Hi folks,
after the previous discussions how to proceed with BATMAN V we came to the conclusion that a runtime switch for changing the routing algorithms would be the best solution. Hence, I drafted a couple of patches that I'd like to get some feedback on. Note, that these patches are not ready to be included. For instance, the rcu locking isn't clean.
I'd like to focus your attention towards the following items:
* How do we design the algorithm registration. At the moment batman_init() calls bat_iv_init() and possibly other algorithms in the future but this is less than ideal. I had hoped to find a mechanism which allows an init function to be declared and called inside of the respective routing algorithm files. These files can be compiled into the module or not.
* The API and its implementation. It is highly likely that this API won't survive for very long as it has been designed to satisfy one routing algorithm only. However, changing it in the future should be easy.
* Other ideas / comments ?
Cheers, Marek
Signed-off-by: Marek Lindner lindner_marek@yahoo.de --- bat_debugfs.c | 8 ++++++ bat_iv_ogm.c | 9 ++++++ main.c | 75 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ main.h | 5 +++ soft-interface.c | 4 +++ types.h | 7 +++++ 6 files changed, 108 insertions(+), 0 deletions(-)
diff --git a/bat_debugfs.c b/bat_debugfs.c index d0af9bf..05e2f1f 100644 --- a/bat_debugfs.c +++ b/bat_debugfs.c @@ -221,6 +221,12 @@ static void debug_log_cleanup(struct bat_priv *bat_priv) } #endif
+static int bat_algorithms_open(struct inode *inode, struct file *file) +{ + struct net_device *net_dev = (struct net_device *)inode->i_private; + return single_open(file, bat_algo_seq_print_text, net_dev); +} + static int originators_open(struct inode *inode, struct file *file) { struct net_device *net_dev = (struct net_device *)inode->i_private; @@ -274,6 +280,7 @@ struct bat_debuginfo bat_debuginfo_##_name = { \ } \ };
+static BAT_DEBUGINFO(routing_algos, S_IRUGO, bat_algorithms_open); static BAT_DEBUGINFO(originators, S_IRUGO, originators_open); static BAT_DEBUGINFO(gateways, S_IRUGO, gateways_open); static BAT_DEBUGINFO(softif_neigh, S_IRUGO, softif_neigh_open); @@ -282,6 +289,7 @@ static BAT_DEBUGINFO(transtable_local, S_IRUGO, transtable_local_open); static BAT_DEBUGINFO(vis_data, S_IRUGO, vis_data_open);
static struct bat_debuginfo *mesh_debuginfos[] = { + &bat_debuginfo_routing_algos, &bat_debuginfo_originators, &bat_debuginfo_gateways, &bat_debuginfo_softif_neigh, diff --git a/bat_iv_ogm.c b/bat_iv_ogm.c index d60e1ba..a2b25ad 100644 --- a/bat_iv_ogm.c +++ b/bat_iv_ogm.c @@ -1169,3 +1169,12 @@ void bat_ogm_receive(const struct ethhdr *ethhdr, unsigned char *packet_buff, } while (bat_ogm_aggr_packet(buff_pos, packet_len, batman_ogm_packet->tt_num_changes)); } + +static struct bat_algo bat_algo_iv = { + .name = "BATMAN IV", +}; + +int __init bat_iv_init(void) +{ + return bat_algo_register(&bat_algo_iv); +} diff --git a/main.c b/main.c index fb87bdc..05b85f0 100644 --- a/main.c +++ b/main.c @@ -37,6 +37,7 @@ /* List manipulations on hardif_list have to be rtnl_lock()'ed, * list traversals just rcu-locked */ struct list_head hardif_list; +static struct hlist_head bat_algo_list;
unsigned char broadcast_addr[] = {0xff, 0xff, 0xff, 0xff, 0xff, 0xff};
@@ -45,6 +46,9 @@ struct workqueue_struct *bat_event_workqueue; static int __init batman_init(void) { INIT_LIST_HEAD(&hardif_list); + INIT_HLIST_HEAD(&bat_algo_list); + + bat_iv_init();
/* the name should not be longer than 10 chars - see * http://lwn.net/Articles/23634/ */ @@ -173,6 +177,77 @@ int is_my_mac(const uint8_t *addr)
}
+static struct bat_algo *bat_algo_get(char *name) +{ + struct bat_algo *bat_algo = NULL, *bat_algo_tmp; + struct hlist_node *node; + + rcu_read_lock(); + hlist_for_each_entry_rcu(bat_algo_tmp, node, &bat_algo_list, list) { + if (strcmp(bat_algo_tmp->name, name) != 0) + continue; + + bat_algo = bat_algo_tmp; + break; + } + rcu_read_unlock(); + + return bat_algo; +} + +int bat_algo_register(struct bat_algo *bat_algo) +{ + struct bat_algo *bat_algo_tmp; + int ret = -1; + + bat_algo_tmp = bat_algo_get(bat_algo->name); + if (bat_algo_tmp) { + pr_info("Trying to register already registered routing " + "algorithm: %s\n", bat_algo->name); + goto out; + } + + INIT_HLIST_NODE(&bat_algo->list); + hlist_add_head_rcu(&bat_algo->list, &bat_algo_list); + ret = 0; + +out: + return ret; +} + +int bat_algo_select(struct bat_priv *bat_priv, char *name) +{ + struct bat_algo *bat_algo; + int ret = -1; + + bat_algo = bat_algo_get(name); + if (!bat_algo) + goto out; + + /* TODO: must lock here - can we highjack an existing spinlock ? */ + rcu_assign_pointer(bat_priv->bat_algo, bat_algo); + ret = 0; + +out: + return ret; +} + +int bat_algo_seq_print_text(struct seq_file *seq, void *offset) +{ + struct bat_algo *bat_algo; + struct hlist_node *node; + + seq_printf(seq, "Available routing algorithms:\n"); + + rcu_read_lock(); + hlist_for_each_entry_rcu(bat_algo, node, &bat_algo_list, list) { + seq_printf(seq, "%s\n", bat_algo->name); + } + rcu_read_unlock(); + + return 0; +} + module_init(batman_init); module_exit(batman_exit);
diff --git a/main.h b/main.h index 464439f..1fa71d8 100644 --- a/main.h +++ b/main.h @@ -159,6 +159,11 @@ void mesh_free(struct net_device *soft_iface); void inc_module_count(void); void dec_module_count(void); int is_my_mac(const uint8_t *addr); +int bat_algo_register(struct bat_algo *bat_algo); +int bat_algo_select(struct bat_priv *bat_priv, char *name); +int bat_algo_seq_print_text(struct seq_file *seq, void *offset); + +int bat_iv_init(void);
#ifdef CONFIG_BATMAN_ADV_DEBUG int debug_log(struct bat_priv *bat_priv, const char *fmt, ...) __printf(2, 3); diff --git a/soft-interface.c b/soft-interface.c index bd8c7cf..a0046ac 100644 --- a/soft-interface.c +++ b/soft-interface.c @@ -855,6 +855,10 @@ struct net_device *softif_create(const char *name) bat_priv->primary_if = NULL; bat_priv->num_ifaces = 0;
+ ret = bat_algo_select(bat_priv, "BATMAN IV"); + if (ret < 0) + goto unreg_soft_iface; + ret = sysfs_add_meshif(soft_iface); if (ret < 0) goto unreg_soft_iface; diff --git a/types.h b/types.h index 35085f4..cfb0cc2 100644 --- a/types.h +++ b/types.h @@ -206,6 +206,7 @@ struct bat_priv { atomic_t gw_reselect; struct hard_iface __rcu *primary_if; /* rcu protected pointer */ struct vis_info *my_vis_info; + struct bat_algo __rcu *bat_algo; };
struct socket_client { @@ -344,4 +345,10 @@ struct softif_neigh { struct rcu_head rcu; };
+ +struct bat_algo { + struct hlist_node list; + char name[20]; +}; + #endif /* _NET_BATMAN_ADV_TYPES_H_ */
Signed-off-by: Marek Lindner lindner_marek@yahoo.de --- bat_iv_ogm.c | 21 +++++++++++++-------- bat_ogm.h | 35 ----------------------------------- hard-interface.c | 16 ++++++++++------ routing.c | 7 +++++-- send.c | 8 +++++--- types.h | 10 +++++++++- 6 files changed, 42 insertions(+), 55 deletions(-) delete mode 100644 bat_ogm.h
diff --git a/bat_iv_ogm.c b/bat_iv_ogm.c index a2b25ad..8e4c81f 100644 --- a/bat_iv_ogm.c +++ b/bat_iv_ogm.c @@ -20,7 +20,6 @@ */
#include "main.h" -#include "bat_ogm.h" #include "translation-table.h" #include "ring_buffer.h" #include "originator.h" @@ -30,7 +29,7 @@ #include "hard-interface.h" #include "send.h"
-void bat_ogm_init(struct hard_iface *hard_iface) +static void bat_ogm_init(struct hard_iface *hard_iface) { struct batman_ogm_packet *batman_ogm_packet;
@@ -47,7 +46,7 @@ void bat_ogm_init(struct hard_iface *hard_iface) batman_ogm_packet->ttvn = 0; }
-void bat_ogm_init_primary(struct hard_iface *hard_iface) +static void bat_ogm_init_primary(struct hard_iface *hard_iface) { struct batman_ogm_packet *batman_ogm_packet;
@@ -56,7 +55,7 @@ void bat_ogm_init_primary(struct hard_iface *hard_iface) batman_ogm_packet->header.ttl = TTL; }
-void bat_ogm_update_mac(struct hard_iface *hard_iface) +static void bat_ogm_update_mac(struct hard_iface *hard_iface) { struct batman_ogm_packet *batman_ogm_packet;
@@ -157,7 +156,7 @@ static void bat_ogm_send_to_if(struct forw_packet *forw_packet, }
/* send a batman ogm packet */ -void bat_ogm_emit(struct forw_packet *forw_packet) +static void bat_ogm_emit(struct forw_packet *forw_packet) { struct hard_iface *hard_iface; struct net_device *soft_iface; @@ -528,7 +527,7 @@ static void bat_ogm_forward(struct orig_node *orig_node, if_incoming, 0, bat_ogm_fwd_send_time()); }
-void bat_ogm_schedule(struct hard_iface *hard_iface, int tt_num_changes) +static void bat_ogm_schedule(struct hard_iface *hard_iface, int tt_num_changes) { struct bat_priv *bat_priv = netdev_priv(hard_iface->soft_iface); struct batman_ogm_packet *batman_ogm_packet; @@ -1140,7 +1139,7 @@ out: orig_node_free_ref(orig_node); }
-void bat_ogm_receive(const struct ethhdr *ethhdr, unsigned char *packet_buff, +static void bat_ogm_receive(const struct ethhdr *ethhdr, unsigned char *packet_buff, int packet_len, struct hard_iface *if_incoming) { struct batman_ogm_packet *batman_ogm_packet; @@ -1170,8 +1169,14 @@ void bat_ogm_receive(const struct ethhdr *ethhdr, unsigned char *packet_buff, batman_ogm_packet->tt_num_changes)); }
-static struct bat_algo bat_algo_iv = { +static struct bat_algo bat_algo_iv __read_mostly = { .name = "BATMAN IV", + .bat_ogm_init = bat_ogm_init, + .bat_ogm_init_primary = bat_ogm_init_primary, + .bat_ogm_update_mac = bat_ogm_update_mac, + .bat_ogm_schedule = bat_ogm_schedule, + .bat_ogm_emit = bat_ogm_emit, + .bat_ogm_receive = bat_ogm_receive, };
int __init bat_iv_init(void) diff --git a/bat_ogm.h b/bat_ogm.h deleted file mode 100644 index 69329c1..0000000 --- a/bat_ogm.h +++ /dev/null @@ -1,35 +0,0 @@ -/* - * Copyright (C) 2007-2011 B.A.T.M.A.N. contributors: - * - * Marek Lindner, Simon Wunderlich - * - * This program is free software; you can redistribute it and/or - * modify it under the terms of version 2 of the GNU General Public - * License as published by the Free Software Foundation. - * - * This program is distributed in the hope that it will be useful, but - * WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU - * General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program; if not, write to the Free Software - * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA - * 02110-1301, USA - * - */ - -#ifndef _NET_BATMAN_ADV_OGM_H_ -#define _NET_BATMAN_ADV_OGM_H_ - -#include "main.h" - -void bat_ogm_init(struct hard_iface *hard_iface); -void bat_ogm_init_primary(struct hard_iface *hard_iface); -void bat_ogm_update_mac(struct hard_iface *hard_iface); -void bat_ogm_schedule(struct hard_iface *hard_iface, int tt_num_changes); -void bat_ogm_emit(struct forw_packet *forw_packet); -void bat_ogm_receive(const struct ethhdr *ethhdr, unsigned char *packet_buff, - int packet_len, struct hard_iface *if_incoming); - -#endif /* _NET_BATMAN_ADV_OGM_H_ */ diff --git a/hard-interface.c b/hard-interface.c index d3e0e32..f4bc753 100644 --- a/hard-interface.c +++ b/hard-interface.c @@ -28,7 +28,6 @@ #include "bat_sysfs.h" #include "originator.h" #include "hash.h" -#include "bat_ogm.h"
#include <linux/if_arp.h>
@@ -147,7 +146,8 @@ static void primary_if_select(struct bat_priv *bat_priv, if (!new_hard_iface) return;
- bat_ogm_init_primary(new_hard_iface); + if (bat_priv->bat_algo->bat_ogm_init_primary) + bat_priv->bat_algo->bat_ogm_init_primary(new_hard_iface); primary_if_update_addr(bat_priv); }
@@ -233,7 +233,8 @@ static void hardif_activate_interface(struct hard_iface *hard_iface)
bat_priv = netdev_priv(hard_iface->soft_iface);
- bat_ogm_update_mac(hard_iface); + if (bat_priv->bat_algo->bat_ogm_update_mac) + bat_priv->bat_algo->bat_ogm_update_mac(hard_iface); hard_iface->if_status = IF_TO_BE_ACTIVATED;
/** @@ -307,7 +308,8 @@ int hardif_enable_interface(struct hard_iface *hard_iface, hard_iface->soft_iface = soft_iface; bat_priv = netdev_priv(hard_iface->soft_iface);
- bat_ogm_init(hard_iface); + if (bat_priv->bat_algo->bat_ogm_init) + bat_priv->bat_algo->bat_ogm_init(hard_iface);
if (!hard_iface->packet_buff) { bat_err(hard_iface->soft_iface, "Can't add interface packet " @@ -527,9 +529,11 @@ static int hard_if_event(struct notifier_block *this, goto hardif_put;
check_known_mac_addr(hard_iface->net_dev); - bat_ogm_update_mac(hard_iface); - bat_priv = netdev_priv(hard_iface->soft_iface); + + if (bat_priv->bat_algo->bat_ogm_update_mac) + bat_priv->bat_algo->bat_ogm_update_mac(hard_iface); + primary_if = primary_if_get_selected(bat_priv); if (!primary_if) goto hardif_put; diff --git a/routing.c b/routing.c index d5ddbd1..478b121 100644 --- a/routing.c +++ b/routing.c @@ -29,7 +29,6 @@ #include "originator.h" #include "vis.h" #include "unicast.h" -#include "bat_ogm.h"
void slide_own_bcast_window(struct hard_iface *hard_iface) { @@ -248,6 +247,7 @@ int window_protected(struct bat_priv *bat_priv, int32_t seq_num_diff,
int recv_bat_ogm_packet(struct sk_buff *skb, struct hard_iface *hard_iface) { + struct bat_priv *bat_priv = netdev_priv(hard_iface->soft_iface); struct ethhdr *ethhdr;
/* drop packet if it has not necessary minimum size */ @@ -274,7 +274,10 @@ int recv_bat_ogm_packet(struct sk_buff *skb, struct hard_iface *hard_iface)
ethhdr = (struct ethhdr *)skb_mac_header(skb);
- bat_ogm_receive(ethhdr, skb->data, skb_headlen(skb), hard_iface); + if (bat_priv->bat_algo->bat_ogm_receive) + bat_priv->bat_algo->bat_ogm_receive(ethhdr, skb->data, + skb_headlen(skb), + hard_iface);
kfree_skb(skb); return NET_RX_SUCCESS; diff --git a/send.c b/send.c index b00a0f5..a8a7fc9 100644 --- a/send.c +++ b/send.c @@ -28,7 +28,6 @@ #include "vis.h" #include "gateway_common.h" #include "originator.h" -#include "bat_ogm.h"
static void send_outstanding_bcast_packet(struct work_struct *work);
@@ -168,7 +167,9 @@ void schedule_bat_ogm(struct hard_iface *hard_iface) if (primary_if) hardif_free_ref(primary_if);
- bat_ogm_schedule(hard_iface, tt_num_changes); + if (bat_priv->bat_algo->bat_ogm_schedule) + bat_priv->bat_algo->bat_ogm_schedule(hard_iface, + tt_num_changes); }
static void forw_packet_free(struct forw_packet *forw_packet) @@ -318,7 +319,8 @@ void send_outstanding_bat_ogm_packet(struct work_struct *work) if (atomic_read(&bat_priv->mesh_state) == MESH_DEACTIVATING) goto out;
- bat_ogm_emit(forw_packet); + if (bat_priv->bat_algo->bat_ogm_emit) + bat_priv->bat_algo->bat_ogm_emit(forw_packet);
/** * we have to have at least one packet in the queue diff --git a/types.h b/types.h index cfb0cc2..e3bdb85 100644 --- a/types.h +++ b/types.h @@ -345,10 +345,18 @@ struct softif_neigh { struct rcu_head rcu; };
- struct bat_algo { struct hlist_node list; char name[20]; + void (*bat_ogm_init)(struct hard_iface *hard_iface); + void (*bat_ogm_init_primary)(struct hard_iface *hard_iface); + void (*bat_ogm_update_mac)(struct hard_iface *hard_iface); + void (*bat_ogm_schedule)(struct hard_iface *hard_iface, + int tt_num_changes); + void (*bat_ogm_emit)(struct forw_packet *forw_packet); + void (*bat_ogm_receive)(const struct ethhdr *ethhdr, + unsigned char *packet_buff, int packet_len, + struct hard_iface *if_incoming); };
#endif /* _NET_BATMAN_ADV_TYPES_H_ */
+static struct bat_algo bat_algo_iv __read_mostly = { .name = "BATMAN IV",
- .bat_ogm_init = bat_ogm_init,
- .bat_ogm_init_primary = bat_ogm_init_primary,
- .bat_ogm_update_mac = bat_ogm_update_mac,
- .bat_ogm_schedule = bat_ogm_schedule,
- .bat_ogm_emit = bat_ogm_emit,
- .bat_ogm_receive = bat_ogm_receive,
};
This needs an .owner = THIS_MODULE,
so you can do reference counting on the module.
Andrew
On Tuesday, November 29, 2011 03:02:45 Andrew Lunn wrote:
+static struct bat_algo bat_algo_iv __read_mostly = {
.name = "BATMAN IV",
- .bat_ogm_init = bat_ogm_init,
- .bat_ogm_init_primary = bat_ogm_init_primary,
- .bat_ogm_update_mac = bat_ogm_update_mac,
- .bat_ogm_schedule = bat_ogm_schedule,
- .bat_ogm_emit = bat_ogm_emit,
- .bat_ogm_receive = bat_ogm_receive,
};
This needs an .owner = THIS_MODULE,
so you can do reference counting on the module.
It is not a separate module ..
Cheers, Marek
On Tue, Nov 29, 2011 at 03:45:29AM +0800, Marek Lindner wrote:
On Tuesday, November 29, 2011 03:02:45 Andrew Lunn wrote:
+static struct bat_algo bat_algo_iv __read_mostly = {
.name = "BATMAN IV",
- .bat_ogm_init = bat_ogm_init,
- .bat_ogm_init_primary = bat_ogm_init_primary,
- .bat_ogm_update_mac = bat_ogm_update_mac,
- .bat_ogm_schedule = bat_ogm_schedule,
- .bat_ogm_emit = bat_ogm_emit,
- .bat_ogm_receive = bat_ogm_receive,
};
This needs an .owner = THIS_MODULE,
so you can do reference counting on the module.
It is not a separate module ..
Which leads to the question, why is it not a separate module?
Andrew
On Tuesday, November 29, 2011 14:09:24 Andrew Lunn wrote:
On Tue, Nov 29, 2011 at 03:45:29AM +0800, Marek Lindner wrote:
On Tuesday, November 29, 2011 03:02:45 Andrew Lunn wrote:
+static struct bat_algo bat_algo_iv __read_mostly = {
.name = "BATMAN IV",
- .bat_ogm_init = bat_ogm_init,
- .bat_ogm_init_primary = bat_ogm_init_primary,
- .bat_ogm_update_mac = bat_ogm_update_mac,
- .bat_ogm_schedule = bat_ogm_schedule,
- .bat_ogm_emit = bat_ogm_emit,
- .bat_ogm_receive = bat_ogm_receive,
};
This needs an .owner = THIS_MODULE,
so you can do reference counting on the module.
It is not a separate module ..
Which leads to the question, why is it not a separate module?
Why should it be ?
Cheers, Marek
On Tue, Nov 29, 2011 at 02:35:09PM +0800, Marek Lindner wrote:
On Tuesday, November 29, 2011 14:09:24 Andrew Lunn wrote:
On Tue, Nov 29, 2011 at 03:45:29AM +0800, Marek Lindner wrote:
On Tuesday, November 29, 2011 03:02:45 Andrew Lunn wrote:
+static struct bat_algo bat_algo_iv __read_mostly = {
.name = "BATMAN IV",
- .bat_ogm_init = bat_ogm_init,
- .bat_ogm_init_primary = bat_ogm_init_primary,
- .bat_ogm_update_mac = bat_ogm_update_mac,
- .bat_ogm_schedule = bat_ogm_schedule,
- .bat_ogm_emit = bat_ogm_emit,
- .bat_ogm_receive = bat_ogm_receive,
};
This needs an .owner = THIS_MODULE,
so you can do reference counting on the module.
It is not a separate module ..
Which leads to the question, why is it not a separate module?
Why should it be ?
I can think of a couple of reasons.
1) You might get asked when you post the code upstream, why you have the ability to register algorithms, but have not taken it further to allow modules to load algorithms. That seems to be the normal case in kernel code, registration implies modules. You probably need an argument ready why you don't want it.
2) It helps shows you have the split between the routing algorithm and the rest of the code correct. If you need to add lots of EXPORT_SYMBOL_GPL() for functions which the routing algorithm needs to call then maybe the abstraction is wrong?
Andrew
Hi Marek,
interesting approach, thanks for the patches! I don't really think that we need modules at this point, but we can easily extend it later - for example, net/mac80211/rate.c has a very similar API and works with modules.
A few comments to the API:
Generally, we should have a few lines per function in types.h to describe what each function is doing - we don't need that for a RFC patch, but that would be useful for the final version.
On Tue, Nov 29, 2011 at 12:28:36AM +0800, Marek Lindner wrote:
struct bat_algo { struct hlist_node list; char name[20];
you can also use a simple pointer here: char *name;
(the name will most probably static in the module anyways)
- void (*bat_ogm_init)(struct hard_iface *hard_iface);
- void (*bat_ogm_init_primary)(struct hard_iface *hard_iface);
- void (*bat_ogm_update_mac)(struct hard_iface *hard_iface);
- void (*bat_ogm_schedule)(struct hard_iface *hard_iface,
int tt_num_changes);
can't we put tt_num_changes somewhere bat_priv?
- void (*bat_ogm_emit)(struct forw_packet *forw_packet);
- void (*bat_ogm_receive)(const struct ethhdr *ethhdr,
unsigned char *packet_buff, int packet_len,struct hard_iface *if_incoming);
you can just pass the SKB instead of ethhdr, packet_buff, and packet_len -> they are derived from the skb anyway.
Then, you have can do the API with always a hard_iface as the first parameter. ;)
I also agree with Andrew on defining (a subset of) mandatory functions which every algo has to provide. The register function could then check and decline algos which don't define mandatory functions, and we don't have to check availability in the real code later ...
best regards, Simon
Hi,
Generally, we should have a few lines per function in types.h to describe what each function is doing - we don't need that for a RFC patch, but that would be useful for the final version.
ok, I will do that.
- void (*bat_ogm_init)(struct hard_iface *hard_iface);
- void (*bat_ogm_init_primary)(struct hard_iface *hard_iface);
- void (*bat_ogm_update_mac)(struct hard_iface *hard_iface);
- void (*bat_ogm_schedule)(struct hard_iface *hard_iface,
int tt_num_changes);can't we put tt_num_changes somewhere bat_priv?
tt_num_changes holds the number of changed TT entries since the last OGM. Obviously that changes all the time. Are you suggesting we store this value in bat_priv instead of a local variable ?
- void (*bat_ogm_emit)(struct forw_packet *forw_packet);
- void (*bat_ogm_receive)(const struct ethhdr *ethhdr,
unsigned char *packet_buff, int packet_len,struct hard_iface *if_incoming);you can just pass the SKB instead of ethhdr, packet_buff, and packet_len -> they are derived from the skb anyway.
Then, you have can do the API with always a hard_iface as the first parameter. ;)
I will send a separate patch for this API change in a minute as it is not really a part of the dynamic routing algo patchset.
Thanks for all the feedback!
Cheers, Marek
On Mon, Dec 05, 2011 at 04:00:48AM +0800, Marek Lindner wrote:
- void (*bat_ogm_init)(struct hard_iface *hard_iface);
- void (*bat_ogm_init_primary)(struct hard_iface *hard_iface);
- void (*bat_ogm_update_mac)(struct hard_iface *hard_iface);
- void (*bat_ogm_schedule)(struct hard_iface *hard_iface,
int tt_num_changes);can't we put tt_num_changes somewhere bat_priv?
tt_num_changes holds the number of changed TT entries since the last OGM. Obviously that changes all the time. Are you suggesting we store this value in bat_priv instead of a local variable ?
Yup. I know that it is changed all the time, just as tt_buff and tt_len changes - it is updated when the packet is generated.
There is not really a technical advantage (or disadvantage), it would just be more clean if information like this is stored in bat_priv IMHO - we fetch other stuff like the tt buffer or the ogm from structs like bat_priv or hard interfaces. And maybe we want to change something in this mechanism in the future.
Just my personal opinion ;) Simon
Signed-off-by: Marek Lindner lindner_marek@yahoo.de --- bat_sysfs.c | 35 +++++++++++++++++++++++++++++++++++ 1 files changed, 35 insertions(+), 0 deletions(-)
diff --git a/bat_sysfs.c b/bat_sysfs.c index c25492f..5375600 100644 --- a/bat_sysfs.c +++ b/bat_sysfs.c @@ -272,6 +272,39 @@ static ssize_t store_vis_mode(struct kobject *kobj, struct attribute *attr, return count; }
+static ssize_t show_bat_algo(struct kobject *kobj, struct attribute *attr, + char *buff) +{ + struct bat_priv *bat_priv = kobj_to_batpriv(kobj); + return sprintf(buff, "%s\n", bat_priv->bat_algo->name); +} + +static ssize_t store_bat_algo(struct kobject *kobj, struct attribute *attr, + char *buff, size_t count) +{ + struct net_device *net_dev = kobj_to_netdev(kobj); + struct bat_priv *bat_priv = netdev_priv(net_dev); + int ret; + + if (buff[count - 1] == '\n') + buff[count - 1] = '\0'; + else + buff[count] = '\0'; + + if (strcmp(bat_priv->bat_algo->name, buff) == 0) + return count; + + ret = bat_algo_select(bat_priv, buff); + if (ret < 0) + bat_info(net_dev, + "Invalid parameter for 'routing algorithm' setting " + "received: %s\n", buff); + else + bat_info(net_dev, "Changing routing algorithm to: %s\n", buff); + + return count; +} + static void post_gw_deselect(struct net_device *net_dev) { struct bat_priv *bat_priv = netdev_priv(net_dev); @@ -382,6 +415,7 @@ BAT_ATTR_BOOL(bonding, S_IRUGO | S_IWUSR, NULL); BAT_ATTR_BOOL(fragmentation, S_IRUGO | S_IWUSR, update_min_mtu); BAT_ATTR_BOOL(ap_isolation, S_IRUGO | S_IWUSR, NULL); static BAT_ATTR(vis_mode, S_IRUGO | S_IWUSR, show_vis_mode, store_vis_mode); +static BAT_ATTR(routing_algo, S_IRUGO | S_IWUSR, show_bat_algo, store_bat_algo); static BAT_ATTR(gw_mode, S_IRUGO | S_IWUSR, show_gw_mode, store_gw_mode); BAT_ATTR_UINT(orig_interval, S_IRUGO | S_IWUSR, 2 * JITTER, INT_MAX, NULL); BAT_ATTR_UINT(hop_penalty, S_IRUGO | S_IWUSR, 0, TQ_MAX_VALUE, NULL); @@ -399,6 +433,7 @@ static struct bat_attribute *mesh_attrs[] = { &bat_attr_fragmentation, &bat_attr_ap_isolation, &bat_attr_vis_mode, + &bat_attr_routing_algo, &bat_attr_gw_mode, &bat_attr_orig_interval, &bat_attr_hop_penalty,
Hi Marek
+static struct bat_algo *bat_algo_get(char *name) +{
- struct bat_algo *bat_algo = NULL, *bat_algo_tmp;
- struct hlist_node *node;
- rcu_read_lock();
- hlist_for_each_entry_rcu(bat_algo_tmp, node, &bat_algo_list, list) {
if (strcmp(bat_algo_tmp->name, name) != 0)continue;bat_algo = bat_algo_tmp;
You should have a try_module_get() here, to stop the module implementing the algorithm from being unloaded.
Andrew
Hi Marek
- How do we design the algorithm registration. At the moment batman_init()
calls bat_iv_init() and possibly other algorithms in the future but this is less than ideal. I had hoped to find a mechanism which allows an init function to be declared and called inside of the respective routing algorithm files. These files can be compiled into the module or not.
I would suggest looking at some other modular system. Generally the core offers a _register() and a _unregister() function. The module implementing an algorithm would call the _register function in its module_init function and the _unregister in module_exit function. It also makes sense to have a naming scheme for these modules. It then becomes possible to use hotplug to load algorithms dynamically, when userspace requests an algorithm which is not already loaded. I guess you first need an algorithm when the soft interface is taken up. If you don't have an algorithm loaded at that point, it makes sense to have a hard coded (or maybe menuconfig) default algorithm, which you try to load via hotplug.
- The API and its implementation. It is highly likely that this API won't
survive for very long as it has been designed to satisfy one routing algorithm only. However, changing it in the future should be easy.
It would be good to comment in the header file which functions can be NULL and which must be implemented. It would be best to assume that new methods added from now on will be NULL in older algorithms. There are a few options for handling this. Check the method is not NULL before calling it. Or the _register function can replace each NULL with a nop function and throw an BUG_ON() when a must have function is NULL.
Andrew
Andrew,
I would suggest looking at some other modular system. Generally the core offers a _register() and a _unregister() function. The module implementing an algorithm would call the _register function in its module_init function and the _unregister in module_exit function. It also makes sense to have a naming scheme for these modules. It then becomes possible to use hotplug to load algorithms dynamically, when userspace requests an algorithm which is not already loaded. I guess you first need an algorithm when the soft interface is taken up. If you don't have an algorithm loaded at that point, it makes sense to have a hard coded (or maybe menuconfig) default algorithm, which you try to load via hotplug.
not sure where we got out of alignment. If you check the patches you will see that it is a modular design. As I described in my initial mail, each routing algorithm is supposed to register itself. The functions / default algorithm / etc already there ..
And no, I am not going to implement a hotplugging system. At first I wanted a compile time option before I was convinced to work on a system which allows to switch the routing algorithm at runtime. This is as far as I will go. Feel free to propose your own patches if want this feature.
- The API and its implementation. It is highly likely that this API won't
survive for very long as it has been designed to satisfy one routing algorithm only. However, changing it in the future should be easy.
It would be good to comment in the header file which functions can be NULL and which must be implemented. It would be best to assume that new methods added from now on will be NULL in older algorithms. There are a few options for handling this. Check the method is not NULL before calling it. Or the _register function can replace each NULL with a nop function and throw an BUG_ON() when a must have function is NULL.
All functions are checked right now and all functions can be NULL. Check the patch #2 ...
Cheers, Marek
All functions are checked right now and all functions can be NULL. Check the patch #2 ...
This seems strange. What sort of routing protocol can you build, which has a NULL receive function? Only static routes. I would perform the check once in _register(), that there is a receive function, and never check it again. Save yourself a few cycles.
There is an LWN article about these ideas. Worth a read. I will post a link later.
Andrew
On Tuesday, November 29, 2011 14:18:10 Andrew Lunn wrote:
All functions are checked right now and all functions can be NULL. Check the patch #2 ...
This seems strange. What sort of routing protocol can you build, which has a NULL receive function? Only static routes. I would perform the check once in _register(), that there is a receive function, and never check it again. Save yourself a few cycles.
There is an LWN article about these ideas. Worth a read. I will post a link later.
Thanks. That will be helpful!
Cheers, Marek
There is an LWN article about these ideas. Worth a read. I will post a link later.
Here is the link, to part 1
https://lwn.net/Articles/444910/
and part 2
https://lwn.net/Articles/446317/
Andrew
b.a.t.m.a.n@lists.open-mesh.org
-
Andrew Lunn -
Marek Lindner -
Simon Wunderlich