From patchwork Thu May 8 15:13:15 2014 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Antonio Quartulli X-Patchwork-Id: 3958 Return-Path: Received-SPF: Pass (sender SPF authorized) identity=mailfrom; client-ip=178.209.62.157; helo=s3.neomailbox.net; envelope-from=antonio@meshcoding.com; receiver=b.a.t.m.a.n@lists.open-mesh.org Received: from s3.neomailbox.net (s3.neomailbox.net [178.209.62.157]) by open-mesh.org (Postfix) with ESMTPS id 49021600888 for ; Thu, 8 May 2014 17:14:17 +0200 (CEST) From: Antonio Quartulli To: b.a.t.m.a.n@lists.open-mesh.org Date: Thu, 8 May 2014 17:13:15 +0200 Message-Id: <1399561995-10066-1-git-send-email-antonio@meshcoding.com> Cc: Antonio Quartulli Subject: [B.A.T.M.A.N.] [PATCHv5 maint] batman-adv: fix TT VLAN inconsistency on VLAN re-add X-BeenThere: b.a.t.m.a.n@lists.open-mesh.org X-Mailman-Version: 2.1.15 Precedence: list Reply-To: The list for a Better Approach To Mobile Ad-hoc Networking List-Id: The list for a Better Approach To Mobile Ad-hoc Networking List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 08 May 2014 15:14:17 -0000 From: Antonio Quartulli When a VLAN interface (on top of batX) is removed and re-added within a short timeframe TT does not have enough time to properly cleanup. This creates an internal TT state mismatch as the newly created softif_vlan will be initialized from scratch with a TT client count of zero (even if TT entries for this VLAN still exist). The resulting TT messages are bogus due to the counter / tt client listing mismatch, thus creating inconsistencies on every node in the network To fix this issue destroy_vlan() has to not free the VLAN object immediately but it has to be kept alive until all the TT entries for this VLAN have been removed. destroy_vlan() still removes the sysfs folder so that the user has the feeling that everything went fine. If the same VLAN is re-added before the old object is free'd, then the latter is resurrected and re-used. Implement such behaviour by increasing the reference counter of a softif_vlan object every time a new local TT entry for such VLAN is created and remove the object from the list only when all the TT entries have been destroyed. Signed-off-by: Antonio Quartulli --- Changes since v4: - improved code in add_vid() - don't leak TT entries in case of vlan re-add failure Changes since v3: - always re-add NO_PURGE local entry on add_vid() Changes since v2: - remove cleanup_work member that is not needed anymore in this approach - reword commit subject - reword commit message (Thanks Marek!) Changes since v1: - destroy and create vlan sysfs folder within softif_vlan_destroy/create() to avoid lock troubles with soft-interface destruction and delayed jobs. soft-interface.c | 60 ++++++++++++++++++++++++++++++++++++++++------------- translation-table.c | 26 +++++++++++++++++++++++ types.h | 2 ++ 3 files changed, 74 insertions(+), 14 deletions(-) diff --git a/soft-interface.c b/soft-interface.c index d962363..5a9dc2f 100644 --- a/soft-interface.c +++ b/soft-interface.c @@ -437,10 +437,15 @@ out: * possibly free it * @softif_vlan: the vlan object to release */ -void batadv_softif_vlan_free_ref(struct batadv_softif_vlan *softif_vlan) +void batadv_softif_vlan_free_ref(struct batadv_softif_vlan *vlan) { - if (atomic_dec_and_test(&softif_vlan->refcount)) - kfree_rcu(softif_vlan, rcu); + if (atomic_dec_and_test(&vlan->refcount)) { + spin_lock_bh(&vlan->bat_priv->softif_vlan_list_lock); + hlist_del_rcu(&vlan->list); + spin_unlock_bh(&vlan->bat_priv->softif_vlan_list_lock); + + kfree_rcu(vlan, rcu); + } } /** @@ -494,6 +499,7 @@ int batadv_softif_create_vlan(struct batadv_priv *bat_priv, unsigned short vid) if (!vlan) return -ENOMEM; + vlan->bat_priv = bat_priv; vlan->vid = vid; atomic_set(&vlan->refcount, 1); @@ -505,6 +511,10 @@ int batadv_softif_create_vlan(struct batadv_priv *bat_priv, unsigned short vid) return err; } + spin_lock_bh(&bat_priv->softif_vlan_list_lock); + hlist_add_head_rcu(&vlan->list, &bat_priv->softif_vlan_list); + spin_unlock_bh(&bat_priv->softif_vlan_list_lock); + /* add a new TT local entry. This one will be marked with the NOPURGE * flag */ @@ -512,10 +522,6 @@ int batadv_softif_create_vlan(struct batadv_priv *bat_priv, unsigned short vid) bat_priv->soft_iface->dev_addr, vid, BATADV_NULL_IFINDEX, BATADV_NO_MARK); - spin_lock_bh(&bat_priv->softif_vlan_list_lock); - hlist_add_head_rcu(&vlan->list, &bat_priv->softif_vlan_list); - spin_unlock_bh(&bat_priv->softif_vlan_list_lock); - return 0; } @@ -527,18 +533,13 @@ int batadv_softif_create_vlan(struct batadv_priv *bat_priv, unsigned short vid) static void batadv_softif_destroy_vlan(struct batadv_priv *bat_priv, struct batadv_softif_vlan *vlan) { - spin_lock_bh(&bat_priv->softif_vlan_list_lock); - hlist_del_rcu(&vlan->list); - spin_unlock_bh(&bat_priv->softif_vlan_list_lock); - - batadv_sysfs_del_vlan(bat_priv, vlan); - /* explicitly remove the associated TT local entry because it is marked * with the NOPURGE flag */ batadv_tt_local_remove(bat_priv, bat_priv->soft_iface->dev_addr, vlan->vid, "vlan interface destroyed", false); + batadv_sysfs_del_vlan(bat_priv, vlan); batadv_softif_vlan_free_ref(vlan); } @@ -556,6 +557,8 @@ static int batadv_interface_add_vid(struct net_device *dev, __be16 proto, unsigned short vid) { struct batadv_priv *bat_priv = netdev_priv(dev); + struct batadv_softif_vlan *vlan; + int ret; /* only 802.1Q vlans are supported. * batman-adv does not know how to handle other types @@ -565,7 +568,36 @@ static int batadv_interface_add_vid(struct net_device *dev, __be16 proto, vid |= BATADV_VLAN_HAS_TAG; - return batadv_softif_create_vlan(bat_priv, vid); + /* if a new vlan is getting created and it already exists, it means that + * it was not deleted yet. batadv_softif_vlan_get() increases the + * refcount in order to revive the object. + * + * if it does not exist then create it. + */ + vlan = batadv_softif_vlan_get(bat_priv, vid); + if (!vlan) + return batadv_softif_create_vlan(bat_priv, vid); + + /* recreate the sysfs object if it was already destroyed (and it should + * be since we received a kill_vid() for this vlan + */ + if (!vlan->kobj) { + ret = batadv_sysfs_add_vlan(bat_priv->soft_iface, vlan); + if (ret) { + batadv_softif_vlan_free_ref(vlan); + return ret; + } + } + + /* add a new TT local entry. This one will be marked with the NOPURGE + * flag. This must be added again, even if the vlan object already + * exists, because the entry was deleted by kill_vid() + */ + batadv_tt_local_add(bat_priv->soft_iface, + bat_priv->soft_iface->dev_addr, vid, + BATADV_NULL_IFINDEX, BATADV_NO_MARK); + + return 0; } /** diff --git a/translation-table.c b/translation-table.c index 959dde7..b689981 100644 --- a/translation-table.c +++ b/translation-table.c @@ -485,6 +485,7 @@ bool batadv_tt_local_add(struct net_device *soft_iface, const uint8_t *addr, struct batadv_priv *bat_priv = netdev_priv(soft_iface); struct batadv_tt_local_entry *tt_local; struct batadv_tt_global_entry *tt_global; + struct batadv_softif_vlan *vlan; struct net_device *in_dev = NULL; struct hlist_head *head; struct batadv_tt_orig_list_entry *orig_entry; @@ -544,6 +545,9 @@ bool batadv_tt_local_add(struct net_device *soft_iface, const uint8_t *addr, if (!tt_local) goto out; + /* increase the refcounter of the related vlan */ + vlan = batadv_softif_vlan_get(bat_priv, vid); + batadv_dbg(BATADV_DBG_TT, bat_priv, "Creating new local tt entry: %pM (vid: %d, ttvn: %d)\n", addr, BATADV_PRINT_VID(vid), @@ -573,6 +577,7 @@ bool batadv_tt_local_add(struct net_device *soft_iface, const uint8_t *addr, if (unlikely(hash_added != 0)) { /* remove the reference for the hash */ batadv_tt_local_entry_free_ref(tt_local); + batadv_softif_vlan_free_ref(vlan); goto out; } @@ -978,6 +983,7 @@ uint16_t batadv_tt_local_remove(struct batadv_priv *bat_priv, { struct batadv_tt_local_entry *tt_local_entry; uint16_t flags, curr_flags = BATADV_NO_FLAGS; + struct batadv_softif_vlan *vlan; tt_local_entry = batadv_tt_local_hash_find(bat_priv, addr, vid); if (!tt_local_entry) @@ -1008,6 +1014,11 @@ uint16_t batadv_tt_local_remove(struct batadv_priv *bat_priv, hlist_del_rcu(&tt_local_entry->common.hash_entry); batadv_tt_local_entry_free_ref(tt_local_entry); + /* decrease the reference held for this vlan */ + vlan = batadv_softif_vlan_get(bat_priv, vid); + batadv_softif_vlan_free_ref(vlan); + batadv_softif_vlan_free_ref(vlan); + out: if (tt_local_entry) batadv_tt_local_entry_free_ref(tt_local_entry); @@ -1080,6 +1091,7 @@ static void batadv_tt_local_table_free(struct batadv_priv *bat_priv) spinlock_t *list_lock; /* protects write access to the hash lists */ struct batadv_tt_common_entry *tt_common_entry; struct batadv_tt_local_entry *tt_local; + struct batadv_softif_vlan *vlan; struct hlist_node *node_tmp; struct hlist_head *head; uint32_t i; @@ -1100,6 +1112,13 @@ static void batadv_tt_local_table_free(struct batadv_priv *bat_priv) tt_local = container_of(tt_common_entry, struct batadv_tt_local_entry, common); + + /* decrease the reference held for this vlan */ + vlan = batadv_softif_vlan_get(bat_priv, + tt_common_entry->vid); + batadv_softif_vlan_free_ref(vlan); + batadv_softif_vlan_free_ref(vlan); + batadv_tt_local_entry_free_ref(tt_local); } spin_unlock_bh(list_lock); @@ -3078,6 +3097,7 @@ static void batadv_tt_local_purge_pending_clients(struct batadv_priv *bat_priv) struct batadv_hashtable *hash = bat_priv->tt.local_hash; struct batadv_tt_common_entry *tt_common; struct batadv_tt_local_entry *tt_local; + struct batadv_softif_vlan *vlan; struct hlist_node *node_tmp; struct hlist_head *head; spinlock_t *list_lock; /* protects write access to the hash lists */ @@ -3106,6 +3126,12 @@ static void batadv_tt_local_purge_pending_clients(struct batadv_priv *bat_priv) tt_local = container_of(tt_common, struct batadv_tt_local_entry, common); + + /* decrease the reference held for this vlan */ + vlan = batadv_softif_vlan_get(bat_priv, tt_common->vid); + batadv_softif_vlan_free_ref(vlan); + batadv_softif_vlan_free_ref(vlan); + batadv_tt_local_entry_free_ref(tt_local); } spin_unlock_bh(list_lock); diff --git a/types.h b/types.h index 78370ab..c4aba2f 100644 --- a/types.h +++ b/types.h @@ -638,6 +638,7 @@ struct batadv_priv_nc { /** * struct batadv_softif_vlan - per VLAN attributes set + * @bat_priv: pointer to the mesh object * @vid: VLAN identifier * @kobj: kobject for sysfs vlan subdirectory * @ap_isolation: AP isolation state @@ -647,6 +648,7 @@ struct batadv_priv_nc { * @rcu: struct used for freeing in a RCU-safe manner */ struct batadv_softif_vlan { + struct batadv_priv *bat_priv; unsigned short vid; struct kobject *kobj; atomic_t ap_isolation; /* boolean */