* [GIT net-next] Open vSwitch
@ 2012-09-04 19:14 Jesse Gross
[not found] ` <1346786049-3100-1-git-send-email-jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
0 siblings, 1 reply; 4+ messages in thread
From: Jesse Gross @ 2012-09-04 19:14 UTC (permalink / raw)
To: David Miller; +Cc: dev-yBygre7rU0TnMu66kgdUjQ, netdev-u79uwXL29TY76Z2rM5mHXA
Two feature additions to Open vSwitch for net-next/3.7.
The following changes since commit 0d7614f09c1ebdbaa1599a5aba7593f147bf96ee:
Linux 3.6-rc1 (2012-08-02 16:38:10 -0700)
are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git master
for you to fetch changes up to 15eac2a74277bc7de68a7c2a64a7c91b4b6f5961:
openvswitch: Increase maximum number of datapath ports. (2012-09-03 19:20:49 -0700)
----------------------------------------------------------------
Pravin B Shelar (2):
openvswitch: Add support for network namespaces.
openvswitch: Increase maximum number of datapath ports.
net/openvswitch/actions.c | 2 +-
net/openvswitch/datapath.c | 375 +++++++++++++++++++++-------------
net/openvswitch/datapath.h | 50 ++++-
net/openvswitch/dp_notify.c | 8 +-
net/openvswitch/flow.c | 11 +-
net/openvswitch/flow.h | 3 +-
net/openvswitch/vport-internal_dev.c | 7 +-
net/openvswitch/vport-netdev.c | 2 +-
net/openvswitch/vport.c | 23 ++-
net/openvswitch/vport.h | 7 +-
10 files changed, 317 insertions(+), 171 deletions(-)
^ permalink raw reply [flat|nested] 4+ messages in thread
* [PATCH net-next 1/2] openvswitch: Add support for network namespaces.
[not found] ` <1346786049-3100-1-git-send-email-jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
@ 2012-09-04 19:14 ` Jesse Gross
2012-09-04 19:14 ` [PATCH net-next 2/2] openvswitch: Increase maximum number of datapath ports Jesse Gross
2012-09-04 19:26 ` [GIT net-next] Open vSwitch David Miller
2 siblings, 0 replies; 4+ messages in thread
From: Jesse Gross @ 2012-09-04 19:14 UTC (permalink / raw)
To: David Miller; +Cc: dev-yBygre7rU0TnMu66kgdUjQ, netdev-u79uwXL29TY76Z2rM5mHXA
From: Pravin B Shelar <pshelar-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
Following patch adds support for network namespace to openvswitch.
Since it must release devices when namespaces are destroyed, a
side effect of this patch is that the module no longer keeps a
refcount but instead cleans up any state when it is unloaded.
Signed-off-by: Pravin B Shelar <pshelar-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
net/openvswitch/datapath.c | 269 ++++++++++++++++++++--------------
net/openvswitch/datapath.h | 19 ++-
net/openvswitch/dp_notify.c | 8 +-
net/openvswitch/vport-internal_dev.c | 7 +-
net/openvswitch/vport-netdev.c | 2 +-
net/openvswitch/vport.c | 22 ++-
net/openvswitch/vport.h | 3 +-
7 files changed, 207 insertions(+), 123 deletions(-)
diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index d8277d2..cad39fc 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -49,12 +49,29 @@
#include <linux/dmi.h>
#include <linux/workqueue.h>
#include <net/genetlink.h>
+#include <net/net_namespace.h>
+#include <net/netns/generic.h>
#include "datapath.h"
#include "flow.h"
#include "vport-internal_dev.h"
/**
+ * struct ovs_net - Per net-namespace data for ovs.
+ * @dps: List of datapaths to enable dumping them all out.
+ * Protected by genl_mutex.
+ */
+struct ovs_net {
+ struct list_head dps;
+};
+
+static int ovs_net_id __read_mostly;
+
+#define REHASH_FLOW_INTERVAL (10 * 60 * HZ)
+static void rehash_flow_table(struct work_struct *work);
+static DECLARE_DELAYED_WORK(rehash_flow_wq, rehash_flow_table);
+
+/**
* DOC: Locking:
*
* Writes to device state (add/remove datapath, port, set operations on vports,
@@ -71,29 +88,21 @@
* each other.
*/
-/* Global list of datapaths to enable dumping them all out.
- * Protected by genl_mutex.
- */
-static LIST_HEAD(dps);
-
-#define REHASH_FLOW_INTERVAL (10 * 60 * HZ)
-static void rehash_flow_table(struct work_struct *work);
-static DECLARE_DELAYED_WORK(rehash_flow_wq, rehash_flow_table);
-
static struct vport *new_vport(const struct vport_parms *);
-static int queue_gso_packets(int dp_ifindex, struct sk_buff *,
+static int queue_gso_packets(struct net *, int dp_ifindex, struct sk_buff *,
const struct dp_upcall_info *);
-static int queue_userspace_packet(int dp_ifindex, struct sk_buff *,
+static int queue_userspace_packet(struct net *, int dp_ifindex,
+ struct sk_buff *,
const struct dp_upcall_info *);
/* Must be called with rcu_read_lock, genl_mutex, or RTNL lock. */
-static struct datapath *get_dp(int dp_ifindex)
+static struct datapath *get_dp(struct net *net, int dp_ifindex)
{
struct datapath *dp = NULL;
struct net_device *dev;
rcu_read_lock();
- dev = dev_get_by_index_rcu(&init_net, dp_ifindex);
+ dev = dev_get_by_index_rcu(net, dp_ifindex);
if (dev) {
struct vport *vport = ovs_internal_dev_get_vport(dev);
if (vport)
@@ -135,6 +144,7 @@ static void destroy_dp_rcu(struct rcu_head *rcu)
ovs_flow_tbl_destroy((__force struct flow_table *)dp->table);
free_percpu(dp->stats_percpu);
+ release_net(ovs_dp_get_net(dp));
kfree(dp);
}
@@ -220,11 +230,12 @@ static struct genl_family dp_packet_genl_family = {
.hdrsize = sizeof(struct ovs_header),
.name = OVS_PACKET_FAMILY,
.version = OVS_PACKET_VERSION,
- .maxattr = OVS_PACKET_ATTR_MAX
+ .maxattr = OVS_PACKET_ATTR_MAX,
+ .netnsok = true
};
int ovs_dp_upcall(struct datapath *dp, struct sk_buff *skb,
- const struct dp_upcall_info *upcall_info)
+ const struct dp_upcall_info *upcall_info)
{
struct dp_stats_percpu *stats;
int dp_ifindex;
@@ -242,9 +253,9 @@ int ovs_dp_upcall(struct datapath *dp, struct sk_buff *skb,
}
if (!skb_is_gso(skb))
- err = queue_userspace_packet(dp_ifindex, skb, upcall_info);
+ err = queue_userspace_packet(ovs_dp_get_net(dp), dp_ifindex, skb, upcall_info);
else
- err = queue_gso_packets(dp_ifindex, skb, upcall_info);
+ err = queue_gso_packets(ovs_dp_get_net(dp), dp_ifindex, skb, upcall_info);
if (err)
goto err;
@@ -260,7 +271,8 @@ err:
return err;
}
-static int queue_gso_packets(int dp_ifindex, struct sk_buff *skb,
+static int queue_gso_packets(struct net *net, int dp_ifindex,
+ struct sk_buff *skb,
const struct dp_upcall_info *upcall_info)
{
unsigned short gso_type = skb_shinfo(skb)->gso_type;
@@ -276,7 +288,7 @@ static int queue_gso_packets(int dp_ifindex, struct sk_buff *skb,
/* Queue all of the segments. */
skb = segs;
do {
- err = queue_userspace_packet(dp_ifindex, skb, upcall_info);
+ err = queue_userspace_packet(net, dp_ifindex, skb, upcall_info);
if (err)
break;
@@ -306,7 +318,8 @@ static int queue_gso_packets(int dp_ifindex, struct sk_buff *skb,
return err;
}
-static int queue_userspace_packet(int dp_ifindex, struct sk_buff *skb,
+static int queue_userspace_packet(struct net *net, int dp_ifindex,
+ struct sk_buff *skb,
const struct dp_upcall_info *upcall_info)
{
struct ovs_header *upcall;
@@ -362,7 +375,7 @@ static int queue_userspace_packet(int dp_ifindex, struct sk_buff *skb,
skb_copy_and_csum_dev(skb, nla_data(nla));
- err = genlmsg_unicast(&init_net, user_skb, upcall_info->pid);
+ err = genlmsg_unicast(net, user_skb, upcall_info->pid);
out:
kfree_skb(nskb);
@@ -370,15 +383,10 @@ out:
}
/* Called with genl_mutex. */
-static int flush_flows(int dp_ifindex)
+static int flush_flows(struct datapath *dp)
{
struct flow_table *old_table;
struct flow_table *new_table;
- struct datapath *dp;
-
- dp = get_dp(dp_ifindex);
- if (!dp)
- return -ENODEV;
old_table = genl_dereference(dp->table);
new_table = ovs_flow_tbl_alloc(TBL_MIN_BUCKETS);
@@ -668,7 +676,7 @@ static int ovs_packet_cmd_execute(struct sk_buff *skb, struct genl_info *info)
packet->priority = flow->key.phy.priority;
rcu_read_lock();
- dp = get_dp(ovs_header->dp_ifindex);
+ dp = get_dp(sock_net(skb->sk), ovs_header->dp_ifindex);
err = -ENODEV;
if (!dp)
goto err_unlock;
@@ -742,7 +750,8 @@ static struct genl_family dp_flow_genl_family = {
.hdrsize = sizeof(struct ovs_header),
.name = OVS_FLOW_FAMILY,
.version = OVS_FLOW_VERSION,
- .maxattr = OVS_FLOW_ATTR_MAX
+ .maxattr = OVS_FLOW_ATTR_MAX,
+ .netnsok = true
};
static struct genl_multicast_group ovs_dp_flow_multicast_group = {
@@ -894,7 +903,7 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
goto error;
}
- dp = get_dp(ovs_header->dp_ifindex);
+ dp = get_dp(sock_net(skb->sk), ovs_header->dp_ifindex);
error = -ENODEV;
if (!dp)
goto error;
@@ -995,7 +1004,7 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
ovs_dp_flow_multicast_group.id, info->nlhdr,
GFP_KERNEL);
else
- netlink_set_err(init_net.genl_sock, 0,
+ netlink_set_err(sock_net(skb->sk)->genl_sock, 0,
ovs_dp_flow_multicast_group.id, PTR_ERR(reply));
return 0;
@@ -1023,7 +1032,7 @@ static int ovs_flow_cmd_get(struct sk_buff *skb, struct genl_info *info)
if (err)
return err;
- dp = get_dp(ovs_header->dp_ifindex);
+ dp = get_dp(sock_net(skb->sk), ovs_header->dp_ifindex);
if (!dp)
return -ENODEV;
@@ -1052,16 +1061,17 @@ static int ovs_flow_cmd_del(struct sk_buff *skb, struct genl_info *info)
int err;
int key_len;
+ dp = get_dp(sock_net(skb->sk), ovs_header->dp_ifindex);
+ if (!dp)
+ return -ENODEV;
+
if (!a[OVS_FLOW_ATTR_KEY])
- return flush_flows(ovs_header->dp_ifindex);
+ return flush_flows(dp);
+
err = ovs_flow_from_nlattrs(&key, &key_len, a[OVS_FLOW_ATTR_KEY]);
if (err)
return err;
- dp = get_dp(ovs_header->dp_ifindex);
- if (!dp)
- return -ENODEV;
-
table = genl_dereference(dp->table);
flow = ovs_flow_tbl_lookup(table, &key, key_len);
if (!flow)
@@ -1090,7 +1100,7 @@ static int ovs_flow_cmd_dump(struct sk_buff *skb, struct netlink_callback *cb)
struct datapath *dp;
struct flow_table *table;
- dp = get_dp(ovs_header->dp_ifindex);
+ dp = get_dp(sock_net(skb->sk), ovs_header->dp_ifindex);
if (!dp)
return -ENODEV;
@@ -1152,7 +1162,8 @@ static struct genl_family dp_datapath_genl_family = {
.hdrsize = sizeof(struct ovs_header),
.name = OVS_DATAPATH_FAMILY,
.version = OVS_DATAPATH_VERSION,
- .maxattr = OVS_DP_ATTR_MAX
+ .maxattr = OVS_DP_ATTR_MAX,
+ .netnsok = true
};
static struct genl_multicast_group ovs_dp_datapath_multicast_group = {
@@ -1210,18 +1221,19 @@ static struct sk_buff *ovs_dp_cmd_build_info(struct datapath *dp, u32 pid,
}
/* Called with genl_mutex and optionally with RTNL lock also. */
-static struct datapath *lookup_datapath(struct ovs_header *ovs_header,
+static struct datapath *lookup_datapath(struct net *net,
+ struct ovs_header *ovs_header,
struct nlattr *a[OVS_DP_ATTR_MAX + 1])
{
struct datapath *dp;
if (!a[OVS_DP_ATTR_NAME])
- dp = get_dp(ovs_header->dp_ifindex);
+ dp = get_dp(net, ovs_header->dp_ifindex);
else {
struct vport *vport;
rcu_read_lock();
- vport = ovs_vport_locate(nla_data(a[OVS_DP_ATTR_NAME]));
+ vport = ovs_vport_locate(net, nla_data(a[OVS_DP_ATTR_NAME]));
dp = vport && vport->port_no == OVSP_LOCAL ? vport->dp : NULL;
rcu_read_unlock();
}
@@ -1235,6 +1247,7 @@ static int ovs_dp_cmd_new(struct sk_buff *skb, struct genl_info *info)
struct sk_buff *reply;
struct datapath *dp;
struct vport *vport;
+ struct ovs_net *ovs_net;
int err;
err = -EINVAL;
@@ -1242,15 +1255,14 @@ static int ovs_dp_cmd_new(struct sk_buff *skb, struct genl_info *info)
goto err;
rtnl_lock();
- err = -ENODEV;
- if (!try_module_get(THIS_MODULE))
- goto err_unlock_rtnl;
err = -ENOMEM;
dp = kzalloc(sizeof(*dp), GFP_KERNEL);
if (dp == NULL)
- goto err_put_module;
+ goto err_unlock_rtnl;
+
INIT_LIST_HEAD(&dp->port_list);
+ ovs_dp_set_net(dp, hold_net(sock_net(skb->sk)));
/* Allocate table. */
err = -ENOMEM;
@@ -1287,7 +1299,8 @@ static int ovs_dp_cmd_new(struct sk_buff *skb, struct genl_info *info)
if (IS_ERR(reply))
goto err_destroy_local_port;
- list_add_tail(&dp->list_node, &dps);
+ ovs_net = net_generic(ovs_dp_get_net(dp), ovs_net_id);
+ list_add_tail(&dp->list_node, &ovs_net->dps);
rtnl_unlock();
genl_notify(reply, genl_info_net(info), info->snd_pid,
@@ -1302,34 +1315,20 @@ err_destroy_percpu:
err_destroy_table:
ovs_flow_tbl_destroy(genl_dereference(dp->table));
err_free_dp:
+ release_net(ovs_dp_get_net(dp));
kfree(dp);
-err_put_module:
- module_put(THIS_MODULE);
err_unlock_rtnl:
rtnl_unlock();
err:
return err;
}
-static int ovs_dp_cmd_del(struct sk_buff *skb, struct genl_info *info)
+/* Called with genl_mutex. */
+static void __dp_destroy(struct datapath *dp)
{
struct vport *vport, *next_vport;
- struct sk_buff *reply;
- struct datapath *dp;
- int err;
rtnl_lock();
- dp = lookup_datapath(info->userhdr, info->attrs);
- err = PTR_ERR(dp);
- if (IS_ERR(dp))
- goto exit_unlock;
-
- reply = ovs_dp_cmd_build_info(dp, info->snd_pid,
- info->snd_seq, OVS_DP_CMD_DEL);
- err = PTR_ERR(reply);
- if (IS_ERR(reply))
- goto exit_unlock;
-
list_for_each_entry_safe(vport, next_vport, &dp->port_list, node)
if (vport->port_no != OVSP_LOCAL)
ovs_dp_detach_port(vport);
@@ -1345,17 +1344,32 @@ static int ovs_dp_cmd_del(struct sk_buff *skb, struct genl_info *info)
rtnl_unlock();
call_rcu(&dp->rcu, destroy_dp_rcu);
- module_put(THIS_MODULE);
+}
+
+static int ovs_dp_cmd_del(struct sk_buff *skb, struct genl_info *info)
+{
+ struct sk_buff *reply;
+ struct datapath *dp;
+ int err;
+
+ dp = lookup_datapath(sock_net(skb->sk), info->userhdr, info->attrs);
+ err = PTR_ERR(dp);
+ if (IS_ERR(dp))
+ return err;
+
+ reply = ovs_dp_cmd_build_info(dp, info->snd_pid,
+ info->snd_seq, OVS_DP_CMD_DEL);
+ err = PTR_ERR(reply);
+ if (IS_ERR(reply))
+ return err;
+
+ __dp_destroy(dp);
genl_notify(reply, genl_info_net(info), info->snd_pid,
ovs_dp_datapath_multicast_group.id, info->nlhdr,
GFP_KERNEL);
return 0;
-
-exit_unlock:
- rtnl_unlock();
- return err;
}
static int ovs_dp_cmd_set(struct sk_buff *skb, struct genl_info *info)
@@ -1364,7 +1378,7 @@ static int ovs_dp_cmd_set(struct sk_buff *skb, struct genl_info *info)
struct datapath *dp;
int err;
- dp = lookup_datapath(info->userhdr, info->attrs);
+ dp = lookup_datapath(sock_net(skb->sk), info->userhdr, info->attrs);
if (IS_ERR(dp))
return PTR_ERR(dp);
@@ -1372,7 +1386,7 @@ static int ovs_dp_cmd_set(struct sk_buff *skb, struct genl_info *info)
info->snd_seq, OVS_DP_CMD_NEW);
if (IS_ERR(reply)) {
err = PTR_ERR(reply);
- netlink_set_err(init_net.genl_sock, 0,
+ netlink_set_err(sock_net(skb->sk)->genl_sock, 0,
ovs_dp_datapath_multicast_group.id, err);
return 0;
}
@@ -1389,7 +1403,7 @@ static int ovs_dp_cmd_get(struct sk_buff *skb, struct genl_info *info)
struct sk_buff *reply;
struct datapath *dp;
- dp = lookup_datapath(info->userhdr, info->attrs);
+ dp = lookup_datapath(sock_net(skb->sk), info->userhdr, info->attrs);
if (IS_ERR(dp))
return PTR_ERR(dp);
@@ -1403,11 +1417,12 @@ static int ovs_dp_cmd_get(struct sk_buff *skb, struct genl_info *info)
static int ovs_dp_cmd_dump(struct sk_buff *skb, struct netlink_callback *cb)
{
+ struct ovs_net *ovs_net = net_generic(sock_net(skb->sk), ovs_net_id);
struct datapath *dp;
int skip = cb->args[0];
int i = 0;
- list_for_each_entry(dp, &dps, list_node) {
+ list_for_each_entry(dp, &ovs_net->dps, list_node) {
if (i >= skip &&
ovs_dp_cmd_fill_info(dp, skb, NETLINK_CB(cb->skb).pid,
cb->nlh->nlmsg_seq, NLM_F_MULTI,
@@ -1459,7 +1474,8 @@ static struct genl_family dp_vport_genl_family = {
.hdrsize = sizeof(struct ovs_header),
.name = OVS_VPORT_FAMILY,
.version = OVS_VPORT_VERSION,
- .maxattr = OVS_VPORT_ATTR_MAX
+ .maxattr = OVS_VPORT_ATTR_MAX,
+ .netnsok = true
};
struct genl_multicast_group ovs_dp_vport_multicast_group = {
@@ -1525,14 +1541,15 @@ struct sk_buff *ovs_vport_cmd_build_info(struct vport *vport, u32 pid,
}
/* Called with RTNL lock or RCU read lock. */
-static struct vport *lookup_vport(struct ovs_header *ovs_header,
+static struct vport *lookup_vport(struct net *net,
+ struct ovs_header *ovs_header,
struct nlattr *a[OVS_VPORT_ATTR_MAX + 1])
{
struct datapath *dp;
struct vport *vport;
if (a[OVS_VPORT_ATTR_NAME]) {
- vport = ovs_vport_locate(nla_data(a[OVS_VPORT_ATTR_NAME]));
+ vport = ovs_vport_locate(net, nla_data(a[OVS_VPORT_ATTR_NAME]));
if (!vport)
return ERR_PTR(-ENODEV);
if (ovs_header->dp_ifindex &&
@@ -1545,7 +1562,7 @@ static struct vport *lookup_vport(struct ovs_header *ovs_header,
if (port_no >= DP_MAX_PORTS)
return ERR_PTR(-EFBIG);
- dp = get_dp(ovs_header->dp_ifindex);
+ dp = get_dp(net, ovs_header->dp_ifindex);
if (!dp)
return ERR_PTR(-ENODEV);
@@ -1574,7 +1591,7 @@ static int ovs_vport_cmd_new(struct sk_buff *skb, struct genl_info *info)
goto exit;
rtnl_lock();
- dp = get_dp(ovs_header->dp_ifindex);
+ dp = get_dp(sock_net(skb->sk), ovs_header->dp_ifindex);
err = -ENODEV;
if (!dp)
goto exit_unlock;
@@ -1638,7 +1655,7 @@ static int ovs_vport_cmd_set(struct sk_buff *skb, struct genl_info *info)
int err;
rtnl_lock();
- vport = lookup_vport(info->userhdr, a);
+ vport = lookup_vport(sock_net(skb->sk), info->userhdr, a);
err = PTR_ERR(vport);
if (IS_ERR(vport))
goto exit_unlock;
@@ -1658,7 +1675,7 @@ static int ovs_vport_cmd_set(struct sk_buff *skb, struct genl_info *info)
reply = ovs_vport_cmd_build_info(vport, info->snd_pid, info->snd_seq,
OVS_VPORT_CMD_NEW);
if (IS_ERR(reply)) {
- netlink_set_err(init_net.genl_sock, 0,
+ netlink_set_err(sock_net(skb->sk)->genl_sock, 0,
ovs_dp_vport_multicast_group.id, PTR_ERR(reply));
goto exit_unlock;
}
@@ -1679,7 +1696,7 @@ static int ovs_vport_cmd_del(struct sk_buff *skb, struct genl_info *info)
int err;
rtnl_lock();
- vport = lookup_vport(info->userhdr, a);
+ vport = lookup_vport(sock_net(skb->sk), info->userhdr, a);
err = PTR_ERR(vport);
if (IS_ERR(vport))
goto exit_unlock;
@@ -1714,7 +1731,7 @@ static int ovs_vport_cmd_get(struct sk_buff *skb, struct genl_info *info)
int err;
rcu_read_lock();
- vport = lookup_vport(ovs_header, a);
+ vport = lookup_vport(sock_net(skb->sk), ovs_header, a);
err = PTR_ERR(vport);
if (IS_ERR(vport))
goto exit_unlock;
@@ -1741,7 +1758,7 @@ static int ovs_vport_cmd_dump(struct sk_buff *skb, struct netlink_callback *cb)
u32 port_no;
int retval;
- dp = get_dp(ovs_header->dp_ifindex);
+ dp = get_dp(sock_net(skb->sk), ovs_header->dp_ifindex);
if (!dp)
return -ENODEV;
@@ -1766,28 +1783,6 @@ static int ovs_vport_cmd_dump(struct sk_buff *skb, struct netlink_callback *cb)
return retval;
}
-static void rehash_flow_table(struct work_struct *work)
-{
- struct datapath *dp;
-
- genl_lock();
-
- list_for_each_entry(dp, &dps, list_node) {
- struct flow_table *old_table = genl_dereference(dp->table);
- struct flow_table *new_table;
-
- new_table = ovs_flow_tbl_rehash(old_table);
- if (!IS_ERR(new_table)) {
- rcu_assign_pointer(dp->table, new_table);
- ovs_flow_tbl_deferred_destroy(old_table);
- }
- }
-
- genl_unlock();
-
- schedule_delayed_work(&rehash_flow_wq, REHASH_FLOW_INTERVAL);
-}
-
static struct genl_ops dp_vport_genl_ops[] = {
{ .cmd = OVS_VPORT_CMD_NEW,
.flags = GENL_ADMIN_PERM, /* Requires CAP_NET_ADMIN privilege. */
@@ -1872,6 +1867,59 @@ error:
return err;
}
+static void rehash_flow_table(struct work_struct *work)
+{
+ struct datapath *dp;
+ struct net *net;
+
+ genl_lock();
+ rtnl_lock();
+ for_each_net(net) {
+ struct ovs_net *ovs_net = net_generic(net, ovs_net_id);
+
+ list_for_each_entry(dp, &ovs_net->dps, list_node) {
+ struct flow_table *old_table = genl_dereference(dp->table);
+ struct flow_table *new_table;
+
+ new_table = ovs_flow_tbl_rehash(old_table);
+ if (!IS_ERR(new_table)) {
+ rcu_assign_pointer(dp->table, new_table);
+ ovs_flow_tbl_deferred_destroy(old_table);
+ }
+ }
+ }
+ rtnl_unlock();
+ genl_unlock();
+
+ schedule_delayed_work(&rehash_flow_wq, REHASH_FLOW_INTERVAL);
+}
+
+static int __net_init ovs_init_net(struct net *net)
+{
+ struct ovs_net *ovs_net = net_generic(net, ovs_net_id);
+
+ INIT_LIST_HEAD(&ovs_net->dps);
+ return 0;
+}
+
+static void __net_exit ovs_exit_net(struct net *net)
+{
+ struct ovs_net *ovs_net = net_generic(net, ovs_net_id);
+ struct datapath *dp, *dp_next;
+
+ genl_lock();
+ list_for_each_entry_safe(dp, dp_next, &ovs_net->dps, list_node)
+ __dp_destroy(dp);
+ genl_unlock();
+}
+
+static struct pernet_operations ovs_net_ops = {
+ .init = ovs_init_net,
+ .exit = ovs_exit_net,
+ .id = &ovs_net_id,
+ .size = sizeof(struct ovs_net),
+};
+
static int __init dp_init(void)
{
struct sk_buff *dummy_skb;
@@ -1889,10 +1937,14 @@ static int __init dp_init(void)
if (err)
goto error_flow_exit;
- err = register_netdevice_notifier(&ovs_dp_device_notifier);
+ err = register_pernet_device(&ovs_net_ops);
if (err)
goto error_vport_exit;
+ err = register_netdevice_notifier(&ovs_dp_device_notifier);
+ if (err)
+ goto error_netns_exit;
+
err = dp_register_genl();
if (err < 0)
goto error_unreg_notifier;
@@ -1903,6 +1955,8 @@ static int __init dp_init(void)
error_unreg_notifier:
unregister_netdevice_notifier(&ovs_dp_device_notifier);
+error_netns_exit:
+ unregister_pernet_device(&ovs_net_ops);
error_vport_exit:
ovs_vport_exit();
error_flow_exit:
@@ -1914,9 +1968,10 @@ error:
static void dp_cleanup(void)
{
cancel_delayed_work_sync(&rehash_flow_wq);
- rcu_barrier();
dp_unregister_genl(ARRAY_SIZE(dp_genl_families));
unregister_netdevice_notifier(&ovs_dp_device_notifier);
+ unregister_pernet_device(&ovs_net_ops);
+ rcu_barrier();
ovs_vport_exit();
ovs_flow_exit();
}
diff --git a/net/openvswitch/datapath.h b/net/openvswitch/datapath.h
index c1105c1..771c11e 100644
--- a/net/openvswitch/datapath.h
+++ b/net/openvswitch/datapath.h
@@ -27,8 +27,7 @@
#include <linux/u64_stats_sync.h>
#include "flow.h"
-
-struct vport;
+#include "vport.h"
#define DP_MAX_PORTS 1024
#define SAMPLE_ACTION_DEPTH 3
@@ -63,6 +62,7 @@ struct dp_stats_percpu {
* @port_list: List of all ports in @ports in arbitrary order. RTNL required
* to iterate or modify.
* @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.
@@ -80,6 +80,11 @@ struct datapath {
/* Stats. */
struct dp_stats_percpu __percpu *stats_percpu;
+
+#ifdef CONFIG_NET_NS
+ /* Network namespace ref. */
+ struct net *net;
+#endif
};
/**
@@ -108,6 +113,16 @@ 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;
diff --git a/net/openvswitch/dp_notify.c b/net/openvswitch/dp_notify.c
index 36dcee8..5558350 100644
--- a/net/openvswitch/dp_notify.c
+++ b/net/openvswitch/dp_notify.c
@@ -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;
}
diff --git a/net/openvswitch/vport-internal_dev.c b/net/openvswitch/vport-internal_dev.c
index 4061b9e..5d460c3 100644
--- a/net/openvswitch/vport-internal_dev.c
+++ b/net/openvswitch/vport-internal_dev.c
@@ -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;
diff --git a/net/openvswitch/vport-netdev.c b/net/openvswitch/vport-netdev.c
index 6ea3551..3c1e58b 100644
--- a/net/openvswitch/vport-netdev.c
+++ b/net/openvswitch/vport-netdev.c
@@ -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;
diff --git a/net/openvswitch/vport.c b/net/openvswitch/vport.c
index 6140336..9873ace 100644
--- a/net/openvswitch/vport.c
+++ b/net/openvswitch/vport.c
@@ -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;
@@ -170,14 +173,17 @@ 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;
}
}
diff --git a/net/openvswitch/vport.h b/net/openvswitch/vport.h
index aac680c..97cef08 100644
--- a/net/openvswitch/vport.h
+++ b/net/openvswitch/vport.h
@@ -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 *);
--
1.7.9.5
^ permalink raw reply related [flat|nested] 4+ messages in thread
* [PATCH net-next 2/2] openvswitch: Increase maximum number of datapath ports.
[not found] ` <1346786049-3100-1-git-send-email-jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
2012-09-04 19:14 ` [PATCH net-next 1/2] openvswitch: Add support for network namespaces Jesse Gross
@ 2012-09-04 19:14 ` Jesse Gross
2012-09-04 19:26 ` [GIT net-next] Open vSwitch David Miller
2 siblings, 0 replies; 4+ messages in thread
From: Jesse Gross @ 2012-09-04 19:14 UTC (permalink / raw)
To: David Miller; +Cc: dev-yBygre7rU0TnMu66kgdUjQ, netdev-u79uwXL29TY76Z2rM5mHXA
From: Pravin B Shelar <pshelar-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
Use hash table to store ports of datapath. Allow 64K ports per switch.
Signed-off-by: Pravin B Shelar <pshelar-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
net/openvswitch/actions.c | 2 +-
net/openvswitch/datapath.c | 110 +++++++++++++++++++++++++++++++-------------
net/openvswitch/datapath.h | 33 ++++++++++---
net/openvswitch/flow.c | 11 ++---
net/openvswitch/flow.h | 3 +-
net/openvswitch/vport.c | 1 +
net/openvswitch/vport.h | 4 +-
7 files changed, 113 insertions(+), 51 deletions(-)
diff --git a/net/openvswitch/actions.c b/net/openvswitch/actions.c
index f3f96ba..0da6877 100644
--- a/net/openvswitch/actions.c
+++ b/net/openvswitch/actions.c
@@ -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;
diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index cad39fc..105a0b5 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -116,7 +116,7 @@ static struct datapath *get_dp(struct net *net, int dp_ifindex)
/* Must be called with rcu_read_lock or RTNL lock. */
const char *ovs_dp_name(const struct datapath *dp)
{
- struct vport *vport = rcu_dereference_rtnl(dp->ports[OVSP_LOCAL]);
+ struct vport *vport = ovs_vport_rtnl_rcu(dp, OVSP_LOCAL);
return vport->ops->get_name(vport);
}
@@ -127,7 +127,7 @@ static int get_dpifindex(struct datapath *dp)
rcu_read_lock();
- local = rcu_dereference(dp->ports[OVSP_LOCAL]);
+ local = ovs_vport_rcu(dp, OVSP_LOCAL);
if (local)
ifindex = local->ops->get_ifindex(local);
else
@@ -145,9 +145,30 @@ static void destroy_dp_rcu(struct rcu_head *rcu)
ovs_flow_tbl_destroy((__force struct flow_table *)dp->table);
free_percpu(dp->stats_percpu);
release_net(ovs_dp_get_net(dp));
+ kfree(dp->ports);
kfree(dp);
}
+static struct hlist_head *vport_hash_bucket(const struct datapath *dp,
+ u16 port_no)
+{
+ return &dp->ports[port_no & (DP_VPORT_HASH_BUCKETS - 1)];
+}
+
+struct vport *ovs_lookup_vport(const struct datapath *dp, u16 port_no)
+{
+ struct vport *vport;
+ struct hlist_node *n;
+ struct hlist_head *head;
+
+ head = vport_hash_bucket(dp, port_no);
+ hlist_for_each_entry_rcu(vport, n, head, dp_hash_node) {
+ if (vport->port_no == port_no)
+ return vport;
+ }
+ return NULL;
+}
+
/* Called with RTNL lock and genl_lock. */
static struct vport *new_vport(const struct vport_parms *parms)
{
@@ -156,9 +177,9 @@ static struct vport *new_vport(const struct vport_parms *parms)
vport = ovs_vport_add(parms);
if (!IS_ERR(vport)) {
struct datapath *dp = parms->dp;
+ struct hlist_head *head = vport_hash_bucket(dp, vport->port_no);
- rcu_assign_pointer(dp->ports[parms->port_no], vport);
- list_add(&vport->node, &dp->port_list);
+ hlist_add_head_rcu(&vport->dp_hash_node, head);
}
return vport;
@@ -170,8 +191,7 @@ void ovs_dp_detach_port(struct vport *p)
ASSERT_RTNL();
/* First drop references to device. */
- list_del(&p->node);
- rcu_assign_pointer(p->dp->ports[p->port_no], NULL);
+ hlist_del_rcu(&p->dp_hash_node);
/* Then destroy it. */
ovs_vport_del(p);
@@ -1248,7 +1268,7 @@ static int ovs_dp_cmd_new(struct sk_buff *skb, struct genl_info *info)
struct datapath *dp;
struct vport *vport;
struct ovs_net *ovs_net;
- int err;
+ int err, i;
err = -EINVAL;
if (!a[OVS_DP_ATTR_NAME] || !a[OVS_DP_ATTR_UPCALL_PID])
@@ -1261,7 +1281,6 @@ static int ovs_dp_cmd_new(struct sk_buff *skb, struct genl_info *info)
if (dp == NULL)
goto err_unlock_rtnl;
- INIT_LIST_HEAD(&dp->port_list);
ovs_dp_set_net(dp, hold_net(sock_net(skb->sk)));
/* Allocate table. */
@@ -1276,6 +1295,16 @@ static int ovs_dp_cmd_new(struct sk_buff *skb, struct genl_info *info)
goto err_destroy_table;
}
+ dp->ports = kmalloc(DP_VPORT_HASH_BUCKETS * sizeof(struct hlist_head),
+ GFP_KERNEL);
+ if (!dp->ports) {
+ err = -ENOMEM;
+ goto err_destroy_percpu;
+ }
+
+ for (i = 0; i < DP_VPORT_HASH_BUCKETS; i++)
+ INIT_HLIST_HEAD(&dp->ports[i]);
+
/* Set up our datapath device. */
parms.name = nla_data(a[OVS_DP_ATTR_NAME]);
parms.type = OVS_VPORT_TYPE_INTERNAL;
@@ -1290,7 +1319,7 @@ static int ovs_dp_cmd_new(struct sk_buff *skb, struct genl_info *info)
if (err == -EBUSY)
err = -EEXIST;
- goto err_destroy_percpu;
+ goto err_destroy_ports_array;
}
reply = ovs_dp_cmd_build_info(dp, info->snd_pid,
@@ -1309,7 +1338,9 @@ static int ovs_dp_cmd_new(struct sk_buff *skb, struct genl_info *info)
return 0;
err_destroy_local_port:
- ovs_dp_detach_port(rtnl_dereference(dp->ports[OVSP_LOCAL]));
+ ovs_dp_detach_port(ovs_vport_rtnl(dp, OVSP_LOCAL));
+err_destroy_ports_array:
+ kfree(dp->ports);
err_destroy_percpu:
free_percpu(dp->stats_percpu);
err_destroy_table:
@@ -1326,15 +1357,21 @@ err:
/* Called with genl_mutex. */
static void __dp_destroy(struct datapath *dp)
{
- struct vport *vport, *next_vport;
+ int i;
rtnl_lock();
- list_for_each_entry_safe(vport, next_vport, &dp->port_list, node)
- if (vport->port_no != OVSP_LOCAL)
- ovs_dp_detach_port(vport);
+
+ for (i = 0; i < DP_VPORT_HASH_BUCKETS; i++) {
+ struct vport *vport;
+ struct hlist_node *node, *n;
+
+ hlist_for_each_entry_safe(vport, node, n, &dp->ports[i], dp_hash_node)
+ if (vport->port_no != OVSP_LOCAL)
+ ovs_dp_detach_port(vport);
+ }
list_del(&dp->list_node);
- ovs_dp_detach_port(rtnl_dereference(dp->ports[OVSP_LOCAL]));
+ ovs_dp_detach_port(ovs_vport_rtnl(dp, OVSP_LOCAL));
/* rtnl_unlock() will wait until all the references to devices that
* are pending unregistration have been dropped. We do it here to
@@ -1566,7 +1603,7 @@ static struct vport *lookup_vport(struct net *net,
if (!dp)
return ERR_PTR(-ENODEV);
- vport = rcu_dereference_rtnl(dp->ports[port_no]);
+ vport = ovs_vport_rtnl_rcu(dp, port_no);
if (!vport)
return ERR_PTR(-ENOENT);
return vport;
@@ -1603,7 +1640,7 @@ static int ovs_vport_cmd_new(struct sk_buff *skb, struct genl_info *info)
if (port_no >= DP_MAX_PORTS)
goto exit_unlock;
- vport = rtnl_dereference(dp->ports[port_no]);
+ vport = ovs_vport_rtnl_rcu(dp, port_no);
err = -EBUSY;
if (vport)
goto exit_unlock;
@@ -1613,7 +1650,7 @@ static int ovs_vport_cmd_new(struct sk_buff *skb, struct genl_info *info)
err = -EFBIG;
goto exit_unlock;
}
- vport = rtnl_dereference(dp->ports[port_no]);
+ vport = ovs_vport_rtnl(dp, port_no);
if (!vport)
break;
}
@@ -1755,32 +1792,39 @@ static int ovs_vport_cmd_dump(struct sk_buff *skb, struct netlink_callback *cb)
{
struct ovs_header *ovs_header = genlmsg_data(nlmsg_data(cb->nlh));
struct datapath *dp;
- u32 port_no;
- int retval;
+ int bucket = cb->args[0], skip = cb->args[1];
+ int i, j = 0;
dp = get_dp(sock_net(skb->sk), ovs_header->dp_ifindex);
if (!dp)
return -ENODEV;
rcu_read_lock();
- for (port_no = cb->args[0]; port_no < DP_MAX_PORTS; port_no++) {
+ for (i = bucket; i < DP_VPORT_HASH_BUCKETS; i++) {
struct vport *vport;
-
- vport = rcu_dereference(dp->ports[port_no]);
- if (!vport)
- continue;
-
- if (ovs_vport_cmd_fill_info(vport, skb, NETLINK_CB(cb->skb).pid,
- cb->nlh->nlmsg_seq, NLM_F_MULTI,
- OVS_VPORT_CMD_NEW) < 0)
- break;
+ struct hlist_node *n;
+
+ j = 0;
+ hlist_for_each_entry_rcu(vport, n, &dp->ports[i], dp_hash_node) {
+ if (j >= skip &&
+ ovs_vport_cmd_fill_info(vport, skb,
+ NETLINK_CB(cb->skb).pid,
+ cb->nlh->nlmsg_seq,
+ NLM_F_MULTI,
+ OVS_VPORT_CMD_NEW) < 0)
+ goto out;
+
+ j++;
+ }
+ skip = 0;
}
+out:
rcu_read_unlock();
- cb->args[0] = port_no;
- retval = skb->len;
+ cb->args[0] = i;
+ cb->args[1] = j;
- return retval;
+ return skb->len;
}
static struct genl_ops dp_vport_genl_ops[] = {
diff --git a/net/openvswitch/datapath.h b/net/openvswitch/datapath.h
index 771c11e..129ec54 100644
--- a/net/openvswitch/datapath.h
+++ b/net/openvswitch/datapath.h
@@ -29,7 +29,9 @@
#include "flow.h"
#include "vport.h"
-#define DP_MAX_PORTS 1024
+#define DP_MAX_PORTS USHRT_MAX
+#define DP_VPORT_HASH_BUCKETS 1024
+
#define SAMPLE_ACTION_DEPTH 3
/**
@@ -57,10 +59,8 @@ 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.
*
@@ -75,8 +75,7 @@ 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;
@@ -87,6 +86,26 @@ struct datapath {
#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.
diff --git a/net/openvswitch/flow.c b/net/openvswitch/flow.c
index b7f38b1..f9f211d 100644
--- a/net/openvswitch/flow.c
+++ b/net/openvswitch/flow.c
@@ -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);
@@ -1000,7 +997,7 @@ 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. */
@@ -1143,7 +1140,7 @@ 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) {
@@ -1180,7 +1177,7 @@ 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;
diff --git a/net/openvswitch/flow.h b/net/openvswitch/flow.h
index 9b75617..d92e22a 100644
--- a/net/openvswitch/flow.h
+++ b/net/openvswitch/flow.h
@@ -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 {
diff --git a/net/openvswitch/vport.c b/net/openvswitch/vport.c
index 9873ace..1abd960 100644
--- a/net/openvswitch/vport.c
+++ b/net/openvswitch/vport.c
@@ -127,6 +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) {
diff --git a/net/openvswitch/vport.h b/net/openvswitch/vport.h
index 97cef08..c56e483 100644
--- a/net/openvswitch/vport.h
+++ b/net/openvswitch/vport.h
@@ -70,10 +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;
@@ -83,10 +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;
--
1.7.9.5
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [GIT net-next] Open vSwitch
[not found] ` <1346786049-3100-1-git-send-email-jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
2012-09-04 19:14 ` [PATCH net-next 1/2] openvswitch: Add support for network namespaces Jesse Gross
2012-09-04 19:14 ` [PATCH net-next 2/2] openvswitch: Increase maximum number of datapath ports Jesse Gross
@ 2012-09-04 19:26 ` David Miller
2 siblings, 0 replies; 4+ messages in thread
From: David Miller @ 2012-09-04 19:26 UTC (permalink / raw)
To: jesse-l0M0P4e3n4LQT0dZR+AlfA
Cc: dev-yBygre7rU0TnMu66kgdUjQ, netdev-u79uwXL29TY76Z2rM5mHXA
From: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
Date: Tue, 4 Sep 2012 12:14:07 -0700
> Two feature additions to Open vSwitch for net-next/3.7.
>
> The following changes since commit 0d7614f09c1ebdbaa1599a5aba7593f147bf96ee:
>
> Linux 3.6-rc1 (2012-08-02 16:38:10 -0700)
>
> are available in the git repository at:
>
> git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git master
>
> for you to fetch changes up to 15eac2a74277bc7de68a7c2a64a7c91b4b6f5961:
>
> openvswitch: Increase maximum number of datapath ports. (2012-09-03 19:20:49 -0700)
Pulled, thanks Jesse.
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2012-09-04 19:26 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2012-09-04 19:14 [GIT net-next] Open vSwitch Jesse Gross
[not found] ` <1346786049-3100-1-git-send-email-jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
2012-09-04 19:14 ` [PATCH net-next 1/2] openvswitch: Add support for network namespaces Jesse Gross
2012-09-04 19:14 ` [PATCH net-next 2/2] openvswitch: Increase maximum number of datapath ports Jesse Gross
2012-09-04 19:26 ` [GIT net-next] Open vSwitch David Miller
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.