B.A.T.M.A.N July 2008

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

11 participants
9 discussions

[B.A.T.M.A.N.] batman-adv-kernelland: Possible kernel deadlock?
by Scott Raynel 02 Aug '08

02 Aug '08

Hi all, I'm running batman-adv-kernelland r1102 on linux-2.6.21.5. I've found I can fairly reliably deadlock the kernel by disabling and re-enabling batman a few times, e.g. $ echo ath0 > /proc/net/batman-adv/interfaces wait a while... $ echo > /proc/net/batman-adv/interfaces wait a while and repeat. It appears that this is deadlocking the kernel somehow, as syscalls never return, e.g. 'ps' prints the header and then freezes. Our interface configuration system brings interfaces up and down on re- configure, so we'd like the batman interface to be able to withstand this use case. I will try to start looking through the code, but I thought I'd throw the problem out there to see what others more experienced with it think. Thanks all, -- Scott Raynel WAND Network Research Group Department of Computer Science University of Waikato New Zealand

3 4

[B.A.T.M.A.N.] Multi-card MESH
by Derek C 31 Jul '08

31 Jul '08

Hi, Is it possible to use BATMAN with two backbone network cards? I am hoping to make nodes with three radio cards - two at 5Ghz and one 2Ghz card for customer access. I have been playing around with openwrt on gateworks boards (4 miniPCI slots) and three radio cards. In my testing I've only been using typical AP / STA setups on the "backbone" radio cards and I'm using OSPF to handle dynamic routing. I'm handing my MESH "intelligence" (making one node's STA card associate to a close-by strong signal node's AP card) with Perl scripts. I've got problems with my test system not avoiding circular loops and ignoring the bandwidth supply node(s). I think I've gone on the wrong path trying to re-invent the MESH wheel so I'm now looking into BATMAN as a possibility. thanks for any information! Derek -- Derek C

2 2

[B.A.T.M.A.N.] Batman 0.3 + Freifunk setup questions
by Don Davis 28 Jul '08

28 Jul '08

Hello, I had batman 0.2 + freifunk working. Each router was 10.0.0.* and had the same wlan address. Something happened and then batman just started silently dying on the gateway router 10.0.0.1 (with the address 10.0.2.146 from the router connected to the internet). I erased nvram and reinstalled with separate 10.0.5.* addresses for the wlan and that seemed to work (with no http access to the routers and ssh via the 10.0.5.* addresses). I had wanted to stay with batman 0.2 because of the web interface. That's no longer practical. 0.3 is the preferred batman, so I should use it. I've taken a look at the docs: https://dev.open-mesh.net/svn/batman/trunk/batman/INSTALL and "RoutingVoodoo." assuming a fresh install of freifunk with *no OLSR* running what parameters would I want to set: adsl | 10.0.2.1 [home router ppoe] | freifunk 1 [gateway] does subnet matter? lan 10.0.0.1 wlan do I want 10.0.0.1 or 10.0.5.1? wan via dhcp? essid, bssid, channel all the same for ff routers myessid, 02:CA:FF:EE:BA:BE, 12, ad-hoc what batman parameters will I want to pass? shall I create an init script or is there one? 0.2 had the init and the minutely 'watchout4batman' | | | freifunk n non-gateway nodes lan 10.0.0.n? wlan? wan via dhcp? batman -p 10.0.0.1 -r 3 (or 10.0.2.146 freifunk 1's wan address) eth1 br0 vlan1? Anything else I should be aware of? Any other recommendations? (e.g. "Scrap that this is exactly how you do it: 1. 2. 3. "

3 9

[B.A.T.M.A.N.] [PATCH] Don't process the icmp packet further when we could not read it
by Sven Eckelmann 18 Jul '08

18 Jul '08

__copy_from_user returns the number of bytes it could not read when it fails. It makes no sense to process the data further if we don't have it in a complete form. access_ok will only check if it may be valid but not if it is definitly valid. Signed-off-by: Sven Eckelmann <sven.eckelmann(a)gmx.de> --- batman-adv-kernelland/batman-core/device.c | 4 +++- 1 files changed, 3 insertions(+), 1 deletions(-) diff --git a/batman-adv-kernelland/batman-core/device.c b/batman-adv-kernelland/batman-core/device.c index 6608f13..c401634 100644 --- a/batman-adv-kernelland/batman-core/device.c +++ b/batman-adv-kernelland/batman-core/device.c @@ -229,7 +229,9 @@ ssize_t bat_device_write(struct file *file, const char __user *buff, size_t len, if (!access_ok(VERIFY_READ, buff, sizeof(struct icmp_packet))) return -EFAULT; - __copy_from_user(&icmp_packet, buff, sizeof(icmp_packet)); + if (__copy_from_user(&icmp_packet, buff, sizeof(icmp_packet))) { + return -EFAULT; + } if ((icmp_packet.packet_type == BAT_ICMP) && (icmp_packet.msg_type == ECHO_REQUEST)) { -- 1.5.6.3

2 1

[B.A.T.M.A.N.] [PATCH] Move more compatibility code to compat.h
by Sven Eckelmann 18 Jul '08

18 Jul '08

Use the madwifi-like compat.h to keep as much of the #ifLINUX_VERSION.. stuff out of the code to increase the readability. Parts which weren't moved are the device creation/deletion, which is to different to move it to compat.h, and the module count. Signed-off-by: Sven Eckelmann <sven.eckelmann(a)gmx.de> --- batman-adv-kernelland/batman-core/compat.h | 16 ++++++++++++++-- batman-adv-kernelland/batman-core/device.c | 12 +----------- batman-adv-kernelland/batman-core/main.c | 1 + batman-adv-kernelland/batman-core/main.h | 1 - 4 files changed, 16 insertions(+), 14 deletions(-) diff --git a/batman-adv-kernelland/batman-core/compat.h b/batman-adv-kernelland/batman-core/compat.h index cb94665..5f58f8f 100644 --- a/batman-adv-kernelland/batman-core/compat.h +++ b/batman-adv-kernelland/batman-core/compat.h @@ -20,8 +20,9 @@ * of the Linux kernel. */ +#include <linux/version.h> /* LINUX_VERSION_CODE */ -#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,22) +#if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 22) #define skb_set_network_header(_skb, _offset) \ do { (_skb)->nh.raw = (_skb)->data + (_offset); } while (0) @@ -32,4 +33,15 @@ #define list_first_entry(ptr, type, member) \ list_entry((ptr)->next, type, member) -#endif +#endif /* LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 22) */ + + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 26) + +#define device_create_drvdata(_cls, _parent, _devt, _device, _fmt) \ + class_device_create(_cls, _parent, _devt, _device, _fmt) + +#define device_destroy(_cls, _device) \ + class_device_destroy(_cls, _device) + +#endif /* LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 26) */ diff --git a/batman-adv-kernelland/batman-core/device.c b/batman-adv-kernelland/batman-core/device.c index 18b026a..6608f13 100644 --- a/batman-adv-kernelland/batman-core/device.c +++ b/batman-adv-kernelland/batman-core/device.c @@ -30,9 +30,7 @@ #include "compat.h" -#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) -#include <linux/devfs_fs_kernel.h> -#else +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 0) static struct class *batman_class; #endif @@ -85,11 +83,7 @@ void bat_device_setup(void) debug_log(LOG_TYPE_WARN, "Could not register class 'batman-adv' \n"); return; } else { -#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,25) device_create_drvdata(batman_class, NULL, MKDEV(tmp_major, 0), NULL, "batman-adv"); -#else - class_device_create(batman_class, NULL, MKDEV(tmp_major, 0), NULL, "batman-adv"); -#endif } #endif @@ -106,11 +100,7 @@ void bat_device_destroy(void) #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) devfs_remove("batman-adv", 0); #else -#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,25) device_destroy(batman_class, MKDEV(Major, 0)); -#else - class_device_destroy(batman_class, MKDEV(Major, 0)); -#endif class_destroy(batman_class); #endif diff --git a/batman-adv-kernelland/batman-core/main.c b/batman-adv-kernelland/batman-core/main.c index bf2a52c..7628865 100644 --- a/batman-adv-kernelland/batman-core/main.c +++ b/batman-adv-kernelland/batman-core/main.c @@ -30,6 +30,7 @@ #include "vis.h" #include "types.h" #include "hash.h" +#include "compat.h" diff --git a/batman-adv-kernelland/batman-core/main.h b/batman-adv-kernelland/batman-core/main.h index 517fbc6..4208ae5 100644 --- a/batman-adv-kernelland/batman-core/main.h +++ b/batman-adv-kernelland/batman-core/main.h @@ -78,7 +78,6 @@ #include <linux/mutex.h> /* mutex */ #include <linux/module.h> /* needed by all modules */ -#include <linux/version.h> /* LINUX_VERSION_CODE */ #include <linux/netdevice.h> /* netdevice */ #include <linux/timer.h> /* timer */ #include <linux/if_ether.h> /* ethernet header */ -- 1.5.6.3

1 0

[B.A.T.M.A.N.] batman and Kernel 2.6.26-rc8&rc9
by Mark Kelly 18 Jul '08

18 Jul '08

The batman layer3 kernel module does not appear to compile on the 2.6.26-rc8 & rc9 kernels. Is this a known bug?

3 4

[B.A.T.M.A.N.] [PATCH] batman-adv-kernelland linux-2.6.21 compatibility
by Scott Raynel 18 Jul '08

18 Jul '08

Hi there, Not sure how you guys want patches submitted, so point me in the right direction if this is not it. Here's a simple patch that enables compilation on linux-2.6.21 (and possibly earlier, though I've tested 2.6.16 and it doesn't work). Cheers, -- Scott Raynel WAND Network Research Group Department of Computer Science University of Waikato New Zealand

3 3

[B.A.T.M.A.N.] Multipath-Routing BATMAN-ADV
by "Daniel Roßbach" 16 Jul '08

16 Jul '08

Hello, I try to research some possibilities to implement multipath-routing (MPR) to BATMAN Advanced. There are two different ways: 1. one-channel-MPR and 2. multi-channel-MPR. To the first: The idea is to use two paths through the mesh and to avoid crossed. I have build a simple model to show that it make sense. The model shows that it can take 67% of the time that needs BATMAN with a single path. I still implemented it into userspace of revision 1060. The next step is to proof, that it works...or that it doesn't work...! To the second: The idea is the use of multi paths with multi interfaces. Packets can be transmitted over two different ways without any interference problems. At the moment I'm analyzing the problems and the requirements for this. The better way is of course the second one. The advantage of using one channel is that there is just one wireless interface needed. Daniel -- Psssst! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger

2 1

[B.A.T.M.A.N.] BATMAN ADVANCED KERNEL LAND LOOP
by cipollone 03 Jul '08

03 Jul '08

Hi, I'm testing BATMAN Advanced Kernel Land and I've found a problem. I'm using BATMAN-adv over OpenWRT. My test is: 3 virtual machine 2 Ethernet card for each virtual machine 3 Vlan So we have a loop and I hoped that BATMAN solve this but... What's happen is that each virtual machine can send message to itself using the 2 interface. It passes because the BATMAN message are re-broadcasted and at a certain point 1 of machine start to communicate only with itself, disappearing from the network. I need helppp!!! Paolo Cipollone

3 4

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

B.A.T.M.A.N July 2008