Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / net / switchdev / switchdev.c
diff --git a/kernel/net/switchdev/switchdev.c b/kernel/net/switchdev/switchdev.c
new file mode 100644 (file)
index 0000000..055453d
--- /dev/null
@@ -0,0 +1,404 @@
+/*
+ * net/switchdev/switchdev.c - Switch device API
+ * Copyright (c) 2014 Jiri Pirko <jiri@resnulli.us>
+ * Copyright (c) 2014-2015 Scott Feldman <sfeldma@gmail.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/init.h>
+#include <linux/mutex.h>
+#include <linux/notifier.h>
+#include <linux/netdevice.h>
+#include <net/ip_fib.h>
+#include <net/switchdev.h>
+
+/**
+ *     netdev_switch_parent_id_get - Get ID of a switch
+ *     @dev: port device
+ *     @psid: switch ID
+ *
+ *     Get ID of a switch this port is part of.
+ */
+int netdev_switch_parent_id_get(struct net_device *dev,
+                               struct netdev_phys_item_id *psid)
+{
+       const struct swdev_ops *ops = dev->swdev_ops;
+
+       if (!ops || !ops->swdev_parent_id_get)
+               return -EOPNOTSUPP;
+       return ops->swdev_parent_id_get(dev, psid);
+}
+EXPORT_SYMBOL_GPL(netdev_switch_parent_id_get);
+
+/**
+ *     netdev_switch_port_stp_update - Notify switch device port of STP
+ *                                     state change
+ *     @dev: port device
+ *     @state: port STP state
+ *
+ *     Notify switch device port of bridge port STP state change.
+ */
+int netdev_switch_port_stp_update(struct net_device *dev, u8 state)
+{
+       const struct swdev_ops *ops = dev->swdev_ops;
+       struct net_device *lower_dev;
+       struct list_head *iter;
+       int err = -EOPNOTSUPP;
+
+       if (ops && ops->swdev_port_stp_update)
+               return ops->swdev_port_stp_update(dev, state);
+
+       netdev_for_each_lower_dev(dev, lower_dev, iter) {
+               err = netdev_switch_port_stp_update(lower_dev, state);
+               if (err && err != -EOPNOTSUPP)
+                       return err;
+       }
+
+       return err;
+}
+EXPORT_SYMBOL_GPL(netdev_switch_port_stp_update);
+
+static DEFINE_MUTEX(netdev_switch_mutex);
+static RAW_NOTIFIER_HEAD(netdev_switch_notif_chain);
+
+/**
+ *     register_netdev_switch_notifier - Register notifier
+ *     @nb: notifier_block
+ *
+ *     Register switch device notifier. This should be used by code
+ *     which needs to monitor events happening in particular device.
+ *     Return values are same as for atomic_notifier_chain_register().
+ */
+int register_netdev_switch_notifier(struct notifier_block *nb)
+{
+       int err;
+
+       mutex_lock(&netdev_switch_mutex);
+       err = raw_notifier_chain_register(&netdev_switch_notif_chain, nb);
+       mutex_unlock(&netdev_switch_mutex);
+       return err;
+}
+EXPORT_SYMBOL_GPL(register_netdev_switch_notifier);
+
+/**
+ *     unregister_netdev_switch_notifier - Unregister notifier
+ *     @nb: notifier_block
+ *
+ *     Unregister switch device notifier.
+ *     Return values are same as for atomic_notifier_chain_unregister().
+ */
+int unregister_netdev_switch_notifier(struct notifier_block *nb)
+{
+       int err;
+
+       mutex_lock(&netdev_switch_mutex);
+       err = raw_notifier_chain_unregister(&netdev_switch_notif_chain, nb);
+       mutex_unlock(&netdev_switch_mutex);
+       return err;
+}
+EXPORT_SYMBOL_GPL(unregister_netdev_switch_notifier);
+
+/**
+ *     call_netdev_switch_notifiers - Call notifiers
+ *     @val: value passed unmodified to notifier function
+ *     @dev: port device
+ *     @info: notifier information data
+ *
+ *     Call all network notifier blocks. This should be called by driver
+ *     when it needs to propagate hardware event.
+ *     Return values are same as for atomic_notifier_call_chain().
+ */
+int call_netdev_switch_notifiers(unsigned long val, struct net_device *dev,
+                                struct netdev_switch_notifier_info *info)
+{
+       int err;
+
+       info->dev = dev;
+       mutex_lock(&netdev_switch_mutex);
+       err = raw_notifier_call_chain(&netdev_switch_notif_chain, val, info);
+       mutex_unlock(&netdev_switch_mutex);
+       return err;
+}
+EXPORT_SYMBOL_GPL(call_netdev_switch_notifiers);
+
+/**
+ *     netdev_switch_port_bridge_setlink - Notify switch device port of bridge
+ *     port attributes
+ *
+ *     @dev: port device
+ *     @nlh: netlink msg with bridge port attributes
+ *     @flags: bridge setlink flags
+ *
+ *     Notify switch device port of bridge port attributes
+ */
+int netdev_switch_port_bridge_setlink(struct net_device *dev,
+                                     struct nlmsghdr *nlh, u16 flags)
+{
+       const struct net_device_ops *ops = dev->netdev_ops;
+
+       if (!(dev->features & NETIF_F_HW_SWITCH_OFFLOAD))
+               return 0;
+
+       if (!ops->ndo_bridge_setlink)
+               return -EOPNOTSUPP;
+
+       return ops->ndo_bridge_setlink(dev, nlh, flags);
+}
+EXPORT_SYMBOL_GPL(netdev_switch_port_bridge_setlink);
+
+/**
+ *     netdev_switch_port_bridge_dellink - Notify switch device port of bridge
+ *     port attribute delete
+ *
+ *     @dev: port device
+ *     @nlh: netlink msg with bridge port attributes
+ *     @flags: bridge setlink flags
+ *
+ *     Notify switch device port of bridge port attribute delete
+ */
+int netdev_switch_port_bridge_dellink(struct net_device *dev,
+                                     struct nlmsghdr *nlh, u16 flags)
+{
+       const struct net_device_ops *ops = dev->netdev_ops;
+
+       if (!(dev->features & NETIF_F_HW_SWITCH_OFFLOAD))
+               return 0;
+
+       if (!ops->ndo_bridge_dellink)
+               return -EOPNOTSUPP;
+
+       return ops->ndo_bridge_dellink(dev, nlh, flags);
+}
+EXPORT_SYMBOL_GPL(netdev_switch_port_bridge_dellink);
+
+/**
+ *     ndo_dflt_netdev_switch_port_bridge_setlink - default ndo bridge setlink
+ *                                                  op for master devices
+ *
+ *     @dev: port device
+ *     @nlh: netlink msg with bridge port attributes
+ *     @flags: bridge setlink flags
+ *
+ *     Notify master device slaves of bridge port attributes
+ */
+int ndo_dflt_netdev_switch_port_bridge_setlink(struct net_device *dev,
+                                              struct nlmsghdr *nlh, u16 flags)
+{
+       struct net_device *lower_dev;
+       struct list_head *iter;
+       int ret = 0, err = 0;
+
+       if (!(dev->features & NETIF_F_HW_SWITCH_OFFLOAD))
+               return ret;
+
+       netdev_for_each_lower_dev(dev, lower_dev, iter) {
+               err = netdev_switch_port_bridge_setlink(lower_dev, nlh, flags);
+               if (err && err != -EOPNOTSUPP)
+                       ret = err;
+       }
+
+       return ret;
+}
+EXPORT_SYMBOL_GPL(ndo_dflt_netdev_switch_port_bridge_setlink);
+
+/**
+ *     ndo_dflt_netdev_switch_port_bridge_dellink - default ndo bridge dellink
+ *                                                  op for master devices
+ *
+ *     @dev: port device
+ *     @nlh: netlink msg with bridge port attributes
+ *     @flags: bridge dellink flags
+ *
+ *     Notify master device slaves of bridge port attribute deletes
+ */
+int ndo_dflt_netdev_switch_port_bridge_dellink(struct net_device *dev,
+                                              struct nlmsghdr *nlh, u16 flags)
+{
+       struct net_device *lower_dev;
+       struct list_head *iter;
+       int ret = 0, err = 0;
+
+       if (!(dev->features & NETIF_F_HW_SWITCH_OFFLOAD))
+               return ret;
+
+       netdev_for_each_lower_dev(dev, lower_dev, iter) {
+               err = netdev_switch_port_bridge_dellink(lower_dev, nlh, flags);
+               if (err && err != -EOPNOTSUPP)
+                       ret = err;
+       }
+
+       return ret;
+}
+EXPORT_SYMBOL_GPL(ndo_dflt_netdev_switch_port_bridge_dellink);
+
+static struct net_device *netdev_switch_get_lowest_dev(struct net_device *dev)
+{
+       const struct swdev_ops *ops = dev->swdev_ops;
+       struct net_device *lower_dev;
+       struct net_device *port_dev;
+       struct list_head *iter;
+
+       /* Recusively search down until we find a sw port dev.
+        * (A sw port dev supports swdev_parent_id_get).
+        */
+
+       if (dev->features & NETIF_F_HW_SWITCH_OFFLOAD &&
+           ops && ops->swdev_parent_id_get)
+               return dev;
+
+       netdev_for_each_lower_dev(dev, lower_dev, iter) {
+               port_dev = netdev_switch_get_lowest_dev(lower_dev);
+               if (port_dev)
+                       return port_dev;
+       }
+
+       return NULL;
+}
+
+static struct net_device *netdev_switch_get_dev_by_nhs(struct fib_info *fi)
+{
+       struct netdev_phys_item_id psid;
+       struct netdev_phys_item_id prev_psid;
+       struct net_device *dev = NULL;
+       int nhsel;
+
+       /* For this route, all nexthop devs must be on the same switch. */
+
+       for (nhsel = 0; nhsel < fi->fib_nhs; nhsel++) {
+               const struct fib_nh *nh = &fi->fib_nh[nhsel];
+
+               if (!nh->nh_dev)
+                       return NULL;
+
+               dev = netdev_switch_get_lowest_dev(nh->nh_dev);
+               if (!dev)
+                       return NULL;
+
+               if (netdev_switch_parent_id_get(dev, &psid))
+                       return NULL;
+
+               if (nhsel > 0) {
+                       if (prev_psid.id_len != psid.id_len)
+                               return NULL;
+                       if (memcmp(prev_psid.id, psid.id, psid.id_len))
+                               return NULL;
+               }
+
+               prev_psid = psid;
+       }
+
+       return dev;
+}
+
+/**
+ *     netdev_switch_fib_ipv4_add - Add IPv4 route entry to switch
+ *
+ *     @dst: route's IPv4 destination address
+ *     @dst_len: destination address length (prefix length)
+ *     @fi: route FIB info structure
+ *     @tos: route TOS
+ *     @type: route type
+ *     @nlflags: netlink flags passed in (NLM_F_*)
+ *     @tb_id: route table ID
+ *
+ *     Add IPv4 route entry to switch device.
+ */
+int netdev_switch_fib_ipv4_add(u32 dst, int dst_len, struct fib_info *fi,
+                              u8 tos, u8 type, u32 nlflags, u32 tb_id)
+{
+       struct net_device *dev;
+       const struct swdev_ops *ops;
+       int err = 0;
+
+       /* Don't offload route if using custom ip rules or if
+        * IPv4 FIB offloading has been disabled completely.
+        */
+
+#ifdef CONFIG_IP_MULTIPLE_TABLES
+       if (fi->fib_net->ipv4.fib_has_custom_rules)
+               return 0;
+#endif
+
+       if (fi->fib_net->ipv4.fib_offload_disabled)
+               return 0;
+
+       dev = netdev_switch_get_dev_by_nhs(fi);
+       if (!dev)
+               return 0;
+       ops = dev->swdev_ops;
+
+       if (ops->swdev_fib_ipv4_add) {
+               err = ops->swdev_fib_ipv4_add(dev, htonl(dst), dst_len,
+                                             fi, tos, type, nlflags,
+                                             tb_id);
+               if (!err)
+                       fi->fib_flags |= RTNH_F_OFFLOAD;
+       }
+
+       return err;
+}
+EXPORT_SYMBOL_GPL(netdev_switch_fib_ipv4_add);
+
+/**
+ *     netdev_switch_fib_ipv4_del - Delete IPv4 route entry from switch
+ *
+ *     @dst: route's IPv4 destination address
+ *     @dst_len: destination address length (prefix length)
+ *     @fi: route FIB info structure
+ *     @tos: route TOS
+ *     @type: route type
+ *     @tb_id: route table ID
+ *
+ *     Delete IPv4 route entry from switch device.
+ */
+int netdev_switch_fib_ipv4_del(u32 dst, int dst_len, struct fib_info *fi,
+                              u8 tos, u8 type, u32 tb_id)
+{
+       struct net_device *dev;
+       const struct swdev_ops *ops;
+       int err = 0;
+
+       if (!(fi->fib_flags & RTNH_F_OFFLOAD))
+               return 0;
+
+       dev = netdev_switch_get_dev_by_nhs(fi);
+       if (!dev)
+               return 0;
+       ops = dev->swdev_ops;
+
+       if (ops->swdev_fib_ipv4_del) {
+               err = ops->swdev_fib_ipv4_del(dev, htonl(dst), dst_len,
+                                             fi, tos, type, tb_id);
+               if (!err)
+                       fi->fib_flags &= ~RTNH_F_OFFLOAD;
+       }
+
+       return err;
+}
+EXPORT_SYMBOL_GPL(netdev_switch_fib_ipv4_del);
+
+/**
+ *     netdev_switch_fib_ipv4_abort - Abort an IPv4 FIB operation
+ *
+ *     @fi: route FIB info structure
+ */
+void netdev_switch_fib_ipv4_abort(struct fib_info *fi)
+{
+       /* There was a problem installing this route to the offload
+        * device.  For now, until we come up with more refined
+        * policy handling, abruptly end IPv4 fib offloading for
+        * for entire net by flushing offload device(s) of all
+        * IPv4 routes, and mark IPv4 fib offloading broken from
+        * this point forward.
+        */
+
+       fib_flush_external(fi->fib_net);
+       fi->fib_net->ipv4.fib_offload_disabled = true;
+}
+EXPORT_SYMBOL_GPL(netdev_switch_fib_ipv4_abort);