From a93e3b17227ed8b0db7e44d0302b4da7d07f9a35 Mon Sep 17 00:00:00 2001 From: Petr Machata Date: Thu, 22 Nov 2018 23:28:25 +0000 Subject: switchdev: Add a blocking notifier chain In general one can't assume that a switchdev notifier is called in a non-atomic context, and correspondingly, the switchdev notifier chain is an atomic one. However, port object addition and deletion messages are delivered from a process context. Even the MDB addition messages, whose delivery is scheduled from atomic context, are queued and the delivery itself takes place in blocking context. For VLAN messages in particular, keeping the blocking nature is important for error reporting. Therefore introduce a blocking notifier chain and related service functions to distribute the notifications for which a blocking context can be assumed. Signed-off-by: Petr Machata Reviewed-by: Jiri Pirko Reviewed-by: Ido Schimmel Signed-off-by: David S. Miller --- net/switchdev/switchdev.c | 26 ++++++++++++++++++++++++++ 1 file changed, 26 insertions(+) (limited to 'net') diff --git a/net/switchdev/switchdev.c b/net/switchdev/switchdev.c index 74b9d916a58b..e109bb97ce3f 100644 --- a/net/switchdev/switchdev.c +++ b/net/switchdev/switchdev.c @@ -535,6 +535,7 @@ int switchdev_port_obj_del(struct net_device *dev, EXPORT_SYMBOL_GPL(switchdev_port_obj_del); static ATOMIC_NOTIFIER_HEAD(switchdev_notif_chain); +static BLOCKING_NOTIFIER_HEAD(switchdev_blocking_notif_chain); /** * register_switchdev_notifier - Register notifier @@ -576,6 +577,31 @@ int call_switchdev_notifiers(unsigned long val, struct net_device *dev, } EXPORT_SYMBOL_GPL(call_switchdev_notifiers); +int register_switchdev_blocking_notifier(struct notifier_block *nb) +{ + struct blocking_notifier_head *chain = &switchdev_blocking_notif_chain; + + return blocking_notifier_chain_register(chain, nb); +} +EXPORT_SYMBOL_GPL(register_switchdev_blocking_notifier); + +int unregister_switchdev_blocking_notifier(struct notifier_block *nb) +{ + struct blocking_notifier_head *chain = &switchdev_blocking_notif_chain; + + return blocking_notifier_chain_unregister(chain, nb); +} +EXPORT_SYMBOL_GPL(unregister_switchdev_blocking_notifier); + +int call_switchdev_blocking_notifiers(unsigned long val, struct net_device *dev, + struct switchdev_notifier_info *info) +{ + info->dev = dev; + return blocking_notifier_call_chain(&switchdev_blocking_notif_chain, + val, info); +} +EXPORT_SYMBOL_GPL(call_switchdev_blocking_notifiers); + bool switchdev_port_same_parent_id(struct net_device *a, struct net_device *b) { -- cgit v1.2.3 From 2b239f678079fd8423c3cd7c5b804abadd2fe898 Mon Sep 17 00:00:00 2001 From: Petr Machata Date: Thu, 22 Nov 2018 23:29:05 +0000 Subject: net: dsa: slave: Handle SWITCHDEV_PORT_OBJ_ADD/_DEL Following patches will change the way of distributing port object changes from a switchdev operation to a switchdev notifier. The switchdev code currently recursively descends through layers of lower devices, eventually calling the op on a front-panel port device. The notifier will instead be sent referencing the bridge port device, which may be a stacking device that's one of front-panel ports uppers, or a completely unrelated device. DSA currently doesn't support any other uppers than bridge. SWITCHDEV_OBJ_ID_HOST_MDB and _PORT_MDB objects are always notified on the bridge port device. Thus the only case that a stacked device could be validly referenced by port object notifications are bridge notifications for VLAN objects added to the bridge itself. But the driver explicitly rejects such notifications in dsa_port_vlan_add(). It is therefore safe to assume that the only interesting case is that the notification is on a front-panel port netdevice. Therefore keep the filtering by dsa_slave_dev_check() in place. To handle SWITCHDEV_PORT_OBJ_ADD and _DEL, subscribe to the blocking notifier chain. Dispatch to rocker_port_obj_add() resp. _del() to maintain the behavior that the switchdev operation based code currently has. Signed-off-by: Petr Machata Acked-by: Jiri Pirko Signed-off-by: David S. Miller --- net/dsa/slave.c | 56 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 56 insertions(+) (limited to 'net') diff --git a/net/dsa/slave.c b/net/dsa/slave.c index 7d0c19e7edcf..d00a0b6d4ce0 100644 --- a/net/dsa/slave.c +++ b/net/dsa/slave.c @@ -1557,6 +1557,44 @@ err_fdb_work_init: return NOTIFY_BAD; } +static int +dsa_slave_switchdev_port_obj_event(unsigned long event, + struct net_device *netdev, + struct switchdev_notifier_port_obj_info *port_obj_info) +{ + int err = -EOPNOTSUPP; + + switch (event) { + case SWITCHDEV_PORT_OBJ_ADD: + err = dsa_slave_port_obj_add(netdev, port_obj_info->obj, + port_obj_info->trans); + break; + case SWITCHDEV_PORT_OBJ_DEL: + err = dsa_slave_port_obj_del(netdev, port_obj_info->obj); + break; + } + + port_obj_info->handled = true; + return notifier_from_errno(err); +} + +static int dsa_slave_switchdev_blocking_event(struct notifier_block *unused, + unsigned long event, void *ptr) +{ + struct net_device *dev = switchdev_notifier_info_to_dev(ptr); + + if (!dsa_slave_dev_check(dev)) + return NOTIFY_DONE; + + switch (event) { + case SWITCHDEV_PORT_OBJ_ADD: /* fall through */ + case SWITCHDEV_PORT_OBJ_DEL: + return dsa_slave_switchdev_port_obj_event(event, dev, ptr); + } + + return NOTIFY_DONE; +} + static struct notifier_block dsa_slave_nb __read_mostly = { .notifier_call = dsa_slave_netdevice_event, }; @@ -1565,8 +1603,13 @@ static struct notifier_block dsa_slave_switchdev_notifier = { .notifier_call = dsa_slave_switchdev_event, }; +static struct notifier_block dsa_slave_switchdev_blocking_notifier = { + .notifier_call = dsa_slave_switchdev_blocking_event, +}; + int dsa_slave_register_notifier(void) { + struct notifier_block *nb; int err; err = register_netdevice_notifier(&dsa_slave_nb); @@ -1577,8 +1620,15 @@ int dsa_slave_register_notifier(void) if (err) goto err_switchdev_nb; + nb = &dsa_slave_switchdev_blocking_notifier; + err = register_switchdev_blocking_notifier(nb); + if (err) + goto err_switchdev_blocking_nb; + return 0; +err_switchdev_blocking_nb: + unregister_switchdev_notifier(&dsa_slave_switchdev_notifier); err_switchdev_nb: unregister_netdevice_notifier(&dsa_slave_nb); return err; @@ -1586,8 +1636,14 @@ err_switchdev_nb: void dsa_slave_unregister_notifier(void) { + struct notifier_block *nb; int err; + nb = &dsa_slave_switchdev_blocking_notifier; + err = unregister_switchdev_blocking_notifier(nb); + if (err) + pr_err("DSA: failed to unregister switchdev blocking notifier (%d)\n", err); + err = unregister_switchdev_notifier(&dsa_slave_switchdev_notifier); if (err) pr_err("DSA: failed to unregister switchdev notifier (%d)\n", err); -- cgit v1.2.3 From f30f0601eb934dda107decd2e57b37168096fd74 Mon Sep 17 00:00:00 2001 From: Petr Machata Date: Thu, 22 Nov 2018 23:29:44 +0000 Subject: switchdev: Add helpers to aid traversal through lower devices After the transition from switchdev operations to notifier chain (which will take place in following patches), the onus is on the driver to find its own devices below possible layer of LAG or other uppers. The logic to do so is fairly repetitive: each driver is looking for its own devices among the lowers of the notified device. For those that it finds, it calls a handler. To indicate that the event was handled, struct switchdev_notifier_port_obj_info.handled is set. The differences lie only in what constitutes an "own" device and what handler to call. Therefore abstract this logic into two helpers, switchdev_handle_port_obj_add() and switchdev_handle_port_obj_del(). If a driver only supports physical ports under a bridge device, it will simply avoid this layer of indirection. One area where this helper diverges from the current switchdev behavior is the case of mixed lowers, some of which are switchdev ports and some of which are not. Previously, such scenario would fail with -EOPNOTSUPP. The helper could do that for lowers for which the passed-in predicate doesn't hold. That would however break the case that switchdev ports from several different drivers are stashed under one master, a scenario that switchdev currently happily supports. Therefore tolerate any and all unknown netdevices, whether they are backed by a switchdev driver or not. Signed-off-by: Petr Machata Reviewed-by: Ido Schimmel Signed-off-by: David S. Miller --- net/switchdev/switchdev.c | 100 ++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 100 insertions(+) (limited to 'net') diff --git a/net/switchdev/switchdev.c b/net/switchdev/switchdev.c index e109bb97ce3f..099434ec7996 100644 --- a/net/switchdev/switchdev.c +++ b/net/switchdev/switchdev.c @@ -621,3 +621,103 @@ bool switchdev_port_same_parent_id(struct net_device *a, return netdev_phys_item_id_same(&a_attr.u.ppid, &b_attr.u.ppid); } EXPORT_SYMBOL_GPL(switchdev_port_same_parent_id); + +static int __switchdev_handle_port_obj_add(struct net_device *dev, + struct switchdev_notifier_port_obj_info *port_obj_info, + bool (*check_cb)(const struct net_device *dev), + int (*add_cb)(struct net_device *dev, + const struct switchdev_obj *obj, + struct switchdev_trans *trans)) +{ + struct net_device *lower_dev; + struct list_head *iter; + int err = -EOPNOTSUPP; + + if (check_cb(dev)) { + /* This flag is only checked if the return value is success. */ + port_obj_info->handled = true; + return add_cb(dev, port_obj_info->obj, port_obj_info->trans); + } + + /* Switch ports might be stacked under e.g. a LAG. Ignore the + * unsupported devices, another driver might be able to handle them. But + * propagate to the callers any hard errors. + * + * If the driver does its own bookkeeping of stacked ports, it's not + * necessary to go through this helper. + */ + netdev_for_each_lower_dev(dev, lower_dev, iter) { + err = __switchdev_handle_port_obj_add(lower_dev, port_obj_info, + check_cb, add_cb); + if (err && err != -EOPNOTSUPP) + return err; + } + + return err; +} + +int switchdev_handle_port_obj_add(struct net_device *dev, + struct switchdev_notifier_port_obj_info *port_obj_info, + bool (*check_cb)(const struct net_device *dev), + int (*add_cb)(struct net_device *dev, + const struct switchdev_obj *obj, + struct switchdev_trans *trans)) +{ + int err; + + err = __switchdev_handle_port_obj_add(dev, port_obj_info, check_cb, + add_cb); + if (err == -EOPNOTSUPP) + err = 0; + return err; +} +EXPORT_SYMBOL_GPL(switchdev_handle_port_obj_add); + +static int __switchdev_handle_port_obj_del(struct net_device *dev, + struct switchdev_notifier_port_obj_info *port_obj_info, + bool (*check_cb)(const struct net_device *dev), + int (*del_cb)(struct net_device *dev, + const struct switchdev_obj *obj)) +{ + struct net_device *lower_dev; + struct list_head *iter; + int err = -EOPNOTSUPP; + + if (check_cb(dev)) { + /* This flag is only checked if the return value is success. */ + port_obj_info->handled = true; + return del_cb(dev, port_obj_info->obj); + } + + /* Switch ports might be stacked under e.g. a LAG. Ignore the + * unsupported devices, another driver might be able to handle them. But + * propagate to the callers any hard errors. + * + * If the driver does its own bookkeeping of stacked ports, it's not + * necessary to go through this helper. + */ + netdev_for_each_lower_dev(dev, lower_dev, iter) { + err = __switchdev_handle_port_obj_del(lower_dev, port_obj_info, + check_cb, del_cb); + if (err && err != -EOPNOTSUPP) + return err; + } + + return err; +} + +int switchdev_handle_port_obj_del(struct net_device *dev, + struct switchdev_notifier_port_obj_info *port_obj_info, + bool (*check_cb)(const struct net_device *dev), + int (*del_cb)(struct net_device *dev, + const struct switchdev_obj *obj)) +{ + int err; + + err = __switchdev_handle_port_obj_del(dev, port_obj_info, check_cb, + del_cb); + if (err == -EOPNOTSUPP) + err = 0; + return err; +} +EXPORT_SYMBOL_GPL(switchdev_handle_port_obj_del); -- cgit v1.2.3 From d17d9f5e5143125f9274194d8f7368f76b9d141f Mon Sep 17 00:00:00 2001 From: Petr Machata Date: Thu, 22 Nov 2018 23:32:57 +0000 Subject: switchdev: Replace port obj add/del SDO with a notification Drop switchdev_ops.switchdev_port_obj_add and _del. Drop the uses of this field from all clients, which were migrated to use switchdev notification in the previous patches. Add a new function switchdev_port_obj_notify() that sends the switchdev notifications SWITCHDEV_PORT_OBJ_ADD and _DEL. Update switchdev_port_obj_del_now() to dispatch to this new function. Drop __switchdev_port_obj_add() and update switchdev_port_obj_add() likewise. Signed-off-by: Petr Machata Reviewed-by: Ido Schimmel Signed-off-by: David S. Miller --- net/dsa/slave.c | 2 -- net/switchdev/switchdev.c | 67 ++++++++++++++++++----------------------------- 2 files changed, 25 insertions(+), 44 deletions(-) (limited to 'net') diff --git a/net/dsa/slave.c b/net/dsa/slave.c index d00a0b6d4ce0..268119cf7117 100644 --- a/net/dsa/slave.c +++ b/net/dsa/slave.c @@ -1050,8 +1050,6 @@ static const struct net_device_ops dsa_slave_netdev_ops = { static const struct switchdev_ops dsa_slave_switchdev_ops = { .switchdev_port_attr_get = dsa_slave_port_attr_get, .switchdev_port_attr_set = dsa_slave_port_attr_set, - .switchdev_port_obj_add = dsa_slave_port_obj_add, - .switchdev_port_obj_del = dsa_slave_port_obj_del, }; static struct device_type dsa_type = { diff --git a/net/switchdev/switchdev.c b/net/switchdev/switchdev.c index 099434ec7996..fe23fac4dc4b 100644 --- a/net/switchdev/switchdev.c +++ b/net/switchdev/switchdev.c @@ -353,30 +353,29 @@ static size_t switchdev_obj_size(const struct switchdev_obj *obj) return 0; } -static int __switchdev_port_obj_add(struct net_device *dev, - const struct switchdev_obj *obj, - struct switchdev_trans *trans) +static int switchdev_port_obj_notify(enum switchdev_notifier_type nt, + struct net_device *dev, + const struct switchdev_obj *obj, + struct switchdev_trans *trans) { - const struct switchdev_ops *ops = dev->switchdev_ops; - struct net_device *lower_dev; - struct list_head *iter; - int err = -EOPNOTSUPP; - - if (ops && ops->switchdev_port_obj_add) - return ops->switchdev_port_obj_add(dev, obj, trans); + int rc; + int err; - /* Switch device port(s) may be stacked under - * bond/team/vlan dev, so recurse down to add object on - * each port. - */ + struct switchdev_notifier_port_obj_info obj_info = { + .obj = obj, + .trans = trans, + .handled = false, + }; - netdev_for_each_lower_dev(dev, lower_dev, iter) { - err = __switchdev_port_obj_add(lower_dev, obj, trans); - if (err) - break; + rc = call_switchdev_blocking_notifiers(nt, dev, &obj_info.info); + err = notifier_to_errno(rc); + if (err) { + WARN_ON(!obj_info.handled); + return err; } - - return err; + if (!obj_info.handled) + return -EOPNOTSUPP; + return 0; } static int switchdev_port_obj_add_now(struct net_device *dev, @@ -397,7 +396,8 @@ static int switchdev_port_obj_add_now(struct net_device *dev, */ trans.ph_prepare = true; - err = __switchdev_port_obj_add(dev, obj, &trans); + err = switchdev_port_obj_notify(SWITCHDEV_PORT_OBJ_ADD, + dev, obj, &trans); if (err) { /* Prepare phase failed: abort the transaction. Any * resources reserved in the prepare phase are @@ -416,7 +416,8 @@ static int switchdev_port_obj_add_now(struct net_device *dev, */ trans.ph_prepare = false; - err = __switchdev_port_obj_add(dev, obj, &trans); + err = switchdev_port_obj_notify(SWITCHDEV_PORT_OBJ_ADD, + dev, obj, &trans); WARN(err, "%s: Commit of object (id=%d) failed.\n", dev->name, obj->id); switchdev_trans_items_warn_destroy(dev, &trans); @@ -471,26 +472,8 @@ EXPORT_SYMBOL_GPL(switchdev_port_obj_add); static int switchdev_port_obj_del_now(struct net_device *dev, const struct switchdev_obj *obj) { - const struct switchdev_ops *ops = dev->switchdev_ops; - struct net_device *lower_dev; - struct list_head *iter; - int err = -EOPNOTSUPP; - - if (ops && ops->switchdev_port_obj_del) - return ops->switchdev_port_obj_del(dev, obj); - - /* Switch device port(s) may be stacked under - * bond/team/vlan dev, so recurse down to delete object on - * each port. - */ - - netdev_for_each_lower_dev(dev, lower_dev, iter) { - err = switchdev_port_obj_del_now(lower_dev, obj); - if (err) - break; - } - - return err; + return switchdev_port_obj_notify(SWITCHDEV_PORT_OBJ_DEL, + dev, obj, NULL); } static void switchdev_port_obj_del_deferred(struct net_device *dev, -- cgit v1.2.3 From ab4a16869f25648570976a4d55923d0c9e7effb8 Mon Sep 17 00:00:00 2001 From: Petr Machata Date: Thu, 22 Nov 2018 23:48:05 +0000 Subject: rocker, dsa, ethsw: Don't filter VLAN events on bridge itself Due to an explicit check in rocker_world_port_obj_vlan_add(), dsa_slave_switchdev_event() resp. port_switchdev_event(), VLAN objects that are added to a device that is not a front-panel port device are ignored. Therefore this check is immaterial. Signed-off-by: Petr Machata Acked-by: Jiri Pirko Signed-off-by: David S. Miller --- net/dsa/port.c | 3 --- 1 file changed, 3 deletions(-) (limited to 'net') diff --git a/net/dsa/port.c b/net/dsa/port.c index ed0595459df1..2d7e01b23572 100644 --- a/net/dsa/port.c +++ b/net/dsa/port.c @@ -252,9 +252,6 @@ int dsa_port_vlan_add(struct dsa_port *dp, .vlan = vlan, }; - if (netif_is_bridge_master(vlan->obj.orig_dev)) - return -EOPNOTSUPP; - if (br_vlan_enabled(dp->bridge_dev)) return dsa_port_notify(dp, DSA_NOTIFIER_VLAN_ADD, &info); -- cgit v1.2.3