datapath: rename brcompat.c to brcompat_main.c
authorChris Wright <chrisw@sous-sol.org>
Wed, 7 Mar 2012 18:28:02 +0000 (10:28 -0800)
committerJesse Gross <jesse@nicira.com>
Thu, 8 Mar 2012 21:29:38 +0000 (13:29 -0800)
This just makes it easier to subsequently rename the brcompat_mod module
to brcompat without introducing circular dependencies.

Signed-off-by: Chris Wright <chrisw@sous-sol.org>
Signed-off-by: Jesse Gross <jesse@nicira.com>
datapath/brcompat.c [deleted file]
datapath/brcompat_main.c [new file with mode: 0644]
datapath/linux/.gitignore
datapath/linux/Modules.mk

diff --git a/datapath/brcompat.c b/datapath/brcompat.c
deleted file mode 100644 (file)
index d4a0aca..0000000
+++ /dev/null
@@ -1,578 +0,0 @@
-/*
- * Copyright (c) 2007-2012 Nicira Networks.
- *
- * This program is free software; you can redistribute it and/or
- * modify it under the terms of version 2 of the GNU General Public
- * License as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but
- * WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
- * General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
- * 02110-1301, USA
- */
-
-#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
-
-#include <linux/module.h>
-#include <linux/kernel.h>
-#include <linux/uaccess.h>
-#include <linux/completion.h>
-#include <linux/etherdevice.h>
-#include <linux/if_bridge.h>
-#include <linux/netdevice.h>
-#include <linux/rtnetlink.h>
-#include <net/genetlink.h>
-
-#include "openvswitch/brcompat-netlink.h"
-#include "datapath.h"
-
-static struct genl_family brc_genl_family;
-static struct genl_multicast_group brc_mc_group;
-
-/* Time to wait for ovs-vswitchd to respond to a datapath action, in
- * jiffies. */
-#define BRC_TIMEOUT (HZ * 5)
-
-/* Mutex to serialize ovs-brcompatd callbacks.  (Some callbacks naturally hold
- * br_ioctl_mutex, others hold rtnl_lock, but we can't take the former
- * ourselves and we don't want to hold the latter over a potentially long
- * period of time.) */
-static DEFINE_MUTEX(brc_serial);
-
-/* Userspace communication. */
-static DEFINE_SPINLOCK(brc_lock);    /* Ensure atomic access to these vars. */
-static DECLARE_COMPLETION(brc_done); /* Userspace signaled operation done? */
-static struct sk_buff *brc_reply;    /* Reply from userspace. */
-static u32 brc_seq;                 /* Sequence number for current op. */
-
-static struct sk_buff *brc_send_command(struct net *,
-                                       struct sk_buff *,
-                                       struct nlattr **attrs);
-static int brc_send_simple_command(struct net *, struct sk_buff *);
-
-static struct sk_buff *brc_make_request(int op, const char *bridge,
-                                       const char *port)
-{
-       struct sk_buff *skb = genlmsg_new(NLMSG_GOODSIZE, GFP_KERNEL);
-       if (!skb)
-               goto error;
-
-       genlmsg_put(skb, 0, 0, &brc_genl_family, 0, op);
-       if (bridge)
-               NLA_PUT_STRING(skb, BRC_GENL_A_DP_NAME, bridge);
-       if (port)
-               NLA_PUT_STRING(skb, BRC_GENL_A_PORT_NAME, port);
-       return skb;
-
-nla_put_failure:
-       kfree_skb(skb);
-error:
-       return NULL;
-}
-
-static int brc_send_simple_command(struct net *net, struct sk_buff *request)
-{
-       struct nlattr *attrs[BRC_GENL_A_MAX + 1];
-       struct sk_buff *reply;
-       int error;
-
-       reply = brc_send_command(net, request, attrs);
-       if (IS_ERR(reply))
-               return PTR_ERR(reply);
-
-       error = nla_get_u32(attrs[BRC_GENL_A_ERR_CODE]);
-       kfree_skb(reply);
-       return -error;
-}
-
-static int brc_add_del_bridge(struct net *net, char __user *uname, int add)
-{
-       struct sk_buff *request;
-       char name[IFNAMSIZ];
-
-       if (!capable(CAP_NET_ADMIN))
-               return -EPERM;
-
-       if (copy_from_user(name, uname, IFNAMSIZ))
-               return -EFAULT;
-
-       name[IFNAMSIZ - 1] = 0;
-       request = brc_make_request(add ? BRC_GENL_C_DP_ADD : BRC_GENL_C_DP_DEL,
-                                  name, NULL);
-       if (!request)
-               return -ENOMEM;
-
-       return brc_send_simple_command(net, request);
-}
-
-static int brc_get_indices(struct net *net,
-                          int op, const char *br_name,
-                          int __user *uindices, int n)
-{
-       struct nlattr *attrs[BRC_GENL_A_MAX + 1];
-       struct sk_buff *request, *reply;
-       int *indices;
-       int ret;
-       int len;
-
-       if (n < 0)
-               return -EINVAL;
-       if (n >= 2048)
-               return -ENOMEM;
-
-       request = brc_make_request(op, br_name, NULL);
-       if (!request)
-               return -ENOMEM;
-
-       reply = brc_send_command(net, request, attrs);
-       ret = PTR_ERR(reply);
-       if (IS_ERR(reply))
-               goto exit;
-
-       ret = -nla_get_u32(attrs[BRC_GENL_A_ERR_CODE]);
-       if (ret < 0)
-               goto exit_free_skb;
-
-       ret = -EINVAL;
-       if (!attrs[BRC_GENL_A_IFINDEXES])
-               goto exit_free_skb;
-
-       len = nla_len(attrs[BRC_GENL_A_IFINDEXES]);
-       indices = nla_data(attrs[BRC_GENL_A_IFINDEXES]);
-       if (len % sizeof(int))
-               goto exit_free_skb;
-
-       n = min_t(int, n, len / sizeof(int));
-       ret = copy_to_user(uindices, indices, n * sizeof(int)) ? -EFAULT : n;
-
-exit_free_skb:
-       kfree_skb(reply);
-exit:
-       return ret;
-}
-
-/* Called with br_ioctl_mutex. */
-static int brc_get_bridges(struct net *net, int __user *uindices, int n)
-{
-       return brc_get_indices(net, BRC_GENL_C_GET_BRIDGES, NULL, uindices, n);
-}
-
-/* Legacy deviceless bridge ioctl's.  Called with br_ioctl_mutex. */
-static int old_deviceless(struct net *net, void __user *uarg)
-{
-       unsigned long args[3];
-
-       if (copy_from_user(args, uarg, sizeof(args)))
-               return -EFAULT;
-
-       switch (args[0]) {
-       case BRCTL_GET_BRIDGES:
-               return brc_get_bridges(net, (int __user *)args[1], args[2]);
-
-       case BRCTL_ADD_BRIDGE:
-               return brc_add_del_bridge(net, (void __user *)args[1], 1);
-       case BRCTL_DEL_BRIDGE:
-               return brc_add_del_bridge(net, (void __user *)args[1], 0);
-       }
-
-       return -EOPNOTSUPP;
-}
-
-/* Called with the br_ioctl_mutex. */
-static int
-#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23)
-brc_ioctl_deviceless_stub(unsigned int cmd, void __user *uarg)
-{
-       struct net *net = NULL;
-#else
-brc_ioctl_deviceless_stub(struct net *net, unsigned int cmd, void __user *uarg)
-{
-#endif
-       switch (cmd) {
-       case SIOCGIFBR:
-       case SIOCSIFBR:
-               return old_deviceless(net, uarg);
-
-       case SIOCBRADDBR:
-               return brc_add_del_bridge(net, uarg, 1);
-       case SIOCBRDELBR:
-               return brc_add_del_bridge(net, uarg, 0);
-       }
-
-       return -EOPNOTSUPP;
-}
-
-static int brc_add_del_port(struct net_device *dev, int port_ifindex, int add)
-{
-       struct sk_buff *request;
-       struct net_device *port;
-       int err;
-
-       if (!capable(CAP_NET_ADMIN))
-               return -EPERM;
-
-       port = __dev_get_by_index(dev_net(dev), port_ifindex);
-       if (!port)
-               return -EINVAL;
-
-       /* Save name of dev and port because there's a race between the
-        * rtnl_unlock() and the brc_send_simple_command(). */
-       request = brc_make_request(add ? BRC_GENL_C_PORT_ADD : BRC_GENL_C_PORT_DEL,
-                                  dev->name, port->name);
-       if (!request)
-               return -ENOMEM;
-
-       rtnl_unlock();
-       err = brc_send_simple_command(dev_net(dev), request);
-       rtnl_lock();
-
-       return err;
-}
-
-static int brc_get_bridge_info(struct net_device *dev,
-                              struct __bridge_info __user *ub)
-{
-       struct __bridge_info b;
-
-       memset(&b, 0, sizeof(struct __bridge_info));
-
-       /* First two bytes are the priority, which we should skip.  This comes
-        * from struct bridge_id in br_private.h, which is unavailable to us.
-        */
-       memcpy((u8 *)&b.bridge_id + 2, dev->dev_addr, ETH_ALEN);
-       b.stp_enabled = 0;
-
-       if (copy_to_user(ub, &b, sizeof(struct __bridge_info)))
-               return -EFAULT;
-
-       return 0;
-}
-
-static int brc_get_port_list(struct net_device *dev, int __user *uindices,
-                            int num)
-{
-       int retval;
-
-       rtnl_unlock();
-       retval = brc_get_indices(dev_net(dev), BRC_GENL_C_GET_PORTS, dev->name,
-                                uindices, num);
-       rtnl_lock();
-
-       return retval;
-}
-
-/*
- * Format up to a page worth of forwarding table entries
- * userbuf -- where to copy result
- * maxnum  -- maximum number of entries desired
- *            (limited to a page for sanity)
- * offset  -- number of records to skip
- */
-static int brc_get_fdb_entries(struct net_device *dev, void __user *userbuf,
-                              unsigned long maxnum, unsigned long offset)
-{
-       struct nlattr *attrs[BRC_GENL_A_MAX + 1];
-       struct sk_buff *request, *reply;
-       int retval;
-       int len;
-
-       /* Clamp size to PAGE_SIZE, test maxnum to avoid overflow */
-       if (maxnum > PAGE_SIZE/sizeof(struct __fdb_entry))
-               maxnum = PAGE_SIZE/sizeof(struct __fdb_entry);
-
-       request = brc_make_request(BRC_GENL_C_FDB_QUERY, dev->name, NULL);
-       if (!request)
-               return -ENOMEM;
-       NLA_PUT_U64(request, BRC_GENL_A_FDB_COUNT, maxnum);
-       NLA_PUT_U64(request, BRC_GENL_A_FDB_SKIP, offset);
-
-       rtnl_unlock();
-       reply = brc_send_command(dev_net(dev), request, attrs);
-       retval = PTR_ERR(reply);
-       if (IS_ERR(reply))
-               goto exit;
-
-       retval = -nla_get_u32(attrs[BRC_GENL_A_ERR_CODE]);
-       if (retval < 0)
-               goto exit_free_skb;
-
-       retval = -EINVAL;
-       if (!attrs[BRC_GENL_A_FDB_DATA])
-               goto exit_free_skb;
-       len = nla_len(attrs[BRC_GENL_A_FDB_DATA]);
-       if (len % sizeof(struct __fdb_entry) ||
-           len / sizeof(struct __fdb_entry) > maxnum)
-               goto exit_free_skb;
-
-       retval = len / sizeof(struct __fdb_entry);
-       if (copy_to_user(userbuf, nla_data(attrs[BRC_GENL_A_FDB_DATA]), len))
-               retval = -EFAULT;
-
-exit_free_skb:
-       kfree_skb(reply);
-exit:
-       rtnl_lock();
-       return retval;
-
-nla_put_failure:
-       kfree_skb(request);
-       return -ENOMEM;
-}
-
-/* Legacy ioctl's through SIOCDEVPRIVATE.  Called with rtnl_lock. */
-static int old_dev_ioctl(struct net_device *dev, struct ifreq *rq, int cmd)
-{
-       unsigned long args[4];
-
-       if (copy_from_user(args, rq->ifr_data, sizeof(args)))
-               return -EFAULT;
-
-       switch (args[0]) {
-       case BRCTL_ADD_IF:
-               return brc_add_del_port(dev, args[1], 1);
-       case BRCTL_DEL_IF:
-               return brc_add_del_port(dev, args[1], 0);
-
-       case BRCTL_GET_BRIDGE_INFO:
-               return brc_get_bridge_info(dev, (struct __bridge_info __user *)args[1]);
-
-       case BRCTL_GET_PORT_LIST:
-               return brc_get_port_list(dev, (int __user *)args[1], args[2]);
-
-       case BRCTL_GET_FDB_ENTRIES:
-               return brc_get_fdb_entries(dev, (void __user *)args[1],
-                                          args[2], args[3]);
-       }
-
-       return -EOPNOTSUPP;
-}
-
-/* Called with the rtnl_lock. */
-static int brc_dev_ioctl(struct net_device *dev, struct ifreq *rq, int cmd)
-{
-       int err;
-
-       switch (cmd) {
-       case SIOCDEVPRIVATE:
-               err = old_dev_ioctl(dev, rq, cmd);
-               break;
-
-       case SIOCBRADDIF:
-               return brc_add_del_port(dev, rq->ifr_ifindex, 1);
-       case SIOCBRDELIF:
-               return brc_add_del_port(dev, rq->ifr_ifindex, 0);
-
-       default:
-               err = -EOPNOTSUPP;
-               break;
-       }
-
-       return err;
-}
-
-
-static struct genl_family brc_genl_family = {
-       .id = GENL_ID_GENERATE,
-       .hdrsize = 0,
-       .name = BRC_GENL_FAMILY_NAME,
-       .version = 1,
-       .maxattr = BRC_GENL_A_MAX,
-        SET_NETNSOK
-};
-
-static int brc_genl_query(struct sk_buff *skb, struct genl_info *info)
-{
-       int err = -EINVAL;
-       struct sk_buff *ans_skb;
-       void *data;
-
-       ans_skb = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
-       if (!ans_skb)
-               return -ENOMEM;
-
-       data = genlmsg_put_reply(ans_skb, info, &brc_genl_family,
-                                0, BRC_GENL_C_QUERY_MC);
-       if (data == NULL) {
-               err = -ENOMEM;
-               goto err;
-       }
-       NLA_PUT_U32(ans_skb, BRC_GENL_A_MC_GROUP, brc_mc_group.id);
-
-       genlmsg_end(ans_skb, data);
-       return genlmsg_reply(ans_skb, info);
-
-err:
-nla_put_failure:
-       kfree_skb(ans_skb);
-       return err;
-}
-
-/* Attribute policy: what each attribute may contain.  */
-static struct nla_policy brc_genl_policy[BRC_GENL_A_MAX + 1] = {
-       [BRC_GENL_A_ERR_CODE] = { .type = NLA_U32 },
-       [BRC_GENL_A_FDB_DATA] = { .type = NLA_UNSPEC },
-};
-
-static int brc_genl_dp_result(struct sk_buff *skb, struct genl_info *info)
-{
-       unsigned long int flags;
-       int err;
-
-       if (!info->attrs[BRC_GENL_A_ERR_CODE])
-               return -EINVAL;
-
-       skb = skb_clone(skb, GFP_KERNEL);
-       if (!skb)
-               return -ENOMEM;
-
-       spin_lock_irqsave(&brc_lock, flags);
-       if (brc_seq == info->snd_seq) {
-               brc_seq++;
-
-               kfree_skb(brc_reply);
-               brc_reply = skb;
-
-               complete(&brc_done);
-               err = 0;
-       } else {
-               kfree_skb(skb);
-               err = -ESTALE;
-       }
-       spin_unlock_irqrestore(&brc_lock, flags);
-
-       return err;
-}
-
-static struct genl_ops brc_genl_ops[] = {
-       { .cmd = BRC_GENL_C_QUERY_MC,
-         .flags = GENL_ADMIN_PERM, /* Requires CAP_NET_ADMIN privelege. */
-         .policy = NULL,
-         .doit = brc_genl_query,
-       },
-       { .cmd = BRC_GENL_C_DP_RESULT,
-         .flags = GENL_ADMIN_PERM, /* Requires CAP_NET_ADMIN privelege. */
-         .policy = brc_genl_policy,
-         .doit = brc_genl_dp_result,
-       },
-};
-
-static struct sk_buff *brc_send_command(struct net *net,
-                                       struct sk_buff *request,
-                                       struct nlattr **attrs)
-{
-       unsigned long int flags;
-       struct sk_buff *reply;
-       int error;
-
-       mutex_lock(&brc_serial);
-
-       /* Increment sequence number first, so that we ignore any replies
-        * to stale requests. */
-       spin_lock_irqsave(&brc_lock, flags);
-       nlmsg_hdr(request)->nlmsg_seq = ++brc_seq;
-       INIT_COMPLETION(brc_done);
-       spin_unlock_irqrestore(&brc_lock, flags);
-
-       nlmsg_end(request, nlmsg_hdr(request));
-
-       /* Send message. */
-       error = genlmsg_multicast_netns(net, request, 0,
-                                       brc_mc_group.id, GFP_KERNEL);
-       if (error < 0)
-               goto error;
-
-       /* Wait for reply. */
-       error = -ETIMEDOUT;
-       if (!wait_for_completion_timeout(&brc_done, BRC_TIMEOUT)) {
-               pr_warn("timed out waiting for userspace\n");
-               goto error;
-       }
-
-       /* Grab reply. */
-       spin_lock_irqsave(&brc_lock, flags);
-       reply = brc_reply;
-       brc_reply = NULL;
-       spin_unlock_irqrestore(&brc_lock, flags);
-
-       mutex_unlock(&brc_serial);
-
-       /* Re-parse message.  Can't fail, since it parsed correctly once
-        * already. */
-       error = nlmsg_parse(nlmsg_hdr(reply), GENL_HDRLEN,
-                           attrs, BRC_GENL_A_MAX, brc_genl_policy);
-       WARN_ON(error);
-
-       return reply;
-
-error:
-       mutex_unlock(&brc_serial);
-       return ERR_PTR(error);
-}
-
-static int __init brc_init(void)
-{
-       int err;
-
-       pr_info("Open vSwitch Bridge Compatibility, built "__DATE__" "__TIME__"\n");
-
-       /* Set the bridge ioctl handler */
-       brioctl_set(brc_ioctl_deviceless_stub);
-
-       /* Set the openvswitch_mod device ioctl handler */
-       ovs_dp_ioctl_hook = brc_dev_ioctl;
-
-       /* Randomize the initial sequence number.  This is not a security
-        * feature; it only helps avoid crossed wires between userspace and
-        * the kernel when the module is unloaded and reloaded. */
-       brc_seq = net_random();
-
-       /* Register generic netlink family to communicate changes to
-        * userspace. */
-       err = genl_register_family_with_ops(&brc_genl_family,
-                                           brc_genl_ops, ARRAY_SIZE(brc_genl_ops));
-       if (err)
-               goto error;
-
-       strcpy(brc_mc_group.name, "brcompat");
-       err = genl_register_mc_group(&brc_genl_family, &brc_mc_group);
-       if (err < 0)
-               goto err_unregister;
-
-       return 0;
-
-err_unregister:
-       genl_unregister_family(&brc_genl_family);
-error:
-       pr_emerg("failed to install!\n");
-       return err;
-}
-
-static void brc_cleanup(void)
-{
-       /* Unregister ioctl hooks */
-       ovs_dp_ioctl_hook = NULL;
-       brioctl_set(NULL);
-
-       genl_unregister_family(&brc_genl_family);
-}
-
-module_init(brc_init);
-module_exit(brc_cleanup);
-
-MODULE_DESCRIPTION("Open vSwitch bridge compatibility");
-MODULE_AUTHOR("Nicira Networks");
-MODULE_LICENSE("GPL");
-
-#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,36)
-/*
- * In kernels 2.6.36 and later, Open vSwitch can safely coexist with
- * the Linux bridge module, but it does not make sense to load both bridge and
- * brcompat_mod, so this prevents it.
- */
-BRIDGE_MUTUAL_EXCLUSION;
-#endif
diff --git a/datapath/brcompat_main.c b/datapath/brcompat_main.c
new file mode 100644 (file)
index 0000000..d4a0aca
--- /dev/null
@@ -0,0 +1,578 @@
+/*
+ * Copyright (c) 2007-2012 Nicira Networks.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of version 2 of the GNU General Public
+ * License as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/uaccess.h>
+#include <linux/completion.h>
+#include <linux/etherdevice.h>
+#include <linux/if_bridge.h>
+#include <linux/netdevice.h>
+#include <linux/rtnetlink.h>
+#include <net/genetlink.h>
+
+#include "openvswitch/brcompat-netlink.h"
+#include "datapath.h"
+
+static struct genl_family brc_genl_family;
+static struct genl_multicast_group brc_mc_group;
+
+/* Time to wait for ovs-vswitchd to respond to a datapath action, in
+ * jiffies. */
+#define BRC_TIMEOUT (HZ * 5)
+
+/* Mutex to serialize ovs-brcompatd callbacks.  (Some callbacks naturally hold
+ * br_ioctl_mutex, others hold rtnl_lock, but we can't take the former
+ * ourselves and we don't want to hold the latter over a potentially long
+ * period of time.) */
+static DEFINE_MUTEX(brc_serial);
+
+/* Userspace communication. */
+static DEFINE_SPINLOCK(brc_lock);    /* Ensure atomic access to these vars. */
+static DECLARE_COMPLETION(brc_done); /* Userspace signaled operation done? */
+static struct sk_buff *brc_reply;    /* Reply from userspace. */
+static u32 brc_seq;                 /* Sequence number for current op. */
+
+static struct sk_buff *brc_send_command(struct net *,
+                                       struct sk_buff *,
+                                       struct nlattr **attrs);
+static int brc_send_simple_command(struct net *, struct sk_buff *);
+
+static struct sk_buff *brc_make_request(int op, const char *bridge,
+                                       const char *port)
+{
+       struct sk_buff *skb = genlmsg_new(NLMSG_GOODSIZE, GFP_KERNEL);
+       if (!skb)
+               goto error;
+
+       genlmsg_put(skb, 0, 0, &brc_genl_family, 0, op);
+       if (bridge)
+               NLA_PUT_STRING(skb, BRC_GENL_A_DP_NAME, bridge);
+       if (port)
+               NLA_PUT_STRING(skb, BRC_GENL_A_PORT_NAME, port);
+       return skb;
+
+nla_put_failure:
+       kfree_skb(skb);
+error:
+       return NULL;
+}
+
+static int brc_send_simple_command(struct net *net, struct sk_buff *request)
+{
+       struct nlattr *attrs[BRC_GENL_A_MAX + 1];
+       struct sk_buff *reply;
+       int error;
+
+       reply = brc_send_command(net, request, attrs);
+       if (IS_ERR(reply))
+               return PTR_ERR(reply);
+
+       error = nla_get_u32(attrs[BRC_GENL_A_ERR_CODE]);
+       kfree_skb(reply);
+       return -error;
+}
+
+static int brc_add_del_bridge(struct net *net, char __user *uname, int add)
+{
+       struct sk_buff *request;
+       char name[IFNAMSIZ];
+
+       if (!capable(CAP_NET_ADMIN))
+               return -EPERM;
+
+       if (copy_from_user(name, uname, IFNAMSIZ))
+               return -EFAULT;
+
+       name[IFNAMSIZ - 1] = 0;
+       request = brc_make_request(add ? BRC_GENL_C_DP_ADD : BRC_GENL_C_DP_DEL,
+                                  name, NULL);
+       if (!request)
+               return -ENOMEM;
+
+       return brc_send_simple_command(net, request);
+}
+
+static int brc_get_indices(struct net *net,
+                          int op, const char *br_name,
+                          int __user *uindices, int n)
+{
+       struct nlattr *attrs[BRC_GENL_A_MAX + 1];
+       struct sk_buff *request, *reply;
+       int *indices;
+       int ret;
+       int len;
+
+       if (n < 0)
+               return -EINVAL;
+       if (n >= 2048)
+               return -ENOMEM;
+
+       request = brc_make_request(op, br_name, NULL);
+       if (!request)
+               return -ENOMEM;
+
+       reply = brc_send_command(net, request, attrs);
+       ret = PTR_ERR(reply);
+       if (IS_ERR(reply))
+               goto exit;
+
+       ret = -nla_get_u32(attrs[BRC_GENL_A_ERR_CODE]);
+       if (ret < 0)
+               goto exit_free_skb;
+
+       ret = -EINVAL;
+       if (!attrs[BRC_GENL_A_IFINDEXES])
+               goto exit_free_skb;
+
+       len = nla_len(attrs[BRC_GENL_A_IFINDEXES]);
+       indices = nla_data(attrs[BRC_GENL_A_IFINDEXES]);
+       if (len % sizeof(int))
+               goto exit_free_skb;
+
+       n = min_t(int, n, len / sizeof(int));
+       ret = copy_to_user(uindices, indices, n * sizeof(int)) ? -EFAULT : n;
+
+exit_free_skb:
+       kfree_skb(reply);
+exit:
+       return ret;
+}
+
+/* Called with br_ioctl_mutex. */
+static int brc_get_bridges(struct net *net, int __user *uindices, int n)
+{
+       return brc_get_indices(net, BRC_GENL_C_GET_BRIDGES, NULL, uindices, n);
+}
+
+/* Legacy deviceless bridge ioctl's.  Called with br_ioctl_mutex. */
+static int old_deviceless(struct net *net, void __user *uarg)
+{
+       unsigned long args[3];
+
+       if (copy_from_user(args, uarg, sizeof(args)))
+               return -EFAULT;
+
+       switch (args[0]) {
+       case BRCTL_GET_BRIDGES:
+               return brc_get_bridges(net, (int __user *)args[1], args[2]);
+
+       case BRCTL_ADD_BRIDGE:
+               return brc_add_del_bridge(net, (void __user *)args[1], 1);
+       case BRCTL_DEL_BRIDGE:
+               return brc_add_del_bridge(net, (void __user *)args[1], 0);
+       }
+
+       return -EOPNOTSUPP;
+}
+
+/* Called with the br_ioctl_mutex. */
+static int
+#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23)
+brc_ioctl_deviceless_stub(unsigned int cmd, void __user *uarg)
+{
+       struct net *net = NULL;
+#else
+brc_ioctl_deviceless_stub(struct net *net, unsigned int cmd, void __user *uarg)
+{
+#endif
+       switch (cmd) {
+       case SIOCGIFBR:
+       case SIOCSIFBR:
+               return old_deviceless(net, uarg);
+
+       case SIOCBRADDBR:
+               return brc_add_del_bridge(net, uarg, 1);
+       case SIOCBRDELBR:
+               return brc_add_del_bridge(net, uarg, 0);
+       }
+
+       return -EOPNOTSUPP;
+}
+
+static int brc_add_del_port(struct net_device *dev, int port_ifindex, int add)
+{
+       struct sk_buff *request;
+       struct net_device *port;
+       int err;
+
+       if (!capable(CAP_NET_ADMIN))
+               return -EPERM;
+
+       port = __dev_get_by_index(dev_net(dev), port_ifindex);
+       if (!port)
+               return -EINVAL;
+
+       /* Save name of dev and port because there's a race between the
+        * rtnl_unlock() and the brc_send_simple_command(). */
+       request = brc_make_request(add ? BRC_GENL_C_PORT_ADD : BRC_GENL_C_PORT_DEL,
+                                  dev->name, port->name);
+       if (!request)
+               return -ENOMEM;
+
+       rtnl_unlock();
+       err = brc_send_simple_command(dev_net(dev), request);
+       rtnl_lock();
+
+       return err;
+}
+
+static int brc_get_bridge_info(struct net_device *dev,
+                              struct __bridge_info __user *ub)
+{
+       struct __bridge_info b;
+
+       memset(&b, 0, sizeof(struct __bridge_info));
+
+       /* First two bytes are the priority, which we should skip.  This comes
+        * from struct bridge_id in br_private.h, which is unavailable to us.
+        */
+       memcpy((u8 *)&b.bridge_id + 2, dev->dev_addr, ETH_ALEN);
+       b.stp_enabled = 0;
+
+       if (copy_to_user(ub, &b, sizeof(struct __bridge_info)))
+               return -EFAULT;
+
+       return 0;
+}
+
+static int brc_get_port_list(struct net_device *dev, int __user *uindices,
+                            int num)
+{
+       int retval;
+
+       rtnl_unlock();
+       retval = brc_get_indices(dev_net(dev), BRC_GENL_C_GET_PORTS, dev->name,
+                                uindices, num);
+       rtnl_lock();
+
+       return retval;
+}
+
+/*
+ * Format up to a page worth of forwarding table entries
+ * userbuf -- where to copy result
+ * maxnum  -- maximum number of entries desired
+ *            (limited to a page for sanity)
+ * offset  -- number of records to skip
+ */
+static int brc_get_fdb_entries(struct net_device *dev, void __user *userbuf,
+                              unsigned long maxnum, unsigned long offset)
+{
+       struct nlattr *attrs[BRC_GENL_A_MAX + 1];
+       struct sk_buff *request, *reply;
+       int retval;
+       int len;
+
+       /* Clamp size to PAGE_SIZE, test maxnum to avoid overflow */
+       if (maxnum > PAGE_SIZE/sizeof(struct __fdb_entry))
+               maxnum = PAGE_SIZE/sizeof(struct __fdb_entry);
+
+       request = brc_make_request(BRC_GENL_C_FDB_QUERY, dev->name, NULL);
+       if (!request)
+               return -ENOMEM;
+       NLA_PUT_U64(request, BRC_GENL_A_FDB_COUNT, maxnum);
+       NLA_PUT_U64(request, BRC_GENL_A_FDB_SKIP, offset);
+
+       rtnl_unlock();
+       reply = brc_send_command(dev_net(dev), request, attrs);
+       retval = PTR_ERR(reply);
+       if (IS_ERR(reply))
+               goto exit;
+
+       retval = -nla_get_u32(attrs[BRC_GENL_A_ERR_CODE]);
+       if (retval < 0)
+               goto exit_free_skb;
+
+       retval = -EINVAL;
+       if (!attrs[BRC_GENL_A_FDB_DATA])
+               goto exit_free_skb;
+       len = nla_len(attrs[BRC_GENL_A_FDB_DATA]);
+       if (len % sizeof(struct __fdb_entry) ||
+           len / sizeof(struct __fdb_entry) > maxnum)
+               goto exit_free_skb;
+
+       retval = len / sizeof(struct __fdb_entry);
+       if (copy_to_user(userbuf, nla_data(attrs[BRC_GENL_A_FDB_DATA]), len))
+               retval = -EFAULT;
+
+exit_free_skb:
+       kfree_skb(reply);
+exit:
+       rtnl_lock();
+       return retval;
+
+nla_put_failure:
+       kfree_skb(request);
+       return -ENOMEM;
+}
+
+/* Legacy ioctl's through SIOCDEVPRIVATE.  Called with rtnl_lock. */
+static int old_dev_ioctl(struct net_device *dev, struct ifreq *rq, int cmd)
+{
+       unsigned long args[4];
+
+       if (copy_from_user(args, rq->ifr_data, sizeof(args)))
+               return -EFAULT;
+
+       switch (args[0]) {
+       case BRCTL_ADD_IF:
+               return brc_add_del_port(dev, args[1], 1);
+       case BRCTL_DEL_IF:
+               return brc_add_del_port(dev, args[1], 0);
+
+       case BRCTL_GET_BRIDGE_INFO:
+               return brc_get_bridge_info(dev, (struct __bridge_info __user *)args[1]);
+
+       case BRCTL_GET_PORT_LIST:
+               return brc_get_port_list(dev, (int __user *)args[1], args[2]);
+
+       case BRCTL_GET_FDB_ENTRIES:
+               return brc_get_fdb_entries(dev, (void __user *)args[1],
+                                          args[2], args[3]);
+       }
+
+       return -EOPNOTSUPP;
+}
+
+/* Called with the rtnl_lock. */
+static int brc_dev_ioctl(struct net_device *dev, struct ifreq *rq, int cmd)
+{
+       int err;
+
+       switch (cmd) {
+       case SIOCDEVPRIVATE:
+               err = old_dev_ioctl(dev, rq, cmd);
+               break;
+
+       case SIOCBRADDIF:
+               return brc_add_del_port(dev, rq->ifr_ifindex, 1);
+       case SIOCBRDELIF:
+               return brc_add_del_port(dev, rq->ifr_ifindex, 0);
+
+       default:
+               err = -EOPNOTSUPP;
+               break;
+       }
+
+       return err;
+}
+
+
+static struct genl_family brc_genl_family = {
+       .id = GENL_ID_GENERATE,
+       .hdrsize = 0,
+       .name = BRC_GENL_FAMILY_NAME,
+       .version = 1,
+       .maxattr = BRC_GENL_A_MAX,
+        SET_NETNSOK
+};
+
+static int brc_genl_query(struct sk_buff *skb, struct genl_info *info)
+{
+       int err = -EINVAL;
+       struct sk_buff *ans_skb;
+       void *data;
+
+       ans_skb = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
+       if (!ans_skb)
+               return -ENOMEM;
+
+       data = genlmsg_put_reply(ans_skb, info, &brc_genl_family,
+                                0, BRC_GENL_C_QUERY_MC);
+       if (data == NULL) {
+               err = -ENOMEM;
+               goto err;
+       }
+       NLA_PUT_U32(ans_skb, BRC_GENL_A_MC_GROUP, brc_mc_group.id);
+
+       genlmsg_end(ans_skb, data);
+       return genlmsg_reply(ans_skb, info);
+
+err:
+nla_put_failure:
+       kfree_skb(ans_skb);
+       return err;
+}
+
+/* Attribute policy: what each attribute may contain.  */
+static struct nla_policy brc_genl_policy[BRC_GENL_A_MAX + 1] = {
+       [BRC_GENL_A_ERR_CODE] = { .type = NLA_U32 },
+       [BRC_GENL_A_FDB_DATA] = { .type = NLA_UNSPEC },
+};
+
+static int brc_genl_dp_result(struct sk_buff *skb, struct genl_info *info)
+{
+       unsigned long int flags;
+       int err;
+
+       if (!info->attrs[BRC_GENL_A_ERR_CODE])
+               return -EINVAL;
+
+       skb = skb_clone(skb, GFP_KERNEL);
+       if (!skb)
+               return -ENOMEM;
+
+       spin_lock_irqsave(&brc_lock, flags);
+       if (brc_seq == info->snd_seq) {
+               brc_seq++;
+
+               kfree_skb(brc_reply);
+               brc_reply = skb;
+
+               complete(&brc_done);
+               err = 0;
+       } else {
+               kfree_skb(skb);
+               err = -ESTALE;
+       }
+       spin_unlock_irqrestore(&brc_lock, flags);
+
+       return err;
+}
+
+static struct genl_ops brc_genl_ops[] = {
+       { .cmd = BRC_GENL_C_QUERY_MC,
+         .flags = GENL_ADMIN_PERM, /* Requires CAP_NET_ADMIN privelege. */
+         .policy = NULL,
+         .doit = brc_genl_query,
+       },
+       { .cmd = BRC_GENL_C_DP_RESULT,
+         .flags = GENL_ADMIN_PERM, /* Requires CAP_NET_ADMIN privelege. */
+         .policy = brc_genl_policy,
+         .doit = brc_genl_dp_result,
+       },
+};
+
+static struct sk_buff *brc_send_command(struct net *net,
+                                       struct sk_buff *request,
+                                       struct nlattr **attrs)
+{
+       unsigned long int flags;
+       struct sk_buff *reply;
+       int error;
+
+       mutex_lock(&brc_serial);
+
+       /* Increment sequence number first, so that we ignore any replies
+        * to stale requests. */
+       spin_lock_irqsave(&brc_lock, flags);
+       nlmsg_hdr(request)->nlmsg_seq = ++brc_seq;
+       INIT_COMPLETION(brc_done);
+       spin_unlock_irqrestore(&brc_lock, flags);
+
+       nlmsg_end(request, nlmsg_hdr(request));
+
+       /* Send message. */
+       error = genlmsg_multicast_netns(net, request, 0,
+                                       brc_mc_group.id, GFP_KERNEL);
+       if (error < 0)
+               goto error;
+
+       /* Wait for reply. */
+       error = -ETIMEDOUT;
+       if (!wait_for_completion_timeout(&brc_done, BRC_TIMEOUT)) {
+               pr_warn("timed out waiting for userspace\n");
+               goto error;
+       }
+
+       /* Grab reply. */
+       spin_lock_irqsave(&brc_lock, flags);
+       reply = brc_reply;
+       brc_reply = NULL;
+       spin_unlock_irqrestore(&brc_lock, flags);
+
+       mutex_unlock(&brc_serial);
+
+       /* Re-parse message.  Can't fail, since it parsed correctly once
+        * already. */
+       error = nlmsg_parse(nlmsg_hdr(reply), GENL_HDRLEN,
+                           attrs, BRC_GENL_A_MAX, brc_genl_policy);
+       WARN_ON(error);
+
+       return reply;
+
+error:
+       mutex_unlock(&brc_serial);
+       return ERR_PTR(error);
+}
+
+static int __init brc_init(void)
+{
+       int err;
+
+       pr_info("Open vSwitch Bridge Compatibility, built "__DATE__" "__TIME__"\n");
+
+       /* Set the bridge ioctl handler */
+       brioctl_set(brc_ioctl_deviceless_stub);
+
+       /* Set the openvswitch_mod device ioctl handler */
+       ovs_dp_ioctl_hook = brc_dev_ioctl;
+
+       /* Randomize the initial sequence number.  This is not a security
+        * feature; it only helps avoid crossed wires between userspace and
+        * the kernel when the module is unloaded and reloaded. */
+       brc_seq = net_random();
+
+       /* Register generic netlink family to communicate changes to
+        * userspace. */
+       err = genl_register_family_with_ops(&brc_genl_family,
+                                           brc_genl_ops, ARRAY_SIZE(brc_genl_ops));
+       if (err)
+               goto error;
+
+       strcpy(brc_mc_group.name, "brcompat");
+       err = genl_register_mc_group(&brc_genl_family, &brc_mc_group);
+       if (err < 0)
+               goto err_unregister;
+
+       return 0;
+
+err_unregister:
+       genl_unregister_family(&brc_genl_family);
+error:
+       pr_emerg("failed to install!\n");
+       return err;
+}
+
+static void brc_cleanup(void)
+{
+       /* Unregister ioctl hooks */
+       ovs_dp_ioctl_hook = NULL;
+       brioctl_set(NULL);
+
+       genl_unregister_family(&brc_genl_family);
+}
+
+module_init(brc_init);
+module_exit(brc_cleanup);
+
+MODULE_DESCRIPTION("Open vSwitch bridge compatibility");
+MODULE_AUTHOR("Nicira Networks");
+MODULE_LICENSE("GPL");
+
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,36)
+/*
+ * In kernels 2.6.36 and later, Open vSwitch can safely coexist with
+ * the Linux bridge module, but it does not make sense to load both bridge and
+ * brcompat_mod, so this prevents it.
+ */
+BRIDGE_MUTUAL_EXCLUSION;
+#endif
index 0aee74651b49b72be74727ce04eb63699155e21f..e4b708ef99fb985570023406eabded019ce6145d 100644 (file)
@@ -6,7 +6,7 @@
 /addrconf_core-openvswitch.c
 /brc_sysfs_dp.c
 /brc_sysfs_if.c
-/brcompat.c
+/brcompat_main.c
 /checksum.c
 /dev-openvswitch.c
 /dp_sysfs_dp.c
index 86341ad07d0dff15c79e6eae4fda453b9847f853..d9a85ccb7c6c579a6b50db0eb46ff03ec097d470 100644 (file)
@@ -70,5 +70,5 @@ openvswitch_headers += \
        linux/compat/genetlink.inc
 
 both_modules += brcompat
-brcompat_sources = linux/compat/genetlink-brcompat.c brcompat.c
+brcompat_sources = linux/compat/genetlink-brcompat.c brcompat_main.c
 brcompat_headers =