B.A.T.M.A.N May 2011

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

14 participants
59 discussions

[B.A.T.M.A.N.] [PATCH] batctl: update readme
by Marek Lindner 26 May '11

26 May '11

Signed-off-by: Marek Lindner <lindner_marek(a)yahoo.de> --- README | 41 ++++++++++++++++++++--------------------- 1 files changed, 20 insertions(+), 21 deletions(-) diff --git a/README b/README index d11c6d3..442f8b6 100644 --- a/README +++ b/README @@ -18,8 +18,8 @@ settings. How does it work ? ================== -batctl uses the /dev/batman-adv device provided by the B.A.T.M.A.N. advanced -kernel module to inject custom icmp packets into the data flow. That's why +batctl uses the debugfs/batman_adv/bat0/socket device provided by the B.A.T.M.A.N. +advanced kernel module to inject custom icmp packets into the data flow. That's why ping and traceroute work almost like their IP based counterparts. Tcpdump was designed because B.A.T.M.A.N. advanced encapsulates all traffic within batman packets, so that the normal tcpdump would not recognize the packets. @@ -30,8 +30,8 @@ The bat-hosts file This file is simliar to the /etc/hosts file. You can write one MAC address and one host name per line. batctl will analyze the file to find the matching MAC -address to your provided host name. Host names are much easier to remember -than MAC addresses. ;) +address to your provided host name. Host names are much easier to remember than +MAC addresses. ;) batctl ping @@ -45,6 +45,7 @@ options: -h print this help -i interval in seconds -t timeout in seconds + -R record route Example: @@ -95,8 +96,9 @@ options: 4 - batman unicast packets 8 - batman broadcast packets 16 - batman vis packets - 32 - non batman packets - 33 - batman ogm & non batman packets + 32 - batman fragmented packets + 64 - non batman packets + 65 - batman ogm & non batman packets Example output for tcpdump: @@ -241,12 +243,12 @@ batctl interface display or modify the interface settings -Usage: batctl interface|if [none|interface] +Usage: batctl interface|if [add|del iface(s)] Example: $ batctl interface -[ active] eth0 fe:fe:00:00:01:01 +eth0: active batctl interval =============== @@ -270,7 +272,6 @@ Usage: batctl log|l Example: $ batctl log -r) [ 400] Received BATMAN packet via NB: fe:fe:00:00:02:01 IF: eth0 [fe:fe:00:00:01:01] (from OG: fe:fe:00:00:01:01 via prev OG: fe:fe:00:00:01:01 seqno 670, tq 245, TTL 49, V 8, IDF 1) [ 400] Drop packet: originator packet from myself (via neighbour) [ 400] Received BATMAN packet via NB: fe:fe:00:00:02:01 IF: eth0 [fe:fe:00:00:01:01] (from OG: fe:fe:00:00:02:01 via prev OG: fe:fe:00:00:02:01 seqno 545, tq 255, TTL 50, V 8, IDF 0) @@ -285,15 +286,14 @@ batctl loglevel display or modify the log level -Usage: batctl |ll [level] +Usage: batctl loglevel|ll [level] Example: $ batctl loglevel -[x] critical (0) -[x] warnings (1) -[x] notices (2) -[x] batman (4) -[x] routes (8) +[x] all debug output disabled (0) +[ ] messages related to routing / flooding / broadcasting (1) +[ ] messages related to route or hna added / changed / deleted (2) +[ ] all debug messages (3) batctl aggregation ================== @@ -334,18 +334,17 @@ Globally announced TT entries received via the mesh (translation table): * 86:4b:b2:d2:00:00 via fe:fe:00:00:07:01 * ca:a1:5b:e5:00:00 via fe:fe:00:00:06:01 -batctl vis_server +batctl vis_mode ================= display or modify the status of the VIS server -Usage: batctl vis_server|vs [enabled|disabled] +Usage: batctl vis_mode|vm [mode] Example: -$ batctl vis_server -[x] client mode (server disabled) -[ ] server mode (server enabled) +$ batctl vis_mode +client batctl vis_data @@ -410,7 +409,7 @@ vis-dot to png -------------- The vis dot output could then further be converted to an image of the topology -graph, e.g. with the help of the graphviz-tools. The according commands could +graph, e.g. with the help of the graphviz-tools. The according commands could then look like this: $ batctl vis_data dot > /tmp/graph.dot -- 1.7.2.3

2 1

[B.A.T.M.A.N.] Random bat_dbg cleanup patches
by Sven Eckelmann 22 May '11

22 May '11

Hi, I would like to propose following patches, but hadn't much time to check if they hide the real problem. Kind regards, Sven

2 3

[B.A.T.M.A.N.] [PATCH 01/14] batctl: Remove optimisation and debugging options
by Sven Eckelmann 22 May '11

22 May '11

2 27

[B.A.T.M.A.N.] [PATCH] batman-adv: a multiline comment should precede the variable it is describing
by Antonio Quartulli 22 May '11

22 May '11

This comment has been wrongly put after the variable it refers to and was also bad indented Signed-off-by: Antonio Quartulli <ordex(a)autistici.org> --- types.h | 8 ++++---- 1 files changed, 4 insertions(+), 4 deletions(-) diff --git a/types.h b/types.h index fab70e8..65b3222 100644 --- a/types.h +++ b/types.h @@ -246,10 +246,10 @@ struct frag_packet_list_entry { }; struct vis_info { - unsigned long first_seen; - struct list_head recv_list; - /* list of server-neighbors we received a vis-packet - * from. we should not reply to them. */ + unsigned long first_seen; + /* list of server-neighbors we received a vis-packet + * from. we should not reply to them. */ + struct list_head recv_list; struct list_head send_list; struct kref refcount; struct hlist_node hash_entry; -- 1.7.3.4

3 2

[B.A.T.M.A.N.] [PATCH] batman-adv: use is_broadcast_ether_addr() instead of compare_eth(.., brd_addr)
by Antonio Quartulli 22 May '11

22 May '11

Instead of comparing mac addresses with the broadcast address by means of compare_eth(), the is_broadcast_ether_addr() kernel function has to be used. Signed-off-by: Antonio Quartulli <ordex(a)autistici.org> --- routing.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/routing.c b/routing.c index 07f23ba..90ae6f0 100644 --- a/routing.c +++ b/routing.c @@ -661,7 +661,7 @@ void receive_bat_packet(const struct ethhdr *ethhdr, hard_iface->net_dev->dev_addr)) is_my_oldorig = 1; - if (compare_eth(ethhdr->h_source, broadcast_addr)) + if (is_broadcast_ether_addr(ethhdr->h_source)) is_broadcast = 1; } rcu_read_unlock(); -- 1.7.3.4

3 2

[B.A.T.M.A.N.] [PATCH] batman-adv: move smallest_signed_int(), seq_before() and seq_after() into main.h
by Antonio Quartulli 22 May '11

22 May '11

smallest_signed_int(), seq_before() and seq_after() are very useful functions that help to handle comparisons between sequence numbers. However they were only defined in vis.c. With this patch every batman-adv function will be able to use them. Signed-off-by: Antonio Quartulli <ordex(a)autistici.org> --- main.h | 16 ++++++++++++++++ vis.c | 16 ---------------- 2 files changed, 16 insertions(+), 16 deletions(-) diff --git a/main.h b/main.h index 3ca3941..90db244 100644 --- a/main.h +++ b/main.h @@ -181,4 +181,20 @@ static inline int compare_eth(void *data1, void *data2) #define atomic_dec_not_zero(v) atomic_add_unless((v), -1, 0) +/* Returns the smallest signed integer in two's complement with the sizeof x */ +#define smallest_signed_int(x) (1u << (7u + 8u * (sizeof(x) - 1u))) + +/* Checks if a sequence number x is a predecessor/successor of y. + * they handle overflows/underflows and can correctly check for a + * predecessor/successor unless the variable sequence number has grown by + * more then 2**(bitwidth(x)-1)-1. + * This means that for a uint8_t with the maximum value 255, it would think: + * - when adding nothing - it is neither a predecessor nor a successor + * - before adding more than 127 to the starting value - it is a predecessor, + * - when adding 128 - it is neither a predecessor nor a successor, + * - after adding more than 127 to the starting value - it is a successor */ +#define seq_before(x, y) ({typeof(x) _dummy = (x - y); \ + _dummy > smallest_signed_int(_dummy); }) +#define seq_after(x, y) seq_before(y, x) + #endif /* _NET_BATMAN_ADV_MAIN_H_ */ diff --git a/vis.c b/vis.c index c39f20c..e1ed552 100644 --- a/vis.c +++ b/vis.c @@ -30,22 +30,6 @@ #define MAX_VIS_PACKET_SIZE 1000 -/* Returns the smallest signed integer in two's complement with the sizeof x */ -#define smallest_signed_int(x) (1u << (7u + 8u * (sizeof(x) - 1u))) - -/* Checks if a sequence number x is a predecessor/successor of y. - * they handle overflows/underflows and can correctly check for a - * predecessor/successor unless the variable sequence number has grown by - * more then 2**(bitwidth(x)-1)-1. - * This means that for a uint8_t with the maximum value 255, it would think: - * - when adding nothing - it is neither a predecessor nor a successor - * - before adding more than 127 to the starting value - it is a predecessor, - * - when adding 128 - it is neither a predecessor nor a successor, - * - after adding more than 127 to the starting value - it is a successor */ -#define seq_before(x, y) ({typeof(x) _dummy = (x - y); \ - _dummy > smallest_signed_int(_dummy); }) -#define seq_after(x, y) seq_before(y, x) - static void start_vis_timer(struct bat_priv *bat_priv); /* free the info */ -- 1.7.3.4

3 6

[B.A.T.M.A.N.] [PATCH 01/13] vis: Remove optimisation and debugging options
by Sven Eckelmann 21 May '11

21 May '11

Distributions like Gentoo or Debian have to strip the options regarding the optimisation levels and debugging information to fulfil their policies. There is currently no valid reason why the Makefile should override it. Signed-off-by: Sven Eckelmann <sven(a)narfation.org> --- This patch depends on other patches submitted earlier: - vis: Remove obsolete creation of source packages - vis: Remove subversion/svk revision information Makefile | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/Makefile b/Makefile index c808391..6867a5f 100644 --- a/Makefile +++ b/Makefile @@ -26,7 +26,7 @@ endif endif CC = gcc -CFLAGS += -pedantic -Wall -W -O1 -g3 -std=gnu99 +CFLAGS += -pedantic -Wall -W -std=gnu99 EXTRA_CFLAGS = -DDEBUG_MALLOC -DMEMORY_USAGE -DREVISION_VERSION=$(REVISION_VERSION) LDFLAGS += -lpthread -- 1.7.5.1

1 12

[B.A.T.M.A.N.] [PATCH 01/14] batmand: Remove optimisation and debugging options
by Sven Eckelmann 21 May '11

21 May '11

Distributions like Gentoo or Debian have to strip the options regarding the optimisation levels and debugging information to fulfil their policies. There is currently no valid reason why the Makefile should override it. Signed-off-by: Sven Eckelmann <sven(a)narfation.org> --- This patch depends on other patches submitted earlier: - batmand: Fix FTBFS on second build attempt - batmand: Remove obsolete creation of source packages - batmand: Remove subversion/svk revision information - batmand: Use nproc to get number of available processors Makefile | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/Makefile b/Makefile index 4df2872..49342ac 100644 --- a/Makefile +++ b/Makefile @@ -29,7 +29,7 @@ endif #NO_POLICY_ROUTING = -DNO_POLICY_ROUTING CC = gcc -CFLAGS += -pedantic -Wall -W -Os -g3 -std=gnu99 +CFLAGS += -pedantic -Wall -W -std=gnu99 EXTRA_CFLAGS = -DDEBUG_MALLOC -DMEMORY_USAGE -DPROFILE_DATA $(NO_POLICY_ROUTING) -DREVISION_VERSION=$(REVISION_VERSION) LDFLAGS += -lpthread -- 1.7.5.1

1 13

Re: [B.A.T.M.A.N.] batman-adv bug 149: odd broadcast behavior -- generated CRC errors (Sven Eckelmann)
by David Beberman 20 May '11

20 May '11

First, my apologies on not monitoring the irc channel. A bit consumed at work. (if anybody happens to know how to get the broadcom bcm4329 driver to change channels in ad-hoc mode, please feel free to share. iwconfig ... channel XX command fails, but driver code looks correct. Target is HTC Desire). Regarding fragmentation -- this is strictly coming from the UPD/IPv6. It it not fragmenting in the wlan driver. I have fragmentation threshold set off. I'm assuming it is responding to the configuration parameters correctly. I'm using Omnipeek on a PC to sniff the packets. The antennas are about 2 feet apart. The network traffic is relatively low, in the 1 - 10% on average, with virtually no indication of corruption from hidden nodes or other sources under regular operation. That is, I only see the CRC error flags set with the broadcast packets at 802.11g data rates. This is the 802.11 packet CRC, nothing internal to it. The Omnipeek does not know about the batman-adv protocol. Also, I have other nodes setup with batman-adv which are repeating the broadcasts. I can see that only the packets without the CRC error flags are being rebroadcast. This indicates to me that the packets flagged as CRC errors by Omnipeek are also being dropped as error'ed packets by the other nodes. I do not see this problem with packets sent directly to the bcm4329 device. However, I will continue to try to see if I can get that to happen, would then mean it is bcm4329 driver and hardware problem. At the moment, using UDP broadcast with IPv4 directly onto the wlan driver without batman-adv (bat0), I do not see the behavior. I've been using my own code for this. Noticed netcat and nc last night. Will try piping with those tools today in both modes, with and without batman-adv to reverify what I'm seeing. I do not know how to do kernel debugging with kdbg/gdb across a USB port on the target side. I do not have JTAG or serial access. Google'd for how to setup the kernel to use USB port for kdbg, but did not find anything. Just lots of info on how to debug the USB driver itself. I'm using the cyanogenmod version of Android with the HTC Desire version of the Linux kernel, and the insertion of modules like batman-adv. I did verify that the modules are being built against the correct Linux kernel sources, just in case there was something silly going on. I'm building with NETCONSOLE enabled today and hope to be able to see driver messages sent over the WLAN using netcat on the host side. That should allow me to see both batman-adv and the bcm4329 trace messages. I should also be able to send the console to ttyUSB0, assuming I can get the HTC Desire USB to behave properly. I'll change the line to skb_copy as was suggested by <T_x>? and Sven. Will post my results. Thanks for the follow up. David Beberman

1 0

[B.A.T.M.A.N.] kfree_rcu patches
by Sven Eckelmann 20 May '11

20 May '11

Hi, Paul E. McKenney <paulmck(a)linux.vnet.ibm.com> proposed some changes which will reduce the amount of "kfree only" rcu-callback functions. The batman-adv compatibility layer has to add a similar functionality to support older kernels. I wanted to propose a solution which should be discussed before the kfree_rcu functionality enters Linus' kernel and our master branch. The amount of kfree_rcu_* function was reduced to support only a maximum offset of 256 to any 'struct rcu_head' inside a struct. This should be enough for the current version and supported architectures. Offsets of 4096 are supported by Paul E. McKenney's implementation, but our workaround would increase the size of the final module too much (personal opinion). ~~~ These patches should *NOT* enter the batman-adv repo right now ~~~~~ Kind regards, Sven

2 16

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 2011