msg->length - sizeof *msg;
+ flow_extract(&payload, msg->port, &flow);
+
+ rule = lookup_valid_rule(p, &flow);
+ if (!rule) {
+ /* Don't send a packet-in if OFPPC_NO_PACKET_IN asserted. */
+ struct ofport *port = port_array_get(&p->ports, msg->port);
+ if (port) {
+ if (port->opp.config & OFPPC_NO_PACKET_IN) {
+ COVERAGE_INC(ofproto_no_packet_in);
+ /* XXX install 'drop' flow entry */
+ ofpbuf_delete(packet);
+ return;
+ }
+ } else {
+ VLOG_WARN_RL(&rl, "packet-in on unknown port %"PRIu16, msg->port);
+ }
+
+ COVERAGE_INC(ofproto_packet_in);
+ pinsched_send(p->miss_sched, in_port, packet, send_packet_in_miss, p);
+ return;
+ }
+
+ if (rule->cr.wc.wildcards) {
+ rule = rule_create_subrule(p, rule, &flow);
+ rule_make_actions(p, rule, packet);
+ } else {
+ if (!rule->may_install) {
+ /* The rule is not installable, that is, we need to process every
+ * packet, so process the current packet and set its actions into
+ * 'subrule'. */
+ rule_make_actions(p, rule, packet);
+ } else {
+ /* XXX revalidate rule if it needs it */
+ }
+ }
+
+ rule_execute(p, rule, &payload, &flow);
+ rule_reinstall(p, rule);
+ ofpbuf_delete(packet);
+ }
+ \f
+ static void
+ revalidate_cb(struct cls_rule *sub_, void *cbdata_)
+ {
+ struct rule *sub = rule_from_cls_rule(sub_);
+ struct revalidate_cbdata *cbdata = cbdata_;
+
+ if (cbdata->revalidate_all
+ || (cbdata->revalidate_subrules && sub->super)
+ || (tag_set_intersects(&cbdata->revalidate_set, sub->tags))) {
+ revalidate_rule(cbdata->ofproto, sub);
+ }
+ }
+
+ static bool
+ revalidate_rule(struct ofproto *p, struct rule *rule)
+ {
+ const flow_t *flow = &rule->cr.flow;
+
+ COVERAGE_INC(ofproto_revalidate_rule);
+ if (rule->super) {
+ struct rule *super;
+ super = rule_from_cls_rule(classifier_lookup_wild(&p->cls, flow));
+ if (!super) {
+ rule_remove(p, rule);
+ return false;
+ } else if (super != rule->super) {
+ COVERAGE_INC(ofproto_revalidate_moved);
+ list_remove(&rule->list);
+ list_push_back(&super->list, &rule->list);
+ rule->super = super;
+ rule->hard_timeout = super->hard_timeout;
+ rule->idle_timeout = super->idle_timeout;
+ rule->created = super->created;
+ rule->used = 0;
+ }
+ }
+
+ rule_update_actions(p, rule);
+ return true;
+ }
+
+ static struct ofpbuf *
+ compose_flow_exp(const struct rule *rule, long long int now, uint8_t reason)
+ {
+ struct ofp_flow_expired *ofe;
+ struct ofpbuf *buf;
+
+ ofe = make_openflow(sizeof *ofe, OFPT_FLOW_EXPIRED, &buf);
+ flow_to_match(&rule->cr.flow, rule->cr.wc.wildcards, &ofe->match);
+ ofe->priority = htons(rule->cr.priority);
+ ofe->reason = reason;
+ ofe->duration = (now - rule->created) / 1000;
+ ofe->packet_count = rule->packet_count;
+ ofe->byte_count = rule->byte_count;
+
+ return buf;
+ }
+
+ static void
+ send_flow_exp(struct ofproto *p, struct rule *rule,
+ long long int now, uint8_t reason)
+ {
+ struct ofconn *ofconn;
+ struct ofconn *prev;
+ struct ofpbuf *buf;
+
+ /* We limit the maximum number of queued flow expirations it by accounting
+ * them under the counter for replies. That works because preventing
+ * OpenFlow requests from being processed also prevents new flows from
+ * being added (and expiring). (It also prevents processing OpenFlow
+ * requests that would not add new flows, so it is imperfect.) */
+
+ prev = NULL;
+ LIST_FOR_EACH (ofconn, struct ofconn, node, &p->all_conns) {
+ if (ofconn->send_flow_exp && rconn_is_connected(ofconn->rconn)) {
+ if (prev) {
+ queue_tx(ofpbuf_clone(buf), prev, ofconn->reply_counter);
+ } else {
+ buf = compose_flow_exp(rule, now, reason);
+ }
+ prev = ofconn;
+ }
+ }
+ if (prev) {
+ queue_tx(buf, prev, ofconn->reply_counter);
+ }
+ }
+
+ static void
+ uninstall_idle_flow(struct ofproto *ofproto, struct rule *rule)
+ {
+ assert(rule->installed);
+ assert(!rule->cr.wc.wildcards);
+
+ if (rule->super) {
+ rule_remove(ofproto, rule);
+ } else {
+ rule_uninstall(ofproto, rule);
+ }
+ }
+
+ static void
+ expire_rule(struct cls_rule *cls_rule, void *p_)
+ {
+ struct ofproto *p = p_;
+ struct rule *rule = rule_from_cls_rule(cls_rule);
+ long long int hard_expire, idle_expire, expire, now;
+
+ hard_expire = (rule->hard_timeout
+ ? rule->created + rule->hard_timeout * 1000
+ : LLONG_MAX);
+ idle_expire = (rule->idle_timeout
+ && (rule->super || list_is_empty(&rule->list))
+ ? rule->used + rule->idle_timeout * 1000
+ : LLONG_MAX);
+ expire = MIN(hard_expire, idle_expire);
+ if (expire == LLONG_MAX) {
+ if (rule->installed && time_msec() >= rule->used + 5000) {
+ uninstall_idle_flow(p, rule);
+ }
+ return;
+ }
+
+ now = time_msec();
+ if (now < expire) {
+ if (rule->installed && now >= rule->used + 5000) {
+ uninstall_idle_flow(p, rule);
+ }
+ return;
+ }
+
+ COVERAGE_INC(ofproto_expired);
+ if (rule->cr.wc.wildcards) {
+ /* Update stats. (This code will be a no-op if the rule expired
+ * due to an idle timeout, because in that case the rule has no
+ * subrules left.) */
+ struct rule *subrule, *next;
+ LIST_FOR_EACH_SAFE (subrule, next, struct rule, list, &rule->list) {
+ rule_remove(p, subrule);
+ }
+ }
+
+ send_flow_exp(p, rule, now,
+ (now >= hard_expire
+ ? OFPER_HARD_TIMEOUT : OFPER_IDLE_TIMEOUT));
+ rule_remove(p, rule);
+ }
+
+ static void
+ update_used(struct ofproto *p)
+ {
+ struct odp_flow *flows;
+ size_t n_flows;
+ size_t i;
+ int error;
+
+ error = dpif_flow_list_all(p->dpif, &flows, &n_flows);
+ if (error) {
+ return;
+ }
+
+ for (i = 0; i < n_flows; i++) {
+ struct odp_flow *f = &flows[i];
+ struct rule *rule;
+
+ rule = rule_from_cls_rule(
+ classifier_find_rule_exactly(&p->cls, &f->key, 0, UINT16_MAX));
+ if (!rule || !rule->installed) {
+ COVERAGE_INC(ofproto_unexpected_rule);
+ dpif_flow_del(p->dpif, f);
+ continue;
+ }
+
+ update_time(rule, &f->stats);
+ rule_account(p, rule, f->stats.n_bytes);
+ }
+ free(flows);
+ }
+
+ static void
+ do_send_packet_in(struct ofconn *ofconn, uint32_t buffer_id,
+ const struct ofpbuf *packet, int send_len)
+ {
+ struct ofp_packet_in *opi;
+ struct ofpbuf payload, *buf;
+ struct odp_msg *msg;
+
+ msg = packet->data;
+ payload.data = msg + 1;
+ payload.size = msg->length - sizeof *msg;
+
+ send_len = MIN(send_len, payload.size);
+ buf = ofpbuf_new(sizeof *opi + send_len);
+ opi = put_openflow_xid(offsetof(struct ofp_packet_in, data),
+ OFPT_PACKET_IN, 0, buf);
+ opi->buffer_id = htonl(buffer_id);
+ opi->total_len = htons(payload.size);
+ opi->in_port = htons(odp_port_to_ofp_port(msg->port));
+ opi->reason = msg->type == _ODPL_ACTION_NR ? OFPR_ACTION : OFPR_NO_MATCH;
+ ofpbuf_put(buf, payload.data, MIN(send_len, payload.size));
+ update_openflow_length(buf);
+ rconn_send_with_limit(ofconn->rconn, buf, ofconn->packet_in_counter, 100);
+ }
+
+ static void
+ send_packet_in_action(struct ofpbuf *packet, void *p_)
+ {
+ struct ofproto *p = p_;
+ struct ofconn *ofconn;
+ struct odp_msg *msg;
+
+ msg = packet->data;
+ LIST_FOR_EACH (ofconn, struct ofconn, node, &p->all_conns) {
+ if (ofconn == p->controller || ofconn->miss_send_len) {
+ do_send_packet_in(ofconn, UINT32_MAX, packet, msg->arg);
+ }
+ }
+ ofpbuf_delete(packet);
+ }
+
+ static void
+ send_packet_in_miss(struct ofpbuf *packet, void *p_)
+ {
+ struct ofproto *p = p_;
+ struct ofconn *ofconn;
+ struct ofpbuf payload;
+ struct odp_msg *msg;
+
+ msg = packet->data;
+ payload.data = msg + 1;
+ payload.size = msg->length - sizeof *msg;
+ LIST_FOR_EACH (ofconn, struct ofconn, node, &p->all_conns) {
+ if (ofconn->miss_send_len) {
+ uint32_t buffer_id = pktbuf_save(ofconn->pktbuf, &payload,
+ msg->port);
+ int send_len = (buffer_id != UINT32_MAX ? ofconn->miss_send_len
+ : UINT32_MAX);
+ do_send_packet_in(ofconn, buffer_id, packet, send_len);
+ }
+ }
+ ofpbuf_delete(packet);
+ }
+
+ static uint64_t
+ pick_datapath_id(const struct ofproto *ofproto)
+ {
+ const struct ofport *port;
+
+ port = port_array_get(&ofproto->ports, ODPP_LOCAL);
+ if (port) {
+ uint8_t ea[ETH_ADDR_LEN];
+ int error;
+
+ error = netdev_get_etheraddr(port->netdev, ea);
+ if (!error) {
+ return eth_addr_to_uint64(ea);
+ }
+ VLOG_WARN("could not get MAC address for %s (%s)",
+ netdev_get_name(port->netdev), strerror(error));
+ }
+ return ofproto->fallback_dpid;
+ }
+
+ static uint64_t
+ pick_fallback_dpid(void)
+ {
+ uint8_t ea[ETH_ADDR_LEN];
+ eth_addr_random(ea);
+ ea[0] = 0x00; /* Set Nicira OUI. */
+ ea[1] = 0x23;
+ ea[2] = 0x20;
+ return eth_addr_to_uint64(ea);
+ }
+ \f
+ static bool
+ default_normal_ofhook_cb(const flow_t *flow, const struct ofpbuf *packet,
+ struct odp_actions *actions, tag_type *tags,
+ void *ofproto_)
+ {
+ struct ofproto *ofproto = ofproto_;
+ int out_port;
+
+ /* Drop frames for reserved multicast addresses. */
+ if (eth_addr_is_reserved(flow->dl_dst)) {
+ return true;
+ }
+
+ /* Learn source MAC (but don't try to learn from revalidation). */
+ if (packet != NULL) {
+ tag_type rev_tag = mac_learning_learn(ofproto->ml, flow->dl_src,
+ 0, flow->in_port);
+ if (rev_tag) {
+ /* The log messages here could actually be useful in debugging,
+ * so keep the rate limit relatively high. */
+ static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(30, 300);
+ VLOG_DBG_RL(&rl, "learned that "ETH_ADDR_FMT" is on port %"PRIu16,
+ ETH_ADDR_ARGS(flow->dl_src), flow->in_port);
+ ofproto_revalidate(ofproto, rev_tag);
+ }
+ }
+
+ /* Determine output port. */
+ out_port = mac_learning_lookup_tag(ofproto->ml, flow->dl_dst, 0, tags);
+ if (out_port < 0) {
+ add_output_group_action(actions, DP_GROUP_FLOOD);
+ } else if (out_port != flow->in_port) {
+ odp_actions_add(actions, ODPAT_OUTPUT)->output.port = out_port;
+ } else {
+ /* Drop. */
+ }
+
+ return true;
+ }
+
+ static const struct ofhooks default_ofhooks = {
+ NULL,
+ default_normal_ofhook_cb,
+ NULL,
+ NULL
+ };
--- /dev/null
+ /*
+ * Copyright (c) 2009 Nicira Networks.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+ #ifndef OFPROTO_H
+ #define OFPROTO_H 1
+
+ #include <stdbool.h>
+ #include <stddef.h>
+ #include <stdint.h>
+ #include "flow.h"
+ #include "tag.h"
+
+ struct odp_actions;
+ struct ofhooks;
+ struct ofproto;
+ struct svec;
+
+ struct ofexpired {
+ flow_t flow;
+ uint64_t packet_count; /* Packets from *expired* subrules. */
+ uint64_t byte_count; /* Bytes from *expired* subrules. */
+ long long int used; /* Last-used time (0 if never used). */
+ long long int created; /* Creation time. */
+ uint8_t tcp_flags; /* Bitwise-OR of all TCP flags seen. */
+ uint8_t ip_tos; /* Last-seen IP type-of-service. */
+ };
+
+ int ofproto_create(const char *datapath, const struct ofhooks *, void *aux,
+ struct ofproto **ofprotop);
+ void ofproto_destroy(struct ofproto *);
+ int ofproto_run(struct ofproto *);
+ int ofproto_run1(struct ofproto *);
+ int ofproto_run2(struct ofproto *, bool revalidate_all);
+ void ofproto_wait(struct ofproto *);
+ bool ofproto_is_alive(const struct ofproto *);
+
+ /* Configuration. */
+ void ofproto_set_datapath_id(struct ofproto *, uint64_t datapath_id);
+ void ofproto_set_mgmt_id(struct ofproto *, uint64_t mgmt_id);
+ void ofproto_set_probe_interval(struct ofproto *, int probe_interval);
+ void ofproto_set_max_backoff(struct ofproto *, int max_backoff);
+ void ofproto_set_desc(struct ofproto *,
+ const char *manufacturer, const char *hardware,
+ const char *software, const char *serial);
+ int ofproto_set_in_band(struct ofproto *, bool in_band);
+ int ofproto_set_discovery(struct ofproto *, bool discovery,
+ const char *accept_controller_re,
+ bool update_resolv_conf);
+ int ofproto_set_controller(struct ofproto *, const char *controller);
+ int ofproto_set_listeners(struct ofproto *, const struct svec *listeners);
+ int ofproto_set_snoops(struct ofproto *, const struct svec *snoops);
+ int ofproto_set_netflow(struct ofproto *, const struct svec *collectors,
+ uint8_t engine_type, uint8_t engine_id, bool add_id_to_iface);
+ void ofproto_set_failure(struct ofproto *, bool fail_open);
+ void ofproto_set_rate_limit(struct ofproto *, int rate_limit, int burst_limit);
+ int ofproto_set_stp(struct ofproto *, bool enable_stp);
+ int ofproto_set_remote_execution(struct ofproto *, const char *command_acl,
+ const char *command_dir);
+
+ /* Configuration querying. */
+ uint64_t ofproto_get_datapath_id(const struct ofproto *);
+ uint64_t ofproto_get_mgmt_id(const struct ofproto *);
+ int ofproto_get_probe_interval(const struct ofproto *);
+ int ofproto_get_max_backoff(const struct ofproto *);
+ bool ofproto_get_in_band(const struct ofproto *);
+ bool ofproto_get_discovery(const struct ofproto *);
+ const char *ofproto_get_controller(const struct ofproto *);
+ void ofproto_get_listeners(const struct ofproto *, struct svec *);
+ void ofproto_get_snoops(const struct ofproto *, struct svec *);
++void ofproto_get_all_flows(struct ofproto *p, struct ds *);
+
+ /* Functions for use by ofproto implementation modules, not by clients. */
+ int ofproto_send_packet(struct ofproto *, const flow_t *,
+ const union ofp_action *, size_t n_actions,
+ const struct ofpbuf *);
+ void ofproto_add_flow(struct ofproto *, const flow_t *, uint32_t wildcards,
+ unsigned int priority,
+ const union ofp_action *, size_t n_actions,
+ int idle_timeout);
+ void ofproto_delete_flow(struct ofproto *, const flow_t *, uint32_t wildcards,
+ unsigned int priority);
+ void ofproto_flush_flows(struct ofproto *);
+
+ /* Hooks for ovs-vswitchd. */
+ struct ofhooks {
+ void (*port_changed_cb)(enum ofp_port_reason, const struct ofp_phy_port *,
+ void *aux);
+ bool (*normal_cb)(const flow_t *, const struct ofpbuf *packet,
+ struct odp_actions *, tag_type *, void *aux);
+ void (*account_flow_cb)(const flow_t *, const union odp_action *,
+ size_t n_actions, unsigned long long int n_bytes,
+ void *aux);
+ void (*account_checkpoint_cb)(void *aux);
+ };
+ void ofproto_revalidate(struct ofproto *, tag_type);
+ struct tag_set *ofproto_get_revalidate_set(struct ofproto *);
+
+ #endif /* ofproto.h */
}
}
- struct dpif dpif;
- char dpif_name[IF_NAMESIZE];
-
- if (dpif_open(all_dps.names[i], &dpif)) {
- continue;
- }
- if (!dpif_get_name(&dpif, dpif_name, sizeof dpif_name)) {
- printf("%s\n", dpif_name);
+static void
+do_dump_dps(int argc UNUSED, char *argv[] UNUSED)
+{
+ struct svec all_dps;
+ unsigned int i;
+ int error;
+
+ svec_init(&all_dps);
+ error = dp_enumerate(&all_dps);
+
+ for (i = 0; i < all_dps.n; i++) {
- dpif_close(&dpif);
++ struct dpif *dpif;
++ if (!dpif_open(all_dps.names[i], &dpif)) {
++ printf("%s\n", dpif_name(dpif));
++ dpif_close(dpif);
+ }
+ }
+
+ svec_destroy(&all_dps);
+ if (error) {
+ exit(EXIT_FAILURE);
+ }
+}
+
static void
do_dump_flows(int argc UNUSED, char *argv[])
{
--- /dev/null
-The default is 15 seconds, and the minimum value is 5 seconds.
+ .TH ovs\-openflowd 8 "March 2009" "Open vSwitch" "Open vSwitch Manual"
+ .ds PN ovs\-openflowd
+
+ .SH NAME
+ ovs\-openflowd \- OpenFlow switch implementation
+
+ .SH SYNOPSIS
+ .B ovs\-openflowd
+ [\fIoptions\fR] \fIdatapath\fR [\fIcontroller\fR]
+
+ .SH DESCRIPTION
+ The \fBovs\-openflowd\fR program implements an OpenFlow switch using a
+ flow-based datapath. \fBovs\-openflowd\fR connects to an OpenFlow controller
+ over TCP or SSL.
+
+ The mandatory \fIdatapath\fR argument argument specifies the local datapath
+ to relay. It takes one of the following forms:
+
+ .so lib/dpif.man
+
+ .PP
+ The optional \fIcontroller\fR argument specifies how to connect to
+ the OpenFlow controller. It takes one of the following forms:
+
+ .RS
+ .IP "\fBssl:\fIip\fR[\fB:\fIport\fR]"
+ The specified SSL \fIport\fR (default: 6633) on the host at the given
+ \fIip\fR, which must be expressed as an IP address (not a DNS name).
+ The \fB--private-key\fR, \fB--certificate\fR, and \fB--ca-cert\fR
+ options are mandatory when this form is used.
+
+ .IP "\fBtcp:\fIip\fR[\fB:\fIport\fR]"
+ The specified TCP \fIport\fR (default: 6633) on the host at the given
+ \fIip\fR, which must be expressed as an IP address (not a DNS name).
+
+ .TP
+ \fBunix:\fIfile\fR
+ The Unix domain server socket named \fIfile\fR.
+ .RE
+
+ .PP
+ If \fIcontroller\fR is omitted, \fBovs\-openflowd\fR attempts to discover the
+ location of the controller automatically (see below).
+
+ .SS "Contacting the Controller"
+ The OpenFlow switch must be able to contact the OpenFlow controller
+ over the network. It can do so in one of two ways:
+
+ .IP out-of-band
+ In this configuration, OpenFlow traffic uses a network separate from
+ the data traffic that it controls, that is, the switch does not use
+ any of the network devices added to the datapath with \fBovs\-dpctl
+ add\-if\fR in its communication with the controller.
+
+ To use \fBovs\-openflowd\fR in a network with out-of-band control, specify
+ \fB--out-of-band\fR on the \fBovs\-openflowd\fR command line. The control
+ network must be configured separately, before or after \fBovs\-openflowd\fR
+ is started.
+
+ .IP in-band
+ In this configuration, a single network is used for OpenFlow traffic
+ and other data traffic, that is, the switch contacts the controller
+ over one of the network devices added to the datapath with \fBovs\-dpctl
+ add\-if\fR. This configuration is often more convenient than
+ out-of-band control, because it is not necessary to maintain two
+ independent networks.
+
+ In-band control is the default for \fBovs\-openflowd\fR, so no special
+ command-line option is required.
+
+ With in-band control, the location of the controller can be configured
+ manually or discovered automatically:
+
+ .RS
+ .IP "controller discovery"
+ To make \fBovs\-openflowd\fR discover the location of the controller
+ automatically, do not specify the location of the controller on the
+ \fBovs\-openflowd\fR command line.
+
+ In this mode, \fBovs\-openflowd\fR will broadcast a DHCP request with vendor
+ class identifier \fBOpenFlow\fR across the network devices added to
+ the datapath with \fBovs\-dpctl add\-if\fR. It will accept any valid DHCP
+ reply that has the same vendor class identifier and includes a
+ vendor-specific option with code 1 whose contents are a string
+ specifying the location of the controller in the same format used on
+ the \fBovs\-openflowd\fR command line (e.g. \fBssl:192.168.0.1\fR).
+
+ The DHCP reply may also, optionally, include a vendor-specific option
+ with code 2 whose contents are a string specifying the URI to the base
+ of the OpenFlow PKI (e.g. \fBhttp://192.168.0.1/openflow/pki\fR).
+ This URI is used only for bootstrapping the OpenFlow PKI at initial
+ switch setup; \fBovs\-openflowd\fR does not use it at all.
+
+ The following ISC DHCP server configuration file assigns the IP
+ address range 192.168.0.20 through 192.168.0.30 to OpenFlow switches
+ that follow the switch protocol and addresses 192.168.0.1 through
+ 192.168.0.10 to all other DHCP clients:
+
+ default-lease-time 600;
+ .br
+ max-lease-time 7200;
+ .br
+ option space openflow;
+ .br
+ option openflow.controller-vconn code 1 = text;
+ .br
+ option openflow.pki-uri code 2 = text;
+ .br
+ class "OpenFlow" {
+ .br
+ match if option vendor-class-identifier = "OpenFlow";
+ .br
+ vendor-option-space openflow;
+ .br
+ option openflow.controller-vconn "tcp:192.168.0.10";
+ .br
+ option openflow.pki-uri "http://192.168.0.10/openflow/pki";
+ .br
+ option vendor-class-identifier "OpenFlow";
+ .br
+ }
+ .br
+ subnet 192.168.0.0 netmask 255.255.255.0 {
+ .br
+ pool {
+ .br
+ allow members of "OpenFlow";
+ .br
+ range 192.168.0.20 192.168.0.30;
+ .br
+ }
+ .br
+ pool {
+ .br
+ deny members of "OpenFlow";
+ .br
+ range 192.168.0.1 192.168.0.10;
+ .br
+ }
+ .br
+ }
+ .br
+
+ .IP "manual configuration"
+ To configure in-band control manually, specify the location of the
+ controller on the \fBovs\-openflowd\fR command line as the \fIcontroller\fR
+ argument. You must also configure the network device for the OpenFlow
+ ``local port'' to allow \fBovs\-openflowd\fR to connect to that controller.
+ The OpenFlow local port is a virtual network port that \fBovs\-openflowd\fR
+ bridges to the physical switch ports. The name of the local port for
+ a given \fIdatapath\fR may be seen by running \fBovs\-dpctl show
+ \fIdatapath\fR; the local port is listed as port 0 in \fBshow\fR's
+ output.
+
+ .IP
+ Before \fBovs\-openflowd\fR starts, the local port network device is not
+ bridged to any physical network, so the next step depends on whether
+ connectivity is required to configure the device's IP address. If the
+ switch has a static IP address, you may configure its IP address now
+ with a command such as
+ .B ifconfig of0 192.168.1.1
+ and then invoke \fBovs\-openflowd\fR.
+
+ On the other hand, if the switch does not have a static IP address,
+ e.g. it obtains its IP address dynamically via DHCP, the DHCP client
+ will not be able to contact the DHCP server until the OpenFlow switch
+ has started up. Thus, start \fBovs\-openflowd\fR without configuring
+ the local port network device, and start the DHCP client afterward.
+ .RE
+
+ .SH OPTIONS
+ .SS "Controller Discovery Options"
+ .TP
+ \fB--accept-vconn=\fIregex\fR
+ When \fBovs\-openflowd\fR performs controller discovery (see \fBContacting
+ the Controller\fR, above, for more information about controller
+ discovery), it validates the controller location obtained via DHCP
+ with a POSIX extended regular expression. Only controllers whose
+ names match the regular expression will be accepted.
+
+ The default regular expression is \fBssl:.*\fR (meaning that only SSL
+ controller connections will be accepted) when any of the SSL
+ configuration options \fB--private-key\fR, \fB--certificate\fR, or
+ \fB--ca-cert\fR is specified. The default is \fB^tcp:.*\fR otherwise
+ (meaning that only TCP controller connections will be accepted).
+
+ The \fIregex\fR is implicitly anchored at the beginning of the
+ controller location string, as if it begins with \fB^\fR.
+
+ When controller discovery is not performed, this option has no effect.
+
+ .TP
+ \fB--no-resolv-conf\fR
+ When \fBovs\-openflowd\fR performs controller discovery (see \fBContacting
+ the Controller\fR, above, for more information about controller
+ discovery), by default it overwrites the system's
+ \fB/etc/resolv.conf\fR with domain information and DNS servers
+ obtained via DHCP. If the location of the controller is specified
+ using a hostname, rather than an IP address, and the network's DNS
+ servers ever change, this behavior is essential. But because it also
+ interferes with any administrator or process that manages
+ \fB/etc/resolv.conf\fR, when this option is specified, \fBovs\-openflowd\fR
+ will not modify \fB/etc/resolv.conf\fR.
+
+ \fBovs\-openflowd\fR will only modify \fBresolv.conf\fR if the DHCP response
+ that it receives specifies one or more DNS servers.
+
+ When controller discovery is not performed, this option has no effect.
+
+ .SS "Networking Options"
+ .TP
+ \fB--datapath-id=\fIdpid\fR
+ Sets \fIdpid\fR, which must consist of exactly 12 hexadecimal digits,
+ as the datapath ID that the switch will use to identify itself to the
+ OpenFlow controller.
+
+ If this option is omitted, the default datapath ID is taken from the
+ Ethernet address of the datapath's local port (which is typically
+ randomly generated).
+
+ .TP
+ \fB--mgmt-id=\fImgmtid\fR
+ Sets \fImgmtid\fR, which must consist of exactly 12 hexadecimal
+ digits, as the switch's management ID.
+
+ If this option is omitted, the management ID defaults to 0, signaling
+ to the controller that management is supported but not configured.
+
+ .TP
+ \fB--fail=\fR[\fBopen\fR|\fBclosed\fR]
+ The controller is, ordinarily, responsible for setting up all flows on
+ the OpenFlow switch. Thus, if the connection to the controller fails,
+ no new network connections can be set up. If the connection to the
+ controller stays down long enough, no packets can pass through the
+ switch at all.
+
+ If this option is set to \fBopen\fR (the default), \fBovs\-openflowd\fR will
+ take over responsibility for setting up flows in the local datapath
+ when no message has been received from the controller for three times
+ the inactivity probe interval (see below), or 45 seconds by default.
+ In this ``fail open'' mode, \fBovs\-openflowd\fR causes the datapath to act
+ like an ordinary MAC-learning switch. \fBovs\-openflowd\fR will continue to
+ retry connection to the controller in the background and, when the
+ connection succeeds, it discontinues its fail-open behavior.
+
+ If this option is set to \fBclosed\fR, then \fBovs\-openflowd\fR will not
+ set up flows on its own when the controller connection fails.
+
+ .TP
+ \fB--inactivity-probe=\fIsecs\fR
+ When the OpenFlow switch is connected to the controller, the
+ switch waits for a message to be received from the controller for
+ \fIsecs\fR seconds before it sends a inactivity probe to the
+ controller. After sending the inactivity probe, if no response is
+ received for an additional \fIsecs\fR seconds, the switch
+ assumes that the connection has been broken and attempts to reconnect.
-time is 15 seconds.
++The default and the minimum value are both 5 seconds.
+
+ When fail-open mode is configured, changing the inactivity probe
+ interval also changes the interval before entering fail-open mode (see
+ above).
+
+ .TP
+ \fB--max-idle=\fIsecs\fR|\fBpermanent\fR
+ Sets \fIsecs\fR as the number of seconds that a flow set up by the
+ OpenFlow switch will remain in the switch's flow table without any
+ matching packets being seen. If \fBpermanent\fR is specified, which
+ is not recommended, flows set up by the switch will never
+ expire. The default is 15 seconds.
+
+ Most flows are set up by the OpenFlow controller, not by the
+ switch. This option affects only the following flows, which the
+ OpenFlow switch sets up itself:
+
+ .RS
+ .IP \(bu
+ When \fB--fail=open\fR is specified, flows set up when the
+ switch has not been able to contact the controller for the configured
+ fail-open delay.
+
+ .IP \(bu
+ When in-band control is in use, flows set up to bootstrap contacting
+ the controller (see \fBContacting the Controller\fR, above, for
+ more information about in-band control).
+ .RE
+
+ .IP
+ As a result, when both \fB--fail=closed\fR and \fB--out-of-band\fR are
+ specified, this option has no effect.
+
+ .TP
+ \fB--max-backoff=\fIsecs\fR
+ Sets the maximum time between attempts to connect to the controller to
+ \fIsecs\fR, which must be at least 1. The actual interval between
+ connection attempts starts at 1 second and doubles on each failing
+ attempt until it reaches the maximum. The default maximum backoff
++time is 8 seconds.
+
+ .TP
+ \fB-l\fR, \fB--listen=\fImethod\fR
+ Configures the switch to additionally listen for incoming OpenFlow
+ connections for switch management with \fBovs\-ofctl\fR. The \fImethod\fR
+ must be given as one of the passive OpenFlow connection methods listed
+ below. This option may be specified multiple times to listen to
+ multiple connection methods.
+
+ .RS
+ .TP
+ \fBpssl:\fR[\fIport\fR][\fB:\fIip\fR]
+ Listens for SSL connections on \fIport\fR (default: 6633). The
+ \fB--private-key\fR, \fB--certificate\fR, and \fB--ca-cert\fR options
+ are mandatory when this form is used.
+ By default, \fB\*(PN\fR listens for connections to any local IP
+ address, but \fIip\fR may be specified to listen only for connections
+ to the given \fIip\fR.
+
+ .TP
+ \fBptcp:\fR[\fIport\fR][\fB:\fIip\fR]
+ Listens for TCP connections on \fIport\fR (default: 6633).
+ By default, \fB\*(PN\fR listens for connections to any local IP
+ address, but \fIip\fR may be specified to listen only for connections
+ to the given \fIip\fR.
+
+ .TP
+ \fBpunix:\fIfile\fR
+ Listens for connections on Unix domain server socket named \fIfile\fR.
+ .RE
+
+ .TP
+ \fB--snoop=\fImethod\fR
+ Configures the switch to additionally listen for incoming OpenFlow
+ connections for controller connection snooping. The \fImethod\fR must
+ be given as one of the passive OpenFlow connection methods listed
+ under the \fB--listen\fR option above. This option may be specified
+ multiple times to listen to multiple connection methods.
+
+ If \fBovs\-ofctl monitor\fR is used to connect to \fImethod\fR specified on
+ \fB--snoop\fR, it will display all the OpenFlow messages traveling
+ between the switch and its controller on the primary OpenFlow
+ connection. This can be useful for debugging switch and controller
+ problems.
+
+ .TP
+ \fB--in-band\fR, \fB--out-of-band\fR
+ Configures \fBovs\-openflowd\fR to operate in in-band or out-of-band control
+ mode (see \fBContacting the Controller\fR above). When neither option
+ is given, the default is in-band control.
+
+ .TP
+ \fB--netflow=\fIip\fB:\fIport\fR
+ Configures the given UDP \fIport\fR on the specified IP \fIip\fR as
+ a recipient of NetFlow messages for expired flows. The \fIip\fR must
+ be specified numerically, not as a DNS name.
+
+ This option may be specified multiple times to configure additional
+ NetFlow collectors.
+
+ .SS "Rate-Limiting Options"
+
+ These options configure how the switch applies a ``token bucket'' to
+ limit the rate at which packets in unknown flows are forwarded to an
+ OpenFlow controller for flow-setup processing. This feature prevents
+ a single OpenFlow switch from overwhelming a controller.
+
+ .TP
+ \fB--rate-limit\fR[\fB=\fIrate\fR]
+ .
+ Limits the maximum rate at which packets will be forwarded to the
+ OpenFlow controller to \fIrate\fR packets per second. If \fIrate\fR
+ is not specified then the default of 1,000 packets per second is used.
+
+ If \fB--rate-limit\fR is not used, then the switch does not limit the
+ rate at which packets are forwarded to the controller.
+
+ .TP
+ \fB--burst-limit=\fIburst\fR
+ .
+ Sets the maximum number of unused packet credits that the switch will
+ allow to accumulate during time in which no packets are being
+ forwarded to the OpenFlow controller to \fIburst\fR (measured in
+ packets). The default \fIburst\fR is one-quarter of the \fIrate\fR
+ specified on \fB--rate-limit\fR.
+
+ This option takes effect only when \fB--rate-limit\fR is also specified.
+
+ .SS "Remote Command Execution Options"
+
+ .TP
+ \fB--command-acl=\fR[\fB!\fR]\fIglob\fR[\fB,\fR[\fB!\fR]\fIglob\fR...]
+ Configures the commands that remote OpenFlow connections are allowed
+ to invoke using (e.g.) \fBovs\-ofctl execute\fR. The argument is a
+ comma-separated sequence of shell glob patterns. A glob pattern
+ specified without a leading \fB!\fR is a ``whitelist'' that specifies
+ a set of commands that are that may be invoked, whereas a pattern that
+ does begin with \fB!\fR is a ``blacklist'' that specifies commands
+ that may not be invoked. To be permitted, a command name must be
+ whitelisted and must not be blacklisted;
+ e.g. \fB--command-acl=up*,!upgrade\fR would allow any command whose name
+ begins with \fBup\fR except for the command named \fBupgrade\fR.
+ Command names that include characters other than upper- and lower-case
+ English letters, digits, and the underscore and hyphen characters are
+ unconditionally disallowed.
+
+ When the whitelist and blacklist permit a command name, \fBovs\-openflowd\fR
+ looks for a program with the same name as the command in the commands
+ directory (see below). Other directories are not searched.
+
+ .TP
+ \fB--command-dir=\fIdirectory\fR
+ Sets the directory searched for remote command execution to
+ \fBdirectory\fR. The default directory is
+ \fB@pkgdatadir@/commands\fR.
+
+ .SS "Daemon Options"
+ .so lib/daemon.man
+
+ .SS "Public Key Infrastructure Options"
+
+ .TP
+ \fB-p\fR, \fB--private-key=\fIprivkey.pem\fR
+ Specifies a PEM file containing the private key used as the switch's
+ identity for SSL connections to the controller.
+
+ .TP
+ \fB-c\fR, \fB--certificate=\fIcert.pem\fR
+ Specifies a PEM file containing a certificate, signed by the
+ controller's certificate authority (CA), that certifies the switch's
+ private key to identify a trustworthy switch.
+
+ .TP
+ \fB-C\fR, \fB--ca-cert=\fIcacert.pem\fR
+ Specifies a PEM file containing the CA certificate used to verify that
+ the switch is connected to a trustworthy controller.
+
+ .TP
+ \fB--bootstrap-ca-cert=\fIcacert.pem\fR
+ When \fIcacert.pem\fR exists, this option has the same effect as
+ \fB-C\fR or \fB--ca-cert\fR. If it does not exist, then \fBovs\-openflowd\fR
+ will attempt to obtain the CA certificate from the controller on its
+ first SSL connection and save it to the named PEM file. If it is
+ successful, it will immediately drop the connection and reconnect, and
+ from then on all SSL connections must be authenticated by a
+ certificate signed by the CA certificate thus obtained.
+
+ \fBThis option exposes the SSL connection to a man-in-the-middle
+ attack obtaining the initial CA certificate\fR, but it may be useful
+ for bootstrapping.
+
+ This option is only useful if the controller sends its CA certificate
+ as part of the SSL certificate chain. The SSL protocol does not
+ require the controller to send the CA certificate, but
+ \fBcontroller\fR(8) can be configured to do so with the
+ \fB--peer-ca-cert\fR option.
+
+ .SS "Logging Options"
+ .so lib/vlog.man
+ .SS "Other Options"
+ .so lib/common.man
+ .so lib/leak-checker.man
+
+ .SH "SEE ALSO"
+
+ .BR ovs\-appctl (8),
+ .BR ovs\-controller (8),
+ .BR ovs\-discover (8),
+ .BR ovs\-dpctl (8),
+ .BR ovs\-ofctl (8),
+ .BR ovs\-pki (8),
+ .BR ovs\-vswitchd.conf (5)
--- /dev/null
- s->max_backoff = 15;
+ /*
+ * Copyright (c) 2008, 2009 Nicira Networks.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+ #include <config.h>
+ #include <assert.h>
+ #include <errno.h>
+ #include <getopt.h>
+ #include <inttypes.h>
+ #include <netinet/in.h>
+ #include <stdlib.h>
+ #include <signal.h>
+ #include <string.h>
+
+ #include "command-line.h"
+ #include "compiler.h"
+ #include "daemon.h"
+ #include "dirs.h"
+ #include "dpif.h"
+ #include "fault.h"
+ #include "leak-checker.h"
+ #include "list.h"
+ #include "netdev.h"
+ #include "ofpbuf.h"
+ #include "ofproto/ofproto.h"
+ #include "openflow/openflow.h"
+ #include "packets.h"
+ #include "poll-loop.h"
+ #include "rconn.h"
+ #include "svec.h"
+ #include "timeval.h"
+ #include "unixctl.h"
+ #include "util.h"
+ #include "vconn-ssl.h"
+ #include "vconn.h"
+
+ #include "vlog.h"
+ #define THIS_MODULE VLM_openflowd
+
+ /* Behavior when the connection to the controller fails. */
+ enum fail_mode {
+ FAIL_OPEN, /* Act as learning switch. */
+ FAIL_CLOSED /* Drop all packets. */
+ };
+
+ /* Settings that may be configured by the user. */
+ struct ofsettings {
+ /* Overall mode of operation. */
+ bool discovery; /* Discover the controller automatically? */
+ bool in_band; /* Connect to controller in-band? */
+
+ /* Datapath. */
+ uint64_t datapath_id; /* Datapath ID. */
+ const char *dp_name; /* Name of local datapath. */
+
+ /* Description strings. */
+ const char *mfr_desc; /* Manufacturer. */
+ const char *hw_desc; /* Hardware. */
+ const char *sw_desc; /* Software version. */
+ const char *serial_desc; /* Serial number. */
+
+ /* Related vconns and network devices. */
+ const char *controller_name; /* Controller (if not discovery mode). */
+ struct svec listeners; /* Listen for management connections. */
+ struct svec snoops; /* Listen for controller snooping conns. */
+
+ /* Failure behavior. */
+ enum fail_mode fail_mode; /* Act as learning switch if no controller? */
+ int max_idle; /* Idle time for flows in fail-open mode. */
+ int probe_interval; /* # seconds idle before sending echo request. */
+ int max_backoff; /* Max # seconds between connection attempts. */
+
+ /* Packet-in rate-limiting. */
+ int rate_limit; /* Tokens added to bucket per second. */
+ int burst_limit; /* Maximum number token bucket size. */
+
+ /* Discovery behavior. */
+ const char *accept_controller_re; /* Controller vconns to accept. */
+ bool update_resolv_conf; /* Update /etc/resolv.conf? */
+
+ /* Spanning tree protocol. */
+ bool enable_stp;
+
+ /* Remote command execution. */
+ char *command_acl; /* Command white/blacklist, as shell globs. */
+ char *command_dir; /* Directory that contains commands. */
+
+ /* Management. */
+ uint64_t mgmt_id; /* Management ID. */
+
+ /* NetFlow. */
+ struct svec netflow; /* NetFlow targets. */
+ };
+
+ static void parse_options(int argc, char *argv[], struct ofsettings *);
+ static void usage(void) NO_RETURN;
+
+ int
+ main(int argc, char *argv[])
+ {
+ struct unixctl_server *unixctl;
+ struct ofproto *ofproto;
+ struct ofsettings s;
+ int error;
+
+ set_program_name(argv[0]);
+ register_fault_handlers();
+ time_init();
+ vlog_init();
+ parse_options(argc, argv, &s);
+ signal(SIGPIPE, SIG_IGN);
+
+ die_if_already_running();
+ daemonize();
+
+ /* Start listening for ovs-appctl requests. */
+ error = unixctl_server_create(NULL, &unixctl);
+ if (error) {
+ ovs_fatal(error, "Could not listen for unixctl connections");
+ }
+
+ VLOG_INFO("Open vSwitch version %s", VERSION BUILDNR);
+ VLOG_INFO("OpenFlow protocol version 0x%02x", OFP_VERSION);
+
+ /* Start OpenFlow processing. */
+ error = ofproto_create(s.dp_name, NULL, NULL, &ofproto);
+ if (error) {
+ ovs_fatal(error, "could not initialize openflow switch");
+ }
+ error = ofproto_set_in_band(ofproto, s.in_band);
+ if (error) {
+ ovs_fatal(error, "failed to configure in-band control");
+ }
+ error = ofproto_set_discovery(ofproto, s.discovery, s.accept_controller_re,
+ s.update_resolv_conf);
+ if (error) {
+ ovs_fatal(error, "failed to configure controller discovery");
+ }
+ if (s.datapath_id) {
+ ofproto_set_datapath_id(ofproto, s.datapath_id);
+ }
+ if (s.mgmt_id) {
+ ofproto_set_mgmt_id(ofproto, s.mgmt_id);
+ }
+ ofproto_set_desc(ofproto, s.mfr_desc, s.hw_desc, s.sw_desc, s.serial_desc);
+ error = ofproto_set_listeners(ofproto, &s.listeners);
+ if (error) {
+ ovs_fatal(error, "failed to configure management connections");
+ }
+ error = ofproto_set_snoops(ofproto, &s.snoops);
+ if (error) {
+ ovs_fatal(error,
+ "failed to configure controller snooping connections");
+ }
+ error = ofproto_set_netflow(ofproto, &s.netflow, 0, 0, false);
+ if (error) {
+ ovs_fatal(error, "failed to configure NetFlow collectors");
+ }
+ ofproto_set_failure(ofproto, s.fail_mode == FAIL_OPEN);
+ ofproto_set_probe_interval(ofproto, s.probe_interval);
+ ofproto_set_max_backoff(ofproto, s.max_backoff);
+ ofproto_set_rate_limit(ofproto, s.rate_limit, s.burst_limit);
+ error = ofproto_set_stp(ofproto, s.enable_stp);
+ if (error) {
+ ovs_fatal(error, "failed to configure STP");
+ }
+ error = ofproto_set_remote_execution(ofproto, s.command_acl,
+ s.command_dir);
+ if (error) {
+ ovs_fatal(error, "failed to configure remote command execution");
+ }
+ if (!s.discovery) {
+ error = ofproto_set_controller(ofproto, s.controller_name);
+ if (error) {
+ ovs_fatal(error, "failed to configure controller");
+ }
+ }
+
+ while (ofproto_is_alive(ofproto)) {
+ error = ofproto_run(ofproto);
+ if (error) {
+ ovs_fatal(error, "unrecoverable datapath error");
+ }
+ unixctl_server_run(unixctl);
+ dp_run();
+ netdev_run();
+
+ ofproto_wait(ofproto);
+ unixctl_server_wait(unixctl);
+ dp_wait();
+ netdev_wait();
+ poll_block();
+ }
+
+ return 0;
+ }
+ \f
+ /* User interface. */
+
+ static void
+ parse_options(int argc, char *argv[], struct ofsettings *s)
+ {
+ enum {
+ OPT_DATAPATH_ID = UCHAR_MAX + 1,
+ OPT_MANUFACTURER,
+ OPT_HARDWARE,
+ OPT_SOFTWARE,
+ OPT_SERIAL,
+ OPT_ACCEPT_VCONN,
+ OPT_NO_RESOLV_CONF,
+ OPT_BR_NAME,
+ OPT_FAIL_MODE,
+ OPT_INACTIVITY_PROBE,
+ OPT_MAX_IDLE,
+ OPT_MAX_BACKOFF,
+ OPT_SNOOP,
+ OPT_RATE_LIMIT,
+ OPT_BURST_LIMIT,
+ OPT_BOOTSTRAP_CA_CERT,
+ OPT_STP,
+ OPT_NO_STP,
+ OPT_OUT_OF_BAND,
+ OPT_IN_BAND,
+ OPT_COMMAND_ACL,
+ OPT_COMMAND_DIR,
+ OPT_NETFLOW,
+ OPT_MGMT_ID,
+ VLOG_OPTION_ENUMS,
+ LEAK_CHECKER_OPTION_ENUMS
+ };
+ static struct option long_options[] = {
+ {"datapath-id", required_argument, 0, OPT_DATAPATH_ID},
+ {"manufacturer", required_argument, 0, OPT_MANUFACTURER},
+ {"hardware", required_argument, 0, OPT_HARDWARE},
+ {"software", required_argument, 0, OPT_SOFTWARE},
+ {"serial", required_argument, 0, OPT_SERIAL},
+ {"accept-vconn", required_argument, 0, OPT_ACCEPT_VCONN},
+ {"no-resolv-conf", no_argument, 0, OPT_NO_RESOLV_CONF},
+ {"config", required_argument, 0, 'F'},
+ {"br-name", required_argument, 0, OPT_BR_NAME},
+ {"fail", required_argument, 0, OPT_FAIL_MODE},
+ {"inactivity-probe", required_argument, 0, OPT_INACTIVITY_PROBE},
+ {"max-idle", required_argument, 0, OPT_MAX_IDLE},
+ {"max-backoff", required_argument, 0, OPT_MAX_BACKOFF},
+ {"listen", required_argument, 0, 'l'},
+ {"snoop", required_argument, 0, OPT_SNOOP},
+ {"rate-limit", optional_argument, 0, OPT_RATE_LIMIT},
+ {"burst-limit", required_argument, 0, OPT_BURST_LIMIT},
+ {"stp", no_argument, 0, OPT_STP},
+ {"no-stp", no_argument, 0, OPT_NO_STP},
+ {"out-of-band", no_argument, 0, OPT_OUT_OF_BAND},
+ {"in-band", no_argument, 0, OPT_IN_BAND},
+ {"command-acl", required_argument, 0, OPT_COMMAND_ACL},
+ {"command-dir", required_argument, 0, OPT_COMMAND_DIR},
+ {"netflow", required_argument, 0, OPT_NETFLOW},
+ {"mgmt-id", required_argument, 0, OPT_MGMT_ID},
+ {"verbose", optional_argument, 0, 'v'},
+ {"help", no_argument, 0, 'h'},
+ {"version", no_argument, 0, 'V'},
+ DAEMON_LONG_OPTIONS,
+ VLOG_LONG_OPTIONS,
+ LEAK_CHECKER_LONG_OPTIONS,
+ #ifdef HAVE_OPENSSL
+ VCONN_SSL_LONG_OPTIONS
+ {"bootstrap-ca-cert", required_argument, 0, OPT_BOOTSTRAP_CA_CERT},
+ #endif
+ {0, 0, 0, 0},
+ };
+ char *short_options = long_options_to_short_options(long_options);
+
+ /* Set defaults that we can figure out before parsing options. */
+ s->datapath_id = 0;
+ s->mfr_desc = NULL;
+ s->hw_desc = NULL;
+ s->sw_desc = NULL;
+ s->serial_desc = NULL;
+ svec_init(&s->listeners);
+ svec_init(&s->snoops);
+ s->fail_mode = FAIL_OPEN;
+ s->max_idle = 0;
+ s->probe_interval = 0;
- ovs_fatal(0, "-f or --fail argument must be \"open\" "
- "or \"closed\"");
++ s->max_backoff = 8;
+ s->update_resolv_conf = true;
+ s->rate_limit = 0;
+ s->burst_limit = 0;
+ s->accept_controller_re = NULL;
+ s->enable_stp = false;
+ s->in_band = true;
+ s->command_acl = "";
+ s->command_dir = NULL;
+ svec_init(&s->netflow);
+ s->mgmt_id = 0;
+ for (;;) {
+ int c;
+
+ c = getopt_long(argc, argv, short_options, long_options, NULL);
+ if (c == -1) {
+ break;
+ }
+
+ switch (c) {
+ case OPT_DATAPATH_ID:
+ if (strlen(optarg) != 12
+ || strspn(optarg, "0123456789abcdefABCDEF") != 12) {
+ ovs_fatal(0, "argument to --datapath-id must be "
+ "exactly 12 hex digits");
+ }
+ s->datapath_id = strtoll(optarg, NULL, 16);
+ if (!s->datapath_id) {
+ ovs_fatal(0, "argument to --datapath-id must be nonzero");
+ }
+ break;
+
+ case OPT_MANUFACTURER:
+ s->mfr_desc = optarg;
+ break;
+
+ case OPT_HARDWARE:
+ s->hw_desc = optarg;
+ break;
+
+ case OPT_SOFTWARE:
+ s->sw_desc = optarg;
+ break;
+
+ case OPT_SERIAL:
+ s->serial_desc = optarg;
+ break;
+
+ case OPT_ACCEPT_VCONN:
+ s->accept_controller_re = optarg;
+ break;
+
+ case OPT_NO_RESOLV_CONF:
+ s->update_resolv_conf = false;
+ break;
+
+ case OPT_FAIL_MODE:
+ if (!strcmp(optarg, "open")) {
+ s->fail_mode = FAIL_OPEN;
+ } else if (!strcmp(optarg, "closed")) {
+ s->fail_mode = FAIL_CLOSED;
+ } else {
- " attempts (default: 15 seconds)\n"
++ ovs_fatal(0, "--fail argument must be \"open\" or \"closed\"");
+ }
+ break;
+
+ case OPT_INACTIVITY_PROBE:
+ s->probe_interval = atoi(optarg);
+ if (s->probe_interval < 5) {
+ ovs_fatal(0, "--inactivity-probe argument must be at least 5");
+ }
+ break;
+
+ case OPT_MAX_IDLE:
+ if (!strcmp(optarg, "permanent")) {
+ s->max_idle = OFP_FLOW_PERMANENT;
+ } else {
+ s->max_idle = atoi(optarg);
+ if (s->max_idle < 1 || s->max_idle > 65535) {
+ ovs_fatal(0, "--max-idle argument must be between 1 and "
+ "65535 or the word 'permanent'");
+ }
+ }
+ break;
+
+ case OPT_MAX_BACKOFF:
+ s->max_backoff = atoi(optarg);
+ if (s->max_backoff < 1) {
+ ovs_fatal(0, "--max-backoff argument must be at least 1");
+ } else if (s->max_backoff > 3600) {
+ s->max_backoff = 3600;
+ }
+ break;
+
+ case OPT_RATE_LIMIT:
+ if (optarg) {
+ s->rate_limit = atoi(optarg);
+ if (s->rate_limit < 1) {
+ ovs_fatal(0, "--rate-limit argument must be at least 1");
+ }
+ } else {
+ s->rate_limit = 1000;
+ }
+ break;
+
+ case OPT_BURST_LIMIT:
+ s->burst_limit = atoi(optarg);
+ if (s->burst_limit < 1) {
+ ovs_fatal(0, "--burst-limit argument must be at least 1");
+ }
+ break;
+
+ case OPT_STP:
+ s->enable_stp = true;
+ break;
+
+ case OPT_NO_STP:
+ s->enable_stp = false;
+ break;
+
+ case OPT_OUT_OF_BAND:
+ s->in_band = false;
+ break;
+
+ case OPT_IN_BAND:
+ s->in_band = true;
+ break;
+
+ case OPT_COMMAND_ACL:
+ s->command_acl = (s->command_acl[0]
+ ? xasprintf("%s,%s", s->command_acl, optarg)
+ : optarg);
+ break;
+
+ case OPT_COMMAND_DIR:
+ s->command_dir = optarg;
+ break;
+
+ case OPT_NETFLOW:
+ svec_add(&s->netflow, optarg);
+ break;
+
+ case OPT_MGMT_ID:
+ if (strlen(optarg) != 12
+ || strspn(optarg, "0123456789abcdefABCDEF") != 12) {
+ ovs_fatal(0, "argument to --mgmt-id must be "
+ "exactly 12 hex digits");
+ }
+ s->mgmt_id = strtoll(optarg, NULL, 16);
+ if (!s->mgmt_id) {
+ ovs_fatal(0, "argument to --mgmt-id must be nonzero");
+ }
+ break;
+
+ case 'l':
+ svec_add(&s->listeners, optarg);
+ break;
+
+ case OPT_SNOOP:
+ svec_add(&s->snoops, optarg);
+ break;
+
+ case 'h':
+ usage();
+
+ case 'V':
+ OVS_PRINT_VERSION(OFP_VERSION, OFP_VERSION);
+ exit(EXIT_SUCCESS);
+
+ DAEMON_OPTION_HANDLERS
+
+ VLOG_OPTION_HANDLERS
+
+ LEAK_CHECKER_OPTION_HANDLERS
+
+ #ifdef HAVE_OPENSSL
+ VCONN_SSL_OPTION_HANDLERS
+
+ case OPT_BOOTSTRAP_CA_CERT:
+ vconn_ssl_set_ca_cert_file(optarg, true);
+ break;
+ #endif
+
+ case '?':
+ exit(EXIT_FAILURE);
+
+ default:
+ abort();
+ }
+ }
+ free(short_options);
+
+ argc -= optind;
+ argv += optind;
+ if (argc < 1 || argc > 2) {
+ ovs_fatal(0, "need one or two non-option arguments; "
+ "use --help for usage");
+ }
+
+ /* Local and remote vconns. */
+ s->dp_name = argv[0];
+ s->controller_name = argc > 1 ? xstrdup(argv[1]) : NULL;
+
+ /* Set accept_controller_regex. */
+ if (!s->accept_controller_re) {
+ s->accept_controller_re
+ = vconn_ssl_is_configured() ? "^ssl:.*" : "^tcp:.*";
+ }
+
+ /* Mode of operation. */
+ s->discovery = s->controller_name == NULL;
+ if (s->discovery && !s->in_band) {
+ ovs_fatal(0, "Cannot perform discovery with out-of-band control");
+ }
+
+ /* Rate limiting. */
+ if (s->rate_limit && s->rate_limit < 100) {
+ VLOG_WARN("Rate limit set to unusually low value %d", s->rate_limit);
+ }
+ }
+
+ static void
+ usage(void)
+ {
+ printf("%s: an OpenFlow switch implementation.\n"
+ "usage: %s [OPTIONS] DATAPATH [CONTROLLER]\n"
+ "DATAPATH is a local datapath (e.g. \"dp0\").\n"
+ "CONTROLLER is an active OpenFlow connection method; if it is\n"
+ "omitted, then ovs-openflowd performs controller discovery.\n",
+ program_name, program_name);
+ vconn_usage(true, true, true);
+ printf("\nOpenFlow options:\n"
+ " -d, --datapath-id=ID Use ID as the OpenFlow switch ID\n"
+ " (ID must consist of 12 hex digits)\n"
+ " --mgmt-id=ID Use ID as the management ID\n"
+ " (ID must consist of 12 hex digits)\n"
+ " --manufacturer=MFR Identify manufacturer as MFR\n"
+ " --hardware=HW Identify hardware as HW\n"
+ " --software=SW Identify software as SW\n"
+ " --serial=SERIAL Identify serial number as SERIAL\n"
+ "\nController discovery options:\n"
+ " --accept-vconn=REGEX accept matching discovered controllers\n"
+ " --no-resolv-conf do not update /etc/resolv.conf\n"
+ "\nNetworking options:\n"
+ " --fail=open|closed when controller connection fails:\n"
+ " closed: drop all packets\n"
+ " open (default): act as learning switch\n"
+ " --inactivity-probe=SECS time between inactivity probes\n"
+ " --max-idle=SECS max idle for flows set up by switch\n"
+ " --max-backoff=SECS max time between controller connection\n"
++ " attempts (default: 8 seconds)\n"
+ " -l, --listen=METHOD allow management connections on METHOD\n"
+ " (a passive OpenFlow connection method)\n"
+ " --snoop=METHOD allow controller snooping on METHOD\n"
+ " (a passive OpenFlow connection method)\n"
+ " --out-of-band controller connection is out-of-band\n"
+ " --netflow=HOST:PORT configure NetFlow output target\n"
+ "\nRate-limiting of \"packet-in\" messages to the controller:\n"
+ " --rate-limit[=PACKETS] max rate, in packets/s (default: 1000)\n"
+ " --burst-limit=BURST limit on packet credit for idle time\n"
+ "\nRemote command execution options:\n"
+ " --command-acl=[!]GLOB[,[!]GLOB...] set allowed/denied commands\n"
+ " --command-dir=DIR set command dir (default: %s/commands)\n",
+ ovs_pkgdatadir);
+ daemon_usage();
+ vlog_usage();
+ printf("\nOther options:\n"
+ " -h, --help display this help message\n"
+ " -V, --version display version information\n");
+ leak_checker_usage();
+ exit(EXIT_SUCCESS);
+ }
}
}
+ unixctl_command_register("bridge/dump-flows", bridge_unixctl_dump_flows);
+
+ bond_init();
bridge_reconfigure();
}
bridge_get_all_ifaces(br, &want_ifaces);
svec_diff(&want_ifaces, &cur_ifaces, &add_ifaces, NULL, NULL);
- next_port_no = 1;
for (i = 0; i < add_ifaces.n; i++) {
const char *if_name = add_ifaces.names[i];
- for (;;) {
- bool internal;
- int error;
-
- /* It's an internal interface if it's marked that way, or if
- * it's a bonded interface for which we're faking up a network
- * device. */
- internal = cfg_get_bool(0, "iface.%s.internal", if_name);
- if (cfg_get_bool(0, "bonding.%s.fake-iface", if_name)) {
- struct port *port = port_lookup(br, if_name);
- if (port && port->n_ifaces > 1) {
- internal = true;
- }
- }
- int internal = cfg_get_bool(0, "iface.%s.internal", if_name);
- int flags = internal ? ODP_PORT_INTERNAL : 0;
- int error = dpif_port_add(br->dpif, if_name, flags, NULL);
++ bool internal;
++ int error;
+
- /* Add to datapath. */
- error = dpif_port_add(&br->dpif, if_name, next_port_no++,
- internal ? ODP_PORT_INTERNAL : 0);
- if (error != EEXIST) {
- if (next_port_no >= 256) {
- VLOG_ERR("ran out of valid port numbers on dp%u",
- dpif_id(&br->dpif));
- goto out;
- }
- if (error) {
- VLOG_ERR("failed to add %s interface to dp%u: %s",
- if_name, dpif_id(&br->dpif), strerror(error));
- }
- break;
++ /* It's an internal interface if it's marked that way, or if
++ * it's a bonded interface for which we're faking up a network
++ * device. */
++ internal = cfg_get_bool(0, "iface.%s.internal", if_name);
++ if (cfg_get_bool(0, "bonding.%s.fake-iface", if_name)) {
++ struct port *port = port_lookup(br, if_name);
++ if (port && port->n_ifaces > 1) {
++ internal = true;
+ }
+ }
++
++ /* Add to datapath. */
++ error = dpif_port_add(br->dpif, if_name,
++ internal ? ODP_PORT_INTERNAL : 0, NULL);
+ if (error == EXFULL) {
+ VLOG_ERR("ran out of valid port numbers on %s",
+ dpif_name(br->dpif));
+ break;
+ } else if (error) {
+ VLOG_ERR("failed to add %s interface to %s: %s",
+ if_name, dpif_name(br->dpif), strerror(error));
+ }
}
- out:
svec_destroy(&cur_ifaces);
svec_destroy(&want_ifaces);
svec_destroy(&add_ifaces);
if (port->is_mirror_output_port) {
continue;
}
- for (j = 0; j < port->n_ifaces; j++) {
- struct iface *iface = port->ifaces[j];
- uint8_t iface_ea[ETH_ADDR_LEN];
+
+ /* Choose the MAC address to represent the port. */
+ iface_ea_u64 = cfg_get_mac(0, "port.%s.mac", port->name);
+ if (iface_ea_u64) {
+ /* User specified explicitly. */
+ eth_addr_from_uint64(iface_ea_u64, iface_ea);
+
+ /* Find the interface with this Ethernet address (if any) so that
+ * we can provide the correct devname to the caller. */
+ iface = NULL;
+ for (j = 0; j < port->n_ifaces; j++) {
+ struct iface *candidate = port->ifaces[j];
+ uint8_t candidate_ea[ETH_ADDR_LEN];
- if (!netdev_nodev_get_etheraddr(candidate->name, candidate_ea)
++ if (!netdev_get_etheraddr(candidate->netdev, candidate_ea)
+ && eth_addr_equals(iface_ea, candidate_ea)) {
+ iface = candidate;
+ }
+ }
+ } else {
+ /* Choose the interface whose MAC address will represent the port.
+ * The Linux kernel bonding code always chooses the MAC address of
+ * the first slave added to a bond, and the Fedora networking
+ * scripts always add slaves to a bond in alphabetical order, so
+ * for compatibility we choose the interface with the name that is
+ * first in alphabetical order. */
+ iface = port->ifaces[0];
+ for (j = 1; j < port->n_ifaces; j++) {
+ struct iface *candidate = port->ifaces[j];
+ if (strcmp(candidate->name, iface->name) < 0) {
+ iface = candidate;
+ }
+ }
+
+ /* The local port doesn't count (since we're trying to choose its
+ * MAC address anyway). Other internal ports don't count because
+ * we really want a physical MAC if we can get it, and internal
+ * ports typically have randomly generated MACs. */
if (iface->dp_ifidx == ODPP_LOCAL
|| cfg_get_bool(0, "iface.%s.internal", iface->name)) {
continue;
}
- error = netdev_nodev_get_etheraddr(iface->name, iface_ea);
+
+ /* Grab MAC. */
- if (!error) {
- if (!eth_addr_is_multicast(iface_ea) &&
- !eth_addr_is_reserved(iface_ea) &&
- !eth_addr_is_zero(iface_ea) &&
- memcmp(iface_ea, ea, ETH_ADDR_LEN) < 0) {
- memcpy(ea, iface_ea, ETH_ADDR_LEN);
- *hw_addr_iface = iface;
- }
- } else {
+ error = netdev_get_etheraddr(iface->netdev, iface_ea);
+ if (error) {
static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 5);
VLOG_ERR_RL(&rl, "failed to obtain Ethernet address of %s: %s",
iface->name, strerror(error));
+ continue;
}
}
- memcpy(ea, iface_ea, ETH_ADDR_LEN);
- *devname = iface ? iface->name : NULL;
+
+ /* Compare against our current choice. */
+ if (!eth_addr_is_multicast(iface_ea) &&
+ !eth_addr_is_reserved(iface_ea) &&
+ !eth_addr_is_zero(iface_ea) &&
+ memcmp(iface_ea, ea, ETH_ADDR_LEN) < 0)
+ {
++ *hw_addr_iface = iface;
+ }
}
if (eth_addr_is_multicast(ea) || eth_addr_is_vif(ea)) {
memcpy(ea, br->default_ea, ETH_ADDR_LEN);
if (slave->up) {
bond.up = true;
}
- memcpy(slave->mac, iface->mac, ETH_ADDR_LEN);
+ netdev_get_etheraddr(iface->netdev, slave->mac);
}
+
proc_net_compat_update_bond(port->name, &bond);
free(bond.slaves);
}
}
unixctl_command_register("vswitchd/reload", reload);
- cfg_read();
+ retval = cfg_read();
+ if (retval) {
+ ovs_fatal(retval, "could not read config file");
+ }
mgmt_init();
bridge_init();
- port_init();
mgmt_reconfigure();
need_reconfigure = false;
import traceback
import time
import re
- import pickle
+import random
+ from xml.dom.minidom import getDOMImplementation
+ from xml.dom.minidom import parse as parseXML
output_directory = None
def interface_exists(i):
return os.path.exists("/sys/class/net/" + i)
+def get_netdev_mac(device):
+ try:
+ return read_first_line_of_file("/sys/class/net/%s/address" % device)
+ except:
+ # Probably no such device.
+ return None
+
+def get_netdev_tx_queue_len(device):
+ try:
+ return int(read_first_line_of_file("/sys/class/net/%s/tx_queue_len"
+ % device))
+ except:
+ # Probably no such device.
+ return None
+
+def get_netdev_by_mac(mac):
+ maybe = None
+ for device in os.listdir("/sys/class/net"):
+ dev_mac = get_netdev_mac(device)
+ if dev_mac and mac.lower() == dev_mac.lower():
+ if get_netdev_tx_queue_len(device):
+ return device
+ if not maybe:
+ # Probably a datapath internal port.
+ maybe = device
+ return maybe
+
+ #
+ # Helper functions for encoding/decoding database attributes to/from XML.
+ #
+ def str_to_xml(xml, parent, tag, val):
+ e = xml.createElement(tag)
+ parent.appendChild(e)
+ v = xml.createTextNode(val)
+ e.appendChild(v)
+ def str_from_xml(n):
+ def getText(nodelist):
+ rc = ""
+ for node in nodelist:
+ if node.nodeType == node.TEXT_NODE:
+ rc = rc + node.data
+ return rc
+ return getText(n.childNodes).strip()
+
+
+ def bool_to_xml(xml, parent, tag, val):
+ if val:
+ str_to_xml(xml, parent, tag, "True")
+ else:
+ str_to_xml(xml, parent, tag, "False")
+ def bool_from_xml(n):
+ s = str_from_xml(n)
+ if s == "True":
+ return True
+ elif s == "False":
+ return False
+ else:
+ raise Error("Unknown boolean value %s" % s);
+
+ def strlist_to_xml(xml, parent, ltag, itag, val):
+ e = xml.createElement(ltag)
+ parent.appendChild(e)
+ for v in val:
+ c = xml.createElement(itag)
+ e.appendChild(c)
+ cv = xml.createTextNode(v)
+ c.appendChild(cv)
+ def strlist_from_xml(n, ltag, itag):
+ ret = []
+ for n in n.childNodes:
+ if n.nodeName == itag:
+ ret.append(str_from_xml(n))
+ return ret
+
+ def otherconfig_to_xml(xml, parent, val, attrs):
+ otherconfig = xml.createElement("other_config")
+ parent.appendChild(otherconfig)
+ for n,v in val.items():
+ if not n in attrs:
+ raise Error("Unknown other-config attribute: %s" % n)
+ str_to_xml(xml, otherconfig, n, v)
+ def otherconfig_from_xml(n, attrs):
+ ret = {}
+ for n in n.childNodes:
+ if n.nodeName in attrs:
+ ret[n.nodeName] = str_from_xml(n)
+ return ret
+
+ #
+ # Definitions of the database objects (and their attributes) used by interface-reconfigure.
+ #
+ # Each object is defined by a dictionary mapping an attribute name in
+ # the xapi database to a tuple containing two items:
+ # - a function which takes this attribute and encodes it as XML.
+ # - a function which takes XML and decocdes it into a value.
+ #
+ # other-config attributes are specified as a simple array of strings
+
+ PIF_XML_TAG = "pif"
+ VLAN_XML_TAG = "vlan"
+ BOND_XML_TAG = "bond"
+ NETWORK_XML_TAG = "network"
+
+ ETHTOOL_OTHERCONFIG_ATTRS = ['ethtool-%s' % x for x in 'autoneg', 'speed', 'duplex', 'rx', 'tx', 'sg', 'tso', 'ufo', 'gso' ]
+
+ PIF_ATTRS = { 'uuid': (str_to_xml,str_from_xml),
+ 'management': (bool_to_xml,bool_from_xml),
+ 'network': (str_to_xml,str_from_xml),
+ 'device': (str_to_xml,str_from_xml),
+ 'bond_master_of': (lambda x, p, t, v: strlist_to_xml(x, p, 'bond_master_of', 'slave', v),
+ lambda n: strlist_from_xml(n, 'bond_master_of', 'slave')),
+ 'bond_slave_of': (str_to_xml,str_from_xml),
+ 'VLAN': (str_to_xml,str_from_xml),
+ 'VLAN_master_of': (str_to_xml,str_from_xml),
+ 'VLAN_slave_of': (lambda x, p, t, v: strlist_to_xml(x, p, 'VLAN_slave_of', 'master', v),
+ lambda n: strlist_from_xml(n, 'VLAN_slave_Of', 'master')),
+ 'ip_configuration_mode': (str_to_xml,str_from_xml),
+ 'IP': (str_to_xml,str_from_xml),
+ 'netmask': (str_to_xml,str_from_xml),
+ 'gateway': (str_to_xml,str_from_xml),
+ 'DNS': (str_to_xml,str_from_xml),
+ 'MAC': (str_to_xml,str_from_xml),
+ 'other_config': (lambda x, p, t, v: otherconfig_to_xml(x, p, v, PIF_OTHERCONFIG_ATTRS),
+ lambda n: otherconfig_from_xml(n, PIF_OTHERCONFIG_ATTRS)),
+ }
+
+ PIF_OTHERCONFIG_ATTRS = [ 'domain', 'peerdns', 'defaultroute', 'mtu', 'static-routes' ] + \
+ [ 'bond-%s' % x for x in 'mode', 'miimon', 'downdelay', 'updelay', 'use_carrier' ] + \
+ ETHTOOL_OTHERCONFIG_ATTRS
+
+ VLAN_ATTRS = { 'uuid': (str_to_xml,str_from_xml),
+ 'tagged_PIF': (str_to_xml,str_from_xml),
+ 'untagged_PIF': (str_to_xml,str_from_xml),
+ }
+
+ BOND_ATTRS = { 'uuid': (str_to_xml,str_from_xml),
+ 'master': (str_to_xml,str_from_xml),
+ 'slaves': (lambda x, p, t, v: strlist_to_xml(x, p, 'slaves', 'slave', v),
+ lambda n: strlist_from_xml(n, 'slaves', 'slave')),
+ }
+
+ NETWORK_ATTRS = { 'uuid': (str_to_xml,str_from_xml),
+ 'bridge': (str_to_xml,str_from_xml),
+ 'PIFs': (lambda x, p, t, v: strlist_to_xml(x, p, 'PIFs', 'PIF', v),
+ lambda n: strlist_from_xml(n, 'PIFs', 'PIF')),
+ 'other_config': (lambda x, p, t, v: otherconfig_to_xml(x, p, v, NETWORK_OTHERCONFIG_ATTRS),
+ lambda n: otherconfig_from_xml(n, NETWORK_OTHERCONFIG_ATTRS)),
+ }
+
+ NETWORK_OTHERCONFIG_ATTRS = [ 'mtu', 'static-routes' ] + ETHTOOL_OTHERCONFIG_ATTRS
+
class DatabaseCache(object):
+ def __read_xensource_inventory(self):
+ filename = "/etc/xensource-inventory"
+ f = open(filename, "r")
+ lines = [x.strip("\n") for x in f.readlines()]
+ f.close()
+
+ defs = [ (l[:l.find("=")], l[(l.find("=") + 1):]) for l in lines ]
+ defs = [ (a, b.strip("'")) for (a,b) in defs ]
+
+ return dict(defs)
+ def __pif_on_host(self,pif):
+ return self.__pifs.has_key(pif)
+
+ def __get_pif_records_from_xapi(self, session, host):
+ self.__pifs = {}
+ for (p,rec) in session.xenapi.PIF.get_all_records().items():
+ if rec['host'] != host:
+ continue
+ self.__pifs[p] = {}
+ for f in PIF_ATTRS:
+ self.__pifs[p][f] = rec[f]
+ self.__pifs[p]['other_config'] = {}
+ for f in PIF_OTHERCONFIG_ATTRS:
+ if not rec['other_config'].has_key(f): continue
+ self.__pifs[p]['other_config'][f] = rec['other_config'][f]
+
+ def __get_vlan_records_from_xapi(self, session):
+ self.__vlans = {}
+ for v in session.xenapi.VLAN.get_all():
+ rec = session.xenapi.VLAN.get_record(v)
+ if not self.__pif_on_host(rec['untagged_PIF']):
+ continue
+ self.__vlans[v] = {}
+ for f in VLAN_ATTRS:
+ self.__vlans[v][f] = rec[f]
+
+ def __get_bond_records_from_xapi(self, session):
+ self.__bonds = {}
+ for b in session.xenapi.Bond.get_all():
+ rec = session.xenapi.Bond.get_record(b)
+ if not self.__pif_on_host(rec['master']):
+ continue
+ self.__bonds[b] = {}
+ for f in BOND_ATTRS:
+ self.__bonds[b][f] = rec[f]
+
+ def __get_network_records_from_xapi(self, session):
+ self.__networks = {}
+ for n in session.xenapi.network.get_all():
+ rec = session.xenapi.network.get_record(n)
+ self.__networks[n] = {}
+ for f in NETWORK_ATTRS:
+ self.__networks[n][f] = rec[f]
+ self.__networks[n]['other_config'] = {}
+ for f in NETWORK_OTHERCONFIG_ATTRS:
+ if not rec['other_config'].has_key(f): continue
+ self.__networks[n]['other_config'][f] = rec['other_config'][f]
+
+ def __to_xml(self, xml, parent, key, ref, rec, attrs):
+ """Encode a database object as XML"""
+ e = xml.createElement(key)
+ parent.appendChild(e)
+ if ref:
+ e.setAttribute('ref', ref)
+
+ for n,v in rec.items():
+ if attrs.has_key(n):
+ h,_ = attrs[n]
+ h(xml, e, n, v)
+ else:
+ raise Error("Unknown attribute %s" % n)
+ def __from_xml(self, e, attrs):
+ """Decode a database object from XML"""
+ ref = e.attributes['ref'].value
+ rec = {}
+ for n in e.childNodes:
+ if n.nodeName in attrs:
+ _,h = attrs[n.nodeName]
+ rec[n.nodeName] = h(n)
+ return (ref,rec)
+
def __init__(self, session_ref=None, cache_file=None):
if session_ref and cache_file:
raise Error("can't specify session reference and cache file")
pifrec = db.get_pif_record(pif)
return bridge_name(pif)
-def physdev_names(pif):
- """Return the name(s) of the physical network device(s) associated with pif.
-For a VLAN PIF, the physical devices are the VLAN slave's physical devices.
-For a bond master PIF, the physical devices are the bond slaves.
-For a non-VLAN, non-bond master PIF, the physical device is the PIF itself.
+def get_physdev_pifs(pif):
+ """Return the PIFs for the physical network device(s) associated with pif.
+For a VLAN PIF, this is the VLAN slave's physical device PIF.
+For a bond master PIF, these are the bond slave PIFs.
+For a non-VLAN, non-bond master PIF, the PIF is its own physical device PIF.
"""
-
pifrec = db.get_pif_record(pif)
if pifrec['VLAN'] != '-1':
projects / cascardo / ovs.git / commitdiff