B.A.T.M.A.N December 2009

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

22 participants
42 discussions

[B.A.T.M.A.N.] [PATCH] batman-adv: Use forw_bcast_list_lock always in disabled interrupt context
by Sven Eckelmann 17 Dec '09

17 Dec '09

forw_bcast_list_lock is spin_locked in both process and softirq context. SoftIRQ calls the spinlock with disabled IRQ and normal process context with enabled IRQs. When process context is inside an spin_locked area protected by forw_bcast_list_lock and gets interrupted by an IRQ, it could happen that something tries to lock forw_bcast_list_lock again in SoftIRQ context. It cannot proceed further since the lock is already taken somewhere else, but no reschedule will happen inside the SoftIRQ context. This leads to an complete kernel hang without any chance of resurrection. All functions called in process context must disable IRQs when they try to get get that lock to to prevent any reschedule due to IRQs. Signed-off-by: Sven Eckelmann <sven.eckelmann(a)gmx.de> --- batman-adv-kernelland/send.c | 19 +++++++++++-------- 1 files changed, 11 insertions(+), 8 deletions(-) diff --git a/batman-adv-kernelland/send.c b/batman-adv-kernelland/send.c index fc4953f..49b1534 100644 --- a/batman-adv-kernelland/send.c +++ b/batman-adv-kernelland/send.c @@ -338,12 +338,13 @@ static void forw_packet_free(struct forw_packet *forw_packet) static void _add_bcast_packet_to_list(struct forw_packet *forw_packet, unsigned long send_time) { + unsigned long flags; INIT_HLIST_NODE(&forw_packet->list); /* add new packet to packet list */ - spin_lock(&forw_bcast_list_lock); + spin_lock_irqsave(&forw_bcast_list_lock, flags); hlist_add_head(&forw_packet->list, &forw_bcast_list); - spin_unlock(&forw_bcast_list_lock); + spin_unlock_irqrestore(&forw_bcast_list_lock, flags); /* start timer for this packet */ INIT_DELAYED_WORK(&forw_packet->delayed_work, @@ -382,10 +383,11 @@ void send_outstanding_bcast_packet(struct work_struct *work) container_of(work, struct delayed_work, work); struct forw_packet *forw_packet = container_of(delayed_work, struct forw_packet, delayed_work); + unsigned long flags; - spin_lock(&forw_bcast_list_lock); + spin_lock_irqsave(&forw_bcast_list_lock, flags); hlist_del(&forw_packet->list); - spin_unlock(&forw_bcast_list_lock); + spin_unlock_irqrestore(&forw_bcast_list_lock, flags); /* rebroadcast packet */ rcu_read_lock(); @@ -436,24 +438,25 @@ void purge_outstanding_packets(void) { struct forw_packet *forw_packet; struct hlist_node *tmp_node, *safe_tmp_node; + unsigned long flags; bat_dbg(DBG_BATMAN, "purge_outstanding_packets()\n"); /* free bcast list */ - spin_lock(&forw_bcast_list_lock); + spin_lock_irqsave(&forw_bcast_list_lock, flags); hlist_for_each_entry_safe(forw_packet, tmp_node, safe_tmp_node, &forw_bcast_list, list) { - spin_unlock(&forw_bcast_list_lock); + spin_unlock_irqrestore(&forw_bcast_list_lock, flags); /** * send_outstanding_bcast_packet() will lock the list to * delete the item from the list */ cancel_delayed_work_sync(&forw_packet->delayed_work); - spin_lock(&forw_bcast_list_lock); + spin_lock_irqsave(&forw_bcast_list_lock, flags); } - spin_unlock(&forw_bcast_list_lock); + spin_unlock_irqrestore(&forw_bcast_list_lock, flags); /* free batman packet list */ spin_lock(&forw_bat_list_lock); -- 1.6.5.4

3 4

[B.A.T.M.A.N.] We made LWN!
by Andrew Lunn 17 Dec '09

17 Dec '09

Hi Folks For those of you with an LWN subscription, take a look at: http://lwn.net/Articles/366084/ We get a one line mention in the section about the 2.6.33 merge window :-) Andrew

1 0

[B.A.T.M.A.N.] Typo in change set 1501
by Gus Wirth 16 Dec '09

16 Dec '09

The fixes for typos in changeset 1501 missed one. On line 183 the "Neighbour" should be "Neighbor" to be consistent. It probably got missed because it is capitalized. Gus

2 1

[B.A.T.M.A.N.] required network to run
by Conner, Eric 16 Dec '09

16 Dec '09

I am new to ad-hoc mobile networking, and have a few questions. I am just a little confused as to where batman fits in. Why do I need an active network to attach the batmand to? If I'm in a network can't I see all the other nodes already? What is batman buying for me if I have to create my own ad-hoc network? If I don't have to create my own ad-hoc network, how do I set up my linux boxes so that they have an active network that will discover the nodes and allow the batmand to run? Thanks for your help Eric

1 0

[B.A.T.M.A.N.] [PATCH] [batman-debug4] Fix small typo "unkown" -> "unknown"
by Sven Eckelmann 16 Dec '09

16 Dec '09

Signed-off-by: Sven Eckelmann <sven.eckelmann(a)gmx.de> --- batman-debug4.pl | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/batman-debug4.pl b/batman-debug4.pl index b2f9f45..1322b32 100755 --- a/batman-debug4.pl +++ b/batman-debug4.pl @@ -100,7 +100,7 @@ while ( <BATMANLOG> ) { $receive_hash{ $last_orig }{ $last_neigh }{ "uni_flag" }++; - } elsif ( m/OGM\ via\ unkown\ neighbor/ ) { + } elsif ( m/OGM\ via\ unknown\ neighbor/ ) { $receive_hash{ $last_orig }{ $last_neigh }{ "unknown" }++; -- 1.6.5.4

2 1

[B.A.T.M.A.N.] Does batman-adv support gateway ?
by wu xiaofei 15 Dec '09

15 Dec '09

Does batman-adv support gateway ? Should all nodes' interfaces(batman-adv interface , non batman-adv interface) be in the same network(e.g. 192.168.1.0/24) or use "HNA" to announce different networks? (some nodes have multi interfaces) Olsrd can use "HNA" to announce 'non-olsrd interface' in different network from 'olsrd interface' , so other nodes know how to get to that network. How to config the IP address of a node with two interfaces running batman-adv at the same time? -- Wu

3 3

[B.A.T.M.A.N.] [PATCH] [batman-adv] Remove NULL pointer check before calling kfree
by Sven Eckelmann 14 Dec '09

14 Dec '09

It is safe to call kfree(NULL) which makes this extra check unneeded. It was found using checkpatch.pl from linux-2.6 Signed-off-by: Sven Eckelmann <sven.eckelmann(a)gmx.de> --- batman-adv-kernelland/hard-interface.c | 3 +-- 1 files changed, 1 insertions(+), 2 deletions(-) diff --git a/batman-adv-kernelland/hard-interface.c b/batman-adv-kernelland/hard-interface.c index e9cb977..28d02f6 100644 --- a/batman-adv-kernelland/hard-interface.c +++ b/batman-adv-kernelland/hard-interface.c @@ -399,8 +399,7 @@ int hardif_add_interface(char *dev, int if_num) return 1; out: - if (batman_if->packet_buff) - kfree(batman_if->packet_buff); + kfree(batman_if->packet_buff); kfree(batman_if); kfree(dev); return -1; -- 1.6.5.4

2 1

[B.A.T.M.A.N.] [PATCH] [batman-adv] Remove whitespace style errors
by Sven Eckelmann 14 Dec '09

14 Dec '09

checkpatch.pl found different smaller errors regarding trailing whitespaces or trailing extra newlines before EOF. These should be removed to follow linux-2.6 CodingStyle. Signed-off-by: Sven Eckelmann <sven.eckelmann(a)gmx.de> --- batman-adv-kernelland/CHANGELOG | 2 +- batman-adv-kernelland/originator.c | 2 -- batman-adv-kernelland/originator.h | 1 - batman-adv-kernelland/proc.c | 2 -- batman-adv-kernelland/routing.c | 1 - batman-adv-kernelland/routing.h | 6 +++--- batman-adv-kernelland/vis.c | 1 - 7 files changed, 4 insertions(+), 11 deletions(-) diff --git a/batman-adv-kernelland/CHANGELOG b/batman-adv-kernelland/CHANGELOG index 18d82f5..8a18163 100644 --- a/batman-adv-kernelland/CHANGELOG +++ b/batman-adv-kernelland/CHANGELOG @@ -19,7 +19,7 @@ batman-adv 0.1: * support latest kernels (2.6.20 - 2.6.28) * LOTS of cleanup: locking, stack usage, memory leaks * Change Ethertype from 0x0842 to 0x4305 - unregistered at IEEE, if you want to sponsor an official Ethertype ($2500) + unregistered at IEEE, if you want to sponsor an official Ethertype ($2500) please contact us -- Sun, 28 Dec 2008 00:44:31 +0100 diff --git a/batman-adv-kernelland/originator.c b/batman-adv-kernelland/originator.c index 9962af7..80106f7 100644 --- a/batman-adv-kernelland/originator.c +++ b/batman-adv-kernelland/originator.c @@ -239,5 +239,3 @@ void purge_orig(struct work_struct *work) start_purge_timer(); } - - diff --git a/batman-adv-kernelland/originator.h b/batman-adv-kernelland/originator.h index 6ef7a05..c1a9c2d 100644 --- a/batman-adv-kernelland/originator.h +++ b/batman-adv-kernelland/originator.h @@ -28,4 +28,3 @@ struct orig_node *get_orig_node(uint8_t *addr); struct neigh_node * create_neighbor(struct orig_node *orig_node, struct orig_node *orig_neigh_node, uint8_t *neigh, struct batman_if *if_incoming); - diff --git a/batman-adv-kernelland/proc.c b/batman-adv-kernelland/proc.c index ed4c734..f476d9c 100644 --- a/batman-adv-kernelland/proc.c +++ b/batman-adv-kernelland/proc.c @@ -827,5 +827,3 @@ int setup_procfs(void) return 0; } - - diff --git a/batman-adv-kernelland/routing.c b/batman-adv-kernelland/routing.c index f4be6be..2873468 100644 --- a/batman-adv-kernelland/routing.c +++ b/batman-adv-kernelland/routing.c @@ -1063,4 +1063,3 @@ void batman_data_ready(struct sock *sk, int len) atomic_set(&data_ready_cond, 1); wake_up_interruptible(&thread_wait); } - diff --git a/batman-adv-kernelland/routing.h b/batman-adv-kernelland/routing.h index 985728f..890a4f5 100644 --- a/batman-adv-kernelland/routing.h +++ b/batman-adv-kernelland/routing.h @@ -27,9 +27,9 @@ extern atomic_t exit_cond; void slide_own_bcast_window(struct batman_if *batman_if); void batman_data_ready(struct sock *sk, int len); int packet_recv_thread(void *data); -void receive_bat_packet(struct ethhdr *ethhdr, - struct batman_packet *batman_packet, - unsigned char *hna_buff, int hna_buff_len, +void receive_bat_packet(struct ethhdr *ethhdr, + struct batman_packet *batman_packet, + unsigned char *hna_buff, int hna_buff_len, struct batman_if *if_incoming); void update_routes(struct orig_node *orig_node, struct neigh_node *neigh_node, diff --git a/batman-adv-kernelland/vis.c b/batman-adv-kernelland/vis.c index cdb8aab..c26d48c 100644 --- a/batman-adv-kernelland/vis.c +++ b/batman-adv-kernelland/vis.c @@ -559,4 +559,3 @@ static void start_vis_timer(void) queue_delayed_work(bat_event_workqueue, &vis_timer_wq, (atomic_read(&vis_interval)/1000) * HZ); } - -- 1.6.5.4

2 1

[B.A.T.M.A.N.] [PATCH] [batctl] Fix array out of bound exception
by Linus Lüssing 14 Dec '09

14 Dec '09

This patch fixes an array out of bound exception in line 201 of functions.c. We need to reserve one extra character for the appended null character. Signed-off-by: Linus Lüssing <linus.luessing(a)web.de> --- Index: batctl/functions.c =================================================================== --- batctl/functions.c (revision 1491) +++ batctl/functions.c (working copy) @@ -161,10 +161,10 @@ if (read_opt & USE_READ_BUFF) { read_ptr = read_buff; - read_len = sizeof(read_buff); + read_len = sizeof(read_buff)-1; } else { read_ptr = lbuff; - read_len = sizeof(lbuff); + read_len = sizeof(lbuff)-1; } open:

2 1

[B.A.T.M.A.N.] wireshark-batman-adv: Displaying issue with vis-packets
by Linus Lüssing 14 Dec '09

14 Dec '09

The second originator field of a vis client update packet looks a bit odd in wireshark. Sven :)? (have a look at the attached cap-file for an example) I was also wondering about renaming this field maybe, in the beginning I got quite confused by seeing two identical originator fields in most vis packets. Maybe it should be renamed to previous sender or forwarding originator instead? Cheers, Linus

2 1

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

B.A.T.M.A.N December 2009