B.A.T.M.A.N May 2013

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

16 participants
80 discussions

[B.A.T.M.A.N.] pull request net: batman-adv 2013-05-09
by Antonio Quartulli 11 May '13

11 May '13

here you have four fixes intended for net. 1/4 fixes the parsing of a string sent from userspace in order to avoid random memory access in case of string length of 0. 2/4 adds a check for the return value of pskb_trim_rcsum() in order to stop processing the skb in case of failure. 3/4 prevents DAT (the Distributed ARP Table) to send cached ARP replies when both the source and the destination of the snooped ARP request are local clients (meaning: directly connected to the node). This can confuse a bridge where batman-adv is enslaved. 4/4 fix a race condition in the main clean up procedure by reordering sub-components freeing function invocations. Please pull or let me know if there is any problem. Thanks a lot, Antonio The following changes since commit 4f924b2aa4d3cb30f07e57d6b608838edcbc0d88: if_cablemodem.h: Add parenthesis around ioctl macros (2013-05-08 13:13:30 -0700) are available in the git repository at: git://git.open-mesh.org/linux-merge.git tags/batman-adv-fix-for-davem for you to fetch changes up to a4361860351e87876aebd9595906d928ce8572c6: batman-adv: reorder clean up routine in order to avoid race conditions (2013-05-09 12:39:45 +0200) ---------------------------------------------------------------- Included changes: - fix parsing of user typed protocol string to avoid random memory access in some cases - check pskb_trim_rcsum() return value - prevent DAT from sending ARP replies when not needed - reorder the main clean up routine to prevent race conditions ---------------------------------------------------------------- Antonio Quartulli (2): batman-adv: make DAT drop ARP requests targeting local clients batman-adv: reorder clean up routine in order to avoid race conditions Marek Lindner (2): batman-adv: check proto length before accessing proto string buffer batman-adv: check return value of pskb_trim_rcsum() net/batman-adv/distributed-arp-table.c | 13 +++++++++++++ net/batman-adv/main.c | 18 +++++++++++++----- net/batman-adv/network-coding.c | 8 ++++++-- 3 files changed, 32 insertions(+), 7 deletions(-)

2 6

Re: [B.A.T.M.A.N.] Performance Comparison (Batman-adv. vs. 802.11s)
by Yeoh Chun-Yeow 11 May '13

11 May '13

Hi, Ajeet You can take a look on this paper [1] published in year 2010. [1] http://wwwtlc.iet.unipi.it/research/ggt-iswpc10.pdf ---- Chun-Yeow

1 0

[B.A.T.M.A.N.] Performance Comparison (Batman-adv. vs. 802.11s)
by ajeet singh 10 May '13

10 May '13

Dear all, I have done the set-up of Batman-adv. and 802.11s in my lab. I am comparing their working functionalities and performance measurements (throughput etc.). Have anyone tested batman-adv. and 802.11s with same test setup (I mean with same wireless chipsets)? Please provide me any related URLs. Thanks in advance. Thanks Ajeet

3 2

[B.A.T.M.A.N.] [PATCHv2 maint] batman-adv: fix VLAN Ethernet header access
by Antonio Quartulli 10 May '13

10 May '13

From: Antonio Quartulli <antonio(a)open-mesh.com> In interface_rx() skb->data cannot be directly used to access the Ethernet header because it point to the batman one. Signed-off-by: Antonio Quartulli <antonio(a)open-mesh.com> --- v2: - use vlan_eth_hdr() soft-interface.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/soft-interface.c b/soft-interface.c index 6f20d33..114d10d 100644 --- a/soft-interface.c +++ b/soft-interface.c @@ -320,7 +320,7 @@ void batadv_interface_rx(struct net_device *soft_iface, switch (ntohs(ethhdr->h_proto)) { case ETH_P_8021Q: - vhdr = (struct vlan_ethhdr *)skb->data; + vhdr = vlan_eth_hdr(skb); vid = ntohs(vhdr->h_vlan_TCI) & VLAN_VID_MASK; if (vhdr->h_vlan_encapsulated_proto != ethertype) -- 1.8.1.5

2 4

[B.A.T.M.A.N.] [PATCH next 1/2] batman-adv: use VLAN_VID_MASK instead of BATADV_VID_MASK
by Antonio Quartulli 09 May '13

09 May '13

VLAN_VID_MASK is already diefined in linux/if_vlan.h and therefore does not need to be privately re-defined Introduced in 0b1da1765fdb00ca5d53bc95c9abc70dfc9aae5b ("batman-adv: change VID semantic in the BLA code") Signed-off-by: Antonio Quartulli <ordex(a)autistici.org> --- bridge_loop_avoidance.c | 2 +- main.h | 2 -- 2 files changed, 1 insertion(+), 3 deletions(-) diff --git a/bridge_loop_avoidance.c b/bridge_loop_avoidance.c index e95cf4b..e9d8e0b 100644 --- a/bridge_loop_avoidance.c +++ b/bridge_loop_avoidance.c @@ -344,7 +344,7 @@ static void batadv_bla_send_claim(struct batadv_priv *bat_priv, uint8_t *mac, if (vid & BATADV_VLAN_HAS_TAG) skb = vlan_insert_tag(skb, htons(ETH_P_8021Q), - vid & BATADV_VID_MASK); + vid & VLAN_VID_MASK); skb_reset_mac_header(skb); skb->protocol = eth_type_trans(skb, soft_iface); diff --git a/main.h b/main.h index d575436..68c7ee0 100644 --- a/main.h +++ b/main.h @@ -172,8 +172,6 @@ enum batadv_vlan_flags { BATADV_VLAN_HAS_TAG = BIT(15), }; -/* mask needed to extract the vlan ID (12bits) from a 16bits variable */ -#define BATADV_VID_MASK 0x0FFF #define BATADV_PRINT_VID(vid) (vid & BATADV_VLAN_HAS_TAG ? \ (int)(vid & BATADV_VID_MASK) : -1) -- 1.8.1.5

2 3

[B.A.T.M.A.N.] [PATCH maint] batman-adv: fix VLAN Ethernet header access
by Antonio Quartulli 09 May '13

09 May '13

From: Antonio Quartulli <antonio(a)open-mesh.com> In interface_rx() skb->data cannot be directly used to access the Ethernet header because it point to the batman one. Signed-off-by: Antonio Quartulli <antonio(a)open-mesh.com> --- soft-interface.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/soft-interface.c b/soft-interface.c index 6f20d33..c8e1f8f 100644 --- a/soft-interface.c +++ b/soft-interface.c @@ -320,7 +320,7 @@ void batadv_interface_rx(struct net_device *soft_iface, switch (ntohs(ethhdr->h_proto)) { case ETH_P_8021Q: - vhdr = (struct vlan_ethhdr *)skb->data; + vhdr = (struct vlan_ethhdr *)ethhdr; vid = ntohs(vhdr->h_vlan_TCI) & VLAN_VID_MASK; if (vhdr->h_vlan_encapsulated_proto != ethertype) -- 1.8.1.5

1 0

[B.A.T.M.A.N.] [PATCHv2 maint] batman-adv: reorder clean up routine in order to avoid race conditions
by Antonio Quartulli 09 May '13

09 May '13

nc_worker accesses the originator table during its periodic work, but since the originator table is freed before stopping the worker this leads to a global protection fault. Fix this by killing the worker (in nc_free) before freeing the originator table. Moreover tidy up the entire clean up routine by running all the subcomponents freeing procedures first and then killing the TT and the originator tables at the end. Signed-off-by: Antonio Quartulli <ordex(a)autistici.org> --- v2: - based on top of maint - commit message reworded main.c | 16 ++++++++++++---- 1 file changed, 12 insertions(+), 4 deletions(-) diff --git a/main.c b/main.c index 9c620cd..6fefcb1 100644 --- a/main.c +++ b/main.c @@ -163,14 +163,22 @@ void batadv_mesh_free(struct net_device *soft_iface) batadv_vis_quit(bat_priv); batadv_gw_node_purge(bat_priv); - batadv_originator_free(bat_priv); batadv_nc_free(bat_priv); + batadv_dat_free(bat_priv); + batadv_bla_free(bat_priv); + /* free the TT and the originator tables only after having terminated all + * the other minor components which may use these structures for their + * purposes + */ batadv_tt_free(bat_priv); - batadv_bla_free(bat_priv); - - batadv_dat_free(bat_priv); + /* since the originator clean up routine is the one accessing the TT + * tables as well, invoke it as last step so that any race condition is + * avoided when the RCU callbacks scheduled by this function access the + * TT data + */ + batadv_originator_free(bat_priv); free_percpu(bat_priv->bat_counters); -- 1.8.1.5

2 1

[B.A.T.M.A.N.] [PATCH maint] batman-adv: make DAT drop ARP requests targeting local clients
by Antonio Quartulli 09 May '13

09 May '13

In the outgoing ARP request snooping routine in DAT, ARP Request sent by local clients which are supposed to be replied by other local clients can be silently dropped. The destination host will reply by itself through the LAN and therefore there is no need to involve DAT. Reported-by: Carlos Quijano <carlos(a)crqgestion.es> Signed-off-by: Antonio Quartulli <ordex(a)autistici.org> Tested-by: Carlos Quijano <carlos(a)crqgestion.es> --- distributed-arp-table.c | 13 +++++++++++++ 1 file changed, 13 insertions(+) diff --git a/distributed-arp-table.c b/distributed-arp-table.c index 8e15d96..2399920 100644 --- a/distributed-arp-table.c +++ b/distributed-arp-table.c @@ -837,6 +837,19 @@ bool batadv_dat_snoop_outgoing_arp_request(struct batadv_priv *bat_priv, dat_entry = batadv_dat_entry_hash_find(bat_priv, ip_dst); if (dat_entry) { + /* If the ARP request is destined for a local client the local + * client will answer itself. DAT would only generate a + * duplicate packet. + * + * Moreover, if the soft-interface is enslaved into a bridge, an + * additional DAT answer may trigger kernel warnings about + * a packet coming from the wrong port. + */ + if (batadv_is_my_client(bat_priv, dat_entry->mac_addr)) { + ret = true; + goto out; + } + skb_new = arp_create(ARPOP_REPLY, ETH_P_ARP, ip_src, bat_priv->soft_iface, ip_dst, hw_src, dat_entry->mac_addr, hw_src); -- 1.8.1.5

2 1

[B.A.T.M.A.N.] [PATCH maint] batman-adv: check return value of pskb_trim_rcsum()
by Marek Lindner 09 May '13

09 May '13

Reported-by: Sven Eckelmann <sven(a)narfation.org> Signed-off-by: Marek Lindner <lindner_marek(a)yahoo.de> --- network-coding.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/network-coding.c b/network-coding.c index f7c5430..e84629e 100644 --- a/network-coding.c +++ b/network-coding.c @@ -1514,6 +1514,7 @@ batadv_nc_skb_decode_packet(struct batadv_priv *bat_priv, struct sk_buff *skb, struct ethhdr *ethhdr, ethhdr_tmp; uint8_t *orig_dest, ttl, ttvn; unsigned int coding_len; + int err; /* Save headers temporarily */ memcpy(&coded_packet_tmp, skb->data, sizeof(coded_packet_tmp)); @@ -1568,8 +1569,11 @@ batadv_nc_skb_decode_packet(struct batadv_priv *bat_priv, struct sk_buff *skb, coding_len); /* Resize decoded skb if decoded with larger packet */ - if (nc_packet->skb->len > coding_len + h_size) - pskb_trim_rcsum(skb, coding_len + h_size); + if (nc_packet->skb->len > coding_len + h_size) { + err = pskb_trim_rcsum(skb, coding_len + h_size); + if (err) + return NULL; + } /* Create decoded unicast packet */ unicast_packet = (struct batadv_unicast_packet *)skb->data; -- 1.7.10.4

2 2

[B.A.T.M.A.N.] batman-adv 2013.2.0 released
by Marek Lindner 09 May '13

09 May '13

The B.A.T.M.A.N. team is delighted to announce its latest release - batman-adv 2013.2.0 - introducing a complete new submodule for optimized packet transport, new features and our regular set of bugfixes. 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-2013.2.0/ as well as prepackaged binaries in your distribution. Thanks ------ Thanks to all people sending in patches: * Antonio Quartulli <ordex(a)autistici.org> * Marek Lindner <lindner_marek(a)yahoo.de> * Martin Hundebøll <martin(a)hundeboll.net> * Matthias Schiffer <mschiffer(a)universe-factory.net> * Sasha Levin <sasha.levin(a)oracle.com> * Simon Wunderlich <siwu(a)hrz.tu-chemnitz.de> * Sven Eckelmann <sven(a)narfation.org> batman-adv ---------- The highlight of this release certainly is the newly added network coding support. Network Coding aims to improve transmission efficiency to preserve precious air time. The principle behind network coding builds upon wireless being a shared medium and exploiting this fact to our advantage (given that the situation allows to do so). The increased efficiency is achieved by combining two packets into a single transmission that is received by 2 destinations at the same time (due to the aforementioned shared medium characteristics). Tests have shown an increase in performance of up to 60% due to this clever mechanism. Exhaustive documentation has been made available for the curious reader explaining every detail regarding configuration, implementation and future work. Henceforward, batman-adv interfaces can be configured via the rtnl interface configuration API in addition to the known sysfs API. The rtnl API offers a generic approach to configure all modern Linux interfaces which is used by utilities like 'ip'. As part of the interface API restructuring batman-adv also gained the ability to automatically resolve duplicate interface usage. For example, adding an interface to batman-adv while the interface still is part of a bridge will now result in the removal of the interface from the bridge without manual intervention. The identification of own mac addresses in multi-meshes setups (multiple batX interfaces on the same host) has been fixed and a necessary length check when parsing OGMs has been added. batctl ------ To complement the network coding submodule a variety of new features went into batctl. It is now possible to enable/disable the network coding at runtime via a new commandline switch. Furthermore, batctl is able to print the network coding neighbor table and allows to configure the network coding log level for advanced debugging. The minimal unicast 4addr dissector had incorrect size expectations which was fixed. The 4addr dissector learned to print the unicast 4addr subtype in order to facilitate keeping the individual 4addr types apart. Happy routing, The B.A.T.M.A.N. team

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

B.A.T.M.A.N May 2013