Staging: batman-adv: layer2 unicast packet fragmentation

This patch implements a simple layer2 fragmentation to allow traffic
exchange over network interfaces with a MTU smaller than 1500 bytes. The
fragmentation splits the big packets into two parts and marks the frames
accordingly. The receiving end buffers the packets to reassemble the
orignal packet before passing it to the higher layers. This feature
makes it necessary to modify the batman-adv encapsulation for unicast
packets by adding a sequence number, flags and the originator address.
This modifcation is part of a seperate packet type for fragemented
packets to keep the original overhead as low as possible. This patch
enables the feature by default to ensure the data traffic can travel
through the network. But it also prints a warning to notify the user
about the performance implications.

Note: Fragmentation should be avoided at all costs since it has a
dramatic impact on the performance, especially when it comes wifi
networks. Instead of a single packet, 2 packets have to be sent! Not
only valuable airtime is wasted but also packetloss decreases the
throughput. A link with 50% packetloss and fragmentation enabled is
pretty much unusable.

Signed-off-by: Andreas Langer <an.langer@gmx.de>
[sven.eckelmann@gmx.de: Rework on top of current version]
Signed-off-by: Sven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

drivers/staging/batman-adv/bat_sysfs.c

@@ -134,6 +134,58 @@ static ssize_t store_bond(struct kobject *kobj, struct attribute *attr,
 	return count;
 }
 
+static ssize_t show_frag(struct kobject *kobj, struct attribute *attr,
+			     char *buff)
+{
+	struct device *dev = to_dev(kobj->parent);
+	struct bat_priv *bat_priv = netdev_priv(to_net_dev(dev));
+	int frag_status = atomic_read(&bat_priv->frag_enabled);
+
+	return sprintf(buff, "%s\n",
+		       frag_status == 0 ? "disabled" : "enabled");
+}
+
+static ssize_t store_frag(struct kobject *kobj, struct attribute *attr,
+			  char *buff, size_t count)
+{
+	struct device *dev = to_dev(kobj->parent);
+	struct net_device *net_dev = to_net_dev(dev);
+	struct bat_priv *bat_priv = netdev_priv(net_dev);
+	int frag_enabled_tmp = -1;
+
+	if (((count == 2) && (buff[0] == '1')) ||
+	    (strncmp(buff, "enable", 6) == 0))
+		frag_enabled_tmp = 1;
+
+	if (((count == 2) && (buff[0] == '0')) ||
+	    (strncmp(buff, "disable", 7) == 0))
+		frag_enabled_tmp = 0;
+
+	if (frag_enabled_tmp < 0) {
+		if (buff[count - 1] == '\n')
+			buff[count - 1] = '\0';
+
+		bat_err(net_dev,
+			"Invalid parameter for 'fragmentation' setting on mesh"
+			"received: %s\n", buff);
+		return -EINVAL;
+	}
+
+	if (atomic_read(&bat_priv->frag_enabled) == frag_enabled_tmp)
+		return count;
+
+	bat_info(net_dev, "Changing fragmentation from: %s to: %s\n",
+		 atomic_read(&bat_priv->frag_enabled) == 1 ?
+		 "enabled" : "disabled",
+		 frag_enabled_tmp == 1 ? "enabled" : "disabled");
+
+	atomic_set(&bat_priv->frag_enabled, (unsigned)frag_enabled_tmp);
+
+	update_min_mtu();
+
+	return count;
+}
+
 static ssize_t show_vis_mode(struct kobject *kobj, struct attribute *attr,
 			     char *buff)
 {
@@ -279,6 +331,7 @@ static ssize_t store_log_level(struct kobject *kobj, struct attribute *attr,
 static BAT_ATTR(aggregated_ogms, S_IRUGO | S_IWUSR,
 		show_aggr_ogms, store_aggr_ogms);
 static BAT_ATTR(bonding, S_IRUGO | S_IWUSR, show_bond, store_bond);
+static BAT_ATTR(fragmentation, S_IRUGO | S_IWUSR, show_frag, store_frag);
 static BAT_ATTR(vis_mode, S_IRUGO | S_IWUSR, show_vis_mode, store_vis_mode);
 static BAT_ATTR(orig_interval, S_IRUGO | S_IWUSR,
 		show_orig_interval, store_orig_interval);
@@ -289,6 +342,7 @@ static BAT_ATTR(log_level, S_IRUGO | S_IWUSR, show_log_level, store_log_level);
 static struct bat_attribute *mesh_attrs[] = {
 	&bat_attr_aggregated_ogms,
 	&bat_attr_bonding,
+	&bat_attr_fragmentation,
 	&bat_attr_vis_mode,
 	&bat_attr_orig_interval,
 #ifdef CONFIG_BATMAN_ADV_DEBUG
@@ -308,6 +362,7 @@ int sysfs_add_meshif(struct net_device *dev)
 		  routine as soon as we have it */
 	atomic_set(&bat_priv->aggregation_enabled, 1);
 	atomic_set(&bat_priv->bonding_enabled, 0);
+	atomic_set(&bat_priv->frag_enabled, 1);
 	atomic_set(&bat_priv->vis_mode, VIS_TYPE_CLIENT_UPDATE);
 	atomic_set(&bat_priv->orig_interval, 1000);
 	atomic_set(&bat_priv->log_level, 0);

drivers/staging/batman-adv/hard-interface.c

@@ -166,6 +166,11 @@ int hardif_min_mtu(void)
 	/* allow big frames if all devices are capable to do so
 	 * (have MTU > 1500 + BAT_HEADER_LEN) */
 	int min_mtu = ETH_DATA_LEN;
+	/* FIXME: each batman_if will be attached to a softif */
+	struct bat_priv *bat_priv = netdev_priv(soft_device);
+
+	if (atomic_read(&bat_priv->frag_enabled))
+		goto out;
 
 	rcu_read_lock();
 	list_for_each_entry_rcu(batman_if, &if_list, list) {
@@ -175,7 +180,7 @@ int hardif_min_mtu(void)
 				      min_mtu);
 	}
 	rcu_read_unlock();
-
+out:
 	return min_mtu;
 }
 
@@ -261,8 +266,30 @@ int hardif_enable_interface(struct batman_if *batman_if)
 	orig_hash_add_if(batman_if, bat_priv->num_ifaces);
 
 	atomic_set(&batman_if->seqno, 1);
+	atomic_set(&batman_if->frag_seqno, 1);
 	bat_info(soft_device, "Adding interface: %s\n", batman_if->dev);
 
+	if (atomic_read(&bat_priv->frag_enabled) && batman_if->net_dev->mtu <
+		ETH_DATA_LEN + BAT_HEADER_LEN)
+		bat_info(soft_device,
+			"The MTU of interface %s is too small (%i) to handle "
+			"the transport of batman-adv packets. Packets going "
+			"over this interface will be fragmented on layer2 "
+			"which could impact the performance. Setting the MTU "
+			"to %zi would solve the problem.\n",
+			batman_if->dev, batman_if->net_dev->mtu,
+			ETH_DATA_LEN + BAT_HEADER_LEN);
+
+	if (!atomic_read(&bat_priv->frag_enabled) && batman_if->net_dev->mtu <
+		ETH_DATA_LEN + BAT_HEADER_LEN)
+		bat_info(soft_device,
+			"The MTU of interface %s is too small (%i) to handle "
+			"the transport of batman-adv packets. If you experience"
+			" problems getting traffic through try increasing the "
+			"MTU to %zi.\n",
+			batman_if->dev, batman_if->net_dev->mtu,
+			ETH_DATA_LEN + BAT_HEADER_LEN);
+
 	if (hardif_is_iface_up(batman_if))
 		hardif_activate_interface(soft_device, bat_priv, batman_if);
 	else
@@ -495,6 +522,11 @@ int batman_skb_recv(struct sk_buff *skb, struct net_device *dev,
 		ret = recv_unicast_packet(skb, batman_if);
 		break;
 
+		/* fragmented unicast packet */
+	case BAT_UNICAST_FRAG:
+		ret = recv_ucast_frag_packet(skb, batman_if);
+		break;
+
 		/* broadcast packet */
 	case BAT_BCAST:
 		ret = recv_bcast_packet(skb);

drivers/staging/batman-adv/originator.c

@@ -27,6 +27,7 @@
 #include "translation-table.h"
 #include "routing.h"
 #include "hard-interface.h"
+#include "unicast.h"
 
 static DECLARE_DELAYED_WORK(purge_orig_wq, purge_orig);
 
@@ -95,6 +96,7 @@ static void free_orig_node(void *data)
 		kfree(neigh_node);
 	}
 
+	frag_list_free(&orig_node->frag_list);
 	hna_global_del_orig(orig_node, "originator timed out");
 
 	kfree(orig_node->bcast_own);
@@ -157,6 +159,10 @@ struct orig_node *get_orig_node(uint8_t *addr)
 
 	size = bat_priv->num_ifaces * sizeof(uint8_t);
 	orig_node->bcast_own_sum = kzalloc(size, GFP_ATOMIC);
+
+	INIT_LIST_HEAD(&orig_node->frag_list);
+	orig_node->last_frag_packet = 0;
+
 	if (!orig_node->bcast_own_sum)
 		goto free_bcast_own;
 
@@ -271,6 +277,10 @@ void purge_orig(struct work_struct *work)
 			hash_remove_bucket(orig_hash, &hashit);
 			free_orig_node(orig_node);
 		}
+
+		if (time_after(jiffies, (orig_node->last_frag_packet +
+			msecs_to_jiffies(FRAG_TIMEOUT))))
+			frag_list_free(&orig_node->frag_list);
 	}
 
 	spin_unlock_irqrestore(&orig_hash_lock, flags);

drivers/staging/batman-adv/packet.h

@@ -24,14 +24,15 @@
 
 #define ETH_P_BATMAN  0x4305	/* unofficial/not registered Ethertype */
 
-#define BAT_PACKET    0x01
-#define BAT_ICMP      0x02
-#define BAT_UNICAST   0x03
-#define BAT_BCAST     0x04
-#define BAT_VIS       0x05
+#define BAT_PACKET       0x01
+#define BAT_ICMP         0x02
+#define BAT_UNICAST      0x03
+#define BAT_BCAST        0x04
+#define BAT_VIS          0x05
+#define BAT_UNICAST_FRAG 0x06
 
 /* this file is included by batctl which needs these defines */
-#define COMPAT_VERSION 11
+#define COMPAT_VERSION 13
 #define DIRECTLINK 0x40
 #define VIS_SERVER 0x20
 #define PRIMARIES_FIRST_HOP 0x10
@@ -47,6 +48,9 @@
 #define VIS_TYPE_SERVER_SYNC		0
 #define VIS_TYPE_CLIENT_UPDATE		1
 
+/* fragmentation defines */
+#define UNI_FRAG_HEAD 0x01
+
 struct batman_packet {
 	uint8_t  packet_type;
 	uint8_t  version;  /* batman version field */
@@ -96,6 +100,16 @@ struct unicast_packet {
 	uint8_t  ttl;
 } __attribute__((packed));
 
+struct unicast_frag_packet {
+	uint8_t  packet_type;
+	uint8_t  version;  /* batman version field */
+	uint8_t  dest[6];
+	uint8_t  ttl;
+	uint8_t  flags;
+	uint8_t  orig[6];
+	uint16_t seqno;
+} __attribute__((packed));
+
 struct bcast_packet {
 	uint8_t  packet_type;
 	uint8_t  version;  /* batman version field */

drivers/staging/batman-adv/routing.c

@@ -32,6 +32,7 @@
 #include "ring_buffer.h"
 #include "vis.h"
 #include "aggregation.h"
+#include "unicast.h"
 
 static DECLARE_WAIT_QUEUE_HEAD(thread_wait);
 
@@ -1105,43 +1106,43 @@ struct neigh_node *find_router(struct orig_node *orig_node,
 	return router;
 }
 
-int recv_unicast_packet(struct sk_buff *skb, struct batman_if *recv_if)
+static int check_unicast_packet(struct sk_buff *skb, int hdr_size)
 {
-	struct unicast_packet *unicast_packet;
-	struct orig_node *orig_node;
-	struct neigh_node *router;
 	struct ethhdr *ethhdr;
-	struct batman_if *batman_if;
-	struct sk_buff *skb_old;
-	uint8_t dstaddr[ETH_ALEN];
-	int hdr_size = sizeof(struct unicast_packet);
-	unsigned long flags;
 
 	/* drop packet if it has not necessary minimum size */
 	if (skb_headlen(skb) < hdr_size)
-		return NET_RX_DROP;
+		return -1;
 
 	ethhdr = (struct ethhdr *) skb_mac_header(skb);
 
 	/* packet with unicast indication but broadcast recipient */
 	if (is_bcast(ethhdr->h_dest))
-		return NET_RX_DROP;
+		return -1;
 
 	/* packet with broadcast sender address */
 	if (is_bcast(ethhdr->h_source))
-		return NET_RX_DROP;
+		return -1;
 
 	/* not for me */
 	if (!is_my_mac(ethhdr->h_dest))
-		return NET_RX_DROP;
+		return -1;
 
-	unicast_packet = (struct unicast_packet *) skb->data;
+	return 0;
+}
 
-	/* packet for me */
-	if (is_my_mac(unicast_packet->dest)) {
-		interface_rx(skb, hdr_size);
-		return NET_RX_SUCCESS;
-	}
+static int route_unicast_packet(struct sk_buff *skb, struct batman_if *recv_if,
+		int hdr_size)
+{
+	struct orig_node *orig_node;
+	struct neigh_node *router;
+	struct batman_if *batman_if;
+	struct sk_buff *skb_old;
+	uint8_t dstaddr[ETH_ALEN];
+	unsigned long flags;
+	struct unicast_packet *unicast_packet =
+		(struct unicast_packet *) skb->data;
+	struct ethhdr *ethhdr = (struct ethhdr *) skb_mac_header(skb);
 
 	/* TTL exceeded */
 	if (unicast_packet->ttl < 2) {
@@ -1172,7 +1173,7 @@ int recv_unicast_packet(struct sk_buff *skb, struct batman_if *recv_if)
 	spin_unlock_irqrestore(&orig_hash_lock, flags);
 
 	/* create a copy of the skb, if needed, to modify it. */
-	if (!skb_clone_writable(skb, sizeof(struct unicast_packet))) {
+	if (!skb_clone_writable(skb, hdr_size)) {
 		skb_old = skb;
 		skb = skb_copy(skb, GFP_ATOMIC);
 		if (!skb)
@@ -1191,6 +1192,80 @@ int recv_unicast_packet(struct sk_buff *skb, struct batman_if *recv_if)
 	return NET_RX_SUCCESS;
 }
 
+int recv_unicast_packet(struct sk_buff *skb, struct batman_if *recv_if)
+{
+	struct unicast_packet *unicast_packet;
+	int hdr_size = sizeof(struct unicast_packet);
+
+	if (check_unicast_packet(skb, hdr_size) < 0)
+		return NET_RX_DROP;
+
+	unicast_packet = (struct unicast_packet *) skb->data;
+
+	/* packet for me */
+	if (is_my_mac(unicast_packet->dest)) {
+		interface_rx(skb, hdr_size);
+		return NET_RX_SUCCESS;
+	}
+
+	return route_unicast_packet(skb, recv_if, hdr_size);
+}
+
+int recv_ucast_frag_packet(struct sk_buff *skb, struct batman_if *recv_if)
+{
+	struct unicast_frag_packet *unicast_packet;
+	struct orig_node *orig_node;
+	struct frag_packet_list_entry *tmp_frag_entry;
+	int hdr_size = sizeof(struct unicast_frag_packet);
+	unsigned long flags;
+
+	if (check_unicast_packet(skb, hdr_size) < 0)
+		return NET_RX_DROP;
+
+	unicast_packet = (struct unicast_frag_packet *) skb->data;
+
+	/* packet for me */
+	if (is_my_mac(unicast_packet->dest)) {
+
+		spin_lock_irqsave(&orig_hash_lock, flags);
+		orig_node = ((struct orig_node *)
+			hash_find(orig_hash, unicast_packet->orig));
+
+		if (!orig_node) {
+			pr_warning("couldn't find orig node for "
+				"fragmentation\n");
+			spin_unlock_irqrestore(&orig_hash_lock, flags);
+			return NET_RX_DROP;
+		}
+
+		orig_node->last_frag_packet = jiffies;
+
+		if (list_empty(&orig_node->frag_list))
+			create_frag_buffer(&orig_node->frag_list);
+
+		tmp_frag_entry =
+			search_frag_packet(&orig_node->frag_list,
+			unicast_packet);
+
+		if (!tmp_frag_entry) {
+			create_frag_entry(&orig_node->frag_list, skb);
+			spin_unlock_irqrestore(&orig_hash_lock, flags);
+			return NET_RX_SUCCESS;
+		}
+
+		skb = merge_frag_packet(&orig_node->frag_list,
+			tmp_frag_entry, skb);
+		spin_unlock_irqrestore(&orig_hash_lock, flags);
+		if (!skb)
+			return NET_RX_DROP;
+
+		interface_rx(skb, hdr_size);
+		return NET_RX_SUCCESS;
+	}
+
+	return route_unicast_packet(skb, recv_if, hdr_size);
+}
+
 int recv_bcast_packet(struct sk_buff *skb)
 {
 	struct orig_node *orig_node;