Commit cefd81cf authored by David S. Miller's avatar David S. Miller
parents 3731a334 15eac2a7
......@@ -266,7 +266,7 @@ static int do_output(struct datapath *dp, struct sk_buff *skb, int out_port)
if (unlikely(!skb))
return -ENOMEM;
vport = rcu_dereference(dp->ports[out_port]);
vport = ovs_vport_rcu(dp, out_port);
if (unlikely(!vport)) {
kfree_skb(skb);
return -ENODEV;
......
This diff is collapsed.
......@@ -27,10 +27,11 @@
#include <linux/u64_stats_sync.h>
#include "flow.h"
#include "vport.h"
struct vport;
#define DP_MAX_PORTS USHRT_MAX
#define DP_VPORT_HASH_BUCKETS 1024
#define DP_MAX_PORTS 1024
#define SAMPLE_ACTION_DEPTH 3
/**
......@@ -58,11 +59,10 @@ struct dp_stats_percpu {
* @list_node: Element in global 'dps' list.
* @n_flows: Number of flows currently in flow table.
* @table: Current flow table. Protected by genl_lock and RCU.
* @ports: Map from port number to &struct vport. %OVSP_LOCAL port
* always exists, other ports may be %NULL. Protected by RTNL and RCU.
* @port_list: List of all ports in @ports in arbitrary order. RTNL required
* to iterate or modify.
* @ports: Hash table for ports. %OVSP_LOCAL port always exists. Protected by
* RTNL and RCU.
* @stats_percpu: Per-CPU datapath statistics.
* @net: Reference to net namespace.
*
* Context: See the comment on locking at the top of datapath.c for additional
* locking information.
......@@ -75,13 +75,37 @@ struct datapath {
struct flow_table __rcu *table;
/* Switch ports. */
struct vport __rcu *ports[DP_MAX_PORTS];
struct list_head port_list;
struct hlist_head *ports;
/* Stats. */
struct dp_stats_percpu __percpu *stats_percpu;
#ifdef CONFIG_NET_NS
/* Network namespace ref. */
struct net *net;
#endif
};
struct vport *ovs_lookup_vport(const struct datapath *dp, u16 port_no);
static inline struct vport *ovs_vport_rcu(const struct datapath *dp, int port_no)
{
WARN_ON_ONCE(!rcu_read_lock_held());
return ovs_lookup_vport(dp, port_no);
}
static inline struct vport *ovs_vport_rtnl_rcu(const struct datapath *dp, int port_no)
{
WARN_ON_ONCE(!rcu_read_lock_held() && !rtnl_is_locked());
return ovs_lookup_vport(dp, port_no);
}
static inline struct vport *ovs_vport_rtnl(const struct datapath *dp, int port_no)
{
ASSERT_RTNL();
return ovs_lookup_vport(dp, port_no);
}
/**
* struct ovs_skb_cb - OVS data in skb CB
* @flow: The flow associated with this packet. May be %NULL if no flow.
......@@ -108,6 +132,16 @@ struct dp_upcall_info {
u32 pid;
};
static inline struct net *ovs_dp_get_net(struct datapath *dp)
{
return read_pnet(&dp->net);
}
static inline void ovs_dp_set_net(struct datapath *dp, struct net *net)
{
write_pnet(&dp->net, net);
}
extern struct notifier_block ovs_dp_device_notifier;
extern struct genl_multicast_group ovs_dp_vport_multicast_group;
......
......@@ -41,19 +41,21 @@ static int dp_device_event(struct notifier_block *unused, unsigned long event,
case NETDEV_UNREGISTER:
if (!ovs_is_internal_dev(dev)) {
struct sk_buff *notify;
struct datapath *dp = vport->dp;
notify = ovs_vport_cmd_build_info(vport, 0, 0,
OVS_VPORT_CMD_DEL);
ovs_dp_detach_port(vport);
if (IS_ERR(notify)) {
netlink_set_err(init_net.genl_sock, 0,
netlink_set_err(ovs_dp_get_net(dp)->genl_sock, 0,
ovs_dp_vport_multicast_group.id,
PTR_ERR(notify));
break;
}
genlmsg_multicast(notify, 0, ovs_dp_vport_multicast_group.id,
GFP_KERNEL);
genlmsg_multicast_netns(ovs_dp_get_net(dp), notify, 0,
ovs_dp_vport_multicast_group.id,
GFP_KERNEL);
}
break;
}
......
......@@ -203,10 +203,7 @@ struct sw_flow_actions *ovs_flow_actions_alloc(const struct nlattr *actions)
int actions_len = nla_len(actions);
struct sw_flow_actions *sfa;
/* At least DP_MAX_PORTS actions are required to be able to flood a
* packet to every port. Factor of 2 allows for setting VLAN tags,
* etc. */
if (actions_len > 2 * DP_MAX_PORTS * nla_total_size(4))
if (actions_len > MAX_ACTIONS_BUFSIZE)
return ERR_PTR(-EINVAL);
sfa = kmalloc(sizeof(*sfa) + actions_len, GFP_KERNEL);
......@@ -992,7 +989,7 @@ int ovs_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_lenp,
swkey->phy.in_port = in_port;
attrs &= ~(1 << OVS_KEY_ATTR_IN_PORT);
} else {
swkey->phy.in_port = USHRT_MAX;
swkey->phy.in_port = DP_MAX_PORTS;
}
/* Data attributes. */
......@@ -1135,7 +1132,7 @@ int ovs_flow_metadata_from_nlattrs(u32 *priority, u16 *in_port,
const struct nlattr *nla;
int rem;
*in_port = USHRT_MAX;
*in_port = DP_MAX_PORTS;
*priority = 0;
nla_for_each_nested(nla, attr, rem) {
......@@ -1172,7 +1169,7 @@ int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb)
nla_put_u32(skb, OVS_KEY_ATTR_PRIORITY, swkey->phy.priority))
goto nla_put_failure;
if (swkey->phy.in_port != USHRT_MAX &&
if (swkey->phy.in_port != DP_MAX_PORTS &&
nla_put_u32(skb, OVS_KEY_ATTR_IN_PORT, swkey->phy.in_port))
goto nla_put_failure;
......
......@@ -43,7 +43,7 @@ struct sw_flow_actions {
struct sw_flow_key {
struct {
u32 priority; /* Packet QoS priority. */
u16 in_port; /* Input switch port (or USHRT_MAX). */
u16 in_port; /* Input switch port (or DP_MAX_PORTS). */
} phy;
struct {
u8 src[ETH_ALEN]; /* Ethernet source address. */
......@@ -161,6 +161,7 @@ int ovs_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_lenp,
int ovs_flow_metadata_from_nlattrs(u32 *priority, u16 *in_port,
const struct nlattr *);
#define MAX_ACTIONS_BUFSIZE (16 * 1024)
#define TBL_MIN_BUCKETS 1024
struct flow_table {
......
......@@ -144,7 +144,7 @@ static void do_setup(struct net_device *netdev)
netdev->tx_queue_len = 0;
netdev->features = NETIF_F_LLTX | NETIF_F_SG | NETIF_F_FRAGLIST |
NETIF_F_HIGHDMA | NETIF_F_HW_CSUM | NETIF_F_TSO;
NETIF_F_HIGHDMA | NETIF_F_HW_CSUM | NETIF_F_TSO;
netdev->vlan_features = netdev->features;
netdev->features |= NETIF_F_HW_VLAN_TX;
......@@ -175,9 +175,14 @@ static struct vport *internal_dev_create(const struct vport_parms *parms)
goto error_free_vport;
}
dev_net_set(netdev_vport->dev, ovs_dp_get_net(vport->dp));
internal_dev = internal_dev_priv(netdev_vport->dev);
internal_dev->vport = vport;
/* Restrict bridge port to current netns. */
if (vport->port_no == OVSP_LOCAL)
netdev_vport->dev->features |= NETIF_F_NETNS_LOCAL;
err = register_netdevice(netdev_vport->dev);
if (err)
goto error_free_netdev;
......
......@@ -83,7 +83,7 @@ static struct vport *netdev_create(const struct vport_parms *parms)
netdev_vport = netdev_vport_priv(vport);
netdev_vport->dev = dev_get_by_name(&init_net, parms->name);
netdev_vport->dev = dev_get_by_name(ovs_dp_get_net(vport->dp), parms->name);
if (!netdev_vport->dev) {
err = -ENODEV;
goto error_free_vport;
......
......@@ -16,10 +16,10 @@
* 02110-1301, USA
*/
#include <linux/dcache.h>
#include <linux/etherdevice.h>
#include <linux/if.h>
#include <linux/if_vlan.h>
#include <linux/jhash.h>
#include <linux/kernel.h>
#include <linux/list.h>
#include <linux/mutex.h>
......@@ -27,7 +27,9 @@
#include <linux/rcupdate.h>
#include <linux/rtnetlink.h>
#include <linux/compat.h>
#include <net/net_namespace.h>
#include "datapath.h"
#include "vport.h"
#include "vport-internal_dev.h"
......@@ -67,9 +69,9 @@ void ovs_vport_exit(void)
kfree(dev_table);
}
static struct hlist_head *hash_bucket(const char *name)
static struct hlist_head *hash_bucket(struct net *net, const char *name)
{
unsigned int hash = full_name_hash(name, strlen(name));
unsigned int hash = jhash(name, strlen(name), (unsigned long) net);
return &dev_table[hash & (VPORT_HASH_BUCKETS - 1)];
}
......@@ -80,14 +82,15 @@ static struct hlist_head *hash_bucket(const char *name)
*
* Must be called with RTNL or RCU read lock.
*/
struct vport *ovs_vport_locate(const char *name)
struct vport *ovs_vport_locate(struct net *net, const char *name)
{
struct hlist_head *bucket = hash_bucket(name);
struct hlist_head *bucket = hash_bucket(net, name);
struct vport *vport;
struct hlist_node *node;
hlist_for_each_entry_rcu(vport, node, bucket, hash_node)
if (!strcmp(name, vport->ops->get_name(vport)))
if (!strcmp(name, vport->ops->get_name(vport)) &&
net_eq(ovs_dp_get_net(vport->dp), net))
return vport;
return NULL;
......@@ -124,6 +127,7 @@ struct vport *ovs_vport_alloc(int priv_size, const struct vport_ops *ops,
vport->port_no = parms->port_no;
vport->upcall_pid = parms->upcall_pid;
vport->ops = ops;
INIT_HLIST_NODE(&vport->dp_hash_node);
vport->percpu_stats = alloc_percpu(struct vport_percpu_stats);
if (!vport->percpu_stats) {
......@@ -170,14 +174,17 @@ struct vport *ovs_vport_add(const struct vport_parms *parms)
for (i = 0; i < ARRAY_SIZE(vport_ops_list); i++) {
if (vport_ops_list[i]->type == parms->type) {
struct hlist_head *bucket;
vport = vport_ops_list[i]->create(parms);
if (IS_ERR(vport)) {
err = PTR_ERR(vport);
goto out;
}
hlist_add_head_rcu(&vport->hash_node,
hash_bucket(vport->ops->get_name(vport)));
bucket = hash_bucket(ovs_dp_get_net(vport->dp),
vport->ops->get_name(vport));
hlist_add_head_rcu(&vport->hash_node, bucket);
return vport;
}
}
......
......@@ -20,6 +20,7 @@
#define VPORT_H 1
#include <linux/list.h>
#include <linux/netlink.h>
#include <linux/openvswitch.h>
#include <linux/skbuff.h>
#include <linux/spinlock.h>
......@@ -38,7 +39,7 @@ void ovs_vport_exit(void);
struct vport *ovs_vport_add(const struct vport_parms *);
void ovs_vport_del(struct vport *);
struct vport *ovs_vport_locate(const char *name);
struct vport *ovs_vport_locate(struct net *net, const char *name);
void ovs_vport_get_stats(struct vport *, struct ovs_vport_stats *);
......@@ -69,10 +70,10 @@ struct vport_err_stats {
* @rcu: RCU callback head for deferred destruction.
* @port_no: Index into @dp's @ports array.
* @dp: Datapath to which this port belongs.
* @node: Element in @dp's @port_list.
* @upcall_pid: The Netlink port to use for packets received on this port that
* miss the flow table.
* @hash_node: Element in @dev_table hash table in vport.c.
* @dp_hash_node: Element in @datapath->ports hash table in datapath.c.
* @ops: Class structure.
* @percpu_stats: Points to per-CPU statistics used and maintained by vport
* @stats_lock: Protects @err_stats;
......@@ -82,10 +83,10 @@ struct vport {
struct rcu_head rcu;
u16 port_no;
struct datapath *dp;
struct list_head node;
u32 upcall_pid;
struct hlist_node hash_node;
struct hlist_node dp_hash_node;
const struct vport_ops *ops;
struct vport_percpu_stats __percpu *percpu_stats;
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment