B.A.T.M.A.N June 2024

b.a.t.m.a.n@lists.open-mesh.org

8 participants
7 discussions

[PATCH v2] batman-adv: add dynamic, bridged-in TT VID detection support
by Linus Lüssing 14 Nov '24

14 Nov '24

So far, if we wanted to bridge VLAN tagged frames into the mesh one would need to manually create an according VLAN interface on top of bat0 first, to trigger batman-adv to create the according structures for a VID. With this change the VLAN from bridged-in clients is now automatically detected and added to the translation table on the fly. Signed-off-by: Linus Lüssing <linus.luessing(a)c0d3.blue> --- v2: fix a typo, a missing "to" in the commit message net/batman-adv/hard-interface.c | 2 +- net/batman-adv/multicast.c | 8 +- net/batman-adv/soft-interface.c | 121 ++++++++++++++++------------- net/batman-adv/soft-interface.h | 6 +- net/batman-adv/translation-table.c | 19 ++--- net/batman-adv/translation-table.h | 4 +- 6 files changed, 88 insertions(+), 72 deletions(-) diff --git a/net/batman-adv/hard-interface.c b/net/batman-adv/hard-interface.c index 96a412beab2d..f5826dd8752c 100644 --- a/net/batman-adv/hard-interface.c +++ b/net/batman-adv/hard-interface.c @@ -946,7 +946,7 @@ static int batadv_hard_if_event_softif(unsigned long event, switch (event) { case NETDEV_REGISTER: bat_priv = netdev_priv(net_dev); - batadv_softif_create_vlan(bat_priv, BATADV_NO_FLAGS); + batadv_softif_create_vlan_own(bat_priv, BATADV_NO_FLAGS); break; } diff --git a/net/batman-adv/multicast.c b/net/batman-adv/multicast.c index 38fab5e46ae2..61e765352e29 100644 --- a/net/batman-adv/multicast.c +++ b/net/batman-adv/multicast.c @@ -724,6 +724,7 @@ static void batadv_mcast_mla_tt_add(struct batadv_priv *bat_priv, { struct batadv_hw_addr *mcast_entry; struct hlist_node *tmp; + int ret; if (!mcast_list) return; @@ -733,9 +734,10 @@ static void batadv_mcast_mla_tt_add(struct batadv_priv *bat_priv, &bat_priv->mcast.mla_list)) continue; - if (!batadv_tt_local_add(bat_priv->soft_iface, - mcast_entry->addr, BATADV_NO_FLAGS, - BATADV_NULL_IFINDEX, BATADV_NO_MARK)) + ret = batadv_tt_local_add(bat_priv->soft_iface, + mcast_entry->addr, BATADV_NO_FLAGS, + BATADV_NULL_IFINDEX, BATADV_NO_MARK); + if (ret <= 0) continue; hlist_del(&mcast_entry->list); diff --git a/net/batman-adv/soft-interface.c b/net/batman-adv/soft-interface.c index 30ecbc2ef1fd..fb647798e5c8 100644 --- a/net/batman-adv/soft-interface.c +++ b/net/batman-adv/soft-interface.c @@ -141,6 +141,10 @@ static int batadv_interface_set_mac_addr(struct net_device *dev, void *p) rcu_read_lock(); hlist_for_each_entry_rcu(vlan, &bat_priv->softif_vlan_list, list) { + /* we don't use this VID ourself, avoid adding us to it */ + if (!batadv_is_my_client(bat_priv, old_addr, vlan->vid)) + continue; + batadv_tt_local_remove(bat_priv, old_addr, vlan->vid, "mac address changed", false); batadv_tt_local_add(dev, addr->sa_data, vlan->vid, @@ -549,13 +553,15 @@ struct batadv_softif_vlan *batadv_softif_vlan_get(struct batadv_priv *bat_priv, } /** - * batadv_softif_create_vlan() - allocate the needed resources for a new vlan + * batadv_softif_create_vlan() - create a softif vlan struct * @bat_priv: the bat priv with all the soft interface information * @vid: the VLAN identifier * - * Return: 0 on success, a negative error otherwise. + * Return: a pointer to the newly allocated softif vlan struct on success, NULL + * otherwise. */ -int batadv_softif_create_vlan(struct batadv_priv *bat_priv, unsigned short vid) +static struct batadv_softif_vlan * +batadv_softif_create_vlan(struct batadv_priv *bat_priv, unsigned short vid) { struct batadv_softif_vlan *vlan; @@ -563,15 +569,14 @@ int batadv_softif_create_vlan(struct batadv_priv *bat_priv, unsigned short vid) vlan = batadv_softif_vlan_get(bat_priv, vid); if (vlan) { - batadv_softif_vlan_put(vlan); spin_unlock_bh(&bat_priv->softif_vlan_list_lock); - return -EEXIST; + return vlan; } vlan = kzalloc(sizeof(*vlan), GFP_ATOMIC); if (!vlan) { spin_unlock_bh(&bat_priv->softif_vlan_list_lock); - return -ENOMEM; + return NULL; } vlan->bat_priv = bat_priv; @@ -584,34 +589,71 @@ int batadv_softif_create_vlan(struct batadv_priv *bat_priv, unsigned short vid) hlist_add_head_rcu(&vlan->list, &bat_priv->softif_vlan_list); spin_unlock_bh(&bat_priv->softif_vlan_list_lock); + return vlan; +} + +/** + * batadv_softif_vlan_get_or_create() - retrieve or create a softif vlan struct + * @bat_priv: the bat priv with all the soft interface information + * @vid: the VLAN identifier + * + * Return: the softif vlan struct if found or created or NULL otherwise. + */ +struct batadv_softif_vlan * +batadv_softif_vlan_get_or_create(struct batadv_priv *bat_priv, + unsigned short vid) +{ + struct batadv_softif_vlan *vlan = batadv_softif_vlan_get(bat_priv, vid); + + if (vlan) + return vlan; + + return batadv_softif_create_vlan(bat_priv, vid); +} + +/** + * batadv_softif_create_vlan_own() - add our own softif to the local TT + * @bat_priv: the bat priv with all the soft interface information + * @vid: the VLAN identifier + * + * Adds the MAC address of our own soft interface with the given VLAN ID as + * a permanent local TT entry. + * + * Return: 0 on success, a negative error otherwise. + */ +int batadv_softif_create_vlan_own(struct batadv_priv *bat_priv, + unsigned short vid) +{ + int ret; + /* add a new TT local entry. This one will be marked with the NOPURGE * flag */ - batadv_tt_local_add(bat_priv->soft_iface, - bat_priv->soft_iface->dev_addr, vid, - BATADV_NULL_IFINDEX, BATADV_NO_MARK); - - /* don't return reference to new softif_vlan */ - batadv_softif_vlan_put(vlan); + ret = batadv_tt_local_add(bat_priv->soft_iface, + bat_priv->soft_iface->dev_addr, vid, + BATADV_NULL_IFINDEX, BATADV_NO_MARK); + if (ret < 0) + return ret; return 0; } /** - * batadv_softif_destroy_vlan() - remove and destroy a softif_vlan object + * batadv_softif_destroy_vlan_own() - remove our own softif from the local TT * @bat_priv: the bat priv with all the soft interface information - * @vlan: the object to remove + * @vid: the VLAN identifier + * + * Removes the MAC address of our own soft interface with the given VLAN ID from + * the local TT. */ -static void batadv_softif_destroy_vlan(struct batadv_priv *bat_priv, - struct batadv_softif_vlan *vlan) +static void batadv_softif_destroy_vlan_own(struct batadv_priv *bat_priv, + unsigned short vid) { /* 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_softif_vlan_put(vlan); + batadv_tt_local_remove(bat_priv, bat_priv->soft_iface->dev_addr, vid, + "vlan interface destroyed", false); } /** @@ -629,7 +671,6 @@ 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; /* only 802.1Q vlans are supported. * batman-adv does not know how to handle other types @@ -639,25 +680,7 @@ static int batadv_interface_add_vid(struct net_device *dev, __be16 proto, vid |= BATADV_VLAN_HAS_TAG; - /* 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); - - /* 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; + return batadv_softif_create_vlan_own(bat_priv, vid); } /** @@ -676,7 +699,6 @@ static int batadv_interface_kill_vid(struct net_device *dev, __be16 proto, unsigned short vid) { struct batadv_priv *bat_priv = netdev_priv(dev); - struct batadv_softif_vlan *vlan; /* only 802.1Q vlans are supported. batman-adv does not know how to * handle other types @@ -684,15 +706,7 @@ static int batadv_interface_kill_vid(struct net_device *dev, __be16 proto, if (proto != htons(ETH_P_8021Q)) return -EINVAL; - vlan = batadv_softif_vlan_get(bat_priv, vid | BATADV_VLAN_HAS_TAG); - if (!vlan) - return -ENOENT; - - batadv_softif_destroy_vlan(bat_priv, vlan); - - /* finally free the vlan object */ - batadv_softif_vlan_put(vlan); - + batadv_softif_destroy_vlan_own(bat_priv, vid | BATADV_VLAN_HAS_TAG); return 0; } @@ -1099,7 +1113,6 @@ static void batadv_softif_destroy_netlink(struct net_device *soft_iface, { struct batadv_priv *bat_priv = netdev_priv(soft_iface); struct batadv_hard_iface *hard_iface; - struct batadv_softif_vlan *vlan; list_for_each_entry(hard_iface, &batadv_hardif_list, list) { if (hard_iface->soft_iface == soft_iface) @@ -1107,11 +1120,7 @@ static void batadv_softif_destroy_netlink(struct net_device *soft_iface, } /* destroy the "untagged" VLAN */ - vlan = batadv_softif_vlan_get(bat_priv, BATADV_NO_FLAGS); - if (vlan) { - batadv_softif_destroy_vlan(bat_priv, vlan); - batadv_softif_vlan_put(vlan); - } + batadv_softif_destroy_vlan_own(bat_priv, BATADV_NO_FLAGS); unregister_netdevice_queue(soft_iface, head); } diff --git a/net/batman-adv/soft-interface.h b/net/batman-adv/soft-interface.h index 9f2003f1a497..7050ccd304df 100644 --- a/net/batman-adv/soft-interface.h +++ b/net/batman-adv/soft-interface.h @@ -21,10 +21,14 @@ void batadv_interface_rx(struct net_device *soft_iface, struct batadv_orig_node *orig_node); bool batadv_softif_is_valid(const struct net_device *net_dev); extern struct rtnl_link_ops batadv_link_ops; -int batadv_softif_create_vlan(struct batadv_priv *bat_priv, unsigned short vid); +int batadv_softif_create_vlan_own(struct batadv_priv *bat_priv, + unsigned short vid); void batadv_softif_vlan_release(struct kref *ref); struct batadv_softif_vlan *batadv_softif_vlan_get(struct batadv_priv *bat_priv, unsigned short vid); +struct batadv_softif_vlan * +batadv_softif_vlan_get_or_create(struct batadv_priv *bat_priv, + unsigned short vid); /** * batadv_softif_vlan_put() - decrease the vlan object refcounter and diff --git a/net/batman-adv/translation-table.c b/net/batman-adv/translation-table.c index b21ff3c36b07..ca74a46c1d93 100644 --- a/net/batman-adv/translation-table.c +++ b/net/batman-adv/translation-table.c @@ -591,8 +591,8 @@ static void batadv_tt_global_free(struct batadv_priv *bat_priv, * * Return: true if the client was successfully added, false otherwise. */ -bool batadv_tt_local_add(struct net_device *soft_iface, const u8 *addr, - unsigned short vid, int ifindex, u32 mark) +int batadv_tt_local_add(struct net_device *soft_iface, const u8 *addr, + unsigned short vid, int ifindex, u32 mark) { struct batadv_priv *bat_priv = netdev_priv(soft_iface); struct batadv_tt_local_entry *tt_local; @@ -604,10 +604,10 @@ bool batadv_tt_local_add(struct net_device *soft_iface, const u8 *addr, struct hlist_head *head; struct batadv_tt_orig_list_entry *orig_entry; int hash_added, table_size, packet_size_max; - bool ret = false; bool roamed_back = false; u8 remote_flags; u32 match_mark; + int ret = 0; if (ifindex != BATADV_NULL_IFINDEX) in_dev = dev_get_by_index(net, ifindex); @@ -658,21 +658,22 @@ bool batadv_tt_local_add(struct net_device *soft_iface, const u8 *addr, net_ratelimited_function(batadv_info, soft_iface, "Local translation table size (%i) exceeds maximum packet size (%i); Ignoring new local tt entry: %pM\n", table_size, packet_size_max, addr); + ret = -E2BIG; goto out; } tt_local = kmem_cache_alloc(batadv_tl_cache, GFP_ATOMIC); - if (!tt_local) + if (!tt_local) { + ret = -ENOMEM; goto out; + } /* increase the refcounter of the related vlan */ - vlan = batadv_softif_vlan_get(bat_priv, vid); + vlan = batadv_softif_vlan_get_or_create(bat_priv, vid); if (!vlan) { - net_ratelimited_function(batadv_info, soft_iface, - "adding TT local entry %pM to non-existent VLAN %d\n", - addr, batadv_print_vid(vid)); kmem_cache_free(batadv_tl_cache, tt_local); tt_local = NULL; + ret = -ENOMEM; goto out; } @@ -769,7 +770,7 @@ bool batadv_tt_local_add(struct net_device *soft_iface, const u8 *addr, if (remote_flags ^ (tt_local->common.flags & BATADV_TT_REMOTE_MASK)) batadv_tt_local_event(bat_priv, tt_local, BATADV_NO_FLAGS); - ret = true; + ret = 1; out: batadv_hardif_put(in_hardif); dev_put(in_dev); diff --git a/net/batman-adv/translation-table.h b/net/batman-adv/translation-table.h index d18740d9a22b..bbdda8488c14 100644 --- a/net/batman-adv/translation-table.h +++ b/net/batman-adv/translation-table.h @@ -16,8 +16,8 @@ #include <linux/types.h> int batadv_tt_init(struct batadv_priv *bat_priv); -bool batadv_tt_local_add(struct net_device *soft_iface, const u8 *addr, - unsigned short vid, int ifindex, u32 mark); +int batadv_tt_local_add(struct net_device *soft_iface, const u8 *addr, + unsigned short vid, int ifindex, u32 mark); u16 batadv_tt_local_remove(struct batadv_priv *bat_priv, const u8 *addr, unsigned short vid, const char *message, bool roaming); -- 2.45.1

3 6

Re: bat0 not capable of MTU > 1500
by Rich Chase 03 Sep '24

03 Sep '24

I am trying to create an unimpeded path MTU of 1500 on my OpenWRT router (Belkin RT-3200). I built the firmware from git Master, with kernel version 6.1, and batman-adv version 2023.2. I have a batmesh hard-if device mastering an 802.1S mesh, which are both set to MTU=2304. Using Luci (OpenWRT web gui for uci), I set the MTU of bat0 to 1504 to accommodate .1q vlans on bat0 -- bat0.5, bat0.110, etc. I see in the system log: kern.err kernel: [751585.375855] bat0: mtu greater than device maximum Also, if I do: #> ip link set bat0 mtu 1504 I get: ip: SIOCSIFMTU: Invalid argument and if I test: #> ip link show bat0 ...it says that my mtu is 1500 still. For my life I can't see anything I'm doing wrong. Why can't bat0 have MTU greater than 1500? I don't have any other devices mastered by bat0, other than my .1s wifi mesh link. Throwing myself at your mercy. Help?

3 3

configuration for mobile nodes
by BEAL Romain 29 Jun '24

29 Jun '24

Hi, I want to use batman-adv for highly mobile nodes. All of my nodes are robot or radio module embedded on a human, for example. Losing a node is very common and batman-adv needs to self-reconfigure quickly in my case. Adding a node is very common too, maybe for only few seconds or minutes. The default configuration doesn't work very well for my use case. But when I use this configuration : #define BATADV_PURGE_TIMEOUT 2000 #define BATADV_TT_LOCAL_TIMEOUT 6000 #define BATADV_TT_CLIENT_ROAM_TIMEOUT 6000 #define BATADV_TT_CLIENT_TEMP_TIMEOUT 6000 #define BATADV_ORIG_WORK_PERIOD 500 #define BATADV_DAT_ENTRY_TIMEOUT (5*600) The result is better. I have some unstable connection for only 5 seconds when I loss a node. Have you done any testing for this type of use ? Thanks for help Ce message électronique et tous les fichiers qui lui sont attachés sont confidentiels et destinés exclusivement à l'usage de la personne à laquelle ils sont adressés. Si vous avez reçu ce message par erreur, merci d'en avertir immédiatement son émetteur et de supprimer ce message de votre système sans en conserver de copie. La copie, la rediffusion à tout tiers et l'utilisation du contenu de ce message ou des fichiers attachés est strictement interdite sans l'accord préalable écrit de l'émetteur. Ce document est susceptible de contenir de l'information soumise à contrôle (ML22) relevant de la réglementation européenne quant aux exportations des matériels de guerre et matériels assimilés. Toute exportation ou réexportation est soumise à autorisation écrite du gouvernement français. -------------------------------------------------------------------------------- This email and any attachments are confidential and intended solely for the use of the individual to whom they are addressed. If you have received this e-mail in error, please inform the sender immediately without keeping any copy thereof and delete it from your system. Copy, use or distribution of contents of this message or attachments to any other person is strictly prohibited without sender's prior written permission. This document may contains controlled data submitted to EU military lists regulations. Any export or reexport is submitted to written authorization of French Government.

2 1

[PATCH 0/2] pull request for net: batman-adv 2024-06-21
by Simon Wunderlich 22 Jun '24

22 Jun '24

Hi David, hi Jakub, here are two bugfixes for batman-adv which we would like to have integrated into net. Please pull or let me know of any problem! Thank you, Simon The following changes since commit 1613e604df0cd359cf2a7fbd9be7a0bcfacfabd0: Linux 6.10-rc1 (2024-05-26 15:20:12 -0700) are available in the Git repository at: git://git.open-mesh.org/linux-merge.git tags/batadv-net-pullrequest-20240621 for you to fetch changes up to 6bfff3582416b2f809e6b08c6e9d57b18086bdbd: Revert "batman-adv: prefer kfree_rcu() over call_rcu() with free-only callbacks" (2024-06-12 20:18:00 +0200) ---------------------------------------------------------------- Here are some batman-adv bugfixes: - Don't accept TT entries for out-of-spec VIDs, by Sven Eckelmann - Revert "batman-adv: prefer kfree_rcu() over call_rcu() with free-only callbacks", by Linus Lüssing ---------------------------------------------------------------- Linus Lüssing (1): Revert "batman-adv: prefer kfree_rcu() over call_rcu() with free-only callbacks" Sven Eckelmann (1): batman-adv: Don't accept TT entries for out-of-spec VIDs net/batman-adv/originator.c | 27 ++++++++++++++++++++++ net/batman-adv/translation-table.c | 47 +++++++++++++++++++++++++++++++++++--- 2 files changed, 71 insertions(+), 3 deletions(-)

2 3

[PATCH] batman-adv: add dynamic, bridged-in TT VID detection support
by Linus Lüssing 12 Jun '24

12 Jun '24

So far, if we wanted bridge VLAN tagged frames into the mesh one would need to manually create an according VLAN interface on top of bat0 first, to trigger batman-adv to create the according structures for a VID. With this change the VLAN from bridged-in clients is now automatically detected and added to the translation table on the fly. Signed-off-by: Linus Lüssing <linus.luessing(a)c0d3.blue> --- net/batman-adv/hard-interface.c | 2 +- net/batman-adv/multicast.c | 8 +- net/batman-adv/soft-interface.c | 121 ++++++++++++++++------------- net/batman-adv/soft-interface.h | 6 +- net/batman-adv/translation-table.c | 19 ++--- net/batman-adv/translation-table.h | 4 +- 6 files changed, 88 insertions(+), 72 deletions(-) diff --git a/net/batman-adv/hard-interface.c b/net/batman-adv/hard-interface.c index 96a412beab2d..f5826dd8752c 100644 --- a/net/batman-adv/hard-interface.c +++ b/net/batman-adv/hard-interface.c @@ -946,7 +946,7 @@ static int batadv_hard_if_event_softif(unsigned long event, switch (event) { case NETDEV_REGISTER: bat_priv = netdev_priv(net_dev); - batadv_softif_create_vlan(bat_priv, BATADV_NO_FLAGS); + batadv_softif_create_vlan_own(bat_priv, BATADV_NO_FLAGS); break; } diff --git a/net/batman-adv/multicast.c b/net/batman-adv/multicast.c index 38fab5e46ae2..61e765352e29 100644 --- a/net/batman-adv/multicast.c +++ b/net/batman-adv/multicast.c @@ -724,6 +724,7 @@ static void batadv_mcast_mla_tt_add(struct batadv_priv *bat_priv, { struct batadv_hw_addr *mcast_entry; struct hlist_node *tmp; + int ret; if (!mcast_list) return; @@ -733,9 +734,10 @@ static void batadv_mcast_mla_tt_add(struct batadv_priv *bat_priv, &bat_priv->mcast.mla_list)) continue; - if (!batadv_tt_local_add(bat_priv->soft_iface, - mcast_entry->addr, BATADV_NO_FLAGS, - BATADV_NULL_IFINDEX, BATADV_NO_MARK)) + ret = batadv_tt_local_add(bat_priv->soft_iface, + mcast_entry->addr, BATADV_NO_FLAGS, + BATADV_NULL_IFINDEX, BATADV_NO_MARK); + if (ret <= 0) continue; hlist_del(&mcast_entry->list); diff --git a/net/batman-adv/soft-interface.c b/net/batman-adv/soft-interface.c index 30ecbc2ef1fd..fb647798e5c8 100644 --- a/net/batman-adv/soft-interface.c +++ b/net/batman-adv/soft-interface.c @@ -141,6 +141,10 @@ static int batadv_interface_set_mac_addr(struct net_device *dev, void *p) rcu_read_lock(); hlist_for_each_entry_rcu(vlan, &bat_priv->softif_vlan_list, list) { + /* we don't use this VID ourself, avoid adding us to it */ + if (!batadv_is_my_client(bat_priv, old_addr, vlan->vid)) + continue; + batadv_tt_local_remove(bat_priv, old_addr, vlan->vid, "mac address changed", false); batadv_tt_local_add(dev, addr->sa_data, vlan->vid, @@ -549,13 +553,15 @@ struct batadv_softif_vlan *batadv_softif_vlan_get(struct batadv_priv *bat_priv, } /** - * batadv_softif_create_vlan() - allocate the needed resources for a new vlan + * batadv_softif_create_vlan() - create a softif vlan struct * @bat_priv: the bat priv with all the soft interface information * @vid: the VLAN identifier * - * Return: 0 on success, a negative error otherwise. + * Return: a pointer to the newly allocated softif vlan struct on success, NULL + * otherwise. */ -int batadv_softif_create_vlan(struct batadv_priv *bat_priv, unsigned short vid) +static struct batadv_softif_vlan * +batadv_softif_create_vlan(struct batadv_priv *bat_priv, unsigned short vid) { struct batadv_softif_vlan *vlan; @@ -563,15 +569,14 @@ int batadv_softif_create_vlan(struct batadv_priv *bat_priv, unsigned short vid) vlan = batadv_softif_vlan_get(bat_priv, vid); if (vlan) { - batadv_softif_vlan_put(vlan); spin_unlock_bh(&bat_priv->softif_vlan_list_lock); - return -EEXIST; + return vlan; } vlan = kzalloc(sizeof(*vlan), GFP_ATOMIC); if (!vlan) { spin_unlock_bh(&bat_priv->softif_vlan_list_lock); - return -ENOMEM; + return NULL; } vlan->bat_priv = bat_priv; @@ -584,34 +589,71 @@ int batadv_softif_create_vlan(struct batadv_priv *bat_priv, unsigned short vid) hlist_add_head_rcu(&vlan->list, &bat_priv->softif_vlan_list); spin_unlock_bh(&bat_priv->softif_vlan_list_lock); + return vlan; +} + +/** + * batadv_softif_vlan_get_or_create() - retrieve or create a softif vlan struct + * @bat_priv: the bat priv with all the soft interface information + * @vid: the VLAN identifier + * + * Return: the softif vlan struct if found or created or NULL otherwise. + */ +struct batadv_softif_vlan * +batadv_softif_vlan_get_or_create(struct batadv_priv *bat_priv, + unsigned short vid) +{ + struct batadv_softif_vlan *vlan = batadv_softif_vlan_get(bat_priv, vid); + + if (vlan) + return vlan; + + return batadv_softif_create_vlan(bat_priv, vid); +} + +/** + * batadv_softif_create_vlan_own() - add our own softif to the local TT + * @bat_priv: the bat priv with all the soft interface information + * @vid: the VLAN identifier + * + * Adds the MAC address of our own soft interface with the given VLAN ID as + * a permanent local TT entry. + * + * Return: 0 on success, a negative error otherwise. + */ +int batadv_softif_create_vlan_own(struct batadv_priv *bat_priv, + unsigned short vid) +{ + int ret; + /* add a new TT local entry. This one will be marked with the NOPURGE * flag */ - batadv_tt_local_add(bat_priv->soft_iface, - bat_priv->soft_iface->dev_addr, vid, - BATADV_NULL_IFINDEX, BATADV_NO_MARK); - - /* don't return reference to new softif_vlan */ - batadv_softif_vlan_put(vlan); + ret = batadv_tt_local_add(bat_priv->soft_iface, + bat_priv->soft_iface->dev_addr, vid, + BATADV_NULL_IFINDEX, BATADV_NO_MARK); + if (ret < 0) + return ret; return 0; } /** - * batadv_softif_destroy_vlan() - remove and destroy a softif_vlan object + * batadv_softif_destroy_vlan_own() - remove our own softif from the local TT * @bat_priv: the bat priv with all the soft interface information - * @vlan: the object to remove + * @vid: the VLAN identifier + * + * Removes the MAC address of our own soft interface with the given VLAN ID from + * the local TT. */ -static void batadv_softif_destroy_vlan(struct batadv_priv *bat_priv, - struct batadv_softif_vlan *vlan) +static void batadv_softif_destroy_vlan_own(struct batadv_priv *bat_priv, + unsigned short vid) { /* 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_softif_vlan_put(vlan); + batadv_tt_local_remove(bat_priv, bat_priv->soft_iface->dev_addr, vid, + "vlan interface destroyed", false); } /** @@ -629,7 +671,6 @@ 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; /* only 802.1Q vlans are supported. * batman-adv does not know how to handle other types @@ -639,25 +680,7 @@ static int batadv_interface_add_vid(struct net_device *dev, __be16 proto, vid |= BATADV_VLAN_HAS_TAG; - /* 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); - - /* 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; + return batadv_softif_create_vlan_own(bat_priv, vid); } /** @@ -676,7 +699,6 @@ static int batadv_interface_kill_vid(struct net_device *dev, __be16 proto, unsigned short vid) { struct batadv_priv *bat_priv = netdev_priv(dev); - struct batadv_softif_vlan *vlan; /* only 802.1Q vlans are supported. batman-adv does not know how to * handle other types @@ -684,15 +706,7 @@ static int batadv_interface_kill_vid(struct net_device *dev, __be16 proto, if (proto != htons(ETH_P_8021Q)) return -EINVAL; - vlan = batadv_softif_vlan_get(bat_priv, vid | BATADV_VLAN_HAS_TAG); - if (!vlan) - return -ENOENT; - - batadv_softif_destroy_vlan(bat_priv, vlan); - - /* finally free the vlan object */ - batadv_softif_vlan_put(vlan); - + batadv_softif_destroy_vlan_own(bat_priv, vid | BATADV_VLAN_HAS_TAG); return 0; } @@ -1099,7 +1113,6 @@ static void batadv_softif_destroy_netlink(struct net_device *soft_iface, { struct batadv_priv *bat_priv = netdev_priv(soft_iface); struct batadv_hard_iface *hard_iface; - struct batadv_softif_vlan *vlan; list_for_each_entry(hard_iface, &batadv_hardif_list, list) { if (hard_iface->soft_iface == soft_iface) @@ -1107,11 +1120,7 @@ static void batadv_softif_destroy_netlink(struct net_device *soft_iface, } /* destroy the "untagged" VLAN */ - vlan = batadv_softif_vlan_get(bat_priv, BATADV_NO_FLAGS); - if (vlan) { - batadv_softif_destroy_vlan(bat_priv, vlan); - batadv_softif_vlan_put(vlan); - } + batadv_softif_destroy_vlan_own(bat_priv, BATADV_NO_FLAGS); unregister_netdevice_queue(soft_iface, head); } diff --git a/net/batman-adv/soft-interface.h b/net/batman-adv/soft-interface.h index 9f2003f1a497..7050ccd304df 100644 --- a/net/batman-adv/soft-interface.h +++ b/net/batman-adv/soft-interface.h @@ -21,10 +21,14 @@ void batadv_interface_rx(struct net_device *soft_iface, struct batadv_orig_node *orig_node); bool batadv_softif_is_valid(const struct net_device *net_dev); extern struct rtnl_link_ops batadv_link_ops; -int batadv_softif_create_vlan(struct batadv_priv *bat_priv, unsigned short vid); +int batadv_softif_create_vlan_own(struct batadv_priv *bat_priv, + unsigned short vid); void batadv_softif_vlan_release(struct kref *ref); struct batadv_softif_vlan *batadv_softif_vlan_get(struct batadv_priv *bat_priv, unsigned short vid); +struct batadv_softif_vlan * +batadv_softif_vlan_get_or_create(struct batadv_priv *bat_priv, + unsigned short vid); /** * batadv_softif_vlan_put() - decrease the vlan object refcounter and diff --git a/net/batman-adv/translation-table.c b/net/batman-adv/translation-table.c index b21ff3c36b07..ca74a46c1d93 100644 --- a/net/batman-adv/translation-table.c +++ b/net/batman-adv/translation-table.c @@ -591,8 +591,8 @@ static void batadv_tt_global_free(struct batadv_priv *bat_priv, * * Return: true if the client was successfully added, false otherwise. */ -bool batadv_tt_local_add(struct net_device *soft_iface, const u8 *addr, - unsigned short vid, int ifindex, u32 mark) +int batadv_tt_local_add(struct net_device *soft_iface, const u8 *addr, + unsigned short vid, int ifindex, u32 mark) { struct batadv_priv *bat_priv = netdev_priv(soft_iface); struct batadv_tt_local_entry *tt_local; @@ -604,10 +604,10 @@ bool batadv_tt_local_add(struct net_device *soft_iface, const u8 *addr, struct hlist_head *head; struct batadv_tt_orig_list_entry *orig_entry; int hash_added, table_size, packet_size_max; - bool ret = false; bool roamed_back = false; u8 remote_flags; u32 match_mark; + int ret = 0; if (ifindex != BATADV_NULL_IFINDEX) in_dev = dev_get_by_index(net, ifindex); @@ -658,21 +658,22 @@ bool batadv_tt_local_add(struct net_device *soft_iface, const u8 *addr, net_ratelimited_function(batadv_info, soft_iface, "Local translation table size (%i) exceeds maximum packet size (%i); Ignoring new local tt entry: %pM\n", table_size, packet_size_max, addr); + ret = -E2BIG; goto out; } tt_local = kmem_cache_alloc(batadv_tl_cache, GFP_ATOMIC); - if (!tt_local) + if (!tt_local) { + ret = -ENOMEM; goto out; + } /* increase the refcounter of the related vlan */ - vlan = batadv_softif_vlan_get(bat_priv, vid); + vlan = batadv_softif_vlan_get_or_create(bat_priv, vid); if (!vlan) { - net_ratelimited_function(batadv_info, soft_iface, - "adding TT local entry %pM to non-existent VLAN %d\n", - addr, batadv_print_vid(vid)); kmem_cache_free(batadv_tl_cache, tt_local); tt_local = NULL; + ret = -ENOMEM; goto out; } @@ -769,7 +770,7 @@ bool batadv_tt_local_add(struct net_device *soft_iface, const u8 *addr, if (remote_flags ^ (tt_local->common.flags & BATADV_TT_REMOTE_MASK)) batadv_tt_local_event(bat_priv, tt_local, BATADV_NO_FLAGS); - ret = true; + ret = 1; out: batadv_hardif_put(in_hardif); dev_put(in_dev); diff --git a/net/batman-adv/translation-table.h b/net/batman-adv/translation-table.h index d18740d9a22b..bbdda8488c14 100644 --- a/net/batman-adv/translation-table.h +++ b/net/batman-adv/translation-table.h @@ -16,8 +16,8 @@ #include <linux/types.h> int batadv_tt_init(struct batadv_priv *bat_priv); -bool batadv_tt_local_add(struct net_device *soft_iface, const u8 *addr, - unsigned short vid, int ifindex, u32 mark); +int batadv_tt_local_add(struct net_device *soft_iface, const u8 *addr, + unsigned short vid, int ifindex, u32 mark); u16 batadv_tt_local_remove(struct batadv_priv *bat_priv, const u8 *addr, unsigned short vid, const char *message, bool roaming); -- 2.45.1

1 0

[PATCH] Revert "batman-adv: prefer kfree_rcu() over call_rcu() with free-only callbacks"
by Linus Lüssing 12 Jun '24

12 Jun '24

This reverts commit cf9845d4566698731c23cd7a02db956706d058a1. This change seems to result in a memory leak / RCU race and the following kernel splat when the batman-adv kernel module is unloaded: ``` [ 112.208633] ============================================================================= [ 112.210359] BUG batadv_tl_cache (Tainted: G OE ): Objects remaining in batadv_tl_cache on __kmem_cache_shutdown() [ 112.211943] ----------------------------------------------------------------------------- [ 112.212517] Slab 0xffffe8afc0216d00 objects=16 used=1 fp=0xffff93f4085b4340 flags=0xfffffc0000a00(workingset|slab|node=0|zone=1|lastcpupid=0x1fffff) [ 112.212517] CPU: 1 PID: 776 Comm: rmmod Tainted: G OE 6.8.12-amd64 #1 Debian 6.8.12-1 [ 112.212517] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.3-debian-1.16.3-2 04/01/2014 [ 112.212517] Call Trace: [ 112.212517] <TASK> [ 112.212517] dump_stack_lvl+0x64/0x80 [ 112.212517] slab_err+0xe6/0x120 [ 112.212517] __kmem_cache_shutdown+0x160/0x2e0 [ 112.212517] kmem_cache_destroy+0x55/0x160 [ 112.220849] batadv_tt_cache_destroy+0x15/0x60 [batman_adv] [ 112.220849] __do_sys_delete_module+0x1d5/0x320 [ 112.220849] do_syscall_64+0x83/0x190 [ 112.220849] ? do_syscall_64+0x8f/0x190 [ 112.220849] ? exc_page_fault+0x7f/0x180 [ 112.220849] entry_SYSCALL_64_after_hwframe+0x78/0x80 [ 112.224478] RIP: 0033:0x7f2ac8434977 [ 112.224478] Code: 73 01 c3 48 8b 0d a9 94 0c 00 f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 b8 b0 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d 79 94 0c 00 f7 d8 64 89 01 48 [ 112.224478] RSP: 002b:00007ffe0adf6138 EFLAGS: 00000206 ORIG_RAX: 00000000000000b0 [ 112.224478] RAX: ffffffffffffffda RBX: 000055db9018e770 RCX: 00007f2ac8434977 [ 112.224478] RDX: 0000000000000000 RSI: 0000000000000800 RDI: 000055db9018e7d8 [ 112.224478] RBP: 0000000000000000 R08: 1999999999999999 R09: 0000000000000000 [ 112.224478] R10: 00007f2ac84a6ac0 R11: 0000000000000206 R12: 00007ffe0adf6390 [ 112.224478] R13: 000055db9018e770 R14: 000055db9018d2a0 R15: 0000000000000000 [ 112.233961] </TASK> [ 112.233961] Disabling lock debugging due to kernel taint [ 112.233961] Object 0xffff93f4085b4140 @offset=320 [ 112.233961] Allocated in batadv_tt_local_add+0x297/0xa20 [batman_adv] age=15835 cpu=1 pid=755 [ 112.233961] batadv_tt_local_add+0x297/0xa20 [batman_adv] [ 112.233961] batadv_interface_set_mac_addr+0xf6/0x120 [batman_adv] [ 112.233961] dev_set_mac_address+0xde/0x140 [ 112.233961] dev_set_mac_address_user+0x30/0x50 [ 112.233961] do_setlink+0x261/0x12d0 [ 112.233961] rtnl_setlink+0x11f/0x1d0 [ 112.233961] rtnetlink_rcv_msg+0x152/0x3c0 [ 112.241772] netlink_rcv_skb+0x5b/0x110 [ 112.241772] netlink_unicast+0x1a6/0x290 [ 112.241772] netlink_sendmsg+0x223/0x490 [ 112.241772] __sys_sendto+0x1df/0x1f0 [ 112.241772] __x64_sys_sendto+0x24/0x30 [ 112.241772] do_syscall_64+0x83/0x190 [ 112.241772] entry_SYSCALL_64_after_hwframe+0x78/0x80 [ 112.245994] ------------[ cut here ]------------ [ 112.246650] kmem_cache_destroy batadv_tl_cache: Slab cache still has objects when called from batadv_tt_cache_destroy+0x15/0x60 [batman_adv] [ 112.246668] WARNING: CPU: 1 PID: 776 at mm/slab_common.c:493 kmem_cache_destroy+0x14d/0x160 [ 112.249584] Modules linked in: veth batman_adv(OE-) cfg80211 rfkill bridge stp llc libcrc32c crc32c_generic crc16 rpcsec_gss_krb5 auth_rpcgss nfsv4 dns_resolver binfmt_misc pcspkr button joydev evdev serio_raw loop dm_mod efi_pstore nfnetlink vsock_loopback vmw_vsock_virtio_transport_common vmw_vsock_vmci_transport vsock vmw_vmci qemu_fw_cfg ip_tables x_tables autofs4 nfsv3 nfs_acl nfs lockd grace sunrpc 9pnet_rdma rdma_cm iw_cm ib_cm ib_core configfs 9p netfs ata_generic ata_piix libata psmouse scsi_mod 9pnet_virtio i2c_piix4 9pnet e1000 scsi_common floppy crypto_simd cryptd [ 112.256555] CPU: 1 PID: 776 Comm: rmmod Tainted: G B OE 6.8.12-amd64 #1 Debian 6.8.12-1 [ 112.258457] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.3-debian-1.16.3-2 04/01/2014 [ 112.260410] RIP: 0010:kmem_cache_destroy+0x14d/0x160 [ 112.261687] Code: 00 eb be 5b 5d 41 5c 41 5d c3 cc cc cc cc 48 8b 53 60 48 8b 4c 24 20 48 c7 c6 60 d5 e3 98 48 c7 c7 b8 ec 2d 99 e8 43 0d d8 ff <0f> 0b e9 e2 fe ff ff c3 cc cc cc cc 0f 1f 80 00 00 00 00 90 90 90 [ 112.265219] RSP: 0018:ffffb3b2806e7e48 EFLAGS: 00010282 [ 112.266044] RAX: 0000000000000000 RBX: ffff93f4270a2640 RCX: 0000000000000027 [ 112.267157] RDX: ffff93f43c521708 RSI: 0000000000000001 RDI: ffff93f43c521700 [ 112.268268] RBP: 000055db9018e7d8 R08: 0000000000000000 R09: ffffb3b2806e7cd8 [ 112.269418] R10: ffffb3b2806e7cd0 R11: 0000000000000003 R12: 0000000080012d00 [ 112.270572] R13: ffffb3b2806e7f58 R14: 0000000000000000 R15: 0000000000000000 [ 112.271699] FS: 00007f2ac8308440(0000) GS:ffff93f43c500000(0000) knlGS:0000000000000000 [ 112.273001] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 112.273923] CR2: 00005584ef830110 CR3: 000000000787c000 CR4: 00000000000006f0 [ 112.275050] Call Trace: [ 112.275464] <TASK> [ 112.275810] ? kmem_cache_destroy+0x14d/0x160 [ 112.276518] ? __warn+0x81/0x130 [ 112.277043] ? kmem_cache_destroy+0x14d/0x160 [ 112.277730] ? report_bug+0x171/0x1a0 [ 112.278315] ? prb_read_valid+0x1b/0x30 [ 112.278919] ? handle_bug+0x3c/0x80 [ 112.279467] ? exc_invalid_op+0x17/0x70 [ 112.280071] ? asm_exc_invalid_op+0x1a/0x20 [ 112.280741] ? kmem_cache_destroy+0x14d/0x160 [ 112.281603] ? kmem_cache_destroy+0x14d/0x160 [ 112.282489] batadv_tt_cache_destroy+0x15/0x60 [batman_adv] [ 112.283373] __do_sys_delete_module+0x1d5/0x320 [ 112.284080] do_syscall_64+0x83/0x190 [ 112.284696] ? do_syscall_64+0x8f/0x190 [ 112.285315] ? exc_page_fault+0x7f/0x180 [ 112.285970] entry_SYSCALL_64_after_hwframe+0x78/0x80 [ 112.286768] RIP: 0033:0x7f2ac8434977 [ 112.287355] Code: 73 01 c3 48 8b 0d a9 94 0c 00 f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 b8 b0 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d 79 94 0c 00 f7 d8 64 89 01 48 [ 112.290282] RSP: 002b:00007ffe0adf6138 EFLAGS: 00000206 ORIG_RAX: 00000000000000b0 [ 112.291465] RAX: ffffffffffffffda RBX: 000055db9018e770 RCX: 00007f2ac8434977 [ 112.292595] RDX: 0000000000000000 RSI: 0000000000000800 RDI: 000055db9018e7d8 [ 112.293724] RBP: 0000000000000000 R08: 1999999999999999 R09: 0000000000000000 [ 112.294863] R10: 00007f2ac84a6ac0 R11: 0000000000000206 R12: 00007ffe0adf6390 [ 112.295982] R13: 000055db9018e770 R14: 000055db9018d2a0 R15: 0000000000000000 [ 112.297103] </TASK> [ 112.297465] ---[ end trace 0000000000000000 ]--- ``` So far, after some debugging, the actual cause for this could not immediately be found within the batman-adv code. Therefore reverting this for now until the underlying issue can be found and better understood. Some additional debugging information and discussions can be found on our Redmine bugtracker, linked below. Link: https://www.open-mesh.org/issues/428 Signed-off-by: Linus Lüssing <linus.luessing(a)c0d3.blue> --- Some help / guidance for further debugging from people more knowledgeable in the RCU core internals would be appreciated. Specifically why the rcu_barrier() call in batadv_exit() does not seem to catch the kfree_rcu() which the reverted patch was originally adding. compat.h | 8 ----- net/batman-adv/translation-table.c | 47 ++++++++++++++++++++++++++++-- 2 files changed, 44 insertions(+), 11 deletions(-) diff --git a/compat.h b/compat.h index d823eeb7bc70..638bfc54f3d0 100644 --- a/compat.h +++ b/compat.h @@ -16,14 +16,6 @@ #include "compat-autoconf.h" -#if LINUX_VERSION_IS_LESS(6, 4, 0) - -#if IS_ENABLED(CONFIG_SLOB) -#error kfree_rcu for kmem_cache not supported when SLOB is enabled -#endif - -#endif /* LINUX_VERSION_IS_LESS(6, 4, 0) */ - #endif /* __KERNEL__ */ #endif /* _NET_BATMAN_ADV_COMPAT_H_ */ diff --git a/net/batman-adv/translation-table.c b/net/batman-adv/translation-table.c index b21ff3c36b07..2243cec18ecc 100644 --- a/net/batman-adv/translation-table.c +++ b/net/batman-adv/translation-table.c @@ -208,6 +208,20 @@ batadv_tt_global_hash_find(struct batadv_priv *bat_priv, const u8 *addr, return tt_global_entry; } +/** + * batadv_tt_local_entry_free_rcu() - free the tt_local_entry + * @rcu: rcu pointer of the tt_local_entry + */ +static void batadv_tt_local_entry_free_rcu(struct rcu_head *rcu) +{ + struct batadv_tt_local_entry *tt_local_entry; + + tt_local_entry = container_of(rcu, struct batadv_tt_local_entry, + common.rcu); + + kmem_cache_free(batadv_tl_cache, tt_local_entry); +} + /** * batadv_tt_local_entry_release() - release tt_local_entry from lists and queue * for free after rcu grace period @@ -222,7 +236,7 @@ static void batadv_tt_local_entry_release(struct kref *ref) batadv_softif_vlan_put(tt_local_entry->vlan); - kfree_rcu(tt_local_entry, common.rcu); + call_rcu(&tt_local_entry->common.rcu, batadv_tt_local_entry_free_rcu); } /** @@ -240,6 +254,20 @@ batadv_tt_local_entry_put(struct batadv_tt_local_entry *tt_local_entry) batadv_tt_local_entry_release); } +/** + * batadv_tt_global_entry_free_rcu() - free the tt_global_entry + * @rcu: rcu pointer of the tt_global_entry + */ +static void batadv_tt_global_entry_free_rcu(struct rcu_head *rcu) +{ + struct batadv_tt_global_entry *tt_global_entry; + + tt_global_entry = container_of(rcu, struct batadv_tt_global_entry, + common.rcu); + + kmem_cache_free(batadv_tg_cache, tt_global_entry); +} + /** * batadv_tt_global_entry_release() - release tt_global_entry from lists and * queue for free after rcu grace period @@ -254,7 +282,7 @@ void batadv_tt_global_entry_release(struct kref *ref) batadv_tt_global_del_orig_list(tt_global_entry); - kfree_rcu(tt_global_entry, common.rcu); + call_rcu(&tt_global_entry->common.rcu, batadv_tt_global_entry_free_rcu); } /** @@ -379,6 +407,19 @@ static void batadv_tt_global_size_dec(struct batadv_orig_node *orig_node, batadv_tt_global_size_mod(orig_node, vid, -1); } +/** + * batadv_tt_orig_list_entry_free_rcu() - free the orig_entry + * @rcu: rcu pointer of the orig_entry + */ +static void batadv_tt_orig_list_entry_free_rcu(struct rcu_head *rcu) +{ + struct batadv_tt_orig_list_entry *orig_entry; + + orig_entry = container_of(rcu, struct batadv_tt_orig_list_entry, rcu); + + kmem_cache_free(batadv_tt_orig_cache, orig_entry); +} + /** * batadv_tt_orig_list_entry_release() - release tt orig entry from lists and * queue for free after rcu grace period @@ -392,7 +433,7 @@ static void batadv_tt_orig_list_entry_release(struct kref *ref) refcount); batadv_orig_node_put(orig_entry->orig_node); - kfree_rcu(orig_entry, rcu); + call_rcu(&orig_entry->rcu, batadv_tt_orig_list_entry_free_rcu); } /** -- 2.45.1

4 8

[PATCH v4] batman-adv: Add flex array to struct batadv_tvlv_tt_data
by Erick Archer 07 Jun '24

07 Jun '24

The "struct batadv_tvlv_tt_data" uses a dynamically sized set of trailing elements. Specifically, it uses an array of structures of type "batadv_tvlv_tt_vlan_data". So, use the preferred way in the kernel declaring a flexible array [1]. At the same time, prepare for the coming implementation by GCC and Clang of the __counted_by attribute. Flexible array members annotated with __counted_by can have their accesses bounds-checked at run-time via CONFIG_UBSAN_BOUNDS (for array indexing) and CONFIG_FORTIFY_SOURCE (for strcpy/memcpy-family functions). In this case, it is important to note that the attribute used is specifically __counted_by_be since variable "num_vlan" is of type __be16. The following change to the "batadv_tt_tvlv_ogm_handler_v1" function: - tt_vlan = (struct batadv_tvlv_tt_vlan_data *)(tt_data + 1); - tt_change = (struct batadv_tvlv_tt_change *)(tt_vlan + num_vlan); + tt_change = (struct batadv_tvlv_tt_change *)((void *)tt_data + + flex_size); is intended to prevent the compiler from generating an "out-of-bounds" notification due to the __counted_by attribute. The compiler can do a pointer calculation using the vlan_data flexible array memory, or in other words, this may be calculated as an array offset, since it is the same as: &tt_data->vlan_data[num_vlan] Therefore, we go past the end of the array. In other "multiple trailing flexible array" situations, this has been solved by addressing from the base pointer, since the compiler either knows the full allocation size or it knows nothing about it (this case, since it came from a "void *" function argument). The order in which the structure batadv_tvlv_tt_data and the structure batadv_tvlv_tt_vlan_data are defined must be swap to avoid an incomplete type error. Also, avoid the open-coded arithmetic in memory allocator functions [2] using the "struct_size" macro and use the "flex_array_size" helper to clarify some calculations, when possible. Moreover, the new structure member also allow us to avoid the open-coded arithmetic on pointers in some situations. Take advantage of this. This code was detected with the help of Coccinelle, and audited and modified manually. Link: https://www.kernel.org/doc/html/next/process/deprecated.html#zero-length-an… [1] Link: https://www.kernel.org/doc/html/next/process/deprecated.html#open-coded-ari… [2] Reviewed-by: Kees Cook <keescook(a)chromium.org> Signed-off-by: Erick Archer <erick.archer(a)outlook.com> --- Changes in v4: - Add the #include of <linux/stddef.h> for the "__counted_by_be" helper (Sven Eckelmann). - Add the "Reviewed-by:" tag. Changes in v3: - Address from the base pointer tt_data to avoid an "out-of-bounds" notification (Kees Cook). - Update the commit message to explain the new change. Changes in v2: - Add the #include of <linux/overflow.h> for the "flex_array_size" helper (Sven Eckelmann). - Add the "ntohs" function to the "flex_array_size" helper removed by mistake during the conversion (Sven Eckelmann). - Add the "__counted_by_be" attribute. Previous versions: v3 -> https://lore.kernel.org/linux-hardening/AS8PR02MB72371F89D188B047410B755E8B… v2 -> https://lore.kernel.org/linux-hardening/AS8PR02MB723756E3D1366C4E8FCD14BF8B… v1 -> https://lore.kernel.org/linux-hardening/AS8PR02MB7237987BF9DFCA030B330F658B… Hi, As it was decided in version 3, I resend this path since -r2 has been released. I hope that this time the patch can be applied ;) Regards, Erick --- include/uapi/linux/batadv_packet.h | 29 ++++++++++-------- net/batman-adv/translation-table.c | 49 ++++++++++++------------------ 2 files changed, 36 insertions(+), 42 deletions(-) diff --git a/include/uapi/linux/batadv_packet.h b/include/uapi/linux/batadv_packet.h index 6e25753015df..439132a819ea 100644 --- a/include/uapi/linux/batadv_packet.h +++ b/include/uapi/linux/batadv_packet.h @@ -9,6 +9,7 @@ #include <asm/byteorder.h> #include <linux/if_ether.h> +#include <linux/stddef.h> #include <linux/types.h> /** @@ -592,19 +593,6 @@ struct batadv_tvlv_gateway_data { __be32 bandwidth_up; }; -/** - * struct batadv_tvlv_tt_data - tt data propagated through the tt tvlv container - * @flags: translation table flags (see batadv_tt_data_flags) - * @ttvn: translation table version number - * @num_vlan: number of announced VLANs. In the TVLV this struct is followed by - * one batadv_tvlv_tt_vlan_data object per announced vlan - */ -struct batadv_tvlv_tt_data { - __u8 flags; - __u8 ttvn; - __be16 num_vlan; -}; - /** * struct batadv_tvlv_tt_vlan_data - vlan specific tt data propagated through * the tt tvlv container @@ -618,6 +606,21 @@ struct batadv_tvlv_tt_vlan_data { __u16 reserved; }; +/** + * struct batadv_tvlv_tt_data - tt data propagated through the tt tvlv container + * @flags: translation table flags (see batadv_tt_data_flags) + * @ttvn: translation table version number + * @num_vlan: number of announced VLANs. In the TVLV this struct is followed by + * one batadv_tvlv_tt_vlan_data object per announced vlan + * @vlan_data: array of batadv_tvlv_tt_vlan_data objects + */ +struct batadv_tvlv_tt_data { + __u8 flags; + __u8 ttvn; + __be16 num_vlan; + struct batadv_tvlv_tt_vlan_data vlan_data[] __counted_by_be(num_vlan); +}; + /** * struct batadv_tvlv_tt_change - translation table diff data * @flags: status indicators concerning the non-mesh client (see diff --git a/net/batman-adv/translation-table.c b/net/batman-adv/translation-table.c index b21ff3c36b07..b44c382226a1 100644 --- a/net/batman-adv/translation-table.c +++ b/net/batman-adv/translation-table.c @@ -28,6 +28,7 @@ #include <linux/net.h> #include <linux/netdevice.h> #include <linux/netlink.h> +#include <linux/overflow.h> #include <linux/rculist.h> #include <linux/rcupdate.h> #include <linux/skbuff.h> @@ -815,8 +816,7 @@ batadv_tt_prepare_tvlv_global_data(struct batadv_orig_node *orig_node, num_entries += atomic_read(&vlan->tt.num_entries); } - change_offset = sizeof(**tt_data); - change_offset += num_vlan * sizeof(*tt_vlan); + change_offset = struct_size(*tt_data, vlan_data, num_vlan); /* if tt_len is negative, allocate the space needed by the full table */ if (*tt_len < 0) @@ -835,7 +835,7 @@ batadv_tt_prepare_tvlv_global_data(struct batadv_orig_node *orig_node, (*tt_data)->ttvn = atomic_read(&orig_node->last_ttvn); (*tt_data)->num_vlan = htons(num_vlan); - tt_vlan = (struct batadv_tvlv_tt_vlan_data *)(*tt_data + 1); + tt_vlan = (*tt_data)->vlan_data; hlist_for_each_entry(vlan, &orig_node->vlan_list, list) { tt_vlan->vid = htons(vlan->vid); tt_vlan->crc = htonl(vlan->tt.crc); @@ -895,8 +895,7 @@ batadv_tt_prepare_tvlv_local_data(struct batadv_priv *bat_priv, total_entries += vlan_entries; } - change_offset = sizeof(**tt_data); - change_offset += num_vlan * sizeof(*tt_vlan); + change_offset = struct_size(*tt_data, vlan_data, num_vlan); /* if tt_len is negative, allocate the space needed by the full table */ if (*tt_len < 0) @@ -915,7 +914,7 @@ batadv_tt_prepare_tvlv_local_data(struct batadv_priv *bat_priv, (*tt_data)->ttvn = atomic_read(&bat_priv->tt.vn); (*tt_data)->num_vlan = htons(num_vlan); - tt_vlan = (struct batadv_tvlv_tt_vlan_data *)(*tt_data + 1); + tt_vlan = (*tt_data)->vlan_data; hlist_for_each_entry(vlan, &bat_priv->softif_vlan_list, list) { vlan_entries = atomic_read(&vlan->tt.num_entries); if (vlan_entries < 1) @@ -2875,7 +2874,6 @@ static bool batadv_send_tt_request(struct batadv_priv *bat_priv, { struct batadv_tvlv_tt_data *tvlv_tt_data = NULL; struct batadv_tt_req_node *tt_req_node = NULL; - struct batadv_tvlv_tt_vlan_data *tt_vlan_req; struct batadv_hard_iface *primary_if; bool ret = false; int i, size; @@ -2891,7 +2889,7 @@ static bool batadv_send_tt_request(struct batadv_priv *bat_priv, if (!tt_req_node) goto out; - size = sizeof(*tvlv_tt_data) + sizeof(*tt_vlan_req) * num_vlan; + size = struct_size(tvlv_tt_data, vlan_data, num_vlan); tvlv_tt_data = kzalloc(size, GFP_ATOMIC); if (!tvlv_tt_data) goto out; @@ -2903,12 +2901,10 @@ static bool batadv_send_tt_request(struct batadv_priv *bat_priv, /* send all the CRCs within the request. This is needed by intermediate * nodes to ensure they have the correct table before replying */ - tt_vlan_req = (struct batadv_tvlv_tt_vlan_data *)(tvlv_tt_data + 1); for (i = 0; i < num_vlan; i++) { - tt_vlan_req->vid = tt_vlan->vid; - tt_vlan_req->crc = tt_vlan->crc; + tvlv_tt_data->vlan_data[i].vid = tt_vlan->vid; + tvlv_tt_data->vlan_data[i].crc = tt_vlan->crc; - tt_vlan_req++; tt_vlan++; } @@ -2960,7 +2956,6 @@ static bool batadv_send_other_tt_response(struct batadv_priv *bat_priv, struct batadv_orig_node *res_dst_orig_node = NULL; struct batadv_tvlv_tt_change *tt_change; struct batadv_tvlv_tt_data *tvlv_tt_data = NULL; - struct batadv_tvlv_tt_vlan_data *tt_vlan; bool ret = false, full_table; u8 orig_ttvn, req_ttvn; u16 tvlv_len; @@ -2983,10 +2978,9 @@ static bool batadv_send_other_tt_response(struct batadv_priv *bat_priv, orig_ttvn = (u8)atomic_read(&req_dst_orig_node->last_ttvn); req_ttvn = tt_data->ttvn; - tt_vlan = (struct batadv_tvlv_tt_vlan_data *)(tt_data + 1); /* this node doesn't have the requested data */ if (orig_ttvn != req_ttvn || - !batadv_tt_global_check_crc(req_dst_orig_node, tt_vlan, + !batadv_tt_global_check_crc(req_dst_orig_node, tt_data->vlan_data, ntohs(tt_data->num_vlan))) goto out; @@ -3329,7 +3323,6 @@ static void batadv_handle_tt_response(struct batadv_priv *bat_priv, struct batadv_orig_node *orig_node = NULL; struct batadv_tvlv_tt_change *tt_change; u8 *tvlv_ptr = (u8 *)tt_data; - u16 change_offset; batadv_dbg(BATADV_DBG_TT, bat_priv, "Received TT_RESPONSE from %pM for ttvn %d t_size: %d [%c]\n", @@ -3342,10 +3335,7 @@ static void batadv_handle_tt_response(struct batadv_priv *bat_priv, spin_lock_bh(&orig_node->tt_lock); - change_offset = sizeof(struct batadv_tvlv_tt_vlan_data); - change_offset *= ntohs(tt_data->num_vlan); - change_offset += sizeof(*tt_data); - tvlv_ptr += change_offset; + tvlv_ptr += struct_size(tt_data, vlan_data, ntohs(tt_data->num_vlan)); tt_change = (struct batadv_tvlv_tt_change *)tvlv_ptr; if (tt_data->flags & BATADV_TT_FULL_TABLE) { @@ -3944,10 +3934,10 @@ static void batadv_tt_tvlv_ogm_handler_v1(struct batadv_priv *bat_priv, u8 flags, void *tvlv_value, u16 tvlv_value_len) { - struct batadv_tvlv_tt_vlan_data *tt_vlan; struct batadv_tvlv_tt_change *tt_change; struct batadv_tvlv_tt_data *tt_data; u16 num_entries, num_vlan; + size_t flex_size; if (tvlv_value_len < sizeof(*tt_data)) return; @@ -3957,17 +3947,18 @@ static void batadv_tt_tvlv_ogm_handler_v1(struct batadv_priv *bat_priv, num_vlan = ntohs(tt_data->num_vlan); - if (tvlv_value_len < sizeof(*tt_vlan) * num_vlan) + flex_size = flex_array_size(tt_data, vlan_data, num_vlan); + if (tvlv_value_len < flex_size) return; - tt_vlan = (struct batadv_tvlv_tt_vlan_data *)(tt_data + 1); - tt_change = (struct batadv_tvlv_tt_change *)(tt_vlan + num_vlan); - tvlv_value_len -= sizeof(*tt_vlan) * num_vlan; + tt_change = (struct batadv_tvlv_tt_change *)((void *)tt_data + + flex_size); + tvlv_value_len -= flex_size; num_entries = batadv_tt_entries(tvlv_value_len); - batadv_tt_update_orig(bat_priv, orig, tt_vlan, num_vlan, tt_change, - num_entries, tt_data->ttvn); + batadv_tt_update_orig(bat_priv, orig, tt_data->vlan_data, num_vlan, + tt_change, num_entries, tt_data->ttvn); } /** @@ -3998,8 +3989,8 @@ static int batadv_tt_tvlv_unicast_handler_v1(struct batadv_priv *bat_priv, tt_data = tvlv_value; tvlv_value_len -= sizeof(*tt_data); - tt_vlan_len = sizeof(struct batadv_tvlv_tt_vlan_data); - tt_vlan_len *= ntohs(tt_data->num_vlan); + tt_vlan_len = flex_array_size(tt_data, vlan_data, + ntohs(tt_data->num_vlan)); if (tvlv_value_len < tt_vlan_len) return NET_RX_SUCCESS; -- 2.25.1

2 1

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

B.A.T.M.A.N June 2024