B.A.T.M.A.N November 2018

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

14 participants
23 discussions

[B.A.T.M.A.N.] [PATCH 0/2] pull request for net: batman-adv 2018-11-14
by Simon Wunderlich 17 Nov '18

17 Nov '18

Hi David, here are some bugfixes which we would like to see integrated into net. Please pull or let me know of any problem! Thank you, Simon The following changes since commit 651022382c7f8da46cb4872a545ee1da6d097d2a: Linux 4.20-rc1 (2018-11-04 15:37:52 -0800) are available in the git repository at: git://git.open-mesh.org/linux-merge.git tags/batadv-net-for-davem-20181114 for you to fetch changes up to d7d8bbb40a5b1f682ee6589e212934f4c6b8ad60: batman-adv: Expand merged fragment buffer for full packet (2018-11-12 10:41:29 +0100) ---------------------------------------------------------------- Here are two batman-adv bugfixes: - Explicitly pad short ELP packets with zeros, by Sven Eckelmann - Fix packet size calculation when merging fragments, by Sven Eckelmann ---------------------------------------------------------------- Sven Eckelmann (2): batman-adv: Use explicit tvlv padding for ELP packets batman-adv: Expand merged fragment buffer for full packet net/batman-adv/bat_v_elp.c | 6 ++++-- net/batman-adv/fragmentation.c | 2 +- 2 files changed, 5 insertions(+), 3 deletions(-)

2 3

[B.A.T.M.A.N.] [PATCH 00/16] pull request for net-next: batman-adv 2018-11-14
by Simon Wunderlich 16 Nov '18

16 Nov '18

Hi David, here is a feature/cleanup pull request of batman-adv to go into net-next. Please pull or let me know of any problem! Thank you, Simon The following changes since commit 651022382c7f8da46cb4872a545ee1da6d097d2a: Linux 4.20-rc1 (2018-11-04 15:37:52 -0800) are available in the git repository at: git://git.open-mesh.org/linux-merge.git tags/batadv-next-for-davem-20181114 for you to fetch changes up to 016fd285682952b943641c074d1cc0d02b3e6889: batman-adv: enable MCAST by default at compile time (2018-11-12 10:41:51 +0100) ---------------------------------------------------------------- This feature/cleanup patchset includes the following patches: - Bump version strings, by Simon Wunderlich - Fixup includes, by Sven Eckelmann (3 patches) - Separate BATMAN_ADV_DEBUG from DEBUGFS, by Sven Eckelmann - Fixup tracing log documentation, by Sven Eckelmann - Use exclusive locks to secure netlink information dump transfers, by Sven Eckelmann (8 patches) - Move CRC16 dependency, by Sven Eckelmann - Enable MCAST by default, by Linus Luessing ---------------------------------------------------------------- Linus Lüssing (1): batman-adv: enable MCAST by default at compile time Simon Wunderlich (1): batman-adv: Start new development cycle Sven Eckelmann (14): batman-adv: Drop unused lockdep include batman-adv: Add includes for deprecation warning batman-adv: Improve includes for trace functionality batman-adv: Allow to use BATMAN_ADV_DEBUG without BATMAN_ADV_DEBUGFS batman-adv: Fix description for BATMAN_ADV_DEBUG batman-adv: Add inconsistent gateway netlink dump detection batman-adv: Add inconsistent hardif netlink dump detection batman-adv: Store modification counter via hash helpers batman-adv: Add inconsistent backbone netlink dump detection batman-adv: Add inconsistent claim netlink dump detection batman-adv: Add inconsistent dat netlink dump detection batman-adv: Add inconsistent local TT netlink dump detection batman-adv: Add inconsistent multicast netlink dump detection batman-adv: Move CRC16 dependency to BATMAN_ADV_BLA net/batman-adv/Kconfig | 10 +++-- net/batman-adv/bat_iv_ogm.c | 25 ++++++----- net/batman-adv/bat_v.c | 26 +++++++---- net/batman-adv/bridge_loop_avoidance.c | 82 +++++++++++++++++++--------------- net/batman-adv/debugfs.c | 2 + net/batman-adv/distributed-arp-table.c | 42 +++++++++-------- net/batman-adv/gateway_client.c | 3 ++ net/batman-adv/hard-interface.c | 3 ++ net/batman-adv/hash.c | 2 + net/batman-adv/hash.h | 6 +++ net/batman-adv/log.c | 60 ++++++++++++++----------- net/batman-adv/main.c | 3 ++ net/batman-adv/main.h | 3 +- net/batman-adv/multicast.c | 51 +++++++++++---------- net/batman-adv/netlink.c | 24 +++++----- net/batman-adv/trace.c | 2 - net/batman-adv/trace.h | 6 +++ net/batman-adv/translation-table.c | 41 +++++++++-------- net/batman-adv/types.h | 5 ++- 19 files changed, 236 insertions(+), 160 deletions(-)

2 17

[B.A.T.M.A.N.] broadcast storms
by Jake.Harris＠zf.com 13 Nov '18

13 Nov '18

I'm sure a similar question to this has been answered, but I am new to this mailing list format and don't know an efficient way to search https://lists.open-mesh.org/pipermail/b.a.t.m.a.n/ I'm having problems with broadcast messages effectively echoing around the network of 50ish nodes. I attached a few seconds of the batctl tcpdump output. I can't seem to find a pattern to what causes this, it tends to happen once every two or three weeks, the storm causes problems with the batman program where during the storm nodes drop all their neighbors (batctl n shows an empty list) indefinitely, which I have worked around that issue via a batch script that reloads batman if the neighbor list is empty. Reloading successfully reconnects to the network but the storm still persists. The only way I've found to fix this is to reboot all the nodes at the same time such that the whole network is down to kill the echos. I believe I had this problem much more frequently (every 4 days or so) a while ago on the same network when using discrete tcp destinations for the nodes to communicate, the storm frequency was reduced to what it is now by using broadcast packets and reducing the communication rate from 12 seconds to once every 40 seconds. Rebooting the nodes that are responsible for the echoing messages has no effect, I rebooted 192.168.1.230 before running tcpdump that is attached and as it shows packets from 230 continued to bounce around while the node was powered off and after it rejoined the network. It doesn't appear broadcast uses a time-to-live parameter to limit the hops the packets will make. I'm at a loss for a way to remedy this, there seems to only be multicast optimizations.

2 7

[B.A.T.M.A.N.] [PATCH 3.16 142/366] cfg80211: initialize sinfo in cfg80211_get_station
by Ben Hutchings 11 Nov '18

11 Nov '18

3.16.61-rc1 review patch. If anyone has any objections, please let me know. ------------------ From: Sven Eckelmann <sven(a)narfation.org> commit 3c12d0486856b9eb89c2a9ac336713cba90813e3 upstream. Most of the implementations behind cfg80211_get_station will not initialize sinfo to zero before manipulating it. For example, the member "filled", which indicates the filled in parts of this struct, is often only modified by enabling certain bits in the bitfield while keeping the remaining bits in their original state. A caller without a preinitialized sinfo.filled can then no longer decide which parts of sinfo were filled in by cfg80211_get_station (or actually the underlying implementations). cfg80211_get_station must therefore take care that sinfo is initialized to zero. Otherwise, the caller may tries to read information which was not filled in and which must therefore also be considered uninitialized. In batadv_v_elp_get_throughput's case, an invalid "random" expected throughput may be stored for this neighbor and thus the B.A.T.M.A.N V algorithm may switch to non-optimal neighbors for certain destinations. Fixes: 7406353d43c8 ("cfg80211: implement cfg80211_get_station cfg80211 API") Reported-by: Thomas Lauer <holminateur(a)gmail.com> Reported-by: Marcel Schmidt <ff.z-casparistrasse(a)mailbox.org> Cc: b.a.t.m.a.n(a)lists.open-mesh.org Signed-off-by: Sven Eckelmann <sven(a)narfation.org> Signed-off-by: Johannes Berg <johannes(a)sipsolutions.net> Signed-off-by: Ben Hutchings <ben(a)decadent.org.uk> --- net/wireless/util.c | 2 ++ 1 file changed, 2 insertions(+) --- a/net/wireless/util.c +++ b/net/wireless/util.c @@ -1566,6 +1566,8 @@ int cfg80211_get_station(struct net_devi if (!rdev->ops->get_station) return -EOPNOTSUPP; + memset(sinfo, 0, sizeof(*sinfo)); + return rdev_get_station(rdev, dev, mac_addr, sinfo); } EXPORT_SYMBOL(cfg80211_get_station);

1 0

[B.A.T.M.A.N.] [PATCH maint] batman-adv: Expand merged fragment buffer for full packet
by Sven Eckelmann 07 Nov '18

07 Nov '18

The complete size ("total_size") of the fragmented packet is stored in the fragment header and in the size of the fragment chain. When the fragments are ready for merge, the skbuff's tail of the first fragment is expanded to have enough room after the data pointer for at least total_size. This means that it gets expanded by total_size - first_skb->len. But this is ignoring the fact that after expanding the buffer, the fragment header is pulled by from this buffer. Assuming that the tailroom of the buffer was already 0, the buffer after the data pointer of the skbuff is now only total_size - len(fragment_header) large. When the merge function is then processing the remaining fragments, the code to copy the data over to the merged skbuff will cause an skb_over_panic when it tries to actually put enough data to fill the total_size bytes of the packet. The size of the skb_pull must therefore also be taken into account when the buffer's tailroom is expanded. Fixes: 9b3eab61754d ("batman-adv: Receive fragmented packets and merge") Reported-by: Martin Weinelt <martin(a)darmstadt.freifunk.net> Co-authored-by: Linus Lüssing <linus.luessing(a)c0d3.blue> Signed-off-by: Sven Eckelmann <sven(a)narfation.org> --- Cc: Martin Weinelt <martin(a)darmstadt.freifunk.net> --- net/batman-adv/fragmentation.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/net/batman-adv/fragmentation.c b/net/batman-adv/fragmentation.c index 0fddc171..5b71a289 100644 --- a/net/batman-adv/fragmentation.c +++ b/net/batman-adv/fragmentation.c @@ -275,7 +275,7 @@ batadv_frag_merge_packets(struct hlist_head *chain) kfree(entry); packet = (struct batadv_frag_packet *)skb_out->data; - size = ntohs(packet->total_size); + size = ntohs(packet->total_size) + hdr_size; /* Make room for the rest of the fragments. */ if (pskb_expand_head(skb_out, 0, size - skb_out->len, GFP_ATOMIC) < 0) { -- 2.19.1

1 0

[B.A.T.M.A.N.] [PATCH] batman-adv: enable MCAST by default at compile time
by Linus Lüssing 06 Nov '18

06 Nov '18

Thanks to rigorous testing in wireless community mesh networks several issues with multicast entries in the translation table were found and fixed in the last 1.5 years. Now we see the first larger networks (a few hundred nodes) with a batman-adv version with multicast optimizations enabled arising, with no TT / multicast optimization related issues so far. Therefore it seems safe to enable multicast optimizations by default not only in kernel module builds but also for in-kernel builds. Signed-off-by: Linus Lüssing <linus.luessing(a)c0d3.blue> --- net/batman-adv/Kconfig | 1 + 1 file changed, 1 insertion(+) diff --git a/net/batman-adv/Kconfig b/net/batman-adv/Kconfig index f75816f5..b70b96cd 100644 --- a/net/batman-adv/Kconfig +++ b/net/batman-adv/Kconfig @@ -82,6 +82,7 @@ config BATMAN_ADV_NC config BATMAN_ADV_MCAST bool "Multicast optimisation" depends on BATMAN_ADV && INET && !(BRIDGE=m && BATMAN_ADV=y) + default y help This option enables the multicast optimisation which aims to reduce the air overhead while improving the reliability of -- 2.11.0

2 1

[B.A.T.M.A.N.] [RFC 0/2] batman-adv: netlink restructuring, part 2
by Sven Eckelmann 05 Nov '18

05 Nov '18

Hi, Jiri Pirko called the batman-adv configuration interface a while back "a huge mistake" [1] and suggested genl as alternative. But it is not completely clear what he meant by that. I had a look at the implementation in team and devlink (he is maintainer of both). Unfortunately, both work in slightly different ways. But they seem to have in common that there are different classes of configuration options/params which then have their own set of commands. But each set of command works for multiple options/params. I have tried to replicate something similar with this implementation. It might be important to point out that it requires multiple messages when trying to change multiple options (e.g. modification of gw_bandwidth_down and gw_bandwidth_up require two messages). This should be more like devlink and not like team. But it directly sends the new options back via genlmsg_multicast to all subscribers of the "config" mcast group. There is also currently no nesting of attributes. The code is basically around the triplet name-type-value. All three are required for set operations, name is required for get operations and nothing of the triplet is required for dump operations (because it will create messages containing this triplet). The different classes of options also require as input: * mesh interface (the batadv device) - meshif_index * hard interface specific (the enslaved devices) - meshif_index - hardif_index * vlan specific (vlan on top of the mesh interface) - meshif_index - vlan id Luckily, Jiri Pirko told us that he will have a look at patches. Here are two patches with an initial implementation. I might have to split them futher - maybe three patches for the infrastructure part and a couple for the import of the sysfs options in the new infrastructure. I am happy to receive suggestions how I should split them. There is also some potential in reducing the shared code for the .doit commands by moving the BATADV_ATTR_MESH_IFINDEX -> bat_priv code to .pre and .post handlers. Kind regards, Sven [1] https://www.open-mesh.org/issues/300 Sven Eckelmann (2): batman-adv: Add infrastructure for netlink config manipulation batman-adv: Convert existing sysfs options to netlink include/uapi/linux/batman_adv.h | 55 + net/batman-adv/Makefile | 1 + net/batman-adv/netlink.c | 50 +- net/batman-adv/netlink.h | 6 + net/batman-adv/netlink_cfg.c | 2096 +++++++++++++++++++++++++++++++ net/batman-adv/netlink_cfg.h | 78 ++ 6 files changed, 2280 insertions(+), 6 deletions(-) create mode 100644 net/batman-adv/netlink_cfg.c create mode 100644 net/batman-adv/netlink_cfg.h -- 2.19.1

2 7

[B.A.T.M.A.N.] [RFC 0/2] batctl: netlink restructuring, part 3
by Sven Eckelmann 04 Nov '18

04 Nov '18

As mentioned earlier, the netlink config restructuring in batctl will be splitted in multiple steps. The first step was already merged and the RFC changes for the netlink interface in batman-adv were just posted. Since it doesn't make sense to have a full implementation in batctl when the batman-adv part is not yet in a finished state. But it also is useless to have only the batman-adv RFC part without a userspace tool. The (by default disabled) config subcommand is (my) solution for this problem. It allows developers to play around with this interface and assign new values for specific keys and to read the current config state. Make sure that you enable it during the build: $ make clean; make CONFIG_config=y Kind regards, Sven Sven Eckelmann (2): batctl: Add generic netlink configuration subcommand batctl: Add support for config mcast group in event monitor Makefile | 1 + batman_adv.h | 55 ++++++ config.c | 542 +++++++++++++++++++++++++++++++++++++++++++++++++++ event.c | 100 ++++++++++ man/batctl.8 | 37 ++++ netlink.c | 3 + 6 files changed, 738 insertions(+) create mode 100644 config.c -- 2.19.1

1 2

[B.A.T.M.A.N.] [PATCH] batctl: Fix minor typos in manpage
by Sven Eckelmann 03 Nov '18

03 Nov '18

Signed-off-by: Sven Eckelmann <sven(a)narfation.org> --- man/batctl.8 | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/man/batctl.8 b/man/batctl.8 index d5fc5302ffd5d6a5ab1dfcb164c57a9f7f8e398f..3e210281fcf4d5d3c1d76b1f9c02459f4828979e 100644 --- a/man/batctl.8 +++ b/man/batctl.8 @@ -338,7 +338,7 @@ given batctl will not replace the MAC addresses with bat\-host names in the outp This command starts a throughput test entirely controlled by batman module in kernel space: the computational resources needed to align memory and copy data between user and kernel space that are required by other user space tools may -represent a bootleneck on some low profile device. +represent a bottleneck on some low profile device. The test consist of the transfer of 14 MB of data between the two nodes. The protocol used to transfer the data is somehow similar to TCP, but simpler: some @@ -346,9 +346,9 @@ TCP features are still missing, thus protocol performances could be worst. Since a fixed amount of data is transferred the experiment duration depends on the network conditions. The experiment can be interrupted with CTRL + C. At the end of a successful experiment the throughput in KBytes per second is returned, -togheter with the experiment duration in millisecond and the amount of bytes +together with the experiment duration in millisecond and the amount of bytes transferred. If too many packets are lost or the specified MAC address is not -reachable, a message notifing the error is returned instead of the result. +reachable, a message notifying the error is returned instead of the result. .RE .br .SH FILES

1 1

[B.A.T.M.A.N.] [PATCH v3 00/42] batctl: pre-netlink restructuring, part 1
by Sven Eckelmann 03 Nov '18

03 Nov '18

Hi, the batctl command already has support to get (previously debugfs) tables via netlink. It does this while still being able to fall back to the old debugfs tables on kernels which don't support the new generic netlink family. Something similar should be done in the future for the settings which are currently part of sysfs. But we can already see that the current integration of netlink in batctl only done in a single file - netlink.c But this file is already starting to be so big that working with it is rather cumbersome. So as first steps: * restructure command registration * move commands in separate files (which should store their actual implementation) * add a new helper command to receive multicast group messages from the kernel More things will follow in the future. Changes: v3: * correctly split ap_isolation and bonding patch * fix copyright headers for new files * move more routing_algo specific code to its command specific file * fix build problems with musl in OpenWrt 18.06 * switch Makefile's OBJ + OBJ_BISECT to obj-y and obj-n * add way to "freely" disable some commands for size optimization reasons - it is for example now possible to reduce the size of the batctl on mips_24kc_musl from 64604 bytes to 14252 bytes when only the interface commands are required (I know, stupid example) - the binary to generate the debug tables output for the same architectur used 26988 bytes v2: * Add missing orig_interval.c Kind regards, Sven Sven Eckelmann (42): batctl: Drop unused define SOCKET_PATH_FMT batctl: Use common code organization for statistics batctl: Drop legacy vis_* related warning messages batctl: Move loglevel command to separate file batctl: Move log command to separate file batctl: Move gw_mode command to separate file batctl: Move routing_algo command to separate file batctl: Rename tp_meter to throughputmeter batctl: Introduce datastructure for subcommands batctl: Add per command flags batctl: Use command structure for remaining subcommands batctl: Use getopt to parse main options batctl: Store usage line next to command batctl: Prepare command infrastructure for shared functions batctl: Add type to command to structure usage output batctl: Convert debug table to command infrastructure batctl: Convert sysfs settings to command infrastructure batctl: Move backbonetable debug table to own file batctl: Move claimtable debug table to own file batctl: Move dat_cache debug table to own file batctl: Move gateways debug table to own file batctl: Move mcast_flags debug table to own file batctl: Move nc_nodes debug table to own file batctl: Move neighbors debug table to own file batctl: Move originators debug table to own file batctl: Move transglobal debug table to own file batctl: Move translocal debug table to own file batctl: Move aggregation setting to own file batctl: Move ap_isolation setting to own file batctl: Move bonding setting to own file batctl: Move bridge_loop_avoidance setting to own file batctl: Move distributed_arp_table setting to own file batctl: Move fragmentation setting to own file batctl: Move isolation_mark setting to own file batctl: Move multicast_mode setting to own file batctl: Move network_coding setting to own file batctl: Move orig_interval setting to own file batctl: Use external netlink socket for debug tables batctl: Move routing_algo specific code it command source file batctl: Add command to monitor for netlink events batctl: Simplify enabling for bisect_iv subcommand batctl: Allow to enable/disable subcommands Makefile | 91 ++- translate.h => aggregation.c | 15 +- ap_isolation.c | 33 + backbonetable.c | 126 ++++ bisect_iv.c | 6 +- bisect_iv.h | 4 - traceroute.h => bonding.c | 15 +- bridge_loop_avoidance.c | 33 + claimtable.c | 131 ++++ dat_cache.c | 147 +++++ debug.c | 218 +------ debug.h | 38 +- distributed_arp_table.c | 33 + event.c | 246 ++++++++ ping.h => fragmentation.c | 15 +- functions.c | 28 +- functions.h | 1 - gateways.c | 175 +++++ gw_mode.c | 171 +++++ interface.c | 23 +- ioctl.h | 28 - isolation_mark.c | 35 + log.c | 73 +++ loglevel.c | 147 +++++ main.c | 274 ++++---- main.h | 65 +- man/batctl.8 | 4 + mcast_flags.c | 180 ++++++ multicast_mode.c | 33 + interface.h => nc_nodes.c | 14 +- neighbors.c | 142 +++++ netlink.c | 1332 ++++----------------------------------- netlink.h | 55 +- network_coding.c | 33 + orig_interval.c | 35 + originators.c | 228 +++++++ ping.c | 10 +- routing_algo.c | 263 ++++++++ ioctl.c => statistics.c | 10 +- sys.c | 432 +------------ sys.h | 39 +- tcpdump.c | 5 +- tcpdump.h | 2 - tp_meter.c => throughputmeter.c | 10 +- tp_meter.h | 28 - traceroute.c | 10 +- transglobal.c | 162 +++++ translate.c | 8 +- translocal.c | 158 +++++ 49 files changed, 3156 insertions(+), 2208 deletions(-) rename translate.h => aggregation.c (67%) create mode 100644 ap_isolation.c create mode 100644 backbonetable.c rename traceroute.h => bonding.c (68%) create mode 100644 bridge_loop_avoidance.c create mode 100644 claimtable.c create mode 100644 dat_cache.c create mode 100644 distributed_arp_table.c create mode 100644 event.c rename ping.h => fragmentation.c (67%) create mode 100644 gateways.c create mode 100644 gw_mode.c delete mode 100644 ioctl.h create mode 100644 isolation_mark.c create mode 100644 log.c create mode 100644 loglevel.c create mode 100644 mcast_flags.c create mode 100644 multicast_mode.c rename interface.h => nc_nodes.c (73%) create mode 100644 neighbors.c create mode 100644 network_coding.c create mode 100644 orig_interval.c create mode 100644 originators.c create mode 100644 routing_algo.c rename ioctl.c => statistics.c (90%) rename tp_meter.c => throughputmeter.c (97%) delete mode 100644 tp_meter.h create mode 100644 transglobal.c create mode 100644 translocal.c -- 2.11.0

1 43

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

B.A.T.M.A.N November 2018