/*
- * Copyright (c) 2009, 2010 Nicira Networks.
+ * Copyright (c) 2009, 2010, 2011 Nicira Networks.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
#include <linux/ip.h>
#include <linux/types.h>
#include <linux/ethtool.h>
+#include <linux/mii.h>
#include <linux/pkt_sched.h>
#include <linux/rtnetlink.h>
#include <linux/sockios.h>
#include <unistd.h>
#include "coverage.h"
+#include "dpif-linux.h"
#include "dynamic-string.h"
#include "fatal-signal.h"
#include "hash.h"
#include "netdev-provider.h"
#include "netdev-vport.h"
#include "netlink.h"
+#include "netlink-socket.h"
#include "ofpbuf.h"
#include "openflow/openflow.h"
#include "packets.h"
#include "poll-loop.h"
#include "rtnetlink.h"
+#include "rtnetlink-link.h"
#include "socket-util.h"
#include "shash.h"
#include "svec.h"
/* Create rtnetlink socket. */
if (!status) {
- status = nl_sock_create(NETLINK_ROUTE, 0, 0, 0, &rtnl_sock);
+ status = nl_sock_create(NETLINK_ROUTE, &rtnl_sock);
if (status) {
VLOG_ERR_RL(&rl, "failed to create rtnetlink socket: %s",
strerror(status));
static void
netdev_linux_run(void)
{
- rtnetlink_notifier_run();
+ rtnetlink_link_notifier_run();
}
static void
netdev_linux_wait(void)
{
- rtnetlink_notifier_wait();
+ rtnetlink_link_notifier_wait();
}
static void
-netdev_linux_cache_cb(const struct rtnetlink_change *change,
+netdev_linux_cache_cb(const struct rtnetlink_link_change *change,
void *aux OVS_UNUSED)
{
struct netdev_dev_linux *dev;
}
}
-/* Creates the netdev device of 'type' with 'name'. */
+/* Creates system and internal devices. */
static int
-netdev_linux_create_system(const struct netdev_class *class OVS_UNUSED,
+netdev_linux_create(const struct netdev_class *class,
const char *name, const struct shash *args,
struct netdev_dev **netdev_devp)
{
int error;
if (!shash_is_empty(args)) {
- VLOG_WARN("%s: arguments for system devices should be empty", name);
+ VLOG_WARN("%s: arguments for %s devices should be empty",
+ name, class->type);
}
if (!cache_notifier_refcount) {
- error = rtnetlink_notifier_register(&netdev_linux_cache_notifier,
- netdev_linux_cache_cb, NULL);
+ error = rtnetlink_link_notifier_register(&netdev_linux_cache_notifier,
+ netdev_linux_cache_cb, NULL);
if (error) {
return error;
}
cache_notifier_refcount++;
netdev_dev = xzalloc(sizeof *netdev_dev);
- netdev_dev_init(&netdev_dev->netdev_dev, name, &netdev_linux_class);
+ netdev_dev_init(&netdev_dev->netdev_dev, name, args, class);
*netdev_devp = &netdev_dev->netdev_dev;
return 0;
/* Create tap device. */
ifr.ifr_flags = IFF_TAP | IFF_NO_PI;
- strncpy(ifr.ifr_name, name, sizeof ifr.ifr_name);
+ ovs_strzcpy(ifr.ifr_name, name, sizeof ifr.ifr_name);
if (ioctl(state->fd, TUNSETIFF, &ifr) == -1) {
VLOG_WARN("%s: creating tap device failed: %s", name,
strerror(errno));
goto error;
}
- netdev_dev_init(&netdev_dev->netdev_dev, name, &netdev_tap_class);
+ netdev_dev_init(&netdev_dev->netdev_dev, name, args, &netdev_tap_class);
*netdev_devp = &netdev_dev->netdev_dev;
return 0;
netdev_linux_destroy(struct netdev_dev *netdev_dev_)
{
struct netdev_dev_linux *netdev_dev = netdev_dev_linux_cast(netdev_dev_);
- const char *type = netdev_dev_get_type(netdev_dev_);
+ const struct netdev_class *class = netdev_dev_get_class(netdev_dev_);
if (netdev_dev->tc && netdev_dev->tc->ops->tc_destroy) {
netdev_dev->tc->ops->tc_destroy(netdev_dev->tc);
}
- if (!strcmp(type, "system")) {
+ if (class == &netdev_linux_class || class == &netdev_internal_class) {
cache_notifier_refcount--;
if (!cache_notifier_refcount) {
- rtnetlink_notifier_unregister(&netdev_linux_cache_notifier);
+ rtnetlink_link_notifier_unregister(&netdev_linux_cache_notifier);
}
- } else if (!strcmp(type, "tap")) {
+ } else if (class == &netdev_tap_class) {
destroy_tap(netdev_dev);
+ } else {
+ NOT_REACHED();
}
free(netdev_dev);
netdev->fd = -1;
netdev_init(&netdev->netdev, netdev_dev_);
- error = netdev_get_flags(&netdev->netdev, &flags);
- if (error == ENODEV) {
- goto error;
+ /* Verify that the device really exists, by attempting to read its flags.
+ * (The flags might be cached, in which case this won't actually do an
+ * ioctl.)
+ *
+ * Don't do this for "internal" netdevs, though, because those have to be
+ * created as netdev objects before they exist in the kernel, because
+ * creating them in the kernel happens by passing a netdev object to
+ * dpif_port_add(). */
+ if (netdev_dev_get_class(netdev_dev_) != &netdev_internal_class) {
+ error = netdev_get_flags(&netdev->netdev, &flags);
+ if (error == ENODEV) {
+ goto error;
+ }
}
if (!strcmp(netdev_dev_get_type(netdev_dev_), "tap") &&
return error;
}
+static int
+netdev_linux_do_miimon(const struct netdev *netdev, int cmd,
+ const char *cmd_name, struct mii_ioctl_data *data)
+{
+ struct ifreq ifr;
+ int error;
+
+ memset(&ifr, 0, sizeof ifr);
+ memcpy(&ifr.ifr_data, data, sizeof *data);
+ error = netdev_linux_do_ioctl(netdev_get_name(netdev),
+ &ifr, cmd, cmd_name);
+ memcpy(data, &ifr.ifr_data, sizeof *data);
+
+ return error;
+}
+
+static int
+netdev_linux_get_miimon(const struct netdev *netdev, bool *miimon)
+{
+ const char *name = netdev_get_name(netdev);
+ struct mii_ioctl_data data;
+ int error;
+
+ *miimon = false;
+
+ memset(&data, 0, sizeof data);
+ error = netdev_linux_do_miimon(netdev, SIOCGMIIPHY, "SIOCGMIIPHY", &data);
+ if (!error) {
+ /* data.phy_id is filled out by previous SIOCGMIIPHY miimon call. */
+ data.reg_num = MII_BMSR;
+ error = netdev_linux_do_miimon(netdev, SIOCGMIIREG, "SIOCGMIIREG",
+ &data);
+
+ if (!error) {
+ *miimon = !!(data.val_out & BMSR_LSTATUS);
+ } else {
+ VLOG_WARN_RL(&rl, "%s: failed to query MII", name);
+ }
+ } else {
+ struct ethtool_cmd ecmd;
+
+ VLOG_DBG_RL(&rl, "%s: failed to query MII, falling back to ethtool",
+ name);
+
+ memset(&ecmd, 0, sizeof ecmd);
+ error = netdev_linux_do_ethtool(name, &ecmd, ETHTOOL_GLINK,
+ "ETHTOOL_GLINK");
+ if (!error) {
+ struct ethtool_value eval;
+
+ memcpy(&eval, &ecmd, sizeof eval);
+ *miimon = !!eval.data;
+ } else {
+ VLOG_WARN_RL(&rl, "%s: ethtool link status failed", name);
+ }
+ }
+
+ return error;
+}
+
/* Check whether we can we use RTM_GETLINK to get network device statistics.
* In pre-2.6.19 kernels, this was only available if wireless extensions were
* enabled. */
const char *type = netdev_dev_get_type(&netdev_dev->netdev_dev);
netdev_dev->is_tap = !strcmp(type, "tap");
- netdev_dev->is_internal = false;
- if (!netdev_dev->is_tap) {
- struct ethtool_drvinfo drvinfo;
- int error;
-
- memset(&drvinfo, 0, sizeof drvinfo);
- error = netdev_linux_do_ethtool(name,
- (struct ethtool_cmd *)&drvinfo,
- ETHTOOL_GDRVINFO,
- "ETHTOOL_GDRVINFO");
-
- if (!error && !strcmp(drvinfo.driver, "openvswitch")) {
- netdev_dev->is_internal = true;
- }
- }
-
+ netdev_dev->is_internal = (!netdev_dev->is_tap
+ && dpif_linux_is_internal_device(name));
netdev_dev->cache_valid |= VALID_IS_PSEUDO;
}
}
* bitmap of "enum ofp_port_features" bits, in host byte order. Returns 0 if
* successful, otherwise a positive errno value. */
static int
-netdev_linux_get_features(struct netdev *netdev,
+netdev_linux_get_features(const struct netdev *netdev,
uint32_t *current, uint32_t *advertised,
uint32_t *supported, uint32_t *peer)
{
int ioctl_nr, const char *ioctl_name, struct in_addr addr)
{
struct ifreq ifr;
- strncpy(ifr.ifr_name, netdev_get_name(netdev), sizeof ifr.ifr_name);
+ ovs_strzcpy(ifr.ifr_name, netdev_get_name(netdev), sizeof ifr.ifr_name);
make_in4_sockaddr(&ifr.ifr_addr, addr);
return netdev_linux_do_ioctl(netdev_get_name(netdev), &ifr, ioctl_nr,
return ENXIO;
}
+static int
+netdev_linux_get_status(const struct netdev *netdev, struct shash *sh)
+{
+ struct ethtool_drvinfo drvinfo;
+ int error;
+
+ memset(&drvinfo, 0, sizeof drvinfo);
+ error = netdev_linux_do_ethtool(netdev_get_name(netdev),
+ (struct ethtool_cmd *)&drvinfo,
+ ETHTOOL_GDRVINFO,
+ "ETHTOOL_GDRVINFO");
+ if (!error) {
+ shash_add(sh, "driver_name", xstrdup(drvinfo.driver));
+ shash_add(sh, "driver_version", xstrdup(drvinfo.version));
+ shash_add(sh, "firmware_version", xstrdup(drvinfo.fw_version));
+ }
+
+ return error;
+}
+
/* Looks up the ARP table entry for 'ip' on 'netdev'. If one exists and can be
* successfully retrieved, it stores the corresponding MAC address in 'mac' and
* returns 0. Otherwise, it returns a positive errno value; in particular,
memcpy(&r.arp_pa, &sin, sizeof sin);
r.arp_ha.sa_family = ARPHRD_ETHER;
r.arp_flags = 0;
- strncpy(r.arp_dev, netdev_get_name(netdev), sizeof r.arp_dev);
+ ovs_strzcpy(r.arp_dev, netdev_get_name(netdev), sizeof r.arp_dev);
COVERAGE_INC(netdev_arp_lookup);
retval = ioctl(af_inet_sock, SIOCGARP, &r) < 0 ? errno : 0;
if (!retval) {
}
static void
-netdev_linux_poll_cb(const struct rtnetlink_change *change,
+netdev_linux_poll_cb(const struct rtnetlink_link_change *change,
void *aux OVS_UNUSED)
{
if (change) {
struct list *list;
if (shash_is_empty(&netdev_linux_notifiers)) {
- int error = rtnetlink_notifier_register(&netdev_linux_poll_notifier,
- netdev_linux_poll_cb, NULL);
+ int error;
+ error = rtnetlink_link_notifier_register(&netdev_linux_poll_notifier,
+ netdev_linux_poll_cb, NULL);
if (error) {
return error;
}
/* If that was the last notifier, unregister. */
if (shash_is_empty(&netdev_linux_notifiers)) {
- rtnetlink_notifier_unregister(&netdev_linux_poll_notifier);
- }
-}
-
-const struct netdev_class netdev_linux_class = {
- "system",
-
- netdev_linux_init,
- netdev_linux_run,
- netdev_linux_wait,
-
- netdev_linux_create_system,
- netdev_linux_destroy,
- NULL, /* reconfigure */
-
- netdev_linux_open,
- netdev_linux_close,
-
- netdev_linux_enumerate,
-
- netdev_linux_recv,
- netdev_linux_recv_wait,
- netdev_linux_drain,
-
- netdev_linux_send,
- netdev_linux_send_wait,
-
- netdev_linux_set_etheraddr,
- netdev_linux_get_etheraddr,
- netdev_linux_get_mtu,
- netdev_linux_get_ifindex,
- netdev_linux_get_carrier,
- netdev_linux_get_stats,
- netdev_vport_set_stats,
-
- netdev_linux_get_features,
- netdev_linux_set_advertisements,
- netdev_linux_get_vlan_vid,
-
- netdev_linux_set_policing,
- netdev_linux_get_qos_types,
- netdev_linux_get_qos_capabilities,
- netdev_linux_get_qos,
- netdev_linux_set_qos,
- netdev_linux_get_queue,
- netdev_linux_set_queue,
- netdev_linux_delete_queue,
- netdev_linux_get_queue_stats,
- netdev_linux_dump_queues,
- netdev_linux_dump_queue_stats,
-
- netdev_linux_get_in4,
- netdev_linux_set_in4,
- netdev_linux_get_in6,
- netdev_linux_add_router,
- netdev_linux_get_next_hop,
- netdev_linux_arp_lookup,
-
- netdev_linux_update_flags,
-
- netdev_linux_poll_add,
- netdev_linux_poll_remove,
-};
-
-const struct netdev_class netdev_tap_class = {
- "tap",
-
- netdev_linux_init,
- netdev_linux_run,
- netdev_linux_wait,
-
- netdev_linux_create_tap,
- netdev_linux_destroy,
- NULL, /* reconfigure */
-
- netdev_linux_open,
- netdev_linux_close,
-
- NULL, /* enumerate */
-
- netdev_linux_recv,
- netdev_linux_recv_wait,
- netdev_linux_drain,
-
- netdev_linux_send,
- netdev_linux_send_wait,
-
- netdev_linux_set_etheraddr,
- netdev_linux_get_etheraddr,
- netdev_linux_get_mtu,
- netdev_linux_get_ifindex,
- netdev_linux_get_carrier,
- netdev_linux_get_stats,
- NULL, /* set_stats */
-
- netdev_linux_get_features,
- netdev_linux_set_advertisements,
- netdev_linux_get_vlan_vid,
-
- netdev_linux_set_policing,
- netdev_linux_get_qos_types,
- netdev_linux_get_qos_capabilities,
- netdev_linux_get_qos,
- netdev_linux_set_qos,
- netdev_linux_get_queue,
- netdev_linux_set_queue,
- netdev_linux_delete_queue,
- netdev_linux_get_queue_stats,
- netdev_linux_dump_queues,
- netdev_linux_dump_queue_stats,
-
- netdev_linux_get_in4,
- netdev_linux_set_in4,
- netdev_linux_get_in6,
- netdev_linux_add_router,
- netdev_linux_get_next_hop,
- netdev_linux_arp_lookup,
-
- netdev_linux_update_flags,
-
- netdev_linux_poll_add,
- netdev_linux_poll_remove,
-};
+ rtnetlink_link_notifier_unregister(&netdev_linux_poll_notifier);
+ }
+}
+
+#define NETDEV_LINUX_CLASS(NAME, CREATE, ENUMERATE, SET_STATS) \
+{ \
+ NAME, \
+ \
+ netdev_linux_init, \
+ netdev_linux_run, \
+ netdev_linux_wait, \
+ \
+ CREATE, \
+ netdev_linux_destroy, \
+ NULL, /* set_config */ \
+ \
+ netdev_linux_open, \
+ netdev_linux_close, \
+ \
+ ENUMERATE, \
+ \
+ netdev_linux_recv, \
+ netdev_linux_recv_wait, \
+ netdev_linux_drain, \
+ \
+ netdev_linux_send, \
+ netdev_linux_send_wait, \
+ \
+ netdev_linux_set_etheraddr, \
+ netdev_linux_get_etheraddr, \
+ netdev_linux_get_mtu, \
+ netdev_linux_get_ifindex, \
+ netdev_linux_get_carrier, \
+ netdev_linux_get_miimon, \
+ netdev_linux_get_stats, \
+ SET_STATS, \
+ \
+ netdev_linux_get_features, \
+ netdev_linux_set_advertisements, \
+ netdev_linux_get_vlan_vid, \
+ \
+ netdev_linux_set_policing, \
+ netdev_linux_get_qos_types, \
+ netdev_linux_get_qos_capabilities, \
+ netdev_linux_get_qos, \
+ netdev_linux_set_qos, \
+ netdev_linux_get_queue, \
+ netdev_linux_set_queue, \
+ netdev_linux_delete_queue, \
+ netdev_linux_get_queue_stats, \
+ netdev_linux_dump_queues, \
+ netdev_linux_dump_queue_stats, \
+ \
+ netdev_linux_get_in4, \
+ netdev_linux_set_in4, \
+ netdev_linux_get_in6, \
+ netdev_linux_add_router, \
+ netdev_linux_get_next_hop, \
+ netdev_linux_get_status, \
+ netdev_linux_arp_lookup, \
+ \
+ netdev_linux_update_flags, \
+ \
+ netdev_linux_poll_add, \
+ netdev_linux_poll_remove \
+}
+
+const struct netdev_class netdev_linux_class =
+ NETDEV_LINUX_CLASS(
+ "system",
+ netdev_linux_create,
+ netdev_linux_enumerate,
+ NULL); /* set_stats */
+
+const struct netdev_class netdev_tap_class =
+ NETDEV_LINUX_CLASS(
+ "tap",
+ netdev_linux_create_tap,
+ NULL, /* enumerate */
+ NULL); /* set_stats */
+
+const struct netdev_class netdev_internal_class =
+ NETDEV_LINUX_CLASS(
+ "internal",
+ netdev_linux_create,
+ NULL, /* enumerate */
+ netdev_vport_set_stats);
\f
/* HTB traffic control class. */
int mtu;
netdev_get_mtu(netdev, &mtu);
+ if (mtu == INT_MAX) {
+ VLOG_WARN_RL(&rl, "cannot set up HTB on device %s that lacks MTU",
+ netdev_get_name(netdev));
+ return EINVAL;
+ }
memset(&opt, 0, sizeof opt);
tc_fill_rate(&opt.rate, class->min_rate, mtu);
const char *priority_s = shash_find_data(details, "priority");
int mtu;
+ netdev_get_mtu(netdev, &mtu);
+ if (mtu == INT_MAX) {
+ VLOG_WARN_RL(&rl, "cannot parse HTB class on device %s that lacks MTU",
+ netdev_get_name(netdev));
+ return EINVAL;
+ }
+
/* min-rate. Don't allow a min-rate below 1500 bytes/s. */
if (!min_rate_s) {
/* min-rate is required. */
* doesn't include the Ethernet header, we need to add at least 14 (18?) to
* the MTU. We actually add 64, instead of 14, as a guard against
* additional headers get tacked on somewhere that we're not aware of. */
- netdev_get_mtu(netdev, &mtu);
hc->burst = burst_s ? strtoull(burst_s, NULL, 10) / 8 : 0;
hc->burst = MAX(hc->burst, mtu + 64);
{
struct ifreq ifr;
- strncpy(ifr.ifr_name, netdev_name, sizeof ifr.ifr_name);
+ ovs_strzcpy(ifr.ifr_name, netdev_name, sizeof ifr.ifr_name);
COVERAGE_INC(netdev_get_ifindex);
if (ioctl(af_inet_sock, SIOCGIFINDEX, &ifr) < 0) {
VLOG_WARN_RL(&rl, "ioctl(SIOCGIFINDEX) on %s device failed: %s",
int hwaddr_family;
memset(&ifr, 0, sizeof ifr);
- strncpy(ifr.ifr_name, netdev_name, sizeof ifr.ifr_name);
+ ovs_strzcpy(ifr.ifr_name, netdev_name, sizeof ifr.ifr_name);
COVERAGE_INC(netdev_get_hwaddr);
if (ioctl(af_inet_sock, SIOCGIFHWADDR, &ifr) < 0) {
VLOG_ERR("ioctl(SIOCGIFHWADDR) on %s device failed: %s",
struct ifreq ifr;
memset(&ifr, 0, sizeof ifr);
- strncpy(ifr.ifr_name, netdev_name, sizeof ifr.ifr_name);
+ ovs_strzcpy(ifr.ifr_name, netdev_name, sizeof ifr.ifr_name);
ifr.ifr_hwaddr.sa_family = hwaddr_family;
memcpy(ifr.ifr_hwaddr.sa_data, mac, ETH_ADDR_LEN);
COVERAGE_INC(netdev_set_hwaddr);
struct ifreq ifr;
memset(&ifr, 0, sizeof ifr);
- strncpy(ifr.ifr_name, name, sizeof ifr.ifr_name);
+ ovs_strzcpy(ifr.ifr_name, name, sizeof ifr.ifr_name);
ifr.ifr_data = (caddr_t) ecmd;
ecmd->cmd = cmd;
netdev_linux_do_ioctl(const char *name, struct ifreq *ifr, int cmd,
const char *cmd_name)
{
- strncpy(ifr->ifr_name, name, sizeof ifr->ifr_name);
+ ovs_strzcpy(ifr->ifr_name, name, sizeof ifr->ifr_name);
if (ioctl(af_inet_sock, cmd, ifr) == -1) {
VLOG_DBG_RL(&rl, "%s: ioctl(%s) failed: %s", name, cmd_name,
strerror(errno));