batman-adv: Fragment and send skbs larger than mtu

Non-broadcast packets larger than MTU are fragmented and sent with
an encapsulating header. Up to 16 fragments are supported, which are
sent in reverse order on the wire to allow minimal memory copying when
creating fragments.

Signed-off-by: Martin Hundebøll <martin@hundeboll.net>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Antonio Quartulli <antonio@meshcoding.com>

5 files changed, 154 insertions, 4 deletions

diff --git a/net/batman-adv/fragmentation.c b/net/batman-adv/fragmentation.c
index c829d3cee89c..271d321b3a04 100644
--- a/net/batman-adv/fragmentation.c
+++ b/net/batman-adv/fragmentation.c
@@ -368,3 +368,124 @@ out:
 		batadv_neigh_node_free_ref(neigh_node);
 	return ret;
 }
+
+/**
+ * batadv_frag_create - create a fragment from skb
+ * @skb: skb to create fragment from
+ * @frag_head: header to use in new fragment
+ * @mtu: size of new fragment
+ *
+ * Split the passed skb into two fragments: A new one with size matching the
+ * passed mtu and the old one with the rest. The new skb contains data from the
+ * tail of the old skb.
+ *
+ * Returns the new fragment, NULL on error.
+ */
+static struct sk_buff *batadv_frag_create(struct sk_buff *skb,
+					  struct batadv_frag_packet *frag_head,
+					  unsigned int mtu)
+{
+	struct sk_buff *skb_fragment;
+	unsigned header_size = sizeof(*frag_head);
+	unsigned fragment_size = mtu - header_size;
+
+	skb_fragment = netdev_alloc_skb(NULL, mtu + ETH_HLEN);
+	if (!skb_fragment)
+		goto err;
+
+	skb->priority = TC_PRIO_CONTROL;
+
+	/* Eat the last mtu-bytes of the skb */
+	skb_reserve(skb_fragment, header_size + ETH_HLEN);
+	skb_split(skb, skb_fragment, skb->len - fragment_size);
+
+	/* Add the header */
+	skb_push(skb_fragment, header_size);
+	memcpy(skb_fragment->data, frag_head, header_size);
+
+err:
+	return skb_fragment;
+}
+
+/**
+ * batadv_frag_send_packet - create up to 16 fragments from the passed skb
+ * @skb: skb to create fragments from
+ * @orig_node: final destination of the created fragments
+ * @neigh_node: next-hop of the created fragments
+ *
+ * Returns true on success, false otherwise.
+ */
+bool batadv_frag_send_packet(struct sk_buff *skb,
+			     struct batadv_orig_node *orig_node,
+			     struct batadv_neigh_node *neigh_node)
+{
+	struct batadv_priv *bat_priv;
+	struct batadv_hard_iface *primary_if;
+	struct batadv_frag_packet frag_header;
+	struct sk_buff *skb_fragment;
+	unsigned mtu = neigh_node->if_incoming->net_dev->mtu;
+	unsigned header_size = sizeof(frag_header);
+	unsigned max_fragment_size, max_packet_size;
+
+	/* To avoid merge and refragmentation at next-hops we never send
+	 * fragments larger than BATADV_FRAG_MAX_FRAG_SIZE
+	 */
+	mtu = min_t(unsigned, mtu, BATADV_FRAG_MAX_FRAG_SIZE);
+	max_fragment_size = (mtu - header_size - ETH_HLEN);
+	max_packet_size = max_fragment_size * BATADV_FRAG_MAX_FRAGMENTS;
+
+	/* Don't even try to fragment, if we need more than 16 fragments */
+	if (skb->len > max_packet_size)
+		goto out_err;
+
+	bat_priv = orig_node->bat_priv;
+	primary_if = batadv_primary_if_get_selected(bat_priv);
+	if (!primary_if)
+		goto out_err;
+
+	/* Create one header to be copied to all fragments */
+	frag_header.header.packet_type = BATADV_UNICAST_FRAG;
+	frag_header.header.version = BATADV_COMPAT_VERSION;
+	frag_header.header.ttl = BATADV_TTL;
+	frag_header.seqno = htons(atomic_inc_return(&bat_priv->frag_seqno));
+	frag_header.reserved = 0;
+	frag_header.no = 0;
+	frag_header.total_size = htons(skb->len);
+	memcpy(frag_header.orig, primary_if->net_dev->dev_addr, ETH_ALEN);
+	memcpy(frag_header.dest, orig_node->orig, ETH_ALEN);
+
+	/* Eat and send fragments from the tail of skb */
+	while (skb->len > max_fragment_size) {
+		skb_fragment = batadv_frag_create(skb, &frag_header, mtu);
+		if (!skb_fragment)
+			goto out_err;
+
+		batadv_inc_counter(bat_priv, BATADV_CNT_FRAG_TX);
+		batadv_add_counter(bat_priv, BATADV_CNT_FRAG_TX_BYTES,
+				   skb_fragment->len + ETH_HLEN);
+		batadv_send_skb_packet(skb_fragment, neigh_node->if_incoming,
+				       neigh_node->addr);
+		frag_header.no++;
+
+		/* The initial check in this function should cover this case */
+		if (frag_header.no == BATADV_FRAG_MAX_FRAGMENTS - 1)
+			goto out_err;
+	}
+
+	/* Make room for the fragment header. */
+	if (batadv_skb_head_push(skb, header_size) < 0 ||
+	    pskb_expand_head(skb, header_size + ETH_HLEN, 0, GFP_ATOMIC) < 0)
+		goto out_err;
+
+	memcpy(skb->data, &frag_header, header_size);
+
+	/* Send the last fragment */
+	batadv_inc_counter(bat_priv, BATADV_CNT_FRAG_TX);
+	batadv_add_counter(bat_priv, BATADV_CNT_FRAG_TX_BYTES,
+			   skb->len + ETH_HLEN);
+	batadv_send_skb_packet(skb, neigh_node->if_incoming, neigh_node->addr);
+
+	return true;
+out_err:
+	return false;
+}
diff --git a/net/batman-adv/fragmentation.h b/net/batman-adv/fragmentation.h
index 883a6f46005e..ca029e2676e7 100644
--- a/net/batman-adv/fragmentation.h
+++ b/net/batman-adv/fragmentation.h
@@ -27,6 +27,9 @@ bool batadv_frag_skb_fwd(struct sk_buff *skb,
 			 struct batadv_orig_node *orig_node_src);
 bool batadv_frag_skb_buffer(struct sk_buff **skb,
 			    struct batadv_orig_node *orig_node);
+bool batadv_frag_send_packet(struct sk_buff *skb,
+			     struct batadv_orig_node *orig_node,
+			     struct batadv_neigh_node *neigh_node);
 
 /**
  * batadv_frag_check_entry - check if a list of fragments has timed out
diff --git a/net/batman-adv/send.c b/net/batman-adv/send.c
index b8356ec067bf..1a1aa59d78ee 100644
--- a/net/batman-adv/send.c
+++ b/net/batman-adv/send.c
@@ -28,8 +28,7 @@
 #include "gateway_client.h"
 #include "originator.h"
 #include "network-coding.h"
-
-#include <linux/if_ether.h>
+#include "fragmentation.h"
 
 static void batadv_send_outstanding_bcast_packet(struct work_struct *work);
 
@@ -109,7 +108,19 @@ int batadv_send_skb_to_orig(struct sk_buff *skb,
 	/* batadv_find_router() increases neigh_nodes refcount if found. */
 	neigh_node = batadv_find_router(bat_priv, orig_node, recv_if);
 	if (!neigh_node)
-		return ret;
+		goto out;
+
+	/* Check if the skb is too large to send in one piece and fragment
+	 * it if needed.
+	 */
+	if (atomic_read(&bat_priv->fragmentation) &&
+	    skb->len > neigh_node->if_incoming->net_dev->mtu) {
+		/* Fragment and send packet. */
+		if (batadv_frag_send_packet(skb, orig_node, neigh_node))
+			ret = NET_XMIT_SUCCESS;
+
+		goto out;
+	}
 
 	/* try to network code the packet, if it is received on an interface
 	 * (i.e. being forwarded). If the packet originates from this node or if
@@ -123,7 +134,9 @@ int batadv_send_skb_to_orig(struct sk_buff *skb,
 		ret = NET_XMIT_SUCCESS;
 	}
 
-	batadv_neigh_node_free_ref(neigh_node);
+out:
+	if (neigh_node)
+		batadv_neigh_node_free_ref(neigh_node);
 
 	return ret;
 }
diff --git a/net/batman-adv/soft-interface.c b/net/batman-adv/soft-interface.c
index dd189e6bf05a..18b1fd915d1f 100644
--- a/net/batman-adv/soft-interface.c
+++ b/net/batman-adv/soft-interface.c
@@ -442,6 +442,7 @@ static void batadv_softif_destroy_finish(struct work_struct *work)
 static int batadv_softif_init_late(struct net_device *dev)
 {
 	struct batadv_priv *bat_priv;
+	uint32_t random_seqno;
 	int ret;
 	size_t cnt_len = sizeof(uint64_t) * BATADV_CNT_NUM;
 
@@ -491,6 +492,10 @@ static int batadv_softif_init_late(struct net_device *dev)
 	bat_priv->tt.last_changeset = NULL;
 	bat_priv->tt.last_changeset_len = 0;
 
+	/* randomize initial seqno to avoid collision */
+	get_random_bytes(&random_seqno, sizeof(random_seqno));
+	atomic_set(&bat_priv->frag_seqno, random_seqno);
+
 	bat_priv->primary_if = NULL;
 	bat_priv->num_ifaces = 0;
 
@@ -758,6 +763,8 @@ static const struct {
 	{ "mgmt_tx_bytes" },
 	{ "mgmt_rx" },
 	{ "mgmt_rx_bytes" },
+	{ "frag_tx" },
+	{ "frag_tx_bytes" },
 	{ "frag_rx" },
 	{ "frag_rx_bytes" },
 	{ "frag_fwd" },
diff --git a/net/batman-adv/types.h b/net/batman-adv/types.h
index 5a2cc7a9a620..d517d5dde6ad 100644
--- a/net/batman-adv/types.h
+++ b/net/batman-adv/types.h
@@ -300,6 +300,8 @@ struct batadv_bcast_duplist_entry {
  * @BATADV_CNT_MGMT_TX_BYTES: transmitted routing protocol traffic bytes counter
  * @BATADV_CNT_MGMT_RX: received routing protocol traffic packet counter
  * @BATADV_CNT_MGMT_RX_BYTES: received routing protocol traffic bytes counter
+ * @BATADV_CNT_FRAG_TX: transmitted fragment traffic packet counter
+ * @BATADV_CNT_FRAG_TX_BYTES: transmitted fragment traffic bytes counter
  * @BATADV_CNT_FRAG_RX: received fragment traffic packet counter
  * @BATADV_CNT_FRAG_RX_BYTES: received fragment traffic bytes counter
  * @BATADV_CNT_FRAG_FWD: forwarded fragment traffic packet counter
@@ -341,6 +343,8 @@ enum batadv_counters {
 	BATADV_CNT_MGMT_TX_BYTES,
 	BATADV_CNT_MGMT_RX,
 	BATADV_CNT_MGMT_RX_BYTES,
+	BATADV_CNT_FRAG_TX,
+	BATADV_CNT_FRAG_TX_BYTES,
 	BATADV_CNT_FRAG_RX,
 	BATADV_CNT_FRAG_RX_BYTES,
 	BATADV_CNT_FRAG_FWD,
@@ -542,6 +546,7 @@ struct batadv_priv_nc {
  * @aggregated_ogms: bool indicating whether OGM aggregation is enabled
  * @bonding: bool indicating whether traffic bonding is enabled
  * @fragmentation: bool indicating whether traffic fragmentation is enabled
+ * @frag_seqno: incremental counter to identify chains of egress fragments
  * @ap_isolation: bool indicating whether ap isolation is enabled
  * @bridge_loop_avoidance: bool indicating whether bridge loop avoidance is
  *  enabled
@@ -585,6 +590,7 @@ struct batadv_priv {
 	atomic_t aggregated_ogms;
 	atomic_t bonding;
 	atomic_t fragmentation;
+	atomic_t frag_seqno;
 	atomic_t ap_isolation;
 #ifdef CONFIG_BATMAN_ADV_BLA
 	atomic_t bridge_loop_avoidance;