B.A.T.M.A.N July 2013

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

14 participants
38 discussions

[B.A.T.M.A.N.] [PATCHv8 0/2] per-VLAN TT CRC
by Antonio Quartulli 30 Jul '13

30 Jul '13

This is the second patchset of the TT-VLAN improvement. It brings a not negligible change into the TT logic since it is splitting the CRC mechanism on a per-VLAN basis. Since it is not so easy to digest I decided to send it alone. This change is required by an improvement which will come right after: VLAN based TT filtering for BLA2 enabled nodes (third patchset of the TT-VLAN feature). In a few words.... Right now a backbone node rejects any TT entry announced by another backbone. With this change it will be possible to filter out only those TT entries belonging to the VLAN where BLA2 is running on. In this way, the others VLANs will still be able to contact each other. Before implementing the new feature, the first patch is introducing a new lock to make any change to the local and global TT table atomic. In this way a TT Reply cannot be sent during a writing operation on one of the TT tables. This will prevent the TT Reply from containing any inconsistent data. Thanks everybody for their feedback! Changes from v7: - improved kernel doc and commit message in 1/2 - fixed some kernel doc typ0s in 2/2 - moved change_offset computation inside batadv_tt_prepare_tvlv_global_data() and batadv_tt_prepare_tvlv_local_data() to avoid code duplication in 2/2 Changes from v6: - remove orig_node_vlan from the list after freeing - introduce locking around TT operations with 1/2 - fixed refcounting - fixed style Cheers, Antonio Quartulli (2): batman-adv: lock around TT operations to avoid sending inconsistent data batman-adv: make the TT CRC logic VLAN specific compat.c | 9 + compat.h | 1 + main.c | 1 + originator.c | 102 +++++++- originator.h | 7 + packet.h | 18 +- translation-table.c | 738 ++++++++++++++++++++++++++++++++++++++++++---------- types.h | 52 +++- 8 files changed, 772 insertions(+), 156 deletions(-) -- 1.8.1.5

1 2

[B.A.T.M.A.N.] batman-adv 2013.3.0 released
by Marek Lindner 29 Jul '13

29 Jul '13

The B.A.T.M.A.N. team proudly presents its newest release - batman-adv 2013.3.0 - a stability and bugfix release uniquely focused on ironing out bugs and annoyances. As the kernel module always depends on the Linux kernel it was compiled against, it does not make sense to provide binaries on our website. As usual, you will find the signed tarballs in our download section: http://downloads.open-mesh.org/batman/releases/batman-adv-2013.3.0/ as well as prepackaged binaries in your distribution. Important changes ----------------- For the first time, a batman-adv and batctl release is accompanied by A.L.F.R.E.D. (Almighty Lightweight Fact Remote Exchange Daemon) - a user space daemon conceived for the purpose of replacing the in-kernel visualization component with an easy to extend user space application. It bears the striking advantage of not only distributing visualization data but all kinds of data within a mesh network, thereby filling an often felt gap. Note: The alfred vis support is not compulsory for obtaining visualization data as the in-kernel code still is enabled. Both systems are compatible and can even be deployed side by side. It is planned to remove the in-kernel code by the end of the year which is why it is recommended to check out alfred soon, in the interest of making sure the transition goes as smooth as possible. Thanks ------ Thanks to all people sending in patches: * Antonio Quartulli <ordex(a)autistici.org> * Jiri Pirko <jiri(a)resnulli.us> * Linus Lüssing <linus.luessing(a)web.de> * Marek Lindner <lindner_marek(a)yahoo.de> * Martin Hundebøll <martin(a)hundeboll.net> * Matthias Schiffer <mschiffer(a)universe-factory.net> * Patrick McHardy <kaber(a)trash.net> * Simon Wunderlich <siwu(a)hrz.tu-chemnitz.de> * Sven Eckelmann <sven(a)open-mesh.com> batman-adv ---------- The batman-adv kernel module takes special care when transporting payload broadcast traffic across the mesh network to increase the likelihood of a successful transmission. To reduce overhead this mechanism was tweaked to only be enabled on wireless interfaces as it is safe to assume wired connections and VPNs suffer less from packet loss. The internal handling of VLAN IDs has been unified to prepare for the upcoming translation table VLAN support. DAT has learned not to reply to ARP requests sent by local clients destined for other local clients to avoid duplicate ARP replies. The batman-adv protocol (OGM) duplicate check was applied too strictly in certain situations which could lead to route starvation of better routes. To mitigate this effect the duplicate policy is applied on a per neighbor basis. Upon configuring an interface batman-adv would try to acquire the in-kernel network configuration lock or return with ERESTARTSYS in case the lock was held by some other party. To avoid having to abort an interface activation this part of the code was reworked and the need to acquire the lock removed. A crash on kernel module unload triggered by a double free of the traffic statistic counters was fixed. The network coding neighbor table won't display neighbors unsuitable for network coding anymore. batctl ------ A misleading warning about an uninitialized variable when compiling with "O2" was fixed and a few typographic errors in the bisect output were corrected. alfred ------ Alfred is a user space daemon for distributing arbitrary local information over the mesh/network in a decentralized fashion. This data can be anything which appears to be useful - originally designed to replace the batman-adv visualization (vis), you may distribute hostnames, phone books, administration information, DNS information, the local weather forecast, etc. Alfred runs as daemon in the background of the system. A user may insert information by using the alfred binary on the command line, or use special programs to communicate with alfred (done via unix sockets). The daemon then takes care of distributing the local information to other alfred servers on other nodes. This is done via IPv6 link-local multicast, and does not require any configuration. A user can request data from alfred, and will receive the information available from all alfred servers in the network. Happy routing, The B.A.T.M.A.N. team

1 0

[B.A.T.M.A.N.] [PATCH] batman-adv: set skb priority according to content
by Simon Wunderlich 29 Jul '13

29 Jul '13

From: Simon Wunderlich <simon(a)open-mesh.com> The skb priority field may help the wireless driver to choose the right queue (e.g. WMM queues). This should be set in batman-adv, as this information is only available here. This patch adds support for IPv4/IPv6 DS fields and VLAN PCP. Note that only VLAN PCP is used if a VLAN header is present. Also initially set TC_PRIO_CONTROL only for self-generated packets, and keep the priority set by higher layers. Signed-off-by: Simon Wunderlich <simon(a)open-mesh.com> --- bat_iv_ogm.c | 1 + fragmentation.c | 2 ++ icmp_socket.c | 1 + main.c | 58 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ main.h | 1 + routing.c | 20 ++++++++++++++++++- send.c | 1 - soft-interface.c | 2 ++ 8 files changed, 84 insertions(+), 2 deletions(-) diff --git a/bat_iv_ogm.c b/bat_iv_ogm.c index 5ff1a6b..43bd0e0 100644 --- a/bat_iv_ogm.c +++ b/bat_iv_ogm.c @@ -478,6 +478,7 @@ static void batadv_iv_ogm_aggregate_new(const unsigned char *packet_buff, kfree(forw_packet_aggr); goto out; } + skb->priority = TC_PRIO_CONTROL; skb_reserve(forw_packet_aggr->skb, ETH_HLEN); skb_buff = skb_put(forw_packet_aggr->skb, packet_len); diff --git a/fragmentation.c b/fragmentation.c index ab85e75..271d321 100644 --- a/fragmentation.c +++ b/fragmentation.c @@ -393,6 +393,8 @@ static struct sk_buff *batadv_frag_create(struct sk_buff *skb, if (!skb_fragment) goto err; + skb->priority = TC_PRIO_CONTROL; + /* Eat the last mtu-bytes of the skb */ skb_reserve(skb_fragment, header_size + ETH_HLEN); skb_split(skb, skb_fragment, skb->len - fragment_size); diff --git a/icmp_socket.c b/icmp_socket.c index 512efcc..82ac647 100644 --- a/icmp_socket.c +++ b/icmp_socket.c @@ -183,6 +183,7 @@ static ssize_t batadv_socket_write(struct file *file, const char __user *buff, goto out; } + skb->priority = TC_PRIO_CONTROL; skb_reserve(skb, ETH_HLEN); icmp_packet = (struct batadv_icmp_packet_rr *)skb_put(skb, packet_len); diff --git a/main.c b/main.c index 58cd339..95e5f04 100644 --- a/main.c +++ b/main.c @@ -20,6 +20,9 @@ #include <linux/crc32c.h> #include <linux/highmem.h> #include <linux/if_vlan.h> +#include <net/ip.h> +#include <net/ipv6.h> +#include <net/dsfield.h> #include "main.h" #include "sysfs.h" #include "debugfs.h" @@ -277,6 +280,60 @@ int batadv_max_header_len(void) return header_len; } +/** + * batadv_skb_set_priority - sets skb priority according to packet content + * @skb: the packet to be sent + * @offset: offset to the packet content + * + * This function sets a value between 256 and 263 (802.1d priority), which + * can be interpreted by the cfg80211 or other drivers. + */ +void batadv_skb_set_priority(struct sk_buff *skb, int offset) +{ + struct iphdr ip_hdr_tmp, *ip_hdr; + struct ipv6hdr ip6_hdr_tmp, *ip6_hdr; + struct ethhdr ethhdr_tmp, *ethhdr; + struct vlan_ethhdr *vhdr, vhdr_tmp; + u32 prio; + + /* already set, do nothing */ + if (skb->priority >= 256 && skb->priority <= 263) + return; + + ethhdr = skb_header_pointer(skb, offset, sizeof(*ethhdr), &ethhdr_tmp); + if (!ethhdr) + return; + + switch (ethhdr->h_proto) { + case htons(ETH_P_8021Q): + vhdr = skb_header_pointer(skb, offset + sizeof(*vhdr), + sizeof(*vhdr), &vhdr_tmp); + if (!vhdr) + return; + prio = htons(vhdr->h_vlan_TCI) & VLAN_PRIO_MASK; + prio = prio >> VLAN_PRIO_SHIFT; + break; + case htons(ETH_P_IP): + ip_hdr = skb_header_pointer(skb, offset + sizeof(*ethhdr), + sizeof(*ip_hdr), &ip_hdr_tmp); + if (!ip_hdr) + return; + prio = (ipv4_get_dsfield(ip_hdr) & 0xfc) >> 5; + break; + case htons(ETH_P_IPV6): + ip6_hdr = skb_header_pointer(skb, offset + sizeof(*ethhdr), + sizeof(*ip6_hdr), &ip6_hdr_tmp); + if (!ip6_hdr) + return; + prio = (ipv6_get_dsfield(ip6_hdr) & 0xfc) >> 5; + break; + default: + return; + } + + skb->priority = prio + 256; +} + static int batadv_recv_unhandled_packet(struct sk_buff *skb, struct batadv_hard_iface *recv_if) { @@ -1057,6 +1114,7 @@ void batadv_tvlv_unicast_send(struct batadv_priv *bat_priv, uint8_t *src, if (!skb) goto out; + skb->priority = TC_PRIO_CONTROL; skb_reserve(skb, ETH_HLEN); tvlv_buff = skb_put(skb, sizeof(*unicast_tvlv_packet) + tvlv_len); unicast_tvlv_packet = (struct batadv_unicast_tvlv_packet *)tvlv_buff; diff --git a/main.h b/main.h index 7b68db9..e860f1e 100644 --- a/main.h +++ b/main.h @@ -187,6 +187,7 @@ int batadv_is_my_mac(struct batadv_priv *bat_priv, const uint8_t *addr); struct batadv_hard_iface * batadv_seq_print_text_primary_if_get(struct seq_file *seq); int batadv_max_header_len(void); +void batadv_skb_set_priority(struct sk_buff *skb, int offset); int batadv_batman_skb_recv(struct sk_buff *skb, struct net_device *dev, struct packet_type *ptype, struct net_device *orig_dev); diff --git a/routing.c b/routing.c index e63b05d..149ef57 100644 --- a/routing.c +++ b/routing.c @@ -659,7 +659,7 @@ static int batadv_route_unicast_packet(struct sk_buff *skb, struct batadv_orig_node *orig_node = NULL; struct batadv_unicast_packet *unicast_packet; struct ethhdr *ethhdr = eth_hdr(skb); - int res, ret = NET_RX_DROP; + int res, hdr_len, ret = NET_RX_DROP; unicast_packet = (struct batadv_unicast_packet *)skb->data; @@ -684,6 +684,22 @@ static int batadv_route_unicast_packet(struct sk_buff *skb, unicast_packet = (struct batadv_unicast_packet *)skb->data; unicast_packet->header.ttl--; + switch (unicast_packet->header.packet_type) { + case BATADV_UNICAST_4ADDR: + hdr_len = sizeof(struct batadv_unicast_4addr_packet); + break; + case BATADV_UNICAST: + hdr_len = sizeof(struct batadv_unicast_packet); + break; + default: + /* other packet types not supported - yet */ + hdr_len = -1; + break; + } + + if (hdr_len > 0) + batadv_skb_set_priority(skb, hdr_len); + res = batadv_send_skb_to_orig(skb, orig_node, recv_if); /* translate transmit result into receive result */ @@ -1133,6 +1149,8 @@ int batadv_recv_bcast_packet(struct sk_buff *skb, if (batadv_bla_check_bcast_duplist(bat_priv, skb)) goto out; + batadv_skb_set_priority(skb, sizeof(struct batadv_bcast_packet)); + /* rebroadcast packet */ batadv_add_bcast_packet_to_list(bat_priv, skb, 1); diff --git a/send.c b/send.c index d2cc507..66d09a5 100644 --- a/send.c +++ b/send.c @@ -66,7 +66,6 @@ int batadv_send_skb_packet(struct sk_buff *skb, ethhdr->h_proto = htons(ETH_P_BATMAN); skb_set_network_header(skb, ETH_HLEN); - skb->priority = TC_PRIO_CONTROL; skb->protocol = htons(ETH_P_BATMAN); skb->dev = hard_iface->net_dev; diff --git a/soft-interface.c b/soft-interface.c index 25662d9..a377f4d 100644 --- a/soft-interface.c +++ b/soft-interface.c @@ -235,6 +235,8 @@ static int batadv_interface_tx(struct sk_buff *skb, } } + batadv_skb_set_priority(skb, 0); + /* ethernet packet should be broadcasted */ if (do_bcast) { primary_if = batadv_primary_if_get_selected(bat_priv); -- 1.7.10.4

1 0

[B.A.T.M.A.N.] [PATCHv7 0/2] per-VLAN TT CRC
by Antonio Quartulli 29 Jul '13

29 Jul '13

This is the second patchset of the TT-VLAN improvement. It brings a not negligible change into the TT logic since it is splitting the CRC mechanism on a per-VLAN basis. Since it is not so easy to digest I decided to send it alone. This change is required by an improvement which will come right after: VLAN based TT filtering for BLA2 enabled nodes (third patchset of the TT-VLAN feature). In a few words.... Right now a backbone node rejects any TT entry announced by another backbone. With this change it will be possible to filter out only those TT entries belonging to the VLAN where BLA2 is running on. In this way, the others VLANs will still be able to contact each other. Before implementing the new feature, the first patch is introducing a new lock to make any change to the local and global TT table atomic. In this way a TT Reply cannot be sent during a writing operation on one of the TT tables. This will prevent the TT Reply from containing any inconsistent data. Changes from v6: - remove orig_node_vlan from the list after freeing - introduce locking around TT operations with 1/2 - fixed refcounting - fixed style Cheers, Antonio Quartulli (2): batman-adv: lock around TT operations to avoid sending inconsistent data batman-adv: make the TT CRC logic VLAN specific compat.c | 9 + compat.h | 1 + main.c | 1 + originator.c | 102 +++++++- originator.h | 7 + packet.h | 18 +- translation-table.c | 730 ++++++++++++++++++++++++++++++++++++++++++---------- types.h | 45 +++- 8 files changed, 759 insertions(+), 154 deletions(-) -- 1.8.1.5

2 3

[B.A.T.M.A.N.] Complete netifd integration in openwrt
by Gui Iribarren 29 Jul '13

29 Jul '13

Hello folks, i finally took the time to finish the netifd integration in openwrt. since 2013.0.0, the "slave interfaces" moved from batman-adv.bat0.interfaces to network.meshblah.proto=batadv, so that netifd could properly manage the setup and teardown, avoiding some race conditions i faced at that time (about a year ago) now i did the same for the "mesh interface", batX, which until now had the parameters still set in /etc/config/batman-adv. ### So, what in 2012.4.0 was # cat /etc/config/batman-adv config 'mesh' 'bat0' option 'interfaces' 'mesh0' option 'ap_isolation' '1' option 'gw_mode' 'server' # cat /etc/config/network config interface 'mesh0' option ifname 'eth2' option proto 'none' option auto '1' ### now fully turns into: # cat /etc/config/network config interface 'bat0' option proto 'batmesh' option ifname 'bat0' option ap_isolation '1' option gw_mode 'server' config interface 'mesh0' option ifname 'eth2' option proto 'batadv' option mesh 'bat0' [about the proto name... since proto=batadv was already taken for slave interfaces, i couldn't come up with anything better than "batmesh" :( I was talking with Nico just now, and he suggested "batadvmesh" in any case, suggestions are much welcome. Anyway, proto=bat.*mesh is for setting the mesh parameters] and well, to avoid breaking current configs unnecesarily [over sysupgrades (or opkg upgrades?)] i included a small uci-defaults migration script to take care of that; it transparently migrates all current /etc/config/batman-adv settings to netifd-style. Why the integration? reliability in the setup; no more race conditions whatsoever at boot, or runtime: i even tried rmmoding the module, and doing "ifup mesh0" brought everything up again cleanly. while netifd had some stigmas when it was born, now (in my experience) it has grown to a pretty solid nifty daemon. I made a pull request on https://github.com/openwrt-routing/packages/pull/4 comments welcome :D if you prefer [PATCHES] over mail, just ask :) as always, cheers, and thanks for being so cool people! Gui

2 2

[B.A.T.M.A.N.] Important warnings from alfred 2013.3.0 release source
by Braden Aran 28 Jul '13

28 Jul '13

Hi, I get following important warnings when compiling alfred 2013.3.0 unix_sock.c: In function ‘unix_sock_req_data_finish’: unix_sock.c:261:7: warning: ignoring return value of ‘write’, declared with attribute warn_unused_result [-Wunused-result] write(client_sock, &status, sizeof(status)); ^ unix_sock.c: In function ‘unix_sock_req_data_reply’: unix_sock.c:187:8: warning: ignoring return value of ‘write’, declared with attribute warn_unused_result [-Wunused-result] write(client_sock, buf, sizeof(push->header) + len); ^

2 1

[B.A.T.M.A.N.] A.L.F.R.E.D.+vis manpages missing
by Braden Aran 28 Jul '13

28 Jul '13

Hi, it seems the A.L.F.R.E.D. component and his vis is getting released soon http://git.open-mesh.org/alfred.git/commit/fac591b6d8a8a9eaca33a0d7614433cb… But where are the man pages? Don't you think it would be nice for people when they can read how it is used in a standard unix way? At least they need to use it because this functionality is removed in batman-adv soon. http://www.open-mesh.org/projects/open-mesh/wiki/2013-07-29-batman-adv-2013… I can't find the manpages for both binaries in the tarball http://downloads.open-mesh.org/batman/stable/sources/alfred/alfred-2013.3.0… And isn't the name vis potentially problematic and bad named after all? "vis" doesn't tell anyone about the connection to batman-adv and it sounds more like a GUI program to visualize stuff (mathematical equations?). And it clashes directly with the vis of batmand. So I cannot install both on the same system in a unix manner.

2 1

[B.A.T.M.A.N.] Mangling broadcast packets
by Jan Huwald 27 Jul '13

27 Jul '13

Hi, I am currently redesigning hbbp and p2ptbl[1,2]. p2ptbl is a distributed key-value database build on top of HBBP link-local UDP broadcasts. Both are used in some Freifunk communities and elsewhere. To reduce the bandwidth demand of the employed gossip protocol I would like to use something that behaves like a link-local broadcast packet in a batman-adv mesh every respect - except that any mesh node on it's route may decide to drop it or change it's UDP data. Right now only approaches that are ugly or don't work come to my mind. Ugly: * using NFQUEUE on all enslaved interfaces to mangle packets before they are seen by batman; requires out-of-kernel parsing of batman-adv packets and watching enslaved interfaces * add a custom broadcast TVLV; requires kernel code and either bloats OGMs are requires non-OGM broadcast TVLV packets Don't work: * using NFQUEUE on bcast payload (without batman header); does not work because they are not exposed to the iptables * send link-local UDP on all enslaved interfaces; requires to reimplement all the loop-avoidance / routing / resending logic that already exists in batman-adv If you see a smarter way or a reason why manipulating broadcast packets is utter nonsene: please let me know. With best regards, Jan Huwald [1] http://code.sotun.de/git/hbbp/ [2] http://code.sotun.de/git/wrt/p2ptbl/

2 2

[B.A.T.M.A.N.] [PATCH 1/2] batman-adv: check return type of unicast packet preparations
by Linus Lüssing 26 Jul '13

26 Jul '13

batadv_send_skb_prepare_unicast(_4addr) might reallocate the skb's data. And if it tries to do so then this can potentially fail. We shouldn't continue working on this skb in such a case. Signed-off-by: Linus Lüssing <linus.luessing(a)web.de> --- send.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/send.c b/send.c index d2cc507..9c9cab6 100644 --- a/send.c +++ b/send.c @@ -266,11 +266,14 @@ static int batadv_send_skb_unicast(struct batadv_priv *bat_priv, switch (packet_type) { case BATADV_UNICAST: - batadv_send_skb_prepare_unicast(skb, orig_node); + if (!batadv_send_skb_prepare_unicast(skb, orig_node)) + goto out; break; case BATADV_UNICAST_4ADDR: - batadv_send_skb_prepare_unicast_4addr(bat_priv, skb, orig_node, - packet_subtype); + if (!batadv_send_skb_prepare_unicast_4addr(bat_priv, skb, + orig_node, + packet_subtype)) + goto out; break; default: /* this function supports UNICAST and UNICAST_4ADDR only. It -- 1.7.10.4

2 2

[B.A.T.M.A.N.] [PATCHv2] batman-adv: send GW_DEL event when the gw client mode is deselected
by Antonio Quartulli 21 Jul '13

21 Jul '13

From: Antonio Quartulli <antonio(a)open-mesh.com> Whenever the GW client mode is deselected, a DEL event has to be sent in order to tell userspace that the current gateway has been lost. Send the uevent on state change only if a gateway was currently selected. Reported-by: Marek Lindner <lindner_marek(a)yahoo.de> Signed-off-by: Antonio Quartulli <antonio(a)open-mesh.com> --- v2: - fixed typo - fixed kernel doc - fixed comment in sysfs.c gateway_client.c | 27 +++++++++++++++++++++++++++ gateway_client.h | 1 + sysfs.c | 4 ++++ 3 files changed, 32 insertions(+) diff --git a/gateway_client.c b/gateway_client.c index 588d27e..edf61da 100644 --- a/gateway_client.c +++ b/gateway_client.c @@ -188,6 +188,33 @@ next: return curr_gw; } +/** + * batadv_gw_check_client_stop - check if client mode has been switched off + * @bat_priv: the bat priv with all the soft interface information + * + * This function assumes the caller has checked that the gw state *is actually + * changing*. This function is not supposed to be called when there is no state + * change. + */ +void batadv_gw_check_client_stop(struct batadv_priv *bat_priv) +{ + struct batadv_gw_node *curr_gw; + + if (atomic_read(&bat_priv->gw_mode) != BATADV_GW_MODE_CLIENT) + return; + + curr_gw = batadv_gw_get_selected_gw_node(bat_priv); + if (!curr_gw) + return; + + /* if batman-adv is switching the gw client mode off and a gateway was + * already selected, send a DEL uevent + */ + batadv_throw_uevent(bat_priv, BATADV_UEV_GW, BATADV_UEV_DEL, NULL); + + batadv_gw_node_free_ref(curr_gw); +} + void batadv_gw_election(struct batadv_priv *bat_priv) { struct batadv_gw_node *curr_gw = NULL, *next_gw = NULL; diff --git a/gateway_client.h b/gateway_client.h index 7ef955a..b235cbf 100644 --- a/gateway_client.h +++ b/gateway_client.h @@ -20,6 +20,7 @@ #ifndef _NET_BATMAN_ADV_GATEWAY_CLIENT_H_ #define _NET_BATMAN_ADV_GATEWAY_CLIENT_H_ +void batadv_gw_check_client_stop(struct batadv_priv *bat_priv); void batadv_gw_deselect(struct batadv_priv *bat_priv); void batadv_gw_election(struct batadv_priv *bat_priv); struct batadv_orig_node * diff --git a/sysfs.c b/sysfs.c index 6ec145d..096b511 100644 --- a/sysfs.c +++ b/sysfs.c @@ -409,6 +409,10 @@ static ssize_t batadv_store_gw_mode(struct kobject *kobj, curr_gw_mode_str, buff); batadv_gw_deselect(bat_priv); + /* always call batadv_gw_check_client_stop() before changing the gateway + * state + */ + batadv_gw_check_client_stop(bat_priv); atomic_set(&bat_priv->gw_mode, (unsigned int)gw_mode_tmp); batadv_gw_tvlv_container_update(bat_priv); return count; -- 1.8.1.5

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 July 2013