Hi all, i'm starting a B.A.T.M.A.N. mesh testbed, i've read some messages to find some answers but no luck. I'll try to explain what I've done and the errors i get..hope i get some help in here.
From the beginning, I am using batman-adv 3.2 compiled along with batctl.
Only one wireless interface (wlan0) on each linux box (5 boxes by now) Here are commands I fire on each node:
$ insmod batman-adv.ko $ iwconfig wlan0 mode ad-hoc essid BATMAN_NET $ echo wlan0 > /proc/net/batman-adv/interfaces $ ifconfig wlan0 192.168.1.xxx up $ ifconfig bat0 up
I get a new bat0 interface, iwconfig shows nodes properly part of the ad-hoc network. I see correct $batctl o output but i cannot $batctl ping $mac_address
There are no bridges between interfaces. I just want to run batman-adv on wlan0. Do I need to set an IP address for wlan0, for bat0 or for both? Or I have to bridge bat0 and wlan0 together?
Francesco wrote:
Hi all, i'm starting a B.A.T.M.A.N. mesh testbed, i've read some messages to find some answers but no luck. I'll try to explain what I've done and the errors i get..hope i get some help in here.
From the beginning, I am using batman-adv 3.2 compiled along with batctl.
There is nothing like batman-adv 3.2. Where did you got it from?
Only one wireless interface (wlan0) on each linux box (5 boxes by now) Here are commands I fire on each node:
$ insmod batman-adv.ko $ iwconfig wlan0 mode ad-hoc essid BATMAN_NET $ echo wlan0 > /proc/net/batman-adv/interfaces $ ifconfig wlan0 192.168.1.xxx up $ ifconfig bat0 up
You must give bat0 also an ip to use IPv4(/6) over it... in most cases only bat0 as it is just the hardware interface the bat0 interface is using. If you only want to test raw ethernet data over bat0 this should work.
I get a new bat0 interface, iwconfig shows nodes properly part of the ad-hoc network. I see correct $batctl o output but i cannot $batctl ping $mac_address
Can you please give the originator table and send some small dumps (tcpdump - w, batctl td) of both nodes when you try to send a batctl ping? What mac address did you use?
Best regards, Sven
Sven Eckelmann wrote:
You must give bat0 also an ip to use IPv4(/6) over it... in most cases only bat0 as it is just the hardware interface the bat0 interface is using. If you only want to test raw ethernet data over bat0 this should work.
Meant that wlan0 is just the hardware interface bat0 is using.
Or I have to bridge bat0 and wlan0 together?
No, you should not bridge them together.
Best regards, Sven
Hi Sven,
There is nothing like batman-adv 3.2. Where did you got it from?
Sorry, my fault, i meant 0.3.2 release.
Only one wireless interface (wlan0) on each linux box (5 boxes by now) Here are commands I fire on each node:
$ insmod batman-adv.ko $ iwconfig wlan0 mode ad-hoc essid BATMAN_NET $ echo wlan0 > /proc/net/batman-adv/interfaces $ ifconfig wlan0 192.168.1.xxx up $ ifconfig bat0 up
You must give bat0 also an ip to use IPv4(/6) over it... in most cases only bat0 as it is just the hardware interface the bat0 interface is using. If you only want to test raw ethernet data over bat0 this should work.
Ok now i have assigned ip addresses only to bat0 interface, i do ping nodes with ping command. here are some output:
$batctl o
Originator (#/255) Nexthop [outgoingIF]: Potential nexthops ... [B.A.T.M.A.N. Adv 0.1, MainIF/MAC: wlan0/00:0b:6b:34:88:94] 00:0c:42:2c:d8:fc (221) 00:1e:4c:02:94:95 [ wlan0]: 00:0c:42:2c:d8:fc (170) 00:0b:6b:2f:53:09 (148) 00:0b:6b:2f:53:03 (199) 00:1e:4c:02:94:95 (221) 00:0b:6b:2f:53:03 (220) 00:0b:6b:2f:53:03 [ wlan0]: 00:0b:6b:2f:53:03 (220) 00:0b:6b:2f:53:09 (146) 00:0c:42:2c:d8:fc (157) 00:1e:4c:02:94:95 (201) 00:1e:4c:02:94:95 (234) 00:1e:4c:02:94:95 [ wlan0]: 00:1e:4c:02:94:95 (234) 00:0b:6b:2f:53:03 (191) 00:0b:6b:2f:53:09 (146) 00:0c:42:2c:d8:fc (157) 00:0b:6b:2f:53:09 (194) 00:1e:4c:02:94:95 [ wlan0]: 00:0b:6b:2f:53:09 (157) 00:0b:6b:2f:53:03 (183) 00:0c:42:2c:d8:fc (138) 00:1e:4c:02:94:95 (19
It shows 4 of 5 nodes I have online at this very moment.
Say i fire ping from the box missing:
root@francesco-desktop:/home/francesco# ping 192.168.1.2 PING 192.168.1.2 (192.168.1.2) 56(84) bytes of data. 64 bytes from 192.168.1.2: icmp_seq=1 ttl=64 time=3.16 ms 64 bytes from 192.168.1.2: icmp_seq=2 ttl=64 time=3.55 ms 64 bytes from 192.168.1.2: icmp_seq=3 ttl=64 time=8.01 ms ^C --- 192.168.1.2 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 2011ms rtt min/avg/max/mdev = 3.167/4.911/8.012/2.198 ms
Everything ok. But if I try to batctl ping a node that I can see with $batctl o:
root@francesco-desktop:/home/francesco# batctl ping 00:0b:6b:2f:53:03 PING 00:0b:6b:2f:53:03 (00:0b:6b:2f:53:03) 19(47) bytes of data
From 00:0b:6b:2f:53:03: Destination Host Unreachable (icmp_seq 6170) From 00:0b:6b:2f:53:03: Destination Host Unreachable (icmp_seq 6170) From 00:0b:6b:2f:53:03: Destination Host Unreachable (icmp_seq 6170)
^C--- 00:0b:6b:2f:53:03 ping statistics --- 3 packets transmitted, 0 received, 100% packet loss rtt min/avg/max/mdev = 0.000/0.000/0.000/0.000 ms
What am I missing? I will post dump files tomorrow when i get back in the lab.
Thanks.
I get a new bat0 interface, iwconfig shows nodes properly part of the ad-hoc network. I see correct $batctl o output but i cannot $batctl ping $mac_address
Can you please give the originator table and send some small dumps (tcpdump
w, batctl td) of both nodes when you try to send a batctl ping? What mac address did you use?
Best regards, Sven
B.A.T.M.A.N mailing list B.A.T.M.A.N@lists.open-mesh.net https://lists.open-mesh.net/mm/listinfo/b.a.t.m.a.n
Francesco wrote:
Hi Sven,
There is nothing like batman-adv 3.2. Where did you got it from?
Sorry, my fault, i meant 0.3.2 release.
There is no 0.3.2 release. batman-adv hasn't reached 0.2 yet. See http://downloads.open-mesh.net/batman/releases/ for the current releases of batman and batman-adv. The current version of batman-adv in development is <0.2-beta (it is not known yet if there will be real a beta or if the next release is simply 0.2). So the current snapshot of batman-adv would be "B.A.T.M.A.N. advanced 0.2-beta r1453 (compatibility version 8)"
[...]
It shows 4 of 5 nodes I have online at this very moment.
Say i fire ping from the box missing:
root@francesco-desktop:/home/francesco# ping 192.168.1.2 PING 192.168.1.2 (192.168.1.2) 56(84) bytes of data. 64 bytes from 192.168.1.2: icmp_seq=1 ttl=64 time=3.16 ms 64 bytes from 192.168.1.2: icmp_seq=2 ttl=64 time=3.55 ms 64 bytes from 192.168.1.2: icmp_seq=3 ttl=64 time=8.01 ms ^C --- 192.168.1.2 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 2011ms rtt min/avg/max/mdev = 3.167/4.911/8.012/2.198 ms
Everything ok. But if I try to batctl ping a node that I can see with $batctl o:
root@francesco-desktop:/home/francesco# batctl ping 00:0b:6b:2f:53:03 PING 00:0b:6b:2f:53:03 (00:0b:6b:2f:53:03) 19(47) bytes of data
From 00:0b:6b:2f:53:03: Destination Host Unreachable (icmp_seq 6170) From 00:0b:6b:2f:53:03: Destination Host Unreachable (icmp_seq 6170) From 00:0b:6b:2f:53:03: Destination Host Unreachable (icmp_seq 6170)
If you say that normal ping over bat0 works I would guess that you are using a wrong combination of batctl and kernel module. Please use bleeding edge svn batctl only with bleeding edge svn batman-adv-kernelland module.
Best regards, Sven
Hi Sven,now I see what was wrong, I'm really feeling so stupid.. :) It was misleading that I reached the batman-adv download page ( http://downloads.open-mesh.net/batman/stable/sources/batman-adv-kernelland/ ) browsing batman 0.3.2 download page ( http://www.open-mesh.org/wiki/2009-06-13-batman-0-3-2-release ) and I got confused.
Thanks for your kind reply.
I didn't read enough about batctl and its use against most recent batman-adv version. My fault.
As soon as i get back to the lab I'll try latetest version of both batman-adv and batctl and see what comes out.
Francesco wrote:
Hi Sven,now I see what was wrong, I'm really feeling so stupid.. :) It was misleading that I reached the batman-adv download page ( http://downloads.open-mesh.net/batman/stable/sources/batman-adv-kernelland/ ) browsing batman 0.3.2 download page ( http://www.open-mesh.org/wiki/2009-06-13-batman-0-3-2-release ) and I got confused.
No reason to feel stupid. You aren't the first person which got confused by the different implementations of the B.A.T.M.A.N. algorithm.
I didn't read enough about batctl and its use against most recent batman-adv version. My fault.
As soon as i get back to the lab I'll try latetest version of both batman-adv and batctl and see what comes out.
There is currently nothing like a stable branch. So if you want to have 0.1 with some backported stabilization patches you should look at the Debian package of batman-adv-kernelland[1] and batman-adv-battool[2]. The source packages of both packages include the separated patches[3].
But if you are currently in lab testing I would recommend to use both from svn as batctl has some nice new features and changes in batman-adv-kernelland are also looking quite interesting.
Best regards, Sven
[1] http://packages.debian.org/sid/batman-adv-source [2] http://packages.debian.org/sid/batman-adv-battool [3] http://git.debian.org/?p=collab-maint/batman-adv-kernelland.git;a=tree;f=deb...
HI all, I've now downloaded and $make latest batman-adv trunk from $svn co http://downloads.open-mesh.net/svn/batman/trunk/batman-adv-kernelland/ I have issues with batctl trunk: $svn co http://downloads.open-mesh.net/svn/batman/trunk/batctl/
Download is ok but when i fire $ make
root@francesco-desktop:/home/francesco/batctl# make make -j 1 batctl make[1]: Entering directory `/home/francesco/batctl' CC main.o In file included from main.c:34: tcpdump.h:23:25: error: list-batman.h: No such file or directory In file included from main.c:34: tcpdump.h:33: error: field ‘list’ has incomplete type In file included from main.c:35: bisect.h:30: error: field ‘event_list’ has incomplete type bisect.h:31: error: field ‘rt_table_list’ has incomplete type bisect.h:36: error: field ‘list’ has incomplete type bisect.h:47: error: field ‘list’ has incomplete type bisect.h:65: error: field ‘list’ has incomplete type make[1]: *** [main.o] Error 1 make[1]: Leaving directory `/home/francesco/batctl' make: *** [all] Error 2
I suppose there was no need to uninstall old batctl i did compiled..or not?
Thanks.
On Wed, Oct 21, 2009 at 08:54:02AM +0200, Francesco wrote:
HI all, I've now downloaded and $make latest batman-adv trunk from $svn co http://downloads.open-mesh.net/svn/batman/trunk/batman-adv-kernelland/ I have issues with batctl trunk: $svn co http://downloads.open-mesh.net/svn/batman/trunk/batctl/
This unfortunately, does not work. The batctl sources has symbolic links into the batman directory for the files hash.c, list-batman.c, hash.h list-batman.h.
Check out the complete trunk and then you will have everything you need.
Andrew
Hi Andrew, it (obviously) worked, i got the trunk, make and make install batctl after batman-adv.
Weird that: $batctl o BATMAN disabled - primary interface not active
I just run this script:
insmod /home/francesco/trunk/batman-adv-kernelland/batman-adv.ko ifconfig wlan0 down iwconfig wlan0 mode ad-hoc essid BATMAN_NET echo wlan0 > /proc/net/batman-adv/interfaces ifconfig wlan0 up ifconfig bat0 192.168.1.1 up
What is wrong?
I did the same in two mesh nodes with the same result..
On Wed, Oct 21, 2009 at 9:05 AM, Andrew Lunn andrew@lunn.ch wrote:
On Wed, Oct 21, 2009 at 08:54:02AM +0200, Francesco wrote:
HI all, I've now downloaded and $make latest batman-adv trunk from $svn co http://downloads.open-mesh.net/svn/batman/trunk/batman-adv-kernelland/ I have issues with batctl trunk: $svn co http://downloads.open-mesh.net/svn/batman/trunk/batctl/
This unfortunately, does not work. The batctl sources has symbolic links into the batman directory for the files hash.c, list-batman.c, hash.h list-batman.h.
Check out the complete trunk and then you will have everything you need.
Andrew
B.A.T.M.A.N mailing list B.A.T.M.A.N@lists.open-mesh.net https://lists.open-mesh.net/mm/listinfo/b.a.t.m.a.n
Francesco wrote:
Hi Andrew, it (obviously) worked, i got the trunk, make and make install batctl after batman-adv.
Weird that: $batctl o BATMAN disabled - primary interface not active
I just run this script:
insmod /home/francesco/trunk/batman-adv-kernelland/batman-adv.ko ifconfig wlan0 down iwconfig wlan0 mode ad-hoc essid BATMAN_NET echo wlan0 > /proc/net/batman-adv/interfaces ifconfig wlan0 up ifconfig bat0 192.168.1.1 up
What is wrong?
I think that you found a bug in batman-adv. Please add the wlan0 interface to /proc/net/batman-adv/interfaces when wlan0 is up.
Best regards, Sven
Sven Eckelmann wrote:
I think that you found a bug in batman-adv. Please add the wlan0 interface to /proc/net/batman-adv/interfaces when wlan0 is up.
Ok, problem is that the interface starts as IF_INACTIVE, then switches to IF_TO_BE_ACTIVATED when you activate it with ifconfig, but never reaches IF_ACTIVE. It should get the IF_TO_BE_ACTIVATED status in hardif_activate_interface and the IF_ACTIVE in schedule_own_packet. I would guess that this function is never called in that situation.
Following called-by tree: - schedule_own_packet - send_outstanding_bat_packet - new_aggregated_packet (indirect over worker queue) - add_bat_packet_to_list - schedule_own_packet (who sees the loop here?) - schedule_forward_packet - hardif_add_interface
I would guess that either schedule_own_packet must be moved to hardif_activate_interface or the worker queue must be started otherwise when hardif_activate_interface is called in hard_if_event. I would leave the that to Marek as he knows what side effects are problematic in that situation
Best regards, Sven
The module gets in a inactive state when all interfaces are down. This stops the sending of new packets. When a interface gets activated again it must reactivate the module and then start the sending of new originator messages.
Signed-off-by: Sven Eckelmann sven.eckelmann@gmx.de --- batman-adv-kernelland/hard-interface.c | 7 +++++++ 1 files changed, 7 insertions(+), 0 deletions(-)
diff --git a/batman-adv-kernelland/hard-interface.c b/batman-adv-kernelland/hard-interface.c index f04d2db..f954f4e 100644 --- a/batman-adv-kernelland/hard-interface.c +++ b/batman-adv-kernelland/hard-interface.c @@ -425,6 +425,13 @@ static int hard_if_event(struct notifier_block *this, break; case NETDEV_UP: hardif_activate_interface(batman_if); + if ((atomic_read(&module_state) == MODULE_INACTIVE) && + (hardif_get_active_if_num() > 0)) { + activate_module(); + + /* begin sending originator messages on that interface */ + schedule_own_packet(batman_if); + } break; /* NETDEV_CHANGEADDR - mac address change - what are we doing here ? */ default:
Please try this revised patch - it should avoid a race condition in which we may end up with 2 schedule_own_packet() running at the same time.
The module gets in a inactive state when all interfaces are down. This stops the sending of new packets. When a interface gets activated again it must reactivate the module and then start the sending of new originator messages.
Signed-off-by: Sven Eckelmann sven.eckelmann@gmx.de Signed-off-by: Marek Lindner lindner_marek@yahoo.de --- diff --git a/batman-adv-kernelland/hard-interface.c b/batman-adv-kernelland/hard-interface.c index f04d2db..960e86c 100644 --- a/batman-adv-kernelland/hard-interface.c +++ b/batman-adv-kernelland/hard-interface.c @@ -425,6 +425,10 @@ static int hard_if_event(struct notifier_block *this, break; case NETDEV_UP: hardif_activate_interface(batman_if); + if ((atomic_read(&module_state) == MODULE_INACTIVE) && + (hardif_get_active_if_num() > 0)) { + activate_module(); + } break; /* NETDEV_CHANGEADDR - mac address change - what are we doing here ? */ default: diff --git a/batman-adv-kernelland/main.c b/batman-adv-kernelland/main.c index 8fbefc2..4e70c95 100644 --- a/batman-adv-kernelland/main.c +++ b/batman-adv-kernelland/main.c @@ -173,7 +173,7 @@ end: /* shuts down the whole module.*/ void shutdown_module(void) { - atomic_set(&module_state, MODULE_INACTIVE); + atomic_set(&module_state, MODULE_DEACTIVATING);
purge_outstanding_packets(); flush_workqueue(bat_event_workqueue); @@ -199,6 +199,7 @@ void shutdown_module(void)
hardif_remove_interfaces(); synchronize_rcu(); + atomic_set(&module_state, MODULE_INACTIVE); }
void inc_module_count(void) diff --git a/batman-adv-kernelland/main.h b/batman-adv-kernelland/main.h index 21d3919..9493f96 100644 --- a/batman-adv-kernelland/main.h +++ b/batman-adv-kernelland/main.h @@ -63,7 +63,7 @@
#define MODULE_INACTIVE 0 #define MODULE_ACTIVE 1 -/* #define MODULE_WAITING 2 -- not needed anymore */ +#define MODULE_DEACTIVATING 2
/* diff --git a/batman-adv-kernelland/send.c b/batman-adv-kernelland/send.c index 73fe371..01396b4 100644 --- a/batman-adv-kernelland/send.c +++ b/batman-adv-kernelland/send.c @@ -399,7 +399,7 @@ void send_outstanding_bcast_packet(struct work_struct *work) /* if we still have some more bcasts to send and we are not shutting * down */ if ((forw_packet->num_packets < 3) && - (atomic_read(&module_state) != MODULE_INACTIVE)) + (atomic_read(&module_state) != MODULE_DEACTIVATING)) _add_bcast_packet_to_list(forw_packet, ((5 * HZ) / 1000)); else forw_packet_free(forw_packet); @@ -424,7 +424,7 @@ void send_outstanding_bat_packet(struct work_struct *work) * shutting down */ if ((forw_packet->own) && - (atomic_read(&module_state) != MODULE_INACTIVE)) + (atomic_read(&module_state) != MODULE_DEACTIVATING)) schedule_own_packet(forw_packet->if_incoming);
forw_packet_free(forw_packet);
Hi all,
Add the wlan0 interface to /proc/net/batman-adv/interfaces when wlan0 is up actually seems to solve the problem.
Thanks.
On Wed, Oct 21, 2009 at 12:00 PM, Sven Eckelmann sven.eckelmann@gmx.dewrote:
Francesco wrote:
Hi Andrew, it (obviously) worked, i got the trunk, make and make install batctl
after
batman-adv.
Weird that: $batctl o BATMAN disabled - primary interface not active
I just run this script:
insmod /home/francesco/trunk/batman-adv-kernelland/batman-adv.ko ifconfig wlan0 down iwconfig wlan0 mode ad-hoc essid BATMAN_NET echo wlan0 > /proc/net/batman-adv/interfaces ifconfig wlan0 up ifconfig bat0 192.168.1.1 up
What is wrong?
I think that you found a bug in batman-adv. Please add the wlan0 interface to /proc/net/batman-adv/interfaces when wlan0 is up.
Best regards, Sven
B.A.T.M.A.N mailing list B.A.T.M.A.N@lists.open-mesh.net https://lists.open-mesh.net/mm/listinfo/b.a.t.m.a.n
Hi,
From the beginning, I am using batman-adv 3.2 compiled along with batctl.
batman-adv 3.2 ? I guess you mean 0.2 ?
$ insmod batman-adv.ko $ iwconfig wlan0 mode ad-hoc essid BATMAN_NET $ echo wlan0 > /proc/net/batman-adv/interfaces $ ifconfig wlan0 192.168.1.xxx up $ ifconfig bat0 up
The wlan0 device does not need an IP address. It needs to be up & running but the IP goes on the bat0 device.
I get a new bat0 interface, iwconfig shows nodes properly part of the ad-hoc network. I see correct $batctl o output but i cannot $batctl ping $mac_address
Would you mind showing us the batctl o output and the batctl ping call plus error message(s) ?
There are no bridges between interfaces. I just want to run batman-adv on wlan0. Do I need to set an IP address for wlan0, for bat0 or for both? Or I have to bridge bat0 and wlan0 together?
No, don't bridge bat0 and wlan0. Imagine you had another wifi interface in managed mode for non-batman nodes to connect to (e.g. wlan1). Then you would want to make the mesh accessible for the managed clients by bridging bat0 with wlan1.
Regards, Marek
b.a.t.m.a.n@lists.open-mesh.org
-
Andrew Lunn -
Francesco -
Marek Lindner -
Sven Eckelmann